The Self Selection of Complexity in Choice Experiments
研究在离散选择实验中，个体是否会自愿增加任务复杂性，发现30%的样本选择最大选项集，且初始误差方差最低者更可能选择高复杂性，为设计认知效率更高的实验提供启示。
We investigate whether individuals will voluntarily increase the complexity of the tasks they complete within a discrete choice experiment (DCE). We do this via a ‘self selection of complexity’ design in which respondents choose whether to face choice sets comprising 3, 4 or 6 alternatives. We link this approach with the emerging Excessive Choice Effect (ECE) literature. We find that 30% of the sample opt for the largest sets. We test whether this choice of complexity reveals information about respondents' capability/commitment. We find that it does since those with lowest initial error variance levels are most likely to later select the highest level of task complexity. We argue that this result offers insights regarding the design of more cognitively efficient DCE designs. We consider the matching of respondents to the appropriate level of task complexity as analogous to the principal‐agent problem with asymmetric information. Rather than trying to understand respondents' cognitive capability or commitment ex ante we propose that participants self‐select designs that achieve the researcher's objective of minimizing error variance.