At any rate, examination of the various group means shows that two groups may have similar
educational
levels and very different E means, as well as similar E means and very dif- ferent educational levels.
Adorno-T-Authoritarian-Personality-Harper-Bros-1950
) have been shown to be slightly more intelligent, to receive better grades in college, to read more and to have greater intel- lectual curiosity.
While the differences are significant, there is of course much overlap between the two extreme groups.
It was not feasible within the scope of the present research to administer intelligence tests to the groups taking the questionnaire. Fortunately, such tests had already been administered to some or all of the members of four groups: Maritime School Men, Employment Service Men, Psychiatric Clinic Men and Women, and San Quentin Men. No information was obtained in our questionnaire regarding years of education; this question was omitted partly because of the probable unreliability of the answers and partly because of the fear that it might make the less educated subjects defensive about the entire questionnaire. In some cases, e. g. , the college students and the pro- fessionals, the amount of schooling was relatively constant for the entire. group. For two groups, the Psychiatric Clinic patients and the Maritime School, the years of schooling had been determined in a way that seemed fairly (though not entirely) reliable.
The data on intelligence are presented in Tables 1-3 (VIII), on education in Tables 4 (VIII) and 5 (VIII). We may consider intelligence first.
1 For reports and summaries see: Murphy, Murphy and Newcomb (85); Kerr (63).
? 282 THE AUTHORITARIAN PERSONALITY
For all three groups in Tables I-J(VIII), namely the Maritime School Men, Employment Service Men, and Psychiatric Clinic Men and Women, the average IQ is significantly above the general population average (usually by about one sigma); indeed, even the lowest scorer is, except in the third group, above the population mean. This fact, namely the restriction in the "range of talent," must be considered in evaluating the results. The correla- tions for the Maritime School Men, obtained with the AGCT (Army Gen- eral Classification Test), are very similar to those obtained with the Otis Higher Form A Intelligence Test on the Employment Service Veteran Men. The correlations of these tests with the Ethnocentrism scale, Forms 45 and
Ability Test
AGCT
Mechanical Comprehension
Reading Comprehension
Arithmetical Comprehension
Test Properties
Correlation with:b
TABLE 1 (VIII)
CORRELATIONS OF THE E AND F SCALES WITH VARIOUS ABILITY TESTS {MARITIME SCHOOL MEN)
Mean a 126. 7
126. 5
92. 5
81. 2
~ R a n g e 8. 98 102-153
14. 61 66-166
AGCT
? 25
EA -. 02
-. 17
EA+B F -. 20 -. 20
. 00 -. 13
13. 04 57-121 . 55 -. 08 -. 06 -. 20 8. 88 61-105 ? 59 -. 06 -. 16 -. 16
aThe present means may be compared with the following population means: For the general population, the AGCT and Mechanical Comprehension Tests have means of 100. For the "high school graduate? populatiorr the Read- ing comprehension and Arithmetical Comprehension Tests have means of 79. On all but the last-named test, therefore, the present sample is con- siderably above average,
bThe number of cases (N) involved in the correlations are as follows:
Of the 343 subjects in the total sample, 342 received all of the ability tests, with the exception of four individuals who omitted the Reading Comprehension Test, The correlations involving AGCT and F, then, are based on an N of 342. The EA Scale, contained in Form 40, has an N of 168. while EA+B has an N of 178.
40, range between -. 02 and -. p, averaging above -. 2. The correlations ? of E with the Mechanical, Reading and Arithmetical Comprehension Tests2 (Table 1 (VIII)) are slightly lower, averaging about - . 1 . These correla- tions, taken together, are statistically significant, that is, dependably above zero, at the 5 per cent level. It may be noted also that there are no positive correlations. Table 3 (VIII) gives the mean (Wechsler-Bellevue Intelligence
2 The Bennett Mechanical Aptitude Test, the Iowa Silent Reading Test, the Stanford Adult Arithmetical Reasoning Test.
? Otis Raw Score Otis IQ
St anford-Binet
IQb
56. 5 114. 5 129. 5
ETHNOCENTRISM IN RELA TION TO INTELLIGENCE TABLE 2 (VIII)
CORRELATIONS OF THE E, F, AND PEC SCALES WITH 'THE OTIS HIGHER FORM A INTELLIGENCE TEST
(Eli! PLOYMENT SERVICE VETERAN MEN)
Otis Test Properties Meana
Range 34-75 92-133
108-140
Correlation of Otis with:c
EA: -. 32 EA+B: -. 22 F: -. 48 PEC: -. 16
aThe mean Otis IQ of 114. 5 is significantly above the population average of 100. 0 (population S. D. is 10. 0).
bThe conversion of the Otis scores into Stanford-Binet IQ scores was
done by means of a table prepared by Dr. E. E. Ghiselli. For the general population the Stanford-Binet has a mean of 100, an S. D. of 16?
Cotis Test data were available for 104 of the 105 cases in this sample. The N is, then, 104 for the correlations with F and ? PEC. The N is also 104 for EA? since the EA scores of the subjects taking Form 45 as well as of those taking form 40 were used, The N is 50 in the case of EA+B (Form 45).
Form 45 E? Scale Quartiles Low quartile
Low middle quartile
High middle quartile
High quartile
Range on E . ! '! . 10-24 8 25-36 5 37-50 13 51-70 11
37
Mean IQ 125. 3 117. 8 113. 9 107. 3 114. 9
TABLE 3 (VIII)
MEAI~ WECHSLER-BELLEVUE I Q SCORE FOR EACH QUARTILEa OF THE ETHNOCENTRISM SCALE (PSYCHIATRIC CLINIC MEN AND WOMEN)
aThe subjects represent only a part of each quartile. In all, 15 of the 50 men, and 22 of the 71 women, had received Wechsler-Bellevues. The men and women were similar with respect to proportion in each quartile, identical with respect to mean IQ. It is not clear why more upper-half than lower-half subjects have been tested. The mean of 114 approximates that for the patients generally.
? THE AUTHORITARIAN PERSONALITY
Test) IQ for the four E scale quartiles, and we find the equivalent of a low negative correlation.
Tables I-3 (VIII) indicate that, for individuals with IQ's of approximately wo and above, there is a very low but dependable negative relationship be- tween intelligence and ethnocentrism: the most ethnocentric subjects are, on the average, less intelligent than the least ethnocentric, while the middle
scorers on E are intermediate in IQ.
Data on the San Quentin Men, not presented in the above tables, suggest a similar relation between E and IQ in groups having a wider intellectual range. Wechsler-Bellevue Test Scores were available on 77 of the IIO sub- jects in the San Quentin sample. This subsample had a mean E score of 4. 68 and a Standard Deviation of I. 28, as compared with the total-sample mean of 4. 6I and S. D. of 1. 28. The mean Wechsler-Bellevue IQ (full scale) was
I09. o, the S. D. q. 8, and the range 78-q2. This subsample is, then, almost identical with the total (questionnaire) sample in E mean and S. D. ; it is slightly more intelligent than the total prison population, whose mean IQ is just under 100. The obtained r between E and IQ was -. 2 8. This value is of the same order of magnitude as those reported above for samples in which the IQ range was more constricted. It is, of course, not conclusive, since the tested sample may have been systematically biased in its selection. In addi- tion, other factors such as educational and class level are probably con- tributing to this correlation, since they seem to be at least slightly related to both E and IQ. A conclusive study of the relation between IQ and E would have to partial out, or to keep constant, these other factors. Never- theless, the series of negative r's, on a variety of groups and by means of a variety of intelligence tests, provides substantial evidence of a significant relation between E and IQ. That the correlation is greater than zero, and in a negative direction, is in keeping with previous studies as well as with the results of the present study regarding the role of stereotypy and rigidity in ethnocentrism. That the correlation is only moderately dose-apparently in the range of -. 2 to -. 4-is evidence that intelligence is only one of many variables which determine E-scale scores.
Correlations between the Ability Tests and the F scale were computed for the Maritime School Men (Table I (\llll)) and the Employment Service Veterans (Table 2 (VIII)). In the former group the correlations range be- tween -. q and -. 20 while in the latter the extremely high value of -. 48 was obtained. It appears, then, that IQ is more closely related to F than to E although, except for the Veterans, the correlation is relatively small. Further study is required to determine whether or not the r of -. 48 is spurious or exceptional. The higher correlations with F than with E might be explained on the basis of certain of the F clusters, e. g. , superstition and stereotypy; correlations between IQ and the individual F items might well be obtained
? ETHNOCENTRISM IN RELA TION TO INTELLIGENCE 285
in future research. The correlation of -. r6 between IQ and the PEC scale (Table 2 (VIII)) is consistent with other findings.
That the relation between intelligence and ethnocentrism is not very close is suggested also by the over-all group data. Thus, while the three groups in Tables I-3 (VIII) are very similar in average IQ, they vary greatly in aver- age E score. The Psychiatric Clinic patients have an E mean of 3. 7, a full point below the means for the Maritime School and Veteran Men. More- over, the latter groups, while extremely high in average IQ, are also among the most ethnocentric of all groups tested. It would seem, therefore, that high tested intelligence is no guarantee against the overall authoritarian pattern of ideology and personality.
