Both Choice Formats and Labeling Styles Do Matter in Eliciting Preferences and Reducing Attribute Non-Attendance

Comparisons between discrete choice experiments (DCEs) and best-worst scaling (BWS) have been occasionally made in the literature to date. Likewise, labeled and unlabeled experimental designs have been compared under some settings. However, these research findings are basically mixed in terms of difference in parameter estimates and other key aspects of the experiments. The purpose of this study is to compare four different question formats, namely, a labeled DCE, labeled BWS, an unlabeled DCE and unlabeled BWS, so as to contribute to such discussions. A series of econometric estimations in this study revealed that these alternative choice formats and labeling styles resulted in significantly different taste parameters. Specifically, the presence of label would be more impactful than the choice of format. Also, it was found that the impacts of choice formats on the error variances and attribute non-attendance were different depending on the labeling styles. These results suggest that both choice formats and labeling styles do matter in eliciting preferences and reducing attribute non-attendance