Distortions caused by nonobservation of periods of cohabitation before the latest

In retrospective studies of individual life histories, some or all of the information pertaining to periods prior to the latest event interval of interest will sometimes be left out. This simplifies data collection, and it may improve the reliability of the data actually obtained by reducing misreporting. Unfortunately, it may also introduce serious distortion in estimates based on such data. This paper investigates the extent of such distortion in a retrospective study of nonmarital cohabitation, marriage, and childbearing when information on consensual unions is restricted.