Sunday, June 20, 2004

Chapter 41 from: J.J. Ray (Ed.) "Conservatism as Heresy". Sydney: A.N.Z. Book Co., 1974

Are Authoritarians Sick?


Evidence for the view expressed in The Authoritarian Personality (Adorno et al, 1950) that the authoritarian is in some sense 'sick' is far from unequivocal. Incorporating several methodological refinements lacking in earlier studies, the Authoritarianism (F) Scale and the Rokeach Dogmatism Scale were each correlated with a number of clinical indices of psyehopathology. Subjects were 358 day and evening students at the University of Sydney. The results reveal no overall association between authoritarianism and psychopathology, dogmatism and psychopathology or political conservatism and psychopathology, thus presenting disconfirmatory evidence for the hypothesis that authoritarianism is an indicator of psychopathology.

One of the implications of The authoritarian personality (Adorno et al 1950) was that the authoritarian personality type was in some sense 'sick' (see, for example, the concluding chapter to the volume). The projection, rigidity, etc. said to be characteristic of the authoritarian were presented as being psychopathological in degree. They were presented as being mere symptoms of a deep-seated malaise springing from disruptive experiences, particularly with the father, in early childhood. So much so, in fact, that the California authors felt justified, in their concluding chapter, in referring to authoritarianism itself as a 'disease'.

Allowing for the extreme emotional revulsion that these Jewish scholars must have felt for the phenomenon of Nazism --of which they believed authoritarianism to be a precursor-- there does seem to be some grounds in their work for the view that authoritarianism might be related to conventional indices of psychopathology. Unless this is so, it is difficult to see how the terming of authoritarianism as a 'disease' could be anything more than a mere value judgment, particularly since no one has shown that authoritarianism is a sufficient condition for Nazism or Nazi-type behaviour (see Ray, 1971c) or even that it is a necessary condition for this (see Ray, 1971d).

The California work has been so widely questioned as being methodologically inadequate, however, (Christie & Jahoda, 1954) that independent checks on their conclusions are quite essential if those conclusions are to be taken seriously. Some of the checks that have been carried out have been far from encouraging to the hypothesis (e.g. Masling, 1954) but others have purported to present confirming evidence (see Eckhardt & Newcombe, 1969). Although more than twenty years have gone by since the original report, and although the conclusions of that report seem to have passed into popular mythology, the evidence for its conclusions is still far from unequivocal.

In the present paper it is hoped to advance the question somewhat by way of methodological improvements on earlier work. Two of the characteristics of the F scale that have aroused most comment (Christie & Jahoda, 1954) are its one way wording (and the consequent susceptibility to acquiescent response set) and its inbuilt right-wing bias. Some allowance for the first can be made by the use of independent acquiescent indices (Couch & Keniston, 1960; Bass, 1956) and the second can be allowed for by use of Rokeach's (1960) measure of ideology-free authoritarianism --the 'D' or 'Dogmatism' scale. If it is true that the F scale measures acquiescence as well as true authoritarianism, it is at least possible that the correlation of these two variables with psychopathology is not the same. In a society that requires a great deal of conformity and obedience to direction in everyday life, it could be that acquiescence is an adaptive trait or related to an adaptive trait. As such, it might be negatively associated with indices of psychopathology. Thus in cases were acquiescence is operative as a significant influence on F scale scores, it could act as a masking variable on the correlation between authoritarianism and psychopathology. If acquiescence is negatively associated and true authoritarianism is positively associated with psychopathology, the correlation observed with an instrument containing some of both variables might be an overall orthogonality. Thus at least some of the variability in results observed to date might be due to whether or not appreciable amounts of acquiescence were operating on those particular occasions. Acquiescence is presumably fairly sensitive to situational influence and it might be more or less present with particular subjects and on particular occasions. It being something very difficult to control for experimentally with certainty, the only thing we can reasonably do is measure it on each occasion and remove its influence after the event by some statistical method such as the analysis of covariance or partial correlation.

The problem with conservatism as a contaminating variable is also a real one. Eysenck (1954) summarises a great deal of work in his conclusion that, in general, there is no association between conservatism and measures of psychopathology. This being so, to to the extent to which the F scale is responded to simply as a conservatism scale, we would expect again no correlation between it and clinical indices. The existence of the Rokeach 'Dogmatism' scale, however, renders speculation on the matter unnecessary --since the 'D' scale is specifically designed to be an ideology-free measure of authoritarianism.

The following study would then appear to be the first to combine careful control for acquiescence in a study of both the F and D scales as related to independent clinical indices of psychopathology.



