BIAS IN AGE AT MARRIAGE IN FAMILY RECONSTITUTIONS - EVIDENCE FROM FRENCH-CANADIAN DATA

Age-at-marriage estimates from family reconstitutions may be biased do wnward when they are based only on marriages of people who continue to live in their parish of birth, because when the probability of migrat ing rises with age, younger people are selected in preference to older ones. Micro-simulations show that the bias can have dramatic effects. In this paper French-Canadian data are used to investigate the import ance of the bias and to verify empirically the micro-simulation result s. Although a high proportion of people moved between birth and marria ge, the bias had virtually no effect, given the specific characteristi cs of the migrations. If one cannot avoid discussing the timing of mig ration before marriage, when measuring age at first marriage using onl y data on ''stayers'', it is just possible that in most settings, it i s the same for those who lived in their parish of birth, and those who had moved.