Chapter 2 in "The Psychology of Conservatism" edited by G.D. Wilson. London: Academic, 1973
(With a post-publication addendum following the original article)
CONSERVATISM, AUTHORITARIANISM AND RELATED VARIABLES: A Review and Empirical Study*
JOHN J. RAY
University of New South Wales, Australia
B. The existential question
C. Procedures of definition
II. Some "Conceptual" Definitions
A. Conservatism as anti-innovation
B. Structure versus content
C. Conservatism versus reactionism
D. Conservatism and authoritarianism
III. Review of Empirical Evidence
A. Authoritarianism and conservatism
B. Attitudes and personality
C. Rokeach's Dogmatism
D. Eysenck's R and T factors
E. Kerlinger's "criterial referents"
F. Social versus economic conservatism (Lipset)
G. The political conservative (McClosky)
H. Martin and Westie's "Tolerant Personality"
IV. Some New Empirical Evidence
In this chapter the hypothesis that many important social attitudes can be ordered on one bipolar dimension will be investigated. In particular, an attempt will be made to answer the question as to whether there is any ground for distinguishing the construct of "authoritarianism", as measured by the California "F"-Scale, from the traditional notion of "conservatism". This question will be investigated (a) at the conceptual level, (b) in a review of previous empirical studies, and (c) in a new specially designed empirical study.
B. The existential question
The most fundamental question which may be asked in the area would be: "Is there anything there for the words to denote? Do authoritarianism and conservatism actually exist?" Certainly, a vast number of authors seem to believe that they do. Among the many scales that have claimed to tap such a dimension are the "PEC"-Scale of Adorno et al. (1950), the PEC-Scale of Peabody (1961), the conservatism scale of Ekman and Kuennapas (1963), the Tulane Factors of liberalism-conservatism (Kerr, 1955), the conservatism scale of McClosky (1958), the "R" scale of Eysenck (1954) and the PEC-Scale of Anderson and Western (1967). Authoritarianism too has been more frequently scaled than might at first be supposed. There is the Rokeach "D"-Scale (1960), the "D"-Scale of Anderson and Western (1967), the Eysenck "T"-Scale (1944 and 1954), the "F"--Scale of Adorno et al. (1950), the "TFI"-Scale of Levinson and Huffman (1955), the Stereopathy scales of Stern, et al. (1956), the "J" type of Jaensch (1938) and the "A", "AA" and "BF" Scales by Ray (1971, 1972 a, b).
It need not, of course, be claimed that either authoritarianism or conservatism are unidimensional, but it would be expected that a cluster of oblique dimensions would be found that have something in common that justifies use of these names as general rubrics. The Tulane factors of liberalism-conservatism (Kerr, 1955) are five in number, and there is also no shortage of claims that authoritarianism is multidimensional (O'Neill and Levinson, 1954; Camillieri, 1959; Krug, 1961).
C. Procedures of definition
Once a research worker has come to believe that there is some consistency in the expressed attitudes that he wishes to scale, there are two major ways he may proceed. These will be called "conceptual" and "factorial" definition (these words describe operations of defining, not categories of definition). The first is the approach strongly favoured by Christie (1956); the second is identified by him as "typically Eysenckian". Each method may produce more than one consistent solution. For example, factorial definition of intelligence is of several types but few psychologists would want to claim as a consequence that the variable "intelligence" does not exist.
Conceptual definition is of the type employed by Adorno et al. and Rokeach. A construct from some source is developed theoretically, and items are written that have relevance to the construct. The set of items is then scaled by various procedures according to the taste of the authors. The second approach (e.g. Eysenck, 1954) is to use factor analysis on the intercorrelations of a large and hopefully random set of items. The factors that emerge are then "interpreted" in terms of constructs possessed by the researcher.
Both approaches, then, require theorizing in terms of constructs. The difference is that the latter method is ad hoc whereas in the former, hypotheses are subjected to empirical test. While the former may be methodologically preferable, the decision on what items to include in a factor analytic study is often such as to make the two approaches not greatly different. Studies that employ both tend to gain the strengths of both without necessarily adding weaknesses. If several variables are scaled in one factorial study, their relationships may still be compared providing orthogonality was not stipulated for the factor solution.
A point that should not need to be made is that if it is desired to present proof that two variables (e.g. conservatism and authoritarianism) are essentially related, the procedures employed must not be such as to make the finding a necessary artifact. And yet, this was precisely what Adorno et al. did. They required the F- and PEC-Scales to correlate:
"In form 60 the PEC-Scale was shortened to fourteen items and numerous changes made in content and wording ... Two items which worked relatively well in form 78, numbers twenty-seven (Rebellious Ideas) and sixty-one (Security is Bad) were placed in the form 60 F-scale because they seemed on theoretical grounds to fit better there. Several new items have been added. Item thirteen (The American Way) was taken from the form 78 E-scale . . . (p. 163).
Apparently items were freely circulated among the various scales in the study by these authors. An item at one time said to measure authoritarianism is said a little later to measure conservatism, etc. To find that items correlated well with other items said to measure conservatism and then to put those items in the F-Scale makes the subsequent finding of a correlation between the two scales tautological. The average correlation (circa +.5) is in fact surprisingly low -- possibly because of the restricted range in socio-economic status of their respondents.
Eysenck may be said to have made a similar type of error, but in the opposite direction. By stipulating an orthogonal factor solution he required his R and T factors not to correlate. Both authors were apparently interested primarily in demonstrating that their position was one to which certain data can be fitted.
