While these ideas are relatively common today, it would appear that those
individuals
(the high scorers) who take them over most easily are different in their psychological functioning from those who do not.
Adorno-T-Authoritarian-Personality-Harper-Bros-1950
More concretely, can it be expected that people will respond relatively consistently to such varied scale items?
These are questions which must be answered empirically.
The content and generality of anti-Semitic ideology, and the adequacy with which it is measured by the present scale are indicated below by a statistical analysis of scale results.
The validity of the scale will be indicated by correlations of the scale with measures of other, theoretically related, variables, and by analysis of the responses of the two subjects discussed in Chapter II.
C. RESULTS: STATISTICAL ANALYSIS OF THE SCALE
The procedure used for all scales in the present research was to allow six choices of response for each item: slight, moderate, or strong agreement, and the same degrees of disagreement, with no middle or neutral category. Each subject indicated the degree of his agreement by marking +r, +2, or +3, disagreement by -I, -2, or -3.
It seemed likely that three degrees of agreement or disagreement could easily be distinguished by the subjects, and that three degrees gave them the best chance to record clearly felt differences in strength of agreement or disagreement. Certainly the data indicate that all six response categories were used. The "don't know" category has been a source of difficulty and con- troversy in many fields of psychological research ( I2 I). In techniques which permit its use, it tends to be the most frequent choice. Without it, the subject must take a stand one way or the other, although the categories of slight agreement and slight disagreement permit him to be nearly neutral. If a subject is unable to decide, he can, of course, omit the item; but there were never more than 2 to 3 per cent omissions among subjects taking the questionnaire, and never more than I per cent of the group to which it was administered failed to fill it out adequately. Furthermore, the fre- quency with which the "moderate" and "strong" categories were used indi- cates that the items were relatively unambiguous.
The responses were converted into scores by a uniform scoring system. Since higher scores were intended to e~press increasing anti-Semitism, all responses were scored as follows:
? 72
THE AUTHORITARIAN PERSONALITY
-3 =Ipoint -2 =2 points -I =3points
+I=5points +2 =6points +3 =7points
It will be noted that the scoring skips from 3 to 5 points between - I and +1. Four points represented the hypothetical neutral response, and was assigned when the item was omitted. It probably makes little difference statistically that this scheme was used rather than a six-point one in which +I would receive 4 points. This scheme was used mainly because there seemed to be a greater psychological gap between -I and +I responses than between any other two adjacent responses. It was also convenient in marking the omissions.
A person's scale score is simply the sum of his scores on the single items. For the 52 items the scores can range between 52 points (I point on each item, indicating strong opposition to anti-Semitism) and 364 points (7 points on each item, strong anti-Semitism). When the scale score is divided by 52 we obtain the mean score per item; thus, a total score of 78 can also be stated as a score per item of I. 5.
The initial results obtained with the A-S scale have been published else- where (7I). The present discussion will deal with the second administration of the scale; on this occasion the questionnaire administered contained, in addition to the A-S scale, most of the other techniques which were used in subsequent stages of the research. The questionnaire was administered in April, I944, to a class in Introductory Psychology at the University of Cali- fornia. It was given as a routine class activity in two parts, separated by an interval of one week; Part I (Questionnaire A) of the A-S scale was given in the first session, Part II (Questionnaire S) in the second. The class was de- signed for nonmajors in psychology and was rather heterogeneous with re- spect to major subject and year in school.
In view of a possible sex difference, the questionnaires of men and women were separated for statistical purposes. Due to wartime conditions, however, there were fewer than thirty men in the group, so that no statistics on men? were computed. The data presented here are based on the questionnaires of the I44 women subjects, including nineteen members of major minorities: Jews, Negroes, Chinese, and foreign-born. In all subsequent groups the sta- tistical analysis was limited to the questionnaires of native-born, white, non- Jewish Americans.
