Doctoral Dissertation Research in Political Science: Interactions Between Risk, Cooperative Experience, and Distribution Preferences

实验游戏已经成为社会科学的有力工具。潜在的想法很简单：他们的目标是将现实世界的显著特征转化为可控的实验室环境。实验室中的行为可以通过展示人们看重什么以及他们如何推理，来洞察事物在现实世界中是如何发展的。此外，实验室环境使研究人员能够定制实验，检验现实世界提供的数据很少的假设，并操纵与决策相关的独特参数。对于负责拍卖宽带许可证的政府机构或Ebay这样的企业来说，了解某些拍卖规则是否有利于买方或卖方，显然是有价值的。实验室游戏显然对培养这种知识很有用。然而，是什么让实验室游戏与现实世界足够相似，以至于人们有资格认为前者获得的结果与后者有任何关联？这是社会科学界严肃辩论的主题。最近的两项发现突出了所涉及问题的复杂性。一项实验表明，个体在经济博弈中会以系统不同的方式进行，这取决于他们是否认为游戏中的其他参与者是自愿的。这是众多研究结果之一，表明在传统游戏结构中难以控制的“有意情境”可能对行为产生巨大影响。另一项实验表明，人们对“风险”的偏好在不同的拍卖机构中有所不同。“风险”是政治和经济学中一个极其重要的概念。这表明人们？美国的偏好可能取决于引发偏好的制度——这一发现将严重限制许多实验结果对现实世界问题的适用性。本文提出的研究采用了一种新颖的实验方案来检验公众关注的一个重大问题：财富分配偏好的本质是什么？著名的政治哲学家约翰·罗尔斯（John Rawls）将这种偏好作为其正义理论的基石。然而，旨在理解这些偏好的实验产生了各种相互矛盾的结果，不仅是跨文化和跨学科的，而且是在不同时间的学科内部。除了提出一个行为上的难题外，这还对任何分配正义计划的民主本质提出了严肃的问题。基于上述最新发现，本研究采用独特的风险偏好和合作意向性控制，以便更好地理解财富分享偏好。此外，这些控制能够让我们深入了解实验游戏的意图和制度背景的相关性，这本身就是一种重要的方法论进步。最重要的是，这些结果可能阐明了规范理想在经济行为中的本质和作用，并提出了关于有说服力地改变它们的可能性的进一步问题。在这个实验中，两个独立的小组首先参与一个相同的、非竞争性的彩票拍卖游戏的迭代，以获得个人奖励。第一组继续玩两个竞争性拍卖游戏的迭代，同样是为了获得个人奖励。然而，第二组玩家再次玩非竞争性游戏，在每次迭代中，所有玩家都获得任何单个玩家获得的最高收益。这既鼓励冒险的个人游戏，又让参与者通过集体承担风险而获得收益增加的体验，而在第一个池中，较小的奖励是在与他人竞争中获得的。最后，每个小组的成员都拿到了彩票，在他们“刮”彩票之前，他们被要求投票决定他们是希望奖金在小组成员中平均分配（“罗尔斯选项”），还是由幸运的个人保留。这个实验设计使我们能够调查三个问题，即：1)关于制度对风险偏好的影响的早期发现是否在更大的群体中得到复制？ 2)在个人层面上，特定拍卖引发的风险偏好是否预测了分享最终彩票的偏好？ 3)合作收益和风险分担的经历是否使第二组成员比第一组成员更倾向于分享最终彩票？回答第一个问题将以更一般的方式确立一个显著的发现，即某些偏好取决于引发偏好的制度。如果制度以这种方式起作用，博弈论家在从实验室环境中得出的偏好推断到各种现实世界环境时必须谨慎得多。对第二个问题的回答将表明是否存在不同类别的风险偏好，以及普遍的风险厌恶是否是解释“罗尔斯式”偏好的好方法。第三个问题的答案不仅表明过去的有意/制度经历是否会影响财富分配的偏好，而且还将更普遍地确立有意/制度背景与行为的相关性。因此，这项调查有望进一步理解与分配正义相关的偏好，以及总体经济行为的有意和制度背景的相关性。

Experimental games have become a powerful tool in social science. The underlying idea is simple: they aim to translate the salient features of real world situations into controllable laboratory environments. Behavior in the laboratory can provide insights into how things will play out in the real world by showcasing what people value and how they reason. Moreover, a laboratory environment enables investigators to tailor experiments, examining hypotheses for which the real world offers little data and manipulating unique parameters relevant to decision making. For a government agency in charge of auctioning off broadband licenses or for a business like Ebay it is obviously valuable to know, for example, if certain auction rules tend to benefit buyers or sellers. Laboratory games are clearly useful for developing such knowledge.However, what makes a laboratory game sufficiently similar to the real world, such that one is entitled to think that results obtained in the former have any relevance to the latter? This is the subject of a serious debate within the social science community. Two recent findings have highlighted the complexity of the issues involved. One experiment showed that individuals played economic games in systematically different ways based on whether or not they thought the other players in the game were acting voluntarily. This is one of many findings that suggest that "intentional contexts," which are difficult to control for in conventional game structure, may nonetheless exert enormous influence on behavior. Another experiment showed that people's preference for "risk," which is an extremely important concept in politics and economics, differed across various auction institutions. This suggests that people?s preferences may depend on the institutions that elicit them - another finding that would severely qualify the applicability of many experimental results to real world issues.The study proposed