), a militant union, orj\ from the California Labor School, a
strongly
left-wing institution.
Adorno-T-Authoritarian-Personality-Harper-Bros-1950
P .
Rank
2. 41 (6) l""::
t:;l 1. 60 (10. 5) l""::
1. 06 (13) t"' 0
3. 28 1. 50 (12) C)
::tl 4. 04 2. 24 (8) 0
q 3. 08 2. 03 (9) '"d
2. 65 1. 60 (10. 5) 3. 92 2. 08
~
0
. . . . . . . . . ,
(")
0
I
~
(")
0
><
t:;l
l""::
1::0
::tl
"II':
. . . . . .
'"d "'
"
0'1
? 168 THE AUTHORITARIA~ PERSONALITY
These groups are more consen'ati\'e on specific issues than the over-all scale mean of 3. 92 indicates. The over-all means on the items (4, 9, r5, 31, 37, 43) dealing with unions, business, and government functions range from 4? 74 to 5? 35, and these items are also the most discriminating. It would appear, then, that with regard to what is most definitive in liberalism and conserva- tism-mainly ideas regarding power relations among labor, business, and gov- ernment-the liberal position is as yet less crystallized and less militantly held than is the conservative position. There is some question as to how far these results can be generalized beyond the present sample. 5 They are, however, in general accord with numerous other findings and observations regarding the contemporary political scene. What is more difficult, and also more im- portant, to gauge is the psychological potential for future ideological de- velopment in various directions in the face of changing political and economic conditions. Perhaps the other components of political ideology, when sys-
tematically measured and psychologically understood, would provide a basis for the solution of this problem.
4. THE THIRD PEC SCALE (FORMS 45 AND 40)
The construction of Forms 45 and 40 was, as has been discussed in the previous chapter, influenced greatly by considerations of practicality and of administrative expediency. In view of these considerations, and in order to make room for the inclusion of other material, the PEC scale was cut literally to the bone. It was identical in both forms of the questionnaire and contained only five items-not enough to obtain an adequate measure of re- liability, and hardly enough to be called a "scale. " The reasoning behind the use of a five-item E scale was discussed and criticized in Chapter IV; the same criticisms apply to the present PEC scale. It appears now that it would have been wiser to have used a ro-item form; the short form used did, how- ever, make possible the comparison of various groups and the study of rela- tionships between this scale and the others.
The Form 45-40 PEC scale is presented in Table 7(V). It will be seen that the five items were not selected solely on statistical grounds; rather an attempt was made to include items whose D. P. 's were above a minimal level and, more important, which covered as many as possible of the ideological trends previously discussed. The first four items are taken, with occasional slight revisions, from Form 6o. Item 17 (Economic security), has a history of transiency; originally in the Form 78 PEC scale, it was moved to the F scale in Form 6o (see Chapter VII); it has been returned to PEC in an attempt to rid the F scale of all items which might be connected fairly directly with
5 The representativeness of this sample with respect to political pany and other group memberships will be considered later in this chapter (Section E). That the university groups are not unusually conservative is suggested by the fact that E-scale means are rela- tively low in comparison with other middle-class groups (see Chapter IV). Their PEC means can be compared with those for the groups taking Forms 45 and 40, below.
? POLITICO-ECONOMIC IDEOLOGY AND GROUP MEMBERSHIPS 169
TABLE 7 (V)
THE THIRD FoRM oF THE PoLITico-EcoNoMIC CoNsERVATISM (PEC) ScALE (FoRMs 45-40)
3-" Labor unions should become stronger and have more influence generally. 7? America may not be perfect, but the American Way has brought us about
as close as human beings can get to a perfect society.
I I. " Most government controls over business should be continued even though
the war is over.
I4- Men like Henry Ford or J. P. Morgan, who overcame all competition on
the road to success, are models for all young people to admire and imitate. I7. In general, full economic security is bad; most men wouldn't work if they
didn't need the money for eating and living.
? These items are "liberal," the others are "conservative. " A high score is given for agree- ment with the conservative items, disagreement with the liberal items.
existing ideologies regarding politico-economic or minority groups. Item I 7, as well as several others, might be included in any of several scales; proper placement must ultimately be based on statistical analysis.
Forms 45 and 40 were administered late in I945 and in the first part of 1946. The sampling methods and problems, as well as the composition of the groups and their combination for statistical purposes, have been discussed in Chapter IV. It will suffice here to list the groups comprising this sample.
The Form 45 sample contains four groups: (a) Extension Testing Class (adult) Women (N =59) at the University of California; (b) San Quentin Men (N = no), inmates at the California State Prison; (c) and (d) Psychi- atric Clinic Women (N = 7I) and Men (N = so), mostly outpatients at a community clinic in San Francisco.
The following groups are included in the Form 40 sample: (e) George Washington University Women (N = qz), members of day and evening classes in psychology; (f) California Service Club Men (N = 63), obtained at luncheon meetings of San Francisco Bay Area Kiwanis and Rotary clubs;
(g) and (h) Middle-Class Men (N = 69) and W omen (N = I54), mem- bers of various local groups such as church, P. T. A. , women's clubs, etc. ; (i) and (j) Working-Class Men (N = 6I) and Women (N =53), mem- bers of local groups such as United Electrical Workers Union, Warehouse- men's Union (1. L. W. U. ), California Labor School, etc. ; (k) and (I) Los Angeles Men (N = I I 7) and Women (N = qo), a heterogeneous but largely middle-class sample of various local groups in Los Angeles. Data on some of the subgroupings within these statistical units will be considered
in Section E, below.
In addition, there were two groups which were given both Forms 45 and
40. First, the School for Merchant Marine Officers (to be referred to as "Maritime School") (N = 343), half of which was given Form 45, the other half Form 40, the two halves being equated for intelligence (AGCT), time in school, and planned function as officer (deck or engine). Second, veterans
? I]O
THE AUTHORITARIAN PERSONALITY
TABlE 8 (V)
MEANS AND STANDARD DEVIATIONS OF PEC SCAlE SCORES FOR GROUPS TAKING FORMS 45 AND 40
Group
a. Groups taking Form 45 Extension Testing Class
N
Mean Rank S. D.
