A Cautionary Note on the Design of Discrete Choice Experiments: Reply
本文是对2005年一篇关于离散选择实验设计论文的回复，指出不同设计策略对支付意愿估计的效率有差异，但几乎没有证据表明存在系统性偏差，并希望帮助研究者应对审稿中关于设计偏差的批评。
When we became interested in conducting choice experiments (CEs) in the first part of this decade, we were surprised at the lack of guidance on how to construct experimental de-signs for CEs and the lack of systematic anal-ysis in the agricultural economics literature on the consequences of using differing designs for willingness-to-pay (WTP) estimates. As indi-cated in our 2005 paper, of the few CE papers that had been published in the agricultural eco-nomics literature at that time, authors used a myriad of strategies to construct designs, and our 2005 paper represented an effort to, quite frankly, educate ourselves. One of the main things we learned from the simulations reported in our 2005 paper is that, although efficiency of WTP estimates varies quite a bit from design to design, there was lit-tle evidence of bias among the design strategies and utility functions we considered. Although Monte Carlo experiments are, by their very na-ture, open to concerns about lack of generality, we are hopeful that our findings have helped other researchers respond to the kinds of cri-tiques that often arise in reviews of CE papers in which critics wrongly assert bias exists when what they really mean is that more efficient designs exist. It is also worth noting that, even before our paper appeared in print and espe-cially since then, there have been extensive de-velopments in the design of CEs (e.g., Burgess