Comparison of Respondent's profile of Computer-Aided Web Interviewing and Computer-assisted personal interviewing: The case of Ile-de-France mobility Surveys

For the current Global Travel Survey for the Paris region we have a very rich sampling database in order to draw the sample. Indeed, our sampling database is from French national Institute of statistic and made with different administrative data such as: tax data (with demographic information on individuals and household structure and household incomes) and additional contextual data to enrich the tax information with variables to better describe the addresses where people live. We took advantage of this opportunity to test the administration of this survey by two media first a Computer-assisted personal interviewing (CAPI) and second a Computer-Aided Web Interviewing (CAWI) in order to study the effect of both media on the response mechanism. For both media we compute a logit model to model the response mechanism and found that: - Household income, the size of the housing, the age and gender of the reference person, the housing occupancy status and the type of household play a role in the response mechanism for the face to face survey; - Household income, the size of the housing, the age of the reference person, the housing occupancy status and the type of household; the response day and the zone of residence play a role in the response mechanism for the web survey; Based on these results, we have tested different calibration to reweight the respondent samples and tests the significativity of the estimators. The main purpose the paper will be a discussion on the responses mechanism for each media (who are the hard to reach for each media).