4. 33 (4) 1. 28 4. 68a (2} 0. 96 4. 12 (11) 1. 53 4. 14a (10) 1. 40
Mean EA
3. 77 5. 33 4. 23 3. 92
4. 31
4. 04
4. 31 3. 89 3. 64 3. 92 3. 91 3. 82 3. 71
3. 91
4. 95
4. 43 4. 69
4. 13
Rank
(12)
(1)
(5) (7. 5)
(6)
(4) (10) (14) (7. 5) (9) (11) (13)
(2)
(3)
Women 59
San Quentin Men Psychiatric Clinic Women Psychiatric Clinic Men
110
71
50 Over-allb 290
4. 32
4. 30
4. 83 4. 30 4. 26 3. 39 3. 25 3. 91
Los Angeles Women
Over-allb 779 4. 05
1. 29
(6. 5) 1. 13
(1) 1. 31 (6. 5) 1. 52 (8) 1. 62 (13) 1. 58 ( 14) 1. 53 (12) 1. 49
b. Groups taking Form 40 George Washington
University Women 132 California Service Club
Men 63
Middle-Class Men Middle-Class Women Working-Class Men W orking-Glass Women Los Angeles Men
69 154 61 53 117
c. Groups taking both forms Maritime School Men
Form 45
Form 40
Employment Service Men V eterans
179 4. 31) 164 4. 32)
( 5 )
Form 45 Form 40
4. 35) (
51
55 Over-allb 449
(3) (1. 28 4. 34 1. 18
d. Totals for all groups 1518 4. 19 1. 37
130 4. 16 (9)
1. 41 1. 45
~1. 0 8
aThe use of two forms for the Psychiatric Clinic groups complicated the PEC scale results somewhat. The data above are based on 45 women and 29 men taking the Form 45 PEC scale. For the remaining 26 women taking the Form 60 PEC scale (14 items) the mean was 4. 05, and for 21 men the mean was 4. 04. For the combined 47 men and women taking this scale, the reliability was . 77 and the Standard Deviation was 1. 05-- values com- parable to those of the other groups taking Form 60.
brn obtaining the over-all values, the individual group means were not weighted by N.
4. 37)
Statistical Property
? POLITICO-ECONOMIC IDEOLOGY AND GROUP MEMBERSHIPS I 7I
coming for vocational guida. nce to the U. S. Employment Service over a period of several months, the first 51 receiving Form 45, the next 55, Form 40. This procedure had mainly to do with determining possible effects of the presence of certain E items on the F-scale responses and will be discussed in more de- tail in Chapter VII.
In view of the shortness of the PEC scale, no reliabilities were computed. However, means and S. D. 's were obtained for each group and are given in Table 8 (V). The group means appear to fall into three main levels of mag- nitude. The two most conservative groups are the California Service Club Men (4. 83) and the San Quentin Men (4. 68). At an intermediate level, with PEC means of 3. 91 to 4? 37 (significantly lower than the first level and higher than the third) is the bulk of the total sample: University groups, Psy- chiatric Clinic Patients, the Middle-Class and Los Angeles groups. Finally, the most liberal groups-although the means of 3. 25 and 3? 39 are far from extreme, and the variability within each group is large-are the Working- Class Men and Women.
That the Service Club Men make the highest mean is not so much a new discovery as a partial indication that the scale provides a valid measure of conservatism. The program and tradition of these groups are fairly explicit in their support of numerous trends in conservative ideology. It will be re- called that similar results were found with the Oregon Service Club Men
(Form 6o). Once again, however, we must emphasize the variability within this and the other groups.
The great conservatism of the San Quentin Men may come as a surprise to those who conceive of criminals as conscious foes of the social order and to those who assume a psychological affinity between criminality and radical- ism. It might have been expected that those who violate prevailing laws re- garding property and morality would tend to oppose the prevailing social ideology and social authority. Yet this does not appear to be the case. Crimi- nals accept the basic premises of the capitalistic system while at the same time engaging in a pseudorebellion against the formal rules and technicalities. The criminal does not oppose the principles of rugged individualism; he simply carries them ad absurdum. The San Quentin material and the relation of criminality to antidemocracy are considered further in Chapter XXI.
The problem of class differences in conservatism is raised by the fact that the Working-Class Men and Women make a significantly lowerPEC mean than do the Middle-Class Men and Women and the Los Angeles group
(which is largely middle class). There are several reasons for questioning whether these differences can be generalized to the broad middle- and work:- ~ ing-class populations. For one thing, the Working-Class group shows a / distinct sampling bias in a liberal direction: almost half the members of this 1 group are from the United Electrical Workers (C. I. O.
), a militant union, orj\ from the California Labor School, a strongly left-wing institution. The
? THE AUTHORIT ARIAN PERSONALITY
Middle-Class groups are more varied and probably more representative in membership. Furthermore, several groups with PEC means similar to those of the Middle-Class groups contain a large proportion of working-class indi- viduals; these groups are the San Quentin JVIen (almost entirely working class), the Psychiatric Clinic Patients, and the Maritime School and Employ- ment Service Veteran Men. In view of the intergroup as well as the intra- group variability, it seems safe to conclude that over-all class differences in political ideology are not extremely large, and that individual and group differences within each class are'so great that they become the primary problem requiring explanation. How does it happen, for example, that the same working-class background produces a law-abiding conservative worker, a politically conservative criminal, a company union leader, a C. I. O. leader, a Communist? W h y does one middle-class individual join a service club while another becomes a supporter of Henry Wallace? Why is it that some, perhaps most, workers identify with the middle class or with the economic status quo, and some individuals with middle-class background identify with what they conceive to be the true interests of the working class? These may be not so much questions of actual class or group membership as questions of class or group identification-and "identification" is a psychological variable. An individual, in making his social identification, is determining not only his ideology, but also what he is to be like as a person. We shall have occasion to consider further, in the chapters that follow, the deeper emotional trends that help to determine the individual's group memberships and identifications.