Data on the relation of ethnocentrism to amount of education are pre- sented in Tables 4 (VIII) and 5 (VIII). One of the most striking results is
TABLE 4 (VIII)
MEAN NUMBER OF YEARS OF EDUCATION FOR EACH QUARTILE OF THE ETHNOCENTRIS\t SCALE
(PSYCHIATRIC CLINIC MEN AND WOMEN)
Form 45 E Seale Quartiles Low quartile
Low middle quartile
Hi~h middle quartile
High quartile
Range onE 10-24 25-36 37-50 51-70
Na Mean Yrs. Educationb 29 13. 8
28 12. 7
27 11. 8
28 11. 2 112 12. 4
aThese data are based on 45 of the 50 men, 66 of the 71 women, The means for men and women separately were so similar that they were combined in order to increase the N.
bone year has been added to the number of years of education in five cases where the individual had specialized training such as secretarial or accounting school,
that these two variables are much more closely related in the Psychiatric Clinic group than in the Maritime School group. The average number of years of education drops gradually but consistently (from I 3. 8 to I 1. 2) in the Clinic group as score on the E scale increases (Table 4 (VIII)). How- ever, in the Maritime School data, computed in another manner, the changes are not so consistent. On the basis of the combined Forms 45 and 40 data, subjects with less than I2 years of education (i. e. , not high school graduates) have the highest E mean, 4? 9, while those with I2-I4 years (there were no college graduates) had almost identical E means of about 4. 6. However,
? 286
THE AUTHORITARIAN PERSONALITY
Years
Education
Form 45 ( EA+B) Form 40 (Ea> T otal Group
Less than 12 12 years
13 years
14 years
Blank OVer-all
N Mean N
36 4. 38 60 104 4. 28 91 13 4. 75 7 18 4. 34 6 7 4. 63 0 178 4. 36 164
Mean N Mean 5. 21 96 4. 90
5. 04 195 4. 63 4. 40 20 4. 63 5. 17 24 4. 55
7 4. 63 5. 08 342 4. 68
TABLE 5 (VIII)
MEAN E SCORE FOR GROUPS HAVING VARIOUS YEARS OF EDUCATION (MARITIME SCHOOL MEN)
there is some question as to whether the results for the two Forms should be combined, since the results for each Form separately are not so clear-cut. In the Form 45 group there is no consistent trend, the subjects with 13 years of education having the highest E mean and the other educational levels varying only within a range of o. I points. In the Form 40 group, on the other hand, the 13-year level is the least ethnocentric, while the 14-year group vies with the less-than-12 for the most ethnocentric position. The only difference that holds up for both Forms is that between the 12 and the less-than-12 year levels, and this difference borders on the 5 per cent level of statistical significance.
Why is the relation between ethnocentrism and education more consistent in the Psychiatric Clinic group? One possibility is that the Maritime School members who had one or two years of college and then dropped out are systematically atypical, and that an unselected group of college students might be less ethnocentric. To the extent that this is true, a clear-cut relation- ship between ethnocentrism and education does exist. However, the relation in the case of the Psychiatric Clinic may be spuriously high, since the college students and college graduates in this sample are not a random sample of these educational levels. It is possible-though not yet demonstrated-that the col- lege-trained patients are, to a greater degree than those who did not attend college, selected for willingness to recognize the need for, and to seek, psychological aid. To the extent that this is true, the relationship between ethnocentrism and education is less clear-cut than these results indicate.
One might venture the hypothesis that ethnocentrism is at least as closely related to the desired amount of education as to the actual amount. Thus the two Extension Classes (Forms 78 and 45), most of whose members were
? ETHNOCENTRISM IN RELA TION TO INTELLIGENCE 287
adults having only 12 years or less of schooling, but trying to "learn some- thing on the side" and perhaps even to obtain a college degree, had E means as low as those of the University of California students. There is also some clinical evidence to support this hypothesis.
At any rate, examination of the various group means shows that two groups may have similar educational levels and very different E means, as well as similar E means and very dif- ferent educational levels. For example, the University of California students and the George Washington University Students, with similar educational levels, are significantly different in E means (Form 78, A-S scale, and Form 40, E scale). Again, the Working-Class group, though similar in socioeco- nomic and educational background to the San Quentin group, is significantly less ethnocentric (Form 45).
We may tentatively conclude that ethnocentrism shows a slight negative correlation with amount of education. It is likely, though far from a demon- strated fact, that college graduates are less ethnocentric than high school graduates, who are in turn less ethnocentric than those who did not complete high school. It is not clear which is more important: that the correlation is greater than zero, or that it is at best not far from zero. To those who urge education per se as a kind of panacea, the smallness of the correlation ought probably to be stressed. But this is not to deny the importance of education. It is, rather, to emphasize that our educational system, college as well as public school, is still far from realizing its potential strength as a social force in the service of democratic values. The reasons for this are outside the scope of the present research. It may also be pointed out that, even under the best educational conditions, exposure to the classroom is not enough, and that motivation to learn and receptivity to new ideas provide the only psycho- logical soil in which democratic education can develop effectively.
In summary, ethnocentrism seems to have a low but statistically significant relation to both intelligence and education, the most educated and intelligent subjects being, on the average, the least ethnocentric. However, these varia- bles were studied only secondarily in the present research, and convincing determination of their relation to ethnocentrism requires more extended sampling, particularly of the lower educational and intellectual levels. It is also necessary to control more adequately the operation of other variables such as class level, educational opportunity and educational motivation. Nevertheless, the present results do contradict seriously one of the com- monly held theories of prejudice and fascism, namely, that they are sup- ported out of simple stupidity, ignorance or confusion. It would seem, rather, that an autocratic social structure is best suited to the particular type of ra- tionality exercised by the authoritarian personality. A promising field of future research is the study of what might be called "the dynamics of in- telligence. " For example, the intellectual functioning of ethnocentric indi- viduals, even those with above-average IQ's, seems to be relatively rigid,
? THE AUTHORITARIAN PERSONALITY
to work better in relation to things than to people, to be primarily extra- ceptive, and to become disrupted when required to deal with more psy- chological issues, especially those involving personal needs and emotions
(anti-intraceptive).
As has been noted elsewhere (particularly in Chapter IV), the average IQ
and the educational level of the entire sample used in the present research are probably somewhat above those of the general population, or, rather, above those of the urban middle class (our primary reference population). This sampling bias, together with that of age-our sample being somewhat younger than a representative sample would be-has probably resulted in our obtaining over-all means for the various scales which are slightly biased in the direction of being too low. However, the error seems to be less than might have at first been suspected. In addition, it is not likely that such sampling factors have distorted to any appreciable degree the relationships among the variables of ideology, personality, and group membership under investigation. Since we were primarily concerned with the causes and cor- relates of antidemocratic trends, that is, with correlations and differences, rather than with the average amount of any single trend per se, the diverse groups comprising the total sample provide, it would seem, an adequate basis for study.
? CHAPTER IX
THE INTERVIEWS AS AN APPROACH TO THE PREJUDICED PERSONALITY Else Frenkel-Brunswik
A. INTRODUCTION: COMP ARISON OF GROUPS
Reference to the interviews has so far been limited to the discussion of two individual cases, with emphasis mainly on the pattern of social and po- litical issues (Chapter II). A series of five chapters beginning with the present will show the purpose and value of the interviews in their own right, cover- ing in a systematic fashion a variety of topics; furthermore, analysis will be in terms of groups rather than of single individuals (Chapters IX to XIII). Some special individuals or groups as well as some special issues will be taken up once more in later chapters, bringing in additional material from the interviews (Chapters XX to XXII). The problem of ideology as revealed through the interviews is not being considered in the series of chapters which makes up the present Part II; this problem will be taken up in Chap- ters XVI to XIX.
The major advantage of the interview technique lies in the scope and free- dom of expression it offers to the person being studied. Thus we may learn what he thinks about himself, about his hopes, fears and goals, about his childhood and his parents, about members of the other sex, and about peo- ple in general. It is through careful and critical evaluation of sources of this kind that an adequate view of the total personality can perhaps best be ap- proximated.
Rather than making an attempt to establish the dynamic interrelationships of the significant factors for each single individual, however, we shall look for a basis of generalizations within groups which will permit us, it is hoped, to come to grips with the social and psychological trends typical of the highly prejudiced and of the unprejudiced. Does the family constellation differ in the typical prejudiced home as compared with the typical unprejudiced home? Do prejudiced individuals tend to have different images of their parents than do unprejudiced ones? How does the handling of discipline vary
291
? THE AUTHORITARIAN PERSONALITY
? in this respect? Do prejudiced and unprejudiced individuals differ in their sex life, their way of choosing friends, their values, their general cognitive and emotional approach to life? These are some of the questions which will be the prime concern of the present series of chapters. By virtue of its speci- ficity and unique character, the interview may be called upon to yield, in the first place, information of this broader kind on the personality of those scoring high or low on anti-Semitism.
In fact, it was a preliminary review of some of the crucial factors of this kind in a set of exploratory interviews which led to the construction of the F scale (Chapter VII). However, it was this and other scales, consisting of given statements calling for indications of agreement or disagreement only, upon which group comparisons between the personalities and the attitudes of prejudiced and unprejudiced subjects-now to be extended to the richer and more flexible type of evidence as given by an interview-have been thus far exclusively based in the present volume. On this comparatively limited basis, marked differences between these two groups were established.
Questions as to the specific meaning and connotation of the various state- ments for the individuals concerned, however, had to be left open to a con- siderable extent. Further validation of the questionnaire data can be effected by probing in greater detail into the spontaneous elaborations a subject may be willing to make on the topics covered by the questionnaire. To obtain such additional information on results gathered by other techniques is a further goal in the subsequent analysis of the interviews.
While the importance of the interview as source material is generally agreed upon in the social sciences, there are specific difficulties in evaluating such material. This is due mainly to the fact that interview material is highly diversified and unstructured. At the same time, the richness, flexibility, and spontaneity of this material are the features which constitute its major asset; room is left for unanticipated variations. To preserve all of this uniqueness and flavor, we should have to reproduce all, or at least the most outstanding, protocols in full-allowing the reader to form his own impressions and draw his own conclusions.
