Guide on Gender Analysis of Census Data

Full Draft of 6 December 2012

Contents

Introduction: About this Guide.............................................................................................2

Part One Background and Conceptual Clarifications for Gender Analysis of Census Data..4
Chapter 1: Gender in Population and Housing Censuses..................................................5
Chapter 2: Conceptual Clarifications on Gender Equality and Gender-Responsive Data 14
Part Two Key Gender Issues Analysed with Census Data 30
Chapter 3: Fertility.......................................................................................................... 32
Chapter 4: Mortality..........................................................................................................53
Chapter 5: Sex Ratio at Birth and throughout the Life Cycle.............................................62
Chapter 6: Marital Status, Polygamy, Widowhood and Child Marriage.............................80
Chapter 7: Households and Families..............................................................................107
Chapter 8: Income, Poverty and Living Conditions.........................................................139
Chapter 9: Education and Literacy..................................................................................160
Chapter 10: Work, Economic Activities and Social Protection.........................................172
Chapter 11: Migration.....................................................................................................188
Chapter 12: Disability......................................................................................................201
Conclusions.................................................................................................................... 220
References..................................................................................................................... 222
APPENDICES...........................................................................................................243
Appendix 1: Analysis of Census Questionnaires – Gender Issues Included in the 2005-
2014 Census Round.......................................................................................................243
Appendix 2: Mapping of Resources on Gender Statistics, Relevant to the Gender Analysis
of Census Data............................................................................................................... 254
Appendix 3: A Brief Overview of the Evolution of Gender Statistics................................256
Appendix 4: From Understanding the Gender Data Gap to Improving the Production and
Analysis of Gender Statistics..........................................................................................259
Appendix 5: How to Apply this Guide in a Country Context.............................................263

[Forewords and Acknowledgements]

1
Introduction: About this Guide

1. This guide is meant to serve as a tool for the staff of National Statistical Offices (NSOs) -
possibly in collaboration with academic or research institutions -, National Ministries
responsible for gender equality and women’s empowerment, and civil society gender
advocates, to be used in their efforts to promote equality, human rights and equity issues
between women and men through the appropriate analysis of census data. It is also expected
to be utilized by various United Nations Regional, Sub-regional and Country Offices in the
gender analysis of census data, to better support government partners in their formulation of
gender-responsive policies and programmes in all areas and all levels of government.

2. The guide has been produced by the United Nations Population Fund (UNFPA), in close
collaboration with UN Women, the United Nations Children’s Fund and the United Nations
Statistics Division, following two Needs Assessment Conferences for Census Analysis
(Dakar, Senegal in 2009 and Bávaro, Dominican Republic in 2010). As one NSO delegate
noted:

Gender is not systematically analysed [in censuses]. Many countries still do not
consistently disaggregate data by age and sex and do not distinguish between sex
disaggregated data and specific gender analysis from census data. There is a clear
need that countries ensure the production and provision of easy access of age and sex-
disaggregated data for users who require such information. Also, countries should
develop specific gender databases from their censuses and mainstream gender in the
entire national statistical system. UNFPA needs to build capacity for gender analysis
in the form of both technical guidelines and practical training of staff (Dakar
Conference Report, 2009: 8).

Besides responding to a clear national need, there are at least three additional rationales for
producing this guide:

a. Intrinsic Rationale

3. Promoting gender-responsive census analysis is in line with the human rights-based

approach and supports countries in fulfilling their human rights obligations.
 Gender differences and gender inequalities have often been overlooked in statistical
production even though it is now widely acknowledged that access and control over
resources, as well as the various dimensions of wellbeing, is gendered and that
women and men have different needs and interests. It follows that national statistics
that do not reflect these differences do not adequately reflect social reality in a way
that supports sound development policy-making and planning. Gender-responsive
census analysis can therefore contribute to making the national statistics base more
relevant and comprehensive, and improve decision-making for development planning
and programmes.
 Gender analysis focuses on gender-related social inequalities, discriminatory
practices and unjust power relations. Cross-referencing this information and going
beyond sex-disaggregation to consider other factors such as ethnicity, age, place of
residence can help identify additional social issues related to subsets of excluded or
vulnerable groups.

2
 b. Instrumental Rationale

4. Carrying out gender analysis of census data can contribute to better and more sustainable
human development outcomes.
 Gender analysis of census data helps build up the evidence-base informing
development policies and programmes in a way that takes into account the specific
needs of women and men, and girls and boys. Many Least Developed Countries
(LDCs) and countries undergoing humanitarian crises suffer from data scarcity and
planners find it difficult to assess what the population’s needs are. Where in addition
resources are limited, census data are a good place to start: It is sometimes the only
available source of national statistics and some relevant gender analyses can be
performed on the basis of census data, as this manual will show.
 Taking gender differences into account and overcoming gender inequality is an
obligation under international law (CEDAW) and not optional. International
guidelines exist on how gender statistics should be produced (BPFA, Strategic
Objective H.3). This is so because the international community now agrees that
gender equality is a prerequisite to advancing social justice and sustainable
development.

c. Institutional Rationale

5. Promoting gender analysis of census data is timely.

 The vast majority of countries have undertaken a census during the 2010 Population
and Housing Census round or are scheduled to do so soon. Although there are several
manuals on gender statistics, there is not, as yet, a comprehensive orientation on how
to analyse census data for gender purposes.
 Countries have repeatedly requested help in fulfilling their reporting obligations on
gender equality issues (e.g. CEDAW) and in producing thematic census reports.
Although the fact that censuses usually happen only once every ten years is a
limitation in this regard, it can also be interpreted as an opportunity that should be
taken advantage of whenever it presents itself. In particular, NSOs are often at a loss
when it comes to writing reports on gender statistics from censuses, and consequently
the census data on this and many other issues end up being under-utilized. The earlier
a guide on this issue becomes available, the better.

3
 Part One

Background and Conceptual Clarifications

for Gender Analysis of Census Data

4
Chapter 1:
Gender in Population and Housing Censuses
1. Scarcity of relevant, reliable and correctly analysed gender data is a challenge most
development planners and policy-makers have come up against. Yet, globally a lot of data
already have been collected: NGOs carry out needs assessments, university researchers do
surveys, governments take censuses and so forth. The data may not be easily accessible to the
public and may not have been analysed or presented in a format that is easy to be used. But
data often exist on a wide range of topics. It follows that many of the issues that are
commonly assumed to be lacking in information can in fact be examined with existing data.

2. Furthermore, it is now widely acknowledged as good practice to routinely disaggregate

statistical data on individual characteristics by sex and age. This not only provides evidence
of gender differences, but strengthens and improves the whole statistical system. Women and
men have different roles in society, different needs and interests and different access to and
control over resources. National statistics that do not reflect these differences are insufficient
and potentially misleading.

A. The Scope of Census Data

3. Census data are an important resource of data on gender issues. Almost all countries in the
world – including Least Developed Countries (LDCs) and many countries under crisis
conditions – carry out censuses to measure accurately the total number and key
characteristics of women and men, girls and boys in all geographical units of the country.
Indeed, most countries conduct one population census every 10 years, as recommended by
the UN’s Principles and Recommendations for Population and Housing Censuses, Revision 2
(Par. 1.12) (United Nations, 2008 a). Notable exceptions are eight European countries with
excellent administrative databases that allow them to collect the necessary data from these
systems, rather than through census field work.

4. This has not always been the case. Historically, censuses only enumerated male adults
because they existed primarily as a system for revenue control which enabled the rulers to
estimate the wealth of the country with considerable accuracy, and register their subjects for
taxation or military purposes (Dunn, 1940).

5. Censuses are among the most complex peace time operations undertaken by states.
Census-taking requires mapping the entire country, mobilizing and training vast numbers of
enumerators, conducting public information campaigns, collecting individual-level
information and processing millions of questionnaires, monitoring procedures and finally,
analysing the results.

6. Using systematic data collection methods, modern censuses aim to measure accurately the
total number and key characteristics of inhabitants for all administrative units of the country.
As a result, censuses provide universal population data and information on demographic and
social characteristics of the population, such as age, sex, place of usual residence, education
and training, employment and occupation, economic status, disability, migration, household
structure, etc. Censuses tell us about the relationships between household members,
educational attainment, economic activity status, occupation and housing arrangements. As
such, they constitute a key source of information for a variety of policy and research
purposes.

5
Text Box 1: Definitions – Population and Housing Census

A population census is the total process of collecting, compiling, evaluating, analysing and
publishing or otherwise disseminating demographic, economic and social data pertaining, at a
specified time, to all persons in a country or in a well-delimited part of a country.

A housing census is the total process of collecting, compiling, evaluating, analysing and
publishing or otherwise disseminating statistical data pertaining, at a specified time, to all
living quarters and occupants thereof in a country or in a well-delimited part of a country.

The essential features of population and housing censuses are individual enumeration,
universality within a defined territory, simultaneity and defined periodicity.

Population and housing censuses are a principal means of collecting basic population and
housing statistics as part of an integrated programme of data collection and compilation
aimed at providing a comprehensive source of statistical information for economic and social
development planning, for administrative purposes, for assessing conditions in human
settlements, for research and for commercial and other uses.

Principles and Recommendations for Population and Housing Censuses, Articles 1.4, 1.6, 1.8
and 1.20 (United Nations, 2008 a)

7. According to the United Nations Statistics Division, “two hundred twenty eight (228)
countries or areas have scheduled at least one census in the 2010 census round spanning the
period 2005 to 2014”. As of 1 November 2011, “171 have already conducted a census in this
census round. Six countries or areas have not yet scheduled any census until 2014. […] At the
end of the census round in 2014, 98.9 per cent of the world population will have been
enumerated (based on the current census schedule)”.1

Table 1: Progress in census-taking in the 2010 Census Round

Conducted a census Scheduled a census Not scheduled

All countries or
in the 2010 Round in the 2010 Round a census in the
areas
(2005-2012) (2013-2014) 2010 Round
World 235 209 19 7
Africa 58 45 10 3
Americas 52 51 1 0
Asia 50 41 5 4
Europe 50 47 3 0
Oceania 25 25 0 0

Source: Elaborated on the basis of data provided by UN Statistics Division (see footnote 1)

8. A census is one of the most important tools for policymakers. It takes stock of the most
1
http://unstats.un.org/unsd/demographic/sources/census/2010_PHC/censusclockmore.htm; last accessed on 3
December 2012. The data indicates that the majority of countries is succeeding in their census planning and tak-
ing. However, as conducting a census is a complex and costly process that requires great efforts in capacity
building, some countries and regions have been forced to delay or even cancel their censuses. Some of the chal -
lenges that countries are facing include: administrative organization, funding constraints, post-conflict situa-
tions, humanitarian crisis, natural hazards, etc.

6
important asset of a country – its human capital, women and men, girls and boys. Population
data gained from censuses, together with vital registration data and various kinds of
administrative records are critical for ensuring that appropriate policies and programmes are
prioritized at national and local levels.

Text Box 2: Uses of Census Data

Census data can serve

 Planning at the national, regional and local levels
 Decision-making for strategies or policies
 Policy and programme development, (monitoring) and evaluation
 Reporting obligations
 Administrative purposes

9. Therefore, censuses are a rich source of information about the differences between men
and women, girls and boys, or about the needs and requirements of population subgroups
such as elderly men in rural areas or adolescent girls. Their greatest advantage for the purpose
of gender analysis is that censuses allow for disaggregation down to the smallest
geographical unit. Regional NGOs and policy-makers in city councils, for instance, will be
able to extract data specifically on their region/city of interest and understand the population
composition within that restricted area.

10. Censuses can also provide basic national-level development indicators, for instance on
fertility and spatial distribution of men and women. For more complex indicators, census data
often serve as a denominator. They can for instance help uncover gender disparities in
employment, literacy and age of marriage. Where international definitions and classifications
are used, indicators derived from censuses are comparable among countries. Such indicators
can then be used to benchmark progress in achieving the Millennium Development Goals
(MDGs) or the ICPD Programme of Action and to monitor compliance with human rights
obligations such as CEDAW.

Methodology Box 1: Combining Data from Different Sources

The main advantage of census data is their universal coverage. The main drawback is the
generality of the information provided, which is usually lacking in detail for the purposes of
an in-depth gender analysis. However, census data may be combined with other sources to
examine many of the topics discussed in Part Two of this manual. Drawing on multiple data
sources enables one to carry out analyses that cannot be supported by census data alone. The
simplest strategy for doing this is to compute aggregate values for the relevant variables from
both sources separately at the level of relevant population groups. For example, one may be
interested in fertility preferences and income levels for women of different educational levels.
Since education is a variable included in both the census and the DHS, these indicators can be
computed separately using either data source, and the results can be compared. The main
limitations of this approach are that it only works for groups that can be defined in terms of
both data sources and that the number of such groups cannot be too large, as the DHS does
not allow much disaggregation.

In order to go beyond such simple comparisons of groups, one needs to integrate the two data
bases. To this end, there are two main strategies: construction of proxy variables and

7
statistical matching. The construction of proxy variables consists in developing a regression
model or other multivariate model based on the survey data and using explanatory variables
that are common to the survey and the census, to predict the value of the variable that one
would like to include in the census data base. The census value of the variable is then
constructed by using the same equation on the explanatory variables, as found in the census.
Typically, this approach has been used for the construction of household income data for
censuses that do not have this information, by regressing household characteristics such as
ownership of consumer durables or the quality of construction of the home on income data
from a Living Standards Measurement Survey or other kind of household survey that
provides income data (Elbers, Lanjouw and Lanjouw, 2002). The primary objective, in this
case, is to construct poverty estimates for smaller geographic areas than is feasible with the
income survey itself. But the approach is not necessarily limited to this application. In the
particular case mentioned above, one might predict desired family sizes based on, for
example, the age and number of living children, level of education and urban/rural residence
of the woman and then apply the same equation in the census, in order to relate the desired
number of children to typical census variables.

In the statistical matching or “data borrowing” approach, one uses the variables that are
common to the census and the survey to construct a measure of similarity or distance between
individual cases of the census and survey files. Each individual case found in the census is
then matched to its closest neighbour in the survey file. In some cases one may want to divide
the data into different subsets, in order to avoid, for example, the matching of men to women
or persons from very different parts of the country. The survey data of the closest neighbour
are then simply imputed to the individual census records.

When a survey is done shortly after a census it may be possible to establish a match between
census records and survey records on the basis of common geographical identifiers. Since
surveys typically use a census-based master sample frame such a match is technically quite
feasible, as long as the time interval between census and survey is not too long (say, less than
2-3 years). After appending the two data sets the desired survey variables can be estimated for
households or persons that were not covered by the survey on the basis of the relationships
found amongst those records where both census and survey data is available.

Both methods are not without their pitfalls and complications. Both the construction of
proxies and the statistical matching approach assume that once the common variables have
been controlled, the remaining variables from the survey are statistically independent from
those in the census. The fact that this is often not the case may introduce systematic biases. A
number of procedures have been proposed in the literature to deal with this problem (e.g.
Rubin, 1986; Moriarty and Scheuren, 2001). Because of these complications, either of the
two main strategies should not be applied without calling in appropriate technical support.

B. Gender Analysis of Census Data

11. Gender analysis as a way of interpreting census data has emerged in response to growing
need of gender information of countries, e.g. to report on progress made in terms of gender
equality and the empowerment of women in line with international obligations. As a way of
working with data, gender analysis is more than simply analysing quantitative data by ‘sex’
using standard descriptive statistical techniques. Gender analysis includes a gender-
responsive selection of questions to be posed to the data and in the interpretation of sex-

8
disaggregated data in the context of power relations between the sexes, i.e. in a way that
includes other sources of knowledge such as qualitative data, knowledge of cultural factors,
or further multivariate analyses shedding light on socio-economic realities.

Text Box 3: What is Gender Analysis?

Gender analysis is a critical examination of how differences in gender roles, activities,

needs, opportunities and rights/entitlements affect men, women, girls and boys in certain
situation or contexts. Gender analysis examines the relationships between females and
males and their access to and control of resources and the constraints they face relative to
each other.

Source: “Gender Equality, UN Coherence & You – Glossary: Definitions A-Z)

12. Gender analysis does the following:

 Critically examines the differences in women's and men's lives;

 Searches for the underlying causes of inequality between women and men and boys
and girls;
 Highlights gender-specific variables and is generally (though not exclusively) used to
achieve positive change for women and girls.

13. Gender analysis illuminates the extent to which:

 Women's and men's lives and therefore experiences, needs, interests, priorities, and
capacities are different.
 Women's lives are not all the same – each woman’s life is also shaped by a host of
other social characteristics such as ethnicity, religion, income level, immigration
status, sexual orientation, age, etc. The same holds for men.
 Women's life experiences, needs, issues and priorities are different for different
groups of women.
 Men and women have triple roles with regards to work:
o Reproductive work: including household maintenance and childrearing;
o Productive work: generating income or goods;
o Community work: activities in the public sphere undertaken for the good of
the community.
 To a much larger extent than is the case for men, the work of women is often unpaid.

14. Gender analysis goes beyond interpreting data. As part of gender-mainstreaming (see Par.
30), it is also a practical, programmatic tool that seeks to be participatory and holistic. Gender
analysis should place great importance on empowerment, consultation and participation of
those concerned. In addition, a comprehensive gender analysis of census data may require
multivariate techniques that go beyond the usual practices of NSOs and that require the
involvement of academic or research institutions. Therefore, gender analysis should not be
carried out by National Statistics Offices (NSOs) in isolation. NSOs can identify, plan,
implement, monitor and evaluate gender equality with data analysis projects – for example a
publication on the status of Women and Men in their countries – with a) representatives of

9
women’s machineries (Ministry for Family/Women; national CEDAW committee ...) and b)
with representatives of Civil Society (women’s movement, NGOs), ideally including
community members themselves who can testify as to the lived experiences of women and
men in the county. This type of publication can then inform national planning and policy
development initiatives.

15. The following sums up some of the strengths of census data with regards to gender
analysis (Meena and Chaudhury, 2010; Schkolnik, 2011):

a) Censuses provide a basic set of sex-disaggregated data at the smallest

geographical level. Thanks to their universal geographical coverage and routine
inclusion of demographic categories such as age, sex and marital status, censuses bear
great potential for disaggregation. Censuses provide data on the entire resident population
in a country at a given reference period by administrative area. They can thus inform local
planners about population composition and characteristics for instance, how many
widows, adolescent girls or polygamous families are living in a given administrative
area? Thus, census data are key to identifying “vulnerable groups” for targeted
interventions (e.g. rural vs. urban adolescent girls). Moreover, the fact that census
mapping culminates in the delimitation of the entire national territory into small
enumeration areas also means that censuses are used to demarcate constituencies – a basic
requirement for election processes, ensuring representation based on accurate numbers
and social participation. In the event of natural disaster or human-made crisis, more
realistic estimates of the women and men, girls and boys affected can be reconstituted,
using information on the enumeration areas affected.

b) Censuses provide insights into the private and community spheres and
(indirectly) into time-use of women and girls, men and boys. Feminists have long
criticized that the public (=male) vs. private (=female) dichotomy allows government to
clean its hands of responsibility for the state of the ‘private’ world. The so-called private
sphere (sexuality, reproduction, gender relations including gender-based violence,
women’s unpaid care- and housework, etc.) is a notoriously sensitive issue and often
under-studied. By entering into households and providing details on household and
housing characteristics as well as on social infrastructure, the real-life living conditions of
girls and boys, men and women including elements of vulnerability are exposed. What
does it mean for a widow to head a household composed of herself and her orphaned
grandchildren? What impact does the lack of a water source and access to
telecommunications have on girls’ education? What does it mean for women of
reproductive age to live in a locality with limited vital social infrastructure, such as health
facilities, schools, churches, community halls, markets and roads? If analysed with a
gender lens, one can learn a lot from censuses about gendered differentials in access to
resources and services.

c) Census data for advocacy: A local-level “early warning system” on gender

inequalities? Analyses of census data may uncover sex ratio imbalances or
unconventional population structures that are symptomatic of growing inequalities in a
country, region or municipality. In Viet Nam, for instance, gender advocates are alarmed
about the decreasing number of girls in some provinces. As census results are usually
published and widely disseminated for development planning, they can – if properly
analysed – serve media, civil society actors, NGOs, researchers and individuals in their
advocacy efforts, as a recognized and official source of information. Note however, that

10
 although the published reports from censuses may be easily available, they often provide
only aggregate information. The raw data files, which should enable breakdown of data to
smaller units, are often not easily accessible, especially to non-State actors (more on this
issue below). At worst, they may be so poorly conserved that they are no longer
retrievable or they may be rendered inaccessible for political reasons, for instance in
multi-ethnic countries.

d) Censuses provide essential background information allowing for further

research on women and men, girls and boys. Most importantly, the census lays the
groundwork for population projections (e.g. how many boys and girls will need schooling
in 2025?) by providing details on the key elements of population dynamics, i.e. fertility,
mortality and migration. Second, most surveys (e.g. labour force surveys or studies on
maternal mortality) draw their samples from master sampling frames provided by the
most recent census.

16. Among the weaknesses of census data for the purpose of gender analysis, the following
can be highlighted (Meena and Chaudhury, 2010; Schkolnik, 2011):

a) Census data may not have been produced in a gender-responsive way: In most
countries, statisticians without specific training in gender are responsible for producing
census questionnaires, defining concepts, variables and classifications and for managing
the field operations including enumerator training. As a consequence, the data collected in
censuses may not lend themselves easily to gender analysis but may in fact already be
gender-biased (ECLAC, 2006 a). For instance, the concept of “head of household” is
problematic in several ways, but in particular where question wording (or indeed, an
enumerator) refers to the head of household as “he”, respondents are likely to underreport
on female-headed households.

b) Census data are of very limited scope and depth: Census data do not provide all the
information needed for gender analysis. For instance, census questionnaires do not
generally ask (and indeed, given their objectives and constraints, cannot be expected to
ask) about issues such as women’s unpaid domestic work or gender-based discrimination
in public decision-making. Nor do they ask questions about fertility preferences, time-use,
sexual behaviour and many other gender-relevant issues. Questions on gender-based
violence (GBV) require specific ethnical and safety standards in data collection to protect
the victims, making their inclusion in censuses unrealistic. The mode of census
enumeration, calling upon a very high number of interviewers, allows neither a rigorous
selection of high-level personnel nor a thorough training that would adequately protect
the respondents. In addition, the population may be reluctant to answer a long
questionnaire and might feel that asking detailed questions on sensitive issues (income,
ethnicity, etc.) is an intrusion into privacy. Census offices are generally reluctant to
increase the number of questions on census forms, especially if the issues may elicit
controversy. Not only does each additional question imply a substantial cost increase, but
there is also a risk that it will deteriorate the quality of the core information. Where
information is collected on maternal mortality, time-use or violence, it is therefore often
neither sufficiently detailed nor accurate.

c) Gender-related discrimination is not explicitly measured by censuses: Census

data do not reveal information related to, for example, behavioural consequences of laws
or policy. Thus, one of the key tenets of gender analysis – that women, men, girls and

11
boys have different needs and are affected differently by policies and programmes –
cannot be examined on the basis of census data. Linked to this, it must be acknowledged
that census data can also be misleading about gender relations depending on how
questionnaires are designed and administered because respondents may be influenced by
gender-related power dynamics. There are a number of good practices regarding gender
responsive questionnaires and training of enumerators.

d) The level of analysis for census data is sex, not gender while policy interest tends
to be on the gender differentiated needs of men and women, the relational socio-cultural
construct, not sex, the biological concept (see Chapter 2.A for extended definitions). As
sex-disagreggation is merely a first step to making gender-based analyses, additional
effort is needed to unearth women’s and men’s different needs and aspirations as well as
the power differentials and relational factors that explain women’s and men’s access to
resources and services. For instance, the gender pay gap is a measure of earnings
differentials between women and men. Even if censuses ask about individual income
(many don't), this only tells us how much, in monetary terms, women take out of
employment compared with men, as a male/female difference expressed as a percentage
of male earnings. To make more specific statements about gender and inform policy-
making, this indicator not only has to be further disaggregated – by age groups, by
occupation, by part-time, full-time, etc. – but other factors have to be considered such as
the availability of child care, social norms about child-rearing and female employment,
and the gendered division of labour in routine housework. While the census can be
helpful in the former, it does facilitate the latter.

e) The census data may be outdated or of low quality (e.g. due to underreporting on
women): Many countries, particularly countries prone to humanitarian crises, can not
respect the 10-year interval for census-taking. Even in those countries that do, the last
census may be several years old and its figures may no longer reflect the lived reality of
women and men, girls and boys. Complex projection and estimation techniques are
needed to estimate the actual situation on the ground. This is not the case in those
countries (especially in the European Union) that rely heavily on continuously updated
population registers. In terms of data quality, underreporting on women is a well-
documented phenomenon, especially in countries like China where female births are often
hidden to get around the official one child policy. In some parts of South Asia, unmarried
women are less likely to be counted, whereas the under-counting of young male adults
and young babies of either sex is widespread in much of the world. Female household
headship, numbers of children under age 5, and numbers of young male migrants are also
routinely under-reported, whereas age data by single ages may be inaccurate. Finally,
many censuses suffer from incomplete coverage such that the results have to be adjusted
before publication. Under such circumstances, the raw data files contain information that
is different from the adjusted and published information, rendering reconstruction of
information for smaller geographical units problematic. In particular, the imputation of
missing data or the correction of inconsistencies in the information may be done
according to criteria that are not gender-neutral and that, in some cases, actually introduce
serious distortions.

f) Data access and the capacity to analyse census data in the appropriate ways may
be problematic: Census data bases are typically much larger than the data sets produced
by most surveys, making their analysis more difficult. Moreover, NSOs are generally not
at liberty to distribute them to potential users in their raw, unabridged form, due to

12
 problems of data confidentiality. This is very different from the situation of, for example,
the Demographic Health Surveys, most of which are easily accessible to individual
researchers. To deal with this situation, NSOs typically adopt one of three strategies: i)
They analyse as much of the data as they can in-house; ii) They prepare user samples of
1%, 5% or 10% that have been processed so as to make the identification of individual
households impossible, for use by academic and other research institutions; or iii) They
distribute the data to the general public in the form of data bases such as REDATAM
which allow users to prepare their own tables without having direct access to the micro-
data. Each of these strategies has potential limitations. If the gender analysis of census
data is carried out in-house, it will usually be guided by the need to produce certain
essential tables, but depending on the analytical capacity of the NSO, it will often not go
into in-depth studies of particular relationships, particularly if they involve multivariate
analysis. Preparing user samples can be costly2 and may run into limitations on the user
end if users need to produce detailed analyses of very specific population groups.
Information management systems such as REDATAM, on the other hand, are extremely
useful for the flexible production of tables based on the entire population, but they
generally do not allow for multivariate data analysis.

International Conventions and Conferences cited:

 International Conference on Population and Development (ICPD, 1994)
 Beijing Platform for Action (BPA, Fourth World Conference on Women, 1995)
 Convention on the Elimination of All Forms of Discrimination Against Women
(CEDAW, 1979)
Websites consulted:
 United Nations Millennium Declaration (2000) on
http://www.un.org/millennium/declaration/ares552e.htm
 Millennium Development Goals (MDGs) on http://www.un.org/millenniumgoals/
 Gender Equality, UN Coherence & You – Glossary: Definitions A-Z on
http://www.unicef.org/gender/training/content/scoIndex.html
 2010 World Population and Housing Census Programme (UN Statistics Division) on
http://unstats.un.org/unsd/demographic/sources/census/censusquest.htm
 UNFPA Census Portal on http://www.unfpa.org/public/op/edit/home/sitemap/pid/6734


2
If the cost is too large for the NSO, it may opt for distributing the information through the IPUMS programme
of the University of Minnesota, which designs user samples for release to the public, guided by the specifica-
tions provided by NSOs.

13
Chapter 2:
Conceptual Clarifications on Gender Equality and Gender-
Responsive Data Analysis
17. The following reviews some core concepts that sometimes lead to misunderstandings
between producers and users of data. By pointing out where differences in meaning exist
between the common statistical usage and that of the gender literature and by offering a
shared definition for the purpose of this manual, dialogue will be enhanced in view of a
shared goal: Making statistics reflect all the national population and measuring progress
towards gender equality.

A. Sex and Gender

18. In its most basic meaning, the concept of gender helps us understand how biological
differences between men and women (sex) acquire cultural and social meanings and produce
identities, differences, and inequalities in a given setting (gender). Sex characteristics at birth
are universal. By contrast, gender refers to socio-cultural differences and social relationships
between women and men that can change, over time for the same individual, and differ
within and among societies. In the English language, it is helpful to think of the terms female
and male as referring to sex differences, and feminine and masculine as referring to gender
differences. Something is “gendered” when socially and culturally defined gender differences
intervene in constructing it. Integrating gender analysis into development work means
analysing the various forms gender differences take and the ways they intersect with other
social markers (race, class, caste, religion, ethnicity, sexuality, etc.).

Text Box 4: Definitions – Sex vs. Gender

Sex: Refers to the classification of people as male or female, based on biological and
physiological characteristics such as chromosomes, hormones, and reproductive organs.

Gender: is a social and cultural construct, which values men’s and women’s, girls’ and boys’
attributes differently. Accordingly, it assigns socially acceptable and often stereotypical roles
and responsibilities to men and women. Gender-based roles and other attributes, therefore,
change over time and vary with different cultural contexts. The concept of gender includes
the expectations held about the characteristics, aptitudes and likely behaviours of both
women and men (femininity and masculinity). This concept is also useful in analysing how
commonly shared practices legitimise discrepancies between sexes.

19. Gender is a social organising principle that influences people’s roles and responsibilities
in the context of other social variables including ethnicity, culture, economic and social class,
religion. It is also a social structure that places women and men in different positions, roles,
and responsibilities. Finally, it is a social stratification that values what women and men do
differently. As a result, there are vast differences in the resources that women and men are
able to access, in the value attributed to their respective contributions, and in their ability to
effectively act on the world and on their own behalf (Lorber, 1994; Kabeer, 2002).

20. In terms of policy, three interrelated points are crucial to understanding the way gender
works; gender affects peoples’ lives with regards to needs, access to power/resources and
differential effects a policy may have on women and men.

14
 a) Needs: One must be aware that women, men, girls and boys may have different needs
as a function of their socio-cultural and economic situations in a given context. For
instance, to enhance the schooling of girls, an increase in the number of female teachers
and separate girls’ bathrooms in mixed schools may be instrumental. Girls have specific
needs in this context as sexual violence may be a real problem in many places and parents
need to be reassured so that they are willing to send their daughters to a school.
b) Power/Resources: Access to, and control of, power and resources including decision-
making is gender-differentiated. For instance, the role of household head is more often
ascribed to men than women due to social bias that men are family leaders, regardless of
whether they are the main income earners. Similarly, the fact that parliaments and local
governments are strongly male-dominated may translate into a bias toward spending
municipal resources on larger roads that support trade and create jobs for men, whereas
less consideration is given for instance to street lighting and police services that increase
safety for women.
c) Effects: Population policies and social programmes may have (unintended) differential
effects on women and men, and girls and boys. For instance, health policies that do not
consider the different – and often lower – income status of women-headed households
may inadvertently restrict poor women from accessing user-paid services and
consequently reinforce poor health outcomes for population subgroups such as
households headed by single women. Women may also not have the opportunity to
participate in decision-making as their time is limited by double or triple roles in
productive and reproductive work.

As a consequence, conceptualizing gender in statistics - beyond the simple disaggregation of

data by sex - is complex. Much of the remainder of this guide will suggest ways of doing this
that are meaningful and consistent – or valid and reliable.

21. The terms gender and sex are often used synonymously. Indeed, the terms “gender
disaggregation” or “disaggregated by gender” continue to be widely used and to confuse
those who were taught during gender training that sex and gender are quite different
concepts. Although the English language (unlike most others) does allow the use of the word
“gender” in the sense of “sex”, there is now broad international consensus that gender is not a
useful category for defining statistical variables: gender statistics are actually disaggregated
by sex and not by gender.

Text Box 5: What is Sex Disaggregated Data?

Sex-disaggregated data is data that is cross-classified by sex, presenting information

separately for men and women, boys and girls. When data is not disaggregated by sex, it is
more difficult to identify real and potential inequalities. Sex-disaggregated data is necessary
for effective gender analysis.

Source: “Gender Equality, UN Coherence & You – Glossary, July 2010”

22. Interpretation is needed to make sense of sex-disaggregated statistics in terms of gender.

Sex-disaggregated data are an indispensible starting-point and should be routinely available.
However, socio-cultural and economic analysis on sex-differences (i.e. gender analysis)
needs to be carried out on those data in order to make them meaningful in terms of gender.
This guide will show you how to do this.

15
B. Measuring sex/gender differences, gender inequality and gender
inequity through gender analysis

23. Gender Statistics is a policy-oriented approach to data, thus focusing on gender (the
socio-cultural construct) rather than sex (the biological marker) as an analytical category.
(Appendix 4 provides a more detailed overview of the evolution of gender statistics). Gender
Statistics is about producing and disseminating statistics that reflect the realities of women
and men of all ages, with a view to informing gender equality initiatives and policies. Current
challenges in gender statistic include financial shortages and capacity gaps that lead to low
data quality, or insufficient data analysis and dissemination, as well as a lack of normative
frameworks in-country, and coherent definitions and methods worldwide that lead to
shortcomings in political will and in implementation on the ground.

24. It is customary to distinguish between two categories of gender statistics, namely sex-
disaggregated statistics and other gender-relevant statistics. The latter refer to data that
provide information on the situation of either sex or the gender relations between men and
women, but that cannot be meaningfully compared between the sexes. For example, maternal
mortality is an inherently female phenomenon, whereas the incidence of prostate cancer is an
inherently male phenomenon. Although fertility can, in principle, be quantified for both men
and women, the former is much more difficult than the latter. In addition, men and women
have inherent biological differences with respect to the way in which they are affected by
fertility. Therefore, fertility is probably better thought of as a gender-related statistic, rather
than in terms of sex disaggregation.

25. Statisticians use words like “variance”, “variation”, “disparity”, “dispersion”, “inequality”
or “differentiation” descriptively, without much regard for the inherent fairness or injustice of
the differences observed between individuals. Sociologists make distinctions between such
concepts as “inequality of opportunities” and “inequality of outcomes”. And economists may
distinguish between kinds or degrees of inequality that are functional or dysfunctional to
economic growth.3 The gender area has its own terminology for describing disparities
between the sexes in terms of their fairness or lack thereof. Sex/gender difference is a
descriptive, value-neutral concept. Sex/gender difference refers to disparities or lack of
similarities between men and women – as social groups – in their respective status and
livelihood conditions. For instance, if women and men have different consumption
preferences, needs and aspirations, this creates differences in the way they spend their money.
Similarly, they may have different inclinations with respect to their career paths or ways to
spend their leisure time. Preferences, needs and aspirations are generally influenced by
representations of femininity and masculinity and can therefore be described as “gender
differences”. Where differences are related to biological traits – think of reproductive health
issues for instance – one can speak of “sex differences.” Where no differences exist, there is a
situation of “parity”. But parity is not the ultimate goal of gender equality as some differences
between the sexes are acceptable. For example, the goal that men and women should have the
same distribution of occupations may not be a pertinent one, while it is pertinent to require
the occupational status and incomes associated with these occupations to be similar.

26. Gender inequality is a human-rights-related, normative concept; it implies an assessment

of a given gender difference as unfair. Gender inequality refers to women’s and men’s, girls’
and boys’ unequal “conditions, treatment and opportunities for realizing their full potential,
3
In addition, economists often use the term “equity” in a sense that is completely distinct from the one ex-
plained in this section, namely to refer to debt-free assets in the form of real estate, bonds or particularly stocks.

16
human rights and dignity, and for contributing to (and benefiting from) economic, social,
cultural and political development” (Gender Equality, UN Coherence & You – Glossary
2010). Unequal valuing by society of the similarities and the differences between men and
women, and of the roles they play, also constitutes a form of inequality. Achieving gender
equality in turn requires women's empowerment to ensure that decision-making at private and
public levels and access to and control of resources are no longer weighted in men's favour.
This will result in women’s and men’s ability to participate as equal partners in productive
and reproductive spheres of life.

27. To illustrate the concepts of gender difference and inequality, think about a village where
the girl’s school starts and finishes after the boy’s school. As long as the variation in time is
not too large, this constitutes a “difference” between them, to be taken into account when
planning transportation systems for example. This difference may or may not be due to
discrimination or unfairness in the way society values men and women. By calling it a
“difference”, the fact that times vary is acknowledged but not judged. Now consider a village
with equal numbers of girls and boys of schooling age but with two boys’ schools and only
one school for girls. Here, girls and boys do not have the same opportunities for realising
their potential as access to education is harder for girls. Hence, there is gender inequality - it
is unfair that on average girls have less educational resources than boys which may
negatively affect their educational achievement.

28. Gender inequity is a policy-oriented concept that considers whether women and men,
girls and boys have the same chances of reaching expected outcomes such as literacy or
decent work. Gender equitable policies “ensure that women and men, girls and boys have
equal chance not only at the starting point but also when reaching the finish line” (Gender
Equality, UN Coherence & You – Glossary, 2010). They cannot be limited to providing equal
access but must also consider the fact that individuals start from unequal positions and hence
may require different degrees of policy intervention in order to achieve the same end result.
In practice, this will often involve some sort of positive discrimination to level the playing
field, such as special scholarships for girls to counter unequal access to secondary or tertiary
education or quotas to promote a balance of women and men in senior management positions
or parliaments.

29. The means to achieve gender equality and gender equity may therefore differ. For
example, regardless of the time girls and boys leave for school – which is a random
inequality –girls on average are more vulnerable to sexual assault and/or sexual harassment
than boys both at and on their way to school. Equity policies would try to ensure that schools
have facilities needed by girls so that they are allowed and willing to attend (i.e. private and
separate toilets) and that the education system aims to mitigate risk factors by providing safe
passages for girls. Equality policies, on the other hand, would try to ensure that girls continue
to be sent to school and that those schools provide them equal opportunity to learn so they
have the potential to achieve equal outcomes.4

30. ‘Gender mainstreaming’ is the chosen approach of the United Nations system and
international community toward realising gender equality. The key concern with regard to
4
This discussion parallels the one of “equality of opportunities” versus “equality of outcomes” in regard to the
role of the school system. While some consider schools as the great equalizers of opportunities between children
of different social backgrounds (or, in this case, different sexes), others (e.g. Jencks, 1972) have demonstrated
that, even in societies like the United States, which places a lot of emphasis on the principle of equal opportu -
nity, only a relatively small proportion of the inequality of outcomes can be explained in terms of differences in
access to education.

17
gender equity, on the other hand, is fairness of opportunities, and the chosen approach is the
empowerment of women and girls through targeted interventions, or special measures.
Equality and equity are inter-related concepts. Discrimination based on sex/gender is both the
root for needing equity policies and the barrier for achieving gender equality.

Text Box 6: What is Gender Mainstreaming?

Gender Mainstreaming: "Mainstreaming a gender perspective is the process of assessing

the implications for women and men of any planned action, including legislation, policies or
programmes, in any area and at all levels. It is a strategy for making the concerns and
experiences of women as well as of men an integral part of the design, implementation,
monitoring and evaluation of policies and programmes in all political, economic and societal
spheres, so that women and men benefit equally, and inequality is not perpetuated. The
ultimate goal of mainstreaming is to achieve gender

Source: United Nations Economic and Social Council (ECOSOC), 1997

31. While sex/gender difference can be measured fairly easily, measuring gender equality and
gender inequity poses greater challenges. To define sex/gender difference, variance, or the
degree to which the objects or individuals being described are different from the mean
(average), can be used. Similarly, to detect inequity and inequality, variance can show where
parity has not been achieved between women and men, girls and boys. Nevertheless, much
caution has to be employed in the construction of gender indicators because, as was pointed
out earlier, parity may not always be relevant or desirable nor does it describe the equitable
access to or fairness of opportunities. Rather what is more typically measured is a
government’s stated commitment to equality and the existence of equity interventions.

32. In most cases, in addition to mainstreaming statistical analysis, qualitative and policy
research is needed in order to assess whether the opportunities of women and men are
unequal. While gender differences in literacy, for example, is fairly easy to pinpoint on the
basis of variance by sex, an investigation of gender inequality in literacy would need to
consider opportunity factors such as the access to scholarships or stipends and explicit criteria
for admission to schools and literacy programmes. In addition, the ability to take advantage
of the opportunities being offered may differ between population groups. For example, even
if boys and girls have access to the same schools, girls may be restricted by the fact that they
need to care for their younger siblings part of the time, or in some other cases boys may be
restricted by the fact that parents count on them from an early age to generate some monetary
income. The field of gender statistics is devoted to holistically assessing such gender
inequalities through gender analysis.

33. Gender analysis is not about women and cannot be carried out on women as an isolated
group. Rather, gender analysis involves looking at power relations between men and women
and it may even focus specifically on men and boys to analyse how certain behaviours come
to be socially perceived as “masculine behaviours”. In this vein, many countries have
underscored the need to improve the quality and dissemination of their gender analysis of
human development indicators and promoted the use of gender-responsive indicators. Such
indicators need to capture and reflect the potentially different impacts of development
strategies and actions on men and women, and boys and girls. This will require going beyond
simply disaggregating data by sex and/or socioeconomic group, ethnicity, race and generation
(i.e. age). It will entail, among other things, efforts to select indicators that are sensitive to

18
possible difference – such as control of and access to resources such as schools– from the
outset of the analysis and mainstreaming gender perspectives into the entire statistical system.

Text Box 7: Conceptual Differences between Statistics on Women, Sex-Disaggregated

Data and Gender Analysis

Statistics on women: Women-only statistics produced mainly to report on the situation of

women in different countries and regions. They are historically connected to the Women in
Development (WID) approach. One limitation is that they do not allow for comparison
between men and women and thus cannot provide data on gender gaps.

Sex-disaggregated data: Describe gender ratios of a certain phenomenon and are a crucial
tool for quantifying differences and inequities between men and women. Historically
connected to the Gender and Development (GAD) approach, sex-disaggregated data,
although crucial, are not sufficient for the development of adequate gender analyses.

Gender analysis is an intellectual effort that involves at least the following fundamental
aspects:
- Sex-disaggregated data for measuring gender differences and different cultural and
socioeconomic realities faced by women and men;
- Multivariate analysis for capturing and interpreting relations that may not be visible if
using sex-disaggregated data only;
- Gender-specific indicators for topics that may be of greater relevance to one sex than the
other;
- In-depth examination and interpretation in order to get a fuller, more valid picture of
what is occurring in context and which are the social constraints that lead to inequality;
- Identifying areas where new data need to be collected in order to fully grasp elements of
inequality;
- Translating data into policy and planning to provide the evidence-based for strategy
formulation.

Text Box 8: Evidence-Based Advocacy

Advocacy is the pursuit of influencing outcomes – including public policy and resource
allocation decisions within political, economic, and social systems and institutions – that
directly affect people’s lives. In practice, it includes the continuous and adaptive process of
gathering, organising and formulating information and data into an effective argument, which
is then communicated to policy-makers through various interpersonal and mass media
communication channels. In the context of gender analysis, advocacy seeks to influence
policymakers, political and social leaders to create an enabling policy and legislative
environment and allocate resources equitably.

Evidence-based advocacy can be contrasted with normative-based advocacy. While

normative-based advocacy draws its legitimacy from national legislation and/or international
norms and standards such as Human Rights conventions, evidence-based advocacy is based
on demonstrable facts and measurable data and information from official statistics, surveys,
experiments, evaluations, direct observation, etc. In terms of presentation, evidence-based
advocacy uses graphs, tables and charts and includes citations to show the strength of the
evidence on which a particular argument is based.

19
In order to be persuasive, evidence must be reliable and relevant to the interests of the
decision-maker or audience targeted. Therefore, different types of evidence have to be
organised and presented differently for different audiences and part of effective advocacy is
understanding and taking into account the interests, needs and prejudice of the various target
groups. Often, evidence-based advocacy provides evidence about the problem, the likely
impact of change, the feasibility of possible solutions, and about who is responsible to make
the change.

UNFPA (2002). Advocacy: Action, Change and Commitment. Distance Learning Courses on
Population Issues, Course 4. United Nations Population Fund (UNFPA) and United Nations
System Staff College (UNSSC), New York and Turin.

UNICEF (2009). Evidence-Based Advocacy for Gender in Education. A Learning Guide.

UNICEF, Bangkok.

Websites consulted:
 Gender Equality, UN Coherence & You – Glossary: Definitions A-Z on
http://www.unicef.org/gender/training/content/scoIndex.html
 2010 World Population and Housing Census Programme (UN Statistics Division) on
http://unstats.un.org/unsd/demographic/sources/census/censusquest.htm
 UNFPA Census Portal on http://www.unfpa.org/public/op/edit/home/sitemap/pid/6734

C. Some issues in data analysis and the construction of indicators

A basic typology of indicators

34. Indicators that compare the situation of women to that of men can be constructed in a
variety of ways and the results may vary according to the specifics of the definitions used.
For some purposes, a particular indicator definition may be ideal, while it may be highly
misleading when used for other ends. Take the example of teenage pregnancy. Reproductive
health providers often use the percentage of deliveries in which the mother is under age 20 as
an indicator for the user profile of maternity clinics. To the extent that young mothers may
require special care for which clinics need to be prepared, this is a perfectly adequate
indicator. However, the same indicator is also often used to quantify the incidence of teenage
pregnancy, a use for which it is ill-suited. This is because the percentage of women under age
20 who become mothers may be stationary or even diminish, but as the fertility rates of older
women decline faster than those of adolescents (as is often the case), the result will be an
increase of the percentage of deliveries in which the mother is under age 20. This is due to
older women having less children, not to younger women having more.

Percentage distribution indicators

35. Problems similar to the ones outlined above often characterize the use of indicators in
which the comparison between men and women is made in terms of absolute numbers. For
example, one might quantify the degree to which unemployment is a problem for men and
women by constructing an indicator consisting of the percentage of the unemployed that are
women. If the objective of this indicator is to define the profile of users of particular services

20
available to the unemployed (e.g. to know if the unemployment agency should make a greater
effort to provide information on job openings or training courses that usually attract a lot of
female applicants), this indicator may be entirely reasonable. But if the objective is to
quantify if women have a higher or a lower risk of becoming unemployed, it is inadequate, at
least in countries where female labour force participation is lower than male labour force
participation, as is usually the case. The absolute number of unemployed women may be
smaller than that of unemployed men, but when computed as a percentage of the female and
male labour forces, the picture may be entirely different as the percentage of economically
active women that are unemployed is often higher than that of men.

36. The same applies to indicators such as school enrolment. The percentage of primary
school students who are girls may be a perfectly adequate indicator for planning purposes,
e.g. to know how many toilet facilities schools need to have for boys or girls, respectively,
but as an indicator for the propensity of boys or girls to enroll in secondary education, it may
be flawed by the fact that the base population of boys and girls in the relevant age group is
not the same, particularly at the local level. An alternative indicator, which quantifies the
relationship not in terms of percentages, but as a ratio between the number of boys and girls
has the same disadvantage. This is why the MDG indicator that deals with differential school
enrolment is not stated in terms of absolute numbers, but rather in terms of Gross Enrolment
Ratios, i.e. the number of girls enrolled (regardless of age) as a proportion of the population
of the relevant age group, divided by the equivalent proportion of boys in the same age group
(gender parity index). This corrects for the problem of unequal base populations. It does not
correct, however, for the the fact that the larger numbers of students of one sex may be due
not to a higher propensity to receive education, but to high repetition rates. In Brazil, for
instance, Gross Enrollment Ratios for girls are higher than those for boys at all levels, except
primary because boys tend to stay longer in primary school, due to their lower educational
performance. The use of the Net Enrollment Ratio, rather than the Gross Enrollment Ratio,
would prevent this bias (Leonardo Athias, of IBGE, personal communication).

Ratio indicators

37. The previous paragraph shows how a ratio indicator based on rates is an improvement
over a ratio indicator based on absolute numbers or a percentage distribution indicator.
Nevertheless, indicators of this type do have one major limitation, namely that the same result
can be brought about either by high rates in the numerator and the denominator or by low
rates in both. Thus, it is difficult to decide whether an enrolment ratio of 1 represents a good
or a bad outcome from the viewpoint of female education. Nor does it provide a clear idea on
how easy or how difficult it would be to correct it. An enrolment ratio of 0.9 is much easier to
correct when it results from a female rate of 18 per cent, compared to a male rate of 20 per
cent than when it results from a female rate of 81 per cent compared to a male rate of 90 per
cent.

Standardized and unstandardized indicators

38. Despite the drawbacks outlined in the previous paragraph, ratio indicators are sufficiently
detailed for many purposes. In those cases in which they are not, there is always the option of
computing the male and female indicators separately, so that one can evaluate their individual
values, rather than just their relative size. However, there are situations in which ratios or

21
even individual male and female indicators can be uninformative or even misleading if no
account is taken of intervening factors.

39. As an example, take the proportion of men and women that experience multiple divorces.
This proportion will usually be higher for men. But how to interpret this ? Is it because men
in their second or third marriage are more likely to divorce again than women in the same
situation ? That may be, but a more likely explanation is that more men than women remarry
after a first divorce or widowhood and consequently, even if the divorce rate for men and
women in second or third marriages is the same, there will still be more men than women
with multiple divorces. In order to take account of this fact, one should either break down the
information by second, third, fourth etc. marriages, or one should weight the divorce rates by
some uniform distribution of second, third, fourth etc. marriages, which does not vary by sex.
This so-called standardization by order of marriage will ensure that the result can be used to
compare the actual propensity of men and women to experience additional divorces in later
unions, rather than depending on an intervening factor (in this case, remarriage). Of course, if
the only objective of the analysis is to estimate the proportion of men and women that ever
experience a multiple divorce, rather than to estimate the risk of a renewed divorce of men or
women in higher order marriages, standardization is not necessary.

40. It is partly for the same reason that the literacy rate in the MDG indicators has been
defined in relation to men and women aged 15-24, rather than 15 and over. Literacy rates are
lower at higher ages and since it is at these ages that women predominate, the male-female
differential is disproportionally weighted in favour of males. Again, this is not a problem if
the only objective is to know how many illiterates there are in the population, by sex. But in
order to get a realistic imagine of male-female differentials, it is better to compute the
indicator by age group, or to standardize. An additional reason for using the 15-24 age
category is that it provides a better measure of the recent performance of the educational
system, rather than a historical assessment of something that is more difficult to correct by
regular educational policies.

41. Standardization is particularly relevant in the case of disability, which is why the chapter
on that subject illustrates it in some detail. A short example out of that chapter may serve to
further clarify the issue. One way to express the differential incidence of disability in men
and women is to compute the number of years that men and women of a certain age can
expect to live with a disability in the future. This number tends to be higher for women. But
the number of years that they can expect to live without a disability is also higher for women.
One solution to this ambiguous relationship is to compute the percentage of the expected
number of remaining years of life that men and women should expect to live with a disability.
But an alternative is to compute the expected number of years with a disability in a mortality-
standardized way, by using the same life table (e.g. an average for both sexes) for men and
women. The latter removes the differential impact of the intervening variable (mortality)
from the indicator of interest.

42. In February of 2012, the UN Statistical Commission approved a Minimum Set of Gender
Indicators, consisting of 52 quantitative indicators and 15 policy indicators related to the
existence of national norms. In the substantive chapters that follow, systematic reference will
be made to this set of indicators and the degree to which they can be estimated from census
data. The operationalization of the indicator set is still being discussed at the time of
publication of this guide, so the the fact that this guide indicates that a certain indicator can be

22
estimated from census data is no guarantee that the methodology that will ultimately be
approved will actually endorse this option.

Multivariate analysis to disentangle intra-group variability and interrelationships

43. Men and women are not homogeneous groups. While women as a group may have lower
educational attainment than men, some sub-groups of women may have higher educational
attainment than some sub-groups of men or even men in general. The relationship may vary
in terms of other intervening socio-economic and demographic factors such as economic
level, rural or urban residence, and age. Therefore, one may want to know if this relationship
of women’s lower educational attainment holds at different economic levels, in both rural and
urban places of residence, and at varying ages. Additionally, when two variables are
correlated, such as lower education and early marriage, the next step is to ascertain whether
one causes the other. Two characteristics may be correlated without being causally related. In
this example, early marriage could be highly correlated with lower education, yet their
relationship could be spurious, i.e. caused by another factor, such as belonging to a certain
ethnic group with prescribed cultural norms regarding both education for girls and early
marriage.

44. Multivariate analysis – meaning analysis with multiple predictor variables – allows, inter
alia, for the measurement of the effects of two or more independent (also called predictor or
explanatory) variables on a dependent or outcome variable. It makes it possible to measure
the effect of each separate explanatory variable on the dependent variable, while controlling
for (i.e. keeping constant, as in the famous condition “ceteris paribus” – all other factors
being equal) the effect of all other explanatory variables being considered. While multivariate
analysis cannot, strictly speaking, demonstrate the existence of causal relationships, it can
approximate the analysis to a causal interpretation in that it provides a more comprehensive
view of the different relationships, thereby making it easier to identify situations in which, for
example, the relationship between two variables can be accounted for by their common
dependence on a third factor.

45. Two types of multivariate analyses which have proved very useful in social studies are
multiple linear regression and logistic regression. Multiple classification analysis (MCA) is
another useful technique, closely related to linear regression.

a) Multiple linear regression uses several explanatory variables (which may be

continuous or discrete) to predict a dependent variable, which has to be continuous
(interval level, meaning that differences between values have a true numerical
interpretation). The relationship has to be linear, but rather than using the original
independent variables, one may transform or combine them, as long as the
transformations do not introduce any non-linear parameters to be estimated. For example,
if the original explanatory variables include age a) and educational attainment e), one may
introduce other variables equal to A2 or A*E or even E/(A-6), but not E/(A-k), where k
has to be estimated from the data. In the end, what is being estimated is a set of
coefficients – one for each independent variable – which quantify the influence of each of
them on the dependent variable.

b) Another attractive technique, which is closely related to linear regression, is Multiple

Classification Analysis (MCA). MCA (also named Factorial ANOVA) quantifies the
interrelationship between a set of predictors and a dependent variable in a linear, additive

23
 model. As is the case with linear regression, the dependent value should be at the interval
scale. It is one of the nice features of MCA that it so easily handles discrete explanatory
variables that do not allow a true numerical interpretation (e.g. ethnic group or level of
agreement with a statement). In the MCA-analysis these categorical independent
variables are called factors. MCA also allows the inclusion of continuous explanatory
variables, called covariates. MCA basically produces the same results as a multiple linear
regression with a set of discrete variables expressed as dummies (e.g. economically active
or not, completed high school or not, etc.). However, the advantage of MCA over linear
regression lies in the way the results are presented. The constant in the MCA analysis is
simply the overall mean of the dependent variable. Each coefficient of the categorical,
independent variables is presented as a deviation from the overall mean. First, unadjusted
deviations are given, and thereafter, adjusted deviations are presented, i.e. after
controlling for the effect of all other independent variables (factors and covariates).

c) Logistic regression also uses several explanatory variables to predict a dependent

variable, but in this case the dependent variable is discrete, taking two dichotomous
values (e.g. yes or no, attending or not attending school). A variant of logistic regression
is multinomial regression, in which the outcome has more than two alternatives (e.g.
marital status). The actual outcome of the logistic or multinomial regression equation is a
number between 0 and 1, which describes the probability that a given outcome will
happen. The same considerations about linearity mentioned above apply here as well,
with the difference that the entire linear expression is linked to this probability through a
logistic function which, by definition, can only take on values between 0 and 1. The slope
coefficients (B) in a logistic regression are so-called log odds ratios, which are hard to
interpret. Therefore, the exponential of the log odds ratio is calculated (e B) which gives
the odds ratio, i.e. the probability that something will happen divided by the probability
that it will not.

46. Multivariate analysis relies upon a set of assumptions about the variables in the model.
Before applying a certain technique, researchers should always test whether these
assumptions hold. For instance, in the case of linear regression, assumptions that are regularly
violated by researchers are: a) The assumption of linearity between the predictor and the
dependent variable, b) The assumption of constant variance of the error terms
(homoscedasticity) c) The assumption of no correlation between the error term and the
predicted variable, and d) The assumption of absence of high correlation between the
predictor variables (multicollinearity). A violation of homoscedasticity would, for instance,
be ‘age’ as a predictor of children ever born by women. The variance of the number of
children ever born among younger women is much smaller than among older women. An
example of multicollinearity would be the use of both weight and height as predictors in a
regression model, as the two are highly correlated. For more detailed information on
multivariate methods, their variations and the ways to deal with the problems that can arise
when applying them, there are numerous standard texts that one can consult, among them
Hoyle, Harris and Judd (2001), Knoke, Bohmstedt and Potter Mee (2002), Linneman (2011)
and Stock and Watson (2010).

47. Some examples of the use of multivariate analysis include:

a) Nguelebe (2005) using 2003 census data on early marriage and schooling in the Central
African Republic (CAR) found that just over one-fourth (26.3 per cent) of girls between
the ages of 12 and 19 were already in a union, compared with just 4.2 per cent of boys in

24
 the same age group. Census data can examine if there are bivariate relationships (using
cross-tabulations) between early marriage and such factors as attending school, literacy,
urban/rural residence, regional of origin, ethnic group, religious affiliation, and
employment status of the mother. But this leaves open the possibility of
misinterpretations, due to the kinds of problems pointed out earlier, e.g. that both early
marriage and low schooling may be the result of belonging to a certain ethnic group or
religion. A multiple regression model to predict early marriage might take the age of first
marriage as an outcome variable or one might use logistic regression if the outcome
variable is whether a given girl is married or not by the time she reaches a certain age.
Factors that share bivariate relationships with early marriage would be included as
independent or predictor variables. Qualitative (discrete, categorical) explanatory
variables, such as ethnic group or religion, have to be broken down into a series of binary
choices (called dummy variables), which are then treated as separate variables, e.g.
Catholic (yes/no), Protestant (yes/no), Muslim (yes/no), etc.

b) Census data could be used to examine the relative effects of the independent variables
of racial or ethnic composition and rural or urban residence on poverty, a dependent
variable. Persons of a specific ethnicity may make up a majority of those residing in the
rural area of a country. Multivariate analysis allows us to determine the relative effects of
living in a rural area and belonging to a specific ethnici group on poverty, while
accounting for the possible interrelations amongst the three variables: 1) place of
residence, 2) ethnicity and 3) poverty. A possible result of this analysis might be that
ethnicity does not have an effect onpoverty outcomes once place of residence is
controlled for.

c) Census data could be used to examine the relative effects of living in a certain region,
type of household (i.e. with or without children), sex and educational attainment of the
head of household or whether the household head is employed full time or not. With this,
the researcher is able to use multivariate analysis to understand to what extent, if any,
being employed full time can be explained by region of residence, the sex of the head of
household, whether the household has children or not, and the educational attainment of
the head of household.

With each of these above examples, the estimation of the effects of the dependent variables is
done simultaneously, so the results show the effect of each independent variable on the
dependent variable, while controlling for the intervening effects of all other independent
variables. The next section focuses on a concrete example using a gendered perspective.

48. A gendered analysis using both bivariate tabulations and multivariate methods can be
found in the work of Snyder, McLaughlin and Findeis (2006) using a 5-percent sample of the
2000 US Census data to examine race and residence as independent variables affecting
poverty prevalence of female-headed households with children. This study first cross-
tabulates census data to learn that cohabiting and grandmother, female-headed households
with children comprise over 25 percent of all female-headed households with children. Using
cross-tabulations again, they then find that household poverty is highest for female-headed
households with children that do not have other adult household earners. They also note the
relative difference in income between these female-headed households 1) with and 2) without
other earners or income, and the average difference between these two groups is substantial;
those with other earners or income are lifted out of poverty.

25
49. Then, multivariate models are used to provide validation to these tabular results. Because
this is done at the level of individual households, where poverty is a categorical variable
(poor versus non-poor households), logistic regression is the method of choice. If it were
done at the level of census tracts or other geographical units, conventional multiple regression
would be more appropriate. Poverty is found to be highest among racial/ethnic minorities and
female-headed households with children in rural areas compared to central cities and
suburban areas. The authors are also able to estimate the relative effects of the independent
variables of race and residence and family type, while controlling for the effects of other
factors, such as region, educational attainment, age, marital status, number of hours worked
last year, and public assistance receipt. These findings – that ethnic minorities and rural
female-headed households have even higher rates of poverty than female-headed households
in general – are net effects and independent of these other factors.

50. Building on other research that finds a steady rise in female-headed households with
children since 1970 (Casper and Bianchi, 2002) and that over one-half of children will live in
a female-headed household (Graefe and Lichter, 1999) during childhood, the tabulations
establishing the poverty link take on real life course implications for women and children.
Understanding the relative effects of defining characteristics, such as race and rural residence,
in predicting poverty can then be useful for policy makers and advocates wishing to
ameliorate the possible outcome of poverty, especially for ethnic minorities and those in rural
areas. Household composition is an important component of the economic support that a
family has. Census data can be used to monitor trends in family household composition on
the one hand, and even control for family household composition while examining income,
poverty status and public assistance receipt on the other.

Geo-spatial analysis

51. With the ever more ubiquitous availability of Geographic Information Systems (GIS),
census data are increasingly being represented and analysed in connection with their
underlying geography. In the case of gender studies, Bosak and Schroeder (2005) discuss
some of the advantages, as well as the pitfalls, of using these techniques. Although geo-
spatial analysis can be done without them, maps are a frequent companion to such endeavors.
Poverty maps, for example, have been around for many years, as a visually appealing tool
that makes it possible for governments to pinpoint the areas where poverty is most acute (see
Chapter 8).

52. A distinction must be made, however, between the mere representation or visualization of
data in the format of a map and the analytical use of geospatial analysis to advance the
understanding of the processes involved. The first merely uses maps as an alternative to the
presentation of data in tables, with the advantage that some characteristics of the data are
more easily grasped that way. For example, it may be that the education gap between boys
and girls is geographically clustered in certain zones of the country that comprise several
geographic base units. This may suggest that the problem has to do with factors that are
common to these base units, maybe the fact that a certain ethnic group predominates in this
part of the country or the fact that a certain type of agriculture is practiced there. Visualizing
the data in the format of a map may help to develop an intuition for such explanations. The
studies of sex ratios (Ebenstein and Leung, 2010) in Chapter 5 and of female enrollment in
Gansu, China (Cao and Lei, 2008) discussed in Chapter 9 are examples of this type of use of
spatial data. For more details on these kinds of maps, see Schultz (2009), among others.

26
53. Geo-spatial analysis, in the strict sense of the word, goes beyond such intuitions by
explicitly linking data that affect each other in ways that allow a spatial interpretation. A
certain district may have a large gender gap in education because of something that is not
characteristic of the district itself, but of a district situated nearby, e.g. the presence of a large
textile factory that employs lots of young girls. Unless the data are analysed taking into
account their spatial structure, such relationships may be missed. In the case of employment
and mobility, there is already a substantial literature in the more developed countries which
links gender factors to the availability of opportunities across space. Some of these analyses
(e.g. Hoogstra, 2012; Tkocz and Kristemen, 2010) can be quite complex, involving
methodological tools such as autoregressive and cross-regressive spatial lags to detect
relations both within and among groups and spatial weights matrices to represent travel times.
These econometric techniques are, unfortunately, beyond the scope of this manual. For more
information in this regard, one may consult a variety of basic (e.g. Kalkhan, 2011) or more
advanced (e.g. Mitze, 2012) texts.

54. Other geo-spatial applications, however, are much simpler, like demonstrating that the
education of women in Lesotho and Ethiopia tends to be higher as they live closer to main
roads (Walker and Vajjhala, 2009). In this case, basically all that is required is overlaying the
population data from the census or (in this case) DHS with a geographical data base showing
the location of roads, elaborating the respective distances and computing their correlation
with population characteristics. There are also several studies that – with or without the use of
maps - relate the ratio of school enrollment rates between boys and girls with the distance to
the nearest school, to explore the assumption that the school enrollment of girls might (or not)
be more negatively affected by a large distance from school than the school enrollment of
boys (see Chapter 9).

Life course approach

55. The life course approach provides a way to examine a person’s life as it is enmeshed
within a social, cultural and structural context over time. In this way, it is possible to
understand a person’s life history as it relates to decisions and actions that lead to later
events. As an example, the experience of poverty as a child may affect educational outcomes
at that point in time, but also later in life and may trigger a trajectory of lowered opportunity
for that individual due to this earlier exposure. Scholars have formally defined life course as
“a sequence of socially defined events and roles that the individual enacts over time” (Giele
and Elder 1998). This approach takes a person’s lived experiences and connects them with a
particular historical context and a specific socioeconomic level over time.

56. A gendered analysis that utilizes the life course approach examines how a woman’s life
experiences are shaped differently from those of men because they are female. For example,
in Table 2, Heise (1994) puts forward a gendered life course analysis of women’s
disadvantaged position as it may explain violence against women at each phase of life – from
Prenatal, Infancy, Childhood, Adolescence, Reproductive and Old Age phases. At each
phase, the social, structural and historical context come together to present women with a
lowered level of health and heightened level of violence against them, ranging from sex
selection at the Prenatal Phase, to female infanticide at the Infancy Phase, to genital cutting at
the Childhood Phase, to dating and courtship violence at the Adolescence Phase, to marital
rape at the Reproductive Phase to abuse of widows at the Old Age Phase.

Table 2: Gender discrimination throughout a woman’s life course

27
Phase Type
Prenatal Prenatal sex selection, battering during pregnancy, coerced pregnancy (rape during war)
Infancy Female infanticide, emotional and physical abuse, differential access to food and medical
care
Childhood Genital cutting; incest and sexual abuse; differential access to food, medical care, and
education; child prostitution
Adolescence Dating and courtship violence, economically coerced sex, sexual abuse in the workplace,
rape, sexual harassment, forced prostitution
Reproductive Abuse of women by intimate partners, marital rape, dowry abuse and murders, partner
homicide, psychological abuse, sexual abuse in the workplace, sexual harassment, rape,
abuse of women with disabilities
Old Age Abuse of widows, elder abuse (which affects mostly women)

Source: Heise (1994).

57. A gendered life course approach considers that disparities may become more pronounced
or attenuated and even reversed in direction as the life cycle progresses. While there may not
be major differences for girls and boys in education in Early Childhood, as a child ages and
becomes a young adult, any minor differences become more and more apparent. UNICEF
(2007) reports in the State of the World’s Children that while the gap between girls’ and
boys’ enrolment in primary school has been declining, nearly 20 per cent of girls will drop
out of primary school by level 5. Thus, for every 100 boys out of school, 115 girls are in the
same situation, and 43 per cent of girls are at the appropriate grade level for their age. Gender
parity in primary enrolment does not carry over into parity in secondary school and overall
educational attainment. Indeed, by the time these girls reach adulthood, as women they will
comprise two-thirds of the world’s illiterate population.

58. Being cross-sectional in scope, census data present a methodological challenge for the life
course approach, which generally uses longitudinal survey data of persons’ social, cultural
and structural contexts. However, using US census data Stevens (1990) creates synthetic
cohorts from the same census year (i.e. computing mean educational attainment for women
and men by specific age groups, 20-29 year olds, 30-39 year olds, 40-40 year olds, etc. to
know if educational attainment is increasing or decreasing over time for women and men),
and Fussell and Furstenberg (2002) follow American cohorts over time using census data
gathered at 10-year increments. This method of following cohorts over time is limited in use
for countries with irregular census data collections or where question wording has changed.
Overall, these cohort construction methods work for some research questions and not others.

28
MEMORY CARD: SUMMARY PART 1

59. In summary, gender and census data are linked in many ways: Gender advocates need
robust and reliable data to sustain their claims and produce more convincing advocacy
materials. Data producers need to understand gender issues to make sure the data they
generate is fully representative of the entire population, including of vulnerable women such
as widows and disabled girls, and of the entire spectrum of issues pertaining to both sexes.
Census data has many limitations as a basis for gender analysis, many of which are linked to
its focus on breadth (full geographical coverage) rather than depth (few questions). The
following chapters will discuss gender issues that can nevertheless be analysed with census
data.

Figure 1: Gender Equality: Normative Ideal and Development Objective

Gender analysis helps to frame questions

Gender refers to the social roles and about women and men's roles and rela-
relations between women and men. This tions in order to avoid making assump-
includes the different responsibilities of tions about who does what, when and
women and men in a given culture orVision of Gender
why. The aim of such an analysis is to
location, and the allocation of power and Equality
formulate development interventions that
resources based on gender-based social are better targeted to meet both women’s
constructions. and men’s needs and constraints.

Gender mainstreaming means Gender sensitive indicators

that attention to gender equality is demonstrate changes in women
given a central part in all development and men’s access to, and control of,
interventions, including analyses, policy different types of resources in a given
advice, advocacy, legislation, research and society or location over a period of time.
planning, implementation, monitoring and They are useful in monitoring and
evaluation of programmes and projects. evaluating progress towards gender
equality.

(Source: DLPI, Course 3, module 3)

29
 Part Two

10 Key Gender Issues Analysed with Census Data

30
60. After having laid the groundwork on gender-responsive data and gender analysis, the
following chapter will show concretely how gender issues can be analysed on the basis of
census data.

61. Ten gender issues were selected for inclusion on the following grounds:
a) Census data are available on the topic (i.e. gender-based violence cannot be analysed)
b) The topic to be analysed at hand has a clear gender-dimension
c) The gender issue is of global relevance or of extraordinary relevance to at least two
regions (i.e. the issue of ‘sex ratios’ was included due to its all-encompassing relevance in
Asia and growing relevance in other regions])
d) A large number of countries are obligated under international law to report on the
gender issue in question. This implies that
o Cooperation between national mechanisms for gender equality and National
Statistical Offices is needed on the issue
o International Human Rights standards and national legislation are available that
indicate how the gender issue should be approached; e.g. child marriage should be
banned, migrants should have access to education and health services regardless of
their regular/irregular status, etc.

62. Each chapter follows the same structure, investigating what the subject is, why it is
important from a gender and human rights perspective, and what data concerns exist. It then
turns to suggested tabulations, suggested indicator and further, more complex analyses that
can be carried out around the gender issue at hand. Each chapter closes with some pointers on
advocacy around the gender issue in favour of greater gender equality and empowerment of
women.

31
Chapter 3:
Fertility
1. What is it?

63. Fertility is the childbearing performance of individuals or couples, measured for

particular groups or populations. It is contrasted with fecundity, the theoretical capacity to
reproduce, which may or may not lead to fertility. 5 Sometimes the term natality is used to
refer to the most general analysis of childbearing. Measures of fertility normally refer only to
live births. Several qualifiers may be applied to fertility to specify more exactly the focus of
analysis, such as marital fertility or fertility by birth order (e.g. the fertility for birth order can
be used to describe the proportion of women by age who have ever had a live-born child).

64. Three types of data are commonly collected in censuses in developing countries: 1) The
number of Children Ever Born alive (CEB); 2) Children born during the past 12 months
before the census; and 3) Survival of children ever born and/or children born in the past 12
months. There is less need for these types of census data in more developed countries,
because they can be more reliably and frequently obtained from civil registration sources.
Applying a variety of analytical techniques, census data on these topics can be converted into
estimates of fertility and infant and child mortality. Although the details of these procedures
are beyond the scope of this manual, a general idea is provided in Methodology Box 4. For
more information, see Manual X of the UN Population Division (United Nations, 1983),
which is currently being updated. 6 It should be pointed out, however, that even with these
conversions and adjustments, fertility data collected from censuses are less detailed and of
lower quality than the fertility data typically collected in specialized surveys such as the
Demographic and Health Surveys (DHS).7

65. Male fertility, which does not fit neatly into the categories above, has also occasionally
been of interest. Conventionally, fertility is investigated only in relation to the age and other
characteristics of women, but for some purposes it is relevant to know how fertility rates vary
by the fathers’ characteristics.

66. Infertility/Childlessness: Demographers define infertility as childlessness in a population

of women of reproductive age, as opposed to infecundity, the biological incapability of
conceiving. The epidemiological definition, on the other hand, is based on “trying for” or
“time to” a pregnancy, generally in a population of women exposed to a probability of
conception. The time that needs to pass (during which the couple has tried to conceive) for
that couple to be diagnosed with infertility varies from country to country The WHO defines
infertility as the inability to conceive a child, clarfying that a couple may be considered
infertile if, after two years of regular sexual intercourse, without contraception, the woman
has not become pregnant (and there is no other reason, such as breastfeeding or postpartum
amenorrhoea). Primary infertility is infertility in a couple who have never had a child.
Secondary infertility is failure to conceive following a previous pregnancy. Infertility may be

5
Note that in French and Spanish, the usage of the terms is the opposite: fécondité or fecundidad for actual re -
productive outcomes and fertilité or fertilidad for biological capacity.
6
The update is being undertaken to incorporate new developments, take account of new trends such as those
brought about by the advent of AIDS, and adapt the older techniques to the possibilities created by more modern
computational tools such as EXCEL. For more information, see http://demographicestimation.iussp.org/.
7
Nevertheless, omission and displacement of births in DHS data are not trivial, and of course, as in any survey,
one has to account for sampling errors, which limits the possibility of using DHS data for small sub-groups.

32
caused by infection in the man or woman, but often there is no obvious underlying cause.
Infertility increases with age. A study on Hutterite women (who do not use birth control)
carried out in the 1950s (Tietze, 1957), found the following progression: By age 30, 7% of
couples were infertile, by age 35, 11% of couples, by age 40, 33% of couples and by age 45,
87% of couples were infertile. Infertility is of particular concern in Africa because of the
extent of the problem and the social stigma attached to it. The highest prevalence of infertility
in Africa occurs south of the Sahara, but some 5 to 8% of couples are estimated to experience
infertility at some point in their reproductive lives (50-80 million people worldwide). The
average infertility in Africa is 10.1% of couples, with a high of 32% in some countries, and
certain tribes have high infertility rates.

2. Why is it important?

67. Fertility and mortality patterns have key consequences for the lives of both men and
women. For example, marriage and childbearing at very young ages or bearing large numbers
of children can limit women’s opportunities for education and employment and can diminish
their chances for advancement in life.

68. Moreover, fertility decisions are part of the key principles of the International Conference
on Population and Development (ICPD) Programme of Action. Principle 8 states: “All
couples and individuals have the basic right to decide freely and responsibly the number and
spacing of their children and to have the information, education and means to do so.” Even
though the right to “found a family” is protected by the Universal Declaration of Human
Rights (United Nations, 1948), women’s reproductive freedom remains less than full.
Fertility stands as an important area to examine, monitor and understand women’s position in
society. From a human rights point of view, one therefore needs to ask: Do current fertility
patterns reflect the free choices of couples and individuals? Do individuals and couples have
sufficient information to formulate and sufficient means to realise those choices? Are there
factors that systematically hinder certain people from being able to make fertility decisions
freely and responsibly, such as by ethnicity, religion, disability or migratory status? How do
fertility decisions relate to marital status and how do these relationships change over time
(see Chapter 6) ?

69. Fertility is a key variable for gender analysis, as it may reveal situations of vulnerability
for women. Wherever possible, fertility should be considered in tandem with household
composition and number of prior children by sex in order to capture some of the complexity
of gender relations within households. Gender relations may be a factor in explaining fertility
behaviour (e.g. the status of women, marital status, son preference and unwanted fertility).
And conversely, fertility may have consequences for the status and empowerment of women
(e.g. childbearing at young ages, labour force participation, differences in birth and death
rates by sex). In this way, women’s lower status shapes fertility outcomes at the same time
that fertility may also adversely affect women’s and girls’ status vis-a-vis their male
counterparts in society.

70. There are several issues generally associated with fertility. First, gender issues can be
understood as explaining fertility behaviour in terms of the status of women, marital status,
son preference and unwanted fertility.

a) Status of women: The status of women is recognized as one of the main determinants
of fertility. This is the case of both academic studies on the effect of, for example, female

33
 education or of female disadvantage in inheritance systems in Sub-Saharan Africa on
fertility rates (e.g. Cosio-Zavala, 2002; Jejeebhoy, 2001; Mason et al., 1995; Presser,
1997; Sathar, Callum and Jejeebhoy, 2001) and of the political consensus of the ICPD. It
provides a leading rationale for the prominent role that gender issues play in the ICPD
Plan of Action.

b) Marital status: Women in polygamous and common law marriages may have fertility
levels different from those of women in formal monogamous unions, whereas the
instability of informal unions, associated with changes in partners, may cause women to
have more children than they otherwise would, due to the perceived need to have a child
with a new partner.

c) Son preference: In societies where there is a strong preference for sons, couples will,
on average, have higher fertility than they would have in the absence of son preference
because they tend to continue having children until at least one of them is a boy (e.g. Chu,
Xie and Yu, 2007). Conversely, the transition to very low fertility may exacerbate the
manifestations of son preference, such as sex selection at birth.

d) Unwanted fertility: Research in recent decades has examined the fertility desires of
women and, to a lesser extent, men, to determine how much of their actual fertility is
wanted and how much is caused by lack of access to fertility control methods or other
factors. Studies have also examined the degree to which unwanted fertility is the result of
women and men having different fertility desires; in particular, whether men have higher
fertility goals than women (e.g. Andro and Hertrich, 2001), so that large numbers of
children correspond to the desires of men, rather than women. Although this issue cannot
be investigated with the use of census data, it will be briefly taken up in the Interpretation
section.

71. In addition, gender issues can be understood by considering the consequences of fertility
for the health and quality of life of men and women. Apt examples discussed in turn include
childbearing at young ages, labour force participation, and differences in birth and death rates
by sex.

a) Childbearing at young ages: Early childbearing has been shown to have a negative
impact on the health and opportunities of both men and women, especially on the
education of young women.

b) Labour force participation: There are possible conflicts between the reproductive and
productive roles of women throughout their life cycle, in particular the length of the
period during which the economic activity of women is limited by the need to care for
dependent children. In countries where men participate more directly in child-raising, it
may make sense to ask the same question for men.

c) Childlessness: In many societies, the inability of a married woman to have children can
negatively affect her standing in the family and in the wider society. In some societies, it
may even be considered a legitimate motive for divorce (Ola, 2009; Inhorn & Van Balen,
2002). Even if the reason for a couple’s childlessness lies with the man, the consequences
of this situation may be more severe for the wife than for her husband. In modern
societies, on the other hand, childlessness is clearly on the rise, as is discussed in some
detail in the results of the 2010 census of Finland, where as many as 21 per cent of all

34
 women aged 40 had never given birth to a child, up from 15 per cent in 1990 (UNECE,
2012 b). Two thirds of these women were unmarried. By comparison, in the Cambodian
census of 2008, 9.9 per cent of women aged 40 were childless, whereas the Ethiopian
census of 2007 enumerated 7.0 per cent of childless women of that age.

3. Data issues

72. Censuses provide some basic information to investigate the status of women, marital
status, and differences of birth and death rates by sex, especially to investigate these issues by
geographic or social subdivisions. Fertility patterns, such as the mean number of children
ever born or the mean age at childbearing can be analysed by women’s level of education and
other socio-economic characteristics that censuses usually provide. Although the Principles
and Recommendations for Population and Housing Censuses, Rev. 2 (United Nations, 2008
a) recommends that tabulations of children ever born should be disaggregated by sex, not all
censuses follow this practice.8 In the case of the 2009 census of Viet Nam, for example, this
information was used for an in-depth study of sex ratios at birth and their variation by
background characteristics (UNFPA, 2010 c) (see Chapter 5).

73. The population for which data on fertility should be collected consists of women 15 years
of age and over, regardless of marital status, unless from a cultural standpoint it is not
feasible to collect information on childbearing from never-married women. The Indian
census, for example, links the fertility question to marital status (only currently married,
widowed, divorced or separated women). Similarly, Indonesia, Mauritius the Occupied
Palestinian Territories, Sudan (northern part of the country) only ask the question to ever-
married women and the Republic of Korea and Nepal only to currently married women.
Asking fertility questions to girls under age 15 may also be problematic and the results tend
to be unreliable. Botswana, Bulgaria, Burundi, Cabo Verde, Cameroon, Colombia, Congo,
Costa Rica, Djibouti, Ecuador, El Salvador, Ghana, Guatemala, Guinea Bissau, Kenya,
Lesotho, Liberia, Malawi, Mali, Mexico, Mozambique, Peru, Sudan, Swaziland, Togo and
Zambia use a lower limit, of 12 years. Nicaragua uses a lower limit of 13. Aruba, Sint
Maarten and Trinidad and Tobago 14, whereas Brazil Burkina Faso, Ethiopia, and Indonesia
ask the question to girls as young as 10. Cook Islands, French Polynesia, Guam, Iran, Ireland,
and Thailand do not specify an age limit.

Methodology Box 4: Measurement of Fertility from Census Data

Unlike specialized fertility surveys, which usually collect the detailed fertility histories of
each woman, the censuses of most developing countries summarize this information in three
or sometimes four questions:

a) How many live-born children9 have you had during your entire life time?
b) How many of these children are currently alive?
c) What was the date of the last live-birth or how many children did you have during
the past 12 months?

8
For example, Argentina, Azerbaijan, Bahamas, Costa Rica, Kazakhstan, Mexico, Palau, Peru, the Seychelles
and Thailand ask the traditional fertility and child survival questions, but do not disaggregate them by sex.
9
For a common definition of live birth, see: http://unstats.un.org/unsd/demographic/sconcerns/natality/ natmeth-
ods.htm.

35
Depending on the particularities of the census, more details can be provided. For example,
most censuses disaggregate the births by sex, but some don’t. Some censuses divide the
children currently alive by whether they live in the household or elsewhere, in order to avoid
the erroneous classification of children living elsewhere as deceased. In addition, some
censuses10 ask if the last-born child or the child born during the past 12 months is still alive.

However, even with these added details, census fertility data suffer from systematic
problems. On the one hand, older women tend to omit some of their children, especially
children that were born long ago and did not survive. On the other hand, women of all ages
may have difficulty in correctly identifying the reference period of 12 months for the most
recent births. Rather than declaring births during the past 12 months, they may declare
children born during the current calendar year, the previous calendar year or both. Finally, if
fertility levels in the country have changed, the question about children ever born provides
data on historical levels of fertility and questions about the date of the last live birth or the
number of children born in the past 12 months provides data on current levels of fertility.

The most common correction that demographers apply to census fertility data is the so-called
P/F ratio correction method, in which it is assumed that current fertility data (F) provide the
correct pattern of fertility by age of the woman, but may be systematically over- or under-
stated at all ages, due to errors in the reference period. It assumes that over- or under-
reporting of births does not change with age. The cumulative fertility data (P) are assumed to
be more reliably reported for younger women (especially the 20-24 age group) than in current
fertility data. The method consists in assessing the correct age pattern based on F and
adjusting its level based on the reported cumulative fertility (P) of women aged 20-24 and/or
25-29. It should also be noted that the reported fertility data for women of ages 15-19 is
usually ignored in indirect estimation techniques because of data quality concerns:
underreporting of (out-of-wedlock) births or age-shifting of young mothers often affect the
data for this age group.

