WOULD YOU LIKE SOME MORE? VOLUNTARY CHOICE SETS AND PROGRESSIVE INCENTIVE ALIGNMENT IN CONJOINT ANALYSIS

摘要

A weakness of analyzing choices in conjoint analysis is that choosing an alternative from a set of products conveys less information than, e.g., rating or ranking each one of them. Thus, for a robust estimation of preferences it is essential to collect a multitude of choices. However, increasing the number of choice sets is not trivial because respondents react negatively to longer surveys and after a certain threshold, are more likely to (1) abandon the survey, which decreases response rates and increases costs, (2) make more errors or answer randomly, which results in higher error variance; or (3) adopt simpler, cognitively less demanding decision strategies (e.g., concentrating on a subset of attributes), leading to biased estimates and observed utility evolution (Liechty, Fong, and DeSarbo 2005; Pocheptsova et al. 2009; Swait and Adamowicz 2001). Although every researcher and practitioner who uses conjoint analysis must make this important trade-off between estimation accuracy and consumer fatigue, little is known about how many choice sets can be established without invoking respondent fatigue.