In the conduct of sample surveys in transport, there will be some level of nonresponse. The issues of nonresponse are fundamentally connected to the questions of reducing survey bias and increasing the accuracy of sample estimates. Nonrespondents in sample surveys often have been shown to have characteristics significantly different from those of respondents. These differences are in sociodemographic characteristics and, more important, travel behavior characteristics. Issues associated with nonresponse to mail-back travel surveys are presented. Previous research into nonresponse in mail-back travel surveys is reviewed, in particular, methods for correcting for nonresponse. A number of alternative mechanisms for the often-observed characteristic of lower trip rates in later response waves to a mail-back survey with multiple reminders are presented. By using data from 3 years of a continuous travel survey in Melbourne, Australia, two of the mechanisms, sociodemographic effects and underreporting of trips, were extensively tested and found not to be the cause of the effect. In comparison of the remaining two mechanisms, lower actual trip rates by nonrespondents and self-selection of travel day by late respondents, a preference is expressed for the self-selection mechanism, and this is based on the results of a survey of nonrespondents and on the outcome of applying correction factors based on the two mechanisms. These results suggest that previous methods for correcting for nonresponse in mail-back travel surveys may be in error. Suggestions are made for steps necessary to fully understand the behavioral mechanisms underlying nonresponse in mail-back travel surveys.
Abstract