Comparing the performance of different approaches to deal with attribute non-attendance in discrete choice experiments: a simulation experiment

Abstract

There is a growing body of literature acknowledging that respondents to DCE often use simplifying strategies, like ignoring one or several attributes to provide with their choices. Two main approaches have appeared to analyse the impact of attribute non-attendance on welfare estimates: the stated non-attendance (SNA) approach and the analytical non-attendance (ANA) approach. Using simulation experiments, this paper investigates the results and reliability of the approaches developed in the recent years in order to deal with attribute non-attendance. The simulation results indicate that the treatments so far proposed are not in all cases suitable. In the absence of correlated errors, the SNA approach seems to provide with unbiased welfare estimates but the ANA approach fails to do so. On the other hand, in the presence of correlated errors, none of the approaches seems to provide with unbiased WTP estimates in all cases.