Response-ordering effects: a methodological issue in conjoint analysis.

Conjoint analysis is a technique relatively new to the evaluation of health care services in the UK. The technique uses data generated from questionnaires. This paper addresses the issue of response-ordering effects that may result from the ordering of dimensions of benefit within a question. Two questionnaires were given to 216 hospital consultants as part of a priority setting exercise. These were identical other than the ordering of the dimensions within each question. The regression analysis was segmented according to questionnaire type and the coefficients of the segmentation were tested for statistically significant differences. The results show no evidence of ordering effects.