In order to obtain a quantitative measure of the relationship between Pro- jective Question scores and standing on the E scale, the
percentage
of PQ-E agreement was computed for each quartile.
Adorno-T-Authoritarian-Personality-Harper-Bros-1950
"Scenic grandeur.
" "Seeing something of real importance.
" ?
5? Omissions are recorded as Nb, scored as H. Of the 6 omissions (9 per cent), 5 were by highs.
Note: Due to various circumstances (seep. 58o), only 65 of the total of 312 subjects received this question. It is likely, therefore, that further experi- ence will suggest modifications and particularly additions to the present scor- ing scheme.
SUMMARY OF PROJECTIVE QUESTION CA TEGORIES
High Categories Low Categories
Question I: What moodsare unpleasant or disturbing?
1. Violations of conventional val- ues
2. Threatening or nonsupporting environment
3? "Rumblings from below"
1. Conscious conflict and guilt
2. Focal dependency and love-seek- ing
3? Open hostility, by self or others, toward love objects
4? Omissions
Question 2: What desires are most difficult to control?
I. N onfocal and/or motor aggression 1. Focal (usually verbal) hostility directed against violators of
2. Ego-alien passivity
3? Impersonal sex
4? Incidental pleasures and viola-
tions of conventional values
5. Omissions
achievement values
2. The tendency to violate achieve-
ment values oneself 3? Miscellaneous
? 1. 2. 3?
4?
1. 2.
3?
1. 2. 3?
1.
2.
1.
2.
Power and control Conservative Americans Parents and relatives
1. The arts and philosophy
2. Physical and biological scientists 3? Social scientists, liberal-radical
political figures
4? Active denial of admiration
3? Passivity 4? Omissions
PROJECTIVE QUESTIONS
Question 3: What great people do you admire most?
579
Miscellaneous
Question 4: What might drive a person nuts?
"Rumblings from below" 1. Inner psychological states Threatening, irritating, or non- 2. Dominating, blocking, rejecting supporting environment. environment
Omissions
Question 5: Worst crimes a person could commit?
Crude aggression and sex 1. Violation of achievement values Other immoral acts
Various legal offenses
Question 6: Most embarrassing moments?
Violations of convention and 1. Hurting another's feelings. etiquette
Blows at exhibitionism and nar- 2. Feelings of inadequacy, failure, cissism being rejected
Question 7: How would you spend your last six months?
Conventional morality and in- 1. Achievement values: creativity hibitions and social contribution Incidental, dilute pleasures 2, Open sensuality and active pleas-
ure
Question 8: What experiences would be most awe-inspiring? .
1. Realization of conventional values 1. Realization of achievement val- ues
2. Power: deference and submis- 2. Power as exemplified in man's
sion toward power figures
3? Destruction-harm of other per-
sons
4 Dilute experiences of nature and
beauty
5. Omissions
achievements and in nature 3? Intense nature experiences
D. RESULTS
Practical considerations prevented quantification of the Projective Ques- tion material from every group to which the questionnaire was administered. (For a complete list and description of these groups, see Chapter IV. ) The responses of the entire high and low quartiles of the following groups were analyzed. (The N's in the parentheses refer to the number of cases in the
? THE AUTHORITARIAN PERSONALITY
extreme high and low quartiles combined, on which the Projection Ques- tion analysis was made. )
I. Psychology Women (N=63): the members of a Psychology class at the University of California. Initial forms of questionnaire, prior to Form 78.
2. Employment Service Veteran Men (N =5I): an unselected sample of the Vet-
eran population going through an office of the United States Employment
Service. Forms 45 and 40.
3? Psychiatric Clinic Men (N =29); and 4? Psychiatric Clinic Women (N =34):
in- and out-patients at a local Community Clinic. Most of these subjects re- ceived Form 45, but some were given a shorter form which included only four projective questions.
5. Middle-Class W omen (N = 70): A highly diversified sample from various middle-class groups: religious, political, business, and the like. Form 40.
6. Middle-Class Men and Service Club Men (combined) (N =65): While the scale responses of these two groups were statisticized separately, it seemed feasible to combine them into a single group for the present purposes. The Middle-Class Men were obtained from largely the same groups as the Middle- Class Women. Form 40.
These groups constitute a fairly representative sample of all those studied. On the basis of the present results, it appears likely that similar quantitative differences between the highs and lows would have been obtained had we analyzed the Projective Question material of the remaining groups. The de- termination of qualitative differences among the highs from various group- ings, and among the lows from various groupings, remains an interesting problem for future research.
Questionnaire Form 45 contained the eight Projective Questions in the order listed above (Section A). As part of the process of cutting Form 40 down to an absolute minimum (see Chapter IV), only the first five of these questions were used. Further complications occurred in the case of the initial form (taken by the Psychology vVomen), which contained only the first seven questions, and in the case of the Psychiatric Clinic Men and Women, some of whom received a shorter form which contained only Questions 3, 4, 5, and 6. The inconsistency is part of the general prob- lem of working out a standard set of questions. However, the incon- sistencies in Form 40 and in the form given to the clinic patients must be regarded as methodological errors, since complete data on all eight questions would have compensated for any possible increase in administrative dif- ficulties. As a result of differences in the number of questions in each form of the questionnaire, there are also differences in the number of subjects receiving a given question. The number of subjects receiving each of the eight questions is indicated and explained in Table I (XV), below. All 3I 2 of the subjects received forms containing Questions 3, 4 and 5? The N varies for the other questions, reaching a low of 65 on Question 8, which was con- tained in Form 45 given to the Employment Service and Clinic groups.