How close is the relation between conservatism and ethnocentrism in the various groups studied? A means for obtaining a preliminary answer to this question is to compare group means on PEC and onE (see Table 8(V)). Since most groups took the short EA scale, the EA means were used even for those groups which took the total EAB scale (see Chapter IV). The rank- order correlation between the PEC means and EA means for the fourteen groups was +-so, indicating a statistically significant but not very close re- lationship. In general, as the degree of group conservatism increases, the degree of ethnocentrism also increases. The four groups with conspicuously high EA means are the San Quentin Men (5. 33), the Maritime School Men
(4? 95), the Employment Service Men Veterans (4? 43), and the California Service Club Men (4. 31). These groups ranked 2, 5, 3, and 1, respectively, on PEC. No groups were conspicuously low onE, the eight lowest groups having no means within the fairly narrow range of 3. 64-3. 92; the most liberal groups were among the least ethnocentric. The over-all EA mean was 4? I 3, almost identical to the over-all PEC mean of 4-19? (The EAB mean is some- what lower, partly because of sampling differences and partly because the EB items-four on Jews and one on Negroes-had lower means. )
The correlation of ranks does not, however, tell the whole story. Many groups made a significantly higher mean on PEC than on E, or vice versa.
? POLITICO-ECONOMIC IDEOLOGY AND GROUP MEMBERSHIPS 17 3
Whether the group mean on PEC is higher than on E, or lower, seems to depend in large measure on the socioeconomic class of the group: the pre- dominantly middle-class groups tend to be lower on E than on PEC, while the working-class groups are, it appears, more ethnocentric than conservative. Consider the middle-class groups: the PEC and E means, respectively, for the Extension Testing Class Women are 4? 33 and 3. 77; for the George Wash- ington University Women, 4. 30 and 4. 04, for the California Service Club Men, 4. 83 and 4. 31; and similarly for the Middle-Class and Los Angeles Men and Women. Only one of these PEC-E differences is below the 5 per cent level of statistical significance. It will be recalled that in the Form 78 and Form 6o samples, largely middle class, the level of conservatism was greater than the level of ethnocentrism. The opposite trend is found in the working-class (or marginal middle-class) groups. Thus the PEC and E means, respectively, are: for the San Quentin Men, 4. 68 and 5. 33; Working- Class Men, 3? 39 and 3. 92; Working-Class Women, 3. 25 and 3. 91; Maritime School Men, 4? 32 and 4? 95; Employment Service Men Veterans, 4. 36 and 4-43? This leaves only the Psychiatric Clinic Men and Women, who are heterogeneous with respect to class and whose PEC and E means differ only slightly (o. 1-o. 2).
Several factors-not mutually exclusive-may help to explain these class differences. First, open prejudice is more accepted on a verbal level in the working class than in the middle class. The higher E means of the former may therefore reflect, in part, the verbal atmosphere rather than a difference in basic outgroup hostility. (This factor would not hold for the pro-ingroup items. ) Then there is the previously discussed "pseudodemocratic fa~ade," which is more characteristic of the middle than of the working class, and which the E-scale items probably only partially circumvented. Also, certain trends in liberal ideology may appeal to some workers not on a truly liberal basis but on a "class-ethnocentric" basis which is an aspect of general ethno- centrism. For example, some workers are strongly prounion and resentful of "bosses," yet at the same time are anti-Negro, anti-foreigner, and con-
servative regarding many political issues.
All in all, the group data lead us to investigate further the relationship
between ethnocentrism and conservatism, with an eye both to what makes the correlation relatively high and to what keeps it from being higher. These problems will be pursued further when we consider the correlations between the PEC and E scales, and the psychological connection between conservatism and ethnocentrism in the individual. But first we must complete the presentation of the PEC-scale data.
Table 9(V) gives a summary of the item analysis of the Form 45-40 PEC scale. Data for the men and the women are summarized separately. The over-all mean for the women, 4. 07, is significantly lower than that of 4. 25 for the men. The women were also, as noted in Chapter IV, slightly but sig-
? 1 74
No.
3.
7. 11. 14. 17.
THE AUTHORIT ARIAN PERSONALITY TABLE 9 (V)
MEANS AND DISCRIMINATORY POWERS OF THE PEC SCALE ITEMS (FORMS 40 AND 45)
Item
Mean D. P.
Rank
(4) (3) (5) (1) (2)
Mean D. P . Rank
4. 67 3. 49 (4) 4. 57 3. 98 (1) 3. 92 2. 97 (5) 3. 56 ? 3. 90 (2) 3. 62 3. 77 (3)
4. 07 3. 62
(Labor unions) 4. 51 3. 16 (American Way) 4. 90 3. 33 (Government controls) 4. 19 3. 08 (Ford and Morgan) 3. 75 3. 58 (Economic security) 3. 93 3. 46
Mean per iteme 4. 25 3. 32
MEN'S GROUPSa (N 869)b WOMEN'S GROUPSc (N 573)d
aThe individual groups of men in this sample are as follows: San Quentin Men Prisoners (N = 110). Employment Service Men Veterans (N = 106). Maritime School Men (N = 343). 'California Service Club Men (N = 63). Middle-Class Men (N = 69). Working-Class Men (N = 61). Los Angeles Men
(N = 117).
bThe over-all N for the PEC scale (men) is 100 less than that for the E scale because two groups were omitted: (1) Of the 50 Psychiatric Clinic Men, only 29 took the regular Form 45; the others took a form equated for E and F, but not for PEC. Because of the small N, no PEC scale analysis was made on this group. (2) The 50 Working-Class Men and Women were use~ as a statistical group for analysis on the E scale because additional data on the total E scale were desired; but their F and PEC scales were not analyzed statistically.
cThe individual groupg of women in this sample are as follows: Extension Testing dlass Women (N =59). George Washington University Women Students (N =132). Psychiatric Clinic Women (N =45). Middle-Class Women (N = 154). Working-Class Women (N = 53). Los Angeles Women
(N = 130).
