Judge of Character Article: The Ability To Judge People By Ronald Taft
(Photo Credit: Kat.B.Photography)
The Article in Full
What are the factors related to the ability to judge accurately such behavioral characteristics as the abilities, traits, action tendencies, motives, and emotions of other people? Are there some persons who consistently demonstrate good ability to judge others accurately and, if so, what are the correlates of such ability?
These are the general questions to which this review is addressed. The practical importance of the above questions in psychology is obvious, especially when we consider the role of the psychologist's personality in determining the validity of the observations and inferences with which he works.
Note: J stands for judge and S for subject. Throughout this review, the term "subject" refers to the person being judged.
METHODS OF MEASURING ABILITY TO JUDGE OTHERS
The number of differing methods of measuring ability to judge others that have been used in the experiments in this area may partly account for the varied and sometimes conflicting results found. The distinction between analytic and non-analytic judgments (Wallin, 75) appears to be a particularly important one. In analytic judgments, the judge (J) is required to conceptualize, and often to quantify, specific characteristics of the subject (S) in terms of a given frame of reference. This mainly involves the process of inference, typical performances of J being rating traits, writing personality descriptions, and predicting the percentage of a group making a given response. In nonanalytic judgments, J responds in a global fashion, as in matching persons with personality descriptions and in making predictions of behavior. An empathetic process is usually involved in nonanalytic judgments.
A Classification of Tests of Ability to Judge Others
The classification that follows is based on that suggested by Notcutt and Silva's review of the experimental approaches (54).
1. Perception of emotional expressions in photographs, drawings, models, and movies.
This method has been used to study ability to judge in a number of studies (2 [ch. 8—10], 3, 13, 18, 25, 26, 30, 33, 37, 38, 39, 73, 76, 78). The required response may be a multiple-choice, a one-word free response, or a completely free response. The criteria are usually S's intention, or the judgment of psychologists. Less controversial criteria were used by Coleman (18), where J had to select from a check list the situation to which S was responding. This type of test usually evokes a nonanalytic judgment, although, as F. H. Allport (2) has demonstrated, these judgments can be made analytically. The method has the advantage of being neat, but the expressions tend to be culturally stereotyped. In real-life situations the expression of emotions may be idiosyncratic (38, 42) and thus their recognition may require a different type of ability from the recognition of stereotyped responses.
2. Rating and ranking of traits.
This is an analytic method, and has the advantage of clear-cut quantification. It also has the virtue of requiring a performance which is frequently used in psychological work. It suffers from all the drawbacks of ratings in general, particularly the lack of consensus about the meaning of terms and the quantitative standards to be used.
A further difficulty with this method of measuring ability to judge others is the establishment of criteria. Two different approaches to these criteria may be distinguished: (a) Peer judgments, i.e., pooled judgments made by the Ss themselves, which may or may not include the self-ratings (1, 6, 17, 24, 27, 32, 60, 67, 73, 81). The use of this type of criterion suffers from the doubt whether we are measuring ability to judge or simply the degree to which J conforms to the criterion group; the nonconformist would score poorly, but might in fact be a good judge. (6) External criteria—these may be judgments made by other observers who may or may not be well acquainted with the Ss; or they may be derived from test results (3, 17, 23, 40, 52, 53, 67, 71, 73, 78). (Only Cogan et al. , Estes , and Vernon  used tests as well as ratings to provide the criteria.) Taft (67) obtained an intercorrelation of .72 between ratings using each of the two types of judgmental criteria, peer judgments and external judgments.
3. Personality descriptions.
The J is provided with some data about S and required to write a description of his personality. The data provided might be a brief interview with S, observation of S in some standard situation, or some descriptive material concerning him. This method, in general, involves analytic judgments, but suffers from the vagueness of J's task; also the criterion lacks precision, and is usually based on the opinions of persons who are arbitrarily regarded as "expert judges" (9, 12, 55).
4. Personality matchings.
In this method, J is required to match some data concerning S with some other data concerning him. Where the Ss are known personally to J, the task may be to match S with the relevant data. The method lends itself readily to nonanalytic judgments, but some Js may use analytic modes of inference to assist them to make the matchings. The matching method (23, 67, 73, 81) has the great advantage over the previously described methods of studying the ability to judge others in that its criterion is completely objective, but it has the weakness that it constitutes an artificial situation not paralleled in everyday life or in psychological practice.
5. Prediction of behavior or life history data.
The J has some acquaintance with S or is given data about him, and his task is to predict S's performance on various test items or his responses to personality and attitude inventories, or to predict specified aspects of his life history. These are the so-called "empathy" tests (20) and are probably primarily nonanalytic. A related test is the "mass-empathy" test (72) in which J predicts the combined responses of a group of people. It is suggested that the "mass-empathy" test of prediction is more likely to be tackled analytically than is the empathy test, as it does not lend itself so readily to empathizing with any particular person. Thus the empathy test will be regarded as nonanalytic, and the mass-empathy test as analytic, wherever such a distinction is cogent to our argument. It is perhaps significant in this reference that a massempathy test (Kerr) and an empathy test (Dymond) were found to be uncorrelated for 87 subjects (8). The empathy method was used in fifteen other studies analyzed (9, 10, 20, 21, 28, 34, 40 [ch. IIIC], 44, 54, 56, 57, 59, 73, 78, 79) and the mass-empathy method in ten (16, 40, 41, 66, 67, 68, 69, 70, 72, 74).
The method of behavioral prediction, like the matching method, has the advantage of possessing an objective criterion. The reliability of the predictions on any one item tends to be low, and, therefore, the test should preferably consist of a number of items or behavioral events to be predicted. One weakness of the empathy method, demonstrated by Hastorf and Bender (11, 35) is the spurious effect of projection; the predictions made by J are often partly the result of projection of his own personality, and, consequently, an accidental resemblance between J and S will render the predictions more accurate than they otherwise would be. Hastorf and Bender suggest the use of an index separating the effect of projection, similarity, and empathy. A further weakness of the "empathy" method is the possibility that judgments may often be made correctly by using cultural stereotype responses without attempting to predict the responses of the particular S. Gage (28) has demonstrated how accuracy of stereotype predictions can be kept separate from accuracy of individual predictions.
A few miscellaneous techniques should also be mentioned, although studies based on them, in general, fall outside the scope of this review. Most of these techniques employ indices that attempt to measure the ability to judge others in an indirect manner, e.g., Dymond's empathy index on the TAT (17), Walton's generalized empathy test (76), and McClelland's "role-playing ability" scale (46). Even more remote are the tests devised by Chapin (15), Moreno and Moreno (49), Sherriffs (61), Moss Social Intelligence Test (50 used in 73). Most of these measures are either assumed to possess face validity or have been validated against one of the first five methods of testing ability to judge others.
IS THERE A GENERAL ABILITY TO JUDGE OTHERS?
Is there sufficient consistency in the ability to judge others for persons to be characterized as good or poor judges? We are here concerned with the generality and specificity of the ability to judge others and with testretest reliability of the measures used. G. W. Allport suggests that the ability to judge others is analogous to artistic ability in that it is neither entirely general nor entirely specific. "It would be unreasonable, therefore, to expect a judge of people to be uniformly successful in estimating every quality of every person. . . It seems more of an error, however, to consider the ability entirely specific than to consider it entirely general" (5, p. 512). Let us now look at the experimental evidence which, on the whole, supports Allport's contention.