II. SOME "CONCEPTUAL" DEFINITIONS
A definitional resource that seems to have been unjustifiably neglected is the Oxford English Dictionary. Its definitions are exhaustive, its historical scope unrivalled and its quotations both extensive and authoritative. It is on the great range of materials collected in this work under the heading "liberal", "liberalism", "conservative" and "conservatism" that the "conceptual definition" of liberalism-conservatism to be given below is built. ("Liberal-conservative" is preferred to "radical-conservative" because "radical" is sometimes used of any belief strongly held or held in the extreme. Thus we hear of the "radical Right.")
A. Conservatism as anti-innovation
The common element that stands out most in the definitions of the Oxford Dictionary under the headings listed is preference or rejection of innovation and new things generally. A liberal prefers the new and a conservative the old. Thus John Stuart Mill is quoted as saying: "A Liberal is he who looks forward for his principles of government; the Tory looks backward". It is this which explains why the radicalism of today becomes the conservatism of tomorrow. Once that to which the belief refers has become accepted and ceases to be an innovation, it is no longer a proper object for liberal agitation and will eventually come to be defended by conservatives. This explains the usefulness of such test items as: "The highest form of government is democracy and the highest form of democracy is a government run by those who are most intelligent" (from Rokeach's D-Scale). Because democracy has become so normative the conservative must defend it, but knowledge of an early elitist era in government still with us in various forms leads the conservative also to defend elitist ideals. The apparent contradiction is in fact a simple defence of the status quo as we actually have it.
Liberalism is not, however, entirely content-free. Liberal thinkers in fact appear to have guided society's evolution rather than passively accept any innovation or new idea that comes along. Nor does the conservative appear to accept any feature of society that becomes normative. The second highest item on Ekman and Kuennapas' (1963) ratio scale of conservatism is "Only those with formal education in the problems of society should be allowed to vote". This is not normative and does in fact appear to be opposed to a norm.
B. Structure versus content
The distinction that seems needed to resolve the apparent contradiction above might be called the difference between "structure" and "content". This is a different distinction to that used by Rokeach. Structure is used here to denote the most general and lasting feature of an attitude system and "content" the specific themes which may be discussed among people (or tapped by test items) at any one point in time.
The content of liberalism appears to be egalitarianism and desire for personal freedom (after all "liberal" is derived from the Latin "liber" meaning "free", and "liberalis" meaning "pertaining to a free man"). Note the following definitions from the Oxford Dictionary:
4. Free from narrow prejudice; open-minded, candid.
So far, then, we have seen that a liberal attitude has the structure of preferring new things, change or innovation and the content of desiring that change be in an egalitarian and libertarian direction. If a conservative attitude has the structure of preferring the status quo and established ways of doing things, what is its content? Is it simply opposition to liberal ideas and rejection of libertarianism, egalitarianism, etc? Kerlinger (1967) would predict not. He claims, in fact, that the issues which concern liberals would be matters of indifference to conservatives, i.e. that liberalism and conservatism are orthogonal -- which contrasts with the usual view that they are opposites. There is a way to reconcile the opposing claims of Kerlinger and other writers who claim that liberalism and conservatism are negatively related. Perhaps Kerlinger is right because the content of liberal and conservative attitudes is unrelated, and perhaps more conventional theorists are right in that the structure of liberal and conservative attitudes is opposed.
5. Of political opinions; favourable to constitutional changes and legal or administrative reforms tending in the direction of freedom or democracy, Hence used as the designation of the party holding such opinions, in England or other states; opposed to Conservative.
A quote from H. M. Williams is given dated 1801, attributing to conservatives:
"The extinction of every vestige of freedom, and of every liberal idea with which they are associated."
A quote from the Pall Mall Gazette dated 1884 is given which says:
"Conservative and Liberal, as we ordinarily use the terms, are distinctions having reference to a particular practical struggle, the gradual substitution of government by the whole body of the people for government by privileged classes."
McClosky (1958) notes that the content of conservative attitudes is surprisingly clear right throughout the time period he samples. Unlike the Oxford Dictionary, his summary gives most weight to the content of conservative beliefs. It deserves to be quoted in full:
(1) Man is a creature of appetite and will, "governed more by emotion than by reason" (Kirk), in whom "wickedness, unreason, and the urge to violence lurk always behind the curtain of civilized behaviour" (Rossiter). He is a fallen creature, doomed to imperfection, and inclined to license and anarchy.
(2) Society is ruled by "divine intent" (Kirk) and made legitimate by Providence and prescription. Religion "is the foundation of civil society" (Huntington) and is man's ultimate defence against his own evil impulses.
(3) Society is organic, plural, inordinately complex, the product of a long and painful evolution, embodying the accumulated wisdom of previous historical ages. There is a presumption in favour of whatever has survived the ordeal of history, and of any institution that has been tried and found to work.
(4) Man's traditional inheritance is rich, grand, endlessly proliferated and mysterious, deserving of veneration, and not to be cast away lightly in favor of the narrow uniformity preached by "sophisters and calculators (Burke)." Theory is to be distrusted since reason, which gives rise to theory, is a deceptive, shallow, and limited instrument.
(5) Change must therefore be resisted and the injunction heeded that "Unless it is necessary to change it is necessary not to change" (Hearnshaw). Innovation "is a devouring conflagration more often than it is a torch of progress" (Kirk).
(6) Men are naturally unequal, and society requires "orders and classes" for the good of all. All efforts at levelling are futile and lead to despair (Kirk and Rossiter), for they violate the natural hierarchy and frustrate man's "longing for leadership". The superior classes must be allowed to differentiate themselves and to have a hand in the direction of the state, balancing the numerical superiority of the inferior classes.
(7) Order, authority and community are the primary defence against the impulse to violence and anarchy. The superiority of duties over rights and the need to strengthen the stabilizing institutions of society, especially the church, the family, and, above all, private property.
Conservatives, then, believe man is not naturally good, are superstitious, and prefer hierarchical social structures. They think highly of order, authority and duty.
