) It was recognized that these five items do not constitute a scale in the more technical sense, but this loss seemed
justified
by the gain in applicability to vanous groups.
Adorno-T-Authoritarian-Personality-Harper-Bros-1950
28 3.
22 4.
05 4.
00 3.
89 2.
87 2.
54 3.
61 1.
50 0.
84
2. 84
Mean
a. 40 3. 74 3. 88 4. 38 3. 93 4. 32 3. 44 2. 13 3. 19 2. 82 2. 10 2. 26
3. 30
D. P. Rank D. P.
2. 89 (9) 3. 26 (7) 4. 00 (2)
(J. businessmen) (Negro rights) (Marry a Jew)
(Negro foreman) (World organization) (Jews alike) (Negroes lazy) (Jewish neighbors) (Foreign ideas)
'"'! :II t"J
>
4. 11 3. 79 4. 94 4. 00 4. 49 3. 70 1. 81 3. 30 2. 74 2. 34
Mean per item 3. 43 3. 49
(No Japs)
(Okies) 2. 62
1. 80 (12) > t"' . . . . . . . . ,
TABLE 15 (IV)
MEANS AND DISCRIMINA'OORY POWERS OF THE E-SCALE ITDIS (FORM 60) a
. . . .
N
0\
aThe four groups on which these data are based are:
Group I: Group II: Group III: Group IV:
University of Oregon Student Women (N =47).
University of Oregon and . University. of California Student Women (N =54). University of Oregon and University of California Student Men (N =57). Oregon Service Club Men (N =68).
Group I Mean D. P.
Group II Mean D. P.
Group III Mean D. P.
Group IV
Over-all
4. 55 4. 15 4. 69 4. 23 4. 38 1. 46 4. 00 3. 54 2. 69 2. 39
3. 70 4. 50 3. 87 4. 54 3. 15 2. 04 3. 07 2. 56 2. 13 2. 41
3. 25
3. 46 (5) ~ :II
4. 22 (1) 0 i;d
. . . . . 3. 69 (4) '"'! >
i;d
3. 41
2. 77 3. 40 3. 46 3. 61 3. 53 3. 91 2. 93 1. 86 a. oo 2. 75 1. 93 2. 19
2. 95
i;d 3. 82 (3) . . . . . > z
2. 02 ( 11)
3. 38 (6) t%j
3. 14 (8) 0
2. 09 (10) 3. 15
~
"0
"'
z
? THE STUDY OF ETHNOCENTRIC IDEOLOGY 127
parable groups of men and women has been found, as may be noted below in the results on additional groups (p. I33 ff. ).
c. ITEM ANALYSIS. The results of the item analysis of theE scale (Form 6o) are presented in Table I5(IV). The average D. P. of 3? I5 is very satis- factory. The three lowest D. P. 's (1. 8-2. I) were obtained by the items having the lowest means (2. I-2. 3). Two of these items, numbers 42 and 55, ob- tained similar means on previous forms of the scale. They were included here, slightly revised, with the expectation that the present groups might agree more strongly. This expectation was not borne out. In view of the relatively strong rejection of Oklahomans in California, the low mean and D. P. of Item 59 are probably due more to faults in formulation than to the inadequacy of the idea which we intended to express. Even the three poorest items, however, differentiate significantly and with a minimum of overlap between the high and low quartiles, the low scorers being strongly opposed (almost uniform responses of - 3 ) , the high scorers tending to disagree only slightly. 9
The rank order of goodness of items is, on the whole, consistent with previous results. The five best items ( I4, 2I, 28, 33, 38) include two referring to Negroes, two to Jews, and one to world organization; these items ranked similarly in earlier forms. Item 3 (Zootsuiters) has a rank of 9, as compared with a rank of I on Form 78. The drop may well be due to the fact that the zootsuiter issue was less focal, and therefore less likely to produce extreme agreement or disagreement, in Oregon than in California. The groups taking Form 6o agree quite well among themselves regarding the relative level of acceptability (mean) and level of discriminability (D. P. ) ofthe items. While rank-order correlations between the groups were not computed, it appears from inspection of Table I5(IV) that the mean or D. P. rank of each item is fairly stable from group to group.
2. THE FOURTH FORM OF THE E SCALE (FORMS 45 AND 40)
The fourth and final form of the E Scale (see Table I6(1V)), as used in Form 45, involved the deletion of two items ("Japs" and "Okies") from the previous form, the other ten items remaining intact. It was understood that this contraction of the E scale would eliminate many of its qualitative functions and would probably lower its reliability to the minimum required by the present research purposes. Nevertheless, the intention to distribute the questionnaire to a wide variety of groups, many of whom would have very little time for filling it out, made every reduction in size seem desirable. The entire Questionnaire Form 45 could, accordingly, be filled out in ap- proximately thirty minutes-short enough so that, for example, a club or
9 The group mean for each item is usually about midway between the means for the high and low quartiles. It is therefore legitimate to infer the quartile means from the item mean and D. P.
? 128
EA 5?
EA ro.
EA rs.
EA 20.
EB 24.
EB 28. EB 32?
EB 36. EB 40. EA 45?
THE AUTHORIT ARIAN PERSONALITY TABLE r6 (IV)
THE FouRTH FoRM OF THE E ScALE (FoRMS 45 AND 40)"'
Zootsuiters prove that when people of their type have too much money and freedom, they just take advantage and cause trouble.
Negroes have their rights, but it is best to keep them in their own dis- tricts and schools and to prevent too much contact with whites.
The worst danger to real Americanism during the last 50 years has come from foreign ideas and agitators.
It would be a mistake ever to have Negroes for foremen and leaders over whites.
One trouble with Jewish businessmen is that they stick together and prevent other people from having a fair chance in competition.
I can hardly imagine myself marrying a Jew.
