Parity specific birth probabilities by duration of marriage: a method of estimation and its application to Kenya

Sarma RSS
Cairo, Egypt, Cairo Demographic Centre, 1985. 26 p. (CDC Working Paper no. 11)

Conditional probabilities of birth for women with a given number of children ever born were estimated by hypothesizing that in a homogeneous population, the occurrence of birth follows a Poisson process, and that the heterogeneity may be described by a Gamma distribution so that the process becomes a mixed Poisson. The distribution of women by number of children ever born reduces to a Negative Binomial. The conditional probability of a birth, given the number of children already born, turns out to be a linear function of the number of children born, involving the estimates of the Negative Binomial parameters and the length of exposure to reproduction. An attempt was made to apply the hypothesized model to data from Kenya. Since there is likely to be understatement of children ever born, which causes underestimation for the mean and even more so for the variance, the reported parity distributions are adjusted by means of appropriate choice of Negative Binomial approximations. The resulting parity-specific annual birth probabilities at different durations of marriage appear reasonable from what is known about reproduction levels of Kenyan women. There is lack of precision regarding the estimation of the length of exposure period in the case of Kenya, but in situations where duration of marriage measures the length of exposure to reproduction adequately, the application of the technique would be straightforward. (author's modified)

Region / Country: 
Document Number: 
Add to my documents. Add to My Documents