1. RELIABILITY
The reliability and related statistical properties of the A-S scale and its subscales are presented in Table 7(III). The total-scale reliability of . 92 meets rigorous statistical standards, especially in view of the fact that Part II was administered a week after Part I. (The reliability of the scale on the
? Range
c
TABLE 7 (III)
RELIABILITY OF THE ANTI-SEMITISM SCALE AND ITS SUBSCALES
Total Part Part SubsQale
Property Scale I II so ST SA ss SI
Reliability
Number of i terns Mean (total) c Mean (odd half) Mean (even half)
S. D. (total )c S. D. (odd half) S. D. (even half)
0 92
52 2. 70
2. 74 2. 66
1. 11 1. 21 1. 12
1. 0-5. 5
0 94
26 2. 74
2. 94 2. 54
1. 21 1. 31 1. 15
1. 0-5. 7
0 91
26 2. 66 2. 86 2. 46
1. 12 1. 19 1. 15
1. 0-5. 8
. 84
12 3. 08 3. 52 2. 65
1. 33 1. 55 1. 30
1. 0-6. 5
0 89
10
2. 59 2. 84 2. 34
1. 23 1. 34 1. 32
1. 0-5. 7
0 89
16 2. 47 2. 48 2. 46
1. 18 1. 21 1. 25
1. 0-5. 8
. 71
8
3. 28 3. 00 3. 55
1. 26 1. 35 1. 45
1. 0-6. 3
0 84
8 2. 55 2. 60 2. 50
1. 24 1. 30 1. 35
1. 0-5. 9
. aThe subscales are abbreviated as follows: subscale "Offensive" (S0), "Threatening" (ST). "Attitudes" (SA),
"Seclusive" (Ss>. "Intrusive" (S ) 1
. bThe reliability of the total scale was obtained by correlating scores on Part I (the half administered first) with scores on Part II (in second half of questionnaire). All other reliabilities are based on correlations between the odd items and the even items. The correlations were corrected by the Spearman-Brown formula to ? give the reliability values in the table.
? CThe values of the means, S. O. 's and ranges are given in terms of mean score per item. If multiplied by the number of items in the scale or subscale in question, they are converted into values representing mean per total scale or subscale.
? 74
THE AUTHORITARIAN PERSONALITY
first group studied, as previously published, was . 98. ) The two parts were equated in terms of the subscales, so that an equal number of items from each subscale appeared in each part. Parts I and II are also roughly equivalent in terms of mean and standard deviation. In view of the high correlation be- tween Parts I and II, as well as their equivalence and their high reliabilities (. 94 and . 91), it would appear that either of them alone provides as good a quantitative measure as does the total scale.
It will be noted that the over-all mean is relatively low ( qo. 2 as compared with a theoretical neutral point of 2o8) and that the obtained range includes extremely low scores but does not include the highest possible scores. The item analysis, as will be seen below, suggests the reason for this: despite our attempt to limit the scale to pseudodemocratic statements numerous items were still too openly or crudely prejudiced and had extremely low means
(below 3. 0). The present group of students was, however, less anti-Semitic on the average than the one studied earlier, the latter having a mean of I 58 and a range of 52-303. The distribution of scores in both cases was fairly symmetrical but platykurtic, with very little clustering of scores around the mean.
The reliabilities of the total scale and of the two parts are almost matched by the high reliabilities of the subscales. Reliabilities of . 8 to ? 9 are very satisfactory even for scales three or four times their length.
With regard to reliability, equivalence of halves, and form of distribution, then, it seems safe to conclude that the A-S scale (as well as the subscales) provides an adequate measuring instrument. It ranks the subjects with a rela- tively small error of measurement along a continuum or dimension. That this dimension may be called general anti-Semitism must still be demonstrated by the data on item analysis and validity which follow. No claim is made that the dimension is "pure" or homogeneous. To the extent that the scale is valid, it provides a measure of anti-Semitism in most of its generality and complexity. More specifically, it may be claimed that the higher an indi- vidual's score, the greater his acceptance of anti-Semitic propaganda and the greater his disposition to engage in anti-Semitic accusations and programs of one form or another.
2. INTERCORREL A TIONS OF T H E SUBSCALES
The above reliability data indicate that people are relatively consistent in their responses to the A-S scale and to the individual subscales dealing with relatively specific kinds of imagery and attitudes. Correlations among the subscales are shown in Table 8(III).
Intercorrelations of ? 74 to . 85 are of considerable significance. The fact that they involve subscales dealing with so great a variety of opinions and attitudes is an important source of support for the hypothesis that anti-
? Subscale
"Offensive" "Threatening" "Attitudes" "Intrus ive?
"Threatening" ? Attitudes"
. 85 . 83 . 84
? seclusive" . 75
. 74 . 74
THE STUDY OF ANTI-SEMITIC IDEOLOGY TABLE 8 (III)
INI'ERCORREIATIONSa OF THE A-8 SUBSCALES
75
Total A-S
92 . 93 . 94
aThese are the raw correlation coefficients. If they were corrected for attenuation to give the maximal value theoretically obtainable (with perfectly reliable instruments), they would all be well over . 90.
