Some methodological considerations in the analysis of current status data

Recent attention has focused on the length of the post partum period of infecundity in those societies where child-spacing has traditionally been achieved through post partum sexual abstinence or prolonged breast-feeding. The tabulation of proportions of women abstaining at interview by the age of the youngest child has been suggested as a technique free of problems of digital preference and memory lapse, while being otherwise equivalent to the distribution of the proportion of women abstaining by completed months since their last confinement, calculated from retrospectively-collected durations of abstinence. Both stochastic and deterministic models demonstrate that the techniques are comparable only when they are applied to non-recurrent events such as first marriage. Childbearing, and the relaxation of the proscription on post partum sexual intercourse, is a recurrent event, and the distribution of proportions of currently abstaining women according to the age of their youngest child is heavily biased. This finding invalidates the use of the post partum current status distribution.