? PROJECTIVE QUESTIONS sSr
The Scoring Manual presented above is a slight modification of the one used by the scorers. The original Manual contained the same categories with substantially the same definitions. It was based on an analysis of the responses of two large groups: University of California Students (male and female) and the University of California Summer Session Adults, Form 6o (adult men and women from various sections of the middle class). As the scoring pro- ceeded, certain additional implications and theoretical points were brought out and incorporated as notes in the original Manual. Also, an attempt was made to clarify certain ambiguities pointed out by the scorers. These notes are in the text of the present Manual. The only further change is the addi- tion of a number of examples from the groups on which the data below are based. While the present Scoring Manual is more articulate than the original, no essentially new theoretical points have been added.
1. RELIABILITY OF SCORING
The critical reader may, after going through the Scoring Manual above,
legitimately ask whether the proposed differences between highs and lows are "really" there, or whether they are not imposed by the writer's bias. A partial answer to this question is offered by means of the controlled scoring procedure. In the last four groups (all but the Psychology Women and the Veteran Men), the following scoring procedure was followed.
The high and low quartiles from a given group were combined and their responses to each item typed in a single, randomly ordered series. Each re- sponse was identified by a code number, so that the scorer did not know whether it was given by an individual scoring high or by an individual scoring low on ethnocentrism. Moreover, the code numbers for each individual varied from item to item in order to prevent halo effect (e. g. , the tendency to give an individual a score of H on Item 2 because he was scored H on Item r). Each rater went through all responses of the combined-high-plus- low grouping for each item, recording her scores of H, L, Na, Nhl, or Nb (see Section A for key to symbols) for each code number. Only after all items had been scored were the code numbers taken away and the identity (with respect to standing on E) of each subject restored. The scorer was, therefore, entirely on her own in deciding whether each response fell into a high, low, or neutral category. This is what is meant by "blind" scoring.
In the case of the Psychology Women and the Employment Service Vet- erans the scoring was not done blindly. This was recognized as a methodo- logical error and corrected on all subsequent groups. However, the advantage in knowing the subjects' standing on E may have been partially counter- balanced by the newness of the task for the scorers, and by the emphasis placed on caution.
As a further check on the dependability of scoring, it was always done independently by two raters. Their degree of agreement in assigning scores
? s8z THE AUTHORIT ARIAN PERSONALITY
gives a measure of reliability, i. e. , of the probability that these results can be duplicated by other raters with similar training. While high reliability does not in itself prove the correctness of the interpretations regarding the deeper meaning of the scoring categories, it does indicate that the scoring categories, as specifically defined, have been objectively measured and are not merely figments of the imagination. Scoring reliability is, then, one index of objec- tivity.
A word ought perhaps to be said about the training of the raters. 6 Both were, when the scoring started, at approximately the level of first-year grad- uate students in psychology. Their learning of the Scoring Manual was part of the process of becoming familiar with the general theoretical orientation of the present research. Neither had had any clinical experience or intensive training in dynamic personality theory, beyond a few undergraduate courses. In addition to studying the Scoring Manual, they had the benefit of several preliminary practice sessions on groups not included in the final statistical treatment. The nature of the scorers' background and training is stressed because it reveals that detailed familiarity with a particular psychological theory is not essential for scoring; theory is, of course, essential for an inte- grated understanding of the total pattern of data.
When the two raters had independently made and recorded their scores for a given group, a conference was held for the purpose of assigning a final score for each response. As has been noted above, each response was scored High, Low, or Neutral; the H or L scores did not specify which particular category (e. g. , high category 3 or low category I) the response repre- sented. The reason for this is that a response might represent variables in more than one category; or it might express in abstract form an underlying high or low trend without falling into a specific category as described. A scoring disagreement was registered whenever the two original scores were not identical. Discussion of the disagreements usually convinced one rater or the other to change in the other's direction. Occasionally a response scored H by one rater and L by the other received a final score of Na or Nhl.
The scoring reliability, that is, the percentage of agreement between raters, was computed as follows. A full error was counted when one rater scored H, the other L. A half error was counted when one rater scored Na or Nhl and the other scored H or L. Thus, if in a group of so there are 8 full errors and 4 half errors, the percentage disagreement is 10/so or zo per cent, or in positive terms there is So per cent agreement.
The. reliability data are presented in Table I (XV). The mean percentage agreement of 90 meets current standards for materials of this sort, and indi- cates that the two sets of scores, independently and blindly derived, agreed very well. In the case of Groups I and z, ? which were scored independently but not blindly (the raters knowing which was the high-on-E subject, which
6 We wish to express our thanks to Anne Morrow and Ellan Ulery for their work "beyond the call of duty" in learning and applying the scoring scheme.