Consistency Between Different Types of Tests
Some persons may be better at judging others on analytical tests, while others may be better on nonanalytic tests. F. H. Allport (2) found that, in judging emotional expression (Rudolph poses), some Js were superior at judging the intended emotion when using a naive type of intuitive method, while other Js were superior after receiving a training in the use of analytic methods of making the same judgments. Using 44 measures of ability to judge, Vernon (73) obtained significant correlations of over .30 for S's ability to rate strangers (analytic) and his performance on subtests of the Moss Social Situations Test involving nonanalytic judgments about people (Social Situations and Observation of Human Behavior). Wedeck (78) obtained significant positive correlations ranging from .18 to .56 (mean .31) between a test of ability to rate the personality traits of verbally described persons and seven other tests of ability to judge emotions and personal qualities of persons depicted in various ways, both pictorial and verbal. The Ss consisted of 203 adolescent school girls. A factor analysis revealed a "psychological factor" with a high saturation in the judgment of emotions tests (nonanalytic) but a negligible one in the trait-rating tests (analytic). Wolf and Murray (81) using four Js found a "fairly consistent" rank order between the accuracy of J's predictions of the 15 Ss' rank order on three objective tests and J's ability to match the Ss with their Dramatic Production Test records. Taft (67) obtained a significant correlation of .36 between the ability to rate others on traits and the ability to predict group responses on an inventory. On the other hand, a test of ability ,to match the Ss with their mosaic productions did not correlate significantly with these two analytic tests. Neither did an empathy test with a mass-empathy test (see p. 3 supra). In general then, significant but low consistency has usually been found between one test of ability to judge others and another, but some studies suggest that analytic and nonanalytic tests tend to differ in their results. We shall therefore need to note in our review the type of test used in the particular experiments reported.
Reliability of Tests
Adams (1) reports an average testretest reliability of .55 for the accuracy of sorority girls in ranking nine of their "sisters" on 63 personality traits, the interval between the two tests being approximately three weeks. The criteria were the pooled rankings. This index of reliability indicates that there is some consistency for a particular judge in making specific judgments, but that the consistency is not high. If this is typical, the influence of attenuation on correlation coefficients using such data would be considerable. In fact, only a few of the studies report even moderately high reliabilities. Dymond (21) reports a split-half reliability for her empathy test (predictions of 5's ratings of himself and of J on six characteristics) of .82 and a testretest reliability after six weeks of .60 (20). Travers (69) asked his Ss to predict the percentage of a specified population who would answer "true" or "false" to each of 25 items, and the split-half reliability of the accuracy scores for this test was .64.
Taft (67) obtained a test-retest reliability coefficient of .82 for a 30- item prediction test similar to that of Travers but this contrasted with a split-half reliability of only .20. Correlations between single items tend to be even lower than .20 (67, 69, 74), a further indication of the low reliability of any one item. We thus see the necessity for tests of this nature to include a large number of items in order to ensure reliability.
Consistency in Accuracy Between Traits and Between Subjects
Travers, in a further experiment (70), found correlations of .44 and .47 respectively for two groups of students (N, 26 and 31) between ability to judge the word knowledge of S's own group and ability to judge that of the general population. "The evidence then indicates that some subjects are generally good at judging what various groups of men know, while others are poor" (70, p. 98). Taft's study yielded a corresponding correlation of .31 for 40 Ss on tests analogous to those of Travers.
Vernon (73) did not compute the reliabilities of the tests which he used, but, on the basis of "logically related aggregates," i.e., clusters, he suggests that his data show four independent dimensions—ability to judge self, to judge acquaintances, to rate strangers, and to judge character sketches of strangers. Kelly and Fiske (40, ch. IIIC) attempted to devise a test of the ability of psychologists to predict inventory responses of two patients whom the latter had diagnosed. The respective accuracies achieved in judging each patient intercorrelated .23 for 100 Js. In contrast to this low consistency, Estes (23) reports that his ten best judges in a test of ability to rate the traits of persons depicted in short movies were consistently more accurate than the ten poorest judges on all 23 variables and for all eight Ss. The criteria of accuracy were the ratings assigned by clinical psychologists.
Gage (28) also reports high generality in the ability of his Js to judge the responses of six strangers on the Kuder Preference Inventory after they had directly observed S's expressive behavior for a short time. He found high consistency in J's ability to judge irrespective of the specific items or Ss on which he was being tested. Consistency (correlation of .71) was also shown between ability to predict the responses of the six strangers and those of three randomly selected classmates. Similar findings, with regard to consistency in J's accuracy between different Ss are reported by Bender and Hastorf (11).
Luft (45) suggests that there might be individual differences in consistency amongst Js. In one of his experiments using 74 Js, all persons with various degrees of training in psychology, he found that 27 per cent of the Js were able to predict the personal inventory responses of both of two Ss with above chance accuracy. The predictions were made on the basis of case-record material. However, the accuracy scores of the total sample of 74 Js did not correlate significantly from one S to the other. These findings may be taken in conjunction with those of Dymond (20) in her experiment referred to above. She found that the capable judges tended to show less variation in the accuracy with which they could judge the S's self-ratings on each of six traits than did poorer judges. Thus, good judges, at least, seem to show some consistency in ability to judge irrespective of the type of S or the type of qualities being judged. We could not expect, however, that even a capable judge would be able to judge members of another culture as well as he can judge members of his own; there is evidence that judgments are more accurate when J and S are similar in cultural backgrounds, also in age and sex (34, 36, 67, 71, 81). Nor could even a good judge be expected to judge all areas of the personality or use different kinds of "clinical evidence" equally well. (Space does not permit the inclusion of a systematic review of the experiments that have been performed on these specific factors in judging [67, pp. 12-23]).
Allport's dictum at the beginning of this section seems to be justified by the data. A reasonable conclusion might be stated as follows: the degree to which a person can make accurate judgments about others is a function of his general ability to judge and of specific situational and interactional factors, but the greater his general ability to judge, the less will be the relative influence of the specific factors. The specific factors concerned are the type of S, the relationship between J and S, the type of judgment demanded, the traits being judged, and the material available to J. In addition, the consistency in the performance of any individual J is limited by the low reliability that characterizes many of the tests used to measure ability to judge. In view of the effect of specific factors we must be cautious in drawing general conclusions from studies that do not require J to make a variety of judgments about a variety of Ss.
THE CHARACTERISTICS OF GOOD JUDGES OF OTHERS
In consequence of the above conclusions, some of the findings quoted in the review that follows should be limited to the actual Ss, judging situations, traits, or modes used in the particular experiments. Wherever similar findings are obtained over several situations, we may expect that the individual characteristics of the good judges concerned can be generalized, but where contradictory findings are obtained between one study and another the variation may merely be due to the specific factors involved in the experiments. Our plan has been to report findings as conclusions where the weight of the evidence clearly supports them, even though some contradictory findings may raise doubts.
Gates (30) found a progressive increase from the ages of 3 to 14 years in the ability to judge the intended emotional expressions in six of the Ruckmick poses. Walton (76), using various tests of empathy with emotional expression, obtained similar results for Js ranging from kindergarten to university age. His published tables seem to suggest that the greatest improvements occur between 9 and 11, and between 14 and adulthood. Further support for the development of empathy in children is provided by Dymond et al. (22) where a marked increase in signs of empathy is reported between 7 and 11 years. The test consisted of predictions by the Js of whether their classmates liked or disliked them.
On the other hand a number of studies using adults as Js failed to find any increase with age in ability to judge others (16, 23, 41, 56, 64, 67). These studies employed Js ranging from 18 to at least the late thirties, but the tests mostly tested their ability to predict the responses of college age Ss. Consequently, the older Js would be less similar to their Ss than would the younger Js and the former may therefore have been under a handicap.
While ability to judge emotional expressions increases with age in children—probably through experience—no increase with age has been found in adults on various tests of ability to judge others. This latter finding, however, is subject to the limitation that the subjects being judged have mostly been closer in age to the younger judges.