The prominence in conservative beliefs of preference for hierarchical social structures is clearly contradictory to the liberal's preference for equality. In the other respects listed, however, there is no obvious opposition to liberal beliefs. Thus the historical evidence does allow the possibility that there is more orthogonality between the content of liberal and conservative beliefs than might at first be supposed, and Kerlinger's (1967) argument becomes feasible.
C. Conservatism versus reactionism
A liberal attitude, then, is one that states a preference for innovation in an egalitarian or humanitarian direction whereas a conservative thinks we are egalitarian enough already and change might be dangerous. As "progress" is, within limits, normative in our culture, the conservative may believe in a certain (slow) rate of change towards greater humanitarianism, etc., whereas the liberal will want a faster rate of change than is normative. In this context we also have a basis for a distinction between conservatives and "reactionaries". A reactionary wants a negative change back to a former state. Liberalism, conservatism, and reactionism form a clear continuum, then, and it is therefore not surprising to find the reactionary desire for a return to more elitist government in such a high position on the ratio scale of Ekman and Kuennapas (1963): Regrettably, authors of attitude scales have tended to ignore the distinction between reactionary and conservative. It may well turn out that reactionaries have the same traits as conservatives, but this cannot be established until the two are first separated. As the quote given in the Oxford Dictionary dated 1862 says: "Let not one presume to identify conservatism with reaction."
D. Conservatism and authoritarianism
By now it must be clear that the picture of the typical conservative which has so far emerged bears startling resemblances to the "authoritarianism" of Adorno et al. (1950). The "rigidity" of the authoritarian and the "opposition to innovation" of the conservative are wholly identifiable. The stress on the importance of duty and the desire for a hierarchical social structure are also identical. Adorno et al. do not provide us with much idea of what the opposite to authoritarianism might be, but in view of our finding that the liberal believes in egalitarianism and individual liberty, the definition of "authoritarianism" given in the Oxford Dictionary is interesting: "Favourable to the principle of authority as opposed to that of individual freedom." An example dated 1884 is: "A lover of liberty, not an authoritarian." Thus it is not surprising that Adorno et al. felt that the F and PEC-Scales must correlate; it is almost impossible to distinguish the two concepts. The only distinction that appears tenable is to say that "authoritarian" is a rather more particular concept than "conservative" as far as content is concerned.
It is not of course clear that the F-Scale does measure authoritarianism in this classic sense. In spite of the title of their book, Adorno et al. (1950) usually refer to the syndrome tapped by their scale as "pre-fascist". Be this as it may, a great deal of research (Kirscht and Dillehay, 1967) on "the new anthropological type" purportedly discovered by these authors has used the F-Scale as the categorizing instrument and it has become widely accepted that this "type" is properly called "authoritarian". Given the residual doubt about whether the F-Scale does measure authoritarianism in the classic sense, the contention that the "authoritarian" (as familiar to the psychologist) is none other than the old fashioned conservative, must be supported by reference to empirical studies in the literature.
III. REVIEW OF EMPIRICAL EVIDENCE
A. Authoritarianism and conservatism
From the beginning, Adorno et al. (1950) acknowledged that conservatism and "pre-fascism" were inevitably interrelated. The inevitability of this relationship was, however, widely questioned (Shils, 1954). An important reason for dissent was that any radical or "left-wing" regimes on the world scene were also held to display markedly authoritarian characteristics (Smith, 1967). Thus the concept of an "authoritarianism of the left" achieved some currency (Brown, 1964). The most constructive attacks on the claim of an invariant relation between authoritarianism and conservatism have been made by Rokeach (1960) and Eysenck (1954). Other major studies to be considered are: Kerlinger (1967), Lipset (1960), McClosky (1958) and Martin and Westie (1959).
B. Attitudes and personality
Before further work is considered, however, an attempt to differentiate "personality" and "attitude" scales is in order. The work of Adorno et al. (1950) implied that there is such a distinction. Their work, in fact, primarily dealt with a personality "type" -- the "authoritarian." Their F Scale, however, was quite clearly like a conventional attitude scale rather than a personality inventory. As a consequence the distinction between attitude and personality might seem blurred. To confuse the two, however, is to ignore a major assertion of the California work --that in their "pre-fascist" attitude scale they have an indirect instrument which can be used to detect authoritarians only because people with authoritarian personalities characteristically happen to hold pre-fascist attitudes.
If their work implies that there is a distinction between personality and attitude statements it is clearly of some importance to have some definition of the difference. The definition used here is that personality items ask the respondent to rate, describe or assess himself or his own behaviour. An attitude item asks his opinion on something in the world outside himself. The distinction preferred by Eysenck (1954) is that attitudes have particularity, i.e. they have a particular "object" such as "the church", "children", or simply "them", whereas personality statements refer to a general predisposition to behave -- largely regardless of object. Whichever definition is adopted, some such distinction is important to the following consideration of work by both Rokeach (1960) and Eysenck (1954).
C. Rokeach's Dogmatism
Rokeach adopts the view that it is more important to consider the "structure" of attitudes as distinct from "content". By "structure" he appears to mean the rigidity with which attitudes are held. The particular structural feature that Rokeach considers to be evident in those attitudes previously characterized as "authoritarian" he has named "dogmatism" and his D-Scale was designed to tap this. Although Rokeach's D-Scale shows consistently high correlations with the F-Scale (Rokeach, 1956; Peabody, 1966) it has also been shown not to discriminate between communistic respondents and fascistic respondents -- which the F-Scale, of course, was designed to do. Rokeach thus is able to claim that dogmatism is orthogonal to liberalism-conservatism and that the F-Scale is probably a confused measure of both factors. Both the F- and D-Scales however, suffer from similar problems of response set, which could account for much of the correlation between them (Peabody, 1961).
