Draft: December 11, 2001

Towards a Social Psychology of Social Judgment:

Implications for Judgmental Ability and Self-knowledge

David C. Funder

University of California, Riverside

Prepared for Sydney Symposium on Social Psychology, 2002

Responding to the social world:

Implicit and explicit processes in social judgments and decisions

Edited by J. Forgas, W. von Hippel, and K. Williams

Psychology Press (Philadelphia)

Draft: please do not cite, quote, or ridicule without permission

e-mail comments to

Towards a Social Psychology of Social Judgment:

Implications for Judgmental Ability and Self-knowledge

Social judgment is a social process. It occurs when people meet, interact, observe each other, and begin to draw conclusions. These conclusions are often personality judgments concerning traits such as sociability, reliability, kindness, or even dangerousness. Such judgments are interesting, useful, and certainly important: they may determine further interaction. A person perceived as sociable will be approached, one perceived as reliable may be offered a job, and one perceived as dangerous may be avoided or even attacked. Because the judgments people make of each other have consequences, it matters whether they are accurate. An important task for psychology is to understand how accurate social judgment is achieved, the circumstances under which accuracy is more and less likely, and how accuracy might be improved.

The thesis of this paper is that a social psychological approach to social judgment can illuminate these issues. This may seem a strange thesis to have to make, but the currently dominant approach to the study of social judgment is not social psychological, it is cognitive, and although sometimes interpreted as having implications for accuracy, it actually has surprisingly few. A social psychological approach can more directly address accuracy issues because it goes beyond cognitive functioning by considering relevant interpersonal processes and social psychological variables.

The present paper has three objectives. First, it considers the differences between cognitive and social psychological approaches to social judgment and accuracy. Second, it outlines a particular social psychological approach, the Realistic Accuracy Model (RAM, Funder, 1995, 1999). Third and finally, it develops some specific implications of RAM for the nature of the “good judge” of personality and the conditions under which self-knowledge may be possible.

Two Approaches to Social Judgment

The Cognitive Approach

The cognitive approach to social judgment can be traced to pioneers such as Solomon Asch (1946) and contributors to the seminal volume by Tagiuri and Petrullo (1958). One reason for the immediate and long-lasting success of the approach was the creativity and persuasiveness of investigators such as these. Another reason is that the historical moment was right. The previously active research literature on social judgment had recently fallen into shambles. To devastating effect, Cronbach (1955; also Gage & Cronbach, 1955) pointed out that one of its major methods, the comparison of self-derived and judge-derived personality profiles, was complicated by several potential artifacts. The artifacts were avoidable, and indeed were not always artifacts (i.e., some were substantive phenomena in their own right), but the critical articles were complex and difficult to understand, and the statistical remedies they prescribed difficult to implement in a pre-computer age. It came as a huge relief, therefore, when, as related by Jones (1985, p. 87), Asch “solved the accuracy problem by bypassing it.”

Asch’s insight was that the process of interpersonal judgment could be studied in much the same way as any other cognitive process, by proposing and testing models that predicted particular relations between stimuli and responses. In his pioneering studies, the stimuli were lists of trait words, the models described processes such as “primacy” (which predicted that the first words presented would be more influential), and the responses were judgmental outputs where the relative influences of the stimulus words could be detected. These studies laid down the paradigm for most of the next 40 years of research on person perception and (as the field later came to be called) social cognition (see Fiske & Taylor, 1991; Kunda, 1999). The approach has been extraordinarily fruitful, demonstrating phenomena such as the primacy effect, the halo effect, illusory correlation, stereotype formation, effects of scripts on perception and recall, and the self-reference effect on memory, to name just a few.

Among the most prominent studies, especially in recent years, have been those demonstrating judgmental error (Funder, 1987, 1995). Studies of error are a slight variation on the standard cognitive paradigm because they propose models of the judgmental outputs that would be produced if subjects were to process the stimulus inputs perfectly. The models that define perfection may derive from abstract principles, formal logic or even mathematical formulae. For example, a sizable sub-literature presents subjects with inputs (such as categorical baserates) that can be processed by Bayesian formulae, and finds that their subsequent judgments never quite equal what the formulae dictate1. Numerous studies following this basic design have shown that subjects incorrectly compute posterior probabilities, correlation coefficients, expected values, and confidence intervals, among other errors.

Although the popularity of such studies surely stems, in part, from their apparently counterintuitive nature, the findings are not really surprising. The research design almost preordains error because the normative standard is a point prediction, which the mean judgment of a sample will never precisely match (Krueger & Funder, 2001). If measurement sensitivity and sample size are adequate, this deviation will be statistically significant, and the discovery of a new error will be proclaimed.