It has been contended that "Experimental studies, so far as they go, establish only a slight margin in favor of women" (5, p. 517). This seems to be a reasonable summary of the many studies that have been conducted on sex differences in ability to judge people. One study (37) reports a significant superiority for women students in judging the emotional expressions in the Rudolph poses, but in the other studies of judging emotional expressions in photographs, models, and movies, no significant or consistent sex differences are reported. In five studies (2, 18, 25, 30, 33) the authors report no differences; in three (13, 26, 39) a slight, but apparently insignificant, superiority for women; and in one (38) a slight superiority for men. The results are not related to the sex of the person whose emotions were being judged.
Similar results have been found in other tests of judging ability. Dymond (20) reports no sex differences at first in the ability of students to predict how other members of their group rated themselves and J; however, in a second administration, six weeks later, the female Js were significantly better than the males. In a further experiment (21), she found female students significantly superior to males in making judgments similar to those in her first experiment. These two studies differed from most of the other ones on sex differences in that the Js were well acquainted with the Ss. However, where husbands and wives predicted each other's responses to an inventory (54), sex differences in accuracy were not significant; if anything, the males were the more accurate.
None of the other five reported experiments (46, 56, 64, 69, 71) which compare the ability of males and females either to rate their 5's traits or to predict their responses found significant differences. It is also interesting to note that no relationship was found between the ability of males on these types of tests of judging and their scores on inventory and projective tests of femininity (67).
The weight of evidence is in favor of no sex difference in ability to judge, or perhaps a slightly superior ability in women. The reported studies do not suggest any convincing explanations for this possible superiority of women, and until further evidence suggests a changed view, it would be wise to conclude that there are no differences.
Family Background and Sibling Rank
In a study of ability to predict how other boys would answer personal inventory questions (mass-empathy), Sweet (66) found a positive relationship between judging ability and the socioeconomic status of the Js. These results, however, could have been due to the influence of intelligence. The only other study reported on this topic, Taft (67), found no correlation among graduate students between ability to judge and socioeconomic status, where the test required J to rate S on six traits, and to predict their personal inventory responses. The criteria for the ratings were both peer and psychological assessments.
Regarding size of family and order of birth, an earlier suggestion by F. H. Allport (2) that Js from larger families are better at judging emotional expression has not been supported. (Allport used only 26 Js and does not report significance.) Estes (23) found no relation between "sibling status" and ability to judge on a trait-rating test and in Taft's study (67) there was actually a negative correlation between ability to judge and number of siblings. In this latter study it was also found that the best Js were only children; then came eldest, middle, and youngest children in that order. The explanation given is that the fewer the older siblings, the more likely is a child to be brought into contact early with adult modes of judging others.
Taft also found that those Js who had a rural background were poorer at judging than the urban ones but this might be explicable again in terms of sibling status.
The findings reported on the influences of background factors on ability to judge are derived from very few studies, and almost entirely from one. Thus no consistencies can be educed until we have more evidence.
Intelligence and Perception
G. W. Allport sums up the studies up to 1937 on the relationship between ability to judge others and intelligence as follows: "Experimental studies have found repeatedly that some relationship exists between superior intelligence and the ability to judge others . . . even within a high and narrow range of intelligence. . . . Understanding people is largely a matter of perceiving relations between past and present activities, between expressive behavior and inner traits, between cause and effect, and intelligence is the ability to perceive just such relations as these" (5, p. 514).
In an early study, Cogan et al. (17) found a significant correlation between J's intelligence and his ability to rate the intelligence of others, and also to rate sense of humor. The criteria were pooled ratings. The correlations between intelligence and ability to rate seven other personality traits were not significant. Adam's study (1) was similar but more complex. He studied 80 female students living together in groups of 10. They ranked themselves and each other on 63 traits, including seven related to intellectual functioning; the pooled rankings acted as the criterion both for measuring the accuracy of the judgments and for measuring the characteristics of the Js. The seven intellectual characteristics had an average correlation of only .12 with ability to judge others, the highest correlations being .21 for "observation" and for "mentally bright" (significant at 5 per cent level). We should note, however, the possibility that the use of peer ratings as the criterion may have provided a measure of J's conformity to his group, rather than of his ability to judge his peers.
Where some independent measure of the criterion was used (21, 55, 67, 73), the accuracy of analytic judgments of personality traits showed significant positive correlations with intelligence fairly generally. Vernon obtained significant correlations of approximately .30 between intelligence and various analytic measures of ability to judge strangers. For other judgments, however, involving nonanalytic modes, the correlations obtained with intelligence were approximately zero.
Taft (67) scored the intelligence of his 40 Js according to their best performance in any of a number of various cognitive tests. This index of intelligence correlated .37 with their ability to judge the traits of their peers and to predict their peers' responses to an inventory (analytic). Academic ability also showed low positive correlations with these tests of ability to judge others but the sample was highly selected in this regard (senior graduate students). Wedeck (78), using 203 girls, reported saturations of g factor ranging from .18 to .34 for his seven varied tests of ability to judge others. There were no consistent differences in this respect between the analytic and the nonanalytic tests.
The results of other studies correlating intelligence and ability on nonanalytic modes of judging are contradictory. In Sweet's study (66), 12- to 14-year-old boys at a YMCA camp predicted how their peers would respond to a questionnaire on 22 different activities. There was a "high relationship" between their accuracy and their CAVI scores. The assessees in the OSS procedures (55) wrote personality sketches of their peers and these were rated by the assessment staff for accuracy. Despite the admitted unreliability of the criteria, the assessees' ability to judge correlated significantly with various verbal intelligence tests (.32 to .48) and .54 with the final staff rating on Effective Intelligence. The latter rating, however, was contaminated both with the scoring of the judging test and with the intelligence test results.
The only other study found that reported a significant positive correlation between judging ability and verbal intelligence (.21) was that of Kanner (38) which correlated the ability of students to judge the intended emotions in the Feleky poses and their Thorndike intelligence scores. Dymond (21) reports a positive correlation between her Js' ability to predict the S's self and other ratings on six traits and Js' scores on the Wechsler Performance Scale. However, a further study repeating the same technique (43) found no correlation with the ACE Group Test. Similarly Kelly and Fiske (40) found a significant correlation between ability to predict inventory responses and score on the fluency subtest of the PMA but not with any of the other subtests or with the Miller Analogies test.
All the other studies correlating nonanalytic modes of judging with intelligence found no correlation, for example, between (a) the accuracy of predictions of the questionnaire responses of eight psychology students and the Js' ACE scores (9), (&) ability to "throw" scores on a masculinityfemininity test and Thorndike "intelligence score" or grade point ratio (41), (c) Army Alpha scores and the ability of Js to match correctly a short behavior record with personality sketches (23), (d) ability to predict group responses and the Otis, Cooperative General Culture Test, or Thorndike Reading Test scores of the Js (69, 70), (e) various intelligence tests and the ability of graduate students to match mosaic productions with acquaintances (67), and (e) various group intelligence tests and ability to predict inventory responses on the basis of a brief observation of 5"s expressive behavior (28).
There seems to be a positive relationship between intelligence and ability to judge others analytically. The highest correlation reported is .55 but the use of intellectually homogeneous groups in most of the studies would have reduced the correlations obtained. The superiority of more intelligent Js is probably the most pronounced in ability to rate the intelligence of others. The results for analytic modes do not appear surprising as such modes require a precise understanding of the meaning and application of abstract terms (traits).
Nonanalytic modes of judging tend to manifest lower correlations between intelligence and accuracy of judgment. It is possible that accurate nonanalytic judgments of others are more a function of good perceptual and judgmental attitudes than of the use of abstract intelligence, provided the mode of making the judgments is clearly within the level of comprehension of J.
