Obstacles to Contraceptive Use in Pakistan: A Study in Punjab

Author(s): John B. Casterline, Zeba A. Sathar and Minhaj ul Haque

Source: Studies in Family Planning, Vol. 32, No. 2 (Jun., 2001), pp. 95-110
Published by: Wiley on behalf of Population Council
Stable URL: https://www.jstor.org/stable/2696340
Accessed: 23-02-2025 05:04 UTC

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected].

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms

Wiley, Population Council are collaborating with JSTOR to digitize, preserve and extend
access to Studies in Family Planning

This content downloaded from 119.157.179.220 on Sun, 23 Feb 2025 05:04:08 UTC
All use subject to https://about.jstor.org/terms
 Obstacles to Contraceptive Use in Pakistan:
A Study in Punjab
John B. Casterline, Zeba A. Sathar, and Minhaj ul Haque

The principal aim of this study is to assess the strength in Pakistan of a set of hypothesized obstacles
to practicing contraception. Survey data are analyzed that were collected in Punjab province in
1996 and that contain unusually detailed measurement of various perceived costs of practicing
contraception, as well as focused measurement offertility motivation. The framework guiding the
research specifies six major obstacles to contraceptive use: the strength of motivation to avoid
pregnancy, awareness and knowledge of contraception, the social and cultural acceptability of
contraception, perceptions of the husband's preferences and attitudes, health concerns, and perceived
access to services. Net effects of each obstacle are estimated through structural equation modeling
of the intention to practice contraception in the near future, in which the six obstacles are treated
as latent variables. The estimates indicate that the two principal obstacles to using a contraceptive
are the woman's perception that such behavior would conflict with her husband's fertility preferences
and his attitudes towardfamily planning and her perception of the social or cultural unacceptability
of contraception. The results confirm the value of taking contraceptive costs seriously, and, in
particular, of attempting to measure these costs in empirical research onfamily planning. (STUDIES
IN FAMILY PLANNING 2001; 32[2]: 95-110)

In 1965, motivated by concerns about rapid population Asia, and Latin America in the period since 1975 have
growth, Pakistan was among the first Asian countries shown that a discrepancy between fertility preferences
to enunciate a national population policy (Rukanuddin and contraceptive behavior, commonly labeled "unmet
and Hardee-Cleaveland 1992; Rosen and Conley 1996). need for family planning," characterizes a sizable frac-
Soon thereafter, the National Impact Survey of 1968-69 tion (typically between 15 percent and 25 percent) of
appeared to demonstrate that this national policy and women of reproductive age in pretransitional and tran-
the programmatic interventions that flowed from it had sitional societies (Westoff and Bankole 2000).
considerable potential for success: The survey revealed What is remarkable about Pakistan is not the exis-
that a large proportion of women of reproductive age tence of this preference-use gap but rather its persis-
expressed a desire to terminate childbearing but were tence at relatively high levels for decades without any
not using any method of contraception (Sirageldin et al. significant change in contraceptive prevalence (and, ac-
1976). This discrepancy between stated fertility prefer- cordingly, in period fertility rates, which in the three
ences and reported contraceptive behavior was inter- decades leading up to the 1990s probably exceeded six
preted as indicative of latent demand for contraception. births per woman [Sathar and Casterline 1998]). Na-
Indeed, surveys carried out in many countries in Africa, tional surveys carried out in 1974-75 (PFS[WFS]), 1984,
and 1990-91 (PDHS) documented roughly the same
prevalence of unmet need-around one-fourth to one-
third of currently married women (Westoff and Bankole
2000)-together with essentially unchanging contracep-
John B. Casterline is Senior Associate, Policy Research tive prevalence (less than 12 percent). Another indica-
Division, Population Council, New York. E-mail: tion of the persistence of a gap between reproductive
[email protected]. Zeba A. Sathar is Deputy aspirations and reproductive outcomes is the relative-
Representative and Program Associate and Minhaj ul Haque ly stable 1.5 - 2.0-child difference between ideal family
is Senior Program Officer, Population Council, Islamabad, size and the total fertility rate (TFR) over this same two-
Pakistan. decade period. Identification of another country in which

Studies in Family Planning Volume 32 Number 2 June 2001 95

This content downloaded from 119.157.179.220 on Sun, 23 Feb 2025 05:04:08 UTC
All use subject to https://about.jstor.org/terms
fertility desires and fertility behavior remained stable has been why a large fraction of Pakistani women ex-
and in sharp contradiction for such a lengthy period press a desire to avoid pregnancy but do not practice
would be difficult. Some inferences might be drawn family planning. This question about the causes of un-
from this persistent gap: that the social and cultural ob- met need for contraception presumes the existence of a
stacles to contraceptive use are unusually powerful and motivation to avoid pregnancy that is not translated into
deep-seated in Pakistani society, and that the provision behavior because of the existence of one or more ob-
of family planning services has been especially poorly stacles to contraceptive use.
designed and implemented in Pakistan (Rosen and Con- To identify the major obstacles to family planning
ley 1996). in Pakistan, we draw both on previous empirical re-
Contraceptive prevalence began to increase in the search in Pakistan and on exploratory qualitative inter-
1990s, rising from 12 percent as reported in the 1990-91 views that we conducted at an early stage in our work.
PDHS (NIPS and IRD/Macro International 1992) to 24 We summarize the main findings from the existing re-
percent in the 1996-97 Pakistan Fertility and Family search literature and discuss conclusions from analysis
Planning Survey (NIPS/LSHTM 1998), a rise of roughly of the qualitative interviews.
2 percentage points per annum. Nevertheless, unmet A series of studies carried out during the past three
need for family planning remains at relatively high lev- decades have sought to explain the discrepancy in Pa-
els by international standards (Westoff and Bankole kistan between fertility preferences and contraceptive
2000). Similarly, fertility rates, although apparently on practice. Among the more important of these studies,
the decline, are also high in relation to other South Asian Sirageldin et al. (1976), in an analysis of the National Im-
nations and in relation to conventional measures of so- pact Survey of 1968-69 that relies mainly on indirect evi-
cioeconomic development, such as income and urban- dence, attribute the high latent demand for contracep-
ization (see Tsui 1996). tion to cultural and social constraints and to inadequate
Our goal in this study is to determine which factors family planning services. Drawing on more direct indi-
stymie the translation of women's desires to avoid preg- cators available in the Pakistan Fertility Survey of 1974-
nancy into contraceptive practice in Pakistan in the late 75, Shah and Shah (1984) conclude that the primary rea-
1990s. To this end, we make use of empirical data col- sons for nonuse of contraceptives among apparently
lected in Punjab province in 1996 that contain many mea- motivated women are the perception that family plan-
sures of the various hypothesized obstacles to contra- ning is unacceptable on religious grounds, fear of con-
ceptive use, measures that are largely unavailable in traceptives' side effects on health, opposition from hus-
other recent national surveys. bands, and, finally, the limited availability of family
planning services. Mahmood (1992) analyzes the 1979-
80 Population, Labour Force and Migration (PLM) sur-
Conceptual Framework vey and infers from the pattern of effects of socioeco-
nomic and demographic variables that psychological
In tackling the question of why some Pakistani women and sociocultural factors, in particular strong religious
practice contraception and others do not, we begin with and social values, are the key to understanding unmet
the Easterlin Synthesis Framework (Easterlin 1975; Herm- need. In a subsequent analysis of the 1990-91 Pakistan
alin 1983), in which fertility regulation is a function of Demographic and Health Survey, Mahmood and Ring-
two classes of factors, the motivation to avoid pregnancy heim (1996) conclude that the primary determinants of
and the costs of fertility regulation. Costs are broadly contraceptive use (not conditional on a desire to avoid
defined to include not simply the time and financial re- pregnancy) are knowledge of a supply source, husband-
sources required to acquire contraceptive supplies and wife communication, and religious attitudes. Hashmi et
services but also the social, psychic, cultural, and health al. (1993) conducted a follow-up study of nonusers in
costs (real and perceived) that accompany adoption and Punjab in the 1990-91 DHS and identify fear of side ef-
continued use of contraceptives. The focus here is this fects on health, religious concerns, lack of knowledge
constellation of factors, which until recently was rela- of a source of contraceptive supplies, and husbands' op-
tively neglected in research on family planning (Bon- position as the chief barriers to use. Other analyses have
gaarts and Bruce 1995). Although substantial declines focused on the national family planning program and
in desired family size are clearly a prerequisite for fer- invariably criticize it harshly for poor management and
tility to fall to replacement level in Pakistan (Shah and the low quality of the services offered to the majority of
Cleland 1993; Sathar and Casterline 1998), as described its clients (Robinson et al. 1981; Mahmood 1992; Ruka-
above, the central question during the past three decades nuddin and Hardee-Cleaveland 1992).

