Measurement Error in Panel Data

Abstract

Measurement error is a pervasive problem in economics and other social and behavioral sciences. Estimators that ignore the issue are generally inconsistent, and more information or stronger assumptions are necessary to correct for measurement error. Panel data are a source of more information, but consistent estimators still require some additional information or assumptions. This chapter gives an overview of the measurement error problem, with an emphasis on the linear regression model for panel data, and shows various ways in which panel data can be helpful to obtain consistent estimators. More advanced topics, such as non linear models and nonclassical measurement error are also discussed.