Training in Psychology
It is often assumed that qualified psychologists are more capable than laymen of making unbiased judgments, since they receive training in the dynamics of personality and also in the correct manner of making judgments, e.g., using fixed standards, considering only relevant evidence, avoiding projection, combining probabilities in their correct weight, etc. On the other hand, some writers argue the opposite. For example, Murray (51) claims that the use of analytic perception and induction together with the repression of emotion and feeling leads to poor ability to judge others.
Let us now turn to the experimental evidence on the question. Are those who have taken courses in psychology more accurate judges than those who have not? In a test of judging emotional expressions in the Boring-Titchener models on an adjective check-list, students who had completed at least one course in psychology were, if anything, less accurate than students just taking their first course (13). Hanks (34) found no relationship between training in psychology and the ability to predict S's answers to inventory questions from biographical and other inventory data. Polansky (56), on the other hand, found graduate students in psychology to be better judges than those without psychological training when judging Ss who were friends of the author. Whether the Ss, as well as the Js, were also trained in psychology is not stated. The tests required Js to make specific behavioral predictions about three Ss whose actual behavior was known.
Studies comparing the ability of professional psychologists with nonpsychologists do not, in general, suggest that those trained in psychology are better judges. If anything, the contrary seems to be the case. In Estes' experiment (23) psychologists were significantly poorer than the average of a wide variety of Js (professional and nonprofessional persons) in judging others on ratings, check lists, and matching tests. The material presented consisted of brief samples of S's expressive behavior recorded in a movie. These findings are supported by Wedell and Smith (79) who compared qualified, experienced clinicians with untrained, inexperienced interviewers on their ability to predict the responses of 200 interviewees to an attitude questionnaire. The untrained Js made the more accurate predictions.
Luft (44, 45) compared the ability of clinicians (psychiatrists, psychologists, and social workers), graduate psychology students, and physical science students on a series of tests in which they were required to predict the responses of individuals to objective and projective test items. The physical scientists were superior to all the other groups of Js on the tasks taken as a whole. On the other hand, Taft (67) found psychologists to be superior to graduate students in various other disciplines. In a study of ability to rate traits and to predict the inventory responses of graduate students participating in an assessment program, judgments made by the assessment staff (psychologists) were more accurate than similar judgments made by the assessees themselves (both staff and peer ratings being used as criteria for the ratings). The assessees did not include psychology graduate students, but the assessment psychologists were "definitely" superior to social science students and "probably" superior to physical science students in judging. In this latter respect, the results appear to contradict those of Luft, but there was some evidence (not reported by Taft ) that the superiority of the psychologists in Taft's study was due primarily to the accuracy of the experimental psychologists rather than the clinical. In Luft's study the psychologists were clinicians.
Let us now consider some studies in which Js with different degrees of training in psychology are compared. Kelly and Fiske (40) did not find advanced graduate students of clinical psychology any more accurate at predicting the personality inventory responses of patients whom they had diagnosed through normal psychological techniques than similar students with one year less training. Kelly and Fiske were also able to compare the validity of the assessments of the professional promise of clinical psychology trainees made by other graduate students in clinical psychology with similar assessments made by professional psychologists and psychiatrists. The judgments were made on the basis of test protocols and interview reports and the Ss were actually unknown to the Js. The validity of the assessments was determined by S"s later performance as a clinical psychologist. The authors conclude that the students "utilized the materials as effectively as the more mature 1947 staff [psychologists]" (40, p. 175). However, the students had certain advantages over the psychologists—they had greater similarity to the Ss and had themselves undergone the experience of assessment; also, being in training themselves, they had a more appropriate frame of reference for interpreting the criterion (as the authors point out). These factors are less likely to explain Soskin's finding (62) that experienced clinical psychologists were not more accurate than graduate students in predicting the inventory responses of a 26-year-old mother, from her projective test protocols.
Professional psychologists were more successful than psychology students in correctly diagnosing the psychiatric patients depicted in the Szondi Test (57). The superiority, however, could well be due to the psychologists' greater familiarity with the diagnostic categories used in the test. Technical knowledge could also account for the clinicians' superiority in Luft's experiment referred to above. Luft found that the clinicians were significantly superior to psychology students in predicting two Ss' responses to a projective test (but not to an objective test) after they had been provided with written interview data about the Ss.
The results on the comparative ability of nonpsychologists, psychology students, and professional psychologists to judge other people are partly obscured by the effect of similarity in age and academic status between J and S. Attempting to allow for this effect in the reported results, physical scientists, and possibly other nonpsychologists, e.g., personnel workers, appear to be more capable of judging others accurately than are either psychology students or clinical psychologists. There is a suggestion that experimental psychologists may be superior as judges to clinical psychologists, but this conclusion must await further evidence.
There is also evidence that suggests that courses in psychology do not improve ability to judge others and there is considerable doubt whether professional psychologists show better ability to judge than do graduate students in psychology. Do these findings necessarily lead to the conclusion that training in psychology blunts a judge's ability? Actually, the results could be wholly attributed to selective factors operating (a) in the accreditation of graduate students as psychologists and (b) in the selection of psychology as a career. On this latter point, there are some interesting but unanswered questions. Perhaps those taking up psychology, especially clinical psychology, are too concerned about social relations (see below, Attitude Toward Social Relations) to be good judges, or perhaps they have had insufficient experience with a wide range of people. Watson (77) has pointed out that many professors and clinicians tend to live in isolation from the general life experiences of the people whom they are endeavoring to understand.
Esthetic Ability and Sensitivity
G. W. Allport claims that of all the characteristics of a good judge of others "esthetic" ability stands above the others, and can even compensate for such things as lack of intelligence and experience (5, p. 538).
Allport and Allport (3) reported that the only correlation found for "susceptibility to social stimuli" (ability to judge emotional expression— Rudolph poses?) was with artistic ability. Of the nine best judges on the test, seven had literary productions to their credit. Unfortunately, no comparable figures are given for those low on the test. Vernon (73) found that the more accurate raters of strangers and of acquaintances (analytic) tended to be more artistic according to their scores on various musical and art judgment tests and were rated high on esthetic values by their peers. Ability to match character sketches with case history material did not correlate with esthetic judgments. Bender (9) reports on a. test of ability to write accurate descriptions of the Ss on the basis of test profiles. The author acted as a referee for accuracy. He states that all the good judges were interested in literary or dramatic activity but again no statement is furnished regarding the poor judges in this respect. On another test by Bender in which J was required to rate S's standing on various opinion scales, ability to judge did not correlate significantly with the Meier- Seashore Art judgment test. Estes (23), however, found clear evidence that good ability to match character sketches was related to painting and dramatic avocations but not musical.
Taft (67) concluded that the ability to rate traits accurately and to predict the Ss' inventory responses (analytic) correlates positively with "simple, traditional, artistic sensitivity." This was measured by art judgment tests similar to those used by Vernon and Bender in which high scores are allotted for ability to follow traditional artistic rules. On the other hand, the same Js' scores, on a test in which their preferences for various patterns were compared with the preferences of actual artists, correlated negatively with analytic judging ability. None of the artistic tests correlated with performance on a nonanalytic test (matching). There was also zero correlation between all the tests of ability to judge others and an index of dramatic ability, i.e., ratings of the Js' ability to empathize with roles in a role-playing test.
The ability to judge others seems to be higher in those persons who have dramatic and artistic interests, but the relationship is not as clear-cut in the case of dramatic and artistic ability.
The evidence on dramatic ability is too limited as yet, but in view of role-playing theory, this would seem to be an important area for study, particularly in its relationship to the ability to make accurate nonanalytic judgments of others.