96 Studies in Family Planning

This content downloaded from 119.157.179.220 on Sun, 23 Feb 2025 05:04:08 UTC
All use subject to https://about.jstor.org/terms
 Although previous research in Pakistan yields what ods and how they work); costs of practicing contracep-
would seem to be a relatively comprehensive listing of tion, specifically perceptions of the social, cultural, and
possible obstacles to contraceptive use, we began this religious acceptability of contraception; perceptions of
research with our own exploratory qualitative inter- the husband's opposition to family planning; health con-
views to assure ourselves that important obstacles had cerns about contraception; and inadequate access to fam-
not been overlooked and to develop a better understand- ily planning services of acceptable quality.
ing of the nature of these obstacles for the purpose of The analytical value of this breakdown is twofold.
improving the formulation of survey questionnaire items. First, it gives structure to the analysis. Equally impor-
Eight focus-group discussions and 34 in-depth inter- tant, it provides an organizing framework for assessment
views were conducted with men and women in eight of the policy and programmatic implications of the re-
different locales in north, central, and southern Punjab. sults. From this standpoint, singling out poor access to
In all, more than 100 individuals were asked their views family planning services is clearly important. The scope
about what factors account for the failure of women or for programmatic initiatives that weaken barriers to con-
couples to translate their reproductive preferences into traceptive use surely extends beyond improving access,
contraceptive practice. In giving both the discussions however. Knowledge of contraception might be im-
and the interviews careful readings, we find the inter- proved through appropriate information, education, and
views to be more revealing of reasons for unmet need. communication efforts; health concerns might be re-
What emerges most clearly from the qualitative inter- duced thereby, and by means of focused counseling and
views with women is the extent to which users and non- follow-up of contraceptive adopters; husband's opposi-
users are polarized in their fear of side effects and in tion might be addressed through community-level cam-
their perceptions of husbands' disapproval of contracep- paigns and by increasing field-workers' sensitivity to
tion. Men, in contrast, invoke a more diverse set of rea- this problem; and so forth. None of these obstacles ap-
sons for nonuse, ranging from religious concerns to fear pears to be immune to appropriately designed interven-
of social ostracism to concern about financial costs and tions, nor is there evidence that some are more suscep-
fear of side effects. A further finding from these inter- tible to intervention than others (although this seems
views is that men cite their wives' disapproval as a rea- likely and should be the subject of further research). In
son for not implementing reproductive intentions less short, programmatic priorities can be derived directly
frequently than women cite their husbands' disapproval. from the relative strength of the various obstacles.
This finding is entirely consistent with the prevailing
view in Pakistani society that decisionmaking about re-
productive matters resides primarily with the husband. Obstacles to Contraceptive Use
Interestingly, the qualitative interviews indicate clearly
that deliberate fertility regulation, through modern fam-
Strength of the Motivation to Avoid Pregnancy
ily planning and other means, is becoming much more
acceptable to both women and men on religious and so- The strength of attachment to the desire to terminate
cial grounds. Old mores on this point are weakening. A childbearing or postpone the next pregnancy clearly var-
further change that respondents perceive is a withdraw- ies among women. Some women who indicate a desire
al of elders and in-laws from the decisionmaking pro- to avoid pregnancy are relatively unconcerned about
cess, with the result that decisions about contraception becoming pregnant, whereas others regard it as a highly
are increasingly the exclusive province of the wife and undesirable outcome and, when asked, admit to numer-
husband. In our survey analysis, we examine some of ous concerns and worries. Women who feel ambivalent
the themes that emerged from the qualitative interviews. about their stated desire may be especially susceptible
On the basis of this exploratory qualitative research to other factors, such as those reviewed below, that act
and of a reading of previous research on Pakistan, we against their practicing contraception. One important
arrived at a framework for indentifying obstacles to con- factor that can weaken the motivation to use a method
traceptive use in Pakistan. This framework influenced is a perception of low risk of conceiving. Responses from
the design of the 1996 survey data collection in Punjab, women who indicate that they can no longer conceive
and it structures the analysis that follows. We posit that, (most of whom are older than 35) are discarded from
among those wishing to avoid pregnancy, the decision analysis here. In quantitative analyses in other settings,
to practice contraception is determined by the follow- weakly held preferences emerge as an important factor
ing: the strength of the motivation to avoid pregnancy; contributing to the preference-use gap in a DHS multi-
knowledge about contraception (the existence of meth- country analysis of unmet need (Westoff and Bankole

Volume 32 Number 2 June 2001 97

This content downloaded from 119.157.179.220 on Sun, 23 Feb 2025 05:04:08 UTC
All use subject to https://about.jstor.org/terms
1995) and in an in-depth study in Luzon, Philippines mi et al. (1993) identify religious concerns as the second
(Casterline et al. 1997). From the Punjab survey data, the most common reason given for nonuse, and in their analy-
following indicators of the strength of motivation to sis of 1990-91 DHS data, Mahmood and Ringheim (1996)
avoid pregnancy are selected: the desire to have no more show that religious conservatism is a strong negative
children, in contrast with wanting to postpone the next correlate of contraceptive use. From this and other re-
birth (those who wish to stop are assumed to have stron- search, Shah and Cleland (1993) conclude that "deep am-
ger motivation); number of concerns about having an- bivalence about family planning" stands among the most
other child; whether the woman would be distressed if imposing barriers to increased practice of contraception
she became pregnant in the near future; and the wom- in Pakistan. In this respect Pakistan may well be excep-
an's perception of her husband's desire for another child, tional: We are not aware of empirical research from other
under the assumption that this perception will affect the settings in which the acceptability of contraception as-
strength of her commitment to her own desires. sumes such importance. Our indicators of the social and
cultural acceptability of contraception are: the respon-
dent's approval of family planning; her perception of the
Knowledge and Awareness of Contraception
approval of in-laws; the number of modern methods the
Many potential informational barriers exist to contracep- woman approves; her approval of withdrawal (use of
tive use. Women must be aware of the methods avail- which increased in the early 1990s, making a substantial
able, must know where supplies of these methods can be contribution to the overall increase in prevalence, and as
obtained and how much they cost (with the exception a non-supply method not affected by shortcomings in
of nonsupply methods such as withdrawal and periodic family planning services); her perception that use of a
abstinence), and they must know how to use the meth- contraceptive method might provoke divine disapproval
od they choose. In several cross-national analyses, these (including death of a child); religious concerns as a stated
types of knowledge are strongly associated with unmet reason for not practicing contraception; and past prac-
need for contraception (Bongaarts and Bruce 1995; Robey tice of contraception, taking this as an indicator that the
et al. 1996), and in India they affect the intention to use woman finds contraceptive use acceptable behavior.'
a method in the future (Mishra et al. 1999). In research
on Pakistan, Siragelden et al. (1976) note that awareness Husband's Opposition

of contraceptive methods was already high in the late Empirical research conducted during the 1990s makes
1960s, a point emphasized again in the research on rea- evident that women's perception that their husbands
sons for nonuse carried out by Hashmi et al. (1993). This oppose family planning is a dominant factor discourag-
finding has led most analysts to conclude that knowl- ing contraceptive practice in a wide variety of settings,
edge barriers are relatively insignificant in Pakistan including Egypt (El-Zanaty et al. 1999), Guatemala
(Shah and Cleland 1993). The explanatory power of the (Asturias de Barrios et al. 1998), India (Viswanathan et
following knowledge indicators is considered here: the al. 1998; Mishra et al. 1999), Philippines (Casterline et
number of modern and traditional methods that the al. 1997), and Nepal (Stash 1999). Although this view is
woman knows and the woman's past use of contracep- also commonly held among researchers and program
tives. (In this analysis, knowledge of sources of family managers in Pakistan, previous research was inconclu-
planning services and of various methods available are sive on this point. In their early analysis of the 1968-69
considered indicators of access to services and are dis- National Impact Survey, Sirageldin et al. (1976) show that
cussed below.) husband's approval of contraception is strongly associ-
ated with contraceptive use among women who want
to avoid further childbearing, that is, women with un-
Costs of Contraceptive Use
met need. Shah and Shah (1984) and Hashmi et al. (1993),
Social and Cultural Acceptability in comparable studies of reasons for nonuse, find, how-
In previous research conducted in Pakistan, the social, ever, that a perception that the husband disapproves is
cultural, and religious unacceptability of contraception provided far less often as a reason for nonuse than are
repeatedly emerged as an important obstacle to using a religious concerns and fear of contraceptives' side effects
method. In an analysis of survey data from the 1960s and on health. A husband's disapproval may reflect his fer-
1970s, Sirageldin et al. (1976) identify what they term "cul- tility preferences or his other adverse feelings about con-
tural and social constraints," and Shah and Shah (1984) traception. On the question of the determining role of
find that the most common reason given for not using a the husband's desire for more children, Mason and
contraceptive is religious concerns (reflecting what they Smith (2000) show that husbands' preferences are more
term a negative climate of opinion). More recently, Hash- strongly associated with contraception than are wives'

