Household Income Reporting: An Analysis of U.S. Consumer Expenditure Survey Data
Thesia I. Garner and Laura A. Blanciforti
When income data are collected from household survey respondents, incomplete responses and nonresponses frequently result. The purpose of this study is to examine the relationship between particular socioeconomic attributes of households and the probability that a household will be a complete income reporter. Data from the Interview portion of the 1987 U.S. Consumer Expenditure Survey (CEX) are analyzed. For this survey, the household does not need to provide a complete accounting of income from all sources to be considered a complete income reporter. Binomial logit analysis is used to model the probability of income response completeness. Unique to this study, in contrast to other studies of income response, is the inclusion of expenditures as an explanatory variable in the estimating equation. Accounting for expenditures is particularly important for CEX data users since published tabulations of expenditures by income are restricted to complete income reporters only. Results from examining only the primary effects reveal that the age, race, education, and occupation of the reference person, and the consumer unit's composition, region of residence, expenditures, and participation in income means tested programs are significantly related to the income reporting status of the unit. This information should be useful for data users not familiar with the implications of using the U.S. CEX#39;s definition of complete income reporter, for statisticians developing income imputation procedures for the survey, and for data collectors devising collection procedures to improve data quality. Specific suggestions for future research directions to further explain issues related to income response are provided.
Incomplete income response; household survey data; socioeconomic characteristics; expenditures; binomial logit analysis.