Impact of interviewing by proxy in travel survey conducted by telephone

Abstract

Telephone-interview surveys are a very efficient way of conducting large-scale travel surveys. Recent advancements in computer technology have made it possible to improve upon the quality of data collected by telephone surveys through computerization of the entire sample-control process, and through the direct recording of the collected data into a computer. Notwithstanding these technological advancements, potential sources of bias still exist, including the reliance on an audit member of the household to report the travel information of other household members. Travel data collected in a recent telephone interview survey in the Toronto region is used to examine this issue. The statistical tool used in the research was the Analysis of Variance (ANOVA) technique as implemented within the general linear model framework in SAS. The study-results indicate that reliance on informants to provide travel information for non-informant members of their respective households led to the underreporting of some categories of trips. These underreported trip categories were primarily segments of home-based discretionary trips, and non home-based trips. Since these latter two categories of trips are made primarily outside the morning peak period, estimated factors to adjust for their underreporting were time-period sensitive. Further, the number of vehicles available to the household, gender, and driver license status respectively were also found to be strongly associated with the underreporting of trips and thus were important considerations in the determination of adjustment factors. Work and school trips were found not to be underreported, a not surprising result giving the almost daily repetitiveness of trips made for these purposes and hence the ability of the informant to provide relatively more precise information on them.