Semitism is a general frame of mind, a way of viewing Jews and Jewish- Gentile interaction. Imagery of Jews as personally offensive and as socially threatening, attitudes of restriction, exclusion and the like, the view that Jews are too assimilative and yet too clannish-these seem to be various facets of a broad ideological pattern. An individual's stand with regard to one of these issues tends to be very similar in direction and degree to his stand with regard to the other~
The correlations of . 92 to ? 94 between each of the three major subscales and the total anti-Semitism scale are high enough so that knowing an indi- vidual's score on any one subscale permits one to predict with considerable accuracy his score on the total A-S scale. In short, while almost every sub- ject varies somewhat in his responses to the individual items (as will be shown below), almost every subject demonstrates a general degree of support or rejection of anti-Semitism which is relatively consistent from one type of accusation or attitude to another. This is not to say that all the ideas con- tained in the scale are of equal importance emotionally to each anti-Semite. It is more probable-and this view is supported by the interviews-that for each high scorer there are a few central opinions (imagery of Jews as cun- ning, power-seeking, sensual, etc. ) and attitudes of primary importance; but these "pet" ideas seem to provide a basis or general readiness for the ac- ceptance of almost any anti-Semitic idea. The fact that this generality is not complete suggests that various patterns of anti-Semitic ideology may exist and might profitably be studied (as variations within the general framework described here).
The correlation of ? 74 between subscales "Seclusive" and "Intrusive" reveals a deep contradiction in anti-Semitic ideology. As a matter of simple logic, it is impossible for most Jews to be both extremely seclusive and aloof and at the same time too intrusive and prying. This categorical, self-con- tradictory rejection of an entire group is, however, more than a matter of
? THE AUTHORITARIAN PERSONALITY
faulty logic. Viewed psychologically, these results suggest a deep-lying ir- rational hostility directed against a stereotyped image to which individual Jews correspond only partially if at all.
The illogical manner in which the hostility operates is illustrated by a comparison of related items from these two subscales. Thus, "Seclusive" Item I I - 2 0 states that rich Jews help "their own people" but not "American causes. " However, "Intrusive" Item Il-25 takes care of any exceptions: Jews donate money not out of generosity but rather out of desire for prestige and fame. Similarly, either Jews do not take enough interest in community and government (Seclusive), or when they do, they have too much control over national politics (Intrusive). Anti-Semitic hostility leads, then, either to a denial of demonstrable facts (Jewish philanthropy, smallness of number, etc. ) or to an interpretation of them which finds the Jews at fault.
The same self-contradictions and the same implications are evident in the high correlation (. 74) between subscales "Seclusive" and "Attitudes. " It is indeed paradoxical to accuse the Jews of being clannish and aloof, and at the same time to urge that they be segregated and restricted. It would seem, then, that a general hostility and readiness to accept negative imagery are an essential part of the psychological functioning of anti-Semitic individ- uals, who can regard a great variety of specific accusations, often mutually contradictory, as valid.
The reliabilities and subscale intercbrrelations, taken together, permit several. conclusions regarding the nature and inner sources of anti-Semitism. It is a general way of thinking in which hostile attitudes and negative opinions toward Jews predominate. Several patterns of imagery brought out by the subscales seem to be partial facets of a single broad ideological framework.
While these ideas are relatively common today, it would appear that those individuals (the high scorers) who take them over most easily are different in their psychological functioning from those who do not. One major char- acteristic of anti-Semites is a relatively blind hostility which is reflected in the stereotypy, self-contradiction, and destructiveness of their thinking about Jews.
3. INTERNAL CONSISTENCY: STATISTICAL ANALYSIS OF THE INDIVIDUAL ITEMS
A critical reader of the A-S scale may feel that certain items are unsatis- factory in one way or another: that they do not measure what the others measure, that everyone agrees with the ideas expressed, that certain items are too ridiculous to be supported by anyone, and so on. He may like a few items particularly and wonder how successful they were. Or he may be con- cerned with shortening and improving the scale and want a statistical basis for item selection and improvement. For these and other reasons a statistical analysis of the items has? considerable value.
