Using Call Records to Study Sources of Nonresponse Bias in Consumer Expenditure Surveys

John S. Dixon and Eric Figueroa

Abstract

Respondent characteristics are known to relate to nonresponse. The record of attempted contacts can shed light on the interview process. The combined relationship between call records and respondent characteristics can be used to examine nonresponse and the impact of nonresponse on estimates. The estimates of consumer expenditure categories for respondents and nonrespondents will be compared using longitudinal data from the Consumer Expenditure Surveys to detect the magnitude and direction of bias. The respondents which were nonrespondents in previous or subsequent panels will be used to provide the estimates of bias using methods related to "gross flows". Call records and household characteristics will be used to suggest some patterns to alleviate nonresponse and to investigate potential bias.