Detecting Discrimination: A signal detection analysis

歧视是基于种族、性别、年龄或性取向等特征对个人和群体的不公平或歧视性待遇。能够发现歧视是反歧视的必要第一步。然而，人们的共识是，在具体观察到的案件中，支持或反对歧视的证据往往是含糊不清的或隐藏的。因此，将结果归因于歧视的情况经常发生在相当不确定的情况下。不确定条件下的决策建模是信号检测理论(SDT)的目标，信号检测理论是歧视归因研究中的经典理论框架。SDT对理解这一判断过程的最重要贡献是分离和测量了影响判断过程的两个组成部分，即敏感度和反应偏差。敏感度是指区分有和没有歧视线索的案件的能力。反应偏差是指相对于其他属性更倾向于某一归因而不是歧视的倾向。尽管SDT在相关文献的理论讨论中发挥了重要作用，但还没有实证研究能够对观察到的歧视归因进行SDT分析。本提案的目的是迈出第一步，以新的模式填补这一空白。SDT分析承诺了两个贡献：(1)它们允许人们更准确地将经验发现描述为敏感性效应、反应偏向效应或两者的组合，以及(2)它们通过澄清中介路径为解决这些发现的理论提供信息。从长远来看，由此产生的见解还有助于设计培训，以提高观察员和决策者的敏感度，以便在歧视案件发生时发现它们。在该项目中，性别偏见、种族偏见及其交集的原因将通过SDT分析进行调查。这使得人们能够解决基于SDT的理论有时已经提供猜测的研究问题，但到目前为止还不能给出经验答案。例如：受歧视群体的成员在敏感性和/或反应上是否与不是歧视受害者或甚至受到歧视偏袒的群体的成员不同？关于典型歧视星座的刻板印象期望在这一过程中起什么作用，尤其是，这种期望是如何通过对敏感性或反应偏差的影响而产生的？

Discrimination is the unfair or prejudicial treatment of people and groups based on characteristics such as race, gender, age or sexual orientation. Being able to detect discrimination is a necessary first step in countering it. Yet, there is consensus that the evidence for or against discrimination in specific observed cases is often ambiguous or hidden. The attribution of an outcome to discrimination therefore regularly occurs in a context of considerable uncertainty. Modeling decisions under uncertainty is the goal of signal detection theory (SDT), that has been a classical theoretical framework in the study of attributions to discrimination. SDT’s most important contribution to understanding this judgmental process is to separate and measure two components that contribute to it, sensitivity and response bias. Sensitivity refers to the ability to distinguish between cases with and without cues to discrimination. Response bias refers to the tendency to prefer an attribution to discrimination relative to other attributions.Despite the prominent role that SDT has played in theoretical discussions in the relevant literature, there are no empirical studies that have made possible SDT analyses of observed attributions to discrimination. The purpose of the present proposal is to take a first step towards filling this gap with a new paradigm. SDT analyses promise two contributions: (1) They allow one to characterize empirical findings more precisely as sensitivity effects, response-bias effects, or combinations of both, and (2) they inform theories addressing these findings by clarifying mediational pathways. In the long term, the resulting insights can also contribute to the design of trainings to sensitize observers and decision makers to detect cases of discrimination when they occur.In the project, attributions to gender bias, racial bias, and their intersection are to be investigated by SDT analyses. This allows one to address research questions for which SDT-based theorizing has already sometimes offered conjectures, but for which empirical answers could so far not be given. Examples are the following: Do members of groups that are discriminated against differ in sensitivity and/or response from members of groups that are not victims of discrimination or are even favored by it? What is the role of stereotypical expectations about typical constellations of discrimination in this process and in particular, how are the effects of such expectations mediated – via effects on sensitivity or response bias?