If Negroes live poorly, it's mainly because they are naturally lazy, ignorant, and without self-control.
There may be a few exceptions, but in general, Jews are pretty much alike.
The trouble with letting Jews into a nice neighborhood is that they gradually give it a typical Jewish atmosphere.
If and when a new world organization is set up, America must be sure that she loses none of her independence and complete power in matters that affect this country.
a The five E A items constitute the entire E scale in Farm 4 0 . The total ten-item scale was given, numbered as it appears above, in Form 45? The reliability of Form 45 was obtained by correlating the 5 E A items with the remaining 5 EB items. It will be noted that there are no items about Jews in EM it being desired as a practical aim to construct a ques- tionnaire which would give an index of anti-Semitism without mentioning Jews at all.
organization could take it during a meeting, just before hearing a talk not directly related to the questionnaire.
Demands of practicality and expediency forced an additional compromise. Questionnaire Form 40 was even shorter than Form 45; in addition to con- tractions of other techniques, the E scale in this form was reduced to five items (as shown in Table r6(1V) ). A primary reason for Form 40 was that certain groups might be unable to spare even the thirty minutes required by Form 45? An additional consideration in the contraction of the E scale, how- ever, was the possibility that, in certain groups at least, the items referring to Jews might be too "controversial" or might focus attention too directly on the issue of prejudice. Accordingly, the five EA items in Form 40 contain no direct reference to Jews. They deal, rather, with Negroes, zootsuiters, foreigners, and "world organization. " (In Form 45 the E scale contains, in addition to these, five EB items, four referring to Jews, one to Negroes.
) It was recognized that these five items do not constitute a scale in the more technical sense, but this loss seemed justified by the gain in applicability to vanous groups.
Our conclusions regarding the advantage of using Form 40 ought perhaps to be noted for those faced with similar problems. Although it avoided focusing atten-
? THE STUDY OF ETHNOCENTRIC IDEOLOGY 129
cion on Jews, the loss in terrns of research aims was not sufficiently compensated for by the small gain in time nor by elimination of resistance. Indeed, the resistance encountered seemed to be based as much on the other phases of the questionnaire as on the E scale. Probably the basic opposition psychologically was to being "investigated" at all in an intensive way. Unlike the usual several-question poll, this questionnaire seemed, to many a subject, to identify him as a total individual even though he knew that his anonymity was preserved. In some cases this was highly anxiety-producing despite our careful attempts at reassurance and at ex- plaining the entire procedure in terms of an impersonal, public opinion, nonindi- vidual approach. In some cases it was impossible to gain the cooperation of the leadership of a group; in other cases cooperative leaders were unable to put the idea across or to have it carried out. Difficulties of this sort were as great with Form 40 as with Form 45? Once a group was induced to cooperate, there were very few omissions of questions or scale items in either form. In short, resistance was related more to the general nature of the questionnaire than to any specific in- dividual items. Form 45 might therefore have been used on practically all of the groups tested. When it is absolutely necessary to delete certain items-e. g. , if one were testing groups with a large Jewish or Negro membership and items referring to these groups might cause friction-probably the best procedure would be to have alternative items to replace those deleted.
While the number of groups which were actively but unsuccessfully ap- proached is not large, there is some indication that resistance of the type men- tioned above is correlated positively with ethnocentrism. For example, among the "Middle-Class Women" (Table 15(V)) there was an exclusive club which "just barely" decided to cooperate and which refused even to consider our request for volunteers to be interviewed. This group obtained one of the highest E means of all groups tested. Such resistance was seldom encountered in less ethnocentric groups. This difficulty might have been ex- pected on the basis of the ethnocentrists' tendency toward self-deception and concern with prying, which was expressed indirectly in the responses on the A-S and E scales, and which is brought out more directly in the chapters that follow.
Considerations of this type are of great importance in any attempt to generalize from a research sample to a broader population. Thus, because of the greater resistance of ethnocentrists to psychological investigation, it is likely that the average degree of ethnocentrism (over-all mean E score) in our total sample is somewhat lower than that which would be found in a truly random or truly representative sample. Even in the more customary public opinion polls, where population areas are often mapped out in ad- vance (stratification or other attempt at representative sampling), an ade- quate sample may not be achieved because, in their door-to-door polling, interviewers cannot reach those subjects who are unreceptive to the idea of being "tested. "lO
10 The common assumption that "any 50 people" within a given area or income level will do, and that errors of sampling on an individual level will cancel each other out, overlooks the likelihood that receptivity may correlate with what is being polled.
? 1 3 0 THE AUTHORITARIAN PERSONALITY
It seems necessary, therefore, in describing the groups on whom data were obtained, to mention briefly the nature and adequacy of the sampling pro- cedure.
a. THE SAMPLE AND THE SAMPLING PROCEDURE. The distribution of Forms 45 and 40 took place during the latter part of 1945 and the first half of 1946, a period of about nine months in all. Form 45 was given to the following groups: 11
1. Testing Class Women (N = 59). This was an adult evening class given by the Extension Division of the University of California. Since it was a class in Psychological Tests, it probably attracted a more diversified group than does the usual adult class in psychology. It was expected to contain not only individuals seriously interested in understanding themselves better-individuals who, as we shall see later, are not likely to be extremely ethnocentric-but also persons in- terested in psychology more as a means of manipulating others. The class was also varied with respect to age (range about 2o-5o), income, and previous education. Therefore, despite the desire to get away from the university groups which pre- dominated in our previous samples, we could not resist taking the opportunity to test this marginal university group. The questionnaire was administered during a class meeting, all members being present. The men were too few to constitute a separate statistical group, and our policy of separating the sexes-perhaps too strictly adhered to-prevented us from combining them.