It should be noted that, in spite of his claim to measure structure rather than content, Rokeach does not use items that appear generically different from those of previous authors. The D-Scale, looks very much like the F-Scale or Eysenck's R-Scale. The one major difference is that Rokeach does include some personality items (whichever criterion one adopts). It would in any case be rather difficult to maintain a distinction between a personality variable and Rokeach's "structural" variable. If, as Rokeach says, structure is independent of attitude content, it does not seem to belong to the realm of attitude measurement at all -- but is certainly comprehensible as a part of personality. Some examples of these obvious "personality" items are: "Once I get wound up in a heated discussion I just can't stop", and "I'd like it if I could find someone who would tell me how to solve my personal problems". On the other hand, self reference in many D-Scale items is often disguised, e.g. "It is only natural for a person to be rather fearful of the future" is probably little more than a palatable version of "I am afraid of the future".
For all that, there are sufficient indisputable attitude items in the D-Scale for us to ask whether there is not some purely attitudinal dimension that is orthogonal to liberalism-conservatism. Using their Australian revision of the D-Scale on a large group of students, Anderson and Western (1967) found a correlation with their liberalism scale of .00. An unpublished study by the present author, however, throws some doubt on the invariance of this relationship. In that study a significant interaction was shown with social class on Eysenck R-Scale scores. Rokeach D-Scale scores were in fact highly related to Eysenck R-Scale scores for the lower class respondents but, as in Anderson and Western's finding, the two were unrelated for upper class respondents. Other reports that D-Scale scores and conservatism are related have been made by Direnzo (1968) and Kirtley and Harkness (1969). It is evident that Rokeach's work does not represent clear proof that there are two orthogonal dimensions in general social attitudes.
D. Eysenck's R and T Factors
In spite of Rokeach's attacks on the (prior) work of Eysenck (Rokeach and Hanley, 1956 a, b), the main divergence between the two authors is in methodology rather than theory. From a large pool of "social attitude" statements, Eysenck was able to extract two orthogonal factors which he named "Tough-tender Mindedness" "T" and "Radicalism-Conservatism" "R". He shows that his T-Scale has similar characteristics to the F-Scale in validation studies. The T-Scale is superior to the F-Scale in being fairly well counterbalanced for response bias (ten "tender" and twelve "tough" items). The R-Scale is completely balanced and also extensively validated.
Eysenck claims that the F-Scale is oblique to both his R and T factors. Schematically, then, Eysenck is able to resolve the problem of "left" versus "right" authoritarianism. The Fascist is identified as a tough-minded conservative and the Communist a tough-minded radical. This schematic solution is supported by some empirical data (Eysenck, 1954) and the major political parties of Great Britain also are assigned positions in this two-factor space. Their main difference lies along the R-C continuum with the T dimension of little importance. A difficulty with Eysenck's two factor solution is that no single item comes near to uniquely defining either end of Eysenck's T-factor, i.e. there is not one item loading highly on the T factor that does not also load highly on the R factor. This could be cured by a 45 degree rotation but Eysenck points out that this would deprive the dimensions of much of their explanatory attractiveness.
Eysenck, however, does not claim that his T-factor is a dimension of social attitudes in its own right. In fact, he makes the radical claim that there is only one fundamental dimension of social attitudes, i.e. radicalism-conservatism. He argues (like Rokeach) that the T-factor is just a projection on to the attitude level of a personality variable ("extraversion" in Eysenck's work, "dogmatism" in Rokeach's). This projection takes different forms in the case of radicals and conservatives, which explains why the T factor is not uniquely defined by any one item. Once this point is comprehended, most of the attacks by Eysenck's critics (Rokeach and Hanley, 1956, a, b; Christie, 1956) lose their point. Rather than attack the "purity" of the T-Scale or Eysenck's arithmetic, they might do better to investigate his claim that both "radical-tough" and "conservative-tough" statements correlate with extraversion. Eysenck's work does at least have the virtue of keeping the personality and attitude levels separate (unlike Rokeach), which makes such a test possible in his case. In any event, note that all the attacks made have been on Eysenck's T-Scale. His findings with respect to the R-Scale have not yet been challenged.
In examining Eysenck's claim that there is only one fundamental dimension of social attitudes, "R", it is profitable to inspect the items defining that dimension as given by him (Eysenck, 1954). It contains such items as: "The Japanese are by nature a cruel people", "Coloured people are innately inferior to white people", "War is inherent in human nature" and "Crimes of violence should be punished by flogging". All these items are again strongly reminiscent of the "authoritarian" syndrome described by Adorno et al. (1950). Rather surprisingly, therefore, Eysenck quotes research by Coulter to show that there is no significant relation between the F- and R-Scales. Some possible reasons for this result might be acquiescence bias in the F-Scale, differences in scoring procedures, or peculiarity of the sample (43 Communists, 43 Fascists and 83 British Soldiers).
Eysenck's assertion that there is only one dimension of social attitudes is certainly unusual in the literature. By contrast, Kerr (1955) reports that even conservatism scores in two slightly different areas (political and economic) are unrelated. In their survey of Australian student attitudes, Anderson and Western (1967) find six intercorrelated factors and one orthogonal politico-economic factor. The difference is at least partly accountable for in terms of the different factor-analytic methods popular in Britain and the U. S.A. Eysenck (1954) reports a British analysis of the original F-Scale data from the California study which "showed a strong general factor throughout". By contrast, Camillieri (1959) and Krug (1961) in the U. S.A. had no difficulty in finding six or seven factors in the F-Scale.
