Reconciling cross-sectional and longitudinal measures of fertility, Quebec 1890-1900.

In the absence of vital registration, studies of the onset and early phases of the fertility transition in North America have been seriously hampered and yet the seemingly early timing of the decline, the multi-ethnic nature of the population and continuous flow of immigrants from Europe suggest that North America has much to offer to this debate. This paper is primarily methodological drawing on parallel data for the city of Montreal and surrounding region. By reconciling cross-sectional census measures of fertility using the own-child methods (1901) with those derived from a longitudinal ten-year panel (1891-1901) using family reconstitution, it exposes some of the weaknesses and the potentials of the two methods most often currently used and the advantages of combining methods. Own-children measures of marital fertility are seriously affected by significant local differences in infant survival between rural and urban areas and between cultural groups as well as by residual effects of duration and timing of marriage, while small-scale longitudinal studies in complex environments cannot always render reliable results for all sub-populations not can they necessarily be 'scaled up.' They suggest that national and even regional averages of fertility may conceal large diversity, which in turn raises questions about the existence of any single transition with uniform characteristics and timing, or universal cause. Instead we argue different groups in different environments may actually have been fine-tuning their fertility behaviour to compensate for the differential effects of mortality through adjustments to both marriage and fertility within marriage.