Regional Estimates in the National Survey on Transport and Travel
The national survey on transport and travel was conducted during 12 months partially covering the years 2007 and 2008. Since it was designed to obtain information concerning the whole national territory, a direct processing of this survey cannot provide estimates concerning smaller areas according to the occasionally insufficient size of regional samples. This study tests several estimation methods through regressions on “small areas”. Such methods use additional information on each region which is derived from the population census. The method based on a mixed regression in which the place of residence is specified as a random effect turns out to be the more appropriate to reduce the risk of model bias.