Presentation in full, though doing maximal justice to the material, has its serious drawbacks. It would not be easy to survey and would of necessity leave to the reader much of the burden of interpretation, or else introduce a potentially arbitrary distribution of emphasis in the process of interpreta- tion. In the same manner, if we were to limit presentation itself to a few select protocols we might easily be criticized on grounds of possible ar- bitrariness in selection.
For these reasons it was decided to attempt some kind of quantification within groups, rather than to present only clinical results based on intensive but more impressionistic case studies as was originally planned. For the kind of evaluation chosen, hypotheses were already formulated on an empirical
? INTERVIEWS AS APPROACH TO PREJUDICED PERSONALITY 293
basis, giving "hunches" for potential validation. The procedure consisted of a careful evaluation of the interview material in terms of an extensive set of scoring categories. These categories had been designed to encompass as much as possible of the richness and intricacy of the material at hand. (See E. Frenkel-Brunswik (3 r, 32, 36) ). They were a product of intensive study of the intervie\vs with full consideration of all the other evidence obtained from the individuals in question, especially their standing on the prejudice scales. The result was establishment of a Scoring Manual comprising about ninety categories and subcategories (see below).
Evaluation of the interview protocols was by raters unfamiliar with the specific ideology of the subject, the Scoring Manual serving as a guide for ratings in terms of the various categories.
In order to offer to the reader as much direct contact with the raw ma- terial of the interviews as possible, numerous quotations are inserted into the subsequent chapters. The raw material is arranged through the medium of the scoring categories, and the relationship of the latter to, or even their de- pendence on, the original material will become evident there. Many of the quotations presented were directly instrumental in designing the categories employed in their evaluation. Actually, the system of scoring categories reflects the theory or the interrelationships between personality and prej- udice which was empirically developed in the course of the exploratory study of the bulk of the interviews, individual by individual. This exploratory study preceded the more elaborate checking procedure in which the indi- vidual lost his identity in a mass of statistical evidence organized in terms of the scoring categories and evaluated in terms of larger groups. It is only through such a statistical procedure that the original hypotheses can be, and in fact have to a considerable extent been, verified.
It was hoped that use of the variables defined by the scoring categories would help to bridge the existing gap between the studies of groups and of individuals and perhaps contribute to the establishment of a mutual give and take of facts and concepts. Indeed, some of the variables and relationships which were originally conceived of in the course of the generalized, sta- tistical establishment of personality patterns in samples of prejudiced vs. un- prejudiced people, were at the same time found to be crucial in the intensive study of single individuals or small groups (see Chapters XX, XXI, XXII).
The subject's view of his own life, as revealed in the course of the inter- view, may be assumed to contain real information together with wishful -and fearful-distortions. Known methods had to be utilized, therefore, and new ones developed to differentiate the more genuine, basic feelings, at- titudes, and strivings from those of a more compensatory character behind which are hidden tendencies, frequently unknown to the subject himself, which are contrary to those manifested or verbalized on a surface level. To cope with such distortions cues are available or may be developed to guide
? 294 THE AUTHORITARIAN PERSONALITY
interpretations. The methodological safeguarding of such interpretations is one of the central problems in the approach to the interviews. The subse- quent analysis of the interview data will include discussion of this point. In general, it endeavors to add to our knowledge of the relationship of surface cues and underlying strivings, with special reference to the problems raised by the personality of the ethnically prejudiced.
In the present chapter the securing of the interview material and the tech- nical aspects of its analysis will be discussed. The sample of the subjects in- terviewed as compared with the total sample will be described first. A characterization of the interviewers in terms of their background, training, and psychological point of view will also be given. Next, the Interview Sched- ule used and the technique employed in interviewing will be presented. This will be followed, in the concluding sections of this chapter, by a discussion of the methods used in the evaluation of the interview data.
In the four following chapters a statistical analysis and discussion of the results gained from the study of the interviews will be presented, first in terms of a detailed set of rating categories (Chapters X to XII), and then in terms of over-all ratings and comprehensive description (Chapter XIII).
B. SELECTION OF SUBJECTS FOR THE INTERVIEWS
1. BASIS OF SELECriON
The selection of the subjects to be interviewed was determined, in the first place, by their responses on the A-S or the E scale. With few exceptions (see below), all interviewees belonged either to the uppermost or to the lowermost quartile in this respect, the proportions of high-scoring and low-
scoring subjects being approximately equal.
Secondly, consideration was given to the response to the three scales of
the questionnaire. Thus, an effort was made to include in the sample inter- viewed not only the most "typical" high scorers and low scorers, i. e. , sub- jects with correspondingly high or low scores on the PEC and F scales, but also some of those more atypical subjects who obtained a high score on the first scales but a relatively low score on one or both of the others.
Thirdly, an effort was made to balance our samples of high-scoring and low-scoring subjects in terms of age, sex, political and religious affiliation, as well as national or regional background.
Of the thirty to forty different socioeconomic groups to which the ques- tionnaire had been administered (see Chapter IV), subjects for interviews were selected from the following twelve: Psychiatric Clinic Patients from the Langley Porter Clinic of the University of California (men and women, abbreviated LPC); University of California Public Speaking Class (men
? INTERVIEWS AS APPROACH TO PREJUDICED PERSONALITY 295
and women, PSM and PSW); Alameda School for Merchant Marine Of- ficers (men, Maritime); San Quentin State Prison Inmates (men, SQ); Uni- versity of California Extension Testing Class (men and women, TC); Uni- versity of California Extension Psychology Class (men and women, EG); University of California Summer Session Education Class (men, EdPs); Stu- dents at the Pacific School of Religion (men, PSR); Employment Service Veterans (men, Vets); Professional Women-public school teachers, social workers, public health nurses (N and R W); University of Oregon Summer Session Students (women, OG); Students at the University of California Medical School (women, Med).
In all, approximately one hundred persons were interviewed. Some of the interviews could not be used in the final scoring, however. One reason for this was that some of the subjects scoring at the very extreme ends of the F scale distribution had been used by the scorers in a last checkup on the scoring manual and had therefore to be excluded later from the main analysis which was to be a "blind" one (see below). Other records had to be discarded because of their brevity or barrenness.
The results to be reported in the subsequent chapters are based on the records of So interviewees, 4 0 men and 4 0 women. Of the men, 2 0 were high extremes on the E scale; and 20 were low extremes. For the women, the corresponding numbers were 2 5 and I 5. The survey presented in Tables I (IX) and 2 (IX) shows for each interviewee the code number, group extrac- tion, standing on responses to the various scales of the questionnaire, with parentheses used to designate membership in one of the middle quartiles.
The rater's "blind" diagnosis of the interview responses makes up the right half of the tables. It is to receive full discussion in Chapter XIII.
2. REPRESENTATIVENESS OF THE INTERVIEWEES
A breakdown with respect to further characteristics of the interviewees, and a comparison of the samples interviewed-approximately one-tenth of the total of the groups mentioned above-with the entire upper and lower quartiles of our over-all samples, will show that our interview samples are fairly representative of the extreme quartiles defined in terms of overt anti- Semitism or ethnocentrism. A quantitative comparison is given in Table 3 (IX). Inspection of the means of all the subjects falling into the upper and lower quartiles with those of corresponding groups of interviewees reveals a sufficiently close agreement. The interviewee samples are, more often than not, somewhat farther to the extreme end of the scale than the correspond- ing total extreme quartiles. This trend holds in spite of the fact that, as seen from Tables I (IX) and 2 (IX), in a few instances interviewees had to be taken from the extreme ends of the middle quartiles of the E scale.
A further breakdown has been undertaken with respect to age, religion,
? THE AUTHORITARIAN PERSONALITY
TABLE 1 (IX)
SURVEY OF 20 PREJUDICED AND 20 UNPREJUDICED MEN INTERVIEWEDa
Standing on
Groupb
Ml LPC h h (h) 1 27 1 H
Code No.
~estionnairec
Interview Scoresd "High" "Low"
A-S E F PEC
M4 Ps. t h h 1 h 47 2 H
M6 PSM h h h h 41 2 H
M7 PEM h h (l) 1 6 32 L L Mll PSM h h h h 52 2 H H Ml3 PSM h h h (h) 52 1 H H Ml4PSM hh(h)(1) 8 32 L L M17Ps. th(h)1h254 H H MIS LPC h h h h 33 3 H H M20Maritime-h11540 L L M40 SQ h h (h) h
M41 SQ h h h h
M43 SQ h h h 1
M45 SQ h h h h
M46 TC h h 1 h
M47 SQ h h h h
M5l SQ h h h h
M52 SQ h h h h
M57 SQ h h h h
M58 TC h h h h
Means of 20 prejudiced men interviewees
M2 EG 1 1 1 1
M3 PEM 1 1 (1) 1
M5 PSM 1 1 1 1 MSPStl1(1)1h544 M9 PSM 1 1 1 (h) 39 2 M1oEdPs 1111
M12EdPs 1111
Ml5 LPC 1 1 1 1
M16 LPC 1 1 1 1 M19Ps. t11(l)(h)350 H H M42Maritime-111157L L
M44 PSR 1 1 1 1 2 54
M48Vets 1111 6 37
M49TC 1111442
M50 SQ 1 1 1 h JO 34 M53Vets1111152
M54 SQ 1 1 1 h 12 24
M55TC 1111456
M56 SQ 1 1 1 1 5 41
M59 SQ 1 1 1 h 14 39 L L
Means of 20 9. 4 38. 0 unprejudiced men
interviewees
"For discussion of the evaluation of the interviews and of the results shown in this table, see Section F of the present chapter, and Chapter XIII.
bFor key to abbreviations, see text, p. 294/95.
cThe upper and lower middle quartiles are indicated by the use of paren-
theses with the letters h and 1.