E. Kerlinger's "Criterial Referents"
Kerlinger (1967) provides some reconciliation of the opposing camps by taking out second order factors from a range of attitude items. As a result of both theoretical considerations and extensive empirical evidence, he comes out in favour of there being only one fundamental parameter of social attitudes -- which he names liberalism-conservatism. He does, however, take the novel position that liberalism and conservatism are orthogonal! While he agrees that all attitudes can be grouped into the broad categories of "favourable to innovation" versus "preference for established institutions and procedures", it is implicit in his theory of "criterial referents" that endorsement of "conservative" statements is not predictive of rejection of "liberal" statements. This is also a part of Peabody's (1961 and 1966) position and, if true, makes futile the attempt to construct balanced scales.
Thus, rather than deplore the one-way wording of the F-Scale, Kerlinger deplores the attempt to construct any other sort of scale. He takes the extreme position that no attitude variable is naturally bipolar and to introduce bipolarity is to introduce artificiality. Acquiescent response set does remain a problem but Kerlinger would prefer to remove its influence by partial correlation with a separate acquiescence measure. This was the procedure apparently employed by Campbell et al. (1960) in their extensive work with McClosky's conservatism scale. The trouble with this method, of course, is that partial correlation cannot restore discriminating power that is not there. This, perhaps, is one reason why McClosky's scale showed a negligible relation to political party choice.
Methodological considerations aside, the evidence of Kerlinger does indicate two well defined orthogonal second order factors-one containing assertions of liberal ideals and the other containing assertions of conservative ideals. On the other hand, we have Eysenck's analysis showing negative correlations between liberal and conservative items. This discrepancy may be accounted for by the samples or items used, or genuine differences between the U.S.A. and Britain. There is some evidence for differences in attitude structure between Australia and the U.S.A. Anderson and Western (1967) and Kerr (1955) both used student samples and both used non-balanced liberalism scales. Yet Kerr in the U. S.A. found economic and political liberalism to be uncorrelated, whereas Western in Australia found them so highly correlated as to make it impossible to retain separate scales.
F. Social versus Economic Conservatism (Lipset)
Although the pattern that exists in social attitudes is unclear, Kerlinger and Eysenck both seem convinced that a liberal-conservative differential is in some sense fundamental. This position is shared by Lipset (1960) who equates conservatism and authoritarianism as both describing "working class ideology". He presents a wide variety of sociological evidence to show that conservative beliefs and preference for authoritarian social structures are characteristic of the working classes -- having as their common cause intellectual, social and economic deprivation, Lipset finds it important, however, to distinguish liberalism in general from liberalism in the one area that most clearly distinguishes the working classes from "them", i.e. economic liberalism. He produces evidence to show that while conservatism generally increases down the social scale (as do F-Scale scores; Brown, 1964), economic conservatism moves in the opposite direction -- being greatest among those who have most to lose from change, i.e. the upper classes. For Lipset the problem of "left wing authoritarianism" on the international scene does not arise. He regards it as a natural consequence of the proletarian base that has, through revolution, given rise to such regimes. For him, middle class authoritarianism or "fascism" is a more difficult problem. He calls this authoritarianism of the centre, identifies it as less thoroughgoing than authoritarianism of the left, and claims that it is only in exceptionally disorganized and deprived circumstances that middle class people turn to authoritarian political and social institutions. Lipset, then, attributes the "radical" nature of working class political movements solely to economic motives; where these are not involved, the workers are naturally the most conservative.
The fact that many people must have some conflict between their economic interests and their general social attitudes makes it again more understandable why conservatism scales such as McClosky's show little relation to political party choice (Campbell et al. 1960). Parties presumably are less free to be inconsistent than are people. For all that, it is a notable confirmation of Lipset's contention that the lower class based Australian Labour Party was historically the strongest supporter of the White Australia Policy. As the studies to be mentioned subsequently show, ethnocentrism is a reliable correlate of conservatism. (The role of the F-Scale as an indirect measure of ethnocentrism, of course, is well known.)
Eysenck's finding that high scores on the R-scale (predominantly a measure on non-economic radicalism) correlate with left wing party preference is not as damaging to Lipset's position as might at first appear. As Lipset points out, this is an era of "Consensus Politics" and the two major parties that are characteristic of the English speaking world tend to separate only on minor details of emphasis. This being so, it is possible that the discriminating power of the scale was contributed entirely by the "economic radicalism" items.
We now have, incidentally, a perfect triad: Anderson and Western claim political and economic liberalism are positively correlated, Kerr claims they are orthogonal, and Lipset claims they are negatively correlated. As Lipset used evidence from far more normal samples than the other two, and quotes a wide, international range of studies, we might be inclined to favour his interpretation. Almost all his data are sociological however, and the one piece of questionnaire evidence that he quotes for the negative correlation between political and economic conservatism is based on responses to three items of the California PEC-Scale by students at the "American University" in Lebanon. Better psychological data are obviously needed.
G. The political conservative (Mc Closky)
Support for Lipset's equation of conservatism and authoritarianism is, however, forthcoming from the study of conservatism by McClosky (1958), A political scientist, McClosky started out with the very sophisticated and historically based definition of conservatism quoted earlier. Basing himself on materials drawn from writers as diverse in time as Edmund Burke and American "New Conservatives", McClosky was able to write items expressing sentiments which undoubtedly are typical of conservative ideology. The resulting scale, together with several clinical-type scales, was then administered to a large and fairly general sample of the population of an American city. When his high conservative scorers were compared with others, the pattern of traits that appeared characteristic for them might almost have been a transcript of "authoritarian" characteristics from Adorno et al. (1950)-although McClosky appears to have been little influenced by that study. The conservatives were high on: hostility, paranoid tendency, contempt for weakness, ego defense, rigidity, obsessive traits and intolerance of human frailty. Conservative beliefs were most frequent "among the uninformed, the poorly educated, and so far as we can determine, the less intelligent" (p. 35). Identification of conservatism with authoritarianism was of course a central tenet of Adorno et al. (1950). This independent confirmation of their work does do something to rebut the attacks on that assumption. Regrettably, however, McClosky's scale has a response set problem identical to that of the F-Scale. The critics still have room to assert that it is "the acquiescent personality" that is being described. If they choose to do so, however, Rorer (1965) has argued that the burden of proof is at least partly on them.
