Using retrospective surveys for estimating the effects of breastfeeding and childspacing on infant and child mortality

Recent contributions to our knowledge about the effects of breastfeeding and pace of childbearing on early child mortality have originated in the analysis of birth histories elicited from retrospective interviews. The validity of these findings has been questioned on the ground that when the timing of events is systematically distorted, estimates of effects will be affected by serious biases. In addition, it has been argued that the results obtained are highly sensitive to the models used and the techniques of estimation applied to calculate estimates of effects. In this paper we review the most important criticisms levelled against recent findings and interpretations and show that they rest on propositions of questionable validity.