Fundamental questions in behavior observation: Amount and overlap of information, sequence effects and perceiver-induced consistency

尽管行为观察在社会科学的许多分支中扮演着重要的角色，但关于这种方法的几个基本问题仍有待回答。目前的研究包含了一个创新的研究设计，使回答这些问题成为可能：(1)信息重叠：不同的感知者是否必须观察一个人的完全相同的行为才能对这个人做出类似的判断，或者如果他们判断同一个人，但不同的行为就足够了？(2)序列效应：如果先观察他人的行为来校准对一个人行为的判断，是否会更符合他人的行为判断，并更符合外部效度标准？观察者是否倾向于在一系列判断的早期阶段避免极端的判断？(3)感知者诱导的一致性：同一感知者行为判断的跨情境一致性在多大程度上反映了目标人特征的实际稳定性（而不是感知者倾向于使他/她的感知彼此一致）？(4)信息量：是否有可能重复基于非常有限的行为信息（即平均6个简短的情况）的相对准确的个人判断？如果所有的判断都是由同一个人提供的，这种情况会发生什么变化？为了解决这些问题，我们将实施一项严格控制的研究设计，其中200人在20种标准化情况下的行为被录像，然后由零认识的观察者判断（即，观察者对目标没有其他信息，也不与他们互动）。研究设计的一个关键因素是将观察者随机分配到两组：在一组中，同一名观察者在不同的情况下判断同一个人，在另一组中，观察者在相同的情况下判断不同的人。此外，不同的观察者接触到相同或不同的目标行为信息的程度，以及这些信息呈现的顺序，都是由实验操纵的。该研究将对我们对行为观察的理解做出重要贡献，从而使该技术在未来的研究中得到更有效的利用。最重要的是，这项研究使得调查一些最具影响力的人类感知模型中一些核心但未经检验的原则的有效性成为可能。

Although behavior observation plays an important role in many branches of social science, several fundamental questions in regard to this method remain yet to be answered. The present study comprises an innovative research designs that makes it possible to answer these questions: (1) Information overlap: Do different perceivers have to observe the exact same behaviors of a person in order to arrive at similar judgments about that person, or does it suffice if they judge the same person, but different behaviors? (2) Sequence effects: Will judgments of a person's behavior agree better with one another, and with external validity criteria, if they were calibrated by observing the behavior of other persons first? Do observers tend to avoid extreme judgments in the early stages of a sequence of judgments? (3) Perceiver-induced consistency: To what extent does the trans-situational consistency of behavior judgments by the same perceiver reflect the actual stability of the characteristics of the target person (as opposed to the perceiver's tendency to make his/her perceptions consistent with one another)? (4) Amount of information: Is is possible to replicate the finding that relatively accurate person judgments are possible based on very limited behavioral information (i.e., six brief situations, on average)? And (how) does this change if all judgments are provided by the same person? To address these questions, we will implement a strictly controlled research design in which the behavior of 200 persons in each of 20 standardized situations is videotaped and then judged by observers at zero acquaintance (i.e., the observers have no other information about the targets and do not interact with them either). A crucial element of the research design is the random assignment of observers to one of two groups: In one group, the same observer judges the same person across various situations, in the other group, an observer judges different persons in the same situation. In addition, the extent to which different observers are exposed to the same or different behavioral information about a target, and the order in which that information is presented, are experimentally manipulated. The study will make an important contribution to our understanding of behavior observation, and thus enable a more efficient use of that technique in future research. Above all, the study makes it possible to investigate the validity of some central yet untested tenets of the most influential models of person perception.