? Group
1. Psychology Womenc
2. Employment Service
N Percentage Agreement
. . 1 . 1. _1. . . i 5 ~ ~ . . ! ! .
Meanb
88
) 95 )
)80 )
)87 )
92
89
90
1859
PROJECTIVE QUESTIONS TABLE 1 (XV)
SCORING RELIABILITY (PERCENTAGE INTERRATER AGREEMENT) FOR THE EIGHT PROJECTIVE QUESTIONS
Veteran Men
3. Psychiatric 29
917190
948887
Clinic Men
4. Psychiatric Clinic Women
5. Middle-Class Womend
6. Middle-Class and Business Club Mend
(17)a
34 (24)a
70
65
76 82
Mean
Percentage 93 Agreementb
T o t a l N 312 290
63 94 75 87 84 95
51 100 96 95 95 94 (24) a
94
90
90
84
85 -
100 87
59 71
96 79
91 82 96 92 97
--- 8594889590 83 83
290 312 312 312 150 128 65
86 83958098
aThe N for Groups 2, 3, and 4 is complicated by the fact that two ques- tionnaire forms were given. Only 24 of the 51 highs and lows in Group 2 feceived Form 45, which contained all 8 questions; the remaining 27 received Form 40, containing only questions 1-5. Similarly, some of the subjects in Groups 3 and 4 filled out Form 45, while o. thers received a modified form containing items 3, 4, 5, and 6 only, The differences between subgrodps within each sample are random.
brhe over-all group and item means are based on single item means weighted by N.
CGroup 1 received Form 78, which did not contain question 8.
dGroups 5 and 6 received Form 40, which did not contain questions 6, 7, and 8.
eKey to questions: 1 (Moods), 2 (Desires), 3 (Great People), 4 (Drive Nuts), 5 (Crimes), 6 (Embarrassing), 7 (Last Six
Months), 8 (A we-inspiring),
81 88 - --
?
? ?
THE AUTHORITARIAN PERSONALITY
the low), only Group 2, with 95 per cent agreement, is above the over-all mean. For only one group, the Psychiatric Clinic Men, does the reliability drop conspicuously below 90 per cent. This drop (to So per cent) seems due in part to certain intrinsic ambiguities in the responses of this group, and in part to the fact that this was the first group to be scored blindly.
The reliabilities for the individual items are also satisfactory, ranging from S3 to 95 per cent. Questions I ("Moods"), 3 ("Great people"), and 5 ("Crimes") were scored most consistently. The average agreement on Ques- tions 7 ("Last six months") and S ("Awe-inspiring") might have been higher had they been filled out by Groups 5 and 6. In only 6 cases out of the total of 4I did an item have a reliability of less than So per cent. It would appear, therefore, that the present scoring scheme is relatively reliable, and in this sense "objective. "
2. PROJECTIVE QUESTION SCORES IN RELATION TO STANDING ON THE E SCALE
To what extent are L scores on the Projective Questions characteristic of the anti-ethnocentric individuals, H scores characteristic of the extremely ethnocentric individuals? The Scoring Manual is based on the hypothesis that the low quartiles on the Ethnocentrism scale will give responses falling mainly in the low categories, whereas the responses of the high quartiles will fall mainly in the high categories.
Data bearing on this question are presented in Table 2 (XV), which indi- cates the degree to which the Projective Questions differentiate the ethno- centric from the anti-ethnocentric subjects.
In order to obtain a quantitative measure of the relationship between Pro- jective Question scores and standing on the E scale, the percentage of PQ-E agreement was computed for each quartile. This is called the L% in the case of the low quartile, H% for the high quartile, and %A for the two quartiles combined (average over-all agreement). Let us use L% to illustrate the pro- cedure. Had scores of Hand L been the only ones assigned, the L% would be simply the number of L scores divided by the total number of cases; thus, if all the low quartile members received L scores on a given item there would be I oo per cent agreement between PQ and E. But this procedure was not feasible because of the neutral scores. 7 Since a neutral score represents a half error, i. e. , it signifies less agreement than an L score but more agreement than an H score, the L% was computed according to the following formula:
Sum of L scores plus ~ sum of neutral scores (Na, Nb, Nhl) L%= Nbfb". l ? 1
urn ero su 1ectsm owquartie
7 One possibility would have been not to consider subjects receiving scores of N, and to get a L/H ratio for the remaining subjects. This would have given higher Lo/o values than those obtained by the method finally used, since in the latter method the N scores were used in such a way as to lower the Lo/o. It was believed that all individuals taking the test should be included in the statistical treatment.
? PROJECTIVE QUESTIONS
where Na means "ambiguous," Nb means "blank" (omitted), Nhl means "mixed high and low trends. "
The only exception to this formula is in the case of Nb on Items 1, 2, 4, 7, and 8, where Nb is equivalent to H and is included among the H scores.
The computation of H% is based on an equivalent formula:
Sum of H scores plus ~ sum of neutral scores (Na, Nb, Nhl)
H%= Nbfb. . h. h? 1 urn ero su Jectsm 1g quart1e
Finally, %A is the average of L% and H%, weighted by N (number of cases in each quartile), and it indicates the over-all PQ-E agreement.
