Sample optimality in the design of stated choice experiments Submitted

Recent research by Bliemer and Rose (2005, 2009, in press) and Rose and Bliemer (2005) suggest as a measure for calculating sample size requirements for models estimated using stated choice data, the S-error statistic. Prior to this, existing sampling theories failed to adequately address the issue of sample size requirements specifically for this type of data and hence researchers have had to resort to simple rules of thumb or ignore the issue and collect samples of arbitrary size, hoping that the sample is sufficiently large enough to produce reliable parameter estimates. In this paper, we explore the sample size calculations proposed by Bliemer and Rose and demonstrate how these measures may be used to suggest a theoretical minimum sample size, assuming prior parameter values used in generating experiments. Sample size requirements for different model types are explored via three different case studies. The paper finds that the S-error statistic provides a robust estimate of the minimum sample size requirements for stated choice studies, however it is recommended that larger sample sizes than suggested by the statistic be collected to allow for different sources of misspecification that can occur during the course of such studies.