Bias in Estimates of Child Obesity from Panel Surveys

Michael S. Rendall, University of Maryland
Peter Brownell, RAND Corporation
Margaret M. Weden, RAND Corporation
Christopher Lau, RAND Corporation
Meena Fernandes, Abt Associates
Zafar Nazarov, Cornell University

Parental reporting of height and weight was evaluated for US children ages 2-13. Obesity prevalence (body-mass-index =95th percentile) and its height and weight components were compared in child supplements of two nationally-representative surveys (the National Longitudinal Survey of Youth, 1996-2008, and the Panel Study of Income Dynamics, 1997-2002). Standardization/decomposition methods were adapted to quantify overall bias in obesity prevalence relative to measured data from the National Health and Nutrition Examination Survey, and to examine the contributions to that bias arising from error in reporting extreme weight versus extreme height values. Parental under-reporting of both height and weight was found. Much greater bias in obesity prevalence estimates, however, arose from height under-reporting. Under-reporting of height decreased with age and under-reporting of weight increased with age. Based on our findings, we offer practical solutions for survey data collection and research on child obesity using parent-reported data.