In practice, this involves some additional complications. For example, the current fertility of
women aged 20-24 does not really refer to ages 20-24 because it captures fertility during the
past year when these women were, on average, half a year younger. Hence it is necessary to
devise a mathematical adjustment to transform these data from the (19.5-24.5) to the (20-25)
interval. It may also be necessary to apply a correction to take account of the fact that the
cumulative fertility of women aged 20-24 may actually refer to a period 2-4 years before the
census, when (at least in a context of declining fertility) it was still slightly higher than at the
date of the census. There is also a variant of the method which uses information on the
duration of marriage instead of the age of the woman. In contexts where almost all births
occur to married women, this method is considered more accurate. In situations where this is
not the case, the date or age of the woman at the time of the first birth is sometimes used as a
proxy. In any case, the number of censuses that ask for either of these data is relatively small.

While the P/F correction method based on the latest census is the most common technique
used by demographers, there are other ways to analyse census fertility data, such as the
comparison of average numbers of children ever born to women in different age groups in
successive censuses. The Own Children Method is also a common method used with census
data. It matches children to mothers at the household level and makes it possible to link the
10
Brazil, Botswana, Burundi, Cayman Islands, China, Djibouti, Dominican Republic, Ecuador, El Salvador, Fiji,
Iran, Liberia, Malawi, Maldives, Occupied Palestinian Territories, Republic of Congo, St. Lucia, Samoa, Sudan,
Swaziland, Tokelau, Trinidad and Tobago.

36
characteristics of mothers and fathers to those of their children. Although the method is
labour-intensive, it results in more robust estimates. Interestingly, the Own Children Method,
while mainly intended as an alternative fertility measurement method for developing
countries, has sometimes been applied in developed countries because it allows analysis of
characteristics of the mother (e.g. race or religion) that are not often included in data from
civil registration systems. See United Nations (1983) for more details on fertility adjustments
and Cho, Retherford and Choe (1986) for the Own Children Method.

74. Despite its limitations, the census can provide small area data that other data sources,
such as fertility surveys, cannot. Even in countries with complete civil registration, some of
the topics addressed in the census, such as the woman's religion, or job status provide
valuable background data on her fertility history that are not available from registration data.
This is particularly the case for some of the additional questions that censuses sometimes ask
on fertility. Apart from the standard questions on children ever born alive and children
currently surviving, these additional questions may include:

a) Date of birth of the last live-born child (instead of or in addition to the number of
children born during the last 12 months);

b) Age of the mother at the time of birth of the last live-born child;

c) Date (year) or age of the mother at the time of birth of the first live-born child (e.g.
Cape Verde, Colombia, Ecuador, Kazakhstan, Peru, Russia, South Africa and six
countries in the Caribbean);

d) Date or age of the woman at the time of first marriage (see Chapter 6).

75. The 2007 census of the Republic of Congo had an unusual format for the question on
recent births in that it asked for births occurred during the past 12 months in the household
and then linked the children to their mothers, rather than asking each individual woman for
her live births during the past 12 months. This format may take away some of the
embarrassment of having to ask this question to very young or single women, but it increases
the risk that children will be attributed to the wrong mothers or that belong to mothers that
currently do not reside in the household.

76. For estimating age-specific fertility rates and other fertility measures, data on “date of
birth of last child born alive” are more accurate than information from questions on the
number of births to a woman during the 12 months immediately preceding the census because
the latter information may be affected by time reference errors.

77. Some censuses ask about the father and/or mother of each person residing in the
household, such as:

a) If the father/mother of the person are still alive (paternal/maternal orphanhood);

b) If the father/mother of the person reside in the same household (e.g. Aruba, Barbados,
Cape Verde, South Africa); and
c) If so, the identification of the person's father/mother.

While these questions are primarily intended for adult mortality analysis, through the so-

37
called orphanhood method (see Chapter 4), they can also be useful in the study of fertility, in
particular for the application of the Own Children Method. In addition, they can be useful for
residential pattern analysis (see Chapter 7).

78. A limited number of countries (i.e. Bhutan, Cambodia, Samoa) also ask what kind of
assistance the woman received during her last childbirth. The skill level of the birth attendant
is an important indicator of the quality of care that is used, for example, in indirect estimates
of maternal mortality. But this information is usually collected through fertility surveys and
the number of countries that have it in their censuses is too small to develop specific
methodologies for its analysis based on census data. The Hungarian census of 2011 is unusual
in that it asks for the number of children ever born not only of women, but also of men. In
addition, it asks for the dates of birth, not only of the last child, but of all children.

79. Regarding male fertility, some countries, such as Norway and Sweden, collect and
publish information in their administrative census on number of children for both men and
women, Unfortunately, measuring the fertility of men requires good vital statistics that
register not only data on the mother, but also on the father of the child. Few countries ask
men about the children they have fathered. Even in the case of Norway, which has such
statistics, there is still a significant percentage (1.9-4.5 per cent in the period 2000-09) of
cases in which no data on the father exist. Even without such information, it is possible to
approximate male fertility with census data, but the results are less reliable and the methods
for doing this are not well developed.

80. The problems of measuring male fertility through the census are basically two:
a) It would require two additional questions for all men above age 15 (or 12), regarding 1) All
children fathered during the man’s life (a few censuses actually do ask this, e.g. the 2010
census of Bermuda); and 2) Idem during the past 12 months or some other recent period.
b) Even in the case of women, a question is often not reliably reported, especially in the case
of children that died while still young or that do not live with the mother. In the case of men,
these problems are likely to be much more severe. Both men and women may want to hide
children had with previous partners from their current spouses, but this problem is likely to be
much more common in men, both because they are more likely to have had previous spouses
and because the children born out of such relationships are less likely to live with them. In
some cases, they may even be unaware of the existence of these children.

Country Example 1: Male Fertility in Norway

Below are male and female age-specific fertility rates (per 1,000) in Norway (2009):

Ages Women Men

15-19 9.5 2.5
20-24 61.4 27.7
25-29 128.1 86.3
30-34 127.2 117.6
35-39 58.3 76.7
40-44 10.2 30.1
45-49 0.5 9.4
50-54 0.0 3.2

Source: Vital registration data from Statistics Norway

38
Inspecting the table above, the main difference between the fertility of men and women is
that men have children at older ages than women. Even for women, fertility in Norway is
already quite late (mean age of 30.0), but the fertility of men, on average, is almost three
years later than that of women (mean age of 32.9). Moreover, whereas the fertility of women
becomes negligible after age 45, some men still continue to have children in their fifties.

However, if male fertility rates are needed and none are available through the civil
registration system, a reasonable proxy can be obtained by shifting the female fertility curve
up by a number of years. The difference between the Singulate Mean Ages at Marriage
(SMAMs) of women and men may be a good initial value for this age shift, although it does
not take account of successively larger age differences in later unions. A better approximation
may be the mean age differences between women aged 15-39 and their spouses (if they live
with a spouse).

4. Tabulations

81. Differences in fertility levels and trends between two or more subgroups of the population
are particularly useful for insights about gender issues. The differences may be between
socioeconomic groups, geographical groups or the same group at two different points of time.
Differences can be categorised as compositional differentials, spatial differentials or temporal
differentials.

82. The Principles and Recommendations for Population and Housing Censuses Rev. 2
(United Nations, 2008 a) recommend three basic tabulations for fertility analysis and list
another six additional possibilities:

Recommended tabulations for population censuses:

1. Female population 10 years of age and over, by age and number of children ever born
alive by sex;
2. Female population 10 years of age and over, by age and number of children living (or
dead) by sex;
3. Female population … to 49 years of age, by age, number of live births, by sex within the
12 months preceding the census, and deaths among these live births, by sex;

The dots in recommended Tabulation 3 indicate an age which may vary from country,
generally 15 years or 12 years, sometimes 10 years. These tables can obviously only be
generated in countries that have this information in their census (see the Text Box preceding
Section 3).

83. The Principles and Recommendations also include the following additional tabulations
for population censuses:
1. Female population 10 years of age and over in their first marriage/union or married only
once, by five-year duration of marriage/union group and number of children ever born
alive by sex;
2. Female population, by age at first birth, by current age and place of residence;
3. Median age at first birth, by current age of women, place of residence and educational
attainment;
4. Mothers 10 years of age and over with at least one child under 15 years of age living in

39
 the same household, by age of mother and by sex and age of children;
5. Female population … to 49 years of age, by age, number of live births by sex within the
12 months preceding the census and educational attainment.
Of these, only 4 and 5 can be compiled in most censuses as the first three require information
that relatively few censuses collect.

84. The table below, generated from the 2008 census of Cambodia, illustrates the cross-
tabulation of fertility data by educational levels. Rather than using live births by sex within
the 12 months preceding the census, the table shows the number of children ever born. The
ideal procedure is to combine both (children ever born and children born during the past 12
months) to compute actual accumulated fertility rates, but because the number of children
ever born is easier to understand, the table has been left in this format. What it shows is that
in Cambodia fertility levels are fairly uniform for women with up to lower secondary
education, but that fertility levels decline substantially as women complete their secondary
education.

Table 3: Cambodia 2008 - Average numbers of children ever born classified by current
age of the mother and highest level of education completed by the mother

Phnom Highest grade completed

Penh
Age None Incomplete Complete Lower Secondary/ Beyond
Group Education Primary Primary Secondary Technical Secondary
15 – 19 0.07 0.04 0.03 0.02 0.02 0.01
20 – 24 0.40 0.28 0.26 0.20 0.09 0.07
25 – 29 0.98 0.87 0.85 0.76 0.62 0.41
30 – 34 1.65 1.56 1.59 1.49 1.37 1.11
35 – 39 2.28 2.24 2.21 2.06 1.70 1.60
40 – 44 2.86 2.80 2.63 2.41 1.98 1.71
45 – 49 3.40 3.22 3.08 3.02 2.37 1.99
50 – 54 3.49 3.38 3.33 3.20 2.67 2.22
55 – 59 3.52 3.44 3.42 3.13 2.45 2.28
Rest of Country
Age None Incomplete Complete Lower Secondary/ Beyond
15 – 19 Education
Group 0.16 0.09
Primary 0.05
Primary 0.03
Secondary 0.03
Technical 0.04
Secondary
20 – 24 0.94 0.74 0.56 0.32 0.13 0.10
25 – 29 1.82 1.60 1.40 1.07 0.72 0.55
30 – 34 2.74 2.55 2.33 1.99 1.69 1.48
35 – 39 3.51 3.36 3.04 2.67 2.18 2.09
40 – 44 4.18 4.01 3.52 3.20 2.53 2.47
45 – 49 4.65 4.45 3.98 4.16 2.96 2.55
50 – 54 4.57 4.64 4.36 4.33 2.95 2.88
55 – 59 4.66 4.74 4.43 4.31 3.01 2.95

85. Beyond the additional tabulations suggested by the Principles and Recommendations,
there may be others of potential relevance for gender issues. In countries where a significant
proportion of unions are informal, consensual or polygamous, it may be relevant to tabulate
the basic fertility data by type of union. Ideally, this information should be compiled by
duration of the union, if available, rather than only by the woman's age. Consensual unions

40
are usually associated with higher fertility than formal marriages (Henriques, 1979), even
after controlling for other factors. Marital instability may play a role in increasing fertility
rates as women feel the need to have at least one child with each new partner (Chen, Wishic
and Scrimshaw, 1974). The latter, however, may be difficult to investigate using census data
because the census generally does not provide any information on marital histories or even
information on whether a woman has been married or in union before her present union.

86. Fertility levels vary by income in ways that are gender-specific. Paternal income tends to
be positively associated with fertility, whereas maternal income is generally inversely
correlated with income. The association is clearer when one of the variables is kept fixed
while the other varies. This is understandable, given that high-income women suffer greater
loss of income by having children. On the other hand, a higher income of the husband, given
a certain income level of the wife, makes it more attractive to have children.

87. Given the social consequences of childlessness for women, it is recommended to prepare
a table of women by number of children (or, at a minimum, women with children or without
children) by 5-year age groups and marital status. Apart from the fact that such a table may
identify the women at risk of social ostracism, it may allow the detection of actual trends,
such as the higher incidence of divorce among women without children. The latter must be
interpreted with caution, however, because the relationship may also go the other way, as
women who divorced at an early age have had less time to become mothers.

88. Polygamous unions generally have lower female fertility rates than monogamous unions,
due to lower coital frequency, especially older spouses, but they tend to have higher male
fertility rates (Anderton and Emigh, 1989; Garenne and Van de Walle, 1989; Lardoux and
Van de Walle, 2002; Shaikh, Aziz and Chowdhury, 1987). There are, however, certain forces
that work in the opposite direction and that may bring about different results in some cases.
For example, polygamy is associated with women's low status and inequality within
marriage, which is often further exacerbated by large age difference between husbands and
their second or third wives. One consequence of male dominance and unequal husband-wife
interaction within a polygamous marriage is lower contraceptive use, mediated by the
husband's disapproval of and lack of spousal communication about family planning (Hogan
et al., 1999). It has also been observed that fertility may vary considerably between the first
and later wives (Bean and Mineau, 1986). In addition, mothers depend on their children in
later years, while fathers can receive support from younger wives. The number of children
ever born, cross-tabulated by age and marital status (monogamous vs. polygamous union)
should show which way these differences end up going.

89. A significant proportion (40-60 per cent) of polygamous unions in the Arab countries
involve widows and divorced women who find it difficult to remarry as a first wife and
therefore accept the status of second wife (Chamie, 1986). Unfortunately, census data
normally do not contain information about the marital status of women before their present
union, so that this relationship cannot be investigated. What can be investigated is the
percentage of polygynous unions by level of education of the husband. Less educated men
are more likely to be in polygynous unions, but some caution is called for as the relationship
is partly explained by age effects. Older men are more likely to be in polygynous unions and
they are also more likely to have a lower level of education. Therefore, it is best to tabulate
the relationship by age groups, to separate the age effect from the educational effect.

90. Early marriage creates the conditions for early pregnancy and higher fertility rates. In

41
societies where premarital childbearing is not socially and culturally accepted, a rising age at
first marriage may play a crucial role in the transition from high to low fertility levels
(Maitra, 2004). A study by UNICEF found that high levels of fertility were associated with
the prevalence of child marriage in 50 countries (UNICEF, 2005). Women who had several
children were significantly more likely to have been married before age 18 than women with
no children. The relation between early marriage and fertility can be studied by cross-
tabulating the age at first marriage, if available, with the number of children ever born, by age
group or cohort. This will show the differences of fertility for women of the same age, or age
group. In addition, the effect of early marriage on early pregnancy can be measured by
tabulating the number of children ever born for young married women, by age.

91. Another tabulation of potential interest is the association between religion (when
available) and fertility. Pronatalist ideologies, associated with different religious traditions,
can have stances against the use of contraception (or certain types of methods) and abortion.
When people have a strong attachment to religious communities, religious directives on
gender roles and sexuality are likely to influence their fertility behaviour, although the extent
to which religious convictions influence behaviour varies greatly. Verifying whether religious
affiliation is associated with fertility rates and family size helps to understand cultural values
with respect to gender, moral communities and social constraints toward women’s
reproductive rights. The precise influence of religion is subject to a certain amount of
controversy. Goldscheider (1999), for instance, maintains that Moslim women in Israel have
higher fertility than the Jewish population not because of any specific prescriptions of the
Moslim faith with respect to fertility and the use of contraception, but because of its views on
the nature of familial relationships and the segregated social of roles of women which tend to
emphasize their roles as mothers and spouses. Jejeebhoy (1995) has collected a lot of
evidence that shows a negative relationship between the autonomy of women and fertility
levels. Obermeyer (1992), on the other hand, does not accept the notion that Moslim women
have less autonomy than women of other religions in similar contexts or that, on account of
this lesser autonomy, they have higher fertility. Religion is sometimes used as an explanatory
variable for contraceptive use (e.g. Addai, 1999 in Ghana; Adsera, 2006 in Spain), but
because the census does not contain any information on contraceptive use, such studies are
limited to data from the DHS or other types of fertility surveys and cannot be replicated with
census data.

92. As an example, consider Table 4.A, which shows how the number of children ever born in
Cambodia (2008) varies by the age of the woman, her area of residence (Phnom Penh or
elsewhere in the country) and religion. Islamic women clearly have the highest fertility. In
Phnom Penh, Christian women have the lowest fertility, but in the rest of the country their
fertility is the next highest, after that of Islamic women. A study by Adsera (2006), on the link
between marital fertility and religion in Spain in women aged 15-49 for the period between
1985 and 1999, found that fertility was particularly high among women in minority religious
communities, such as conservative Protestant denominations and Muslims. It was lower
among women in inter-faith marriages than among women in homogamous unions,
particularly in unions where the husband was not Catholic. The study was based on two
rounds of the Spanish Fertility Survey (SFS), but it would be possible to replicate it using
census data.

Table 4.A: Cambodia 2008 - Average numbers of children ever born classified by
current age of the mother and major religious groups

42
Phnom Penh Religion
Age Group Buddhist Islam Christian Other
20 – 29 0.47 0.70 0.46 0.40
30 – 39 1.85 2.55 1.55 0.93
40 – 49 2.86 4.14 2.28 1.50
50 – 59 3.30 4.54 2.70 2.76
Rest of Country
Age Group Buddhist Islam Christian Other
20 – 29 1.10 1.18 1.10 1.88
30 – 39 2.88 3.24 2.99 4.22
40 – 49 4.19 4.81 4.21 4.81
50 – 59 4.58 5.01 4.76 4.21

Table 4.B: Cambodia 2008 - Average numbers of children ever born classified by
current age of the mother and economic activity status

Phnom Penh Economic Activity Status

Age Group Employed Unemployed Never work- Home Retiree / Other
ed / Student maker Rent Holder
20 – 29 0.42 0.44 0.20 1.15 0.49 1.01
30 – 39 1.69 1.60 1.56 2.31 1.32 1.85
40 – 49 2.70 3.02 2.53 3.23 2.32 2.78
50 – 59 3.14 2.84 3.33 3.59 3.09 3.16
Rest of Country
Age Group Employed Unemployed Never work- Home Retiree / Other
ed / Student maker Rent Holder
20 – 29 1.12 1.13 0.24 1.42 0.44 1.27
30 – 39 2.94 2.93 2.25 3.02 1.51 2.29
40 – 49 4.21 4.20 3.70 4.26 2.49 3.19
50 – 59 4.63 4.67 4.24 4.66 3.29 3.73

93. Table 4.B shows the number of children by activity status of the woman. As expected, the
highest average is found among women classified as “Home makers”, at least in the capital.
In the rest of the country, the differences are minor. The lowest average is found among
retirees and persons living off rents. It is noteworthy that unemployed women have the lowest
fertility in the capital, but the highest in other parts of the country. Employed women have
intermediate fertility in the capital, but are nearer to the top in the rest of the country.

5. Indicators

94. There is no shortage of fertility indicators and a basic understanding of their meaning is
important for gender analysis. While the relevance of most fertility indicators for gender
analysis may be somewhat indirect, the following paragraphs describe the most common
indicators first and then specific indicators that may be more useful for gender analysis. Most
of the indicators discussed here do not follow directly from the tabulations discussed earlier
and need some further computations.

95. Age-Specific Fertility Rates: The Age-Specific Fertility Rate (ASFR) is the number of
births to women of a given age group per 1,000 women in that age group:

43
 Births during the year to women aged (x, x  n)
ASFR(x, x  n)   1000
Women aged (x, x  n) at mid  year

One needs data on the total number of live-born children by current age of the mother, the
numbers of children born during the past 12 months and some amount of model-based

computations, as explained in the Text Box at the beginning of this sub-chapter. A graph of
the ASFRs helps analyse the fertility pattern of a country compared to other countries or over
time. For example, the following graph shows the ASFR in 2005-2010 globally, for the more
developed countries and the least developed countries. It shows that the maximum of fertility
is reached at aged 20-24 for the least developed countries, with 225 children per 1,000
women in that age group, but for the more developed countries the peak of fertility is at 25-29
years, with 99 children per 1,000 women in that age group.

Figure 2: Age-specific fertility rates (2011)

250

200

150

100

50

0
15-19 20-24 25-29 30-34 35-39 40-44 45-49

Developed Developing Excluding LD

Least Developed (LD) World

Source: United Nations (2011 b). World Population Prospects, the 2010 Revision

96. Total Fertility Rate: The Total Fertility Rate (TFR) summarizes the information given by
the ASFRs. It is defined as the number of children that a woman would have over her
childbearing years if, at each age, she experienced the current ASFRs. If the ASFRs are
defined by single ages, it is simply the sum of all of them. If the ASFRs are defined by 5-year
intervals, it is the sum of the ASFRs multiplied by 5. The TFR is a synthetic indicator (it uses
women who live at this moment as members of a fictitious cohort), to compare the level of
fertility over time, and among countries or within a country. In the period 1950–2010 the
TFR in the world decreased from around 5 children per woman to around 2.5, with major
regional differences. In Central America, Eastern Asia and Northern Africa, the TFR
drastically declined (from 6.7 to 2.3, 5.6 to 1.6 and 6.9 to 2.8, respectively), whereas in some
parts of Africa, especially Western and Middle Africa, the decline was relatively modest and

44
the TFR remains above 5 children per woman. Part of the explanation is that contraceptive
use in Africa was in 2007 considerably lower than elsewhere, with only 28 per cent of
women of reproductive age who were married or in union using any method (United Nations,
2011 a).

97. The TFR is different from the life time fertility of a cohort, which represents the average
number of children of all the women born in a given year (who constitute a cohort) and
having survived until the age of 50 at least. In theory, this figure can be directly estimated
from census data from the question on number of children ever born alive. Of course, this
indicator can be calculated only for generations having completed their reproductive life, that
is to say women aged 50 or over. Yet older women often underestimate the number of
children that they have had during their life time, as they tend to omit children that died while
still very young. In the more developed countries, there has been a major divergence between
TFRs and the life time fertility of women in recent decades as younger women, who were
increasingly active in the labour market, postponed childbearing until they were in their
thirties. The short term effect of this was to depress the fertility of younger women, while the
higher fertility of women in their thirties was to materialize only later. Consequently, Europe
passed through a period of extremely low TFRs in the 1990s, even though the life time
fertility of women later turned out to have changed much less.

98. General Fertility Rate: Closely related to the Total Fertility Rate is the General Fertility
Rate (GFR). But unlike the TFR, which is an index, rather than an actual rate, the GFR is a
true rate, defined as the number of births that occur during a year, divided by the average
number of women of reproductive age (15-49 years) during the year. In other words, it is the
proportion of women of reproductive age that will have a childbirth during a given year.

99. The Parity Progression Ratio at parity n is the proportion of women who already have n
children who will go on to have n+1. It can also be refined, to reflect not only the number but
also the composition of existing children (see Chapter 5). In practical terms, it is computed in
the same way as the TFR, but limited to women who have had exactly n live-born children,
regardless of whether they are currently alive or not. It is usually computed from DHS data.
Computing it based on census data requires special care, due to the limitations of census data
explained in Methodology Box 4. A reasonable approximation is to compute the ratio as
described above, based on children born during the past 12 months, and then multiply the
result by the same correction factor that was applied to obtain the corrected TFR from the
apparent overall ASFRs (not specific by birth order) of the census. Alternatively, one may
compute the percentage of women aged 45-49 years who have more than n children among
those that have at least n. The disadvantage of the latter method is that it reflects the fertility
experience of older women, which may not be entirely representative of current fertility. The
following table from the 2000 census of Cape Verde was computed using this latter
procedure.

Table 5: Cape Verde (2000) – Urban and rural parity progression ratios

45
Parity Urban Rural
0 0.947 0.955
1 0.925 0.964
2 0.904 0.953
3 0.874 0.939
4 0.847 0.908
5 0.810 0.865
6 0.758 0.825
7 0.694 0.770
8 0.648 0.701
9 0.648 0.642
10 0.566 0.546

Source: INE, Cape Verde

Given the way the computation was carried out, this means, for instance, that 70.1 per cent of
the rural women in Cape Verde who had at least 8 children actually had more than 8, whereas
29.9 per cent had exactly 8.

100. Adolescent Birth Rate: The adolescent birth rate is the ASFR for women aged 15-19.
Fertility levels among women in this age group are relevant to the status of women, since
women who bear children early in life often forego the opportunity to study or find
employment outside the home, in addition to consequences for health and human rights noted
earlier. Maternal mortality increases steeply for progressively younger mothers under age 18.
The proportion of school leavers among young mothers is also higher than among
adolescents who do not have children and are not pregnant, even though the direction of the
cause-effect relationship in this case is not clear. Because of its potential negative effect on
the education and employment of young women, the adolescent birth rate is one of the
indicators for the monitoring of Millennium Development Goal 5.B. It is also the only
fertility-related indicator in the Minimum Set of Gender Indicators approved by the UN
Statistical Commission in February of 2012 that can be computed from census data. In the
more developed regions, the average adolescent birth rate in 2005-2010 was 24.0 births per
1,000 women aged 15-19. In developing countries the range of variation of the adolescent
birth rate is considerably larger: from below 5 to over 200 births per 1,000 women aged 15-
19. The highest rates are recorded in Niger (207.1), the Democratic Republic of the Congo
(201.4), Mali (186.3), and Angola (171.1) (United Nations, 2011 b).11

101. As was mentioned in Chapter 2, an alternative indicator sometimes used for quantifying
adolescent fertility is the proportion of all births that occur to adolescent women. Unless the
purpose of this indicator is to plan services, to make them more adequate for the age profile
of the typical client, this indicator is not recommended because it can convey seriously
misleading impressions with respect to the time trend of adolescent fertility.

102. The 2010 Human Development Report of UNDP uses the adolescent birth rate as one of
the five indicators of the Gender Inequality Index (GII). One drawback is that no attempt was
made to measure to what extent early fatherhood affects adolescent boys and what
consequences this may have for their future. While it is known that adolescent fatherhood is
less common than adolescent motherhood, as many of the fathers of these children are older
11
Note, however, that the adolescent birth rates used for monitoring MDG 5.B are estimated independently and
not based on the UN population projections.

46
than 20, omitting the information makes it impossible to draw meaningful comparisons with
the situation of men. The GII, therefore, presents adolescent fertility as an exclusively female
problem.

103. In agrarian societies, where children are valued for their labour, family continuity and as
an insurance against the risks of old age, women are valued for their ability to bear children.
Hence, a woman who is unable to bear children is stigmatised as being inadequate and having
failed the husband’s family and clan. Such women are very likely to end up in a polygamous
union as the husband procures another wife who can bear children, or may end up divorced
altogether. Either situation is not desirable for a woman, as it communicates her lower status
in relation to a man’s. In the 2008 census of Malawi, infertility was used to refer to women
who had not had a child by the age of 45 years. In this census, infertility was observed to be
on the decline, from the 4.1 per cent recorded in 1987 to 3.6 per cent recorded in 2008. A
DHS study by Rutstein and Shah (2004) used a stricter criterion, limited to women aged 40-
44 who had been married for at least five years. Consequently, they found a lower figure, of
1.6 per cent. It is recommended that census analyses should at least exclude women who are
not currently married or in unions. In some African countries, such as Chad, the incidence of
infertility, according to the DHS criterion, can be much higher (7.3 per cent). It can also be
substantially higher among some specific ethnic groups.

104. Time spent caring for dependent children. One of the major implications of high fertility
is that it ties one of the adult household members, usually the mother, to the home for many
years, to care for dependent children. How many years of care each child needs depends on
the circumstances of each country. In countries where education is universal and starts at the
pre-school level, women will often be able to take on responsibilities outside the home as
soon as their children reach age 3. In other countries, where pre-school education is non-
existent and primary school education deficient in its coverage, they may have to care for
them full-time until the child is age 8 or 9. Apart from the number of children, the spacing
between children matters. If fertility is concentrated in a relatively short age range, the time
during which there are dependent children in the home is compressed. If fertility is spread out
over a longer age range, women will spend most of their reproductive years caring for
dependent children. Finally, time spent caring for children depends on infant and child
mortality.

105. A measure of the number of years spent caring for dependent children can be computed
from census data. To this end, one needs to determine for each woman whether she has
children of her own, below the accepted age for the country after which they are no longer to
be considered predominantly dependent on the mother. Alternatively, one may assume that
children going to school are no longer (completely) dependent on their mothers. In more
complex households there may be problems in establishing which children belong to each
potential mother.12 Once this has been established, one determines the proportion of women at
each age who are caring for (their own) dependent children. Summing these proportions over
all ages yields the mean number of years that women spend caring for their dependent
children. This assumes both that women are the ones caring for their dependent children and
that they only care for their own children. One may refine the method by applying it not only
to the mothers of children, but generally to all adult women present in the household that do
not work or study, assuming that adult males, even if they do not work or study, generally do
not have a major role in caring for dependent children. Except for a small number of censuses
(Australia, Republic of Korea) that ask questions about this topic, there is generally no way to
12
The same problem affects the Own Children Method for fertility estimation.

47
avoid such assumptions.

Figure 3: Scatter plot of the Total Fertility Rate as a function of the female/male ratio
(x100)
Female/male ratio of population with at least secondary Female/male ratio of shares in parliament
education

8.00 8.00

7.00 7.00

6.00 6.00

5.00 5.00

4.00 4.00

3.00 3.00

2.00 2.00

1.00 1.00

0 0.00
0 20 40 60 80 100 120 140 0 20 40 60 80 100 120 140 160

Sources: UNDP Human Development Report 2010 and UN Population Division World
Population Prospects, 2010 Revision

106. Apart from the Adolescent Fertility Rate, the Minimum Set of Gender Indicators
approved by the UN Statistical Commission in February of 2012 contains three other fertility-
related indicators. However, none of these can be computed from census data:
1. Contraceptive prevalence among women who are married or in a union, aged 15-49;
2. Ante-natal care coverage; and
3. Proportion of births attended by a skilled health professional.

6. Multivariate and further gender analyses

107. Caveats for fertility correlates of geographical/subnational units. If subnational

estimates are available for basic gender data, such as various gender inequality measures,
these can be correlated with fertility levels to show the relationship between gender
inequality and fertility. Fertility can also be related to the percentage of female-headed
households or women that participate in the labour force, at the level of different
geographical units. This is not fundamentally different from some of the tabulations shown in
the previous section, except that the objective is not only to show the difference between
population groups but to propose some sort of systematic quantitative relationship between
the characteristics of those groups (e.g. that groups that exhibit greater gender inequality tend
to have higher fertility). There are, however, some caveats with respect to correlating
different indicators across population groups:
a) Care must be taken to avoid using gender inequality measures which already contain

48
 fertility as one of their components, which would result in a tautological relationship. For
this reason, the Gender Inequality Index of the 2010 Human Development Report is not
recommended, as it is already partly based on the Adolescent Birth Rate. In the example
below, two other indicators from the same report have been used, namely the female/male
ratio of population with at least secondary education and the female/male ratio of shares
in Parliament, both applied to countries, rather than subnational population groups. Note
that the indicator of shares in Parliament may not be applicable to subnational analyses; it
is included only for illustrative purposes.

b) The choice of the particular gender indicator is important. In the example shown here,
the female/male education ratio shows a moderately strong negative relationship with the
Total Fertility Rate across 143 countries (r = -0.69). The female/male ratio of
Parliamentary representation has a very weak relationship with the Total Fertility Rate (r
= -0.11).

c) One has to be alert to the possibility that the relationship found may be due to the fact
that both indicators are perhaps determined by a third one, rather than being the result of
any direct causal relationship between them. This issue will be taken up in the next
sections.

108. In examining gender issues, the researcher begins with univariate and bivariate analyses
to define a potential issue or problem as it may relate to a patterned relationship across
women and men, girls and boys. Then, multivariate data analysis can be useful to
differentiate correlations with causations, and to pinpoint what specific variable or variables
may be the causes for a differential status or level of opportunity by sex. The primer box on
multivariate analysis provided in Box 1 within chapter 2 of part 1 of this manual, entitled
Multivariate Analysis to Disentangle Intra Group Variability and Interrelationships, may be
useful to review prior to considering the following cases.

109. Bivariate analyses of statistical relationships can be misleading if they are interpreted to
show a causal relationship. This is illustrated by the example above of the correlation
between the female/male ratio of population with at least secondary education and the Total
Fertility Rate. Although this relationship is moderately strong (r = -0.69), it is weaker than
the correlation of the overall Human Development Index (HDI) with the Total Fertility Rate
(-0.84). The correlation of the HDI with the female/male education ratio is not quite as strong
(0.72), but nevertheless considerable. This begs the question whether the moderately strong
relationship between the first two variables is not simply due to the fact that both are
reflections of the general level of development of the country, as measured by the HDI, rather
than to a specific relationship between the two of them. This indeed turns out to be the case.
To show this, the TFR was estimated as a quadratic function of the HDI. The residuals of this
HDI-based TFR with respect to the actual TFR were then plotted as a function of the
female/male education ratio. The results in the graph below show that, once the association of
the TFR with the HDI is removed, the residuals display an almost completely random pattern
(r = -0.04). This does not mean that the basic idea that greater gender inequality is associated
with higher fertility is incorrect; it just means that the association cannot be statistically
demonstrated in this manner.

Figure 4: Scatter plot of the Total Fertility Rate as a function of the female/male ratio
(x100) of population with at least secondary education after removing the joint
association with the HDI

49
 2.50
2.00
1.50
1.00
0.50
0
0 20 40 60 80 100 120 140
-0.50
-1.00
-1.50
-2.00
-2.50

Sources: UNDP Human Development Report 2010 and UN Population Division World
Population Prospects, 2010 Revision

110. Another illustration of the importance of using multivariate analysis, rather than simple
cross-tabulations, for drawing conclusions on causal links is a study by McKinnon, Potter and
Garrard-Burnett (2008) on differentials in fertility and family formation among adolescents in
Rio de Janeiro. At first sight, the data from the 2000 census used in this study conveyed the
impression that adolescent fertility among young women without religious affiliation was
more than twice that of Catholics and that Pentecostal Protestants also had higher adolescent
fertility rates than Catholics. However, interpreting this finding as an indication of different
religious dispositions towards adolescent fertility would be quite misleading. As it turned out,
Pentecostal Protestants also have higher rates of having lived with a spouse or partner, have
proportionally more non-white members and reside in areas with lower overall mean
household incomes. Therefore, the researchers used a regression model in which the
probability of giving birth was a function of ever having lived with a partner, migrant status,
educational level, age, race, religious composition, mean income and other indicators to
characterize the relative prosperity level of the place of residence. Once all of these
explanatory factors were considered, Pentecostal Protestants actually had a 23 per cent lower
adolescent fertility than Catholics with similar socioeconomic characteristics. Young women
without religious affiliation continued to have a higher fertility than Catholics, even with
these controls, but the difference fell considerably, from more than double to only 29 per
cent.

111. In the 2001 census of Nepal, 43.5 per cent of divorced women aged 15-49 had never had
children. A likely explanation is that childless women had a higher than normal probability of
being divorced by their husbands. However, it is not entirely impossible that the high
incidence of childlessness among divorcees is a result of the fact that many divorces occur at
an early age, before the woman has had an opportunity to bear a first child. The following
approximate procedure provides a way to quantify by how much being childless raises the
odds that a woman will be divorced. It is easiest to apply if the variables are all computed for
a specific age interval, e.g. ages 35-39 (mean age of 37). Basically what the procedure

50
consists of is the computation of the estimation of the number of divorced and childless
women under the null hypothesis of total independence of these two phenomena.

To this end, one needs the proportions CL(i) of childless women by age (preferably in single
years) in the general population and the equivalent proportions D(i) of women that are
divorced by age i. If it is further assumed that there is no childbearing or divorce before age
15, the expected proportion of women that are both childless and divorced by age 37 will be

CL(15)*D(16)/2 + CL(16)*(D(17)-D(15))/2 + CL(17)*(D(18)-D(16))/2 + …….

……… + CL(36)*(D(37)-D(35))/2 + CL(37)*(D(38)-D(36))/4

This can be compared with the actual proportion of women in this condition. The ratio
between these proportions can be interpreted as the factor by which being childless increases
the probability of being divorced.

Note that the procedure is not entirely conclusive in proving that the high percentage of
childless divorcees is due to the discrimination of childless women. There are at least three
other factors that need to be considered in this context:
1. The computation of the age-specific divorce rates is based on transversal observations
and does not reflect the life experiences of true cohorts, which may bias the results.
2. The higher childlessness of divorced women may not reflect discrimination against
the woman because of her childlessness but the fact that many of these marriages
were unhappy ones in which intercourse rarely took place.
3. The procedure also assumes that childbearing is terminated after a divorce, which
may not be entirely justified in societies where informal unions are common.

112. Other gender-relevant contextual factors that have been related to fertility include the
household structure. Moultrie and Timaeus (2001), for example, studied how the household
composition influences the fertility of women of twenty years or older in South Africa. For
these women, they hypothesised that living arrangements mediate between their socio-
economic and cultural characteristics and the number of children that they have borne. The
focus was on whether women lived with a husband, or with relatives of their parents’
generation, or with relatives of their own generation. Living with relatives from the previous
generation was found to have a negligible net impact on the lifetime fertility of mothers.
However, women who lived with relatives from their own generation had borne about 20 per
cent fewer children than other women of the same age after controlling for the impact of
household income, the woman’s schooling, regional differentials and urban-rural residence.
Unmarried and separated mothers had about 20 per cent fewer children than married mothers
of the same age.

7. Interpretation, Policy and Advocacy

113. Two types of data that can almost never be obtained from censuses are the contraceptive
prevalence rate and data on fertility preferences. Both of these require specialized interviewer
training that can be provided in fertility surveys, but that would be too cumbersome for a
census.13 Yet both the contraceptive prevalence rate and fertility preferences can have major
gender implications. “The ability of women to control their own fertility is absolutely
fundamental to women’s empowerment and equality. When a woman can plan her family,
13
In the 2010 census round, however, two censuses have asked a question how many additional children women
intended to have, namely Republic of Korea (2005) and Kazakhstan (2009).

51
she can plan the rest of her life. When she is healthy, she can be more productive. And when
her reproductive rights—including the right to decide the number, timing and spacing of her
children, and to make decisions regarding reproduction free of discrimination, coercion and
violence—are promoted and protected, she has freedom to participate more fully and equally
in society.” (UNFPA web page on Gender Equality).

114. Conflicting preferences between husbands and wives can affect aggregate fertility
outcomes (Voas, 2003). Note the following table that compares the desired family sizes of
men and women in DHS surveys in different parts of the world:

Table 6: Mean ideal family sizes for men and women in the DHS

Women Men Women Men

Armenia (2005) 2.7 3.1 Lesotho (2004) 3.5 4.1
Azerbaijan (2006) 2.6 3.0 Liberia (2007) 5.4 6.3
Bangladesh (1996-97) 2.5 2.4 Malawi (2004) 4.3 4.3
Benin (2006) 5.2 6.9 Mali (2006) 6.4 8.4
Bolivia (2008) 2.6 3.0 Namibia (2006-07) 3.7 4.7
Burkina Faso (2003) 5.8 7.0 Niger (2006) 9.1 12.6
Chad (2004) 9.2 13.7 Nigeria (2008) 6.7 8.8
Republic of Congo (2005) 5.4 5.9 Philippines (2003) 3.2 3.8
Dem. Rep. of Congo (2007) 6.8 8.2 Rwanda (2005) 4.5 4.2
Ethiopia (2005) 5.1 6.4 Senegal (2005) 5.7 8.3
Ghana (2008) 4.6 5.3 Sierra Leone (2008) 5.3 6.8
Guinea (2005) 5.9 8.8 Swaziland (2006-07) 2.7 3.6
Haiti (2005-06) 3.2 3.3 Tanzania (2004-05) 5.4 5.9
Indonesia (2007) 2.8 3.0 Ukraine (2007) 2.0 2.1
Kenya (2008-09) 5.5 4.6 Zambia (2007) 5.1 5.7

Source: DHS Statcompiler

115. The large differences found in some countries, particularly in Subsaharan Africa, may be
one of the reasons why fertility in these countries remains high, although the table above
indicates that women themselves in Sub-Saharan Africa also have high fertility preferences.
Having little property right and being treated essentially as a form of property to be
exchanged for material goods between families, women in the polygamic system of sub-
Saharan countries are especially vulnerable when they become spouseless or childless.
Without the right to inherit the property of her husband, a wife in this system is motivated to
maintain high fertility, hoping that at least one of the surviving children is a son on whose
inherited field she can continue farming after her husband’s death (Boserup, 1970).

116. On the other hand, fertility preferences in some developed countries are significantly
higher than actual fertility levels and the literature suggests that gender factors may play an
important role in the explanation of these disparities (Sobotka, Goldstein & Jasilionine,
2009). Note also that in some countries (Bangladesh, Haiti, Indonesia, Malawi, Peru,
Rwanda, Turkey, Ukraine) the differences are negligible or even reversed. Depending on the
particular application of these data, it may be possible to construct proxies that work
reasonably well for certain purposes. This is particularly the case if the objective is to
combine information on fertility preferences from a DHS with other kinds of information that
are only available or can only be disaggregated at the desired level by using census data. For
some further explanation of how this may be done, see section 6.

52
117. In order to ensure that “all couples and individuals have the basic right to decide freely
and responsibly the number and spacing of their children and to have the information,
education and means to do so” (ICPD, principle 8), two main advocacy strategies have been
found to be successful:

a) Tackling unwanted early pregnancy by providing reproductive health information and

youth-friendly services to young people;

b) Investing in girls’ education and empowerment more generally, in order to benefit the
young women themselves, their future families, their communities, and their countries.
Both strategies can lead to informed reproductive decision-making and delayed
childbearing.

118. Using census data in the areas of education and fertility, advocacy material may be able
to show educational disparities in who has children at young ages, thus making the case for
investments into girls’ education on demographic grounds. Using census and DHS data on
women’s exposure to pregnancy and child birth health risks, contrasted with information on
government funding for reproductive health care, advocates can lobby Ministries of Health
and other government decision-makers to change their budgeting decisions to improve
women’s health.

53
Chapter 4:
Mortality
1. What is it?

119. Mortality is the one major demographic variable in which men are almost universally at
a disadvantage compared to women. The one major exception to this is the Indian sub-
continent (specifically, India, Bangladesh and Nepal) where, prior to the 1990s, the life
expectancy at birth for males exceeded that of females. Up to this day, the male-female life
expectancy difference in this part of the world, now favourable to women, is still relatively
small by world standards. The opposite situation exists in the countries that make up the
former Soviet Union, some of which (e.g. Belarus, Russia and Ukraine) have life expectancy
gaps of more than 10 years between and women.

120. Life expectancy (at birth) refers to the number of years that an average person can expect
to live, provided that at each age he or she is exposed to the current age specific mortality
rates for that age. For the world as a whole, it is currently about 71.6 years for women and
67.1 years for men. The Age-Specific Mortality Rate or age-specific death rate, in turn, is the
number of person of a given age that die during a year, divided by the average population of
that age during the year. The most significant age-specific mortality rate is the Infant
Mortality Rate, which differes from other age-specific mortality rates in that it uses the
number of births, rather than the average population under age 1, as its denominator. For the
world as a whole, its present value is 43.3 (per 1,000 births) for boys and 40.3 for girls. The
Crude Death Rate, finally, is the total number of deaths divided by the average population
during the year, regardless of age. It is basically a measure of how much populations
diminish as a result of mortality, but it is not a good measure of risk as it is greatly affected
by the age structure of the population. Finally, mortality can be analysed by cause. The most
common cause-specific mortality measures are analogous to the crude or age-specific death
rates, but limited to a specific cause. Because of the smaller numbers, they are generally
expressed as fractions of 100,000, rather than 1,000.

121. Maternal mortality refers to the phenomenon of deaths caused by pregnancy-related

factors. It is typically measured by three indicators:
The Maternal Mortality Ratio - the number of women who die while pregnant or within 42
days of termination of pregnancy, irrespective of the duration and site of the pregnancy, from
any cause related to or aggravated by the pregnancy or its management but not from
accidental or incidental causes, per 100,000 live births14. This is the indicator used with MDG
goal 5B.
The Maternal Mortality Rate – this indicator uses the same numerator, but measures it per
1,000 women of reproductive age in the population. One important implication of this
definition is that this indicator is sensitive to the level of fertility in the population, unlike the
MMRatio.

14
definition adapted from WHO website: http://www.who.int/healthinfo/statistics/indmaternalmortality/en/in-
dex.html

54
122. Less commonly used is a third indicator called Lifetime Risk of Maternal Mortality. This
indicator refers to the probability of maternal death, conditional on survival to age 15 years
(Wilmoth, 2009).

There is widespread confusion about the difference between the Maternal Mortality
Ratio and the Maternal Mortality Rate, and often these terms are used interchange-
ably.

In terms of statistical practice, a ratio is a division of things that have the same unit
of measurement: for instance, the sex ratio divides the number of men by the
number of women. A rate, on the other hand, measures the relationship of two
things that have different units of measurement: for instance, the Infant Mortality
Rate measures the relationship between number of infant deaths and numbers of
births. According to this practice, the Maternal Mortality Ratio is actually a rate.
Nevertheless, it is called a ratio to distinguish it from the existing concept of
Maternal Mortality Rate, as defined above.

123. Although the best and most detailed mortality information depends on civil registration
data, censuses can measure mortality in a number of ways. The most common census
questions are part of the same cluster that is used to measure fertility (see the previous
chapter). It consists in asking women between the ages of 15 (sometimes 12) and 50 about: 1)
Their number of Children Ever Born alive (CEB); 2) Children born during the past 12 months
before the census; and 3) Survival of children ever born and/or children born in the past 12
months. Combining the information from 1) and 3) - and more rarely 2) -, it is possible to
derive estimates of infant and child mortality by using indirect estimation techniques (United
Nations, 1983). Some censuses also ask for the survival of the last birth or the children born
during the past 12 months, in addition to the survival of all Children Ever Born. Other
common census questions regarding mortality include orphanhood questions, questions about
members of the household that died in the recent past and questions about the survival of the
sisters of adult household members, to measure maternal mortality.

2. Why is it important ?

124. Apart from aggregate differences between the mortality of men and women, there are
also major differences in the structure of mortality by cause. To do this requires death
registration data which are not available from censuses, but it may be worthwhile to show
where the major differences are found. Table 7, which is based on global WHO estimates for
2002, summarizes the causes of death in which female death rates (per 100,000 population,
not standardized by age) were at least 20 per cent higher than male death rates or, conversely,
in which male death rates exceeded female death rates by at least 20 per cent.

125. The most important causes in which women face a disadvantage are those that, by
definition, are exclusive (or almost exclusive) to women: maternal mortality (16.5 per
100,000 in 2002), breast cancer (15.3), cervical cancer (7.7), ovarian cancer (4.4) and uterine
cancer (2.3). Because women live longer, they are also more prone to develop diabetes (17.7,
compared to 14.1 per 100,000 in men), and Alzheimer’s disease and other dementias (8.1,
compared to 4.7 per 100,000 in men). Maybe less obviously, because women spend more
time at home, in constructions that are often unsafe, they are at greater risk to die in fires (6.2,
compared to 3.8 per 100,000 in men). Finally, women have higher mortality due to some

55
nutritional deficiencies, especially iron deficiency anemia, and also endocrine/nutritional
disorders, rheumatic heart disease, musculoskeletal diseases, and skin diseases.

126. Men, on the other hand, have considerably higher mortality rates due to most cancers,
especially lung cancer, stomach cancer, liver cancer and – of course – prostate cancer.
Together, these types of cancer are associated with a male death rate of 67.3, compared to a
female death rate of 28.1 per 100,000. In addition, men are much more likely to die of most
types of injury, both intentional and non-intentional. Road traffic accidents, falls, drowning
and poisoning jointly represent a male death rate of 50.9, compared to a female rate of 23.3
per 100,000. The male suicide rate in 2002 was estimated at 17.4 per 100,000, compared to
10.6 for women. Acts of war and violence resulted in 19.2 male, compared to 4.2 female
deaths per 100,000. Finally, male death rates are significantly higher with respect to
tuberculosis (32.9, compared to 17.3 per 100,000 in women), alcohol-induced conditions and
drug abuse, perinatal conditions, hepatitis B and C and some tropical diseases (not including
malaria).

Table 7: Estimated global male and female death rates (per 100,000) by cause of death
in 2002

Female disadvantage Male Female

– Diabetes mellitus 14.1 17.7

– Maternal conditions 0 16.5
J Nutritional deficiencies 6.9 8.7
J.2 Iron deficiency anemia 1.5 2.9
C.5 Breast cancer 0.1 15.3
H.1 Alzheimer's disease and other dementias 4.7 8.1
A.5 Rheumatic heart disease 4.4 6.1
E.5 Fires 3.8 6.2
– Endocrine/nutritional disorders 3.4 4.4
C.11 Cervical cancer 0 7.7
C.14 Ovarian cancer 0 4.4
– Musculoskeletal diseases 1.2 2.2
C.15 Uterine cancer 0 2.3
– Skin diseases 0.8 1.4

Male disadvantage

C Malignant neoplasms (cancers) 126.9 101.7

C.1 Lung cancers 28.4 11.4
C.2 Stomach cancer 16.7 10.5
C.4 Liver cancer 13.6 6.2
C.6 Esophageal cancer 9.1 5.2
C.8 Oral and oropharynx cancers 7.1 3.1
C.9 Prostate cancer 8.6 0
C.10 Leukemia 4.7 3.8
C.13 Bladder cancer 4.0 1.7
E Unintentional injuries 73.7 40.2
E.1 Road traffic accidents 27.8 10.4
E.2 Falls 7.5 5.0
E.3 Drowning 8.4 3.9
E.4 Poisoning 7.2 4.0
– Perinatal conditions 43.7 35.4

56
 F Digestive diseases 34.9 28.2
F.1 Cirrhosis of the liver 16.1 9.1
F.2 Peptic ulcer disease 5.0 3.5
G Intentional injuries (Suicide, Violence, War, etc.) 37 14.9
G.1 Suicide 17.4 10.6
G.2 Violence 14.2 3.7
G.3 War 5.0 0.5
B.4 Tuberculosis 32.9 17.3
B.9 Tropical diseases excluding malaria 2.5 1.6
B.9.1 Leishmaniasis 1.0 0.7
B.9.2 Trypanosomiasis 1.0 0.5
H.2 Epilepsy 2.2 1.8
B.10 Hepatitis B 2.3 1.0
H.4 Alcohol use disorders 2.5 0.4
H.5 Drug use disorders 2.2 0.5
B.11 Hepatitis C 1.1 0.6
I.2 Benign prostatic hyperplasia 1.0 0

Source: WHO (2004): Annex Table 2

127. Out of all of the differences that stand out from the previous listing, two have been of
special concern: violence and accidents, as a major cause of male over-mortality, and
maternal mortality as a cause of mortality that is specific to women. Worldwide, intentional
injuries make about 750,000 more male than female victims annually; the difference with
respect to unintentional injuries (accidents) is 1.2 million. Male disadvantage with respect to
violent deaths is particularly evident in the countries of the former Soviet Union and in much
of Latin America. Gavrilova et al. (2000) comment, for instance, on the overall rise of
mortality that took place in Russia between 1991 and 1994 as a result of the tumultuous
transition from a socialist to a market economy and the devastating effect that this had on
male mortality rates from violent causes. Female mortality from these causes also increased,
but to a lesser extent, thereby exacerbating a male-female difference which was already
among the largest in the world at the time. In particular, male suicide rates increased from
47.7 in 1991 to 76.9 per 100,000 in 1994, as the corresponding female rates increased from
11.2 to 13.6. What this suggests is that men were more psychologically affected by the
uncertainties surrounding the economic transition than women. Similarly, deaths due to
alcohol poisoning - always a problem in the former Soviet Union (see Simpura et al., 1998,
for an account on the Baltic states) - multiplied, from 19.4 to 61.2 per thousand, in the case of
men, and from 4.2 to 15.8 in the case of women, whereas male homicide rates increased from
25.1 to 52.8 per thousand, as female rates went up from 6.9 to 13.6.

128. Male over-mortality from violent causes, particularly homicides, has also been a major
issues in some Latin American countries, such as Brazil. In 2007, there were 45,554
registered homicides in Brazil, 92.1 per cent of which were of male victims, especially men
between the ages of 15 and 40 (Isfeld, 2010). In some more developed countries (Croatia,
Germany, Hungary, Japan, Republic of Korea, Slovenia, Switzerland), on the other hand, the
number of male and female victims is roughly equal. There is a moderately strong positive
relationship between the level of the overall homicide rate in a country and the percentage of
victims that are male. In those countries in which data exist, there is also evidence that the
majority (about 90 per cent globally) of perpetrators of homicides are males (UNODC, 2011:
Fig. 5.12). Homicides in which both the victim and the perpetrator are female are quite rare,
e.g. 2.6 per cent in the US (UNODC, 2011: 72). Whereas men are likelier to be killed in a

57
public place, female victims are murdered mainly at home, as is the case in Europe, where
half of all female victims were killed by a family member. The overwhelming majority of
victims of violence committed by partners and family members are women. In Europe, for
example, women accounted for almost 80 per cent of the total number of persons killed by a
current or former partner in 2008. There is a general sense in the literature that the gender
determinants of violent cause of death are under-studied and that they are too easily attributed
to the innate aggressiveness of males. However, advancing in this area based on census data
is difficult due to the fact that censuses provide no or only minimal cause-specific mortality
data. Two censuses that did attempt to obtain some level of cause-specific mortality data are
the 2008 census of Cambodia and the 2010 census of Zambia. The latter included the
following cause categories: a) Accident; b) Injury; c) Suicide; d) Spousal violence; e) Other
violence; f) Sickness/disease; g) Witchcraft; and h) Other.

129. Studying maternal mortality based on census data, while not ideal, is more viable.
Globally, an estimated 287,000 maternal deaths occurred in 2010
(WHO/UNICEF/UNFPA/World Bank, 2012). Although maternal mortality is only the 20th
most common cause of death for women of all ages worldwide, it is the most important cause
of death for women of reproductive age (usually taken as the age range 15-49) in many
developing countries. In addition, like violent causes of death, it is eminently amenable to
prevention. Maternal mortality by itself is not considered a gender indicator. That does not
mean that it has no linkages with gender, but rather that it is an outcome to which gender
factors contribute15. One publication on UNFPA’s website states: “Preventable maternal
mortality occurs where there is a failure to give effect to the rights of women to health,
equality and non-discrimination. Preventable maternal mortality also often represents a
violation of a woman’s right to life” (Hunt and Bueno de Mesquita, xxxx). Yet, there is little
empirical evidence on the extent to which gender factors contribute to maternal mortality. A
detailed discussion on this subject is beyond the scope of this manual, but one set of results
may serve to illustrate the nature of the relationships.

Table 8: Strength and significance of trends calculated using polynomial

regression analysis for all variables in the study

Number of
Variables observations R-squared value P
Infant mortality rate 1 77.0 0.002
Total fertility rate 14 65.4 0.424
Female literacy rate 1 48.6 0.039
Combined enrolment ratio 1 48.0 0.004
Year of suffrage 1 18.1 0.574
Seats in parliament held by women 1 2.0 0.750
Female professional and technical 6 16.5 0.111
Ratio of estimated female to male 6 7.6 0.075
Female economic activity 1 4.6 0.634
Human development index 1 81.2 0.000
Gender-related development index 1 82.9 0.000
Gender empowerment measure 6 25.0 0.187
(Figure 3) 2
Source: McAlister and Baskett (2006)
15
It is quite easy to show a strong correlation between maternal mortality and selected gender indicators but this
does not imply a causal relationship, as both are correlated with the overall level of development.

58
130. The element to note in the above table is the relatively poor performance of “pure”
gender indicators as predictors of maternal mortality, as compared to indicators that reflect
overall level of development. Of particular note is the finding that the Human Development
Index scores very high, with an R2 of 81.2, and that this improves to 82.9 with the Gender-
related Development Index. Thus, gender is shown to be a dimension of maternal mortality,
but not the principal one.

131. A different perspective, but resulting in similar conclusions, is provided through the
”Three Delays” model. This model proposes that pregnancy-related mortality is
overwhelmingly due to delays in:
1) Deciding to seek appropriate medical help for an obstetric emergency;
2) Reaching an appropriate obstetric facility; and
3) Receiving adequate care when a facility is reached.

Out of these three delays, it is mainly the first one where gender plays an important role, the
other two are more dominated by factors of general development (transport issues) and
development of the health care system (quality and availability of obstetric care).

3. Data issues

132. As was indicated in the first section, censuses can measure mortality in a variety of
ways. The first and most common is through questions to women of reproductive age about
1) Their number of Children Ever Born alive (CEB);
2) Children born during the past 12 months before the census; and
3) Survival of Children Ever Born alive.

133. As was indicated in the previous chapter, most countries ask for this information
disaggregated by sex of the child, but there are still a few countries where this information is
not available. In countries that disaggregate the basic fertility and mortality data by sex,
important information can be obtained about the sex ratio at birth and on differential
mortality between young girls and boys. This issue, although directly related to fertility, will
be discussed in the next sub-chapter. Typically, the information from questions 1) and 2),
disaggregated by the age of the mother, is combined to estimate fertility, whereas 1) and 3)
(more rarely 2) are combined for the purpose of mortality estimation. In addition, some
censuses ask about the survival of the last child born or children born in the past 12 months.

134. A limitation of this method is that it can only provide information for mortality levels up
to age 15 or 20. That means that mortality levels at higher ages (including the life
expectancy) have to be estimated based on extrapolations, using typical relations between the
mortality under age 20 and at higher ages. Such extrapolations contain a good deal of
uncertainty and consequently the life expectancy estimates for many developing countries
(including the sex differential) need to be treated with caution.

135. Some censuses have additional questions that serve primarily to complement the
information on early mortality by adult mortality estimates. One such question is the
orphanhood question, which asks members of the household whether their mother, father or
both are still alive. Based on the age of the respondent and typical fertility patterns in the
country, this allows the estimation of probabilities of death for the parents. A limitation of
this method is that the estimates obtained in this manner refer to deaths that occurred at any

59
time during the birth of the respondent and the present. Especially in the case of older
respondents, these estimates can be quite distinct from current mortality levels. There is also
the possibility that parents live in unspecified areas different from the current residence of the
respondent, thereby making it difficult to use the information for sub-national mortality
estimates. This limitation also applies to the infant and child mortality estimates of the
previous paragraph, but the potential bias is more serious in the case of adult mortality. For
all of these reasons, the questions on orphanhood are generally not considered very effective
and only about 25 countries currently include them in their censuses.

136. Rather than asking about the parents, another option is to ask about the survival of
sisters of adult members of the household. There are two variants if this method. In the direct
sisterhood method, which is the standard method used in the DHS, the detection of deaths of
sisters is followed up by more detailed questions about the year in which the date occurred
and the age of the sister at the time. This method, however, is too laborious for most censuses
which use the indirect sisterhood method, in which only the age of the respondent is used and
the remaining information is attributed based on averages. This makes the indirect variant
much less efficient than the direct variant. Although the sisterhood method can be used to
estimate adult (female) mortality in general, its more typical use is the estimation of maternal
mortality, in which it has to be combined with follow-up questions about the likely cause of
death. However, as will be explained below, its use in censuses for this purpose is generally
not recommended.

137. The other major type of question that can be used to measure adult mortality is the one
that asks about the age and sex of members of the household that died during the past 12
months or another appropriate reference period. The most common problem with this
question is that it tends to systematically under or (more rarely) over-estimate mortality due
to factors such as the following:
 Confusion about the reference period (e.g. current calendar year, rather than past 12
months);
 Confusion about the meaning of “household”, as opposed to “family” or
“community”; or
 Confusion about the meaning of “belonging to this household”, especially in the case
of prolonged hospitalization prior to death.
However, to the extent that these errors affect all age groups more or less equally, the results
can still be used to determine a mortality pattern. In addition, there are methods (see Hill et
al., 2011) to estimate correction factors, based on the observed population sizes by age and
sex, to correct for the systematic errors in estimated mortality levels. By asking appropriate
follow-up questions (see below), this question can also be used to measure maternal
mortality. In the 2010 census round, this method for measuring maternal mortality has been
followed in more than 30 countries that do not have reliable registration data.

138. Measurement of maternal mortality through a population census is recommended for

countries where other sources of maternal mortality information such as the vital registration
system are deficient. In practice this recommendation only applies to countries with at least
500,000 population because of the need to have sufficiently large denominators to reliably
measure this event. In this context it is important to realize that maternal deaths are relatively
rare events and in order to measure them through a sample survey the sample size needs to be
very large, often resulting in prohibitive costs.

60
139. Questions on maternal mortality in a census typically result in information on pregnancy
related deaths, which is not the same as maternal deaths (see also the definition in an earlier
paragraph). Pregnancy related deaths include deaths from any cause, occurring while a
woman was pregnant or within 42 days after delivery. Using this data for analysis of maternal
mortality results in a measure called the Pregnancy-Related Mortality Rate (PRMR).
Comparisons of census-based estimates of the PRMR with survey-based estimates of MMR
found that approximately 85% of pregnancy related deaths are maternal deaths. It is believed
that the correspondence between PRMR and MMR is quite close since the number of
pregnancy related deaths tends to be under-reported in censuses (Hill, 2009; NIPORT; ORC
Macro; Johns Hopkins and ICDDR.B, 2001). Nevertheless, the results from census-based
maternal mortality questions should not be taken at face value and should ideally be followed
up by a survey among the reported pregnancy related deaths to empirically establish the
proportion of pregnancy related deaths that are maternal.

Recommended census questions to estimate maternal deaths:

Q1: Have any residents of this household died during the last 12 months?

For each deceased:

Q2: Sex of the deceased;

Q3: Age of the deceased;
Q4: Date of death;

For female deceased between the ages of 15 and 49:

Q5: Was the deceased pregnant at the time of death or did the death occur within 42
days after delivery

140. The recommended questions to measure maternal mortality in a census are placed in the
household module, and extend the “standard” questions on deaths (by age and sex) in the
household over the past 12 months by one additional question: whether the woman was
pregnant at the time of death, or the death occurred within 42 days after delivery.

141. Some countries (e.g. Lesotho, Malawi, Swaziland) do not use this format, but instead
ask about the survival of the sisters of the respondent. A variant (direct sisterhood) of this so-
called sisterhood method is also used in the DHS, but the difference is that the DHS asks for
additional information on ages and times of occurrence, making the resulting information
much more accurate. The census data, however, is used for indirect estimation of maternal
mortality using the indirect sisterhood method (see above). This method results in estimates
of maternal mortality that refer to approximately 10-15 years before the date of the census. Its
validity is contested by WHO and others as it relies on too many assumptions and the
reference period is too long in the past.

142. The measurement of maternal mortality through census data requires a number of
specialized techniques that are beyond the scope of this manual. For more details, see Hill et
al. (2011).

61
143. Smoking is included in the census of New Zealand, Sint Maarten and some countries in
the Pacific, such as Cook Islands, Kiribati, Niue, Tokelau, Tonga and Vanuatu.

144. Of the health and mortality-related Minimum Set of Gender Indicators approved by the
Statistical Commission in February of 2012, the following can be computed from census
data:
1. Under-5 mortality rate by sex;
2. Maternal mortality ratio (in censuses that ask the appropriate question);
3. Life expectancy at age 60, by sex; and
4. Adult mortality by age group (but not by cause).
The following indicators, which are related to health, rather than mortality, cannot usually be
computed from census data:
1. Smoking prevalence among persons aged 15 and over, by sex;
2. Women's share of the population aged 15-49 living with HIV/AIDS; and
3. Proportion of adults who are obese, by sex.
………..

Bourne and Walker (1991) show for the case of India that, while increased education of
mothers generally favours child survival, the effect is larger for girls than for boys.

62
Chapter 5:
Sex Ratio at Birth and throughout the Life Cycle

1. What is it?

151. The sex ratio at birth (SRB) is defined as the ratio of male to female births in a
population,16 multiplied by 100. The census variables to be analysed to calculate de SRB are
the date of birth and sex of the last live-born child. On this basis, one can compute the
number of girls and boys born over the last 12 months. 17 Although this ratio can vary
somewhat due to biological factors, its natural value is normally not smaller than 104 and not
larger than 106,18 although some would still accept 102 and 107 as normal. The recent inter-
agency publication on preventing gender-based sex selection (OHCHR; UNFPA; UNICEF;
UN Women and WHO, 2011) puts the normal range at 102-106, to take account of the very
low sex ratios found in some African countries. Where observed sex ratios at birth are greater
than 106 (or 107) in census data, this departure from the biological norm may be due either to
under-reporting of female births, or to prenatal sex selection practices and feticide linked to
son preference, or to a combination of the two. The opposite applies if sex ratios at birth are
lower than 104 (or 102). If the difference is real, rather than due to differential under-
enumeration, it is called sex ratio imbalance (at birth). Sex ratio imbalances arise in several
countries that have strong sex preferences – usually for boys – and where prenatal sex
screening and selective abortions have now led to significant distortions due to the
elimination of female fetuses. This is commonly referred to as the problem of the missing
girls (Das Gupta, 2005).

152. The current global average in the sex ratio at birth is about 107 per 100 (United Nations,
2011 b; UNICEF, 2011 b) but regional differences are significant. In Sub-Saharan Africa, sex
16
Note that in India the sex ratio is computed the other way around, as the number of girls over the number of
boys.
17
Where “date of birth of last live-born child born” is not disaggregated by sex, one needs to look at the age and
sex of the youngest child in the household and – if under 1 year old – verify if its age/birthday is compatible
with the declared date of last birth.
18
Some anomalies in sex ratios at birth can be explained in biological terms. For example, a study by the Arctic
Monitoring Assessment Program in 2007 found abnormally low sex ratios, in the order of 50, in some arctic
communities in Russia, Greenland and Canada, which it attributed to high levels of endocrine disruptors in the
blood of inhabitants, particularly PCBs and DDT. Other studies (e.g. Rocheleau et al., 2011), however, have
contested the effect of PCBs on human sex ratios at birth. There is also some discussion among geneticists as to
whether sex ratios vary naturally according to race, maternal and paternal age and birth order (e.g. Erickson,
1976; Imaizumi and Murata, 1979; Ruder, 1985; Chahnazarian, 1988). Historical data from Europe suggest
considerable heterogeneity between families, with boys predominating in some and girls in others, in
proportions that differ from what one would expect if the process were purely random (Garenne, 2008 b). In the
case of Africa, Garenne (2008 a) found that sex ratios declined with maternal age and birth order. Due to the
fact that he used DHS data, no information on paternal age was available. He concluded that these findings are
consistent with James’s (1989, 1996) theories about the biological factors of the sex ratio, in particular, the
effect of concentrations of sex hormones (e.g. progesterone, gonadotropin, estrogen, testosterone). Higher levels
of gonadotropin and progesterone were found to be associated with more female births (lower sex ratios).
Conversely, higher concentrations of male hormones (e.g. testosterone) seem to favour high sex ratios. The
African data do not seem to suggest any deliberate sex selection. Oster (2005) has argued, based on existing
medical literature and analysis of cross country data and vaccination programmes, that parents who are carriers
of hepatitis B have a higher offspring sex ratio (more boys) than non-carrier parents. Since China and some
other countries have high hepatitis B carrier rates, she suggested that hepatitis B could explain up to 50 per cent
of Asia’s “missing women”. However, Lin and Luoh (2008), using data from a large cohort of births in Taiwan,
found only a very small effect of maternal hepatitis carrier status on offspring sex ratio, a conclusion which was
later endorsed by Oster as well (Oster et al., 2008).

63
ratios at birth are generally low (as low as 102 in Kenya, Malawi, Mozambique and South
Africa), with a regional average of 104. However, some countries in Africa, notably Nigeria
and Ethiopia, stand out for having rather high sex ratios, in the order of 107 (Garenne, 2002,
2008 a). In Latin America and the Caribbean, Northern America and Oceania the average is
105, although low sex ratios, in the order of 103 are commonly found in the Caribbean. In
Europe it is 106, while in South Asia it reaches 107 (and in Eastern Asia and the Pacific 113).

153. China is the country with the highest sex ratio at birth (118.7), estimated for the period
2005-10, with values of over 130 in some provinces. Tibet, however, has a very low sex ratio,
of 102. In India, the sex ratio at birth was computed at 904 girls for every 1,000 boys (or, in
more conventional terms, 111.6 boys per 100 girls) in 2006-08 (UNFPA India, 2010). Based
on this information, UNFPA estimated the number of missing girls in India for the 2001-07
time period to be in excess of 6 million. The more commonly used child sex ratio was 976
girls per 1,000 boys (102.5 per 100) in 1961, but had changed to 927 (107.9 per 100, in
conventional terms) in the 2001 census. According to the preliminary results of the 2011
census, there are now about 83.9 million boys under age 7, compared to roughly 75.8 million
girls, implying a child sex ratio of 109.4 per 100, a deterioration with respect to the 2001
census (Jha et al., 2011).

154. Apart from India and China, where the imbalance has been known for a long time (in
China, the sex ratio was more skewed in 1953 than in 2000), in recent years there have also
been increases in the sex ratio in countries like the Republic of Korea (1980s), Albania,
Armenia, Azerbaijan, Georgia (1990s), and most recently Viet Nam. So far the Republic of
Korea is the only country where the sex ratio has returned to normal in the 1990s, after a
period of imbalance (Guilmoto, 2009; Villa, 2006). Chung and Das Gupta (2007) argue that
the trend is due to fundamental changes regarding social norms, but the most recent data for
the country indicate another rise, to 109 in 2010.

Table 9: Sex Ratio – at birth, age-specific, and overall – by region

Regions At birth 0-4 years 5-14 years 15-24 years

Sub-Saharan Africa 104 103 102 101
Middle East and North Africa 105 105 105 105
South Asia 107 108 108 108
South Asia excluding India 105 105 105 104
East Asia and Pacific 113 114 114 109
East Asia and the Pacific excluding 105 105 105 104
China
Latin America and Caribbean 105 104 104 102
CEE/CIS 106 106 105 103
Developing countries 107 107 108 106
World 107 107 107 106
Source: United Nations (2011). World Population Prospects. The 2010 Revision, CD-ROM,
cited in UNICEF (2011 b)

155. The child sex ratio is the ratio of boys compared to girls in the 0-6 year age group. It is
often used as a proxy for the sex ratio at birth because it is easier to compute from census
data, particularly if the census does not disaggregate births by sex. Defining a natural value
for this ratio is more difficult than in the case of the sex ratio at birth because it is affected not
only by the latter, but also by differential infant and child mortality. One would expect the

64
number of boys over girls to decline gradually after birth, as a consequence of higher male
mortality. Depending on the life expectancy at birth, the number of males and females should
equalize among young or older adults, but then it should decline among the elderly as
consequence of lower survival rates among old men.

156. The age-specific sex ratio is the sex ratio that characterizes specific age groups,
including the 0-6 year age group of the child sex ratio. The overall sex ratio refers to all men
in the population, divided by the number of women and multiplied by 100. This is the most
difficult to interpret because it can be influenced by a wide range of phenomena, including
migration.

157. Alternatively, the terms primary, secondary and tertiary sex ratios are sometimes used.
The primary sex ratio is the ratio at the time of conception. The secondary sex ratio at time
of birth and the tertiary sex ratio is the ratio among mature organisms.

2. Why is it important?

158. Looking at the sex ratio at different stages of the life cycle reveals how it is shaped by
the different underlying demographic processes. The sex ratio in the total population depends
on the sex ratio at birth, migration patterns and the conditions of mortality throughout the
life-cycle. For instance, many countries in the Arab States have a significant foreign labour
force, mostly composed of men. As a consequence, their sex ratio is very high, Qatar having
the highest (303, estimated for 2010). In Estonia, Latvia and Lithuania, the male/female ratio
is normal at birth, but declines sharply among 30-40-year olds. Such an imbalance skewed
towards women may be symptomatic of armed conflict or, as is the case in the Baltics,
alcoholism in males. As women live longer than men in most places, a sex ratio of roughly 90
for the age group 60 and over is not uncommon. As a consequence of large differences in
male and female life expectancies (more than 10 years) and differential migration, countries
like Belarus, the Russian Federation and Ukraine also registered very low over-all sex ratios
(below 90) in their 2010 censuses (UNECE, 2012 c).

159. Several South and East Asian countries, such as China, India and South Korea, have
long struggled with the phenomenon of the missing women (Kynch and Sen, 1983; Sen, 1990;
Yi et al., 1993). Using sex ratios in countries particularly affected by gender inequalities, Sen
(1990) calculated that 107 million more women would be alive in the absence of excess fe-
male mortality concerning infant girls in particular. While the figure has been challenged and
recalculated various times (mostly due to problems of undercounting, e.g. of unmarried
women), it is clear that the number of missing women is substantial. Based on typical infant
and child mortality patterns in the country, Jha et al. estimate that 4.53 million sex-selective
abortions took place in the country between 2000 and 2010. In some states, the sex imbalance
is much worse (e.g. 120.5 in Haryana, 118.2 in Punjab, 116.5 in Jammu and Kashmir, and
115.4 in the national capital of New Delhi). These are some of the richest states of the coun-
try, which shows that sex imbalances are not a consequence of poverty and ignorance, but –
to the contrary – show up in places where people have access to the technology that makes it
possible to implement sex selection choices.

160. A recent study from Nepal aptly identifies the core of the problem as “a deeply rooted
preference for sons, which leads parents across cultures and geographic locations to decide
against allowing a girl to live, even before her birth, and the increasing availability of tech-
nology that enables them to do so with ease” (UNFPA, 2007: 2). The most pronounced and

65
well-known case is China (UNICEF 2011 b), where the “one child policy” in place since
1979 has discouraged fertility. Most observers agree that the one child policy exacerbates the
problem, as it is consistently associated with higher ratios of male to female births (Zeng et
al., 1993; Ebenstein, 2010), but it is not the only cause (for the classical model of causation,
see Gu and Roy, 1995), More broadly, early detection of the sex of a foetus since the 1980s in
many countries has led to increased numbers of sex-selective abortions.

161. The sex ratio in infancy is important to monitor because female infanticide and neglect
of new-born girls through differential access to food, vaccination and care has resulted in
higher mortality for girls. In this way, the sex ratio can inform policy by defining a problem
and may also be used to monitor progress toward a solution (e.g. addressing sex imbalances
in vaccination campaigns, free curative care for infants, etc.).

162. A statement recently issued by OHCHR, UNFPA, UNICEF, UN Women and the WHO
(WHO, 2011) reviews the evidence behind the causes, consequences and lessons learned re-
garding “son preference,” and concludes: “Sex selection in favour of boys is a symptom of
pervasive social, cultural, political and economic injustices against women”. On the other
hand, it points out that it is also women who have to bear the consequences of giving birth to
an unwanted girl child. These consequences can include violence, abandonment, divorce or
even death. Son preference can, however, be embedded in wider societal norms and practices.
For instance, in societies where the maintenance of elderly parents falls primarily on the sons,
as married daughters come to belong to the family of their husbands, it is understandable that
parents will want to guarantee a male heir, for the sake of their own sustenance in old age.
This becomes particularly critical in low fertility contexts. The pressure on women to produce
sons also puts women in a position where they in turn perpetuate the lower status of girls
through son preference. As a consequence, various forms of gender-based violence and dis-
crimination and heightened vulnerability of women and girls are associated with sex ratio at
birth imbalances skewed towards boys (UNFPA, 2007, 2010b).

163. Where men of marriageable age are faced with a dramatic shortage of potential brides,
human trafficking, crime and other adverse social consequences are on the rise (Guilmoto,
2007). As the country example highlights just below, men also are likely to suffer social con-
sequences, such as depression, isolation and poverty, as illustrated by the example below. Part
of the shortage of brides on the marriage market, which arises from the lack of young women
at marriageable ages, may also lead to changes in the age difference at marriage between
women and men. In the same way that polygamy often operates in a system where men are
considerably older than women at the time of marriage, imbalances in the sex ratio - because
of son preference - may lead to older men marrying younger women. Son preference also af-
fects the stability of unions. Zeng et al. (2002) show for instance that divorce rates among
women with three or more daughters without a son are 2.2 times higher than among women
with three or more children with at least one son. CAREFUL NOT TO EXAGGERATE THE
ATTENTION DEDICATED TO SON PREFERENCE.

Country Example 2: Generating Vulnerable Elderly Men in China

In 2000 the total number of excess boys and young men up to 20 years of age in China was
almost 21 million. The lack of young women has negatively affected the formation of
families. Poston and Glover (2005) estimate that more than 23 million young men born
between 1980 and 2001 will not be able to find brides in China. If the overall growth of the
population young population were positive, this imbalance might be solved by men marrying

66
younger women, but this is not sustainable in a context of diminishing numbers of young
people. As Judith Banister (2004) points out, sex ratio at birth imbalance was almost
eliminated during the Mao years. That means birth cohorts born up until 1982 were normal,
and women would come of marriageable age 23 years later on average (men 2-3 years later) –
in 2005 onwards. So there has been no squeeze until recently – the ones marrying now were
born in the late 1980s. Das Gupta (2010) argues that the abnormally high sex ratios that have
characterized China since the 1980s will ultimately lead to a situation in which older men,
who did not marry when they were younger, will have no children to support them, so that
during the later years of their lives they will be particularly vulnerable to poverty and social
isolation. Poston and Glover (2005) foresee the formation of “bachelor ghettos” in Beijing,
Shanghai, Guangzhou, Tianjin and other big cities in China, where commercial sex outlets
will be prevalent. They also speculate about the possibility of criminality, as men who do not
marry have a higher probability of turning to crime.

3. Data issues

KIM: AGE HEAPING, BUT IS IT SEX-SPECIFIC ?

164. While analysis of the sex ratio at birth is most reliable when based on data from efficient
vital statistics, the census can provide an estimate if the number of children born during the
last 12 months is differentiated by sex, which is usually the case. Both vital statistics and
census data have the advantage of universal coverage which is important because establishing
the difference between a sex ratio of - say - 108 and a normal ratio of 106 requires at least
33,000 cases in order to obtain statistically significant results, which is too much for most
surveys. Even in the census of a small country such as Vanuatu, the reported high sex ratio at
birth of 111 boys per 100 girls (Vanuatu, 2011) may be based on too few births (about 5,000
per year) to allow solid conclusions.

165. In those countries where the census does not differentiate the number of children born
during the past 12 months by sex (Argentina, Azerbaijan, Bahamas, Costa Rica, Kazakhstan,
Mexico, Palau, Peru, the Seychelles and Thailand), an alternative is to use the child sex ratio
(0-4 or 0-6 years) or the ratio of children under age 1 as proxies. The main methodological
problems associated with this solution are:

a) Children under age 1 are typically poorly enumerated in censuses and under-
enumeration or the misdeclaration of ages are sometimes more pronounced for one sex
than for the other; and

b) By defining “at birth” sex ratios as sex ratios in children less than 1 year old, sex-
selection effects are conflated with sex-specific mortality during the first year of life, in
particular with infanticide and higher perinatal mortality among boys.

The latter will obviously grow more serious as wider age groups are used. In India, for
example, the 2001 census found a sex ratio of 107.1 for the 0-4 age group, while the sex ratio
at birth for 2000-02 was estimated to be 112.1 boys per 100 girls (Kulkarni, 2007). One can
also estimate the sex ratio at birth by taking the sex of the youngest child in the household
and verifying if its age is compatible with the declared date of birth. This analysis, however,
is more complex and there may be difficulties in case where the child has died or no longer
lives with the mother. If there are different mothers in the household, there may also be

67
problems in identifying who a particular child belongs to.

166. Census reporting on children aged 0-6 is typically more reliable, but this age group is
already more exposed to other causes of sex imbalance, such as differential mortality. For
example, it was mentioned earlier that the child sex ratio in India increased from 107.9 to
109.4 per 100 between the 2001 and 2011 censuses. But data from the Indian Sample
Registration System suggest that the sex ratio at birth actually improved somewhat, from
112.1 per 100 in 2000-02 to 111.6 per 100 in 2006-08. If these data are reliable and can be
generalized to the country as a whole (which may not be the case, as they are based on a
sample), the conclusion would have to be that infant and child mortality, rather than sex
selection at birth, was the cause for the deterioration of the child sex ratio between 2001 and
2011. This could be verified using census data collected in 2011 on children born during the
past 12 months by sex, however, this information has not yet been published.

167. Differential under-enumeration is another complicating factor. In many societies, such as

Iran and Libya, under-enumeration of girls and women is known to be widespread in
censuses. This may be due to the respondent (e.g. ‘forgetting’ about girls who do not have a
birth certificate) or enumerator (e.g. taking greater care and time to write down all boys and
men in the household). A recent study of China (Goodkind, 2011) indicates that about 19 per
cent of children at ages 0-4 went unreported in the 2000 census. From this, the author
concludes that, since under-enumeration is much more pronounced for girls than boys, the
reported child sex ratio at birth imbalance is likely to be inflated.

168. Differential under-enumeration is not limited to children under age 1 or aged 0-6. The
declared number of births during the past 12 months, even when disaggregated by sex, may
also contain differential under-enumeration errors. This is illustrated by the case of Malawi,
where the 2008 census counted 268,876 female births, but only 247,753 male births,
implying a sex ratio at birth as low as 92.1. The corresponding numbers of children under age
1 were 255,576 and 247,809, respectively. The latter implies a sex ratio of 97.0, which is
actually more balanced. Because there is no plausible reason to assume that such deviant sex
ratios could be caused by the differential abortion of male foetuses, the only acceptable
explanation is differential under-enumeration of male infants and particularly male births.
What could be causing this phenomenon is a puzzling question that should be reason for
concern for the census authorities. The 2007 census of Ethiopia enumerated 897,827 boys
and 877,627 girls under age 1, implying a sex ratio of 102.3, which is more in line with
expected results than the data from Malawi, but still rather low, especially considering the
results obtained by Garenne (2002, 2008a), cited earlier, according to which the sex ratio at
birth in Ethiopia was found to be relatively high.

4. Tabulations

HERE AND ELSEWHERE, WE NEED TO SEPARATE MORE BY ETHNIC GROUP

169. The Principles and Recommendations for Population and Housing Censuses Rev. 2
(United Nations, 2008 a) recommends to tabulate population by single years of age and sex.
From this tabulation, one can calculate sex ratios for various age groups and geographical
areas. One should bear in mind, however, what was noted in the previous section, about the
possibility of age-specific differences in under-enumeration or age misdeclaration by sex,
particularly under age 1.

68
Figure 5: Hong Kong, 2006 Census – Population Pyramid and Age-Specific Sex-Ratios
5-year age-specific pyramid Graph of age-specific sex-ratios

Men Women

170. This recommended tabulation can also be represented by 5-year age groups in the form
of a population pyramid. However, if the objective is specifically to represent sex ratios, the
population pyramid may be hard to interpret and it may be better to depict the age-specific
ratios directly. As examples, just above are a conventional age pyramid by 5-year age groups
with men on the left and women on the right for the 2006 census of Hong Kong and on the
right, the age-specific sex ratios graphed with the same data.

