Using partial profile choice experiments to handle large numbers of attributes

Introduced by Chrzan & Elrod (1994) as a method for discrete choice experiments to handle large numbers of attributes, partial profile conjoint experiments (PPCE) were subsequently shown to have several benefits. As predicted, PPCE simplifies the respondent task enough to allow experiments with large numbers of attributes. In fact, PPCE makes the respondent task so easy that it results in a substantial net reduction in estimation error relative to full profile discrete choice experiments. Further, PPCE produces choice models with utilities similar to those of full profile discrete choice experiments (after adjusting for differences in scale parameters) and with equal or greater predictive validity. The current empirical study replicates these earlier findings while uncovering another benefit for PPCE: the robustness of its utility estimation in the presence of dominating attributes.