Table 2 (XV) gives the L%, H%, and %A for each group tested and for all groups combined. It also gives the number and percentage of Nb, Na, and Nhl responses. It may be noted first that the over-all PQ-E agreement
(%A) is 74? 9 per cent-a value which indicates a statistically significant rela- tionship between Projective Question scores and high vs. low standing on the E scale. On a purely chance basis, the agreement would be only so per cent. The highs tend, however, to be more consistent in their PQ responses than do the lows: the highs have an over-all H% of 8o. 7, while for the lows the L% is only 69. 1. In other words, the lows received more H scores than the highs did L scores. It is, so to speak, easier to make an H than an L score. The reason for this does not appear to lie primarily in the subjects them- selves, for our general impression, based on other results and on clinical judgment, was that the low quartiles fitted the over-all conception of the "democratic" personality at least as well as the highs approximated the pro- totypic "authoritarian" personality. The preponderance of H scores is prob- ably due to the conditions of testing. Since many of the groups were pressed for time, and since the instructions were not emphatic in suggesting that a fairly detailed answer be given, many of the responses were brief and superficial. It will be recalled from the Scoring Manual that references to "vague, dilute emotional experiences," as well as references only to "be- havior or the situation per se, without consideration of inner meanings and motives," were important cues for assigning a score of H. It may be argued that if the instructions emphasized the giving of more detailed answers, both highs and lows would give more responses meriting a score of L. In the present groups, however, the elaborations of most lows took an L direction, of most highs an H direction. It might be expected, then, that the obtaining of longer responses would clarify and increase the differences between the low and high quartiles. It would also probably reduce somewhat the propor- tion (8. 8 per cent) of responses scored Na and Nhl. We should not, of
course, overlook the likelihood that numerous pressures in our culture, and perhaps the predominant ones, tend to make for authoritarianism in the individual. To the extent that this is true, we should expect some H trends even in individuals attempting to achieve a thoroughly democratic orienta- tion.
? 1. Psychology Women Low quartile
Na Score 1 2 3 4 5 6 7 8 Meanb 63
High
quartile
- 83. 6 13 )
d . . . ,
Total
63 %A 91. 3 81. 7 76. 2 76. 2 75. 4 74. 6 72. 2 - 78. 2
Nb 0 4 3 2 6 4 1 -20) >
2. Employment Service Men Low quartile
51
25 L%
>'d t;::l i;l:) Vl 0 z >
High quartile
26 H% 88. 4 94. 2 84. 6 100. 0 100. 0 75. 0 89. 3 78. 6 90. 1 (14) Nb 1 4 1 3 0 1 0 2 12)
Total
51 %A 83. 3 80. 4 80. 4 91. 2 83. 3 77. 1 79. 2 79. 2 82. 6 (24) Nb 5 9 2 5 1 2 0 3 27 )
TABLE 2 (XV)
PERCENTAGE AGREEMENT BETWEEN PROJECI'IVE QUESTION SCORES AND E-SCALE SCORES
Vt 00
32 L%
31 H%
88. 7 82. 3
Na? NHL
>
90. 6 84. 4 79. 7 13. 4 59. 4 60. 9 62. 5 - 73. 0 2 4 7 13 1 4 10 41
. . . , Nb0100330-7) :I:
91. 9 79. 0
Nb 0 3 3 2 3 1 1 - )14. 3%
72. 6 79. 0 91. 9
Na-NHL 3 3 4 7 0 0 . 1 18 . . . .
Na-NHL 5 7 11 20 1 4 11 59 ) 17. 9% z
78. 0 66. 0 76. 0
(10) Nb 4 5 1 2 1 1 0 1 15)
82. 0 66. 0
Na-N'HL 1 1 2 3 0 1 1 0 9)15. 5%
Na-NHL
0 1 5 0 0 0 1 2 9)12. 2%
Na-NHL 1 2 7 3 0 1 2 2 18)13. 8%
80. 0 65. 0 80. 0
73. 9 I:""' . . . .
) 21. 4%
t;::l
0\
:I: 0 i;l:)
. . . ,
> i;l:) . . . .
. . . ,
~
? 82. 2 50. 0 53. 6 46. 4 50. 0 88. 9 58. 1 1021006)
2 0 3 2 1 0 9)16. 3%
83. 3 86. 7 83. 3 66. 7 93. 8 50. 0 77. 7 0101004)
3 0 3 1 1 0 8)13. 1%
82. 8 69. 0 69. 0 56. 8 . 70. 6 70. 6 67. 9
1 1 2 2 0 0 10) I'd
5 0 6
3 2 0
17 )14. 7% l:l:' 0
70. 6 82. 5 55. 9
70. 6 42. 3 88. 5
71. 3
~
10 1002003) c
3 0 3 4 1 1 12)12. 5% M
. . . . . 50. 0 94. 1 97. 1 58. 8 86. 4 81. 8 74. 5 z0 1 0 1 4 1 3 12) (/)
4 0 0 2 1 0 7)17. 0%
60. 3 88. 3 76. 5 64. 7 62. 5 85. 4 72. 8
2 0 1 6 1 3 15)
7 0 3 6 2 1 19)14. 7%
. . . . . . t"1 () . . . , . . . . .