dThis N is 26 less than the over-all N for women on the E and F scales because only 45 of the 71 Psychiatric Clinic Women took the regular Form 45. The remaining 26 took a form which was equated to Form 45 for E and F, but not for PEC.
eln obtaining the over? all means, the individual group means were not weighted by N.
nificantly less ethnocentric than the men. This may, however, be a differ- ence in the sample rather than in the total population, since we have no female groups comparable to the high-scoring San Quentin, Service Club, and Maritime School male groups. Moreover, as shown in Table S(V), for all comparable sex groups (Psychiatric Clinic, Middle Class, Working Class, Los Angeles) the means for men and women are almost--identical. Since the sampling methods used were not primarily designed to determine the aver- age intensity of any opinions or attitudes in broader populations, it is perhaps safest not to draw inferences about the total male and female population. It
? POLITICO-ECONOMIC IDEOLOGY AND GROUP MEMBERSHIPS I 75
can be said, however, that? for groups of the general type represented here, no sex differences of practical significance seem to exist; and that differences among male groups and among female groups are much greater than the differences between males and females.
The male and female subsamples are also very similar with regard to means and D. P. 's on the individual items. While the scale mean was lower in women, the relative standing of the item means was almost identical for the two sexes, the rank-order correlation being . 90. Women were more con- servative than men on only one item (unions). A similar relation holds also for the D. P. 's, the rank-order correlation being . 70.
The general level of D. P. 's would, other things being equal, be slightly greater for a 5-item than for a I4-item scale, since each item contributes more to the total score. Therefore, the average D. P. of 3? 4-3? 5 for Forms 45 and 40 is comparable or slightly superior to that of 2. I for Forms 78 and 6o. All of the present items seem statistically adequate. Item I I (Government controls over business) had the lowest D. P. , but in view of the greater suc- cess of the "government function" items in earlier forms, improvement should not be difficult. While the five items can hardly claim to be considered a "scale," they show sufficient internal consistency so that one may mean- ingfully speak of "total PEC score" and one may determine the relations between this and various other measures.
The level of internal consistency of the PEC scale is indicated also by a correlational analysis made on a group of 5I7 University of California women students. 6 A mean r of +o. 26 was found between each item and the sum of the remaining items, the range of r's being +o. Io to +o? 33? The rank order of these items, according to the size of the item's correlation with the sum of the others, was identical to the rank order of item D. P. 's for the combined women's groups above (Table 9(V)). The correlations among individual items averaged +o. I4, the range being +o. o2 to +o. 3o. The highest correla- tion, 0. 30, was between Item 7 (American Way ideal) and Item I4 (Ford and Morgan). Only three r's were below . 10 (the I per cent level of significance), and all of them involved Item II (Government controls). These correlations, while far below those for the E scale, indicate that the PEC scale meets the minimum requirements for its present uses, and that a scale of 20 or 30 such items might, without loss of breadth, achieve a re- liability in the neighborhood of . 8.
5. DISCUSSION: SOME P A TTERNS OF CONTEMPORARY LIBERALISM AND CONSERV A TISM
The reliability and internal consistency of the PEC scales suggest, on the one hand, that liberalism and conservatism are relatively organized and meas- urable patterns of current politico-economic thought; and, on the other
6 The group and procedure are discussed more fully in Chapters IV and VII.
? THE AUTHORITARIAN PERSONALITY
hand, that within each of these broad patterns there is considerable subpat- terning, inconsistency, and simple ignorance. To ignore either the relative generality or the relative inconsistency would, it seems, lead to serious mis- understanding of the problem. More detailed exploration of the nature and deeper psychological meaning of these ideological trends, as expressed in the interviews, will be made in Chapter XVII, following presentation of the clinical material. However, at this point we ought briefly to consider, on the basis of the scale data and of some individual patterns of scale response, certain variations within liberalism and within conservatism.
Liberalism was conceived not as a single, unitary attitude, but as an ideo- logical system containing a number of trends or components. The reliability and internal consistency of the initial forms of the scale show that these trends are interrelated significantly but imperfectly in the individual. The prototypic "liberal" is, according to our guiding conception, an individual who actively seeks progressive social change, who can be militantly critical (though not necessarily totally rejective) of the present status quo, who opposes or de-emphasizes numerous conservative values and beliefs regard- ing business success, rugged individualism, human nature, and the like, and who would diminish the power of business by increasing the power of labor and the economic functions of government.
It is clear, however, that many individuals who are generally liberal do not exhibit some of the above trends. While some of the inconsistency- perhaps the largest part-is due to confusion resulting from lacks and dis- tortions in the press and other media of mass communication, part of it seems also to reflect deep-lying emotional trends of considerable intensity and resistance to change. The individual's pattern of thought, whatever its con- tent, reflects his personality and is not merely an aggregate of opinions picked up helter-skelter from the ideological environment.
One variant, particularly common in the groups tested, might be called the politically pacifistic liberal. The guiding idea here seems to be fear of concentration of social power. This individual, who feels keenly the injus- tice of the present social order and who sympathizes with labor and other subordinate groups, nevertheless cannot militantly support their strivings for greater power. He feels that "powerful unions are as dangerous as powerful business. " He is prone to emphasize the idea that unions are no longer weak in relation to organized industry, and he is likely therefore to accept, in one form or another, the conservative argument that unions are all right but their power must be limited. He would like to decrease the power of business but finds difficulty in directly opposing it-"we might, after all, be as bad as they are. " He believes in extending the economic functions of gov- ernment, perhaps even in some degree of nationalization of industry, but fear of government power often leads him to oppose liberal measures or to
? POLITICO-ECONOMIC IDEOLOGY AND GROUP MEMBERSHIPS I 77
support halfway measures which accomplish little. Opposed to force, he tends to confuse force with militancy and to be indecisive, critical of both sides, overly compromising, inept at political action, shocked by the realities of political affairs. He is likely to make a middle rather than low score on the PEC scale, not out of true conservatism but rather out of inhibited liberal- ism; he has, one might say, a "liberal" utopia but he cannot fight for the social changes necessary to realize it.
