Event-history analysis for left-truncated data

Abstract

"This paper aims at providing a practical guidance for coping with social science event-history data that are left-truncated, especially when the length of exposure prior to observation is known. Only the case of single events is treated, although much of the discussion should be applicable to the case of repeated events, in which only the first spell is likely to be left-truncated. Marital dissolution data from the Panel Study of Income Dynamics (PSID) are used as an example...."