Mismeasured variables in econometric analysis: problems from the right and problems from the left.

Part of a special section on econometric tools. Random measurement errors, commonly known as “classical” measurement errors, can cause difficulties in econometric and statistical analysis. If such errors occur in the dependent variable and the independent variable is continuous and unbounded, then...

Descripción completa

Detalles Bibliográficos
Publicado en:Journal of Economic Perspectives Vol. 15; no. 4; pp. 57 - 68
Autor principal: Hausman, Jerry
Formato: Artículo
Publicado: American Economic Association Fall 2001
Materias:
Acceso en línea:Ver este registro en EBSCOhost
Descripción
Sumario:Part of a special section on econometric tools. Random measurement errors, commonly known as “classical” measurement errors, can cause difficulties in econometric and statistical analysis. If such errors occur in the dependent variable and the independent variable is continuous and unbounded, then the measurement errors have a relatively benign effect on ordinary least squares estimates. However, classical measurement error in an explanatory variable will cause bias because a component of the error term is negatively correlated with the observed explanatory variable. Three recent developments relating to the effects of mismeasurement on econometric models are discussed.