Internal Validity of Conjoint Analysis Under Alternative Measurement Procedures
Abstract
This simulation study investigates how conjoint analysis recovers various types of respondents' utility functions for specific attributes when the dependent variable is measured with rank order and n-point rating scales. The results show that: (1) the selected scaling procedure significantly affects various aspects of the utility function recoveries; and (2) the scaling procedure differentially affects the internal validity of conjoint analysis depending upon the chosen conjoint design and some basic characteristics of the underlying “true” utility functions. Implications for marketing researchers are discussed.