Age at first birth in Canada: a hazards model analysis.

Studies on age at 1st birth and its covariates are few in the demographic literature on Canada, mainly due to the paucity of maternity history data. This is the 1st paper on this aspect of fertility, based on the data collected through the Canadian Fertility Survey in 1984 by Balakrishnan and colleagues. Since the majority of childbearing in Canada is still within marriage, this study focuses on ever-married women in the prime reproductive ages, 18-49, in 1984. To circumvent the problems of selectivity and censoring in the analysis of retrospective maternity history collected in a cross-sectional survey, the authors have chosen Cox's proportional hazards model for multivariate analysis for each of the age cohorts, viz. 18-24, 25-29, 30-34, 35-39, 40-44, and 45-49. As the focus is on the 1st birth, women 35 years of age and over can be viewed as having sufficient time for the 1st birth and thus as different from younger age cohorts. Separate analyses for different age groups is done to achieve greater homogeneity of covariate effects. The results indicate that ever married women with lower ages at marriage, premarital pregnancy, low education, not working, and nonuse of contraception before the 1st pregnancy are the potential early family starters. The significance of covariate effects varied for different age cohorts. Also presented are estimates of relative risks of the 1st birth by various combinations of covariates, ranging from high risk to low risk groups utilizing the baseline survival function obtained from the proportional hazards model.