Article Transportation Research Record 2018, Vol. 2672(42) 33–44 Ó National Academy of Sciences: Transportation Research Board 2018 Article reuse guidelines: sagepub.com/journals-permissions DOI: 10.1177/0361198118790376 journals.sagepub.com/home/trr Factors Affecting Interview Duration in Web-Based Travel Surveys Pierre-Le ´o Bourbonnais 1 and Catherine Morency 1 Abstract Historically, travel surveys have been conducted face-to-face, by mail, or by phone. With the increasing share of households having access to the Internet, other survey modes have been deployed. This paper focuses on web surveys. Among other advantages, using the web to conduct surveys reduces costs and helps mitigate poor response rates among young house- holds. Very few studies have been conducted on interview duration and its determinant using paradata from web travel sur- veys. Such knowledge is necessary to validate the context in which travel data are gathered and can be used to understand sample and data quality. Interview duration modeling is also essential for allocating survey servers and monitoring interviews during the data collection phase. This paper models interview duration using paradata from nine web surveys conducted in the Quebec province from 2010 to 2014. The main objectives of the model are to assist the monitoring of interviews by detecting outliers, provide a better estimate of the interview duration to respondents and survey managers during the inter- view, and allow a more precise evaluation of the server performance needed before conducting web travel surveys. Using a multiple regression model, we observed that the most important variables in explaining interview duration were number of car and transit trips as well as number of unique places visited during a day. Conducting the interview on a small-screen device also increased interview duration. The model also provides a baseline estimate of interview duration on the basis of demographic features and questionnaire design. Three main media were traditionally used to conduct origin–destination travel surveys: telephone, mail, and face-to-face. With the increasing sample sizes involved in large-scale and regional surveys, computer-assisted tele- phone interviews (CATI) were considered the best tool available at the time to both reduce costs and increase data quality. Since the beginning of the last decade, how- ever, web questionnaires have been suggested as a way to expand the reach of conventional CATI surveys. Web- based travel surveys were a welcome addition to the transportation analyst’s toolbox, mostly to cope with problems associated with human-resource-intensive CATI surveys, but also to provide a tool based on a medium that an increasing share of the population uses in its daily activities. Declining response rates and diffi- culty constructing a representative sampling frame using only landline phone numbers are now the main draw- backs of conventional telephone-based surveys. According to the Residential Telephone Service Survey (1), 60% of young households 2 18 to 34 years old – use only cell phones, the numbers for which are not usually found in public directories, and do not have a landline phone at home. Although the web does not help in itself to get good and representative samples, it may attract respondents who are not willing to participate in tele- phone surveys, or that are not reachable using conven- tional survey modes or during typical periods in which phone interviews are conducted. Furthermore, for spe- cific types of trip generators, web-based questionnaires are easier to administer and less expensive in the long run. In the last 6 years, web-based surveys have been deployed in the Quebec province to gather travel data from people and households. This paper uses paradata from nine web-based surveys to analyze and model inter- view duration. The paper proposes various descriptive statistics regarding outcome rates and interview duration as well as two interview duration models, one for people- based surveys and the other one for household-based surveys. 1 Department of Civil, Geological and Mining Engineering, Polytechnique Montreal, Montreal, Canada Corresponding Author: Address correspondence to Pierre-Le ´o Bourbonnais: cmorency@polymtl.ca