Estimating fertility rates using the ONS Longitudinal Study--what difference does the inclusion of non-continually resident members make?

Among datasets available for fertility research in England and Wales, the Office for National Statistics (ONS) Longitudinal Study (LS) is unique in its construction and scale. The large number of individuals who are part of the study means that it is an important dataset for estimating fertility trends in England and Wales by age and parity. This article uses the LS to estimate age-specific fertility rates (ASFRs) for England and Wales between 1991 and 2001. This necessitates great care to ensure that the exposure to risk of birth for female LS members is fully understood. To achieve this, two forms of residential history are defined - consistent cases where the residential information for the LS member is potentially complete for the whole decade and inconsistent cases where there is certainly some form of incompleteness in the residence information. By considering 'all consistent cases', that is both the continuously resident plus those who are not continuously resident but appear to have a complete residential history, we obtain ASFRs which are slightly lower than official statistics figures, but closer to these official figures than ASFRs produced when restricting the sample to LS members who remain continuously resident between 1991 and 2001. The fertility of those consistent cases who are not continuously resident is substantially higher than the rates of continuously resident cases.