? THE STUDY OF ANTI-SEMITIC IDEOLOGY
77
The problem can be posed in statistical terms as follows. If an item is good, in terms of the total scale, then item scores ought to correlate well with total scale scores. Since few high scorers agree with all items, and since some low scorers agree with several items, a statistical technique is necessary to determine the closeness of the relationship between item score and scale score. The most extensive technique for item analysis is the computing of correlations between item scores and scale scores, especially if some sort of factor analysis is planned. The Likert "Discriminatory Power" technique, al- though statistically more limited, has a great time-saving advantage. Further- more, Murphy and Likert (84), obtaining both Discriminatory Powers and item-total scale correlations for a single scale, found a correlation of ? 9I between these two measures of item value. In other words, the order of goodness of the items, as determined by the Discriminatory Power tech- nique, is practically the same as the order determined by the correlation technique. The Likert technique was therefore used in the present study.
The Discriminatory Power (D. P. ) of each item is obtained by the follow- ing procedure. Subjects whose total scores fall in the highest 2 5 per cent of the distribution are considered high scorers, while those whose scores fall in the lowest 2 5 per cent of the distribution are considered the low scorers.
The means of the high scorers is obtained for each item and found to vary from item to item. Similarly for the low scorers. If an item measures anti- Semitism well, then anti-Semites (high scorers), as determined by the total scale score, will make higher scores on it than will those who are opposed to anti-Semitism (low scorers). The greater the difference between the item mean for the high scorers and that for the low scorers, the greater the Discriminatory Power of that item, and the better the measure of anti- Semitism it gives. A positive D. P. indicates that the item is anti-Semitic, in the sense that anti-Semites as determined by the total scale agree with the item to a greater degree than do unprejudiced subjects. If an item has a negative D. P. , it has apparently been scored in reverse, since low scorers agree with it more than high scorers do. All items in the present scale have positive D. P. 's.
The data on the item analysis of the A-S scale are presented in Table 9 (III). Each item is identified by a key phrase, and the letters 0, T, A, S, and I refer to the subscales Offensive, Threatening, Attitudes, Seclusive, and In-
trusive respectively.
The most important data on each item are the group mean and the
D. P. The group mean reflects the general group tendency toward agree- . ment or disagreement. A mean near 4. 0 indicates that the group was pretty evenly divided pro and con on the issue. Group means between 3. 0 and 5. 0 are likely to involve scores covering well the entire range from I to 7? Means below 3. 0 indicate a strong group tendency toward disagreement,
? 78
No.
1. 2. 3. 4. 5. 6. 7. 8. 9.
10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26.
THE AUTHORITARIAN PERSONALITY TABlE 9 (III)
ANT! -SEMITISM SCAlE: ITEM MEANS AND DISCRIMINATORY POWERS UNIVERSITY OF CALIFORNIA WOMEN
Item
(0: luxurious)
(T: bad inflrenee)
(A: keep Jews out)
(0: conceit)
(S, T: businessmen)
(A: quota)
(0: basically Jewish)
(T: power and control)
(A: suppress Jews)
(0: dirty districts)
(I, T: Washington)
(A: democratic methods)
(0: all alike)
(T: low living standards) (A: wrong to intermarry) (1,0: prying)
(S: Jews exclude Gentiles) (A: fraternities)
(1: no culture)
(S: give up religion)
(A: apartment houses)
(N: never contented)
(S: foreign element)
(A: don't hire Jews)
(I: hide Jewishness)
(N: war with Germany)
Mean:
Part I
Mean H. Q. --L. Q.
4. 44 2. 03 2. 75 1. 11 4. 25 1. 03 4. 50 1. 30 5. 86 1. 38 2. 89 1. 00 5. 78 1. 99 5. 33 2. 30 3. 61 1. 05 2. 94 1. 24 4. 55 1. 24 4. 75 1. 13 5. 50 1. 67 3. 00 1. 24 4. 19 1. 19 3. 89 1. 03 4. 22 2. 11 5. 89 2. 13 4. 86 1. 73 3. 03 1. 30 4. 47 1. 30 5. 42 1. 22 4. 28 1. 38 5. 30 1. 19 4. 33 1. 62 2. 86 1. 05
4. 34 1. 42
D. P.
2. 41 1. 64 3. 22 3. 20 4. 48 1. 89 3. 79 3. 03 2. 56 1. 70 3. 31 3. 62 3. 83 1. 76 3. 00 2. 86 2. 11 3. 76 3. 13 1. 73 3. 17 4. 20 2. 90 4. 11 2. 71 1. 81
2. 92
Mean for Total Group
3. 11 1. 85 2. 30 2. 71 3. 45 1. 67 3. 59 3. 80 1. 84 1. 98 2. 56 2. 76 3. 64 2. 05 2. 57 2. 24 3. 53 3. 84 3. 19 2. 66 2. 52 3. 17 2. 88 2. 84 2. 87 1. 69
2. 74
? No.