171. Preferably, sex ratios at birth need to be separated by birth order as the perceived need
for sex-selection increases with birth order and the sex composition of the other siblings. This
procedure can be further improved by separating the previous births by sex. For example,
families with one girl and one boy rarely apply sex selection to the third birth, but if both of
the previous children are girls, the sex ratio among the third births tends to be very high. If
the census differentiates the surviving children and children born during the past 12 months
by sex, the procedure is straightforward; otherwise it may be necessary to obtain the sex
composition of the children by analysing the children with the appropriate relationship to the
head of household.

Country Example 3: Sex Ratios at First, Second and Third Births in Viet Nam and
China

The 2009 census of Viet Nam asked all women of reproductive age how many sons and
daughters they had given birth to during their last delivery. Combined with the information
on when this last birth took place, this allowed the computation of the number of girls and
boys born during the last 12 months. This yielded a sex ratio at birth of 110.6, well above the
expected range of 104-106. However, the analysis went further by combining this
information with the data on “children ever born.” This provides a classification of the births
by sex and birth order, which is an important covariate because the perceived need for sex
selection tends to increase with birth order.

69
Couples without sons among their first two children tend to be highly motivated to have a
third child and to want to make sure that it is a boy. In the case of Viet Nam, the sex ratio for
first births found in the census was 110.2, second births 109.0, but among third births it
increased to 115.5. Another interesting finding of the Vietnamese study is that sex selection is
almost non-existent among the poor, while among the medium and higher strata, which have
access to the necessary technology, sex ratios increase to 112 or 113 (UNFPA, 2010 c). This
finding also underscores how income or a proxy for income, such as educational attainment,
is important to consider when interpreting findings.

The same type of analysis was done by Lin and Zhao (2010) on the Chinese censuses of
1982, 1990, 2000 and 2005. They show that the SRBs for first-born children during that
period was actually rather low, varying between a minimum of 100.1 in 2000 and a
maximum of 103.6 in 1982. For higher order births, however, the SRB has increased both by
birth order and over time. In 2005, it was 132.7 for second-order, 152.2 for third-order and
170.6 for fourth-order births, compared to 107.9, 112.9 and 115.1, respectively, in 1982.

172. Another useful tabulation for sex ratio analysis, is the total number of births by sex
during the past 12 months, disaggregated by the sex composition of the preceding surviving
children (e.g. M, F, MM, MF, FF, MMM, MMF, MFF, FFF). This disaggregation may allow
the differentiation between deliberate sex selection and other causes. 19 If the children born
during the past 12 months are not specified by sex, an alternative is to use children by sex
under age 1 in the household instead of children born during the past 12 months, and the total
number and sex composition of the other children with the correct relationship to the head of
household that live there. Jha et al. (2011) show, for example, that there is a clear difference
in sex ratio trends for second births in India, depending on the sex of the first-born child, with
no change in cases where the first-born was a boy and a clearly deteriorating sex ratio over
time in cases where the first-born was a girl.

5. Indicators

173. The transition to very low fertility may exacerbate the manifestations of son preference,
such as sex selection at birth. The clearest example of this is the Chinese one-child policy, but
the tendency for sex ratios at birth to become more unequal as fertility goes down is a more
general phenomenon that has also been noted in India (Das Gupta and Mari Bhat, 1995).

174. Some researchers have argued, therefore, that the sex ratio at birth as such is not an
adequate indicator of son preference because it has to be interpreted in the context of the
Total Fertility Rate (Singh and Singh, 2007). A country with higher fertility and a more equal
sex ratio at birth may actually have higher son preference than another country, with lower
fertility and a more unequal sex ratio at birth. The authors also note that the distribution of
fertility may affect the results, with higher disparities in fertility levels leading to more
unequal sex ratios at birth, given a certain level of son preference. However, formulating a
better indicator of son preference, that is not affected by these intervening factors, is no trivial
task, both because of the variety of factors involved (e.g. fertility preferences in terms of total
numbers, their distribution in the population, cost of sex selection procedures, perceived cost
19
When sex ratios began to rise in Armenia in the 1990s, for example, at first the tendency was to attribute this
to the aftermath of conflict in the region. It was not until further analysis established that the imbalance was lim -
ited to second and third birth orders that the sex selection process was recognized for what it was.

70
of childlessness, of having no surviving son, of having no surviving daughter) and because
the relationships between them are relatively complex.

175. Probably the best strategy to control the intervening effects is the one suggested in the
previous section, namely to control the sex ratio at birth for the number and composition of
existing children. Even then, sex ratios across different countries or regions may not be
strictly comparable due to the fact that they reflect both the strength of son preference and the
cost of early sex detection and abortion.

176. With respect to the sex ratios for the 0-4, 0-6 age groups and others, the general
recommendation is to compute these figures in relation to the “normal” values that one would
expect based on a standard or model life table for the population with the life expectancy of
the country. This procedure is discussed in the next section. The same goes for the estimation
of the number of “missing women.”

6. Multivariate and further gender analyses

177. To illustrate how age-specific sex ratios can be analysed, Table 9, from the 2001 census
of India, may serve as an example. The simplest kind of analysis that one can carry out is to
compare the sex ratio by age in India to that of other countries where presumably female
over-mortality and sex selection are not a problem. Thus, the table below compares the Indian
sex ratios with those of Mexico (2005) which is an example of a country where male sex
selection and female over-mortality should not be problems. Clearly the Mexican sex ratios
are substantially lower in all age groups. A similar kind of comparison can be made with the
Coale and Demeny Model Life Tables, in this case Model West (Coale and Demeny, 1966).
These are theoretical life tables constructed out of the historical experience of a number of
western countries, such as the UK, USA, Australia, New Zealand, etc. For the purposes of the
table, the sex ratio at birth has been set to 105 boys per 100 girls. The expected sex ratios for
particular age groups vary somewhat, depending on the (female) life expectancy, but on the
whole they are fairly stable. India probably conforms best to the middle one (e 0f) = 55),
which is representative of mortality as it was during the early 1980s. Again, the expected sex
ratios are substantially lower than those of India, with the exceptions of the 25-29 and 30-34
age groups.

Table 10: Sex ratios compared between India, Mexico and the Coale & Demeny West
Model

Age India (2001) Idem Model Coale and Demeny West Cross-Multiplied
Ratio Mexico
Group Males Females e0f)=40 e0f)=55 e0f)=70 India Mexico C&D
(2005) e0=55
0-4 57,119,612 53,327,552 107.1 103.3 101.4 103.5 103.9
5-9 66,734,833 61,581,957 108.4 103.2 101.3 102.8 103.6 0.988 1.001 1.007
10-14 65,632,877 59,213,981 110.8 102.6 101.5 102.8 103.5 0.978 1.006 1.000
15-19 53,939,991 46,275,899 116.6 97.7 101.8 102.8 103.3 0.951 1.050 1.000
20-24 46,321,150 43,442,982 106.6 90.3 101.8 102.7 103.0 1.093 1.082 1.001
25-29 41,557,546 41,864,847 99.3 88.6 101.7 102.5 102.7 1.074 1.020 1.002
30-34 37,361,916 36,912,128 101.2 89.4 101.5 102.4 102.5 0.981 0.990 1.001
35-39 36,038,727 34,535,358 104.4 90.1 101.0 102.1 102.3 0.970 0.993 1.003
40-44 29,878,715 25,859,582 115.5 91.3 99.7 101.4 101.9 0.903 0.987 1.007
45-49 24,867,886 22,541,090 110.3 90.9 97.5 100.2 101.3 1.047 1.004 1.012

71
50-54 19,851,608 16,735,951 118.6 92.0 94.6 98.3 100.2 0.930 0.988 1.019
55-59 13,583,022 14,070,325 96.5 92.5 91.2 95.8 98.3 1.229 0.994 1.026
60-64 13,586,347 13,930,432 97.5 90.2 87.5 92.5 95.3 0.990 1.026 1.036
65-69 9,472,103 10,334,852 91.7 89.1 83.3 88.4 91.2 1.064 1.013 1.046
70-74 7,527,688 7,180,956 104.8 88.6 78.5 83.5 86.3 0.874 1.005 1.059
75-79 3,263,209 3,288,016 99.2 88.0 73.2 77.9 80.6 1.056 1.007 1.072
80+ 3,918,980 4,119,738 95.1 78.2 66.1 68.7 69.1

178. One indicator that can be constructed out of the table above is the number of “missing
women” implied by the sex-specific census data and the theoretical numbers of men and
women expected based on the survival ratios derived from a model life table. Comparing the
expected number of women with the number actually enumerated, there is a difference of
31,606,111 (computed as 57,119,612/1.035 + 66,734,833/1.028 + 65,632,877/1.028 + ...... +
3,918,980/0.687 - 53,327,552 - 61,581,957 - .... - 4,119,738) women, representing 6.0 per
cent of the total number expected. Note, however, that the results of analyses of this kind can
be somewhat distorted by migration if the sex ratio of the migrants is highly unbalanced. This
may significantly affect the results in some parts of India, especially in the South, where
many women migrate to the Gulf States as nannies and maids.

Table 11: Comparing actual and expected counts to estimate “missing women”

India (2001) Expected

Age Expected
Group Ratio Women
Males Females e0f)=55
0-4 57,119,612 53,327,552 103.5 55,188,031
5-9 66,734,833 61,581,957 102.8 64,917,153
10-14 65,632,877 59,213,981 102.8 63,845,211
15-19 53,939,991 46,275,899 102.8 52,470,808
20-24 46,321,150 43,442,982 102.7 45,103,359
25-29 41,557,546 41,864,847 102.5 40,543,947
30-34 37,361,916 36,912,128 102.4 36,486,246
35-39 36,038,727 34,535,358 102.1 35,297,480
40-44 29,878,715 25,859,582 101.4 29,466,188
45-49 24,867,886 22,541,090 100.2 24,818,250
50-54 19,851,608 16,735,951 98.3 20,194,922
55-59 13,583,022 14,070,325 95.8 14,178,520
60-64 13,586,347 13,930,432 92.5 14,687,943
65-69 9,472,103 10,334,852 88.4 10,715,049
70-74 7,527,688 7,180,956 83.5 9,015,195
75-79 3,263,209 3,288,016 77.9 4,188,972
80+ 3,918,980 4,119,738 68.7 5,704,483

179. Although the comparisons described in the previous paragraphs indicate substantial
excess mortality of women compared to expected patterns, the age specific numbers, as well
as the number of missing women, are otherwise somewhat difficult to interpret because
differentials in the higher age groups are attributable not only to the current situation in each
age group, but to the entire life history of each age cohort since birth. This makes it difficult
to separate the current situation from historical differences dating much further back in time.
For example, the high sex ratio in the 40-44 age group in India may be due to mortality

72
differentials dating back 45 years. A technique to pinpoint more clearly where the female
over-mortality is concentrated consists in the cross-multiplication demonstrated in the last
three columns. These have been computed by applying the formula:

Index(5-9) = PF(5-9) * PM(0-4) / (PF(0-4) * PM(5-9)) =

= 61,581,957 * 57,119,612 / (53,327,552 * 66,734,833) = 0.988

180. The advantage of this method is that it provides a clearer picture of differential mortality
in specific age groups in the recent past, because it filters out accumulated historical
differences between age groups. It also corrects for differential migration and under-
enumeration of women, as long as this pattern does not vary too much by age. If the index is
larger than 1, it indicates that male mortality in the age group in the recent past has been
higher than female mortality. If it is smaller than 1, the mortality differential is favourable to
men. Normally one would expect the index to be larger than 1, especially in the earliest ages
and after age 50. The Mexican indices are all quite close to 1, mostly slightly larger, but in
some age groups slightly lower. The Indian ratios, on the other hand, display a rather erratic
oscillating behaviour, with rather low values in the 40-44, 50-54, 60-64 and 70-74 age groups
and much higher values in the 45-49, 55-59, 65-69 and 75-79 age groups. This is likely to be
due to errors in the age declaration. It also suggests that it may be somewhat misleading to
accept the earlier sex ratios by age at face value, even if the conclusion that their values are
suspiciously high is not likely to change. In order to obtain more realistic results, it may be
necessary to apply some smoothing. On the whole, however, the indices for India are not
markedly lower than those for Mexico: the average for India is 1.0086, versus 1.0110 for
Mexico. This is consistent with the observation that female mortality in India, at least at
higher ages, is now lower than male mortality, even though the difference in life expectancies
(i.e. 61.3 for women and 59.7 for men in 1995-2000) is still relatively small in comparison to
other countries at similar levels of development. In addition, female mortality in India
continues to be higher from the second to the 60th month of life (Jha et al., 2011).

181. Because of the nature of the indicator, sex ratios tend to be analysed at the macro-level
(i.e. in terms of their variations between geographic or socio-economic groups). This,
however, need not necessarily be so. Like is often the case, the analysis tends to be more
revealing as it becomes more disaggregated. It would be possible to formulate logistic
regression models at the level of individual children, in which the probability of being male
(or female) is formulated as a function of characteristics such as birth order or – better – the
number and composition of elder siblings by sex (i.e. a categorical variable, using the
categories outlined in Section 4), education of the head and/or wealth index of the household,
rural/urban residence, education of the child’s mother and whether she works outside the
home.

182. One way to look at sex ratios as a gender discrimination indicator is through the use of
the sex ratio at last birth. The logic is simple. In a regime, where the total number of children
a woman gives birth to is not governed by nature alone but also by some type of fertility
control, couples will have the tendency to stop having additional children after a child is born
of the preferred sex. Through fertility control people try to maximize their preference for the
number and sex composition of their offspring. In societies with a strong son preference, it
can therefore be expected that the sex ratio at last birth will be high. The measure obviously
works best for women who have passed their reproductive period in life, as their ‘last birth’
really marks the end of their reproductive career. However, also for women below age 50 the

73
sex ratio at last birth can be used as an indicator of sex preference. In this case ‘last birth’ will
be a mix of concluded and non-concluded fertility. If the sex ratio at last birth is higher than
the overall sex ratio at birth, it can be considered a sign that son preference is present.

183. The sex ratio at last birth as a measure of son preference is examined here on the basis of
the 2009 Vanuatu Population and Housing Census. The sex ratio of all ‘Children Ever Born’
among all women aged 15 and over, stood at 1.088 in Vanuatu; for women 15 – 50 years of
age this was 1.084. However, the sex ratio at last birth stands at 115.1 for all women 15 years
of age and older. Among women who are past their reproductive age (50 years and over) the
sex ratio at last birth is 120.2, whereas for women 15-49 years it is 113.6. These figures are
higher than the overall sex ratio at birth and proof that in Vanuatu son preference is clearly a
motive to stop or to continue having additional children.

Table 12: Vanuatu (2009) - Sex ratios of last child by distribution of number of older
siblings for women aged 15 -50 years

Number of girls ever born by the mother, before birth of last child
        0 1 2 3 4
Number of boys   0 116.0 128.6 154.5 165.9 139.6
ever born by the   1 96.2 117.0 134.6 127.6 111.6
mother, before   2 89.0 101.8 100.5 123.4 133.1
birth of last child   3 99.3 106.8 102.3 99.5 159.7
      4 82.9 85.2 127.0 91.6 130.0

Number of girls ever born by the mother, before birth of last child
        0 1 2 3 4
Number of boys   0 86.2 77.7 64.7 60.3 71.6
ever born by the   1 103.9 85.5 74.3 78.3 89.6
mother, before   2 112.3 98.3 99.5 81.0 75.1
birth of last child   3 100.7 93.7 97.8 100.5 62.6
      4 120.6 117.4 78.8 109.2 76.9

Source: Census of Vanuatu (2009)

184. More detail can be brought into the picture by linking the sex ratio at last birth to the
number of older male and female siblings of this last child. These figures are restricted to 4
older brothers and 4 older sisters, because very few cases were available beyond these
numbers. Only children of women between 15 and 50 were considered, to exclude events that
took place too far in the past. The results of this analysis are presented in Table 12. For the
sake of comparability, the number of boys per 100 girls is presented in the upper half of the
table and the reverse (girls per 100 boys) in the lower half.

185. The table clearly shows that the sex distribution of previous children has an important
effect on the sex ratio of the last birth. If the last birth was in fact the first birth (0 older
sisters, 0 older brothers) the sex ratio is 116. As this figure is higher than the overall sex ratio
at birth of 1.084, it indicates that some parents are more eager to stop at one child if that child
is a boy. The sex ratio at last birth for women who have had 2 daughters, but no sons is as
high as 154.5. Many if the women who had 3 children will have continued having another
child. But a far larger proportion of women who had two girls and got a boy as a third child

74
decided to stop than those who gave birth to a girl after two girls. It is interesting to see that
the sex ratio at last birth indicates that women also have the tendency to stop having more
children when they get a baby girl after having only sons before. The figures in the first
column of the lower half of the table show sex ratios which are considerably higher than .92
(i.e. 1/1.084). This means that for some women there is also a desire for a baby girl after the
mother had only sons. However, looking at the sex ratio at last birth for women who had 2
sons and no daughters, one notes that this value (112.3) is considerably lower than the value
for the corresponding category for boys (154.5). Note that analyses of this kind and those of
the following paragraphs make use of the concept of Parity Progression Ratio, i.e. the
proportion of women who will go on to have additional children, given that they already have
a certain number or (in this case) a certain composition of children. This concept is discussed
in the Indicator section of Chapter 3.

186. To look whether these patterns of sex preference are different among various groups in
society, one may set up a logistic regression model with the sex of the last child as the
dependent variable. Earlier it was shown that the number and composition of elder siblings
by gender plays an important role in the sex ratios at the last birth. To bring this variable into
the equation, the number of older male siblings of the last birth was subtracted from the
number of older female sibling. Obviously, this number could be negative, zero or positive. A
number of other predictors were tested in the logistic model (education of mother,
urban/rural, work status of the mother, religion, and ethnicity). In the end only the urban/rural
variable was retained, as all other predictors proved inconclusive. This result is in itself
important because it shows that the pattern of sex preference, described above, exists among
the various subgroups of society. Table 13 shows the results of the logistic regression. Next
to the main effects the interaction between urban/rural and the difference between older
brothers and sisters was included.

Table 13: Vanuatu (2009) - Logistic regression of the probability that the last-born child
is a boy

    B Exp(B)

0.00
Urban/rural Urban 0 .
  Rural -0.042 0.958

Older sisters - Older brothers 0.137 1.147

Interaction (Urban/rural) * (Older sisters - Older brothers) -0.057 0.945

Constant 0.165 1.180

       

Source: Census of Vanuatu (2009)

187. The results show that for each unit difference between ‘older sisters minus older
brothers’ the odds ratio for the last child being a boy increases by a factor 1.147. This
confirms our earlier conclusion that son preference is operating in Vanuatu. Controling for
this sibling difference, the odds ratio for rural areas over urban areas is 0.958. This means
that in rural areas chances of the last birth being a male is lower than in urban areas. Also, the
odds ratios of the interaction term (0.945) indicate that with each extra unit difference
between older brothers and sisters, the gap between the chances of the last birth being a boy
for urban and rural increases. The fact that in rural areas the odds for the last child being a
boy are smaller than in urban areas is not necessarily a proof that son preference in urban
75
areas is higher than in rural areas. It is well possible that women in rural areas follow a more
natural fertility regime than women in urban areas and that regardless of the sex of their latest
child, they will simply go on to have another child. For more details on this methodology,
one may consult the article by Dalla Zuanna and Leone (2001).

188. Mutharayappa et al. (1997) looked at Indian couples who had two or three living
children and analysed their subsequent fertility based on the sex composition of the existing
children. To this effect, they controlled for rural/urban residence, literacy, religion and other
socioeconomic variables. After applying these controls, they looked at the fertility decisions
of couples, depending on whether the existing children were boys or girls. Their conclusion
was that fertility in India would be 8 per cent lower if son preference did not play a role in the
decision to have another child. Although their data came from the National Family Health
Survey, it would be possible to carry out a similar analysis with census data, provided that the
information on Children Ever Born and Children Surviving is disaggregated by the sex of the
children and that it is possible to identify the last-born child, which is the case in India. The
main limitation of using census data for this type of analysis is that it may not be feasible to
obtain the dates of birth of all the surviving children. This has to be done based on the listing
of household members, but this listing may only provide ages, rather than exact dates of birth,
and some children may not live in the same household as their mothers.

189. In addition to demonstrating the effect on the last-born, several studies have attempted to
quantify the effect of son-preference on fertility decisions in other ways. Research by Tu
(1991) in Shaanxi Province, for example, showed that the chance of having a second birth for
a woman whose first child was a girl was 1.5 times that of a woman whose first child was a
boy, and the chance of having a third birth for a woman whose first two children were girls
was 2.9 times that of a woman whose first two children were a boy and a girl. The chance of
having a third birth for a woman whose first two children were boys was slightly higher than
that of a woman whose first two children were a boy and a girl (indicating a slight preference
to have at least one girl, rather than just boys), but the difference was not statistically
significant. In Taiwan, Chu, Xie and Yu (2007) showed that there is a positive relationship
between the proportion of girls in the household and the total number of children. This
suggests that parents continue to have children until they have at least one boy. Similar
studies have been carried out in other countries, such as the Republic of Korea (Park, 1983),
based on World Fertility Survey data.

190. By representing sex ratios spatially, especially for relatively small units, certain patterns
may emerge that may correlate with certain determinants. The following figure shows the sex
ratios for the 0-9 year age group by canton for the 2000 census of China. There is a clear
tendency for sex ratios to be highest in the South and East of the country, with more normal
patterns in the western half of the country, as well as in the north. One also notes a number of
contiguous areas in Anhui, Shanxi and western Hubei Provinces where the ratios are closer to
normal. In their article, the authors of the map correlate such variations with the availability
of rural pension systems.

Figure 6: Sex ratio for age group 0-9 years, by county, in China's 2000 census

76
Source: Ebenstein and Leung, 2010: Figure 2

191. Further analysis may also be carried out with respect to the determinants of imbalances
in the sex ratio at birth. One line of analysis that was suggested earlier is the disaggregation
of births by birth order and by the composition of older siblings. It may be possible to
investigate other determinants. For example, it was mentioned earlier that parents in some
countries need a male heir, to ensure their sustenance in old age. This suggests that parents
who have access to institutional pension systems may have less unbalanced sex ratios among
their offspring than those who depend entirely on their (male) children (for an example of this
kind of analysis see the above study by Ebenstein and Leung, 2010). In some censuses, it
may be possible to differentiate between these situations, either through specific census
questions or indirectly, by looking at the status in employment of the head of household and
his/her spouse.

192. Some of the first regression analyses (e.g. on India: Kishor, 1993 and Murthi et al.,
1995) show that sex ratio imbalances are a function of female economic valuation (using
female labour force participation as a proxy), development level (income/wealth or human
development), male and female educational attainment, cultural factors (using religion and
ethnicity as measures) and urbanization. In accordance with the argument made in Section 4,
one should add to this the effect of overall fertility levels and their distribution in the

77
population. Recent research suggests that greater attention should be paid to comparing sex
ratios at different age groups, to sex-biased migration as an explanation for rural-urban
differences in sex ratios, and to the existing sex composition of the family into which girls are
conceived (Das Gupta, 2005).

7. Interpretation, Policy and Advocacy

193. The examination of sex ratios at birth and for different age groups should be
contextualized by more qualitative analyses. In order to develop an adequate policy or
advocacy response to sex ratio at birth imbalances, gender analysis needs to unearth what
gender inequality or human rights violation is underlying the disparity.

a) Is it differential under-reporting ?
b) Is it sex-selective abortion, based on son preference ? And if so, what is the legal
context of sex-selective abortion in the country ?
c) Is it neglect of newborn girls ?

Although differential under-reporting would the least serious of the three alternatives, it is not
without negative consequences. Under the one child policy, parents in China may be more
likely not to report the birth of a girl than a boy, so as to maintain the option to have another
child, but by doing so they make it impossible for the girl to attend school or have access to a
series of other public benefits that require an official identity. Although under-registration is
not the same as under-count and girls that were never registered may still be counted in the
census, it is likely that parents will keep never registerd girls away from census enumerators
as well, especially if the enumeration involves the presentation of identity documents for all
household members.

194. Any effective strategy for dealing with son preference should be based not only on the
subjective preferences of parents - and how to change them -, but also consider the fact that
parents take rational decisions based on the objective disadvantages that their daughters - and
by extension they themselves - face in a society where women are less valued and where the
ability of women to care for their parents is limited both by economic realities and social
customs. Ebenstein and Leung (2010: 66) express this viewpoint when they consider how
male or female offspring affect the access of parents to care in old age: “The Chinese
government has both re-affirmed the one-child limit and declared that reducing the sex ratio
at birth by 2016 is a national priority (....). Such goals may be in conflict with each other if
economic conditions making sons valuable to parents are not addressed. We find that parents
who fail to produce a son are more likely to participate in old-age pension programmes and
that the number of children in a family is negatively related to pension programme
participation. We also find evidence that the rural old-age pension programme mitigated the
increase in the sex ratio in the areas where the programme was available.”

195. In some countries, the sex ratio seems to start decreasing after the age of 20, only to
equalize around the age of 60. This reflects the high level of maternal mortality. According to
UNICEF (2011 b), based on the analysis of DHS, MICS and Reproductive Health Surveys
for 80 countries, under-5 mortality for girls is typically 4 per cent lower than for boys, except
in East Asia and the Pacific and in South Asia, where it is 5 per cent and 3 per cent higher,
respectively. In Latin America and the Caribbean and in the Central and Eastern
Europe/Commonwealth of Independent States (CEE/CIS) countries, on the other hand, the
mortality of girls is much lower (14 per cent and 22 per cent, respectively) than that of boys

78
under age 5. It is important to distinguish sex differentials in mortality, especially infant and
child mortality, from sex ratio imbalances at birth because their policy implications are very
different. Oster (2009) argues that differential mortality, rather than sex ratio imbalances at
birth, are responsible for the high child sex ratios in India. In practice, however, it is difficult
to disentangle these factors.

Country Example 4: the 2011 Census of India

In India the 2001 census revealed a substantial increase (or decrease, as it would be reported
according to the Indian convention for computing sex ratios) in the child sex ratio of the 0-6
age group, compared to the previous census. This finding was publicized by the media and a
major campaign (‘Save the Girl Child’) to control and monitor female foeticide was
launched, along with a number of remedial measures at national and state levels (UNECE,
2010).

In the 2011 census of India, UNFPA concentrated its support to the government in the area of
gender (see UNFPA India, 2011). Based on the results of the 2001 census, three indicators
were identified to characterize districts with particular gender problems. These were:
a) The overall sex ratio (with a ratio of less than 900 women per 1000 men indicative of a
problem);
b) Low female literacy (30 per cent or lower); and
c) Low female labour force participation (20 per cent or lower).

Likewise, analysis with a different cut-off was done for cities/towns. Based on the results of
the 2001 census, this led to the identification of 260 gender-critical districts (including
cities/towns) out of the 593 districts across the country, for focused attention. These districts
were singled out for additional training of the enumerators, through a special gender module.

More in general, interviewer training focused on seven critical gender elements of census
enumeration:

1. Full coverage of population, to ensure the inclusion of females (elderly, infants, disabled,
etc.);
2. Proper netting of female headed households;
3. Appropriate netting of female work in all economic activities, including informal and
unpaid;
4. Adequate capture of the date of birth, particularly among elders, girls, and illiterates;
5. Adequate capture of mother tongues, especially of married females and non-family
members;
6. Adequate capture of fertility, particularly children born and died in the year before the
census;
7. Instructions to probe the reasons for migration, especially in the case of females.

In order to prevent the misuse of technology, India has institutes the Pre-conception and Pre-
natal Diagnostic Techniques (Prohibition of Sex Selection) Act, which was adopted in 1994
and amended in 2003, but few convictions have been made so far, due to the difficulty of
demonstrating conclusively that the offense was conducted with the consent of the parents
and the service provider (UNFPA India, 2009).

79
196. Some authors consider the term “sex ratio” (to say nothing of “masculinity ratio”, as it is
called in some Romance languages) too slanted towards biology and thus unclear about the
role of cultural differences. While some Australian feminists propose replacing the term “sex
ratio” by “gender ratio” (Lucas, 1985: 7), this usage is not encouraged, for the reasons
discussed in previous chapters regarding the difference between “sex” and “gender.”

197. Sex selection technology providers generally argue that sex selection is an expression of
reproductive rights pursued by women, as well as a sign of female empowerment that
allowed couples to make well-informed family planning decisions, prevented occurrences of
unintended pregnancy and abortion and minimized intimate partner violence and/or child
neglect. In contrast, primary care physicians question whether women could truly express
free choice under pressure from family and community. They voice concerns that sex
selection led to invasive medical intervention in the absence of therapeutic indications,
contributed to gender stereotypes that could result in child neglect of lesser-desired sex, and
was not a solution to domestic violence (Puri and Nachtigall, 2010).

198. Advocacy efforts to reduce sex ratio imbalances should lobby with legislators, the
executive, traditional and religious leaders for enhanced monitoring of technologies that
allow for sex-selective abortions and their application and spread in the private health sector.
More importantly, however, the long-term solution for the problem lies in counterbalancing
the effect of women’s undervaluation in patriarchal systems. This requires various
empowerment measures, tackling the societal level (questioning and reforming systems of
dowry transfers, patrilocal residence and extended patrilineal families, old-age support, ritual
duties, inheritance though sons, etc.) and, where feasible and affordable, the individual level
(support girls and/or all-girls-families through direct subsidies at the time of birth, through
scholarship programmes, and through gender-based quotas or financial incentives aimed at
improving their economic situation and at offsetting the impact of the economic
undervaluation of girls in society).

199. Most important from the viewpoint of this manual, the need for knowledge needs to be
addressed and knowledge needs to be shared. In Viet Nam, a country that has fairly recently
become aware of increasingly skewed sex ratio at birth as a consequence of son-preference
and induced abortions, the following advocacy recommendations were made in this regard
(UNFPA Viet Nam, forthcoming: 7ff):

“To enhance the basis for policy development and dialogue on the forces behind the
increasingly skewed SRB in Viet Nam, there is a need for data of both a quantitative and
a qualitative nature, and for dissemination and public discussion of this evidence.”

The regular analysis of population and birth registration data on sex ratios should be
continued in order to establish and extend the evidence on sex ratios and monitor relevant
trends over time. Further, analyses should be carried using other data sources, such as the
annual Population Change Survey, the Inter-census survey and the 2019 Population and
Housing Census.

80
Chapter 6:
Marital Status, Polygamy, Widowhood and Child Marriage
1. What is it?

200. Marital status is “the personal status of each individual in relation to the marriage laws
or customs of the country” (United Nations, 2008: 159). As they reflect culture, marital status
categories are not universal across censuses and categories in a country may change over
time. However, census data generally allow distinguishing between at least five categories: a)
single, never married; b) married; c) widowed and not remarried; d) divorced and not
remarried; e) married but separated. In some countries, it is necessary to take into account
consensual unions, or socially recognized stable unions that may not have full legal force. In
others, polygamous unions exist that are not always legally recognized or acknowledged by
the census. Some countries join categories d) and e) and thus distinguish only 4 marital status
categories. On the other hand, the Bahamas, in its 2010 census, distinguishes as many as 9: a)
Not in a union; b) Legally married; c) Common-law-union; d) Visiting partner; e) Married
but not in a union; f) Legally separated and not in a union; g) Widowed and not in a union; h)
Divorced and not in a union; and i) Not stated.

201. Marital status is a key variable for gender analysis. It reveals situations of vulnerability
such as polygamous unions, widowhood and child marriage. Wherever possible, marital
status should be considered in tandem with household composition in order to capture some
of the complexities of household composition.

202. For gender analysis, the category of “married,” in particular, needs to be clear.
Monogamous unions should be distinguished from polygamous unions, and unions
recognised by law (i.e. generally marriage) should be distinguished from consensual unions,
which are recognised by tradition. Examples of consensual unions are customary or
“common-law” marriages in much of Africa, visiting unions in the Caribbean and elsewhere,
and cohabitation in de facto unions in Europe. Also, several countries (e.g. Brazil, Croatia,
Germany) allow marriage or registered partnerships for homosexual couples that may be
counted with census data with the categories of marriage, consensual union, or a separate
category.

Text Box 9: Types of Socially Recognised Stable Unions Captured in Population and
Housing Censuses

 Marriages: Unions recognized by civil and/or religious authority:

 Monogamous
 Polygamous
 Same sex marriages

 Consensual Unions: Unions recognized by custom or “common-law”, consensual

unions or “companionate marriages,” free unions, temporary unions, visiting unions or
cohabitations:
 Monogamous
 Polygamous
 Same sex unions

In recognition of the distinction between these categories, some censuses, particularly in the

81
Caribbean countries, distinguish between the “marital status” and the “union status” of
individuals.

Source: Elaborated on the basis of Census Questionnaires of the 2010 Census Round and the
UNDESA Multilingual Demographic Dictionary, 2nd edition 1982.

203. Polygamy is a marriage which includes more than two partners. Its most frequent form is
polygyny where a husband has two or more wives. In a census, the term ‘polygamy’ is
generally used in a de facto sense (i.e. regardless of whether the relationships between the
spouses are recognized by the state). Thus, several African censuses distinguish between
“married – monogamous marriage” or “married – polygamous marriage” and ask for the
number of wives or co-wives (e.g. Benin, Burkina-Faso, Burundi, Côte d’Ivoire, Egypt,
Gambia, Kenya, Lesotho, Niger, South Africa, Togo, Uganda, but not Saudi Arabia). More
generally, there are complexities in gathering information on marital status where marriage is
a process, rather than a single event, as is often the case in Southern Africa. The Data Section
discusses some of these complexities of gathering valid polygamy data in greater detail.

204. Widowhood refers to the marital status of a person whose spouse has died and who has
not remarried. Widowhood affects women disproportionately; in every region of the world, at
least one-third of women age 60 and over are widowed. In Africa, women over age 60 are
over six times more likely to be widowed than men of the same age (United Nations, 2009 a).

205. Child marriage is defined as marriage before age 18 (regardless of national legislation
on the minimum age at marriage, if different), for both girls and boys, which is the minimum
legal age of marriage according to international human rights conventions, notably the
Convention on the Elimination of All Forms of Discrimination Against Women (CEDAW)
(United Nations, 1979), the Convention on the Rights of the Child (CRC) (United Nations,
1989). In practice, because women marry at younger ages than men, child marriage concerns
mainly girls.

2. Why is it important?

DISTINCTION BETWEEN SINGLE AND NEVER MARRIED

ARRANGED MARRIAGES BEFORE BIRTH
WOMEN THAT NEVER MARRY, E.G. BECAUSE THEY HAVE TO CARE FOR
DEPENDENT PARENTS

206. Marital status greatly affects the socio-economic status of women and men and shapes
their experiences in society. One way in which this happens is through the legal or customary
practices that determine the property rights of married women. In many African countries, for
instance, married women can only inherit property from their husbands, even if accumulated
by common labour, through their children. Women in polygamous unions are most likely to
see their economic rights violated due to unequal property rights. In Uganda, it was found
that women considered themselves co-owners of property acquired during marriage because
of the economic importance of their agricultural labour, but men viewed the property as theirs
to use to marry a second wife (Khadiagala, 2002).

207. In addition, women’s marital status may influence their participation in the labour force
and their educational enrolment, with consequences for their risk of dependency and poverty.
For instance, child marriages have a greater impact on the educational career of women than

82
of men, often leading to school drop-out, early pregnancy and the impossibility of ever
entering the labour market. All other things being equal, married women may also be less
likely to work and acquire economic independence. Consequently, divorced or widowed
women are more vulnerable and exposed to the risk of poverty if they did not work while
they were married and have no other personal source of income. When analysing this
relationship, however, it is important to consider other factors, besides marital status, that
may account for the woman's economic activity or inactivity. It may be, for instance, that for
women of a certain educational level, it is the number of dependent children, rather than
marital status per se, that accounts for their labour force participation.

208. Polygamy has several negative impacts on co-wives, e.g. adverse economic and
inheritance consequences, such as loss of property and land ownership, health consequences,
such as an increased risk of contracting HIV/STIs, and psycho-social consequences mediated
by culture, such as disempowerment, low sense of self-worth, and personal betrayal. Even
where culturally or economically accepted by women, it may be experienced as undesirable
and burdensome. For example, in a survey conducted in Cameroon (Cheka, 1996) 66 per cent
of women in polygamous unions did not want their husbands to take another wife. They gave
many reasons for this, most often linked to jealousy and the central role occupied by children.
While other types of relationships (e.g. extra-marital relationships in Western countries) may
have similar impacts, polygamy adds a human rights dimension by legally and socially
sanctioning unequal treatment of men and women. Polygamy also has negative consequences
on families and children, including child poverty and lower educational attainment. In
Swaziland, for instance, polygamy is an important factor in family disruption as conflicts
frequently arise among husband and wives or among co-wives. It also contributes to school
drop-out (Poulsen, 2006).

209. Child marriage has negative physical and mental health consequences for the married
child and her children, and is often closely connected with forms of sexual exploitation and
social isolation. According to Jenson and Thornton (2003), women who married young are
more likely to be beaten or threatened, and more likely to believe that a husband might be
justified in beating his wife. Child brides suffer health risks associated with early sexual
activity and childbearing, leading to high rates of maternal and child mortality as well as
sexually transmitted infections, including HIV, as older spouses may have had multiple
sexual partners before the girl bride.

210. In many societies, widows are socially disadvantaged. Not only do they face several
forms of social, economic, psychological and cultural deprivation, they also lack attention
from policy makers and public interest. For a number of reasons, more females than males
are affected by widowhood. On the one hand, in most countries men suffer higher mortality
(e.g. from chronic life-style diseases or accidents, and violence). In addition, women tend to
marry men slightly older than themselves and not to remarry once widowed. The reasons why
men are more likely to remarry vary from culture to culture, but often include beliefs about a
person’s attractiveness (e.g. linked to virginity), practical considerations (e.g. the need for a
partner that can run a household) and the skewed sex ratio among older adults in many
countries (Carr and Bodnar-Deren, 2009). See the next section, for an actual example.

211. While in developed countries widowhood is experienced primarily by elderly women, in

developing countries it also affects younger women, many of whom are still rearing children.
For example, in the 2004 census of Timor Leste, 19.2 per cent of women were found to be
widows by the time they reached age 50, whereas this was only the case of 3.4 per cent of the

83
women in the Hong Kong census of 2011. Most of this is due to higher mortality, although in
some countries difficulties of remarriage for widows may also play a role. Discrimination in
inheritance, including land or property grabbing, loss of social status, stigma and exclusion
are a few of the human rights violations associated with widowhood in many societies.
Interestingly, in Western countries, research suggests that men are more emotionally affected
by widowhood than women, as they are more dependent on their spouse for social and
emotional support (Lee et al., 1998). (MAY BE AN EXCESSIVE GENERALIZATION)

212. Widowhood and polygamy are interlinked because in societies where widowhood or
singlehood is socially discredited, polygamy may be perceived as a practical alternative. For
instance, Surtees (2003) found evidence of an increased prevalence of polygamy in
Cambodian society, a practice which was not traditionally widespread. In the context of adult
sex ratios skewed by civil war (1970-1975) and ensuing turbulence, many women were
forced to choose between being a second wife or remaining unmarried.

Country Example 5: Widowhood Practices in Nigeria

In Nigeria, family law permits certain widowhood practices that discriminate against women,
particularly those married according to customary rather than statutory law. Some of the
negative practices derive from the belief that “the beauty of a woman is her husband.” At his
death, she is seen as unclean and impure, and her health may be undermined by the customs
she must observe in the weeks after her husband’s death. If she has no male adult children,
she may be ejected from her husband’s house as both it and his land will have been inherited
by his oldest brother. In most cases, the husband’s kin do not provide the widow with any
economic support, particularly if she will not accept the status of being an additional wife to
one of her husband’s brothers. In a study in Imo State, Nigeria, interviews and discussions
were held with traditional rulers, leaders of women’s organizations and widows. Five factors
that have an impact on the health and economic status of widows were identified: a long
period of incarceration during mourning; an obligatory poor standard of hygiene; deprivation
of the husband’s property and maltreatment by his relatives; the enforcement of persistent
wailing; and the practice of demanding that a widow sit in the same room with her husband’s
body until burial.

World Health Organization (1998). Women, Ageing and Health.

213. Child marriage is a harmful traditional practice and one of the most pervasive human
rights abuses worldwide. In some countries in Southern Asia and Sub-Saharan Africa, half of
the girls are married before they turn 18. It violates article 16 of the Universal Declaration of
Human Rights (UDHR) (UN, 1948) which stipulates that “marriage shall be entered into only
with the free and full consent of the intending spouses” as boys and girls lack the maturity to
make an informed and free decision and, worse, are sometimes married off by their families
without being consulted. The UN Convention on Consent to Marriage, Minimum Age for
Marriage and Registration of Marriages (1962) reiterates the right to free and full consent at
marriage and holds duty-bearers accountable for specifying a minimum age for marriage and
making sure that all marriages are officially registered. CEDAW goes further by stating that
“[t]he betrothal and the marriage of a child shall have no legal effect” (Article 16.2). In
addition, child marriage is linked to several rights explicitly stated within the Convention on
the Rights of the Child (CRC) (United Nations, 1989), specifically to express their views
freely, to be protected from all forms of abuse and from harmful traditional practices.

84
214. Globally, one-third of women currently aged 20–24 were married or in union before they
turned 18 (UNICEF Child Info, http://www.childinfo.org/marriage_countrydata.php;
accessed 5 May 2011). While the proportion of married girls aged 15 years or less is low, in
some countries it ranges from 1 to 5 per cent – such as El Salvador, Ghana, Malaysia, Nepal,
Nicaragua, Uganda and Zambia – while in Niger the share of very young girls that are
married is above 20 per cent. Married adolescents are generally typified by
 Large spousal age gaps;
 Limited social support, due to social isolation;
 Limited educational attainment and no schooling options;
 Intense pressure to become pregnant;
 Increased risk of maternal and infant mortality;
 Increased vulnerability to HIV and other STIs;
 Restricted social mobility/freedom of movement;
 Little access to modern media (TV, radio, newspapers);
 Lack of skills to be viable to the labour market. (SOURCE ?)

3. Data issues

215. Marital status data gathered from censuses may not capture the complexity of human
experiences and processes that make up the living and union characteristics of women and
men. In some of these cases, persons may fit more than one category at the same time. For
instance, in some European countries couples may live together, but for legal or fiscal reasons
maintain separate addresses. It is not always clear whether to classify persons as “single” or
“consensual union” if indeed these categories exist. On the other hand, in many emigration
countries, couples may not be legally separated or divorced, but de facto live separate lives as
they work in different countries. It is unclear whether to record them as “married” or
“married but separated.” In some African countries (e.g. Angola), polygamy is illegal, but
men often take informal additional wives. It is not clear whether the informal wife’s marital
status should be tallied as “single” or “married.” Finally, child marriages may not be
formalized due to legal reasons, yet the arrangements are binding between families. Should
informal child brides then be recorded as “married” or “single” ?

IT’S FINE TO RAISE ALL OF THESE ISSUES/QUESTIONS, BUT CAN WE PROVIDE

SOME ANSWERS ?

216. Few countries ask questions about the previous marital status of individuals. Some, like
Mauritius and Nepal in their 2011 censuses, ask whether the person has been married more
than once. One country that includes more detailed information is Ireland. The 2006 census
of Ireland distinguishes seven marital status categories, namely a) Single (never married); b)
Married (first marriage); c) Remarried following widowhood; d) Remarried following
divorce/annulment; e) Separated (including deserted); f) Divorced; and g) Widowed. This
allows some interesting analyses, such as quantifying the propensity of widowed or divorced
men and women to remarry. According to the Irish data, 9.04 per cent of women over the age
of 15 had been widowed and of those only 2.89 per cent had remarried. In the case of men
over age 15, only 2.78 per cent had been widowed, but of those a much higher percentage
(11.29 per cent) had remarried. To some extent, these results are affected by the age structure,
but even if this is taken into account, men are still more likely to remarry. Of the widows
aged 40-49, for example, 12.4 per cent had remarried, but the equivalent figure for widowers
was 21.2 per cent. Men were also more likely to remarry after a divorce, although here the

85
difference was much smaller. Of those that had been married before, 39.67 per cent had
remarried, compared to 30.08 per cent in the case of women. These data confirm the common
perception that divorced women and particularly widows are less likely to remarry than men
in similar situations.

Country Example 6: Marital Status Categories in Côte d’Ivoire

In Côte d'Ivoire, an effort was made for marital status categories in the census to capture
social reality rather than the legal status quo. Legally speaking, only marriages performed by
a registry are valid. The law further prohibits the payment and acceptance of a bride-price,
polygamy is outlawed and the marriageable age is 18 for women and 20 for men. However,
according to the 1998 DHS, 35 per cent of women lived in a polygamous marriage and
traditional marriages are commonly performed, even with girls as young as 14, in
conservative communities in the North.

In order to adequately capture social reality, the Côte d’Ivoire census of 1998 offered
“consensual/free union (union libre)” and “polygamous marriage with 1, 2, etc. co-wives” as
answer categories for marital status. It also included the following “types of marriage”: legal
marriage, customary marriage, religious marriage, legal and customary marriage, legal and
religious marriage, religious and customary marriage, legal, and customary and religious
marriage.

217. In countries where consensual unions are frequent and not condemned by social norms,
the census is an invaluable tool to report on them, as they are generally not registered.
However, it is likely that undocumented/customary marriages and consensual unions will be
under-reported in the census, especially in countries where they are socially stigmatized.

218. A marital status category that may be relevant from a gender viewpoint but that has to be
constructed from the census data, rather than being directly obtained from the standard
options, is that of married women who are not living with their partners. This status may
reflect a variety of situations. In many cases, it will refer to women whose spouses are living
abroad or elsewhere in the country for work-related reasons. It may also reflect a situation in
which spouses are living apart but do not yet consider themselves permanently separated. In
censuses taken based on the “de facto”, rather than the “de iure” criterion, it may also indicate
that the husband was temporarily absent on census night, even though he normally lives in
the household. It is important to be aware of the enumeration criterion as this third situation is
obviously very different from the other two.

219. The validity of marital status data is compromised by misreporting and various forms of
census editing. False information may be given on marital status for reasons of social
desirability (e.g. in societies where divorce or separation is not socially acceptable).
Estimates by age may also be affected by age misreporting. Also, under-age spouses are often
not asked marital status questions. Where they are asked and report to be married, marital
status is in some countries set to “single” to conform to national legislation. Question
wording and answer categories on consensual unions and polygamy are not harmonised
internationally. Thus, extensive recoding is necessary to make data comparable.

220. Regarding polygamy in particular, there are two important challenges to gender analysis.
First, in many Muslim-majority (e.g. Iraq, Kuwait, Qatar, Mauritania) and some African
countries (e.g. Togo, Uganda), only husbands are asked how many wives they have while

86
wives are only asked whether they are married, without mention of the number of co-wives.
To identify women living in polygamous marriages, one has to select households where a
man declared to be in a polygamous union. Then, if the man is head of the household, one can
identify the women whose relationship to the head of household is declared as “spouse”. This
does not work in situations where the various wives live in different households. Second, in
some African countries, polygamous consensual unions, although frequent, may be
underreported when answer categories do not explicitly include unions involving persons
who are cohabiting or married by customary marriage.

221. Text Box 10 below shows census questions on polygamy that a) ask about polygamy for
men only, b) include answer categories for women married with men who have several wives
c) include polygamous consensual unions.

Text Box 10: Census Questions on Polygamy

a) Qatar Census, 2010

Marital Status
1. Never married
2. Married
3. Divorced
4. Widowed
For married males only: Number of wives

b) Burundi Census, 2008

Is … married? (Est-ce que …est marié ?)
0. Single (Célibataire)
1. Man married monogamously or woman in monogamous marriage (Homme marié
monogame ou femme en mariage monogamique)
2. Man married to two wives or woman married to a man having two wives (Homme marié à
2 épouses ou femme marié à un homme ayant 2 épouses)
3. Man married to three or more wives or woman married to a man having three or more
wives (Homme marié à 3 épouses ou plus ou femme mariée à un homme ayant 3 épouses ou
plus)
4. Free union (Union libre)
5. Divorced (Divorcé)
6. Separated (Séparé)
7. Widowed (Veuf)

c) Uganda Census, 2002

What is NAME's marital status?
1. Never married
2. Currently married (monogamous)
3. Currently married (polygamous)
4. Widowed
5. Divorced
6. Separated
7. Cohabiting (Monogamous)
8. Cohabiting (Polygamous)

Source: UNFPA Census Portal

87
222. Deficiencies in enumerator training and census editing may further affect the validity of
data on polygamy. The 2001 South African Census, for instance, used question wording of
type c) above. However, Table 14 shows that even though marital status included the item
"polygamous marriage" for both sexes, not a single woman was reported in category 3.
Enumerators were instructed that a man with more than one wife should be recorded as
‘polygamous’ rather than civil or customary marriage and that their wives were to be
classified in categories 1 or 2 (married civil/religious or traditional/customary).CAREFUL
HERE: THIS MAY HAVE BEEN DONE THROUGH DATA CLEANING RATHER THAN
THROUGH ENUMERATOR INSTRUCTIONS. THERE ARE OTHER EQUALLY
SERIOUS PROBLEMS IN SOUTH AFRICA REGARDING TRANSLATIONS, PROCESS
MARRIAGES, ETC.

Table 14: South Africa (2008) - Population by sex and marital status

Male Married civil/religious 3,603,016

Married traditional/customary 1,458,342
Polygamous marriage 31,380
Living together like married partners 1,161,375
Never married 14,627,092
Widower/widow 253,148
Separated 117,474
Divorced 182,210
Female Married civil/religious 3,766,062
Married traditional/customary 1,668,682
Polygamous marriage 0
Living together like married partners 1,228,330
Never married 14,601,746
Widower/widow 1,550,031
Separated 200,824
Divorced 370,058
Report of the Census Sub-Committee to the South African Statistics
Council on Census 2001; reproduced on
http://www.statssa.gov.za/extract.htm

223. Censuses can reveal patterns of widowhood in a country. Indeed, often widows are
enumerated as heads of households in census data. Remarriage, on the other hand, cannot be
determined as censuses normally do not ask if a person has been married before. While some
censuses ask about the data of first marriage and current marital status, the “married”
category is not split into “married to 1 st, 2nd, 3rd, … spouse”. The 2006 census of the Maldives
is one of the few that asked how many times each individual had been married and to how
many people, but the quality of the information obtained was not very good. This is a major
obstacle for the identification of gender differences in marriage behaviour.

224. To measure child marriage, “age at first marriage” is the key census variable to analyse.
However, relatively few censuses include it (e.g. Algeria, Azerbaijan, Bermuda, China,
Democratic People’s Republic of Korea, Guinea-Bissau, India, Israel, Kazakhstan, Lesotho,
Malta, Maldives, Occupied Palestine Territories, Republic of Korea, St. Lucia, Sudan,
Swaziland). Where age at first marriage is not collected in a census, a proxy, called Singulate
Mean Age at Marriage (SMAM), can be calculated from the population distribution by
marital status (see the Indicators section below for details). However, the SMAM only

88
provides an aggregate measure for the entire population or sub-groups thereof, but not for
individuals. It is also possible to compare successive censuses, to see how the distribution of
marital status among individuals aged (x,x+n) years in the first censuses has changed t years
later, when these same individuals are aged (x+t,x+n+t).

4. Tabulations

225. Marital status should be tabulated for persons of all ages, irrespective of the national
minimum legal marriageable age. In this way, persons who were married below the minimum
age, persons who married in another country with a different minimum marriageable age and
persons who were permitted to marry below the legal minimum age because of special
circumstances are not excluded. Even then, there is still a risk that child marriages will be
underdeclared or even purged from the data by the census authorities.

226. The Principles and Recommendations for Population and Housing Censuses Rev. 2
(United Nations, 2008 a) recommends basic and additional tabulations with regard to marital
status. The latter can only be applied where the relevant information is available. The most
important basic table is the following:

P4.2-R Population, by marital status, age and sex

227. When compiled for successive censuses, this table can help to detect important trends. In
the case of several East and South-east Asian countries, for example, there has been a
dramatic increase in the proportions of women remaining single in their 30s and 40s,
especially in the big cities. In 2000, 17 per cent of women aged 45-49 in Bangkok remained
single, 13 per cent in Singapore and 10 per cent of Chinese women in Kuala Lumpur. The
2000 census data also show sharp increases in the proportions of non-married among women
in their 30s in some countries where non-marriage rates were traditionally low, notably the
Republic of Korea and Indonesia (Jones, 2003). The following table illustrates this with data
from the 1960-2000 censuses of several countries in the region.

Table 15: Percentages of never married women by age group for consecutive censuses in
East and Southeast Asia

1960 1970 1980 1990 2000

Thailand
30-34 6.7 8.1 11.8 14.1 16.1
35-39 4.2 5.2 7.3 9.6 11.6
40-44 3.1 3.9 5.3 7.0 9.3
45-49 2.6 3.0 4.1 5.2 8.0
Peninsular Malaysia Chinese
30-34 3.8 9.5 13.3 15.8 18.2
35-39 2.7 5.7 7.6 9.1 10.5
40-44 2.6 3.4 5.8 6.4 8.4
45-49 2.5 2.4 4.6 5.7 7.2
Peninsular Malaysia Malays
30-34 1.1 3.3 7.9 10.2 9.7
35-39 0.8 1.9 3.8 5.8 6.0
40-44 0.6 1.1 2.2 4.1 4.4

89
 45-49 0.6 0.7 1.7 2.3 3.2
Japan
30-34 9.6 7.2 9.1 13.9 26.6
35-39 5.6 5.8 5.5 7.5 13.8
40-44 3.1 5.3 4.4 5.8 8.6
45-49 1.9 4.0 4.4 4.6 6.3
Republic of Korea
30-34 0.5 1.4 2.7 5.3 10.7
35-39 0.2 0.4 1.0 2.4 4.3
40-44 0.1 0.2 0.5 1.1 2.6
45-49 0.1 0.1 0.3 0.6 1.7
Singapore Chinese
30-34 4.7 11.1 17.8 22.4 21.9
35-39 4.3 5.8 9.3 15.6 16.2
40-44 5.2 3.8 6.7 12.3 14.1
45-49 6.2 3.3 4.6 7.9 12.6
Hong Kong (1996)
30-34 6.0 5.6 11.0 24.8 26.5
35-39 5.0 3.0 4.5 11.6 14.6
40-44 5.9 2.9 2.7 7.3 9.0
45-49 7.4 3.8 2.3 3.9 5.9

Source: Jones (2003): Tables 1 and 2

Non-marriage has also increased among males, although the age pattern and the timing of the
increase in the various countries have been different from the patterns found in women.

228. Very high percentages of single women aged 35-39 can be found not only in East Asia,
but also in Europe, Australia and the Caribbean region. In these cases, the reason is that
women in this part of the world often do not marry, but live together with their partners
without a formal marriage contract. In Jamaica (2001), according to the World Marriage Data
Base of the UN Population Division (2008), 64.5 per cent of women aged 35-39 had never
been formally married; in Dominica, this was 58.1 per cent. In Sweden (2006), the number
was 40.6 per cent. Compared to these percentages, the number of older single women in East
Asia seems relatively low, but it must be borne in mind that its interpretation is quite different
as unmarried women in East Asia are unlikely to be cohabiting with their partners.

229. Marital status by sex and age should also be tabulated in combination with religion,
school attendance, educational attainment, fertility levels, life expectancy, migratory status,
employment status and disability status. The Principles and Recommendations for Population
and Housing Censuses Rev. 2 – while not explicitly recommending the above tabulations for
religion, fertility, life expectancy and educational variables – suggest the following four as
“additional” under the respective chapter headings:

P2.1-A Foreign-born population, by marital status, age and sex

P7.7-A Usually (or currently) active population, by main occupation, marital status
and age

90
 P7.8-A Usually (or currently) active population, by main status in employment,
marital status and age

P8.3-A Total population 15 of age years and over, by disability status, cross-classified
by marital status, urban/rural residence, age and sex

Table 16: Cambodia (2008) - Age of the (male) head of household and the (female)
spouse

Age of the Female Spouse

70- 75-
Head 10-19 20-24 25-29 30-34 35-39 40-44 45-49 50-54 55-59 60-64 65-69 74 79 80+
10-19 2097 1528 251 44 23 8 16 14 13 9 8 8 2 2
6013
20-24 15167 4 16788 1755 473 151 79 52 12 13 13 7 6 15
9293
25-29 7217 1 139040 19499 4679 1042 306 221 59 102 25 67 15 89
1861
30-34 1177 3 93559 77755 25428 5297 1288 302 79 88 18 50 23 124
13239
35-39 577 7764 42841 90813 1 32201 6511 1203 393 172 94 45 28 78
40-44 356 2403 9760 23988 94998 116058 26801 4803 1147 231 96 27 23 48
45-49 298 1164 3393 7720 27692 84674 103909 19138 3135 496 200 55 37 39
50-54 183 551 1264 2065 5612 16899 63412 63449 9607 1346 300 56 34 31
55-59 127 341 747 901 2482 6608 20807 56889 41017 4988 829 127 37 23
2241
60-64 47 173 495 468 993 2212 5177 19072 38487 8 3077 447 128 47
2233
65-69 30 103 279 266 605 1159 2635 7127 16688 0 14011 1727 301 75
70-74 27 79 183 138 286 491 1006 2495 4525 8975 13721 7657 960 176
75-79 13 35 102 69 172 275 491 952 1893 2835 5336 6429 3720 385
80+ 13 64 151 92 142 137 199 403 635 900 1228 2181 3092 1998

73.5
Mean 24.70 26.72 30.68 35.45 39.59 44.04 49.04 54.73 60.09 65.00 69.55 2 77.20 75.55

230. Table 16 shows the simultaneous distribution of the ages of heads of household and their
spouses, in the case where the head of household is a man, for the 2008 census of Cambodia.
An analogous table can be constructed for the opposite case, where the head of household is
female and the spouse male. On the whole, this particular table does not show any major age
differences between heads and their spouses, with the possible exception of the first column
(spouses under age 20), whose husbands are, on average, 24.70 years old. The mean ages of
the husbands (head of households) for the other age categories are remarkably close to those
of the spouses. On the whole, the difference even tends to diminish with age, which indicates
a relatively low incidence of remarriages of older men with (much) younger women. Spouses
over age 75 tend to have husbands slightly younger than themselves, due to the higher
mortality of men compared to women at higher ages. There are, of course, exceptions (the
numbers most distant from the diagonal), but 82.4 per cent of the spouses have ages within
the same 5-year age bracket as their husbands or the one adjacent to that. Cases of young
women living with much older men (the lower left corner of the table) do occur, but are not
very common. The results in other countries may, of course, be quite different.

231. The study by Teachman, Tedrow and Crowder (2000) in the Country Example below
illustrates the usefulness of census data for investigating long-term trends and change in
marriage and divorce.

91
Country Example 7: Gender and Long-term Shifts in Rates of Marriage and Divorce in
the United States

Teachman et al. (2000) used three national censuses and a variety of sources to examine
changes in marital status of American women over a period of three decades. Overall, they
report a general decline over time in the early formation of marriages, a growing tendency
to never marry (especially for some racial and ethnic groups), and an increase over time in
the dissolution of marriages. The more specific findings fall into three areas: marriage,
singlehood and divorce.

a) Regarding early marriage, among both Whites and African Americans, the proportion of
women aged 20-24 ever married declined by about 32 percentage points between 1975 and
1998. By the late 1990s only one third of White women and 15 per cent of African
American women aged 20-24 had ever married. The observed change in the prevalence of
marriage among Hispanic women has been much more moderate. The percentage of
Hispanic women aged 20-24 ever married changed little until 1985, but then declined
markedly, by 11 percentage points between 1985 and 1990. Between 1990 and 1995, the
percentage ever married rose slightly before experiencing another 9-percentage-point drop
between 1995 and 1998. Still, Hispanic women in the 1990s were more likely than either
Whites or African Americans to have formed a marriage by age 20-24.

b) The likelihood of permanent singlehood has also increased for all groups, with a more
substantial increase in the likelihood of permanent singlehood for African American women
(nevertheless, the study acknowledges that it does not take into consideration the proportion
of women who form nonmarital unions).

c) Divorce has increased in all three groups over the period of the study; the proportion of
ever-married women divorced from their first marriage by age 40-44 rose sharply between
1975 and 1990 for all three race and ethnic groups. For Whites, the increase was
particularly large from 1975 to 1985 (from 20 per cent to 32 per cent) with some slowing in
the 1985–1990 period (32 per cent to 35 per cent). For African Americans, the increase in
the per cent divorced has been more steady, rising from slightly less than 30 per cent in
1975 to 45 per cent in 1990. Hispanic women have also experienced an increase in the per
cent of women aged 40-44 divorced from their first marriage (from just less than 20 per cent
in 1980 to 27 per cent in 1990).

The authors conclude that that marriage has become less valued and less important as a
source of economic stability and exchange, particularly for women in American society.
Gender roles have changed along with economic expectations for women, these changes
have resulted in a shift away from marriage as the dominant marital status category.

232. Marital status by sex, age, religion and/or ethnicity can indicate a relation between belief
and nuptiality, which reflects attitudes towards marriage and divorce as well as different legal
provisions, especially in countries where a “Personal Status Law” - and not a civil code -
regulates marriage, divorce, custody, inheritance and so forth (Israel, most Muslim-majority
countries, with the notable exception of Tunisia). Such tabulations have been used to diffuse
stereotypes, e.g. to show that in the US ‘born-again’ Christians, despite their emphasis on

92
family values, actually have similar divorce rates as other Christians or non-affiliated persons
(Lehrer et al., 1993). The same information can also be used to determine typical differences
in the age at marriage between men and women according to religion, using the concept of
SMAM referred to earlier.

WE WILL DO A VISUALIZATION HERE OF AGE AT FIRST MARRIAGE OR MAYBE

THE DIFFERENCE IN AGE AT FIRST MARRIAGE BETWEEN MEN AND WOMEN
FOR MALAWI, MAYBE DIFFERENTIATED BY CURRENT AGE GROUPS, TO SEE
TRENDS. MAYBE WE CAN SPATIALLY CORRELATE THAT WITH EDUCATION OR
SOME OTHER EXPLANATORY VARIABLE.

233. To understand the impact of education in a context culturally favourable to polygamy,

one may further disaggregate table P4.2-R above, i.e. “Population, by marital status, age and
sex,” by educational attainment and/or literacy. If this does not yield clear results on
polygamy, one may select areas where polygamy is prevalent and compare, for women of the
same age group, ethnic group and religion, if the proportion of women polygamously married
is negatively related to educational attainment/literacy. Past research suggests that women
who have received no formal education are more likely to be in polygamous unions than
women who have received primary or secondary education. Where both spouses have not
received formal education, polygamy is most widespread (UNICEF, 2005). Comparing the
educational level of women in polygamous unions to women in monogamous unions can
confirm these findings in-country. It is necessary to compare women of the same age group to
control for the effect of changes in education.

234. Understanding widowhood can contribute to explaining certain social phenomena such
as poverty. The widowed population can be tabulated by age and crossed with variables such
as household headship, socio-economic level (using the Basic Needs Approach) and receipt
of state benefits (where such data are available). In addition to their larger numbers, widows
are more likely than widowers to co-reside with their children. In Vanuatu (2011), for
example, 33.9 per cent of widowed women over age 60 are living with their children,
compared to 23.9 per cent of widowed men and 18.5 per cent of women over age 60 who are
not widowed. It is important to remember that marriage data uncontrolled for age will give a
distorted image. While there are usually many more widows than widowers this is partially
due to the fact that there are, numerically speaking, more older women than older men in a
population. When cross-classified by age, the proportions are less disparate.

235. Although there is a statistical relationship between the age at first marriage or the marital
status of women at any given age and their level of education, it is generally impossible to
demonstrate the causal direction of this relationship, at least with census data. It may be that
early marriage is an obstacle to further schooling, but it is equally possible that the early
marriage is a consequence of having dropped out of school at an early age. It takes
longitudinal data of a kind that is normally not available in censuses to disentangle the
causality of the relationship. At the very least one should know when the woman got married
and when she left school, but not many censuses ask the former and almost none ask the
latter. One could try to estimate the age of leaving school from the highest grade attended, but
this can be deceptive as it is exactly those women who are most delayed in their education
who are most likely to drop out.

93
Table 17: Malawi (2008) – SMAM by educational attainment

SMAM
Characteristics
Male Female
No Education 23.0 18.2
Primary 23.0 19.5
Secondary 25.2 22.1
Post-Secondary 28.5 26.2

236. The 2008 census of the DPR of Korea is among the ones that collected data on the age at
first marriage. This age is actually quite high in the country, with a mean of 28.4 years for
men and 25.5 years for women. Nevertheless, there is some association between the age at
first marriage and the level of educational attainment. Only 15.7 per cent of women currently
aged 30-39 who married at age 20 or below had post-secondary or higher education. For
those who married between age 21 and 24, the percentage was 18.3 per cent; for those
married between ages 25 and 29, it was 20.9 per cent; for those married between ages 30 and
34, it was 21.5 per cent and for those married after age 35 it was 22.4 per cent. The
corresponding percentages for men currently aged 30-39 were 23.2 per cent, 25.1 per cent,
25.2 per cent, 27.1 per cent and 30.4 per cent, respectively. As was observed earlier, these
data do not allow a strictly causal interpretation. It is difficult to imagine, however, that
marriage between ages 25 and 29 would have had any direct effect on women’s (or men’s)
ability to complete their post-secondary or higher level studies. Therefore, the higher post-
secondary or higher education completion rates of those who married later are more likely to
have a different explanation, like the lower propensity to marry of those who are still
investing in their studies or early post-university careers, perhaps due to the difficulty of
saving for marriage while still in the university. The other conclusion that emerges from these
data, of course, is that, regardless of the age at first marriage, there was still a gap of about 8
per cent between male and female completion rates of post-secondary or higher education.

IT WOULD BE GOOD TO REPEAT THIS ANALYSIS FOR A COUNTRY WITH MUCH

LOWER AGES AT MARRIAGE.

237. Child marriage may result in difficulties entering the labour market, especially where
child brides are taking care of their families. A first step to understanding this issue could be
to study whether among women of a given age group, for example 25-29 years old who
married before 15, or before 20, are economically active (or employed) in lower numbers
than those from their cohort who married later. Obviously, this is only possible if the age at
first marriage is declared in the census. It is important to conduct the analysis by age group,
because behaviours change from one generation to the next due to changes in social norms
and beliefs over time. Adding the number of children, presence of children, or educational
attainment may also further explain labour market participation. The unemployment rate
across women who married as children compared to women who did not marry as children,
with similar background characteristics, can then serve as a measure of women’s status in
society that can be monitored over time with future census data. In addition to the problems
associated with the age at first marriages, one limitation to keep in mind with this analysis is
the possible under-reporting of unemployment for young women, as they are more likely than
young men to be classified as not economically active.

94
5. Indicators

238. Proportion of ever-married (and never-married) men and women. Differences in

women's and men's behaviour towards marriage can be analysed by looking at the proportion
of ever-married women and men and its complement, the proportion of never married women
and men. The proportion of ever married and never married women and men, and their trend,
can reveal important gender differences. In the situation referred to in Country Example 2, for
example, where the sex ratio at birth is considerably imbalanced, the proportion of women
and men ever married and never married below a certain age is skewed since men have fewer
opportunities to find a spouse as women are less numerous. Polygamy can have this same
effect, especially among men of lesser means. Note that, in some countries, it is necessary to
take into account consensual unions and same-sex partnerships to have a complete picture.

239. Proportion of men and women in polygamous unions (prevalence and evolution). The
proportion of men/women (global or by age group) living in polygamous unions should be
interpreted with caution as there is important underreporting of polygamy for women. Often,
analysts therefore chose to look at the marital status of men only to determine the prevalence
and trend of polygamy. The evolution of polygamy can be studied on the basis of successive
censuses, as long as the definitions and questionnaire are comparable. If not, it is possible to
estimate the trend by comparing the behaviour of successive generations, bearing in mind the
limitations of this approach that were commented on in the Box on Visualization of Spatial
Data.

240. Age at first marriage. The Principles and Recommendations (United Nations, 2008 a) do
not recommend any tabulations specifically to measure age at first marriage. Where age at
first marriage is included as a census question (e.g. Algeria, Azerbaijan, Bermuda, China,
Democratic People’s Republic of Korea, Guinea-Bissau, India, Israel, Kazakhstan, Lesotho,
Maldives, Malta, Occupied Palestinian Territories, Republic of Korea, St. Lucia, Sudan,
Swaziland; the 2000 census of Switzerland asked how long people had been in their current
union), misreporting may be widespread, to conceal illegal early marriage. Household
surveys are generally believed to be better suited to analysing child marriage, but face the
same underreporting limitation. Censuses can provide complementary information on girls
and boys, who are under the age of 18 and reported as currently married, at the time of the
census. However, in some countries marital status is not collected for household members
under the minimum legal marriageable age. The Minimum Gender Indicator Set approved by
the UN Statistical Commission in February of 2012 contains one marriage indicator, which
can be computed from census data if the relevant question was asked, namely the percentage
of women aged 20-24 years old who were married at or in a union before age 18.

241. Singulate Mean Age at Marriage. As noted above, when age at first marriage is not
collected in a census, it is advisable to calculate Singulate Mean Age at Marriage as a proxy.
The SMAM is the average length of single life, expressed in years, among those who marry
before age 50. It is calculated from the proportion of single persons (not including persons
separated, divorced or widowed) by age. The main disadvantage of the SMAM in comparison
to individual data on age at marriage is that it is an aggregate indicator. It can be broken down
by major population groups, but it cannot be related to individual characteristics. The other
major limitation of the SMAM is that it does not function well in circumstances where there
are a lot of informal unions and where those leaving such unions tend to declare themselves

95
as “single”, rather than “separated”, “divorced” or “widowed”.

Methodology Box 5: Calculating the SMAM

The steps for calculating the Singulate Mean Age at Marriage are the following

The following illustrates the computational steps with the proportion of never-married
women by age group, in the 2008 census of Malawi:
15-19 70.6 per cent
20-24 17.4
25-29 6.7
30-34 3.0
35-39 1.9
40-44 1.5
45-49 1.2

Step 1. Calculation of the person years lived in a single state:

15*100+5*70.6+5*17.4+5*6.7+5+5*3.0+5*1.9+5*1.5+5*1.2 = 2004.5 (A).
Step 2. Estimation of the proportion remaining single at age 50: 0.9 per cent.
B
(S 4549S 5054)
2

Step 3. Estimation of the proportion ever marrying by age 50: 99.1 per cent (C).


Step 4. Calculation of the number of person-years lived by the proportion not marrying:
50*0.9=45 (D).
Step 5. Calculation of Singulate Mean Age at Marriage (SMAM):
SMAM = (A - D)/C = (2004.5 - 45)/99.1 = 19.77.

242. In some countries, mostly in Africa, the age at first marriage for women is below 20,
such as in Niger, where the SMAM was 17.6 in 2006. At the other end of the spectrum,
Northern European countries have the highest age at first marriage for women, as in Sweden
where the SMAM was 32.2 in 2006. The analysis of the trend is also necessary to understand
the dynamic: in Niger, the age at first marriage for women has increased from 16.2 in 1977 to
17.6 in 2006 (United Nations, 2009 a).

243. Proportion of women married below the legal age, by age . To answer the question of
whether child marriage is decreasing in a country or not, a graph can be presented showing
the proportion of women married below the legal age, by their current age. If the curve is
increasing with age, it means that younger women get married less early than their elders. If
only marital status is available in the census questionnaire, it is necessary to combine
successive censuses to analyse the trend. Attention should be paid to the comparability of
these censuses, in terms of age, marital status reporting (definition and methodology), and
coverage.

244. Age at first marriage of women and men, by age group. Comparing the age at first
marriage of men and women, by age group or cohort, is important for understanding the
scope of early marriage.
________________________________________________________________________

Country Example 9: Using the Singulate Mean Age at Marriage to Examine Early
Marriage in Malawi

96
Table 18 shows how the general proportionate SMAM increased over time for all age
categories and for both sexes in Malawi. For women, the SMAM increased from 17.8 in 1977
to 19.8 in 2008, while that for males increased from 22.9 in 1977 to 23.9 in 2008, indicating
that generally, more women still marry younger than their male counterparts. Consequently,
the proportion of the population staying single increased, with that of males aged 15-19
increasing from 93.8 per cent in 1977 to 95.2 per cent in 2008, while that of females aged 15-
19 years increased from 48.9 per cent in 1977 to 70.6 per cent in 2008.

Nonetheless, the proportion of females getting married early is still much higher than that of
their male counterparts and requires addressing. For example the proportion of females aged
15-19 getting married was 29.4 per cent while that of their male counterparts was 4.8 per
cent. Similarly, the proportion of females aged 20-24 years and 25-29 years (17.4 per cent
and 6.7 per cent respectively) staying single is still much lower than that of their male
counterparts (54 per cent and 21.2 per cent respectively). These trends suggest that girls are
still marrying young and strategies to prevent this, such as retaining girls in school, especially
secondary school, would go a long way in keeping girls in school longer, increasing the age
at which they marry and reducing their fertility rate.

Table 18: Malawi - Proportion single and Singulate Mean Age at Marriage: 1977, 1987,
1998 and 2008

Percentage Single
Age Group Male Female
1977 1987 1998 2008 1977 1987 1998 2008
15-19 93.8 91.1 91.7 95.2 48.9 55.1 61.8 70.6
20-24 49.3 51.4 53.0 54.0 7.4 11.5 14.6 17.4
25-29 13.3 17.4 18.0 21.2 2.2 3.5 4.8 6.7
30-34 4.9 6.3 6.0 7.7 1.3 1.6 2.1 3.0
35-39 2.9 3.4 3.4 4.1 1.0 0.9 1.3 1.9
40-44 2.3 2.3 2.6 2.7 1.0 0.8 1.1 1.5
45-49 1.8 1.7 1.7 2.1 0.9 0.7 1.0 1.2
SMAM 22.9 23.2 23.4 23.9 17.8 18.4 19.0 19.8

Source: Malawi. Gender in Malawi. Analytical Report 3 of the 2008 Census: Table 4.5

Table 18 confirms the earlier statement that females generally entered marriage at a younger
age than their male counterparts, regardless of residence and educational attainment. It shows
that rural women entered marriage 2.1 years earlier than urban women, while women with no
education entered marriage 8 years earlier than those who had post secondary education. This
suggests that being rural and being uneducated or less educated renders a young woman more
vulnerable to early marriage. It confirms previous assertions that education and residence
have an impact on a women’s entry into marriage and consequently their fertility.

Table 19: Malawi - Singulate Mean Age at Marriage by residence and educational
attainment

SMAM
Characteristics
Male Female
Residence

97
 Urban 25.9 21.9
Rural 23.4 19.8
Educational Attainment
No Education 23.0 18.2
Primary 23.0 19.5
Secondary 25.2 22.1
Post-Secondary 28.5 26.2

Source: Malawi. Gender in Malawi. Analytical Report 3 of the 2008 Census: Table 4.6
_________________________________________________________________________

245. Analogously to the SMAM, other measures can be defined for the timing of events. It
has been suggested, for example, to define a Mean or Median Age at Widowhood and to
compare this measure between men and women. Although this measure can be constructed
from most census data, it has the following limitations:
1. Women may remarry, so that the mean/median age will be over-stated. In Ireland
(2006), for example, the mean age of widowhood for ever-married women was 55.9
years if remarriage is taken into account and 56.2 if not.
2. Most women marry at some point, but a lot of women (and particularly men) never
become widow(er)s. In the Ethiopian census of 2007, for example, widowhood in the
highest age group (75+) was 62.1 per cent for women and only 11.2 per cent for men.
3. Because widowhood is most prevalent in the very highest age groups, the results will
be sensitive to where the cut-off point for the last age group is placed.
4. The concept of widowhood is problematic in contexts where a high proportion of
unions are informal.
5. If there is differential mortality of widows, the results will be distorted, especially in
the highest age groups.

246. Mean difference in the age at first marriage of the spouses. An indicator that can be
derived from the mean age at first marriage (measured directly through the relevant census
question or indirectly through the SMAM) is the mean difference in age at first marriage of
the spouses. This is relevant from a gender perspective because women who are much
younger than their husbands generally have less autonomy and authority in the marital
relationship. By and large, differences in age at first marriage between the spouses have been
diminishing, but they remain large in some countries in West Africa, such as Mauritania (7.6
years in 2001) and Sierra Leone (6.8 years in 2004) (United Nations, 2009a). The indicator is
less adequate in societies in which remarriage is frequent or where polygamy is widespread
because it does not measure the age differences in these later unions. In second and third
unions or marriages, age differences between the spouses tend to increase as men often
remarry with substantially younger wives. Consequently, the gap between the mean ages of
husbands and wives tends to widen as they grow older, which increases the probability of
widowhood and its economic and social consequences for women, as discussed earlier.

247. A simpler measure to compute is the difference between the mean age of married men
and married women. This measure can be compared to the SMAM or to the average age at
first marriage if these data are gathered. In the 2008 census of Mozambique, for instance, the
SMAM was 18.1 years for women and 22.4 years for men, a difference of 4.3 years. But the
average age of women that were married or living in consensual unions was 33.4 years,
compared to age 40 for men, which means a difference of 6.6 years (NSO Mozambique
website, accessed 8 April 2011). It should be pointed out, however, that these two indicators

98
measure different things. The SMAM only refers to first marriages, but the average age
difference of married persons mixes first marriages with remarriages.
Further national level interpretation on this issue can be found in the CEDAW Committee
concluding comments for its countries, at
(http://www2.ohchr.org/english/bodies/cedaw/cedaws).

All of these indicators should be analysed by region within a country and by religious/ethnic
group, if available, as the prevalence of child marriage will be higher where a culture of
gender inequality prevails, as well as in regions prone to conflict or natural disaster.

6. Multivariate and further gender analyses

248. An obvious use of logistic regression is to analyse the marital status of women based on
certain explanatory variables such as age, educational attainment and/or literacy of both
spouses, religion/ethnicity, and place of residence (rural/urban). Where available (the SMAM
will not do in this case), the age at first marriage should also be used. Widowhood, for
example, is associated with early marriage, male over-mortality, and social norms regarding
remarriage. Where female age at marriage increases, levels of widowhood decline (UNICEF,
2005).

249. Measuring the scope and frequency of early marriage and its trend over time is essential
for developing national policies and legislation. In particular, knowing what individual-level
characteristics are associated with child marriage may be useful to plan policy interventions
to prevent it. In multivariate analysis, age at marriage (where it is available) could then be
treated as a dependent variable in order to model the factors that affect age at marriage.
Taking this line of analysis, Maitra (2004) finds that ethnicity, religion and parental education
all are significantly associated with age at marriage. In a cross-country study with 50
countries, UNICEF (2005) found that the educational level of girls was significantly
associated with higher ages at marriage. The spousal age gap was negatively associated with
the woman's age at marriage: women more than four years younger than their partners were
more likely to be married as children.

250. An excellent example of the use of these kinds of methods for the analysis of marital
status comes out of the 2009 census of Viet Nam (Viet Nam, 2011), which performed a series
of logistic regressions of different marital status categories. As an illustration, the following
reproduces the table with regression coefficients and the comments of the report on the
probability of never marriage among population aged 40–69.

251. “In this analysis, based on social norms and distribution of marital status by age in Viet
Nam, delayed marriage is defined as the situation of individuals who delay marrying till after
the age of 40. The term delayed marriage is used for convenience, but in fact, includes also
people who will never marry. In addition, it should be noted that delaying marriage, as
defined in this monograph, does not necessarily correspond to the level of SMAM in the
population. According to estimates from the Census sample survey data, by the time of the
1999 Census in Viet Nam, there were more than 84,000 males and 371,000 females aged 40
and older who had never been married, accounting for 1.1% and 3.8% of males and females
respectively in this age cohort. Ten years later, by the time of the 2009 Census, the
corresponding numbers had increased to 210,000 males and 635,000 with the proportions at
1.7% and 4.4%, respectively. The absolute size of the never-married population increased
greatly over the past ten years not only because of the increases in the size of the total

99
population but also because of increases in the proportion never-married in the population.
Particularly, from 1999 to 2009, the proportions never-married among males aged 40–49 and
of both sexes aged 50–59 and 60–69 had all increased. Only the proportion never-married
among females aged 40–49 had decreased (from 6.2% to 5.7%), most likely because of the
recent decline in the population sex ratio. However, in general, the number and the proportion
delaying marriage among females are much higher than among males, reflecting the situation
of low sex ratio of the population in Viet Nam in the last several decades.

From the birth cohort perspective, the size of the never-married population has decreased
during the period 1999-2009. In 1999, about 58,000 males aged 40–49 were never-married,
accounting for 1.6% of the cohort. In 2009, this cohort now aged 50–59 years had only
42,000 never married males, accounting for 1.2% of this birth cohort. The numbers declined
not only because of marriage, but also because of mortality and international emigration.
However, if mortality and international migration rates are not much different by marital
status, the decline of about one third (from 1.6% to1.2%) would be close to the proportion
getting married in this cohort over the 10 years between the Censuses. For other cohorts
(except the cohort 70+ because of the strong effects of mortality), the probabilities of getting
married in the ages 40 and older for males (about 25% after 10 years) are higher than for
females (less than 15% after 10 years).

Figure 7: Viet Nam (2009) - Maps of the proportion never married among the popula-
tion aged 40 and older by province

100
Comparing the status of never-marriage after age 40 of males and female in urban and rural
areas in 2009. The results show that, delayed marriage is more frequent in urban areas than in
rural areas in all four age groups of both sexes. The proportion never-married among males in
the age group 40–49 in urban areas is about three times higher than in rural areas (5% versus
1.7%), and the proportion never-married among females in urban areas is about 1.7 times
higher than in rural areas (7.9% versus 4.6%). This corresponds to the general pattern that
delayed marriage or never marriage is becoming more common in regions with higher levels
of economic development and industrialization. Figure 7 presents the maps for the proportion
never married for both sexes among people age 40 and older in all provinces in Viet Nam in
2009. It is likely that the pattern of “delayed marriage becoming more common in areas with
higher levels of economic development and industrialization” is more consistent with the
situation in provinces from Da Nang and further south. The proportion never married among
the population aged 40 and older is highest in more industrialized provinces such as Da Nang,
Ho Chi Minh City, and Binh Duong. In the North, the situation is different when the highest
proportions delaying marriage are not found in Hanoi or Hai Phong, but in Ha Giang (for
males) and Thai Binh, Ha Nam, Nam Dinh, and Ninh Binh (for females). Thus, it can be
concluded that delayed marriage in Viet Nam is not only related to the level of
industrialization but also depends on other socio-cultural factors.

