Among the
reliabilities
between raters A and B this variable ranked only about fifth best
(76 per cent).
(76 per cent).
Adorno-T-Authoritarian-Personality-Harper-Bros-1950
3 ?
5 4
? Rated Rated
No. of Agreements
T otal Percentage Agreement
81. 4
72. 0
65. 3
73. 7
L by L by Both N by
6 4 20 1
2 2 15 3
1 2 10 7
1 - 11 1
2 5 16 7
12 181
2 22
Percentage Agreement
18 + 4 = 22
25 + 1 = 26 83. 9
-
20 + 1 = 21 42. 5
20+1. 5=21. 5 69. 4
1 15 1
75. 0
1. 9 2. 3 15. 0 3. 0
23 27
21. 1 24. 5
50
45. 5
82. 1 84. 7 87. 1
75. 5 77. 2 78. 8
PSYCHOLOGICAL ILL HEALTH AND POTENTIAL FASCISM 93 I
PSYCHIATRIC CLINIC PATIENTS: MEN AND WOMEN COMBINED (N =59)
One Other
H and L Respectively
H and L Combined
12 + 3 = 15 38. 5
21+2. 5=23. 5 75. 8
16+4=20 43. 5 18+5. 5=23. 5
71. 4 75. 8
27 29
56 96. 4 94. 9 93. 5
48 78. 6
14+5=19 43. 5 67. 9 73. 7 20+4. 5=24. 5 79. 0
244785. 779. 7 23 74. 2
53. 6
? 932
THE AUTHORITARIAN PERSONALITY
there being only 54 per cent agreement. This figure is the result of a rather large number of disagreements between the two raters. The control rater said that she was quite unsure of her ratings because she felt that the inter- view material reflected more the physician's than the patient's attitude toward the complaints. Also, she conceived of "ego-alienness" as something unfavor- able from the mental hygiene point of view and was therefore more often willing to score "absence" of the trait than "presence. " Possibly the most important factor, which lowered both sets of reliabilities, is the relative broadness with which this variable was defined, including few behavioral criteria but leaving much to the rater's intuition.
With the exception of variable III, the ratings for all variables, as well as the over-all estimates, proved to be reliable measures.
5. RELA TIONSHIP BETWEEN RA TINGS AND ETHNOCENTRISM SCORE
The idea of the present rating method is the following: If the "blind" ratings of a given trait correlate significantly with another trait about which the raters had no information (in this case ethnocentrism), it is likely that the two traits are correlated in reality. The relationships to be described below are probably valid because we have eliminated many of the factors inherent in the rating process which could have influenced the results.
There are, however, many factors inherent in the rating method which, can still influence the "blind" ratings in such a way that statistical relation- ships of two traits are artificially raised or lowered. The possibility of arti- ficially high correlations is present particularly when several traits are rated at the same time. Here, the rating of one trait can easily influence the ratings of other traits. This factor (called halo effect) could have influenced the ratings of judges A and B. It was primarily for this reason that control raters, each of whom was trained to pay attention to and to recognize only one of
. the seven single traits, were used.
The ratings of (I) over-all highness and lowness, and (2) of the single
traits, were related to ethnocentrism in the following manner: For a given trait, we counted the number of subjects in the high E quartile who were judged to be "high" by the raters. To these were added half of the high-scor- ing subjects receiving a "neutral" rating. This sum, divided by the total num- ber of high scorers in the group, yielded the percentage of high scorers rated in the high direction. By the same procedure the percentage of low scorers rated in the low direction was obtained. The weighted average of these two percentages was then computed; this indicated the percentage of subjects related "correctly," that is, the percentage whose ratings were re- lated to E score in the expected way.
a. THE OvER-ALL RATINGs. Table I I (XXII) gives the agreements between the subject's standing on E and A's ratings. Rater A's guesses as to highness and lowness agreed with the subject's actual E score 86 per cent of the
? Variable
I. Main emphasis on somatic complaints
II. Intraception III. Ego-alienness
IV. Externalized theory of onset and causes
v. Spontaneous mention of unhappy childhood and family relations
ACtual Standing on E Scale
High Low
High
Low
High
Low
High Low
High Low
Rated Rated H L
Rated N
No. of Agreements
18-1/2 24-1/2
25 24
24-1/2 21
20-1/2 19
27 13-1/2
22-1/2 . 24:. . 112
No. o f Agreements Combined
43
49 45-1/2 39-1/2
40-1/2
47
Percentage Agreement
66. 1 'T9:lY
89. 3 77. 4
87. 5 67. 7
73. 2 61. 3
96. 4 43. 5
80. 4 7 9 . 0
T o t a l Percentage Agreement
72. 9 83. 1 77. 1 66. 9
68. 6 79. 7 84. 7 86. 4
VI. Cues regarding charac- High
13
2 23
1 23
3 26
1
7 3
2 2
0 0
ter structure
VII. Predominant type of symptoms
OV er-all A verages
Low
Hhh Low
High Low
High Low
26 50 92. 9 24 77. 4
25 51 89. 3
TABLE 11 (XXII)
THE AMOUNI' OF AGREEMENT BETWEEN RATER A' S ESTIMATE OF HIGH OR LOW ETHNOCENI'RISM, BASED ON ANALYSIS OF INTAKE INTERVIEWS, AND ETHNOCENTRISM AS MEASURED BY
THEE SCALE
PSYCHIATRIC CLINIC PATIENTS: MEN AND WOMEN COMBINED (N =59)
1567 5233
2422 5224
2323 9202
17 7
27 17
19 5
25 6
25 5
47 14 10
10
26
lJ3. 9
83. 7
09. ~
? I.