1. 2. 3. 4. 5. 6. 7. 8. 9.
C. RESULTS: STATISTICAL ANALYSIS OF THE SCALE
The procedure used for all scales in the present research was to allow six choices of response for each item: slight, moderate, or strong agreement, and the same degrees of disagreement, with no middle or neutral category. Each subject indicated the degree of his agreement by marking +r, +2, or +3, disagreement by -I, -2, or -3.
It seemed likely that three degrees of agreement or disagreement could easily be distinguished by the subjects, and that three degrees gave them the best chance to record clearly felt differences in strength of agreement or disagreement. Certainly the data indicate that all six response categories were used. The "don't know" category has been a source of difficulty and con- troversy in many fields of psychological research ( I2 I). In techniques which permit its use, it tends to be the most frequent choice. Without it, the subject must take a stand one way or the other, although the categories of slight agreement and slight disagreement permit him to be nearly neutral. If a subject is unable to decide, he can, of course, omit the item; but there were never more than 2 to 3 per cent omissions among subjects taking the questionnaire, and never more than I per cent of the group to which it was administered failed to fill it out adequately. Furthermore, the fre- quency with which the "moderate" and "strong" categories were used indi- cates that the items were relatively unambiguous.
The responses were converted into scores by a uniform scoring system. Since higher scores were intended to e~press increasing anti-Semitism, all responses were scored as follows:
? 72
THE AUTHORITARIAN PERSONALITY
-3 =Ipoint -2 =2 points -I =3points
+I=5points +2 =6points +3 =7points
It will be noted that the scoring skips from 3 to 5 points between - I and +1. Four points represented the hypothetical neutral response, and was assigned when the item was omitted. It probably makes little difference statistically that this scheme was used rather than a six-point one in which +I would receive 4 points. This scheme was used mainly because there seemed to be a greater psychological gap between -I and +I responses than between any other two adjacent responses. It was also convenient in marking the omissions.
A person's scale score is simply the sum of his scores on the single items. For the 52 items the scores can range between 52 points (I point on each item, indicating strong opposition to anti-Semitism) and 364 points (7 points on each item, strong anti-Semitism). When the scale score is divided by 52 we obtain the mean score per item; thus, a total score of 78 can also be stated as a score per item of I. 5.
The initial results obtained with the A-S scale have been published else- where (7I). The present discussion will deal with the second administration of the scale; on this occasion the questionnaire administered contained, in addition to the A-S scale, most of the other techniques which were used in subsequent stages of the research. The questionnaire was administered in April, I944, to a class in Introductory Psychology at the University of Cali- fornia. It was given as a routine class activity in two parts, separated by an interval of one week; Part I (Questionnaire A) of the A-S scale was given in the first session, Part II (Questionnaire S) in the second. The class was de- signed for nonmajors in psychology and was rather heterogeneous with re- spect to major subject and year in school.
In view of a possible sex difference, the questionnaires of men and women were separated for statistical purposes. Due to wartime conditions, however, there were fewer than thirty men in the group, so that no statistics on men? were computed. The data presented here are based on the questionnaires of the I44 women subjects, including nineteen members of major minorities: Jews, Negroes, Chinese, and foreign-born. In all subsequent groups the sta- tistical analysis was limited to the questionnaires of native-born, white, non- Jewish Americans.
