Make your own free website on Tripod.com
**************************************************************************************

British Journal of Political Science 1972, 2, 244-251.

THE MEASUREMENT OF POLITICAL DEFERENCE: Some Australian Data



JOHN RAY

University of New South Wales

The concept of political deference as an explanation for English voting behaviour has been both a popular one and one that has recently come under attack. Writers who have found it useful in explanation include McKenzie and Silver [1] and Nordlinger [2]. Writers who have attacked it include Parkin [3] and Kavanagh [4]. A somewhat intermediate position is taken by Jessop [5].

Much of the controversy has centred around the variety of meanings assigned to the term and the poor operationalizations of any of these meanings in previous empirical research. Both these points are made at length by Kavanagh. Parkin's contribution, on the other hand, centres around a rejection of psychological (more specifically, attitudinal) enquiry in favour of purely sociological (or, institutional) explanation. The work to be reported below does represent one attempt to overcome the sort of problem raised by Kavanagh and it is also felt that, in spite of the undoubted value of Parkin's proposals, an analysis of how people see themselves in relation to the world will always be a necessary complement to analyses of how they are in fact related to the world.

Jessop defines four main senses of deference at the outset of his article. His sense (b) is the one used by Nordlinger and McKenzie and Silver in their widely quoted research on working-class conservatism. This sense of the word he qualifies as: 'Ascriptive sociopolitical deference towards a socially ascribed elite as uniquely qualified for high political office'. Additional to these four senses Jessop introduces later on in his paper a fifth sense of 'meritocratic deference' -- defined as deference towards an intellectual elite rather than the traditional ruling class. He presents data leading to the conclusion that deference in his sense (b) should be treated separately from meritocratic deference. In the present work it is these two senses of deference that will be considered. As was mentioned, both Nordlinger and McKenzie and Silver find political 'deference' to be a most significant element in the explanation of British working class conservative voting. Depending where the cutting-point for 'working class' is made, the proportionate contribution of the working class to the Tory vote is roughly the same (about one-quarter) in Australia and Great Britain [6]. This being so, we have perhaps reason to expect that some sort of social deference might be of relevance to the explanation of voting in Australia also. What follows is an attempt to garner evidence for the existence in Australia of social deference as a consistent disposition to respond.

Political sociologists characteristically class people as 'deferentials' or 'non-deferentials' on the basis of a response to one or two questions. Following a social-psychological line of argument, this is here deemed inadequate. If an open-ended question is used, one has the problem of coder subjectivity (or inter-coder reliability); and if a closed-ended question is used, one does not know whether the inferences we make about the wording of a question are the same as those our respondents make. Psychologists do not normally feel free to speak of a concept as a real variable unless it has been shown that a large number of statements expressing aspects of this concept can objectively be shown to hang together empirically. One of the ways this demonstration is frequently accomplished to is to take a large number of such statements (items) and treat them as a 'scale'. The assumption of so doing is that if all items do tap the one concept, the people who agree with one ought also to tend to agree with the others.

There are two lines of evidence that are normally sought in establishing whether a set of items form a Likert-type scale. These are the reliability coefficient 'alpha' and the correlation of each item with the summed score of all items. The list of item-total correlations provides a means of examining which items detract most from the homogeneity of the total set whereas the reliability is a ratio between the number of items and the actual homogeneity (mean inter-item correlation) of the items [7]. It is normally found that this statistic ('alpha') corresponds very closely to the repeatability of measurements made using the scale. Possible values range between 0.00 and 1.00 but a value between 0.75 and 0.90 is the normal expectation [8].

STUDY I

Using the above methodology, a set of twelve items was selected that closely reflect deference and anti-deference as it is usually conceived (see Items for Scale 1). Because of the absence of a clear Australian equivalent for the British aristocracy, it was thought that deference might be more strongly directed towards leaders as technically competent people. This is reflected in items 1, 2, 3 and 4. Items 5, 6, 7 and 12 reflect considerations of hereditary breeding and items 8, 9, 10 and 11 deal with equalitarianism as such.

Items for Scale 1

1. Most decisions should be left to experts.
2. Only those who are competent on an issue should speak about it.
3. Participation of the people is not necessary if decision-making is left in the hands of a few trusted and competent leaders.
4. It is most important to have the participation of everybody in making decisions, regardless of their knowledge of the issues involved.
5. People from good families who have been to good schools will naturally make better leaders.
6. The country is best run by those who have been brought up for it.
7. Even if given the chance, I would not have the background to help in running the country.
8. There should be no class distinction.
9. Jack is as good as his master.
10. The son of a labourer is just as likely to make a good Prime Minister as the son of a banker or doctor.
11. State schools are every bit as good as private schools.
12. The children of a tradesman are likely to be really no different from the children of a rich grazier.


