1 The reconstruction of fertility trends with DHS birth histories Application of Poisson regression to person-period data

Context and objective The reconstitution of birth histories—or maternity histories—is a widely used approach for collecting data on fertility in developing countries. Since the 1970s with the World Fertility Survey (WFS), and even more so since the mid-1980s with the program of Demographic and Health Surveys (DHS), birth histories have become an indispensable source of data for studying fertility levels, trends, and determinants. The principle is well-known: a sample of women are asked about their reproductive history, and the birth dates of each of their children, from the first birth until the time of the survey, are recorded. These birth histories are mainly used to calculate the classic indicators of fertility, in particular fertility rates and total fertility rates, and to reconstitute fertility trends over the past ten to fifteen years. When combined with socio-economic data collected by fertility surveys, they can also be used for explicative analyses of fertility behaviour. Finally, although these data are most often used to study recent fertility, occasionally researchers carry out explicative analyses that exploit their longitudinal nature. In this paper, we present a simple method that can be used for a variety of analyses of fertilit