The conclusion on artistic ability must turn on the definition of such ability; while there may be a positive correlation between ability to judge others on analytic modes and ability to endorse traditionally accepted esthetic rules this may possibly be a function of intelligence and interest. The above relationship is not found where the sophisticated standards of professional artists are used and it might well be that professional artists, if tested, would perform poorly on tests of ability to judge others. The accuracy of nonanalytic judgments seems to be even less dependent on artistic ability—with the possible exception of literary ability— than are analytic judgments. (This is explicit in Taft and implicit in Vernon.)
Emotional Stability and Character Integration
It may be argued that the welladjusted person is less subject to projecting himself into others than a poorly adjusted person and therefore he is able to judge them better. However, it is also possible to argue that a poorly adjusted person, who at the same time is aware of his emotional difficulties, is more sensitive to similar difficulties in others. Most of the findings in this area indicate that the first argument is the more correct, but the evidence is not unequivocal.
Where the tests of ability to judge are analytic in nature, the studies have reported a significant positive relationship between accuracy and emotional adjustment. The only possible exception is that of Adams (1) who reports that his good raters of personality traits tend to be rated high by their peers on the following: touchy, lack courage, work for present, independent, talkative, egotistic. These are signs of poor superficial adjustment, but do not necessarily point to a fundamental maladjustment. The other experiments reveal a positive relationship between analytic tests of judging and (a) the Bell Adjustment Inventory (69, 70), (b) the Character Education Test (66), (c) teacher's ratings that children need psychological help (32), (d) steel-workers' accident-proneness (63), (e) the California Authoritarian Scale (59), and (f) ratings of graduate students by their faculty on "personal soundness" (67). The latter study also reported a positive correlation between ability to make accurate analytic judgments and the "psychotic" scales on the MM PI (Pa, Pt, Sc).
The relationship between various measures of adjustment and nonanalytic tests of judging are more equivocal. Studies using the MMPI support the results quoted above: the ability of clinical psychologists to predict the inventory responses of their patients correlated negatively with their scores on the Pt scale (41); good judges (undergraduates) on Dymond's empathy test (43) tended to be low on Pd, Pa, Pt, and Sc, and psychology graduate students who were rated high by their peers on "role-playing ability" (nonanalytic?) also were low on these latter scales and on the D scale (46).
Emotional adjustment as measured by the Bell Scale correlated positively with ability to match the Ruckmick and Frois-Wittman emotional expression pictures (26) but not with ability to predict inventory responses after observing S's expressive behavior (28). A test of matching a short film of expressive behavior with character sketches showed zero correlation with the Bernreuter (23) whereas matching mosaic productions with the acquaintances who produced them showed a significant negative correlation with ratings on "personal soundness" (67).
Ability to judge others on analytic modes correlates positively with emotional adjustment; presumably the more psychologically significant aspect of this correlation is that poor judges tend to be poorly adjusted, and therefore, probably more likely to allow personal biases to affect their judgments. Studies are still needed in this area using Ss who are more heterogeneous on emotional adjustment than the highly selected groups used in most of the experiments reported. Such studies would perhaps throw light on the connection between poor adjustment and poor ability to judge. The need for further studies is even more evident in the case of nonanalytic modes of judging in which the evidence is more contradictory. There is, however, a clear trend for the MMPI on both analytic and nonanalytic judgments —the poorer judges tend to be elevated on the psychotic scales, in particular, "psychasthenia."
Is there any relationship between having a good knowledge of oneself and being an accurate judge of others? The inference that there is such a relationship seems reasonable if we accept the theory that we learn to know ourselves by our acceptance of the attitudes of others toward ourselves, and that we learn to know others by observations and inferences deriving from our introjection of the behavior of others. According to this formulation the acquiring of selfknowledge and knowledge of others are indispensable to each other. Writers who have stressed this viewpoint include Mead (48), Sullivan (65), and Cottrell (19).
Before looking at the experimental evidence on this question, we should refer briefly to a difficulty involved in handling the concept of insight experimentally. For experimental purposes self-insight is usually defined operationally as the ability of J to judge himself accurately, using as the criteria of accuracy judgments made by some other persons. However, it is possible, especially in the case of persons with complex, recondite personalities, that the self-judgments may be more veridical than those made by others. The use of tests, objective and projective, as the criteria can somewhat ameliorate this difficulty (e.g., 73). Other steps that might be taken are: (a) to use, as criteria, judgments made by "experts" (eg., 53) denned either as persons who know the person well or as professional psychologists—although as we have seen there is considerable doubt whether these latter Js are, in fact, more expert than lay Js; (&) to require S to judge himself as he predicts he will be judged by the other Ss, and then to use these latter judgments as the criteria (e.g., 20)—this method has the disadvantage of contaminating judgments of self with judgments of others; and (c) to use over-all clinical judgments of the person's "self-insight" (e.g., 73, 80).
In many studies it has been found that the Js tend to rate themselves high on admirable traits and low on reprehensible ones (3, 17, 32, 60, 67). Consequently, those who are actually high on admirable or low on reprehensible traits will tend to be scored higher than others on self-insight. This artifact operates in all studies of self-insight, no matter how measured, and could affect the relationship found between self-insight and ability to judge others.
Let us now look at the results of experiments in this area:
1. Do good judges of self possess the same personality characteristics as good judges of others?
Adams tested the ability of girls to rank themselves and acquaintances on 63 traits, using the pooled rankings as the criterion (1). He concluded that good judges of self and good judges of others possess quite different personality syndromes—strangely enough, the former were oriented towards society, whereas the latter were more egotistic. Compared with the good judge of others, the good self-judge is more intelligent, possesses the more desirable emotional attributes, and is much more socially minded. Vernon (73) obtained substantially the same findings as Adams, using ratings by the Js of themselves, friends, and strangers. He used both pooled judgments and tests as criteria, but note the weakness that these tests were mainly of the inventory, i.e., self-rating, type. When he grouped the intercorrelations between his data into "logically related aggregates," self-ratings formed a different cluster from ratings of others. The good self-raters tended to have a good sense of humor and good abstract intelligence and were sociable, whereas the good judges of others were less sociable and intelligent, but more artistic than good self-raters.
The Ss (graduate students taking part in a "living-in" assessment procedure) in Taft's experiment (67) predicted their sociometric standing, their relative status in the group on inventory item responses, and also how their peers would rate them on six traits. Good judges of self were taken as those assessees who were accurate in making these predictions. The assessees were rated by the staff on the Gough check list of 279 descriptive adjectives, and the characteristics of good judges of self were thus compared with the characteristics of those assessees who were good judges of the traits and the inventory responses of their fellow assessees. Sixteen of the adjectives distinguished good judges of self from good judges of others, at the 1 per cent level of confidence. Similarly, ten adjectives distinguished the poor judges of self from the poor judges of others. The qualitative differences confirmed Adams' and Vernon's finding that good judges of self tend to be more sociable than good judges of others.