98 Studies in Family Planning

This content downloaded from 119.157.179.220 on Sun, 23 Feb 2025 05:04:08 UTC
All use subject to https://about.jstor.org/terms
preferences (especially in the "most feudal" communi- erable difficulty in obtaining low-cost, high-quality fam-
ties in Punjab), but that nevertheless they account for a ily planning services (see reviews in Robinson et al. 1981,
relatively small fraction (around 10 percent) of unmet Rukanuddin and Hardee-Cleaveland 1992, and Rosen
need for limiting family size. Another line of research and Conley 1996). Mahmood and Ringheim (1996) find
focuses on spousal communication (about contraception that women's knowledge of sources of supplies is the
and related issues) and finds that it is a good predictor strongest predictor of contraceptive use, as reported in
of contraceptive use (Mahmood and Ringheim 1996) and the 1990-91 DHS data. In research conducted in the early
of fertility preferences (Mahmood and Ringheim 1997).2 1990s, contraceptive prevalence rose substantially in six
From the 1996 Punjab data, we select the following as areas served by intensive community-based distribution
indicators of the husband's support of (or opposition to) projects, a finding consistent with the argument that
contraception (all of these indicators are taken from in- poor services are a major barrier to use (Shelton et al.
terviews with women, and hence represent the wives' 1999). Our 1996 data collection in Punjab province did
perceptions of their husbands' views): husband's desire not include direct measurement of the service environ-
for another child; husband's approval of family plan- ment. Therefore, we must rely on respondents' reports.
ning; the number of modern methods of which he ap- The indicators selected for this analysis are: the number
proves; husband's approval of withdrawal; ease of dis- of methods for which women know a source of supply;
cussing family planning with the husband; husband's the proximity of the nearest services; and whether or not
opposition as a stated reason for not practicing contra- the respondent can visit a health facility unaccompanied.
ception; and past use of a contraceptive, assuming that The aim in this analysis is to assess, in quantitative
a woman is unlikely to have used a method in the past terms, the relative magnitude of these various obstacles
without her husband's approval. to the practice of contraception. We treat the six obstacles
as separable and competing reasons for nonuse, but do-
Health Concerns ing so is undoubtedly a simplification of reality. The
The same empirical studies carried out in diverse set- qualitative research indicates clearly that each of these
tings in the 1990s that highlight the determining role of barriers is multidimensional in nature, a conclusion that
women's perceptions that their husbands oppose contra- also emerges from Stash's (1999) investigation of the
ception also reveal that a set of related health concerns causes of unmet need in Nepal (see also Nag 1984).
constitutes a powerful obstacle to using a method (Bon- Health concerns are a good example of this multidimen-
gaarts and Bruce 1995; Casterline et al. 1997; Asturias sionality: As we learned from our exploratory qualita-
de Barrios et al. 1998; Viswanathan et al. 1998; Yinger tive interviews, fear of the side effects of contraceptives
1998; El-Zanaty et al. 1999; Stash 1999). Judging from dissuades women from using them not only because of
prior research conducted in Pakistan, this conclusion aversion to the expected physical discomfort but also
applies in this setting as well. In both Shah and Shah because of the expected time and financial costs of man-
(1984) and Hashmi et al. (1993), fear of side effects is aging the side effects, the potential loss of work time,
identified as one of the two most important explanations the possibility of interference with spousal sexual rela-
for nonuse. Here, we make use of the following mea- tions, and a sense that the( side effects signify divine dis-
sures: number of contraceptive methods thought to have approval. The multiple facets of health concerns, there-
bad side effects; side effects identified as an obstacle to fore, touch on other obstacles treated as distinct in the
use; and side effects as a stated reason for not practic- framework, such as anxiety about the acceptability of
ing contraception. contraceptive use and respondents' perceptions that
their husbands oppose family planning. Hence the de-
Access to Contraceptive Services sign of this analysis entails a regrettable simplification,
One of the principal aims of this study is to broaden the but we feel this cost is more than offset by our gaining
investigation of the costs of contraceptive use to include the ability to offer a quantitative assessment of the rela-
the social, cultural, and health costs described above, tive strength of each obstacle. An explicit ranking of the
which typically have not been well measured in survey obstacles, which we provide below, should be especially
research on contraception. In so doing, however, we do helpful for the setting of policy and program priorities.
not intend to downplay the potential significance of poor
access to high-quality family planning services. Access
remains a pervasive problem in most developing soci- Data and Methods
eties (Robey et al. 1996), and Pakistan is no exception.
Numerous studies have revealed that large segments of The data analyzed in this article were collected in 1996
the population, concentrated in rural areas, face consid- in Punjab province, Pakistan, by the Population Coun-

Volume 32 Number 2 June 2001 99

This content downloaded from 119.157.179.220 on Sun, 23 Feb 2025 05:04:08 UTC
All use subject to https://about.jstor.org/terms
cil, under the direction of the second and third authors. various perceived costs of contraception and the strength
The survey used a probability sample of the entire prov- of motivation affect contraceptive practice. We are aware
ince, and yielded a sample of 23 clusters (eight urban, of no other survey data collected in Pakistan in recent
15 rural) from which 1,310 currently married women years that contain the same combination of variables
aged 20-44 and 554 of their husbands were interviewed and, indeed, only a few equivalent data sets are avail-
successfully. Communities were sampled with unequal able from any developing country setting.
probability of selection, and therefore sampling weights In an investigation of barriers to contraceptive use,
are applied throughout the analysis unless otherwise the natural choice of dependent variable is current con-
noted. For further details about the design of the sur- traceptive-use status. From the standpoint of causal
vey, see Population Council (1997). modeling, however, the cross-sectional nature of the sur-
The principal objective of the 1996 study was to in- vey data presents a problem. Contraceptive use is mea-
vestigate the causes of unmet need for family planning. sured contemporaneously with the various perceptions
To this end, information was obtained on each of the fac- listed above, and therefore, some ambiguity about cause
tors identified in the previous section. To obtain such and effect is inevitable: Have the woman's perceptions
information, the survey questionnaire asked contracep- led to her decision to practice (or not practice) contra-
tive-method-specific queries (Robinson and Cleland ception, or vice versa? For example, are concerns about
1992) concerning the respondent's knowledge, approval side effects a determinant, or a result, of using a meth-
(respondent and spouse), perceptions of access to services, od? The same can be asked about most of the measured
and concerns about side effects; a block of items concern- indicators of central interest in this analysis (Robinson
ing the respondent's views about whether or not certain and Cleland 1992). Indeed, since contraceptive-use sta-
factors were important hindrances to practicing contra- tus may have been stable for some time preceding the
ception in her community; and follow-up questions to the interview, the contraceptive behavior of many women
standard fertility-preference items that probed the degree may have preceded the formation of the perceptions ex-
of the respondent's attachment to the expressed prefer- pressed in the survey interview. In all likelihood, the
ences. These items were administered to all women, users causality has operated in both directions, whereas our aim
and nonusers alike. In addition, nonusers were asked why is to estimate one set of causal effects, that is, the effects
they were not using a method at the time of the survey of the obstacles to the decision to practice contraception.
and, if they indicated no intention to use a method in the To diminish this threat to the validity of the find-
future, why this was so. ings, we depart from most of the previous literature on
Additional information was collected on demo- reasons for nonuse of contraceptives and analyze the in-
graphic and socioeconomic characteristics of the respon- tention to use in the future among women not currently
dent and her household, including measures of eco- using a method. Several longitudinal analyses have
nomic status that have been shown to be associated with shown women's stated intentions to be strongly associ-
use of modern contraceptives in Pakistan (Mahmood 1992; ated with their later contraceptive behavior (for example,
Agha 2000). Two blocks of items inquired about the Bhatia 1982; Adler et al. 1990; Curtis and Westoff 1996).
woman's degree of autonomy and about the extent of To the extent that the intention to use is an accurate
her decisionmaking power in various domains of every- proxy for future contraceptive behavior, the causality is
day and family life. These measures may be particularly clarified and simplified: Contraceptive behavior is ex-
important in Pakistan, given the overall low levels of amined as a function of women's perceptions at an ear-
female autonomy and participation in activities outside lier time (that is, the time of the survey interview). In
the home (Shah 1986; United Nations 1993; Sathar and reality, stated intentions to use are not a pure proxy for
Kazi 1997). Assuming that the measured demographic future behavior; rather they are also influenced by other
and socioeconomic variables accurately reflect the stage unmeasured attitudes and past experiences affecting the
of the respondent's reproductive career and her social perceived obstacles that are the independent variables
class, these further items concerning mobility and parti- of central interest in this research. For this reason, in
cipation in household decisionmaking capture differ- modeling the intention to use, we have not entirely re-
ences in women's ability to formulate and express repro- solved the problem of sorting out cause and effect. Nev-
ductive intentions, their authority to make reproductive ertheless, by excluding current users and examining in-
decisions, and their capacity to access the information tentions about future behavior, we feel that we are on
and services required to implement those decisions. relatively firm footing in positing causal relationships.
Collectively, these items provide an unusually solid We devise an intention-to-use variable akin to the
empirical foundation for testing hypotheses about how unmet-need indicators commonly used in the research