2. 41 (6) l""::
t:;l 1. 60 (10. 5) l""::
1. 06 (13) t"' 0
3. 28 1. 50 (12) C)
::tl 4. 04 2. 24 (8) 0
q 3. 08 2. 03 (9) '"d
2. 65 1. 60 (10. 5) 3. 92 2. 08
~
0
. . . . . . . . . ,
(")
0
I
~
(")
0
><
t:;l
l""::
1::0
::tl
"II':
. . . . . .
'"d "'
"
0'1
? 168 THE AUTHORITARIA~ PERSONALITY
These groups are more consen'ati\'e on specific issues than the over-all scale mean of 3. 92 indicates. The over-all means on the items (4, 9, r5, 31, 37, 43) dealing with unions, business, and government functions range from 4? 74 to 5? 35, and these items are also the most discriminating. It would appear, then, that with regard to what is most definitive in liberalism and conserva- tism-mainly ideas regarding power relations among labor, business, and gov- ernment-the liberal position is as yet less crystallized and less militantly held than is the conservative position. There is some question as to how far these results can be generalized beyond the present sample. 5 They are, however, in general accord with numerous other findings and observations regarding the contemporary political scene. What is more difficult, and also more im- portant, to gauge is the psychological potential for future ideological de- velopment in various directions in the face of changing political and economic conditions. Perhaps the other components of political ideology, when sys-
tematically measured and psychologically understood, would provide a basis for the solution of this problem.
4. THE THIRD PEC SCALE (FORMS 45 AND 40)
The construction of Forms 45 and 40 was, as has been discussed in the previous chapter, influenced greatly by considerations of practicality and of administrative expediency. In view of these considerations, and in order to make room for the inclusion of other material, the PEC scale was cut literally to the bone. It was identical in both forms of the questionnaire and contained only five items-not enough to obtain an adequate measure of re- liability, and hardly enough to be called a "scale. " The reasoning behind the use of a five-item E scale was discussed and criticized in Chapter IV; the same criticisms apply to the present PEC scale. It appears now that it would have been wiser to have used a ro-item form; the short form used did, how- ever, make possible the comparison of various groups and the study of rela- tionships between this scale and the others.
The Form 45-40 PEC scale is presented in Table 7(V). It will be seen that the five items were not selected solely on statistical grounds; rather an attempt was made to include items whose D. P. 's were above a minimal level and, more important, which covered as many as possible of the ideological trends previously discussed. The first four items are taken, with occasional slight revisions, from Form 6o. Item 17 (Economic security), has a history of transiency; originally in the Form 78 PEC scale, it was moved to the F scale in Form 6o (see Chapter VII); it has been returned to PEC in an attempt to rid the F scale of all items which might be connected fairly directly with
5 The representativeness of this sample with respect to political pany and other group memberships will be considered later in this chapter (Section E). That the university groups are not unusually conservative is suggested by the fact that E-scale means are rela- tively low in comparison with other middle-class groups (see Chapter IV). Their PEC means can be compared with those for the groups taking Forms 45 and 40, below.
? POLITICO-ECONOMIC IDEOLOGY AND GROUP MEMBERSHIPS 169
TABLE 7 (V)
THE THIRD FoRM oF THE PoLITico-EcoNoMIC CoNsERVATISM (PEC) ScALE (FoRMs 45-40)
3-" Labor unions should become stronger and have more influence generally. 7? America may not be perfect, but the American Way has brought us about
as close as human beings can get to a perfect society.
I I. " Most government controls over business should be continued even though
the war is over.
I4- Men like Henry Ford or J. P. Morgan, who overcame all competition on
the road to success, are models for all young people to admire and imitate. I7. In general, full economic security is bad; most men wouldn't work if they
didn't need the money for eating and living.
? These items are "liberal," the others are "conservative. " A high score is given for agree- ment with the conservative items, disagreement with the liberal items.
existing ideologies regarding politico-economic or minority groups. Item I 7, as well as several others, might be included in any of several scales; proper placement must ultimately be based on statistical analysis.
Forms 45 and 40 were administered late in I945 and in the first part of 1946. The sampling methods and problems, as well as the composition of the groups and their combination for statistical purposes, have been discussed in Chapter IV. It will suffice here to list the groups comprising this sample.
The Form 45 sample contains four groups: (a) Extension Testing Class (adult) Women (N =59) at the University of California; (b) San Quentin Men (N = no), inmates at the California State Prison; (c) and (d) Psychi- atric Clinic Women (N = 7I) and Men (N = so), mostly outpatients at a community clinic in San Francisco.
The following groups are included in the Form 40 sample: (e) George Washington University Women (N = qz), members of day and evening classes in psychology; (f) California Service Club Men (N = 63), obtained at luncheon meetings of San Francisco Bay Area Kiwanis and Rotary clubs;
(g) and (h) Middle-Class Men (N = 69) and W omen (N = I54), mem- bers of various local groups such as church, P. T. A. , women's clubs, etc. ; (i) and (j) Working-Class Men (N = 6I) and Women (N =53), mem- bers of local groups such as United Electrical Workers Union, Warehouse- men's Union (1. L. W. U. ), California Labor School, etc. ; (k) and (I) Los Angeles Men (N = I I 7) and Women (N = qo), a heterogeneous but largely middle-class sample of various local groups in Los Angeles. Data on some of the subgroupings within these statistical units will be considered
in Section E, below.
In addition, there were two groups which were given both Forms 45 and
40. First, the School for Merchant Marine Officers (to be referred to as "Maritime School") (N = 343), half of which was given Form 45, the other half Form 40, the two halves being equated for intelligence (AGCT), time in school, and planned function as officer (deck or engine). Second, veterans
? I]O
THE AUTHORITARIAN PERSONALITY
TABlE 8 (V)
MEANS AND STANDARD DEVIATIONS OF PEC SCAlE SCORES FOR GROUPS TAKING FORMS 45 AND 40
Group
a. Groups taking Form 45 Extension Testing Class
N
Mean Rank S. D.
