Intérêt du modèle " Hurdle " pour la comparaison des comportements de mobilité déclarée dans un protocole d'enquête mixte

Relevance "Hurdle" model to compare declared travel behaviour in a mixed-mode survey. - Due to decline in response rate to household travel surveys, it is essential that survey methods research explore the use of mixed-mode surveys. With this in mind, the Laboratoire d'économie des transports (Transport Economics Laboratory) conducted a web-based survey on non-respondents to a face-to-face household travel survey administered in the Lyon urban area in 2005-2006. Comparative analysis of the figures shows that, overall, face-to-face respondents travel more than web-based respondents. This difference is partly explained by a high proportion of individuals who did not travel on the survey reference day and also by a lower trip level for those who travel. Classical regression methods can explain the individual mobility level with individuals socioeconomic characteristics, but do not take account of individuals who did not travel during the survey reference period. Using two econometric methods (simple Poisson and Hurdle Poisson regression) we examined factors affecting the decision to travel on one hand and the level of mobility on the other hand. This involves estimating the average daily mobility of individuals as a function of their socioeconomic characteristics while taking into account the fact that for a considerable part of the survey population, daily mobility is nil. Our approach intends firstly to define the causes of immobility better and secondly to characterise the mobility level with the help of a Hurdle model, so as to compare the results obtained in the web-based and faceto-face samples.