Searching for More Effective Variables to Use in a Household Survey’s Nonresponse Adjustment Process

• Activity Number: 556 - Revisions
• Type: Contributed
• Date/Time: Thursday, August 11, 2022 : 10:30 AM to 12:20 PM
• Sponsor: Government Statistics Section
• Author(s): Lauren Vermeer* and Sharon Krieger and David Swanson
• Companies: Bureau of Labor Statistics and Bureau of Labor Statistics and Bureau of Labor Statistics
• Keywords: non-response; non-response bias; consumer expenditure ; effective variables; R^2

Abstract

Like most household surveys, the Consumer Expenditure Survey’s response rate is decreasing. Also like most household surveys, it has a risk of bias, including nonresponse bias. The Consumer Expenditure Survey has a nonresponse adjustment procedure designed to remove the bias generated by nonresponding households from its data. However, as response rates decrease, any imperfection in the survey’s nonresponse adjustment procedure increases, illustrating the importance of keeping the procedure current and accurate. The Consumer Expenditure Survey uses the traditional cell adjustment method for its nonresponse adjustments, and in this paper we describe a technique we recently employed to systematically search for a more effective set of cell-defining variables.