Determining the quality of conjoint analysis results using violation of a priori signs

Researchers in the field of conjoint analysis have suggested the use of sign violations of the estimated part-worth coefficients as a measure of the quality of the results. A priori sign violation is the mismatch between the ordinality of the true and estimated part-worth utilities. Based on the a priori sign criterion, LINMAP outperforms OLS for each type of evaluation model used. Validation using the a priori sign violation should be the preferred method when some or all of these conditions hold (the number of profiles is relatively small and the number of attributes is relatively large), when the dominant evaluative model is used, and the internal fit is poor (or evaluative error is large). The results suggest that the extent of sign violations should be used as an indication of the quality of results only within the context of the study design.