Empirical Evaluation of Preference Elicitation Techniques from Marketing and Decision Analysis

Within empirical preference measurement, conjoint analysis (CA) and analytic hierarchy process (AHP) are common methods with CA dominating marketing research and practice and AHP becoming more and more relevant as a tool of decision analysis. Both methods mainly differ with respect to their basic conception: AHP is a compositional method whereas CA is designed in a decompositional manner. Our study aims at comparing the methods as instruments of preference measurement on a fair basis therefore being designed as similar as possible. As decision problem we use the question which university to prefer or how to design preferable universities, respectively. The results of our empirical study show a high degree of predictive validity of both methods with slight advantages for AHP. Furthermore, they seem to have a high convergent validity. However, inspecting the results in detail reveals considerable discrepancies between them, in particular with respect to the computed part-worths and attribute weights. Due to not knowing the real preferences exactly we cannot undoubtedly state which method takes the better model of the preferences of the respondents but AHP seems to perform a little better and may be a reasonable alternative to CA especially in case of more complex marketing studies.