II. III. IV. V.
VI.
VII.
Main emphasis on somatic complaints
I n t r a c e p t i o n
E g o ? a l i e n n e s s
E x t e r n a l i z e d
of onset and causes
Spontaneous mention
of unhappy childhood and family relations
C u e s r e g a r d i n g character structure
Predominant type of symptoms
Means
TABLE 12 (XXII)
THE AMOUNT OF AGREEMENT BETWEEN ESTIMATES OF ETHNOCENTRISM, BASED ON RATINGS OF SINGLE VARIABLES FROM INTAKE INTERVIEWS, AND ETHNOCENTRISM AS MEASURED BY THE E SCALE
Variable Composite Ratinga Single V ariables:
Psychiatric Clinic Patients: Men And Women Combined (N =59)
Actual No. No. No. No. of Total Standing Rated Rated Rated No. of Agreements Percentage Percentage
on E Scale H L N Agreements Combined Agreement Agreement fH~i~g~h________~2~2~----~6~----*o________~2~2~. *o________~4~4~-~o________~7~1~-~o________~7~4~-~6---
Low 9 22 0 22. 0 78. 6
~H~i~g~h________~1~5~----~1~2r-----T1________~1~5~-~5________~3~8~? ~o________~5~5~. 4~------~6~4~-~0--- Low 8 22 1 22. 5 72. 6
~H~i~g~h~------~1~9~----~5~----~4~------~2~1r. ~o~------~3~8~-~5~------~7~5~-~o~------~6~5~-~3_ _ _ Low 13 17 1 17. 5 56. 5
;H~i~g~h~------~9~-----T14~----~5~------~1~1~. ? 5_ _ _ _ _ _ _ _ ~3~3~-~o~------~4~lr. ~1_ _ _ _ _ _ _ _ ~5~5~-~9_ _ _ _ Low 6 18 7 21. 5 69. 4
+H~i~g~h--------~2~1~-----i4~----*3_ _ _ _ _ _ _ _ ~2~2~. *5_ _ _ _ _ _ _ _ 24~2~-~o_ _ _ _ _ _ _ _ ~8~0~? ~4_ _ _ _ _ _ _ _ ~7~1~? ~2--- Low 7 15 9 19. 5 62. 9
~H~i~gh~-------72*6______~2~----0~-------72T6~. 0~------~37~. 5~------~9;2~. 9~------~6~3~. 6~-- Low 19 11 1 11. 5 37. 1
~H~i~gh~_ _ _ _ _ _ _ ; 1 ; 2 _ _ _ _ _ _ ~1~1~--~5~------~14~. 5~------~3~6~. 5~------~5T1~. 8~------~6~1~. 9~--
t h e o r y
Low
High Low
High Low
5 18 8 22. 0
22 6 0 22. 0 10 21 0 21. 0
17. 71 7. 7 2. 6 19. 0 9. 71 17. 4 3. 9 19. 4
43. 0
38. 4
71. 0
78. 6 72. 9 67. 7
67. 9 65. 0 62. 5
acomposite rating based on ratings by 7 independent raters on the -variables listed. A "High rating" received point, a "Neutral rating? ~ point; 4 points or more constituted a composite "High" score; everything below 4 a ? ? Low" score.
? PSYCHOLOGICAL ILL HEALTH AND POTENTIAL FASCISM 935
time. For the high scorers alone the agreement was 89 per cent, for the low scorers 84 per cent. These figures show again how closely ethnocentrism is related to personality factors, although the relationships are by no means perfect.
The remaining problem now is: How did raters A and B arrive at their rather accurate guesses about ethnocentrism from a short paragraph of inter- view material dealing almost exclusively with the subjects' complaints? Did they base their guesses mainly on the variables described in the manual, or did they inadvertently use other cues, such as type of language used and other cues that have not been made explicit but which they learned to associate with highness or lowness while interviewing subjects with known ethno- centrism scores?
