An empirical comparison of conjoint and best-worst scaling case III methods

Abstract

The Best-Worst Scaling (BWS) case III method, also called the BWS ‘multi-profile case.’ has been widely used to characterize survey respondent preferences for market goods. The BWS method is similar to conjoint analysis methods in that respondents select from a set of hypothetical item profiles with different attribute levels. Unlike conjoint methods, which allow respondents to select their best/most preferred profile, the BWS case III method asks respondents to select ‘best’ and ‘worst’ profiles in each choice set. This study compares consumer willingness to pay (WTP) estimates from conjoint and BWS case III survey formats. Data on consumer preferences for single-use eating-ware products made from biobased materials were collected. Results suggest that for the most preferred attribute levels, WTPs estimates are similar in magnitude and consistent for signs across methods. For least-preferred attributes, WTP estimates from the conjoint method are higher than those of the BWS method. However, the BWS WTP estimates have smaller confidence intervals.