A note on synthethic cohort estimates of average desired family size

Abstract This note considers the problem of estimating average desired family size from parity-specific data on proportions wanting another child. Three procedures proposed in the literature are reviewed and shown to give seriously biased estimates, when women implement their fertility preferences. An alternative procedure is proposed which overcomes most of the deficiencies of previous methods. An example from Sri Lanka illustrates the fact that the differences among the methods may be quite substantial.