Given McClosky's results and the original reliable correlation between the F- and PEC-Scale reported by Adorno et al. (1950), it is not really clear how one is to interpret Peabody's (1961) finding that these scales do not correlate at all. He did use balanced versions of the two scales but before one hastens to a response set interpretation of the original results, let it be remembered that the PEC-Scale was partly balanced from the beginning.
H. Martin and Westie's "Tolerant Personality"
The final major independent study in this area is "The Tolerant Personality" by Martin and Westie (1959). Their study is notable in that it seems relatively free of the methodological doubts that have plagued the other studies reported here. They used a sophisticated measure of tolerance in which a "tolerant" response was not simply rejection of an ethnocentric item. Thus neither "response set" nor "criterial referent" (Kerlinger, 1967) criticisms can be levelled against it. Like McClosky, they also administered a large range of other questionnaires to their respondents-differing only in that they were specifically chosen to allow examination of the hypotheses in Adorno et al. (1950). As with McClosky, the pattern found might well have been a transcript from "The Authoritarian Personality". The intolerant were more: nationalistic, intolerant of ambiguity, superstitious, threat-oriented, authoritarian, religious, child-punitive, distrustful of politicians and venerative of their mothers. They were less interested in politics, of lower social class and less educated.
The sample was an almost random one obtained by house to house calls. There was, however, the inevitable limitation of such methods that a large number of the people contacted (two out of three) failed to cooperate fully. It might also be questioned whether the results for one American city could be generalized to other cities, states or countries.
In the field of general social attitude measurement there is some evidence for the pre-eminent importance of one underlying conservatism cum authoritarianism dimension. The studies of Adorno et al. (1950), Lipset (1960), Martin and Westie (1956) and McClosky (1958) point to the identity of authoritarian and conservative attitudes, while the work of Lipset (1950), Eysenck (1954), and Kerlinger (1967) point to this as being the one fundamental dimension that there is. At issue with the first point is the finding of Peabody (1961) that conservatism and authoritarianism are unrelated. At issue with the second, are a number of studies that have found several orthogonal factors in this area (e.g. Ferguson, 1941; Kerr 1955; Anderson and Western, 1967). There is also Kerlinger's unusual qualification that, although conservatism and liberalism are opposed, liberal and conservative tests are unrelated. Lipset too has a qualification about economic conservatism, but this has had no appreciable support from the psychological literature.
IV. SOME NEW EMPIRICAL EVIDENCE
In the present study a deliberate attempt was made to fit a unifactor model to empirical data that is reasonably free of the usual doubts about sample adequacy and representativeness of scale items.
A questionnaire consisting of items from the F-Scale, the Eysenck R-Scale, an ethnocentrism E-Scale and the D-Scale was administered to a group of 474 Australian regular army conscripts. This battery (henceforth known as the F.R.E.D. battery) included the Eysenck R-Scale (sixteen items) and the Australian D-Scale revision (nine items) by Anderson and Western (1967) in their entirety. The E-Scale consisted of the ten "strongest" items of the Beswick and Hills (1969) Australian Ethnocentrism Scale (selected on the basis of the item analyses provided by the authors). From the F-Scale, only those ten items having the most direct reference to authoritarianism in the classical sense were selected. The items expressing superstition, projection, etc., were omitted (cf. Prothro and Melikian, 1953; French and Ernest, 1955). Thus the scales and constructs represented in this study are the ones that have been most discussed in the field of social attitude measurement.
The sample chosen was unusually representative. In Australia, conscripts are chosen randomly on the basis of a birth date procedure from the entire male 20-year-old population. Eligibility is not influenced by area of residence and non-citizens are equally liable with Australian nationals. Students may have their eligibility deferred but may not be exempt. The questionnaire was administered on the first day of the October 1968 intake at No. 3 Training Battalion, Singleton, New South Wales. Perhaps the most useful feature of this sample is the absence of any "volunteer effect" such as occurs in door-to-door samples.
Out of 474 protocols returned, seventy were discarded due to various defects such as partial incompleteness, zigzag response set, etc. The 404 left for analysis were coded for computer processing and tested for single-factoredness by the procedure of Morrison et al. (1967). This program is a modification of the Lawley maximum likelihood procedure which seeks the maximum amount of variance explainable by a single factor and provides a chi square test for the residual. While this program has been especially written to test the sort of model being examined here, it does have two defects common to all maximum likelihood methods: 1. As Nunnally (1967) points out, the maximum likelihood method produces more and more significant factors as N is increased. If N is large enough, the number of significant factors approaches the number of variables. 2. The basic data for the maximum likelihood procedures are principal components -- and principal components, as a method of analysis, is biased towards multiple factor solutions. While there is no easy cure for the latter defect, a rough correction for the first has been tentatively suggested by Law (1967). He proposes that the chi square value obtained be divided by the number of hundreds in the sample. (The examples of the program's use given in the manual by its authors are with a sample where N = 100). In spite of the above limitations, however, this was the best program available for the task.
For any test of single-factoredness an important theoretical requirement is that all scales included should be of approximately equal length. Otherwise at least part of the single-factoredness observed may be said to have been obtained by over-representing one scale. It may be seen that the scales of nine, ten, ten and sixteen items approach this ideal fairly closely, especially when it is realized that the sixteen item R-Scale was expected on the basis of Anderson and Western's (1967) recent Australian study to be resolvable into two eight item scales -- one containing the "political" and "economic" items and the other the "social" and "moral" items.
