Retrospective Versus Panel Data in Analyzing Lifecycle Events

This paper compares lifecycle data from a retrospective marital history with those derived for the same individuals from panel information, utilizing data from the Young Women's cohort of the National Longitudinal Survey of Work Experience (NLS) which was initiated in 1968. The results indicate that when a marital event is reported in both sources there is substantial agreement about the date of the event. The errors are systematic and seem to relate to factors that increase the difficulty of recall in retrospective histories. The two data sources seem to do equally well in estimating hazard rate models of first marriage and give qualitatively similar results for hazard rate models of divorce and remarriage.