Threshold Determination Using Extensions of Best-Worst Scaling
(A2021-93642)

AUTHORS

Sven Beisecker, WHU - Otto Beisheim School of Management; Christian Schlereth, WHU - Otto Beisheim School of Management; Felix Eggers, University of Groningen

ABSTRACT

Best-worst scaling (BWS) is a popular method that seeks to measure preferences for multiple items on a continuous scale between two extremes (e.g., “best” and “worst”). Yet, BWS suffers from the threshold identification problem, i.e., the obtained scores and rankings provide insights into each item’s relative preferences, but not into the overall acceptability of an item. For example, firms applying BWS to score different slogans will not know which of these, if any, are acceptable. The present paper (i) proposes different threshold identification approaches and (ii) develops models for the corresponding multinomial Hierarchical Bayes estimation. In two empirical studies we compare the approaches’ choice consistency, response time, cognitive ease of survey completion, and resulting parameter estimates. Although simulation results seem to advocate the elicitation of more information, empirical evidence shows that the simplest indirect threshold identification approach is on par.