Doctoral Dissertation Research in Economics: An Experimental Study of Confirmation Bias

事实证明，人类在吸收信息时是远远不够客观的。随着各种媒体和社交媒体平台深入融入我们的日常生活，了解人们对新信息的真实反应变得比以往任何时候都更加重要。确认偏误是指对证据偏向于某一假设的处理，在人们的信息解释过程中起着重要作用。然而，确认偏误并没有被完全理解，因为最近的实证研究已经产生了关于确认偏误如何被触发以及它如何影响人们的信念的混合证据。这项研究将调查并提供更多的见解，确认偏误如何影响个人在获得信息时更新他们对概率事件的信念的能力。本研究的结果将有助于决策者设计最优政策，帮助人们形成正确的信念，并防止信息来源操纵人们的信念。确认偏误有两种形式：有动机的确认偏误，指的是出于潜在偏好而对信息的偏见处理；无动机的确认偏误，指的是仅仅因为个人认为自己的假设比其他假设更有可能正确而对信息的偏见处理。这项研究将是第一个利用受控实验室实验来单独识别和检查两种形式的确认偏差之间的潜在关系和相互作用的研究之一。偏差的识别在于引出实验对象对概率事件的信念，用二元评分规则激励，并测量他们与贝叶斯后验信念的偏差。分离两种形式的确认偏误依赖于改变治疗变量，即动机。这项研究将两个引起越来越多关注的主题联系起来——确认偏误和动机推理。此外，本研究将测试一个新的基于理论的框架的有效性来分析确认偏误。该奖项反映了美国国家科学基金会的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

Humans have proven to be far from objective when they assimilate information. As various media and social media platforms have become deeply integrated within our everyday lives, it has become more important than ever to understand how people truly react to new information. Confirmation bias refers to the biased treatment of evidence in favor of a certain hypothesis, and plays an influential role in people’s information interpretation process. However, confirmation bias is not fully understood as recent empirical research have produced mixed evidence in regard to how confirmation bias is triggered and how it affects people’s beliefs. This study will investigate, and provide more insight into, how confirmation bias affects individuals’ abilities to update their beliefs about probabilistic events when they are presented with information. The results from this study will help policymakers design optimal policies that aid people in forming correct beliefs, and prevent information sources from manipulating people’s beliefs. There are two forms of confirmation bias: motivated confirmation bias, which refers to the biased treatment of information that is motivated by underlying preferences, and unmotivated confirmation bias, which refers to the biased treatment of information that arises simply because the individual believes her hypothesis is more likely to be true than alternative ones. This study will be among the first to utilize a controlled lab experiment to separately identify, and examine the potential relationship and interactions between the two forms of confirmation bias. Identification of the biases lies within eliciting experimental subjects’ beliefs about probabilistic events, incentivized with a binary scoring rule, and measuring their deviations from the Bayesian posterior beliefs. Separation of the two forms of confirmation bias relies on varying the treatment variable that is motivation. This study connects two topics that have attracted growing attention – confirmation bias and motivated reasoning. Additionally, this study will test the validity of a new theory-based framework for analyzing confirmation bias.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.