? Number of ratings other than "Neutral. " Number of Neutrals is ob-
tained by subtracting that of "High" and of "Low" from 72 (on Table 1 (IX)) or 65 (on Table 2 (IX)). For selection of categories, seep. 335.
Composite Standing on 72 Categories
Intuitive Over-all Rating of Interviews
H H H
55 2 H H 49 2 H H 43 3 H H 48 2 H H 42 6 H H 44 2 H H 36 5 H H 51 4 H H 56 0 H H 54 2 H H
38. 7 7. 3
5 52 0 53 3 34
L L L L L L L L H ff H H L L L L L L
337 245 2 43 6 44
L L LL L L L L L L L L L L L L
? F22 PSW
F24 PSW
F25 LPC
h (h) h (h) 28 1
INTERVIEWS AS APPROACH TO PREJUDICED PERSONALITY 297 TABLE 2 (IX)a
SURVEY OF 25 PREJUDICED AND 15 UNPREJUDICED WOMEN INTERVIEWED
Group
H hhhh3'1 4 H hhhh3) 7 H F26N hh(l)h922 L
F28 RW hhhh197H F31 PSW hh(h)h510H
Intuitive over-all Rating of Interviews
H H H L H H H L H H H H H H L H L H H H L H H H H
H L L L L L L H L L L L L L L
U>de No.
standing on Questionnaire
A-S E F PEC
Interview Scores
"High" "Low"
Q:>mposite standing on 65 categories
F32N hhhh34 3 F33 TC h h h h 3 32
H L H H H H H H? L H L H H H H H H
F36 TC F37 EXi
hh(h)(l)'n 5 h(h)hh21 10 hhhh25 14 hhh(1)245 hhh(h)30 6 hh11238
F38 PSW
F39a N
FOO RW
F61 LPC
F64 RW h h h h 0 26 F66 PSW hhh1357 F67 RW hhhh328 F68N h(h)1h32 4
F69 PSW
F71 PSW
F72 LPC
F74 PSW
F77 LPC
F78 PSW
F79 OG
Means of 25 prejudiced women interviewees
hhhh37 7 hhhh47 2 hhhh17 26 hhhh45 4 hhh13'1 4 h h h (h) 44
h h h (h) 36
27. 4 9. 5
F21PSW 111h24 0
H L L L L L L H
L L L L L L
F23 TC 1 1 (1) 1
F'n PSW 1 (l) 1 1
F29 LPC 1 1 1 1
F30 RW 1 1 1 1
F34 PSW 1 1 (l) 1
F35 TC 1 1 1 1
F39 PSW 1 1 1 1
F62 PSW 1 1 1 h 1 44 L F63 LPC 1 1 1 1
F65 PSW 1 1 h 1 F70 Med 1 1 1 1 F73 PSW 1 (1) 1 1 F75 PW 1 (l) (h) 1 F76 PSW 1 ( 1 ) ( h ) ( l )
Means of 15 unprejudiced women interviewees
385
4 42 6 44 0 38 1 35 343
14 2 2 7. 0 36. 2
asee footnotes to Table 1 (IX).
2 46 0 53 7 36 3 44 149 1 42
1 H 4 H
? Questionnaire form
T otal of a ll taking Forms Range Mean
subjects
High-scoring interviewees Mean N
66. 5 7 70. 6 18
58. 6 12 59. 0 6
Low- scoring Low-quartile interviewees
78:
45:
Men 17-82 46. 4 Women 14-86 46. 2
Men 10-70 43. 2 Women 10-70 35. 4
Mean
29. 1 28. 2
25. 0 15. 6
Mean N
30. 3 6 33. 6 10
15. 8 11 13. 6 5
TABLE 3(IX)
REPRESENTATIVENESS OF INTERVIEWEES IN TERMS OF SCORES ON THE ETHNOCENTRISM SCALEa
78
or 45 N
52 243
390 130
High~ quartile Mean
64. 6 69. 5
59. 7 54. 8
aSince 75 out of the 80 interviewees are from among the subjects tested by Form 78 or Form 45, comparisons are here limited to these two groups.
N
00 "'
? Religious
A ffiliation
Catholic Protestant None
Blank
Sums
High Quartile
Low Quartile
INTERVIEWS AS APPROACH TO PREJUDICED PERSONALITY 299
and politics. To maintain anonymity, these data are not included in the tables just mentioned which deal with individual subjects, but are presented in a statistical manner in Tables 4 (IX) to 6 (IX). For men, a few data on religion and politics are missing; hence the discrepancies in the sums relating to the total quartiles.
On the whole, the distributions of the interviewees and of corresponding extreme quartiles are not at too great odds with one another, considering the difficulties in finding subjects with the exact combination of qualifica- tions. The more striking deviations from close correspondence may be listed as follows:
\Vith respect to age (Table 4 (IX)), there are no low-scoring women interviewees in the age bracket of "46 and over"; the share of this bracket
TABLE 4 (IX)
AGE DIS1RIBUTION IN TOTAL EXTREME QUARTILES AND INTERVIEWEES (NUMBERS OF THE LATTER ARE SPECIFIED IN PARENTHESES)
Age Groups
16-22
23 - 30 31- 45 46andover
Sums I
High Quartile
Low Quartile
Men
59 (6) 88 (8) 78 (5) 53 (1)
278 (20)
Women
70 (8) 55 (5) 59 (6) 51 (6)
235 (25)
Men
60 (4) 90 (10) 93 (4) 36 (2)
279 (a>)
Women
60 (9) 61 (3) 75 (3) 49 (0)
245 (15)
is added to the youngest age group. Furthermore, there is only one high- scoring male interviewee in this highest age bracket. Our interviewee sample is therefore on the younger side when compared with all the subjects.
The major deviation with respect to religion (Table 5 (IX)) is that three (i. e. , 20 per cent) of the low-scoring women interviewees are Catholics while the corresponding figure for the "low" women in our total sample is only
TABLE 5 (IX)
RFLIGIOUS AFFILIATION IN TOTAL EX1REli'IE QUARTILES AND INTERVIEWEES (NUMBERS OF THE LATTER ARE SPECIFIED IN PARENTHESES)
Men
47 (7) 185 (11) 15 (2)
13 (0) 260 ( 20)
Women
40 (7) 184 (17) 9 (0)
2 (1) 235 ( 25)
Men
29 (1) 156 (14) 65 (5)
11 (0) 261 (20)
Women
10 (3) 156 (7) 70 (5) 9 (0)
245 (15)
? JOO THE AUTHORITARIAN PERSONALITY
10 (4 per cent). On the positive side, we may single out for special mention the fact that there is close agreement of corresponding figures for both high scorers and low scorers when the categories "None" and "Blank" are pooled; it may thus be said that both high-scoring and low-scoring interviewees are representative of their extreme quartiles with respect to indifference to or rejection of religion.
As to politics (Table 6 (IX)), "liberal" women are more numerous among the interviewees than among the corresponding quartiles, especially so far
TABLE 6 (IX)
POLITICAL OUTLOOK IN TOTAL EXTREME QYARTILES AND INTERVIEWEES
Political
O u t l o o k
Liberal Conservative Leftist
Misc. and blank
Sums
High Quartile
Low Quartile
(NUMBERS OF THE LATTFB ARE SPECIFIED IN PAR~THESES)
Men
99 (7) 112 (10) 0 (0)
49 (3) 200 (~)
Women
98 (17) 109 (5) 0 (0) 28 (3)
235 ( 2 5 )
Men
156 (14) 45 (2) 22 (3) 38 (1)
261 (~)
Women
173 (12) 23 (0) 21 (2) 28 (1)
245 ( 1 5 )
as the high scorers are concerned (98 to 17, i. e. , 68 to 42 per cent). This latter fact, however, does not hold for men. The comparatively small group of leftists or radicals (covering those who gave their attitude as "socialist" or as "communist") is represented with relatively greater frequency among in- terviewees (two women and three men, all low scorers).
3. APPROACHING THE INTERVIEWEES
An effort was made to maintain anonymity for all those interviewed as well as to convince them of the fact that they would remain unidentified. Pains were taken to conceal from the interviewee the true basis of selection.
In particular, the following procedure was adopted in securing the co- operation of the prospective interviewee: After the questionnaire responses had been evaluated, the person who had administered the questionnaire appeared at one of the next meetings of the group in question and an- nounced that further information was required of some of those who had answered the questionnaire. Those selected were identified in terms of their birthdates only and asked to arrange for an appointment after the meet- mg.
At the beginning of the actual interview they were told that they had been selected on the basis of age and regional origin. The interviewers gained the impression that in this way the anxiety as to the basis of selection was sue-
? INTERVIEWS AS APPROACH TO PREJUDICED PERSONALITY 301
cessfully removed. Actually, not one of even the highest scorers ever showed signs of knowing the true reason for his or her selection, although some of them showed signs of suspiciousness of a more general nature. The reason for this naivete seems to lie primarily in the fact that most high scorers do not think of themselves as particularly prejudiced.
In most cases an invitation to be interviewed was readily accepted. The motivation seemed to be primarily the desire to talk about oneself and the implicit hope of receiving some advice in the process. To some of the subjects the added incentive of a remuneration ($3 per test or interview) seemed not unimportant.
The interviews lasted from one and a half to three hours and were usually conducted in one session. As a rule they were held in one of the offices of the Berkeley Public Opinion Study, in an atmosphere of comfort and quiet. When it was impractical or impossible for the subject to come to the office (as was the case especially with the prison group) the interviewer went to see the subject.