1. RELIABILITY
The reliability and related statistical properties of the A-S scale and its subscales are presented in Table 7(III). The total-scale reliability of . 92 meets rigorous statistical standards, especially in view of the fact that Part II was administered a week after Part I. (The reliability of the scale on the
? Range
c
TABLE 7 (III)
RELIABILITY OF THE ANTI-SEMITISM SCALE AND ITS SUBSCALES
Total Part Part SubsQale
Property Scale I II so ST SA ss SI
Reliability
Number of i terns Mean (total) c Mean (odd half) Mean (even half)
S. D. (total )c S. D. (odd half) S. D. (even half)
0 92
52 2. 70
2. 74 2. 66
1. 11 1. 21 1. 12
1. 0-5. 5
0 94
26 2. 74
2. 94 2. 54
1. 21 1. 31 1. 15
1. 0-5. 7
0 91
26 2. 66 2. 86 2. 46
1. 12 1. 19 1. 15
1. 0-5. 8
. 84
12 3. 08 3. 52 2. 65
1. 33 1. 55 1. 30
1. 0-6. 5
0 89
10
2. 59 2. 84 2. 34
1. 23 1. 34 1. 32
1. 0-5. 7
0 89
16 2. 47 2. 48 2. 46
1. 18 1. 21 1. 25
1. 0-5. 8
. 71
8
3. 28 3. 00 3. 55
1. 26 1. 35 1. 45
1. 0-6. 3
0 84
8 2. 55 2. 60 2. 50
1. 24 1. 30 1. 35
1. 0-5. 9
. aThe subscales are abbreviated as follows: subscale "Offensive" (S0), "Threatening" (ST). "Attitudes" (SA),
"Seclusive" (Ss>. "Intrusive" (S ) 1
. bThe reliability of the total scale was obtained by correlating scores on Part I (the half administered first) with scores on Part II (in second half of questionnaire). All other reliabilities are based on correlations between the odd items and the even items. The correlations were corrected by the Spearman-Brown formula to ? give the reliability values in the table.
? CThe values of the means, S. O. 's and ranges are given in terms of mean score per item. If multiplied by the number of items in the scale or subscale in question, they are converted into values representing mean per total scale or subscale.
? 74
THE AUTHORITARIAN PERSONALITY
first group studied, as previously published, was . 98. ) The two parts were equated in terms of the subscales, so that an equal number of items from each subscale appeared in each part. Parts I and II are also roughly equivalent in terms of mean and standard deviation. In view of the high correlation be- tween Parts I and II, as well as their equivalence and their high reliabilities (. 94 and . 91), it would appear that either of them alone provides as good a quantitative measure as does the total scale.
It will be noted that the over-all mean is relatively low ( qo. 2 as compared with a theoretical neutral point of 2o8) and that the obtained range includes extremely low scores but does not include the highest possible scores. The item analysis, as will be seen below, suggests the reason for this: despite our attempt to limit the scale to pseudodemocratic statements numerous items were still too openly or crudely prejudiced and had extremely low means
(below 3. 0). The present group of students was, however, less anti-Semitic on the average than the one studied earlier, the latter having a mean of I 58 and a range of 52-303. The distribution of scores in both cases was fairly symmetrical but platykurtic, with very little clustering of scores around the mean.
The reliabilities of the total scale and of the two parts are almost matched by the high reliabilities of the subscales. Reliabilities of . 8 to ? 9 are very satisfactory even for scales three or four times their length.
With regard to reliability, equivalence of halves, and form of distribution, then, it seems safe to conclude that the A-S scale (as well as the subscales) provides an adequate measuring instrument. It ranks the subjects with a rela- tively small error of measurement along a continuum or dimension. That this dimension may be called general anti-Semitism must still be demonstrated by the data on item analysis and validity which follow. No claim is made that the dimension is "pure" or homogeneous. To the extent that the scale is valid, it provides a measure of anti-Semitism in most of its generality and complexity. More specifically, it may be claimed that the higher an indi- vidual's score, the greater his acceptance of anti-Semitic propaganda and the greater his disposition to engage in anti-Semitic accusations and programs of one form or another.
2. INTERCORREL A TIONS OF T H E SUBSCALES
The above reliability data indicate that people are relatively consistent in their responses to the A-S scale and to the individual subscales dealing with relatively specific kinds of imagery and attitudes. Correlations among the subscales are shown in Table 8(III).
Intercorrelations of ? 74 to . 85 are of considerable significance. The fact that they involve subscales dealing with so great a variety of opinions and attitudes is an important source of support for the hypothesis that anti-
? Subscale
"Offensive" "Threatening" "Attitudes" "Intrus ive?
"Threatening" ? Attitudes"
. 85 . 83 . 84
? seclusive" . 75
. 74 . 74
THE STUDY OF ANTI-SEMITIC IDEOLOGY TABLE 8 (III)
INI'ERCORREIATIONSa OF THE A-8 SUBSCALES
75
Total A-S
92 . 93 . 94
aThese are the raw correlation coefficients. If they were corrected for attenuation to give the maximal value theoretically obtainable (with perfectly reliable instruments), they would all be well over . 90.
