Nonresponse Bias in a Travel Survey of Nontelephone Households*

The Bureau of Transportation Statistics (BTS) co-sponsored two national random digit dial (RDD) telephone surveys during the summer of 2002. Stakeholders expressed concern about the potential bias in travel-related behavior estimates because the sample designs of these surveys excluded the four percent of nontelephone (non land line) households. Nontelephone households may have different travel-related behaviors compared to telephone households. While there have been other studies of nontelephone households, they have not focused on travel-related behaviors. BTS responded to this concern by conducting the Supplemental Survey, a travel-behavior-related nontelephone household survey designed to investigate possible differences. However, the survey only yielded an unweighted CASRO response rate of 21 percent. Because data for both respondents and nonrespondents were available from the sample frame, we conducted a nonresponse bias analysis for this unique survey. One of the purposes of this work was to identify whether nonrespondents in travel surveys are different compared to nonrespondents in other surveys. The analysis includes both bivariate and multivariate analyses to ascertain predictors of nonresponse to the survey. Multivariate analyses indicate that in this nontelephone household mail survey, nonrespondents were more likely to be under 35, male, have less than a high school education, and live in a household with more than one person, children, and no vehicle