4. 33 (4) 1. 28 4. 68a (2} 0. 96 4. 12 (11) 1. 53 4. 14a (10) 1. 40
Mean EA
3. 77 5. 33 4. 23 3. 92
4. 31
4. 04
4. 31 3. 89 3. 64 3. 92 3. 91 3. 82 3. 71
3. 91
4. 95
4. 43 4. 69
4. 13
Rank
(12)
(1)
(5) (7. 5)
(6)
(4) (10) (14) (7. 5) (9) (11) (13)
(2)
(3)
Women 59
San Quentin Men Psychiatric Clinic Women Psychiatric Clinic Men
110
71
50 Over-allb 290
4. 32
4. 30
4. 83 4. 30 4. 26 3. 39 3. 25 3. 91
Los Angeles Women
Over-allb 779 4. 05
1. 29
(6. 5) 1. 13
(1) 1. 31 (6. 5) 1. 52 (8) 1. 62 (13) 1. 58 ( 14) 1. 53 (12) 1. 49
b. Groups taking Form 40 George Washington
University Women 132 California Service Club
Men 63
Middle-Class Men Middle-Class Women Working-Class Men W orking-Glass Women Los Angeles Men
69 154 61 53 117
c. Groups taking both forms Maritime School Men
Form 45
Form 40
Employment Service Men V eterans
179 4. 31) 164 4. 32)
( 5 )
Form 45 Form 40
4. 35) (
51
55 Over-allb 449
(3) (1. 28 4. 34 1. 18
d. Totals for all groups 1518 4. 19 1. 37
130 4. 16 (9)
1. 41 1. 45
~1. 0 8
aThe use of two forms for the Psychiatric Clinic groups complicated the PEC scale results somewhat. The data above are based on 45 women and 29 men taking the Form 45 PEC scale. For the remaining 26 women taking the Form 60 PEC scale (14 items) the mean was 4. 05, and for 21 men the mean was 4. 04. For the combined 47 men and women taking this scale, the reliability was . 77 and the Standard Deviation was 1. 05-- values com- parable to those of the other groups taking Form 60.
brn obtaining the over-all values, the individual group means were not weighted by N.
4. 37)
Statistical Property
? POLITICO-ECONOMIC IDEOLOGY AND GROUP MEMBERSHIPS I 7I
coming for vocational guida. nce to the U. S. Employment Service over a period of several months, the first 51 receiving Form 45, the next 55, Form 40. This procedure had mainly to do with determining possible effects of the presence of certain E items on the F-scale responses and will be discussed in more de- tail in Chapter VII.
In view of the shortness of the PEC scale, no reliabilities were computed. However, means and S. D. 's were obtained for each group and are given in Table 8 (V). The group means appear to fall into three main levels of mag- nitude. The two most conservative groups are the California Service Club Men (4. 83) and the San Quentin Men (4. 68). At an intermediate level, with PEC means of 3. 91 to 4? 37 (significantly lower than the first level and higher than the third) is the bulk of the total sample: University groups, Psy- chiatric Clinic Patients, the Middle-Class and Los Angeles groups. Finally, the most liberal groups-although the means of 3. 25 and 3? 39 are far from extreme, and the variability within each group is large-are the Working- Class Men and Women.
That the Service Club Men make the highest mean is not so much a new discovery as a partial indication that the scale provides a valid measure of conservatism. The program and tradition of these groups are fairly explicit in their support of numerous trends in conservative ideology. It will be re- called that similar results were found with the Oregon Service Club Men
(Form 6o). Once again, however, we must emphasize the variability within this and the other groups.
The great conservatism of the San Quentin Men may come as a surprise to those who conceive of criminals as conscious foes of the social order and to those who assume a psychological affinity between criminality and radical- ism. It might have been expected that those who violate prevailing laws re- garding property and morality would tend to oppose the prevailing social ideology and social authority. Yet this does not appear to be the case. Crimi- nals accept the basic premises of the capitalistic system while at the same time engaging in a pseudorebellion against the formal rules and technicalities. The criminal does not oppose the principles of rugged individualism; he simply carries them ad absurdum. The San Quentin material and the relation of criminality to antidemocracy are considered further in Chapter XXI.
The problem of class differences in conservatism is raised by the fact that the Working-Class Men and Women make a significantly lowerPEC mean than do the Middle-Class Men and Women and the Los Angeles group
(which is largely middle class). There are several reasons for questioning whether these differences can be generalized to the broad middle- and work:- ~ ing-class populations. For one thing, the Working-Class group shows a / distinct sampling bias in a liberal direction: almost half the members of this 1 group are from the United Electrical Workers (C. I. O.
), a militant union, orj\ from the California Labor School, a strongly left-wing institution. The
? THE AUTHORIT ARIAN PERSONALITY
Middle-Class groups are more varied and probably more representative in membership. Furthermore, several groups with PEC means similar to those of the Middle-Class groups contain a large proportion of working-class indi- viduals; these groups are the San Quentin JVIen (almost entirely working class), the Psychiatric Clinic Patients, and the Maritime School and Employ- ment Service Veteran Men. In view of the intergroup as well as the intra- group variability, it seems safe to conclude that over-all class differences in political ideology are not extremely large, and that individual and group differences within each class are'so great that they become the primary problem requiring explanation. How does it happen, for example, that the same working-class background produces a law-abiding conservative worker, a politically conservative criminal, a company union leader, a C. I. O. leader, a Communist? W h y does one middle-class individual join a service club while another becomes a supporter of Henry Wallace? Why is it that some, perhaps most, workers identify with the middle class or with the economic status quo, and some individuals with middle-class background identify with what they conceive to be the true interests of the working class? These may be not so much questions of actual class or group membership as questions of class or group identification-and "identification" is a psychological variable. An individual, in making his social identification, is determining not only his ideology, but also what he is to be like as a person. We shall have occasion to consider further, in the chapters that follow, the deeper emotional trends that help to determine the individual's group memberships and identifications.
