An experiment on simplifying conjoint analysis designs for measuring preferences
通过增加属性水平重叠来简化联合分析中的选择任务，并与标准最小重叠设计比较，发现虽未显著改善一致性或效率，但影响了陈述偏好，有助于理解受访者如何回答联合分析问题。
In conjoint analysis (CA) studies, choosing between scenarios with multiple health attributes may be demanding for respondents. This study examined whether simplifying the choice task in CA designs, by using a design with more overlap of attribute levels, provides advantages over standard minimal-overlap methods. Two experimental conditions, minimal and increased-overlap discrete choice CA designs, were administered to 353 respondents as part of a larger HIV testing preference survey. In the minimal-overlap survey, all six attribute levels were allowed to vary. In the increased-overlap survey, an average of two attribute levels were the same between each set of scenarios. We hypothesized that the increased-overlap design would reduce cognitive burden, while minimally impacting statistical efficiency. We did not find any significant improvement in consistency, willingness to trade, perceived difficulty, fatigue, or efficiency, although several results were in the expected direction. However, evidence suggested that there were differences in stated preferences. The results increase our understanding of how respondents answer CA questions and how to improve future surveys.