Words or graphics to present a Discrete Choice Experiment: Does it matter?

Abstract

Objective: To test whether presenting attribute levels in words or graphics generates different results with respect to attribute level interpretation, relative importance and participation probabilities. Methods: Parents of 959 newborns completed a DCE questionnaire that contained two versions of the same nine choice tasks in which the attribute levels were presented in words or graphics. Five attributes related to the decision of parents to vaccinate their newborn against rotavirus were included. Mixed-logit models were conducted to estimate the relative importance of the attribute levels. Results: Respondents who started with the choice tasks in words produced the most consistent answer patterns. All respondents significantly preferred words to graphics. Part-worth utilities and the relative importance of the attribute levels differed based on the words and graphics data, resulting in different probabilities to participate in vaccination. Conclusions: Words were preferred over graphics, resulted in higher choice consistency, and showed more valid attribute level estimates. Graphics did not improve respondents' understanding of the attribute levels. Practice implications: Future research on the use of either words or graphics is recommended in order to establish guidelines on how to develop a valid presentation method for attribute levels in the choice tasks of a DCE.