Non-coverage Errors in Travel Surveys Due to Mobile Phone Only Households

摘要

National and regional household travel surveys have conventionallysampled landline telephone households through list-assisted random digit dialing.However, a recent increase in “mobile phone only” households results in either non7coverage or under-coverage of a growing segment of the population. This canpotentially cause a substantial bias in the representativeness of travel behaviortoward the target population. To cover mobile phone only households, an address10based sampling method is of interest. This study explores whether the characteristicsand travel behavior of mobile phone only households differ from those ofhouseholds with landline telephones. In addition, this study quantifies the extent ofnon-coverage errors in the surveys in terms of the respondents’ travel behavior.Along with census data, the mobile phone only sample (N=2,988) was comparedwith the landline telephone sample (N=7,774) drawn from the 2008 National CapitolRegion Household Travel Survey. Results show that the mobile phone only sampleconsists of relatively more single-person households, younger individuals, andBlacks/Asians/Hispanics, which are generally identified as hard-to-reach groups.Statistical models were developed to examine differences in travel behavior (e.g.,trip-making), suggesting that the mobile phone only households make more transit(27%) and walking (18%) trips. This is partly due to the spatial distribution of theresidential locations between the two groups, which are found to be statisticallysignificant. Regarding non-coverage errors, results show that the inclusion of themobile phone only households can reduce the errors, especially for transit andwalking travel behaviors. The implications for travel survey methods are furtherdiscussed.