100 Studies in Family Planning

This content downloaded from 119.157.179.220 on Sun, 23 Feb 2025 05:04:08 UTC
All use subject to https://about.jstor.org/terms
literature. The analysis is restricted to nonusers who ex- tions in the structural model regresses the latent variables
press a desire to avoid pregnancy, that is, women who on the demographic and socioeconomic variables, which
state that they want no more children or that they want are assumed to be exogenous in this analysis. A further
to postpone the next birth for at least two years. For these equation consists of the intention to use regressed on the
women, we construct an indicator of whether or not they six obstacles to contraception (the latent variables), and it
intend to use a contraceptive during the next two years. is this equation that is of paramount concern in this re-
Women who want another child soon, women who per- search. Although the three sets of equations are sepa-
ceive themselves to be infecund, women who are prac- rated for presentation purposes, they are estimated
ticing contraception at the time of the survey, and wom- jointly. Values of the latent variables are not calculated
en with missing information on a few explanatory directly, as they would be if indexes or scales were con-
variables are all dropped from the analysis, leaving 579 structed from the measured indicators. Rather, given the
unweighted cases for analysis. Thirty-five percent of posited relationships among the measured indicators
these women indicate an intention to use a contracep- and the latent variables, all coefficients of interest can
tive within two years.3 be calculated, along with their standard errors.4
To estimate the effects of interest in an efficient and In this analysis, the latent variables are assumed to
parsimonious fashion, we carry out structural equations be continuous in nature. Our measured indicators, how-
modeling, treating the various obstacles to contracep- ever, are a combination of continuous and categorical
tion as unobserved ("latent") variables (Bollen 1989). The variables, and therefore, it is not appropriate to assume
full model contains several distinct components and is linear regressions throughout the model and conven-
admittedly complex (with estimation requiring special- tional Gaussian multivariate normality. The estimation
ized software), but each component is relatively straight- is, therefore, more challenging. We use the approach de-
forward and resembles familiar regression analysis. The veloped by Muthen over the past 15 years (Muthen 1983
key feature of the approach is the assumption of the ex- and 1984; Muthen et al. forthcoming), as implemented
istence of latent variables, not observed directly but in- in the software package Mplus (Muthen and Muthen
stead expressing themselves through responses to sur- 1998). The categorical indicators used in the measure-
vey items. Any one of these items may be imperfectly ment model must be either dichotomous or ordinal, a
measured, but collectively they are assumed to provide requirement met in the Punjab data. For these categori-
an adequate representation of the unobserved variable. cal indicators, the regression model can be viewed as a
This approach takes maximum advantage of the Punjab probit (binary or ordinal), yielding merely one param-
survey data, which contain multiple items associated eter estimate for each effect of the latent variables even
with each of the hypothesized obstacles that may ac- when the measured indicators are polytomous. In the
count for unmet need. portion of the structural model in which the latent vari-
The full model consists of three components, each ables serve as the dependent variables, because the la-
of which is a set of regression equations: (1) measured tent variables are assumed to be continuous, all the re-
indicators regressed on the latent variables (that is, the gressions are linear regressions. To make the regression
obstacles to contraception) (27 equations); (2) latent vari- results more comparable across variables and equations,
ables regressed on the background variables (six equa- we present standardized coefficients. The statistical tests
tions); and (3) intention to use regressed on the latent employ robust standard errors that take into account
variables (one equation). heteroskedasticity in the data. For details on the meth-
The first set of equations constitutes the so-called od of estimation (which is maximum likelihood via
measurement model. As in all commonly used forms of weighted least squares), see Muthen and Muthen (1998).
factor analysis, in the measurement model the measured The following seven background variables are selec-
indicators should be viewed as the dependent variables, ted: number of living sons, size of residence, household
and the latent variables as the independent variables. wealth, years of schooling, the woman's decisionmak-
The regression coefficients estimated in the measure- ing autonomy, exposure to family planning messages
ment model give a sense of the relative strength, or on radio, and exposure to family planning messages on
weighting, of the relationships between the latent vari- television. (Past use of contraceptives, which in the mod-
able and its measured indicators. These coefficients are, eling is treated as one of the measured indicators of sev-
however, of limited interest for our purposes; they eral of the obstacles to the practice of contraception, can
should be viewed as building blocks for the estimation be regarded as a further background variable.) In ex-
of the "structural model." The second two sets of equa- ploratory regressions, several other background vari-
tions constitute the structural model. One set of equa- ables measured in the 1996 survey proved to have no

Volume 32 Number 2 June 2001 101

This content downloaded from 119.157.179.220 on Sun, 23 Feb 2025 05:04:08 UTC
All use subject to https://about.jstor.org/terms
net effects: age of the respondent, the husband's years hold matters, and among those who recall family plan-
of schooling, and the dwelling's construction material ning advertisements on radio or television.
(assumed to be indicative of household wealth). Some The one set of differentials that do not conform to
further comments are in order about the number of liv- expectations are those for indicators of health concerns
ing sons, which is known to be a strong predictor of con- about contraception. Two of the indicators produce no
traceptive use in Pakistan, reflecting the prevailing pref- differential in the intention to use, and the association
erence for sons over daughters (Khan and Sirageldin with the number of methods thought to have bad health
1977; United Nations 1993; Mahmood and Ringheim side effects is sharply positive. Although this latter re-
1996; Winkvist and Akhtar 2000). Most previous stud- sult appears counterintuitive, in fact empirical data from
ies of the effects of son preference on reproductive be- contemporary developing societies commonly show an
havior in Pakistan have examined effects that operate association between fear of methods' side effects and
through fertility preferences. A different hypothesis is contraceptive use that works in this direction (for ex-
considered here: that among those who have already ex- ample, Biddlecom and Kaona 1998). The most plausible
pressed a desire to stop or postpone childbearing, the explanation for this empirical regularity is that famil-
strength of attachment to this preference is positively iarity with contraception, which naturally should be
associated with the number of living sons. greater among experienced contraceptive users (former
or current) and among those seriously contemplating
future use, leads to greater awareness of the possible
Results health consequences. Moreover, the structure of the 1996
inquiry in Punjab was such that only those women who
The measured indicators of the six latent factors-strength were aware of specific methods of contraception were
of the motivation to avoid pregnancy, knowledge of con- asked about whether they thought that their use caused
traception, and the four costs of practicing contracep- bad side effects to health, a questionnaire design feature
tion, along with the socioeconomic, demographic, and that induces a positive association between number of
women's autonomy variables-are listed in Table 1, methods known (itself positively associated with the in-
which also shows the percentage distribution of the tention to use) and number of methods thought to cause
sample (unweighted) across the categories of each indi- bad side effects. This finding is an excellent example of
cator (right-hand column) and the proportion intend- the problem of reverse causality-familiarity with con-
ing to practice contraception, the converse of unmet need traception's affecting perceptions of contraception-dis-
(middle column). With only a few exceptions noted be- cussed above. Evidently the decision to analyze inten-
low, none of these bivariate relationships runs counter tion to use, rather than contraceptive use per se, has not
to expectations. Large differentials in the intention to alleviated this problem entirely. The analysis includes
use, amounting to 20 percentage points or more, are ap- indicators of knowledge and awareness of contraception,
parent for one or more measured indicators of each of however. In the regression analysis, adjustment for this
the six factors. The differentials are somewhat smaller set of indicators should help contain some of the bias
for motivation to avoid pregnancy (comparatively weak evident in Table 1.
associations are found between intention to use and con- The size of Table 1 makes vividly clear the large
cerns and worries about having another child) and health number of measured indicators available for this analy-
concerns. Among the specific indicators, the differentials sis (23 indicators in all, plus the background variables),
are especially large for the respondent's approval of con- a distinctive and advantageous feature of the Punjab
traception, the number of modern methods approved, data.5 Each of the six latent variables is represented by
religious concerns as a reason for not using, husband's three to seven measured indicators, as specified in the
approval of contraception, the number of modern meth- measurement model presented in Table 2. The structure
ods the husband is thought to approve, the ease of dis- of the measurement model is not quite as simple as Table
cussing family planning with the husband, and the num- 1 suggests: The husband's desire for another child is used
ber of methods for which sources of supply are known. as an indicator of both motivation and husbands' attitudes,
Looking at the background variables in the lower por- and past use of contraceptives is used as an indicator of
tion of the right-hand column, the proportion intend- knowledge, acceptability, and husbands' attitudes.6
ing to use a contraceptive is substantially higher among A close look at the estimates in Table 2 reveals that
women with more living sons, urban women, women every one of the coefficients in the measurement model
in wealthier households (that is, households with more is significant at the 0.001 level. From the signs of the co-
possessions), women who have attended school, wom- efficients, the nature of the unobserved obstacle can be
en who report greater decisionmaking power in house- inferred. It is thereby apparent that the strength of mo-

102 Studies in Family Planning

This content downloaded from 119.157.179.220 on Sun, 23 Feb 2025 05:04:08 UTC
All use subject to https://about.jstor.org/terms
Table 1 Among women surveyed, percentage intending to practice contraception in the future and percentage distribution of

Percentage distribution Percentage

respondents, by categories of measured indicators, 1996 Punjab Unmet Need Survey

distribution
Category and Percent intending of respondents Category and Percent intending of respondents
measured indicator to use a method' (N = 579)b measured indicator to use a method' (N =579)b

Wants more
more224238
62No
Yes38
12 88
Motivation to avoid pregnancy Reason for not using a method: husband's
Desire for children opposition

1 28
36 50
Wants no
2+ 17 0
1 25
28 18
29
12
Number of concerns about having another child Respondent's health concerns

No 35 60 3+ 47 27
0 45 33 Number of methods having bad side effects

Would be worried if she became pregnant 2 40 26

Yes 36 40 Side effects are an obstacle to use
Husband's
Wants more desiresoonfor children
22 28 Yes No 36
32 84
16
Wants no more 48 47 NoYes 363283
Wants more later 27 25 Reason for not using a method: health concerns

5 38 31 0-1 8 15
Knowledge/awareness of contraception 17

6+ 50 15 2-3 34 30
Number of modern methods known Access to services
4 36 27 are known
0 27 56 5+46 31
0-3 24 27 Number of methods for which sources of supplies

