IPUMS-International

International Integrated Microdata Access System

Because the Australian public use samples are drawn from a sample of complete households, rather than a sample of unrelated individuals, information on family relationships within households can be used to construct a data file for couples.

The mechanics of file construction differ from one census to the next, depending on how the hierarchical public use sample was prepared prior to release, as well as on the basic information contained in the sample file.

For example, the 1981 public use sample has a single record for each person in the household, covering both personal, family and household data. To identify related couples (e.g. for analyses of religious, ethnic or educational homogamy), the researcher needs first to filter according to a range of criteria: income unit type (IUT -- head and spouse only, plus head, spouse and dependants); marital status (MST -- now married, including de facto); family type (FMC -- a complex set of codes distinguishing families with both head and spouse, with and without dependants); family number within each household (FNO -- primary or secondary family); and relationship to family head (RLF -- head or spouse).

An example of an application of 1981 sample data to issues pertaining to educational homogamy and social closure is available in F.L. Jones, "Marriage patterns and the stratification system: trends in educational homogamy since the 1930s", Australian and New Zealand Journal of Sociology, 23, No. 2 (July 1987), pp. 185-198 (local copy - 1.3 MB). Some of the basic SPSS command codes used to create the working data file for this analysis are attached (c81pairs.sps).

A more complex example using multilevel modeling and the 1986 public use sample is given in F.L. Jones, "Individual and group effects on ethnic intermarriage: a multilevel analysis", Australian Journal of Social Research, 3 (January 1997), pp. 17-37 (local copy - 1.5 MB). The task of constructing data on couples is somewhat less complex with this data set because it contains "markers" for spouse location (SPL) on census night (present/absent) and a "couple" indicator (MDC -- married or de facto). However, we constructed a new variable, household number (HHNO), a serial number for all households, to facilitate the correct identification of married couples within households. An example of SPSS code used to construct a data set restricted to persons married in Australia, still in their first marriage, and aged less than 60 is attached (c86pairs.sps). This data set was constructed to be as similar as possible to a special 1986 matrix tape commissioned for an analysis of ethnic intermarriage. See F.L. Jones and R. Luijkx, "Post-war patterns of intermarriage in Australia: the mediterranean experience", European Sociological Review, 12 (May 1996), pp. 67-86 (local copy - 2.5 MB).

Brief details of this marriage tape, and several other ancestry tapes, can be seen by consulting a file on this site (c86anc.txt) or by searching through the 1986 Census files at the ASSDA site.