Review of 2021 Census Topics

The inclusion of Children Ever Born (TISP) on the census is essential to the understanding of population change both at the national and at the subnational levels. I argue for the retention of Children Ever Born on a five-yearly basis. The potential advantages of extending of the scope of the children ever born data collection to include the provision of information on children for men and disaggregation of children into sons and daughters are also mentioned.
Fertility, along with mortality and in and out migration is one of the components of population growth. The formulation of fertility assumptions is integral to the preparation of population projections. Such projections underpin budgetary and other resource allocation to government departments and are integral to a wide range of planning and policy decisions. Children are also of fundamental importance to most Australian families. In view of its importance in both the public sphere and in the private sphere, research on the determinants and personal-level and societal-level consequences of fertility is critically important.
Decisions to have additional children depend in part on the number of children the partners already have, their numbers of sons and the numbers of daughters. For these reasons the collection of children ever born in each five yearly census can enhance the understanding and projection of fertility over the subsequent period. The linkage of data from successive five yearly censuses has the potential to considerably improve the projection of fertility levels.
Data on children ever born from the census is the only data source on fertility patterns for the multitude of population subgroups which are identifiable in the census. The collection of data on children ever born by the census has been proven to be unproblematic.
Fertility will remain of fundamentally importance to most Australian families in the future. It will inevitably be remain integral to determining future population and hence State and local government funding and budgets, electoral representation and boundaries, and resource allocation by a wide range of public sector, private sector and not-for-profit sector organsiations. The collection of data on children ever born every five years by the census would provide critically important time series.

Fertility, along with mortality and in and out migration is one of the components of population growth. The formulation of fertility assumptions is integral to the preparation of population projections, at the national, State or Territory and local levels. Such projections underpin budgetary and other resource allocation to government departments and are integral to a wide range of planning and policy decisions, including Treasury budget planning, urban planning, transport planning, education planning, and health service planning. Fertility decision making is also of fundamental importance to most Australian families. Both because of its importance in the public sphere and because of its importance in the private sphere, research on fertility and its determinants and personal-level and societal-level consequences is critical importance.
The more commonly-presented measures of fertility relate to particular time periods. In recent decades Australia’s Total Fertility Rate has undulated up and down. Age-Specific Fertility Rates for different ages have often moved in different directions over the same period of time: a pattern of decreases across the 15-29 age range and increases across the 30-49 age range has been evident for most of the last 30 years. These patterns are most readily understood in terms of the patterns within birth cohorts; as numbers of children produced below the age of 30 have generally reduced over time and there have been (almost) commensurate increases in birth rates above age 30 within the same cohort (a ‘postponement-catch up’ pattern) (Parr and Guest 2011). Knowing the numbers of women (and ideally their partners) with different numbers of children ever born at different ages can enhance the understanding and anticipation of such “catch up” effects. Immigration has an important bearing on the percentages of women (and men) of differing ages with different numbers of children. The percentages of newly arrived immigrants with zero children or one child are generally significantly greater than those for longer-established residents, and the ‘catch-up’ effect among migrants is more rapid (Parr 2014). We depend on census data on children ever born to reveal such patterns. Research has also shown the likelihood of a woman having another child depends not only on her current number of existing children but also on the number of sons, number of daughters and on her partner’s children (Gray and Evans 2005, Kippen et al. 2007, Thomson et al. 2014). Thus the collection of the number of children by sex and for men has the potential to improve the understanding of family formation patterns and the projection of future fertility.
In addition to considering the distributions of women of different ages by parity (number of children ever born), the analysis and projection of fertility should factor in intercensal probabilities of transition between parities. Linked data between successive censuses both of which include a question on numbers of children ever born can provide the estimates of such parity-transition probabilities. Linked census data on children ever born and birth registration data can also be used to estimate parity progression patterns.
Data on children ever born from the census is the only data source on fertility patterns for the multitude of population subgroups which are identifiable in the census. Thus, for example, exaggerated urban myths about the fertility of Muslim Australians are readily dispelled by fact (Allen and Parr 2017).
The ages and parity distribution of children among men differs from that among women. Men have children at later ages and both childlessness and having a relatively large number of children are more common among men (Parr 2010). Collecting census data on numbers of children for men would facilitate research on how in male parenting ages and parity vary between across the population.
It is therefore critically important to the understanding of fertility trends and to improving the forecasting of fertility that a question on children ever born features on the census every five years.