Table 20 presents the regression model of probability of never-marriage among the

population aged 40–69 in Viet Nam in 2009. The age group 70+ is excluded from the model
because of the small proportions never married and strong influence of mortality. The
dependent variable is delayed marriage status of individuals (never-married=1, ever-
married=0). The independent variables are the same as earlier but include also four types of
disability status: vision, hearing, walking, and memory. Other technical specifications and
interpretations are similar to the previous regression model. In general, the results indicate
that, in comparison to the previous model on early marriage, the independents variables in
this regression model explain smaller part of the variation of the dependent variable,
especially in female group. The reason is that delayed marriage may be more strongly related
to other omitted variables. Below can be found the analysis of the relationship between each
independent variable and the probability of never-marriage among the population aged 40–
69. The results show that among males and females aged 40–69, the probability of never-
marriage is higher than the Northern Midlands and Mountains for all five remaining regions,
especially for the Southeast and Mekong River Delta. Only the regression coefficient for
females in the Central Highlands is negative. Thus, holding constant the other independent
variables in the model, females in the Central Highlands and males in the Northern Midlands
and Mountains are most likely to be married by age 40 in comparison with other regions. On
the other hand, for both males and females, the probability of delayed marriage in the
Southeast was significantly higher than in other regions.

Table 20: Viet Nam (2009) - Logistic regression of probability of never marriage among
population aged 40–69

Male Female
Region

Northern Midlands and Mountains - -

Red River Delta 0.714 0.718
North and South Central Coast 0.519 0.563
Central Highlands 0.446 - 0.185

101
Southeast 1.545 1.080
Mekong River Delta 0.944 0.688
Urban (Rural =0) 0.917 0.438

Age Group

40-44 - -
45-49 - 0.556 - 0.045
50-54 - 1.216 - 0.136
55-59 - 1.794 - 0.309
60-64 - 2.446 - 0.709
65-69 - 3.044 - 1.495
In-migrant 0.216 0.147
Ethnic minority 0.141 - 0.077
Religious adherent 0.220 0.321

Educational Attainment

Below Primary - -
Below Lower Secondary - 1.314 - 0.608
Below Upper Secondary - 1.518 - 0.712
Upper Secondary - 1.443 - 0.651
Post-Secondary - 1.573 - 0.564
Working - 1.238 - 0.038
Vision Disability 0.399 0.722
Hearing Disability 0.087 0.166
Walking Disability 0.084 0.363
Memory Disability 1.676 1.150

Constant - 2.279 - 3.081

Source: Viet Nam (2011): Table A.20

Regarding urban and rural areas, the regression model one again confirms the results
analysed earlier. Population aged 40–69 in urban area is more likely to be never-married than
in rural areas, and the difference is clearer for males than for females. Third, regarding age,
the probability of being never-married decreases quickly as age increases, especially for
males. That means as age increases, the proportion delaying marriage decreases because
many individuals get married after they turn 40 (not because old people can get married more
easily than the young). Only for females, the difference between the age group 40–44 and
45–49 is not statistically significant. Forth, regarding migration status, the results show that
for males and females aged 40–69, in-migrants have a higher probability of delaying
marriage than non-migrants. Combined with the results in the regression model in Table20, it
can be concluded that migration is relevant to both early and delayed marriage of females in
contemporary Viet Nam.

Fifth, regarding ethnicity, it is interesting that, the probability of delayed marriage among
ethnic minority males is higher than among Kinh males. In contrast, the probability of
delaying marriage among ethnic minority females is lower than for Kinh females, holding
other variables constant. The difference is small but it is statistically significant. One of the

102
possible reasons is that the sex ratio among the young and middle-aged people in the ethnic
minorities is lower (more balance) than in the Kinh population.

Sixth, concerning religion, the probability of being unmarried among both males and females
who are religious adherents is higher than in the non-religious groups. This seems reasonable
as some people do not marry because they are religious adherents, while some people become
religious because they are unable to get married.

Regarding educational attainment, people with higher educational levels are less likely to
delay marriage compared to those with less than primary education, and the difference is
stronger for males than for females. Thus, low educational levels may be the direct or indirect
cause of delaying marriage for people aged 40–69, especially for males. However, the
regression coefficients do not vary much between the level at “less than lower secondary
school” and the higher levels, especially for females. This shows that the probability of
delaying marriage among people aged 40–69 is not significantly related to educational
achievement, except for the group “less than primary school” that are more likely to be
married at the age 40-69. If high educational attainments leads to later marriage, it must be
very high educational achievement such as post-university, not the educational levels
considered in the regression.

The results on working status show that there is a significant difference between males and
females. The probability of delaying marriage among working males is significantly lower
than for nonworking males. However, the working status of females aged 40–69 is not
significantly related to their probability of being unmarried. This result corresponds with the
general view that working males can more easily get married than unemployed males and
vice versa, married males are more responsible than unmarried males so they find jobs in
order to be the breadwinners for their families.

And last but not least are the results on disability status. As predicted, people with disabilities
have a higher probability of delaying marriage than people without disabilities. The highest
probability of delaying marriage is for people with memory disability (difficulty with
memory and concentration), followed by people with vision disability (difficulty in seeing
even with glasses). Males with walking disabilities (difficulty in moving around) and hearing
disabilities (hard of hearing) are more likely to delay marriage than people without
disabilities, but the differences are small. Compared to the female model, the male model
reports a higher coefficient for memory disability, lower coefficients for vision and walking
disabilities, and a similar coefficient for hearing disability.

In short, delayed marriage (defined as being unmarried among the age group age 40 to 69) is
most correlated to low educational attainment, disability (especially memory and vision
disability), religious adherence, in-migration status, and residence in the Southeast and the
Mekong River Delta.”

252. By using appropriate multivariate regression techniques, one may underpin the marital
status-education-work relationship for women. The basic question in this relationship is
whether the marital status of a woman has a direct effect on her labour force participation,
after controling for other intervening factors. The following logistic regression, based on the
Aruba 2010 Population and Housing Census, was used to study the relationship. The
dependent variable in the analysis was a dichotomy: whether the woman worked or not. The
analysis was restricted to women aged 25-50 years, because below age 25 many women are

103
still in school and above age 50 many women on Aruba withdraw from the labour market and
most mothers have grown up children. The predictors used in the analysis were: age, number
of children ever born (CEB), household income excluding that of the women in question,
country of birth, educational attainment, marital status and a variable which indicated if the
woman was living together with a partner or not. Country of birth was included in the
analysis as many foreign women come to work in the Aruban hotel sector. Next to marital
status, the variable indicating if the woman was living together with a partner on a durable
basis was added, because on Aruba, consensual unions are very common. The variable on
household income (excluding that of the women in question) is included to control for the
economic necessity of the female respondent to work or not.

253. The results of the analysis are presented in Table 21. As the analysis is based on census
data – and not a survey - the standard errors and significance levels of the regression
coefficients are irrelevant and left out of the table. Among the categorical variables, the
following reference categories were used: Aruba (country of birth), less than primary/no
education (educational attainment), never married (marital status) and living together. The
values in the exp(B) column show the odds ratios. These ratios are computed by raising e to
the power of the regression coefficient.

254. The logistic regression shows some interesting results in terms of the position of women
and their labour force participation. First, the odds ratio for CEB (0.935) shows that on Aruba
the odds of being at work for a woman is 6.5 per cent lower for each additional child she gave
birth to. Second, participation in the labour force varies quite significantly by country of
birth. The highest participation is among women from the Dominican Republic and the
lowest among women from the USA. Note that no coefficient is entered for Aruba, as this is
the residual category against which all the others are measured. Also, the higher a woman’s
educational attainment, the higher her chances of being at work. Note the very low value of
women with a PhD, which group consists only of a few women. Living together with a
partner has some effect on the chances of having a job, but not substantially (1.08). However,
marital status plays a much more important role to determine the work status of a woman.
The odds of being at work for married women on Aruba is only 0.653 that of never-married
women, after controlling for all other predictors. Divorced women, on the other hand, have
higher odds (1.158), while widowed women score lower (0.704).

Table 21: Aruba (2010) - Logistic regression of the probability of working for women
aged 25-50, by selected explanatory variables

Explanatory variable Category B exp(B)

Constant -0.114 0.892

Age 0.021 1.021
Number of children ever born -0.067 0.935
Total income of other household members 0.000 1.000
Country of birth Aruba
Colombia -0.294 0.746
USA -0.899 0.407
Dominican Republic 0.196 1.217
Venezuela -0.804 0.447
Curaçao -0.100 0.905
Netherlands -0.273 0.761

104
 Other -0.101 0.904
Educational attainment None/Less than primary
Primary 0.490 1.633
Lower vocational 0.765 2.150
High school (4 year cycle) 1.089 2.971
High school (5 year cycle) 0.978 2.659
High school (6 year cycle) 0.697 2.008
Intermediate vocational 1.401 4.059
Higher (Bachelor) 1.656 5.236
Higher (Master's) 1.474 4.368
Higher (PhD) -0.142 0.867
Marital status Never married
Married -0.426 0.653
Divorced/legally separated 0.147 1.158
Widowed -0.352 0.704
Living together Yes
No 0.077 1.080

Source: Population and Housing Census Aruba 2010

7. Interpretation, Policy and Advocacy

255. When interpreting data on marital status for the purpose of gender analysis, it is
important to remember that “being married” may not mean the same thing to women and men
in terms of lived experiences. Particularly in countries where laws governing married status
differ by religious denomination, “being married” diverges even in its legal meaning. For
instance, the level of difficulty involved in passing on religious denomination and nationality
or securing custody for their children differs for Muslim, Christian and Druze women in
Lebanon.

Text Box 11: Marriage and Divorce from a Gender Perspective

Gender advocates have struggled for decades to make divorce an option for women. While
one reason for this is the possibility to escape an abusive relationship, another is that the mere
possibility of divorce provides women with leverage to gain a more equal status within
marriage (Yodanis, 2005).

One example in support of this view (i.e. the possibility of divorce leads to better marriages)
is Indonesia: Here, divorce rates have been declining not as a consequence of conservative
gender ideologies, but due to increased free choice in marriage, educational expansion,
delayed marriage, urbanization, increasing employment before marriage, and legislative
change (Heaton et al, 2001).

This example also illustrates the importance of contextual information in analysing the data.
Thus, a decline in divorce rates can be interpreted in different ways. Additional research and
qualitative studies are often useful to correctly interpret the findings.

105
256. At the time of writing, divorce is legal in all countries globally, except for Filipino non-
Muslims. However, in many Muslim-majority countries, obtaining a divorce is significantly
more difficult for women than for men.

257. Polygamy is a contentious issue in many societies, with all countries influenced by
Islamic law, except Tunisia, permitting polygamy. However, some countries restrict
polygamy by requiring court permission (Syria, Morocco, Iraq), or, in the case of Pakistan,
the permission of an arbitration council. Also, Jordan has enacted legislation permitting a
wife at the time of marriage to include a stipulation in her contract that gives her the right to
divorce her husband if he marries another woman (Mashhour, 2005). Polygamy, typically
polygyny by nature and practice, confers power, status and privilege to a man over and above
that of a woman. Hence, the CEDAW Committee in its General Recommendation 21 notes,
“Polygamous marriage contravenes a woman's right to equality with men, and can have such
serious emotional and financial consequences for her and her dependents that such marriages
ought to be discouraged and prohibited.”20

258. A high prevalence of girls married under the age of 18, when their male peers remain
single, is an indicator for gender inequality in that country. Of note, many countries set legal
marriageable ages that differ from the internationally agreed benchmark of age 18 for both
women and men. Moreover, governments set different marriageable ages for females and
males (e.g. Senegal, 20 for men and 16 for women; state of Ohio in the US, 18 for men and
16 for women; Bangladesh, 21 for men and 18 for women) and some countries such as
Kenya, Jordan and Paraguay set the minimum age for marriage below 18 years for both
sexes.

259. Child marriage takes place almost exclusively within the context of poverty and gender
inequality and has important social, cultural and economic dimensions. While impoverished
rural parents may believe that child marriage will protect their daughters, it in fact results in
lost development opportunities and limited life options. Often, child brides are pulled out of
school, depriving them of an education and meaningful work, and increasing their
dependency on their husbands (Manda and Meyer, 2005). Early widowhood is also associated
with child marriage as many girls are married to older men, and men’s life-expectancy is
lower than that of women in most countries.

260. As is always the case, ultimately the reduction of child marriage can be brought about by
providing better alternatives to women. For example, the Government of Malawi has decided
to enhance educational spending on girls nationwide in order to curb the negative social and
economic consequences of child marriage. In parallel, Malawi has introduced targeted
programmes in some regions to boost women’s employment and support family planning
services (Manda and Meyer, 2005). For countries with important ethnic cleavages, the
targeting of child marriage prevention efforts might be refined to focus on girls from
communities most at risk of marrying their girls as children, be they majority or minority
groups.
20
This is based on Article 5a) of the original CEDAW text, which states that “States parties shall take all appro -
priate measures to modify the social and cultural patterns of conduct of men and women, with a view to achiev -
ing the elimination of prejudices and customary and all other practices which are based on the idea of the inferi -
ority or the superiority of either of the sexes or on stereotyped roles for men and women.” Similarly, General
Comment 14 of the CEDAW states that “States parties' reports also disclose that polygamy is practiced in a
number of countries. Polygamous marriage contravenes a woman’s right to equality with men, and can have
such serious emotional and financial consequences for her and her dependants that such marriages ought to be
discouraged and prohibited.”

106
261. Uganda recently unveiled its long awaited proposed revisions to the Marriage and
Divorce Bill. Seems the updates have been about 4 decades in the making. The bill, which
gives women the right to divorce an impotent husband, also establishes equitable distribution
of property between spouses upon divorce, providing co-habiting couples with the same
rights to property as married people. For the first time in Uganda, it also establishes marital
rape as a crime. The wide ranging piece of legislation outlaws the practice of widow
inheritance and makes it an offense to demand the return of bride price upon dissolution of a
marriage. The most important thing is that the laws in Uganda finally recognize the “non-
monetary” contributions of aggrieved women to a broken marriage, i.e. childrearing, and
finally gives them means of redress in a divorce. It prohibits widow inheritance, in
conformity with Article 32(2) of the Ugandan Constitution stating that laws, cultures,
customs and traditions which are against the dignity, welfare and interest of women or which
undermine their status, are prohibited by the Constitution. It reforms and consolidates the law
relating to civil, Christian, Hindu, Bahai and customary marriages in terms of marital rights
and duties, separation and divorce legislation. Data from the Ugandan Bureau of Statistics
(UBOS) was used to show how the principles of equality and non-discrimination are violated
with the current state of affairs: Women from certain backgrounds (less educated, certain
ethnic groups) are affected by polygamy more than others.

262. Showing the effect of child marriage on girls’ education, economic status and other
indicators of women’s wellbeing can help highlight the loss for a national economy and the
consequences on public health. All of the tabulations and multivariate analyses described
above can be reproduced at local geographical level, which allows identifying areas such as
rural areas or regions in the country that should be targeted by specific measures or
campaigns.

107
Chapter 7:
Households and Families
1. What is it?

263. Household and family compositions are core topics in censuses. As the household is
based on the arrangements of people to provide them with food, shelter and other essentials,
it is the basis to study the position of each member within the group. The Principles and
Recommendations for Population and Housing Censuses, Rev. 2 (United Nations, 2008 a)
defines household and family as follows:

‘A household may be either:

a) A one-person household, that is to say, a person who makes provision for his or her
own food or other essentials for living without combining with any other person to
form part of a multi-person household; or

b) A multi-person household, i.e. a group of two or more persons living together who
make common provision for food or other essentials for living. The persons in the
group may pool their resources and have a common budget; they may be related or
unrelated persons or a combination of persons both related and unrelated. This
arrangement exemplifies the “housekeeping” concept.’

264. ‘The family within the household, a concept of particular interest, is defined as those
members of the household who are related, to a specified degree, through blood, adoption or
marriage. The degree of relationship used in determining the limits of the family in this sense
is dependent upon the uses to which the data are to be put and so cannot be established for
worldwide use. (The Principles and Recommendations for Population and Housing
Censuses, Rev. 2 - United Nations, 2008 a)

A family nucleus may take one of the following forms consisting of persons living in the
same household:

a. A married couple without children;

b. A married couple with one or more unmarried children;
c. A father with one or more unmarried children;
d. A mother with one or more unmarried children.

In countries were consensual unions are common; two additional categories may be added:

e. Partners in consensual with child(ren)

f. Partners in consensual without children

265. Unfortunately, the Principles and Recommendations for Population and Housing
Censuses do not provide a clear advice how family nuclei should be mapped in households
consisting of persons living in polygamous unions. In principle, countries are left to make
their own decisions in this respect. The Principles and Recommendations for Population and
Housing Censuses suggest classifying households at least in four categories, i.e. one-person
household, nuclear household, extended household and composite household. A nuclear
household consists of a family nucleus, but no other persons. A household can be defined as
either extended or composite, when next to the family nucleus other persons are present. The

108
difference between extended and composite households lies in the fact that in an extended
household all members are related to each other, while in composite households one or more
members are not related to at least one of the other members of the household. Next to the
four main household categories, countries can decide to subdivide nuclear, extended and
composite households further into a number of subcategories.

266. Traditionally in population censuses the type of household and family composition is
determined on the basis of each member’s relationship to the reference person or head of
household. At the onset of the census interview, the enumerator identifies the reference
person (or head of household) on the basis of pre-defined specific criteria. For all other
remaining persons in the household their relationship to the reference persons is then
recorded.

267. The concept of ‘housing unit’ is closely related to household. A housing unit is a
separate and independent place of abode intended for habitation by one household. As women
and girls generally spend more time in domestic consumption work than men and boys, the
condition of the living quarter is an important aspect to determine their well-being and health.

2. Why is it important?

268. The International Conference on Population and Development (IDPD) Programme of

Action (United Nations, 1994) recognizes the importance of understanding the implications
of household composition when it declares (in Par. 5.6) that “Governments should maintain
and further develop mechanisms to document changes and undertake studies on family
composition and structure, especially on the prevalence of one-person households, and
single-parent and multi-generational families.” Moreover, it stresses (in its Par. 5.9) that
governments must “develop the capacity to monitor the impact of social and economic
decisions and actions on the well-being of families, on the status of women within families,
and on the ability of families to meet the basic needs of their members.” In this context,
censuses become a crucial data source for evidence-based policies and interventions. Finally,
according to par. 3.70 of the Principles and Recommendations, “the household, a basic socio-
economic unit in all countries, is often central to the study of social and economic
development.”

269. The number, size and structure of households and changes in the rate of household
formation are useful for planning and for developing special policies formulated for
vulnerable population segments. Roughly speaking, the issues can be broken down according
to three criteria, namely:

a. To identify changes in the size, structure and characteristics of family nuclei and
households and determine the stages in the life cycle of families and households (e.g. families
with young children, households with one or more retired members...).
b. To indicate the presence of members of the household with specific key characteristics, for
instance: young dependent children, adult income earners other than the head, elderly persons
with special needs, household members with disabilities and domestic servants
c. To look for the possibility to debunk certain ideological conceptions about what constitutes
a “normal” household composition, by confronting such notions with what is actually
observed.

270. The characteristics of households then may highlight potential needs and problems,

109
specifically in the areas of income and poverty, the education of children, and the ability of
some household members -such as women- to work outside the home. Families with children
present a higher likelihood of vulnerability and poverty than families without children
(OECD, 2011). Other research finds enormous variations in poverty across female-headed
households with specific compositional, residential and racial characteristics (Snyder,
McLaughlin and Findeis, 2006). The World’s Women 2010 (United Nations, 2010 a) notes
that while generalizations between “female-headed households” and “male-headed
households” are not ideal, in Latin America and other developing regions, households of lone
mothers with children have higher poverty rates than those of lone fathers with children, and
that poverty rates for women living in one-person households are higher than the
corresponding rates for men. In addition, the analysis of household composition may help in
understanding other phenomena, such as fertility, differential remarriage of men and women,
and son preference.

271. The Beijing Platform for Action (Par. 22) recognizes, for example, that “female-
maintained households are very often among the poorest because of wage discrimination,
occupational segregation patterns in the labour market and other gender-based barriers.”
Moreover, the Beijing Platform for Action (Par. 46) adds, “Many women encounter specific
obstacles related to their family status, particularly as single parents.” Monitoring the
situation of the most vulnerable families such as female-headed or female-maintained
households is crucial for the advancement of more effective policies, not only regarding the
promotion of gender equity – such as the elimination of the ‘feminization of poverty’ or the
phenomenon in which women experience poverty at far higher rates than men – but also
regarding general social and economic development.

272. Household and family composition and structure may also be associated with gender
inequalities in employment and education. In most cultures it is still predominantly women’s
responsibility to perform daily family tasks. In the same vein, girls are more likely than boys
to perform household work (e.g. care-giving, cooking and cleaning). The care-taking
activities of girls in the household may impair their school attendance and social life.
Depending on household composition, such tendencies may be intensified and jeopardize
women’s and girls’ life opportunities. In several countries, high fertility limits women’s
opportunities for education and employment (United Nations, 2010 a; see also the chapter on
fertility).

273. Different living arrangements often imply different needs. For example, lone parents
may have needs that are substantially different from married parents, and the needs of lone
mothers may differ from those of lone fathers (e.g. economics, employment, social support,
education, and parenting). Censuses may provide information on the diversity within family
types across gender, and also by factors such as race/ethnicity, education, and age. Indeed, in-
depth analyses of household composition, based on census data may be fundamental for
policies designed to address the growing diversity of families and their specific
circumstances, characteristics and needs, including the most vulnerable family arrangements
(e.g. mother-only families in minority groups and families of teenage mothers).

274. Population and Housing Censuses may also provide valuable information about the
quality of the environment that families live and work in. For example, a topic of serious
concern is indoor air pollution. According to the World Health Organization, indoor air
pollution is responsible for an estimated 1.6 million deaths per year
(http://www.who.int/heli/risks/indoorair/indoorair/en/). About half these deaths are among

110
small children under the age of 5. Indoor air pollution is mostly caused by burning biomass
fuels (wood, crop waste or dung) in badly ventilated kitchens and rooms for cooking or
heating purposes. Women spend a considerable time of the day inside the house preparing
food and performing other household chores. They are –next to small children- most at risk
for adverse health effects of indoor air pollution. The problem is most severe among the
poorest segments of society, as they are more likely to live in poorly ventilated housing units
and use biomass fuels. As many Population and Housing Censuses gather data on fuel for
cooking, kitchen facilities and the physical characteristics of the housing unit, they are an
important source of information on this issue.

275. Census data on household and family composition can be useful to define poverty and
vulnerability and to plan policies targeting these areas, despite the fact that poverty
measurement based on census data has its limitations (see Chapter 8). Household
composition characteristics, such as sex, age and number of family members, may also
influence the risk of poverty and vulnerability. As an example, poverty risks are often highest
in single-earner families and lowest in dual-earner families (OECD, 2011). Using “dual” or
“double” earner families, similarly to a family structure type, may allow for the analysis of
women’s participation in the labour market, and to consider the possible overlap of women’s
roles and their status at home. In this way, the double burden of women’s work can be
considered.

276. In addition, grandparents, especially grandmothers, can have an important impact on the
wellbeing of children (see Aubel, Pandey and Rijal, 1999; Aubel et al., 2000a; Aubel et al.,
2000b, 2003; Bedri, 1995). Research shows that the presence of a grandmother in the
household is associated with 1) mothers of children working outside the home, 2) improving
the school attendance of children, and 3) improving the reproductive health of both mothers
and children. Nevertheless, the presence of a grandmother is not always a positive sign. If the
grandmother is the father's mother, or if the father's father also lives in the same household
and particularly if the latter is the head of household, it characterizes a situation in which the
mother of the children may have little autonomy because she is subject to the authority of the
parents-in-law. This limits not only the woman's choice, but may also have negative
consequences for the health of her children. Gÿrsoy-Tezcan (1992), for example, found these
kinds of living arrangements to be one of the explanations for relatively high infant mortality
in Turkey.

277. The prevalence of co-resident domestic servants, who tend to be predominantly female,
provides additional information on gender and family relations within national contexts, as
well as on the provision of care institutions by the government. Recent ILO estimates, based
on national surveys and/or censuses of 117 countries, placed the number of domestic workers
at around 53 million. However, experts say that due to the fact that this kind of work is often
hidden and unregistered, the total number of domestic workers could be as high as 100
million. In developing countries, they comprise from 4 per cent to 12 per cent of wage
employment. Around 83 per cent of these workers are women or girls, and many are migrant
workers (ILO Convention on Domestic Workers, 2011). Because domestic servants often are
in vulnerable situations (e.g. the lack of labour inspection and legislation enforcement – see
Ramírez-Machado, 2003), data on co-resident domestic servants may be useful for policies
addressing gender inequalities in the labour market. Unfortunately, only 20-to-25 per cent of
censuses allow the identification of domestic servants; in others they are simply lumped
together with other non-family household members.

111
Country Examples 10: Co-Resident Domestic Servants in Kuwait and Colombia

In some countries, such as Kuwait, co-resident domestic servants correspond to a

considerable proportion of household members, as indicated by the study conducted by Shah
et al. (2002). Using data from a nationally representative survey on households, they found
17.3 per cent of the 14,835 individuals residing in the investigated households were domestic
servants who were unrelated to the Kuwaiti residents. The prevalence of co-resident domestic
servants was particularly high in households with elderly persons; about 90 per cent of all
households with an older adult had at least one co-resident domestic servant. These numbers,
however, are exceptionally high and owe a lot to the availability of cheap domestic servants
from abroad. By comparison, in the 2005 census of Colombia less than 1 per cent (i.e. only
0.40 per cent) of all household residents were found to be co-resident domestic servants. The
percentage was somewhat higher in larger households, with a maximum of 0.55 per cent in
households with 5 members and a minimum of 0.30 per cent in households with only 2
members.

278. Finally, analyses based on different time measurements, such as comparisons between
successive censuses, may not only indicate changes in the social structure of the households
but also changes in gender relations. Data analyses indicate, for instance, that increases in the
number of women living alone are related to increasing numbers of female university
students and female workers. This was noticed in urban areas of several, -especially-
developed countries. In line with this, changes in household size may also be associated with
increased access to sexual and reproductive services, which are recognized to be crucial for
the promotion of gender equality (UNFPA/IPEA, 2007).

Country Example 11: The Changing Nature of Household Composition and its Link to
Changes in Gender Roles in France

Ogden and Hall (2004) studied household change in France during the 1990s, using data from
the 1999 French population census and comparing them with data from earlier censuses.
Their results indicate a decline in the mean household size and a rise in the number and
proportion of people living alone. These observations may be associated, among other
factors, with changes in gender roles and sexual norms (e.g. fertility trends, feminization of
the labour force, growing student population). The tendency for women to be heads of one-
parent households goes along with the fact that men during young adulthood and early middle
age (25-49 years) are more likely to live alone. Nevertheless, the rise in the number of young
women in their twenties and thirties who live alone is also remarkable. From 1975 to 1999
the number of men aged 25-39 who were living alone rose by a factor of 1.7, while the
number of women living alone rose by a factor of 2.0. At older ages, the overwhelming
predominance of women in one-person households was found to be due to earlier male
deaths. Of the 1.7 million people living alone over the age of 75 in 1999, 81 per cent were
women. Another finding was that of the 57 million people recorded as living within
households in the 1999 census, just a little over 30 million were living as ‘traditional’ families
(i.e. as a couple with children), whereas almost 27 million did not fit that description.

3. Data issues

279. The use of census data for gender analysis has both strengths and weaknesses with
regards to household and family composition and structure. In terms of strengths, censuses
are perhaps the most complete source of information on household and family composition,

112
because they enumerate the highest number of people in a population or country. Related to
this, due to the universality of coverage, family structures can be divided into much finer
categories than would be possible with most surveys. Also, because of the same universality,
the census is less susceptible to the sampling biases that may affect surveys such as the
Demographic Health Surveys (DHS), which are designed to capture women living in unions.
Finally, another advantage is that censuses allow for international comparability of household
and family composition on a regular basis for gender issues when the internationally agreed
upon recommendations on definitions and classifications are used (e.g. the Principles and
Recommendations).

280. Censuses collect data by enumerating all the individuals that normally live in households
(i.e. de iure census) or that happen to be there at the time of enumeration (i.e. de facto
census).21 In addition, the census collects data on the physical structure of living quarters and
on the people living in institutional or collective-household arrangements, such as army
barracks, prisons, hotels or pensions or on the street. Among the weaknesses, differences in
methodologies of census taking, notably between de facto and de iure censuses, may affect
the definition of a household and who is considered to be a member of it or not. In a strictly
de facto census any person who did not spend the night of the census with the household
would be considered to be a member of another household. In de iure censuses absentee
household members continue to be considered as part of the household for up to six months.
Interviewers often have difficulties with household membership in the face of recurring
absences of persons that may stretch over several years (e.g. students). Another limitation of
census data for gender analyses is that the relationships between members are usually only
described with respect to the head of household or reference person. This limits the extent to
which all the relationships between household members can be mapped. For instance, in a
household where there are two daughters of the reference person, each in their twenties, and
one man described as “son-in-law,” it may not be possible to determine to whom this son-in-
law is married. The marital status of the daughters may give a clue, but this will not always
work, especially if the marriage is consensual.

281. To organize household data, almost all censuses require that one of the household
members be selected as the “head of household” or “reference person.” There are a number
of reasons why this is done:

a. To assign a person as primary informant, responsible for providing information on

all household members, including those who are not present at the time of the
interview;
b. To detect the relations within the household, i.e. to identify family nuclei, and
describe the household structure (one-person household, nuclear household,
extended or composite household));
c. As a key informant, to ensure that all household members are enumerated ;
d. As a source person to check information pertaining to other household members;
and
e. To “summarize” the characteristics of the household by identifying them with
those of the head of household, e.g. to analyse the living conditions of household
21
This difference in enumeration methods can have implications from the viewpoint of a gender analysis. For
example, if the census coincides with a period of the year in which many of the men are temporarily absent from
their homes because of seasonal activities (temporary harvesting labour) and if the census is conducted accord -
ing to the de facto criterion, more women than usual will be classified as living without their husbands. If the
census is a de jure census, the missing men will be counted in their households of usual residence, but they may
not be considered heads of households, even if they normally act as such.

113
 members according to the occupation or the income of the head or reference
person. This last use of the head of household concept is quite controversial,
because there is no guarantee that the characteristics of the head of household or
the reference person will adequately represent those of the household.

282. Using headship of household - female against male headed households - to analyse
gender issues is hampered by a number of theoretical shortcomings and practical problems.
Even though statistics differentiated by female or male household head can be useful for
defining and planning policies for vulnerable population segments, statisticians and gender
researchers note that this type of analysis may have its limitations, and these limitations
should be taken into account when interpreting data results and planning policies. The
following describes five problems related to the use of female/male headship to study gender
differences.

1. The first problem lies in the fact that the definition and operationalization of the
notion ‘head of household’ is vague and in no way uniform. Contrary to many other
variables that are strictly defined, the Principles and Recommendations for
Population and Housing Censuses, Rev. 2 (United Nations, 2008 a), leave the
definition and appointment of the head of household wide open: ‘Countries may use
the term they deem most appropriate to identify this person (household reference
person, head of household, householder, among others) as long as the person so
identified is used solely to determine relationships between household members. It is
recommended that each country present, in published reports, the concepts and
definitions that are used’. Appendix 1 illustrates the variety of definitions of head of
household that a number of countries used in the 2010 round of censuses. As Hedman
et al. (1996: 6) rightly state, the term ‘head of household’ “is used to cover a number
of different concepts referring to the chief economic provider, the chief decision
maker, the person designated by other members as the head”. Despite the diversity of
approaches that may be assumed, questionnaires often do not properly inform which
of the term of household head is specifically referred to. At least five different
concepts of head of household can be found in censuses: 1) main breadwinner, 2)
householder, 3) main authority, 4) reference person, and 5) questionnaire respondent.
These concepts are discussed below in points a) through e), also the relative
advantages and disadvantages of each possibility are discussed.

a) Main breadwinner. This economic approach to household headship may be useful for
gender analysis if clear definitions are presented in the census taking process and in
data tabulation and dissemination. It identifies the primary person responsible for the
economic maintenance of the household, so data could be presented on the main
contributors to the household’s economy, by sex. Further cross-tabulations could
include marital status, family composition, owning or renting house, number of
family nuclei, and other socio-economic characteristics of the head of the household
(e.g. job sector). One country that used the notion ‘main breadwinner’ was Anguilla
in the 2001 census, in which the head of the household was the major economic
provider. Using the breadwinner metric allows the analysis to avoid under-
representing women’s household contributions in practice, as may be the case if
relying on authority and traditional gender stereotypes. However, one disadvantage to
the breadwinner option is that the question about an economic head may not be
adequate for societies where spouses share economic responsibilities more or less
equally. In such cases, the Principles and Recommendations recommend the use of

114
 the term ‘reference person,’ with no implication of headship or, alternatively, that
provision be made for designation of joint headship. A second disadvantage of using
the breadwinner definition is that a gender bias may underlie this definition as it does
not capture women’s hidden household economic contributions; as female labour is
often unpaid, they may not be defined as “breadwinner” even in cases where the
market value of their labour exceeds that of their male partners or if her activities are
the main source of family livelihood. Additionally, female respondents may answer
that they have no ‘occupation’ or ‘work’ although they have been working on craft
activities at home or in agriculture. Also, using the breadwinner or economic head for
identifying economic conditions in the household may be a pitfall, because household
members may not equally share resources and may present different socioeconomic
characteristics (Hedman et al, 1996). Finally, ownership of resources does not
necessarily imply access to and use of resources (United Nations, 2008).

b) Householder. This term refers to the person who owns or pays the rent of the home,
and may be defined as the person in whose name the household dwelling is
registered, adding that “this approach is more objective than household head and may
relate in some ways to power relationships in the household” (UNECE/World Bank
2010: 16). Australia, New Zealand and American Samoa use some of the countries
that use the concept of householder. In terms of advantages, the householder option is
similar to the breadwinner criterion, but offers greater specificity. In terms of
disadvantages, the householder concept may be ambiguous in the context of
developing countries, where homes are often rented without formal contracts, where
occupants pay no rent at all, or where people live in homes that have been temporarily
borrowed or that are makeshift temporary accommodations. Also, the person paying
the rent may not always be the one earning the money with which the rent is paid.

c) Main authority. This approach identifies the person that the other household members
recognize as the main authority for all sorts of household decisions. This approach
has become quite popular, particularly because it is viewed as an indicator of
decision-making (UNECE/ World Bank, 2010). In terms of advantages, the main
authority approach can provide a measure (i.e. using sex ratios) on the main authority
in the household. This can also be used in gender analysis to identify the ways in
which power relations are featured in the household structure. In terms of
disadvantages, the main authority approach carries the assumption that a hierarchical
relationship exists between household members that may not represent reality.
Related to this, authority and decision-making typically are not formal but continually
negotiated processes. Further, in practice, the census taker may confuse the decision-
maker with breadwinner, and decision-making may differ by the type of decision
being made (UNECE/World Bank, 2010). Another limitation to the main authority
approach is that authority, responsibilities and decision-making may be shared,
especially considering the changes in gender and family relations in the last decades.
Decision-making power may be shared by a couple or even several household
members. Finally, household members may not have the same opinion on who is the
main decision-maker or who exercises the authority to run the household. Moreover,
because authority is defined as “perceived authority,” it may not be possible to come
up with an objective criterion. The Brazilian country example below illustrates how
the main authority approach can be used in practice and how some of these perceived
disadvantages can be surmounted.

115
d) Reference person. The reference person approach implies selecting an adult
household member at random. It has the advantage of making the process explicit in
the sense that it does not allow any undue interpretations of what the headship
criterion actually represents. Because it is random, it is not necessarily the main
income earner, nor the person taking the major decisions regarding the household or
the person that owns or pays the home. This concept focuses on the second function
of the household head assignment, a reference for mapping family relationships, more
than the third, a breadwinner or person in authority who can represent the household.
The 1996 census of New Zealand, for example, did not use the term ‘household head,’
but replaced it by a sentence which referred to the person responsible for the answers.
In the census form, it stated: “In every dwelling, one person must take responsibility
for this form. An adult who lives here would be best, but any person can be the one
who must: fill in this form; make sure that everyone fills in an individual form (the
blue one); and have all the forms ready for the enumerator to collect.” The 2001 and
2006 census forms also used similar expressions. In some cases, this interpretation is
made explicit by choosing the reference person randomly among the adult household
members. This practice is currently used in Denmark, the UK, Canada, and several
other countries. In terms of disadvantages, the reference person approach may
generate confusion. Further, choosing a reference person randomly increases the
chance that this person may not be related to the other household members, in which
case, information on related persons and family nuclei will be lost.

e) Questionnaire respondent. This conceptual approach means that the ‘household head’
is the person responsible for answering the census form no matter what his/her status
may be within the household. For instance, in the United Kingdom Population Census
of 2011 a ‘householder’ is defined in order to indicate who is responsible for the
completion of the questionnaire. It is not defined to produce any outputs as there is
nothing on the questionnaire to identify this person. The advantages and
disadvantages of the ‘questionnaire respondent’ are roughly the same as just
discussed for the ‘reference person’ concept, i.e. the notion may be confusing and
chances are that the person is not related to other members of the household.

This conceptual vagueness or ambiguity of the headship concept may lead to

divergences between census workers’ assumptions and the responses given by
household members. Each one will tend to define the head according to their own
cultural representations and, naturally, there is a great risk that premises differ
substantially from one another, or may not adequately reflect household realities. A
census may supposedly use a “breadwinner” definition, but household
members/enumerators may decide simply to put down the oldest male, without
questioning the situation. As enumerators, respondents and analysts may not be
talking about the same concept, and there is no guarantee of what is being measured.
Without a clear definition, would two census workers define the same head? And
would household members designate the same person as the head? Without adequate
orientation and interview training, internal/external reliability becomes quite unlikely.
Indeed, if the concept is vague or ambiguous, it is very likely that data and analysis
will not adequately represent reality or provide accurate and consistent evidence for
policy-making.

2. A second problem of using the female/male headship dichotomy in gender studies is

that gender inequality may take place at the intra-household level (e.g. unequal

116
 distribution of earnings and consumption among members of the household).
Therefore, focusing on female-headed households may not capture these inequalities
and be misleading. The shortcoming of using female/male headship of household is
well illustrated in a study on 11 Latin American countries by Deere, Alvarado and
Twyman (2010). Drawing on the recent Living Standard Measurement Studies for
Latin America and the Caribbean, they present baseline indicators of the degree of
gender inequality in individual asset ownership. Disaggregated data on housing
ownership suggests that the distribution of asset ownership by sex within households
is much more equitable than a headship analysis would suggest. The authors estimate
that in Nicaragua women own 36-41 per cent of household physical wealth. However,
if the analysis of household wealth were conducted by sex of the head, female-headed
households would own only 20 to 23 per cent of household wealth, significantly less
than the share of female-headed households in that country. This different vision of
relative female poverty is largely due to the fact that women in male-headed
households often own property, either in their own right or as joint property with their
spouses. As will be argued in more detail in the Chapter 8, the differences in poverty
rates between male and female-headed households, if not broken down into finer
categories, are typically small and tend to be associated with other demographic
differences between these households, such as the number of children and adult male
and female household members.

3. A third problem has to do with the limited possibilities for analysis and cross-country
comparisons. Obviously, if the researcher is not sure of what the ‘household head’
variable of a given census is actually measuring, this may put into question how the
results should be interpreted. Further confusion arises when census data on household
headship from different countries are compared, with a high likelihood of putting side
by side different measurements in cross-national comparisons.

Even when the concept is clear in the original questionnaire form, the definition may
not be presented in census tables or other census reports. In this case, particularly
when making use of secondary sources, analysis and comparisons must thus be
conducted with caution to ensure that assumptions made by data users correspond to
the respondents’ approach. Sometimes, misguidedly vocabulary changes also do take
place. The original questionnaire form may have used the term ‘reference person’,
while in tabulations or in analyses this was replaced by ‘household head’. This may
lead to biased interpretations. In this regard, UNDESA warns that:

‘It is important to recognize that many countries use the concept of reference person
in listing household members and that this person may or may not be the “household
head”. Where this is the practice, the “household head” identified in tabulations is, in
reality, the reference person and should be treated with caution’. (United Nations,
2005: 135)

4. Focusing on female-headed households may lead to biased policy priorities. In the

words of Sylvia Chant (2003: 30), “Placing excessive emphasis on the economic
disadvantage of female heads misrepresents and devalues their enormous efforts to
overcome gender obstacles.” The ‘feminization of poverty’ thesis “...precludes an
analytical consideration of the social dimensions of gender and poverty...” and “tends
to translate into single-issue, single-group policy interventions.” These narrow policy
interventions may in turn fail to affect and reshape the embedded structures of gender

117
 inequality found in the home, the labour market and other institutions.

5. A fifth consequence of the methodological bias is the reproduction or reinforcement

of gender stereotypes. Stereotypes may not only be guiding questionnaire formulation
and census execution, but also respondents’ answers, and the researcher’s analysis.

Participants of the Second Global Forum on Gender Statistics expressed concern that
“using the conventional classification of household headship (i.e. whether household
is female- or male-headed) implies a kind of hierarchy within the household that
suggests subordination” (United Nations, 2009 b: 17). Questionnaires may also be
designed under the assumption that gender norms naturally place men as the ‘head;’ in
this case, it is quite possible that questions are formulated in a way that leaves little
room for respondents to indicate a female-headed household. Even if the
questionnaire was well formulated, the training of enumerators may not adequately
address this topic, allowing for biased interview approaches.

As Budlender (1997: 4) points out, in many African cultures “the household head will
be the oldest male,” adding that “in a multi-generational household, he will often not
be the highest income-earner, or even control the resources of the household.” He is
recognized as the ‘head’ in terms of respect, but it would be inappropriate to say that
he is the economic head.

The First World Conference on Women (1975) had already pointed out, in its Plan of
Action, the risk of methodological bias relating the operationalization of the concept
of household head: “among other data biased by preconceptions are those on heads of
households or families, when it is assumed that a woman can be the head only in the
absence of a man” (Par. 164). These concerns were further incorporated in census
manuals. The Principles and Recommendations (Pars. 2.118, 2.119), for instance, state
that “the most common assumption that can distort the facts is that no woman can be
the head of any household that also contains an adult male. Enumerators and even
respondents may simply take such an assumption for granted”. It is also argued that
this gender stereotype “often reflects circumstances that may have been true in the
past but are true no longer, insofar as the household and economic roles of women are
changing.” The Principles and Recommendations recommend, in their paragraph
2.119, that “It is important that clear instructions be provided as to who is to be
treated as the head of the household so as to avoid the complications of enumerator or
respondent preconceptions on the subject. The procedure to follow in identifying a
head when the members of the household are unable to do so should be clear and
unambiguous and should avoid sex-based bias.”

Nepal provides an example of a country which is currently making efforts to avoid

gender bias regarding household headship. In the 2001 census questionnaire,
illustrative examples in the census instruction manual were revised in order to move
away from the traditional type of responses: “descriptive examples like ‘female or
male’ in place of ‘male or female’ in earlier census manuals or the example of ‘female
name’ as the household head in place of only ‘male name’ in earlier census manuals
were included”. Additionally, women were shown as household heads in specific
publicity materials (posters), which also presented women performing various non-
marketed economic activities, such as “food processing, agricultural cropping,
fetching water and fuel wood collection” (UNFPA, 2004: 17).

118
283. An illustrative example of divergent data outcomes regarding different definitions of
household headship is found in a study based on the 1997 LSMS (Living Standard
Measurement Study) data for Panama (Fuwa, 2000). Here, female-headed households were
classified according to three criteria: self-reporting of a woman as the head; a “potential”
female head, if no working-age male was present; and a “working head” definition, in cases
where more than half of the total hours worked were contributed by a single female member.
With respect to poverty rates, “the study showed that the overlap between these three sets of
households was low, around 40-60 per cent”. Also, “the corresponding poverty rates were
different: 29 per cent for the self-declared female-headed households; 23 per cent for the
“potential” female-headed households; and 21 per cent for the households headed by a
“working female” (United Nations, 2010 a: 164). In this case, different definitions in fact
measured different social situations.

Country example 13: Brazilian Census Using Decision-maker to Designate the

Household Head

The 2010 Brazilian census included the possibility of more than one household head
(“one person”/“more than one person”) which could be considered as shared
responsibilities in the household – but without using the term ‘household head’, as this
term has been considered increasingly inappropriate. In the early 90’s, surveys
conducted by the NSO (IBGE) started using the term “reference person”; later, after pilot
tests, it was replaced with the term “responsible person”, which was widely accepted by
respondents. The 2000 and 2010 censuses incorporated this change but didn’t establish a
criterion to the selection of household “responsible person”. In the 2010 census of Brazil,
the “responsible person” is chosen among the household residents. Preliminary results of
2010 Census show that 70.4% of the households have only one “responsible person”.
The questionnaire of the 2011 census of South Africa provided a definition of the
household head as “the person who is the main decision-maker in the household”. It
considered the possibility of joint responsibility, in which case it instructed census
interviewers to take the oldest person.

284. Despite these limitations, the head of household measure has been used frequently in
gender studies to document the erosion of male exclusive providers, and has given new
meaning to this aspect of domestic living. An example of this ambivalence is illustrated in the
World’s Women 2010 (United Nations, 2010 a), which cautions that generalizations between
“female-headed households” and “male-headed households” are impossible because of the
contextual differences in women’s and men’s statuses as well as the variety of headship
definitions and the households that may be included under these labels. Therefore,
researchers must proceed cautiously to compare across households headed by women and
men, in order to understand patterned poverty and vulnerability in different types of
household structures. This caution often translates into additional precision in the analysis
and household specification, as the World’s Women 2010 then reports that Latin American
households headed by lone mothers with children have higher poverty rates than those of lone
fathers with children, and that poverty rates for women living in one-person households are
higher than the corresponding rates for men.

Methodology Box 6: Mixing Different Levels of Analysis

119
Like many social relationships, gender issues can be analysed at different levels: individual,
household, community, geographical, and cross-country. While each of these levels of
analysis may have their own legitimacy, the interpretation of the results will differ. A
relationship between households or geographical units does not always translate directly into
a relationship between individuals. When the units are geographical, this problem is known as
the ecological fallacy. A typical finding is that in US elections districts with a higher
proportion of Afro-Americans in their population often vote more strongly for white
supremacist candidates (e.g. George Wallace, in 1968). Obviously, this does not mean that
Afro-Americans are likely to vote for white supremacist candidates, but rather that race
relations are more conflictive in districts that have a high percentage of Afro-American
voters, so that white voters in these districts are more inclined to vote for these kinds of
candidates. Something similar may happen at the household level. The greater poverty or
vulnerability of households with female heads or high proportions of female members may
not be directly related to the characteristics of these women, but reflect that these are special
kinds of households where some of the male members are either incapacitated or absent. This
may still be a worthwhile phenomenon to investigate, as long as it is kept in mind that the
results characterize households, rather than individuals.

285. Because of the problems involved, the household head concept should be clarified, and
perhaps it should be replaced by several concepts to provide a better understanding of
household structure and processes. Statisticians and scholars alike have made the case to
clarify the concept of head of household. Although most censuses seek to define one concept
of household head, it would actually help the analysis if the different concepts were treated
separately. As Budlender (1997: 15) states:

Instead of insisting on one definition, one could several questions so as to get at multiple
definitions of household and multiple axes of household headship, and then leave it to
users/analysts of the data to pick up the particular definitions they find useful in
addressing a specific question. Multiple questions would also allow for cross-tabulation
and comparison of the definitions, to see where they differ, why they differ and by how
much. These questions are not only of academic interest. They are also of interest to
policy-makers who are eager to target their programmes where they will be most
effective, and at those most in need, while avoiding the allocation of scarce resources to
less ‘appropriate’ beneficiaries’.

Failure to clarify the concept of head of households has a number of potentially undesirable
consequences for the presentation and interpretation of results – specifically in the areas of 1)
what is being measured; 2) problems for analysis and cross-country comparisons; and 3)
gender stereotypes.

286. Some countries, like Ireland and Aruba, have replaced the simple relationship with the
household head by a more complex matrix of relationships between household members, but
this can be time-consuming to fill out, even though the matrix is limited only to relationships
with the first four household members, as follows:

Relationship of person 5 (or 6, 7, etc.) to persons 1 2 3 4

120
 1. Husband or wife _ _ _ _
2. Partner (including same-sex partner) _ _ _ _
3. Son or daughter _ _ _ _
4. Step-child _ _ _ _
5. Brother or sister _ _ _ _
6. Mother or father _ _ _ _
7. Grandparent _ _ _ _
8. Step-mother or father _ _ _ _
9. Son or daughter-in-law _ _ _ _
10. Grandchild _ _ _ _
11. Other related _ _ _ _
12. Unrelated (including foster child) _ _ _ _

287. Aruba came up with an alternative approach to classify family nuclei and household
types in its 2010 population census. Through a set of questions it was possible to identify the
categories and sub-categories of household type and family type. Moreover, the questions
allowed making a household classification using both a formal approach (married couples) as
well as a sociological approach (married + consensual unions). The questions used to make
this classification were:

 Is the person related (also by marriage) to everyone in this household?

 Does the father of this person live in this household? If yes what is the person number
of the father?
 Does the mother of this person live in this household? If yes what is the person
number of the mother?
 What is the marital status of this person?
 Is this person currently living on a durable basis with a partner (married or not)? If
yes, what is the person number of this person?
 If living together, is person married to this partner?

This approach worked well, especially with digital questionnaires, where the person number
of related people could be chosen from an answer box containing the names of all eligible
candidates. However, Aruba is a small country and it is doubtful whether the same technique
could be used on a larger scale in censuses in much bigger countries.

288. The lack of detail, and inconsistency between countries among the relationship with
head categories in the census presents another limitation. International recommendations
support the collection of detailed information on relationships, with a sufficient number of
pre-coded categories. Nevertheless, in some countries the number of options is extremely
limited (e.g. only 5 relationship categories in Malawi and Bangladesh). As was mentioned in
Chapter 3, some censuses ask about the father and mother of each person residing in the
household, i.e. a) If the father/mother of the person reside in the same household (e.g. Aruba,
Barbados, Cape Verde, South Africa); and b) If so, the identification of the person's
father/mother. This information can be quite helpful to analyse residential patterns, e.g. by
providing an indicator of the likely distribution of unpaid care work, because in many
countries, more children live with their mothers than with their fathers. The 2006 census of
the Maldives went even further and asked for each child under the age of 18 whether he/she
lived with:

1. Both parents;

121
 2. Mother and stepfather;
3. Father and stepmother;
4. Mother only;
5. Father only;
6. Other relatives;
7. Only unrelated household members.

Similarly, for people over age 65, it asked whether they were living with:

1. Children;
2. Spouse;
3. Stepchildren;
4. Other relatives;
5. Unrelated individuals;
6. Without guardian.

4. Tabulations

289. The Principles and Recommendations consider household and family characteristics as
an essential topic to be investigated and suggest that NSOs should construct the following
tabulations relevant for gender analysis:

Recommended tabulations for population censuses:

1. Population in households, by relationship to head or other reference member of

household, marital status and sex, and size of institutional population (i.e. persons
who are not members of households)
2. Head or other reference member of household, by age and sex; and other household
members, by age, sex and relationship to head or other reference member

Additional tabulations for population censuses:

1. Population in households, by household status (or type of household), age and sex,
and institutional population by age and sex
2. Household population under 18 years of age, by age and sex and by whether living
with both parents, mother alone, father alone, or neither parent
3. Households and population in households, by sex, by size and type of household and
number of persons 60 years of age and over

290. To illustrate (part of) the first kind of table, one may take the 2004 census of Timor
Leste as an example, as displayed in Table 22. Among other things, this table shows that a
surprisingly high percentage of female heads of households (38.1 per cent) are actually
married; only the number of widowed female heads of household (48.5 per cent) is larger.
Most of the remaining female heads of household are single, with very few cases of divorce
or separation. Note also that couples are about as likely to live with the parents of the woman
(1,385) as with the parents of the man (1,410). Surprisingly, in almost half of the former
(602) cases, the couple is not formally married and in 113 cases the man continues to live
with his parents-in-law even after the wife has died. The latter construction is even more
common in the case of widowed daughters-in-law (379). Mothers live with their children
much more often (7,008 cases) than fathers (1,994).

122
Table 22: Timor Leste - Population in private households by marital status according to
sex and relationship to the head of household

Never
Males Total Married Widowed Divorced Separated
Married
Head 158,063 10,107 139,371 7,822  416 347
Wife/Husband 4,590 0 4,590 0  0 0
Daughter/Son 237,881 236,731 722 339  52 37
Stepchild/Adopted Child 12,024 11,870 106 28 16 4
Daughter/Son-in-Law 1,385 602 657 113 5 8
Mother/Father 1,994 136 738 1,092 9 19
Sister/Brother 18,628 16,798 1,165 524 80 61
Grandchild 10,109 10,018 70 11 7 3
Grandparent 482 65 116 285 7 9
Other Relative 20,009 17,676 1,652 579 53 49
Non-Relative 1,798 1,331 434 21 8 4
Never
Females Total Married Widowed Divorced Separated
Married
Head 36,899 3,317 14,054 17,882  843 803
Wife/Husband 134,207 0 134,207 0  0 0
Daughter/Son 218,046 215,168 1,610 760  282 226
Stepchild/Adopted Child 10,323 10,051 124 95 340 19
Daughter/Son-in-Law 1,410 405 610 379 70 9
Mother/Father 7,008 327 1,670 4,890 55 66
Sister/Brother 15,628 12,479 1,444 1,426 159 120
Grandchild 9,119 8,957 103 35 17 7
Grandparent 1,343 87 292 946 8 10
Other Relative 16,459 12,839 1,723 1,689 115 93
Non-Relative 1,053 858 124 51 15 5

Source: Population and Housing Census of Timor Leste (2004): Table 3.1

291. The following is an illustration of another kind of table that can be extracted from a
census (in this case, the 2008 census of Cambodia) by combining information about the sex
and age of the head of household with information on the relationship of the household
members with the head of household.

Table 23: Cambodia 2008 - Average numbers of household members classified by age
and sex of the head of household and by relationship to the head of household (all
numbers multiplied by 100)

Other
Other
Male Grand Non
Head Spouse Child Parent Relativ Total
Head Child Related
e
.
0-14 100 2 1 6   101 31 240
15-19 100 54 35 6   53 42 290
20-24 100 90 104 6   25 20 344
25-29 100 94 167 8 0 24 13 406
30-34 100 96 247 9 0 24 9 485
35-39 100 96 298 9 1 24 9 537

123
 40-44 100 96 328 8 6 24 7 568
45-49 100 96 324 6 15 27 8 575
50-54 100 94 295 5 28 30 8 559
55-59 100 93 257 4 41 29 7 530
60-64 100 90 208 2 57 28 7 492
65-69 100 88 169 2 70 27 8 463
70-74 100 84 130 1 80 26 10 431
75-79 100 80 110 1 81 23 15 409
80+ 100 71 98 1 74 19 40 404
Total 100 93 247 7 15 26 10 498
                 
Other Other
Female Grand
Head Spouse Child Parent Relativ Non Total
Head Child
e Related
0-14 100 1 0 10   114 18 243
15-19 100 20 28 9   81 80 319
20-24 100 37 94 9   52 46 338
25-29 100 40 160 12 0 44 24 380
30-34 100 37 225 15 0 31 11 419
35-39 100 32 250 14 2 30 10 439
40-44 100 25 250 13 11 32 5 437
45-49 100 20 229 10 28 36 5 427
50-54 100 14 195 8 49 39 5 410
55-59 100 10 159 5 71 38 4 387
60-64 100 8 129 3 92 34 4 369
65-69 100 6 106 1 105 28 3 350
70-74 100 5 88 1 106 23 2 325
75-79 100 3 79 0 98 19 3 303
80+ 100 3 73 1 88 19 4 287
Total 100 21 179 9 38 36 11 394

292. Some of the conclusions that can be drawn from this table – at least for the case of
Cambodia – are the following:

 Female-headed households with young heads have about the same size as male-
headed households, but as the age of the head increases, female-headed households
become progressively smaller with respect to male-headed households.
 Roughly the same pattern is observed with respect to the number of children, although
at higher ages there is a slight recovery of the relative number of children among
female-headed households.
 The highest percentage of female heads of household with spouses (35-40 per cent) is
found among those aged 20-34; after age 35, the percentage declines quickly.
 Female-headed households have a slightly higher tendency to have parents, other
relatives and especially grandchildren living with them.
 Especially among households with older heads, the percentage of household members
in female-headed households that are children or grandchildren tends to be higher
than among male-headed households.

293. As was mentioned before, much use has been made of the criterion of whether the head
of household is male or female. Some of the limitations of this distinction were discussed
earlier in this sub-chapter. Tabulating census data on the basis of male/female headship only

124
gives a first impression of the social and economic position of women, but should be
complemented by the comparison of the structure of households. Sometimes the percentage
of women in the household is used as a criterion. For example, The World’s Women 2010
(United Nations, 2010 a: 159) mentions that “households with an overrepresentation of
women might be more likely to be found below the poverty line”, but immediately points out
that this may be due to two rather different causes, namely: 1) in some types of households
where the share of women is higher, the earnings per capita tend to be lower because
women’s participation in the labour market and their earnings are lower than men’s; and 2) as
the ratio of women to men increases with age, the presence of non-earning older persons in
extended households may depress per capita household income. While the first explanation
points to a genuine gender disparity, the second is a demographic composition factor that one
would like to control for, rather than confounding it with the first.

294. When using household information for gender analysis, special attention should be
dedicated to the situation of elderly women compared to elderly men. The situation of elderly
household members varies considerably among countries. In Brazil, for example, the
presence of an elderly person with a retirement pension or allowance was found to reduce the
likelihood of family vulnerability to poverty, and having an elderly person in the family was
even strategically more effective in reducing vulnerability than having a spouse (Lavinas and
Nicoll, 2007). The structure of gender imbalances at older ages may be different from the
typical pattern at younger ages and care must be taken not to assume automatically that all of
them will favour men over women. As Knodel and Ofstedal (2003: 693) point out,

In many settings, perhaps even in most, older women may be disadvantaged relative to
older men on some or most dimensions of well-being. Clearly, too, there are numerous
exceptions as indicated by the examples presented above. A major hindrance to making a
more definitive statement about gender and aging is that systematic empirical
assessments comparing the situation of older men and women are inadequate for drawing
a firm conclusion-especially for the developing world, where the large majority of older
persons live. Most broad statements asserting a generalized female disadvantage in old
age appear to be based on presumptions and an incomplete allowance for the full set of
influences over the life course, including later stages, that determine well-being in old
age. In any event, generalizations about which sex is more disadvantaged are of limited
value. To more fully understand the effects of gender on the well-being and support of
older persons and their families, research must move beyond assumptions of universal
gender inequality and the disadvantaged situation of older women to examine the
experiences of older men and women within the contexts in which they live. Such research
should recognize that well-being at older ages is multidimensional and that gender
differences may go in either direction or, for that matter, be largely absent, depending on
the aspect of well-being under consideration.

295. Migration is another significant factor that has to be controlled, for any comparisons to
be meaningful. Klasen, Lechtenfeld and Povel (2010), found, for example, that in Thailand
and Viet Nam households with female headship that had a male family member residing
elsewhere were, on average, better off than the general population, whereas female-headed
households without such external remittances were poorer and more vulnerable than average.

---------------------------------------------------------
Methodology Box 7: Kinship classification

125
One way to organize a more detailed classification is to compare across different household
structures, first organizing across those with and without other adults in the household, then
by male or female head, and finally by meaningful comparisons across different family
nuclei. See the classification structure just below, illustrated with data from Cambodia (2008)
and El Salvador (2007).

Table 24: Household composition by headship for Cambodia (2008) and El Salvador
(2007)

Cambodia (2008) Without Other Adults With Other Adults

Male head Female head Male head Female head
Head without spouse or children 30,274 68,377 52,970 174,078
Couple without children 121,031 10,135 256,785 19,225
Couple with 1-2 children under 15 485,038 38,463 568,448 45,617
Couple with 3+ children under 15 246,319 18,834 288,206 22,632
Lone parent with 1-2 children under 15 12,286 81,563 32,561 173,868
Lone parent with 3+ children under 15 2,835 25,275 9,601 49,643
Other structure or unknown 5,204 2,629

El Salvador (2007) Without Other Adults With Other Adults

Male head Female head Male head Female head
Head without spouse or children 69,568 52,601 38,872 108,311
Couple without children 85,439 9,687 112,138 14,037
Couple with 1-2 children under 15 214,715 16,282 178,165 22,952
Couple with 3+ children under 15 88,305 7,301 83,394 10,522
Lone parent with 1-2 children under 15 6,358 54,673 26,867 123,728
Lone parent with 3+ children under 15 1,578 22,474 9,479 47,289
Other structure or unknown 1,222 528

Source: Computed based on REDATAM data base, ECLAC/CELADE

296. The proposed division by number of children is merely a suggestion. Depending on the
level of fertility in a country, it may be more appropriate to divide parents by whether they
have 0, 1 to 3, or 4 children or more. In other cases, a mere classification in terms of whether
the family nucleus does or does not have children may be sufficient. Note that lone female
heads of households with children and no other adults in both of the tables above account for
only about 15 per cent of all female-headed households. About double that percentage is
made up by female heads of household without spouse, but with children and other adult
household members. However, about half or slightly over half of all female-headed
households consist of women living alone, with a spouse or a spouse and children, or with
other adults and no children under age 15. In particular, note the large number of female
household heads living without spouse or children under age 15, but with other adults. In El
Salvador these make up about 22 per cent of the female-headed households and in Cambodia
almost 24 per cent.

297. In Cambodia, about 40 per cent of these households have people over 60 living in them,
compared to the average of 23.5 per cent for all households (not shown in Table 24). These
can consist of older women without spouses and with adult children, or younger women
without spouse or children caring for elderly parents. Overall, there are 33,724 households
where one woman between the ages of 15 and 50 lives together with older adults, without any

126
children under age 15 or other adults. This is about 5 per cent of all households that have
people over age 60 living in them. The number of households where one man between the
same ages lives with adults over the age of 60, without children under age 15 or other adults,
is smaller, namely 23,274. This suggests some tendency for the care of older persons to fall
disproportionally on women, although the absolute size of the numbers and the difference
between them is not as large as one might expect. In countries where this is an issue,
especially in Eastern and Southeast Asia, it is recommended to produce tables that elaborate
on these kinds of household compositions.

298. The age of the heads of household and possibly their marital status may be taken into
consideration, as well as the existence of household members living abroad, which may
indicate that the household is receiving remittances. One may also wish to further subdivide
the households with “Other adults”, to allow the inclusion of structures that may be of
particular interest, such as those that include the parents of the head of household or the
spouse. Households headed by grandmothers that care for their grandchildren are a group of
growing importance, not only in Sub-Saharan Africa (due to the impact of AIDS), but even in
the United States, where they now comprise more than one fourth of all female-headed
households with children (US Bureau of the Census, 2003). The “Other structures” in the
above tables include households of several unrelated individuals living together and
grandparents or aunts/uncles with grandchildren or nephews/nieces, without the parents.

299. Obviously, if all the relevant distinctions are made, the resulting table will end up being
quite complex. The actual decision on how detailed the table should be will require some
compromise between comprehensiveness and relevance of the possible subdivisions. For
some purposes, it may be sufficient to disaggregate by broad age categories (e.g. less than 25,
25-49, 50-64, 65+) of the head of household or to omit the age disaggregation altogether. In
countries with little international migration, the distinction between households that do or do
not have members residing abroad may not be important. In other countries, where extended
households are rare, it may not pay off to go into much detail about the identity of the "other
adults" co-residing with the basic family nucleus.
_____________________________________________

300. Vanuatu (2011) provides an example of a country that did extensive tabulations of
household composition by headship. Among others, it also disaggregated the data by rural-
urban residence and by main source of income. The classification of household structures is
different from the ones used above. Next to the number of male and female-headed
households a detailed classification of household type is made (see Table 25). For nuclear
households consisting of mothers and children and fathers and children, a sub-division was
made according to the number of children under 15 belonging to the household (0 children, 1
– 3 children and 4 or more children). Also, the mean number of persons in the household per
household type and sex of the head is given. Finally, the sex ratio (i.e. the number of
males/females*100) for the heads of households of each household type is presented. For
nuclear, single parent households, the sex ratios refer to the corresponding class: e.g. (the
number of fathers with 0 children under 15)/(the number of mothers with 0 children under
15) *100.

Table 25: Vanuatu (2009) - Mean number of persons by type of household and sex of

127
head

No. of persons in house-

  Sex of head of household Sex
hold, by sex of head
ratio
  Male Female Total Male Female Total
One person household 1,887 1,110 2,997 1.0 1.0 1.0 170.0
Nuclear, Couple, no children 2,563 191 2,754 2.0 2.0 2.0 1341.9
Nuclear, Couple with children 18,065 1,259 19,324 4.8 4.6 4.8 1434.9
Nuclear, mother without children under 15 0 610 610 . 2.5 2.5 47.9
Nuclear, mother with 1-3 children under 15 0 1,735 1,735 . 3.5 5.8 29.6
Nuclear, mother with >3 children under 15 0 298 298 . 5.8 5.6 11.1
Nuclear, father without children under 15 292 0 292 2.5   2.5 47.9
Nuclear, father with 1-3 children under 15 513 0 513 3.3 . 3.3 29.6
Nuclear, father with >3 children under 15 33 0 33 5.8 . 5.8 11.1
Extended family 13,615 4,587 18,202 6.4 5.3 6.1 296.8
Composite household 434 77 511 7.0 4.5 6.6 563.6
Cannot determine 40 64 104 6.5 6.7 6.6 62.5
Total 37,442 9,931 47,373 5.0 4.2 4.9 377.0

Source: Population and Housing Census of Vanuatu (2009)

301. The sex ratios of head of household show that in nuclear families where a couple is
present, males are 13 to 14 times more likely to be chosen as head of households than
women. Because women are more likely than men to be a lone parent, sex ratios are very low.
Note that there are almost ten times more women with more than 3 children under 15 than
men (sex ratio = 11.1). Also among extended and composite households the sex ratios show
that men or much more likely than women to be chosen as head. However, this chance is
much lower than among nuclear household containing a couple. While the number of persons
in household headed by males and females is more or less the same among nuclear
households, extended and composite household headed by women are smaller.

302. The information on main source of income of Vanuatu households (not shown here)
suggests that male-headed households are somewhat more likely to depend on wages or
salaries (81.5 per cent in urban areas and 18.8 per cent in rural areas) than female-headed
households (79.2 and 16.0 per cent, respectively). This is particularly the case for lone heads
of households with children and extended households. In the case of couples in nuclear
households with female heads are actually slightly more dependent on wages or salaries.
Households that depend predominantly on remittances are relatively few (1-3 per cent), but
increase to 5-8 per cent in the case of a few household categories, such as rural grandfathers
or grandmothers caring for grandchildren, lone male household heads in rural areas living
with parents or in-laws, and women living by themselves.

303. Although a more in-depth investigation of the household determinants of female labour
force participation and the school attendance of children requires multivariate methods, there
are some basic tabulations that one may produce in order to get at least an idea on how these
characteristics vary between different types of households. These tabulations might include
the following:

 School attendance of young girls and boys (e.g. ages 6-11) by

128
 a) Presence of a grandmother (i.e. the head's mother or mother-in-law) in the
household;
b) Presence of a live in domestic servant;
c) Presence of one or both parents (father and mother, father only, mother only,
neither);
d) Number of older- younger siblings in the household.

 Women’s labour force activity status by...

a) Household size;
b) Household composition (see next section);
c) Presence of elderly household members, particularly grandmothers (of the
children in the household);
d) Presence of a live in domestic servant.

Figure 8: Swaziland (2007) – Percentage of female-headed households by tinkhundla

(from lightest to darkest, the colour gradations indicate 6-11, 12-18 and 19-34 per cent)

Source: Map 2.0, Census Atlas of the 2007 Population and Housing Census of Swaziland

5. Indicators

304. There are a number of conventional indicators of household composition that may be
relevant for different kinds of gender analysis. Several of these go back to the household
classification schedule introduced in the previous section. Rather than making comparisons in
terms of male and female-headed households in the aggregate, which mix all of these distinct
categories, it is much better to make comparisons between equivalent household composition
categories. The comparison can be made both in terms of how much more common the
female-headed households are represented in the relative incidence of poverty or
vulnerability. Most census reports, however, still limit themselves to the conventional
division of households into male-headed and female-headed, despite the known limitations of

129
these categories. Figure 8 below illustrates the kind of maps that can be constructed with this
information, based on the 2007 census of Swaziland.

305. Another household composition indicator, with more conventional applications, is the
so-called headship ratio, i.e. the percentage of men or women of a given age that are heads of
their households. It is useful primarily for household projections, as the population by age
and sex at any given time, multiplied by the respective projected headship ratios, yields the
number of households. Obviously, it also allows the comparison of the percentage of men and
women of any particular age that are heads of their households. Again, as discussed earlier,
the use of the sex of the head of household has some serious limitations for analytical
purposes. Results obtained by comparing male/female heads of household should be backed
up by more detailed analysis in which type of household is brought in as an additional
explanatory variable.

306. The concept of headship ratio can be extended to other categories of household status,
allowing the construction of descriptive indicators such as the percentage of women of a
certain age who may be the following:

a) Living in their parental household;

b) Living by themselves in a one person household;
c) Living with a husband or partner, without children;
d) Living with a husband or partner, with children;
e) Living without a husband or partner, but with children; or
f) Living with unrelated household members, as a domestic servant.

Comparing the evolution of these indicators between successive censuses can provide
interesting information about changes in gender relations, e.g. an increase in the percentage
of women aged 20-24 who are living alone or with a husband or partner, without children,
which might indicate that young women are having greater opportunities to extend their
education or to work, prior to forming families, than in the past. Obviously, these
comparisons only make sense if the concept of head of household, and the operationalization
of this concept (e.g. the response category, or who decides to designate the head) have not
changed over time. Many countries have adapted their strategies to appoint the head of
household.

307. In section 2, co-residing domestic servants were mentioned as a particularly vulnerable

group. This suggests compiling data on the following indicators:

a) Percentage of households (possibly divided by type, such as one-person, couple

without children, or couple with children –preferably by broad age category) that have co-
residing domestic servants, by sex of the domestic servant; and
b) Percentage of working women that make their living as domestic servants (co-residing
and not).

308. In recent decades many countries have seen a steady growth in the number of one-
person households. Especially in Europe the rate of one-person households increased rapidly
between 1970 and 1990. Sweden had the highest rate of one-person households (39.6 per
cent) in 1990. The growth of one-person households has some important consequences for
policy making. Research showed that for instance, energy consumption per person is higher
for one-person households than for households with more persons (Hanssen; Scherg and

130
Christensen, 2009). Living costs (per person) for one-person households are generally higher
than for multi-person households. Also, a single person may be more vulnerable, as there is
no back-up in case of unemployment or other mishaps. The ‘Worlds Women 2012 (United
Nations, 2010 a) showed that women run a higher risk than men when living in a one-person
household. Therefore, ample attention should be given to this category when studying gender
issues in poverty research.

309. An interesting perspective on household composition is that of the children. Here one
may compute the percentage of children under age 15 (or another relevant age limit) that live:

1. With both parents and no other adult household members;

2. With the mother only;
3. With the father only;
4. Without their parents, cared for by grandparent(s) or other family members;
5. With both parents and a grandmother and/or domestic servant; and
6. With the mother and a grandmother and/or domestic servant.

310. Another, potential source of useful information has to do with the physical
characteristics of the housing unit, in addition to the characteristics of the household as such.
These may provide useful information on the level of comfort and housekeeping chores. For
example, the proportion of girls/women living in a dwelling without easy access to water or
which uses firewood for cooking is a good indicator of the burden they assume to fetch water
and cooking fuel for the household, which may have adverse consequences for girls’ school
attendance and women’s labour market participation. The proportion of women living in
dwellings without a proper kitchen, especially if cooking is done using firewood or charcoal,
is an indicator of the health conditions to which women are exposed in preparing food. The
size of the household may be used as a further qualifier of the amount of work involved in
these household chores.

311. The following indicators from the Minimum Gender Indicator Set approved by the UN
Statistical Commission in February of 2012 can whenever the data are available be computed
from census data:
1. Proportion of population using the internet, by sex (asked, for example, in the 2010
census of Qatar, in addition to computer use);
2. Proportion of population using mobile/cellular telephones, by sex; and
3. Access to mass media and information and communications technology.

6. Multivariate and further gender analyses

312. Household composition, in combination with other explanatory variables, can be an

important explanatory variable for various types of multivariate analyses. As was already
indicated in other parts of this document, the concept of household headship, as implemented
in most censuses, is riddled with ambiguities. It is possible, however, to gain a better
understanding of what factors determine household headship, by decomposing conventional
headship rates by age and sex (particularly female headship rates by age) in terms of some
potential determining factors. Handa (1996) used these methods to look at the determinants of
female household headship in Jamaica. Joshi (2004) used Matlab data to analyse household
headship of widows and married women in rural Bangladesh; her results cannot be easily
replicated with census data.

131
313. In fact, this kind of analysis involves answering two different questions simultaneously,
namely why the composition of the household is as it is and why, given this composition, a
given person has been selected as head. Therefore, it is best to do this kind of analysis by type
of household. In order to explain the incidence of female headship in one person households,
for example, the question is really why this particular person is living alone, rather than with
others, and the relevant characteristics those of the household head him/herself and those of
the geographical setting (e.g. urban/rural residence). In households with more than one adult
member, on the other hand, the characteristics of the other household members may be at
least as important as those of the household head him/herself.

314. One way to carry out the analysis is as a logistic regression, in which the dependent
variable is female headship and the independent variables are household characteristics such
as the number of adult male and female household members, the number of children, the
number of economically active male and female household members, per capita household
income, etc. A more refined, but more complex procedure is to use a multinomial regression
model with explanatory variables such as the age, sex, marital status, economic activity, level
of education and (if available) personal income of each adult household member. The
outcome of such a model is a headship probability for each individual (adult) household
member, rather than just the probability that the head of household will be female.

315. In addition to its greater complexity, one of the problems of the multinomial approach is
that not all of the relevant variables can be easily determined in many censuses. For example,
one of the likely determinants of household headship is how many children a given candidate
for headship has living in the household. But given the way family relationships are
determined in most censuses, this can be hard to establish. It is easier in the case of women
because women over age 15 or between the ages of 15 and 50 are usually asked for their
number of surviving children. Some censuses, like the one of Vanuatu (2009), even ask
whether each person's mother lives in the household. But in the case of fathers, such
questions are less common and only the children of the actual head of household can be
easily determined.

316. To avoid such problems, the following analysis of headship in Vanuatu takes a more
limited approach, by looking only at nuclear families with children and trying to determine
which characteristics of both of the partners determine whether the choice fell upon the
woman, rather than the man. Nuclear households without children were not considered,
because the intention was to look at the effect of the number of children (both sons and
daughters) on the probability that the mother would be chosen as head of household.

317. Instead of separate regression equations for males and females, based on household
records, with the characteristics of both partners. The sex of the head was used as dependent
variable. Only explanatory variables which describe the combined characteristics of both
spouses were included. The results of this analysis are shown below, in Table 26.

Table 26: Vanuatu (2009) - Logistic regression to predict the choice of a female head of
household in nuclear households with children, depending on the characteristics of the
couple

132
     Male
Variable Category B exp(B)
Education male versus female Partners both less than primary . .
  Primary, partner less than primary 0.907 2.477
  More than primary, partner less than primary 0.921 2.513
  Primary, partner less than primary 0.660 1.934
  Both partners primary 0.594 1.812
  Primary, partner more than primary 0.548 1.730
  More than primary, partner less than primary 0.688 1.989
  More than primary, partner primary 1.021 2.776
  Both partners more than primary 0.800 2.226
Worked male versus female Male worked, female worked . .
  Male worked, female did not work -0.043 0.958
  Male did not work, female worked 0.317 1.374
  Male did not work, female did not work -0.075 0.928
Urban/Rural Urban . .
  Rural -1.091 0.336
Age difference (Husband - wife)   -0.007 0.993
Number of sons in household   0.004 1.004
Number of daughters in household   -0.062 0.940
Citizenship Male versus Female Both male and female Vanuatu by birth . .
  Male, Vanuatu by birth, female by naturalization -0.330 0.719
  Male Vanuatu by birth, Female foreign 0.486 1.625
  Male Vanuatu by naturalization female by birth 1.103 3.012
  Both male and female Vanuatu by naturalization -1.214 0.297
  Male Vanuatu by naturalization, female foreign -18.791 0.000
  Male foreign, female Vanuatu by birth 0.538 1.712
  Male foreign, female Vanuatu by naturalization 0.344 1.410
  Both male and female foreign 0.007 1.007
Constant   -3.151 0.043

Source: Population and Housing Census of Vanuatu (2009)

318. An even simpler way to analyse the same data is a logistic model in which the
characteristics of both sexes are analysed separately. Although the previous analysis is
preferable from a theoretical viewpoint, the results are actually quite similar. In this second
approach, a logistic regression model was used in which the dependent variable was ‘0’ if the
husband was chosen as head of the household and ‘1’ if the wife was selected. In Vanuatu,
only 6.5 per cent of nuclear households with children had a female head. The regression was
run separately for males and females. This was done to see if the impact of each of the
explanatory variables was different for males and females. A number of models were tested,
next to characteristics of the respondents; also some predictors of both spouses combined
were included. For instance, an explanatory variable was created which incorporated the
educational attainment of both spouses. In this variable, three educational attainment
categories of the respondent (less than primary, primary and more than primary) were linked
to the same categories for the respondent’s spouse. The same strategy was followed for the
variable ‘worked/did not work’. The values in the exp(B) column present the odds ratio for
women in that particular category to be selected as head compared to the reference group.

133
Table 27: Vanuatu (2009) - Logistic regression to predict the men or women with
certain individual characteristics will be chosen as heads of household in nuclear
households with children

    Male Female
Variable Category B exp(B) B exp(B
Education partners Partners both less than primary . . . ) .
  Primary, partner less than primary -0.616 0.540 0.863 2.370
  More than primary, partner less than idem -0.627 0.534 0.822 2.275
  Primary, partner less than primary -0.873 0.418 0.621 1.861
  Both partners primary -0.536 0.585 0.560 1.750
  Primary, partner more than primary -0.982 0.375 0.482 1.619
  More than primary, partner less than idem -0.791 0.454 0.638 1.893
  More than primary, partner primary -0.456 0.634 0.999 2.714
  Both partners more than primary -0.651 0.521 0.719 2.052
Age 10 year age-groups 15 - 24 yrs. . . . .
  24 - 34 yrs. -0.278 0.757 0.124 1.132
  35 - 44 yrs. -0.206 0.814 -0.132 0.877
  45 - 54 yrs. -0.070 0.932 -0.172 0.842
  55 - 64 yrs. 0.317 1.372 -0.360 0.697
  65 - 74 yrs. -0.136 0.873 -0.075 0.928
  75 - 84 yrs. -0.181 0.834 -0.812 0.444
  85+ yrs. 0.188 1.207 - 0.000
No. of sons in hhold   0.046 1.047 19.481
0.050 1.052
No. of daughters in hhold   0.107 1.113 -0.026 0.975
Urban/Rural Urban . . . .
  Rural 1.276 3.582 -1.215 0.297
Religion Anglican . . . .
  Presbyterian 0.478 1.613 -0.546 0.579
  Catholic 0.009 1.009 -0.116 0.890
  SDA 0.227 1.255 -0.222 0.801
  Church of Christ -0.353 0.702 0.174 1.190
  Assemblies of God 0.336 1.400 -0.459 0.632
  Neil Thomas Minsitry 0.395 1.484 -0.517 0.596
  Apostolic 0.028 1.028 -0.087 0.917
  Customary beliefs 0.671 1.957 -0.972 0.378
  No religion 0.758 2.133 -0.856 0.425
  Refuse to answer 0.745 2.107 -0.518 0.596
  Others 0.159 1.172 -0.290 0.748
Citizenship Vanuatu by birth . . . .
  Vanuatu by naturalisation -0.288 0.750 -0.609 0.544
  Other countires -0.175 0.840 0.181 1.198
Age husband - age wife   0.006 1.006 -0.014 0.986
Working status of couple Both partners worked . . . .
  Worked, partner did not work -0.384 0.681 -0.037 0.964
  Did not work, partner worked 0.015 1.016 0.402 1.495
  Both partners did not work 0.035 1.035 -0.032 0.969
Constant   2.063 7.873 -2.075 0.126

Source: Population and Housing Census of Vanuatu (2009)

319. Table 27 presents the results of this logistic regression. Compared to couples who have
both less than primary education, all other educational combination categories score

134
significantly higher, i.e. women in these educational groups are much more likely to be heads
of household. For instance, if a woman has more than primary education and her husband has
less than primary, her odds to be head of the household are 2.3 times higher than in the case
where both spouses have less than primary education. The largest effect of all variables in the
equation is urban/rural. In rural areas the odds for a male to be selected as head of the
household are 3.6 times higher than for a female to be selected, after controling for other
intervening factors. For women, each extra son increases her likelihood slightly to be selected
as head (odds ratio 1.05), while each additional daughter diminishes her changes (odds
ratio .975). It should not come as a surprise that women who practice a traditional religion
have a much lower likelihood of becoming head of household (odds ratio 0.378). Women
who belong to a traditional religion will most probably live in households that are more
conservative in terms of the position of women within the family. ‘No religion’ also scores
very low (.425). However, it is possible that interviewers interpreted traditional beliefs as ‘no
religion’. Finally, it is interesting to see that men who worked, but whose wife did not, have
much lower odds of being selected as head of the household. This may show that enumerators
did not always use strict rules to assign a head but perhaps made a person head who was
readily available (or that women selected themselves as head when an enumerator showed up
when their husbands were out to work).

320. Based on data from the 1982, 1990, 2000 and 2005 Chinese censuses, Lin and Zhao
(2010) investigated the effect of the sex of the first-born child on his/her chances of living
with the mother, the father or both. They decomposed this probability by the different events
that affect it: whether the parents got married, whether either of the parents migrated, whether
the couple divorced/separated and, if so, who got child custody. Parental mortality was not
considered as it is hard to imagine how the sex of the first-born child might affect it. They
found a small but significant tendency for first-born boys to live with either or both parents
more often than first-born girls. The decompositions indicated that: 1) When unmarried,
having a first-born son increases the probability of subsequent marriage; 2) When married,
having a first-born son decreases the probability of parent migration and reduces the
probability of divorce; 3) When divorced, having a first-born son increases the probability of
custody by the father. They also found that all these effects have become stronger over time,
particularly in the 2005 census. Again, a likely explanation is the preference by parents of
sons over daughters, but other, more subtle possibilities cannot be entirely ruled out. It may
be, for example, that parents, knowing that a first-born girl objectively decreases their
chances for sustenance in old age, regardless of their own preferences in the matter, take
certain compensatory actions (additional children, migration, job changes) that affect their
subsequent chances of staying together as a couple.