2. San Quentin State Prison (California) Inmates (N = IIo). Since these men constitute a particularly important group, psychologically and sociologically, they were studied more intensively than the others; the sampling procedures and results are discussed in detail in Chapter XXI. It may suffice here to say that the sample was well randomized.
3? and 4? Psychiatric Clinic Patients (71 women, 50 men). This group, like the San Quentin group, was considered to have special importance both practically and for a full theoretical understanding of our problem. As a "key group," it seemed to merit thorough study and analysis (Chapter XXII). The questionnaires were administered individually (each subject filling out the questionnaire by him- self) as pan of the clinic routine, and there appeared to be no systematic bias operating in the selection of cases. ,_
5? Working-Class Men and Women. ~ number of small groups were combined to form the "working-class" sample on which statistics were computed. Of the 53 women in this sample, 19 were from the Gelifomia Labor School (an extremely liberal school for working people which has classes in a variety of fields, from trade unionism to ans and crafts), 8 were members of the United Electrical Workers, C. I. O. , 10 were new members of the International Longshoremen and Warehousemen's Union (I. L. W. U. , C. I. O. ), and 16 were office workers obtained
11 The collection of questionnaires from these groups would have been impossible without the generous cooperation of numerous people. We wish to express our gratitude to Dr. Merle H. Elliott, who obtained questionnaires from his class in the Extension Division of the University of Califomia, Dr. David G. Schmidt, who made the necessary arrangements for the San Quentin Group, Dr. Karl Bowman and Dr. Roben Harris, who made it possible for us to obtain subjects at the Langley Poner Clinic, Dr. Barbara Kirch- heimer, who made the arrangements, and Mrs. Emily Moulton, who collected question- naires at the U. S. Employment Service, Captain Malcolm E. Crossman, who gave his suppon, and Dr. Boyd R. McCandless, who gave freely of his time in obtaining question- naires at the Alameda School for Merchant Marine Officers.
? THE STUDY OF ETHNOCENTRIC IDEOLOGY I 3I
through the employer;;:~frhe 61 men were obtained similarly: 15 were from the California Labor Schoof, 12 from the United Electrical \Yorkers, 26 from the I. L. W. U. , and 8 from the United Seamen's Service. All groups were obtained in the San Francisco Bay area. The Labor School subjects constitute the total mem- bership of various classes, the questionnaire being administered in classP The 20 Electrical Workers were obtained in the union hall as they came in on business matters. The 36 I. L. W. U. members were given the questionnaire at the beginning of a class for new members before any indoctrination had started. While the female office workers took the questionnaire at the request of interested employers, it was understood that they maintained their anonymity, and no systematic selec- tive factors appear to have entered in. Less reliance can be placed on the male sample from the Seamen's Service, since the 8 subjects are but a small percentage of those passing in and out of the center. The working-class sample as a whole does not appear to reflect, in either a random or a representative manner, the actual working-class population, and any generalizations from the data must be drawn tentatively and with great caution.
The bulk of the working-class sample was given Form 40, only 19 women and 31 men receiving Form 45? Therefore, for the statistical purpose of relating the E scale to the other scales and measures (see Chapters V through VII), all ques- tionnaires were treated as if they were Form 40, that is, only E. a was statisticized in Form 45? In consequence there are results- on Form 40 for Working-Class Women (N=53) and Working-Class Men (N=61).
However, when additional data were desired on the total Form 45 E scale, it was decided to combine the 19 women and 31 men into a single sample,13 the Working-Class Men and Women (Form 45) (N =5o). This sample is, then, actually a part of the larger Form 40 sample (see below). The men in the Form 45 sample were obtained from the groups mentioned above in almost exactly the same proportions as those taking Form 40. However, the Form 45 women are preponderantly from the Labor School and the United Electrical Workers' Union, and may consequently differ systematically from the others with respect to E.
Form 40 was given to a number of groups forming the following statistical samples:
6. George Washington University Women (N = 132). This group comprises the total female membership of several day and evening classes in psychology at George Washington University, Washington, D. C. 14 (There were so few men that their questionnaires were not statisticized. ) lt was included out of an interest in regional differences between California and the East, even though only limited generalizations can be drawn from so selected a sample.
7? California Service Club Men (N = 63). Two service clubs, Kiwanis and Rotary, comprise this sample. Questionnaires were filled out during a customary luncheon meeting (procedure not previously announced) just prior to the fea- tured talk, given by a member of our staff.
12 It appeared necessary to distinguish "middle-class" from "working-class" members of the Labor School, and to place the former in the broader "middle-class" sample. (See the discussion of the middle-class sample, Form 40. ) The present figures refer only to working- class members. ?
13 This sample was used only for getting the reliability data on the Form 45 E scale; no statistics were computed on the other scales.
14 As mentioned previously, while the questionnaire was administered to all present, only the native-born, white, non-Jewish, American subjects were included in the statistical treatment. The N's reported refer to the number of cases treated statistically.
? IJ2
8. Middle-Class Men (N = 69) and 9? Middle-Class Women (N = 154). These two samples represent the combination, for statistical purposes, of the following groups: The membe~ship at a meeting of the Parent-Teachers' Association in a "solid" middle-class section of Berkeley, California (46 women, 29 men). Again, the questionnaire was administered just before the featured talk on child training. The membership of a Protestant church in a small town just outside of San Fran- cisco (29 women, 31 men). The 15 women in a local Unitarian Church group. The members of the California Labor School who appeared to be "middle class" in terms of occupation (lawyer, engineer, independent businessman, etc. ) and in- come (II women, 9 men); in case of doubt the individual remained in the "working-class" sample discussed above. The 17 women in one division or panel of the local League of Women Voters. Finally, the 36 members of an exclusive, upper middle-class women's club. It would appear, then, that these two samples, particularly the women, represent diverse elements of the middle class.
w. Working-Class Men (N=61) and II. Working-Class Women (N=53). These groups have been described above in connection with the Form 45 sample of Working-Class Men and Women.