C. THE INTERVIEWERS
Certain specifications were also maintained as far as those conducting the interviews were concerned. Men were always interviewed by men, women only by women. All high-scoring subjects were interviewed by American- born Gentiles.
It was not feasible within the scope of the present research to administer intelligence tests to the groups taking the questionnaire. Fortunately, such tests had already been administered to some or all of the members of four groups: Maritime School Men, Employment Service Men, Psychiatric Clinic Men and Women, and San Quentin Men. No information was obtained in our questionnaire regarding years of education; this question was omitted partly because of the probable unreliability of the answers and partly because of the fear that it might make the less educated subjects defensive about the entire questionnaire. In some cases, e. g. , the college students and the pro- fessionals, the amount of schooling was relatively constant for the entire. group. For two groups, the Psychiatric Clinic patients and the Maritime School, the years of schooling had been determined in a way that seemed fairly (though not entirely) reliable.
The data on intelligence are presented in Tables 1-3 (VIII), on education in Tables 4 (VIII) and 5 (VIII). We may consider intelligence first.
1 For reports and summaries see: Murphy, Murphy and Newcomb (85); Kerr (63).
? 282 THE AUTHORITARIAN PERSONALITY
For all three groups in Tables I-J(VIII), namely the Maritime School Men, Employment Service Men, and Psychiatric Clinic Men and Women, the average IQ is significantly above the general population average (usually by about one sigma); indeed, even the lowest scorer is, except in the third group, above the population mean. This fact, namely the restriction in the "range of talent," must be considered in evaluating the results. The correla- tions for the Maritime School Men, obtained with the AGCT (Army Gen- eral Classification Test), are very similar to those obtained with the Otis Higher Form A Intelligence Test on the Employment Service Veteran Men. The correlations of these tests with the Ethnocentrism scale, Forms 45 and
Ability Test
AGCT
Mechanical Comprehension
Reading Comprehension
Arithmetical Comprehension
Test Properties
Correlation with:b
TABLE 1 (VIII)
CORRELATIONS OF THE E AND F SCALES WITH VARIOUS ABILITY TESTS {MARITIME SCHOOL MEN)
Mean a 126. 7
126. 5
92. 5
81. 2
~ R a n g e 8. 98 102-153
14. 61 66-166
AGCT
? 25
EA -. 02
-. 17
EA+B F -. 20 -. 20
. 00 -. 13
13. 04 57-121 . 55 -. 08 -. 06 -. 20 8. 88 61-105 ? 59 -. 06 -. 16 -. 16
aThe present means may be compared with the following population means: For the general population, the AGCT and Mechanical Comprehension Tests have means of 100. For the "high school graduate? populatiorr the Read- ing comprehension and Arithmetical Comprehension Tests have means of 79. On all but the last-named test, therefore, the present sample is con- siderably above average,
bThe number of cases (N) involved in the correlations are as follows:
Of the 343 subjects in the total sample, 342 received all of the ability tests, with the exception of four individuals who omitted the Reading Comprehension Test, The correlations involving AGCT and F, then, are based on an N of 342. The EA Scale, contained in Form 40, has an N of 168. while EA+B has an N of 178.
40, range between -. 02 and -. p, averaging above -. 2. The correlations ? of E with the Mechanical, Reading and Arithmetical Comprehension Tests2 (Table 1 (VIII)) are slightly lower, averaging about - . 1 . These correla- tions, taken together, are statistically significant, that is, dependably above zero, at the 5 per cent level. It may be noted also that there are no positive correlations. Table 3 (VIII) gives the mean (Wechsler-Bellevue Intelligence
2 The Bennett Mechanical Aptitude Test, the Iowa Silent Reading Test, the Stanford Adult Arithmetical Reasoning Test.
? Otis Raw Score Otis IQ
St anford-Binet
IQb
56. 5 114. 5 129. 5
ETHNOCENTRISM IN RELA TION TO INTELLIGENCE TABLE 2 (VIII)
CORRELATIONS OF THE E, F, AND PEC SCALES WITH 'THE OTIS HIGHER FORM A INTELLIGENCE TEST
(Eli! PLOYMENT SERVICE VETERAN MEN)
Otis Test Properties Meana
Range 34-75 92-133
108-140
Correlation of Otis with:c
EA: -. 32 EA+B: -. 22 F: -. 48 PEC: -. 16
aThe mean Otis IQ of 114. 5 is significantly above the population average of 100. 0 (population S. D. is 10. 0).
bThe conversion of the Otis scores into Stanford-Binet IQ scores was
done by means of a table prepared by Dr. E. E. Ghiselli. For the general population the Stanford-Binet has a mean of 100, an S. D. of 16?
Cotis Test data were available for 104 of the 105 cases in this sample. The N is, then, 104 for the correlations with F and ? PEC. The N is also 104 for EA? since the EA scores of the subjects taking Form 45 as well as of those taking form 40 were used, The N is 50 in the case of EA+B (Form 45).
Form 45 E? Scale Quartiles Low quartile
Low middle quartile
High middle quartile
High quartile
Range on E . ! '! . 10-24 8 25-36 5 37-50 13 51-70 11
37
Mean IQ 125. 3 117. 8 113. 9 107. 3 114. 9
TABLE 3 (VIII)
MEAI~ WECHSLER-BELLEVUE I Q SCORE FOR EACH QUARTILEa OF THE ETHNOCENTRISM SCALE (PSYCHIATRIC CLINIC MEN AND WOMEN)
aThe subjects represent only a part of each quartile. In all, 15 of the 50 men, and 22 of the 71 women, had received Wechsler-Bellevues. The men and women were similar with respect to proportion in each quartile, identical with respect to mean IQ. It is not clear why more upper-half than lower-half subjects have been tested. The mean of 114 approximates that for the patients generally.
? THE AUTHORITARIAN PERSONALITY
Test) IQ for the four E scale quartiles, and we find the equivalent of a low negative correlation.
Tables I-3 (VIII) indicate that, for individuals with IQ's of approximately wo and above, there is a very low but dependable negative relationship be- tween intelligence and ethnocentrism: the most ethnocentric subjects are, on the average, less intelligent than the least ethnocentric, while the middle
scorers on E are intermediate in IQ.
Data on the San Quentin Men, not presented in the above tables, suggest a similar relation between E and IQ in groups having a wider intellectual range. Wechsler-Bellevue Test Scores were available on 77 of the IIO sub- jects in the San Quentin sample. This subsample had a mean E score of 4. 68 and a Standard Deviation of I. 28, as compared with the total-sample mean of 4. 6I and S. D. of 1. 28. The mean Wechsler-Bellevue IQ (full scale) was
I09. o, the S. D. q. 8, and the range 78-q2. This subsample is, then, almost identical with the total (questionnaire) sample in E mean and S. D. ; it is slightly more intelligent than the total prison population, whose mean IQ is just under 100. The obtained r between E and IQ was -. 2 8. This value is of the same order of magnitude as those reported above for samples in which the IQ range was more constricted. It is, of course, not conclusive, since the tested sample may have been systematically biased in its selection. In addi- tion, other factors such as educational and class level are probably con- tributing to this correlation, since they seem to be at least slightly related to both E and IQ. A conclusive study of the relation between IQ and E would have to partial out, or to keep constant, these other factors. Never- theless, the series of negative r's, on a variety of groups and by means of a variety of intelligence tests, provides substantial evidence of a significant relation between E and IQ. That the correlation is greater than zero, and in a negative direction, is in keeping with previous studies as well as with the results of the present study regarding the role of stereotypy and rigidity in ethnocentrism. That the correlation is only moderately dose-apparently in the range of -. 2 to -. 4-is evidence that intelligence is only one of many variables which determine E-scale scores.
Correlations between the Ability Tests and the F scale were computed for the Maritime School Men (Table I (\llll)) and the Employment Service Veterans (Table 2 (VIII)). In the former group the correlations range be- tween -. q and -. 20 while in the latter the extremely high value of -. 48 was obtained. It appears, then, that IQ is more closely related to F than to E although, except for the Veterans, the correlation is relatively small. Further study is required to determine whether or not the r of -. 48 is spurious or exceptional. The higher correlations with F than with E might be explained on the basis of certain of the F clusters, e. g. , superstition and stereotypy; correlations between IQ and the individual F items might well be obtained
? ETHNOCENTRISM IN RELA TION TO INTELLIGENCE 285
in future research. The correlation of -. r6 between IQ and the PEC scale (Table 2 (VIII)) is consistent with other findings.
That the relation between intelligence and ethnocentrism is not very close is suggested also by the over-all group data. Thus, while the three groups in Tables I-3 (VIII) are very similar in average IQ, they vary greatly in aver- age E score. The Psychiatric Clinic patients have an E mean of 3. 7, a full point below the means for the Maritime School and Veteran Men. More- over, the latter groups, while extremely high in average IQ, are also among the most ethnocentric of all groups tested. It would seem, therefore, that high tested intelligence is no guarantee against the overall authoritarian pattern of ideology and personality.
