I have relates this article to a contemporary issue. Could you pleasehelp me decide a topic to write my paper using this reading?On the Accuracy of Personality Judgment: A Realistic Approach David C. Funder University of California, Riverside The “accuracy paradigm” for the study of personality judgment provides an important, new complement to the “error paradigm” that dominated this area of research for almost 2 decades. The present article introduces a specific approach within the accuracy paradigm called the Realistic Accuracy Model (RAM). RAM begins with the assumption that personality traits are real attributes of individuals. This assumption entails the use of a broad array of criteria for the evaluation of personality judgment and leads to a model that describes accuracy as a function of the availability, detection, and utilization of relevant behavioral cues. RAM provides a common explanation for basic moderators of accuracy, sheds light on how these moderators interact, and outlines a research agenda that includes the reintegration of the study of error with the study of accuracy. Believe truth! Shun error!these, we see, are two materially different laws; and by choosing between them we may end by coloring differently our whole intellectual life. (James, 1897/1915) Judgments of personality are attempts to identify the psychological properties of people, such as personality traits, that help to explain what they have done in the past and to predict what they will do in the future (Funder, 1991). For example, one person might judge the degree to which another is critical, dependable, or energetic. The judgment might be used to explain why he or she insulted an acquaintance or to predict whether he or she will work at a job reliably and well. Judgments like these are not rare in the lab or in daily life and frequently have been the focus of psychological research. For almost two decades, the dominant, implicit motto of this research surely has been “Shun error!” The detection of error has constituted an important approach toward the study of how people judge each others’ personalities, as well as of human judgment more generally (e.g., Kahneman & Tversky, 1973; Ross & Nisbett, 1991). A prolific, ingenious, and influential program of research has exhaustively detailed characteristic shortcomings of human judgment (cf. Ross, 1977, and Nisbett & Ross, 1980), generated suggestions about how to mitigate some of these shortcomings, and yielded insights into processes or “heuristics” of human judgment that can produce error (Funder, 1987; Jones, 1985). Still, successful as it has been, the “error paradigm” can tell no more than half of the story. The avoidance of error is not The research described in this article was supported by National Institute of Mental Health Grant R01-MH42427. I am grateful for the comments of Melinda Blackman, Randy Colvin, Patricia Funder, Jacob Hershey, William Ickes, Robert Kaiser, David Kolar, Dan Ozer, David Schneider, Stephen West, Leslie Wiehl, and Tiffany Wright. Correspondence concerning this article should be addressed to David C. Funder, Department of Psychology-075, University of California, Riverside, California 92521-0426. Electronic mail may be sent via Internet to (underucrac 1 .ucr.edu. quite the same thing as the achievement of accuracy, and explanations of how errors arise shed relatively little light on how correct judgments are ever made. As James (1897/1915) pointed out nearly a century ago, the shunning of error needs to be complemented by a more positively oriented seeking after truth. Just this sort of complement has begun to emerge from an alternative approach to the study of personality judgment that could be called the “accuracy paradigm” (e.g., Funder, 1987; Kenny, 1994; Kruglanski, 1989; Swann, 1984; for a collection of representative research, see Funder & West, 1993b). Followed by an increasing number of researchers in recent years, this new paradigm focuses not so much on what judges cannot do but on what they can make and the circumstances under which they can do it. Although the accuracy paradigm seemsand isrelatively optimistic compared with the error paradigm, it is not really Panglossian. Judgmental mistakes obviously are frequent in daily life. But the accuracy paradigm incorporates a recognition that progress in this area requires examination of when and how people are correct as well as when and how they are mistaken. Varieties of Accuracy The modern accuracy paradigm comprises three major variants: the pragmatic, constructivist, and realistic approaches (Funder & West, 1993a). The pragmatic approach views personality judgments as necessary tools for social living and evaluates their accuracy in terms of their practical value (McArthur & Baron, 1983; Swann, 1984). A judgment is viewed as accurate if it leads to successful social interaction. The constructivist approach views personality judgments as social constructions and evaluates their accuracy in terms of consensus, or agreement between judges (Kruglanski, 1989). In its extreme form, constructivism implies that if judges agree, they must be considered accurate, because no other criterion is available even in principle. In its less extreme form, research focuses exclusively or nearly exclusively on the factors that promote and in652 ACCURACY OF PERSONALITY JUDGMENT 653 hibit interjudge