Income Measurement Error in Surveys: A Review
Jeffrey C. Moore, Linda L. Stinson, and Edward J. Welniak, Jr.
Because income data are germane to a wide array of important policy issues, income questions are almost ubiquitous in government-sponsored surveys. We review research on the quality of survey measures of income, with a particular focus on U.S. government surveys. We briefly examine two of the more typical quality indicators -- "benchmark' comparisons of survey estimates to independent aggregate estimates, and nonresponse -- but focus our attention „primarily on response error research which compares individual survey respondents' reports to external measures of truth, often obtained from independent record systems. The latter investigation reveals a wide range of error properties across income characteristics (sources, amounts received) and income types, which includes high levels of both random error and bias’in some instances. We also examine the recent findings of "cognitive' research into respondents' understanding of the meaning of income questions, their interpretations of the tasks which income questions present, their motivations, etc., and attempt to link what we know about income reporting errors to these cognitive processes.
Data quality; response error; response bias; random error; income source reports; income amount reports; cognitive research.