Data on the relation of ethnocentrism to amount of education are pre- sented in Tables 4 (VIII) and 5 (VIII). One of the most striking results is
TABLE 4 (VIII)
MEAN NUMBER OF YEARS OF EDUCATION FOR EACH QUARTILE OF THE ETHNOCENTRIS\t SCALE
(PSYCHIATRIC CLINIC MEN AND WOMEN)
Form 45 E Seale Quartiles Low quartile
Low middle quartile
Hi~h middle quartile
High quartile
Range onE 10-24 25-36 37-50 51-70
Na Mean Yrs. Educationb 29 13. 8
28 12. 7
27 11. 8
28 11. 2 112 12. 4
aThese data are based on 45 of the 50 men, 66 of the 71 women, The means for men and women separately were so similar that they were combined in order to increase the N.
bone year has been added to the number of years of education in five cases where the individual had specialized training such as secretarial or accounting school,
that these two variables are much more closely related in the Psychiatric Clinic group than in the Maritime School group. The average number of years of education drops gradually but consistently (from I 3. 8 to I 1. 2) in the Clinic group as score on the E scale increases (Table 4 (VIII)). How- ever, in the Maritime School data, computed in another manner, the changes are not so consistent. On the basis of the combined Forms 45 and 40 data, subjects with less than I2 years of education (i. e. , not high school graduates) have the highest E mean, 4? 9, while those with I2-I4 years (there were no college graduates) had almost identical E means of about 4. 6. However,
? 286
THE AUTHORITARIAN PERSONALITY
Years
Education
Form 45 ( EA+B) Form 40 (Ea> T otal Group
Less than 12 12 years
13 years
14 years
Blank OVer-all
N Mean N
36 4. 38 60 104 4. 28 91 13 4. 75 7 18 4. 34 6 7 4. 63 0 178 4. 36 164
Mean N Mean 5. 21 96 4. 90
5. 04 195 4. 63 4. 40 20 4. 63 5. 17 24 4. 55
7 4. 63 5. 08 342 4. 68
TABLE 5 (VIII)
MEAN E SCORE FOR GROUPS HAVING VARIOUS YEARS OF EDUCATION (MARITIME SCHOOL MEN)
there is some question as to whether the results for the two Forms should be combined, since the results for each Form separately are not so clear-cut. In the Form 45 group there is no consistent trend, the subjects with 13 years of education having the highest E mean and the other educational levels varying only within a range of o. I points. In the Form 40 group, on the other hand, the 13-year level is the least ethnocentric, while the 14-year group vies with the less-than-12 for the most ethnocentric position. The only difference that holds up for both Forms is that between the 12 and the less-than-12 year levels, and this difference borders on the 5 per cent level of statistical significance.
Why is the relation between ethnocentrism and education more consistent in the Psychiatric Clinic group? One possibility is that the Maritime School members who had one or two years of college and then dropped out are systematically atypical, and that an unselected group of college students might be less ethnocentric. To the extent that this is true, a clear-cut relation- ship between ethnocentrism and education does exist. However, the relation in the case of the Psychiatric Clinic may be spuriously high, since the college students and college graduates in this sample are not a random sample of these educational levels. It is possible-though not yet demonstrated-that the col- lege-trained patients are, to a greater degree than those who did not attend college, selected for willingness to recognize the need for, and to seek, psychological aid. To the extent that this is true, the relationship between ethnocentrism and education is less clear-cut than these results indicate.
One might venture the hypothesis that ethnocentrism is at least as closely related to the desired amount of education as to the actual amount. Thus the two Extension Classes (Forms 78 and 45), most of whose members were
? ETHNOCENTRISM IN RELA TION TO INTELLIGENCE 287
adults having only 12 years or less of schooling, but trying to "learn some- thing on the side" and perhaps even to obtain a college degree, had E means as low as those of the University of California students. There is also some clinical evidence to support this hypothesis.
At any rate, examination of the various group means shows that two groups may have similar educational levels and very different E means, as well as similar E means and very dif- ferent educational levels. For example, the University of California students and the George Washington University Students, with similar educational levels, are significantly different in E means (Form 78, A-S scale, and Form 40, E scale). Again, the Working-Class group, though similar in socioeco- nomic and educational background to the San Quentin group, is significantly less ethnocentric (Form 45).
We may tentatively conclude that ethnocentrism shows a slight negative correlation with amount of education. It is likely, though far from a demon- strated fact, that college graduates are less ethnocentric than high school graduates, who are in turn less ethnocentric than those who did not complete high school. It is not clear which is more important: that the correlation is greater than zero, or that it is at best not far from zero. To those who urge education per se as a kind of panacea, the smallness of the correlation ought probably to be stressed. But this is not to deny the importance of education. It is, rather, to emphasize that our educational system, college as well as public school, is still far from realizing its potential strength as a social force in the service of democratic values. The reasons for this are outside the scope of the present research. It may also be pointed out that, even under the best educational conditions, exposure to the classroom is not enough, and that motivation to learn and receptivity to new ideas provide the only psycho- logical soil in which democratic education can develop effectively.
In summary, ethnocentrism seems to have a low but statistically significant relation to both intelligence and education, the most educated and intelligent subjects being, on the average, the least ethnocentric. However, these varia- bles were studied only secondarily in the present research, and convincing determination of their relation to ethnocentrism requires more extended sampling, particularly of the lower educational and intellectual levels. It is also necessary to control more adequately the operation of other variables such as class level, educational opportunity and educational motivation. Nevertheless, the present results do contradict seriously one of the com- monly held theories of prejudice and fascism, namely, that they are sup- ported out of simple stupidity, ignorance or confusion. It would seem, rather, that an autocratic social structure is best suited to the particular type of ra- tionality exercised by the authoritarian personality. A promising field of future research is the study of what might be called "the dynamics of in- telligence. " For example, the intellectual functioning of ethnocentric indi- viduals, even those with above-average IQ's, seems to be relatively rigid,
? THE AUTHORITARIAN PERSONALITY
to work better in relation to things than to people, to be primarily extra- ceptive, and to become disrupted when required to deal with more psy- chological issues, especially those involving personal needs and emotions
(anti-intraceptive).
As has been noted elsewhere (particularly in Chapter IV), the average IQ
and the educational level of the entire sample used in the present research are probably somewhat above those of the general population, or, rather, above those of the urban middle class (our primary reference population). This sampling bias, together with that of age-our sample being somewhat younger than a representative sample would be-has probably resulted in our obtaining over-all means for the various scales which are slightly biased in the direction of being too low. However, the error seems to be less than might have at first been suspected. In addition, it is not likely that such sampling factors have distorted to any appreciable degree the relationships among the variables of ideology, personality, and group membership under investigation. Since we were primarily concerned with the causes and cor- relates of antidemocratic trends, that is, with correlations and differences, rather than with the average amount of any single trend per se, the diverse groups comprising the total sample provide, it would seem, an adequate basis for study.
? CHAPTER IX
THE INTERVIEWS AS AN APPROACH TO THE PREJUDICED PERSONALITY Else Frenkel-Brunswik
A. INTRODUCTION: COMP ARISON OF GROUPS
Reference to the interviews has so far been limited to the discussion of two individual cases, with emphasis mainly on the pattern of social and po- litical issues (Chapter II). A series of five chapters beginning with the present will show the purpose and value of the interviews in their own right, cover- ing in a systematic fashion a variety of topics; furthermore, analysis will be in terms of groups rather than of single individuals (Chapters IX to XIII). Some special individuals or groups as well as some special issues will be taken up once more in later chapters, bringing in additional material from the interviews (Chapters XX to XXII). The problem of ideology as revealed through the interviews is not being considered in the series of chapters which makes up the present Part II; this problem will be taken up in Chap- ters XVI to XIX.
The major advantage of the interview technique lies in the scope and free- dom of expression it offers to the person being studied. Thus we may learn what he thinks about himself, about his hopes, fears and goals, about his childhood and his parents, about members of the other sex, and about peo- ple in general. It is through careful and critical evaluation of sources of this kind that an adequate view of the total personality can perhaps best be ap- proximated.
Rather than making an attempt to establish the dynamic interrelationships of the significant factors for each single individual, however, we shall look for a basis of generalizations within groups which will permit us, it is hoped, to come to grips with the social and psychological trends typical of the highly prejudiced and of the unprejudiced. Does the family constellation differ in the typical prejudiced home as compared with the typical unprejudiced home? Do prejudiced individuals tend to have different images of their parents than do unprejudiced ones? How does the handling of discipline vary
291
? THE AUTHORITARIAN PERSONALITY
? in this respect? Do prejudiced and unprejudiced individuals differ in their sex life, their way of choosing friends, their values, their general cognitive and emotional approach to life? These are some of the questions which will be the prime concern of the present series of chapters. By virtue of its speci- ficity and unique character, the interview may be called upon to yield, in the first place, information of this broader kind on the personality of those scoring high or low on anti-Semitism.
In fact, it was a preliminary review of some of the crucial factors of this kind in a set of exploratory interviews which led to the construction of the F scale (Chapter VII). However, it was this and other scales, consisting of given statements calling for indications of agreement or disagreement only, upon which group comparisons between the personalities and the attitudes of prejudiced and unprejudiced subjects-now to be extended to the richer and more flexible type of evidence as given by an interview-have been thus far exclusively based in the present volume. On this comparatively limited basis, marked differences between these two groups were established.
Questions as to the specific meaning and connotation of the various state- ments for the individuals concerned, however, had to be left open to a con- siderable extent. Further validation of the questionnaire data can be effected by probing in greater detail into the spontaneous elaborations a subject may be willing to make on the topics covered by the questionnaire. To obtain such additional information on results gathered by other techniques is a further goal in the subsequent analysis of the interviews.
While the importance of the interview as source material is generally agreed upon in the social sciences, there are specific difficulties in evaluating such material. This is due mainly to the fact that interview material is highly diversified and unstructured. At the same time, the richness, flexibility, and spontaneity of this material are the features which constitute its major asset; room is left for unanticipated variations. To preserve all of this uniqueness and flavor, we should have to reproduce all, or at least the most outstanding, protocols in full-allowing the reader to form his own impressions and draw his own conclusions.
