Abstract: When forming impressions about other people, stereotypes about the individual’s social group often affect the resulting impression. There are 2 distinguishable processes underlie stereotypic impression formation: stereotype activation and stereotype application. Most previous research has used implicit measures to assess stereotype activation and explicit measures to assess stereotype application, which has several disadvantages. The authors introduce a measure of stereotypic impression formation, the stereotype misperception task (SMT), together with a multinomial model that quantitatively disentangles the contributions of stereotype activation and application to responses in the SMT. For the SMT, we introduce its procedure and assumptions. In the SMT, participants are asked to form impressions of people whose faces are shown in blurred drawings. Participants work through several trials and judge each person, one at a time, on a certain trait. For instance, if one wants to assess the extent to which African Americans are stereotyped as threatening, one would ask participants to judge the faces on threat. Immediately before the presentation of the target drawing, a non-blurred prime picture of a real face is presented. On some trials, the prime picture shows the face of a person belonging to the stereotyped group of interest (e.g., a Black face). On other trials, the prime picture shows the face of a person belonging to another social group (e.g., a Caucasian face) or a neutral stimulus, such as a facelike shape. For the purpose of the multinomial modeling, the SMT requires participants to make a dichotomous judgment of each drawing (i.e., whether the drawing is high or low on a particular trait) and the target drawings vary slightly in the extent to which they convey the respective trait. This allows us to estimate the extent to which participants are able to base their judgment on the target (instead of on the prime). The rationale behind the SMT is the following. When participants form an impression of the person shown in the blurred drawing, they may apply the activated stereotypes. Because the blurred drawings provide only ambiguous information, participants may use any accessible knowledge to make their judgment. As a consequence, stereotypes about the person shown in the prime picture may bias the impression of the person shown in the blurred drawing. At the same time, participants might be able to correct for any bias if they are motivated to do so. Because participants have as much time as they want to give their response, it should be possible to adjust the judgment on the basis of a theory about the direction of the bias. The SMT can be characterized as an indirect measure of stereotyping because participants are not asked to self assess the to-be-measured construct and the stereotypicality of participants’ impression of the primes is inferred from their responses to the target drawings as a function of the preceding primes. For SMT’s multinomial model, we introduce its parameters, structure and data analysis. There are 4 parameters in its multinomial model: SAC (stereotype activation), SAP (stereotype application), D (detection of target trait) and G (tendency to guess), each of which represent a particular process. The 4 process all together contribute to the final impression. Each of the 4 parameter values indicates the possibility of the process which may take place during the impression formation. The parameters are estimated from the observed frequencies of the two response options given a particular prime and target. Each branch of the process tree demonstrates a possible way to form impressions about other people and the sum of the parameter values of this branch is the possibility of this way among all the possible ways to form impressions. In the final section of this article, we review the differences and similarities between SMT, affect misattribution procedure (AMP) and the quad model. The SMT is modeled after AMP, and the most significant difference between them is the task. SMT asks participants to make trait judgments about the target people while AMP asks them to make evaluative judgments about the Chinese characters, which participants have no idea of their actual meaning. The multinomial model of SMT and the quad model may look the same at first glance, but when comparing them carefully, we will find that there are many differences between them. Despite the differences between their parameters, some of the parameters may have the same meaning under some particular circumstances. Finally, we provide some limitations of SMT and some future directions about the study of SMT. Many open questions in the realm of stereotyping might be solved by SMT, such as the extent to which stereotype activation and stereotype application are automatic processes and the extent to which stereotype activation versus application is moderated by motivational, emotional, and contextual factors. We hope that SMT could be a useful measurement tool that separates multiple processes underlying impression formation and with SMT, the researchers can advance research on stereotyping.

Key words: stereotype misperception task, stereotype activation, stereotype application, affect misattribution procedure, the quad model

摘要： 刻板印象错误知觉任务（Stereotype Misperception Task,简称 SMT）是一种可以对刻板印象激活和刻板印象应用进行有效区分的研究范式，它所对应的多项式模型可以对刻板印象激活和刻板印象应用各自贡献量的大小进行测算与评估。依托对刻板印象激活与刻板印象应用的关系的解构，分析与评估了刻板印象错误知觉任务的操作程序、原理、所对应的多项式模型及其数据分析，比较了刻板印象错误知觉任务和情感错误归因程序以及四重模型内在的关联以及本质区别，评价与展望了刻板印象错误知觉任务的发展方向。

关键词: 刻板印象错误知觉任务, 刻板印象激活, 刻板印象应用, 情感错误归因程序, 四重模型