12. Los Angeles Men (N = 117) and 13? Los Angeles Women (N = 130). 15 In an attempt to obtain greater regional diversity for the total sample, a group of men and women was tested in the Los Angeles area. Because of time limitations the sampling procedure was not thoroughly controlled, and exact figures are not available on the number of subjects in each of the groups comprising the sample. Subjects were obtained from the following groups (not more than 25 per cent of the total N from any one group): parents of college students (volunteers), high school teachers, veterans at a counseling center, Radio Writers Guild (tested during a meeting), League of Women Voters, Boy Scout leaders, members of an anti-Semitic organization (12 responders out of some 100 questionnaires mailed out), and several small local clubs and neighborhood groups. The sample is pri- marily middle class in composition, although it cannot be considered clearly representative of the middle-class population. Moreover, its mean may be syste- matically lowered by the relatively high educational level and by the fact that many of the subjects were obtained on a volunteer basis. It was suitable for the present research purposes, however, since it appeared highly diverse with respect to ethnocentrism and with respect to the social and psychological characteristics whose relations to ethnocentrism were being investigated.
In addition to the above groups, the following two groups received both Forms 45 and 40:
14. Employment Service Men Veterans (N = 106). It seemed likely, early in 1946, that the questionnaire, particularly the F scale (see Chapter VII) and the projective questions (Chapter XV) could reveal much that was of interest to the clinician and the vocational counselor. Thus, when the questionnaire was given to veterans seeking vocational guidance at the local U. S. Employment Service, it was with the thought that it would be an aid to the agency as well as to the research. With a few exceptions, all (white, Christian) male veterans coming in for counseling during a several-month period starting early in 1946 were given the questionnaire, the first 51 receiving Form 45, the next 55, Form 40. The excep- tions were men who seemed not to have enough education to handle the question- naire and men in whose case a convenient time could not be arranged. This group can thus be considered a relatively random sample of the counselees. However,
15 Thes? questionnaires were collected by Dr. J. F. Brown with the assistance of Emily Gruen and Carol Creedon.
THE AUTHORITARIAX PERSONALITY
? THE STUDY OF ETHNOCENTRIC IDEOLOGY I33
it may well be that counselees as a group are not representative of the veteran population. Thus, our sample is above average in socioeconomic level (see Chap- ter V) and in education and intelligence (see Chapter VIII). Furthermore, on the basis of evidence to be presented in later chapters, particularly Chapter XI, it appears likely that willingness to seek guidance, and especially to accept the mild psychotherapy going with it, is more common in nonethnocentrists than in others. How serious a sampling bias this produces depends in part on other factors which might impel ethnocentric individuals to seek help (e. g. , external pressures, or a tendency to conceive of the Service as benevolent authority). A t any rate, it is not unlikely that the mean E score for this sample may be somewhat lower than for the veteran population generally.
15. Maritime School Men (N = 343). This group comprises the entire mem- bership of a government training school for Merchant Marine officers. The school is located in Alameda (San Francisco Bay area), but its students come from all parts of the country. Upon admission all of them must have had at least fourteen months of active service as unlicensed seamen. The questionnaires were adminis- tered during the study periods, under well-controlled conditions, by members of the Psychology staff who seemed to be on excellent terms with the men. Half of the study sections received Form 45, the other half Form 40, the halves being roughly equated in terms of ability and time in school. This group, like the one described immediately above, cannot be considered a fully representative sample of the armed services population. It is selected in at least the following ways: predominantly lower middle-class background, relatively few members coming from the lower socioeconomic strata or from the upper middle class or above; above average in upward social mobility-in the desire to "raise oneself socially and financially"; above average in intelligence, this being a primary qualification for admission (mean AGCT score of 126. 2, range of 102-153). 16 Despite these relative uniformities, the group is extremely diverse in most other ways.
b. RELIABILITY AND GROUP DIFFERENCES. The reliability data for Forms 45 and 40 are presented in Table 17(IV). As noted above, the 5-item EA scale in Form 40 contained no items referring to Jews; Form 45 contained these five items plus five EB items, four of which are from the former A-S scale. Since the small number of items in Form 40 made it unfeasible to compute are- liability coefficient, it was decided to determine the reliability of the total scale by correlating EA with EB rather than by correlating odd-even or equivalent halves. This procedure gave some indication of the degree of equivalence between scores on Form 40 and scores on Form 45; it provided, for example, a partial answer to the question: of a gJ;oup scoring in the low quartile on the EA scale, what percentage would score in the low quartile on EA + B? The average reliability of ? 79 for the seven groups taking Form 45 (Table I 7(IV) A, C) indicates that the overlap is relatively great-although it also brings out the advantage of using the longer scale.
The present method of computing reliability, while it was helpful in de- termining the degree of relationship between EA and EA + B, and in showing
16 No detailed description of the social and psychological properties of the various groups will be presented in this chapter. Instead, each set of properties will be presented and discussed in the appropriate chapter, e. g. , politico-economic properties in Chapter V, religion in Chapter VI, and so on.
? 1 34 THE AUTHORIT ARIAN PERSONALITY
the great, though incomplete, unity in ethnocentric ideology, had neverthe- less the disadvantage of yielding lower reliabilities than would have been ob- tained by a division into odd-even or equivalent halves. Two halves equated for content are certainly likely to intercorrelate more highly than two halves, such as EA and EB, which differ in content. This hypothesis was tested on two groups. In the case of the San Quentin Men, who obtained an EA - EB reliability of . 65, the lowest of any group tested, the reliability rose to ? 79 when odd-even halves were used. In a group of 5r7 women, students at the University of California,l1 the reliability based on EA vs. EB was . 79, while the odd-even reliability was . 87. Since in its usual meaning "reliability" refers to the relation between '"equivalent measures of the same thing," the reliability of the total E scale is probably around .