2. What relationship has been found between self-insight and ability to judge others?
In the most simplified instance, where the correlations are based on J's ability to judge himself and his peers on single traits, a high positive relationship has been found consistently. Traits on which this relationship has been established include both socially approved and socially disapproved characteristics as follows: beauty (58), leadership (32), empathy (19), obstinacy and disorderliness (60). Evidence has also been presented demonstrating that Js possessing disapproved traits but lacking insight into them tend to project these traits into persons whom they are judging more than do other Js who possess the same traits but do not lack insight (60, 80), Where self-insight has been measured over a number of traits, or in a "global" fashion, the results quoted are conflicting. For example, in a further experiment, Vernon used peer ratings of his Js' self-insight as the measure; this measure correlated significantly with the objective measure of ability to rate self accurately described above, and correspondingly these peer-ratings showed no correlation with various measures of ability to judge (ability to rate traits, to make predictions about the behavior of friends and strangers, and to match character sketches with case study material). Similarly, Frenkel- Brunswik (27) found "no consistent relationship" between three psychologists' self-insight and their accuracy in rating adolescents on the Murray needs. The criteria for measuring accuracy of both self and other ratings were the judgments made by the other two psychologists. Taft (67) also failed to find a significant correlation between Ja' scores on the overall index of ability to judge others (rating the Ss on six traits and predicting their responses to inventory items) and the index of ability to judge self, described above. On the other hand, Norman (53) did obtain significant correlations in a very similar study, using as Ss 72 graduate students in psychology who were taking part in an assessment program. Their over-all ability to judge themselves on 31 personality traits correlated positively with their ability to judge their peers on the same traits, both when peer ratings and when staff ratings were used as the criteria.
The reason for the clear-cut differences in the results of the above experiments is not obvious. One difference between Norman's study and the others is that, in his experiment, ability to judge self and others is measured by J's accuracy in rating himself and his peers on the same list of traits. In Vernon's and Taft's studies the measure of ability to judge included more than tests of ability to rate self and others on a list of traits. Frenkel-Brunswik's study differs in that the Js were not rating their peers.
Taft's results also suggest one subtlety not mentioned by Norman; a chi-square analysis revealed a significant relationship that would not show up in either a linear or a curvilinear correlation: dividing the Js into three groups on the basis of their accuracy, it was found that whereas good judges of others tended to be average on ability to judge self and average judges of others tended to be good judges of self, poor judges of others tended to be either good or poor judges of self (rather than average).
Persons who show insight into their own status with respect to their peers on individual traits tend also to rate their peers accurately on those traits. However, when over-all indices are obtained of the subject's self-insight and of ability to judge others, using a variety of tests of these abilities, the relationship is not so clear-cut.
Good judges of self have been shown to possess a number of traits that differentiate them from good judges of others: in particular, greater social orientation. A more mature consideration is required in future work on this topic of the extent to which motivational forces that cause J to be accurate in his judgments of himself also cause him to distort his judgments of others.
Social Relations Social Skill and Popularity
Common sense would suggest that a person who possesses good ability to judge people is able to use this ability to advantage in situations requiring social skill, e.g., in leadership or salesmanship. It may be argued that ability to empathize with others and to play their roles, particularly in their relationships to oneself, is positively related to social skill. While there is some evidence to support this viewpoint, there is sufficient evidence to the contrary to force us to seek a more sophisticated attitude. In a study related to the area of role playing and social judgment, Moreno and Moreno (49) found that those children who were able to perceive social roles more accurately than others (according to the psychologist's judgment) were not necessarily those who could enact them best.
Vernon (73) found that the ability to judge strangers is related positively and significantly to scores on the Social Situations and the Memory for Names and Faces subtests of the George Washington Test of Social Intelligence, and negatively to scores on the Observation of Human Behavior subtest. In the OSS study (55), however, scores on the Judgment of Others Test (ability to write accurate personality descriptions of one's peers) had such a low correlation with the staff ratings of the candidates on Social Relations that it was discarded as a measure of this trait. Taft (67) used several tests and ratings to measure the social skills of his subjects; in a role-playing test the good judges of others were rated significantly lower on "ingenuity" than were the poor judges; there were no differences between good and poor judges in the ratings made by the assessment staff on "persuasiveness" in discussion nor on "likeability" as defined by peer sociometric choices. Both analytic and nonanalytic judgments showed the same pattern.
Norman, in the Clinical Psychologist Assessment Study described above, also failed to obtain a linear correlation between ability to judge and a sociometric measure of the acceptability of his Js as professional co-workers. There was, however, a curvilinear relationship; those who received a medium score on acceptability were significantly better at judging than those who received a low score on acceptability. There is a hint here that a study which included more socially maladjusted persons than is usually found among Js in these types of investigation would reveal the expected positive correlation between social skill and ability to judge others.
Studies of the ability to predict group responses have found almost without exception a positive relationship between ability to judge and popularity; also good leaders and salesmen tend to be good judges. The study by Chowdhry and Newcomb (16) finds that those chosen as leaders in a sociometric test are good judges of group opinion only on topics relevant to the group. Ability on the "mass empathy" test (Psychometric Affiliates) correlates positively with various measures of the popularity and efficiency of union officials (72), with the ability of new car salesmen to sell (68), with the sociometric scores of steel workers as working companions (63), and with sociometric ratings of students on leadership after a half-hour discussion (8). In this latter study the Dymond test also correlated significantly with these ratings on leadership. Gage (29) reviews several studies (by Gage and Suci, Wood, and Sprunger) that confirm the positive relationship between the degree to which a group accepts J and his ability to predict the responses of this group to various opinion items. The one exception quoted is a Naval Leadership Project study which failed to find a correlation between the popularity of leaders on a ship and their ability to predict their crew's opinions on an attitude scale. Gage also found a positive correlation with popularity among high school students when the item responses were predicted by the Js for each individual S separately instead of for the group as a whole (28). Conclusions. The ability to predict how Ss will respond to opinion items shows a consistent positive relationship with measures of social skill, such as leadership, salesmanship, and popularity. This relationship would follow logically from the probability that these types of social skill are aided by the ability to predict how people will behave. Unfortunately, this relationship could also be due simply to a combination of J's tendency to project his own responses onto his S and his being in fact similar to S in his responses. Until these factors have been more carefully isolated, it is impossible to know which explanation is the correct one. Recently a method has been proposed (11) for separating the effects of projection, similarity, and empathy in such tests of judging ability. Other tests of ability to judge others, e.g., rating traits or matching expressive behavior, do not show this same consistent relationship with social skills.
Attitude Towards Social Relations In addition to considering the ability to judge others as a social "capacity" it should also be considered as an "attitude." An interest in social relations does not, however, mean good ability to judge the personality of others.
Adams working with the judgment of traits (see above) found that the good judges were independent but "talkative." "The good judge tends towards the egotistic . . . is cold-blooded towards others and not interested in them . . . he develops a shrewd ability to measure others, not as human beings, but as tools" (1, p. 181). This is consistent with the findings already mentioned that good judges of others tend to be less sociable than good judges of self (1, 67, 73).
Dymond (21) reports characteristics for her "high empathy" group that were more comparable to Adams' good self-raters than his good raters of others. On the basis of a TAT analysis, they are described as "outgoing, optimistic, warm, emotional people, who have a strong interest in others," while those low in empathy are "either self-centered and demanding in their emotional contacts or else lone wolves who prefer to get along without strong ties to other people" (21, p. 349). Possibly this finding results from Dymond's method of measuring "empathy" which includes in the score the ability to predict how the Ss will rate J as well as how they will rate themselves. This former measure is close to what we have called ability to judge self.
Taft asked his Js to check their own traits on the Gough Adjective Check List; the poor judges selected socially relevant traits almost exclusively, e.g., egotistical, noisy, whereas the good judges put far more stress on traits concerned with executing tasks, e.g., industrious, patient. In the tests of judging, the poor judges were significantly more likely to make their errors in the direction of generosity to their Ss than were the good judges. These and other data lead to the conclusion that the poor judges were more socially oriented than the good judges, while the latter were more task oriented. The social dependence of the poor judges makes them unwilling - or, perhaps, unable - to judge their Ss in a "hardheaded," extraceptive manner.
The evidence supports the contention that social detachment is a necessary prerequisite for making accurate judgments of others. This social detachment of the good judge of others could well account for the superior ability in this respect of physical scientists, referred to previously (45, 67), since this group might be thought to be less concerned with social matters than psychologists or social scientists.