The results obtained from the control raters, who had had no such previous experience with high- and low-scoring subjects, should help to decide this question. These results are shown in Table I2 (XXII).
As discussed above, a composite "high" or "low" score was computed, based only on the ratings of the seven single variables by the different raters. This composite rating agreed with E score 75per cent of the time, indicating a statistically significant relationship. This figure is I I points lower than the validity figure obtained by rater A. Some of this difference could undoubt- edly have been eliminated by more extensive training of the control raters regarding the concepts and the cues in the material they had to use. Never- theless, in view of the control raters' unfamiliarity with the over-all theory, and their knowledge of only the single variable being rated in each case, their achievement of 75 per cent accuracy takes on added significance.
Prediction of ethnocentrism score (high or low) from clinical material was made more accurate (86 per cent for rater A) when the rater had more train- ing, was acquainted with the concepts and materials of the total study, and could form a picture of the subject by looking for a whole pattern or syn- drome of responses.
b. THE SINGLE VARIABLES. For purposes of this discussion all results for each of the single variables have been summarized in Table I 3 (XXII). This table shows: (I) Percentage agreements between rater A and each control rater; (z) Percentage agreements between raters A and B; (3) Percentage agreements between A's ratings and E score; (4) Percentage agreements be- twen each control rater and E score.
Variable 1: Main Emphasis on Somatic Complaints Rather than on Psy- chological Problems. Among the reliabilities between rater A and the con- trol raters, this variable had the second highest, 8I per cent.
Among the reliabilities between raters A and B this variable ranked only about fifth best
(76 per cent). This relatively low, but still acceptable agreement between A and B was caused not so much by disagreements but by a relatively large number of "neutral" scores (omissions) on the part of B.
? V ariable
I. Main emphasis on somatic complaints
II. Intraception
III. Ego-alienness
IV. Externalized theory of onset and causes
v. Spontaneous mention of unhappy childhood and family relations
VI. Cues regarding character structure
VII. Predominant type of symptoms
A verages
Over-all rating
Actual
on E Scale
Hi~h
Low
High Low
High Low
Hi&! ! Low
Hi! :! h Low
High Low
Hij! h Low
High Low
High I:Ow
Rater A and Control Raters
Raters B and A
Rater A and E Score
Control Raters and ~ E Score ~
TABLE 13 (XXII)
SUMMARY OF DATA F1WM 1l! E RATING OF INTAKE INTERVIEWS
A. RELIABILITY: PERCENTAGE AGREEH:NT AMONG RATERS FUR SEVEN VARIABLES B. VALIDITY: PERCENTAGE AGREEMENT BETWEEN RATINGS AND SalRE ON 1l! E E SCALE
Ps~chiatric Clinic Patients: Men and WOmen Combined (N :;; 59)
"' 0\
A. Reliabilit~ B. Validit;):
Standing Percentage Agreement: Percentage Agreement: Percentage Agreement: Percentage Agreement:
78. 6 83. 9
75. 0 69. 4
53. 6 75. 8
71. :1
75. 8
96. 4 93. 5
67. 9 79. 0
85. 7 74. 2
75. 5 78. 8
82. 1a S7. 1a:
81. 4
72. 0
65. 3
:za. 7
94. 9
73. 7
79. 7
77. 2
84. 7a
82. 1 71. 0
78. 6 83. 9
76. 8 64. 5
:Z:i,O 77. 4
89. 3 91. 9
71. 4 83. 9
80. 3 82. 3
79. 3 79. 7
82. 1 90. 3
76. 3
83. 1
66. 1 79. 0
89. 3 77. 4
72. 9
83. 1
77. 1
fifi l!
68. 6
79. 7
84. 7
76. 1
8! ,! . ~
55. 4 64. 0 72. 6
75. 0 65. 3 56. 5
> C1
apercentage agreement between Rater A's "over-all rating" and a composite score based on 7 independent ratings by control raters. bpercentaae agreement between E score and composite rating.
70. 3 87. 5 67. 7
41. 1 69. 4
! ! lld
62. 9
. . . . . 55. 9 :;2
~
> :Zl 2 z >1;j
:Zfi a
90. 7
7! j. O
81. 4
79. 5
:13. 2 61. 3
96. 4 43. 5
80. ~
79. 0
92. 9 77. 4
83. 7 69. 3
i:'j
~
(J) 0
~
t"' ? 1. ! 1 . . . . .
86. 4 89. 3 83. 9
92. 9 63. 6 37. 1
:! H
71. 0
78. 6 72. 9 67. 7
67. 9 65. 0 62. 5
:Zl? ob :a. ! lb 78. 6?
