Synthetic Cohorts from Panel Surveys

Abstract

Studies of the determinants and effects of life events such as widowhood and retirement suffer from the fact that the events, despite their common occurrence in the lives of individuals are, from the researcher's perspective, relatively rare. As a result, research designs require inevitable compromises. Many of these compromises lead to biases of one kind or another. One possible means of avoiding at least some of the biases is to form synthetic cohorts of individuals who have experienced events by pooling observations from panel surveys. This approach yields a multiwave data set centered on the event of interest with fixed amounts of pre- and postevent data. The data can be analyzed using a variety of techniques and the results can be interpreted in light of the interrupted times series design.