Are good judges those who are themselves more difficult to judge? On two grounds it might be expected that this would be the case: first, "As a rule people cannot comprehend others who are more complex and subtle than they" (5, p. 515), and, second, good judges are less socially expressive than poor judges, as we have just seen. Actually, Bender states that the better judges in his study were, in his opinion, persons whom others would have found more difficult to judge (9). His test required the Js to predict, on the basis of character sketches of eight Ss, their scores on seven opinion scales. Taft also found that his good judges were less accurately judged by their peers than were the poor judges.
Dymond found the contrary in her study, but since her good judges were more sociable than the poor ones, this suggests that the Judgability of good and poor judges is possibly a function of their respective sociability. The validity of this explanation is borne out by the fact that, in an unreported part of Taft's experiment, the sociability of the Ss was positively correlated with the accuracy with which they were judged by both the assessment staff and their peers.
OVER-ALL SUMMARY AND CONCLUSIONS
The ability to judge others has been considered as a personality trait, and its correlates have been discussed. Five different methods of measuring this ability have been described, and it was suggested that the results of the studies quoted may vary according to the operational definition used. This would seem to apply particularly to the distinction between analytic and nonanalytic techniques, although, when we review the findings on the correlates of the ability to judge, we find few that reveal a definite difference between these two types of techniques.
The contradictions found between studies may be due partly to the low reliability of the measures used, and partly to the effect of specific factors such as the type of judgment required, the traits being judged, and the Ss used. This problem of specificity arises with all traits, but it seems to be particularly marked in the case of the ability to judge others; nevertheless, there does seem to be sufficient generality on this ability to justify describing at least some judges as "good" or "poor."
A great deal of carefully designed investigation is obviously required in the area of judging ability, both with respect to this ability as a general trait, and with respect to differences in the ability under specific conditions. Future experiments should employ longer, more reliable tests of the ability to judge the characteristics of others, and should systematically vary the types of tests used, the types of Ss being judged, and the range of Js. Studies are also required in which the effect of motivational factors can be observed.
Our review of the literature suggests that the following characteristics are fairly consistently found to be positively correlated with the ability to judge the personality characteristics of others: (a) age (children), (b) high intelligence and academic ability (with analytic judgments especially), (c) specialization in the physical sciences, (d) esthetic and dramatic interests, (e) insight into one's status with respect to one's peers on specific traits, (f) good emotional adjustment and integration (analytic tests only), and (g) social skill (only with tests of ability to predict S's behavior). The ability to judge correlates negatively with J's social dependence and his "psychasthenic" score on the MM PI. Characteristics showing fairly consistent lack of correlation are age (in adults), sex, and training in psychology. Some possible relationships on which more evidence is still required before we can substantiate the possible correlation with ability to judge mentioned in parentheses are number of older siblings (negative), literary ability (positive with analytic judgments), and being a clinical psychologist (negative). Two characteristics that may hold for especially poor judges only are poor social adjustment, and either good or poor (not average) ability to judge self.
The main attributes of the ability to judge others seem to lie in three areas: possessing appropriate judgmental norms, judging ability, and motivation. Where J is similar in background to S he has the advantage of being readily able to use appropriate norms for making his judgment. The relevant judging ability seems to be a combination of general intelligence and social intelligence, with the possibility of an additional specific factor for nonanalytic judgments ("intuition") - so far only Wedeck has distinguished such a factor. But probably the most important area of all is that of motivation: if the judge is motivated to make accurate judgments about his subject and if he feels himself free to be objective, then he has a good chance of achieving his aim, provided of course that he has the requisite ability and can use the appropriate judgmental norms. The act of judging one's fellows is a purposive piece of behavior that involves not only conscious motivation but also ingrained attitudes towards social relationships, including the relationships inherent in the act of judging itself.
1. ADAMS, H. F. The good judge of personality. J. abnorm. soc. Psychol., 1927, 22, 172-181.
2. ALLPORT, F. H. Social psychology. Cambridge; Riverside Press, 1924.
3. ALLPORT, F. H., & ALLPORT, G. W. Personality traits: their classification and measurement. J. abnorm. soc. Psychol., 1921, 16, 6-40.
4. ALLPORT, F. H., & MUSGRAVE, R. S. Teleonomic description in the study of behavior. Charact. & Pers., 1941, 9, 326-343.
5. ALLPORT, G. W. Personality: a psychological interpretation. New York: Henry Holt, 1937.
6. ARGELANDER, ANNALIES. The personal factor in judging human character. Charact. & Pers., 1937, 5, 285-296.
7. BARRON, F. Personality style and perceptual choice. J. Pers., 1952, 20, 385-401.
8. BELL, G.B..& HALL, H.E., JR. The relationship between leadership and empathy. J. abnorm. soc. Psychol., 1954, 49, 156-157.
9. BENDER, I. E. A study in integrations of personalities by prediction and matching. Unpublished doctor's dissertation, Univer. of Syracuse, 1935.
10. BENDER, I. E., & HASTORF, A. H. The perception of persons: forecasting another person's responses on three personality scales. J. abnorm. soc. Psychol., 1950, 45, 556-561.
11. BENDER, I. E., & HASTORF, A. H. On measuring generalized empathic ability (social sensitivity). J. abnorm. soc. Psychol., 1953, 48, 503-506.
12. BRACKEN VON, H. Personlichkeitsfassung auf Grund von Personlichkeitsbeschreibungen. Jenaer. Beit. Z. Jugend und Erzihungspsychologie, 1925, 1, 1-50. Reported in reference 5, 459-460.
13. BUZBY, D. E. The interpretation of facial expression. Amer. J. Psychol., 1924, 35, 602-604.
14. CANTOR, R. R. An experimental study of a human relations training program. Unpublished doctor's dissertation. The Ohio State Univer., 1949.
15. CHAPIN, F. S. Preliminary standardization of a social insight scale. Amer, soc. Rev., 1942, 7, 214-225.
16. CHOWDHRY, K., & NEWCOMB, T. M. The relative abilities of leaders and nonleaders to estimate opinions of their own group. J. abnorm. soc. Psychol., 1952, 47, 51-57.
17. COGAN, L. G., CONKLIN, R. M., & HOLLINGWORTH, H. L. An experimental study of self-analysis, estimates of associates and the results of tests. Sch. & Soc., 1915, 2, 171-179.
18. COLEMAN, J. C. Facial expressions of emotion. Psychol, Monogr., 1949, 63, No. 1 (Whole No. 296).
19. DYMOND, ROSALIND F. A preliminary investigation of the relationship of insight and empathy. J. consult. Psychol, 1948, 12, 228-233.
20. DYMOND, ROSALIND F. A scale for the measurement of empathic ability. J. consult. Psychol, 1949, 13, 127-133.
21. DYMOND, ROSALIND, F. Personality and empathy. J. consult Psychol, 1950, 14, 343-350.
22. DYMOND, ROSALIND F., HUGHES, ANNE S., & RAABE, VIRGINIA L. Measurable changes in empathy with age. J. consult. Psychol, 1952, 16, 202-206.
23. ESTES, S. G. Judging personality from expressive behavior. J. abnorm. soc. Psychol., 1938, 33, 217-236.
24. FERGUSON, L. W. The value of acquaintance ratings in criteria research. Personnel Psychol, 1949, 2, 93-102.
25. FERNBERGER, S. W. False suggestion and the Piderit model. Amer. J. Psychol, 1928,40, 562-568. 22 RONALD TAFT
26. FIELDS, S. J. Discrimination of facial expression and its relation to personal adjustment. J. soc. Psychol., 1953, 38, 63-71.