~
v. ;
t"l
. . . . ,
~
0
~
? . PSYCHOLOGICAL ILL HEALTH AND POTENTIAL FASCISM 937
There was a significant relationship between variable I and ethnocentrism. According to rater A, 66 per cent of the high scorers emphasized their physical complaints, whereas about 79 per cent of the low scorers failed to do so. The control rater's figures are lower: 55 per cent for the high scorers, 73 per cent for the low scorers. In addition, case studies indicated that this variable is important for differentiating subjects high and low on E.
To be sure, there were some cases of low scorers with tendencies toward conversion symptoms or other psychogenic somatic disturbances. But such symptoms, together with marked anxious concern about bodily integrity was characteristic of high scorers. This anxiety is often extended to the function- ing of the nervous system or "mind. " Thus, high-scoring patients complain and have anxious concern about headaches, various sensory disturbances, loss of memory, nervousness, and "going crazy. " There is also a tendency on the part of the high scorers to develop somatic rather than psychological symptoms. Many of these somatic symptoms, on closer examination, turn out to be expressions of repressed affects. Thus, the tendency to develop and to focus on somatic complaints can be considered part of the defensive activity of the high scorer's narrow ego, which shuts out extensive parts of the in- dividual's inner life and, as an additional defensive measure, causes rejection of any thinking in psychological terms and, instead, an emphasis on thinking in terms of physical causation. Thus, variable I may be an expression of the same processes which underlie variable II, and, in a sense, all the other variables as well.
Variable II: lntraception. This variable had the highest reliability with raters A and B (83 per cent). Among the agreements of A with the control raters, this variable ranked only sixth (72 per cent). As noted above, a differ- ence in training between raters A and B on the one hand and the control rater on the other is probably the cause of the difference in the reliabilities of the two sets of ratings. The fact that A's and B's ratings agreed more closely with E score (83 per cent) than did those of the control rater (65 per cent) is probably to be explained in the same way.
From these data it appears that (r) adequately trained raters can arrive at very reliable ratings of intraception, using patients' statements about their complaints in a first psychiatric interview; (2) intraception is highly corre- lated with lack of ethnocentrism. The latter proposition is supported by a great deal of evidence from other material presented in this volume, in con- nection with the F scale, the Projective Questions, and particularly the Thematic Apperception Test and the interviews.
In the Clinic the difference between high and low scorers on intraception became very clear when any kind of psychotherapy was attempted. Some of the high-scoring subjects whom we interviewed were almost unable to accept the notion of psychological causation of their disturbances, and it took a great deal of time to make them see some very obvious connections
? THE AUTHORITARIAN PERSONALITY
between their symptoms, on the one hand, and some anxiety-producing factors in their life situation and events in the past, on the other. The low scorers either knew these more obvious connections before coming to the Clinic (often reporting about their inner and outer lives with a great deal of awareness of their own and other's psychological processes) or were quick in grasping the therapist's interpretations. Many of these latter patients, at least at first sight, appear to be especially good subjects for psychotherapy. They are cooperative, perceptive, and give excellent histories. But often it is difficult to effect changes in their symptoms because of their characteristic defenses: isolation of affect and intellectualization. It is as if they "can afford" to know more about their inner lives because, among other things, their egos, used to admitting impulses, have developed certain intellectual ways of dealing with drives and emotions.
Variable Ill: Ego-alienness. This variable had the lowest reliabilities. The agreement between A and the control rater was only 65 per cent; the agreement between A and B was 70 per cent. The control rater, as noted above, was quite unsure of her ratings and expressed misgivings about the way in which the variable had been defined. Indeed, it seems likely that the breadth of the category and the absence from it of behavioral criteria lowered the reliability of both sets of ratings. Thus it happened that the control rater tried mainly to judge ego-alienness from the degree of conscious acceptance or rejection of the symptoms as revealed by the interview. Raters A and B also included in their judgments the nature of the symptoms themselves, regardless of the patient's expressed attitude toward them. Thus they judged the presence of predominantly psychosomatic symptoms, or of vague anxiety without content, as more ego-alien than conscious conflicts or feelings of failure.
As was to be expected, the control rater's judgment did not agree very? well with E score (56 per cent). Rater A's ratings, however, showed a fairly high relationship (77 per cent). Examination of the data revealed that some of the low-scoring patients, who on the basis of this variable were judged to be ethnocentric, showed psychotic manifestations. Such manifestations actually have much more ego-alien quality than the neurotic symptoms which generally predominated in our group. The variable probably works better for the high- than for the low-scoring group. ?
Variable IV: Externalized Theory of Onset and Causes of the Ill- ness. The reliabilities here are quite good-74 per cent for A and the control rater, and 76 per cent for A and B. Rater A's agreement with E is her lowest
(67 per cent); the control rater agreed more highly withE (71 per cent). In general, the variable seemed to work better for the high scorers. It is possible that this has to do with the fact that more "neutrals" were scored for this category than for any other, and there were a few more "neutrals" in the low-scoring group. The large number of neutral ratings seemed to be
?