. . . ,
"'
\Jo
00
'-)
?
5? Omissions are recorded as Nb, scored as H. Of the 6 omissions (9 per cent), 5 were by highs.
Note: Due to various circumstances (seep. 58o), only 65 of the total of 312 subjects received this question. It is likely, therefore, that further experi- ence will suggest modifications and particularly additions to the present scor- ing scheme.
SUMMARY OF PROJECTIVE QUESTION CA TEGORIES
High Categories Low Categories
Question I: What moodsare unpleasant or disturbing?
1. Violations of conventional val- ues
2. Threatening or nonsupporting environment
3? "Rumblings from below"
1. Conscious conflict and guilt
2. Focal dependency and love-seek- ing
3? Open hostility, by self or others, toward love objects
4? Omissions
Question 2: What desires are most difficult to control?
I. N onfocal and/or motor aggression 1. Focal (usually verbal) hostility directed against violators of
2. Ego-alien passivity
3? Impersonal sex
4? Incidental pleasures and viola-
tions of conventional values
5. Omissions
achievement values
2. The tendency to violate achieve-
ment values oneself 3? Miscellaneous
? 1. 2. 3?
4?
1. 2.
3?
1. 2. 3?
1.
2.
1.
2.
Power and control Conservative Americans Parents and relatives
1. The arts and philosophy
2. Physical and biological scientists 3? Social scientists, liberal-radical
political figures
4? Active denial of admiration
3? Passivity 4? Omissions
PROJECTIVE QUESTIONS
Question 3: What great people do you admire most?
579
Miscellaneous
Question 4: What might drive a person nuts?
"Rumblings from below" 1. Inner psychological states Threatening, irritating, or non- 2. Dominating, blocking, rejecting supporting environment. environment
Omissions
Question 5: Worst crimes a person could commit?
Crude aggression and sex 1. Violation of achievement values Other immoral acts
Various legal offenses
Question 6: Most embarrassing moments?
Violations of convention and 1. Hurting another's feelings. etiquette
Blows at exhibitionism and nar- 2. Feelings of inadequacy, failure, cissism being rejected
Question 7: How would you spend your last six months?
Conventional morality and in- 1. Achievement values: creativity hibitions and social contribution Incidental, dilute pleasures 2, Open sensuality and active pleas-
ure
Question 8: What experiences would be most awe-inspiring? .
1. Realization of conventional values 1. Realization of achievement val- ues
2. Power: deference and submis- 2. Power as exemplified in man's
sion toward power figures
3? Destruction-harm of other per-
sons
4 Dilute experiences of nature and
beauty
5. Omissions
achievements and in nature 3? Intense nature experiences
D. RESULTS
Practical considerations prevented quantification of the Projective Ques- tion material from every group to which the questionnaire was administered. (For a complete list and description of these groups, see Chapter IV. ) The responses of the entire high and low quartiles of the following groups were analyzed. (The N's in the parentheses refer to the number of cases in the
? THE AUTHORITARIAN PERSONALITY
extreme high and low quartiles combined, on which the Projection Ques- tion analysis was made. )
I. Psychology Women (N=63): the members of a Psychology class at the University of California. Initial forms of questionnaire, prior to Form 78.
2. Employment Service Veteran Men (N =5I): an unselected sample of the Vet-
eran population going through an office of the United States Employment
Service. Forms 45 and 40.
3? Psychiatric Clinic Men (N =29); and 4? Psychiatric Clinic Women (N =34):
in- and out-patients at a local Community Clinic. Most of these subjects re- ceived Form 45, but some were given a shorter form which included only four projective questions.
5. Middle-Class W omen (N = 70): A highly diversified sample from various middle-class groups: religious, political, business, and the like. Form 40.
6. Middle-Class Men and Service Club Men (combined) (N =65): While the scale responses of these two groups were statisticized separately, it seemed feasible to combine them into a single group for the present purposes. The Middle-Class Men were obtained from largely the same groups as the Middle- Class Women. Form 40.
These groups constitute a fairly representative sample of all those studied. On the basis of the present results, it appears likely that similar quantitative differences between the highs and lows would have been obtained had we analyzed the Projective Question material of the remaining groups. The de- termination of qualitative differences among the highs from various group- ings, and among the lows from various groupings, remains an interesting problem for future research.
Questionnaire Form 45 contained the eight Projective Questions in the order listed above (Section A). As part of the process of cutting Form 40 down to an absolute minimum (see Chapter IV), only the first five of these questions were used. Further complications occurred in the case of the initial form (taken by the Psychology vVomen), which contained only the first seven questions, and in the case of the Psychiatric Clinic Men and Women, some of whom received a shorter form which contained only Questions 3, 4, 5, and 6. The inconsistency is part of the general prob- lem of working out a standard set of questions. However, the incon- sistencies in Form 40 and in the form given to the clinic patients must be regarded as methodological errors, since complete data on all eight questions would have compensated for any possible increase in administrative dif- ficulties. As a result of differences in the number of questions in each form of the questionnaire, there are also differences in the number of subjects receiving a given question. The number of subjects receiving each of the eight questions is indicated and explained in Table I (XV), below. All 3I 2 of the subjects received forms containing Questions 3, 4 and 5? The N varies for the other questions, reaching a low of 65 on Question 8, which was con- tained in Form 45 given to the Employment Service and Clinic groups.