agreement independent of accuracy (Kenny, 1994). Both of these approaches provide useful insights, and the research that emanates from them constitutes an important part of the accuracy paradigm. However, the present article is organized around and will argue specifically for a third approach, to be called the Realistic Accuracy Model (RAM). RAM begins with a general and consequential assumption that sets it apart from the error paradigm as well as the other two approaches to accuracy. Its basic, “neo-Allportian” assumption is that personality traits are real characteristics of individuals (Allport, 1937; Funder, 1991,1993b). This assumption yields two implications. The first is that the accuracy of personality judgment is an extremely complex matter. It goes beyond relatively convenient operational definitions and into complex issues concerning the construct validity of personality traits (Cronbach & Meehl, 1955; Ozer, 1989). The evaluation of accuracy requires consideration of the widest range of data that an investigator can gather and moves the study of social judgment off the familiar territory of social psychology onto some of the traditional turf of personality psychology. The second implication is that accuracy in personality judgment is a joint product of the attributes and behavior of the target as well as of the observations and perspicacity of the judge. In other words, accuracy stems from the relevance, availability, detection, and utilization of behavioral cues. Other approaches to error and accuracy do not yield this implication because they pay much more attention to the judge than to the person who is judged. The assumption that traits are real determines the distinctive way research within RAM seeks data, evaluates accuracy, and organizes results. The Importance of Accuracy Not all judgments of personality concern traits, but many certainly do. Consider the typical letter of recommendation. Trait constructs abound (e.g., “the candidate is conscientious, energetic, insightful, and pleasant”), and if these constructs have any meaning at all, then their application might be right, wrong, or someplace in between.1 Or, a college student home on vacation might be asked by her mother to describe her new roommate. “She is friendly, but sort of sloppy,” says the student, “and very hardworking.” Presumably, these terms are meant to describe something real. Or, to be more precise, each of these terms intends to describe two real things: a pattern of behavior and an inferred attribute of the person who performed it (Funder, 1991). Sometimes personality judgments are made and communicated in an attempt to predict the future behavior of the person whose attributes are described. For example, the reader of the letter of recommendation may have to make a decision about whether to admit or hire the person in question. The basis of this decision is in large part a prediction of the person’s behavior in school or on the job that the reader bases on the traits described in the letter. But very often the goal of trait description is as much intrinsic as it is pragmatic. People are naturally curious about each other, and the college student’s mother may want to know about who her daughter is living with, not really to predict the roommate’s behavior, but just because she cares. The nature of much interpersonal gossip demonstrates how interested people are in judging and communicating about each others’ salient characteristics, sometimes for what seems like no particular reason, which is precisely what shows the interest to be intrinsic. The accuracy of personality judgment is also important for methodological, theoretical, and philosophical reasons (Funder & West, 1993a). From a methodological perspective, human judgments of individual characteristics are an important source of data for personality, developmental, and clinical psychology. Many studies ask informants, who may be lay peers or clinical assessors, to use a set of rating scales or a Q-sort deck to distill and describe their impressions (e.g., Bern & Funder, 1978; Block, Block, & Harrington, 1974; Block & Robins, 1993; Colvin, Block, & Funder, 1995; Funder & Colvin, 1991). These judgments are then used as data about the personality attributes of the subjects they describe. The quality of these data and the validity of the conclusions drawn from them depend critically on the accuracy of the judgments (Funder, 1993a). From a theoretical perspective, the accuracy of personality judgment is intertwined with the issue of how personality is manifest in behavior. The realistic assumption of RAM implies that to understand when and how a trait can be inferred, one has to understand when and how that trait influences what people do. This latter issue is a traditional concern of personality psychology, and for that reason accuracy research is as relevant to that field as it is to social psychology, the more usual home for research on interpersonal judgment in recent years (Allport, 1937; Funder, 1993b). From a philosophical perspective, the key questions about perception and judgment pertain to the knowledge people possess, the way this knowledge is obtained, and the degree to which it reflects the true state of nature (Hastie & Rasinski, 1988; Kruglanski & Ajzen, 1983). RAM’s realistic approach specifically addresses the connection between social perception and