5 4
? Rated Rated
No. of Agreements
T otal Percentage Agreement
81. 4
72. 0
65. 3
73. 7
L by L by Both N by
6 4 20 1
2 2 15 3
1 2 10 7
1 - 11 1
2 5 16 7
12 181
2 22
Percentage Agreement
18 + 4 = 22
25 + 1 = 26 83. 9
-
20 + 1 = 21 42. 5
20+1. 5=21. 5 69. 4
1 15 1
75. 0
1. 9 2. 3 15. 0 3. 0
23 27
21. 1 24. 5
50
45. 5
82. 1 84. 7 87. 1
75. 5 77. 2 78. 8
PSYCHOLOGICAL ILL HEALTH AND POTENTIAL FASCISM 93 I
PSYCHIATRIC CLINIC PATIENTS: MEN AND WOMEN COMBINED (N =59)
One Other
H and L Respectively
H and L Combined
12 + 3 = 15 38. 5
21+2. 5=23. 5 75. 8
16+4=20 43. 5 18+5. 5=23. 5
71. 4 75. 8
27 29
56 96. 4 94. 9 93. 5
48 78. 6
14+5=19 43. 5 67. 9 73. 7 20+4. 5=24. 5 79. 0
244785. 779. 7 23 74. 2
53. 6
? 932
THE AUTHORITARIAN PERSONALITY
there being only 54 per cent agreement. This figure is the result of a rather large number of disagreements between the two raters. The control rater said that she was quite unsure of her ratings because she felt that the inter- view material reflected more the physician's than the patient's attitude toward the complaints. Also, she conceived of "ego-alienness" as something unfavor- able from the mental hygiene point of view and was therefore more often willing to score "absence" of the trait than "presence. " Possibly the most important factor, which lowered both sets of reliabilities, is the relative broadness with which this variable was defined, including few behavioral criteria but leaving much to the rater's intuition.
With the exception of variable III, the ratings for all variables, as well as the over-all estimates, proved to be reliable measures.
5. RELA TIONSHIP BETWEEN RA TINGS AND ETHNOCENTRISM SCORE
The idea of the present rating method is the following: If the "blind" ratings of a given trait correlate significantly with another trait about which the raters had no information (in this case ethnocentrism), it is likely that the two traits are correlated in reality. The relationships to be described below are probably valid because we have eliminated many of the factors inherent in the rating process which could have influenced the results.
There are, however, many factors inherent in the rating method which, can still influence the "blind" ratings in such a way that statistical relation- ships of two traits are artificially raised or lowered. The possibility of arti- ficially high correlations is present particularly when several traits are rated at the same time. Here, the rating of one trait can easily influence the ratings of other traits. This factor (called halo effect) could have influenced the ratings of judges A and B. It was primarily for this reason that control raters, each of whom was trained to pay attention to and to recognize only one of
. the seven single traits, were used.
The ratings of (I) over-all highness and lowness, and (2) of the single
traits, were related to ethnocentrism in the following manner: For a given trait, we counted the number of subjects in the high E quartile who were judged to be "high" by the raters. To these were added half of the high-scor- ing subjects receiving a "neutral" rating. This sum, divided by the total num- ber of high scorers in the group, yielded the percentage of high scorers rated in the high direction. By the same procedure the percentage of low scorers rated in the low direction was obtained. The weighted average of these two percentages was then computed; this indicated the percentage of subjects related "correctly," that is, the percentage whose ratings were re- lated to E score in the expected way.
a. THE OvER-ALL RATINGs. Table I I (XXII) gives the agreements between the subject's standing on E and A's ratings. Rater A's guesses as to highness and lowness agreed with the subject's actual E score 86 per cent of the
? Variable
I. Main emphasis on somatic complaints
II. Intraception III. Ego-alienness
IV. Externalized theory of onset and causes
v. Spontaneous mention of unhappy childhood and family relations
ACtual Standing on E Scale
High Low
High
Low
High
Low
High Low
High Low
Rated Rated H L
Rated N
No. of Agreements
18-1/2 24-1/2
25 24
24-1/2 21
20-1/2 19
27 13-1/2
22-1/2 . 24:. . 112
No. o f Agreements Combined
43
49 45-1/2 39-1/2
40-1/2
47
Percentage Agreement
66. 1 'T9:lY
89. 3 77. 4
87. 5 67. 7
73. 2 61. 3
96. 4 43. 5
80. 4 7 9 . 0
T o t a l Percentage Agreement
72. 9 83. 1 77. 1 66. 9
68. 6 79. 7 84. 7 86. 4
VI. Cues regarding charac- High
13
2 23
1 23
3 26
1
7 3
2 2
0 0
ter structure
VII. Predominant type of symptoms
OV er-all A verages
Low
Hhh Low
High Low
High Low
26 50 92. 9 24 77. 4
25 51 89. 3
TABLE 11 (XXII)
THE AMOUNI' OF AGREEMENT BETWEEN RATER A' S ESTIMATE OF HIGH OR LOW ETHNOCENI'RISM, BASED ON ANALYSIS OF INTAKE INTERVIEWS, AND ETHNOCENTRISM AS MEASURED BY
THEE SCALE
PSYCHIATRIC CLINIC PATIENTS: MEN AND WOMEN COMBINED (N =59)
1567 5233
2422 5224
2323 9202
17 7
27 17
19 5
25 6
25 5
47 14 10
10
26
lJ3. 9
83. 7
09. ~
? I.
