Interval-Censored Event History Analysis

Résumé 0

Event history analysis was no doubt already in existence when J. Hajnal used census information on marital status to investigate cohort nuptiality or when L. Henry studied cohort fertility from numbers of children ever born. It developed when other elements added to this information: age at first marriage, dates of birth of children; in this respect, the 1946 census in Britain was a turning-point. It has since branched out in two directions: on the one hand, surveys collecting retrospective information on respondent's life course, family and occupational histories, residential mobility...; on the other, individual biographies compiled by extracting information from administrative sources (vital registration data, census schedules, notifications of residence...). Each method has constraints--the second, in particular, because of depending on the availability of administrative data: job changes are never registered and in France, neither are residential moves. That limits observation to the information collected when a census is held or a vital event registered. Daniel Courgeau and Jamal Najim test here the validity of these incomplete data by comparing them to information supplied by an ad hoc survey.