the psychological reality that it presumes to lie beneath. From Accuracy to Error to Accuracy Again The First Wave of Accuracy Research The history of research on personality judgment over the past 60 years shows an interest in accuracy that has waxed, waned, and then recently waxed again.2 Certainly, from a lay, naive, or commonsensical point of view, the most obvious question that might be asked about a personality judgment is, “Is it right or wrong?” It was only natural, therefore, for research on personality judgment to begin, in the mid-1930s, with a focus on ac1 A radical constructivist view would regard words like these as essentially meaningless in that they are not seen as referring to anything other than arbitrary ideas in the mind of the perceiver. Such an approach stops the analysis of accuracy before it even starts and therefore must be rejected here. 2 For much more detailed renditions of this history, see Funder (1987), Funder and West (1993a), and Kenny (1994). 654 DAVID C. FUNDER curacy (e.g., Allport, 1937, chapter 18). A particular concern was the search for the “good judge” of personality, the individual difference characteristics that would make a person more likely to be an accurate judge of others (Estes, 1938; Taft, 1955). A lively research tradition organized around this topic came to an abrupt halt in the mid-1950s, however, for at least three reasons that can now be identified. The first is that the search for the good judge proved by any standard disappointing. Personality correlates of judgmental ability were weak and changeable across studies, and judgmental ability itself often was inconsistent. The same person who was a good judge in one con might be a poor judge in another con that was only slightly different (Cline & Richards, 1960; Crow & Hammond, 1957; Schneider, Hastorf, & Ellsworth, 1979). A second cause for the falloff in interest in accuracy was the publication, in 1955, of a set of severe critiques of the methods used by almost all the research on the topic reported up to that time (Cronbach, 1955; Gage & Cronbach, 1955; see also Hastorf & Bender, 1952). Writing a pair of difficult articles using unconventional statistical notation, Lee Cronbach nonetheless succeeded at convincing his colleagues that nearly all the accuracy research he reviewed was nearly meaningless. The basic reason was that the numbers used to reflect interjudge agreementthe usual criterion for accuracy then as now were potentially influenced by several more-or-less artifactual factors including response sets, actual and assumed similarity between judge and target, and so forth. Cronbach’s intention may have been to improve accuracy research rather than to shut it down, but the latter is very nearly what happened. Both the complex statistical adjustments recommended in a precomputer era and the large amount of data gathering that Cronbach argued was required apparently proved daunting to many investigators. As Schneider et al. (1979) noted, the topic suddenly “lost some of its intuitive charm” (p. 222). For more than two decades (until approximately 1980),3 very few studies of accuracy followed Cronbach’s critique. The Focus on Cognitive Process A third reason for the falloff in accuracy research in the mid1950s was the growth of an alternative paradigm within the study of what became called person perception (e.g., Bruner & Tagiuri, 1954;Tagiuri&Petrullo, 195 8). This paradigm moved the study of interpersonal judgment into the laboratory, where subjects could be induced to make judgments of artificial stimuli such as lists of trait words rather than of real people. Asch (1946) and others discovered that such studies could reveal interesting aspects of the way information is combined on the way to personality judgments, without requiring any concern with (and without providing any information about) the social content or accuracy of these judgments. Consider one of the pioneering studies in this tradition, the one by Asch (1946). In cleverly designed experiments, Asch presented subjects with bits of descriptions of hypothetical stimulus persons (e.g., a list of trait words). He showed how the degree to which a judgment was affected by a given piece of information depended on whether the information was presented first, in the middle, or last and whether it seemed consistent or contradictory relative to the rest of the information presented. This was and remains important work, but notice how its conclusions are not specifically relevant to personality. Analogous findings could beand arefound concerning other kinds of information leading to other kinds of judgment (cf. J. R. Anderson, 1990; Glanzer & Cunitz, 1966; Koppenaal & Glanzer, 1990). The research is in that sense “content-free”: It addresses the cognitive process rather than social substance of judgment, and this process occurs wholly in the mind of the judge rather than in the interpersonal world. By the same token, such research and that which it inspired say little or nothing about the variables that might affect accuracy. Asch’s (1946) “primacy effect,” for example, will enhance accuracy if the content of the information presented first happens to be diagnostic of what is judged, and the very same effect will lessen accuracy if the information is undiagnostic or misleading. However, accuracy was never the point of this kind of research on