II. III. IV. V.
VI.
VII.
Main emphasis on somatic complaints
I n t r a c e p t i o n
E g o ? a l i e n n e s s
E x t e r n a l i z e d
of onset and causes
Spontaneous mention
of unhappy childhood and family relations
C u e s r e g a r d i n g character structure
Predominant type of symptoms
Means
TABLE 12 (XXII)
THE AMOUNT OF AGREEMENT BETWEEN ESTIMATES OF ETHNOCENTRISM, BASED ON RATINGS OF SINGLE VARIABLES FROM INTAKE INTERVIEWS, AND ETHNOCENTRISM AS MEASURED BY THE E SCALE
Variable Composite Ratinga Single V ariables:
Psychiatric Clinic Patients: Men And Women Combined (N =59)
Actual No. No. No. No. of Total Standing Rated Rated Rated No. of Agreements Percentage Percentage
on E Scale H L N Agreements Combined Agreement Agreement fH~i~g~h________~2~2~----~6~----*o________~2~2~. *o________~4~4~-~o________~7~1~-~o________~7~4~-~6---
Low 9 22 0 22. 0 78. 6
~H~i~g~h________~1~5~----~1~2r-----T1________~1~5~-~5________~3~8~? ~o________~5~5~. 4~------~6~4~-~0--- Low 8 22 1 22. 5 72. 6
~H~i~g~h~------~1~9~----~5~----~4~------~2~1r. ~o~------~3~8~-~5~------~7~5~-~o~------~6~5~-~3_ _ _ Low 13 17 1 17. 5 56. 5
;H~i~g~h~------~9~-----T14~----~5~------~1~1~. ? 5_ _ _ _ _ _ _ _ ~3~3~-~o~------~4~lr. ~1_ _ _ _ _ _ _ _ ~5~5~-~9_ _ _ _ Low 6 18 7 21. 5 69. 4
+H~i~g~h--------~2~1~-----i4~----*3_ _ _ _ _ _ _ _ ~2~2~. *5_ _ _ _ _ _ _ _ 24~2~-~o_ _ _ _ _ _ _ _ ~8~0~? ~4_ _ _ _ _ _ _ _ ~7~1~? ~2--- Low 7 15 9 19. 5 62. 9
~H~i~gh~-------72*6______~2~----0~-------72T6~. 0~------~37~. 5~------~9;2~. 9~------~6~3~. 6~-- Low 19 11 1 11. 5 37. 1
~H~i~gh~_ _ _ _ _ _ _ ; 1 ; 2 _ _ _ _ _ _ ~1~1~--~5~------~14~. 5~------~3~6~. 5~------~5T1~. 8~------~6~1~. 9~--
t h e o r y
Low
High Low
High Low
5 18 8 22. 0
22 6 0 22. 0 10 21 0 21. 0
17. 71 7. 7 2. 6 19. 0 9. 71 17. 4 3. 9 19. 4
43. 0
38. 4
71. 0
78. 6 72. 9 67. 7
67. 9 65. 0 62. 5
acomposite rating based on ratings by 7 independent raters on the -variables listed. A "High rating" received point, a "Neutral rating? ~ point; 4 points or more constituted a composite "High" score; everything below 4 a ? ? Low" score.
? PSYCHOLOGICAL ILL HEALTH AND POTENTIAL FASCISM 935
time. For the high scorers alone the agreement was 89 per cent, for the low scorers 84 per cent. These figures show again how closely ethnocentrism is related to personality factors, although the relationships are by no means perfect.
The remaining problem now is: How did raters A and B arrive at their rather accurate guesses about ethnocentrism from a short paragraph of inter- view material dealing almost exclusively with the subjects' complaints? Did they base their guesses mainly on the variables described in the manual, or did they inadvertently use other cues, such as type of language used and other cues that have not been made explicit but which they learned to associate with highness or lowness while interviewing subjects with known ethno- centrism scores?
The results obtained from the control raters, who had had no such previous experience with high- and low-scoring subjects, should help to decide this question. These results are shown in Table I2 (XXII).