Number of traditional methods known 43 24

1?+ 45 44 Perceived proximity of nearest services
Acceptability of contraception On foot,<30 minutes 45 22
No 144
2 30 More distant 20
42 39
No 23 50No 33 56
Approves of family planning On foot, 30+ minutes 36 10
Yes 70 No facility 29
Yes 45 i-aYes 38 44
Perceives that in-laws approve Can visit health facility alone

1 372 37
2420 Never usedused266472
0 10 25 Past use of contraceptives
Number of modern methods respondent approves of Background variables

No 30of 77 1 29 30
~~~~~~~~~~~~~Ever 28
Yes 51
Approves
23 2
withdrawal
05 0 25 13
3+ 52 31 Number of living sons

disapproval Rural 31 74
No 38 39 Urban 50 26
No 41 78 1 34 36
Thinks contraceptive use might provoke divine Area of residence

Perceptions of husband's attitudes 3+ 23

Yes 12 22 2 43
Yes 33 61 Number of household items owned
Reason for not using a method: religious concerns 0 27 26

No 11
Yes 49 430337
57 1+ 42 26
41 15

0 16 48 0 25 25
Husband approves of family planning Schooling (years)

1
2
3+ 44
4265 20
14
18 1
2
3+ 33
4139 19
20
36
Number of modern methods of which husband Number of matters in which respondent
approves can make household decision on her ownc

No 32 83 No 30 30
Yes 52 17 Yes 48 48
Husband approves of withdrawal Exposed to family planning information on radio

Difficult 10 23 No 22 22
Easy 42 77 Total
Yes 44 44
35 100
Discussion of family planning with husband is Exposed to family planning information on television

Note: Sample consists of currently married women aged 20-44 not currently practicing contraception who perceive themselves to be fecund and who want to post-
pone the next birth or have no more children. Four respondents were dropped who lacked information on one or more of the measured indicators.
a Weighted. b Unweighted. c Respondents were asked if they could decide on their own how to spend part of the household income, whether to send their children to
school, where to take a sick child, and whether they could buy new clothing for themselves.

Volume 32 Number 2 June 2001 103

This content downloaded from 119.157.179.220 on Sun, 23 Feb 2025 05:04:08 UTC
All use subject to https://about.jstor.org/terms
tivation to curtail childbearing should have a positive Table 3 consist of the effects of the seven background
effect on the intention to use a contraceptive; that knowl- variables on the latent variables (that is, the obstacles to
edge is a matter of familiarity with contraception and contraception). With the exception of husband's atti-
also should have a positive effect on the intention to use tudes and health concerns, all of the obstacles vary sig-
a method; that social acceptability concerns approval of nificantly according to four or more of these background
contraception and therefore should have a positive ef- variables. None of these effects is surprising. Overall,
fect on the intention to use; that the wife's perception of the respondent's schooling is a weaker explanatory vari-
husband's attitudes captures aspects of the husband's able than are urban residence and the woman's decision-
views that should facilitate contraceptive use, and hence making autonomy. From the latter result, the conclusion
should have a positive effect on the intention to use; that could be drawn that greater insight about the determin-
the health-concerns indicator is a positive function of ing role of women's status is obtained from focused mea-
fears about detrimental side effects of contraceptives, sures of female autonomy than from more global and
and hence should have a negative effect on the intention diffuse variables such as schooling (a conclusion con-
to use; and, finally, that access to services is a matter of sistent with other recent empirical research, for example,
greater availability of services, and therefore should Jejeebhoy 1996; Sathar and Kazi 1996).
have a positive effect on the intention to use a method. For our purposes, the key coefficients are those ex-
In short, five of the six latent variables are expected to pressing the effects of the six obstacles on the intention
show positive effects on the intention to use. to use (shown in the lower panel of Table 3). Three of these
The measurement model in Table 2 is of lesser in- are statistically significant-the effects of knowledge,
terest in this research than the structural model pre- acceptability, and husband's attitudes. Of the remain-
sented in Table 3. The six equations in the top panel of ing three, the effects of motivation and health concerns

Table 2 Coefficients of measured indicators of latent variables that may act as obstacles to contraceptive use, by indicator, 1996
Punjab Unmet Need Survey

Latent variable (N = 579') Coefficientb

Motivation to avoid pregnancy Desires no more children 0Q77c
Measured indicator

Number of concerns about having another child 0.68**

Would be worried if became pregnant 0.82**
Husband desires no more children 0.61**
Knowledge Number of modern methods known 0.56c
Number of traditional methods known 0.78**
Past use of contraceptives 0.49**
Acceptability Respondent approves of family planning 0.85c
Perceives that in-laws approve 0.58*
Number of modern methods of which respondent approves 0.64**
Respondent approves of withdrawal 0.44**
Thinks contraceptive use might provoke divine disapproval -0.21**
Religious concerns given as reason for not using a method -0.79**
Past use of contraceptives 0.25**
Respondent's perception of husband's attitudes Husband approves of family planning 0Q95c
Number of modern methods of which husband approves 0.81**
Husband approves of withdrawal 0.31
Husband desires no more children 0.26**
Ease of discussing family planning with husband 0.81 **
Husband's opposition given as reason for not using a method -0.68**
Past use of contraceptives 0.26**
Respondent's health concerns Number of methods thought to have bad side effects 0Q75c
Side effects are obstacle to use 0.54**
Health concerns given as reason for not using a method 0.29**
Access to services Number of methods for which respondent knows source 0.1 6c
Perceived proximity to nearest services -0.26**
Respondent can visit health facility alone 0.47*
*Significant at p<0.001 (one-sided).
aFor sample-selection criteria, see Note to Table 1. bStandardized coefficients. CUnstandardized coefficient set to 1.00, for model identification; therefore,
statistical test is inappropriate (see Maruyama 1998).

104 Studies in Family Planning

This content downloaded from 119.157.179.220 on Sun, 23 Feb 2025 05:04:08 UTC
All use subject to https://about.jstor.org/terms
Table 3 Structural model equations for latent variables showing coefficients of the effects of background variables, and for
intention to use a contraceptive showing coefficients for the effects of latent variables, 1996 Punjab Unmet Need Survey
Latent variablea
Variable Motivation Knowledge Acceptability Husband's attitude Health concerns Access to services

Number of living sons 0.54** - - 0.12** - -

Background variable

Urban residence 0.08* 0.28** 0.12** 0.03 0.14* 0.32*

Household items (2+) - - - 0.12* - -
Schooling (1 + years) 0.00 0.22** 0.1 1* 0.00 -0.01 0.45**
2 0.18**
0.18**0.1 7** 0.12** 0.04 -0.01
Makes own household decisionsb

3+ 0.28** 0.08* 0.10* 0.16* 0.39**

0.79**
on radio - 0.07* 0.06* -
Exposed to family planning information

on television - 0.14** 0.29**

Exposed to family planning information

Motivation -0.03
Latent variable Effects on intention to usea
Knowledge/awareness of contraception 0.30**
Acceptability of contraception 0.47**
Husband's attitudes 0.46**
Health concerns 0.04
Access to services -0.25
* Significant at p<0.05; ** p<0.001. - Not in equation. (N) = 579.
Note: For sample-selection criteria, see note to Table 1.
aStandardized coefficients are given. bRespondents were asked if they could decide on their own how to spend part of the household income, whether to send
their children to school, where to take a sick child, and whether they could buy new clothing for themselves.

are close to zero, and the effect of access is negative. The husbands' views is matched only by their concerns about
major conclusion from Table 3, therefore, is that the key the acceptability of contraception as a predictor of their
obstacles to contraceptive use in Punjab are the perceived intention to use a contraceptive in the future. The large
social and cultural unacceptability of contraception and effect of acceptability is expected; many studies con-
wives' perceptions that their husbands are opposed to ducted in Pakistan, reviewed above, have stressed the
(or, at the least, not supportive of) family planning. salience of social and religious concerns.
These two obstacles are essentially equal in strength. The variables concerning husbands' views are de-
About two-thirds as strong, according to the standard- rived entirely from interviews with their wives; that is,
ized coefficients, is lack of contraceptive knowledge. these are wives' perception of the attitudes and prefer-
The dominance of acceptability and husband's atti- ences of their husbands. The 1996 data collection in Pun-
tudes is consistent with our interpretation of the quali- jab included interviews with a subset of husbands,7 who
tative interviews we conducted and, on the whole, with were administered essentially the same questionnaire as
previous research results for Pakistan. Missing from re- that presented to their wives. With these data, the hus-
sults shown in Table 3, but prominent in our conclusions bands' responses can be compared with their wives'
drawn from the qualitative interviews and from the ex- perceptions to assess their accuracy. Table 4 shows this
isting research literature, are women's fears of the det- comparison for the variables used in this analysis. Con-
rimental side effects of contraceptives on health. That siderable discrepancy is found between the wives' per-
outcome is discussed below. The existing literature is, ceptions and the responses provided by their husbands.
to some degree, ambivalent as to whether husbands' at- Clearly, such spousal discordance is not limited to Pa-
titudes, expressed or perceived, constitute a major bar- kistan (Thompson 1990; Becker 1996; Bankole and Singh
rier to Pakistani women's ability to implement their fer- 1997). Because gender stratification is especially sharp