321. Type of household is an important predictor for most of the topics covered in this
manual. Below, some examples from the literature are presented in which the type of
household is used as an explanatory variable in the gender study of fertility, mortality and
education. As mentioned in the box above, Chu, Xie and Yu (2007) showed in the case of
Taiwan that there is a positive relationship between the proportion of girls in the household
and the total number of children. This suggests that in countries with marked son-preference
parents continue to have children until they have at least one boy. In such countries,
computing this correlation coefficient is recommended as a standard practice, together with
the sex ratios at birth by parity, to be discussed in the next sub-chapter. Note, however, that
this makes sense only for one-family households. Other ways in which household
composition data can be used to study fertility behaviour include the Own Children Method,
which was briefly mentioned in the sub-chapter on fertility

135
322. Households may vary in terms of poverty, health status or school attendance of the
children, depending on whether certain kinds of household members are present or not. There
is a lot of public debate on whether the absence of a father, a mother or both parents has a
negative impact on the development of the children. One-parent families, particularly
families consisting of a mother without husband and several dependent children, are thought
to pose greater risks for the health of children. In fact, the health indicators of such families
are often more unfavourable, but so are their socioeconomic characteristics, so that it is not
clear if it is the latter or the former that increases the risk. Blakely et al. (2003) have carried
out multivariate analyses on data from the 1991 census of New Zealand, which were linked to
mortality records in order to control for the socioeconomic determinants. Their conclusion is
that there does not appear to be notable variation in relative risk terms of socioeconomic
differences in child mortality by age or cause of death and that any association of one-parent
families with child mortality is due to associated low socioeconomic position. It may be
appropriate to replicate this kind of study in other contexts, to see if similar results are
obtained in developing countries.

323. A similar issue that has come up in the literature is whether children are better cared for
in female-headed households than in households with male headship. The argument is that
mothers will usually make household decisions based on the best interests of their children,
but may not be in a position to do so if they are subject to the authority of a male head of
household (e.g. Castle, 1993). However, female-headed households are different from male-
headed households in a variety of ways and many factors have to be controlled for in order to
conclude that this is indeed the determining factor. One study that attempted to do this, for
example, is the one by Adhikari and Podhisita (2010), on household headship and child
deaths in Nepal, based on the 2006 Demographic and Health Survey of that country. Using a
binary logistic regression model which contains the age at first marriage, children ever born,
place of residence, ecological zone, literacy status, religion, wealth status, use of family
planning methods, visits to a health facility, and antenatal care for the last pregnancy as
predictors. Controlling such factors, the authors found that deaths among children born
during the last five years were 31 per cent less common in female-headed households than in
male-headed households. Basing studies of this kind on DHS, rather than census data has
certain advantages because the information on dates of birth and certain potential
determinants of child health is more extensive in the DHS than in the census. Nevertheless, to
the extent that the census contains the basic data on children ever born and surviving, it
should be possible to carry out similar analyses using census data.

324. In some national contexts, the number of siblings and the presence of a grandmother
have been identified as having an impact on girls’ school attendance. Parker (2005) used
16,000 aged 6-14 sampled by the 2001 Lesotho Demographic Survey to look at the
relationship between residence with a grandmother and current school enrolment for children,
ages 6-14, in Lesotho. Logistic regression was used to establish whether having a
grandmother living in the household was associated with school attendance. The results
showed this association to be positive. Taking this analysis, which is equally feasible with
census data as with DHS data, one step further, one might differentiate between 16
categories, depending on the presence of the father, the mother, the maternal grandmother and
the paternal grandmother, controlling perhaps for urban/rural residence and the level of
education of the head of household to investigate school attendance of boys and girls by age
in each category.

136
325. The importance of studying one-person households was mentioned earlier. The basic
question is what type of persons live in such households. The 2009 Vanuatu Population and
Housing Census was used to investigate the factors that determine whether a person lives in a
one-person household or not. A logistic regression was set up in which the dependent variable
was whether the person was head of a one-person household or living in another type of
household (with multiple persons). The results of this analysis are presented in Table 28. To
show the importance of setting up the right multivariate equation, two models were
constructed. In model I, next to sex the following predictors were included: Age, Age 2,
Citizenship, Urban/rural residence, Education (3 categories), Residence 5 years ago and
whether the person worked the week before the census. This model shows that the two
categories with the highest odds for a person to be found in a one person household are:
‘being born outside Vanuatu’ and ‘female’. A person who is born outside Vanuatu and who is
not naturalized is about 4 times more likely to live in a one-person household than a person
born in Vanuatu. Women have a 2.2 larger odds ratio to live alone than men. This finding
confirms the results obtained by using a simple cross tabulation between sex and one
person/multiple person household. In model II, Marital status is introduced. The results show
that this now becomes the most discriminatory variable in the equation. Compared to never
married persons, all other marital statuses have much lower probabilities of living alone.
Somebody who is in the married state has an odds ratio of almost 0.034 of living alone
compared to a never married person. However, the effect of sex becomes relatively minor and
is even less than 1 (0.886), which means that, after controling for marital status, the
likelihood that a woman lives in a one-person household is less than that of a man.

Table 28: Vanuatu (2009) - Logistic regression for whether the head of household
conforms a one-person household or not, by selected explanatory variables

Predictors Categories Model I Model II Male only Female only

    B Exp(B) B Exp(B) B Exp(B) B Exp(B)
Age   -.062 .940 .004 1.004 .016 1.016 .034 1.034
Age2   .001 1.001 .000 1.000 .000 1.000 .000 1.000
Marital status Never married                
  Legally married     -3.357 .035 -3.980 .019 -1.547 .213
  Defacto     -2.841 .058 -3.387 .034 -1.431 .239
  Divorced     -.436 .647 -.323 .724 -.115 .892
  Separated     -.762 .467 -.597 .551 -.663 .515
  Widowed     -1.209 .298 -1.106 .331 -.729 .482
Citizenship Vanuatu by birth                
  Vanuatu by naturalisation .412 1.510 .360 1.433 .315 1.371 .340 1.405
  Other countires 1.437 4.207 1.396 4.038 1.405 4.074 1.124 3.077
Urban/Rural Urban                
  Rural -.018 .983 .070 1.072 .101 1.107 .170 1.185
Sex Male                
  Female .802 2.229 -.121 .886        

Education (3 categories) Less than primary                

  Primary education -.268 .765 -.161 .851 .019 1.019 -.604 .547

  More than primary .110 1.117 .124 1.132 .210 1.234 -.019 .982

Worked week before census Worked                

  Did not work -.142 .867 -.116 .890 -.212 .809 -.111 .895

Residence 5 years ago Sample place as census                

  Other place .593 1.809 .522 1.685 .591 1.806 .391 1.479

Children Surviving               -.200 .819

Constant   -2.189 .112 -1.279 .278 -1.186 .305 -2.518 .081

137
Source: Population and Housing Census of Vanuatu (2009)

326. There is a good chance that the likelihood to live alone is less for men and women who
have children with whom they could go and live at an older age. The census provides data on
the number of children ever born (CEB) and on the number of children who have deceased,
although only for women. To measure the effect of having children on the likelihood of living
alone, the number of surviving children of woman was calculated. Then two logistic
regressions were carried out – one for each sex - with the same set of explanatory variables as
before, but including the number of surviving children in the regression for women (see last
two columns of Table 28). The results clearly show that having surviving children, the
likelihood to live alone becomes considerably smaller. With each additional surviving child,
the odds ratio of living in a one-person household decreases by about 18 per cent.

7. Interpretation, Policy and Advocacy

327. Advocacy efforts to reduce gender inequalities regarding household and family
composition should include some or all of the following elements.

a) Addressing diverse living arrangements

As the ICPD Programme of Action notes (Par. 5.1), when policies and programmes
that affect the family ignore the existing diversity of family forms, or are
insufficiently sensitive to the needs and rights of women and children, parents may
face great difficulties in reconciling work and family responsibilities. The Beijing
Platform for Action (Par. 46) also stresses that since “many women encounter specific
obstacles related to their family status, particularly as single parents,” policies must
pay special attention to address their needs and to support family stability.

Special attention must also be paid to the needs of widows and orphans (ICPD
Programme of Action, Par. 5.13).

b) Addressing the burden of household and family activities

Promotion of changes in traditional notions of gender-based division of parental and

domestic functions, in order to reduce the burden of care and household activities that
women and girls frequently face and that jeopardize their educational and professional
opportunities.

Along with this, governments should promote quality and comprehensive sexual and
reproductive health services in order to ensure women and men the opportunity to
balance the size of their families with their needs, desires, and goals. Affordable and
physically accessible care facilities (e.g. child-care facilities, kindergartens, care
services for those who are ill, disable, elder, etc.) must also be provided to support the
different types of families in their efforts to reconcile productive and reproductive
roles. Another important action to be taken is the promotion of family-friendly work
environments, including the right to flexible working hours and schedules, paid
parental and maternal leave, maternal protection, health insurance and social security
(Beijing Platform for Action, several articles).

138
 Regarding education, increase enrolment and retention rates of girls may also be
supported “by allocating appropriate budgetary resources and by enlisting the support
of parents and the community, as well as through campaigns, flexible school
schedules, incentives, scholarships and other means to minimize the costs of girls’
education to their families and to facilitate parents’ ability to choose education for the
girl child” (Beijing Platform for Action, Par. 80 f.).

The ICPD Programme of Action (Par. 5.2.b) emphasizes the importance of

establishing social security measures that address the social, cultural and economic
factors behind the increasing costs of child-rearing.

c) Addressing family poverty risk

According to ICPD Programme of Action (Par. 5.4), “when formulating socio-

economic development policies, special consideration should be given to increasing
the earning power of all adult members of economically deprived families, including
the elderly and women who work in the home, and to enabling children to be educated
rather than compelled to work”. The text also adds that “particular attention should be
paid to needy single parents, especially those who are responsible wholly or in part
for the support of children and other dependants, through ensuring payment of at least
minimum wages and allowances, credit, education, funding for women’s self-help
groups and stronger legal enforcement of male parental financial responsibilities.”

139
Chapter 8:
Income, Poverty and Living Conditions
1. What is it?

328. Despite some recent setbacks, caused by the ongoing global economic, financial and
food crises, many countries in the developing world have made significant progress in the
reduction of poverty during the last decade. According to the 2011 UN Millennium
Development Goals Report, it is expected that the world will move below its target level of
23 per cent poverty by 2015. Despite this progress - which to some degree is guided by a
continued rapid growth in Eastern Asia (China) - many countries continue to struggle to
provide the basic needs for their populations. To monitor this progress, it is important that
high quality information is provided to intensify actions to combat poverty. The United
Nations does not provide a standard definition of poverty that applies to all countries.
However, the World Bank poverty limit of USD 1.25 (which is computed based on Parity of
Purchasing Power – PPP22) per capita is still generally used as a numeric measure of absolute
poverty, despite its limitations. This measure is also used for the purpose of measuring
progress in the achievement of Millennium Development Goal I: ‘Eradicate extreme poverty
and hunger. (Target: Halve, between 1990 and 2015, the proportion of people whose income
is less than USD 1 a day)’.23

329. Going beyond a pure financial notion of poverty, the Report of the World Summit on
Social Development in Copenhagen (1995) differentiated between two levels of poverty:
absolute poverty and overall poverty.
a. Absolute poverty is defined as “severe deprivation of basic human needs, including food,
safe drinking water, sanitation facilities, health, shelter, education and information,” and
depends on income and/or access to services.
b. Overall poverty is defined as “lack of income and productive resources to ensure
sustainable livelihoods; hunger and malnutrition; ill health; limited or lack of access to
education and other basic services; increased morbidity and mortality from illness;
homelessness and inadequate housing; unsafe environments and social discrimination and
exclusion.” Overall poverty is found as pockets of poverty amid wealth in richer countries
and as mass poverty in poorer countries (United Nations, 1995 a).

At the Copenhagen Summit, 117 countries adopted the commitment to eradicate absolute
poverty and reduce overall poverty. Countries were urged to develop national strategies to
reduce overall poverty substantially and to eradicate absolute poverty before a fixed point in
time.

330. The Beijing Platform for Action of the Fourth World Conference on Women describes
poverty as “multidimensional” and as a relative lack of income or productive resources to
ensure adequate food, shelter and housing, or through the human conditions associated with
poverty such as hunger and malnutrition, ill health, limited access to education, increased
morbidity and mortality, inadequate housing and homelessness, unsafe environments, and
social discrimination and exclusion (United Nations, 1995 b).

22
PPP refers to purchasing price parity, which measures the relative purchasing power of different countries’
currencies over the same types of goods and services, adjusting for inflation. PPP helps provide an accurate
comparison of standards of living across countries (World Bank, 2011).
23
The change from USD 1 to USD 1.25 was introduceed to correct for inflation of the US dollar.

140
331. After the World Summits in Rio, Cairo, Copenhagen and Beijing, various expert groups
were installed under the auspices of the United Nations Statistical Commission to follow up
and evaluate the progress made in the implementation of the action plans and provide advice
on the recommendations agreed to at these summits. As part of this process, in 1997 an
Expert Group on Poverty Statistics (Rio Group) was set up, chaired by the Brazilian Institute
for Geography and Statistics (IBGE), with its Secretariat at the UN Economic Commission
for Latin America (ECLAC). The Compendium produced by the Rio group presents a set of
poverty measurement approaches and methodologies.

332. Based on the preceding paragraphs, one can distinguish between two broad approaches
to measure poverty. In the first approach the income of individuals or households is
compared to a given poverty line. ‘The poverty line represents the aggregate value of all the
goods and services considered necessary to satisfy the household’s basic needs’ (Expert
Group on Poverty Statistics Compendium of best practices in poverty measurement. Rio
Group, 2006: 12). Two different perspectives exist within this approach. The ‘absolute’
poverty view only uses the basic necessities to guarantee the subsistence of the members of a
household. The ‘relative’ poverty perspective takes into account a person’s need to actively
take part in society and transcends the use of mere subsistence needs.

333. The second approach to measure poverty, usually called the Unmet Basic Needs
approach, makes use of a set of deprivation factors, which are established in advance, as
minimum conditions to be met to fulfill the basic needs of individuals and households.
Individuals and households are identified as poor if they do not meet the minimum in terms
of one or more of the deprivation factors. This second approach looks into the real
satisfaction of needs, while the first one looks at the availability of financial resources to meet
these needs. The second approach has been widely used, especially, in Latin America
(ECLAC, 2006 b: 101). The advantage of this methodology in the context of this manual is
that, while censuses do not always have income information, they do provide a wealth of data
to measure poverty using Unmet Basic Needs, including characteristics of the housing unit,
education and ownership of assets.

334. The monetary approach and the Unmet Basic Needs approach do not measure exactly
the same poverty dimensions. In particular, it is thought that Unmet Basic Needs change
more slowly than monetary poverty. Bearing this in mind, one can combine both criteria to
define a more dynamic poverty concept. For instance, those who are poor according to the
Unmet Basic Needs criterion, but whose current income/consumption places them above the
poverty line are sometimes referred to as the "inertially poor", meaning that their current
income would be sufficient to rise out of poverty, but that they will need more time to
overcome the deficiency of basic needs that they carry with them from the past.

335. Apart from such objective approaches, poverty can also be seen as a state of mind,
which depends on individual perceptions of one's position relative to others. The ‘subjective’
approach, therefore, leaves the determination of poverty in the hands of the respondent. It is
also possible to combine such subjective perceptions with income data, to generate
subjectively defined poverty lines. In general, countries that introduced direct questions on
income level in their censuses are able to use the ‘absolute’ or ‘relative’ approach. However,
censuses generally do not provide information for the subjective approach.

141
2. Why is it important?

336. Articles 23 and 25 of the Universal Declaration of Human Rights (1948) first recognized
poverty as a violation of human rights. Specifically, Article 23 states that working persons
have “the right to just and favourable remuneration ensuring for himself and his family an
existence worthy of human dignity.” Article 25 provides for the “right to a standard of living
adequate for the health and well-being of himself and his family, including food, clothing,
housing and medical care and necessary social services.”

337. In 2005, 1.4 billion people from developing countries were living below the World Bank
international poverty line of $1.25 a day (United Nations, 2010 a). Gender differences in the
incidence of poverty are widespread. Overall in less developed regions, fewer women than
men have access to cash income, and in most countries in Africa and about half of all
countries in Asia fewer women have access to land and property. In more developed
countries, older women are more likely to be poor than older men. Also, single mothers with
young children are more likely to live in poverty than single fathers with young children
(United Nations, 2010 a), especially in the absence of public transfer programmes.

338. The Beijing Declaration of the Fourth World Conference on Women (United Nations,
1995c) affirmed the international commitment to eliminate the burden of poverty for women
by addressing the structural causes of poverty and by providing equal access for both rural
and urban women to productive resources, opportunities and public services.

339. The inclusion of poverty-related questions in censuses is important to support policies

focused on poverty reduction and gender equality. The Millennium Development Goals set
forth within the Millennium Declaration (United Nations, 2000) to halve, over the period
1990 to 2015, the proportion of people whose income is less than one dollar a day. The
Declaration’s focus on poverty eradication has created a need for regular and timely
collection of data to monitor and evaluate levels of poverty. Some censuses provide income
data, but many others offer information related to poverty and economic well-being such as
health status, ownership of assets, educational attainment, living and housing conditions and
morbidity and mortality.

3. Data Issues

The following paragraphs will first elaborate on the first approach (poverty line) and next
deal with the ‘unmet basic needs’ approach.

340. The level of personal and/or household income plays a crucial role in determining
whether a household falls below the poverty line or not. According to the Principles and
Recommendations for Population and Housing Censuses. Revision 2 (United Nations, 2008
a), income may be defined as:

‘a) Income, in cash or kind, received by each household member;

b) Total household income in cash and in kind from all sources.

The preferred reference period for income data should be the preceding 12 months or past
year. The income could be classified as income from paid employment, self-employment,
property and other investment, transfers from governments, other households and non-profit
institutions’.

142
341. Census data on income can be used in many fields of interest. However, the use of
income data from censuses does not come without problems. Several shortcomings are
present that may jeopardize the quality of income information from a census.
1. Income is probably the most private question in the census, which can provoke a lot of
resistance and is therefore often placed at the end of the questionnaire to avoid a premature
end of the interview. Respondents are often very suspicious about any government agency
showing too much interest in the level of their earnings. Consequently, many respondents
refuse to give information about their income or, even worse, provide false information.
Specifically, people with high non-salary earnings have the tendency to underreport their true
level of income.
2. Questions on income are often gathered at the household level. This may pose problems
because in many societies household members may not know exactly what others earn. This
is certainly the case in composite household were one or more members are not related to
each other.
3. Incomes may be made up of a lot of different components, some of which may not be
readily remembered by the respondent, especially if they refer to occasional or informal
activities that imply benefits in kind or in cash money, on which no taxes are paid.
4, People may simply not know the exact amount they earn. For instance, shop keepers may
not know at the end of the month what their ‘net gain’ is.

342. To minimize these problems, censuses often rely on the use of income bands. These
bands reduce respondents’ burden and allow simple tabulations with other contextual social
and demographic variables. However, a drawback of this approach is that it becomes very
difficult to calculate household income from individually gathered information. Some
counties have taken measures to reduce problems of non-response for the income questions.
For instance, in Aruba, two direct questions (without bands) on income from main job and
other sources of income were asked to each member of the household. In the case a
respondent refused to (or could not) give exact information on his/her level of income, a flash
card was shown in which the respondent could indicate in what income category his/her
income fell. During the editing stage, data from the persons who answered the direct, detailed
income questions were then used to make hot deck 24 imputations for those who had only
given their income in a category. This allowed the calculation of the household income, while
taking away a lot of resistance against the direct question on income.

343. The first step to measure the proportion of persons or households below the poverty line
is, obviously, to establish a cutting point below which individuals or households are
considered to be poor. Many countries have established poverty lines which are used in their
social and economic planning. It lies outside the scope of this manual to go into the
methodology used to establish poverty lines. The interested reader is referred to the
publications by the World Bank25 or the UN Statistics Division26. Countries where no official
poverty line has been established can rely on the poverty limit of USD 1.25 per day set by the
World Bank and used to measure the progress in the MDGs. The UN 2011 Millennium
Development Goals Report indicates that in developing countries the proportion of persons
24
In a hot deck imputation information from other respondents with similar characteristics is used to make im-
puted that are best suited for the missing information. See: United Nations, (2008 a) Principles and Recommen-
dations for Population and Housing Censuses Revision 2: 70.
25
http://web.worldbank.org/WBSITE/EXTERNAL/TOPICS/EXTPOVERTY/EXTPA/0,,contentMDK:
20242879~menuPK:435055~pagePK:148956~piPK:216618~theSitePK:430367~isCURL:Y~isCURL:
Y,00.html.
26
http://unstats.un.org/unsd/methods/poverty/chapters.htm.

143
living on less than USD 1.25 has dropped significantly from a level of 45 per cent in 1990 to
27 per cent in 2005. All regions, except the Caucasus and Central Asia have currently lower
levels than in 1990.

344. Usually poverty lines are determined at the household level, rather than at the individual
level. Because of the large variety in the size and composition of households, it is necessary
to apply equivalence scaling to enable household income to be tested to a given poverty line.
This is usually done by assigning different weights to different persons in the household.
Weighting is important because of two reasons: 1) Different members of households have
different needs; e.g. an adult member’s nutritional needs are higher than a child’s; and 2)
There are economies of scale operating. For instance, ceteris paribus, a family of three will
spend less on energy costs per person than three individual persons. The OECD proposes an
equation for weighting that is commonly used for this purpose (Haughton and Khandker,
2009: 29):

AE = 1 + 0.7 (Nadults − 1) + 0.5 Nchildren

where AE stands for ‘Adult Equivalent’, ‘N adults’ for the total number of adults and ‘N children’
for the number of children. A person living alone would have an AE of 1, while a household
of two adults would have an AE of 1.7. A nuclear household consisting of a mother, father
and 2 children would have an AE of 2.7. By using these AE’s it is possible to compare
households of each possible composition and size to an established poverty line, but also to
other households.

345. In gender research on levels of poverty one may compare income to a pre-set poverty
line, but it is also possible make direct comparisons in income and poverty levels between
males and females Ideally, a gendered research would seek to examine differences in intra-
household level of income, resources allocation, and ownership of assets or appliances,
between women and men. For persons living with others in the same household, the
information on personal income should generally not be used to measure individual poverty,
as income is usually shared within the household. A real limitation of census data (where
income information is gathered at the household level) is that at the individual level, only
persons living in a one-person household can be compared. In most countries, women living
in a one-person household have higher poverty rates than men. Previous analyses have
shown, for example, that divorced or widowed women, living alone or as lone mothers, have
a higher prevalence of poverty than married women (United Nations, 2010 a).

346. As in most surveys, censuses do not inform in detail how household income is spent or
consumed at the individual level within the household or how resources are distributed to
each household member. Therefore, to address this data limitation, it is important to collect
individual income data and cross-tabulate them with household or family characteristics to
analyse both individual and household patterns. Some censuses allow this (see the last
paragraph of this section). But even this is not a guarantee that the distribution of resources
within the household will be accurately captured, as all too often the income of some
household members is appropriated by others

347. Below two country examples examining female-headed households in Mozambique and
Brazil are presented. These examples find different outcomes using the female-headship
variable, indicating that while female-headed household may be a useful measure of gender

144
inequality in some cases, findings and their interpretations may differ depending upon
context.

Country Example 12: Female-Headed Households and Poverty Risk in Mozambique

and Brazil

Mozambique. Fox et al. (2005) investigated the feminization of poverty in Mozambique using
data from the 1997 census and a household survey (2003). They found that the proportion of
female-headed households had increased in the poorest quintile (from 19 per cent to 24 per
cent), as well as in the second and third poorest quintiles (with 1.6 per cent and 1.9 per cent,
respectively), but decreased in the best- and second-best-off quintiles (with 4.1 per cent and
3.9 per cent, respectively). Most female-headed households are headed by widows and
divorcees, while a small proportion are single mothers. A higher proportion of female-headed
households stated that their situation had worsened in the last five years, and, this perception
of deteriorating conditions was found to be more pronounced among rural than urban women
who head households, suggesting better opportunities for female household heads in urban
areas. The underlying gender issue is that women predominate both in the agricultural sector
and unskilled labour, where returns to labour are low.

Brazil. Lavinas and Nicoll (2007) examined which type of family structure represented the
most vulnerable or ‘at-risk’ family arrangement. Using disaggregated employment data by
sex among women, then classified as head of family or wives, the results suggest that even in
the lowest income brackets, family arrangements involving lone mothers with children were
not necessarily the most vulnerable. The sex of the family head (i.e. ‘responsible person’)
was not a strong determinant of vulnerability; a family headed by a woman (often on her
own) or by a man (the overwhelming majority with a spouse) were almost equally likely to be
vulnerable, all other things being equal. Likewise, neither the sex of a family head, nor the
family type (i.e. two-parent or single-parent), made almost no difference in vulnerability.
This finding stands in contrast to results based on data from other countries, which has
identified that single-parent families with children were much more exposed to the risk of
vulnerability than two-parent families with children. Further, this study found that having
children in the household increased the likelihood of a family being vulnerable.

348. In recent years, due to studies such as the one above, the emphasis on household
headship as the differentiating gender variable has come under increasing criticism from both
statisticians and gender researchers, for the following reasons:

a) Most women live in male-headed households, so that what happens in female-headed

households imperfectly represents the situation of individual women;
b) To the extent that gender inequality is reflected in the relationship between male and
female members of the same households (e.g. unequal appropriation of earnings), focusing on
female-headed households may be misleading;
c) Focusing on female-headed households may lead to biased policy priorities (see also the
discussion in Chapter 7).

349. Additionally, the differences in poverty rates between male and female-headed
households, if not broken down into finer categories, are typically small and tend to be
associated with other demographic differences between these households, such as the number
of children and adult male and female household members. Medeiros and Costa (2006: 8),

145
assert that “the relationship between poverty and female headship of households seems not to
be direct and univocal, as poverty appears to have a stronger correlation with the presence of
children in the family and other characteristics of family members than with the type of head
of household.” Again, what seems to be the need then is to understand how income is earned,
allocated and spent within the household in order to understand the processes at play in
poverty as it may be gendered.

350. Most censuses do not directly measure income, even at the household level, so that the
poverty status of a household has to be ascertained by means of deprivation factors, among
which the ‘Unmet Basic Needs’ criterion is most often used. These deprivation factors
cannot be specified at the individual level, but are always connected to the household.

351. In the case census questionnaires do not include questions on income, combining census
information with income data from surveys still makes it possible to use small area estimation
techniques, based on the determination of characteristics of the population living in poverty,
to identify areas where poverty is high. Methodology Box 1 in Chapter 1 contains more
details on these methods. With this methodology - developed by the World Bank - household
surveys that have measures of income or consumption can be used to estimate statistical
models that can be applied to census data in order to estimate poverty at small geographical
levels. Once the geographic areas of poverty are identified, it is possible to analyse the socio-
economic characteristics of the population of these areas compared to the rest of the country.
A line of inquiry could then be as follows: ‘How different is the gender gap in a subject field
like education in an economically poor geographic area, compared with the national average?
And how do poor areas compare with the national average on other specific women-related
issues, such as the total fertility rate , sex ratios or unemployment?’

352. The Unmet Basic Needs approach uses standard indicators of the household’s
socioeconomic level that do not yield precise income estimates, but only broad classifications
of the household's situation. The typical components of the Unmet Basic Needs Index are the
following:

1. Crowding: number of persons per bedroom;

2. Quality of the dwelling: earthen floor or use of sub-standard materials for the walls or
roof;
3. Sanitation: Absence of an indoor toilet or running water inside the home;
4. Educational attendance: number of school-age children not attending school;
5. High dependency: number of persons per working household member;
6. Education: level of education of the household head.

The choice of the right components to be included in the equation is critical. In some
countries, the ownership of a transistor radio may still differentiate between households of
distinct income levels whereas in other countries it is so common that it provides little or no
information on the household's socioeconomic situation.

353. For each of these components, critical limits are defined (e.g. more than three persons
per bedroom for ‘Crowding’, or less than two years of formal education for ‘Education of the
household head’). This defines the number of Unmet Basic Needs. Finally, all households
that have more than one, two, or three (i.e. depending on the country) Unmet Basic Needs are
considered to be poor. This makes it possible to prepare detailed poverty maps and other

146
analytical instruments based on census data, without having any information on income
levels.

354. On the other hand, the approach has also been criticized for using a somewhat arbitrary
set of indicators with equally arbitrary cut-off points. In particular, there are alternative
methodologies (e.g. the wealth index of the DHS) that make use of the ownership of selected
consumer durables. However, this methodology is difficult to use in censuses (although some
countries do use this information) because there is so much variation between countries in the
way the information on consumer durables is collected. Moreover, the Unmet Basic Needs
Index is subject to the same criticism that applies to composite indices in general: ‘What does
one gain by defining several disparate measures of household wellbeing, rather than
investigating the different dimensions separately’. For more detailed information, see Chapter
3 of the Compendium of Best Practices in Poverty Measurement (ECLAC, 2006 b).

355. Censuses can provide additional information on gender deprivation factors, because they
provide essential information on the characteristics of the living unit (whether a hut, a house
or an apartment), in terms of comfort, equipment and status of occupation. Usually collected
during the same operation as the population census, these data can be easily cross-classified
with individual data. The Principles and Recommendations for Population and Housing
Censuses, rev. 2 (United Nations, 2008 a) recommend the following core topics be included
in a housing census. While all of these items may not be asked on the census, they can be
useful in providing a general picture of the well-being of the household and its inhabitants.
a. Type of living quarters
b. Occupancy status or tenure (own, rent, or occupy without cash payment)
c. Type of ownership, who is the legal owner?
d. Number of rooms
e. Water supply system
f. Main source of drinking water
g. Type of toilet
h. Sewage disposal
i. Bathing facilities
j. Availability of kitchen
k. Fuel used for cooking
l. Type of lighting and/or electricity
m. Main type of solid waste disposal
n. Number of occupants
o. Construction material of outer walls of household dwelling

356. A gendered focus would then examine these quality-of-living characteristics and spend
more attention on the types of characteristics with a potential gender component. For
example, access to water, source of drinking water, and fuel used for cooking all have a
potential gender component, because they can mean markedly different levels of work for
women.

357. A gendered analysis of poverty at the household level may examine the level of income
or characteristics of the living unit by the sex of the reference person listed as head of
household noting differences across women and men. At the individual level, if the census
questions on ownership of assets and cash income are asked, these data may be used to
examine the relative individual material situations of women and men. Individual questions,

147
such as education attendance or attainment, may also provide insight into how resources are
allocated within the household across male and female children.

358. As can be gleaned from the summary table in Annex 1, the information available in
censuses for the construction of poverty indicators varies considerably between countries.
Some (e.g. Bahamas, Brazil, Croatia, French Polynesia, Jamaica, Netherlands Antilles, St.
Lucia, and Singapore, among others) collect data on personal income. In the case of the
Bahamas, the individual amount is even broken down into 10 possible sources of income. St.
Lucia, on the other hand, focuses on the person's main job or source of income, whereas
Singapore only asks for income from work and the Netherlands Antilles asks for income from
the two most important sources. In other countries, income is declared only at the household
level. Mauritania asks for household expenditures, rather than household income. Still others,
like Albania or Turkmenistan, define only the sources of income, without asking for specific
amounts. But the most common situation is the one where no direct income/expenditure data
are available, but where the census asks the questions necessary to construct Unmet Basic
Needs indicators. This is the situation in, for example, Cambodia, Honduras, India, Kenya,
Lesotho, Mali, Mexico, Montenegro, Malaysia, Nicaragua, Pakistan, Peru, Romania, South
Africa, Thailand, Vanuatu and Zambia, among others.

4. Tabulations

359. Tabulations focused on poverty should disaggregate the data on relevant demographic,
social and economic characteristics by sex, even when bearing in mind that the poverty of
men and women may not be an individual characteristic, but rather characteristics of the
households in which they live. However, because poverty is relative to a country’s level of
economic development, salient tabulations will differ from one country to the next to show
the multiple dimensions of poverty. The Principles and Recommendations (United Nations,
2008 a) do not contain a recommended set of poverty tabulations. Possible tabulations that
could be adopted across most countries include the following:

1. Proportion of female and male headed households that are considered to be poor.
2. Proportion of single-parent households with children by sex of reference parent that
are considered to be poor.
3. Proportion of women and men by geographic area that are considered to be poor
4. Education levels of women and men that are considered to be poor.
5. Number of households that are considered to be poor by type of households.
6. Proportion of population below the poverty line by educational attainment and gender.
7. Proportion of persons in certain age groups (children, adults at various age groups,
elderly) and gender who are considered to be poor.
8. Persons by marital status and gender who are considered to be poor
9. Labour force participation and economic activities of men and women who are poor.
10. Specific sub-groups of the population relevant to a nation, such as migrant labourers,
refugees, and ethnic groups who are are consideredas poor.

360. In the above tables ‘considered to be poor’ refers to either individuals or to households
that have been labeled ‘poor’ on the basis of a test to an established poverty line or on the
basis of a set of deprivation factors (e.g. unmet basic needs). Next to measuring the number
of persons/households that are poor, it is also important to compare the socio-economic
characteristics of the poor with the non-poor to understand the dynamics of poverty and to
identify possible strategies for poverty eradication.

148
361. The census data of Vanuatu allowed for making an analysis of unmet basic needs. The
following deprivation factors were chosen:
 Dwellings with floors and walls from traditional/makeshifts or improvised materials.
 An average of more than three persons per room.
 No electricity used for lighting, no running water or water from well.
 Households with children 5 – 12 years old, not attending school.
 Head of household with 3 or less years of education in households with 3 or more
people per employed person

362. Households were considered to be poor when they scored positive to at least one of the
five deprivation factors. In total 43.5 per cent of all households scored positive on at least one
of these five factors. This figure comes close to the ‘Intensity of Deprivation’ provided by the
explanatory note of the 2011 Human Development Report on Vauatu, which was 42.7 per
cent. Figure 9 shows the percentages of poor households by type of household. One-person
households had the highest percentage of poor persons.. No less than 62.5 per cent of persons
living on their own had at least one deprivation factor. It is often stated that single mothers
with children have much higher poverty. Our analysis shows that in fact they score about the
same as nucear households with children(47.3 against 46.7 per cent). However, compared to
lone fathers with children (38.7 per cent) they score higher. Extended families that include
grandchildren score as high as single mothers. Composite households where the head forms
part of a nucleus, but where other not related persons are present score lowest (15.9 per cent).
This should come as no surprise as these are often the households where one or more
housemaids are present.

Figure 9: Vanuatu (2009) - Percentage of households scoring on at least one deprivation

factor

Source: Population and Housing Census of Vanuatu (2009)

149
363. An aspect of particular interest is the sex of the household head. As was emphasized in
Chapter 7, dividing households by the sex of the head of household is insufficient for the
purpose of poverty differentiation. Table 29 shows that the differences within different
categories of male or female-headed households, depending on the other particularities of
their compositions, are much greater than those between the totality of male-headed versus
the totality of female-headed households.

Table 29: Vanuatu (2009) - Percentage of households considered to be poor, by type of

household and sex of the head of household

  Nunber of households Percentage poor

Type of Household Male Female Male Female

headed headed headed headed
One person hh – male 1,887 . 62.3 .
One person hh – female . 1,110 . 62.5
Couple without children - male head 2,563 . 38.1 .
Couple without children - female head . 191 . 14.1
Couple with 1 -2 children < 15 9,631 744 44.7 25.1
Couple with 3+ children < 15 6,867 446 53.6 31.2
Couple with children, no children < 15 1,567 69 43.5 29.0
Single mother, 1 - 2 children < 15 . 1,170 . 45.1
Single mother, 3+ children < 15 . 168 . 41.1
Single mother, no children < 15 . 1,296 . 50.1
Single father, 1 - 2 children < 15 444 . 38.1 .
Single father, 3+ children < 15 102 . 52.9 .
Single father, no children < 15 291 . 35.4 .
Extended hh, 1 -2 children < 15 5,272 1,544 35.6 34.8
Exteded houshold, 3+ children < 15 2,448 584 40.1 36.0
Extended hh, no children < 15 5,895 2,459 39.8 41.1
Composite household 434 77 29.7 18.2
HH not to be determined 40 64 37.5 34.4
Total 37,441 9,922 44.1 41.4

Source: Population and Housing Census of Vanuatu (2009)

364. The table shows the percentages of households in Vanuatu that are considered to be
poor, by type of household. As in Figure 9, a household is considered to be poor if it scores
positive on one of five deprivation factors. The type of household was extended from the
previous division to 18 distinct categories to control for the effect of having dependent young
children living in the household. The table shows that both men and women who are living
on their own have the highest percentage of poverty. Actually, the figures for each sex are
nearly identical (62 per cent). About a third of these one-person households have heads over
age 60. The latter category of households has a higher poverty rate, but again the percentages
for male and female-headed households are not markedly different: 73.2 and 71.4 per cent,
respectively. Vanuatu may not be typical in this respect as generally women living alone,
especially widows, are poorer than men in the same situation (United Nations, 2010 a).

150
365. Lone mothers who have 1 or 2 children below age 15 have somewhat higher levels of
poverty than lone fathers: 45.1 per cent, compared to 38.1. However, it seems that single
mothers who have 3 children or more below age 15 have a lower chance of being poor. One
should take into account, however, that the number of single fathers with 3 children or more
is very small (102). It is very interesting to see that in nuclear households consisting of a
married couple, with or without young children, levels of poverty are lower when the head is
a woman and not a man. For instance, if a woman is head in a household consisting of a
married couple with 1 or 2 children below age 15 the headcount index is 25.1 per cent,
against 44.7 percent when the head is a man. If the head of household is over 60 years old,
poverty rates are somewhat higher in both cases, but the difference between male and female-
headed households does not change appreciably. The percentage of persons that are poor
among extended households is much closer between both sexes. Here, as in most other
household categories, the difference between the poverty rates of households with heads over
and under age 60 is larger (with higher poverty among the former) than between those with
male or female heads. This situation may be different in other parts of the world, such as
Latin America, where the poverty of older persons is lower.

366. These results show that a simple comparison between male and female headed
household would miss all the nuances that are present. Such an analysis would indicate that
the headcount index for both sexes would come very close (44.1 and 41.4 per cent for male
and female-headed households, respectively). It is well possible that a selection criterion is
operating making the position of households that are headed by a female (with the exception
of single mothers) different from other households. It may be that the underlying reason that
makes a female head of the household is also the reason why their chance of being poor is
smaller, i.e. their economic position.

367. Living conditions can also be analysed along different gendered dimensions. In terms of
health and sanitation, data on the source of drinking water, fuel for cooking, sewage disposal,
toilet, and main type of solid waste disposal may be relevant to use. Men’s and women’s
roles and inequalities can be related with living conditions.

368. Additionally, women’s time use may be shaped by the household’s main source of
drinking water and water supply system, the availability of a kitchen, and the number of
household occupants that may carry out or add to household chores. At this level, though,
cultural knowledge about women’s and men’s roles will also inform the construction of
relevant tabulations.

5. Indicators

369. Relevant poverty indicators vary across countries. At the country level it is first
necessary to examine the census form to identify items relevant to differentiating absolute
and overall poverty within that country’s setting.

370. A number of aggregate indices of poverty are available (see Haughton and Khandker,
2009: 69):

1. The Headcount Index is the index most frequently used. The index simply states the
number of persons that are considered poor per 100 total population. The number of
persons considered poor is again the result of a .test using either a poverty line or
deprivation factors. Although widely used, this indicator has some weaknesses. It does

151
 not measure the intensity of poverty; i.e. it does not look into how poor the poor really
are. Another disadvantage is that it is calculated at the individual level, while most
data on poverty are collected at the household level.
2. The Poverty Gap Index presents the average difference between the poverty line and
persons’ actual income. Only persons falling below the poverty line are included in
the equation. The Poverty Gap Index (P) is calculated as:

Gi = (z – yi) × I(yi < z)

Where Gi: is the poverty gap for person i;

Z is the poverty line;
Yi is the income of person i and
I takes value 0 or 1. 0 if a person’s income is greater than z and vice versa.

And subsequently:

P = 1/N x Σ(Gi/z)

Apart from computing whether female-headed households (or sub-categories thereof)

are or are not poorer than male-headed households, it may be worthwhile to compare
the respective Poverty Gap Indices, to assess whether poverty is deeper in female-
headed households (or sub-categories thereof) than in the respective male-headed
households.
3. Next to these general measures a number of specialized poverty indicators have been
developed (e.g. the Squared Poverty Gap (Poverty Severity) Index, the Sen index, the
Sen-Shorrocks-Thon Index.erc.). The methodological aspexts of these measures can
be found in the ‘Handbook on Poverty and Inequality’ (2009) by the World Bank.27

All these poverty indices use individual data and not household characteristics. In those
countries were census data on individual income are available, these measures could be
calculated separately for males and females.

371. The Millenium Development Goals (MDGs) set out measurable indicators of poverty
that can be used to define and monitor it over time.

Within the first goal of the MDGs (Eradicate extreme poverty and hunger) three targets have
been specified. Progress in each of these targets can be measured by some specific indicators.
The official United Nations site for MDG-indicators discerns the following indices28:

Target 1.A: Halve, between 1990 and 2015, the proportion of people whose income is less
than one dollar a day
1.1 Proportion of population below $1 (PPP) per daya
1.2 Poverty gap ratio
1.3 Share of poorest quintile in national consumption
Target 1.B: Achieve full and productive employment and decent work for all, including
women and young people
1.4 Growth rate of GDP per person employed
27
http://web.worldbank.org/WBSITE/EXTERNAL/TOPICS/EXTPOVERTY/EXTPA/0,,contentMDK:
22405907 ~menuPK:6626650~pagePK:148956~piPK:216618~theSitePK:430367,00.html.
28
http://mdgs.un.org/unsd/mdg/Host.aspx?Content=indicators/officiallist.htm.

152
1.5 Employment-to-population ratio
1.6 Proportion of employed people living below $1 (PPP) per day
1.7 Proportion of own-account and contributing family workers in total employment
Target 1.C: Halve, between 1990 and 2015, the proportion of people who suffer from hunger
1.8 Prevalence of underweight children under-five years of age
1.9 Proportion of population below minimum level of dietary energy consumption

Only a few of these indices can be calculated on the basis of census data. For the purpose of
gendered research on poverty, where possible these indicators should be calculated separately
for each sex.

372. Because many censuses do not provide the necessary information for computing
monetary poverty metrics, the Unmet Basic Needs (UBN) approach with its different
variants, is the most frequently used method for estimating poverty levels for small
geographic areas. For example, Skoufias (2005), in a slight variation on the UBN-
methodology explained before, used data from the 2002 Population and Housing Census of
Guyana related to the access of households to basic services, like water, electricity and
garbage disposal to construct a Living Conditions Index (LCI). This index is based on the
assignment of the response codes into levels: level 1 for high quality to level 5 to denote low
quality. Each level was assigned a number of points (i.e. 100 points for level 1, 75 for level 2,
50 for level 3, 25 for level 4 and 0 for level 5 or no access). These points were then summed
across six areas:

1) Access and quality of a household’s water source;

2) Source of drinking water;
3) Type of toilet facility;
4) Type of lighting;
5) Main method of garbage disposal; and
6) The extent of crowding in the household (the number of people residing in the
household divided by the number of bedrooms in the dwelling).

373. For each household, the LCI value was computed as the sum of points across the six
categories: the lower the sum, the poorer the household. The household-specific index was
then averaged by an enumeration geography unit, such as a tract, village, district, or region, to
provide a measure of the relative quality of services by that geographic unit. Because the LCI
is a number that is derived at the household level, it can be used to rank households within a
geographic unit.

374. A somewhat under-utilized resource are the often extensive lists of assets asked for in
censuses that may or may not be present in the household. The 2007 census of Swaziland, for
example, asked for the presence of 13 items, including cars, vans, motorcycles, computers,
mobile phones, internet connections, refrigerators, radios and TVs. On average, male-headed
households possessed items in 3.2 of the 13 categories, compared to 2.8 for female-headed
households.

375. Some indicators of poverty, such as availability of water, type of heating may be
important for research on gender differentials. As an example, fetching or pulling water for
the household in most developing countries is more likely to be performed by girls or women
than by boys or men, which may have consequences for girls' school attendance. In sub-
Saharan Africa, only 54 per cent of households are within 15 minutes from a source of

153
drinking water, and girls under 15 years are more likely than boys of the same age to be in
charge of water collection (United Nations, 2010 a). From this, it is relevant to tabulate
source of drinking water across boys and girls, and then compare the levels of school
attendance across boy and girl groups.

6. Multivariate and further gender analyses

376. Multivariate regression techniques provide the means to examine differentials in levels
of income and poverty between different subgroups in society. Census data from two island
countries, Vanuatu and Aruba, will be used to illustrate the application of multivariate
regression techniques to research gender differentials in income and poverty. In the case of
Vanuatu an analysis closely related to the Unmet Basic Needs will be done, while for Aruba
income differentials between males and females will be examined, while controlling for
intervening factors. Both analyses will use a Multiple Classification Analysis (MCA).

377. In Vanuatu, five deprivation factors were first calculated for households (see above).
Whenever a household scored positive the deprivation factors obtained value ‘1’, in all other
cases the factor remained ‘0’. The sum of these five deprivation factors varies between 0 and
5. This score can be interpreted as in indication of the overall deprivation of the household.
Then this sum of the deprivation factors (intensity of poverty) for the households were
assigned to all the persons living in the household. The overall mean of this indicator for all
persons was 0.55. Table 30 shows the result of an MCA-analysis. The last colomn shows
deviations from the overall mean after controling for other factors and covariates. A positive
sign of the deviation means that the particular category has a higher intensity of poverty as
the reference category, and a negative coefficient means a lower level of poverty. In addition
to a number of categorical variables, the age of the respondent and the number of children in
the household younger than 15 were introduced as control variables.

378. The results in Table 30 clearly show that levels of poverty are much higher in rural than
in urban areas. The intensity of poverty score was 0.106 in urban areas and 0.696 in rural
areas. Poverty also varied across the three categories of citizenship. Persons who had a
citizenship from another country had on average a lower poverty level than persons who were
born on Vanuatu or who were naturalized (difference -0.217). The results on the type of
household confirm our conclusion based on the cross tabulation depicted in Figure 9, i.e.
persons residing in one-person households have a higher degree of poverty than the other
households. Also, slightly higher levels of poverty are present among nuclear households
with a single mother and with parents with children and in extended households with
grandchildren. Differences between male and female heads of households turn out to be
almost non-existing. However, this does not prove that there are no difference in income and
poverty levels between males and females in Vanuatu. Because, as stated by The World’s
Women 2010 (United Nations, 2010 a: 159), “if the total number of poor is disaggregated by
sex (i.e. the sex of the household members), the results are not going to reflect possible
gender inequality within the households but merely the distribution of population by sex in
poor households.”. To really disentangle the existence of poverty and deprivation between the
sexes on Vanuatu, more in-depth research is necessary on individual income levels and the
distribution of wealth between members of both sexes in households.

154
Table 30: Vanuatu (2009) - MCA analysis number of deprivation factors with selected
explanatory variables

    Predicted Mean Deviation

Adjusted Adjusted
for Fact- for Fact-
Unadjust- Unadjust
Variables Categories ors + ors +
ed -ed
Cova- Cova-
N riates riates
UrbanRural Urban 54,830 0.084 0.106 -0.468 -0.447
Rural 170,671 0.703 0.696 0.150 0.144
Sex Male 114,401 0.552 0.554 -0.001 0.002
Female 111,100 0.553 0.550 0.001 -0.002
Citizenship Vanuatu by birth 215,790 0.558 0.554 0.006 0.001
Vanuatu by naturalisation 7,785 0.503 0.554 -0.050 0.002
Other countires 1,926 0.128 0.425 -0.424 -0.127
Marital status Never married 128,909 0.560 0.550 0.007 -0.002
Legally Married 71,927 0.570 0.559 0.017 0.007
Defacto 18,706 0.401 0.522 -0.152 -0.031
Divorced 539 0.573 0.623 0.021 0.071
Separated 1,223 0.549 0.606 -0.004 0.054
Widowed 4,197 0.703 0.626 0.151 0.073
Hhold type detailed One person household 2,997 0.775 0.830 0.222 0.278
Nuclear, couple, no children 5,508 0.439 0.533 -0.113 -0.020
Nuclear couple, with children 92,755 0.625 0.573 0.072 0.020
Nuclear, mother with children 9,211 0.653 0.612 0.100 0.060
Nuclear, father with children 2,656 0.513 0.506 -0.040 -0.047
Extended: head in nucleus+gchildren 19,546 0.614 0.578 0.062 0.026
Extd: head in nucleus+gchildren+other 24,276 0.482 0.530 -0.071 -0.022
Extd: head in nucleus+others 60,981 0.447 0.501 -0.106 -0.051
Extd: lone head+parents in law 225 0.502 0.529 -0.050 -0.024
Extd: lone head+extended family 6,656 0.479 0.574 -0.074 0.022
Cannot determine 690 0.436 0.532 -0.116 -0.020

Source: Population and Housing Census of Vanuatu

379. Another approach is presented for Aruba. This analysis does not focus directly on levels
of poverty, but tries to differentiate between the levels of income from work between both
sexes. In the 2010 Aruban population census, two questions were asked about income: one on
monthly income received from the person's main job, and one on any other form of monthly
income. To analyse the results of the first question, i.e. income from main job, a Multiple
Classification Analysis (MCA) was set up to examine the gender gap in income. To control
for intervening factors and covariates, the following predictors were incorporated in the
model: gender, age, educational attainment, country of birth, occupational category (ISCO –
main categories). ‘Hours of work’ was not included, as in Aruba there is hardly any
difference between both sexes. Only persons between 15 and 65, who were working at the
time of the census, were included. The results of the MCA- analysis are presented in Table
28.

155
Table 31: Aruba (2010) - MCA analysis of income from main job by main explanatory
variables

    Predicted Mean Deviation from overall mean

Variable Category Unadjusted Unadjusted Adjusted

Gender Male 3420.4 351.4 347.6

  Female 2726.2 -342.8 -339.1
Educational Less than primary/None 1808.0 -1261.0 -645.0
attainment Primary (special) education 2075.0 -994.0 -505.9
  Lower vocational education 2569.8 -499.3 -439.4
  High School 4 yr cycle. 2937.7 -131.4 -146.2
  High school 5 yr. cycle 2588.0 -481.0 -70.5
  High school 6 yr cycle 4787.3 1718.3 855.5
Vocational Education, Intermediate
  3578.5 509.5 233.0
  Higher education (Bachelor) 5058.0 1989.0 1035.4
  Higher Education ( Master) 7287.5 4218.5 2952.4
  Higher Education (PhD) 9867.3 6798.3 5219.3
Country of birth Aruba 3395.8 326.7 191.7
  Colombia 1990.7 -1078.4 -417.7
  China 2257.0 -812.0 -653.9
  Haiti 1552.9 -1516.1 -381.9
  Jamaica 1985.8 -1083.3 -191.3
  Netherlands 4848.5 1779.5 378.3
  Phillippines 2138.6 -930.4 -803.0
  USA 5339.0 2269.9 705.4
  Venenuela 2501.4 -567.7 -327.6
  Other country 2833.7 -235.4 -156.5
Occupation Managers 5815.1 2746.0 2066.4
Main categories Professionals 5247.1 2178.1 1063.4
Technicians & Associate
  Professionals 3948.4 879.4 563.9
  Clerical Support Workers 2800.4 -268.6 -106.7
  Service and Sales Workers 2139.0 -930.1 -530.2
Skilled Agricultural, Forestry &
  Fishery Workers 2156.0 -913.0 -998.2
  Craft & related trade workers 2433.1 -635.9 -564.5
Plant & Machine operators and
  Assemblers 2405.8 -663.2 -692.4
  Elementary Occupations 1593.2 -1475.8 -935.3
 
R2     0.375

Source: Aruba census 2010

380. The second column in the MCA-table shows the (unadjusted) mean for each category.
The overall average monthly income per person is Afl. 3,069 (USD 1,724, at the fixed rate of
Afl. 1.78 Afl. per US dollar). The third column gives the unadjusted deviations from the

156
main. It shows that still a significant income gap exists between both sexes, even after
controlling for intervening variables. The unadjusted deviations for both sexes are
respectively Afl. 351.4 for men and – Afl. 342.8 for women. From these deviations it is clear
that the difference in monthly income from main job between men and women is Afl. 694.2
(USD 390). It is interesting to see that the deviations from the overall mean for men and
women, after controlling for all intervening factors and covariates, is almost the same as the
unadjusted deviations. Some initial tests showed that in fact three of the predictors were
working in opposite directions. Controlling only for educational attainment in the model
resulted in a wider gender income gap (Afl. 789). Adding country of birth to the equation led
to a reduction in the income gap (Afl. 744) and occupational category further reduced the
difference (Afl. 689).

Figure 10: Aruba (2010) - Differential in income from main job between men and
women by age, controling for education, country of birth and occupational category

8000
Income Afl.

7000

6000

5000

4000

3000
Male
2000 Female

1000

0
15 20 25 30 35Age 40 45 50 55 60

Source: Aruba census 2010

381. A further analysis into the income differences between both sexes was done by
constructing a linear regression with the same variables, but some extra transformations of
age. To see if there was an interrelationship between age and sex, an interaction term ‘age x
sex’ was included. Also, the square of age was added, to check for non-linearity. The results
of this analysis are presented in Figure 10. All values for the dummy variables (educational
attainment, country of birth and occupational category) were kept at their mean. Figure 5
clearly shows that a) The income gap between males and females gets wider by age. The
regression coefficient for the interaction terms was -14.8, meaning that per year of life the
income gap between men and women widens by Afl. 14.8. At age 20, the fitted income gap is

157
Afl. 377, at age 30 it is Afl. 525 and at age 50 it is Afl. 673; and b) In Aruba the relationship
between income and age is quite linear. The regression coefficient for the square of age was
only -1.3.

382. Beyond the analysis shown in the preceding paragraphs, there is a standard literature on
the decomposition of male-female income differences by various contextual variables, such
as occupational differentiation, differences in levels of education, part-time versus full time
work and other factors that might account for the income differences. Typically, these
methods use the separately estimated (log) wage equations for two groups of workers to
decompose the difference in their (geometric) mean wages into a discrimination
(unexplained) portion and a human capital (endowments or explained) portion. The simplest
decomposition procedure is to adopt one of the estimated wage structures as the
nondiscriminatory norm. Often researchers select the wage structure for the group of workers
believed to be dominant in the labor market (at least relative to the comparison group).
Differences in the mean characteristics of the two groups are weighted by the estimated
coefficients for the nondiscriminatory wage standard and summed to obtain the human capital
portion of the overall wage differential. The discrimination portion of the overall wage
differential is the residual left over after netting out the human capital portion. Equivalently,
the discrimination portion can be directly obtained as the summed difference in estimated
coeffcients between the two groups of workers weighted by the mean characteristics of the
subordinate group. An implication of this procedure is that all of the discriminatory wage
differential is ascribed to underpayment of the subordinate group rather than to overpayment
of the dominant group (Neuman and Oaxaca, 1998).

383. Because this methodology is fairly complex and ideally requires more detailed
information than what is readily available from the census, it is not discussed here in any
detail. Nevertheless, for those who wish to go deeper into the econometric analysis of male-
female income differences, even with census data, it is probably necessary to get acquainted
with this literature, especially the articles by Oaxaca (1973), Binder (1973) and Oaxaca and
Ransom (1994). In his original article, Oaxaca demonstrated that in the US 74 per cent of the
male-female income difference between white workers and 92 between black workers should
be considered as based on “pure” discrimination within the occupational categories that he
used. As one uses finer occupational categories, this percentage tends to diminish because a
greater portion of the income difference is accounted for by variations between occupations,
rather than within occupations. Fresneda (2012), who used a finer occupational
differentiation for the case of Brazil, found that this significantly affected the results. There is
some discussion, however, as to whether such detailed occupational categories should be
used because as ever more detailed distinctions between categories are introduced, there is a
real possibility that the categories themselves will be instruments of labour force
segmentation that discriminate against certain categories of workers, such as women.

7. Interpretation, Policy and Advocacy

384. Regarding the point mentioned at the end of the previous paragraph, Anker (1998)
established that the largest contributor to the work and income differential between women
and men is that women and men tend to concentrate in different occupations, which he refers
to as horizontal occupational segregation. He also finds that even within an industry, women
tend to concentrate lower in the hierarchy, which he refers to as vertical occupational
segregation. In addition, cultural norms shape perceptions about what occupations are
suitable for women and men, and further, that men are typically the breadwinners within the

158
household, and hence their labour is increased in value while women’s labour value is
diminished.

385. The key is to understand the relative and absolute poverty situation within a given
country. Understanding poverty with a gendered lens involves asking questions such as:
a) What is the proportion of women or men who are poor?
b) Where are the poorest areas of the country? Where are the most affluent?
c) Are women who are poor less educated or literate than the men, compared with the the
national average or sub-region (urban, rural, other levels of government) average?
d) What is the proportion of poor women in informal sector employment compared to men,
and to the national average?
e) How does poverty vary across the nation, and are geographic differences similar for men
and women? Using a geographic parameter relevant for that nation (e.g. rural/urban,
grassland/desert), what proportion of women and men are poor, and then compare this to the
national average.
f) To measure crowding, what is the average size of households and number of children
across female and male headed households or family units?
g) What is the school attendance rate of poor young girls and boys compared to the
national or sub-regional rate?
h) What are the important differences between women and men who are poor and those who
are not poor – is it their level of education, their economic activity (or lack of it); their ethnic
group or another factor or even a combination of factors?

386. Advocacy in the areas of poverty and living conditions was most recently supported in
the Millennium Declaration, whose first stated goal is to eradicate extreme poverty and
hunger. However, the right to be protected from a life of poverty has been codified in
international law since 1948, with Articles 23 and 25 of the Universal Declaration of Human
Rights. Still, poverty in both ‘absolute’ and ‘overall’ types persists. It is also women in our
world, regardless of residing in a rich or poor country, who are more likely to be poor (United
Nations, 2010 a). Further, governments have acknowledged their commitment since the
Beijing Platform (1995b) to address 1) the burden of poverty that continues to fall upon
women, and 2) educational inequalities that underlie disparities in poverty. Still, poverty and
inequality patterned by sex persist.

387. Access to productive resources, particularly in rural areas where poverty is higher,
remains another important issue to gender equality. Rural women’s access to productive
resources – such as land, irrigation equipment, and other inputs necessary for cultivating their
own plots and earning their own money – remains a barrier to surmount to take care of family
and children’s needs. Women’s empowerment and economic independence are keys in the
fight against poverty.

388. Gendered analysis of poverty requires that statisticians, planners and economists remain
allied to idea that poverty data should be sex disaggregated to allow for an engendered
approach to poverty analysis. Also, women’s machineries need to be involved in the
coordinating bodies or steering committees to ensure that the definitions of poverty used, the
poverty data collected and the results meet their information needs for policy making,
monitoring, evaluation and advocacy. For example, engendered poverty analysis needs to
consider the time period the income statistics apply to, whether one year, one month or
another period of time. Is this period of time (or reference period) suitable to reflect the way

159
women generate income in the country considering things like irregular or sporadic economic
activity, income from part-time work, and income from informal sector activities?

389. In addition, gender poverty analysis should lobby for other supporting statistics about
resources available to the household (e.g. land, assets, tenure, utilities), because purely
quantitative measures like poverty lines have greater validity when complemented by other
measures of wellbeing or deprivation. As the Data section (i.e. E.3.) underscored, it is not
clear using census data how household income is allocated, spent or consumed, so gender
inequality is likely to be underestimated using household level data. Information on
household decision-making and resource allocation as collected in other household surveys,
such as the Demographic Health Surveys, would provide the requisite data to understand
women’s economic position vis-à-vis men at the household level.

160
Chapter 9:
Education and Literacy

1. What is it?

390. Population censuses directly collect educational information from individuals in three
primary areas: literacy, educational attainment and school attendance.
a. Literacy distinguishes those who have the ability to read and write as “literate,” and those
who do not have the ability to read and write as “illiterate.”
b. Educational attainment is defined as the highest grade completed, or alternatively as the
highest grade attended, in the educational system of the country where the education was
received. According to the International Standard Classification of Education (ISCED),
education includes all deliberate and systematic activities designed to meet learning needs.
c. School attendance is defined as regular attendance of any regular, accredited programme of
organised learning, either public or private, at the time of the census data collection, or
alternatively, during the last school year (United Nations, 2008 a). School attendance is
different from school enrollment in that it refers to children who are actually attending
school, rather than to children who have been registered as students at the beginning of the
school year. The latter is the basis for the enrollment statistics of the Ministries of Education,
which is often different from the data found in the census.

2. Why is it important?

391. Education is a key element in the analysis of gender issues. Educational indicators can
identify gender gaps in literacy, access to schooling, and in educational attainment. Because
educational status patterns both family roles and work roles, understanding how education
may be different for girls and boys, and for women and men in a given society also has
implications for other areas of administrative and policy concern. A gendered analysis then
considers how sex can serve as a primary and overall classification, and how disaggregating
the data by sex can help define and monitor an inequality in education. This type of analysis
is useful for individuals involved in public policy planning and implementation as well as
those associated with advocacy and equality for all, regardless of sex. Moreover,
understanding education through a gendered lens makes it possible to highlight, target, and
monitor inequality across men and women, and boys and girls.

392. Education is recognized and codified as a fundamental human right, and systematic
unequal access to it by sex may limit the position and life chances of some rights holders
when compared with others (United Nations, 2008 a). The Beijing Platform calls upon
governments to take action if there are inequalities and inadequacies in and unequal access to
education and training (United Nations, 1995). In even more specific terms, the MDG frame
gender equality in education as integral to economic development. For example, Goal 2 of
achieving universal primary education, can be assessed with two measures generally gathered
within census data, the literacy rate and school attendance. Goal 3 to promote gender equality
and empower women, also utilizes the educational indicators (i.e. the ratio of literate women
to men 15-to-24 years old, ratios of girls to boys in primary, secondary, and tertiary
education) to monitor progress. Despite the multiple international human rights instruments
that obligate nations to respect the right to education, 100 million children - at least 60 per
cent of them girls - do not have access to primary education and nearly two-thirds (64 per
cent) of the 774 million illiterate adults worldwide are women (United Nations, 2010 b –
UPDATE stats for final draft).

161
393. There is a strong link between education and gender equality; the global number of
illiterate adults has declined slightly during the past two decades, and increases in women’s
education have also been associated with many changes in women’s roles and positions in
society. Women’s increased literacy and educational attainment are related to individual
behaviours, such as later marriage and childbearing, fewer children over their lifetime, and
higher income. Women’s improved literacy and educational attainment are also related to
demographic trends at the societal level, such as lower fertility, and decreasing maternal
mortality and child mortality ratios, and social patterns such as greater formal labour force
participation and higher status for women overall in society. Educational and schooling
decisions are the basis upon which women negotiate family roles and work roles (Bianchi and
Spain, 1986). Examining educational factors through a gender lens highlights issues that may
affect or challenge women disproportionately in a given society as well as inform policy that
seeks to provide for and improve the wellbeing of all citizens.

394. Educational data are important to collect in a census because they are used to compute
adult literacy rates and average educational attainment of the population. Household sample
surveys are also sources. While not offering full coverage as in the census, surveys may be
more timely for policy-making purposes if it has been several years since the last census.
Education data also serve as components of several widely used, international indexes, such
as the Human Development Index (e.g. mean years of schooling of adults), Gender Inequality
Index (e.g. educational attainment), the Gender-Related Development Index (e.g. adult
literacy and school enrolment) or the World Economic Forum’s Global Gender Gap Index
(e.g. adult literacy and school enrolment). These data are important for policy-making and
planning purposes because they cover the entire population and can be used to identify areas
of need which can be better targetted for additional support.

3. Data issues

395. NSOs are recommended to collect census data on the core topics of literacy, school
attendance, educational attainment, and the optional topics of educational field and
educational qualifications. In addition, literacy status, school attendance and educational
attainment data should be collected and tabulated separately and independently of each other
without assuming any relationships among them (United Nations, 2008 a). A census usually
provides a proxy for literacy based upon assessment by the respondent or the household
head/informant, which can be expressed as a per cent literate or a literacy rate. This proxy
variable is less reliable than the procedure used in some surveys, where the respondent is
instructed to read a sentence or paragraph in common language about everyday events.

396. Across these educational topics, the challenge is to measure accurately the topic (i.e.
literacy) while also collecting data that can be used in international comparisons. As an
example, the literacy question currently varies across countries, so data may not always be
used for valid international comparisons. NSOs may consult the UNESCO Institute for
Statistics website, www.uis.unesco.org, for guidance. Also, Par. 2.215 of the
Recommendations from the International Standard Classification of Education (ISCED) (i.e.
http://www.uis.unesco.org/Education/Pages/international-standard-classification-of-
education.aspx) provides sample educational questions – especially in the areas of literacy,
school attendance and educational attainment (e.g. three levels of education, primary,
secondary and post-secondary) – in order to harmonize and standardize these measures for
international comparison (UNESCO, 2008). Further, any differences between national and

162
international definitions and classifications of education should be explained in the census
publications to facilitate comparison and analysis (United Nations, 2008 a).

397. Using meaningful tabulations represents another data challenge. First, literacy data
should be tabulated for all persons 10 years of age and over. Literacy cannot be accurately
computed from educational attainment, because persons may leave school with partial
literacy skills or lose them due to lack of practice. UNESCO recommends that literacy tests
should be administered to verify and improve the quality of literacy data. However,
administering a literacy test to all household members is practical and may affect
participation, hence limiting the utility of the results. Nonetheless, NSOs should evaluate and
report on the quality of literacy statistics published using census data. Second, educational
attainment should be classified as grades or years of education in primary, secondary and
post-secondary school. If the educational structure has changed over time, the data should be
coded or organized to make provisions for persons educated at a time when the national
educational system may have been different than the current system. Also, while educational
attainment is classified into seven levels, persons with no schooling should be included, and
adjustments should be made if necessary to accurately capture the situation of those who
were educated in another country. Finally, attendance at an educational institution should be
collected for all persons even thought it relates in particular to the population of official
school age, typically 5 to 29 years old, yet this range may vary depending on a country’s
national educational structure (United Nations, 2008 a).

398. Additionally, a distinction should be made between school attendance, which is

collected in the population census, and school enrolment, which typically is derived from
administrative data, such as school registration records. Because a child can be enrolled in
school and not necessarily be attending, results from censuses and administrative data may
differ. Further, because of this difference between “attending school” and “enrolled in
school,” these concepts should be clearly defined using internationally harmonized questions
as measures. This distinction between the administrative enrolment data and the census
attendance data illustrates how census data can serve as a useful complement to
administrative data in identifying those children in a population, who are enrolled but not
attending school (United Nations, 2008 a).

399. Finally, a major consideration for analysis is that educational indicators are usually
calculated by age-groups, to neutralize generation effects. In a country where access to
education is improving, basing assessment of school attainment on the older generations or
on indicators that mix different generations can be seriously misleading. Literacy, in
particular, is affected both by the fact that the percentage of illiterates increases sharply with
age and by the much larger number of women at higher ages. If the objective is simply to
quantify the number of illiterate males and females, this is not a problem, but in order to
assess the current performance of the school system in promoting equality between boys and
girls, the literacy rates of men and women aged 15-24 is a more appropriate measure. The
following graph illustrates the inequality profile by age for the case of the census of Malawi
(2008).

163
Figure 11: Malawi (2008) – Literacy by age and sex

100

90

80

70

60

50 Female
Male
40
Both Sexes
30

20

10

0
10- 15- 20- 25- 30- 35- 40- 45- 50- 55- 60- 65- 70- 75- 80- 85+
14 19 24 29 34 39 44 49 54 59 64 69 74 79 84

Source: Malawi. Gender in Malawi. Analytical Report 3 of the 2008 Census: Figure 5.4

4. Tabulations

400. The Principles and Recommendations for Population and Housing Censuses Rev. 2
(United Nations, 2008 a) recommend three tabulations to use in the analysis of educational
characteristics.

a) Population 10 years of age and over, by literacy, age group and sex. This tabulation
describes the rate of literacy by age group across females and males in society. From this
tabulation, shifts in fertility over age groups can be observed. To tabulate adult literacy
comparing across women and men, ten-year increments of age are suggested for each age
grouping beginning at age 15 (e.g. 15-24 years, 25-34 years, and so on).

b) Population, over 15 years of age not attending school, by educational attainment, age
group and sex. This tabulation describes the educational attainment, or highest grade
completed, of women and men by age-specific groups (e.g. 15-24 years, 25-34 years, and
so on) among those who are no longer attending school in the population. Educational
attainment may be presented in years of school or in other relevant groups, such as
primary, secondary or tertiary levels, for a society. This tabulation could also be scaled to
the age at the last year of primary school, age 11 or 12, in order to describe the proportion
of girls and boys, separately, who have completed primary school by the expected age for
a given society.

164
 c) Population 5 to 29 years of age, by school attendance, single years of age and sex.
This tabulation describes regular school attendance overall and in single years of age by
sex at the time of the census data collection or the last school year.

401. The following tabulations allow us to examine within group variability of women’s
differential outcomes and differences across men and women as they may be related to, or
even caused by, educational factors.

a) Female educational attainment (primary, secondary or tertiary level) by the marriage

rate, for age-specific cohorts;

b) Female educational attainment by number of children, for age-specific cohorts;

c) Educational attainment by median age of first marriage (if available) or SMAM, for
age-specific cohorts of women and men;

d) Educational attainment by labour market participation, for age-specific cohorts of

women and men.

402. Cross-classifying women’s literacy or educational attainment by whether women are in

wage employment illustrates how unequal access to education and later consequences, such
as lower income and labour market segregation, become apparent. In this way, unequal
access to education is an indicator of gender disparity as well as an underlying cause of its
persistence from one generation to the next, as constrained women then may offer more
limited educational life chances for their own girls.

5. Indicators

403. Minimum Gender Indicator Set

Computable from census data

Literacy rate of persons aged 15-24 years old, by sex;

Adjusted net enrolment ratio in primary education, by sex;
Gross enrolment ratio in secondary education, by sex;
Gross enrolment ratio in tertiary education, by sex;
Gender parity index in enrolment at primary, secondary and tertiary levels
Graduates from lower secondary education, by sex
Education attainment of population aged 25 and over, by sex

Actually, the census measures attendance, not enrollment.

Not computable from census data (??)

Share of female science, engineering, manufacturing and construction graduates at tertiary

level
Proportion of females among third-level teachers or professors
Net intake in first grade of primary education, by sex B.1
Primary education completion rate, by sex B.1
Transition rate to secondary education, by sex B.1

165
404. Literacy, educational attainment and school attendance provide three important
indicators, each in its own right, to measure access to education and progress toward the
Millennium Development Goal 2, achieving universal primary education. Goal two of the
Millennium Declaration (United Nations, 2008 a), to achieve universal primary education,
and the Beijing Platform (United Nations, 1995), to advocate equality in opportunity across
women and men, suggest several indicators that can be computed using census data.

a) Ratio of literate women to men. The literacy rate discussed in the tabulations section
can be used to calculate the ratio of literate women to men. Given a balanced population
of both women and men, if the overall literacy rate for persons is .85 (i.e. meaning 85 per
cent of the population can read and write), but the ratio of literate women to men is .75,
then 7.5 women compared with every 10 men in the population are literate. Government
services and advocacy groups may work in tandem to target the increased school
attendance and literacy of girls. In this way, this one statistic becomes meaningful as a
benchmark to raise awareness and provide an impetus for change.

From this benchmark, administrators, public policy makers and advocacy groups may
enact programmes to promote girls’ educational equality and well being. This benchmark
literacy statistic can then be recalculated using census data five or ten years later, or
household sample surveys collecting data on literacy, to inform progress on the issue of
women’s literacy by itself, and also as a component of the overall literacy rate. This
statistic can be calculated as an overall rate, but it should also be calculated for age-
specific cohorts, so that women 15-to-24 years of age can be compared with women 25-
to-34 years of age when the next census enumeration is conducted ten years, and so on.
This statistic becomes even more meaningful if there is a policy change, as the population
can be divided into cohorts corresponding to the time of the policy change. This is just
one example of the usefulness of a gendered analysis of the literacy statistic.

[Insert “Example, Senegal” box here with case of Senegal, showing longitudinal progress.]

b) Ratio of primary school educational attainment for girls and boys. The census question
on educational attainment (i.e. the highest grade completed) can be used to compute the
ratio of girls to boys who have completed primary school. If the primary school
completion rate (computed above in the tabulations section) in the age range of 12-15
is .50 for girls and .60 for boys, then the ratio of primary school educational attainment
for girls and boys (12-to-15-years old) is .5/.6 or .833. Just over eight girls (i.e. .833) for
every 10 boys in this age range have finished primary school education in the population.

In this way, this measure may be used as an indicator of Millennium Development Goal 2
(i.e. achieving universal primary education for children). This measure should be
specified so that it is age appropriate. The results of one census enumeration can be used
to evaluate the longitudinal progress toward closing a gender gap in primary education by
computing this indicator for age-specific groups (e.g. 10-14 years, 15-19 years) within the
population, and then comparing the oldest generation to the younger ones. If the ratio of
girls to boys and women to men completing primary school increases over time, this
measure indicates gendered progress in education or even a “catching up” in education as
girls over the official age respond to government policies encouraging school attendance
for girls.

166
 Additionally, the educational attainment question may be used to create other educational
completion ratios, for example at the secondary and tertiary post-secondary levels, to
examine difference by sex. This measure can then be computed over time and compared
across census enumerations or across cohorts within the same census enumeration to
provide a longitudinal measure indicating progress over time.

c. “Out-of-school” girls and boys. The census question on school attendance can be used
to measure the percentage of “out-of-school” or “ever-in-school” children, and this can be
disaggregated for girls and boys. This indicator is the complement of the currently
attending school measure presented in the tabulations section.

Within the context of the UNESCO Education for All Goal 2, the number of out-of-
school children has grown in public awareness (United Nations, 2008 a). This measure
can be computed from one decennial census to provide a snapshot description of the
issue, and it can be used to monitor progress or the effectiveness of a policy
implementation over time by comparing to household survey data asking about school
attendance and enrolment, and long-term progress can be monitored from census
enumeration to census enumeration.

Figure 12: Regional distribution of minority population and middle school

attendance rate (Gansu Province, China), Population and Housing Census 2000

Source: Feng Jing (2005), cited in Cao and Lei (2008), Figure 4

The following is a relatively simple example of how spatial data can be used to elucidate
the relationship between education and gender. It is from a study by Cao and Lei (2008),
on female attendance of primary and middle school education in Gansu Province, China.
The data for this study came from the 2000 Population and Housing Census,
complemented by the 2000 Education Census, which provided information on
characteristics such as the density of schools for each of the 60 counties of the Province.
A multivariate analysis of these data, using the county as unit of analysis, reveals that the

167
primary determinants of female non-attendance at the middle school level are: a) The
illiteracy rate of women over age 15 (suggesting that girls from families where the mother
is illiterate are less likely to attend school themselves); b) The percentage of the
population consisting of ethnic minorities (see the graph below); c) Poverty level of the
county; and d) Density of schools. Other variables, such as the proportion of non-
agricultural population and rural income, proved to be less important.

The study could be detailed in a number of ways. In its present form, it only looks at
female non-attendance and does not compare this with the non-attendance of boys. From
a gender perspective, this is a limitation; it would be better if the dependent variable were
some measure of differential non-attendance between boys and girls. Also, note that the
use of spatial information is limited to the choice of the county as the unit of analysis, but
no attempt is made to relate the characteristics of one county to others, in the
neighbourhood. It may be that the county level is too aggregated for this kind of analysis,
but in a more detailed study of the geography of non-attendance one would want to
consider interactions such as the influence of a high concentration of schools in one
geographical unit with the non-attendance in other units nearby.

d. Gender Parity Index (GPI) to measure parity in education. The Gender Parity Index
(GPI) can be computed and included as an additional statistic in a table to provide a
gendered analysis of the literacy rate, school enrollment, or school attendance. The ratio
of female to male is interpreted in the same way by group, whether examining literacy,
the proportion having completed primary or second school, or the percentage currently
attending school.

For a given measure, the GPI is calculated as the ratio of the value for females to that for
males. A value distinctly less than one indicates disparity in favour of men or boys,
whereas a value distinctly greater than one indicates disparity in favour of women or
girls. For example, a GPI for literacy close to 1.00, between 0.97 and 1.03, indicates
parity in literacy of a specified age group (UNESCO, 2006). Note that in the case of
enrollment (or attendance) at the primary, secondary or tertiary level, the GPI is
computed based on the Gross Enrollment (or Attendance) Ratios of each sex and not the
raw numbers of boys and girls enrolled (or attending). This is to correct for the fact that
the base populations of boys and girls of school age may be different. The resulting index
may, however, still be biased if the repetition rates of one sex are markedly higher than
the other (see the explanation in Section C of Chapter 2).

For primary and secondary education attendance, the GPI is still below 0.95 in Sub-
Saharan Africa overall, especially evident in countries like Chad (GPI=0.48, 1996 census
data), Guinea (GPI=0.55, 1997), Niger (GPI=0.60, 1996), the Democratic Republic of
Congo (GPI=0.66, 1994), and Sierra Leone (GPI=0.67, 1990). It is also marked low in
other countries such as Afghanistan (GPI=0.66, 2008), Turkey (GPI=0.83, 1996),
Pakistan (GPI=0.83, 2008), Iraq (GPI=0.84, 2005) and Papua New Guinea (GPI=0.84,
2006). However, there are increasingly more countries with GPIs larger than 1.03, such as
Iran (GPI=1.40, 2008), Bangladesh (GPI=1.07, 2007), Venezuela (GPI=1.05, 1990),
Namibia (GPI=1.05, 1997), as well as some Caribbean countries. In these countries, more
girls than boys are attending primary and secondary education, which may suggest
several things: 1) a gender disparity for boys’ educational opportunities; 2) boys dropping
out because of better economic opportunities; 3) the fact that girls need more education
than boys to be competitive in the labour market; 4) educational systems playing ‘catch

168
up’ to enrol often larger numbers of girls who did not enter school at the correct age, etc.
This cluster of countries reinforces the need and usefulness to undertake a gendered
analysis of data, so that this type of analysis is mainstreamed and used to provide equal
opportunities regardless of gender to both women and men, both girls and boys.

Additionally, the Gender Parity Index (GPI) can be calculated over time as a useful
measure of progress toward gender parity on a specific educational measure. Table 32
provides a longitudinal use of the Gender Parity Index using primary net enrolment rates
by region of the world and shows marked increases in gender parity for the world in
general, changing from 0.93 to 0.97 over the period 1999 through 2007. This increase was
fueled by less developed regions, which collectively had an increase from 0.92 to 0.97
over this eight-year period, while more developed regions remained at parity in terms of
the primary school net enrolment rate. Table 32 shows the GPI longitudinally by region,
but this same statistic could be calculated at the sub-country level and compared with the
country level, or calculated at the country level and then compared with neighboring
countries, a regional cluster of countries, the continent or the world overall.

Table 32: Gender Parity Index (GPI) based on primary net enrolment rates by
region, 1999 and 2007

1999 2007
World 0.93 0.97
Less developed regions 0.92 0.97
More developed regions 1.00 1.00
Africa 0.89 0.93
Eastern Africa 0.92 0.98
Middle Africa 0.86 0.86
Northern Africa 0.92 0.94
Southern Africa 1.02 1.01
Western Africa 0.81 0.88
Asia 0.93 0.97
Eastern Asia 1.01 1.01
South-Central Asia 0.85 0.96
South-Eastern Asia 0.97 0.99
Western Asia 0.90 0.93
Europe 0.99 1.00
Eastern Europe 0.99 1.00
Northern Europe 1.00 1.01
Southern Europe 0.99 0.99
Western Europe 1.00 1.00
L. America & Caribbean 0.98 1.00
Caribbean 0.99 0.98
Central America 1.00 0.99
South America 0.97 1.00
Northern America 1.00 1.01
Oceania 0.98 0.97

Source: Taken from The World’s Women 2010 using data from the UNESCO Institute for Statistics (2009
a).

169
 Note: Net enrolment rates are not calculated from censuses, so a better example should be placed here in the
next revision.

The GPI should be used to compare rates for girls and boys, such as school attendance
rates, and not absolute numbers, such as the numbers of girls and boys in primary
education. Due to the sex ratio at birth there are generally more boys than girls of school-
age, so calculating the GPI for an absolute number would be artificially underestimated.

405. For monitoring of the MDGs, the GPI of the gross enrolment ratios (explain the
difference between gross and net enrolment ratios) is generally monitored (and not the net
rates). Care should be taken to do the same with attendance rates, because it is more
meaningful to monitor gender parity for all children in school (i.e. gross rates) and not just
those who are of the correct age for the given level of education (i.e. net rates). Also, data on
school attendance can be used to compute net attendance rates, thus complementing net
enrolment rates. Attendance is the main indicator used to monitor the universal primary
education Millennium Development Goal. Similarly, gender parity indices for attendance of
primary, secondary and tertiary education can be used in this same way.

406. Other indicators, that are not part of the MDG framework include the average age at
which boys and girls enter primary school, the proportions of boys and girls that finish
primary education and the average time that it takes them to do so. It has been found, for
example, that in many countries girls tend to enter primary school later than boys, but that,
once they are in school, they finish quicker than boys do (UNESCO, 2010). In the absence of
direct information on the ages at which boys and girls enter school, a synthetic mean age can
be computed based on the information on ever having been in school. The computation of
this synthetic mean age is analogous to the computation of the Singulate Mean Age at
Marriage (SMAM) described in Chapter 5.

6. Multivariate and further gender analyses

Go over this section and see how to make it more operational. Many examples in this section
(here and in other chapters) are vague references to things that people have investigated with
data of different kinds and that may or may not be replicable using country census data. If
they are replicable, it has to be explained how this is done; if not, the examples shouldn't be
here. Also add your own multivariate education example.

407. Education was found to have a positive effect on egalitarian attitudes in data from
China. Shu (2004) finds that education influences gender attitudes in multiple ways at both
the micro- and macro levels. Better-educated individuals hold more egalitarian gender
attitudes, and this positive effect of individual education is larger for women than for men,
indicating a strong empowerment effect of education for women. Shu also finds a trickle-
down effect of egalitarianism through education at the community level; individuals in
communities with high education at the community level are socialized toward more
egalitarian attitudes. In this way, education is a useful tool to spread egalitarian gender
attitudes, which may be useful for advocates interested in increasing women’s position. This
example highlights the usefulness of educational measures can be conceived as causal factors
that shape progress on women’s issues.