As discussed above, a composite "high" or "low" score was computed, based only on the ratings of the seven single variables by the different raters. This composite rating agreed with E score 75per cent of the time, indicating a statistically significant relationship. This figure is I I points lower than the validity figure obtained by rater A. Some of this difference could undoubt- edly have been eliminated by more extensive training of the control raters regarding the concepts and the cues in the material they had to use. Never- theless, in view of the control raters' unfamiliarity with the over-all theory, and their knowledge of only the single variable being rated in each case, their achievement of 75 per cent accuracy takes on added significance.
Prediction of ethnocentrism score (high or low) from clinical material was made more accurate (86 per cent for rater A) when the rater had more train- ing, was acquainted with the concepts and materials of the total study, and could form a picture of the subject by looking for a whole pattern or syn- drome of responses.
b. THE SINGLE VARIABLES. For purposes of this discussion all results for each of the single variables have been summarized in Table I 3 (XXII). This table shows: (I) Percentage agreements between rater A and each control rater; (z) Percentage agreements between raters A and B; (3) Percentage agreements between A's ratings and E score; (4) Percentage agreements be- twen each control rater and E score.
Variable 1: Main Emphasis on Somatic Complaints Rather than on Psy- chological Problems. Among the reliabilities between rater A and the con- trol raters, this variable had the second highest, 8I per cent.
Among the reliabilities between raters A and B this variable ranked only about fifth best
(76 per cent). This relatively low, but still acceptable agreement between A and B was caused not so much by disagreements but by a relatively large number of "neutral" scores (omissions) on the part of B.
? V ariable
I. Main emphasis on somatic complaints
II. Intraception
III. Ego-alienness
IV. Externalized theory of onset and causes
v. Spontaneous mention of unhappy childhood and family relations
VI. Cues regarding character structure
VII. Predominant type of symptoms
A verages
Over-all rating
Actual
on E Scale
Hi~h
Low
High Low
High Low
Hi&! ! Low
Hi! :! h Low
High Low
Hij! h Low
High Low
High I:Ow
Rater A and Control Raters
Raters B and A
Rater A and E Score
Control Raters and ~ E Score ~
TABLE 13 (XXII)
SUMMARY OF DATA F1WM 1l! E RATING OF INTAKE INTERVIEWS
A. RELIABILITY: PERCENTAGE AGREEH:NT AMONG RATERS FUR SEVEN VARIABLES B. VALIDITY: PERCENTAGE AGREEMENT BETWEEN RATINGS AND SalRE ON 1l! E E SCALE
Ps~chiatric Clinic Patients: Men and WOmen Combined (N :;; 59)
"' 0\
A. Reliabilit~ B. Validit;):
Standing Percentage Agreement: Percentage Agreement: Percentage Agreement: Percentage Agreement:
78. 6 83. 9
75. 0 69. 4
53. 6 75. 8
71. :1
75. 8
96. 4 93. 5
67. 9 79. 0
85. 7 74. 2
75. 5 78. 8
82. 1a S7. 1a:
81. 4
72. 0
65. 3
:za. 7
94. 9
73. 7
79. 7
77. 2
84. 7a
82. 1 71. 0
78. 6 83. 9
76. 8 64. 5
:Z:i,O 77. 4
89. 3 91. 9
71. 4 83. 9
80. 3 82. 3
79. 3 79. 7
82. 1 90. 3
76. 3
83. 1
66. 1 79. 0
89. 3 77. 4
72. 9
83. 1
77. 1
fifi l!
68. 6
79. 7
84. 7
76. 1
8! ,! . ~
55. 4 64. 0 72. 6
75. 0 65. 3 56. 5
> C1
apercentage agreement between Rater A's "over-all rating" and a composite score based on 7 independent ratings by control raters. bpercentaae agreement between E score and composite rating.
70. 3 87. 5 67. 7
41. 1 69. 4
! ! lld
62. 9
. . . . . 55. 9 :;2
~
> :Zl 2 z >1;j
:Zfi a
90. 7
7! j. O
81. 4
79. 5
:13. 2 61. 3
96. 4 43. 5
80. ~
79. 0
92. 9 77. 4
83. 7 69. 3
i:'j
~
(J) 0
~
t"' ? 1. ! 1 . . . . .
86. 4 89. 3 83. 9
92. 9 63. 6 37. 1
:! H
71. 0
78. 6 72. 9 67. 7
67. 9 65. 0 62. 5
:Zl? ob :a. ! lb 78. 6?
~
v. ;
t"l
. . . . ,
~
0
~
? . PSYCHOLOGICAL ILL HEALTH AND POTENTIAL FASCISM 937
There was a significant relationship between variable I and ethnocentrism. According to rater A, 66 per cent of the high scorers emphasized their physical complaints, whereas about 79 per cent of the low scorers failed to do so. The control rater's figures are lower: 55 per cent for the high scorers, 73 per cent for the low scorers. In addition, case studies indicated that this variable is important for differentiating subjects high and low on E.