Semitism is a general frame of mind, a way of viewing Jews and Jewish- Gentile interaction. Imagery of Jews as personally offensive and as socially threatening, attitudes of restriction, exclusion and the like, the view that Jews are too assimilative and yet too clannish-these seem to be various facets of a broad ideological pattern. An individual's stand with regard to one of these issues tends to be very similar in direction and degree to his stand with regard to the other~
The correlations of . 92 to ? 94 between each of the three major subscales and the total anti-Semitism scale are high enough so that knowing an indi- vidual's score on any one subscale permits one to predict with considerable accuracy his score on the total A-S scale. In short, while almost every sub- ject varies somewhat in his responses to the individual items (as will be shown below), almost every subject demonstrates a general degree of support or rejection of anti-Semitism which is relatively consistent from one type of accusation or attitude to another. This is not to say that all the ideas con- tained in the scale are of equal importance emotionally to each anti-Semite. It is more probable-and this view is supported by the interviews-that for each high scorer there are a few central opinions (imagery of Jews as cun- ning, power-seeking, sensual, etc. ) and attitudes of primary importance; but these "pet" ideas seem to provide a basis or general readiness for the ac- ceptance of almost any anti-Semitic idea. The fact that this generality is not complete suggests that various patterns of anti-Semitic ideology may exist and might profitably be studied (as variations within the general framework described here).
The correlation of ? 74 between subscales "Seclusive" and "Intrusive" reveals a deep contradiction in anti-Semitic ideology. As a matter of simple logic, it is impossible for most Jews to be both extremely seclusive and aloof and at the same time too intrusive and prying. This categorical, self-con- tradictory rejection of an entire group is, however, more than a matter of
? THE AUTHORITARIAN PERSONALITY
faulty logic. Viewed psychologically, these results suggest a deep-lying ir- rational hostility directed against a stereotyped image to which individual Jews correspond only partially if at all.
The illogical manner in which the hostility operates is illustrated by a comparison of related items from these two subscales. Thus, "Seclusive" Item I I - 2 0 states that rich Jews help "their own people" but not "American causes. " However, "Intrusive" Item Il-25 takes care of any exceptions: Jews donate money not out of generosity but rather out of desire for prestige and fame. Similarly, either Jews do not take enough interest in community and government (Seclusive), or when they do, they have too much control over national politics (Intrusive). Anti-Semitic hostility leads, then, either to a denial of demonstrable facts (Jewish philanthropy, smallness of number, etc. ) or to an interpretation of them which finds the Jews at fault.
The same self-contradictions and the same implications are evident in the high correlation (. 74) between subscales "Seclusive" and "Attitudes. " It is indeed paradoxical to accuse the Jews of being clannish and aloof, and at the same time to urge that they be segregated and restricted. It would seem, then, that a general hostility and readiness to accept negative imagery are an essential part of the psychological functioning of anti-Semitic individ- uals, who can regard a great variety of specific accusations, often mutually contradictory, as valid.
The reliabilities and subscale intercbrrelations, taken together, permit several. conclusions regarding the nature and inner sources of anti-Semitism. It is a general way of thinking in which hostile attitudes and negative opinions toward Jews predominate. Several patterns of imagery brought out by the subscales seem to be partial facets of a single broad ideological framework.
While these ideas are relatively common today, it would appear that those individuals (the high scorers) who take them over most easily are different in their psychological functioning from those who do not. One major char- acteristic of anti-Semites is a relatively blind hostility which is reflected in the stereotypy, self-contradiction, and destructiveness of their thinking about Jews.
3. INTERNAL CONSISTENCY: STATISTICAL ANALYSIS OF THE INDIVIDUAL ITEMS
A critical reader of the A-S scale may feel that certain items are unsatis- factory in one way or another: that they do not measure what the others measure, that everyone agrees with the ideas expressed, that certain items are too ridiculous to be supported by anyone, and so on. He may like a few items particularly and wonder how successful they were. Or he may be con- cerned with shortening and improving the scale and want a statistical basis for item selection and improvement. For these and other reasons a statistical analysis of the items has? considerable value.
? THE STUDY OF ANTI-SEMITIC IDEOLOGY
77
The problem can be posed in statistical terms as follows. If an item is good, in terms of the total scale, then item scores ought to correlate well with total scale scores. Since few high scorers agree with all items, and since some low scorers agree with several items, a statistical technique is necessary to determine the closeness of the relationship between item score and scale score. The most extensive technique for item analysis is the computing of correlations between item scores and scale scores, especially if some sort of factor analysis is planned. The Likert "Discriminatory Power" technique, al- though statistically more limited, has a great time-saving advantage. Further- more, Murphy and Likert (84), obtaining both Discriminatory Powers and item-total scale correlations for a single scale, found a correlation of ? 9I between these two measures of item value. In other words, the order of goodness of the items, as determined by the Discriminatory Power tech- nique, is practically the same as the order determined by the correlation technique. The Likert technique was therefore used in the present study.
