The existing empirical evidence shows that both contingent valuation and discrete choice experiment (DCE) methods are susceptible to various ordering effects. However, very few studies have analysed attribute-ordering effects in DCEs, and no study has investigated their potential influence on information-processing strategies, such as attribute non-attendance (ANA). This paper tests for attribute-ordering effects and examines whether the order of attributes describing the alternatives affects respondents’ propensity to attend to or ignore an attribute. A split-sample approach is used, where one sample received a DCE version in which the positions of the first and last non-monetary attributes are switched across the sequence of choice tasks compared with the other sample. The results show that attribute order does not affect welfare estimates in a significant way under the standard assumption of full attribute attendance, thus rejecting the notion of procedural bias. However, the welfare estimates for the attributes whose order was reversed and the share of respondents who ignored them differ significantly between the two attribute-ordering treatments once ANA behaviour is accounted for in the estimated choice models. These results highlight the important role of information-processing strategies in the design and evaluation of DCEs.
This study uses eye-tracking measures to account for attribute non-attendance (ANA) in choice experiments. Using the case of sustainability labelling on coffee, we demonstrate various approaches to account for ANA based on the fixation count cut-offs, definitions for detecting ignored attributes, and methods for modelling ANA. Some of the sustainability attributes identified through eye-tracking measures as being ‘visually ignored’ were truly ignored, whereas in none of the tested approaches was price truly ignored. The adequacy of eye tracking as a visual ANA measure might thus depend on the type of attribute. Further, the study unveiled inconsistencies in identifying non-attenders using visual ANA and the coefficient of variation. Based on our results, we cannot conclude that eye tracking always adequately identifies ANA. However, we identified several major challenges that can assist in further optimising the use of eye tracking in the context of ANA.
We seek to disentangle the effect of knowledge about an environmental good on respondents' propensity to ignore one or more attributes on the choice cards in a discrete choice experiment eliciting people's preferences for increased protection of cold-water corals in Norway. We hypothesize that a respondent's level of knowledge influences the degree to which she ignores attributes. Respondents participated in a quiz on cold-water coral prior to the valuation task and we use the result of the quiz as an ex-ante measure of their knowledge. Our results suggests that a high level of knowledge, measured by a high quiz score, is associated with higher probabilities of attendance to the three non-cost attributes, although this effect is only significant for one of them. A higher quiz score is also associated with a significantly lower probability of attending to the cost attribute. Furthermore, although being told your score has mixed directional effects on attribute non-attendance, it does not significantly affect the probability of attending to any of the attributes. Finally, allowing for attribute non-attendance leads to substantially lower conditional willingness-to-pay estimates. This highlights the importance of measuring how much people know about the goods over which they are choosing, and underlines that more research is needed to understand how information influences the degree to which respondents ignore attributes.
In this study, we investigate Scottish postgraduate economics students’ preferences for module design. Using a multi-profile best-worst scaling survey, we find that students have clear preferences on how they wish their modules to be delivered, taught and assessed. Furthermore, using a discrete mixtures modelling approach we explain the heterogeneous nature of preferences for the module attributes and the students’ lexicographic preference orderings. We show how failing to address this leads to erroneous results and limits the ability to derive reliable prediction. The findings in this study should appeal to university staff involved in the design of postgraduate (as well as undergraduate) courses as it should help them better establish a coherent learning experience for students, through which students can attain their full academic potential.