Nonresponse and Recall Biases in a Survey of Traveler Spending

Abstract

This study assesses the extent of nonresponse and recall biases in a travel spending survey of 1,647 individuals. Detailed data from actual nonrespondents were collected. Although the response rate to the mailback survey was relatively high (79%), nonresponse bias in some key variables was detected. Additional findings cast doubt on the use of either wave analysis or late respondent comparisons as proxies for data from actual nonrespondents. Recall bias was also observed but was found to be related as much to trip complexity as to the passage of time. Findings support procedures that obtain complete responses either during or immediately upon the completion of a respondent's trip.