Presentation in full, though doing maximal justice to the material, has its serious drawbacks. It would not be easy to survey and would of necessity leave to the reader much of the burden of interpretation, or else introduce a potentially arbitrary distribution of emphasis in the process of interpreta- tion. In the same manner, if we were to limit presentation itself to a few select protocols we might easily be criticized on grounds of possible ar- bitrariness in selection.
For these reasons it was decided to attempt some kind of quantification within groups, rather than to present only clinical results based on intensive but more impressionistic case studies as was originally planned. For the kind of evaluation chosen, hypotheses were already formulated on an empirical
? INTERVIEWS AS APPROACH TO PREJUDICED PERSONALITY 293
basis, giving "hunches" for potential validation. The procedure consisted of a careful evaluation of the interview material in terms of an extensive set of scoring categories. These categories had been designed to encompass as much as possible of the richness and intricacy of the material at hand. (See E. Frenkel-Brunswik (3 r, 32, 36) ). They were a product of intensive study of the intervie\vs with full consideration of all the other evidence obtained from the individuals in question, especially their standing on the prejudice scales. The result was establishment of a Scoring Manual comprising about ninety categories and subcategories (see below).
Evaluation of the interview protocols was by raters unfamiliar with the specific ideology of the subject, the Scoring Manual serving as a guide for ratings in terms of the various categories.
In order to offer to the reader as much direct contact with the raw ma- terial of the interviews as possible, numerous quotations are inserted into the subsequent chapters. The raw material is arranged through the medium of the scoring categories, and the relationship of the latter to, or even their de- pendence on, the original material will become evident there. Many of the quotations presented were directly instrumental in designing the categories employed in their evaluation. Actually, the system of scoring categories reflects the theory or the interrelationships between personality and prej- udice which was empirically developed in the course of the exploratory study of the bulk of the interviews, individual by individual. This exploratory study preceded the more elaborate checking procedure in which the indi- vidual lost his identity in a mass of statistical evidence organized in terms of the scoring categories and evaluated in terms of larger groups. It is only through such a statistical procedure that the original hypotheses can be, and in fact have to a considerable extent been, verified.
It was hoped that use of the variables defined by the scoring categories would help to bridge the existing gap between the studies of groups and of individuals and perhaps contribute to the establishment of a mutual give and take of facts and concepts. Indeed, some of the variables and relationships which were originally conceived of in the course of the generalized, sta- tistical establishment of personality patterns in samples of prejudiced vs. un- prejudiced people, were at the same time found to be crucial in the intensive study of single individuals or small groups (see Chapters XX, XXI, XXII).
The subject's view of his own life, as revealed in the course of the inter- view, may be assumed to contain real information together with wishful -and fearful-distortions. Known methods had to be utilized, therefore, and new ones developed to differentiate the more genuine, basic feelings, at- titudes, and strivings from those of a more compensatory character behind which are hidden tendencies, frequently unknown to the subject himself, which are contrary to those manifested or verbalized on a surface level. To cope with such distortions cues are available or may be developed to guide
? 294 THE AUTHORITARIAN PERSONALITY
interpretations. The methodological safeguarding of such interpretations is one of the central problems in the approach to the interviews. The subse- quent analysis of the interview data will include discussion of this point. In general, it endeavors to add to our knowledge of the relationship of surface cues and underlying strivings, with special reference to the problems raised by the personality of the ethnically prejudiced.
In the present chapter the securing of the interview material and the tech- nical aspects of its analysis will be discussed. The sample of the subjects in- terviewed as compared with the total sample will be described first. A characterization of the interviewers in terms of their background, training, and psychological point of view will also be given. Next, the Interview Sched- ule used and the technique employed in interviewing will be presented. This will be followed, in the concluding sections of this chapter, by a discussion of the methods used in the evaluation of the interview data.
In the four following chapters a statistical analysis and discussion of the results gained from the study of the interviews will be presented, first in terms of a detailed set of rating categories (Chapters X to XII), and then in terms of over-all ratings and comprehensive description (Chapter XIII).
B. SELECTION OF SUBJECTS FOR THE INTERVIEWS
1. BASIS OF SELECriON
The selection of the subjects to be interviewed was determined, in the first place, by their responses on the A-S or the E scale. With few exceptions (see below), all interviewees belonged either to the uppermost or to the lowermost quartile in this respect, the proportions of high-scoring and low-
scoring subjects being approximately equal.
Secondly, consideration was given to the response to the three scales of
the questionnaire. Thus, an effort was made to include in the sample inter- viewed not only the most "typical" high scorers and low scorers, i. e. , sub- jects with correspondingly high or low scores on the PEC and F scales, but also some of those more atypical subjects who obtained a high score on the first scales but a relatively low score on one or both of the others.
Thirdly, an effort was made to balance our samples of high-scoring and low-scoring subjects in terms of age, sex, political and religious affiliation, as well as national or regional background.
Of the thirty to forty different socioeconomic groups to which the ques- tionnaire had been administered (see Chapter IV), subjects for interviews were selected from the following twelve: Psychiatric Clinic Patients from the Langley Porter Clinic of the University of California (men and women, abbreviated LPC); University of California Public Speaking Class (men
? INTERVIEWS AS APPROACH TO PREJUDICED PERSONALITY 295
and women, PSM and PSW); Alameda School for Merchant Marine Of- ficers (men, Maritime); San Quentin State Prison Inmates (men, SQ); Uni- versity of California Extension Testing Class (men and women, TC); Uni- versity of California Extension Psychology Class (men and women, EG); University of California Summer Session Education Class (men, EdPs); Stu- dents at the Pacific School of Religion (men, PSR); Employment Service Veterans (men, Vets); Professional Women-public school teachers, social workers, public health nurses (N and R W); University of Oregon Summer Session Students (women, OG); Students at the University of California Medical School (women, Med).
In all, approximately one hundred persons were interviewed. Some of the interviews could not be used in the final scoring, however. One reason for this was that some of the subjects scoring at the very extreme ends of the F scale distribution had been used by the scorers in a last checkup on the scoring manual and had therefore to be excluded later from the main analysis which was to be a "blind" one (see below). Other records had to be discarded because of their brevity or barrenness.
The results to be reported in the subsequent chapters are based on the records of So interviewees, 4 0 men and 4 0 women. Of the men, 2 0 were high extremes on the E scale; and 20 were low extremes. For the women, the corresponding numbers were 2 5 and I 5. The survey presented in Tables I (IX) and 2 (IX) shows for each interviewee the code number, group extrac- tion, standing on responses to the various scales of the questionnaire, with parentheses used to designate membership in one of the middle quartiles.
The rater's "blind" diagnosis of the interview responses makes up the right half of the tables. It is to receive full discussion in Chapter XIII.
2. REPRESENTATIVENESS OF THE INTERVIEWEES
A breakdown with respect to further characteristics of the interviewees, and a comparison of the samples interviewed-approximately one-tenth of the total of the groups mentioned above-with the entire upper and lower quartiles of our over-all samples, will show that our interview samples are fairly representative of the extreme quartiles defined in terms of overt anti- Semitism or ethnocentrism. A quantitative comparison is given in Table 3 (IX). Inspection of the means of all the subjects falling into the upper and lower quartiles with those of corresponding groups of interviewees reveals a sufficiently close agreement. The interviewee samples are, more often than not, somewhat farther to the extreme end of the scale than the correspond- ing total extreme quartiles. This trend holds in spite of the fact that, as seen from Tables I (IX) and 2 (IX), in a few instances interviewees had to be taken from the extreme ends of the middle quartiles of the E scale.
A further breakdown has been undertaken with respect to age, religion,
? THE AUTHORITARIAN PERSONALITY
TABLE 1 (IX)
SURVEY OF 20 PREJUDICED AND 20 UNPREJUDICED MEN INTERVIEWEDa
Standing on
Groupb
Ml LPC h h (h) 1 27 1 H
Code No.
~estionnairec
Interview Scoresd "High" "Low"
A-S E F PEC
M4 Ps. t h h 1 h 47 2 H
M6 PSM h h h h 41 2 H
M7 PEM h h (l) 1 6 32 L L Mll PSM h h h h 52 2 H H Ml3 PSM h h h (h) 52 1 H H Ml4PSM hh(h)(1) 8 32 L L M17Ps. th(h)1h254 H H MIS LPC h h h h 33 3 H H M20Maritime-h11540 L L M40 SQ h h (h) h
M41 SQ h h h h
M43 SQ h h h 1
M45 SQ h h h h
M46 TC h h 1 h
M47 SQ h h h h
M5l SQ h h h h
M52 SQ h h h h
M57 SQ h h h h
M58 TC h h h h
Means of 20 prejudiced men interviewees
M2 EG 1 1 1 1
M3 PEM 1 1 (1) 1
M5 PSM 1 1 1 1 MSPStl1(1)1h544 M9 PSM 1 1 1 (h) 39 2 M1oEdPs 1111
M12EdPs 1111
Ml5 LPC 1 1 1 1
M16 LPC 1 1 1 1 M19Ps. t11(l)(h)350 H H M42Maritime-111157L L
M44 PSR 1 1 1 1 2 54
M48Vets 1111 6 37
M49TC 1111442
M50 SQ 1 1 1 h JO 34 M53Vets1111152
M54 SQ 1 1 1 h 12 24
M55TC 1111456
M56 SQ 1 1 1 1 5 41
M59 SQ 1 1 1 h 14 39 L L
Means of 20 9. 4 38. 0 unprejudiced men
interviewees
"For discussion of the evaluation of the interviews and of the results shown in this table, see Section F of the present chapter, and Chapter XIII.
bFor key to abbreviations, see text, p. 294/95.
cThe upper and lower middle quartiles are indicated by the use of paren-
theses with the letters h and 1.
