Homogeneous versus heterogeneous designs for stated choice experiments: Ain't homogeneous designs all bad?

Some recent attempts on constructing heterogeneous designs for stated choice experiments where different respondents or groups of respondents get different subdesigns have proven successful. Compared to homogeneous designs where all respondents get the same choice sets, heterogeneous designs allow for more variation in the attribute levels resulting in a larger amount of information on the respondents' preferences. Homogeneous designs have remained popular, however, because they are easier to generate and implement. In this paper, the question is raised about when homogeneous designs perform almost as well as heterogeneous designs under the Bayesian multinomial logit design framework. A simulation study is presented to identify the situations where the losses in estimation efficiency from using a homogeneous design are small and where they are large. When the residual degrees of freedom from using a homogeneous design are large and, to a lesser extent, the number of attributes and attribute levels are small, the efficiency losses are negligible and the use of a homogeneous design can be justified.