To be sure, there were some cases of low scorers with tendencies toward conversion symptoms or other psychogenic somatic disturbances. But such symptoms, together with marked anxious concern about bodily integrity was characteristic of high scorers. This anxiety is often extended to the function- ing of the nervous system or "mind. " Thus, high-scoring patients complain and have anxious concern about headaches, various sensory disturbances, loss of memory, nervousness, and "going crazy. " There is also a tendency on the part of the high scorers to develop somatic rather than psychological symptoms. Many of these somatic symptoms, on closer examination, turn out to be expressions of repressed affects. Thus, the tendency to develop and to focus on somatic complaints can be considered part of the defensive activity of the high scorer's narrow ego, which shuts out extensive parts of the in- dividual's inner life and, as an additional defensive measure, causes rejection of any thinking in psychological terms and, instead, an emphasis on thinking in terms of physical causation. Thus, variable I may be an expression of the same processes which underlie variable II, and, in a sense, all the other variables as well.
Variable II: lntraception. This variable had the highest reliability with raters A and B (83 per cent). Among the agreements of A with the control raters, this variable ranked only sixth (72 per cent). As noted above, a differ- ence in training between raters A and B on the one hand and the control rater on the other is probably the cause of the difference in the reliabilities of the two sets of ratings. The fact that A's and B's ratings agreed more closely with E score (83 per cent) than did those of the control rater (65 per cent) is probably to be explained in the same way.
From these data it appears that (r) adequately trained raters can arrive at very reliable ratings of intraception, using patients' statements about their complaints in a first psychiatric interview; (2) intraception is highly corre- lated with lack of ethnocentrism. The latter proposition is supported by a great deal of evidence from other material presented in this volume, in con- nection with the F scale, the Projective Questions, and particularly the Thematic Apperception Test and the interviews.
In the Clinic the difference between high and low scorers on intraception became very clear when any kind of psychotherapy was attempted. Some of the high-scoring subjects whom we interviewed were almost unable to accept the notion of psychological causation of their disturbances, and it took a great deal of time to make them see some very obvious connections
? THE AUTHORITARIAN PERSONALITY
between their symptoms, on the one hand, and some anxiety-producing factors in their life situation and events in the past, on the other. The low scorers either knew these more obvious connections before coming to the Clinic (often reporting about their inner and outer lives with a great deal of awareness of their own and other's psychological processes) or were quick in grasping the therapist's interpretations. Many of these latter patients, at least at first sight, appear to be especially good subjects for psychotherapy. They are cooperative, perceptive, and give excellent histories. But often it is difficult to effect changes in their symptoms because of their characteristic defenses: isolation of affect and intellectualization. It is as if they "can afford" to know more about their inner lives because, among other things, their egos, used to admitting impulses, have developed certain intellectual ways of dealing with drives and emotions.
Variable Ill: Ego-alienness. This variable had the lowest reliabilities. The agreement between A and the control rater was only 65 per cent; the agreement between A and B was 70 per cent. The control rater, as noted above, was quite unsure of her ratings and expressed misgivings about the way in which the variable had been defined. Indeed, it seems likely that the breadth of the category and the absence from it of behavioral criteria lowered the reliability of both sets of ratings. Thus it happened that the control rater tried mainly to judge ego-alienness from the degree of conscious acceptance or rejection of the symptoms as revealed by the interview. Raters A and B also included in their judgments the nature of the symptoms themselves, regardless of the patient's expressed attitude toward them. Thus they judged the presence of predominantly psychosomatic symptoms, or of vague anxiety without content, as more ego-alien than conscious conflicts or feelings of failure.
As was to be expected, the control rater's judgment did not agree very? well with E score (56 per cent). Rater A's ratings, however, showed a fairly high relationship (77 per cent). Examination of the data revealed that some of the low-scoring patients, who on the basis of this variable were judged to be ethnocentric, showed psychotic manifestations. Such manifestations actually have much more ego-alien quality than the neurotic symptoms which generally predominated in our group. The variable probably works better for the high- than for the low-scoring group. ?
Variable IV: Externalized Theory of Onset and Causes of the Ill- ness. The reliabilities here are quite good-74 per cent for A and the control rater, and 76 per cent for A and B. Rater A's agreement with E is her lowest
(67 per cent); the control rater agreed more highly withE (71 per cent). In general, the variable seemed to work better for the high scorers. It is possible that this has to do with the fact that more "neutrals" were scored for this category than for any other, and there were a few more "neutrals" in the low-scoring group. The large number of neutral ratings seemed to be
?