(The balanced 12-item Deference 'Scale' Items 1, 2, 3, 5, 6 and 8 were scored 5 to 1 and the rest 1 to 5).

As part of a large set of items, these were presented to ninety-six National Servicemen [9]. National Servicemen were chosen because they are randomly selected from the male population of that age. They are superior to a door-to-door sample in that there is no 'volunteer artifact', i.e. all recruits approached did fill out the questionnaire. They had the option of responding to each item from 'strongly agree', 'agree', 'not sure', 'disagree' to 'strongly disagree'. The results of this study are shown in Table 1.

The reliability obtained of 0.43 is far below the level normally acceptable in a group of items to be used as a scale. The question we are most interested in here, however, is: does this result differ significantly from zero? Using Hoyt's [10] approach to the coefficient we may derive it from an analysis of variance of the 'persons' main effect tested against the persons by items interaction. The relationship between this F and our coefficient is alpha = I - I /F. The degrees of freedom are 95 and 1045 (95 x 11). The critical level of F at the 0.01 level of probability is thus 1.38 -- or a reliability of 0.28. Our figure is higher than this and, consequently, significant.

In accordance with normal practice, the three items showing the lowest item-total correlation (and hence detracting most from the homogeneity) were dropped and all statistics recomputed for a 9-item scale. The reliability remained at 0.43. After then dropping the next three weakest items the reliability rose to only 0.45.

TABLE 1

Means, Standard Deviations and Corrected Item-Total Correlations for the 12 Item 'Scale'

Item No.....Mean........S.D....Correlation

1.................2.95........1.18.....0.235
2.................2.93........1.24....0.168
3.................2.34 .......1.06....0.159
4.................3.03........1.07.....0.106
5.................2.08........0.92.....0.279
6.................3.05........1.07.....0.131
7.................3.05........1.14.....0.074
8.................2.50........1.12.....0.061
9.................2.35........1.22.....0.129
10...............1.86........0.93.....0.256
11...............2.46........1.16.....0.269
12...............2.39........1.13.....0.080

TOTAL......31.00........4.95

Reliability Coefficient Alpha = 0.43

Discussion

The implication of the fact that the 'Deference' items did group together significantly yet failed to fulfil the normal requirements for a scale is that deference appears to exist as a real variable but not perhaps in the precise form that we have conceived it. The meaning of the alpha of 0.45 for six items is that were we to readminister the items to the same people after allowing an interval sufficient for forgetting the specific behaviour emitted at first administration, the correlation of scores on the two occasions would be about 0.45 [11]. Thus only about 20 per cent (0.45 squared x 100) of the variability in answers to questions about deference is accountable for by the content of the items. The rest of the variability is due to random or 'error' factors. If this is the case where subjects have six chances to make their position clear, how much more error is introduced where (as in the past) only one question has been asked?

STUDY II

Since we now have reason for believing that deference does exist in Australia, our failure to construct a really usable scale gives rise to the need for a new approach to the task. A new item format in fact seems called for. The items simply expressing a belief that were used in the first study are probably inadequate for tapping what is essentially an unfamiliar concept to many Australians. The disposition to defer may be there but the awareness of deferring is probably slight. This being so, the justification for deferring is probably little discussed and little known. In the new scale to be presented here therefore, the items not only express a preference but also offer reasons in justification of that preference. A typical item is: 'It is best to vote for a man who is already rich, because if he gets in he is less likely to be tempted by opportunities of making money on the sly'. Twenty items in this format were written -- half positively expressed and ten negatively. As a pretest these were administered, together with several other scales, to the fifth and sixth forms of Meadowbank Boys' High School (n for analysis = 110). The items of this version of the scale have already been presented in Ray [12]. The reliability obtained was 0.60 and six items were identified as weak -- five of these being negatively worded. On the basis of this pretest, the weak items were revised and four new items added. The twenty-four item scale was then included in a larger survey being carried out by students in a second-year social psychology course at Macquarie University. Also included in the survey was the Ray [12] 'Attitude to Authority' scale.

In this survey, each student was given three schedules to have filled out -- one subject to come from the 18-25 age range and one from the 40-55-Year range. The third could be from either. The instructions 'choose subjects of as low a social class as possible' were used to counter the usual middle-class bias of samples so obtained. It was expected that this sample could be used not only to standardize the scale but also to give evidence of its validity. The obvious validity check one would make is to show that working-class conservatives (L-CP voters) are more deferential on the whole (get higher scores on the scale) than are working-class radicals (ALP voters).

