EFFICIENT VERSUS NON-EFFICIENT STATED CHOICE DESIGNS. A COMPARISON IN A MODE CHOICE CONTEXT

This paper evaluates gains in efficiency produced by the use of efficient designs to analyze stated choice (SC) data. Based on a standard experiment used in a previous research, we compare the efficiency of this design with that of the efficient design obtained according to the minimization of the D-error, considering different modelling strategies. The experiment was conducted in the context of the choice between the plane and the new high speed train in the route Madrid-Barcelona. As the levels assigned to some attributes in the stated choice exercise were customized to each respondent experience, pivoting the information provided by preliminary revealed preference questions around the reference alternative (the plane, in this case), a different efficient design was created for every respondent in the sample. Results of the analysis demonstrate that substantial gains in the significance level of the parameter estimates could have been attained if the efficient design had been used to analyze SC data.