Automatic and Controlled Components of Implicit Prejudice

人们可能没有意识到他们自己重要的潜在态度和信念，特别是在刻板印象和偏见领域。然而，这些“隐性偏见”对日常社会交往中的群体间行为有很大的影响。事实上，这些偏见的影响往往大于明确表明的群体间态度和信仰，从而影响少数群体成员的重要结果。因此，理解隐性偏见的性质和运作，以及可能改变它们或降低它们影响的因素是至关重要的。为了实现这些目标，加州大学戴维斯分校的杰弗里·谢尔曼博士使用了他在过去的研究中开发的名为四方模型的工具进行数学建模。这一工具的应用表明，内隐态度和信念有时可能反映出记忆中有偏见的关联的自动激活，或者可能反映出未能调节此类关联对行为的影响。Quad模型可以用来估计这些因素中的每一个在产生有偏见和无偏见的行为中的独立作用。本研究有四个具体目标。首先，将应用四方模型来阐明影响内隐偏见程度的背景因素和个人因素。隐性偏见已被证明在不同的社会背景和不同的个人之间有很大的不同，通常反映了个人的目标和动机。本研究将对内隐偏向的背景和个体差异进行建模，以便更好地了解增加或减少偏向关联自动激活的因素以及调节这些关联表达的能力。第二个目标是应用一种“治疗”的方法来减少内隐偏见。在这些研究中，专门为影响潜在关联或监管这些关联的能力而设计的干预措施将适用于与增强关联激活或削弱自我调节能力相关的个人和背景。其目标是展示如何调整干预措施，以解决与隐性偏见增加相关的处理过程中的特定缺陷(过度激活关联与监管失败)。第三个目标是使用Quad模型来更好地理解刻板印象和偏见的不同措施之间的关系，这些措施往往彼此不对应。本研究提出，这些不同措施之间的分离可能反映出这些措施在多大程度上反映了关联的自动激活和未能对关联进行监管。应用该模型将有助于具体说明不同措施何时会产生相应的结果以及为什么不会产生相应的结果。最后，第四个目标是使用Quad模型更好地预测组间设置中的行为。刻板印象和偏见的衡量标准有时不能很好地预测人们对少数群体成员的实际行为。Quad模型可以通过独立评估自动关联的角色、管理这些关联的能力以及这些组件之间的交互来提高这些测量预测行为的能力。总之，本研究的目的是改进刻板印象和偏见的测量，增加对增加或减少这种偏见的因素的理解，并提高偏见测量对人们行为的预测能力。内隐态度会影响生活中重要领域的行为，包括执法、健康和就业。因此，更好地了解它们的性质和运作是至关重要的。

People may be unaware of their own important underlying attitudes and beliefs, particularly in the domains of stereotyping and prejudice. Nevertheless, these "implicit biases" significantly influence inter-group behavior in everyday social interactions. Indeed, the influence of these biases often is greater than that of explicitly stated inter-group attitudes and beliefs, affecting important outcomes for members of minority groups. As such, it is critical to understand the nature and operation of implicit bias, as well as the factors that may change them or reduce their influence. To achieve these goals, Dr. Jeffrey Sherman of the University of California - Davis uses mathematical modeling with a tool called the Quad model that he developed in his past research. Application of this tool has shown that implicit attitudes and beliefs sometimes may reflect the automatic activation of biased associations in memory or may reflect failures to regulate the influences of such associations on behavior. The Quad model can be used to estimate the independent roles of each of these factors in producing biased and unbiased behavior. There are four specific goals in this research. First, the Quad model will be applied to shed light on the contextual and individual factors that influence the extent of implicit bias. Implicit bias has been shown to vary significantly across social contexts and across individuals, often reflecting personal goals and motives. This research will model contextual and individual differences in implicit bias in order to better understand the factors that increase or decrease the automatic activation of biased associations and the ability to regulate the expression of those associations. The second goal is to apply a "treatment" approach to reducing implicit bias. In these studies, interventions that are designed specifically to influence either underlying associations or the ability to regulate those associations will be applied to individuals and contexts associated with enhanced activation of associations or diminished ability to self-regulate. The goal is to show how interventions can be tailored to address specific deficits in processing (over-activation of associations vs. failure of regulation) associated with increases in implicit bias. The third goal is to use the Quad model to better understand the relationships among different measures of stereotyping and prejudice, which frequently fail to correspond to one another. The present research proposes that dissociations among these different measures may reflect differences in the extents to which the measures reflect the automatic activation of associations versus the failure to regulate the associations. Application of the model will help to specify when and why different measures will and will not produce corresponding results. Finally, the fourth goal is to use the Quad model to better predict behavior in inter-group settings. Measures of stereotyping and prejudice are sometimes poor predictors of people's actual behavior towards members of minority groups. The Quad model can improve the ability of these measures to predict behavior by independently assessing the roles of automatic associations, the ability to regulate those associations, and interactions between these components. In sum, the purpose of this research is to improve the measurement of stereotyping and prejudice, increase understanding of the factors that increase or decrease such biases, and improve the ability of measures of bias to predict people's behavior. Implicit attitudes influence behavior in important domains of life including law enforcement, health, and employment. It is therefore critical to gain a better understanding of their nature and operation.