Non-resident parents: Why are they hard to capture in surveys and what can we do about it?

Caroline Bryson and Stephen McKay

Published 25 October 2018

The under-representation of non-resident parents in surveys has long hindered research on family separation, leaving key evidence gaps for those making policy and practice decisions related to separating and separated families, including (but not restricted to) issues around child support, child arrangements, welfare benefits and housing. In this paper, we articulate the importance of robust quantitative data collected directly from non-resident parents. We review the methods previously employed to attempt to achieve this, and we use the UK Household Longitudinal Study (UKHLS) (University of Essex, ISER, 2017) to demonstrate where and how response biases occur. The main body of the paper reports findings from an experiment run on Wave 10 of the UKHLS Innovation Panel (Al Baghal et al., 2018; University of Essex, ISER, 2018) in which we compare two approaches to identifying non-resident parents from among the panel members. One method, a variant of that currently used in the UKHLS, asks panel members about living relatives with whom they do not live.

The second method modifies the UKHLS standard fertility history questions collecting information on past births and then asks whether any such children are under 18 and living outside the household. Our findings are necessarily tentative, with around 100 non-resident parents identified across both arms of the experiment from among the 2,570 panel members interviewed in Wave 10. They nonetheless point towards a potential to improve the survey representativeness of non-resident parents, at least to some degree. While we found no statistically significant differences in the non-resident parent prevalence rates between the two methods, in combination they increased the non-resident parent sample by one quarter. Moreover, the data suggest that the fertility history approach improves the representativeness of the non-resident parent sample, in terms of both their socio-demographic profile and their levels of parental involvement. That said, even the combined approach results in a large underrepresentation of non-resident parents and a continued bias towards those who are more involved with their children.