here uses a novel experimental protocol to examine a question of substantial public concern: what is the nature of preferences for wealth distribution? The political philosopher John Rawls famously made such preferences a cornerstone of his theory of justice. However, experiments aimed at understanding these preferences have come up with various conflicting results, not only cross culturally and cross subject pool, but also within subjects over time. In addition to presenting a behavioral puzzle, this raises serious questions about the democratic nature of any program of distributive justice. Building on the recent findings mentioned above, the present study employs unique controls for both risk preference and cooperative intentionality in order to better understand wealth sharing preferences. Moreover, these controls promise to grant new insight into the relevance of intentional and institutional contexts for experimental games, which would be a significant methodological advance in itself. Most importantly, the results may illuminate the nature and role of normative ideals in economic behavior and pose further questions about possibility of persuasively transforming them.In this experiment, two independent groups first participate in iterations of an identical, noncompetitive lottery-auction game for individual rewards. The first group proceeds to play iterations of two competitive auction games, also for individual rewards. The second group, however, plays the noncompetitive game again, and in each iteration all players receive the highest payoff achieved by any individual player. This both encourages riskier individual play and gives subjects the experience of increased earnings through the collective bearing of risk, whereas in the first pool smaller rewards are earned in competition with others. Finally, members of each group arepresented with lottery tickets and, before they "scratch" them, are asked to vote on whether they want the winnings to be divided equally amongst the group members (the "Rawlsian option"), or retained by the individuals who get lucky.This experimental design enables the investigation of three questions, viz: 1) Are earlier findings about the institutional effects on risk preference replicated for larger sized groups, 2) On an individual level do risk preferences elicited by particular auctions predict preferences for sharing the final lottery, 3) Does the experience of cooperative earnings and risk sharing dispose members of the second group to prefer sharing the final lottery more than the first? Answering the first question will establish in more general terms the remarkable finding that certain preferences depend on the institutions that elicit them. If institutions matter in this way, game theorist must be much more cautious about extrapolating from preferences derived in laboratory settings to various real world environments. Answers to the second question will suggest whether there are distinct classes of risk preference as well as whether general risk aversion is a good way of explaining "Rawlsian" preferences. Answers to the third question will not only suggest whether past intentional/institutional experiences can shape preferences for wealth distribution, but also establish the relevance of intentional/institutional contexts for behavior more generally. This investigation thus promises to further understanding both of preferences related to distributive justice as well as the relevance of intentional and institutional contexts for economic behavior at large.