? PROJECTIVE QUESTIONS sSr
The Scoring Manual presented above is a slight modification of the one used by the scorers. The original Manual contained the same categories with substantially the same definitions. It was based on an analysis of the responses of two large groups: University of California Students (male and female) and the University of California Summer Session Adults, Form 6o (adult men and women from various sections of the middle class). As the scoring pro- ceeded, certain additional implications and theoretical points were brought out and incorporated as notes in the original Manual. Also, an attempt was made to clarify certain ambiguities pointed out by the scorers. These notes are in the text of the present Manual. The only further change is the addi- tion of a number of examples from the groups on which the data below are based. While the present Scoring Manual is more articulate than the original, no essentially new theoretical points have been added.
1. RELIABILITY OF SCORING
The critical reader may, after going through the Scoring Manual above,
legitimately ask whether the proposed differences between highs and lows are "really" there, or whether they are not imposed by the writer's bias. A partial answer to this question is offered by means of the controlled scoring procedure. In the last four groups (all but the Psychology Women and the Veteran Men), the following scoring procedure was followed.
The high and low quartiles from a given group were combined and their responses to each item typed in a single, randomly ordered series. Each re- sponse was identified by a code number, so that the scorer did not know whether it was given by an individual scoring high or by an individual scoring low on ethnocentrism. Moreover, the code numbers for each individual varied from item to item in order to prevent halo effect (e. g. , the tendency to give an individual a score of H on Item 2 because he was scored H on Item r). Each rater went through all responses of the combined-high-plus- low grouping for each item, recording her scores of H, L, Na, Nhl, or Nb (see Section A for key to symbols) for each code number. Only after all items had been scored were the code numbers taken away and the identity (with respect to standing on E) of each subject restored. The scorer was, therefore, entirely on her own in deciding whether each response fell into a high, low, or neutral category. This is what is meant by "blind" scoring.
In the case of the Psychology Women and the Employment Service Vet- erans the scoring was not done blindly. This was recognized as a methodo- logical error and corrected on all subsequent groups. However, the advantage in knowing the subjects' standing on E may have been partially counter- balanced by the newness of the task for the scorers, and by the emphasis placed on caution.
As a further check on the dependability of scoring, it was always done independently by two raters. Their degree of agreement in assigning scores
? s8z THE AUTHORIT ARIAN PERSONALITY
gives a measure of reliability, i. e. , of the probability that these results can be duplicated by other raters with similar training. While high reliability does not in itself prove the correctness of the interpretations regarding the deeper meaning of the scoring categories, it does indicate that the scoring categories, as specifically defined, have been objectively measured and are not merely figments of the imagination. Scoring reliability is, then, one index of objec- tivity.
A word ought perhaps to be said about the training of the raters. 6 Both were, when the scoring started, at approximately the level of first-year grad- uate students in psychology. Their learning of the Scoring Manual was part of the process of becoming familiar with the general theoretical orientation of the present research. Neither had had any clinical experience or intensive training in dynamic personality theory, beyond a few undergraduate courses. In addition to studying the Scoring Manual, they had the benefit of several preliminary practice sessions on groups not included in the final statistical treatment. The nature of the scorers' background and training is stressed because it reveals that detailed familiarity with a particular psychological theory is not essential for scoring; theory is, of course, essential for an inte- grated understanding of the total pattern of data.
When the two raters had independently made and recorded their scores for a given group, a conference was held for the purpose of assigning a final score for each response. As has been noted above, each response was scored High, Low, or Neutral; the H or L scores did not specify which particular category (e. g. , high category 3 or low category I) the response repre- sented. The reason for this is that a response might represent variables in more than one category; or it might express in abstract form an underlying high or low trend without falling into a specific category as described. A scoring disagreement was registered whenever the two original scores were not identical. Discussion of the disagreements usually convinced one rater or the other to change in the other's direction. Occasionally a response scored H by one rater and L by the other received a final score of Na or Nhl.
The scoring reliability, that is, the percentage of agreement between raters, was computed as follows. A full error was counted when one rater scored H, the other L. A half error was counted when one rater scored Na or Nhl and the other scored H or L. Thus, if in a group of so there are 8 full errors and 4 half errors, the percentage disagreement is 10/so or zo per cent, or in positive terms there is So per cent agreement.
The. reliability data are presented in Table I (XV). The mean percentage agreement of 90 meets current standards for materials of this sort, and indi- cates that the two sets of scores, independently and blindly derived, agreed very well. In the case of Groups I and z, ? which were scored independently but not blindly (the raters knowing which was the high-on-E subject, which
6 We wish to express our thanks to Anne Morrow and Ellan Ulery for their work "beyond the call of duty" in learning and applying the scoring scheme.
? Group
1. Psychology Womenc
2. Employment Service
N Percentage Agreement
. . 1 . 1. _1. . . i 5 ~ ~ . . ! ! .