2. 84
Mean
a. 40 3. 74 3. 88 4. 38 3. 93 4. 32 3. 44 2. 13 3. 19 2. 82 2. 10 2. 26
3. 30
D. P. Rank D. P.
2. 89 (9) 3. 26 (7) 4. 00 (2)
(J. businessmen) (Negro rights) (Marry a Jew)
(Negro foreman) (World organization) (Jews alike) (Negroes lazy) (Jewish neighbors) (Foreign ideas)
'"'! :II t"J
>
4. 11 3. 79 4. 94 4. 00 4. 49 3. 70 1. 81 3. 30 2. 74 2. 34
Mean per item 3. 43 3. 49
(No Japs)
(Okies) 2. 62
1. 80 (12) > t"' . . . . . . . . ,
TABLE 15 (IV)
MEANS AND DISCRIMINA'OORY POWERS OF THE E-SCALE ITDIS (FORM 60) a
. . . .
N
0\
aThe four groups on which these data are based are:
Group I: Group II: Group III: Group IV:
University of Oregon Student Women (N =47).
University of Oregon and . University. of California Student Women (N =54). University of Oregon and University of California Student Men (N =57). Oregon Service Club Men (N =68).
Group I Mean D. P.
Group II Mean D. P.
Group III Mean D. P.
Group IV
Over-all
4. 55 4. 15 4. 69 4. 23 4. 38 1. 46 4. 00 3. 54 2. 69 2. 39
3. 70 4. 50 3. 87 4. 54 3. 15 2. 04 3. 07 2. 56 2. 13 2. 41
3. 25
3. 46 (5) ~ :II
4. 22 (1) 0 i;d
. . . . . 3. 69 (4) '"'! >
i;d
3. 41
2. 77 3. 40 3. 46 3. 61 3. 53 3. 91 2. 93 1. 86 a. oo 2. 75 1. 93 2. 19
2. 95
i;d 3. 82 (3) . . . . . > z
2. 02 ( 11)
3. 38 (6) t%j
3. 14 (8) 0
2. 09 (10) 3. 15
~
"0
"'
z
? THE STUDY OF ETHNOCENTRIC IDEOLOGY 127
parable groups of men and women has been found, as may be noted below in the results on additional groups (p. I33 ff. ).
c. ITEM ANALYSIS. The results of the item analysis of theE scale (Form 6o) are presented in Table I5(IV). The average D. P. of 3? I5 is very satis- factory. The three lowest D. P. 's (1. 8-2. I) were obtained by the items having the lowest means (2. I-2. 3). Two of these items, numbers 42 and 55, ob- tained similar means on previous forms of the scale. They were included here, slightly revised, with the expectation that the present groups might agree more strongly. This expectation was not borne out. In view of the relatively strong rejection of Oklahomans in California, the low mean and D. P. of Item 59 are probably due more to faults in formulation than to the inadequacy of the idea which we intended to express. Even the three poorest items, however, differentiate significantly and with a minimum of overlap between the high and low quartiles, the low scorers being strongly opposed (almost uniform responses of - 3 ) , the high scorers tending to disagree only slightly. 9
The rank order of goodness of items is, on the whole, consistent with previous results. The five best items ( I4, 2I, 28, 33, 38) include two referring to Negroes, two to Jews, and one to world organization; these items ranked similarly in earlier forms. Item 3 (Zootsuiters) has a rank of 9, as compared with a rank of I on Form 78. The drop may well be due to the fact that the zootsuiter issue was less focal, and therefore less likely to produce extreme agreement or disagreement, in Oregon than in California. The groups taking Form 6o agree quite well among themselves regarding the relative level of acceptability (mean) and level of discriminability (D. P. ) ofthe items. While rank-order correlations between the groups were not computed, it appears from inspection of Table I5(IV) that the mean or D. P. rank of each item is fairly stable from group to group.
2. THE FOURTH FORM OF THE E SCALE (FORMS 45 AND 40)
The fourth and final form of the E Scale (see Table I6(1V)), as used in Form 45, involved the deletion of two items ("Japs" and "Okies") from the previous form, the other ten items remaining intact. It was understood that this contraction of the E scale would eliminate many of its qualitative functions and would probably lower its reliability to the minimum required by the present research purposes. Nevertheless, the intention to distribute the questionnaire to a wide variety of groups, many of whom would have very little time for filling it out, made every reduction in size seem desirable. The entire Questionnaire Form 45 could, accordingly, be filled out in ap- proximately thirty minutes-short enough so that, for example, a club or
9 The group mean for each item is usually about midway between the means for the high and low quartiles. It is therefore legitimate to infer the quartile means from the item mean and D. P.
? 128
EA 5?
EA ro.
EA rs.
EA 20.
EB 24.
EB 28. EB 32?
EB 36. EB 40. EA 45?
THE AUTHORIT ARIAN PERSONALITY TABLE r6 (IV)
THE FouRTH FoRM OF THE E ScALE (FoRMS 45 AND 40)"'
Zootsuiters prove that when people of their type have too much money and freedom, they just take advantage and cause trouble.
Negroes have their rights, but it is best to keep them in their own dis- tricts and schools and to prevent too much contact with whites.
The worst danger to real Americanism during the last 50 years has come from foreign ideas and agitators.
It would be a mistake ever to have Negroes for foremen and leaders over whites.
One trouble with Jewish businessmen is that they stick together and prevent other people from having a fair chance in competition.
I can hardly imagine myself marrying a Jew.
If Negroes live poorly, it's mainly because they are naturally lazy, ignorant, and without self-control.
There may be a few exceptions, but in general, Jews are pretty much alike.