The Lawley chi square value produced was 3717.84 with d.f. of 945. This corresponds to a Z of 22.7 and a highly significant departure from single-factoredness is at first indicated. If Law's (1967) correction is applied however, chi square is reduced to 929.46 and Z becomes 0.3 which indicates no significant departure from single-factoredness.
The "coefficient alpha" reliability of the entire forty-five items was 0.72. This is comparable with the internal consistency coefficients usually observed in groups of items deliberately written to tap only one concept. The first eight eigenvalues were 5.4, 2.7, 2,3, 1.99, 1.73, 1.51, 1.49 and 1.34. Note that the "natural break" occurs after the first eigenvalue. By this criterion as well then, single-factoredness (unidimensionality) in F.R.E.D. is indicated.
The result of the above study may be seen to confirm our findings from the literature review concerning the appropriateness and usefulness of a single factor account of the social attitude domain. The social attitude items of the F-, R-, E- and D-Scales are all orderable along a single dimension, best called liberalism-conservatism. Neither conceptually nor empirically does there appear to be any ground for distinguishing authoritarianism and conservatism -- except that the former may be regarded as a somewhat more particular case of the latter.
Adorno, T. W., Frenkel-Brunswik, E., Levinson, J. and Sanford, R. N. (1950). "The Authoritarian Personality". Science Editions, Wiley, New York.
Anderson, D. S. and Western, J. S. (1967). An inventory to measure students' attitudes. University of Queensland Papers, Vol. 1 No. 3. University of Queensland Press, St. Lucia, Brisbane.
Beswick, D. G. and Hills, M. D. (1969). An Australian ethnocentrism scale. Aust. Y. Psychol. 21, 211-226.
Block, J. (1965). "The Challenge of Response Sets." Appleton-Century-Crofts. New York.
Brown, R. (1964). "Social Psychology." Free Press, New York.
Camillieri, S. F. (1959). A factor analysis of the F scale. Social forces, 37, 316-323.
Campbell, A., Converse, P. E., Miller, W. E. and Stokes, D. E. (1960). "The American Voter." Wiley, New York.
Christie, R. and Jahoda, Marie. (eds) (1954). "Studies in the Scope and Method of the Authoritarian Personality." Free Press, Glencoe, Illinois.
Christie, R. (1956). Some abuses of psychology. Psychol. Bull. 53, 439-451.
Direnzo, G. J. (1968). Dogmatism and presidential preferences in the 1964 elections. Psychol. Rep. 22, 119?-1202.
Ekman, G. and Kuennapas, T. (1963). Scales of conservatism. Percept. mot. Skills. 16, 329-334.
Eysenck, H, J. (1944). General social attitudes. ,7. Soc. Psychol. 19.
Eysenck, H. J. (1954). "The Psychology of Politics." Routledge, London. Ferguson, L. (1941). The stability of the primary social attitudes: I. Religionism and humanitarianism. J. Psychol. 12, 283-288.
Fishbein, M. (1967). A consideration of beliefs and their role in attitude measurement. In "Readings in Attitude Theory and Measurement". (Fishbein, M., ed.). Wiley, New York.
French, Elizabeth and Ernest, R. R. (1955). The relation between authoritarianism and acceptance of military ideology. J. Pers. 24, 181-191.
Jaensch, E. R. (1938). "Der Gegentypus." Barth, Leipzig.
Kaiser, H. F. (1968). A measure of the average intercorrelation. Educ. psych. Meas. 28, 245-247.
Kerlinger, F. N. (1967). Social attitudes and their criterial referents: A structural theory. Psychol. Rev. 74, 110-122.
Kerr, W. A. (1955). "Tulane Factors of Liberalism-Conservatism (manual)." Psychometric Affiliates, Chicago.
Krug, R. (1961). An analysis of the F scale: I. Item factor analysis. Y. soc. Psych. 53, 285-391.
Kirscht, J. P. and Dillehay, R. C. (1967). "Dimensions of Authoritarianism: A Review of Research and Theory." University of Kentucky Press, Lexington.
Kirtley, D. and Harkness, R. (1969). Some personality and attitude correlates of dogmatism. Psychol. Rep. 24, 851-854.
Law, H. G. (1967). The measurement of theological belief. Unpublished B.A. thesis, University of Queensland.
Levinson, D. J. and Huffman, P. E. (1955). Traditional family ideology and its relation to personality. Y. Pers. 23, 251-278.
Lipset, S. M. (1960). "Political Man." Doubleday, New York.
Martin, J. G. and Westie, F. R. (1959). The tolerant personality. Amer. Sociol. Rev. 24, 521-828.
McClosky, H. (1958). Conservatism and personality. Am. pol. Sci. Rev. 52, 27-45. -
Morrison, D. G., Campbell, D. T. and Wolins, L. A. (1967). Fortran IV program for evaluating internal consistency single-factoredness in sets of multilevel attitude items. Educ. Psychol. Meas., 27, 201.
Nunnally, J. C. (1967). "Psychometric Theory." McGraw-Hill, New York.
O'Neill, W. M. and Levinson, D. J. (1954). A factorial exploration of authoritarianism and some of its ideological correlates. Y. Pers., 22, 449-463.
Peabody, D. (1961). Attitude content and agreement set in scales of authoritarianism, dogmatism, anti-semitism and economic conservatism. J7. abnorm. soc. Psychol., 63, 1-12.
Peabody, D. (1966). Authoritarianism scales and response bias. Psychol. Bull. 65, 11-23.
Prothro, E. T. and Melikian, L. (1953). The California public opinion scale in an authoritarian culture. Public Opinion Quarterly, 17, 353-362.
Ray, J.J. (1971) An "Attitude to Authority" scale. Australian Psychologist, 6, 31-50.