Census data on children ever born is the only source of information on the fertility patterns of a range of small population groups including occupations down to the 4-digit code level, second generation migrant groups, ancestry groups, language groups, and religious groups (all also down to the 4 digit level). Data on these other types of small population groups is generally collected by the census. The ABS TableBuilder facility also allows researchers to cross-tabulate such variables. Patterns of preference between male and female children are apparent in some overseas countries, and for Australia there is evidence of a preference for ’one of each’; couples with, for example, two children of the same sex are more likely to have a third child than those with one boy and one girl. Census data on numbers of sons and numbers of daughters would facilitate the extension of research to consider patterns for migrant, second generation, ancestry and religion groups which form small percentages of the total population and for small geographic areas.
Census data on numbers of children can also bolster the understanding of other behaviours. For example, some of my research links some of differences in women’s employment and income patterns between country of birth groups and between religious groups to their differing numbers of children ever born (De Alwis and Parr under review).
Census data on children ever born for small geographic areas, such as SA2s, can also enhance the understanding of fertility trends and the preparation of fertility assumptions for population projections for these areas (Parr 2014, 2018) and understanding of such geographic patterns of disadvantage is restricted by our inability to consider underemployment. The provision of census data for Commonwealth and State electorates may enhance the understanding of local representatives and candidates of the issues faced by the local community.
The value of census data on children ever born is enhanced though the ready combination of such data with data on age, sex, marital status, birthplace, ancestry, language, religion, labour force status, occupation, and income by the census. It will be enhanced even further by the linkage of data from successive censuses and between census and birth registration.

A question on number of children ever born has been asked successfully administered on a number of previous censuses, including the last three censuses. This shows it can be readily understood by respondents.
The wording used has been:
“For each female, how many babies has she ever given birth to?”
The question could be additionally applied to men by revising the wording to
“How many have you ever given birth to/fathered?”
Whichever form of the question is preferred the additional instructions “Include live births only” and “Exclude foster, step and adopted children” should be included.
All women should know how many children they have had. Among men there may be some misattributed paternity. However, according to Gilding (2005) in Western societies this is around 1 per cent and no more than 3 per cent.
The question requires provision of a number or the marking of a box for “none”, and thus is quick and easy to answer.

Questions on children ever born/fathered have been administered successfully in numerous censuses and surveys. For the overwhelming majority such questions are not intrusive, offensive or controversial. It is possible that memories of children who have died or who were still born may cause some sadness for a small minority of respondents.
Almost all women should know how many children they have had. Among men there may be some misattributed paternity. However, according to Gilding (2005) in Western societies this is around 1 per cent and no more than 3 per cent.
Questions on number of sons and numbers of daughters should be similarly acceptable.
There are no other consideration for specific groups.

The answers will be in integer values, mostly below 6 and almost all below 20. Thus it will neither present major coding difficulties nor will it require extensive processing. It will not significantly add to the cost of the census. The instruction needed (see previous section) is brief and straightforward. The data on the number of children will require only one number to be recorded for each person aged over 15. The extension to a number of sons and a number of daughters would extend the data collection by only two simply-coded numbers.

A question on children ever born has been asked on the 2006, 2011 and 2016 censuses. Time series allow the change in completed family size (i.e. for age 45-49) over time to be tracked. This is a critically important aspect for fertility patterns. The availability of data on successive five-yearly censuses not only allows changes to numbers of children at other ages to be tracked but also and even more crucially it allows changes in numbers of children within cohorts over time to be identified. Such age, period and cohort time series are vitally important to understanding fertility change. Time series on numbers of children ever born among migrants are important for understanding fertility trends over time and trends in the composition of recent immigrant arrivals and overseas-born populations over time. Linked data between successive censuses will vastly enhance the value of census data on children ever born by allowing intercensal propensities for change in numbers of children to be estimated.
Fertility will remain of fundamentally importance to most Australian families in the future. It will inevitably be an integral to determining future population and hence State and local government funding and budgets, electoral representation and boundaries, and resource allocation by a wide range of public sector, private sector and not-for-profit sector organsiations.

There is no nationwide, comprehensive collection of data on numbers of children ever born.
Some sample surveys e.g HILDA do collect data on children ever born. However the sample sizes of such surveys are not large enough to allow robust estimates for large numbers of small geographical areas or small population subgroups to be estimated.
Data on children ever born cannot be produced through integration of existing data sources.

References
Allen, L. and Parr N. (2017) FactCheck Q&A: the facts on birth rates for Muslim couples and non-Muslim couples in Australia. (2017) The Conversation. 24 July 2017.
Gilding, M. (2005) Rampant misattributed paternity: the creation of an urban myth. People and Place. 13(2): 1-11.
Gray, E. and Evans, A. (2005) Parity progression in Australia: What role does sex of existing children play? Australian Journal of Social Issues 40(4):505-520.
Kippen, R., Evans, A. and Gray, E. (2007) Parental preference for sons and daughters in a Western Industrial setting: evidence and implications. Journal of Biosocial Science .39(4): 583-597
Parr, N. (2010) Childlessness among men in Australia. Population Research and Policy Review. 29(3): 319-338.
Parr, N. (2014) Fertility levels and intentions in New South Wales. Expert Report Commissioned by the NSW Department of Planning and Environment.
Parr, N. (2018) Spatial Patterns of Fertility in New South Wales. Expert Report Commissioned by the NSW Department of Planning and Environment.
Parr, N. and Guest, R. (2011) The Contribution of Increases in Family Benefits to Australia’s Early 21st Century Fertility Increase: An Empirical Analysis. Demographic Research. 25:6: 214-244.
Thomson, E., Lappegard, T., Carlson, M., Evans, A. and Gray, E. (2014) Childbearing across partnerships in Australia, the United States, Norway and Sweden. Demography 51(2): 485-508.