Estimating the Total Fertility Rate with alternative data sources

CPC and CG members, led by Dr Joanne Ellison, have highlighted the importance of accurately estimating fertility rates to make appropriate comparisons of fertility levels across time and space, and to inform fertility projections.

The Total Fertility Rate (TFR) is an indicator of the fertility level in a particular country or region, and measures the average number of children a woman would have if she were to experience the age-specific fertility rates of a given period throughout her reproductive career.

The study team, from the universities of Southampton and St Andrews, compared estimates of the 2011 TFR for all UK-born and non-UK-born women in England and Wales, and Scotland using three data sources: vital registration data, longitudinal studies (linked census and vital events data), and census household microdata samples.

In the UK, official estimates of the TFR use vital registration data and the census. Sample surveys tend to be used to compute the TFR when vital registration data is not sufficiently complete, such as in some less-developed countries. However, survey data is also useful because it contains additional characteristics, such as education or ethnicity, and may allow individuals to be traced over time. It is of interest, then, to compare TFR estimates from these sources with the estimates from vital registration data for the purposes of validation, calibration and assessing representativeness.

The key findings from the study show that:

TFR estimates using census microdata tend to be closest to those from vital registration data, especially for Scotland. For England and Wales, the census estimates are lower than those from vital registration data, especially for non-UK-born women.
The longitudinal study estimates are the lowest among the three data sources for Scotland, while for England and Wales, they are lower or higher than the corresponding vital registration estimate, with this generally depending on the precise estimation method used.

Overall, the study finds some small variation in the total fertility rate estimates from these different sources, owing to their contrasting coverage, mode of collection, and sample size.
The reasonable consistency of the census-linked data and the census household microdata with vital registration estimates show that they are an important source of information that allows the examination of subgroup differences in childbearing behaviour.

Dr Ellison commented: "This study highlights the importance of comparing estimates of fertility measures across a range of data sources to get a handle on the associated uncertainty.

"The fact that the census-linked data has proved reasonably consistent with vital registration data, and given the ability to obtain methodologically sound estimates as well as the availability of 13 other census variables such as education and ethnicity, the longitudinal studies are an important data source for fertility analyses."

Read CPC-CG Working Paper 106: 'Estimating the 2011 total fertility rate for England & Wales and Scotland using alternative data sources' to find out more.

Posted 26/01/2024 11:28