EAGER: Field Experiments and Formal Models of Arbitrary Social Inequality

社会学家长期以来一直怀疑，大量的社会不平等是从过去遗传下来的，即通过一种成功-繁殖-成功的机制，或“累积优势”。当一群被挑选出来的人在一大堆反复失败的人中积累了一连串的成功时，这很可能是因为这些幸运的人经历了小的、随机的初始优势，这些优势通过积极的反馈变得越来越大，从而使他们与其他人区别开来。然而，要确定任何给定的社会不平等都是武断的是很困难的，因为一旦明确了赢家和输家，就很容易通过选择性地将其归因于某些难以量化的个人品质来证明这种不同的成功是合理的。这项研究结合了一种实验策略，令人信服地展示了累积优势，这种优势可以在人们寻求某种形式的成功的许多现实情况下被现场部署。实验设计在成功积累的三个具体领域中实施：经济收益、社会地位和社会支持。在每个领域，实验者从现实世界的人群中随机选择个体，并通过互联网转账给予他们金钱、地位或支持。将随后的成功与非接受者进行比较，可以清楚地检验累积优势。此外，从这些实验中获得的不同成功数量中，我们了解到有多少过去的成功转化为多少未来的成功，这将用于构建累积优势的正式模型并将其参数化。反过来，这些模型将被用来计算一个系统中有多少不平等是由累积优势产生的。所提出的实验策略将揭示不同群体成员之间成功差异的很大一部分是否可归因于先前的成功差异。本项目提出的方法的一个关键特征是在不同领域应用一致的实验框架来测试一个共同的假设，并得出关于社会不平等动态的一般结论。这种方法有望使我们对社会不平等如何产生和持续的理解发生质的转变。更广泛的影响本研究的结果可能会提出新的政策干预措施，专门设计来纠正累积优势的影响，通过使用对自身的反馈机制。例如，研究可能表明，可以通过累积优势产生的收益偿还贷款的方式，有效地启动项目。此外，该项目中提出的大部分工作将通过提出一个可应用于其他重要社会学问题的协作框架，在计算机科学和社会学社区之间建立联系。此外，该项目主要包括研究生和本科生的培训和指导。

SES-1340122PI: Arnout van de RijtSUNY at Stony BrookSociologists have long suspected that a great deal of social inequality is inherited from the past, namely through a mechanism of success-breeds-success, or, "cumulative advantage." When a select group of individuals accumulates long strings of successes amongst large numbers of repeatedly unsuccessful others, this may well be because these fortunate individuals experienced small, random initial advantages that grew larger and larger through positive feedback, thus increasingly setting them apart from the rest. However, establishing that any given instance of social inequality is arbitrary is difficult, because once clear winners and losers are identified, it is easy to justify that differential success by selectively attributing it to certain hard-to-quantify personal qualities. This research incorporates an experimental strategy for compellingly demonstrating cumulative advantage that can be deployed live in many real-world situations where people seek some form of success. The experimental design is implemented in three specific live domains of success accumulation: financial gain, social status, and social support. In each domain, the experimenter selects individuals at random from a real-world population and bestows upon them money, status, or support through an internet-enabled transfer. Comparison of subsequent success with non-recipients provides a clean test of cumulative advantage. Further, from varying the amount of success bestowed in these experiments we learn how much past success translates into how much future success, which will be used to construct and parameterize formal models of cumulative advantage. These models, in turn, will be used to calculate how much inequality in a system is generated by cumulative advantage. The experimental strategy proposed will reveal whether a large portion of the variation in success across members of various populations is causally attributable to prior differences in success. A key feature of the approach proposed in this project is the application of a consistent experimental framework across different domains to test a common hypothesis and draw general conclusions about the dynamics of social inequality. This approach promises a qualitative shift in our understanding of how societal inequalities emerge and are sustained. Broader Impacts The results of this research may suggest novel policy interventions specifically designed to rectify the effects of cumulative advantage by using the very feedback mechanism against itself. For example, the research may suggest that it would be possible to effectively jump-start projects by means of loans that are repayable through the proceeds that cumulative advantage generates. Further, much of the work proposed in the project will build links between the computer science and sociology communities by suggesting a collaborative framework that can be applied to other important sociological questions. Additionally, the project centrally includes the training and mentoring of graduate and undergraduate students.