tility preferences by practicing contraception. Studies in Pakistan, however (as compared, for example, with
that explicitly or implicitly downplay the dominance of that in other South and Southeast Asian countries-see
these variables include Hashmi et al. (1993), Shah and Mason and Smith 2000; see also United Nations 1993 and
Shah (1984), and Mason and Smith (2000). Our results Sathar and Kazi 1997), one would expect wives' mis-
are emphatic on this point: Among women who wish perceptions of husbands' views to be unusually pro-
to avoid becoming pregnant, their perceptions of their nounced. Most dramatic is the discrepancy regarding

Volume 32 Number 2 June 2001 105

This content downloaded from 119.157.179.220 on Sun, 23 Feb 2025 05:04:08 UTC
All use subject to https://about.jstor.org/terms
husband's approval of family planning: Among couples power with multivariate controls. The confounding ef-
in which the wives felt that their husbands did not ap- fects of obstacles such as the acceptability of contracep-
prove of contraception, 61 percent of the husbands in- tion and the husband's views of contraception appear
dicated that they did approve of it. The discrepancies in to account for the bivariate associations between the in-
Table 4 are larger for the variables that couples are less tention to use and strength of motivation (as represented
likely to have discussed: For example, wives more ac- by items such as the desire for another child and con-
curately perceive their husbands' desires to stop than cerns about having another child). In this instance, mul-
to space childbearing. Because fertility limitation is the tivariate modeling provides a revised and clearer pic-
more consequential step, we suspect that couples are ture of the relative magnitude of hypothesized effects.
more likely to discuss it than to discuss spacing their The message from the regression results in Table 3 is
children. Interestingly, husbands' disapproval of with- that strength of motivation to avoid pregnancy, as mea-
drawal (associated with a lower likelihood of intending sured in the Punjab survey, has little bearing on contra-
to use, as shown in Table 1), is more likely than their ceptive intentions, once other determining factors are
approval of it to be perceived accurately by their wives. taken into account.
Finally, whereas most husbands report that communi- In the discussion of results shown in Table 1 with
cation about family planning is easy, they are more likely regard to respondents' health concerns, we pointed out
to express this opinion if their wives hold that view as that the bivariate associations were not as expected, in
well. The wife-husband comparisons in Table 4 suggest all likelihood because of a reverse causality in which fa-
that the effects of the husbands' variables in the regres- miliarity with contraception makes women more aware
sion analysis reflect both genuine attitudes and prefer- of their negative side effects on health. With controls for
ences of the husband that work against adopting a con- other variables in Table 3, including knowledge of con-
traceptive method and erroneous perceptions on the part traceptives, the positive effects of the indicators of health
of the wife. concerns evident in the bivariate analysis of Table 1 have
Surprising results in the structural model of Table 3 been nullified. Nevertheless, our hypothesis that anxi-
are the weak effects on the intention to use of strength eties about detrimental health consequences are a seri-
of motivation to avoid pregnancy and health concerns. ous impediment to contraception is not substantiated.
In the case of motivation, the moderately strong bivari- We are not convinced that these results provide a valid
ate associations evident in Table 1 do not retain their assessment of the determining role of health concerns
in contraceptive decisionmaking in Pakistan. The wide-
spread concern about adverse health repercussions that
is evident in both the qualitative interviews and in the
Table 4 Percentage of husband's reported attitudes toward survey data collected in Punjab in 1996 must, we feel,
contraception that match wife's perceptions of husband's act as a serious obstacle to the adoption and continued
attitudes, 1996 Punjab Unmet Need Survey
use of contraceptives. This concern may be expressed
As reported
as much, or more, in the choice of a contraceptive meth-

No 39
31 61 100
by husband
Wife's perception (n = 248)B Total od as in the decision to practice contraception per se.

Yes 69 100
Husband approves of family planning No Yes
Methods such as withdrawal, the condom, and periodic
abstinence have been relatively popular in Pakistan in

0
1-2583223
5319
15 100
husband approves 0 1-2 3+
Number of modern methods of which the 1990s, contributing significantly to the recent in-

100
3+ 48 31 21 100
crease in contraceptive prevalence (NIPS/LSHTM 1998).
In-depth research has revealed that a common reason

No 86
65 14 100
for choosing withdrawal is the absence of side effects as

Yes 35 100
Husband approves of withdrawal No Yes compared, for example, with hormonal and surgical
methods (Ministry of Population Welfare and Popula-

Soon 49 13 38 100
Husband desires another child Soon Later No more tion Council 1998). By this reasoning, the results in the

Later 50 31 19 100
Wantsno more 8 10 82 100
lower panel of Table 3 may not be misleading: As of the
mid-1990s, fear of side effects may have far greater in-
fluence on method choice than on the decision to use a

Difficult
Easy 32 406860100
100
Spousal discussion of family planning is Difficult Easy
contraceptive, hence the negligible effect on the inten-
tion to use a method.
aThe selection criteria-for women are the same as those specified in note to Table In the case of access to services, again questions can
1. This table is calculated for the subset of those women whose husbands were
interviewed successfully. See Population Council (1997) for details on sampling
be raised about the adequacy of measurement in the
procedures. Punjab survey data, as mentioned above. Complaints

106 Studies in Family Planning

This content downloaded from 119.157.179.220 on Sun, 23 Feb 2025 05:04:08 UTC
All use subject to https://about.jstor.org/terms
about inaccessible and low-quality services were com- enough, the woman or couple will resist contraceptive
mon in the qualitative interviews we conducted in 1996, practice despite a desire to avoid another pregnancy (in
and other independent studies carried out in the past the short term or indefinitely).
five years have continued to document a seriously defi- In view of the dominant status of this framework,
cient, albeit improving, family planning service environ- surprisingly little research has been conducted (espe-
ment (Rosen and Conley 1996; Ministry of Population cially quantitative empirical research) on the full array
Welfare 1999). Inadequate access to services is especially of hypothesized contraceptive costs (Robinson and Cle-
pronounced in rural areas. Moreover, intervention stud- land 1992). In part, the lack of such research can be ex-
ies show that improvements in services result in sub- plained by the difficulty of measuring these costs; in-
stantial increases in contraceptive prevalence (Shelton deed, some of them are intrinsically nearly impossible
et al. 1999). Table 3 may, nevertheless, be correct in show- to measure. The conclusion is hard to avoid, however,
ing that access to services constitutes a less imposing bar- that the scant empirical attention to the magnitude of
rier to contraceptive use for Punjabi women than do contraceptive costs and their effects on contraceptive
some of the other obstacles examined here, such as the decisionmaking reflects less than full respect for the po-
social and cultural acceptability of contraceptives and tential power of the various possible obstacles to con-
users' perceptions of their husbands' views regarding traceptive use. In the design of the 1996 survey in Punjab
family planning. That the modest effect of access is nega- province, we made a concerted effort to devise indica-
tive-that is, that greater access discourages future con- tors of a wide range of obstacles. As a result of this ef-
traceptive use-as suggested by the estimates in Table fort, the regression analysis can draw on a large num-
3 is harder to accept, however. Perhaps women who in- ber of measures. In the Punjab data, the measurement
tend to use a method are more cognizant of the inacces- of motivation might be viewed as unsatisfactory-a
sibility of family planning services and are therefore point that we would not entirely dispute; and the mea-
more likely to find existing services less than ideally situ- surement of contraceptive costs is far from complete-a
ated. In any case, this finding for Punjab resembles re- point we would also readily concede. Nevertheless, this
sults from other settings: When asked directly, women research has an advantage over previous approaches in
rarely give much weight to poor access to services as an its systematic treatment of contraceptive costs. We en-
explanation for their unwillingness to adopt contracep- courage other researchers to explore a wide range of ob-
tion or for their discontinuation of contraceptive prac- stacles to contraceptive use, and, if possible, to go be-
tice (Bongaarts and Bruce 1995; Westoff and Bankole yond those detailed here.
1995; Mishra et al. 1999). As noted above, our findings are generally consis-
tent with other research conducted in Pakistan during
the past two decades. What sets this research apart from
Conclusion and Implications previous research is the assignment of explicit weights
to the obstacles, with women's perceptions of their hus-
Although the primary motivation for undertaking this bands' attitudes and preferences assuming a more domi-
research was to improve our understanding of the de- nant position in our results. In part, this result may re-
terminants of contraceptive behavior in Pakistan, the re- flect differences in research methodology, but it might
sults have larger implications for the field. For 25 years, be indicative of changes that have occurred in Pakistan