A questionnaire was made up to contain the following tests among others: the twenty-eight item California 'F' scale (Adorno, 1950, p. 255); the forty item Rokeach 'Dogmatism' scale (Rokeach, 1960, p. 73); the Bass 'Social Acquiescence' scale (SA), (Bass, 1956); the Couch & Keniston 'Agreement Response Set Scale' (ARS), (Couch & Keniston, 1960); the Maudsley Personality Inventory (twelve item short form) (Eysenck, 1958); and the Gordon Personal Profile (Gordon, 1953).

The use of two scales to measure acquiescence was dictated by findings such as McGee's (1962) that acquiescence measures do not intercorrelate highly --possibly because some or all of them are in fact sensitive to other things than pure acquiescence. Paradoxically, the requirement of writing a group of items with no coherent theme or consistent content appears difficult of achievement. In this situation, the two scales that appear to have had widest acceptance in previous work were both included here.

The Maudsley Personality Inventory and the Gordon Personal Profile provided the indices of psychopathology. Inclusion of the former scale was suggested by the large amount of evidence summarised in Eysenck & Eysenck (1969) to the effect that extraversion (E) and neuroticism (N) (which the Inventory measures) are the two primary or most general dimensions of personality description. Additionally, the scale has had widespread clinical use and extensive validation.

Gordon (1953) considers that his test gives quick, reliable measures of five aspects of personality which are especially significant in the daily functioning of the normal person -- Ascendancy (A), Responsibility (R), Emotional Stability (ES), Sociability (S) and Total (T) or overall self-evaluation. Radcliffe, in reviewing the test (Buros, 1965) concludes that "In all, this is a carefully constructed test which is easily administered and scored. For a personality test, reliabilities are encouraging, especially those for the total score". It was for these reasons, viz. significance of the variables in the functioning of the normal person, ease of administration, and good reliability and validity, that the Inventory was included.


The basic hypothesis here of course, is that authoritarianism is dysfunctional, not in the sense of failing to adjust to a societal norm (there was at no point any suggestion that authoritarian attitudes were contrary to the social norms of post-war California) but rather in the sense of being generally handicapping in any social or non-social situation. Paranoia, neuroticism and projection are obvious clinical examples of such disease entities. Note however, that in actual practice it would be hard to sort out the several sorts of psychopathology. A subject who, in a personality inventory, reports personal feelings of distress, may be doing so, not because he cannot function among people generally but rather because the people among whom he finds himself are ones who, for some reason or other, have preferences for forms of conduct which are contrary to those which are customary for him. Being neurotic may be an effect of social disapproval, not a cause of it.

It is possible then that authoritarianism is not itself a disease or symptomatic of a disease but that the attitudes people have to authoritarianism may alone cause the observed variations in the correlation between authoritarianism and psychopathology indices. Where authoritarianism is disapproved of, authoritarians will tend to become neurotic (or, putting it the other way, only neurotics will remain authoritarian). Where authoritarianism is not disapproved of, there will be no observed relationship with psychopathology. Where authoritarianism is positively approved of, it will be the low authoritarians who show up as more poorly adjusted.

There is in fact already some strong evidence for the latter contention in the work of Orpen (1972)-- who found an extremely high positive correlation between authoritarianism and good psychological health among white South Africans --i.e. among people who were members of an authoritarian culture.

Therefore although all our present subjects do nominally come from the same culture, it was felt to be a desirable refinement to consider both political allegiance and religious affiliation as possibly defining separate sub-cultural groups. For this reason the analyses below will be split up in these terms.

All subjects were students in the Introductory Psychology course at the University of Sydney- -of whom approximately one third were older evening students. To facilitate the religious sub-group analysis, data was recorded only for students of Roman Catholic, Anglican, Methodist or Presbyterian background, though subjects were not necessarily practicing Christians at the time of testing. The few students of other backgrounds would have fallen into groups too small for meaningful analysis.

Testing was carried out in groups sessions with about thirty subjects at a time. The total testing time was three hours for each subject, subjects being required to come for one hour each week over a period of three weeks. The order of presentation of the tests was counterbalanced.


A total of 353 subjects with denominational background in one of the four categories was obtained. Of these, a total of 292 favoured one of the two major Australian political parties: Liberal and Labor. In Table 1, the correlations for all groups are given separately. It will be noted that there is in fact little difference of relevance to our present concerns between the correlations originating from any of these groups. In no group is there a significant relationship between the two authoritarianism measures and the two main psychopathology indices (Eysenck 'N' score and Gordon 'T' score). Turning to the sub-scales of the Gordon inventory, there was a negative relationship between both indices of authoritarianism and emotional stability for Methodists only. Against this, Authoritarians were less 'ascendant' among Methodists and more 'responsible' among Liberal voters. Among Labor voters, less dogmatic people were ascendant.