How close is the relation between conservatism and ethnocentrism in the various groups studied? A means for obtaining a preliminary answer to this question is to compare group means on PEC and onE (see Table 8(V)). Since most groups took the short EA scale, the EA means were used even for those groups which took the total EAB scale (see Chapter IV). The rank- order correlation between the PEC means and EA means for the fourteen groups was +-so, indicating a statistically significant but not very close re- lationship. In general, as the degree of group conservatism increases, the degree of ethnocentrism also increases. The four groups with conspicuously high EA means are the San Quentin Men (5. 33), the Maritime School Men
(4? 95), the Employment Service Men Veterans (4? 43), and the California Service Club Men (4. 31). These groups ranked 2, 5, 3, and 1, respectively, on PEC. No groups were conspicuously low onE, the eight lowest groups having no means within the fairly narrow range of 3. 64-3. 92; the most liberal groups were among the least ethnocentric. The over-all EA mean was 4? I 3, almost identical to the over-all PEC mean of 4-19? (The EAB mean is some- what lower, partly because of sampling differences and partly because the EB items-four on Jews and one on Negroes-had lower means. )
The correlation of ranks does not, however, tell the whole story. Many groups made a significantly higher mean on PEC than on E, or vice versa.
? POLITICO-ECONOMIC IDEOLOGY AND GROUP MEMBERSHIPS 17 3
Whether the group mean on PEC is higher than on E, or lower, seems to depend in large measure on the socioeconomic class of the group: the pre- dominantly middle-class groups tend to be lower on E than on PEC, while the working-class groups are, it appears, more ethnocentric than conservative. Consider the middle-class groups: the PEC and E means, respectively, for the Extension Testing Class Women are 4? 33 and 3. 77; for the George Wash- ington University Women, 4. 30 and 4. 04, for the California Service Club Men, 4. 83 and 4. 31; and similarly for the Middle-Class and Los Angeles Men and Women. Only one of these PEC-E differences is below the 5 per cent level of statistical significance. It will be recalled that in the Form 78 and Form 6o samples, largely middle class, the level of conservatism was greater than the level of ethnocentrism. The opposite trend is found in the working-class (or marginal middle-class) groups. Thus the PEC and E means, respectively, are: for the San Quentin Men, 4. 68 and 5. 33; Working- Class Men, 3? 39 and 3. 92; Working-Class Women, 3. 25 and 3. 91; Maritime School Men, 4? 32 and 4? 95; Employment Service Men Veterans, 4. 36 and 4-43? This leaves only the Psychiatric Clinic Men and Women, who are heterogeneous with respect to class and whose PEC and E means differ only slightly (o. 1-o. 2).
Several factors-not mutually exclusive-may help to explain these class differences. First, open prejudice is more accepted on a verbal level in the working class than in the middle class. The higher E means of the former may therefore reflect, in part, the verbal atmosphere rather than a difference in basic outgroup hostility. (This factor would not hold for the pro-ingroup items. ) Then there is the previously discussed "pseudodemocratic fa~ade," which is more characteristic of the middle than of the working class, and which the E-scale items probably only partially circumvented. Also, certain trends in liberal ideology may appeal to some workers not on a truly liberal basis but on a "class-ethnocentric" basis which is an aspect of general ethno- centrism. For example, some workers are strongly prounion and resentful of "bosses," yet at the same time are anti-Negro, anti-foreigner, and con-
servative regarding many political issues.
All in all, the group data lead us to investigate further the relationship
between ethnocentrism and conservatism, with an eye both to what makes the correlation relatively high and to what keeps it from being higher. These problems will be pursued further when we consider the correlations between the PEC and E scales, and the psychological connection between conservatism and ethnocentrism in the individual. But first we must complete the presentation of the PEC-scale data.
Table 9(V) gives a summary of the item analysis of the Form 45-40 PEC scale. Data for the men and the women are summarized separately. The over-all mean for the women, 4. 07, is significantly lower than that of 4. 25 for the men. The women were also, as noted in Chapter IV, slightly but sig-
? 1 74
No.
3.
7. 11. 14. 17.
THE AUTHORIT ARIAN PERSONALITY TABLE 9 (V)
MEANS AND DISCRIMINATORY POWERS OF THE PEC SCALE ITEMS (FORMS 40 AND 45)
Item
Mean D. P.
Rank
(4) (3) (5) (1) (2)
Mean D. P . Rank
4. 67 3. 49 (4) 4. 57 3. 98 (1) 3. 92 2. 97 (5) 3. 56 ? 3. 90 (2) 3. 62 3. 77 (3)
4. 07 3. 62
(Labor unions) 4. 51 3. 16 (American Way) 4. 90 3. 33 (Government controls) 4. 19 3. 08 (Ford and Morgan) 3. 75 3. 58 (Economic security) 3. 93 3. 46
Mean per iteme 4. 25 3. 32
MEN'S GROUPSa (N 869)b WOMEN'S GROUPSc (N 573)d
aThe individual groups of men in this sample are as follows: San Quentin Men Prisoners (N = 110). Employment Service Men Veterans (N = 106). Maritime School Men (N = 343). 'California Service Club Men (N = 63). Middle-Class Men (N = 69). Working-Class Men (N = 61). Los Angeles Men
(N = 117).
bThe over-all N for the PEC scale (men) is 100 less than that for the E scale because two groups were omitted: (1) Of the 50 Psychiatric Clinic Men, only 29 took the regular Form 45; the others took a form equated for E and F, but not for PEC. Because of the small N, no PEC scale analysis was made on this group. (2) The 50 Working-Class Men and Women were use~ as a statistical group for analysis on the E scale because additional data on the total E scale were desired; but their F and PEC scales were not analyzed statistically.
cThe individual groupg of women in this sample are as follows: Extension Testing dlass Women (N =59). George Washington University Women Students (N =132). Psychiatric Clinic Women (N =45). Middle-Class Women (N = 154). Working-Class Women (N = 53). Los Angeles Women
(N = 130).