the dominant paradigm in research on family planning during the 1990s, when women became better informed
in contemporary developing societies has been the about contraception and more prepared to regard it as
Easterlin Synthesis Framework (Easterlin 1975), which a viable option. We suspect that such changes were rela-
posits that two sets of factors bear directly on the deci- tively more intense in Punjab province than elsewhere
sion to exercise deliberate fertility regulation, namely the in Pakistan. The data indicate that women's perceptions
desire to avoid becoming pregnant and the costs (de- of their husbands' views are not entirely accurate, that
fined broadly) of birth control. Among its many virtues, they perceive their husbands to have more negative
the synthesis framework offers a parsimonious expla- views toward family planning than is the case. This find-
nation for the widespread lack of fit between the desire ing provides some basis for suggesting that husbands'
to limit or stop childbearing and contraceptive practice: opposition will become a weaker obstacle in the com-
that the gap will occur when the motivation to avoid ing years in Pakistan. Indeed, we should expect the vari-
pregnancy is overwhelmed by one or more costs of fer- ous contraceptive costs to rise and fall with the passage
tility regulation. These costs already may have been ex- of time. We stress that the findings from this study per-
perienced by women or couples, or they may be antici- tain to a specific historical juncture and, moreover, are
pated. In either case, if the perceived costs are high based on data from only one province (albeit the most

Volume 32 Number 2 June 2001 107

This content downloaded from 119.157.179.220 on Sun, 23 Feb 2025 05:04:08 UTC
All use subject to https://about.jstor.org/terms
populous). The ranking of factors may differ in more re- tion of misperceptions of husbands' views and genuine
mote provinces such as Baluchistan, where access to ser- opposition on the part of men, we believe that two types
vices in particular may rank higher in the ladder of rea- of strategies to overcome the obstacle that men present
sons for unmet need. are implied: one set to improve spousal communication
The implications of these findings for policy formu- (about contraception, childbearing, and related issues)
lation and program development are clear. Pakistani (Mahmood and Ringheim 1997), and the second to
women face several obstacles to implementing their fer- change men's attitudes (about the desirability of large
tility preferences. Some consensus has been reached on families and about contraception). The former might in-
how the availability and quality of family planning ser- clude activities that "break the ice" where communica-
vices can be improved (Shelton et al. 1999), but achiev- tion is especially constrained (possibly through well-
ing this goal is, of course, another matter. Less is known designed IEC campaigns), whereas the latter would in-
about how policies and programs can overcome the non- clude efforts to educate men about the risks associated
access obstacles. In particular, it is clear that husbands' with having large families compared with the risks of
attitudes (their approval and intentions) and wives' as- contraceptive use. A different, although not contradic-
sessment of them are of overriding importance, and in- tory, strategy is to weaken men's predominant role in
deed some would view this factor as an almost insur- reproductive decisionmaking through activities that nur-
mountable obstacle to satisfactory resolution of women's ture the empowerment of women to negotiate choices.
preference-use gap. This view would be too pessimis- Although a shift in the gender balance of power may
tic. Ample evidence indicates that an increasing fraction follow from larger social change, including broadened
of Pakistani men are strongly motivated to implement opportunities for schooling and paid employment, this
their own and their wives' fertility preferences through transition can be facilitated by the expansion of and im-
contraceptive practice. Moreover, husbands' and wives'
provement in reproductive health services that, among
preferences tend to be in close agreement. A discernable
other things, provide women with knowledge about
transformation of men's views toward fertility regula-
how, when, and where family planning methods can be
tion is under way in Pakistan; the conviction is growing
obtained. One means of making providers, especially
among men that family size must be limited, above all
community-based health workers, more sensitive to the
for economic reasons but also for health reasons (Levack
implications of gender stratification as an obstacle to
and Rahim 1998; Sathar and Casterline 1998). One con-
women's reproductive health, is to train them to address
crete indication is that nearly one-half of couples prac-
issues that encompass their clients' family and social
ticing contraception are using methods that require male
situations. At present, such services are driven by pro-
cooperation (the condom, periodic abstinence, and with-
viders' own perceptions of their clients' needs rather
drawal) (NIPS/ LSHTM 1998), and, as noted above, in-
than by an assessment based on information gleaned
crease in the use of these methods has made a substan-
from clients about their home realities and constraints.
tial contribution to the increase in contraceptive preva-
Overcoming men's and women's entrenched mis-
lence during the 1990s.
givings about the social and cultural acceptability of con-
Although the majority of husbands surveyed ap-
traception requires a different set of initiatives. Focused
prove of family planning, nevertheless Pakistani men
IEC campaigns, through the mass media and local ef-
continue to harbor many misgivings about limiting fer-
forts, can emphasize that family planning is not contrary
tility and practicing contraception (Bhatti and Hakim
to Islam and, more important, can be in the best inter-
1996). Although women appear to have an exaggerated
ests of women, men, and their children, and that it is
impression of their husbands' opposition, undoubtedly
ethically sound behavior. The most powerful counter-
a significant portion of perceived male resistance is real
force to concerns about the acceptability of contracep-
(Population Council 1997). Exacerbating the problem,
tion, however, may be the increasing fraction of couples
and to some degree accounting for it, is the segregation
of domains of women and men, with reproduction who have practiced contraception. Increased prevalence

clearly falling into the sphere of women (Kazi and Sathar that results from the reduction of other obstacles such

1997). Therefore, communication about reproduction as health concerns and poor services will, in turn, height-

and responsibility for day-to-day decisions pertaining en the general perception that contraception is an accept-
to reproduction and to reproductive health have been able choice for couples to make. Another consequence
largely restricted to women. For decades, women in Pa- of expanding prevalence is likely to be the diffusion of
kistan have been the focus of government services and information about contraception, which will help to
of information, education, and communication (IEC) overcome obstacles associated with limited awareness
campaigns. Recognizing that the problem is a combina- of methods, side effects, and sources of supply.

108 Studies in Family Planning

This content downloaded from 119.157.179.220 on Sun, 23 Feb 2025 05:04:08 UTC
All use subject to https://about.jstor.org/terms
Notes Biddlecom, Ann E. and Fred Kaona. 1998. "The nature of unmet need
for contraception in an urban African setting." Paper presented

1 Past use is also considered as an indicator of knowledge and aware- at the International Union for the Scientific Study of Population
ness, as indicated above, and of husband's opposition, as indicated seminar on Fertility Transition in sub-Saharan Africa, Nairobi,
in the next paragraph. The statistical approach used in this analy- 2-4 November.
sis permits us to regard specific measured indicators as indicative Bollen, Kenneth A. 1989. Structural Equations with Latent Variables.
of more than one underlying obstacle to contraception. New York: John Wiley & Sons.

2 Mahmood and Ringheim concede that the direction of causality Bongaarts, John and Judith Bruce. 1995. "The causes of unmet need
underlying the association between spousal communication and for contraception and the social content of services." Studies in
contraception is uncertain. Family Planning 26(2): 57-75.

3 The number of women dropped from the total sample of 1,310 Casterline, John B., Aurora E. Perez, and Ann E. Biddlecom. 1997.
because of each of the specified restrictions, applied sequentially, "Factors underlying unmet need for family planning in the Phil-
is as follows: wants a child soon, 285; infecund, 64; using a meth- ippines." Studies in Family Planning 27(3): 173-191.
od, 378; and missing information, 4.
Curtis, Sian L. and Charles F. Westoff. 1996. "Intentions to use con-
4 For a more complete presentation of this approach, including a traceptives and subsequent contraceptive behavior in Morocco."
discussion of technical issues, the reader is referred to textbooks Studies in Family Planning 27(5): 239-250.
such as Bollen (1989). Maruyama (1998) offers a less technical in- Easterlin, Richard. 1975. "An economic framework for fertility analy-
troduction.
sis." Studies in Family Planning 6(2): 54-63.
5 Only a few other measured indicators of latent factors were con- El-Zanaty, Fatma, Ann Way, Sunita Kishor, and John Casterline. 1999.
sidered, but they were dropped in the interest of parsimony once
Egypt Indepth Study on the Reasons for Nonuse of Family Planning.
their net effects were found to be not statistically significant. Cairo: National Population Council.
Given the large number of parameters to be estimated in rela-
Hashmi, Sultan S., Khushnood Alam, and Aysha Sheraz. 1993. Non-
tion to the number of observations and the large proportion of
users and Unmet Need for Contraception. Islamabad: National In-
measured indicators that are categorical (which further reduces
stitute of Population Studies.
