Possible Explanations of Correspondence Bias
in the Jones and Harris Paradigm

Much of the literature review below is based on Jones (1990),
Interpersonal Perception, Chapter Six, Correspondence Bias.

What follows this overview is a list of some factors possibly contributing to the claimed finding of "correspondence bias" using the Jones and Harris research paradigm.

You should know, by way of background, that research paradigms are extremely important in the development of scientific knowledge in psychology (and in other disciplines). Essentially what happens is that a research team employs a particular method (the paradigm--in this case the attitude-attribution research approach Jones and Harris developed) to document the existence of an important phenomenon (here, correspondence bias). Other scientists then use and modify this same paradigm to replicate the finding, to assess the conditions under which it occurs, and to attempt to account for or explain the phenomenon. Taken together, these exact or modified replications constitute what scientists refer to as a literature on the topic. What follows represents an introduction to such a literature, that on correspondence bias using the Jones and Harris paradigm.

One important purpose of such a literature is to assess whether the research outcome is only due to a peculiarity of the research method itself. If so, then the finding may be a possibly uninformative by-product of the method and would be labeled by psychologists as an artifact of the method. Are there explanations you generated for the Jones and Harris findings that point to this possibility? Be certain, later, to check what research discussed below suggests about the validity of those explanations.

If the findings of a paradigm are not artifactual, then the paradigm and its findings provide fertile grounds for exploring general features of social perception and competing explanations for the research outcome. Much of what follows lists possible explanations for correspondence bias and provides links to short reports of the methods and results of other studies evaluating those explanations. Spend some time exploring this abbreviated literature review to see how the possible factors you have generated to account for the correspondence bias have fared in light of subsequent research.

1. Who the subjects are.
2. What attitude is studied.
3. How the essayist's choice is constrained.
4. What the subject assumes the purpose of the experiment is.
5. How subjects have difficulty ignoring the starting position in the essay.
6. How subjects need cognitive effort to alter dispositional assumptions.

• Is correspondence bias a tendency only shown by distinct groups of people?
  
  
• Although most of the subjects in these studies are American college students (and students in introductory psychology classes at that), the correspondence bias effect has been replicated using the Jones and Harris paradigm for over 30 years. Moreover, findings on a variety of topics in social psychology show that the responses of American college students are generally consistent with those of other groups in our population in standard research paradigms.
  
  
• Is correspondence bias a peculiarly Western tendency? Consider the results of a recent study by Choi and Nisbett (1998).
  

• Is there something about the particular attitude studied (e.g., attitudes toward Castro assessed in the 1960's) that accounts for the outcome? If so, what might that influence be?
  
  
• Apparently this is not a factor. The same tendency has been shown on numerous other topics, including marijuana legalization, attitudes toward desegregation, attitudes toward capital punishment, attitudes toward abortion, and even attitudes toward Castro held by students in the 1990's.
  
  
• A variant of this argument is that it isn't the content of the attitude that accounts for correspondence bias but the fact that an attitude is being judged. Were something else being judged (e.g., a personality trait), perhaps subjects wouldn't show correspondence bias. This has been studied in other studies using different paradigms (e.g., Ross, Amabile, & Steinmetz, 1977) and in a study using a variant of the Jones and Harris paradigm: Miller, Jones, and Hinckle (1981). What do the results of Miller, Jones, and Harris suggest about this issue?
  
  

• Is there something about how situational constraint was introduced (that the essay was written as an assigned position in a writing course) that accounts for the outcome? What about this situational constraint may possibly make the paradigm a poor test of perceivers' willingness to adjust their conclusions about others in light of situational influences on behavior?
  
  
• This issue has been examined by employing alternative situational constraints in the Jones and Harris paradigm. For example, other cover stories have cited the constraints imposed by the assignment of course examination questions and of topics in a debate, or by the inducement provided by high monetary compensation.
  
