APPENDIX B SUMMARY OF WEIGHTINGS PROCEDURES

Surveys of the type used in this study require weightings in order to produce accurate population
estimates. This is done by assigning each respondent with a weight that specifies the respondents
contribution to a population estimate (thus if the ith respondent has weight wi and a value of Xi on

variable X, the population mean is estimated as X wi X i / wi ; if all weights were

replaced with the value of 1, the resulting statistic would be X i / n , which is the simple sample

mean for X).

The use of respondent weights compensates for disproportionate sampling (for example due to
interviewing only one person from each household) and can also adjust departures from demographic
representativeness by the sample due to differential non-response between various demographic
groups.

The estimates produced in this report derive from data collected in two surveys: the Survey of Older
New Zealanders aged 65 years and over and the Survey of the Working-Age Population aged 18-64
years25.

The survey of older New Zealanders was administered through the Household Labour Force Survey
(HLFS) using the HLFS sampling frame and was concerned with the civilian, usually resident, non-
institutionalised population aged 65 years and over living in permanent private dwellings. One eligible
person per household was interviewed.

The survey of the working age population (18-64 years) involved house-to-house sampling where only
one person per household was interviewed and included people aged 18-64 years living in permanent
private dwellings.

The results from the two surveys have been combined to create a single dataset on the total
population. Weights are used to derive estimates for the total population and sub populations.

Population estimates have been calculated using respondent weights to represent the adult population
and child weights to represent the children in the respondents economic family unit. Children in this
study were not surveyed in their own right but are counted in the economic family units of which they

25
Data was also available from a third survey which was a supplementary sample of older Maori aged 65-69
years. Incorporating this data with the two main surveys would have provided so little gain in the accuracy of
population and sub-group estimates that it was judged not to justify the added complications that would have
resulted to making estimates.

191
are members. A dependent child is defined as being aged less than 18 years and with no partner or
child of their own.

The living standard score assigned to the relevant economic family unit is assigned to the child or
children in the unit. The children in the sampled economic family units in the Survey of Working Age
People were weighted to represent the count of children living with a working age parent or parents. A
child weight was not calculated for the Survey of Older People however only a very small proportion of
children live in an economic family unit with a parent aged 65 years or over. No information was
collected on people aged less than 18 years who did not live with their parents or who had a child or
partner aged less than 18 years of their own.

The population estimates presented in these reports are for adults and children living with adults in
private dwellings.

The specification of respondent weights to combine the two samples involved dealing with some
complications, the primary one being overlaps between the two populations. The overlap results from
a small sub-population of people over 65 years of age with partners less than 65 years of age who
may be double counted by the weightings. As noted earlier, dependent children amongst the 65 plus
population have not been accounted for by the weightings. These limitations are expected to have
only negligible effects on results produced by the weights.

Users of the dataset will find that the respondent weights to produce population estimates are given as
part of the documentation.

192
 APPENDIX C COMPARISON BETWEEN COMBINED
WEIGHTED RESPONSES AND CENSUS 2001

This appendix provides a comparison of population estimates obtained from the composite weighted
living standards survey dataset with the 2001 Population Census.

Table C1: Demographic and social characteristics

Weighted Responses Census 2001

Age groups26
Dependent children (under 18 years) 25.9 27.0
18-24 years 10.7 9.2
25-44 years 31.5 29.7
45-64 years 20.5 22.1
65+ years 11.4 12.1
Total 100.0 100.0

Gender
Male 49.8 48.8
Female 50.2 51.2
Total 100.0 100.0

Ethnic groups27
Mori 14.0 14.7
NZ Pacific 5.8 6.5
Chinese 2.0 2.8
Indian 1.4 1.7
Other 3.8 0.7
European 79.7 80.0

Regions
Auckland 27.3 28.8
Wellington 9.2 9.1
Other major urban 32.9 33.2
Secondary and minor urban 16.1 14.7
Rural 14.4 14.3
Total 100.0 100.0

26
The procedure for specifying the weights involved matching the weighted composite dataset to census based
population estimates of age and gender which forced close correspondence between the weighted survey and
census distributions on these variables.
27
Ethnicity is based on total responses to the ethnicity question; therefore the ethnic categories are not mutually
exclusive.

193
Table C1 continued: Demographic and social characteristics
Weighted Responses Census 2001

Housing tenure28
Owned family trust/ family and/or other 10.6 NA
Owned family unit 76.4 67.8
Rented private 10.4 19.6
Rented local authority 0.4 1.1
Rented HNZ 2.3 4.1
Total 100.0 100.0

Higher educational qualifications29

No formal qualifications 20.5 27.5
School qualifications 29.7 37.6
Occupational cert. dips. 34.8 21.9
Bachelors degree or higher qualifications 15.0 13.0
Total 100.0 100.0

Table C2: Economic and financial characteristics

Weighted Responses Census 2001

Major occupational groups of employed population aged 18-64 years 30

Leg., Adm. & Managers 15.1 14.1

Professionals and technicians 27.7 27.9
Agric. and Fisheries 9.5 7.9
Clerks, Service, Sales 16.9 26.4
Trades, plant, machinery 26.3 18.1
Elementary occupations 4.5 5.6
Total 100.0 100.0

Income source31 32
Income-tested benefits 16.0 21.8
New Zealand Superannuation 12.4 18.8
Market sources 71.6 73.4
Total 100.0

28
Census information on home ownership refers to those who own or part-own their home. The Census does not
include information on homes that are owned by a family trust. The comparison with the Census here is not
directly comparable as the Census distribution is based on a tenure/landlord distribution of households; the living
standards distribution is based on a tenure/landlord distribution of the population in Economic Family Units.
29
For the Living Standards Survey this information pertained to those 18 years of age and over; for Census 2001,
information was gathered for those aged 20 and over. As such, differences in educational qualifications between
the 2001 Census and Living Standards Survey may reflect differences in the classification and reporting of
qualifications.
30
The 2001 Census estimate of occupational distributional is not strictly comparable with the distribution obtained
from the living standards survey. This is due to a variety of factors including: the Census distribution is based on
usually resident population aged 20-64 years while the living standards survey distribution is based on the
occupational distribution of the main income earner in the economic family unit aged 18-64 years. The Census
distribution is likely to under-represent armed forces as they are less likely to live in private dwellings.
31
Census information on income source is based on total response and persons can be included in more than
one category. In contrast the estimates from the living standards survey are based on mutually exclusive
categories.
32
Comparisons were not possible between the survey and census income distributions. The income data
obtained from the surveys related to EFUs while the Census produced income data for individuals and
households but not EFUs.

194