An introduction to multivariate life table analysis of birth intervals.

Abstract

The availability through cross-sectional surveys of individual-level data on birth histories has led to a view of fertility as a sequential process in which the rate of transition from 1 parity to the next is determined by parity-specific factors. The contribution of factors such as age or contraceptive usage can be assessed more accurately by studying birth intervals; however, this approach is subject to the methodological problems of reporting errors and selectivity. To overcome these difficulties, demographers rely on multivariate life tables. In the hazard model, covariates are used to determine the likelihood that a woman exposed to the risk of having a birth of a given order will do so in a certain month after initiating the interval. The 2nd model, which considers time in its discrete rather than continuous form, uses a log linear model to describe the relationship between the proportion of women having a birth within a specific duration segment and the woman's characteristics. The choice of life table method depends both on the type of available data and the availability of suitable computer software. Although this paper focuses on the application of life table analysis to birth intervals, this approach is generalizable to the study of contraceptive efficacy, infant mortality, marriage dissolution, and other demographic phenomena.