A sampling of such errors is shown in Table 1. Some of these errors are actually alternative labels for the same phenomenon, while some refer to several different phenomena (the “jingle-jangle” effect, Block, 1995). Others appear to be contradictory, as might be expected given that errors are identified by significant deviation on either side of the normative point prediction. For example, subjects have been observed to overestimate the effect of individual differences on behavior relative to the effect of the situation (fundamental attribution error), and also to overestimate the effect of the situation relative to individual differences (false consensus effect, external agency illusion). They are too prone to expect runs to end (gambler’s fallacy) and to expect they will last (hot hand bias). They give too little weight to beliefs about categorical baserates (baserate neglect error) as well as too much weight (stereotyping bias). Positivity and negativity biases are both widely reported, as are undue optimism and pessimism.

The errors form a long list, but no particular pattern. They are typically studied one at a time (which tends to protect contradictions from notice); some of them are effectively the property of particular labs or investigators. Singly or together, they do not yield a general theory of judgment, nor explain how accurate judgment is ever achieved (Anderson, 1990; Funder, 1995; Krueger & Funder, 2001). Indeed, the sheer number of errors – and every issue of PSPB and JPSP seems to proffer more2 – can make it seem as if human inferential processes are so basically flawed that accurate judgment is rare or impossible (e.g., Shaklee, 1991).

Any such implication is false. Recall that the express purpose of the cognitive paradigm, upon which demonstrations of error are based, was to bypass accuracy considerations (Jones, 1985). The processes discovered by the cognitive paradigm, including those that distort stimulus information, may lead to inaccurate judgments; then again, they may, like optical illusions, reveal adaptive processes that enhance accuracy in real life (Funder, 1987). Discoveries of error, notwithstanding their (ever increasing) number, do not speak to the accuracy issue one way or the other; they bypass it, just as Jones said. To address the degree to which, the circumstances under which, and the process by which judgmental accuracy is ever achieved, a fundamentally different approach is necessary.

The Social Psychology of Social Judgment

If the cognitive psychology of social judgment can be traced to the 1950’s, the social psychology of this topic can be traced back even further, to the 1930’s. Gordon Allport and several of his contemporaries examined connections between personality traits and behavior, and between behavioral observations and interpersonal perception. For example, Allport and Vernon (1933) showed that extraverts tend to take large steps, and that lay observers regard people who take large steps as extraverts. Underlying the approach was the assumption that people are, by and large, accurate. Allport noted that usually we are able “to select gifts that our friends will like, to bring together a congenial group at dinner, to choose words that will have the desired effect upon an acquaintance, or to pick a satisfactory employee, tenant, or room-mate” (Allport, 1937, p. 353). He viewed the job of psychology as being to explain how this is possible.

Most of the early research used self-other agreement as the criterion for accuracy; a judgment was presumed accurate to the degree that it matched a target’s self-description (e.g., Taft, 1955). The statistical problems entailed by such profile correlations received wide attention and helped undermine the whole approach, as has already been mentioned. A further issue, never really considered by Cronbach and other critics during the 1950s, is that self-description is less than a sure criterion for truth. But the quote from Allport, above, shows that he envisioned a much wider array of criteria (and he used them in some of his own work). The accuracy of a personality judgment resides not just in the degree to which it matches the person’s self-perception, but also the degree to which it predicts his or her social behavior (e.g., reaction to a gift, compatibility at a dinner party) and performance (e.g., as a tenant, employee, or room-mate).

A critical difference between the cognitive and social psychological approach to social judgment lies in the use of wide-ranging criteria such as these. The cognitive paradigm, as we have seen, entails assessment of whether subjects’ judgments of artificial stimuli match the prescriptions of a normative model. The social paradigm uses “correspondence” criteria, the degree to which judgments match the actual properties of real stimuli (Hammond, 1996). Of course, there is no single gold standard for assessing what the actual personality properties of a target person are, but there are several silver standards (if you will). Many investigators assess accuracy via inter-judge agreement, either between judges (e.g., Kenny, 1994) or between judges and targets (self-other agreement, e.g., Funder, 1980; Kenny & DePaulo, 1993). Judgmental accuracy is also sometimes assessed according to whether it can predict outcomes such as job performance (e.g., Mayfield, 1972; Waters & Waters, 1970) or directly observed social behavior (e.g., Kolar, Funder & Colvin, 1996).

