Reliability of union histories in social science surveys: Blurred memory, deliberate misreporting, or true tales?

This paper examines the reliability of biographical information gathered retrospectively. It draws on data from the German Family Panel (pairfam), which collected information on the partnership status at first birth using two different strategies. The first strategy was to collect partnership and fertility histories separately in an event history calendar. The second strategy was to ask the respondents directly about their partnership status at first childbirth. We find that in almost 20 percent of the cases, the information collected using the two different strategies did not correspond. The dissolution of a partnership and having a complex partnership biography are strong predictors for discrepancies in the information gathered through the two different strategies. We conclude by discussing the factors that lead to the different outcomes produced by each of the two methods, and the implications of these discrepancies for the study of partnership and fertility behavior in general.