The Discriminatory Power (D. P. ) of each item is obtained by the follow- ing procedure. Subjects whose total scores fall in the highest 2 5 per cent of the distribution are considered high scorers, while those whose scores fall in the lowest 2 5 per cent of the distribution are considered the low scorers.
The means of the high scorers is obtained for each item and found to vary from item to item. Similarly for the low scorers. If an item measures anti- Semitism well, then anti-Semites (high scorers), as determined by the total scale score, will make higher scores on it than will those who are opposed to anti-Semitism (low scorers). The greater the difference between the item mean for the high scorers and that for the low scorers, the greater the Discriminatory Power of that item, and the better the measure of anti- Semitism it gives. A positive D. P. indicates that the item is anti-Semitic, in the sense that anti-Semites as determined by the total scale agree with the item to a greater degree than do unprejudiced subjects. If an item has a negative D. P. , it has apparently been scored in reverse, since low scorers agree with it more than high scorers do. All items in the present scale have positive D. P. 's.
The data on the item analysis of the A-S scale are presented in Table 9 (III). Each item is identified by a key phrase, and the letters 0, T, A, S, and I refer to the subscales Offensive, Threatening, Attitudes, Seclusive, and In-
trusive respectively.
The most important data on each item are the group mean and the
D. P. The group mean reflects the general group tendency toward agree- . ment or disagreement. A mean near 4. 0 indicates that the group was pretty evenly divided pro and con on the issue. Group means between 3. 0 and 5. 0 are likely to involve scores covering well the entire range from I to 7? Means below 3. 0 indicate a strong group tendency toward disagreement,
? 78
No.
1. 2. 3. 4. 5. 6. 7. 8. 9.
10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26.
THE AUTHORITARIAN PERSONALITY TABlE 9 (III)
ANT! -SEMITISM SCAlE: ITEM MEANS AND DISCRIMINATORY POWERS UNIVERSITY OF CALIFORNIA WOMEN
Item
(0: luxurious)
(T: bad inflrenee)
(A: keep Jews out)
(0: conceit)
(S, T: businessmen)
(A: quota)
(0: basically Jewish)
(T: power and control)
(A: suppress Jews)
(0: dirty districts)
(I, T: Washington)
(A: democratic methods)
(0: all alike)
(T: low living standards) (A: wrong to intermarry) (1,0: prying)
(S: Jews exclude Gentiles) (A: fraternities)
(1: no culture)
(S: give up religion)
(A: apartment houses)
(N: never contented)
(S: foreign element)
(A: don't hire Jews)
(I: hide Jewishness)
(N: war with Germany)
Mean:
Part I
Mean H. Q. --L. Q.
4. 44 2. 03 2. 75 1. 11 4. 25 1. 03 4. 50 1. 30 5. 86 1. 38 2. 89 1. 00 5. 78 1. 99 5. 33 2. 30 3. 61 1. 05 2. 94 1. 24 4. 55 1. 24 4. 75 1. 13 5. 50 1. 67 3. 00 1. 24 4. 19 1. 19 3. 89 1. 03 4. 22 2. 11 5. 89 2. 13 4. 86 1. 73 3. 03 1. 30 4. 47 1. 30 5. 42 1. 22 4. 28 1. 38 5. 30 1. 19 4. 33 1. 62 2. 86 1. 05
4. 34 1. 42
D. P.
2. 41 1. 64 3. 22 3. 20 4. 48 1. 89 3. 79 3. 03 2. 56 1. 70 3. 31 3. 62 3. 83 1. 76 3. 00 2. 86 2. 11 3. 76 3. 13 1. 73 3. 17 4. 20 2. 90 4. 11 2. 71 1. 81
2. 92
Mean for Total Group
3. 11 1. 85 2. 30 2. 71 3. 45 1. 67 3. 59 3. 80 1. 84 1. 98 2. 56 2. 76 3. 64 2. 05 2. 57 2. 24 3. 53 3. 84 3. 19 2. 66 2. 52 3. 17 2. 88 2. 84 2. 87 1. 69
2. 74
? No.
1. 2. 3. 4. 5. 6. 7. 8. 9.