The trouble with letting Jews into a nice neighborhood is that they gradually give it a typical Jewish atmosphere.
If and when a new world organization is set up, America must be sure that she loses none of her independence and complete power in matters that affect this country.
a The five E A items constitute the entire E scale in Farm 4 0 . The total ten-item scale was given, numbered as it appears above, in Form 45? The reliability of Form 45 was obtained by correlating the 5 E A items with the remaining 5 EB items. It will be noted that there are no items about Jews in EM it being desired as a practical aim to construct a ques- tionnaire which would give an index of anti-Semitism without mentioning Jews at all.
organization could take it during a meeting, just before hearing a talk not directly related to the questionnaire.
Demands of practicality and expediency forced an additional compromise. Questionnaire Form 40 was even shorter than Form 45; in addition to con- tractions of other techniques, the E scale in this form was reduced to five items (as shown in Table r6(1V) ). A primary reason for Form 40 was that certain groups might be unable to spare even the thirty minutes required by Form 45? An additional consideration in the contraction of the E scale, how- ever, was the possibility that, in certain groups at least, the items referring to Jews might be too "controversial" or might focus attention too directly on the issue of prejudice. Accordingly, the five EA items in Form 40 contain no direct reference to Jews. They deal, rather, with Negroes, zootsuiters, foreigners, and "world organization. " (In Form 45 the E scale contains, in addition to these, five EB items, four referring to Jews, one to Negroes.
) It was recognized that these five items do not constitute a scale in the more technical sense, but this loss seemed justified by the gain in applicability to vanous groups.
Our conclusions regarding the advantage of using Form 40 ought perhaps to be noted for those faced with similar problems. Although it avoided focusing atten-
? THE STUDY OF ETHNOCENTRIC IDEOLOGY 129
cion on Jews, the loss in terrns of research aims was not sufficiently compensated for by the small gain in time nor by elimination of resistance. Indeed, the resistance encountered seemed to be based as much on the other phases of the questionnaire as on the E scale. Probably the basic opposition psychologically was to being "investigated" at all in an intensive way. Unlike the usual several-question poll, this questionnaire seemed, to many a subject, to identify him as a total individual even though he knew that his anonymity was preserved. In some cases this was highly anxiety-producing despite our careful attempts at reassurance and at ex- plaining the entire procedure in terms of an impersonal, public opinion, nonindi- vidual approach. In some cases it was impossible to gain the cooperation of the leadership of a group; in other cases cooperative leaders were unable to put the idea across or to have it carried out. Difficulties of this sort were as great with Form 40 as with Form 45? Once a group was induced to cooperate, there were very few omissions of questions or scale items in either form. In short, resistance was related more to the general nature of the questionnaire than to any specific in- dividual items. Form 45 might therefore have been used on practically all of the groups tested. When it is absolutely necessary to delete certain items-e. g. , if one were testing groups with a large Jewish or Negro membership and items referring to these groups might cause friction-probably the best procedure would be to have alternative items to replace those deleted.
While the number of groups which were actively but unsuccessfully ap- proached is not large, there is some indication that resistance of the type men- tioned above is correlated positively with ethnocentrism. For example, among the "Middle-Class Women" (Table 15(V)) there was an exclusive club which "just barely" decided to cooperate and which refused even to consider our request for volunteers to be interviewed. This group obtained one of the highest E means of all groups tested. Such resistance was seldom encountered in less ethnocentric groups. This difficulty might have been ex- pected on the basis of the ethnocentrists' tendency toward self-deception and concern with prying, which was expressed indirectly in the responses on the A-S and E scales, and which is brought out more directly in the chapters that follow.
Considerations of this type are of great importance in any attempt to generalize from a research sample to a broader population. Thus, because of the greater resistance of ethnocentrists to psychological investigation, it is likely that the average degree of ethnocentrism (over-all mean E score) in our total sample is somewhat lower than that which would be found in a truly random or truly representative sample. Even in the more customary public opinion polls, where population areas are often mapped out in ad- vance (stratification or other attempt at representative sampling), an ade- quate sample may not be achieved because, in their door-to-door polling, interviewers cannot reach those subjects who are unreceptive to the idea of being "tested. "lO
10 The common assumption that "any 50 people" within a given area or income level will do, and that errors of sampling on an individual level will cancel each other out, overlooks the likelihood that receptivity may correlate with what is being polled.
? 1 3 0 THE AUTHORITARIAN PERSONALITY
It seems necessary, therefore, in describing the groups on whom data were obtained, to mention briefly the nature and adequacy of the sampling pro- cedure.
a. THE SAMPLE AND THE SAMPLING PROCEDURE. The distribution of Forms 45 and 40 took place during the latter part of 1945 and the first half of 1946, a period of about nine months in all. Form 45 was given to the following groups: 11
1. Testing Class Women (N = 59). This was an adult evening class given by the Extension Division of the University of California. Since it was a class in Psychological Tests, it probably attracted a more diversified group than does the usual adult class in psychology. It was expected to contain not only individuals seriously interested in understanding themselves better-individuals who, as we shall see later, are not likely to be extremely ethnocentric-but also persons in- terested in psychology more as a means of manipulating others. The class was also varied with respect to age (range about 2o-5o), income, and previous education. Therefore, despite the desire to get away from the university groups which pre- dominated in our previous samples, we could not resist taking the opportunity to test this marginal university group. The questionnaire was administered during a class meeting, all members being present. The men were too few to constitute a separate statistical group, and our policy of separating the sexes-perhaps too strictly adhered to-prevented us from combining them.
2. San Quentin State Prison (California) Inmates (N = IIo). Since these men constitute a particularly important group, psychologically and sociologically, they were studied more intensively than the others; the sampling procedures and results are discussed in detail in Chapter XXI. It may suffice here to say that the sample was well randomized.
