Risk families and the unequal distribution of deaths in France and Sweden during the 19th century

Catalina Torres, Max-Planck Odense Center on the Biodemography of Aging and University of Southern Denmark
James E. Oeppen, Max-Planck Odense Center on the Biodemography of Aging
Rune Lindahl-Jacobsen, Max-Planck Odense Center on the Biodemography of Aging
James W. Vaupel, Max-Planck Institute for Demographic Research and Max-Planck Odense Center on the Biodemography of Aging

In the past, high levels of infant and child mortality were the main responsibles of low levels of life expectancy at birth. During the last decades, several studies have highlighted the fact that in some historical as wells as in some contemporary populations, infant and child deaths are concentrated in a reduced number of families (e.g. Das Gupta 1990; Zaba and David 1996; Willführ and Gagnon 2012). This phenomenon, known as ‘death clustering’, implies that the risks of experiencing infant and child deaths are unequally distributed among the families in a population. This finding stresses the need of shifting the attention from single individuals to families as the appropriate units of analysis in the study of infant and child mortality (Edvinsson and Janssens 2012). In the present study, we use aggregate mortality data as well as two reliable sources of historical microdata from Sweden and France in order to calculate comparable measures of the concentration of infant and child deaths in different subpopulations, and to examine the impact of the factors affecting the unequal distribution of those deaths. The measures of death clustering in different subpopulations will be illustrated with Lorenz curves, where the observed distributions of deaths will be compared with the corresponding expected binomial distributions. The impact of several factors affecting the distribution of infant and child deaths within families will be analysed with logistic regression, with the mothers as the units of analysis. We expect to find higher levels of death clustering in populations with high mortality. Furthermore, we expect the distribution of deaths to be determined by differences in various biological as well as social characteristics of the mother, in particular the length of the birth interval, the total number of births, and the parental care.