dThis N is 26 less than the over-all N for women on the E and F scales because only 45 of the 71 Psychiatric Clinic Women took the regular Form 45. The remaining 26 took a form which was equated to Form 45 for E and F, but not for PEC.
eln obtaining the over? all means, the individual group means were not weighted by N.
nificantly less ethnocentric than the men. This may, however, be a differ- ence in the sample rather than in the total population, since we have no female groups comparable to the high-scoring San Quentin, Service Club, and Maritime School male groups. Moreover, as shown in Table S(V), for all comparable sex groups (Psychiatric Clinic, Middle Class, Working Class, Los Angeles) the means for men and women are almost--identical. Since the sampling methods used were not primarily designed to determine the aver- age intensity of any opinions or attitudes in broader populations, it is perhaps safest not to draw inferences about the total male and female population. It
? POLITICO-ECONOMIC IDEOLOGY AND GROUP MEMBERSHIPS I 75
can be said, however, that? for groups of the general type represented here, no sex differences of practical significance seem to exist; and that differences among male groups and among female groups are much greater than the differences between males and females.
The male and female subsamples are also very similar with regard to means and D. P. 's on the individual items. While the scale mean was lower in women, the relative standing of the item means was almost identical for the two sexes, the rank-order correlation being . 90. Women were more con- servative than men on only one item (unions). A similar relation holds also for the D. P. 's, the rank-order correlation being . 70.
The general level of D. P. 's would, other things being equal, be slightly greater for a 5-item than for a I4-item scale, since each item contributes more to the total score. Therefore, the average D. P. of 3? 4-3? 5 for Forms 45 and 40 is comparable or slightly superior to that of 2. I for Forms 78 and 6o. All of the present items seem statistically adequate. Item I I (Government controls over business) had the lowest D. P. , but in view of the greater suc- cess of the "government function" items in earlier forms, improvement should not be difficult. While the five items can hardly claim to be considered a "scale," they show sufficient internal consistency so that one may mean- ingfully speak of "total PEC score" and one may determine the relations between this and various other measures.
The level of internal consistency of the PEC scale is indicated also by a correlational analysis made on a group of 5I7 University of California women students. 6 A mean r of +o. 26 was found between each item and the sum of the remaining items, the range of r's being +o. Io to +o? 33? The rank order of these items, according to the size of the item's correlation with the sum of the others, was identical to the rank order of item D. P. 's for the combined women's groups above (Table 9(V)). The correlations among individual items averaged +o. I4, the range being +o. o2 to +o. 3o. The highest correla- tion, 0. 30, was between Item 7 (American Way ideal) and Item I4 (Ford and Morgan). Only three r's were below . 10 (the I per cent level of significance), and all of them involved Item II (Government controls). These correlations, while far below those for the E scale, indicate that the PEC scale meets the minimum requirements for its present uses, and that a scale of 20 or 30 such items might, without loss of breadth, achieve a re- liability in the neighborhood of . 8.
5. DISCUSSION: SOME P A TTERNS OF CONTEMPORARY LIBERALISM AND CONSERV A TISM
The reliability and internal consistency of the PEC scales suggest, on the one hand, that liberalism and conservatism are relatively organized and meas- urable patterns of current politico-economic thought; and, on the other
6 The group and procedure are discussed more fully in Chapters IV and VII.
? THE AUTHORITARIAN PERSONALITY
hand, that within each of these broad patterns there is considerable subpat- terning, inconsistency, and simple ignorance. To ignore either the relative generality or the relative inconsistency would, it seems, lead to serious mis- understanding of the problem. More detailed exploration of the nature and deeper psychological meaning of these ideological trends, as expressed in the interviews, will be made in Chapter XVII, following presentation of the clinical material. However, at this point we ought briefly to consider, on the basis of the scale data and of some individual patterns of scale response, certain variations within liberalism and within conservatism.
Liberalism was conceived not as a single, unitary attitude, but as an ideo- logical system containing a number of trends or components. The reliability and internal consistency of the initial forms of the scale show that these trends are interrelated significantly but imperfectly in the individual. The prototypic "liberal" is, according to our guiding conception, an individual who actively seeks progressive social change, who can be militantly critical (though not necessarily totally rejective) of the present status quo, who opposes or de-emphasizes numerous conservative values and beliefs regard- ing business success, rugged individualism, human nature, and the like, and who would diminish the power of business by increasing the power of labor and the economic functions of government.
It is clear, however, that many individuals who are generally liberal do not exhibit some of the above trends. While some of the inconsistency- perhaps the largest part-is due to confusion resulting from lacks and dis- tortions in the press and other media of mass communication, part of it seems also to reflect deep-lying emotional trends of considerable intensity and resistance to change. The individual's pattern of thought, whatever its con- tent, reflects his personality and is not merely an aggregate of opinions picked up helter-skelter from the ideological environment.
One variant, particularly common in the groups tested, might be called the politically pacifistic liberal. The guiding idea here seems to be fear of concentration of social power. This individual, who feels keenly the injus- tice of the present social order and who sympathizes with labor and other subordinate groups, nevertheless cannot militantly support their strivings for greater power. He feels that "powerful unions are as dangerous as powerful business. " He is prone to emphasize the idea that unions are no longer weak in relation to organized industry, and he is likely therefore to accept, in one form or another, the conservative argument that unions are all right but their power must be limited. He would like to decrease the power of business but finds difficulty in directly opposing it-"we might, after all, be as bad as they are. " He believes in extending the economic functions of gov- ernment, perhaps even in some degree of nationalization of industry, but fear of government power often leads him to oppose liberal measures or to
? POLITICO-ECONOMIC IDEOLOGY AND GROUP MEMBERSHIPS I 77
support halfway measures which accomplish little. Opposed to force, he tends to confuse force with militancy and to be indecisive, critical of both sides, overly compromising, inept at political action, shocked by the realities of political affairs. He is likely to make a middle rather than low score on the PEC scale, not out of true conservatism but rather out of inhibited liberal- ism; he has, one might say, a "liberal" utopia but he cannot fight for the social changes necessary to realize it.
