Why fertility is higher in reproductive health surveys?

Abstract

Two main data sources exist for the study of fertility in Mongolia: the vital registration (VR) system and the nationally representative sample surveys (Reproductive Health Survey (RHS)). Yet, a puzzling issue that has remained neglected so far, but deserved further analysis, is the major differences of 0.5 to 1 child per woman in the total fertility rate that these two sources report. In an attempt to reconcile the fertility levels given by the two main sources of fertility estimates in Mongolia, I investigate the possibility that information on fertility collected in RHS surveys produce fertility estimates that are too high. The main results con rm that the discrepanciesbetween fertility estimates in Mongolia stem from selection bias in RHS samples. Once adjusted,survey-based fertility estimates are very close to the VR-based gures. The paper shows therisks of taking survey-based demographic indicators as the ‘Gold Standard’ and draw attention tothe fact that demographers will always need to evaluate critically demographic gures.