  
• Perhaps the strongest test of this argument is provided by studies in which the perceivers themselves first engage in situationally-dictated misrepresentations of their own beliefs or personalities. Presumably, those subjects would then clearly appreciate the constraints operating on others who wrote similar essays. How do the results of the Miller, Jones, and Hinckle (1981) study reflect on this issue?
  
  

"Demand characteristics" are what psychologists call those features of an experiment that implicitly convey to subjects something about what is reasonable to assume in the experimental situation. These assumptions may, in turn, make the subjects' judgments much more reasonable than they otherwise appear to be. What might some of the demand characteristics of the Jones and Harris paradigm be?

• Arthur Miller argued that subjects assume researchers would not have asked subjects to judge the essays unless the essays conveyed something detectable about the real attitude of the essayist. The fact that Jones and Harris described their study to subjects as "an attempt to determine if people can make valid judgments of another's personality and attitudes on the basis of very limited information" may be taken as implying that the limited information of the essays should influence subjects' attributions.
  
  
• This argument was evaluated in a study by Miller, Jones, and Hinckle (1981). How did the design of this study examine demand characteristics? What did the results of the study show?
  
  

• One general tendency in human judgment is anchoring, that is, our susceptibility to the influence of an arbitrary starting point in judgments. For example, Tversky and Kahneman have shown that if you ask subjects one of the following questions: (a) Is the percentage of countries in the United Nations that are African more or less than 10%? or (b) Is the percentage of countries in the United Nations that are African more or less than 65% ? most subjects will respond to (a) by saying "more" and to (b) by saying "less." If you next ask all of these subjects to estimate the percentage of African countries in the United Nations, the subjects who answered (a) guess, on average, 25% and the subjects who answered (b) guess, on average, 45%. In other words, the subjects are unduly influenced by the arbitrary anchor in the "more or less" question first posed to them. They fail to ignore the anchor.
  
  
• How could this tendency account for correspondence bias?
  
  
• A psychologist named Quattrone suggested that subjects judging an essay in the Jones and Harris paradigm might begin by taking the position advocated in the essay as the anchor. This might then be adjusted by the subject's assumptions about the likely positions of people like the essayist on the issue in question. This adjustment may be insufficient, and the subject may be unduly influenced by the arbitrary anchor. This, in turn, would lead to the differences in the attitudes believed to be the true positions of the essayists in the "pro" and "con" conditions of the Jones and Harris study.
  
  
• How compelling do think this account is? How might it be tested?
  
  

• Generally speaking, it may be that our tendency to make correspondent inferences is basic. We may begin with correspondent (dispositional) social judgments for a variety of reasons, including the following:
  
  
  • We perceive an outcome associated with a person (e.g., view the person acting), and this perceptual reality is sufficient to produce the attribution. Relatedly, the influences of situations and settings may be less salient and concrete. They are less easily noticed.
    
    
  • We assume and believe that people believe what they say and act as they prefer. Whether this is generally accurate is another and important matter. It may be that this belief is validly grounded in everyday experience; alternatively, it may be a common misperception. Nonetheless, the possibility that this is our base rate assumption could be a powerful feature of interpersonal judgment.
    
    
  • We have been socialized to hold others and ourselves responsible for their lots in life. Correspondent inferences may simply reflect this.
    
    
• Perceivers may find it difficult to adjust their correspondence starting point in light of information about situational constraints. Such an adjustment may require great cognitive effort. In other words, in the absence of special cognitive effort, we assume correspondence.
  
  
• Social psychologist Dan Gilbert has been a proponent of a variant of the above view. Gilbert asserts that correspondent inferences occur automatically and nonconsciously. They follow directly from the perception of an individual's behavior (or traces of that behavior, e.g., in the form of the essay). Correspondent bias requires a correction of this perception, according to Gilbert, and that requires directed cognitive effort on the part of the perceiver to accomplish.
  
  
• How might this approach fit the Jones and Harris findings? How might a psychologist evaluate this account of correspondent bias in an experiment? To see how Gilbert, Pelham, and Krull did so, click here.