Results

The final form of the scale (with four weak items discarded) is presented in Items for Scale 2. Its reliability is 0.77 and the mean and standard deviation of the total scores are 50.85 and 9.17 respectively. Of the 20 items, 9 are positively scored and 11 negatively. The correlation between this scale and the 'Attitude to Authority' scale also included in this survey (reliability 0.81) was 0.167 -- which is significant at the < 0.05 level. The correlation between self-assigned class (measured on a 7-point scale) and the Deference Scale was 0.335 -- which is also significant and means that the more people see themselves as upper class the more deferential to social position they are.

Items for Scale 2

1. The son of a rich businessman or grazier has been brought up to take on leadership and responsibility and is therefore the one best fitted to represent the people of his district in Parliament.
2. If a poor man were elected to Parliament he wouldn't know what to do in those high circles.
3. Ruling the country is a task for which only the highly educated are fit.
4. It is best to vote for a man who is already rich because if he gets in he is less likely to be tempted by opportunities for making money on the sly.
5. Just because a man is a successful businessman it doesn't mean he will be any good at running a country. (R)
6. Men who were educated at one of our better private schools would usually be the ones best fitted to run this country.
7. This country would be best run by men who have had a university education.
8. A Prime Minister does not need to be a highly educated man because if he is in doubt about something he can always call on the advice of outside experts. (R)
9. What we need in Parliament is people who have led the same sort of life that the ordinary man has to put up with every day. (R)
10. We would have a lot more honesty and fair dealing in the Government if more working-men were elected to Parliament. (R)
11. I would prefer to be represented in Parliament by someone of my own social class. (R)
12. People who have been to private schools are no better trained than the ones who only went to State-run schools. (R)
13. It is best that this country should be run by upper-class people.
14. The ordinary man would not make a good Prime Minister even if he were given the opportunity.
15. People who are born into rich families are just as likely to be unfit to govern as anybody else. (R)
16. A man from the working class is more likely to make a good member of Parliament than someone from the upper class. (R)
17. I would prefer to be represented in Parliament by a man respected for his social position.
18. There is no reason why the son of a worker shouldn't rise to make a good Prime Minister of the country. (R)
19. If I had the choice of voting for an upper-class person and a man from the working classes, I would prefer to vote for the working-class candidate. (R)
20. Just because a man is way high up socially it doesn't mean he knows any better what is good for the country. (R)


(The Deference Scale - Final form of 20 items reduced from an original 24 on a sample of 96 conscripts. Items marked '(R)' are scored 1 to 5. Other items are scored 5 to 1.)

For the final validity check the sample was divided up in terms of political preference and then subdivided into those in manual and those in non-manual occupations. Alford [13] argues for this latter division as the most significant in assigning subjects to a social class. See also Ray [14] for a more extended treatment of the reasons for this procedure -- which has been challenged by Kavanagh. To the political preference question in the survey 49 subjects gave no reply, 15 replied 'not sure' and 8 preferred the DLP. This left 83 ALP supporters and 48 L-CP supporters. On the manual/non-manual criterion these two latter groups were further subdivided as follows: Labour - 58 lower class and 25 upper class; Liberal-Country Party - 17 lower class and 31 upper class. The mean scores on the Deference Scale for these four groups were: 60.82 (SD 9.07), 67.48 (7.76), 65.47 (4.87), and 64.07 (6.26) respectively. Note that, as indexed by the standard deviation, there was most variability in the replies of the lower-class ALP voters and least variability in the replies of the lower-class conservatives. The t statistic for the difference between the means of the working-class conservatives and the working-class Labour voters was 2.7 -- which is significant at the 0.005 level (one tailed).

Discussion

There seems to be some conflict between the Australian findings reported here and the British findings reported by Jessop [15]. Jessop reports an important distinction between meritocratic deference and deference on purely social grounds whereas in the present work both types of utterance go together to form a satisfactory attitude scale with all the usual indications of homogeneity ('alpha' = 0.77). This could be because Australia lacks anything like the British aristocracy and consequently the only thing available to which people might possibly defer is in fact earned merit. On the other hand there also seems to be a conceptual fault in Jessop's distinction. Surely the only reason even the British defer (consciously, that is) to the aristocracy would be out of some belief that their particular background gives them a peculiar competence to govern. Competence and social position are not alternative reasons for deference at all. The alternatives surely are between varying types of competence or between varying sources of competence. Thus the present scale taps solely deference to competence -- but competence that is inferred on several different types of grounds. A reading of the scale's items reveals that deference is expressed to successful businessmen, to highly educated people, to 'graziers' (the nearest Australia comes to a landed gentry), to rich men, to upper-class people and to people who have been educated at private schools. That there is this general factor of deference to competence the success of the scale attests. This is not to say that separate scales to tap deference to education and deference to social class separately could not be developed. For some particular research applications that might in fact be necessary. It is felt, however, that a measure of general deference to conventional indices of competence in political choice is likely to be the most widely useful. By avoiding references to specifically British themes (such as 'peers' and 'Oxford University'), the present scale also facilitates international comparisons. The one word of specifically Australian reference in the scale is 'grazier'. This could perhaps be replaced for British use by 'one of the gentry' and for American use by 'rancher'.