408. Women’s education and family formation. Mother’s increased education is correlated
with many factors such as lower teen pregnancy, later ages of childbearing, lower fertility,

170
lower rates of marriage in some cases and delayed marriage in others. In Cameroon
(Eloudou-Enyegue, 2004), girls drop out of school after becoming pregnant, while in Brazil
(Chagas de Almeida and Aquino, 2009) mother’s lower education explains her daughter’s
increased likelihood of teen pregnancy. Using census data over thirty years in the US, Bianchi
and Spain (1986) show increased education is associated with the rise in age of first marriage,
decline in fertility and increased labour market participation overall. These studies point to
several tabulations using census data as enumerated just below. In all of these studies,
educational attainment is an independent variable that shapes an outcome or dependent
variable, such as age of marriage, teen pregnancy, fertility, or labour market participation.

409. Maternal education and child health. Correlations exist between maternal education and
markers of child health – such as infant mortality and immunization status – yet a causal
relationship is not firmly established. Desai and Alva (1998) explain that mother’s education
acts as a proxy for the economic level of the family and geographic area of residence. In their
multivariate analysis, they find that the education effect on infant mortality is lessened when
husband’s education, access to piped water and toilet are included as control variables.
However, they find that maternal education remains statistically significant as a predictor of
children’s immunization status, net of control variables. Similarly, Hobcraft (1993) finds that
more educated women are likely to have initiated immunization and have completed
vaccination of their children, compared with less educated women. In this way, the health
benefit of increased maternal education for children’s improved health or reduced mortality
seems to be mediated through the household’s economic status and possibly immunizations.
These studies suggest that NSOs can monitor the effect of maternal education on children’s
health status over time.

410. Another study using data from India (Kravdal, 2004), the average education of women
in a census numeration area has a strong impact on child mortality independently of the effect
of the mother’s own education. This finding speaks to a community education effect, similar
to a mother’s individual education, which translates into the use of maternity and other
preventive health services, increased child’s nutrition and effective care of a sick child by the
mother. This finding builds on earlier research establishing a negative relationship between
increased mother’s education and decreased child mortality in Latin America (Haines and
Avery, 1978), Africa (Caldwell, 1979) and Asia (Cochrane, 1980). The strength of the study
from India is that it augments regional survey data with census data, and as a result is able to
test for the relative effects of women’s education at both the individual and community
levels.

411. The preceding studies point to the usefulness of both cross-tabulations to establish
relationships, and to multivariate analysis to disentangle the effects of education on an
outcome variable. The type of multivariate analysis is then determined by the nature of the
variables of interest.

Cross-tabulate education of the wife with that of the husband. This has important implications
for divorce rates in some countries (says Eduard), although it is not clear how you would
measure the latter using only census data.

7. Interpretation, Policy and Advocacy

412. Awareness campaigns – such as billboards, radio and television – can be used to
sensitize the population about girls’ school enrolment in geographic areas where girls are not

171
enroled in the same numbers as boys. Development strategies focused on keeping girls in
school need to address the diverse constraints (e.g. fetching water and gathering wood in
many rural areas) pushing the mothers to take out their daughters from school. If
development strategies can alleviate women’s work burden, through access to labour saving
equipment, then girls’ educational prospects stand to improve.

172
Chapter 10:
Work, Economic Activities and Social Protection
1. What is it?

413. Work refers to the participation of individuals in economic activities, both paid and
unpaid production for a period longer than one-hour during the reference period, and which
delineates employment from unemployment. The population census directly collects data on
the economic activities and characteristics of individuals in the areas of employment status,
type of work, and the sources and/or amounts of income received by individuals and/or
households (United Nations, 2008 a).

414. The economically active population, or equivalently the formal labour force, is made up
of all persons above the minimum age specified, generally 15 years of age or older, who
report being available to furnish the supply of labour for the production of goods and services
during a short reference period (e.g. one week, one day), and are usually registered as being
“unemployed.” The economically active population can be categorized into two groups:
employed and unemployed.

a) Employed comprises all persons of a specified age, generally 15 and over, who work
for pay, profit or family gain in cash or kind or who do nonpaid work for a family
business or farm. If persons are temporarily absent during the reference period, they
are considered to be employed??????.

b) Unemployed comprises all individuals above a specified age, generally 15, who do not
work, are available for work, and who did actively seek work during a reference
period.

415. The group, inactive or not currently active persons or persons not in the labour force, is
comprised of persons not fitting the categories of employed or unemployed during the
reference period, and further, those who are outside of the specified age defined for the
economically active population. A person may be economically inactive because one is in
school, training or college, engaging in household duties, retired or aged out of the labour
force, or sick, infirm or disabled. Women are more likely than men to fall into this category,
as they typically produce services – such as preparing and serving meals or caring for
children, the sick and elderly within a household – that are consumed by the household.

416. Occupation is defined as the type of work done in a job by employed persons. It is
recommended that NSOs collect information on occupation in accordance with the latest
revision available of the International Standard Classification of Occupations (ISCO) at the
following URL: http://www.ilo.org/public/english/bureau/stat/isco/index.htm (ILO 2010).

417. Industry refers to the kind of production or economic activity where persons work. It is
recommended to collect information on industry in accordance with the most recent revision
of the International Standard Industrial Classification of All Economic Activities (ISIC)
(United Nations, 2008). Important gender differences are observed in terms of economic
sector. Women are predominantly and increasingly employed in the service sector and the
relative importance of the industrial sector as a source of employment for women continued
to decline in the last two decades in all regions (United Nations, 2010 b). This preponderance
of women in service work may be due to the general development of service work as women

173
have entered formal sector employment en masse after 1960, or it could also be linked to
industry or employer discrimination on the one hand, or to a gender preference of women to
work in the service sector. For example, this preference could be driven by social
expectations for what types of work are acceptable for women, or by different childhood
socialisation by sex (i.e. caretaking among girls who then are represented in caretaking
professions such as nursing and social work in many societies). By nation, these factors may
differ and are potential engendered analysis points.

418. Informal sector work includes persons who are employed in a reference period in the
untaxed and unregulated part of an economy [not the official definition]. The informal sector
economic activity is not included in the gross national product (GNP) [not true: it should be
included and often is], yet it comprises more than one-half of the economically active
population in low-income countries and is present in middle- and high-income countries, too
(International Labour Organization, 2005).

419. Status in employment describes the type of explicit or implicit employment contract
[self-employed people (and others) do not have contracts] the economically active person
has. It is recommended that the economically active population be classified by employment
status into the following groups: 1) Employees, 2) Employers, 3) Own-account workers, 4)
Contributing family workers, 5) Members of producers' cooperatives, and 6) Persons not
classifiable by status.

420. Time worked is a measure that typically asks the number of hours per week employed
persons work. ‘Time worked’ data help to ensure a more accurate measurement of the
relative contributions of men and women to the workforce and identify gender gaps. Time
worked is the total time actually spent producing goods and services, within regular working
hours and as overtime, during the reference period adopted for economic activity in the
census. It is recommended that the reference period is short, for example, the week preceding
the census. If the reference period is long, for example, the 12 months preceding the census,
time worked should be measured in larger units such as weeks. If a person has more than one
job during the reference period, it is recommended to record both total time worked from all
jobs and the time worked in the main job for which occupation is being registered. In many
countries, especially in Europe, women having young children may choose or may have
limited options other than to work part-time. Part-time employment is much more common
among women than among men globally, with the prevalence rate for women exceeding
twice that for men in about three quarters of the countries (United Nations, 2010 a).

421. These higher participation rates of women in part-time work are due in part to the
difficulties women experience in combining family responsibilities and work life. In those
countries where child care services are very expensive (i.e. as costly as a salary earned by the
woman), or poor quality and/or not widely available, it is difficult for women to work full-
time in the formal labour market. It is not necessarily a “free choice” for women, but rather
women’s participation in the labour market while also juggling family responsibilities and
perhaps even a limited child care and support structure. Where affordable, quality, full-time
child care is accessible, women typically work full-time (e.g. Sweden, Norway). For
example, the Netherlands has the highest percentage of women working part-time, which is a
response to an environment of limited, affordable, full-time child care. This is a gender
concern.

422. Full-time and part-time are defined as follows:

174
 a) Full-time work typically refers to working at least 35 hours in a week [is this the
internationally agreed definition ? check with ILO].

b) Part-time work refers to a national-level definition of work that is less than a full-
time threshold and more than one hour in a typical week. [if 35 hours is the limit, why
not say that part time is less than 35 hours ?]

423. The institutional sector of employment describes the legal organization and primary
functions, behaviour and objectives of the enterprise that are associated with a job as defined
by the System of National Accounts (SNA) as the following categories: 1) corporation, 2)
general government, 3) non-profit institutions serving households (e.g. churches, cultural and
sports clubs and charitable institutions), and 4) households.

424. Place of work is the location of an employed person’s main job and typically takes the
following categories: 1) work at home, 2) no fixed place of work, and 3) with a fixed place of
work outside the home. However, the place of work categories may differ by nation
determined by local relevancy (e.g. categories for Serbia include 1) at home, 2) on the farm,
3) no fixed place of work, 4) abroad, and 5) fixed place outside the home). In the 2011 census
of Albania, 2) and 5) were merged and 4) was disaggregated to identify neighbouring
countries. Mauritius (2011) distinguishes between those working at a fixed place outside the
home, those without a fixed place of work, those working at home and those working outside
the country. Swizerland (2010) distinguishes between those working from home, at a fixed
place of work and those working at varying locations. The Costa Rican census actually tries
to establish how far away the place of work is from the home.

425. The right to social protection is enshrined in Article 22 of the Universal Declaration of
Human Rights (United Nations, 1948) which states that every human being has the right to
social security. Article 25 protects the right to security in the event of unemployment,
sickness, disability, widowhood, old age or other lack of livelihood in circumstances beyond
an individual’s control. Social security has two main dimensions, specifically “income
security” and the “availability of medical care.” The preface to the World Social Security
Report 2010/11 (ILO, 2010 b) notes that while some level of social security protection exists
in all countries, just one-third of countries and 28 per cent of the world’s population have
comprehensive social protection systems covering all branches of social security defined in
the Social Security (Minimum Standards) Convention (No. 102) (ILO, 1952), and about one-
third have no access to any health facilities or services at all. Statutory unemployment social
security protection exists in just 78 countries (or 42 per cent).

426. Of persons over age 65 in high-income countries 75 per cent receive pension benefits,
compared with less than 20 per cent of those in low-income countries (ILO, 2010 b). In less
developed regions, old-age pension or social insurance coverage is found in the formal and
public sectors, and women are often not covered because they are less likely to be employed
in these sectors than men. Informal employment is the main source of work for women
outside agriculture in these regions (United Nations, 2010 a). As a result, most of the world’s
elderly, a majority of whom are women, depend either on social security systems or the
support of their children in order to stay out of poverty.

427. Although the social security concept is not intrinsically linked to work or employment,
most social security systems in the world are organized around employment. Being or having

175
been employed, usually only in the formal sector, is often a precondition for becoming
eligible for benefits, e.g. health benefits for employees and their families, old age pensions
based on pension contributions accumulated during one’s working life, or unemployment
benefits based on previous income and the number of years worked prior to dismissal.
Typically, employment in the formal sector is also the precondition for maternity benefits.
Some countries (New Zealand, Namibia, several European countries), however, have a flat-
rate minimum old age pension benefit based on the number of years of residence in the
country, rather than the number of years worked.

428. The Principles and Recommendations for Population and Housing Censuses, Rev. 2
(2008 a) do not provide specific recommendations on gathering and analysing data on social
security, yet many countries gather data relating to employment, receipt of social security,
contributions to social security, and health insurance coverage. Of the 99 countries examined
within the most recent, long form census round (2005-14) as presented in Appendix X, 55.6
per cent of countries ask whether respondents are currently working or employed, 13 per cent
ask about receipt of social security, 4 per cent ask about contributions to social security, and
12.1 per cent ask whether the respondents have health insurance. Update this information.

2. Why is it important?

429. Women’s economic independence through employment and productive resources is

protected within the CEDAW (United Nations, 1979). In addition, the Beijing Platform (i.e.
Fourth World Conference on Women) issued a commitment for women’s full economic
independence and access to work and productive resources, and the Millennium Declaration
linked full and productive employment and decent work for both women and men as a
necessary step to promote economic development and do away with poverty and hunger.

430. The Fourth World Conference on Women (United Nations, 1995) commits to promote
women's economic independence, specifically through employment and access to productive
resources and opportunities, as well as through the coordination of work and family
responsibilities for women and men. Further, the Millennium Declaration commits to full and
productive employment and decent work for all, both men and women, as part of Goal 1, to
eradicate extreme poverty and hunger. And legally binding because most countries have
ratified it, the Convention on the Elimination of Discrimination Against Women (CEDAW)
(1979) protects women’s access to work.

431. Statistics on the economic characteristics of employed, unemployed and inactive persons
gathered in the population census can be used in combination with other demographic and
social factors (e.g. education, household and dwelling information) to construct a
comprehensive picture of the socio-economic situation. Economic data gathered in the
population census also provide baseline information from which survey data and
administrative data can be related and evaluated.

432. There is a marked difference, 25 percentage points, in the global labour force
participation rate between men and women in 2010; 77 per cent of men participate in the
labour force compared with 52 per cent of women (United Nations, 2010 a). In addition to
this large gap in participation, which has persisted for 20 years, there is considerable
occupational segregation and a gender wage gap that continue to persist in all regions of the
world. Further, women increasingly are employed in the services sector and within informal
economic activities, which generally commands less in remuneration than the industrial

176
sector (United Nations, 2010 a). In some countries, e.g. in Central America, Colombia and
Venezuela, average salaries for women are the same or even higher than those for men, but
non-salary incomes are substantially lower, due to the informal natural of much of the work
carried out by women (ECLAC, 2011). Beyond sector differences on how women and men
are engaged, there remains a need to examine the different conditions of employment for
women and men. Working from home may allow women who would not otherwise have the
opportunity to be employed to exercise an economic activity, but if more women work from
home than men, it can reinforce the marginalization and devaluation of female work (Chant
and Pedwell, 2008).

433. Adding to this, women actually work longer hours on average than men when both paid
and unpaid work are considered. Women spend at least twice as much time as men in unpaid
domestic work (Bianchi et al., 2000). The number of hours that women spend on housework
and community and volunteer work exceeds those spent by men for the same purposes. The
average number of hours per day women spend in unpaid housework and community
activities ranges from around three (in Denmark) to over six (in Turkey) (United Nations,
2010 a). At the same time, men spend less than one hour on these activities in several
countries such as Cambodia and Pakistan. This is important, because women’s unpaid
domestic and caretaking work is considered household consumption and not counted as being
employed as part of the economically active population. Because unpaid work mainly done
by women is not included in the national accounting systems, it goes under-estimated.
However, understanding this under-estimated household work situation through census
questions (i.e. whether water is fetched from a well or wood collected for cooking compare to
less labour-intensive alternatives) allows us to understand women’s roles, opportunities and
constraints, and overall life chances compared with men. Finally, women who do not
participate in the labour market depend on the resources provided by their spouse or partner
or another relative, and therefore are more vulnerable in case of widowhood or separation
(United Nations, 2010 a).

434. Social security protection mechanisms are important for women because they are more
likely than men to live longer and have more interruptions in their work than men. The latter
may result in lower benefits during retirement, especially in systems with a tight link between
contributions and benefits (Ståhlberg et al., 2008). In such systems, women who do full-time
housekeeping and child rearing work are left vulnerable. In more developed countries,
women rely on survivor benefits through their husband’s work, but in less developed
countries there is often no benefit. Even in developed countries, women often earn less than
men and spend more time out of the work force due to caregiving responsibilities, which
results in “gaps” in their earnings histories referred to as “the cost of caring.” At the birth of a
child, maternity leave is an important social protection for the mother as well as the child and
others living in the household. When women take time off from formal employment to care
for children and the elderly, it is again a situation when women are more likely than men not
to earn income or save for the future. Retired women generally have less access to other
sources of income, such as pensions, assets and earnings, and hence rely on social security
more than men to avoid poverty in older age (Hayes, Hartmann and Lee, 2010). As a result of
women marrying older spouses and living longer than men, they are more likely to be
widowed and then must rely on old-age pensions and social security payments for longer
periods than men. Women are also more likely to deplete their financial resources caring for a
spouse. In all, these demographic patterns may place women in a vulnerable economic
situation (Hayes, Hartmann and Lee, 2010).

177
435. Systems based on years of residence rather than work, which generally imply substantial
transfers from contributors to non-contributors, are particularly beneficial to women who do
not have a long employment history. For example, in Japan the national pension scheme
implemented in the 1960s succeeded in expanding pension insurance to more than 18 million
previously uninsured workers, the majority of whom were women (Chen, Jhabvala and Lund,
2002). ILO (2010) maintains that a big challenge of social protection is extending maternity
benefits. At present, cash benefits before and after birth are limited to formal sector
employees, and there are huge differences in the amount of benefit across rich and poor
countries, and across rural and urban areas. Adding to this, while less than 35 per cent of
women in rural areas are estimated to have access to professional health services in low-
income countries, about 70 per cent of urban women have access. However, this higher rate is
still substantially less, over 20 per cent lower, than health services access in high-income
countries (ILO, 2010 b).

436. Access to social security is considered requisite for gender equality. The Plan of Action
of the first World Conference on Women (1975), for instance, explicitly referred to the need
of ensuring women the provision of social security protection, equal in all aspects to that of
men. Other human rights treaties, such as the ICPD Programme of Action, have also
expressed concern with the increasing numbers and proportions of elderly people in the world
and emphasized the need of developing and ensuring quality systems of economic and social
security in old age.

3. Data issues

437. The way work is defined and measured is crucial to measuring how employment and
income opportunities may be different by sex. Men are more often employed outside the
house with a paid job, while women do most of the unpaid domestic work and are more likely
to be misclassified as homemakers when only basic questions are asked. Because women are
generally engaged in homemaking duties, or cultural perceptions relating to sex roles on the
part of the respondent or the enumerator, women’s economic activity status may be
misclassified and therefore under-counted. Indeed, there remains a need to collect data on
time-use to capture time spent by men and women on paid and unpaid work, both in the home
and the labour market.

Kim raised the issue of village work as a category of work.

438. Occupational data should be coded at the lowest possible level of the classification in
order to identify gender differences in occupation. For example, the ISCO 08 (ILO, 2010 a)
classifies "Health professionals" under item 22, but differentiate at the level of three digits
“Medical doctors” (item 221), “Nursing and midwifery professionals” (222), “Traditional and
complementary medicine professionals” (223), “Paramedical practitioners” (224),
“Veterinarians” (225), and “other health professionals” (226). While all of these categories
are health professionals, a significantly greater number of women are employed as nurses
compared with men. Failure to collect specific categories may result in categories that are too
broad to capture the differences in prevalence of men and women within a particular
occupational category, and further risks conflating two categories which may command
vastly different incomes and statuses. 1. You can go up to the fourth digit; 2. How feasible is
it to do that in censuses ? 3. We are not really trying to give advice to countries on how they
should design their census forms.

178
439. Comparisons between the wages of men and women are often made in weekly or
monthly terms, without considering that part-time work is more common among women. In
order to make meaningful comparisons, wages have to be computed on the basis of the
number of hours worked. Additionally, income data such as self-employment, property
income and non-cash or in kind income can be difficult to collect in population censuses.
Because income is typically more easily gathered in a sample survey of households,
depending on the national requirements countries may ask limited information on cash
income. This paragraph is out of place

440. Because the census data for institutional sector of employment are captured with pre-
coded alternatives, borderline classifications may be categorized according to the subjective
understanding of the respondent rather than the intended distinctions. Any analysis and
resulting statistics should keep this in mind as a possible limitation of the data.

441. Even though the Principles and Recommendations do not make specific
recommendations on the analysis of data relating to social security as defined by the ILO,
many countries gather these data (e.g. currently working or employed, receipt of social
security, contributions to social security, and health insurance coverage). A question to
capture maternity benefit or loss of work after the birth of a child may provide a useful
measure of women’s status within a country. If this question is repeated longitudinally,
cohorts can be followed over time. Alternatively, cohorts of women by age and by number of
children including the last birth could be constructed with data from one census
administration to examine loss of work due to the birth of a child or a critical number of
children.

4. Tabulations

442. The Principles and Recommendations for Population and Housing Censuses Rev 2
(United Nations, 2008 a) recommend several tabulations to describe how persons’ work and
economic lives may be shaped differently by gender. The usual or current activity status is
the basis for these tabulations, yet the activity rates used to monitor labour force participation
exclude unpaid work, such as own-account production or caring for children and the elderly
[careful: own-account production is employment, caring for children is not]. Hence, women’s
labour force participation is lower and likely to be underestimated compared with that of men
at all stages of the life cycle.

P7.1-R Population ... years of age and over, by usual (or current) activity status,
educational attainment, age and sex*
P7.2-R Usually (or currently) active population by activity status, main occupation,
age and sex*
P7.3-R Usually (or currently) active population by activity status, main industry, age
and sex*
P7.4-R Usually (or currently) active population by activity status, main status in
employment, age and sex*
P7.5-R Usually (or currently) active population by activity status, main status in
employment, main industry and sex
P7.6-R Usually (or currently) active population by activity status, main status in
employment, main occupation and sex
P7.7-R Usually (or currently) active population by activity status, main industry, main
occupation and sex*

179
 P7.8-R Population not usually (or currently) active, by functional categories, age and
sex*

443. As discussed above, the interpretation of these tables is likely to show a lower labour
market participation of women compared with men. Additional tabs might explore whether
the gender gap in labour market participation is explained by educational attainment. An
engendered perspective would ask: ‘Do women and men at similar levels of educational
attainment have similar labour market participation rates?’ Are women and men at similar
levels of educational attainment represented in the same occupations or industries at similar
rates? Are women and men at similar levels of educational attainment represented in the same
functional categories (e.g. management, factory line) within work organisations. Some gaps
in labour market participation may be the result of women being less educated compared with
men, or alternatively, there may be cultural reasons (e.g. discrimination, social norms, laws
limiting women’s work) that limit women’s participation at the same rate as of men. For
many women, it is not a choice.

444. As mentioned above, some censuses, such as the Australian and Canadian censuses of
2006 and the Korean censuses of 2005 and 2010, have specific questions on care-giving
activities, including child care. The 2005 census of the Republic of Korea established, for
instance, that 5.4 per cent of children between ages 0 and 12 are cared for by their
grandparents during the daytime. Australia asked questions on unpaid domestic work carried
out and care given to family members or others because of disability, a long term illness or
problems related to age. With respect to child care, it asked: “In the last two weeks did the
person spend time looking after a child, without pay? The data results, which were
computed at the national level and excluding cases where the information was not stated or
not applicable, were as follows:

Table 33: Australia (2006) - Percentage of men and women who spent time providing
unpaid child care

Type of unpaid child Men Women

care provided
Own child(ren) 19.7 22.9
Other child(ren) 5.3 10.0
Both own and other 0.6 1.6

Source: Australian Bureau of Statistics CDATA Online (2006 Census)

In those rare cases where this kind of information is available from the census, it can be used
to fine-tune either of the two methods outlined above.

445. Some potential tabulations with a gendered perspective regarding social security
include:
a) Participation in the labour market of persons 65 years and over, by sex and age (in five-
year increments);
b) Social security or old-age pension receipt of persons 65 years and over, by sex and age
(in five-year increments);
c) Health insurance receipt of persons 65 years and over, by sex and age (in five-year
increments);

180
 d) Contributions to social security or an old-age pension of persons 15 years and over, by
sex and five-year increments of age; and
e) Health insurance status of persons 15 years and over, by sex and five-year increments
of age.

446. Additional factors, such as race, ethnicity, marital status or rural/urban status, may also
be relevant dependent on the composition of the national population and should be added to
the above tabulations when relevant.

5. Indicators

447. Minimum Set of Gender Indicators

Computable with census data

Labour force participation rates for persons aged 15-24 and 15+, by sex
Proportion of employed who are own-account workers, by sex
Proportion of employed who are working as contributing family workers, by sex
Proportion of employed who are employers, by sex
Percentage distribution of employed population by sector, each sex
Informal employment as a percentage of total non-agricultural employment, by sex (??)
Youth unemployment, by sex
Proportion of employed working part-time, by sex (??)
Employment rate of persons aged 25-49 with a child under 3 living in a household and with
no children living in the household, by sex

448. Not computable with census data

Average number of hours spent on unpaid domestic work, by sex (separate housework and
child care if possible)
Average number of hours spent on paid and unpaid work combined (total work burden), by
sex
Percentage of firms owned by women
Proportion of children under age 3 in formal care;
Women’s share of managerial positions;
Percentage of female police officers;
Percentage of female judges.

449. A gendered analysis of the factors related to paid and unpaid work begins with the core
topic areas that describe persons’ economic situations, and then considers how these may
vary in a systematic manner across men and women in the population. This section outlines
several indicators that may be useful to measure and describe persons’ economic situations.
a) Unemployment rate for women and men. To calculate the unemployment rate, the
economically active population is divided into employed and unemployed population, and
the unemployment rate is the percentage among the economically active who are not
employed (i.e. among those who are registered as such with the government. Promoting
gender equality in employment is widely recognized as an essential component of
economic and social development. Women’s participation in employment increases their
contribution to household income and their control over the allocation of those resources.

181
This leads to greater economic independence and self-determination, which are both
important for women’s empowerment. In the vast majority of countries, adult
unemployment is higher among women compared to men, with important regional
differences [the bigger problem is that many unemployed women are classified as not
economically active]. Unemployment is also prevalent among the youth population,
especially young women. Northern Africa had the highest gap – seven points – between
women’s and men’s employment rate overall in 2007, and had a gap of 12 points across
young men and women (United Nations, 2010 a). Finland, on the other hand, reported
that its register-based census of 2010 had found that in 2009, for the first time, the
employment rate of women exceeded that of men. The economic downturn of 2009
mainly affected export industries and brought men’s employment down more than
women’s. The occupational structures among women and men differ from each other, so
that the majority of women work in the public or services sector, which are less
vulnerable to economic trends (UNECE, 2012 c). Larger numbers of unemployed men
compared to women were also found in the 2010 censuses of Belarus, the Republic of
Moldova, Russian Federation, Tajikistan, and Ukraine (UNECE, 2012 c).

b) Occupation and industry for women and men. The complete analysis of the
distributions of women and men by status in employment, occupation and economic
sector of activity reveal gender differences and economic segregation. Cross-tabulations,
for example, of sector of activity and status in employment are necessary to answer
questions such as how men and women are distributed across employment status and
whether they differ in any way from one branch of activity to another. But beyond mere
sex-disaggregation, the analysis of the causes and consequences of gender differences
should be analysed in depth. Women are usually found in specific (i.e. female)
occupations and sectors that also have lower status and lower pay. This is part of the
existing occupational segregation and discrimination in employment.

c) Informal sector work. Two indicators on informal sector work can be obtained from
population census data. The first is the percentage of economically active women and
men employed in the informal sector by branch of activity; it shows the differences in
how women and men are engaged in the informal sector. The second is the sex
composition of informal sector workers by branch of activity, and illustrates the relative
importance of women and men within the informal sector. In many countries, the
informal sector provides women with the only opportunity for work in a situation of
limited access to formal sector employment. However, those working in the informal
sector lack protection, rights and representation, and represent a vulnerable population.

d) Average hours worked for women and men. One way in which the work of men and
women may be different is that women tend to work part-time more often than men. The
average number of hours worked, when asked in the census, can be compared between
women and men, and even analysed by occupation, branch of activity or urban and rural
areas. This provides a measure of intensity of work involvement. The main utility of this
indicator is that it is one of the elements by which income differentials between men and
women need to be qualified.

Average hours worked by women in occupation or branch i x 100

Average hours worked by men in occupation or branch i

d) Average number of hours spent on unpaid domestic work for women and men. This

182
 measure can be computed similarly to the average hours worked indicator above to
examine the branch of activity, whether unpaid domestic work, across women and men.

e) Percentage of women and men in part-time work. Part-time work is one way
employed women balance paid work with family responsibilities. In many countries,
employed women typically assume most of the responsibility for domestic work.

f) Average income by occupation for women and men. The gender pay gap reflects
inequalities that affect mainly women. A simple indicator is the ratio of women’s average
earnings to men’s average earnings, expressed per 100 and for the same period of time
(monthly, yearly…). From the information collected on income, when included in the
census questions and identified by category (salary and other), it is possible to estimate
the ratio by occupation and branch of activity using the following method.

Average income of women in occupation i x 100

Average income of men in occupation i

Additionally, it is important to investigate whether the gender wage gap is due to

discrimination or not. An engendered perspective would ask: for a similar position with
the same education and experience, do men earn more than women? Some gender wage
gaps may be the result of women being less educated, occupying lower positions, and
spending more time out of the labour market long term. Considering women who are
equal to men in the areas of education, occupational rank and experience, what is the
average wage by sex? The above calculation for income should therefore be computed
with these “educational,” “occupational rank” and “experience” qualifications. Income
information should not be here but in chapter 8

h) Average household income by female and male household head. If income data are
not gathered at the individual level within households, examining the average household
income by female or male household head may be another useful indicator of economic
well being. It can be computed similarly to average income of women and men by
occupation above. If valid individual income data are available within households,
women’s income could be presented as a per cent of men’s income similarly to average
income of women to men in an occupation discussed just above.

450. Some useful indicators for gender analyses of social security are:

a) Percentage of female and male beneficiaries, by age appropriate to retired population

In the US, women comprise 57 per cent of all Social Security beneficiaries age 62 and
older, and 68 per cent of beneficiaries age 85 and older.
b) Average monthly social security income received by women and men, by appropriate
age range

c) Percentage of economically active women and men protected by social security, by age
range

d) Average years of contribution and average years of benefits, by sex

6. Multivariate and further gender analyses

183
451. Occupational Feminization and Pay in the US. A longitudinal study (Levanon, England,
and Allison, 2009) using US decennial census data found that occupations with a greater
share of females pay less than those with a lower share, even when controlling for education
and skill. Therefore, at each level of education and skill level within an occupation where
there were more women than men working, the women were paid less than their male
counterparts of the same education and skill level. The authors used census data to test two
theories about why women’s work is paid less. The first theory, a gendered labour queue for
certain occupations reasons that employers’ preference for men creates the greater propensity
of women to be represented in lower paid occupations, while the second theory reasons that it
is the proportion of women in the occupation that drives down wages or devalues women’s
work. They used fixed-effects regression models, which allow the researcher to control for
the stable characteristics of occupations over 50 years. Their findings largely support the
devaluation view over the queuing view. Similarly, Blackwell and Glover (2008) use linked
census and longitudinal study data to examine women’s participation in science, engineering
and technology fields. They find that 80 per cent of women in health-related occupations (e.g.
nursing) were mothers, compared with only 40 per cent in science, engineering and
technology. These results show a connection between occupation and family life choices for
women.

452. Labour Force Participation of Married Women in China. Research (Maurer-Fazio,

Connelly, Chen and Tang, 2011) using longitudinal population census data from 1982-2000
in China examined married, urban women’s labour force participation. They find that while
having a parent, parent-in-law, or person above 75 years old in the household increases
women’s likelihood of being employed, while having a preschool-age child in the household
decreases women’s propensity to be employed. These older people in the household may take
on some of the unpaid household work that women generally do in China, thus freeing these
women of working age up to take paid employment outside the home. When rural-to-urban
migrant status is included in the analysis, the negative effect on women’s labour force
participation of having young children in the household is substantially larger for married,
rural-urban migrants than for their non-migrant urban counterparts. These rural-urban
migrant women may not have other supports for child care if they are new to the urban area,
so they stay at home with their preschool-aged children. Indeed, the study finds that the
positive effect of co-residence with elders is greater for the rural-urban migrant women than
for the non-migrants. These rural-urban migrant women are in a position where the kin-based
help is likely to live in the same household, whereas non-migrant urban women with elder
family members established in the urban area may receive kin-based help from kin who do
not live with them.

453. Agricultural holdings in households, fertility patterns and women’s labour force
participation as unpaid family workers. In the 2010 census round, several countries,
following FAO recommendations, have included a question on whether households serve as
agricultural production units, with their own plot of land and/or livestock. This opens up
some gender-relevant opportunities for analysis as rural households with their own
agricultural holdings are expected to be different from those that do not have such holdings.
Women belonging to such households are expected to have higher labour force participation
rates, although almost exclusively as unpaid family workers, almost all male heads of
households are expected to have a wife to help them in the production, fertility is expected to
be higher, and children are expected to have lower school attendance rates: boys because they
need to help on the family holding, girls because they must replace their mothers in

184
household duties. However, in order to bear out such relationships, certain statistical controls
have to be included. It may be appropriate to control for the socio-economic level of
households, for example, by introducing some sort of wealth index based on the quality of the
dwelling and the ownership of consumer durables. Moreover, it is probably advisable to
control for the presence of other adults, such as grandparents, in the household as these may
take over some of the household tasks of spouses.

454. On the basis of the 2010 Aruba Population and Housing Census, two separate analyses
were done to look into the occupational status of women and men: a) the type of organization
of work and b) the status of employment. For both analyses multinomial logistic regressions
were set up. The multinomial logistic regression is the extension of the simple logistic model
with a dichotomous dependent variable. The multinomial model allows for a categorical
dependent outcome with more than two levels. In the analysis, one of these categories has to
be chosen as a residual (or reference) category. In the analysis, all other categories of the
dependent variable are then compared to this category. The regression coefficients and odds
ratios of the predictors in the multinomial regression are equivalent to those of the simple
logistic model, i.e. each category of a given predictor is compared to the residual category of
that predictor in terms of their probability of occurring.

455. In the Aruban census the following categories of type of work were used: 1) Limited
corporation, 2) One-person business, 3) Foundation, 4) General partnership, 5) Association, 6)
Government institution, 7) Government company and 8) Other. In the analysis, categories
‘General partnership’ and ‘Association’ were placed in the category’ others’, as they had very
few cases. The reference category for type of work was ‘Limited Corporation’.

456. Table 34 presents the results of the multinomial logistic regression for type of
organization worked for. The results show that, after controling for age, education, marital
status and country of birth, large differences remain between male and females in terms of the
type of organization for which they work. Compared to employment in a limited corporation,
women are less likely than men to be economically active in a one-person business (odds ratio
= 0.728) or in a government company (odds ratio = 0.391). Their chances are almost equal to
those of men to find work in government department (odds ratio = 0.986), but they are much
more likely to work for a foundation (odds ratio = 3.441) or the ‘other type’ (odds ratio =
1.306). Differences between males and females are highest for ‘Government Company’ and
‘Foundation’. On Aruba, utilities (water, gas, electricity…) are placed in government
companies. More men than women work here. On the other hand, many of the organizations
in public service (elderly homes, health organizations) and education are foundations. On
Aruba, jobs in these sectors are clearly dominated by women.

457. Table 35 sheds some light on the gender differences in status of employment. Categories
for status employment in the analysis are: 1) Employer (3 or more employees), 2) Small
independent, 3) Small independent, without employees, 4) Temporary employee deployed by
a job agency and 5) Temporary employee, volunteer, non-paid family member. The last
category actually consists of three response categories in the census questionnaire. As there
were only a small number of cases in these categories, they were brought together.

458. Again the same predictors were chosen. The reference category for status of employment
is ‘salary earner’. Compared to this reference category, women are less likely than men to be
found in any of the other categories of employment status. The differences between men and
women are biggest in those categories that involve independent entrepreneurship, i.e.

185
employer (3 or more employees), small independent, small independent without employees.
The odds of being an employer with 3 or more employees are about 2.5 times bigger for males
than for females (odds ratio = 0.403). Also, men are about three times more likely than
women of being a small independent without employees (odds ratio = 0.339) and twice more
likely to be a small independent with one or two employees (odds ratio = 0.512).

Table 34: Aruba (2010) - Multinomial logistic regression of type of organization that
women work for, compared to men, by various explanatory variables

  Reference category = One-p. business Foundation Govt.dept. Govt. company Other

  Limited corporation B exp(B) B exp(B) B exp(B) B exp(B) B exp(B)
Intercept -2.124   -4.308   -2.065   -2.779   -3.636  
Age 0.008 1.008 0.026 1.027 0.007 1.007 -0.005 0.995 0.028 1.029
Sex Male 0.000 . 0.000 . 0.000 . 0.000 . 0.000 .
  Female -0.318 0.728 1.236 3.441 -0.014 0.986 -0.940 0.391 0.181 1.306
Marital status Never married 0.000 . 0.000 . 0.000 . 0.000 . 0.000 .
  Married 0.152 1.164 0.046 1.047 0.166 1.180 0.332 1.393 -0.871 0.460
  Divorced/Legally sep. 0.074 1.077 -0.161 0.851 0.145 1.156 -0.013 0.987 -0.637 0.604
  Widowed 0.432 1.540 -0.338 0.713 0.034 1.035 0.351 1.421 -0.513 0.777
Educ. attainment None 0.000 . 0.000 . 0.000 . 0.000 . 0.000 .
  Primary 0.038 1.039 -0.204 0.816 -0.104 0.901 0.126 1.135 -0.151 1.001
  Lower vocational -0.100 0.905 0.219 1.245 0.276 1.318 0.770 2.160 -0.397 0.813
  High school (4 - 6 yrs) -0.169 0.845 -0.079 0.924 0.437 1.548 0.501 1.650 -0.546 0.682
  Higher vocational -0.256 0.774 1.150 3.157 0.904 2.469 1.373 3.949 -0.612 0.665
  Higher (BA - MA - PhD) -0.354 0.702 2.289 9.862 1.468 4.339 1.076 2.934 -0.540 0.714
Country of birth Aruba 0.000 . 0.000 . 0.000 . 0.000 . 0.000 .
  Colombia 0.458 1.580 -2.169 0.114 -2.845 0.058 -3.225 0.040 1.633 5.738
  USA 0.415 1.515 -1.474 0.229 -1.646 0.193 -1.576 0.207 0.919 4.479
  Dominican Republic 0.282 1.326 -1.505 0.222 -2.482 0.084 -2.123 0.120 0.967 3.102
  Venezuela 0.390 1.478 -1.824 0.161 -2.569 0.077 -3.302 0.037 1.095 3.629
  Curaçao -0.169 0.844 -0.461 0.631 -0.752 0.471 -0.859 0.424 0.221 1.750
  Netherlands -0.014 0.986 -0.181 0.835 -0.561 0.571 -1.709 0.181 0.354 1.868
  Other 0.124 1.132 -1.172 0.310 -2.240 0.106 -2.272 0.103 1.118 3.452

Table 35: Aruba (2010) - Multinomial logistic regression of women’s status in

employment, compared to men, by various explanatory variables

Temporary
Small Temporary employee,
Employer (3 or independent, employee volunteer,
more Small without deployed by a unpaid family
  Reference category = employees) independent employees job agency member
  Salary earner B exp(B) B exp(B) B exp(B) B exp(B) B exp(B)
Intercept -5.982   -5.148   -4.143   -2.080   -1.008  
Age 0.032 1.033 0.025 1.025 0.029 1.030 -0.021 0.980 -0.019 0.981
Sex Male 0.000 . 0.000 . 0.000 . 0.000 . 0.000 .
  Female -0.908 0.403 -0.670 0.512 -1.080 0.339 -0.117 0.890 -0.137 0.872
Marital status Never married 0.000 . 0.000 . 0.000 . 0.000 . 0.000 .
  Married 0.720 2.054 0.776 2.172 0.177 1.194 -0.256 0.774 -0.341 0.711
  Divorced/Legally sep. 0.386 1.471 0.596 1.815 0.230 1.259 -0.132 0.877 -0.167 0.846
  Widowed 0.863 2.370 0.867 2.379 0.447 1.564 -0.085 0.918 0.190 1.209
Educ. attainment None 0.000 . 0.000 . 0.000 . 0.000 . 0.000 .
  Primary 0.179 1.196 0.377 1.458 0.056 1.057 -0.392 0.676 -0.262 0.769
  Lower vocational -0.075 0.928 0.221 1.248 0.045 1.046 -0.711 0.491 -0.531 0.588
  High school (4 - 6 yrs.) 0.729 2.073 0.466 1.594 -0.049 0.952 -0.538 0.584 -0.503 0.605

186
  Higher vocational 0.619 1.858 0.378 1.460 -0.128 0.880 -0.801 0.449 -0.905 0.404
  Higher (BA - MA - PhD) 1.243 3.466 0.581 1.787 -0.281 0.755 -1.174 0.309 -0.792 0.453
Country of birth Aruba 0.000 . 0.000 . 0.000 . 0.000 . 0.000 .
  Colombia -0.024 0.976 0.208 1.232 0.493 1.638 0.393 1.481 0.512 1.668
  USA 1.291 3.638 1.055 2.873 0.670 1.954 0.157 1.170 1.059 2.885
  Dominican Republic 0.217 1.242 0.080 1.083 0.426 1.531 0.488 1.629 0.625 1.869
  Venezuela 0.568 1.764 0.743 2.102 0.512 1.669 0.320 1.377 0.733 2.082
  Curaçao 0.367 1.444 0.382 1.465 0.099 1.104 -0.394 0.674 0.212 1.237
  Netherlands 0.892 2.440 0.624 1.865 0.333 1.396 0.203 1.225 0.448 1.565
  Other 0.812 2.253 0.508 1.663 0.268 1.308 0.125 1.134 0.412 1.510

Source: Population and Housing Census Aruba, 2010

7. Interpretation, Policy and Advocacy

459. Compared to their male counterparts, women participate in the labour market at a lower
rate and are represented in higher numbers in less lucrative occupations and sectors of the
economy. This is not a coincidence, but a pattern all over the world, which reflects
discriminatory practices in the labour force (i.e. education, selection, promotion, etc.). And,
women still earn less than men even after controlling for hours worked, education and skills
over 15 years after the Beijing Declaration (United Nations, 1995 b) affirming women’s right
to employment and productive resources, and the Millennium Declaration’s further
commitment to full, productive employment for both women and men.

460. Unpaid work mainly carried out by women need to measured, valued and accounted in
the national accounting systems. In this regard, data should be collecetd in such a way that
household and caretaking work, predominantly done by women across countries, is not
misclassified and underestimated. Related to this, policies should be enacted that pay women
for doing domestic work (e.g. Canada pays an allowance for unpaid and caretaking work) and
provide access to daycare to help families manage work and household responsibilities.

461. Advocates should inform policymakers and the general public about the importance of
this unpaid work done by women, when in turn allows men to do paid work. Also, advocates
can alleviate women’s domestic burden by sensitising the general public to inequalities in the
amount of domestic work by sex among those couples where both women and men work in
the labour market. Finally, the women’s domestic burden may be lightened by providing
basic infrastructure (e.g. clean, running water) and labour-saving equipment and technologies
(e.g. cooking, grinding and cleaning appliances) accessible to all.

462. The Institute for Women’s Policy Research (IWPR, 2011) has used US Census Bureau
data to examine women’s disadvantaged economic position with respect to social security
benefits and retirement. The institute finds that older American women are more likely to
face poverty than older men, especially unmarried or widowed women. Women’s median
annual social security benefits reach just 70 per cent of that of men. Further, these social
security benefits mean the difference between living in poverty or not for over two-thirds of
unmarried women living alone. With Medicare health insurance covering individuals at age
65, few women or men lack health insurance. The IWPR combines analysis with policy
activism in order to represent the interests of women. Their research on social security
pensions and Medicare health insurance raises awareness of the income inequalities within
the US pension system.

187
463. The World’s Women report (United Nations, 2010 a) found that one-half of the countries
worldwide meet the new international standard for minimum duration of maternity leave and
that 40 per cent meet the minimum standard for cash benefits, but there remains a gap
between statutory law and what is practiced. Many women, in particular those who do not
work in the formal or public sectors, are not covered by the legislation. Oun and Trujillo
(2005) make the case that where the maternity benefit funds come from is the reason for this
inequality towards women. They suggest that payment with public funds or social insurance
could reduce this inequality and gap between law and practice. Employers no longer bear the
direct costs of maternity. Currently, about one in four countries, especially in Africa, Asia
and the Arab States, continue to provide payment during maternity leave through the
employer with no public or social security assistance.

188
Chapter 11:
Migration
1. What is it?

464. Demographers differentiate across local migration or residential mobility, internal

migration, and international migration, all of which entail a change in an individual’s usual
place of residence lasting at least six or twelve months.
a). Local migration or residential mobility refers to a change of residence, either in the same
city or town, or between communities or cities.
b). Internal migration refers to a change of residence within the same country. An internal
migrant is a person who moves to a different administrative territory within the same country.
This kind of migration is often associated with urbanization (i.e. rural-to urban migration) or
with the forced movement of people fleeing violent conflict or natural disaster. In a census,
information on internal migration and the place of origin and destination of migrants is
derived from cross-classification of variables on past and current residence, such as place of
birth, place of previous residence, place of current residence and duration of stay. According
to the Principles and Recommendations for Population and Housing Censuses, Rev. 2
(United Nations, 2008 a), “geographical and internal migration characteristics” include: 1)
place of usual residence; 2) place where present at time of census; 3) place of birth; 4)
duration of residence; 5) place of previous residence; 6) place of residence at a specified date
in the past; 7) total population; 8) locality; and 8) urban and rural.
c). International migration refers to a person crossing an international border. An
international migrant is any person who changes his or her country of usual residence. The
census topics to be analysed in this context are: 1) country of birth; 2) citizenship; and 3) year
or period of arrival (United Nations, 2009 c). Most international migrants, nearly 90 per cent,
move for employment reasons (ILO, 2007: 4), be it as workers or as dependent family
members of a labour migrant. “Migrant workers” are defined as those who are engaged or
have been engaged in a remunerated activity in a State of which they are not a national.
Migrants and members of their families are protected by international law under the
Convention on the Protection of the Rights of All Migrant Workers and Members of Their
Families (CMW). International migration is analysed chiefly through the proxy of foreign-
born populations or, when this is not available, through foreign populations (i.e. non-citizens
residing in the country). Some censuses also include information on emigrants such as
number of household members living abroad, sex/age of departed and amount of remittances
sent back to the country of origin.

465. However, these categories do not capture the many complex and gendered features of
population movements in the 21st century, especially in developing countries, where motives
for migration differ vastly, and temporary moves have gained great social significance. The
research literature includes other types of migration, such as internal and international,
transitory, circular and permanent, regular (conforming to legal requirements) and irregular,
and voluntary and forced (such as human trafficking, conflict-induced or climate migration).
Due to the kinds of questions asked, censuses are most useful for traditional analyses in terms
of internal and international migration.

466. Gender is a key component in shaping migrants’ experiences and social realities.
Women and men may migrate for different reasons – employment, education, marriage,
human trafficking, etc. – that are related to gender and lead to different social outcomes for
women and men. Migration laws affect women and men differently, and their social and

189
economic realities as migrants may also differ greatly (Docquier et al., 2007). Recent
research suggests that female and male migrants also make different social contributions to
their host and home countries, for instance in terms of remittances (Semyonov and
Gorodzeisky, 2005; Worldbank, 2006).

2. Why is it important?

467. The UN (2010) estimates that there are 214 million international migrants in the world
today. This means that a substantial percentage, 3.1 per cent of the world’s 7 billion persons,
live outside their country of birth. The 2009 Human Development Report (UNDP, 2009)
makes the case that migration is a key element of human freedom. Additionally, it is hugely
effective in improving the income, education and participation of individuals, as well as
enhancing their children’s future prospects.

468. Migration is an issue of great concern in human rights treaties and has played an
important role in development policy agendas. Guarantee of all basic human rights to all
migrants – regardless of regular or irregular status in the receiving country – is one of the key
elements in the International Convention on the Protection of the Rights of All Migrant
Workers and Members of Their Families as well as one of 15 principles of the ICPD
Programme of Action. It has been recognized that, although every nation has the right to
decide who can enter and stay in the territory, such right should be exercised in the absence of
discriminatory actions and policies, when it comes to ensuring equal pay for equal work or
accessing health and education services (i.e. see ICPD Programme of Action, Section X).

469. Migration is one of three standard variables for demographic analysis; “data on internal
and international migration, together with fertility and mortality, are needed to prepare
population estimates for planning purposes and for determining policies on migration and
assessing their effectiveness” (UN 2009, para. 3.69). It has long been the missing component
in discussions of the demographic transition in Western Europe in the 19th century, asking for
instance whether the decline of the birth rates in Europe would have been faster in the
absence of substantial migration to the United States and elsewhere.

470. From a human rights perspective, a person’s status as migrant in a country is sometimes
associated with human rights violations and other forms of discrimination that exacerbate
existing gender inequalities. Where women and girls are seen as vulnerable groups, migrant
women and girls can be regarded as bearing a double burden. Specific vulnerabilities of
migrant women and girls include:

a) Violation of personal integrity and basic rights such as access to education and health

b) Dependence upon male family members, employers or others and lack of collective
organization (e.g. trade unions), leading to low levels of empowerment and high risk of
exploitation, physical and sexual abuse, gender-based violence and human trafficking
(ICPD, 1994; IOM, 2005; Omelaniuk, 2002);

c) Loss of status, discrimination and low occupational mobility, which translates into
difficulty accessing jobs with education and training of their home country;

d) Isolation and lack of social and cultural connection with their home lands;

190
 e) Invisibility on statistics and policy-making.

471. A gender lens specifically highlights how migration may mean empowerment and/or
disempowerment both for the migrant and her or his family members:

a) Empowerment: Women and men may find better living conditions in the receiving
city or country, including improvement of their economic, educational and social
position and increased decision-making power. Moving away from patriarchal systems,
women may have more freedom. They may self-actualize as they seek opportunities that
challenge traditional gender roles and norms. Migration may also be a means to escape
conflict and personal violence, as well as to enhance one’s access to education and
health care (Martin 2007).

b) Disempowerment: Women and men may face discriminatory practices in the receiving
country, as well as isolation, limited or no access to education and health services and
informal and unregulated work arrangements, among others. In several cases, migration
may put an added burden on migrant women who are often expected to transmit cultural
and traditional values, including patriarchal beliefs, to their children. Moreover, many
migrant women may find themselves at a greater risk of abuse and exploitation (such as
being forced into sex work or protection sex) because of their legal status and their
vulnerable social position.

472. Female migrants are a numerically important group; almost half of all international
migrants globally are women and girls. While in the 1960s, women comprise 46.8 per cent of
the world’s migrants, this percentage rose to 49.4 per cent by 2000 (UN-DESA 2009),
leading some researchers to talk of a feminisation of migration (see Castles and Miller, 1993).
With young women and girls underrepresented among migrants, especially in developed
countries, “the preponderance of women in the overall migrant population is the result of
high proportions of women at older ages, an outcome that probably owes more to the higher
longevity of women with respect to men than to the feminization of migration” (UN-DESA
2011). Overall, an analysis of international migration and gender can breathe understanding
into the interplay across migration, economic growth and poverty reduction, and further serve
as a measure toward the third Millennium Development Goal of promoting gender equality
and the empowerment of women.

473. As Yinger (2006) notes, “What has changed more dramatically than the numbers are the
reasons why females migrate.” While in 1960 more women were classified as dependents and
moved for family reunification purposes, today a higher percentage migrates for economic
opportunities. Female labour migration is diversifying in terms of destination countries, skill
levels and occupations (Jolly and Reeves, 2005). That said, a gendered division of
occupations – such as domestic and hotel work, care activities, sex work, and entertainment
industry for women, and construction, mining, truck driving, and security for men – remains
pronounced for migrant workers.

474. Overall, there is substantial evidence which supports the utility of a gendered
examination of migration. In addition to the sheer numbers of women migrants and the
different reasons for migration across women and men, research suggests that the nature of
migrant networks are qualitatively different for women and men, and this can, in turn, shape
their participation in the labour force (e.g. see Curran et al. 2005). Additionally, marital status
might be significant as it interacts with migration, as Kanaiaupani’s (2000) research shows

191
how marriage has a positive effect on the migration of men but a negative effect on the
migration of women.

475. Finally, studying migration is important because the impact of migration is not restricted
to the migrant only. It is also necessary to assess the situation of those who are left behind by
emigrant family members and the overall consequences of female and male emigration for
sending countries (e.g. brain drain of qualified medical personnel).

3. Data issues

476. The Principles and Recommendations note that “as interest in the movement of people
across national boundaries […] has grown steadily among countries, census items and
tabulations relative to international migration have grown in importance” (United Nations
2009, para 3.77). Despite this growing interest, censuses – like other data sources – are
limited in measuring emigration and often revert to citizenship as a proxy to analyse
immigration. Irregular immigrants, making up between 10 and 15 per cent of immigrants
globally, are rarely captured in censuses as undocumented migrants often refuse to be counted
out of fear of prosecution or deportation or fear of reprisals from traffickers (International
Organisation for Migration, 2010).

477. Migration is treated as a ‘transition’ in a census, rather than as an ‘event’ as it would be

in population registry data. A census is therefore limited when it comes to capturing and
measuring circular or seasonal movement of persons. Further, other gender issues related to
migration cannot be investigated on the basis of census data alone, including the nature of
migration – whether it is regular, irregular, forced, or voluntary –and the link between
migration and gender-based violence. Additionally, language barriers may also create the
circumstances for the undercounting of internal migration.

478. Few censuses probe respondents on their reasons for migrating. Among other things, this
is because the reasons are often linked or inter-dependent, so that they do not lend themselves
to neat categorizations. For example, a couple may decide that a move to a larger city might
help both of them to get a better job. However, the husband might go first to find work and
establish a base, and then send for his family. In such a situation, the wife, when asked for her
reason to migrate, might – not unreasonably – answer that she is following her family, when
actually the reasons for moving are more complex and intertwined. Some censuses (e.g.
Nepal, 2011) ask the respondents not only for their reasons for migrating, but also on their
reasons for staying where they are, e.g. agriculture, business, services, study, marriage,
dependent, conflict. Interpreting the answers to such a question may be even more difficult
than in the case of reasons for migrating.

479. Table 36 shows the results of the simple question on reasons for migration for Cambodia
(2008), one of the countries that does collect this information. The most evident difference
between men and women is the much higher proportion (48.7 per cent, compared to 27.3 per
cent) of women that mention family reasons as their primary motivation for moving. Once
this category is removed, the largest difference is in the first category (transfer of work
place), which is less common for women, but in part this may be because - for the reasons
explained in the previous paragraph, some of these moves are censored by the “family
moved” category. Repatriation and visiting are more common among women, for reasons
which are somewhat less obvious.

192
 Table 36: Cambodia (2008) - Reasons for migration by sex

Reason for migration Men Per cent Women Per cent

Transfer of work place 246,962 13.8 79,175 4.5
Search of employment 431,877 24.2 329,572 18.8
Education 62,926 3.5 33,591 1.9
Marriage 331,578 18.6 185,647 10.6
Family moved 487,006 27.3 852,827 48.7
Lost land or house 22,351 1.3 19,793 1.1
Natural calamity 2,577 0.1 2,254 0.1
Insecurity 31,604 1.8 36,640 2.1
Repatriation 92,429 5.2 111,604 6.4
Orphaned 6,811 0.4 6,028 0.3
Visiting 47,901 2.7 82,006 4.7
Other 21,499 1.2 13,472 0.8
Total 1,785,521 100 1,752,60 100
9

Source: National Statistical Institute of Cambodia. Results of the 2008 Population Census.

480. Although about one in four countries include questions in their censuses on former
members of the household who are now living abroad, identifying households where a
member is an emigrant is often difficult. Among other reasons, this is because migrations that
involve entire families cannot be captured this way. Another possibility is to look at those
censuses that ask about the nature of household income and to consider remittances as a
measure. At present, only about 10 per cent of censuses in the current census round directly
ask for remittances received. In those countries whose censuses are conducted on a de iure
basis, one can also use married women living without their partners as a proxy for male
emigration. The best alternative, however, to measure aggregate emigration figures, is to use
the censuses or registration systems of the primary destination countries of emigrants.

481. With respect to immigration, the main problem is the identification of immigrants and
immigrant households. When migrants do not speak the host country language, language
stands out as a reason that migrants are underrepresented in censuses. Additionally, migrants
may not wish to be enumerated if they do not have legal residency documents. While there is
much policy interest in households and individuals with a “migration background,” these
cannot always be interpreted as foreign citizens, because citizenship legislation differs from
country to country. For instance, a person of Jewish descent or related to someone of Jewish
descent can claim Israeli nationality (i.e. ius sanguini), while anyone born in France is French
by birth right (i.e. ius soli). Thus, migration background may be hard to identify unless
specifically asked, as in the German census which asks whether the respondent’s mother or
father has migrated to Germany and from where. Most other countries simply use foreign
citizenship as a proxy for immigrants, thus producing a number of false positives (for
example, persons born in a country under ius sanguini law but who have never relocated from
one place to another) and false negatives (for example, persons who have changed their
country of usual residence but hold the nationality of the census-taking country). Where
international migration is analysed using the variables citizenship (or, alternatively, place of
birth), extrapolating to “migration” language should be avoided or done with caution. Special
cases (stateless persons, naturalised persons, people who have dual citizenship) pose
additional problems. Finally, where place of birth is used as a criterion to identify
“immigrants,” moves in-between birth and current residence are left unaccounted.

193
482. The main strength of census data is that they allow for a detailed analysis of the
immigrant stock and its characteristics, which is – at least in some countries – a rare group in
the overall population (United Nations, 2007). When data are available, given that censuses
collect a wide range of information on each individual, they permit cross-tabulation of
migration-related characteristics (such as citizenship, duration of stay and place of residence
in the receiving country by sex) with a combination of demographic and socio-economic
variables (including age, educational attainment, marital status, labour force participation and
occupation by sex) (United Nations, 2007). This helps explain the influence of these factors
on the decision to migrate as well as allowing a comparison of the female migrants’
experience with that of male migrants.

483. Population censuses yield the most comparable data on international migration at the
global level. Census data offer near universal coverage, a vast amount of person-specific and
geographic information, and regularity of data collection. At the same time, availability,
timeliness and accuracy are limitations that should be addressed at the national level in order
to maximise census data usefulness.

484. In terms of internal migration, identifying past and current residence may be difficult in
the case of slum-dwellers without standard addresses and people fleeing violent conflict or
natural disaster. Further, significant under-reporting can occur with students, domestic
personnel, temporary or circular labour migrants and other groups that have de facto
relocated their usual place of residence if they have been present for six months or longer in
the new location, as well as with regards to non-residents such as visitors that may be
erroneously classified as recent in-migrants. Finally, there may be confusion for persons with
two or more residences and members of the armed forces. Usual residence should in any case
be based on the 12-month limit.

485. There has to be mention here of the fact that migrants are likely to be undercounted
because they prefer evading enumeration. Also, we need to say something about the
migration of nurses, although it is not clear how we can use census data to produce
information on that in the countries of origin.

4. Tabulations

486. Regarding geographical and internal migration characteristics, the Principles and
Recommendations recommend that NSOs construct the following essential (*), recommended
(R) and additional a) tabulations:

 P1.1-R: Total population and population of major and minor civil divisions, by
urban/rural distribution and by sex*;
 P1.2-R: Population by size-class of locality, by major civil division and by sex*;
 P1.3-R: Population of principal localities and of their urban agglomerations, by sex*;
 P1.4-R: Native and foreign-born population, by geographical division, by age and
sex*;
 P1.5-R: Population, by duration of residence in locality and major civil division, age
and sex*;
 P1.6a-R: Population by place of usual residence, duration of residence, place of
previous residence, by major civil division and sex;
 P1.6b-R: Population … years of age and over, by place of usual residence, place of

194
 residence at a specified date in the past, by major civil division, age and sex;
 P1.1-A: Native population, by major civil division of birth, age and sex.

487. The essential, recommended and additional tabulations on international migration and
immigrant stock, in turn, are the following:

 P2.1-R: Foreign-born population, by country of birth, age and sex;

 P2.2-R: Foreign-born population, by period of arrival, country of birth, age and sex*;
 P2.3-R: Population, by country of birth and citizenship, age and sex;
 P2.4-R: Economically active foreign-born population … years of age and over, by
period of arrival, occupation and sex;
 P2.1-A: Foreign-born population, by marital status, age and sex;
 P2.2-A: Foreign-born population … years of age and over, by usual (or current)
activity status, age and sex;
 P2.3-A: Foreign-born population … years of age and over, by educational attainment,
age and sex.

All of the latter concern immigrants and even more specifically those born abroad. Some
censuses also ask about members of the household who are currently living abroad. While
this question misses some emigrants (e.g. those that went abroad with their entire household),
it is currently the best instrument for measuring emigration at the origin. The results can be
tabulated by age and sex, as in the following example from the 2000 census of Cape Verde.

Table 37: Cape Verde (2000) – Emigrants by age and sex declared by the remaining
household members

Males Females
0-4 79 80
5-9 143 150
10-14 242 314
15-19 720 760
20-24 1,156 1,127
25-29 1,096 770
30-34 906 545
35-39 638 385
40-44 365 205
45-49 168 157
50-54 79 103
55-59 68 101
60-64 87 184
65-69 78 122
70-74 50 97
75-79 40 55
80+ 11 28

Source: INE, Cape Verde

What this table shows is that, while emigration is a predominantly male phenomenon in Cape
Verde, women actually are the majority of emigrants below age 20 and after age 50.

195
5. Indicators

488. Internal Migration: The proportion of migrants (m/f) to an area a, who migrated from an
area b, and conversely. This indicator can be used to measure the intensity of migration, by
sex, and their direction of migration flow across two regions, between rural and urban areas
or to measure internal migration (in terms of change of place of residence) within the same
administrative area (for example in the same region).

489. Immigration: a) Foreign population (m/f) as a percentage of total population (m/f) and b)
Proportion of women in the foreign population. Indicator a) provides the estimated number of
female and male international migrants expressed as a percentage of the total female and
male population; indicator b) shows sex ratios in migration.
490. Emigration [where data are available]: a) Emigration rate (m/f) and b) emigration rate of
migrants with tertiary education (m/f). Indicator a) measures the stock of female and male
emigrants from a country at a particular point in time expressed as a percentage of the sum of
the resident population in the country of origin and the emigrant population. Indicator b)
includes only those with a university education and thus indicates brain drain.

491. Labour force participation of immigrants: Economically active foreign-born population

by occupation, age, sex and urban/rural residence, and may serve as a measure of upward or
downward social mobility. This indicator provides the estimated number of female and male
international migrants that are participating in the labour force by age, occupation and place
of residence.

6. Multivariate and further gender analyses

492. Several studies provide in-depth analyses on reasons for migration.

Country example 19: Marriage Migration in Nepal and Iran

Acharya and Chaudhury (2010) use census data from Nepal and Iran find that both countries
list marriage as one of the causes of both internal and international migration. They find that
the largest proportion of women have moved because of marriage, in both internal and
international migration. They explain that traditionally, women have to move onto their
household of marriage, which is the household of the husband so they move to a different
village, district and even country. This migration in effect isolates women from their parental
household and its support system. This can create a context of diminished social status for
women in the new marital household. In contrast, men seldom incur a move after a marriage,
so they do not experience this same social disruption. Because men and women migrate for
different reasons, their migration and integration process in the new milieu may be different.
A female migrant may experience a different trajectory of labour force participation and
health outcomes compared to her male migrant counterpart due to the different reasons that
one may move, which seems to be patterned by gender. A next step would be to examine
many factors, such as labour force participation, health outcomes, and poverty status, to
understand to what extent they are similar and different across women and men.

493. Marriage – voluntary or forced – or family reunification has traditionally played a

significant role in migration, now exacerbated by increasing world globalization. Marriage
may be a driving force for women to migrate within their countries or internationally – be it

196
for constituting a new family, or for escaping abusive marriages that limited their freedom.
Where “reasons for migration” are asked in the census (e.g. Cambodia, Colombia, Nepal),
long form questionnaires or the combination of censuses and migration surveys can be used
to produce simple tabulations such as the example from Cambodia below, or to carry out
multivariate analyses such as the example from China below.

494. A first tabulation may be to examine the per cent of migrants as a result of marriage, by
age, sex, ethnic group, rural/urban, educational level, and country or city of origin. From this,
it may then make sense to example using a multivariate logistic model what factors – such as
age, sex, ethnic group, rural or urban background, educational level, and city or country of
origin – to know what proportion of the variation in the per cent of migrants as a result of
marriage is associated with each of these factors, while considering the interrelations among
these factors.

495. Fan and Huang (1998), for instance, based on statistical analyses of a 1 per cent sample
of China’s 1990 Census, analysed interprovincial female migration as an economic strategy
and, thus, for empowerment. The results indicated that women in disadvantaged positions –
regarding institutional, structural, and socioeconomic factors – were more likely to pursue
marriage as a strategy to achieve migration and to improve their social and economic
mobility. Female migrants with agricultural hukou (household) classification, for instance,
were 5.0 times more likely to be marriage migrants than those with non-agricultural hukou.
Further, female migrants with college or above level of education were 77.3 per cent less
likely to be marriage migrants than those without. Estimates of the economic variables also
suggested that female migrants from wealthier provinces were less likely (having a 41.3 per
cent lower likelihood or propensity), and female migrants to coastal provinces more likely
(1.2 times), to be marriage migrants.

496. Another study may be to examine gender barriers to labour force participation. Female
migrant workers are among the least protected by labour and immigration laws (cf. Beijing
Platform for Action) and face additional barriers to the enjoyment of their labour rights due to
language, ethnicity, culture, religion, or socio-economic status. Household composition may
pose an additional difficulty. Although many female migrant workers contribute to the
economies of both the sending and the receiving countries or cities – through their
participation in the labour force and remittances – in many receiving countries they
experience higher levels of unemployment compared with both non-migrant workers and
male migrant workers. A logistic regression could be carried out with ‘status in employment’
or ‘currently working or employed’ (recoded as a binary variable: unemployed or not
unemployed) as a dependent variable. The various above-cited socio-demographic variables,
as well as ‘disposition to work’ and other work-related census variables (as long as no
collinearity exists) could be entered as predictors to assess their relative weight. As
governments are accountable for facilitating migrant’s full integration into the labour force
and for assuring full access to economic opportunities, a better understanding of barriers to
employment can help formulate policy-responses. As examples, are language courses or child
care needed, or does one particular ethnic group need more social support?

497. A study by Stone, Purkayastha and Berdahl (2006) looking at differences in earnings of
female migrants in the US according to their country of origin, found that some Asian
subgroups require specific policy responses. Data from the 1 per cent 2000 Integrated Public-
Use Microdata Series including Filipina, Asian Indian, and non-Hispanic white women living
in New York, Chicago, and Los Angeles, revealed that earnings inequality among highly

197
educated migrant women of the same age and occupation, proficient in English and working
the same amount of hours, is associated with ethnic origin and period of arrival in the US.
While Filipinas and non-Hispanic white women’s earnings are found not to differ
significantly, Asian Indian women’s earnings are lower. Women who arrived in the United
States in the 1990s earn significantly less compared to natives, while women who migrated
before the 1980s report higher earnings compared to natives, thus defying the supposed
“double burden” of being a migrant and a woman.

498. Immigration and household composition. Information on the living arrangements of

migrant women and men is useful for understanding migrants’ lives and experiences,
including:

a) Household composition (e.g. sex, age, etc. of those who live alone, live with other
migrants, live with people born in the receiving country) which can be further
disaggregated by age at arrival, years of residence, place of birth, highest level of
education, etc. and should, at least for internal migration, be compared with women and
men who live in their previous place of residence (e.g. rural areas);

b) Family size and fertility rate among immigrants; and

c) The relationship between migratory status and employment as domestic personnel.

499. An example investigating household composition and living arrangements among

migrants is Thomas’ (2001) study “Evolving family living arrangements of Canada’s
immigrants.” Combining data from the 1986, 1991 and 1996 Canadian Population Censuses
and data from the Landed Immigrant Data System (LIDS), the author finds that migrant men
and women have different living arrangements at different stages in their lives in the
receiving country. Regarding those who migrated in their twenties, there is a one in three
chance that a woman will live with an immigrant who was already established in Canada,
comparing with a one in four chance for men. Around 20 per cent of the men in this age-
group will live with a person who followed them to Canada, contrasting with 11 per cent of
the women. Also, 15 per cent of men who immigrate in their twenties will live alone. As the
age at migration increases, the probability that an immigrant, either man and woman, will live
with relatives that are already established in the receiving country increases. Nevertheless,
there is a 48 per cent likelihood that those who arrived as pensioners will be living with
immigrants who preceded them. Also, marriage migration from the US to Canada is found to
be a frequent phenomenon, and Americans and Europeans have the highest probabilities of
living with Canadian-born adults.

500. Emigration and changes in family structure. Emigration can create changes in family
structure. To analyse the situation, emigrant households need to be identified; then, issues
such as unmet basic needs (see poverty section), employment opportunities of the spouse
staying behind and children’s education can be analysed. For example, women left behind
may need to spend more time in productive activities outside the house and give more
responsibilities of childcare and housework to their daughters. It is therefore interesting to
analyse school enrolment rates of boys and girls in emigrant households and to compare them
to the rest of the population. Other issues for analysis include:

a) The household structure and headship of households that have members living abroad
vis à vis the general population;

198
 b) The economic characteristics of female heads of households that have members
residing abroad, such as the amount of financial support received from people living
outside of the household within the country or outside, particularly anything suggestive of
the way they manage the remittances received from abroad, small businesses that they
may head, homes that they may have acquired, and consumer durables in the home;

c) The age and sex of household members residing abroad, if these are known, cross-
classified by the socioeconomic characteristics of the households, which may be
indicative of the amount of resources sent by different kinds of migrants.

501. Gender and the brain drain. Highly skilled women are on the move as a result in part
due to the rise of female education. Docquier et al. (2007), for instance, computed sex-
disaggregated indicators of brain drain as a proportion of the total educated population born
in the source country for 195 countries. Using census data (1990 and 2000) of the receiving
OECD countries, the authors restricted their study to the foreign-born adult population aged
25 and over, classified into three educational level groups. The findings suggest that, between
1990 and 2000, the number of skilled women immigrants to OECD countries increased by 74
per cent, and that the share of women in the skilled immigrant population also increased. For
the vast majority of source regions, the growth rates of skilled women emigrants were higher
than the growth rates for unskilled women emigrants or skilled male emigrants; indeed, on
average, women’s brain drain was 17 per cent above that of men. According to the authors,
this feminization of the South-North brain drain mostly reflects gendered changes in the
supply of education.

502. The gendered nature and consequences of remittance sending. As women are
increasingly migrating independently and as income-earners, remittances are also
increasingly being sent by female migrant workers. Investigating the gendered nature of
remittance sending helps elucidate the contributions women and men make to their families
and communities of origin, to GDP in their home countries and thus to poverty reduction and
economic growth. There is evidence from Cuba showing that women are more likely to send
remittances (cash, goods, or both) than male migrants (Blue, 2004), be it in small amounts or
in kind. In some countries, such as Sri Lanka, the amount of remittances sent by females
outweighs that sent by men (UNFPA, 2006), while in others, such as the Philippines, men
send more money back home than women, even when taking into consideration earnings
differentials between the sexes (Semyonov and Gorodzeisky, 2005). In many countries,
remittances sent by women differ from those sent by men in amount, frequency, and
orientation on how they should be spent. (Blue, 2004; Semyonov and Gorodzeisky, 2005).

503. The effect of remittances – regardless of the sender – is another research topic with
gender implications. For instance, studying rural areas in Pakistan, Mansuri (2006) found a
positive impact of remittances on children’s schooling. Not only are children in migrant
households more likely to attend school, the effect is also more pronounced for girls than for
boys. In contrast, Haveman and Wolfe (1995) studied children left behind by highly skilled
female migrants and observe that these children are more likely to drop out of school than
their peers. Their explanation is that these children tend to have higher levels of human
capital which makes them attractive assets to the domestic labour market and able to
significantly contribute to household income. Where such data is available from the census,
NSOs could tabulate:

a) Average shares of remittances as part of total household income, by sex of remitter;

199
b) Educational attainment, literacy, school attendance (m/f) of children in remittance-
receiving households, by sex of remitter.

7. Interpretation, Policy and Advocacy

504. If sex ratios of internal migrants and international emigrants and immigrants are
significantly skewed, analysts need to ask follow-up questions, such as:

a) If internal migrants are mainly males moving to the bigger cities (e.g. Ghana): What
are the characteristics of male migrants (e.g. age, education, employment, marital status)?
What are the implications on urban women (e.g. employment options), rural women (e.g.
marriage options) and families left behind (e.g. remittances, children’s education)?

b) If international emigrants are mainly female (e.g. the Philippines), what are the
characteristics of the females who migrate (e.g. age, education, employment, marital
status)? What are the reasons for migration (e.g. marriage, types of work)? What are the
effects (e.g. remittances, education) to be felt by families, especially children, left behind?

505. Vulnerabilities associated with gender and migration can be mutually reinforcing. This is
known double burden or double disadvantage. For example, migrant women are less likely to
be employed in their European Union host country than are native women or migrant men.
Neither being female nor being a migrant alone can explain this result. Rather, gender and
migratory status reinforce vulnerability in this case.

506. Many gender issues related to migration cannot be invesigated using census data alone.
At present just a few countries ask about reasons for migration – such as marriage, poverty,
or military conflict – and census data are not useful to measure the circular or seasonal
movement of persons, or whether the migration is regular, irregular, forced, voluntary, or
possibly related to gender-based violence. Migrants’ issues can be better understood if more
replete data are collected on the process of migration itself.

507. The Philippines is one of the world’s largest labour exporting countries. The number of
women migrants outnumbers that of men. The National Statistical Coordination Board
(NSCB) plays an important role in supporting emigrants by providing sex-disaggregated data
on issues such as: 1) the distribution of overseas Filipino workers by place of work and
occupation; 2) the average cash remittance of overseas Filipino workers by place of work;
and 3) the distribution of overseas Filipino workers by major occupation and average cash
remittance. The NSCB has regularly publishing and disseminated reports and Factsheets on
Women and Men, as well as to making sex-disaggregated census data available online
(www.census.gov.ph). Among the data users are NGOs, the Central Bank and the
Commission on Overseas Filipinos. The latter registers emigrants, provides pre-departure
orientation seminars, supports families with relatives abroad, and provides community
assistance or referrals to cases involving trafficking and domestic violence. The Central Bank
uses NSCB data to enhance the financial products and services available to migrants and their
families, while NGOs learn more about their constituencies through census data. For
instance, the United Filipinos in Hong Kong group, that monitors the working conditions of
foreign domestic workers, was able to estimate the size of its target population (UNIFEM,
2008). NSCB in turn collaborates with NGOs and other stakeholders to broaden
implementation of Objective A.4. of the Beijing Platform for Action, which suggests that

200
national and international statistical organizations should devise suitable statistical means to
recognize and make visible women’s participation in the unremunerated and domestic
sectors, to migrant workers.

201
Chapter 12:
Disability
1. What is it?

508. The International Classification of Functioning, Disability and Health (ICF) (WHO,
2001) defines disability “as an umbrella term for impairments, activity limitations or
participation restrictions” and embraces at least three dimensions: body, individual and
society.

509. The Convention on the Rights of Persons with Disabilities (CRPD) states, “Persons with
disabilities include those who have long-term physical, mental, intellectual or sensory
impairments which in interaction with various barriers may hinder their full and effective
participation in society on an equal basis with others” (United Nations 2006). The Principles
and Recommendations for Population and Housing Censuses, Rev. 2 (Par. 2.376) (United
Nations, 2008 a) recognize that disability is a complex concept and recommends that
countries adopt an explicit definition based on the ICF when developing census or survey
questions that will be used to identify disability status and level of functioning.
a. Persons with disabilities are defined as “those persons who are at greater risk than the
general population for experiencing restrictions in performing specific tasks or participating
in role activities” per the Principles and Recommendations (Par. 2.351).
b. Human functioning refers to the dynamic interaction across personal factors, health
conditions and environmental factors.

2. Why is it important?

510. Persons with disabilities are protected by the Convention on the Rights of Persons with
Disabilities (2006). Other programmes and standards in place to protect persons with
disabilities include the World Programme of Action Concerning Disabled Persons (1982) and
the Standard Rules on the Equalization of Opportunities for Persons with Disabilities (1993).

511. The World Report on Disability (WHO/World Bank, 2011) suggests (using 2010 global
population estimates) that over one billion people in the world, or about 1 in 7 persons,
experience disability today. This estimate of 15 per cent of the persons living with disabilities
is higher than prior World Health Organisation estimates, which were about 10 per cent
(WHO/UNFPA, 2009). Further, disability prevalence is higher among already vulnerable
populations, specifically those within low income countries, those in the lowest quintile of
income within a country, older persons and women. Older women living in low-income
countries, therefore, are especially vulnerable to disability.

512. Article 31 of the Convention on the Rights of Persons with Disabilities (CRPD) (United
Nations, 2006 b) addresses issues relating to statistics and data collection on persons with
disabilities. States Parties are obligated to collect and disaggregate data, while also ensuring
confidentiality and protecting human rights and ethical principles, in order to inform and
implement policies within the Convention’s scope as well as monitor and assess progress
toward its full implementation.

513. As stated in the Principles and Recommendations, Par. 2.350, census data on disability
“can be utilized for general planning programmes and services (prevention and
rehabilitation), monitoring selected aspects of disability trends in the country, evaluation of

202
national programmes and services concerning the equalization of opportunities, and for
international comparison of selected aspects of disability prevalence in countries”. In
addition, census data on disability can be used to identify households for a more
comprehensive survey of persons with disabilities.

514. To understand the experience of disability, it is crucial to consider the interaction

between disability and gender; women and men, boys and girls generally face specific
disability experiences and may require particular social support services and policies.
Although not always made explicit, policies towards people with disabilities are not gender-
neutral: in practice, gender inequalities often combine with disadvantages experienced by
disabled women to create particular forms of exclusion and multiple discrimination. In
several social contexts, compared to their disabled male peers or non-disabled female peers,
women with disabilities are less educated, poorer, more isolated, more likely to be
unemployed, and more likely to be abused, as well as having lower health outcomes and
lower social status (Stubbs and Tawake, 2009). Disadvantages are, therefore, due to a
combination of gender, disability, other horizontal inequalities, such as migratory status and
ethnicity, as they intersect with social class and economic status.

515. Regarding education, international human rights instruments, such as the Convention on
the Rights of Persons with Disabilities, emphasize that persons with disabilities should be
guaranteed the right to equitable educational experiences at all levels, including lifelong
learning. It is therefore crucial to monitor educational attainment of persons with and without
disabilities and by sex.

516. Some crucial topics on gender and disability that can be examined using census data are
addressed in international instruments are listed below.

a) Education. Access to appropriate education and skills-training for women and girls
with disabilities is stressed by the Beijing Platform for Action. Strategic objective B.2,
to eradicate illiteracy among women, stresses that governments should take actions to
reduce the female illiteracy rate, with particular emphasis on women with disabilities.

b) Work. The Convention on the Rights of Persons with Disabilities addresses the urge
for the recognition of the skills, merits and abilities of persons with disabilities, and of
their contributions to the workplace and the labour market. Strategic objective F.5 of
the Beijing Platform for Action, to “eliminate occupational segregation and all forms
of employment discrimination,” also emphasizes the importance of implementing and
monitoring positive public and private-sector employment, equity and positive action
programmes to address systemic discrimination against women in the labour force, in
particular women with disabilities, with respect to hiring, retention and promotion,
and vocational training of women in all sectors.

c) Poverty. The Beijing Platform for Action recognizes that women with disabilities are
among the most economically disadvantaged groups and states that more effective
anti-poverty programmes directed towards them are a need.

d) Health. The Beijing Platform for Action calls for governments to ensure that girls and
women of all ages with any form of disability receive supportive health services.

e) Ageing. The Beijing Platform for Action recognizes that, with the increase in life

203
 expectancy and the growing numbers of older women, their health concerns require
particular attention. The incidence of impairment and disability increases with age.
Older women are particularly vulnerable to disability in old age due to gender
differences in life expectancy and gender inequalities over the life course.

f) Care. The ICPD Programme of Action declares that governments should support and
develop the appropriate mechanisms to assist families caring for children, the
dependent elderly and family members with disabilities.

g) Marriage and family formation. ICPD recognizes the reproductive rights for women
and men with disabilities and their right to household and family formation. The
document further stresses that adequate assistance must be provided to persons with
disabilities in the exercise of their family and reproductive rights and responsibilities.

h) Violence. The Beijing Platform for Action recognizes that women with disabilities are
among those women particularly vulnerable to violence. The Convention on the
Rights of Persons with Disabilities further declares that national governments should
take all appropriate legislative, administrative, social, educational and other measures
to protect persons with disabilities, both within and outside the home, from all forms
of exploitation, violence and abuse, including their gender-based aspects.

517. As stated in the Principles and Recommendations, access to paid work is crucial to
achieving self-reliance and ensuring the well-being of the adult population, both of persons
with disabilities as well as of those without disabilities. While women, in general, face
discrimination related to work opportunities and income gaps, women with disabilities may
face a double disadvantaged position on the labour market. Men with disabilities also tend to
be economically disadvantaged, compared to their non-disabled peers. Despite such
economic disadvantages, there are also additional costs of living connected with a disability,
such as medical treatments, transportation costs, and support services. As such, people with a
disability are more likely to be at risk of deprivation and poverty than other people.

518. The Convention on the Rights of Persons with Disabilities emphasises the need to
eliminate discrimination against persons with disabilities in all matters relating to marriage,
family, parenthood and relationships. While disability may stigmatize both men and women,
social constructions of femininity and masculinity may lead to specific discrimination
outcomes. Women with disabilities may be more vulnerable to social isolation, to sexuality
repression, and to denials of family formation. According to UNESCAP, “universally, the
incidence of marriage for disabled women is lower than that for disabled men.” For instance,
in Nepal, where marriage is a social norm for women, 80 per cent of women with disabilities
reported to be unmarried (Paudel, 1995). Another analysis on the 2008 Tanzania Disability
Survey showed that 54.8 per cent of persons with disabilities were in marital union, with
more males being involved in a relationship (62.5 per cent) than females (47.4 per cent)
(Tanzania National Bureau of Statistics, 2010). However, one should be cautious about these
conclusions as there is a serious selection bias operating. Because women (disabled or not)
live longer than men, more women than men are outside a relationship, as they have lost their
partner at an earlier stage in life. Another issue of concern, which has increasingly been
discussed, is forced marriages among people with learning disabilities or those who lack the
capacity to consent to a marriage (Clawson, 2010).

519. The World’s Women 2010 (United Nations, 2010 a) notes that actively caring for

204
disabled household members tends to be a time-consuming task primarily undertaken by
women, especially in less developed countries with few public services for such care. For
example, census data from the 2000 United States Census found a consistent pattern of living
arrangements that leaves children with disabilities disproportionately under the roofs of their
mothers or other women (Cohen and Petrescu-Prahova, 2006). Children with disabilities are
more likely to live with single parents, and especially their mothers, than other children.
Further, those who do not live with either parent are more likely to live in households headed
by women than other children. The authors argue that gendered living arrangements among
children with disabilities are a neglected aspect of gender inequality in caring labour.