3? and 4? Psychiatric Clinic Patients (71 women, 50 men). This group, like the San Quentin group, was considered to have special importance both practically and for a full theoretical understanding of our problem. As a "key group," it seemed to merit thorough study and analysis (Chapter XXII). The questionnaires were administered individually (each subject filling out the questionnaire by him- self) as pan of the clinic routine, and there appeared to be no systematic bias operating in the selection of cases. ,_
5? Working-Class Men and Women. ~ number of small groups were combined to form the "working-class" sample on which statistics were computed. Of the 53 women in this sample, 19 were from the Gelifomia Labor School (an extremely liberal school for working people which has classes in a variety of fields, from trade unionism to ans and crafts), 8 were members of the United Electrical Workers, C. I. O. , 10 were new members of the International Longshoremen and Warehousemen's Union (I. L. W. U. , C. I. O. ), and 16 were office workers obtained
11 The collection of questionnaires from these groups would have been impossible without the generous cooperation of numerous people. We wish to express our gratitude to Dr. Merle H. Elliott, who obtained questionnaires from his class in the Extension Division of the University of Califomia, Dr. David G. Schmidt, who made the necessary arrangements for the San Quentin Group, Dr. Karl Bowman and Dr. Roben Harris, who made it possible for us to obtain subjects at the Langley Poner Clinic, Dr. Barbara Kirch- heimer, who made the arrangements, and Mrs. Emily Moulton, who collected question- naires at the U. S. Employment Service, Captain Malcolm E. Crossman, who gave his suppon, and Dr. Boyd R. McCandless, who gave freely of his time in obtaining question- naires at the Alameda School for Merchant Marine Officers.
? THE STUDY OF ETHNOCENTRIC IDEOLOGY I 3I
through the employer;;:~frhe 61 men were obtained similarly: 15 were from the California Labor Schoof, 12 from the United Electrical \Yorkers, 26 from the I. L. W. U. , and 8 from the United Seamen's Service. All groups were obtained in the San Francisco Bay area. The Labor School subjects constitute the total mem- bership of various classes, the questionnaire being administered in classP The 20 Electrical Workers were obtained in the union hall as they came in on business matters. The 36 I. L. W. U. members were given the questionnaire at the beginning of a class for new members before any indoctrination had started. While the female office workers took the questionnaire at the request of interested employers, it was understood that they maintained their anonymity, and no systematic selec- tive factors appear to have entered in. Less reliance can be placed on the male sample from the Seamen's Service, since the 8 subjects are but a small percentage of those passing in and out of the center. The working-class sample as a whole does not appear to reflect, in either a random or a representative manner, the actual working-class population, and any generalizations from the data must be drawn tentatively and with great caution.
The bulk of the working-class sample was given Form 40, only 19 women and 31 men receiving Form 45? Therefore, for the statistical purpose of relating the E scale to the other scales and measures (see Chapters V through VII), all ques- tionnaires were treated as if they were Form 40, that is, only E. a was statisticized in Form 45? In consequence there are results- on Form 40 for Working-Class Women (N=53) and Working-Class Men (N=61).
However, when additional data were desired on the total Form 45 E scale, it was decided to combine the 19 women and 31 men into a single sample,13 the Working-Class Men and Women (Form 45) (N =5o). This sample is, then, actually a part of the larger Form 40 sample (see below). The men in the Form 45 sample were obtained from the groups mentioned above in almost exactly the same proportions as those taking Form 40. However, the Form 45 women are preponderantly from the Labor School and the United Electrical Workers' Union, and may consequently differ systematically from the others with respect to E.
Form 40 was given to a number of groups forming the following statistical samples:
6. George Washington University Women (N = 132). This group comprises the total female membership of several day and evening classes in psychology at George Washington University, Washington, D. C. 14 (There were so few men that their questionnaires were not statisticized. ) lt was included out of an interest in regional differences between California and the East, even though only limited generalizations can be drawn from so selected a sample.
7? California Service Club Men (N = 63). Two service clubs, Kiwanis and Rotary, comprise this sample. Questionnaires were filled out during a customary luncheon meeting (procedure not previously announced) just prior to the fea- tured talk, given by a member of our staff.
12 It appeared necessary to distinguish "middle-class" from "working-class" members of the Labor School, and to place the former in the broader "middle-class" sample. (See the discussion of the middle-class sample, Form 40. ) The present figures refer only to working- class members. ?
13 This sample was used only for getting the reliability data on the Form 45 E scale; no statistics were computed on the other scales.
14 As mentioned previously, while the questionnaire was administered to all present, only the native-born, white, non-Jewish, American subjects were included in the statistical treatment. The N's reported refer to the number of cases treated statistically.
? IJ2
8. Middle-Class Men (N = 69) and 9? Middle-Class Women (N = 154). These two samples represent the combination, for statistical purposes, of the following groups: The membe~ship at a meeting of the Parent-Teachers' Association in a "solid" middle-class section of Berkeley, California (46 women, 29 men). Again, the questionnaire was administered just before the featured talk on child training. The membership of a Protestant church in a small town just outside of San Fran- cisco (29 women, 31 men). The 15 women in a local Unitarian Church group. The members of the California Labor School who appeared to be "middle class" in terms of occupation (lawyer, engineer, independent businessman, etc. ) and in- come (II women, 9 men); in case of doubt the individual remained in the "working-class" sample discussed above. The 17 women in one division or panel of the local League of Women Voters. Finally, the 36 members of an exclusive, upper middle-class women's club. It would appear, then, that these two samples, particularly the women, represent diverse elements of the middle class.
w. Working-Class Men (N=61) and II. Working-Class Women (N=53). These groups have been described above in connection with the Form 45 sample of Working-Class Men and Women.
