Problems concerning the parametric analysis of the age at first birth

The application of parametric split models to analyse the birth of the first child is discussed by applying the model of Coale and McNeil and the log-logistic model. We show that serious problems of estimating the final survival probability may occur when the empirical age distribution of the analysed event is not fully known and the model deviates considerably from the empirical distribution. We suggest strategies to handle these problems in a pragmatic way.