? Number of ratings other than "Neutral. " Number of Neutrals is ob-
tained by subtracting that of "High" and of "Low" from 72 (on Table 1 (IX)) or 65 (on Table 2 (IX)). For selection of categories, seep. 335.
Composite Standing on 72 Categories
Intuitive Over-all Rating of Interviews
H H H
55 2 H H 49 2 H H 43 3 H H 48 2 H H 42 6 H H 44 2 H H 36 5 H H 51 4 H H 56 0 H H 54 2 H H
38. 7 7. 3
5 52 0 53 3 34
L L L L L L L L H ff H H L L L L L L
337 245 2 43 6 44
L L LL L L L L L L L L L L L L
? F22 PSW
F24 PSW
F25 LPC
h (h) h (h) 28 1
INTERVIEWS AS APPROACH TO PREJUDICED PERSONALITY 297 TABLE 2 (IX)a
SURVEY OF 25 PREJUDICED AND 15 UNPREJUDICED WOMEN INTERVIEWED
Group
H hhhh3'1 4 H hhhh3) 7 H F26N hh(l)h922 L
F28 RW hhhh197H F31 PSW hh(h)h510H
Intuitive over-all Rating of Interviews
H H H L H H H L H H H H H H L H L H H H L H H H H
H L L L L L L H L L L L L L L
U>de No.
standing on Questionnaire
A-S E F PEC
Interview Scores
"High" "Low"
Q:>mposite standing on 65 categories
F32N hhhh34 3 F33 TC h h h h 3 32
H L H H H H H H? L H L H H H H H H
F36 TC F37 EXi
hh(h)(l)'n 5 h(h)hh21 10 hhhh25 14 hhh(1)245 hhh(h)30 6 hh11238
F38 PSW
F39a N
FOO RW
F61 LPC
F64 RW h h h h 0 26 F66 PSW hhh1357 F67 RW hhhh328 F68N h(h)1h32 4
F69 PSW
F71 PSW
F72 LPC
F74 PSW
F77 LPC
F78 PSW
F79 OG
Means of 25 prejudiced women interviewees
hhhh37 7 hhhh47 2 hhhh17 26 hhhh45 4 hhh13'1 4 h h h (h) 44
h h h (h) 36
27. 4 9. 5
F21PSW 111h24 0
H L L L L L L H
L L L L L L
F23 TC 1 1 (1) 1
F'n PSW 1 (l) 1 1
F29 LPC 1 1 1 1
F30 RW 1 1 1 1
F34 PSW 1 1 (l) 1
F35 TC 1 1 1 1
F39 PSW 1 1 1 1
F62 PSW 1 1 1 h 1 44 L F63 LPC 1 1 1 1
F65 PSW 1 1 h 1 F70 Med 1 1 1 1 F73 PSW 1 (1) 1 1 F75 PW 1 (l) (h) 1 F76 PSW 1 ( 1 ) ( h ) ( l )
Means of 15 unprejudiced women interviewees
385
4 42 6 44 0 38 1 35 343
14 2 2 7. 0 36. 2
asee footnotes to Table 1 (IX).
2 46 0 53 7 36 3 44 149 1 42
1 H 4 H
? Questionnaire form
T otal of a ll taking Forms Range Mean
subjects
High-scoring interviewees Mean N
66. 5 7 70. 6 18
58. 6 12 59. 0 6
Low- scoring Low-quartile interviewees
78:
45:
Men 17-82 46. 4 Women 14-86 46. 2
Men 10-70 43. 2 Women 10-70 35. 4
Mean
29. 1 28. 2
25. 0 15. 6
Mean N
30. 3 6 33. 6 10
15. 8 11 13. 6 5
TABLE 3(IX)
REPRESENTATIVENESS OF INTERVIEWEES IN TERMS OF SCORES ON THE ETHNOCENTRISM SCALEa
78
or 45 N
52 243
390 130
High~ quartile Mean
64. 6 69. 5
59. 7 54. 8
aSince 75 out of the 80 interviewees are from among the subjects tested by Form 78 or Form 45, comparisons are here limited to these two groups.
N
00 "'
? Religious
A ffiliation
Catholic Protestant None
Blank
Sums
High Quartile
Low Quartile
INTERVIEWS AS APPROACH TO PREJUDICED PERSONALITY 299
and politics. To maintain anonymity, these data are not included in the tables just mentioned which deal with individual subjects, but are presented in a statistical manner in Tables 4 (IX) to 6 (IX). For men, a few data on religion and politics are missing; hence the discrepancies in the sums relating to the total quartiles.
On the whole, the distributions of the interviewees and of corresponding extreme quartiles are not at too great odds with one another, considering the difficulties in finding subjects with the exact combination of qualifica- tions. The more striking deviations from close correspondence may be listed as follows:
\Vith respect to age (Table 4 (IX)), there are no low-scoring women interviewees in the age bracket of "46 and over"; the share of this bracket
TABLE 4 (IX)
AGE DIS1RIBUTION IN TOTAL EXTREME QUARTILES AND INTERVIEWEES (NUMBERS OF THE LATTER ARE SPECIFIED IN PARENTHESES)
Age Groups
16-22
23 - 30 31- 45 46andover
Sums I
High Quartile
Low Quartile
Men
59 (6) 88 (8) 78 (5) 53 (1)
278 (20)
Women
70 (8) 55 (5) 59 (6) 51 (6)
235 (25)
Men
60 (4) 90 (10) 93 (4) 36 (2)
279 (a>)
Women
60 (9) 61 (3) 75 (3) 49 (0)
245 (15)
is added to the youngest age group. Furthermore, there is only one high- scoring male interviewee in this highest age bracket. Our interviewee sample is therefore on the younger side when compared with all the subjects.
The major deviation with respect to religion (Table 5 (IX)) is that three (i. e. , 20 per cent) of the low-scoring women interviewees are Catholics while the corresponding figure for the "low" women in our total sample is only
TABLE 5 (IX)
RFLIGIOUS AFFILIATION IN TOTAL EX1REli'IE QUARTILES AND INTERVIEWEES (NUMBERS OF THE LATTER ARE SPECIFIED IN PARENTHESES)
Men
47 (7) 185 (11) 15 (2)
13 (0) 260 ( 20)
Women
40 (7) 184 (17) 9 (0)
2 (1) 235 ( 25)
Men
29 (1) 156 (14) 65 (5)
11 (0) 261 (20)
Women
10 (3) 156 (7) 70 (5) 9 (0)
245 (15)
? JOO THE AUTHORITARIAN PERSONALITY
10 (4 per cent). On the positive side, we may single out for special mention the fact that there is close agreement of corresponding figures for both high scorers and low scorers when the categories "None" and "Blank" are pooled; it may thus be said that both high-scoring and low-scoring interviewees are representative of their extreme quartiles with respect to indifference to or rejection of religion.
As to politics (Table 6 (IX)), "liberal" women are more numerous among the interviewees than among the corresponding quartiles, especially so far
TABLE 6 (IX)
POLITICAL OUTLOOK IN TOTAL EXTREME QYARTILES AND INTERVIEWEES
Political
O u t l o o k
Liberal Conservative Leftist
Misc. and blank
Sums
High Quartile
Low Quartile
(NUMBERS OF THE LATTFB ARE SPECIFIED IN PAR~THESES)
Men
99 (7) 112 (10) 0 (0)
49 (3) 200 (~)
Women
98 (17) 109 (5) 0 (0) 28 (3)
235 ( 2 5 )
Men
156 (14) 45 (2) 22 (3) 38 (1)
261 (~)
Women
173 (12) 23 (0) 21 (2) 28 (1)
245 ( 1 5 )
as the high scorers are concerned (98 to 17, i. e. , 68 to 42 per cent). This latter fact, however, does not hold for men. The comparatively small group of leftists or radicals (covering those who gave their attitude as "socialist" or as "communist") is represented with relatively greater frequency among in- terviewees (two women and three men, all low scorers).
3. APPROACHING THE INTERVIEWEES
An effort was made to maintain anonymity for all those interviewed as well as to convince them of the fact that they would remain unidentified. Pains were taken to conceal from the interviewee the true basis of selection.
In particular, the following procedure was adopted in securing the co- operation of the prospective interviewee: After the questionnaire responses had been evaluated, the person who had administered the questionnaire appeared at one of the next meetings of the group in question and an- nounced that further information was required of some of those who had answered the questionnaire. Those selected were identified in terms of their birthdates only and asked to arrange for an appointment after the meet- mg.
At the beginning of the actual interview they were told that they had been selected on the basis of age and regional origin. The interviewers gained the impression that in this way the anxiety as to the basis of selection was sue-
? INTERVIEWS AS APPROACH TO PREJUDICED PERSONALITY 301
cessfully removed. Actually, not one of even the highest scorers ever showed signs of knowing the true reason for his or her selection, although some of them showed signs of suspiciousness of a more general nature. The reason for this naivete seems to lie primarily in the fact that most high scorers do not think of themselves as particularly prejudiced.
In most cases an invitation to be interviewed was readily accepted. The motivation seemed to be primarily the desire to talk about oneself and the implicit hope of receiving some advice in the process. To some of the subjects the added incentive of a remuneration ($3 per test or interview) seemed not unimportant.
The interviews lasted from one and a half to three hours and were usually conducted in one session. As a rule they were held in one of the offices of the Berkeley Public Opinion Study, in an atmosphere of comfort and quiet. When it was impractical or impossible for the subject to come to the office (as was the case especially with the prison group) the interviewer went to see the subject.
C. THE INTERVIEWERS
Certain specifications were also maintained as far as those conducting the interviews were concerned. Men were always interviewed by men, women only by women. All high-scoring subjects were interviewed by American- born Gentiles.
