The effect of endogeneity and measurement error bias on models of the risk of child stunting

The impact of endogeneity and measurement error on models that estimate the risk of child stunting is demonstrated. Stunting occurs when poor living environments cause short physical stature and is a major health problem in developing countries. The literature modelling the effect of various policies on the risk of stunting suffers from uncertainty about the strength of income versus maternal education effects. Results are based on a household survey from Papua New Guinea (PNG), where repeated within-year observations on households allow calculation of each variable's reliability ratio. Both measurement error and endogeneity bias are shown to affect conclusions about whether raising incomes or maternal education is the best way to reduce the risk of stunting. (C) 2002 IMACS. Published by Elsevier Science B.V. All rights reserved.