Meanb
88
) 95 )
)80 )
)87 )
92
89
90
1859
PROJECTIVE QUESTIONS TABLE 1 (XV)
SCORING RELIABILITY (PERCENTAGE INTERRATER AGREEMENT) FOR THE EIGHT PROJECTIVE QUESTIONS
Veteran Men
3. Psychiatric 29
917190
948887
Clinic Men
4. Psychiatric Clinic Women
5. Middle-Class Womend
6. Middle-Class and Business Club Mend
(17)a
34 (24)a
70
65
76 82
Mean
Percentage 93 Agreementb
T o t a l N 312 290
63 94 75 87 84 95
51 100 96 95 95 94 (24) a
94
90
90
84
85 -
100 87
59 71
96 79
91 82 96 92 97
--- 8594889590 83 83
290 312 312 312 150 128 65
86 83958098
aThe N for Groups 2, 3, and 4 is complicated by the fact that two ques- tionnaire forms were given. Only 24 of the 51 highs and lows in Group 2 feceived Form 45, which contained all 8 questions; the remaining 27 received Form 40, containing only questions 1-5. Similarly, some of the subjects in Groups 3 and 4 filled out Form 45, while o. thers received a modified form containing items 3, 4, 5, and 6 only, The differences between subgrodps within each sample are random.
brhe over-all group and item means are based on single item means weighted by N.
CGroup 1 received Form 78, which did not contain question 8.
dGroups 5 and 6 received Form 40, which did not contain questions 6, 7, and 8.
eKey to questions: 1 (Moods), 2 (Desires), 3 (Great People), 4 (Drive Nuts), 5 (Crimes), 6 (Embarrassing), 7 (Last Six
Months), 8 (A we-inspiring),
81 88 - --
?
? ?
THE AUTHORITARIAN PERSONALITY
the low), only Group 2, with 95 per cent agreement, is above the over-all mean. For only one group, the Psychiatric Clinic Men, does the reliability drop conspicuously below 90 per cent. This drop (to So per cent) seems due in part to certain intrinsic ambiguities in the responses of this group, and in part to the fact that this was the first group to be scored blindly.
The reliabilities for the individual items are also satisfactory, ranging from S3 to 95 per cent. Questions I ("Moods"), 3 ("Great people"), and 5 ("Crimes") were scored most consistently. The average agreement on Ques- tions 7 ("Last six months") and S ("Awe-inspiring") might have been higher had they been filled out by Groups 5 and 6. In only 6 cases out of the total of 4I did an item have a reliability of less than So per cent. It would appear, therefore, that the present scoring scheme is relatively reliable, and in this sense "objective. "
2. PROJECTIVE QUESTION SCORES IN RELATION TO STANDING ON THE E SCALE
To what extent are L scores on the Projective Questions characteristic of the anti-ethnocentric individuals, H scores characteristic of the extremely ethnocentric individuals? The Scoring Manual is based on the hypothesis that the low quartiles on the Ethnocentrism scale will give responses falling mainly in the low categories, whereas the responses of the high quartiles will fall mainly in the high categories.
Data bearing on this question are presented in Table 2 (XV), which indi- cates the degree to which the Projective Questions differentiate the ethno- centric from the anti-ethnocentric subjects.
In order to obtain a quantitative measure of the relationship between Pro- jective Question scores and standing on the E scale, the percentage of PQ-E agreement was computed for each quartile. This is called the L% in the case of the low quartile, H% for the high quartile, and %A for the two quartiles combined (average over-all agreement). Let us use L% to illustrate the pro- cedure. Had scores of Hand L been the only ones assigned, the L% would be simply the number of L scores divided by the total number of cases; thus, if all the low quartile members received L scores on a given item there would be I oo per cent agreement between PQ and E. But this procedure was not feasible because of the neutral scores. 7 Since a neutral score represents a half error, i. e. , it signifies less agreement than an L score but more agreement than an H score, the L% was computed according to the following formula:
Sum of L scores plus ~ sum of neutral scores (Na, Nb, Nhl) L%= Nbfb". l ? 1
urn ero su 1ectsm owquartie
7 One possibility would have been not to consider subjects receiving scores of N, and to get a L/H ratio for the remaining subjects. This would have given higher Lo/o values than those obtained by the method finally used, since in the latter method the N scores were used in such a way as to lower the Lo/o. It was believed that all individuals taking the test should be included in the statistical treatment.
? PROJECTIVE QUESTIONS
where Na means "ambiguous," Nb means "blank" (omitted), Nhl means "mixed high and low trends. "
The only exception to this formula is in the case of Nb on Items 1, 2, 4, 7, and 8, where Nb is equivalent to H and is included among the H scores.
The computation of H% is based on an equivalent formula:
Sum of H scores plus ~ sum of neutral scores (Na, Nb, Nhl)
H%= Nbfb. . h. h? 1 urn ero su Jectsm 1g quart1e
Finally, %A is the average of L% and H%, weighted by N (number of cases in each quartile), and it indicates the over-all PQ-E agreement.
