Regression-like Analysis of Birth Interval Sequences

Abstract

Employing a family of generalized linear models for the gamma distribution, the structure of birth interval data derived from three historical populations is examined. The relative importance of such factors as current age, parity of the mother, and duration of marriage in determining the length of succeeding intervals is investigated. The results suggest that the three data sets have a common, rather simple structure and various applications of such empirical models are considered.