Ray, J.J. (1972a) Non-ethnocentric authoritarianism. Australian & New Zealand Journal of Sociology 8(June), 96-102.
Ray, J.J. (1972b) A new balanced F scale -- And its relation to social class. Australian Psychologist 7, 155-166.
Rokeach, M. (1956). A factorial study of dogmatism and related concepts. J. abnorm. soc. psych. 53, 356-350.
Rokeach, M. and Hanley, C. (1956). Eysenck's tender-mindedness dimension: A critique. Psychol. Bull. 43, 169-176.
Rokeach, M. and Hanley, C. (1956). Care and carelessness in psychology. Psychol. Bull. 53, 183-86(b).
Rokeach, M. (1960). "The Open and Closed Mind." Basic Books, New York.
Rorer, L. G. (1965). The great response-style myth. Psych. Bull. 63, 129-156.
Shils, E. A. (1954). Authoritarianism: Right and left. In "Studies in the Scope and Method of "the Authoritarian Personality"". (Christie, R. and Jahoda, M., eds.). Free Press, Glencoe, Illinois.
Smith, M. B. (1967). Foreword in: Kirscht, J. P. and Dillehay, R. C. "Dimensions of Authoritarianism: A Review of Research and Theory". University of Kentucky Press, Lexington.
Stern, G. G. (1956). Stein, M. I. and Bloom, B. S. "Methods in Personality Assessment." Free Press, Glencoe, Illinois.
*The assistance and co-operation of the Department of the Army in the empirical research reported here is gratefully acknowledged.
Another finding in the literature of unidimensionality in the conservatism/authoritarianism domain is by Suziedelis & Lorr (1973)
The paper above is the published version of my M.A. dissertation, written in 1968 -- 35 years ago compared to when this addendum was written. The information that has become available to me since 1968 would lead me to be much more critical than I was above about the various theories of conservatism that I survey in the paper. In particular, I think it is now clear that opposition to change per se is no part of conservatism. Conservatives do oppose some changes --- what they see as the destructive changes favoured by Leftists -- but also advocate changes of their own -- such as a move to more market-based economic and social arrangements. Anybody who thinks that conservatives do not want any changes to the world as it now is has not talked to many conservatives lately.
The basic empirical finding of the above work -- that all social and political attitudes do group on a single Left-Right dimension -- has however been resoundingly confirmed by later work. See Ray (1982 & 1984a). For an explanation of WHY the attitude domain is one-dimensional, see here and here.
It is however possible to express any view of anything in more or less stupid ways and in more or less implausible ways and Leftist psychologists have -- perhaps unwittingly -- tended to use the less credible expressions of conservatism when they construct scales to measure it. The gain they make from that is, of course, to show that "conservatism" (as they measure it) correlates with a whole variety of discreditable characteristics -- some of which are described in the paper above. The penalty that they pay for their biased approach to the measurement of conservatism, however, is that the scales of Right-wing orientation or conservatism that they construct turn out to be irrelevant to real-life political party choice. It was, for instance, noted above that the McClosky scale of alleged conservatism did not predict vote in the general population and the same has been shown to be true of the F scale (Ray, 1983) and the RWA scale (Ray, 1990). In other words, half of the people that Leftist psychologists define as "conservative" in fact vote for Leftist political parties and candidates! One has to laugh.
By contrast, my own attempts to express both Leftist and Rightist views in a reasonable way have produced scales that DO predict vote -- with Pearsonian correlations between vote and conservatism of attitudes as high as .50 (Ray, 1984a) and multiple Rs as high as .70 (Ray, 1984b).
The biased approach of Leftist psychologists, therefore, has simply condemned their work to irrelevance.
Ray, J.J. (1982) Authoritarianism/libertarianism as the second dimension of social attitudes. Journal of Social Psychology, 117, 33-44.
Ray, J.J. (1983). Half of all authoritarians are Left-wing: A reply to Eysenck and Stone. Political Psychology, 4, 139-144.
Ray, J.J. (1984a) Attitude to abortion, attitude to life and
conservatism in Australia. Sociology & Social Research 68, 236-246.
Ray, J.J. (1984b) Combining demographic and attitude variables to predict vote. Journal of Social Psychology, 122, 145-146.
Ray, J.J. (1990) Book Review: Enemies of freedom by R. Altemeyer. Australian Journal of Psychology, 42, 87-111.
Suziedelis, A. & Lorr, M. (1973) Conservative attitudes and authoritarian values. The Journal of Psychology, 83, 287-294.
Go to Index page for this site
Go to John Ray's "Tongue Tied" blog (Backup here or here)
Go to John Ray's "Dissecting Leftism" blog (Backup here or here)
Go to John Ray's "Australian Politics" blog (Backup here or here)
Go to John Ray's "Gun Watch" blog (Backup here or here)
Go to John Ray's "Education Watch" blog (Backup here or here)
Go to John Ray's "Socialized Medicine" blog (Backup here or here)
Go to John Ray's "Political Correctness Watch" blog (Backup here or here)
Go to John Ray's "Greenie Watch" blog (Backup here or here)
Go to John Ray's "Food & Health Skeptic" blog (Backup here)
Go to John Ray's "Leftists as Elitists" blog (Not now regularly updated -- Backup here)
Go to John Ray's "Marx & Engels in their own words" blog (Not now regularly updated -- Backup here)
Go to John Ray's "A scripture blog" (Not now regularly updated -- Backup here)
Go to John Ray's recipe blog (Not now regularly updated -- Backup here or or here)
Go to John Ray's Main academic menu
Go to Menu of recent writings
Go to John Ray's basic home page
Go to John Ray's pictorial Home Page (Backup here)
Go to Selected pictures from John Ray's blogs (Backup here)
Go to Another picture page (Best with broadband)