12. Los Angeles Men (N = 117) and 13? Los Angeles Women (N = 130). 15 In an attempt to obtain greater regional diversity for the total sample, a group of men and women was tested in the Los Angeles area. Because of time limitations the sampling procedure was not thoroughly controlled, and exact figures are not available on the number of subjects in each of the groups comprising the sample. Subjects were obtained from the following groups (not more than 25 per cent of the total N from any one group): parents of college students (volunteers), high school teachers, veterans at a counseling center, Radio Writers Guild (tested during a meeting), League of Women Voters, Boy Scout leaders, members of an anti-Semitic organization (12 responders out of some 100 questionnaires mailed out), and several small local clubs and neighborhood groups. The sample is pri- marily middle class in composition, although it cannot be considered clearly representative of the middle-class population. Moreover, its mean may be syste- matically lowered by the relatively high educational level and by the fact that many of the subjects were obtained on a volunteer basis. It was suitable for the present research purposes, however, since it appeared highly diverse with respect to ethnocentrism and with respect to the social and psychological characteristics whose relations to ethnocentrism were being investigated.
In addition to the above groups, the following two groups received both Forms 45 and 40:
14. Employment Service Men Veterans (N = 106). It seemed likely, early in 1946, that the questionnaire, particularly the F scale (see Chapter VII) and the projective questions (Chapter XV) could reveal much that was of interest to the clinician and the vocational counselor. Thus, when the questionnaire was given to veterans seeking vocational guidance at the local U. S. Employment Service, it was with the thought that it would be an aid to the agency as well as to the research. With a few exceptions, all (white, Christian) male veterans coming in for counseling during a several-month period starting early in 1946 were given the questionnaire, the first 51 receiving Form 45, the next 55, Form 40. The excep- tions were men who seemed not to have enough education to handle the question- naire and men in whose case a convenient time could not be arranged. This group can thus be considered a relatively random sample of the counselees. However,
15 Thes? questionnaires were collected by Dr. J. F. Brown with the assistance of Emily Gruen and Carol Creedon.
THE AUTHORITARIAX PERSONALITY
? THE STUDY OF ETHNOCENTRIC IDEOLOGY I33
it may well be that counselees as a group are not representative of the veteran population. Thus, our sample is above average in socioeconomic level (see Chap- ter V) and in education and intelligence (see Chapter VIII). Furthermore, on the basis of evidence to be presented in later chapters, particularly Chapter XI, it appears likely that willingness to seek guidance, and especially to accept the mild psychotherapy going with it, is more common in nonethnocentrists than in others. How serious a sampling bias this produces depends in part on other factors which might impel ethnocentric individuals to seek help (e. g. , external pressures, or a tendency to conceive of the Service as benevolent authority). A t any rate, it is not unlikely that the mean E score for this sample may be somewhat lower than for the veteran population generally.
15. Maritime School Men (N = 343). This group comprises the entire mem- bership of a government training school for Merchant Marine officers. The school is located in Alameda (San Francisco Bay area), but its students come from all parts of the country. Upon admission all of them must have had at least fourteen months of active service as unlicensed seamen. The questionnaires were adminis- tered during the study periods, under well-controlled conditions, by members of the Psychology staff who seemed to be on excellent terms with the men. Half of the study sections received Form 45, the other half Form 40, the halves being roughly equated in terms of ability and time in school. This group, like the one described immediately above, cannot be considered a fully representative sample of the armed services population. It is selected in at least the following ways: predominantly lower middle-class background, relatively few members coming from the lower socioeconomic strata or from the upper middle class or above; above average in upward social mobility-in the desire to "raise oneself socially and financially"; above average in intelligence, this being a primary qualification for admission (mean AGCT score of 126. 2, range of 102-153). 16 Despite these relative uniformities, the group is extremely diverse in most other ways.
b. RELIABILITY AND GROUP DIFFERENCES. The reliability data for Forms 45 and 40 are presented in Table 17(IV). As noted above, the 5-item EA scale in Form 40 contained no items referring to Jews; Form 45 contained these five items plus five EB items, four of which are from the former A-S scale. Since the small number of items in Form 40 made it unfeasible to compute are- liability coefficient, it was decided to determine the reliability of the total scale by correlating EA with EB rather than by correlating odd-even or equivalent halves. This procedure gave some indication of the degree of equivalence between scores on Form 40 and scores on Form 45; it provided, for example, a partial answer to the question: of a gJ;oup scoring in the low quartile on the EA scale, what percentage would score in the low quartile on EA + B? The average reliability of ? 79 for the seven groups taking Form 45 (Table I 7(IV) A, C) indicates that the overlap is relatively great-although it also brings out the advantage of using the longer scale.
The present method of computing reliability, while it was helpful in de- termining the degree of relationship between EA and EA + B, and in showing
16 No detailed description of the social and psychological properties of the various groups will be presented in this chapter. Instead, each set of properties will be presented and discussed in the appropriate chapter, e. g. , politico-economic properties in Chapter V, religion in Chapter VI, and so on.
? 1 34 THE AUTHORIT ARIAN PERSONALITY
the great, though incomplete, unity in ethnocentric ideology, had neverthe- less the disadvantage of yielding lower reliabilities than would have been ob- tained by a division into odd-even or equivalent halves. Two halves equated for content are certainly likely to intercorrelate more highly than two halves, such as EA and EB, which differ in content. This hypothesis was tested on two groups. In the case of the San Quentin Men, who obtained an EA - EB reliability of . 65, the lowest of any group tested, the reliability rose to ? 79 when odd-even halves were used. In a group of 5r7 women, students at the University of California,l1 the reliability based on EA vs. EB was . 79, while the odd-even reliability was . 87. Since in its usual meaning "reliability" refers to the relation between '"equivalent measures of the same thing," the reliability of the total E scale is probably around .