person perception. Asch, for one, never claimed that his studies had accuracy implications. A later but equally prominent practitioner of cognitively oriented research on person perception, E. E. Jones, explicitly commented about how studies in the Aschian style, including his own, “solved the accuracy problem by bypassing it” (Jones, 1985, p. 77). In con, Jones was not acknowledging any deficiency. Rather, he viewed it as salutary that research in person perception had found a way around the difficult and perhaps intractable issues involved in the evaluation of judgmental accuracy. Over the next decades, a great deal of research on “attribution theory” and, later, “social cognition” grew out of this approach. Nearly all of it used experimentally manipulated, artificial social stimuli that were ideally suited for testing models of some of the fine cognitive processes involved in person perception. None of it said very much about accuracy one way or the other (e.g., Fiske & Taylor, 1991; Schneider et al., 1979). The Rise of Error What was banished from the front door began creeping in slowly through the back door, however, with the gradual rise of research on judgmental error. This approach originated within 3 The literature on industrial /organizational (I/O) psychology provides an important exception, notably as it has appeared in the pages of the Journal of Applied Psychology. Perhaps because of its relative isolation from mainstream personality and social psychology research, or perhaps because of its pressing concerns with applied issues of personnel selection and placement, I/O psychology maintained a steady interest in accuracy throughout the 1950s and 1960s and into the present day (e.g., Borman, 1977; Gifford, Fan Ng, & Wilkinson, 1985; Hollander, 1957, 1965; Kane & Lawler, 1978; Lewin & Zwany, 1976; Mayfield, 1972; Paunonen & Jackson, 1987; Schmidt, Hunter, Croll, & McKenzie, 1983; Waters & Waters, 1970). The research by Jackson and his coworkers maintained a consistent interest and active research program in accuracy that also addressed clinical and developmental issues (e.g., MacLellan& Jackson, 1985; StrasburgerA Jackson, 1977). ACCURACY OF PERSONALITY JUDGMENT 655 cognitive psychology (e.g., Kahneman & Tversky, 1973) and quickly was extended to the study of person perception (e.g., Nisbett & Ross, 1980). Within both cognitive and social psychology, the strategy was to posit hypothetical models of the process of judgment, which were tested by examining how subjects responded to artificial or contrived stimuli. Some of these modelssuch as Bayesian statistical inference, expected utility theory, and the analysis of variance-based “Kelley cube”began to be ascribed normative status. Rather than just hypotheses about how people might think, these models were treated as prescriptions for how people should think (Gigeren/er, 199 la). Judgments that differed from these prescriptions were seen as errors, and the ease and frequency with which such errors could be demonstrated were interpreted as indicating not only that the models imperfectly described human judgment but that human judgment itself was imperfect. For a considerable period the error paradigm held the field nearly alone in the study of personality judgment. Research identified a large number of errors, described their underlying processes or heuristics, and was widely cited (Ross & Nisbett, 1991). As this work became increasingly popular, its tight focus on error produced a widespread impression that, in general, the accuracy of human judgment is poor (Funder, 1987; Lopes, 1991). The unquestioned dominance of this approach was not to last, however. By the late 1980s, the error paradigm had received its own share of criticism. Critiques of Error Modem critiques of the error paradigm emphasized three points. First, at least a fewand perhaps more than a fewof the errors identified by this research seem on close examination somewhat artificial. They are displayed in experimental cons carefully designed to evoke them, whereas their frequency and meaning in realistic cons often goes unexamined (Gigerenzer, 1991 b). Somesuch as the “fundamental attribution error,” the tendency to see behavior as caused by personal dispositions even in the absence of evidence to that effectmay, as often as not, produce correct judgments in realistic cons (Funder, 1987, 1993a;Jussim, 1993). This observation leads to a second point: Little if any research has shown that the elimination of errors actually improves judgmental accuracy. Such evidence that does exist indicates rather the reverse (Block & Funder, 1986; Borman, 1975, 1979). Eliminating the “halo effect,” for example, has been shown to make judgments of real individuals less accurate (Bernardin & Pence, 1980). This is probably because socially desirable traits usually do co-occur, making the inference of one such trait from the observation of anotherthe halo effect a practice that ordinarily enhances accuracy (Funder & West, 1993a). A third critical point is that, whatever its intrinsic scientific merits, research emanating from the error paradigm sometimes has been guilty of potentially consequential rhetorical excess. Writers have exaggerated the human propensity toward mistakes and underestimated accomplishments of the human judge (see Funder, 1993a; Gigerenzer, 1991b, 199Ic; Lopes, 1991). The result is a