28. GAGE, N. L. Judging interests from expressive behavior. Psychol. Monogr., 1952, 66, No. 18 (Whole No. 350).
29. GAGE, N. L. Explorations in the understanding of others. Educ. psyckol. Measmt, 1953, 13, 14-26.
30. GATES, G. S. An experimental study of the growth of social perception. J. educ. Psychol., 1923, 14, 449-462.
31. GOODENOUGH, FLORENCE L. Sex differences in judging the sex of handwriting. J. soc. Psychol, 1945, 22, 61-68.
32. GREEN, G. H. Insight and group adjustment. J. abnorm. soc. Pyschol., 1948, 43,49-61.
33. GUILFORD, J. P. An experiment in learning to read facial expression. J. abnorm. soc. Psychol., 1929, 24, 191-202.
34. HANKS, L. M., JR. Prediction from case material to personality data. Arch. Psychol, N.Y., 1936, 29, No. 207.
35. HASTORF, A. H., & BENDER, I. E. A caution respecting the measurement of empathic ability. J. abnorm. soc. Psychol, 1952, 47, 574-576.
36. HEYMANS, G., & WIERSMA, E. Beitrage zur speziellen Psychologie auf Grund einer Massenuntersuchung. Z. Psychol, 1906, 42. Reported in reference 6.
37. JENNESS, A. The recognition of facial expressions of emotions. Psychol Bull, 1932, 29, 324-350.
38. KANNER, L. Judging emotions from facial expressions. Psychol Monogr., 1931, 41, No. 3 (Whole No. 186).
39. KELLOGG, W. N., & EAGLESON, B. M. The growth of social perception in different racial groups. J. educ, Psychol, 1931, 22, 367-375.
40. KELLY, E. L., & FISKE, D. W. The prediction of performance in clinical psychology. Ann Arbor: Univer. of Michigan Press, 1951.
41. KELLY, E. L., MILES, C. C., & TERMAN, L. Ability to influence one's score on a typical pencil and paper test of personality. Characl. &• Pers., 1936, 4, 206-215.
42. LANDIS, C. The interpretation of facial expression in emotion. J. gen. Psychol, 1929, 2, 59-72.
43. LINDGREN, H. C., & ROBINSON, JACQUELINE. The evaluation of Dymond's test of insight and empathy. J. consult. Psychol, 1953, 17, 172-176.
44. LUFT, J. Some relationships between clinical specialization and the under- — standing and prediction of an individual's behavior. Unpublished doctor's dissertation, Univer. of California, Los Angeles, 1949.
45. LUFT, J. Implicit hypotheses and clinical predictions. J. abnorm. soc. Psychol, 1950, 45, 756-7'60.
46. MCCLELLAND, W. A preliminary test of role-playing ability. J. consult. Psychol, 1951, 15, 102-108.
47. MARTIN, H. W. Effects of practice on judging various traits of individuals. Psychol Bull, 1938, 35, 690. (Abstract)
48. MEAD, G. H. Mind, self and society. Chicago: Univer. of Chicago Press, 1934.
49. MORENO, J. L., & MORENO, FLORENCE B. Role tests and role diagrams of children. In J. L. Moreno (Ed.), Group psychotherapy. New York: Beacon House, 1945. Pp. 188-203.
50. Moss, F. A., HUNT, T., OMWAKE, K. T., & BONNING, M. M. Social intelligence test. Washington: Center for Psychological Service, 1927.
51. MURRAY, H. A. Explorations in personality. New York: Oxford Univer. Press, 1938.
52. NEWCOMB, T. M. An experiment designed to test the validity of a rating technique. J. educ. Psychol, 1931, 22, 279-289.
53. NORMAN, R. D. The interrelationships among acceptance-rejection, self-other identity, insight into self, and realistic perception of others. J. soc. Psychol, 1953, 37, 205-235.
54. NOTCUTT, B., & SILVA, A. L. M. Knowledge of other people. J. abnorm. soc. Psychol., 1951, 46, 30-37.
55. OSS ASSESSMENT STAFF. Assessment of men. New York: Rinehart, 1948.
56. POLANSKY, N. A. How shall a life history be written? Charact. & Pers., 1941, 9, 188-207.
57. RABIN, A. I. Szondi's pictures: identification of diagnoses. J. abnorm. soc., Psychol, 1950, 45, 392-395.
58. ROKEACH, M. Studies in beauty. II. Some determiners of the perception of beauty in women. J. soc. Psychol., 1945, 22, 155-169.
59. SCODELL, A., & MUSSEN, P. Social THE ABILITY TO JUDGE PEOPLE 23 perceptions of authoritarians and nonauthoritarians. J. abnorm. soc. Psychol,, 1953, 43, 181-184.
60. SEARS, R. R. Experimental studies of projection. I. Attribution of traits. J. soc. Psychol., 1936, 7, 151-163.
61. SHERRIFFS, A. C. The "intuition questionnaire." A new projective test. J. abnorm. soc. Psychol., 1948, 43, 326- 337.
62. SOSKIN, W. F. Bias in postdiction from projective tests. J. abnorm. soc. Psychol., 1954, 49, 69-74.
63. SPEROPF, B.J. Empathic ability and accident rate among steel workers. Personnel Psychol., 1953, 6, 297-300.
64. STEINMETZ, H. C. A study of the ability to predict test responses. Unpublished doctor's dissertation, Purdue Univer., 1947.
65. SULLIVAN, H. S. Conceptions of modern psychiatry. Psychiatry, 1940, 3, 1- 117.
66. SWEET, LENNIG. The measurement of personal attitudes in younger boys. New York: Association Press, 1929.
67. TAFT, R. Some correlates of the ability to make accurate social judgments. Unpublished doctor's dissertation, Univer. of California, 1950.
68. TOBOLSKI, F. P., & KERR, W. A. Predictive value of the empathy test in automobile salesmanship. J. appl. Psychol., 1952, 36, 310-311.
69. TRAVERS, R. M, W. A study in judging the opinions of groups. Arch. Psychol,, N.Y., 1941, No. 266, 1-73.
70. TRAVERS, R. M. W. A study of the ability to judge group-knowledge. Amer. J. Psychol., 1943, 56, 54-65.
71. VALENTINE, C. W. The relative reliability of men and women in intuitive judgments of character. Brit. J. Psychol., 1929, 19, 213-238.
72. VAN ZELST, R. H. Empathy test scores of union leaders. J. appl. Psychol., 1952, 36, 293-295.
73. VERNON, P. E. Some characteristics of the good judge of personality. J. soc. Psychol., 1933, 4, 42-57.
74. WALLEN, R. Individual's estimates of group opinion. J. soc. Psychol., 1943, 17, 269-274.
75. WALLIN, P. The prediction of individual behavior from case studies. Soc. Sci. Res. Coun. Bull., 1941, No. 48, 181- 250.
76. WALTON, W. E. Empathic responses in children. Psychol. Monogr., 1936, 48, No. 1 (Whole No. 213), 40-67.
77. WATSON, D. L. On the role of insight in the study of mankind. Psychoanal. Rev., 1938, 25, 358-371.
78. WEDECK, J. The relationship between personality and "psychological ability." Brit. J. Psychol., 1947, 37, 133- 151.
79. WEDELL, C., & SMITH, K. U. Consistency of interview methods in appraisal of attitudes. J. appl. Psychol., 1951, 35, 392-396.
80. WEINGARTEN, ERICA M. A study of selective perception in clinical judgment. J. Pen., 1949, 17, 396-406.
81. WOLF, R., & MURRAY, H. A. An experiment in judging personality. J. Psychol., 1937, 3, 345-365. Received November 12, 1953.