This shift in criterion is fundamental to the difference between two approaches. From a cognitive perspective, judgmental accuracy is all about the stimulus information. Is it correctly recalled, processed without distortion, and combined according to normative formulae? These questions concern what happens between the ears. From a social perspective, judgmental accuracy is about social interaction. Does Jack agree with Bob about what kind of person Jill is; do they both agree with Jill herself? If they invite Jill to a dinner party, will she make a congenial companion? If they rent her a room, will she trash the place? These questions concern what happens in the social world. Their answers have social consequences, and it is these consequences that make the accuracy issue important.

The Realistic Accuracy Model

The Realistic Accuracy Model (RAM; Funder, 1995, 1999) is a social psychological approach that describes the interpersonal process that allows accurate interpersonal judgment to occur, when it does.

RAM is based on two assumptions. First, personality traits exist. Twenty years of debate notwithstanding (e.g., Kenrick & Funder, 1988), people really do have traits such as sociability, honesty, talkativeness, reliability, and so on. Their behavior relevant to these traits is consistent enough that it is worthwhile to characterize them in such terms, especially if you are intending to give them a gift, invite them to a dinner party, or rent them a room. The second assumption is that people sometimes manage to judge these traits accurately. As Allport might have put it, gifts are sometimes poorly received, dinner parties are disasters, and rented rooms become uninhabitable, but outcomes like these are at least sometimes successfully avoided. If these two assumptions are granted, then a question immediately arises: how is this degree, or any degree, of accurate personality judgment possible?

According to the Realistic Accuracy Model, accurate personality judgment is possible only to the degree that four things happen, as diagrammed in Figure 1. First, the person who is judged must do something relevant to the trait in question. Second, this relevant behavior must be available to the person who would judge the trait. Third, this relevant, available, behavior must be detected by the judge -- it must register, somehow, on his or her nervous system. Fourth and finally, the relevant, available, detected behavioral information must be corrected utilized -- interpreted and judged

The Four Steps to Accurate Judgment

An example may clarify the process. Imagine that a person has the trait of friendliness -- he or she really does like to be with people and seeks out their company. How could someone else judge this trait accurately?

First, the friendly person must do something friendly. All the friendly feelings and impulses in the world are no use, from a judgmental standpoint, if the person never acts on them. A relevant behavior, one produced by and therefore diagnostic of friendliness, is required. Notice that RAM already has an important implication, that traits might exist that are never exhibited and therefore never become judgable. Consider the trait of courage. Most of us have no idea whether we possess this trait or not. In the film Saving Private Ryan the character played by Tom Hanks clearly did have courage, which was exhibited in his clear thinking and ability to plan while under heavy fire on Omaha Beach. It was this extreme situation that produced relevant behavior. Had he stayed home, the opportunity (or need) to exhibit such behavior might never have arisen, and neither he nor anyone else would have had any way of knowing he had the trait.

Getting back to the easier example of friendliness, the relevant friendly behavior, perhaps an approachable smile or a warm greeting, must next be available to the person who judges it. In most cases, this simply means the judge must be present. The consideration is more trivial because people live in a variety of social contexts that influence or limit their behavior. For example, a person may display competent behavior in a workplace, while family members, who see him or her only at home, may have no basis on which to judge this trait. Another person may display warm behavior at home to a degree his co-workers would never suspect. In our dramatic example, the wife of the Tom Hanks character, back in Pennsylvania, probably has no idea about some of the traits he is displaying in France, because the relevant behavior is unavailable to her.

Availability has two important implications. First, the more contexts one shares with the person one is trying to judge, the more likely one is to be able to make a judgment that is generally rather than narrowly accurate (e.g., Blackman & Funder, 1998). Second, the less context-dependent an individual’s behavior is -- the less he or she varies behavior from one situation to the next -- the easier he or she will be to judge (e.g., Colvin, 1993).

At the third stage, once our friendly person has exhibited a relevant behavior in an available context, the judge must successfully detect it. RAM does not make a strong claim that this detection must be conscious -- it might be possible to use behavioral cues detected only implicitly -- but it does seem obvious that the behavioral information must register somehow on the perceiver’s nervous system before anything else can happen.

Detection can be problematic. An observer might be distracted by his or her own thoughts, or unable to do the multi-tasking necessary to simultaneously generate behavior and observe the actions of his or her interaction partner (Lieberman & Rosenthal, 2001). For example, an extremely shy person might be so self-conscious that he or she has little attention to spare for noticing the informative behaviors of anyone else. Or, relevant and available behaviors might be overlooked to the degree that other salient stimuli are present. It is easy to imagine that his companions under fire on Omaha Beach may have had other things on their mind besides being acutely aware of how much courage the Hanks character was demonstrating. Another possibility is that sensory deficits may interfere with detection. For example, older people who begin to lose their hearing may have problems in social relationships, including misperceiving the intentions of others, because they fail to detect informative social cues.