view of human nature that is probably unduly pessimistic. But William James’s discomfort with a one-sided strategy of “shunning error,” expressed nearly a century before, had a deeper basis than the problems identified by these modern critiques. James concluded that after one had successfully identified and eliminated every possible error in one’s judgment, one would be left withnothing! Although James was well aware of the necessity for critical examination of evidence and the elimination of error, he argued that it was necessary also to put some energy into seeking out ideas that one can believe and to have a certain sympathy for the means by which belief can be achieved. The realization that this latter approach is fundamentally different from the detection of error sets the stage for the development of the accuracy paradigm. The New Accuracy Paradigm The Roots of the New Paradigm An alternative approach to the appraisal of accuracy in personality judgment began to develop in the early 1980s. The new paradigm arose in part as a reaction to and correction of the error paradigm and some of its perceived shortcomings. It also can in an important sense be considered a rebirth of the older research on accuracy from the 1940s and 1950s. But, this time around, it has incorporated a greater methodological and conceptual sophistication aimed at helping it to avoid some of the mistakes of the past (Funder & West, 1993a; Kenny, 1993, 1994). The accuracy paradigm also has deeper intellectual roots in the work of Egon Brunswik (1956) and J. J. Gibson (1979). Both of these earlier investigators had in the field of physical perception (and in Brunswik’s case, the study of judgment) long argued for a focus on the connections between perception and reality in real-life settings. Their descendants in personality and social psychology maintain that personality judgment should whenever possible be studied using targets who are real people observed in real settings and that criteria for accuracy should be drawn from the social environment. These criteria include consensus, self-other agreement, and behavioral prediction (Funder & West, 1993a; Kenny, 1994). The Development of Research on Accuracy The very existence of accuracy. Modern research on accuracy can be seen as developing through three overlapping phases. In the first phase, which may be very nearly over, accuracy research put most of its effort into restoring the plausibility of the idea that human judgments of personality could manifest any reasonable degree of validity at all. This seemed necessary because of the impression of almost universal inaccuracy that some readers derived from research on error. In this phase, investigators reported studies showing how personality traits affect behavior and how laypersons can make judgments of such traits that manifest both interjudge agreement and predictive validity (e.g., Cheek, 1982; Funder, 1980a, 1980b, 1982; Kenrick & Stringfield, 1980; Moskowitz & Schwarz, 1982). These two kinds of criteriaagreement and behavioral prediction 656 DAVID C. FUNDER became the basis of accuracy research that followed over the next decade (Funder, 1987; Hinder & West, 1993a). Moderators of accuracy. The second stage of accuracy research, still in progress, is to discover the moderator variables that make accurate judgments of personality more and less likely. Moving beyond “Are personality judgments ever accurate?” accuracy research has begun to address the next question, “When are personality judgments accurate?” Research already has examined numerous variables that might affect accuracy. These variables have been quite diverse, but in previous writings I have proposed that they can be organized into four broad categories: (a) good judge, the possibility that some individuals might be better judges of personality than others; (b) good target, the possibility that some individuals might be more easily judged than others; (c) good trait (or behavior), the possibility that some traits (and therefore some behaviors) might be easier to judge (or to predict) accurately than others; and (d) good information, the possibility that more or certain kinds of information might make accurate judgments more likely. During the past few years, this four-part framework has served to organize numerous data analyses in our research program as well as a convenient way of summarizing principal empirical findings from our lab and elsewhere (see Funder, 1993a, for a review; see Allport, 1937, for a related framework). Recent research has steadily accumulated evidence as to the importance of accuracy for all but the first of these moderator variables. Some of this evidence will be reviewed later in this article. The process of accurate judgment. The four-variable approach to moderators of accuracy has proved itself a useful way to organize a literature that otherwise might have seemed rather chaotic. In addition, it continues to generate research. However, in the broader scheme of things, this list of moderators constitutes just a necessary beginning. Having established that judgments of personality are sometimes accurate, and having begun to document when they are accurate, the necessary third stage for accuracy research is to begin to explain how they become accurate. Such an explanation is the purpose of RAM. The Realistic Accuracy Model The first theoretical