SAMPLE DESIGN FOR DISCRETE CHOICE ANALYSIS OF TRAVEL BEHAVIOR - THE STATE OF THE ART

A review of the state of the art in designing samples for discrete choice analysis of traveller behavior is presented. The basic discrete choice analysis framework is reviewed. It is assumed that any sample used is drawn by a process termed stratified sampling, in which the analyst partitions the population based on attributes and choices made, and then selects the fraction of observations taken within each stratum and the total sample size. Observations within strata are drawn at random. Two related problems, determining the distribution of the attributes in the population and estimating the choice probabilities condititional on the attributes, are explored. Various procedures for solving both these problems are detailed. The role of experimentation in extending the range of attributes in the population is explored. © 1979.