Table 2 (XV) gives the L%, H%, and %A for each group tested and for all groups combined. It also gives the number and percentage of Nb, Na, and Nhl responses. It may be noted first that the over-all PQ-E agreement
(%A) is 74? 9 per cent-a value which indicates a statistically significant rela- tionship between Projective Question scores and high vs. low standing on the E scale. On a purely chance basis, the agreement would be only so per cent. The highs tend, however, to be more consistent in their PQ responses than do the lows: the highs have an over-all H% of 8o. 7, while for the lows the L% is only 69. 1. In other words, the lows received more H scores than the highs did L scores. It is, so to speak, easier to make an H than an L score. The reason for this does not appear to lie primarily in the subjects them- selves, for our general impression, based on other results and on clinical judgment, was that the low quartiles fitted the over-all conception of the "democratic" personality at least as well as the highs approximated the pro- totypic "authoritarian" personality. The preponderance of H scores is prob- ably due to the conditions of testing. Since many of the groups were pressed for time, and since the instructions were not emphatic in suggesting that a fairly detailed answer be given, many of the responses were brief and superficial. It will be recalled from the Scoring Manual that references to "vague, dilute emotional experiences," as well as references only to "be- havior or the situation per se, without consideration of inner meanings and motives," were important cues for assigning a score of H. It may be argued that if the instructions emphasized the giving of more detailed answers, both highs and lows would give more responses meriting a score of L. In the present groups, however, the elaborations of most lows took an L direction, of most highs an H direction. It might be expected, then, that the obtaining of longer responses would clarify and increase the differences between the low and high quartiles. It would also probably reduce somewhat the propor- tion (8. 8 per cent) of responses scored Na and Nhl. We should not, of
course, overlook the likelihood that numerous pressures in our culture, and perhaps the predominant ones, tend to make for authoritarianism in the individual. To the extent that this is true, we should expect some H trends even in individuals attempting to achieve a thoroughly democratic orienta- tion.
? 1. Psychology Women Low quartile
Na Score 1 2 3 4 5 6 7 8 Meanb 63
High
quartile
- 83. 6 13 )
d . . . ,
Total
63 %A 91. 3 81. 7 76. 2 76. 2 75. 4 74. 6 72. 2 - 78. 2
Nb 0 4 3 2 6 4 1 -20) >
2. Employment Service Men Low quartile
51
25 L%
>'d t;::l i;l:) Vl 0 z >
High quartile
26 H% 88. 4 94. 2 84. 6 100. 0 100. 0 75. 0 89. 3 78. 6 90. 1 (14) Nb 1 4 1 3 0 1 0 2 12)
Total
51 %A 83. 3 80. 4 80. 4 91. 2 83. 3 77. 1 79. 2 79. 2 82. 6 (24) Nb 5 9 2 5 1 2 0 3 27 )
TABLE 2 (XV)
PERCENTAGE AGREEMENT BETWEEN PROJECI'IVE QUESTION SCORES AND E-SCALE SCORES
Vt 00
32 L%
31 H%
88. 7 82. 3
Na? NHL
>
90. 6 84. 4 79. 7 13. 4 59. 4 60. 9 62. 5 - 73. 0 2 4 7 13 1 4 10 41
. . . , Nb0100330-7) :I:
91. 9 79. 0
Nb 0 3 3 2 3 1 1 - )14. 3%
72. 6 79. 0 91. 9
Na-NHL 3 3 4 7 0 0 . 1 18 . . . .
Na-NHL 5 7 11 20 1 4 11 59 ) 17. 9% z
78. 0 66. 0 76. 0
(10) Nb 4 5 1 2 1 1 0 1 15)
82. 0 66. 0
Na-N'HL 1 1 2 3 0 1 1 0 9)15. 5%
Na-NHL
0 1 5 0 0 0 1 2 9)12. 2%
Na-NHL 1 2 7 3 0 1 2 2 18)13. 8%
80. 0 65. 0 80. 0
73. 9 I:""' . . . .
) 21. 4%
t;::l
0\
:I: 0 i;l:)
. . . ,
> i;l:) . . . .
. . . ,
~
? 82. 2 50. 0 53. 6 46. 4 50. 0 88. 9 58. 1 1021006)
2 0 3 2 1 0 9)16. 3%
83. 3 86. 7 83. 3 66. 7 93. 8 50. 0 77. 7 0101004)
3 0 3 1 1 0 8)13. 1%
82. 8 69. 0 69. 0 56. 8 . 70. 6 70. 6 67. 9
1 1 2 2 0 0 10) I'd
5 0 6
3 2 0
17 )14. 7% l:l:' 0
70. 6 82. 5 55. 9
70. 6 42. 3 88. 5
71. 3
~
10 1002003) c
3 0 3 4 1 1 12)12. 5% M
. . . . . 50. 0 94. 1 97. 1 58. 8 86. 4 81. 8 74. 5 z0 1 0 1 4 1 3 12) (/)
4 0 0 2 1 0 7)17. 0%
60. 3 88. 3 76. 5 64. 7 62. 5 85. 4 72. 8
2 0 1 6 1 3 15)
7 0 3 6 2 1 19)14. 7%
. . . . . . t"1 () . . . , . . . . .
. . . ,
"'
\Jo
00
'-)
?