claim of RAM, that personality traits are real, has two implications. First, it implies a particular approach to the evaluation of accuracy that requires the use of an especially broad range of criteria. Second, it implies a description of the path between personality and its accurate judgment that leads through the availability, detection, and utilization of relevant behavioral cues. This description allows diverse moderators of accuracy already identified to be explained in terms of a relatively few common factors, suggests new moderators that might be found, and outlines how moderators of accuracy interact. Criteria for Accuracy A distinctive attribute of the accuracy paradigm is that it uses external evidence to indicate the degree to which a judgment might correctly reflect reality. For the study of personality judgment, this strategy becomes possible only when the stimuli to judgment are real people in realistic settings, that is, when there is a reality to be correctly reflected. All three of the basic approaches to accuracy follow this research strategy. Constructivist and pragmatic approaches. The pragmatic point of view, described by Swann (1984), is that an accurate judgment is one that allows the judge to function well in his or her social environment (see also Me Arthur & Baron, 1983). Application of this criterion requires the investigator to gather data about the judgments that subjects make in their natural social environments and to assess how well these individuals are faring. The constructivist point of view, discussed by Kruglanski (1989), is that because reality cannot be known apart from human perceptions of it, accuracy must be considered a function of the collective point of view of a community of judges. Application of this criterion requires that subjects’ judgments of actual acquaintances be compared with each other and their consensus assessed. Both of these approaches step outside of and complement the error paradigm by attending to theoretical considerations and sources of data that go beyond the cognitive process by which judgments are made. Still, compared with a realistic approach, each has the net effect of narrowing what accuracy means and simplifying how it can be evaluated. For the pragmatic approach, the judge’s social success provides a sufficient criterion, and for the constructivist approach, consensus among judges is definitive. For neither approach is truth in any broader or more abstract sense necessarily relevant; accuracy is not viewed as dependent on any properties that the target of judgment actually has. Nearly all the focus is on the judge, not the judged. The realistic approach followed by RAM, in contrast, does not consider either pragmatic success or consensus to provide sufficient criteria for accuracy. Realism requires a broader range of information that, taken together, might indicate what the target of judgment is actually like. In this way, RAM attends to the object of judgment as well as to the judge and as a result makes even greater demands on a prospective investigator. RAM’s realistic approach. RAM evaluates accuracy according to a strategy consistent with a postpositivist philosophy of science variously called critical realism (Cook & Campbell, 1979) and pancritical rationalism (Rorer, 1991). RAM’s approach also resembles the “probabilistic functionalism” of Egon Brunswik (1956). Simply put, the philosophical position underlying RAM is that although truth indeed exists, there is no sure pathway to it. There is only a wide variety of alternative pathways, each of which is extremely unsure. A great temptation under these circumstances is not to venture on the journey at all. Because truth can never be known with certainty, why concern oneself with truth? There are plenty of more tractable issues that can generate useful research.4 To echo the words of Jones (1985), one can indeed “solve the accuracy issue by bypassing it” (p. 77). This “search for the key where the light is best” is not the Jamesian solution, nor is it the approach of RAM. Instead, RAM pre4 This is friendly advice that I have received many times with respect to research on accuracy. ACCURACY OF PERSONALITY JUDGMENT 657 scribes that a portrayal of the actual psychological attributes of the target person be sought in combination with a wide range of information and the accuracy of a judgment of the target’s personality be evaluated in terms of its congruence with this portrayal. The information about the target can and should take many forms. It might include self-ratings of personality; inventory scores; ratings by knowledgeable informants; direct observations of the target person’s behavior (e.g., in a laboratory); and reports of the person’s behavior in daily life, perhaps gathered using diary or “beeper” methods (Csikszentmihalyi & Larson, 1992; Funder, 1993a; Spain, 1994). Multifaceted criteria. This multifaceted approach to the evaluation of accuracy is the fundamental attribute of RAM’s empirical program. Its purpose is to transcend some of the shortcomings inherent in simple operational definitions or other single criteria. For example, consider interjudge agreement. As Kenny (1994) noted, if a group of judges were to be perfectly accurate, they would necessarily manifest perfect agreement. Short of that, ample evidence indicates that the same variables one would expect to