520. Women in poorer communities, particularly those in developing countries, may be more
vulnerable than men to disability. In Fiji, there is a higher incidence of disability among
females than among males; girls comprise 72 per cent of children with disabilities in the 0-5
year old age range and females comprise 56 per cent of persons with disabilities in the 6-20
year old age range (Perry, 2002). Another example of interrelated relationships to be
disentangled with a gender lens comes from research in Somoa by Stubbs and Tawake
(2009). Women with disabilities differ from disabled men and non-disabled women in that
they are less educated, poorer, and of lower health and social status. Being disabled and a
woman bear a disadvantage that is horizontal in nature that results in lowered health and
social status. CAREFUL HERE: THIS IS NOT A UNIVERSAL PHENOMENON. IN
ETHIOPIA, FOR EXAMPLE, THE RELATION IS REVERSED, BUT THE ETHIOPIAN
CENSUS DATA MAY NOT BE REPRESENTATIVE BECAUSE OF THEIR OVER-ALL
LOW INCIDENCE.

521. Ageing processes in men and women with disabilities men may lead to secondary
disabilities. Women live longer than men, and they are more likely to experience disability
and severe disability, which is highest among the oldest old (WHO/World Bank, 2011). The
Convention on the Rights of Persons with Disabilities stresses the right of older women and
men with disabilities to be assisted by adequate social protection programmes, which must be
age-, gender- and disability-sensitive.

3. Data issues

522. The Principles and Recommendations (Par 2.351) indicate that four domains are
essential in determining a person’s disability status: (a) Walking; (b) Seeing; (c) Hearing and
(d) Cognition. Two other functions (self-care and communications) are suggested to be
included in census questionnaires. In 2001, the Washington Group on Disability Statistics
was formed was set up after the United Nations International Seminar on Measurement of
Disability (New York, June 2001) to facilitate international comparison of disability data .
The Washington Group29 suggests the following short set of questions to be incorporated in
national censuses:

1. Do you have difficulty seeing, even if wearing glasses?

2. Do you have difficulty hearing, even if using a hearing aid?
3. Do you have difficulty walking or climbing steps?
4. Do you have difficulty remembering or concentrating?
5. Do you have difficulty (with self-care such as) washing all over or dressing?
6. Using your usual (customary) language, do you have difficulty communicating, (for

Washington Group on Disability Statistics (WG) (s.d.), The Measurement of Disability

29

Recommendations for the 2010 Round of Censuses. p.1.

205
 example understanding or being understood by others)?

523. These questions, or some variations thereof, were included by a number of countries in
their most recent census questionnaire. According to the report of the Washington Group to
the Statistical Commission (Forty-third session, 28 February-2 March 2012) 31 countries
indicated they had used the short set of questions, or at least an adaptation of it. Some
countries used a set of 4 essential questions and others the more extended set of 6 questions.
Many other countries used a large variety of questions related to disability. Some examples:

 In the 2011 Australian Census the following question was included: ‘Does the person
ever need someone to help with, or be with them for, self care activities?' Similar
questions were asked for body movement and communication activities.
 In Botswana (2011) this question was asked ‘Does any listed person in A1 and B1
have any of the following disabilities?’ The interviewer could then choose from a list
of 17 types of disability. Similarly, in Barbados (2011), the interviewer could choose
from a list of 20 disabilities, including “Other”
 In Cambodia (2008) the enumerator was given the following instruction: ‘If the
person is physically/mentally disabled give appropriate code number from the list
below. Otherwise enter dash (-)’.
 The enumerators in the 2006 Egypt were instructed to note down the identifiction
number of the ‘handicapped’ person. Then they had to inquire about the type of
disability (11 categories) and the reason of disability.
 Some countries also allow the respondent to indicate degrees of disability. The 2011
census of Albania, for example, contained four categories, from no disability at all to
severe impairment.
 Some African countries ask whether the respondent is an albino, which is considered
to be a disability.

524. There is no doubt that the wording and contents of questions on disability affect the
answer of a person and ultimately changes the overall results. Because of the wide variety of
questions on disability in population censuses, comparing the prevalence levels of disability
between countries using different questions should be addressed with extreme caution. One
can expect that in the future, when more and more countries adapt the principles set out by
the Washington Group, this will improve. But even among countries, that are using the
Washington Groups’s recommendations some discrepancies may exist. One can expect that a
difference exists between the prevalence of disability among those countries that use the short
set of 4 questions and the group that use the full set of 6 questions. There should also be
differences between countries that only admit yes/no answers and those that allow
respondents to indicate degrees of disability.

525. One weakness with census data is that only a limited number of questions can be
devoted to a single topic, making it difficult to gain specific information on the gaps in health
between women and men. Since only a limited number of disability domains are measured,
“a comprehensive picture of disability can only come from large, national, sample surveys or
administrative data” (Washington Group on Disability Statistics, 2006). However, despite this
and other weaknesses of census data for the analysis of gender issues related to disability,
there are also some clear benefits.

526. First, disability data in censuses can be used to monitor the prevalence of limitations in

206
each disability domain (Washington Group on Disability Statistics, 2006). Those censuses
that ask the relevant questions can provide the absolute number of persons with disabilities
that are otherwise difficult to find, such as the number of blind persons in a country and those
who are deaf or mentally impaired (United Nations, 2001). Moreover, for many countries, the
census is the only source of information on disability at the national, regional and local levels
Additionally, as the Principles and Recommendations (Par. 3.78) state correctly, census data
help to monitor equalization of opportunities, measuring the social and living conditions of
persons with disabilities in terms of school attendance, educational attainment, employment,
marital status and living arrangements, among others.

527. Following international measures of disability, such as the International Classification

of Functioning, Disability and Health (ICF), allows the systematic and comprehensive
identification of persons with disabilities in the population, and allows for the ready
categorization of these disabilities into types and different levels of limited functioning
regarding basic activities regardless of nationality, culture or economic resources
(Washington Group on Disability Statistics, 2006). Further, in addition to identifying
households for a more comprehensive survey of persons with disabilities, collecting the short
set of questions in the census also informs and provides a frame for future surveys on
disability.

528. A potential problem of census data is non-response, which results from the complexity
and sensitivity of questions related to disability status (United Nations, 2001). As an
example, the person replying to the census questionnaire may not report information about a
disabled relative because of shame or stigma. Another cause of non-response is that women
and men with disabilities residing in institutions may not be included in the census population
or at least not in descriptive tabulations because often only non-institutionalized populations
are covered (United Nations, 2001).

4. Tabulations

529. The Principles and Recommendations stress that a census can provide valuable
information on disability and human functioning in a country (Par. 8.2.350) and the need to
develop statistics on the situation of persons with disabilities in order to assess equalization of
opportunities (Par. 3.109). For this purpose, the principal topics in census recommendations
should include items such as a) sex, b) age, c) place of residence, d) type of household, e)
marital status, f) educational attainment and school attendance, g) activity status, h) status of
employment, i) industry and j) occupation. Data permitting, tables can be constructed by
disability status as well as by specific types of disability. As an example, examining the
participation in education for people with and without disability is especially valuable if the
analyst compares among different types of disability within the school-age population.

530. The tabulation plan for disability data should include not only the prevalence rates by
sex and age, but also comparisons between persons with and without disabilities on these key
social and economic characteristics, as well as the type of disability. The Principles and
Recommendations recommend the following basic tabulations on disability characteristics:

 P8.1-R Population with and without disabilities by urban/rural area, age and sex*
 P 8.2-R Population 5 years of age and over, by disability status, educational
attainment, age and sex
 P 8.3-R Population 15 years and over, by disability status, economic activity status,

207
 age and sex

531. Additional tabulations on disability characteristics suggested by the Principles and

Recommendations are:
 P8.1-R Population with and without disabilities by urban/rural area, age and sex*
 P 8.2-R Population 5 years of age and over, by disability status, educational
attainment, age and sex
 P 8.3-R Population 15 years and over, by disability status, economic activity status,
age and sex

532. The following table from the 2006 census of Ireland provides very detailed information
on the prevalence of disabilities by age, sex and type:

Table 38: Ireland (2006) – Prevalence (in percentages) of detailed disability categories
by age and sex

Men Women
Type of Disability <15 15-24 25-44 45-64 65+ <15 15-24 25-44 45-64 65+
Blindness, deafness / severe vision hearing impairment 0.5 0.6 0.9 2.3 8.1 0.5 0.6 0.8 1.7 8.8
Substantial limitation of some basic physical activities 0.6 0.7 1.5 5.6 16.6 0.5 0.7 1.7 5.7 23.0
Difficulty learning, remembering, concentrating 3.2 2.8 1.7 2.5 6.1 1.6 1.7 1.5 2.2 8.5
Difficulty dressing, bathing, getting around in the home 0.8 0.5 0.6 1.6 7.7 0.5 0.4 0.7 2.0 14.0
Difficulty going outside the home alone 0.9 0.8 0.9 2.0 10.7 0.6 0.7 1.1 2.6 18.7
Difficulty in working or attending school/college 0.9 1.4 2.1 5.6 8.9 0.5 1.1 2.2 4.9 11.8
Learning or intellectual disability 3.1 2.8 1.4 1.4 1.7 1.5 1.6 1.0 1.1 2.0
Psychological or emotional condition 0.6 0.8 1.4 2.2 2.6 0.2 0.8 1.7 2.3 3.6
Difficulty in participating in other activities 1.1 1.1 1.5 3.6 10.8 0.7 1.0 1.7 3.9 16.3
Other, including chronic illness 1.1 1.1 1.8 4.9 9.7 0.9 1.2 2.1 4.7 10.8

Source: Population and Housing Census of Ireland (2007), Report 11

Slightly over half of the percentages in Table 38 are higher for men than for women. The
major exception, however, is the 65+ age group, where all disability rates for women are
higher than for men. Because this age category has such a large weight in the overall
disability rate, the latter is higher for women in all disability categories, with the exception of
learning and intellectual disabilities.

533. Tabulations on the distribution of the prevalence of disability by age and sex in the
population should also consider geographical division, urban/rural residence and the living
arrangements of persons with disabilities. If data are available, tabulations should also
disaggregate by income level or poverty status, as well as by recent migrant status.

534. Tabulations on household composition provide useful information for determining the
economic and social provisions that may be needed for persons with disabilities living alone
or with relatives. For this, information on the size of households and the distinction among
the one-person household, the nuclear family household and the extended family household
are essential. Another crucial point is the provision of tabulations for calculating prevalence
of disability per household (number of households with at least one person with disability per
1,000 households). In addition, the marital status of persons with disabilities can serve as a
measure of their social integration, especially if this calculation is compared to persons
without disabilities within the population.

208
535. Besides the basic tabulations on disability suggested in the Principles and
Recommendations (see 4.1 - 4.3), this manual also suggests a set of other tables based on UN
guidelines.

a) Education
o School attendance and educational attainment, by sex and age group.
o Net enrolment rate in primary education, by sex, age group and disability status
(Beijing indicator).
o Net enrolment rate in secondary education, sex, age group and disability status
(Beijing indicator).
o Gender parity index in 1, 2 and 3 levels of education, by sex, age group and
disability status (Beijing indicator).
o Literacy rate for 15-24 years of age by sex, further disaggregated by disability
status (Beijing indicator).

b) Income generation
o Labour force participation rates for 15-24 years of age, by sex and disability status
(derived from Beijing indicator).
o Labour force participation rates for 15 and older (in five-year age groups), by sex
and disability status (derived from Beijing indicator).
o Average total income for adults with disabilities, by sex and age group.
o Disability status as a predictor or explanatory variable in a regression analysis
predicting what is associated with poverty or lower income status.

c) Marriage and family formation

o Total population 15 years of age and over, by disability status, marital status,
number of children ever born alive (if available for men), age group and sex.

d) Care of disabled family members

o Children with disabilities, by the sex of the household head and household
composition. In many national contexts, data reveal that persons with disabilities,
especially children with disabilities, tend to live in families headed by women.
The care of persons with disabilities is mainly a female responsibility.
o Population 60 years of age and over with disabilities, by five-year age groups of
60-64 years, 65-69 years, etc., by the sex of the household head and household
composition.

536. As an example of differential school attendance, consider the following table which was
generated on-line from the REDATAM census data base of the census of El Salvador (2007):

Table 39: El Salvador (2007) - School attendance of 6-14 year olds by sex and type of
disability

Type of Disability Boys Girls

Difficulty Walking or Moving 40.4 41.1
Difficulty in Use of Hands or Arms 38.3 38.0
Sight Impairment, Even Using Glasses 67.6 71.4
Hearing Impairment, Even Using Hearing Aids 48.3 52.5

209
 Speech Impairment 31.6 32.3
Mental Retardation or Deficiency 20.6 19.3
Difficulty Bathing, Clothing, Eating 27.2 31.9
Oher Type of Disability 40.2 44.1
No Disability of Any Type 83.7 84.3

Source: Computed from the On-line REDATAM Data Base of the 2007 Population and
Housing Census of El Salvador

537. The table shows that having a disability significantly reduces children’s chances to
attend formal education, particularly in the case of mental problems, speech impairment or
difficulty in carrying out everyday tasks such as bathing, clothing and eating. Sight
impairment is the least serious disability, from the viewpoint of school attendance. The
differences between boys and girls are small, with school attendance in most categories
slightly higher for girls. This may not be typical of the situation in other parts of the world.
The WHO World Report on Disability (2011) (Table 7.1) lists school completion figures for a
weighted sample of 51 countries, based on survey data, which suggest that disability reduces
a boy’s chances of school completion from 61.3 to 50.6 per cent (i.e. a factor of 0.825) and a
girl’s chances from 52.9 to 41.7 per cent (i.e. a factor 0.788). However, as always it is crucial
that these data be analysed by age group, not only because male and female age structures
may be different, but also because things may have changed and the situation of younger
cohorts may be quite different from older cohorts. The report of the 2007 census of
Swaziland (Volume 4), for example, lists 617 girls with disabilities between the ages of 10
and 19 that had never attended formal education, compared to 665 boys, suggesting a slightly
more favourable situation for girls. But among men and women with disabilities between the
ages of 40 and 49 the overall number and the gender balance were quite different, with 2789
women and 1484 men that had never attended formal education.

Table 40: El Salvador (2007) - Percentage of ever married 30-39 year olds by sex and
type of disability

Type of Disability Men Women

Difficulty Walking or Moving 57.0 49.9
Difficulty in Use of Hands or Arms 53.4 48.0
Sight Impairment, Even Using Glasses 68.8 67.0
Hearing Impairment, Even Using Hearing Aids 39.3 42.5
Speech Impairment 21.4 28.2
Mental Retardation or Deficiency 6.9 16.0
Difficulty Bathing, Clothing, Eating 31.8 38.7
Other Type of Disability 51.9 51.7
No Disability of Any Type 79.1 77.1

Source: Computed from the On-line REDATAM Data Base of the 2007 Population and
Housing Census of El Salvador

538. Table 40, obtained from the same source, compares the probabilities of men and women
with disabilities ever having been married by the time they are in their thirties. The picture
here is somewhat mixed. Although some types of disability affect the marriage chances of
women more than those of men, such as difficulties in walking or in the use of hands or arms,

210
women with hearing or speech impairments or with mental retardation or deficiency or those
who have difficulty in performing everyday activities actually have a better chance of
marrying than men with these disabilities.

539. Census data on voluntary caring for persons with disabilities are hard to get by because
only a handful of censuses ask these questions. The 2007 census of Ireland, which did
address this issue, yielded the following table for men and women by marital status.

Table 41: Ireland (2007) - Voluntary care given by sex and marital status of the
caregiver and the number of hours of care given per week

Men Total 1-14 hours 15-28 hours 29-42 hours 43+ hours
Single 20,190 12,251 2,218 1,762 3,959
Married 36,565 21,853 3,554 2,013 9,145
Separated 2.979 1,800 361 201 617
Widowed 969 496 107 73 293
Women
Single 24,594 15,069 2,725 1,650 5,150
Married 64,054 35,806 6,877 3,181 18,190
Separated 6,723 3,783 758 403 1,779
Widowed 4,843 2,305 493 295 1,750

Source: Population and Housing Census of Ireland (2007), Report 11

540. Some countries publish more detailed tabulations, based on more detailed census
information, beyond the standard questions. The census report of Liberia (2008), for example,
contains a table on the cause of disability, with the following categories:

Table 42: Liberia (2008) – Causes of disability by sex

Males Females

From birth 4,708 4,174

Polio 2,288 1,878
Stroke 1,162 982
Epilepsy 1,195 1,125
War 7,634 6,131
Occupational injury 2,633 1,407
Transportation accident 1,731 923
Other types of accident 4,116 2,303
Ageing process 4,224 5,675
Other causes 11,147 12,476
Other diseases 15,824 16,624

Total 56,562 63,698

541. What this shows is that overall disability numbers are slightly higher for women than for
men, but that this difference is entirely accounted for by the last three causes of disability.
The greater incidence of disabilities associated with ageing is due, to some extent, to the
larger number of elderly women in the population, but this does not explain everything. The

211
female population over age 60 is 6.1 per cent larger than the equivalent male population, but
the number of disabilities related to ageing in women is 34.4 per cent larger. Women,
therefore, do appear to suffer disproportionally from disabilities associated with old age.
Strangely, a category of causes of disability not included here is that of disabilities associated
to childbirth. That this category can be important is shown by Stubbs and Tawahke (2009), in
their study on Samoa, which shows that 7.7 per cent of women with disabilities acquired their
disability through child birth complications. Another important category from a gender
perspective is disability caused by spousal violence. This is one of the 7 categories
(congenital/prenatal, disease/illness, injury/accident, spousal violence, other violence,
unknown, other) included in the 2010 census of Zambia.

542. Investigating a specific topic or factor, such as education, may serve as a useful indicator
of disability and gender issues. For instance, Possi’s (1996) study of Tanzania indicates that
while the gender parity in education is not an issue at the primary school level, the number of
enrolled girls with disabilities decreases as the number of years in secondary school
increases, similarly to the pattern for non-disabled girls in primary and secondary schools in
Tanzania. Further, very few women with disabilities have a college or university education.
(This analysis is relatively meaningless if it doesn’t make a comparison with the statistics for
boys).

5. Indicators

543. Using data from subsequent censuses can provide another gendered measure of
disability status. Using census data from Ireland, it was found that there were more males
than females living with a disability in 2000 (among the disabled population enumerated in
the 2000 census, 20,576 were males and 20,214 were females; the sex-ratio was 101.8).
Nevertheless, although the sex ratios indicated more males than females among the disabled
population (both in 1990 and 2000), the decline in the sex ratio in 2000 denoted that the
female disabled population has been growing at a faster rate than the corresponding male
population. In order to correctly interpret this trend, however, one would need to separate the
components of this increase that are due to the actual increase of the age-specific prevalence
of disabilities from the age effect, which is due to the fact that the female population is ageing
faster than the male population. A solution to this type of problem is by using age
standardization (see the Methodology Box below).

Methodology Box 9: Age Standardization

The 2011 census of Montenegro found that 54 per cent of the people in the country living
with a disability were women and 46 per cent were men (UNECE, 2012 b). If the objective of
computing this statistic is to know whether the care for persons with disabilities should be
organized predominantly to attend to female or to male patients, it is an appropriate indicator.
However, if the objective is to establish whether women are more or less prone to suffer from
disabilities than men, it is flawed by two intervening factors, namely:
1. There are more women than men in the population; and
2. The excess of women over men is concentrated in the oldest ages, where disabilities are
most common.

To take care of the first problem, one may compute a different statistic, namely the percent-
age of men and women that suffer from disabilities. Unlike the previous indicator, this one is
not affected by the total number of men and women. The result is 11.7 per cent for women,

212
compared to 10.2 per cent for men, which still suggests a higher incidence of disabilities
among women. However, when one age-standardizes this difference, i.e. when one computes
the percentages by age and then applies them to the same age distribution (in this case, the
age distribution for both sexes combined), the difference disappears and both percentages
become 11.0 per cent. The incidence of disabilities in men is higher early in life, whereas for
women it is higher at older ages, but given the age distribution for both sexes combined, the
overall incidence is about the same.

The following example from the 2010 census of Mexico illustrates how to carry out the
various steps to standardize the age structure.

Population Percent Disabled Number Disabled Idem Standardized

Male Female Total Male Female Male Female Male Female
0-4 5,346,943 5,181,37 10,528,322 0.87 0.70 46,259 36,323 91,086 73,807
9
5-9 5,604,175 5,443,36 11,047,537 1.87 1.35 104,718 73,380 206,431 148,928
2
10-14 5,547,613 5,392,32 10,939,937 2.00 1.59 110,882 85,523 218,660 173,509
4
15-19 5,520,121 5,505,99 11,026,112 1.81 1.43 99,954 78,713 199,652 157,628
1
20-24 4,813,204 5,079,06 9,892,271 1.89 1.37 91,053 69,806 187,135 135,958
7
25-29 4,205,975 4,582,20 8,788,177 2.08 1.42 87,530 65,043 182,890 124,746
2
30-34 4,026,031 4,444,76 8,470,798 2.45 1.67 98,726 74,296 207,720 141,593
7
35-39 3,964,738 4,328,24 8,292,987 2.78 1.97 110,259 85,073 230,627 163,001
9
40-44 3,350,322 3,658,90 7,009,226 3.59 2.85 120,248 104,229 251,571 199,668
4
45-49 2,824,364 3,104,36 5,928,730 4.77 4.14 134,690 128,499 282,733 245,408
6
50-54 2,402,451 2,661,84 5,064,291 6.40 6.04 153,815 160,884 324,237 306,090
0
55-59 1,869,537 2,025,82 3,895,365 8.50 8.30 158,881 168,187 331,044 323,399
8
60-64 1,476,667 1,639,79 3,116,466 11.44 11.49 168,883 188,340 356,423 357,943
9
65-69 1,095,273 1,221,99 2,317,265 15.32 15.56 167,839 190,138 355,096 360,559
2
70-74 873,893 1,000,04 1,873,934 20.64 21.29 180,328 212,863 386,687 398,875
1
75-79 579,689 665,794 1,245,483 27.54 28.39 159,620 189,031 342,949 353,615
80-84 355,277 443,659 798,936 35.39 36.63 125,739 162,497 282,758 292,623
85+ 298,739 404,556 703,295 46.01 48.40 137,461 195,801 323,612 340,388
Total 54155012 5678412 110939132 4.17 4.00 2256885 2268626 4761313 4297736
0

Source: INEGI, Tabulations from the 2010 census

The percentage of persons with disabilities who are women is 2,268,626 / (2,256,885 +
2,268,626) = 50.1 per cent. The percentage of persons who have a disability is 100 *
2,268,626 / 56,784,120 = 4.00 per cent for women, compared to 100 * 2,256,885 /
54,155,012 = 4.17 per cent for men. The age standardization consists in applying the

213
percentage of persons with a disability, separated by sex, not to the corresponding male or
female population, but to a common population, in this case made up of all individuals, of
both sexes. This yields the hypothetical results displayed in the last two columns, which vary
between the sexes because of the different proportions of disabilities among men and women,
but not because of the different numbers of men in the base population. The age standardized
percentage of women with disabilities is now 100 * 4,297,736 / (54,155,012 + 56,784,120) =
3.87 per cent, compared to 100 * 4,297,736 / (54,155,012 + 56,784,120) = 4.29 per cent for
men. Again, the male rates tend to be higher until age 60, whereas the female rates are higher
at older ages.

544. In the case of Nicaragua, where the overall percentage of persons with disabilities is 9.1
for men and 11.3 for women (Mont, 2007), male disability rates are higher than female rates
until adulthood; men and women are similar until age 39; after that, the rates diverge to a 10
percentage point differential in favour of men. This suggests that there is something about
events during the life course that differentiate the experiences of women and men in how and
when they acquire disabilities. The research by Stubbs and Tawake (2009) in Somoa
showcased just above might suggest that this mid-life higher propensity to develop a
disability for women could be associated with child birth or lower income. In order to
understand the Nicaraguan case better, the researcher may tabulate having a disability by sex,
age group, and poverty status.

545. The following table provides unstandardized and standardized indicators from the 2010
census of Mexico, separated by type of disability.

Table 43: Mexico (2010) – Unstandardized and standardized prevalence of selected

types of disabilities

Percentage Prevalence Standardized

Type of disability
Female Male Female Male Female
Walking or moving 53.3 2.10 2.29 2.19 2.20
Seeing 52.2 1.14 1.19 1.18 1.15
Hearing 45.2 0.50 0.40 0.53 0.38
Speaking or communicating 43.0 0.42 0.30 0.42 0.30
Personal care 52.6 0.20 0.21 0.21 0.20
Paying attention or learning 45.9 0.21 0.17 0.21 0.17
Mental disabilities 43.8 0.47 0.35 0.47 0.34
Total 50.1 4.17 4.00 4.29 3.87

546. Disability-free life expectancy. For addressing the interrelationships of ageing, gender
and disability, the disability-free life expectancy measure may be useful. This concept
provides an indicator of elderly persons’ health condition in order to help plan adequate
services and facilities. The method to calculate this disability-free life expectancy was first
presented in a report of the US Department of Health Education and Welfare (Sullivan, 1971)
and is often referred to as ‘Sullivan Health Expectancy method’. The Sullivan method makes
use of a life table and the age-specific proportions of persons with a disability. The age-
specific proportions of persons with a disability are multiplied by the corresponding number
of person years (Lx) lived between ages x and x + n in the life table. On the basis of these
calculated Lx values, the total after lifetime (Tx) in the disabled state can be calculated. By

214
dividing these age-specific Tx values by the total life table survivors (l x), one obtains the life
expectancy in the disabled (and non-disabled) state.

Table 44: Life table for Aruba 2010-2011 (males and females) life with and without
disability

Males           % Dis- e(x) Not e(x) % e(x)

Ag disable Disable Disable
e l(x) D(x,n) L(x,n) S(x,n) T(x) e(x) abled d d d
0 100000 710 99354 0.99290 7387957 73.9 0.4 67.9 6.0 8.1
1 99290 0 397160 0.99987 7288603 73.4 0.4 67.4 6.0 8.2
5 99290 0 496450 0.99934 6891443 69.4 2.2 63.4 6.0 8.7
10 99290 131 496123 0.99868 6394994 64.4 2.3 58.5 5.9 9.2
15 99159 131 495469 0.99569 5898871 59.5 2.5 53.7 5.8 9.7
20 99028 724 493332 0.99148 5403402 54.6 2.8 48.9 5.7 10.4
25 98305 957 489131 0.99084 4910069 49.9 3.0 44.4 5.6 11.2
30 97348 834 484652 0.98936 4420939 45.4 2.6 39.9 5.5 12.1
35 96513 1228 479494 0.98715 3936287 40.8 3.9 35.4 5.4 13.2
40 95285 1236 473332 0.98499 3456793 36.3 3.7 31.0 5.3 14.5
45 94048 1606 466226 0.97506 2983461 31.7 5.6 26.6 5.2 16.3
50 92442 3046 454597 0.96145 2517234 27.2 7.4 22.3 5.0 18.2
55 89396 3964 437072 0.94223 2062637 23.1 9.6 18.3 4.8 20.6
60 85432 6136 411823 0.89999 1625566 19.0 10.5 14.5 4.5 23.6
65 79297 10339 370636 0.84687 1213742 15.3 16.6 11.0 4.3 28.0
70 68958 12364 313879 0.76874 843106 12.2 21.1 8.2 4.0 33.0
75 56594 16671 241291 0.67552 529227 9.4 30.5 5.6 3.8 40.1
80 39923 14647 162997 0.55034 287936 7.2 39.2 3.7 3.5 48.1
85 25276 14671 89703 0.34571 124939 4.9 56.2 2.0 3.0 59.8
90 10605 8806 31012 0.13622 35236 3.3 67.9 1.0 2.3 68.9
95 1799 1799 4224   4224 2.3 76.9 0.5 1.8 76.9

Females           % Dis- e(x) Not e(x) % e(x)

Ag disable Disable Disable
e l(x) D(x,n) L(x,n) S(x,n) T(x) e(x) abled d d d
0 100000 1591 98552 0.98259 7975199 79.8 0.2 70.8 8.9 11.2
1 98409 150 393276 0.99892 7876647 80.0 0.3 71.0 9.0 11.3
5 98259 0 491296 1.00000 7483371 76.2 1.5 67.1 9.0 11.9
10 98259 0 491296 0.99929 6992075 71.2 1.8 62.2 9.0 12.6
15 98259 139 490948 0.99831 6500779 66.2 2.8 57.3 8.9 13.4
20 98120 192 490120 0.99817 6009831 61.2 2.6 52.5 8.8 14.3
25 97928 167 489222 0.99772 5519712 56.4 2.7 47.7 8.6 15.3
30 97761 278 488108 0.99481 5030490 51.5 3.1 42.9 8.5 16.6
35 97482 735 485573 0.99230 4542382 46.6 3.6 38.2 8.4 18.0
40 96747 761 481834 0.99200 4056809 41.9 4.6 33.7 8.3 19.7
45 95987 782 477978 0.98990 3574975 37.2 5.9 29.1 8.1 21.8
50 95205 1149 473151 0.97807 3096997 32.5 7.6 24.7 7.9 24.2
55 94056 3002 462773 0.96220 2623846 27.9 10.4 20.3 7.6 27.2
60 91053 3994 445282 0.94484 2161073 23.7 11.7 16.4 7.3 30.8
65 87059 5831 420719 0.91657 1715791 19.7 18.4 12.7 7.0 35.8
70 81228 8210 385617 0.86853 1295072 15.9 23.9 9.3 6.6 41.4
75 73019 12069 334921 0.79378 909454 12.5 33.4 6.4 6.1 48.8
80 60950 15558 265855 0.67355 574533 9.4 46.7 4.0 5.5 57.8
85 45392 19158 179066 0.50989 308678 6.8 63.9 2.2 4.6 67.4

215
 90 26234 15946 91305 0.41956 129612 4.9 69.4 1.4 3.6 72.2
95 10288 10288 38308   38308 3.7 78.9 0.8 2.9 78.9

Source: Population and Housing Census Aruba 2010; Helder (2012)

Figure 13: Percentage of life expectancy at age (x) spent with at least one disability, by
sex

Aruba 2010-2011
90.0

80.0

70.0

60.0

50.0

40.0

30.0

20.0
Male

10.0 Female

0.0
0 10 20 30 40 50 60 70 80 90 100

547. Table 44 shows an example of the Sullivan method for Aruba. For each age-category, the
percentage of total remaining life expectancy with a disability was calculated. Differences
between males and females in the percentage of remaining life spent with a disability are
depicted in Figure 13. Life expectancy for men on Aruba at age 60 is 19.0 years. At this age,
an average man can expect to live 14.5 years disability-free and 4.5 with at least one
disability. Women live longer. Their life expectancy at age 60 is 23.7 years, which is 4.7 years
more than men. Of these years, they spend 16.4 years in the disability-free state and 7.3 years
in the disabled state. Compared to men, most of the extra years women live are spent in the
disabled state. At age 60, a man can expect to live with a disability for 23.6 per cent of his

216
remaining years, for a woman this is 30.8 per cent.

548. A study from Thailand (Jitapunkul et al., 2003) showed similar results. Although women
had a longer life expectancy than men, they spent more years in the disabled state. At age 60,
women had a life expectancy of 23.9 years, and on average could expect to spend 18.2 years
free from long-term disability, leaving 5.7 years (or 24 per cent of their remaining life
expectancy) of years lived a disability. By contrast, men aged 60 had a remaining life
expectancy of 20.3 years and a disability-free life expectancy of 16.4 years, resulting in 3.9
years (or 19 per cent of remaining life expectancy) spent in a disabled state. Women,
therefore, had more years to live, both in disabled and in disability-free states. Similar
conclusions were obtained in a study from the City of São Paulo, Brazil (Camargos, Perpétuo
and Machado, 2005). In 2000, 60-year-old men could expect to live, on average, 17.6 years,
of which 14.6 years (83%) would be free of functional disability. Women of the same age
could expect to live 22.2 years, of which 16.4 years (74%) would be free of functional
disability. Men would have a functional disability and be dependent on others for 1.6 years
(9%), while the comparable period for women would be 2.5 years (11%).

549. Using the Sullivan method in censuses, one can at least make some predictions about
gender differences in the incidence of disability on the basis of prevalence data, provided that
one can control for the duration of disability. One potential solution is to compute what
percentage of their remaining life men and women are expected to live with disabilities. In
the example of Aruba in Table 44, this is 30.8 per cent for 60 year old women and 28.9 per
cent for 60 year old men. But this result is still biased by the fact that women, because they
live longer, survive to more advanced ages, where disabilities are more common. A better
alternative is therefore to standardize using the same life table (i.e. the one that characterizes
the mortality experience for both sexes combined) for both men and women. In this way one
can see how the results would change if men and women survived equally. Based on the
female life table, women aged 20 should expect to spend 8.8 years of their remaining life
with a disability, compared to 5.7 for men (see Table 44). But when the life table for both
sexes is used, this result changes to 7.5 years for women, compared to 6.5 for men. This
implies that, although the prevalence of disabilities beyond age 20 is indeed higher for
women than for men, a substantial part of the difference is also accounted for by the fact that
women live longer and are therefore more likely to survive to higher ages, where disabilities
are very common.

550. The Sullivan method provides an insight into gender differences in levels of disability.
However, there are also some shortcomings in using this method. There are at least two
methodological problems linked to this approach. The first one is that the results for both
sexes may be biased because persons with disabilities may have higher mortality than the
general population. This means that the number of years lived with a disability will be over-
estimated in an approach that assumes the same life table for those with and without
disability.

551. The second shortcoming has to do with the nature of the data used in the Sullivan
method. In general, each health condition can be described in terms of its prevalence and
incidence. Incidence refers to new cases of a health condition in a given period and
prevalence to the number of existing cases at a certain point in time. The Sullivan method
uses data on prevalence of disability, but to answer the question whether women are more or
less prone to disabilities than men would also require data on the incidence of disability.
These data are not available in censuses, as it would require a question on the timing when

217
the respondent became disabled. However, prevalence and incidence are closely linked. One
of the basic formulas in epidemiology states that prevalence = incidence x duration(avg). The
study of Oman et al. (1999) sheds some light on the relationship between prevalence and
incidence of disability. In their cohort study of 2,025 residents 55 years of age and older in
Marin County, California, they found that the incidence rates for lower body physical
disability were not significantly different between men and women. However, age-specific
and age-adjusted prevalence rates were consistently higher among women. They attributed
this difference to the longer duration women live with a disability, due to the lower recovery
and mortality rates among females vis à vis males.

6. Multivariate and further gender analyses

MAYBE SOME OF THE STUFF THAT IS NOW IN INDICATORS COULD ALSO BE

PUT HERE.

552. Persons with disabilities generally run a greater risk of being excluded from formal
education. The following example indicates how in Vanuatu children and adolescents
between the ages of 5 and 20 with a disability are less likely to attend education than others
of their age. Table 45 shows the results of a logistic regression in which the dependent
variable was whether a child was going to school (full time or part time – value 0) or not (i.e.
left school or never attended – value 1). Answers to the four questions on disability used in
the population census were included, namely:
Does this person have difficulty in:
a) Seeing, even wearing glasses ?
b) Hearing, even if using a hearing aid ?
c) Walking or climbing steps ?
d) Remembering or concentrating ?
In addition to these four disability conditions, sex, age, rural/urban residence, and citizenship
were included as control variables.

553. The results of the logistic regression clearly show that, for each of the four disability
variables, children with some difficulties have higher probabilities of not attending than those
without difficulties. Children who cannot hear, walk or remember/concentrate at all score
much higher than those who have some difficulty. For instance, a child who cannot walk has
more than 6 times higher odds of not attending school. A child with some difficulty has about
90 per cent higher chance to remain without schooling. Somewhat unexpectedly, the effect is
less pronounced (but present) for children who are visually impaired. Finally, girls have a
slightly higher probability of not attending school than boys, but the difference is only 3 per
cent. This is more or less the same as the difference in the general population, without
controling for intervening factors in a multivariate analysis: among all boys aged 5-19, 72.2
per cent attend school against 71.5 per cent of girls. Note, however, that the education of the
mother has a profound effect on the chance of a child to attend school. The odds of children
to be in school, with a mother who has more than primary education is 2.5 times as high
(1/.406) than among children whose mother has less than primary education.

218
Table 45: Vanuatu (2009) - Logistic regression of school attendance by children and
adolescents aged 5-20, by type of disability and other explanatory variables

Variable Category B exp(B)

Sex Male
Female 0.032 1.033
Age 5 - 7 yrs.
8 - 10 yrs. -0.755 0.470
11 - 13 yrs. -0.374 0.688
14 - 16 yrs. 0.894 2.445
17 - 19 yrs. 2.420 11.248
Urban/Rural Urban
Rural 0.358 1.431
Difficulty seeing No difficulty at all
Some difficulties 0.361 1.435
Cannot do at all 0.528 1.696
Difficulty hearing No difficulty at all
Some difficulties 0.170 1.185
Cannot do at all 1.256 3.512
Difficulty walking No difficulty at all
Some difficulties 0.657 1.929
Cannot do at all 1.867 6.469
Difficulty remembering No difficulty at all
Some difficulties 0.420 1.522
Cannot do at all 1.319 3.739
Citizenship Vanuatu by birth
Vanuatu by naturalisation -0.095 0.909
Other countires -0.287 0.751
Education mother Less than primary
Primary -0.506 0.603
More than primary -0.902 0.406
Constant   -1.274 0.280

Source: Population and Housing Census of Vanuatu (2009)

554. Noteworthy studies that utilize census data to profile persons with disabilities within
their populations are Zambia and Israel. These studies can be viewed online at the following
URLs: Zambia (http://www.statssa.gov.za/census01/html/Disability.pdf), and Israel
(http://www.cdc.gov/nchs/washington_group.htm). This latter study was discussed in the 10 th
Washington Group Meeting in Luxembourg in 2010.

7. Interpretation, Policy and Advocacy

555. Negative stereotypes about persons with disabilities may result in lowered expectations
of their abilities and social policies that do not allow them to realize their full potential.
Negative images and portrayals of persons with disabilities, especially older women with
disabilities, influence society’s view of them, and consequently, their ability to integrate and

219
participate in society. Negative stereotypes also increase their vulnerability to abuse and
discrimination.

556. According to Rousso (2003), boys may have the advantage in obtaining assistive devices
and other rehabilitation services needed to get to and participate at school. Women receive
only one fifth of the rehabilitation in the world and, particularly in developing countries, men
have greater access to rehabilitation services and to prosthetic and orthotic devices than
women. Gender bias in access to rehabilitative services and devices is in itself a barrier to
education for girls with disabilities. He recommends that more reliable data should come
from increased research on such basics as the number of girls with disabilities who are of
school age, their school enrollment levels, and their educational outcomes. This requires
developing a consistent definition of disability as well as disaggregating data on children who
are disabled by sex, and disaggregating data on girls by disability status.

557. Country examples for successful advocacy for data collection include an NGO-alliance
led by North India Cerebral Palsy Association that campaigned in Punjab, India for women
and men not to hide their disabilities during the census.

220
Conclusions
The previous chapters should have provided you with a sound understanding of the strengths
and weaknesses of census data as they relate to gender equality issues and the empowerment
of women and be able to make sense of tabulations, indicators and multivariate analyses on a
variety of gender topics.

The introductory chapter explained how analysing census data in a gender-responsive way is
situated within the overall efforts to produce, disseminate and use gender statistics. You were
introduced to the differences between sex-disaggregated data and gender statistics and
familiarised you with some of the challenges regarding gender statistics. You were also taken
through the key concepts related to gender analysis such as sex, gender, gender equality and
gender mainstreaming. You should now be able to understand some of the differences in the
way statisticians and gender experts use certain terms. Hopefully, this will enable you to
communicate more clearly and collaborate more effectively towards our shared goal, i.e.
providing an evidence-base to monitor progress towards gender equality and the
empowerment of women in your country and to inform policy-making and innovation in this
area.

Chapters 3 to 12 examined how ten issues of known gender relevance can be analysed on the
basis of census data. The ten gender issues were fertility, sex ratios, marital status, household
composition, living conditions/poverty, education, work, migration, disability and social
security. Each chapter followed the same structure: After defining the main concepts, it was
explained why the particular issue at hand is important from a gender and human rights
perspective and point to some of its policy implications. A section on data challenges helped
you appreciate the nature and quality of the census variables at your disposal for analysis.
The text then introduced some of the key tabulations and indicators that can be constructed
around the topic, using census data only, and how these should be interpreted. Finally, it
suggested some further, more advanced, gender analyses that can be carried out as well as
some potential advocacy uses.

The key messages conveyed throughout the ten chapters and in the extended introduction can
be summed up as follows:

 Collaborate. Gender analysis of census data is not a task for NSOs alone. Where
gender experts from various backgrounds (e.g. national mechanisms for gender
equality, universities, media, NGOs and other civil society organisations) work
together with NSOs, gender analysis will be more meaningful on the ground. (This
message is not really clear from the contents of the manual).

 Understand the limitations of your data. The data in front of you have not been
derived from a gender-neutral process. Census-taking is gender biased because of the
questions asked (and not asked) and the way the census questionnaire is constructed
(this is not really shown in the manual), the number and quality of female and male
staff engaged in enumerating, recording, processing and editing the data and for many
other reasons. Furthermore, as the individual chapters on gender issues showed, there
are specific limitations associated with each topic. For instance, to find out about
child marriage, further disaggregation of existing census variables or computation of a
new indicator may be required before embarking on the analysis.

221
  Go beyond sex-disaggregation. While sex-disaggregated data is a key tool to
developing meaningful gender statistics, gender analysis goes far beyond: Gender
analysis includes the development of gender-responsive indicators, in-depth
examination of the gender implications of key social phenomena, efforts towards
multivariate analysis, and efforts towards translating data into policy and planning.

 Tease out gender differences as well as differences based on age, socio-economic

factors, household composition, migration- or disability status, etc. While
tabulations and single indicators provide important overviews of complex issues and
are powerful advocacy tools, analysis should not stop there. Women, men, girls and
boys are not homogeneous groups and a person’s age, disability status, where she/he
lives, with whom, etc. will have important implications for that person’s life chances.
In order to make policies and programmes gender-responsive, meaningful on the
ground and effective, one needs to analyse all the information available, rather than
simply looking at the dichotomy female/male. Again, the collaboration of data experts
and gender experts is required in this process.

 Go beyond census data. In many instances, census data provides only an initial
glimpse of where gender is operating. To fully understand the phenomenon, other data
sources (where available) should be combined with census data or analysed in a
second step. For instance, to understand the gender pay gap in your country, you may
use the census as a sampling frame and then look at Labour Force Survey data for an
accurate measure of women’s and men’s incomes, hours worked, etc. to determine
differences between the sexes. You may choose to also consult other gender data, e.g.
on maternity leave legislation in your country, access to child care, and cultural
attitudes and beliefs regarding women’s labour force participation, child rearing, etc.

At the end of this guide, we provide you with a poster featuring key learning points, a
checklist of questions, and a number of useful “Memory Cards.” These tools aim to help you
in your efforts to promote and support gender analysis of census data in your country.
We’ll have to see about that.

222
References
Acharya, M. and R. H. Chaudhury (2010). Manual on Engendering Census. Bangkok,
UNFPA Asia Pacific Regional Office.
Addai, I. (1999). Does Religion Matter in Contraceptive Use among Ghanaian Women?
Review of Religious Research Vol. 40, No. 3 (Mar., 1999). Religious Research Association,
Inc. http://www.jstor.org/stable/3512371.
Adhikari, Ramesh and Chai Podhisita (2010). Household headship and child death: Evidence
from Nepal. BMC International Health and Human Rights Vol. 10:13.
Adsera, A. (2006). Marital fertility and religion in Spain, 1985 and 1999. Population Studies
Vol. 60, No. 2.
Anderson, M. (1992). The history of women and the history of statistics. Journal of Women’s
History Vol. 4, No. 1 (Spring).
Anderton, D. L. and R. J. Emigh (1989). Polygynous fertility: sexual competition versus
progeny. American Journal of Sociology Vol. 94: 832-855.
Andres, L. and M. Adamuti-Trache (2007). You’ve come a long way, baby? Persistent
gender inequality in university enrolment and completion in Canada, 1979-2004. Canadian
Public Policy Vol. 33, No. 1: 93-116.
Andro, A. and V. Hertrich (2001). La demande contraceptive au Sahel: les attentes des
hommes se rapprochent-elles de celles de leur épouse? Population 5: 721-771.
Anker R. (1998). Gender Jobs: Sex Segregation of Occupations in the World. International
Labour Organisation. Geneva: United Nations.
Arnold, F. (1985). Measuring the effect of son preference on fertility: the case of Korea.
Demography Vol. 22: 280-288.
Asadullah, M. N. and N. Chaudhury (2009). Reverse gender gap in schooling in Bangladesh:
Insights from urban and rural households. Journal of Development Studies Vol. 45, No. 8:
1360-80.
Aubel, J., P. Pandey and S. Rijal (1999). Rapid Assessment: Grandmothers’ Role in Nepali
Families. Kathmandu: NTAG.
Aubel, J., M. Agbendech, F. Brah and A. Madoul (2000a). Rôle des Femmes Agées dans la
Santé Maternelle et Infantile Niger. Niamey: Helen Keller International.
Aubel, J., I.Toure, M. Diagne, L. Kalala, E. Sene, Y. Faye and M. Tandi (2000b). Étude sur
le Rôle des Grands-mères dans la Santé et Nutrition de la Mère et de l’Enfant Dakar:
Christian Children’s Fund.
Aubel, J., N. Muratova, N. Bosorova, L. Djelilova, D. Allabergenova, N. Shaymanov and A.
Kuchimov (2003). The Role of the Grandmothers and Other Household Actors in Maternal
and Child Health: A Qualitative Community Study Uzbekistan: Project Hope.
Ayhan, H. Ö. (2001), Statistics by gender: measures to reduce gender bias in agricultural
surveys. International Statistical Review Vol. 69, No. 3: 447-460.
Banister, J. (2004). Shortage of girls in China today. Journal of Population Research Vol. 21,
No. 1: 19–45.

223
Bean, L. L. and G. P. Mineau (1986). The polygyny-fertility hypothesis: a reevaluation.
Population Studies Vol. 40, No. 1: 67-82.
Bedri, N. M. (1995). Grandmothers’ influence on mother and child health. Ahfad Journal
Vol. 12, No. 1: 74-86.
Bernard Van Leer Foundation (2002). Grandparents as caregivers.2002. Annual Report. The
Hague, Bernard van Leer Foundation.
Bianchi, S.; M. Milkie; L. Sayer and J. Robinson (2000). Is anyone doing the housework –
Trends in the division of household labour. Social Forces Vol. 79, No. 1: 191-228.
Bianchi, S. and D. Spain (1986). American Women in Transition. New York NY, Russell
Sage Foundation.
Blinder, A. S. (1973). Wage discrimination: reduced form and structural estimates. Journal of
Human Resources Vol. 8, No. 4: 436–455.
Blackwell, L. and J. Glover (2008). Women’s scientific employment and family formation: A
longitudinal perspective. Gender, Work & Organization Vol. 15, No. 6: 579-99.
Blakely, Tony; June Atkinson; Cindy Kiro; Alison Blaiklock and Amanda d’Souza (2003).
Child mortality, socioeconomic position, and one-parent families: independ-ent associations
and variation by age and cause of death. International Journal of Epidemiology Vol. 32, No.
3: 410-418.
Blue, S. A. (2004). State policy, economic crisis, gender, and family ties: determinants of
family remittances to Cuba. Economic Geography Vol. 80, No. 1.
Bosak, K. and K. Schroeder (2005). Using geographic information systems (GIS) for gender
and development. Development in Practice Vol. 15, No. 2: 231-237.
Boserup, Ester (1970). Women’s Role in Economic Development. London, St. Mary’s Press.
Bourne, K. L. and G. M. Walker (1991). The differential effect of mothers’ education on
mortality of boys and girls in India. Population Studies Vol. 45, No. 2: 203-220.
Boyce, C. and P. Neale (2006). Using Mystery Clients. A Guide to Using Mystery Clients for
Evaluation Input. Pathfinder International.
Boyd, M. and E. Grieco (2003). Women and migration: Incorporating gender into inter-
national migration theory. http://www.migrationinformation.org/Feature/print.cfm?ID=106
Budlender, D. (1997). The debate about household headship. Pretoria, Statistics South
Africa.
Caldwell. J. (1979). Education as a factor in mortality decline: an examination of Nigerian
data. Population Studies Vol. 3, No. 3: 395-413.
Camargos, M. C.; I. H. Perpétuo and C. J. Machado (2005). Esperança de vida com
discapacidade funcional em pessoas idosas em São Paulo, Brasil. Revista Panameña de Salud
Pública Vol. 17, No. 5-6: 379-386.
Cao, H. and F. L. Lei (2008). Girls Education in Gansu, China. In: Inclusion and Harmony:
Improving Mutual Understanding of Development in Minority Regions. Beijing: The Ethnic
Publishing House: 282-329.
Carr, D. and S. Bodnar-Deren (2009). Gender, aging and widowhood. International
Handbooks of Population 2009 Vol. 1 (Part VII): 705-28.

224
Casper, L. M. and S. M. Bianchi (2002). Continuity and Change in the American Family.
Thousand Oaks CA, Sage Publications.
Castle, S. E. (1993). Intra-household differentials in women's status: household function and
focus as determinants of children's illness management and care in Rural Mali. Health
Transition Review Vol. 3, No. 2: 137–157.
Castles, S. and M. Miller. (1993). The Age of Migration. International Population Movements
in the Modern World. Basingstoke/London, MacMillan.
Center for Gobal Development. (2009). Migrants Count. Five Steps toward Better Migration
Data. Washington, DC.
Chahnazarian, A. (1988). Determinants of the sex ratio at birth: review of recent literature.
Social Biology Vol. 35, No. 3–4: 214–35.
Chamie, J. (1986). Polygyny among Arabs. Population Studies Vol. 40, No. 1: 55-66.
Chant, S. (2003). New contributions to the analysis of poverty: methodological and
conceptual challenges to understanding poverty from a gender perspective. Santiago de
Chile, ECLAC.
Chant, S., and C. Pedwell (2008). Women, gender and the informal economy: An assessment
of ILO research and suggested ways forward, (BIT, Genève). Retrieved from
(http://www.ilo.org/wcmsp5/groups/public/@dgreports/@gender/documents/publication/
wcms_098934.pdf).
Chagas de Almeida, M. da C. and E. M. L. Aquino (2009). The role of education level in the
intergenerational pattern of adolescent pregnancy in Brazil. International Perspectives on
Sexual and Reproductive Health Vol. 35, No. 3: 139-46.
Chaudhry, I. S. (2009). The impact of gender inequality in education on rural poverty in
Pakistan: an empirical analysis. European Journal of Economics, Finance and Administrative
Sciences. Issue 15, http://www.eurojournals.com/EJEFAS.htm.
Cheka, C. (1996). How law and custom serve to disempower women in Cameroon.
Reproductive health matters, Fundamentalism, Women's Empowerment and Reproductive
Rights Vol. 4, No. 8: 41-6.
Chen, M.; R. Jhabvala and F. Lund (2002). Supporting workers in the informal economy: a
policy framework. Geneva: Working Paper on the Informal Economy 2. Geneva,
International Labour Office.
Chen, R. H.; S. M. Wishic and S. Scrimshaw (1974). Effects of unstable sexual unions on fer-
tility in Guayaquil, Ecuador. Social Biology Vol. 21: 353-359.
Cho, L.-J., R. D. Retherford and M. K. Choe (1986). The Own-Children Method of fertility
estimation. Honolulu, University of Hawaii Press for the Population Institute, East-West Cen-
ter.
Chu, C.; Y. Xie and R. Yu (2007). Effects of sibship structure revisited: evidence from
intrafamily resource transfer in Taiwan. Sociology of Education Vol. 80, No. 2. American
Sociological Association. http://www.jstor.org/stable/20452699.
Chu, J. (2001). Prenatal sex determination and sex-selective abortion in Rural Central China.
Population and Development Review Vol. 27: 259-281.

225
Chung, W. and M. Das Gupta (2007). The decline of son preference in South Korea: the roles
of development and public policy. Population and Development Review Vol. 33, No. 4: 757-
783.
Clawson, R. (2010). Forced Marriage and Learning Disabilities: Multi-Agency Practice
Guidelines. London: Foreign and Commonwealth Office, Forced Marriage Unit.
http://www.parliament.uk/deposits/depositedpapers/2010/DEP2010-2162.pdf.
Coale, A. J. and P. Demeny (1966). Regional Model Life Tables and Stable Populations.
Princeton NJ, Princeton University Press.
Cochrane, S. H. and K. C. Zachariah (1983). Infant and Child Mortality as a Determinant of
Fertility. Washington DC, World Bank, WP 556.
Cohen, P. N. and M. Petrescu-Prahova (2006). Gendered living arrangements among children
with disabilities: Evidence from the 2000 census. Journal of Marriage and Family Vol. 68
(August).
Cook, S. and S. Jolly (2000). Gender and poverty in urban China. IDS Research Report 50,
Brighton: Institute of Development Studies. http://server.ntd.co.uk/ids/bookshop/details.asp?
id=608.
Cooke, T. J. (2003). Family migration and the relative earnings of husbands and wives.
Annals of the Association of American Geographers Vol. 93, No. 2.
Corner, L. (2003). From Margins to Mainstream. From Gender Statistics to Engendering
Statistical Systems. UNIFEM. http://www.unifem-ecogov-apas.org/ecogov-apas/
EEGKnowledgeBase/EngenderingNSS/Margins2Mainstream.pdf
Cosío-Zavala, M. E. (2002). Examining changes in the status of women and gender as pre-
dictors of fertility change issues in intermediate-fertility countries. Paper presented at the Ex-
pert Group Meeting on Completing the Fertility Transition, New York, March 11-14.
Curran, S.; F. Garip; C. Ching and K. Tangchonlatip (2005). Gendered migrant social capital:
Evidence from Thailand. Social Forces Vol. 84, No. 1: 225-55.
Dalla Zuanna, G. and T. Leone (2001). A gender preference measure: the sex-ratio at last
birth. Genus Vol. LVII, No. 1: 33-56.
Das Gupta, M. (2005). Explaining Asia’s ‘missing women’? A look at the data. Population
and Development Review Vol. 31, No. 3.
Das Gupta, M. (2010). Son preference and fertility decline in China: consequences for elder
poverty. Paper presented at the 4th Annual Research Conference on Population, Reproductive
Health, and Economic Development, Cape Town.
Das Gupta, M. and P. N. Mari Bhat (1995). Intensified gender bias in India: a consequence of
fertility decline. Cambridge MA, Working Paper 95-02, Harvard Center for Population and
Development Studies.
Deere, Carmen D.; Gina E. Alvarado and Jennifer Twyman (2010). Poverty, headship, and
gender inequality in asset ownership in Latin America. East Lansing, Gender, Development
and Globalization Program, Michigan State University, Working Paper 296.
Desai, S. and S. Alva (1998). Maternal education and child health: Is there a strong causal
relationship? Demography Vol. 35, No. 1: 71-81.
Devereux, S. (2001). Livelihood insecurity and social protection: A re-emerging issue in rural
development. Development Policy Review Vol. 19, No. 4:507-19.

226
Docquier, F.; L. B. Lowell and A. Marfouk (2007). A Gendered Assessment of the Brain
Drain. Discussion Paper No. 3235. Bonn-Germany, IZA (Institute for the Study of Labor).
Dunn, H. L. (1940). Census - past and future. Journal of the American Statistical Association
Vol. 35, No. 209, Part 2: Proceedings of the Centenary Celebration: 242-251.
Ebenstein, A. (2010). The “missing girls” of China and the unintended consequences of the
one child policy. The Journal of Human Resources Vol. 45, No. 1: 87-115.
Ebenstein, A. and S. Leung (2010). Son preference and access to social insurance: evidence
from China's rural pension program. Population and Development Review Vol. 36, No. 1: 47-
70.
ECLAC (2004). Entender la pobreza desde la perspective de género. Santiago de Chile,
ECLAC, Unit on Women and Development, Serie Mujer y Desarrollo 52.
ECLAC (2006 a). Guidelines for Technical Assistance on the Design and Use of Gender
Indicators (Guía de Asistencia Técnica para la Producción y el Uso de Indicadores de
Género). Santiago de Chile, Unidad Mujer y Desarrollo.
ECLAC (2006 b). Compendium of Best Paractices in Poverty Measurement. Rio de Janeiro,
ECLAC/Expert Group on Poverty Statistics, Rio Group.
ECLAC (2011). Social Panorama of Latin America 2010. Santiago, ECLAC.
Elbers, C., J. O. Lanjouw and P. Lanjouw (2002). Micro-Level Estimation of Welfare.
Washington DC, Development Research Group, World Bank.
Eloundou-Enyegue, P. M. (2004). Pregnancy-related dropouts and gender inequality in
education: A life-table approach and application to Cameroon. Demography Vol. 41, No. 3:
509-28.
Elson, D. (1995). Rethinking strategies for development: from male-biased to human-
centered development. In: D. Elson (ed.). Male Bias in the Development Process. Manchester
UK, Manchester University Press.
Engle, L. (2004). The World In Motion; Short Essays on Migration and Gender. Geneva,
International Organisation for Migration.
Erickson, J. D. (1976). The secondary sex ratio in the United States 1969-71: association with
race, parental ages, birth order, paternal education and legitimacy. Annals of Human Genetics
Vol. 40, No. 2: 205-212.
Fan, C. C. and Y. Huang (1998). Department of Geography, University of California, Los
Angeles. Annals of the Association of American Geographers Vol. 88, No. 2.
FAO (1999). Agricultural Censuses and Gender Considerations. Rome, FAO.
FAO (2005). Agricultural Censuses and Gender: Lessons Learned in Africa. FAO Regional
Office for Africa. http://www.fao.org/sd/dim_pe1/pe1_051003_en.htm
Fox, L.; E. Bardasi and K. Van Den Broeck (2005). ‘Poverty in Mozambique. Unravelling
Changes and Determinants’. Poverty Background Paper to the Country Economic
Memorandum 2005. Washington, DC, World Bank.
Fresneda, B. (2012). Desigualdade salarial por gênero: considerações metodológicas. Paper
presented at the XVIII National Meeting of Population Studies, ABEP, 19-23 November.
Fussell, E. and F. F. Furstenberg (2002). Changes in the Tempo and Quantum of Transitions
in to Adult Statuses over the 20th Century in the United States. Report.

227
Fuwa, N. (2000). The poverty and heterogeneity among female-headed households revisited:
the case of Panama. World Development Vol. 28, No. 8: 1515–1542.
Garenne, M. (2002). Sex ratios at birth in African populations: a review of survey data.
Human Biology Vol. 74, No. 6: 889-900.
Garenne, M. (2008 a). Poisson variations in the sex ratio at birth in African demographic
surveys. Human Biology Vol. 80, No. 5: 473-82.
Garenne, M. (2008 b). Heterogeneity in the sex ratio at birth in European populations. Genus
Vol. 64, No. 3-4: 99-108.
Garenne, M. and E. Van de Walle (1989). Polygyny and fertility among the Sereer of
Senegal. Population Studies Vol. 43, No. 2: 267-283.
Gavrilova, N.; V. G. Semyonova; G. N. Evdokushkina and L. A. Gavrilov (2000). The
response of violent mortality to economic crisis in Russia. Population Research and Policy
Review Vol. 19: 397-419.
Giele, J. Z. and G. H. Elder, Jr. (eds.) (1998). Methods of Life Course Research: Qualitative
and Quantitative Approaches. Thousand Oaks CA, Sage Publications.
Global Commission on International Migration (GCIM) (2005). Report of a Workshop on
Gender Dimensions of International Migration, report from the Workshop on Gender
Dimensions of International Migration, GCIM, 23–24 March
http://www.gcim.org/attachements/Gender%20Workshop%20Report.pdf.
Goldscheider, C. (1999). Religious values, dependencies and fertility: evidence and implica-
tions from Israel. In: R. Leete (ed.). Dynamics of Values in Fertility Change. Oxford UK, Ox-
ford University Press: 310-330.
Goldstein, J. R.; T. Sobotka and A. Jasilioniene (2009). The end of “lowest-low” fertility ?
Population and Development Review Vol. 35, No. 4: 663-699.
Goodkind, D. (2011). Child underreporting, fertility, and sex ratio imbalance in China.
Demography Vol. 48, No. 1: 291-316.
Gordon, I. (1995). Migration in a segmented labour market. Transactions of the Institute of
British Geographers, New Series Vol. 20, No. 2.
Graefe, R. and D. T. Lichter (1999). Life course transitions of American children: parental
cohabitation, marriage, and single motherhood. Demography Vol. 36:205-17.
Gu, B. and K. Roy (1995). Sex ratio at birth in China with reference ot other areas of East
Asia: what we know. Asia-Pacific Population Journal Vol. 10, No. 3: 17-42.
Guilmoto, C. Z. (2007). Characteristics of sex-ratio imbalance in India and future scenarios.
Paper presented at the 4th Asia Pacific Conference on Reproductive and Sexual Health and
Rights, Hyderabad, Oct. 29-31.
Guilmoto, C. Z. (2009). The sex ratio transition in Asia. Population and Development Review
Vol. 35, No. 3: 519–549.
Gÿrsoy-Tezcan, A. (1992). Infant mortality: a Turkish puzzle ? Health Transition Review
Vol. 2, No. 2: 131-149.
Haines, M. and R. Avery (1978). Patterns of differential mortality during infance and early
childhood in developing nations. Paper presented at the Population Association of America
Meeting, Atlanta, Georgia.

228
Hakim, C. (2002). Lifestyle preferences as determinants of women's differentiated labor
market careers. Work and Occupations Vol. 29, No. 4: 428-59.
Handa, S. (1996). The determinants of female headship in Jamaica: results from a structural
model. Economic Development and Cultural Change Vol. 44, No. 4: 793-815.
Hanssen, K.G.; R. H. Scherg and R. S. Christensen (2009). One-person households – a
growing challenge for sustainability and housing policy. Paper presented at the ENHR 2009
conference in Prague. Workshop 7: Challenges for housing and urban sustainable
development in the perspective of rapid global change.
http://www.soc.cas.cz/download/1005/paper_Gram-Hanssen_W07.final.pdf.
Haughton, J. and S. R. Khandker (2009). Handbook on Poverty and Inequality. Washington
DC, World Bank.
Haveman, R. and B. Wolfe (1995). The determinants of children’s attainments - a review of
methods and findings. Journal of Economic Literature Vol. 33, No. 4.
Hayes, J.; H. Hartmann and S. Lee (2010). Social Security: Vital to Retirement Security for
35 Million Women and Men. IWPR Publication, No. D487. Washington DC, IWPR.
Heaton, T. B., M. Cammack and L. Young (2001). Why is the divorce rate declining in
Indonesia? Journal of Marriage and Family Vol. 63, No. 2: 480-90.
Hedman, B.; Perucci; F. and P. Sundstrom (1996). Engendering Statistics: A Tool for
Change. Orebro, Statistics Sweden.
Heise, L. (1994). Violence Against Women: The Hidden Health Burden. Washington DC,
World Bank Discussion Paper.
Helder, D. (2012). The cost of disability. (forthcoming)
Helpage/HIV-AIDS Alliance (2003). Forgotten families: older people as carers of orphans
and vulnerable children. London, HelpAge.
Henriques, M. H. F.T. (1979). Legal and consensual unions: their fertility implications in
Latin America. In: Lado T. Ruzicka (ed.). Nuptiality and fertility. Liége, Belgium, Ordina
Editions: 271-286.
Hill, Kenneth et al. (2009). Estimating pregnancy-related mortality from census data: expe-
rience in Latin America. Bulletin of the World Health Organization Vol. 87: 288–295.
Hill, K.; C. Stanton; M. Levin; B. L. Queiroz; Y. Choi; A. W. R. Mahanani and C. AbouZahr
(2011). Measuring Maternal Mortality from a Census: Guidelines for Potential Users. Gene-
va, WHO.
Hobcraft, J. (1993). Women’s education, child welfare and child survival: a review of the evi-
dence. Health Transition Review Vol. 3, No. 2: 159-73.
Hogan, D. P.; B. Berhanu, and A. Hailemariam (1999). Household organization, women's
autonomy, and contraceptive behavior in Southern Ethiopia. Studies in Family Planning Vol.
30, No. 4: 302-14.
Hoogstra, G. J. (2012). Gender, space, and the location changes of jobs and people: a spatial
simultaneous equations analysis. Geographical Analysis No. 44: 47–64.
Hoyle, R.; M. Harris and C. Judd (2001). Research Methods in Social Relations. 7th Ed.
Belmont CA, Wadsworth.
Hu, Y. and N. Goldman (1990). Mortality differentials by marital status: An international

229
comparison. Demography Vol. 27, No. 2: 233-50.
Hunt, P. and J. Bueno de Mesquita (xxxx). Reducing maternal mortality.
http://www.unfpa.org/webdav/site/global/shared/documents/publications/reducing_mm.pdf
ILO (1952). The ILO Social Security (Minimum Standards) Convention, No. 102.
http://www.ilo.org/public/english/protection/secsoc/areas/legal/ conv102.htm.
ILO (2005). Measuring the informal economy: From employment in the informal sector to
informal employment. Working Paper No. 53, Retrieved from
(http://www.ilo.org/public/libdoc/ilo/2005/105B09_37_engl.pdf).
ILO (2007). International Labour Migration and Development: The ILO Perspective.
Geneva, ILO, International Migration Brief.
ILO (2010 a). International Standard Classification of Occupations (ISCO) United Nations.
Retrieved from (http://www.ilo.org/public/english/bureau/stat/isco/index.htm).
ILO (2010 b). World Social Security Report 2010/11. Providing coverage in times of crisis
and beyond. Geneva, ILO, http://www.ilo.org/global/publications/ilo-bookstore/order-
online/books/WCMS_142209/lang--fr/index.htm.
Imaizumi, Y. and M. Murata (1979). The secondary sex ratio, paternal age, maternal age and
birth order in Japan. Annals of Human Genetics Vol. 42, No. 4: 457-65.
Inhorn, M. C. and F. Valen Balen (2002). Infertility around the globe: new thinking on
childlessness, gender, and reproductive technologies. Berkeley CA, University of California
Press.
INSTAT (2004). Gender perspectives in Albania - Population and Housing Census 2001.
Tirane, INSTAT (Instituti i Statistikes).
IOM (2005). Data and Research on Human Trafficking: A Global Survey. Geneva,
International Organization for Migration.
IOM (2010). World Migration Report 2010. The Future of Migration: Building Capacities for
Change. Geneva, International Organization for Migration.
IMF (2007). Senegal: Poverty Reduction Strategy Paper. IMF Country Report No. 07/316
(September).
Institute for Women’s Policy Institute (2011). Women and Social Security. IWPI Factsheet.
Washington, DC. IWPI. http://www.iwpr.org/initiatives/social-security.
INSTAT (2004). Gender perspectives in Albania - Population and Housing Census 2001.
Tiranë: INSTAT (Instituti i Statistikes).
International Labour Conference (1944). Declaration of Philadelphia. ILO Convention.
http://www.ilo.org/ilolex/english/iloconst.htm.
Isfeld, H. (2010). Los Homicidios en Brasil: un Análisis de Género y Diversidad. Washington
DC, PAHO.
James,W. H. (1989). Parental hormone levels and mammalian sex ratios at birth. Journal of
Theoretical Biology Vol. 139, No. 1: 59–67.
James, W. H. (1996). Past sex prejudice and present sex ratios. Theoretical Population
Biology Vol. 50: 394–95.
Jejeebhoy, S. J. (2001). Women's Education, Autonomy, and Reproductive Behaviour:
Experience from Developing Countries. Oxford UK, Clarendon Press.

230
Jejeebhoy, S. J. (2001). Women’s autonomy and reproductive behaviour in India. In: J. Phil-
lips and Z. Sathar (eds.). Fertility transition in South Asia. Oxford, Oxford University Press.
Jencks, C. et al. (1972). Inequality: a Reassessment of the Effect of Family and Schooling in
America. New York NY, Basic Books.
Jenson, R. and R. Thornton (2003). Early female marriage in the developing world, Gender
and Development Vol. 11, No. 2: 9-19.
Jha, P.; M. A. Kesler; R. Kumar; F. Ram; U. Ram; D. G. Bassani; L. Aleksandrowicz; S.
Chandra and J. K. Banthia. (2011). Trends in selective abortions of girls in India: analysis of
nationally representative birth histories from 1990 to 2005 and census data from 1991 to
2011. Lancet Vol. 377, No. 9781: 1921-1928.
Jitapunkul, S.; C. Kunanusont; W. Phoolcharoen; P. Suriyawongpaisal & S. Ebrahim (2003).
"Disability-free life expectancy of elderly people in a population undergoing demographic
and epidemiologic transition". Age and Ageing 32(4), British Geriatrics Society.
Jolly, S. and H. Reeves (2005). Gender and Migration: Overview Report. Brighton, Institute
of Development Studies (IDS).
Jones, G. W. (2003). The “flight from marriage” in South-East Asia. Singapore, Asian
MetaCentre Research Paper Series no. 11.
Jones, M. K.; P. L. Latreille and P. J. Sloane (2006). Disability, gender, and the British labour
market. Oxford Economic Papers 58. Oxford, Oxford University Press.
Jordan (2007). Engendering population censuses. Presentation at the Global Gender Forum,
Rome, 10-12 December.
Joshi, Shareen (2004). Female household-headship in rural Bangladesh: incidence,
determinants and impact on children’s schooling. New Haven CT, Economic Growth Center,
Yale University, Discussion Paper 894.
Junhong, C. (2001). Prenatal sex determination and sex selective abortion in central China.
Population and Development Review Vol. 27, No. 2: 259–80.
Kabeer, N. (2002). Reversed Realities: Gender Hierarchies in Development Thought. London
UK, Verso.
Kalkhan, M. A. (2011). GeoSpatial Information Modeling and Thematic Mapping. Boca
Raton FL, CRC Press.
Kanaiaupani, S. M. (2000). Reframing the migration question: an analysis of men, women,
and gender in Mexico. Social Forces Vol. 78, No. 4: 1311-48.
Khadiagala, L. (2002). Negotiating law and custom: judicial doctrine and women's property
rights in Uganda. Journal of African Law Vol. 46, No. 1: 1-13.
Kim, D. (2004). Missing girls in South Korea: trends, levels and regional variations.
Population Vol. 59: 865–78.
Kishor, S. (1993). May God give sons to all: gender and child mortality in India. American
Sociological Review Vol. 58, No. 2: 247-265.
Klasen, Stephan; Tobias Lechtenfeld and Felix Povel (2010). Female headship, poverty and
vulnerability in Thailand and Vietnam. Unpublished paper on http://www.econstor.eu.
Kleinjans, K. J. (2008). The man of the house: how the use of household head characteristics
may lead to omitted variable bias. California State University.

231
http://business.fullerton.edu/Economics/kkleinjans/Man_of_the_house.pdf.
Knodel, J. and M. B. Ofstedal (2003). Gender and aging in the developing world: where are
the men ? Population and Development Review Vol. 29, No. 4: 677-698.
Knoke, D.; G. Bohrnstedt and A. Potter Mee (2002). Statistics for Social Data Analysis. 4th
Ed. Belmont CA, Wadsworth.
Kongani, R. U. (2009). Incorporating gender perspective in national census and surveys
(Kenya). Global Forum on Gender Statistics. Accra, Ghana, 26-28 January 2009.
ESA/STAT/AC.168/32.
http://unstats.un.org/unsd/censuskb20/Attachments/2009KEN_Gender-
GUID8fe3670aad4e4d45baec26737e014da4.pdf
Kravdal, O. (2004). Child mortality in India: the community-level effect of education.
Population Studies Vol. 58, No. 2: 177-92.
Kulkarni, P. M. (2007). Estimation of missing girls at birth and juvenile ages in India. Study
Commissioned by UNFPA, India: http://www.unfpa.org/gender/docs/missingirlsatbirth_
india.pdf.
Kynch, J. and A. Sen (1983). Indian women: well-being and survival. Cambridge Journal of
Economics Vol. 7: 363-380.
Lardoux, Solène and van de Walle, Etienne. 2002. Polygyny and fertility in rural Senegal.
Population (English edition, 2002) Vol. 58, No. 6: 717-743.
Lavinas, Lena and Marcelo Nicoll (2007). Activity and vulnerability: what family arrange-
ments are at risk? Dados Vol.3 No.se.
Leach, F. (2003). Practising Gender Analysis in Education. Oxford UK, Oxfam.
Lean Lim, L.; K. Landuyt; M. Ebisui; M. Kawar and S. Ameratunga (2003). An Information
Guide – Preventing Discrimination, Exploitation and Abuse of Women Migrant Workers.
Geneva: ILO http://www.ilo.org/public/english/employment/gems/advocacy/protect.htm
Lee, G. R., M.C. Willets and K. Seccombe (1998). Widowhood and depression: gender
differences. Research on Aging Vol. 20: 611–30.
Lehrer, E. L. and C. U. Chiswick (1993) Religion as a determinant of marital stability.
Demography, Vol. 30, No. 3: 385-404.
Levanon, A.; P. England and P. Allison (2009). Occupational feminization and pay: assessing
causal dynamics using 1950-2000 U.S. census data. Social Forces Vol. 88, No. 2: 865-91.
Lin, M.-J. and M.-C. Luoh (2008). Can hepatitis B mothers account for the number of
missing women? Evidence from three million newborns in Taiwan ? American Economic
Review Vol. 98, No. 5: 2259–73.
Lin, W. and Y. Zhao (2010). The Effect of Child Gender on Children’s Living Arrangements
in China: Evidence from China Census Data. Peking University, January 2010.
http://www.wanchuanlin.org/papers/son_china.pdf.
Linneman, T. J. (2011). Social statistics: the basics and beyond. New York, Routledge.
Lorber, J. (1994). Paradoxes of Gender. New Haven CT: Yale University Press.
Maitra, P. (2004). Effect of socioeconomic characteristics on age at marriage and total
fertility in Nepal. The Journal of Health, Population and Nutrition, 22: 84-96.
Malawi (2009). Gender in Malawi. Analytical Report 3 of the 2008 Census.

232
http://www.nso.malawi.net/index.php?
option=com_content&view=article&id=127%3A1998-population-and-housing-
census&catid=8&Itemid=3. Accessed online on July 30, 2011.
Manda, S. and R. Meyer (2005). Age at first marriage in Malawi: A Bayesian multilevel
analysis using a discrete time-to-event. Journal of the Royal Statistical Society. Series A
(Statistics in Society) Vol. 168, No. 2: 439-55.
Mansuri, G. (2006). Migration, school attainment, and child labor: evidence from rural
Pakistan. Policy Research Working Paper Series, No. 3945. Washington DC, World Bank.
Martin, S. (2007). Women, Migration and Development. Washington DC, Institute for the
Study of International Migration, Policy Brief #1/June.
Mashhour, A. (2005). Islamic law and gender equality: Could there be a common ground? A
study of divorce and polygamy in sharia law and contemporary legislation in Tunisia and
Egypt. Human Rights Quarterly Vol. 27, No. 2: 562-96.
Mason, K. O. et al. (1995). Determinants of women’s power and autonomy in five Asian
countries. Paper presented to the Annual Meeting of the Population Association of America,
April, San Francisco.
Maurer-Fazio, M.; R. Connelly; L. Chen and L. Tang (2011). Childcare, eldercare, and labour
force participation of married women in urban China, 1982–2000. Human Resources Vol. 46,
No. 2: 261-94.
McAlister, C. and T. F. Baskett (2006). Female education and maternal mortality: a
worldwide survey. Journal of Obstetric Gynaecology Canada Vol. 28, No. 11: 983–990.
McKinnon, S.; J. E. Potter and V. Garrard-Burnett (2008). Adolescent fertility and religion in
Rio de Janeiro, Brazil in the year 2000: the role of Protestantism. Population Studies Vol. 62,
No. 3.
Medeiros, Marcelo and Joana Costa. (2006). Poverty among women in Latin America:
Feminization or over-representation? Brasília DF, International Poverty Center, Working
Paper 20.
Miller, C. and S. Razavi (1995). From WID to GAD: Conceptual shifts in the women and
development discourse. Geneva, United Nations Research Institute for Social Development,
Report No. 1651.
Miller, C. and S. Razavi (1998). Gender Analysis: Alternative Paradigms. New York NY,
UNDP, Gender in Development Programme, Monograph 6.
Mitze, T. (2012). Empirical Modelling in Regional Science: Towards a Global Time/Space
Structural Analysis. Berlin, Springer.
Möller-Leimkühler, A.M. (2003). The gender gap in suicide and premature death or: why are
men so vulnerable? Eur Arch Psychiatry Clin Neurosci, 253: 1—8.
Mont, D. (2007). Measuring Disability Prevalence. Social Protection Working Paper 0706.
Washington DC, World Bank. http://siteresources.worldbank.org/DISABILITY/Resources/
Data/MontPrevalence.pdf
Moriarty, C. and F. Scheuren (2001). Statistical matching: pitfalls of current procedures. ASA
Proceedings of the Joint Statistical Meetings, American Statistical Association.

233
Moultrie, T. and I. M. Timaeus (2001). Fertility and living arrangements in South Africa.
Journal of Southern African Studies Vol. 27, No. 2, Special Issue on Fertility in Southern
Africa: 207-223.
Mozambique (2011). Report of 2008 Census. National Statistics Office of the Government of
Mozamique. http://www.ine.gov.mz. Accessed online on July 30, 2011.
Murphy, R. (2003). Fertility and distorted sex ratios in a rural Chinese county. Population
and Development Review Vol. 29, No. 4: 595–626.
Murthi, M.; A.-C. Guio and J. Drèze (1995). Mortality, fertility, and gender bias in India: a
district-level analysis. Population and Development Review Vol. 21, No. 4: 745-782.
Mutharayappa, R.; M. Kim Choe; F. Arnold and T. K. Roy (1997). Son preference and its ef-
fects on fertility in India. New Delhi, National Family Health Survey Subject Reports no. 3.
NIPORT; ORC Macro ; Johns Hopkins University and ICDDR.B (2003). Bangladesh Mater-
nal Health Services and Maternal Mortality Survey 2001. Dhaka and Calverton MD, NI-
PORT, ORC Macro, Johns Hopkins University and ICDDR.B.
Neuman, S. and R. L. Oaxaca (1998). Estimating Labour Market Discrimination with Selecti-
vity Corrected Wage Equations: Methodological Considerations and an Illustration from Is-
rael. London UK, Centre for Economic Policy Research Discussion Paper 1915.
Nguelebe, O. (2005). Situation des femmes en Centrafrique en 2003. (Situation of Women in
the Central African Republic in 2003). Ministère du Plan, de l’Économie, des Finances, du
Budget et de la Cooperation International. République Centrafricaine.
Niger and Macro International (2007). Niger: DHS, 2006 – Final Report. Calverton,
Maryland.
Oaxaca R. L. (1973). Male-female wage differentials in urban labor market. International
Economic Review Vol. 14, No. 3: 693-709.
Oaxaca R.L. and M. R. Ransom (1994). On discrimination and the decomposition of wage
differentials. Journal of Econometrics Vol. 61, No. 1: 5-21.
Obermeyer, C. M. (1992). Islam, women and politics: the demography of Arab countries.
Population and Development Review 18: 33-60.
OECD (2011). Doing better for families. Paris, OECD. http://www.oecd.org/document/
49/0,3746,en_21571361_44315115_47654961_1_1_1_1,00.html.
Ogden, Philip E. and Ray Hall (2004). The second demographic transition, new household
forms and the urban population of France during the 1990s. Transactions of the Institute of
British Geographers, New Series Vol. 29, No. 1. Blackwell Publishing and The Royal
Geographical Society. http://www.jstor.org/stable/3804432.
OHCHR; UNFPA; UNICEF; UN Women and WHO (2011). Preventing gender-biased sex
selection. An interagency statement. Geneva, WHO.
Ola, T. M. (2009). The socio-cultural perception and implications of childlessness among
men and women in an urban area, Southwest, Nigeria. Journal of Social Science Vol. 21, No.
3: 205-209.
Oman,D.; D. Reed and A. Ferrara (1999). Do elderly women have more physical disability
than men do? American Journal of Epidemiology Vol. 150, No. 8: 834-842.
Omelaniuk, I. (2002). Gender, Poverty Reduction and Migration. Washington DC, World
Bank.

