Best-worst Scaling Vs. Discrete Choice Experiments
An Empirical Comparison Using Social Care Data
Published In: Social Science & Medicine Vol. 72, no. 10, May 2011, p. 1717-1727
Posted on RAND.org on April 30, 2011
This paper presents empirical findings from the comparison between two principal preference elicitation techniques: discrete choice experiments and profile-based best-worst scaling. Best-worst scaling involves less cognitive burden for respondents and provides more information than traditional "pick-one" tasks asked in discrete choice experiments. However, there is lack of empirical evidence on how best-worst scaling compares to discrete choice experiments. This empirical comparison between discrete choice experiments and best-worst scaling was undertaken as part of the Outcomes of Social Care for Adults project, England, which aims to develop a weighted measure of social care outcomes. The findings show that preference weights from best-worst scaling and discrete choice experiments do reveal similar patterns in preferences and in the majority of cases preference weights—when normalised/rescaled—are not significantly different.