the statistical power in this type of estimation), the fact that so
few effects were found to be insignificant is a testament to the Hermalin, Albert I. 1983. "Fertility regulation and its costs: A critical
overall quality of measurement and supportive of the concept essay." In The Determinants of Fertility in Developing Countries. Eds.

that has guided the design of this research. R. Bulatao and R. Lee. New York: Academic Press. Volume 2,
pp. 1-53.
6 We also estimated models in which past use served as an indica-
tor of health concerns and access. The statistical tests failed to Jejeebhoy, Shireen 1996. "Women's autonomy and reproductive
support the addition of these effects to the measurement model. behaviour in India: Linkages and influence of sociocultural con-
text." Paper presented at the International Union for the Scien-
7 See Population Council (1997) for a discussion of the sample de-
tific Study of Population seminar on Comparative Perspectives
sign and sampling results.
on Fertility Transition in South Asia, Islamabad, 16-19 December.

Kazi, Shahnaz and Zeba Sathar. 1997. "Pakistani husbands and wives:
Different productive and reproductive realities?" Paper presented
References at the annual meeting of the Population Association of America,
Washington, DC, 28-30 March.
Adler, Nancy E., Susan M. Kegeles, Charles E. Irwin, and Charles Khan, M. Ali and Ismail Sirageldin. 1977. "Son preference and the
Wibbelsman. 1990. "Adolescent contraceptive behavior: An assess- demand for additional children in Pakistan." Demography 14(4):
ment of decision processes." Journal of Pediatrics 116(3): 463-471. 481-496.

Agha, Sohail. 2000. "Is low income a constraint to contraceptive use Levack, Andrew and Tariq Rahim. 1998. "The concept of men as part-
among the Pakistani poor?" Journal of Biosocial Science 32(2): 161- ners in Pakistan." Populi 25(4): 6-8.
175.
Mahmood, Naushin. 1992. "Motivation and fertility control behaviour
Asturias de Barrios, Linda, Idalma Mejia de Rodas, Isabel Nieves, in Pakistan." Pakistan Development Review 31(2): 119-144.
Jorge Matute, and Nancy Yinger. 1998. Unmet Needfor Family Plan- Mahmood, Naushin and Karin Ringheim. 1996. "Factors affecting contra-
ning in a Peri-Urban Community of Guatemala City. Washington, ceptive use in Pakistan." Pakistan Development Review 35(1): 1-22.
DC: International Center for Research on Women.
. 1997. "Knowledge, approval and communication about fam-
Bankole, Akinrinola and Susheela Singh. 1997. "Couples' fertility and ily planning as correlates of desired fertility among spouses in
contraceptive decision-making in developing countries: Hearing the Pakistan." International Family Planning Perspectives 23(3): 122-129.
man's voice." International Family Planning Perspectives 24(1): 15-24.
Mason, Karen Oppenheim and Herbert L. Smith. 2000. "Husbands'
Becker, Stan. 1996. "Couples and reproductive behavior: A review of versus wives' fertility goals and use of contraception: The influ-
couple studies." Studies in Family Planning 27(6): 291-306. ence of gender context in five Asian countries." Demography 37(3):

Bhatia, Shushum. 1982. "Contraceptive intentions and subsequent 299-311.

behavior in rural Bangladesh." Studies in Family Planning 13(1): Maruyama, Geoffrey M. 1998. Basics of Structural Equation Modeling.
24-31. Thousand Oaks, California: Sage Publications.
Bhatti, Mansoor Ul Hassan and Abdul Hakim. 1996. Male Attitudes Ministry of Population Welfare. 1999. Population and Development: Pa-
and Motivation for Family Planning in Pakistan. Islamabad: National kistan Country Report for ICPD +5. Islamabad: Ministry of Popu-
Institute of Population Studies. lation Welfare, Government of Pakistan.

Volume 32 Number 2 June 2001 109

This content downloaded from 119.157.179.220 on Sun, 23 Feb 2025 05:04:08 UTC
All use subject to https://about.jstor.org/terms
Ministry of Population Welfare and Population Council. 1998. "A .1997. Women's Autonomy, Livelihood and Fertility in Rural Punjab.
Qualitative Investigation into the Use of Withdrawal." Research Islamabad: Pakistan Institute of Development Economics.
Report No. 6. Islamabad: Population Council. Shah, Iqbal H. and John G. Cleland. 1993. "High fertility in Bangla-
Mishra, Vinod K., Robert D. Retherford, P.S. Nair, and Griffith Feeney. desh, Nepal, and Pakistan: Motives vs. means." In The Revolution
1999. "Reasons for Discontinuing and Not Intending to Use Con- in Asian Fertility: Dimensions, Causes, Implications. Eds. R. Leete
traception in India." National Family Health Survey Subject Reports and I. Alam. Oxford: Clarendon Press. Pp. 175-207.
No. 13. Mumbai, India: International Institute for Population Sci- Shah, Nasra. 1986. Pakistani Women. Islamabad: Pakistan Institute of
ences. Development Economics.
Muthen, Bengt 0. 1983. "Latent variable structural equation model- Shah, Nasra M. and Makhdoom A. Shah. 1984. "From non-use to use:
ing with categorical data." Journal of Econometrica 22: 48-65. Prospects of contraceptive adoption." In Fertility in Pakistan: A
.1984. "A general structural equation model with dichotomous, Review of Findings from the Pakistan Fertility Survey. Eds. I. Alam

ordered categorical, and continuous latent variable indicators." and B. Dinesen. Voorburg, Netherlands: International Statistical
Psychometrika 49: 115-132. Institute. Pp. 149-162.

Muthen, Bengt 0. and Linda K. Muthen. 1998. Mplus: The Compre- Shelton, James D., Lois Bradshaw, Babar Hussein, Zeba Zubair, Tony
hensive Modeling Program for Applied Researchers. Los Angeles: Drexler, and Mark Reade McKenna. 1999. "Putting unmet need
Muthen and Muth6n. to the test: Community-based distribution of family planning in
Pakistan." International Family Planning Perspectives 25(4): 191-195.
Muth6n, Bengt O., S.H.C. du Toit, and D. Spisic. Forthcoming. "Ro-
bust inference using weighted least squares and quadratic esti- Sirageldin, Ismail, Douglas Norris, and J. Gilbert Hardee. 1976. "Fam-

mating equations in latent variable modeling with categorical and ily planning in Pakistan: An analysis of some factors constrain-
ing use." Studies in Family Planning 7(5): 144-154.
continuous outcomes." Psychometrika.
Stash, Sharon. 1999. "Explanations of unmet need for contraception
Nag, Moni. 1984. "Some Cultural Factors Affecting Costs of Fertility
in Chitwan, Nepal." Studies in Family Planning 30(4): 267-287.
Regulation." Population Bulletin No. 17. New York: United Na-
tions. Thompson, Elizabeth. 1990. "Fertility desires and fertility: Hers, his
and theirs." Demography 27 (4): 579-588.
National Institute of Population Studies (NIPS) and IRD/Macro In-
ternational. 1992. Pakistan Demographic and Health Survey 1990/ Tsui, Amy Ong. 1996. "Family Planning Programs in Asia: Approach-
1991. Columbia, MD: IRD/Macro International. ing a Half-Century of Effort." Asia-Pacific Population Research Re-
ports No. 8. Honolulu: East-West Center, Program on Population.
National Institute of Population Studies (NIPS)/London School of
Hygiene & Tropical Medicine (LSHTM). 1998. Pakistan Fertility United Nations. 1993. Women's Status and Fertility in Pakistan: Recent

and Family Planning Survey 1996-1997. Islamabad: National Insti- Evidence. New York: United Nations, Department for Economic
tute of Population Studies. and Social Information and Policy Analysis.

Population Council. 1997. The Gap Between Reproductive Intentions and

Viswanathan, Hema, Sona Godfrey, and Nancy Yinger. 1998. Reach-
ing Women: A Study of Unmet Need in Uttar Pradesh, India. Wash-
Behaviour: A Study of Pakistani Men and Women. Islamabad: Popu-
lation Council. ington, DC: International Center for Research on Women.

Westoff, Charles F. and Akinrinola Bankole. 1995. "Unmet Need:

Robey, Bryant, John Ross, and Indu Bhushan. 1996. "Meeting Unmet
1990-1994. Demographic and Health Surveys Comparative Studies No.
Need: New Strategies." Population Reports, Series J, No. 43. Bal-
16. Calverton, MD: Macro International.
timore: The Johns Hopkins University.
. 2000. "Trends in the demand for family limitation in devel-
Robinson, Warren and John Cleland. 1992. "The influence of contra-
oping countries." International Family Planning Perspectives 26(2):
ceptive costs on the demand for children." In Family Planning
56-62.
Programmes and Fertility. Eds. James Phillips and John Ross. Ox-
ford: Clarendon Press. Pp. 175-207. Winkvist, A. and H.Z. Akhtar. 2000. "God should give daughters to
rich families only: Attitudes towards childbearing among low-
Robinson, Warren C., Makhdoom A. Shah, and Nasra M. Shah. 1981.
income women in Punjab, Pakistan." Social Science and Medicine
"The family planning program in Pakistan: What went wrong?"
51(1): 73-81.
International Family Planning Perspectives 7(3): 85-92.
Yinger, Nancy V. 1998. Unmet Need for Family Planning: Reflecting
Rosen, James E. and Shanti R. Conley. 1996. Pakistan's Population Pro-
Women's Perceptions. Washington, DC: International Center for
gram: The Challenge Ahead. Country Study Series No. 3. Washing- T) __ - __ 'V lAT_ _
ton, DC: Population Action International.

Rukanuddin, Abdul Razzaque and Karen Hardee-Cleaveland. 1992.

"Can family planning succeed in Pakistan?" International Family
Acknowledgments
Planning Perspectives 18(4): 142-146.

Sathar, Zeba and John B. Casterline. 1998. "The onset of fertility tran- This research was supported by the Rockefeller Foundation
sition in Pakistan." Population and Development Review 24(4): 773- through the award "Unmet need for family planning: Its na-
796. ture and causes" and by the United Nations Population Fund
Sathar, Zeba and Shahnaz Kazi. 1996. "Women's autonomy and the through the award "Learning about men." An earlier version
onset of fertility change in rural Pakistan: The significance of gen- of this paper was presented at the annual meeting of the Popu-
der inequality across communities." Paper presented at the an- lation Association of America, New York, 25-29 March 1999.
nual meeting of the Population Association of America, New Or- The authors are grateful to Sajeda Amin, John Bongaarts, James
leans, 9-11 May. Shelton, and Amy Tsui for helpful comments on earlier versions.

110 Studies in Family Planning

This content downloaded from 119.157.179.220 on Sun, 23 Feb 2025 05:04:08 UTC
All use subject to https://about.jstor.org/terms