234
Ono, H. (2003). Women's economic standing, marriage timing, and cross-national contexts of
gender. Journal of Marriage and Family Vol. 65, No. 2: 275-86.
Oster, E. (2005). Hepatitis B and the case of the missing women. Cambridge MA, Harvard
University, Center for International Development Working Paper 7.
Oster, E.; G. Chen; X. Yu and W. Lin (2008). Hepatitis B does not explain male-biased sex
ratios in China. Chicago, University of Chicago Working Paper.
Oster, E. (2009). Proximate sources of population sex imbalance in India. Demography 46
(2): 325–39.
Otoo-Oryorty, N. and S. Pobi (2003). Early marriage and poverty: Exploring links and key
policy issues, Gender and Development Vol. 11, No. 2: 42-51.
Oun, I. and G. Pardo Trujillo (2005). Maternity at Work: A Review of National Legislation.
Geneva, ILO.
Page, H. (1977). Patterns underlying fertility schedules: a decomposition by both age and
marriage duration. Population Studies Vol. 31: 85-108.
Park, C. B. and N. H. Cho (1995). Consequences of son preference in a low-fertility society:
imbalance of the sex ratio at birth in Korea. Population and Development Review Vol. 21: 59-
84.
Parker, E. M. (2006) School enrollment in Lesotho: do grandmothers matter? Poster
presented at the 2006 Meeting of the Population Association of America, Los Angeles, March
2006.
Paudel, S. (1995). Paper presented to the Training Workshop for Women with Disabilities,
22-23 June 1995, Bangkok.
Parrenas, R.S. (2001). Servants of Globalization: Women, Migration and Domestic Work.
Stanford CA, Stanford University Press.
Perry, D. (2002). Status of Employment and Training of Women with Disabilities in Fiji.
Prepared for the International Labour Organisation, Fiji.
Possi, M. K. (1996). Gender and education of people with disabilities in Tanzania. UTAFITI –
Journal of the Faculty of Arts and Social Sciences, University of Dar es Salaam (New Series)
Vol. 3, No. 2. http://archive.lib.msu.edu/DMC/African%20Journals/pdfs/Utafiti/
vol3no2NS/aejp003002NS007.pdf.
Poston, D. L. and K. S. Glover (2005). Too many males: marriage market implications of
gender imbalances in China. Genus Vol. 61, No. 2: 119-140.
Poulsen, H. (2006). The gendered impact of HIV/AIDS on education in South Africa and
Swaziland: Save the children's experiences. Gender and Development Vol. 14, No. 1: 47-56.
Presser, H. (1997). Demography, feminism and science – policy nexus. Population and De-
velopment Review Vol. 23, No. 2: 295-332.
Price, N. L. and K. Hawkins (2002). Researching sexual and reproductive behaviour: a peer
ethnographic approach. Social Science and Medicine Vol. 55, No. 8: 1327-1338.
Puri, S. and R. D. Nachtigall (2010). The ethics of sex selection: a comparison of the attitudes
and experiences of primary care physicians and physician providers of clinical sex selection
services. Fertil. Steril. 93.
Ramírez-Machado, J. M. (2003). Domestic work, conditions of work and employ-ment: a

235
legal perspective. Conditions of Work and Employment Series No. 7. Geneva, ILO.
Rao, J. N. K. (2003). Small Area Estimation. Hoboken NJ, John Wiley & Sons.
Rindfuss, R. R., P. S. Morgan and K. Offutt (1996). Education and the changing age pattern
of American fertility: 1963-1989. Demography Vol. 33, No. 3: 277-290.
Rocheleau, C. M.; S. J. Bertke; J. A. Deddens; A. M. Ruder; C. C. Lawson; M. A. Waters; N.
B. Hopf; M. A. Riggs and E. A. Whelan (2011). Maternal exposure to polychlorinated
biphenyls and the secondary sex ratio: an occupational cohort study. Environmental Health
10: 20. 
Rousso, H. (2003). Education for all: a gender and disability perspective. Paper
commissioned through the World Bank for the UNESCO EFA Global Monitoring Report
2003/4, The Leap to Equality. http://unesdoc.unesco.org/images/0014/001469/146931e.pdf.
Rubin, D. B. (1986). Statistical matching using file concatenation with adjusted weights and
multiple imputations. Journal of Business and Economic Statistics 4: 87-94.
Ruder, A. M. (1985). Paternal-age and birth-order effect on the human secondary sex ratio.
American Journal of Human Genetics Vol. 37: 362-72.
Rutstein, Shea O. and I. Shah (2004). Infecundity, Infertility, and Childlessness in Developing
Countries. Calverton MD, DHS Comparative Reports No. 9.
Sabbadini, L. L. (2008). Official statistics face to the new challenge of gender statistics.
Italian National Statistics Institute (ISTAT). http://unstats.un.org/unsd/statcom/
statcom_08_events/special%20events/New_directions_social/Linda_Laura_Sabbadini_Paper.
pdf.
Sarti, R. (2008). The globalisation of domestic service – An historical perspective. In: Lutz,
Helma (ed.) Migration and Domestic Work. A European Perspective on a Global Theme.
Ashgate, Aldershot.
Sathar, Z., C. Callum and S. Jejeebhoy (2001). Gender, region, religion and reproductive be-
haviour in India and Pakistan. IUSSP, XXIV General Population Conference, Salvador, Bra-
zil, 18-24 August 2001.
Schkolnik, S. (2011). Acerca de la inclusión del enfoque de género en los censos de
población y vivienda. Santiago de Chile, Fondo de Población de las Naciones Unidas
(UNFPA) / Centro Latinoamericano y Caribeño de Demografía (CELADE).
http://www.eclac.cl/publicaciones/xml/3/42923/lcg2484-P_1.pdf.
Schultz, T. (2009). Guidelines on the Presentation of Statistical Maps. UNECE Work Session
on the Communication and Dissemination of Statistics, Warszaw.http://wwwunece.org/stats/
documents/ece/ces/ge.45/2009/crp.1.e.pdf.
Sear, R.; R. Mace and I. A. McGregor (2000). Maternal grandmothers improve nutritional
status and survival of children in rural Gambia. Proc R. Soc Lond 267: 1641-1647.
Semyonov, M. and A. Gorodzeisky (2005). Labor migration, remittances and household
income: A comparison between Filipino and Filipina overseas workers. International
Migration Review Vol. 39, No. 1.
Sen, A. (1990). More than 100 million women are missing. The New York Review of Books,
December 20, 1990.
Sen, G. And G. Grown (1995). Alternative visions, strategies and methods. In: Development
Crisis and Alternative Visions: Third World Women’s Perspectives. New York NY.

236
Shah, N. M.; K. M. Yount; M. A. Shah and I. Menon (2002). Living arrangements of older
women and men in Kuwait. Journal of Cross-Cultural Gerontology 17. Netherlands: Kluwer
Academic Publishers. http://www.springerlink.com/content/ h7x31n5880303317/fulltext.pdf.
Shaikh, K.; K. M. A. Aziz and A. I. Chowdhury (1987). Differentials of fertility between
polygynous and monogamous marriages in rural Bangladesh. Journal of Biosocial Science
19: 49-56.
Shindler, J. and B. Fleisch (2007). Schooling for all in South Africa: closing the gap. Review
of Education Vol. 53, No. 135-57.
Shu, X. (2004). Education and gender egalitarianism: the case of China. Sociology of
Education Vol. 77, No. 41: 311-36.
Simpura, J.; C. Tigerstedt; S. Hanhinen; M. Lagerspitz; H. Leifman; J. Moskalewicz and J.
Törrönen (1998). Alcohol misuse as a health and social issue in the Baltic sea region. a
summary of findings from the Baltica study. Alcohol and Alcoholism 34 (5): 805-24.
Singh, I. and S. Singh (2007). Does sex ratio indicate the extent of male preference ? A quest
for better indicator. MPRA Paper 16470, http://mpra.ub.uni-muenchen .de.
Skoufias, E. (2005). A Poverty Map for Guyana Based on the 2002 Population and Housing
Census. Washington DC, World Bank.Working Paper Retrieved from
(http://web.worldbank.org/WBSITE/EXTERNAL/COUNTRIES/LACEXT/EXTLACREGT
OPPOVANA/
0,,contentMDK:22416595~pagePK:34004173~piPK:34003707~theSitePK:841175,00.html).
Snyder, A. R.; D. K. McLaughlin and J. Findeis (2006). Household composition and poverty
among female-headed households with children: differences by race and residence. Rural
Sociology 71(4):597-624.
Sørensen, N. (2005). Migrant Remittances, Development and Gender. DIIS Brief,
Copenhagen: Danish Institute for International Studies (DIIS)
http://www.diis.dk/graphics/Publications/Briefs2005/nns_migrant_remittances.pdf.
Ståhlberg, A.-C. et al. (2008). Retirement income security for men and women. Technical
Report 23, International Social Security Association – ISSA.
Statistics Canada. (5th edition) (2006). Women in Canada: A Gender-based Statistical Report.
Ottawa: Statistics Canada. http://www.statcan.gc.ca/pub/89-503-x/89-503-x2005001-
eng.pdf.
Stevens, David (1990). New evidence on the timing of early life course transitions: the
United States 1900-1980. Journal of Family History Vol. 15, No. 2:163-78.
Stock, J. H. and M. W. Watson (2010). Introduction to Econometrics. 3rd ed. Boston MA,
Addison-Wesley.
Stone, R.; B. Purkayastha and T. A. Berdahl (2006). Beyond Asian American: Examining
conditions and mechanisms of earnings inequality for Filipina and Asian Indian women.
Sociological Perspectives Vol. 49, No. 2.
Stubbs, D. and S. Tawake. (2009). Pacific Sisters with Disabilities: at the Intersection of
Discrimination. UNDP Pacific Centre.
Sullivan, D. F. (1971). A single Index of mortality and morbidity. In: HSMHA – Health
Reports Vol. 86 No. 4: 347 – 354.
Surtees, R. (2003). Negotiating violence and non-violence in Cambodian marriages. Gender

237
and Development Vol. 11, No. 2: 30-41.
Tanzania National Bureau of Statistics. (2010). Tanzania 2008 Disability Survey Report.
http://www.nbs.go.tz/nada3/index.php/ddibrowser/12/reports.
Teachman, J. D., L. M. Tedrow and K. D. Crowder (2000). The changing demography of
America’s families. Journal of Marriage and Family 62 (4). National Council on Family
Relations. Retrieved from (http://www.jstor.org/stable/1566733).
Tempelman, D. (1999). Integration of Gender Concerns in Agricultural Data Collection -
Conceptual/Methodological Issues. Rome, FAO.
Tenney, F. (1930). Roman Census Statistics from 508 to 225 B.C. American Journal of
Philology Vol. 51, No. 4: 313-324.
Thomas, D. (2001). Evolving family living arrangements of Canada’s immigrants. Canadian
Social Trends. Statistics Canada, Catalogue No. 11-008.
Tietze, C. (1957). Reproductive span and rate of conception among Hutterite women.
Fertility and Sterility 8:89-97.
Tkocz, Z. and G. Kristemen (2010). Commuting distances and gender: a spatial urban model.
Geographical Analysis Vol. 26, No. 1.
Tu, P. (1991). Birth spacing patterns and correlates in Shaanxi, China. Studies in Family
Planning Vol. 22, No. 4: 255-263.
UNDP (2009). 2009 Human Development Report. New York NY, United Nations Develop-
ment Programme. Retrieved at http://hdr.undp.org.
UNDP (2010). Human Development Report 2010. New York, UNDP.
UNECE (2012 a). Analysis of gender indicators based on census data from countries of the
Commonwealth of Independent States: note by the Interstate Statistical Committee of the
Commonwealth of Independent States (CIS-STAT). ECE, Conference of European
Statisticians, Geneva, March 12-14.
UNECE (2012 b). Census data by gender, Montenegro 2011. ECE, Conference of European
Statisticians, Geneva, March 12-14.
UNECE (2012 c). Gender aspect in population census in Finland: note by Statistics Finland.
ECE, Conference of European Statisticians, Geneva, March 12-14.
UNECE/UNDP (2004). Report on the status of official statistics related to gender equality in
Eastern Europe and CIS countries. Geneva, United Nations.
UNECE/World Bank (2010). Developing Gender Statistics: A Practical Tool. Geneva,
United Nations Economic Commission for Europe.
UNESCO (2004). Guide to the Analysis and Use of Household Survey and Census Education
Data. Montreal, UIS.
UNESCO (2006). Education for All in LDCs. Paris, UNESCO. Retrieved from
(http://unesdoc.unesco.org/images/0014/001472/147259M.pdf).
UNESCO (2010). Global Education Digest 2010: Special Focus on Gender. Montréal, UIS/
UNESCO.
UNFPA (2004). Engendering Population Census in South and West Asia – Collected Papers.
Edited by Wasim Zaman & Rafiqul Huda Chaudhury. Kathmandu, Nepal: UNFPA Country
Technical Services Team for South and West Asia.

238
UNFPA. (2006). State of the World Population 2006. A Passage to Hope: Women and
International Migration. New York NY, UNFPA.
UNFPA (2007). Sex selection: pervasiveness and preparedness in Nepal. Paper presented at
the 4th Asia Pacific Conference on Reproductive and Sexual Health and Rights, Hyderabad.
UNFPA (2010 a). Sex ratio at birth in South East Asia with a focus on Vietnam. An annotated
bibliography designed to guide further policy research. Ha Noi, UNFPA.
UNFPA (2010 b). Guidance Note on Prenatal Sex Selection. New York, UNFPA, Publ.
26043.
UNFPA (2010 c). Sex ratio at birth imbalances in Viet Nam: evidence from the 2009 Census.
New York, UNFPA.
UNFPA (2011). China’s 2010 Census: population dynamics and development. Policy Brief
following the Expert Panel for UNFPA Positioning. New York, UNFPA.
UNFPA India (2009). Why do daughters go missing ? Frequently asked questions on pre-
natal sex selection in India. New Delhi, UNFPA.
UNFPA India (2010). Trends in sex ratio at birth and estimates of girls missing at birth in
India. New Delhi, UNFPA.
UNFPA India (2011). Gender mainstreaming in Census of India – 2011. New Delhi,
UNFPA.
UNFPA Uganda (2011). Note to the File: Meeting for the Marriage and Divorce Bill
coalition to review the draft Memorandum of the Bill. Compiled by Peace Acema Badaru.
UNFPA Viet Nam (2011). Son preference in Viet Nam: ancient desires, advancing
technologies. Qualitative Research Report to Better Understand the Rapidly Rising Sex Ratio
at Birth in Viet Nam. Unpublished document. Ha Noi, UNFPA.
UNFPA/Harvard School of Public Health (2010). A Human Rights-Based Approach to
Programming: Practical Implementation Manual and Training Materials.
UNFPA/IOM. (2006). Women on the Move. Expert Group Meeting on "Female Migrants:
Bridging the Gaps throughout the Life Cycle". New York NY, UNFPA and IOM.
UNFPA/IPEA (2007). Potential contributions to the MDG Agenda from the perspective of
ICPD: reference guide to evidence for policy dialogue in the LAC Region.
http://www.unfpa.org/ public/publications/pid/2599.
UNFPA / UN Statistics Division (2009). Conference Report (Dakar, Senegal, 11 to 13
November 2009): Needs Assessment Conference for Census Analysis (NACCA). United
Nations Population Fund, New York.
UNICEF (2001). Early Marriage: Child Spouses, UNICEF Innocenti Research Centre.
Florence: UNICEF.
UNICEF (2005). Early marriage, a harmful traditional practice. A statistical exploration.
New York: UNICEF.
UNICEF (2007). State of the World’s Children. New York NY, UNICEF. Retrieved on
August 4, 2011 (http://www.unicef.org/sowc07/lifecycle/index.html).
UNICEF (2009). CEDAW and the Reporting Process to the Committee on the Elimination of
Discrimination against Women. Gender, Rights and Civic Engagement. Division of Policy
and Practice. New York NY, UNICEF.

239
UNICEF (2011 a). Child Info. Available from http://www.childinfo.org/marriage_
countrydata.php. Accessed 30 August 2011.
UNICEF (2011 b). Boys and girls in the life cycle. Sex-disaggregated data on a selection of
well-being indicators, from early childhood to young adulthood. New York, UNICEF.
UNIFEM. (2008). Progress of the World’s Women 2008/2009. Who Answers to Women?
Gender and Accountability. New York NY, UNIFEM.
United Nations (1948). University Declaration of Human Rights. Retrieved from
(http://www.ony.unu.edu/news/universal-declaration-of-human.html).
United Nations (1975). Plan of Action of the First World Conference on Women, 1975. New
York NY, United Nations.
United Nations (1979). Convention on the Elimination of All Forms of Discrimination
Against Women, CEDAW, http://www.un.org/womenwatch/daw/cedaw/text/fconvention.
htm.
United Nations (1983). Indirect Techniques for Demographic Estimation – Manual X. New
York, UN Population Division, Population Studies, No. 81.
United Nations (1990). Disability Statistics Compendium, Series Y; No. 4. New York, UN
Statistics Division.
United Nations (1994). International Conference on Population and Development
Programme of Action Retrieved from http://www.un.org/popin/icpd2.htm.
United Nations (1995 a). World Summit on Social Development. Copenhagen. Retrieved from
(http://social.un.org/index/Home/WSSD1995.aspx).
United Nations (1995 b). Beijing Platform for Action. United Nations Fourth World Congress
on Women. Retrieved from
(http://www.un.org/womenwatch/daw/beijing/platform/poverty.htm).
United Nations (2000). The Millennium Declaration. http://www.un.org/millennium/
declaration/ares552e.htm.
United Nations (2001). Guidelines and Principles for the Development of Disability
Statistics. New York, UN Statistics Division, Sales No. 01.XVII.15.
United Nations (2006 a). The World’s Women 2005: Progress in Statistics. New York, UN
Statistics Division, Sales No. E.05.XVII.7.
United Nations (2006 b). Convention on the Rights of Persons with Disabilities.
http://www.un.org/disabilities/convention/conventionfull.shtml.
United Nations (2007). Measuring International Migration through Population Censuses.
United Nations Expert Group Meeting on the Use of Censuses and Surveys to Measure
International Migration. 24-28 September 2007. New York NY, UN Statistics Division.
United Nations (2008 a). Principles and Recommendations for Population and Housing
Censuses, revision 2. New York, UN Statistics Division, Statistical Papers, Series M, No.
67/Rev.2.
United Nations (2008 b). Report of the Global Forum on Gender Statistics.
http://unstats.un.org/ unsd/demographic/meetings/wshops/Gender_Statistics_10Dec07_
Rome/docs/Final_report. pdf.

240
United Nations (2008 c). International Standard Industrial Classification of All Economic
Activities Revision 4. New York, International Labour Organization. ST/ESA/STAT/SER.M/
4/Rev.4.
United Nations (2009 a). World Marriage Data 2008 (POP/DB/Marr/Rev2008). New York,
UN Population Division.
United Nations (2009 b). Report of the Second Global Forum on Gender Statistics.
http://unstats.un.org/unsd/demographic/meetings/wshops/Ghana_Jan2009/L3.pdf.
United Nations (2009 c). Trends in International Migrant Stock: the 2008 Revision. New
York NY, UN Population Division.
United Nations (2010 a). The World’s Women 2010: Trends and Statistics. New York, UN
Statistics Division, http://unstats.un.org/unsd/demographic/products/Worldswomen/WW_ full
%20report_color.pdf.
United Nations (2010 b). Collection of economic characteristics in population censuses:
Technical report, in Measuring the Economically Active in Population Censuses: A
Handbook. UN Statistics Division/ILO. Retrieved from
http://unstats.un.org/unsd/demographic/sources/census/census3.htm.
United Nations (2010 c). Trends in International Migrant Stock: Migrants by Age and Sex.
(www.unmigration.org).
United Nations (2011 a). World Contraceptive Use 2010. POP/DB/CP/Rev2010.
http://www.un.org/esa/population/publications/wcu2010/Main.html.
United Nations (2011 b). World Population Prospects: the 2010 Revision. New York, UN
Population Division.
United Nations (2011 c). International Migration in a Globalizing World: The Role of Youth.
New York NY, UN Population Division.
United Nations (2012). Gender Statistics – a manual for integrating a gender perspective into
statistics. New York NY, UN Statistics Division (forthcoming).
United Nations General Assembly (2004). Women and International Migration, World
Survey on the role of women in development. New York: United Nations.
UNODC (2011). Global Study on Homicide. Vienna, UNODC. http://www.unodc.org/
documents/data-and-analysis/statistics/Homicide/Globa_study_on_homicide_2011_web.pdf.
USAID (2005). Grandmothers: a learning institution. Washington DC, USAID.
US Bureau of the Census (2003). Grandparents living with grandchildren: 2000. Washington
DC, US Government Printing Office.
Vanuatu (2011). Women and Men in Vanuatu. Vanuatu National Statistics Office, Gender
Monograph of the 2009 National Population and Housing Census.
Viet Nam (2011). Viet Nam Population and Housing Census 2009: Age-sex Structure and
Marital Status of the Population in Viet Nam. Hanoi, Ministry of Planning and Investment.
http://vn.one.un.org/en/publications/government-agency-publications/doc_details/227-age-
sex-structure-and-marital-status-of-the-population-in-viet-nam-findings-from-the-2009-
census.html.
Villa, A. (2006). Sex preferences and fertility trends in South Korea. Asia-Pacific Social
Science Review 6 (2): 153–61.

241
Voas, David (2003). Conflicting preferences: a reason fertility tends to be too high or too
low. Population and Development Review 29 (4): 627-646.
Walker, W. M. and S. P. Vajjhala (2009). Gender and GIS: mapping the links between spatial
exclusion, transport access, and the Millennium Development Goals in Lesotho, Ethiopia,
and Ghana. Washington DC, Resources for the Future Discussion Paper 9-27.
Washington Group on Disability Statistics. (2006). Overview of Implementation Protocols for
Testing the Washington Group Short Set of Questions on Disability. http://www.cdc.gov/nchs/
data/washington_group/meeting6/main_implementation_protocol.pdf.
Westoff, C. (2003). Trends in marriage and early childbearing in developing countries, DHS
Comparative Reports No. 5, ORC Macro, Calverton MD.
Wilmoth, J. (2009). The lifetime risk of maternal mortality: concept and measurement.
Bulletin of the World Health Organization Vol. 87: 256-262.
WHO (1998). Women, Ageing and Health. Geneva, WHO.
WHO (2001). ICF – International Classification of Functioning, Disability and Health.
Geneva: World Health Organization. http://www.disabilitaincifre.it/documenti/ICF_18.pdf.
WHO (2004). The World Health Report 2004 - changing history. Geneva, WHO.
http://www.who.int/entity/whr/2004/annex/topic/en/annex_2_en.pdf.
WHO (2011). Preventing gender-biased sex selection. An inter-agency statement by
OHCHR, UNFPA, UNICEF, UN Women and WHO. Geneva, WHO.
WHO. 2012. What do we mean by “sex” and gender”? Gender, women and health. <
http://www.who.int/gender/whatisgender/en/>. Accessed: 27 January 2012.
WHO/UNFPA. (2009). Guidance Note. http://www.who.int/reproductivehealth/publications/
general/9789241598682/en/.
WHO/UNICEF/UNFPA/World Bank (2012). Trends in maternal mortality: 1990 to 2010.
Geneva, WHO.
WHO/World Bank (2011). World Report on Disability. http://whqlibdoc.who.int/hq/2011/
WHO_NMH_VIP_11.01_eng.pdf.
Wolfe, A. B. (1932). Population censuses before 1790. Journal of the American Statistical
Association Vol. 27, No. 180: 357-370.
World Bank (1999). Disability and Poverty: A Survey of the Literature. Washington DC,
World Bank.
World Bank (2006). Global Economic Prospects 2006: Economic Implications of Remittan-
ces and Migration. Washington DC, World Bank.
World Bank (2011). Definitions. World Bank Facts. Retrieved from
(http://www.nationsonline.org/oneworld/GNI_PPP_of_countries.htm).
Wu, Z.; K. Vilsainen and E. Hemminki (2006). Determinants of high sex ratio among
newborns: a cohort study from rural Anhui province, China. Reproductive Health Matters
Vol. 14, No. 27: 172–80.
Yi, Z.; T. Ping; G. Baochang; X. Yi; L. Bohua and L. Yongping (1993). Causes and
implications of the recent increase in the reported sex ratio at birth in China. Population and
Development Review Vol. 19, No. 2: 283-302.
Yinger, N. V. (2006). Feminization of Migration. Washington DC, Population Reference
242
Bureau.
Yodanis, C. (2005). Divorce culture and marital gender equality: A cross-national study.
Gender and Society Vol. 19, No. 5 (October): 644-59.
Zeng, Y.; P. T. Schultz; D. D. Wang and D. Gu (2002). Association of divorce with socio-
demographic covariates in China, 1955-1985: event history analysis based on data collected.
Demographic Research Vol. 7, No. 11: 407-432.
Zhu, W.X.; L. Lu and T. Hesketh (2009). China’s excess males, sex selective abortion, and
one child policy: analysis of data from 2005 national intercensus survey. British Medical
Journal 338: 1211.

Next page:
Check line 12 for all countries
Check lines 24-29 for all countries
Check what’s up with the ? and 1 entries
Check 51 for LAO and 37 for SWZ, 33 for TLS
Review PHP
Shouldn't F/M be fm ?
Check coding of the item on relation to pregnancy for consistency

243
APPENDICES
Appendix 1: Gender-Relevant Issues in 2005-2014 Census Forms

Clarifications
Numbers refer to the number of alternatives that can be selected (sometimes involving more than one question),
e.g. 2 means that the question has an answer such as yes/no, male/female, etc. The plus sign (+) is used for
questions that have open-ended answers. A number followed by a plus sign indicates a certain number of
precoded categories plus an open-ended answer.

Fertility/Mortality

Children Ever Born Alive, Surviving and Born During Last 12 Months: S indicates differentiation by sex and D
indicates that the criterion was the date of birth of the last born child. B means that both the date of birth of the
last child and the number of children born during the past 12 months were asked.

Marital status: P after the number of categories indicates that polygamous unions can be identified; C means that
informal/consensual unions can be identified.

Date/Age at first marriage: D = Date, A = Age

Household/Dwelling

Criteria for selection of head of household: N = None specified

A = Authority / decision-making criterion
H = Person who owns home or pays rent
R = Randomly assigned reference person
M = multiple headship considered as a possibility

Relationship to head of household: D after the number of categories indicates that the categorization makes it
possible to identify domestic servants and their families.

Father/mother live in household and also Orphanhood question: F = father, M = mother; fm = either;
FM = both mother and father

In the case of construction materials, the number refers to the number of items that ask about materials used:
external walls, floor, roof, etc. In the case of personal income, the number refers to the number of sources that
can be identified.

Depending on the way the question is asked in different countries, consumer durables may include:
Air conditioner Generator Secondary residence
Bicycle Insurance (home) Sink with piped water
Boat Internet connection Solar collector
Cable Microwave oven Solar water heater
Canoe Motor cycle Stereo set
Cell/mobile phone Motor vehicle Stove
Computer Mower Television
Cuisinière Outboard motor Toaster oven
Daily newspaper Oven (conventional) VCR
Dish washer Parabolic antenna Washing machine
DVD player Radio Water heater (electrical)
Fixed telephone Refrigerator/ freezer Water tank

Income/Poverty

An R after the number of sources for which an amount of personal income was declared indicates that one of the
specific sources declared was Remittances.

244
Disability

The numbers listed here are equal to the number of disability topics addressed by the questionnaire times the
number of alternatives given for each topic (minimum of 2, i.e. does/does not have the particular disability).

ISO codes for the countries in the table

ABW Aruba GUM Guam NPL Nepal
ALB Albania HKG Hong Kong NZL New Zealand
ARE United Arab Emirates HRV Croatia PER Peru
ARG Argentina HUN Hungary PHL Philippines
ASM American Samoa IDN Indonesia PLW Palau
AUS Australia IMN Isle of Man PRK Dem. People’s Rep. of Korea
AZE Azerbaijan IND India PSE Occ. Palestinian Territories
BDI Burundi IRL Ireland PYF French Polynesia
BFA Burkina Faso IRN Iran QAT Qatar
BGD Bangladesh ISR Israel REU Reunion
BGR Bulgaria JAM Jamaica ROM Romania
BHS Bahamas JOR Jordan RUS Russia
BLR Belarus JPN Japan SAU Saudi Arabia
BMU Bermuda KAZ Kazakhstan SDN Sudan
BRA Brazil KEN Kenya SGP Singapore
BRB Barbados KHM Cambodia SLB Solomon Islands
BTN Bhutan KIR Kiribati SLV El Salvador
BWA Botswana KOR Korea, Republic of SRB Serbia
CAN Canada LAO Lao People’s Dem. Rep. SWZ Swaziland
CHE Switzerland LBR Liberia SYC Seychelles
CHN China LCA St. Lucia TGO Togo
COG Congo LSO Lesotho THA Thailand
COK Cook Islands MAC Macao TJK Tajikistan
COL Colombia MDV Maldives TKL Tokelau
CPV Cape Verde MEX Mexico TLS Timor Leste
CRI Costa Rica MLI Mali TON Tonga
CYM Cayman Islands MLT Malta TTO Trinidad & Tobago
CYP Cyprus MNE Montenegro TUR Turkey
CZE Czech Republic MNG Mongolia UKM United Kingdom
DEU Germany MNP N. Mariana Islands USA United States of America
DJI Djibouti MOZ Mozambique VEN Venezuela
DOM Dominican Republic MTQ Martinique VIR US Virgin Islands
DZA Algeria MUS Mauritius VNM Viet Nam
ECU Ecuador MWI Malawi VUT Vanuatu
EGY Egypt MYT Mayotte WLF Wallis and Futuna
ETH Ethiopia NCL New Caledonia WSM Samoa
FJI Fiji NFK Norfolk Island ZAF South Africa
FRA France NGA Nigeria ZMB Zambia
GHA Ghana NIC Nicaragua
GNB Guinea Bissau NIU Niue

245
 Questionnaire Items ALB DZA ASM ARG ABW AUS AZE BHS
1. Children Ever Born Alive (by sex) + +S + +S + + +
2. Children Surviving (by sex) + +S + + +
Fertility/ Mortality

3. Ch. Born in 12 Months / Date of Last (by sex) +S D D +

4. Maternal/paternal orphanhood
5. Number of persons deceased in household +
6. Sex of deceased 2
7. Month/year of birth or age of deceased +
8. Month/year of death of deceased +
9. Relation to pregnancy +
10. Date of / Age at first marriage D A
S
M

11. Marital status 5 4 5 6 4C 5 6C 9C

12. Criterion for headship selection R N H N H N N N
13. Relationship to head of household 16 ? 14 9D 6+ 11 16D
Household / Dwelling

14. Father/mother lives in household F FM

15. Grandchildren living in home / cared for 2 2
16. No. of items asking for construction materials 3 5
17. Presence of kitchen inside the home 5 1 2
2 3
18. Source of heating for cooking 2 6
19. Origin of water consumed in household 3 7
6 6
20. Status of occupation of home 5 5 8 5 7+
4 7 8
5 5+
5+
21. Presence of consumer durables 16 11 8 4 9 2 4 9
22. Health insurance coverage 8 5
Inc. / Poverty

23. Receiving social security payments + 4 2

24. Amount of personal income (no. of sources) 10R 1
25. No. of sources of pers. income without amounts 1 10R
7+
26. Only amount or sources of household income 11
27. Personal ownership of assets 7R
28. Literacy 2 + 2 2
29. School attendance and grade or level + + 9 10
Educ

14 12 2 4
30. Last grade or level attended/completed 11 + 14 9 12 6 11 22
31. Ability to speak national language 2 4
32. Business use of the dwelling 2 2
33. Care-taking of dependent children 3 8
34. Care-taking of disabled/sick household members 8 2
35. Currently working or employed 2 5
3 9+
Work / Economic Activities

36. Reasons for not working 5 7 2 6 2

37. Any work done for pay 8
38. Disposition / availability to work 2 4 2
39. Looking for work 3 2 4
2 2
40. Main occupation 2 3
41. Main activity within occupation 2 + 2 + + 4
42. Hours per week worked + + + + + + 13 +
43. Status in employment + + +
44. Institutional sector + 7 10 2
8 6
45. Sector of activity / Industry 4 6 4 2 6+
46. Contribution to social security + 9+ 11
3 + 4
+ + +
3
47. Place of birth + + + + 8 7+ + +
48. Residence at fixed time in the past + + + + 9 + ? +
Migration

49. Citizenship + + 6 8 8 2 + +
50. Number of persons living abroad + + +
51. Sex of departed 2 2 2
52. Age at time of departure or at present + + +

246
 53. Difficulty hearing / seeing 8 4 4 8
54. Difficulty understanding / concentrating 8 2 2 8
55. Dif. of motion/dressing/bathing/doing errands 8 8 4 8
Disability

56. Other kinds of disability 7+

57. Generic disability question
58. Does the person need help in performing tasks ? 9
9
59. Ethnicity or race + + + 3+ 8+
60. Religion + 9 9+ +
Misc.

61. Language spoken at home + 7 7+

62. Means of transportation to work or school 12 9 7+
6

BGD BRB BLR BMU BTN BWA BRA BGR BFA BDI KHM CAN CPV CYM
1. +S + + +S +S +S +S +S +S +S 8
2. +S +S +S +S +S +S +S +S
Fertility/ Mortality

3. + 6 +S +S B +S +S +S B B
4. FM M FM FM FM FM
5. + + + + + + +
6. 2 2 2 2 2 2 2
7. + + + +
8. 12 + +
9. 2 4 + +
2 2
10. A
S
M

11. 4 6C 6C 7 6C 6C 8C 4C 8P 8P 5 5C 5C 5C
12. N N R R N N M N N N N N N R
13. 5 10 11 18D 22 14 20D 12 10 11 + 13+ 11D
Household / Dwelling

14. FM M FM FM FM
15. 1
16. 2 1 3 3
17. 1 1 2 3 1 3 1
18. 3 4 2
8 8 8
19. 5+ 12+ 7 6+ 4+
20. 7 5+ 11+ 10 4 7 7+ 3 3+
3 6 7+ 5 7
3 11 6 3+ 6+
5+ 6 2
21. 22 3 6 14+ 16 10 16 11 6 ? 12
22. 6 5
23. 3 2
Inc. / Poverty

8 2
24. 6 2
25. 1 11 7
26. 12 11 1
27.
2 2
28. 2 2 2 2 3 2+ 6 2 2
29. 2 5 2 27 3+ 10+ 3
Educ

8 7+ + 4 8
30. ? 9 7 16+ 17 + 14+ 15 11 + 9+ 18 11 18+
31. 4 4 2 4

247
 32. 2 4
33. 8 6
34.
35.
5 10 2
Work / Economic Activities

36. 4 9 3 2 9+ 5 2 2 4 8+
37. 9+ 11 11 5+ 7 8 3 11
38. 2 2 2 + 6
39. 2 2 2 2 6 2
40. 2 2 3 9
10 2 2 2 2
41. 2 + + + + + + 11
42. + + + + +
+ + +
43. + 2 8 +
44. + 4
8 5 9 7 +
45. 4+ 7 6 7 7
46. 6 + + 4+ +
+ 3 8 + +
+ + + + 8 +
+ 12+ + 3 +
3 7 8
47. + + + + + + + + + + 13+ 5 17+
48. 6 + + + + + 3+ 5 9+
Migration

49. 2+ 5 14+ + 3+ 3+ 3+ + 2+ 5 7+
50. + + + + + 2 + + + +
51. 2 2 + 2 2
52. + + + + +
53. 4 4 10 4 8 8 2 4 3 4 4
54. 4 6 8 2 4 2 2 2 3 2 4
Disability

55. 2 18 6 2 22 4 2 3 2 4
56. 2 10+ 42+ 2 2 3 3+
57. 10 10 1
58. 8 7
59. 8 8 5+ 4+ 11+
60. 5 23 3 7 + 7+ 6 + 3+ 11 20
Misc.

61. + 12+ + 2+ 9
62. 2+
19 7 8 8

CHN COL COG COK CRI HRV CYP CZE PRK DJI DOM ECU EGY SLV
1. +S +S + 4 +S +S +S +S
2. +S +S + +S +S +S
Fertility/ Mortality

3. +S +S +S D
4. FM FM
5. + + +
6. 2 + 2 +
2 2
7. + 2 + +
8. + + + +
+ +
9. 2 + 3
+ 4
10. D A
MS

11. 4 6 6P 6 6C 4C 4 6P 6C 6P 6C
12. N A N A N N N N N N N
13. 9 10D 10D 8 12 8 8 13D 11D 13D
Household / Dwelling

14. FM
15. + 1
16. 3 3
17. 1 3 3 5 1 3 1 3
18. 3 3 2
5 6 6+ 6 9
19. 4 4 5 6 8+
20. 24+ 10 4+ 7 10 10
5 4 7 8+ 5 4
8 5 7 7 8
8

248
 21. 24 21 24 4 2 2 15+ 16 22 13
22. 7
23. 7
Inc. / Poverty

6 6 8
24. 12 12
25.
26. 7
27.
2 2
28. 2 3 2 8 2 2
29. 5 3+ 7 3 4+ 3
Educ

2 2 5
30. 7 + + + 8 6 4+ 9 9
31. + 11 2
2
32. 3 2 6
33.
34.
35.
Work / Economic Activities

36. 3 2 2 2 4 2 2 2 2 15 4
37. 9 8 4 7 6
38. 2 7+
2 2 5 6 2 2 6
39. 2 2 2 2 2
40. 7 2 4
2 7 + 2 +
41. + + +
42. + + + + + 7 +
+ + +
43.
44. + + 7 7
+ +
45. 9 9 9 6 3 + 5
46. + 5 + 6
+ + + +
+ +
47. 3+ + + + + 3 + + 3+
48. 2+ + + + + + +
Migration

49. + + 2 +
50. + + +
51. +
2 2 2
52.
+
53. 4 8 2 4 4 4
54. 4 4 4 2
55. 2 4 16 4
Disability

56. 4 16 2
57. 2 6+
58. 8
+ 8 +
4
59. + 5+ + 5+ 10 + 4 10+
60. 9+ 9 9+ 3+ 4
Misc.

4
61.
62. 9

ETH FJI FRA PYF DEU GHA GUM GNB HKG HUN IND IDN IRN IRL
1. +S + +S + +S +S +S +S +
2. +S +S +S +S +S +S
Fertility/ Mortality

3. +S +B +S +S +S
4. +
5. FM
+ + +
6. 2 2
7. 2
+ + +
8. +
9. +
2 2
10. A A
M

11. 6 4PC 4PC 7C 6C 5 6PC 5 5 4 4 7

249
 12. N N R N N N H N R N N N N
13. 11 10 11 14 + 5+D + 9D 12 +
Household / Dwelling

14. FM M
15. + 2
16. +
4 1 3 3 3 3 1
17. 3 2
18. 7+ 3 8+ +
7+ 9+ 7
19. 6+ 6 5 7
20. 7 7+ 3 14+ 6 14
7 5+ 4 6 5
6+ 4 6
9 8
4
21. 3 23 4 11 4 8 10 12 3
22. 7+
23. 2
Inc. / Poverty

24. 3 8R
25. 1R 5
26.
27.
2 2
28. 2 2 7 2 2 4 2
29. 2 4 6 2 3 8 3 + 8 3 3
Educ

30. + 9 17 12 14 5 + + 9 +
31. 18 2 2 2 13
2 2
32. 9+ 2
33. 2
34.
35.
Work / Economic Activities

36. 2 2 9 4 3 8 7+
37. 7 6+ 6 6 6 3 7+ 7
38. 4 2 2 2 7 5 2
39. 2 2 2 2 ? 3 2 2
40. 2 2 2 2 2
2 3 ? 2
41. + + + 10+ + + 10 2
42. + 8 + + +
5 + + + +
43. 2 3 +
44. 4 8 7
45. 6+ 6 10 8 4 4 6
+ 4 6
46. + 6 6 + + 5
+ 23 + + 2
+ + 19
+
47. + + 5+ 23 + + 6 + + + +
48. + 3+ 6+ + + + + +
Migration

49. 3 3+ 3+ 16 6 + 2 + 8+ 2+
50. 5 +
51. 2
52. +
53. 4 4 4 4 4 8 4 2
54. 2 2 2 3 4 2
Disability

55. 2 2 2 8 2 8 2
56. 8 3+ 2+ 4 2
57. 8 2
58. 2+
59. 14+ + + 7+
60. 6 + 9 8+ + 10+ 6 6+ 5 5+
Misc.

61. 2 2 +
62. + +
10 9 20+ 10
5

IMN ISR JAM JPN JOR KAZ KEN KIR LAO LSO LBR MAC MWI MDV

250
 1. + + +S +S +S + +S +S +S
2. + +S +S +S +S +S +S
Fertility/ Mortality

3. +S B +S +S +S
4. FM FM FM FM FM
5. + + + +
6. + 2 2 2
2 2
7. + + +
8. +
9. 2 2
4 2
10. D D A A
M

11. 6 4 5 4 6P 4 7P 5 4 4P
12. N N N N N N N N N N R N N
13. 7+ 21D 12D 9D 11 11 + 6 11D + 9 5 10
Household / Dwelling

14. Fm M 2 FM 7
15.
16. 1 3 3 3
17. 3 3 3
2
18. 4
19. 7+ 8 9+ 5+ 7
14+ 14 5
20. 6 7 10+ 7+ 3
10 5 5 5 3+ 3
5 5 5+
21. 13 13 2 13 15 10 6 5 10 13
22. 5
Inc. / Poverty

23.
24.
25. 5 1 1
26. 1 9+ 5R
27.
2
28. 2 2 7 2 4 4
29. 3 7 3 3 3 4 2 2 3 8
Educ

30. 9 + 10 ? 21 13 29+ + + 6 8
31. 9 4 4 8
32.
33. 2
34.
35.
Work / Economic Activities

36. 2 5 2 3 12 13+
37. 9+ 5 4 + 4+ 6 10+
38. 3 2 2 2
39. 2 2 5 2
40. 3 2 2 2
2 6+ 3
41. 2 + +
42. + + + + + +
? 4+ + +
43. + + +
44. + + 5
45. 8 5 + 4
5 8 + 6 8
46. + +
+ + + + 5+ +
10+ ? + 18 +
47. 6+ + + + + + 12 13+ + 8+ 2+
48. + (+) + + 12 13+ 6+ 2+
Migration

4+
49. + + + 2+ + 12 5 + 7 2
50. + +
51.
52.

251
 53. 4 4 4 4
54. 4 2
55. 16
Disability

2
56. 8 4
57. 3
58. 3 ? +
3
59. + + 7 51 + 7
60. 5 3 6+ 8 + 5 + 4
Misc.

61. 8
62. 10 6+
9 5

MLI MLT MTQ MUS MYT MEX MNG MNE MOZ NPL NCL NZL NIC NGA
1. +S + + + + +S +S +S
2. +S + +S +S +S
Fertility/ Mortality

3. +S +S +S
4. FM
5. + +
6. 2 +
2 2
7. + +
8. + +
+ +
9. 2 6 4
10. D A
M

11. 6CP 7 4C 8+C 6PC 8C 6+ 4C 5C 7 4C 6C 6C 5

12. N R R N R N N N N N R R N N
13. 12D 12 + 8 + 8+ 11 13 + 10+ 16 9
Household / Dwelling

14. 4
15. 1
16. 3 3
17. 2 3 1 3 3 3
3 3 2 + 2
18. 4 2 6 3
19. 8 6 5 6 7 7+
7+ 5+ 6 6 7+ 10
20. 8 8+
7+ 5 4 4 9 7 5+
7 4 4 3
21. 20 13 17 6 2 8 5 16 15
22. 2
23. 2
Inc. / Poverty

24. 5
25. 7 3+
26. 13
27. E
2
28. 2 2 2 2 2 3 3 2
29. 2+ 4 2 5 3 2 2 18 3 3 2 6
Educ

30. + 8+ 18 34+ 12 12 9 13 10 5+ 7 8+
31. 2 2 2 8
2 6 8

252
 32. 2 6
33.
34.
35.
4
Work / Economic Activities

36. 2 7 7 8 3 3 2 6+ 2
37. 9 2 4 5 7 13 6+
8 5 7
38. 2 + 2+ 7 2 2+
39. 2 6
2 2 4+ 2 2 2 9
40. 2 2 2 5 2 2 3 2
41. 3 + 2 6 +
+ + + + + + + + +
42. + + + + +
43. + + + +
+ + + 9
44. 4 6 +
45. 11 7+ 10 + 5 9 4 +
9 4 +
46. 12 + 3 + 7 12 5
12 + + 9 2
5 + + +

47. + 10+ + 5+ 4+ + 4 + 5+ 8+ 3+ +
48. 2+ 3+ 3 3+ 6+ + 4 + 4+ 4+ 3+
Migration

49. + 4 3+ 8 5+ 2+ 2+ 3+
50. 5+
+ + +
51. 2 2
52. + +
53. 4 6 4 4 4 4 2
54. 2 6 2 2 2 2 2
Disability

55. 2 12 4 2 2 9 2
56. + 3 3 2 1
57. +
58. 6 6
59. + 6 + 7+ + 9+ 8+ 17
60. 5 + + 2+ + 11+ 8
Misc.

61. 21 8+ 7+ 2 2
62. 4
5 10+
5

NIU NFK MNP PSE PLW PER PHL QAT KOR REU ROM RUS SGP LCA
1. +S + +S + + +S + +S
2. +S +S + + +S +S
Fertility/ Mortality

3. +S D B
4.
5. + +
6. 2 2
7. + +
8.
9.
2
10. A D A
M

11. 5 6 5 6 6 6C 6C 4 4 4C 5 5
12. N N H N H N N N N R N N
13. + + 14 10 12D 12D 21D 10D 14 + 11 12D+
Household / Dwelling

14. F/M
15. 2 ?
16. 3 1
17. 1 3 2 2 2
+ 3 3 2 3
18. 3 5
19. 4+ 5 6 8 8+
6+ 3 6 5 4
20. 6 6 8 4 7+
4 6 4 6 8
5 4+ 6 4

253
 21. 8 14 11 7 1 2 19
22. 7+ 8 4 7+
23. +
Inc. / Poverty

24. 6
3 8R 12R 2R
25. 1 11R+
26. 11+ 11+
27.
+
28. 3 2 2
29. 6+ 2 3 4 3 2 2 7 5 2 2 2
Educ

30. 4 14 10 20 8 22 12 8 18 12 10+
31. 4 2
32. 2 5+
33. 9+
34.
35.
Work / Economic Activities

36. 5 9 7
37. 5+ 7 5 4+ 14+
38. 7 3 2 2 3
39. 2
2 3 4 2
40. + 2 2 2
41. 2 2 2 2 3
+ + + + + +
42. + + + + +
43. + + +
+ +
44. + 7 + +
45. 6 6 4
5 9 5 9+
46. 5 +
4 + 9 6 +
9 + 5 +
+
47. + + 6 3+ 3 + + + +
48. 3 + + + 5+ 3+ + +
Migration

49. 6 6 ? 3+ 2+ +
50. + + + +
+ +
51. 2 2 2
52. + +
53. 4 16 2 8 2 16
54. 2 16 2 4 2 4
Disability

55. 8 4 2 4 2 16
56. 4 4 2
57.
58. 2 10+
59. 11 + + + 9+
60. + + 3 + 4 9+ 19+
Misc.

61. 6 +
62. 9 10 5

WSM SAU SRB SLB SDN SWZ SYC CHE TJK THA TLS TGO TKL TON
1. +S +S +S +S + + +S +S +S +S
2. +S +S +S +S + + +S +S +S +S
Fertility/ Mortality

3. +S D +S +S + D +S D D
4. FM +S FM FM FM FM
5. + FM + + + +
+ + + +
6. 2 2 2 2 2 2 2
7. + 2 + +
+ + + +
8. + + + +
9. + 3 2 3
2 3
10. A A
M

11. 4 4 6 4 7C 6C 5 5 8C 5 4+

254
 12. N N N N N N R N N R N
13. ? 10D + 10 10 10 14 + 15 9 8 +
Household / Dwelling

14. F/M M
15.
16. 3
17. 3 1 3 1 3 3 3 3
18. 2
19. 9 6 6 6+ 5+ 6+ 4+
9 13 4+
20. 6 4 10 13 7 10+ 8+ 4+ 3+
4 3+ 6 4 6 5+ 6+
4 6+ 3+
5
21. 10 5 15 13 22 10 16 12 11+
22.
Inc. / Poverty

23.
24.
25.
26. ? 9R+
27. 9R 8 + 7R+
8R
28. 8 2 8 2 6 2 2 6 4
29. 4 3 4 3 3 2 + 3 3
Educ

13 4 9
30. 13+ 17 13+ 17 37 13 13 + 33 8+ 7+ 7
31.
32. 2
33.
34.
35.
36. 2 2 9 8
Work / Economic Activities

37. 7+ 7 5 6+
38. 5 2
39. 2
2 2 2 2
40. 2 +
41. 3 3 2+
+ + + + +
42. + + + + +
43. + + +
44. 7
45. + 8 3 4
6 5+
46. 5 9 5 3 10 4
5 + +
+ 9
+ + + + +

47. + + + + + + 2+ +
48. + + + 2+ +
Migration

? +
49. + 2 2+ + + 22 +
50. + +
+
51. 2
52.
53. 2 9 16 4 4 16 4 16
54. 2 3 2 2 4 4
Disability

55. 2 3 16 2 6 4 4
56. 3 2 1
57. 8 8
58. + +
+
59. + 5 11 9+ 9+
60. + 12+ 4 3+ +
Misc.

7+ 11+ + 7 13
61. 5 9 3+ 11+ 3+ 6+
62. 10

TTO TUR ARE UKM USA VIR VUT VEN VNM WLF ZAF ZMB

255
 1. +S + +S +S +S + +S +S
2. +S +S +S +S +S +S
Fertility/ Mortality

3. B +S + +S +S +S
4. FM FM FM
5. + + + +
6. 2 2 2 2
7. + + + +
8. + + +
9. 3 5 3 3
10. A
M

11. 6 4 5 6 6C 5 4C 6C 6C
12. N A N H H N N N R N N
13. 14 9 11D 14 14 + 10 6 + 14 13
Household / Dwelling

14. M
15.
16. 3
17. 2 3 3 3 1 3
18. 2 3 2
19. 6+ 6 5 6+ 13
6 10 10 14
20. 8+ 5 8+ 4+ 25
9+ 4 7 5+ 6
4 5
21. 13 11 15 20 8 10 14 16
22. 4 7+
Inc. / Poverty

23. 2
24. 7 +
25. 7+ 12
26.
27. 3 7R+
28. 2 2 7+ 24 2
29. 2 3 2+ 9 2
Educ

6 3 4
30. 9 14 17+ + 11 12 67 ?
31. 9 2 3
8
32. 2
33. 2
34.
35.
Work / Economic Activities

36. 3 10 8 2 6
37. 12 5 8 7+ 17
38. 4 6 3 2+ 2
39. 3 + 2 2
40. 2 2 2 2
2 + 5 2
41. + + + +
42. + + + + + +
+
43. +
44. 7 3
45. 3 9
4 8 10 3 4
46. 15 + +
9 + 8 +
+ + + +
+
47. 3 + + + 5+ + +
48. 4+ + + + 5+ 7+ +
Migration

49. 6 3 3+ 2 +
50. +
+
51. 2
52. +
53. 4 4 9 4 16 12 5
54. 9 2 3 4 4 6 3
Disability

4
55. 9 2 8 3 4 4 18
56. 3 2 6 4
57. 5 8
58.

256
 59. 11 15+ 15+ 9 4+ + 5 +
60. 12+ 11+ + 8
Misc.

61. 2+ 5 13 +
62. +
10

Appendix 2: Glossary of important gender terms30

Empowerment – A process through which women and men in disadvantaged positions

increase their access to knowledge, resources, and decision-making power and raise their
awareness of participation in their communities in order to reach a level of control over their
own environment.

Gender – The roles and responsibilities of women and men that are socially rather than
biologically determined. Gender explains the way in which social and cultural values and
traditions determine how women and men are perceived and expected to think and act.

Gender Analysis – An analytical tool that examines the differences and disparities in the roles
that women and men play, the power imbalances in their relations, their needs, constraints
and opportunities and the impact of these differences on their lives. Gender analysis may be
conducted at all stages of an intervention, from priority-setting and data collection to the
design, implementation and evaluation of policies or programmes.

Gender and Development – An approach to development based on the premise that all
policies, programmes, and projects should reflect the needs, priorities, roles and the impacts
of development processes on men and women.

Gender –Based Indicators – Indicators that can gauge the extent to which gender objectives
have been achieved.

Gender Bias – The perception that the other sex is not equal and hence does not have the
same rights.

Gender Blindness – Failure to recognize that gender is an essential determinant of social

outcomes, including education, health and the standard of living.

Gender Difference –

Gender (or Sexual) Discrimination – Prejudicial treatment of a person on the basis of a

gender stereotype.

Gender Disparity – Differences in men’s and women’s access to services, status, and power
which usually favour men and are institutionalized through laws and social customs.

Gender Equality – Absence of discrimination on the basis of a person’s sex in opportunities,

the allocation of resources or benefits, or in access to services.

Gender Equity – Fairness and justice in the distribution of responsibilities and benefits
between men and women. It recognizes that men and women have different needs and power
30
Based on UNSD/CAPMAS (2000). Gender and Development: An Information Kit I. Cairo, CAPMAS.

257
and that these differences should be identified and addressed in a manner that rectifies that
imbalance between the sexes. Check if these three correspond to the definitions given in
Chapter 1.

Gender Gap – The disparity measured quantitatively between women and men and girls and
boys in their education, health, access to resources and services. Don’t we want a more
specific definition, with respect to the measurement ?

Gender Mainstreaming – The integration of gender concerns into the analysis, formulation
and monitoring of policies, programmes and projects. The objective is to achieve greater
equality and thereby reduce gaps between women and men in opportunities and benefits.

Gender Perspective – A framework of analysis to assess how women and men affect and are
affected by policies, programmes, projects and activities in any development intervention.

Gender Roles – Specific economic and social roles that a society considers appropriate for
men and women. Gender roles and responsabilities vary between cultures and may change
over time. Men are generally identified with productive roles. Women have a triple role:
reproductive, including domestic responsibilities, productive work and community work.
These are often carried out simultaneously. Women’s roles tend to be undervalued and
excluded from national income accounts in almost all societies.

Gender Sensitivity – The ability and willingness to perceive existing gender issues, gaps and
inequalities and incorporate these into programmes, strategies and actions.

National women machinery – Governmental body (bodies) entrusted with drawing policies,
strategies or plans for improving women’s welfare. This body may also be responsible for
designing and implementing interventions and for coordinating women-focused interventions
at the national level.

Participation – Inclusion of all segments of society, particularly the marginalized, in the

process and benefits of development. The purpose is to improve self-reliance through
increased accessibility to resources and opportunities in order to improve welfare and the
standard of living.

Sex – The genetic and physiological characteristics and traits that indicate whether one is
male or female.

Women in Development (WID) – The development framework that recognizes the distinct
needs and capacities of women. It focuses on developing strategies and action programmes
that will facilitate their participation in the productive sector.

258
Appendix 3: Mapping of Resources on Gender Statistics, Relevant
to the Gender Analysis of Census Data

CensusInfo

User-friendly database system for the dissemination of population and
housing census results, designed by UNSD, UNICEF and UNFPA, adapted
from DevInfo database technology.
 Data is presented in several forms (tables, graphs and maps).
 Useful tool for identifying disparities and priority groups.
 Database administrators can add their own sets of national, regional and
local indicators to their databases.
 Website: http://www.censusinfo.net/
ECLAC Gender Equality Observatory for Latin America and the Caribbean
 Provides a number of regional and national indicators.
 Main areas of concern: Paid and unpaid work, time use and poverty,
Access to decision-making and political representation, gender violence,
health, and reproductive rights.
 Website: http://www.eclac.org/oig/default.asp?idioma=IN
International Statistical Agencies
 Provides a list with links to National Statistics Offices (NSOs) around the
world.
 Website: http://www.census.gov/aboutus/stat_int.html
Gender, Institutions and Development Data Base (GID-DB)
 Data base developed by the Organisation for Economic Co-operation and
Development (OECD), compiled from various sources.
 Combines in a systematic and coherent fashion the current empirical
evidence that exists on the socio-economic status of women.
 Covers a total of 160 countries and comprises an array of 60 indicators on
gender discrimination.
 Measures intra-household behaviour and social norms (family code,
physical integrity, civil liberties, ownership rights, etc).
 Includes the Social Institutions and Gender Index (SIGI). Launched in
March 2009, it complements and improves existing measures in several
ways: while conventional indicators of gender equality capture inequality
outcomes, the SIGI focuses on the root causes behind these inequalities
(www.genderindex.org).
 Website: www.oecd.org/dev/gender
GenderStats
 Electronic database of gender statistics introduced by the World Bank.
 A compilation of data on key gender topics from national statistics
agencies, United Nations databases, and World Bank-conducted or funded
surveys.
 Themes included: demographics, education, health, labour force, and
political participation.

259
  The database is updated three times a year (April, September and
December).
 Website: http://go.worldbank.org/YMPEGXASH0
Gender Statistics Programmes in the Arab Countries (GSP)

  Developed by ESCWA, in coordination with the UNDP, UNICEF,

CAWTAR, AGFUND and IDRC has launched the
 Website: http://www.escwa.un.org/gsp/index.html
Global Resources Database - BRIDGE

The library contains over 3,000 specially selected gender documents
picked from over 1,500 sources.
 Covers longstanding concerns and debate as well as emerging issues.
 Presents records of good practice, lessons learnt and case studies.
 The database is added to monthly.
 Browse by theme or by region.
 Website: http://www.bridge.ids.ac.uk/go/global-resources-database
UNECA African Centre for Gender and Development (ACGS)
African Gender and Development Index (AGDI): is an Africa-specific tool

that measures gender inequalities in the social, economic, political and
human right fields in Africa. It is composed by two parts: 1) the quantita-
tive Gender Status Index (GSI) and 2) the qualitative African Women’s
Progress Scoreboard (AWPS). See: www.uneca.org/eca_programmes/acgd/
 Website: http://www.uneca.org/fr/acgd/en/1024x768/acgd.htm
UNECE Gender Statistics Database
The Database helps to monitor the situation of women and men in all
UNECE member countries.
 Includes a set of Common Gender Indicators based on gender issues
relevant for the UNECE region.
 Data mainly supplied by National Statistical Offices.
 Website: http://www.unece.org/stats/gender/
UNSD Statistics and indicators on women and men
Data compiled from official national sources as well as international
sources.
 six specific fields of concern: population, women and men in families,
health, education, work, and political decision making.
 Website: http://unstats.un.org/unsd/demographic/products/indwm/
UNSD Gender Info
 Global database of gender statistics and indicators on a wide range of
policy areas, including: population, families, health, education, work, and
political participation.
 Easy-to-use tool to shed light on gender issues through customizable tables,
graphs and maps.
 It is an initiative of the United Nations Statistics Division, produced in
collaboration with the United Nations Children’s Fund (UNICEF) and the
United Nations Population Fund (UNFPA).
 Website: www.devinfo.info/genderinfo

260
WIDNET: Women in Development Network
 Offers regional and country-by-country population, family, household,
health, education, labor, and political power information extracted from
statistics compiled by the International Labour Organization, the UN, and
other sources.
 Website: http://www.focusintl.com/widstat0.htm
WomanStats Project
 Data on all countries with a population greater than 200,000 – a total of
174 countries.
 Quantitative and qualitative information on over 310 indicators of women’s
status.
 Access is free of charge
 Information on the site is continually updated as newer information
becomes available.
 Website: http://www.womanstats.org/

261
Appendix 4: A Brief Overview of the Evolution of Gender Statistics
Gender statistics are defined as statistics that adequately reflect differences and inequalitiesin
the situation of women and men in all areas of life (United Nations, 2012). They have
evolved significantly over the last thirty years, beginning with the First World Conference on
Women in Mexico City (1975). At the time, Women in Development (WID) was an
emerging paradigm, and the conference demands mainly focused on the inclusion of women
into statistics, addressing the underreporting of women’s issues and the lack of sex-
disaggregated data.31 The political mobilization announced 1975 the International Women’s
Year, and also proclaimed the following years as the United Nations Decade for Women
(1976-1985). An important book that illustrates the WID paradigm was Women’s Role in
Economic Development, by Ester Boserup (1970), which showed that women made important
economic contributions that were being ignored by governments.

The two subsequent Women’s Conferences – Copenhagen (1980) and Nairobi (1985) –
affirmed a growing demand for the production of sex-disaggregated data. Par. 92 of the
Report of the Copenhagen Conference, for instance, proclaimed that “all data-collecting
agencies should give a sex and age breakdown of any information they gather, wherever
relevant.” In addition, paragraph 95 suggested that “a set of statistical indicators should be
established by which progress towards equality between the sexes can be monitored” (UN,
1980).

Another significant shift was the change in perspective, from statistics on women, in the
1970’s and 1980’s, to statistics on gender – mirroring the evolution from the women in
development (WID) dominant strand of thinking to the gender and development (GAD)
approach. This shift recognized that gender identities are relational and that the roles of all
social actors, both men and women, are constructed within a gendered society – even though
power relations have tended to make women, as compared to men, a more vulnerable
segment.

With the GAD perspective, advocacy stressed not only the requirement of sex-disaggregation
but also the importance of introducing specific questions concerning only men or only
women into the census or survey questionnaires. Moreover, gender mainstreaming
throughout the statistical operations became an end in itself. Historically, this perspective
change was followed by country reports and booklets, as noted by Corner (2003: 2):

The initial work focused primarily on statistics on women and a number of

countries produced booklets on Women in Statistics that primarily confined
themselves to presenting data on women. However, it was soon recognized
that the situation of women could be adequately described and analysed only
in relation to the comparative situation of men. As a result, the focus moved
from statistics on women to statistics on women and men.
In line with the GAD approach, the Platform for Action of the Fourth World Conference 32
(Beijing, 1995) formulated claims for gender statistics by means of a specific strategic
31
The World Plan of Action of the First Conference also stated that regional action should include regional
standing committees of experts from countries in order to “give leadership in the methods of reporting on the
situation of women and in the development of indicators for assessing the progress made towards the goals of
this Plan in conjunction with regional statistical bodies and international efforts to this end” (World Conference
of the International Women’s Year, World Plan of Action, Paragraph 207, 1975).

262
objective (H.3) – along with other references throughout the document. [cite this in full ?
Annex ?] The Beijing +10 declaration introduced in Par. 25 the focus on “men and boys as
gendered persons”, further recognizing their capacities in bringing about changes in attitudes,
relationships and access to resources. More recently, the conceptual approach of The World’s
Women 2010: Trends and Statistics (United Nations, 2010 a) is based on the premise that
“statistics on men figure as prominently as statistics on women.”33

Miller and Razavi (1998) discuss the benefits and limitations of various analytical models
along a continuum of integrationist to transformatory: the Gender Roles Framework (GRF),
the Development Planning Unit (DPU), the Social Relations Framework (SRF), Gender and
Macroeconomics and Alternative Analytical/Training Frameworks.

…….

Finally, there have been noteworthy conceptual shifts in the attempt to capture gender
differences, specificities and disparities with more accuracy. In this vein, conceptual
adjustments were made and new analytical categories were created. The International Labour
Organization (ILO), for instance, revised its definition of economic activity to include
informal sector and non-market activities. Since 1993, the range of activities to be considered
as ‘extended economic activity’ includes some activities that had previously been classified
under ‘household activities’34, thus capturing somewhat more adequately the economic
contributions of many women, particularly in rural areas, to household income and family
businesses. Despite these efforts, many commentators have noted that “the distinction
between ‘subsistence work’ and ‘housework’ is still unclear in ongoing labour force surveys”
(Ayhan, 2001) and that a large amount of women’s (unpaid) work is still excluded from
conventional labour force definitions and statistics (UNECE/World Bank Institute, 2010).
Since the mid-eighties, national statistical offices, too, have started paying more attention to
statistics on women and to gender statistics. Statistics Sweden and Statistics Norway, for
instance, were among the first countries to introduce a gender perspective by designating
specific staff to work on gender statistics (Sweden) and establishing units or departments
dedicated to collecting and analysing gender data (Norway). 35 Ever since there has been much
debate whether it is favourable to 1) have designated staff responsible for gender issues in
national statistical offices; or 2) mainstream gender throughout the system by training
existing staff.

Although designated staff may play an important role in gender analysis, the risk that gender
may be seen as a separate, detached issue within NSOs is not addressed. Mainstreaming
gender issues throughout the national statistical system may enhance system-wide sensitivity
and ensure that a gender perspective is adopted in all stages of data production and analysis.
For this reason, gender advocates tend to support the second approach today: “the production
of gender statistics has to affect the whole official statistic system and not only the single unit

32
The Beijing Platform for Action made reference to the need to develop and strengthen statistical systems in
several issues, such as labor and economic activity (including female contribution in the unremunerated and do -
mestic sectors), health of girls and women of all ages, incidence of violence (including domestic violence, sex -
ual harassment and other different forms of violence against women and girls), and sharing of power and deci -
sion-making.
33
http://unstats.un.org/unsd/demographic/products/Worldswomen/WW_full%20report_color.pdf
34
Activities no longer considered as ‘household activities’ include: production of agricultural produce, gathering
of fruits etc. and their storage; processing of primary products (produced or bought) and the collection of water;
other processing activities, sold or not, like weaving, dress making and furniture making. (Tempelman, 1999).
35
http://www.gender.no/Facts_figures/1322

263
addressing gender, if there is one” (Sabbadini, 2008: 08). This is a bit contradictory with the
calls for statistical data bases on gender made elsewhere in this document.

In recent years, gender-related indicators and indices have also multiplied. The Millennium
Development Goals (MDGs) and the Beijing + 10 Declaration, for instance, reiterated the
need for developing timely and reliable gender statistics, especially at the level of national
statistical systems, suggesting sets of indicators and strategies for engendering the monitoring
and progress reporting. Numerous websites of National Statistical Offices now provide data
and indicators to be used for gender analysis and advocacy (see
http://www.census.gov/aboutus/stat_int.html). Organizations such as UNECE, the World
Bank, ILO, and universities in many countries have developed web-based data resource
centers dedicated specifically to gender statistics. International databases such as GenderInfo
have also been developed, gathering information from different countries.

In 2006, the Global Gender Statistics Programme was established by the United Nations
Statistics Division. As part of this programme, three Global Fora on Gender Statistics were
organized in 2007 (Rome), 2009 (Accra) and 2010 (Manila) with the purpose of promoting
the advancement of gender statistics among decision makers and other user groups. The
Forum also aims at reviewing best practices in the incorporation of a gender perspective in
national statistical systems, at discussing the measurement of women’s participation and
contributions to the economy and gender dimensions of health statistics (UNSD, 2008, 2009,
2010) [This should be United Nations]. The Global Fora have established a regular and
continuing basis for sharing experiences and supporting the implementation of adequate
technical procedures.

Nevertheless, even though gender issues are addressed more systematically now than they
were a few decades ago, there is still a long way to go. Much has to be done to tackle current
challenges and obstacles, such as the gender data gap in particular world regions, the
development of adequate legislation and methods regarding gender statistics and the need to
intensify the interaction between data producers and users at all levels – as discussed in
Appendix 4.

Text Box 12: Evolution of gender statistics

- Historically, gender statistics gradually received increasing importance and attention in

conferences, meeting, forums and other political and advocacy events, especially following
the Fourth World Conference on Women (Beijing, 1995).

- Change in approach from statistics on women alone to statistics on gender equality issues -
following the WID to GAD shift in perspective.

- Past experience along with review and reformulation of strategies led to adjustments in
definitions and to the creation of new analytical categories (ILO, FAO, UNFPA, UNIFEM,
etc).

- A number of NSOs have created specific departments or units for mainstreaming gender
issues in official statistics.

- Increase in the number of gender statistics sources for incomprehensive areas (time-use
surveys, gender-based violence surveys, access to decision-making in all spheres, etc.).

264
- Proliferation and revision of indicators and indices.

- Elimination of gender bias in some data sources.

- Additional questions in surveys and censuses that are useful for gender analysis.

- Review and adjustment of data collection instruments and methodologies for information
from a gender perspective.

- Dissemination of gender statistics increasingly more regular and frequent in many countries,
including through technological devices such as web-based data centers.

- Increase in high level policy dialogues and meetings specifically on gender and statistics.

265
Appendix 5: From Understanding the Gender Data Gap to
Improving the Production and Analysis of Gender Statistics

The phrase gender data gap refers to the fact that the coverage of gender issues, including
women’s lives and realities, is generally inadequate in mainstream data and statistics. This
occurs to some degree everywhere but is worse in some regions (Africa and Oceania) and for
some types of statistics (e.g. social services, disability, time use and unpaid work, gender-
based violence, access to resources, access to decision-making processes, informal economy)
(UNSD, 2005).

There are several reasons for the existence of a gender data gap including:

 Underreporting of women in censuses and surveys in many countries. To get women

out to respond, information campaigns have to explain the purpose and relevance of
data collection; women have to be allowed and feel safe to speak as individuals; and
enumerators, male and female, need to be trained to specifically seek the inputs of
women.
 Questionnaires are primarily developed by men/male statisticians who may bring
gendered biases in the focus and phrasing of questions. There is a tendency to focus
on the more public sphere of men’s lives, as opposed to the more private spheres of
women’s lives (for example unpaid work is usually not reflected nor counted).
Questions may use pronouns that lead to assumptions, such as using the term “he”
when asking about the head of household.
 A lack of routine disaggregation (by age and sex at the very least) which, in
combination with the male norm in language and society, blurs the input and
contribution of women. Speaking of “rural workers”, “smallholders” or “secondary
school students” makes many people think of males rather than females and may hide
important gender differences within these groups. This also leads to certain categories
of women, such as indigenous women, women with disabilities, adolescent girls, and
elderly women, to be particularly underrepresented. Further levels of disaggregation
or multivariate analysis are needed to provide a fuller picture, and in some instances,
additional questions need to be asked to identify special social groups.
 Inadequate budgets, low human and technical capacities, lack of gender
mainstreaming and inadequate concepts and methods hinders the quality of data being
collected and tabulated and also the publishing of the information.
The following will examine some of the major recommendations to producing and using
high-quality gender statistics as well as outline the global standards, potential problems and
country examples.

Recommendation 1: Data must be relevant to Gender Equality Issues- Need for strong
alliances between data users and producers
Why is this important? Gender statistics are the foundation of gender-based analysis and go
well beyond sex-disaggregation. As such, producers of gender statistics are required to know
more about the topics of gender, and users of gender statistics must know about the scope and
limitation of the data. Collaboration between the producers and users of gender statistics will
help ensure that gender statistics are more meaningful and user-friendly and include issues
from multiple angles while also keeping in mind policy and planning imperatives. Raising the
awareness of the institutions that produce statistics about gender issues may also help
increase the demand for gender sensitive data for public policies.

266
Global standards: Strategic objective H.3./206.c) of the Beijing Platform for Action asks
Statistical Services to “Involve centres for women's studies and research organizations in
developing and testing appropriate indicators and research methodologies to strengthen
gender analysis, as well as in monitoring and evaluating the implementation of the goals of
the Platform for Action”. It is now widely understood that other governmental (Ministry of
Fmaily/Women/Social Affairs for example) and civil society organisations (non-
governmental organisations, universities, media, trade unions, etc.) should also be involved in
this process.

Potential problems: A potential problem is the asymetry of power relations: Women’s

machineries are often marginalised in government, suffer from a chronic lack of resources
and have limited/no understanding of the availability of data or how to use data for
programme/policy development. Consultations wth civil society organisations are sometimes
mere window-dressing in response to growing national and international pressure to engage
with them. If data users are not at the table at the time census questionnaires are designed and
tabulations are planned, it will be hard to present gender statistics in a user-friendly way.
Data may not be published “just in time” for crucial policy decisions if there is no interaction
between policy-makers and data-producers.

Country example: Uganda. The Uganda Bureau of Statistics (UBOS) regularly promotes
Public Seminars, organized in collaboration with policy makers and users of statistics
(stakeholders), to present analysis and research findings using data routinely collected by the
Uganda Bureau of Statistics (UBOS), including census data and gender indicators. One of the
main purposes of the seminars is to educate the public about the use of statistics.

Internationally: Over the last ten years, alliances between producers and users of statistics
have been strengthened through international meetings on gender statistics. For example, a
Working Group on Gender Statistics was created during the fourth meeting of the Statistical
Conference of the Americas of the Economic Commission for Latin America and the
Caribbean (ECLAC) in 2007. This inter-institutional and inter-agency (UN Women and the
Gender Affaires Division/ECLAC) initiative has led to progress in statistical activities across
the region.
(Germany example, p. 141, UNECE? What is this ?) You should be included as an
example of alliances between producers and users, the international meetings of gender
statistics that have been carried out for more than ten years (without interruption). They have
been regarded as a good practice, and that is an inter-agency effort (UN Women and Gender
Affairs Division/ECLAC) and inter-institutional. The meeting is part of the activities of the
working group on gender statistics of the Statistical Commission of the Americas.

Recommendation 2: An enabling institutional environment is needed for the production

of gender statistics – political will, a legislative framework, sufficient funding,
commitment and buy-in from senior leaders
Why is this important? Gender statistics emerge from demand, whether from politicians,
specific ministries, women’s departments or the public. Political will for gender statistics can
be thought of as support from political leaders that results in policy change. It generally
originates with a single individual or small group, often situated somewhere within the state
apparatus (best if at a high level) who become the “champion” for gender statistics, build
coalitions and, as part of those coalitions, raise funds for the issue, push for its firm anchoring
in the legislative framework and try to sustain the policy change once achieved. Where
political will is lacking, funds are unlikely to be allocated to gender statistics and an enabling

267
environment (legislative and administrative framework) is unlikely to be created.

Global standards: Many countries incorporate CEDAW provisions on gender equality and
women’s empowerment into domestic law, in particular their civil, penal and labour codes.
Countries also reform their Gender Equality Architecture, conduct Gender Audits and initiate
Gender-Responsive Budgeting to monitor whether and how political translates into funding
and impact. Gender analysis of sex-disaggregated data is needed to monitor progress in
compliance with CEDAW and other international and national instruments at country-level.

Text Box 13: Gender-Responsive Budgeting

Gender-responsive budgeting (GRB) is government planning, programming and budgeting
that contributes to the advancement of gender equality and the fulfillment of
women's rights. It entails identifying and reflecting needed interventions to address gender
gaps in sector and local government policies, plans and budgets. GRB also aims to
analyse the gender-differentiated impact of revenue-raising policies and the allocation of
domestic resources and Official Development Assistance.
 Source: http://www.gender-budgets.org/; accessed 1 March 2011.

Potential problems: As UNSD pointed out, it is essential to “distinguish between a national

plan on gender statistics and a law on gender statistics”, since “without a specific law, there is
no guarantee that gender statistics will be included in the work plan each year” (UNSD
2009). Moreover, the laws on gender statistics must also be relatively detailed, clearly
spelling out administrative and budgetary implications. Where a law on gender statistics
exists, budget negotiations will be easier and the gender statistics architecture is more likely
to dispose of the means it needs to fullfil its functions.
Country examples: Kyrgyz Republic: Obliges State bodies, institutions of local self-
government and heads of legal entities regardless of forms of ownership to submit
appropriate information on gender issues to the National Statistical Committee (Art. 30).
Philippines and Nepal:, CEDAW Monitoring Bodies have been set up that are responsible
for producing and diffusing gender statistics.
Italy: A detailed law on gender statistics makes producing sex-disaggregated data from
population censuses compulsory, including data on households and institutionalized persons,
as well as on different household typologies. (UNECE/UNDP, 2004)
Korea: A new statistical act was released (2007), giving the Korean Statistical Office the
jurisdiction to expand the topics included in data collection, in order to provide more relevant
data for gender analysis (UNSD, 2009).
Spain: The 2007 statistical legislation of the Andalusia region included the provision for
gender mainstreaming which created an enabling environment for the production of gender
statistics, also leading to laws requiring gender budgeting and the monitoring of its
effectiveness (UNSD, 2009). [Colombia, Trinidad & Tobago, Ecuador, Mexico and Peru
(Costa Rica has a law project) have laws requiring the NSO to collect information on time
use and non-remunerated work for unremunerated satellite accounts (set of experiences that
have shared best practices in the context of the work carried out by ECLAC and UN Women
and the institutions mentioned above]

Recommendation 3: Gender data must be reliable and adhere to international

standards – Need for common definitions and quality standards to enhance
comparability
Why is this important? Gender statistics are essential for monitoring progress on
international treaties and agreements as well as providing countries with comparable data to

268
track progress being made in promoting gender equality. In order to do so, comparison of
data must be possible both over time and between countries. However, concepts such as
household and household headship, marriage, economic activity, informal sector, etc. pose
numerous measurement problems. The main benefit of having common definitions and
quality standards is disposing of evidence that is sound and reliable and that is a solid base
for informed policy decisions.
Global standards: While the Principles and Recommendations for Population and Housing
Censuses, Revision 2, form the normative basis for census taking and census analysis, no
such reference framework is available for gender statistics. (From UN Women: the cited
examples have not referred only to censuses. It is important to cite the guide of
ECLAC/UNIFEM/UNFPA, that include some recommendations about census work).
Existing informal guidance documents include a number of publications by the UN Statistics
Division and other international agencies, including development banks (see appendix 2).
Potential problems and benefits: In terms of definitions, the main issue is to balance
between “validity” (measuring what you went out to measure, which implies using a
culturally appropriate approach) and comparability. When standard definitions are too forced,
categories lose their meaning on the ground and hence are no longer useful tools for policy-
making. In terms of quality, the main issues are capacity and resources: To produce highest
quality gender statistics, National Statistics Offices need the appropriate human and financial
resources to pilot test and analyse, train enumerators, invest in data editing, etc.
Regional example: The Gender Statistics programme of the Secretariat of the Pacific
Communities has created a website (PRISM) collating statistics from 22 countries and
territories in the Pacific on common indicators used for monitoring gender equality (see
http://www.spc.int/prism/data/gender). Other regional observatories: Latin America and the
Caribbean Gender Equality Observatory (http://www.cepal.org/oig/) with the support of
UN Women.

Recommendation 4: There is severe underreporting on gender issues – Need for

capacity-building for NSOs
Why is this important? Underreporting on gender exists both in terms of who gets
enumerated/surveyed and in terms of what gets published. Therefore, the institutionalization
of gender mainstreaming is crucial to engendering all statistical operations. On an
organizational level, this means training staff on gender statistics and gender analysis as well
as establishing Gender Units in NSOs. On an operational level, it implies ensuring gender
responsiveness in questionnaire elaboration, field operations and their supervision, as well as
in data management, analysis, publication and dissemination.
Global standards: Data is supposed to adequately reflect the situation of all citizens (male
and female, old and young, indigenous populations and persons with or without disabilities,
etc). Being inclusive of every citizen and asking truly relevant questions is therefore
imperative in order to achieve high data quality. Collecting data on women and girls and on
gender-specific issues is one step into the direction of achieving a fuller, more valid picture of
the situation in a country.
Potential problems:In many countries, high turnover of staff (sometimes due to unattractive
working conditions and remuneration) makes capacity-development a never-ending
challenge. Specialized gender teams, inter-sectoral and inter-agency gender working groups,
and gender departments established in the National Statistics Offices, can ensure that the
participation of gender statistics experts in other relevant official statistical productions is
institutionalized. Capacity-development is also needed for effective data dissemination
strategies: “The analytical capacity of national statistical systems should be strengthened to
ensure that available data are used productively and the findings communicated more

269
successfully to the appropriate audiences, especially to the policy makers and the media […]”
(UNDESA, 2006: 29-30).
Country example: Nepal: Regional training workshops on gender sensitization were
conducted in joint partnership with international agencies and included participants of senior
and mid-level officers of the Central Bureau of Statistics.

Challenge 5: Micro-data on gender issues is dispersed - Need to unify what is available,

upstream small-scale data
Why is this important? At the national and international levels, there is still a large gap with
regards to the availability of data on social services, disability, time use and unpaid work,
gender-based violence, access to resources, and the informal economy, to name just a few
important gender concerns. Such data are needed by in-country NGOs, line ministries and
local governments to monitor projects, prepare advocacy materials and negotiate budgets. If
National Statistics Offices could act as a clearing house for disparate small-scale data at the
sub-national level, or effectively cross-analyse data from various sources (vital statistics,
census, surveys, etc.), much could be gained for understanding gender issues “on the
ground”.
Global standards: The UN’s Inter-Agency and Expert Group Meeting on the Development
of Gender Statistics emphasized the need of two specific issues: 1) “the review of existing
national data collections to identify and develop inventories of sex-disaggregated statistics
and gender related indicators”; and the need to 2) “the re-coding, re-tabulation and re-analysis
of micro-data from surveys or censuses, from a gender perspective” (idem: 08). Similarly,
“one of the strategies recommended by FAO for filling the gender data gap is to promote the
coordination, integration and re-tabulation of agricultural data by sex and age at the sub-
national level (FAO 1999:23)”
Potential problems: A high level of technical expertise is needed for cross-analysing data
collected on small samples of different population groups at different points in time. Despite
huge efforts in combining sources, data quality may be too low for meaningful analysis.
Country examples:
[find]

Recommendation 6: Information is available but not used - Need for better

communication and dissemination of gender data
Why is this important? It is not enough for gender data to simply be available. To make a
difference in the lives of women and men, girls and boys, these data need to be disseminated
– and ideally used – by policy-makers, opinion leaders and the general public. A variety of
formats are required to achieve this. Along with reports and press releases, dictionaries and
metadata are important tools for ensuring that the underlying concepts are actually
comprehended and that the results are well interpreted. Online resources and services, like
query databases and downloadable files, are now a large part of the public’s information
toolkit. Documents, tables, maps and images available on the internet facilitate the access and
visualization of data.
Global standards: Several issues are at play here: The public’s right to information is key, as
is the moral imperative to feed knowledge back into the public domain. Policy makers can be
reminded of norms such as transparency and accountability to invite them to cement the
knowledge-base on which their policy-decisions are based.
Potential problems: If NSOs are increasingly offering a broader range of formats for
dissemination, users of gender statistics also need to be more proactive in accessing and using
the existing data. Benefits include a better-informed public, and – in the best case – better

270
informed public policies. A collateral benefits is heightened visibility for the NSO (which can
have budgetary repercussions).
Country examples:
[find]

271
Appendix 6: How to Apply this Guide in a Country Context

While there are many ways in which this guide can be used, the following provides four key
steps for carrying out a gender analysis project at the country level. It emphasizes
collaboration between the producers and users of gender statistics by involving NSO and
gender experts from the government, academia and civil society. As such, these steps will
help ensure that gender statistics are more meaningful, user-friendly and address the key
gender issues relevant to that country.

Step One: Selection of Gender Issues:

Participants: NSO, Gender experts from government and civil society, including research
institutions
Format: Workshop
Documentation: 10 Key Question Tool, Census Questionnaire, this manual (table of
contents: chapter 3)
Purpose: To identify the key gender issues that can be analysed with the census data
obtained in country X.
Roles: Gender Experts provide an evidence-based overview of the key gender issues in the
country, ideally using the 10 Key Question Tool below and considering the 10 gender issues.
Statisticians explain what can and cannot be measured with census data on the basis of the
country’s census questionnaire.
Expected Outcome: Consensus on what statisticians should compute

Box : The 10 Key Questions Tool

1. Who does what? [activities]
2. How? With what? [access to resources]
3. Who owns what? [ownership of assets]
4. Who is responsible for what? [obligations]
5. Who is entitled to what? [claims, rights]
6. Who controls what? [income, spending]
7. Who decides what? [power]
8. Who gets what? [distribution]
9. Who gains and who loses? [redistribution]
10. Why - What is the basis for the situation? [rules, norms,
customs]
(Questions 1-9 can be combined with the additional question, "And With Whom?' in order to capture
the social relations involved)

Step Two: Census Analysis – Preparation of Tabulations and Computation of Sex-

Disaggregated Indicators

Participants: NSO, Gender experts from research institutions, external consultants

Format: Desktop study/in-depth statistical analysis
Documentation: “Principles and Recommendations”, this manual (sections 4 and 5 of each
chapter in Part II)
Purpose: To provide the raw data, tabulations and indicators needed for answering the key
gender questions identified in Step One.
Roles: Statisticians and researches perform high quality data analysis
Expected Outcome: Tabulations and indicators are available and of high quality

272
Step Three: Interpretation of Data, Suggesting further Analyses

Participants: NSO, Gender experts from government and civil society, including research
institutions
Format: Workshop
Documentation: Tabulations and indicators produced by NSO, this manual (sections 2 and
6)
Purpose: To make sense of the data and suggest further analyses going into more depth with
some key findings
Roles: Statisticians walk participants through the analyses carried out, outline problems
encountered and summarise the gender differences identified; Gender Experts discuss what
may be underlying the gender differences documented
Expected Outcome: Consensus on additional variables that need to be taken into
consideration (and technically can be) in order to shed light upon the findings

Step Four: Advocacy Material is devised

Participants: Gender experts from government and civil society

Format: Workshop
Documentation: Key national policies, this manual (section 8)
Purpose: To identify how the indicators, tabulations and results of multivariate analysis can
be used to inform and advocate for current and future national gender equality policies and
initiatives, or for reporting purposes
Roles: Experts provide an overview of the key gender policies and initiatives currently on-
going and planned in-country and select critical data for evidence-based advocacy

Expected Outcome: An advocacy plan with clearly defined

roles/responsibilities/timeline (cite key references)

273