To see the full table above, click here

To see the full table above, click here

The correlations with the two acquiescence scales are presented in Table 2. It will be noted that the ARS scale correlates consistently, and in some cases highly, with the Eysenck 'N' scale and the Gordon 'T' score. This is not so for the Social Acquiescence scale. Since an inspection of the items in the ARS scale suggests a strong component of 'sensation seeking' of some sort (e.g. 'novelty has a great appeal to me' and 'I crave excitement') it does appear that the'content-free' ideal seems to have been poorly realised in the ARS scale. Both impressionistically and empirically there is a clear contamination with psychopathological sentiments. For this reason, it is the Social Acquiescence scale upon which principal reliance must be placed here.

The implications this has for our hypothesis is that, if we use the SA scale, acquiescence is not a masking variable and could not hence affect the correlation between authoritarianism and psychopathology. If we accept the ARS at face value on the other hand, acquiescence (being positively correlated with both authoritarianism and psychopathology ) is a masking variable and correction for this would make the correlation between authoritarianism and psyehopathology significantly negative instead of non-significant. This contrasts with the California hypothesis of a positive correlation between authoritarianism and psychopathology.

For instance, if we take the 229 Liberal voters, the correlation between the F and N scales (-.07, NS) becomes -.175 (significant <.05) when ARS score is partialled out. A similar result would hold for the group of sixty-three Labor voters. It is however doubtful that the ARS scale should be used in this way.

A final contrast of relevance concerns the relationship between political conservatism and psychopathology. This can simply be assessed by comparing the mean scores on the psychopathology indices for our two groups of Liberal and Labor voters. Results for this comparison are given in Table 3. On no psychopathology index is there any significant difference.




Agreement Response Set....60.48....12.31............57.46.....11.72......1.74...<.05
Social Acquiescence............25.09....8.81..............22.31....10.04......1.76...<.05
Emotional Stability...................4.11....6.54..............4.68.......6.28.......-.62.....N.S.
Gordon Total..........................12.70...17.12............13.68....17.50......-.40......N.S.

Table 3: Mean, S.D. and t value on each variable, for subjects classified as either Liberal supporters or Labor supporters.


The present evidence represents clear disconfirmation for the hypothesis --widely accepted as a fact --that authoritarians are in some sense mentally 'sick'. Using two personality inventories, two measures of authoritarianism, and two potential controls for acquiescence, no trace of the predicted association could be found. Considering that the evidence in favour of the hypothesis adduced by Eckhardt & Newcombe (1969) was at best indirect --being inferences from higher 'intolerance of ambiguity' scores among the more authoritarian members of a group of peace activists --the present results must be accepted as better evidence on the question.

The isolated association among Methodists between authoritarianism and the Gordon ES scale is probably explicable in terms of the 'social pressure' considerations treated earlier. Sydney Methodists do have a notable record of humanitarian programs and socially radical thinking. A person who, by reason of early training and present religious beliefs feels obliged to accept the Methodist church as the most theologically 'correct' one and who is yet not himself humanitarian or radical in inclination might find himself in a conflictful or stressful situation. This effect is not strong enough, however, to produce an overall association between psychopathology (as measured by the Gordon T score) and authoritarianism. The above interpretation is lent credence by the negative association between ascendance and authoritarianism among Methodists. An authoritarian in Methodist company feels himself obliged to take a back seat. Since the indices are of personality traits rather than of situational responses, we might even say that an authoritarian in a Methodist family is made to feel oppressed. Such a combination produces lasting maladjustment.

A closely similar explanation could apply for the finding that among Labor (radical) voters, dogmatic people were less ascendant. Although it is no news that for various reasons some dogmatic people are attracted to, and may even gain some prominence in, politically radical movements it is equally true that a humanitarian/radical ideology is fundamentally opposed to dogmatism. In general, then, we have the finding that dogmatic people can not be ascendant in radical company.

Yet again, the remaining finding --of a correlation between responsibility and authoritarianism among Liberal (conservative) voters --seem susceptible of a situational explanation. In this company, authoritarian sentiments are normative and an impression of oneself as responsible in such company is probably a true one.

It must be said, however, that all the explanations of findings drawn from the Gordon sub-scales are essentially ad hoc. The point of them is to stress and exemplify the influence that situational factors could have in the apparent presence or absence of psychopathology. Any examination of a relationship between some other variable (such as authoritarianism) and psychopathology must take them into account.

The present work has taken account of such factors and has also employed a variety of other methodological controls, at least some of which appear to have been neglected on most previous occasions. For this reason, it is felt that the results with our two major indices of psychopathology allow us to conclude with some confidence that there is no overall association between authoritarianism and psychopathology, dogmatism and psychopathology or political conservatism and psychopathology.


Note that in Australia, the "Liberal" party is the major conservative party and the "Labor" party is the major Leftist party.

Full citation details for all references used above can be found here.