The second point of conflict with Jessop's findings is with his conclusion that deference to competence (of whatever sort) is not in fact related to working-class conservative voting. The present results, on the other hand, are in accordance both with the theoretical prediction (e.g. Parkin [16]) and with previous British empirical findings (e.g. Nordlinger [17] and McKenzie and Silver [18]). How then may we explain this variability in findings? An important part of the explanation must lie in the measurement procedure of previous authors -- who categorize people against some arbitrary cutting point simply as 'deferentials' or 'non-deferentials'. The present study, on the other hand, treats degrees of deference and shows that there is less of it among working-class Labour voters. Obviously, if the questions centering around this concept are to be much advanced, use of a scale such as the present one ought to be considered in future British studies.

An interesting (but not unanticipated - see Parkin's [19] theory) 'political' finding that emerges very clearly with the measurement refinements introduced here is that working-class conservatives are not a-typically deferential. Rather it is the working-class Labourites who are a-typically non-deferential. (See the means of the four groups above). In other words, both groups of upper-class people also respect social position and expertise in the people they vote for. It is this effect which also accounts for the overall positive correlation between deference and self-assigned class. We do then have support for Parkin's account of deference as representing a normative cultural value from which working-class Labour voters are especially (but institutionally) insulated.

A slightly surprising finding is the low relationship between deference and authoritarianism. A similar low relationship was observed in the Meadowbank pre-test of the scale -- where the correlation was 0.109. It is quite clear then that deference cannot now be viewed as simply a particular instance of attitude to authority in the political field. It is a quite separate determinant of voting behaviour in its own right. Deferentials defer not because of their attitude to authority but because of their beliefs about the causes and efficacy of social position. They are not browbeaten people. This does represent clear empirical support for the division by Jessop of his senses (a) and (d) from the sense currently under discussion.

It is apparent therefore that the refinement in measurement that the new scale represents has already proved its worth in clarifying the nature and distribution of an important political variable.

Since the final scale is suitable for Australian use, it should a fortiori be suitable for British use. We ought to be able to expect in fact that the reliability would rise among British respondents -- from whom the concept underlying the scale emanates. The concept should be 'more real' to them and hence the scale items intrinsically more meaningful. Only empirical test can confirm this, however.

It also follows (as was mentioned before) that international comparisons of deference are a possibility offered by the new scale -- comparisons that could be made objectively and with a known degree of measurement reliability.

----------------------------

Footnotes:

1. R. T. McKenzie and A. Silver, Angels in Marble (London: Heinemann, I968).
2. E. A. Nordlinger, The Working Class Tories (London: MacGibbon and Kee, I967).
3. F. Parkin, `Working Class Conservatism: the Theory of Political Deviance', British Journal of Sociology, XVIII (I967), 278-90.
4. D. Kavanagh, `The Deferential English: A Comparative Critique', Government and Opposition, VI (1971), 333-60.
5. R. D. Jessop, `Civility and Traditionalism in English Political Culture', British Journal of Political Science, I (197I), 1-24.
6. A. F. Davies and S. Encel, Australian Society (Melbourne: Cheshire, 1965).
7. F. M. Lord and M. R. Novick, Statistical Theories of Mental Test Scores (Reading, Mass.: Addison-Wesley, 1968), p. 90.
8. M. E. Shaw and J. M. Wright, Scales for the Measurement of Attitudes (New York: McGraw Hill, I967).
9. Data collected by Miss Janet Ross as part of the pre-test for a joint project.
10. C. J. Hoyt, 'Note on a Simplified Method of Computing Test Reliability', Educational and Psychological Measurement, 10 94I) 93-5.
11. L. J. Cronbach, 'Coefficient Alpha and the Internal Structure of Tests', Psychometrika, XVI (1951), 297-334.
12. Ray, J.J. (1971) An "Attitude to Authority" scale. Australian Psychologist, 6, 31-50.
13. R. R. Alford, 'A Suggested Index of the Association of Social Class and Voting,' Public Opinion Quarterly, XXVI (I962), 417-25.
14. Ray, J.J. (1971) The questionnaire measurement of social class. Australian & New Zealand J. Sociology 7(April), 58-64.
15. Jessop, 'Civility and Traditionalism'.
16. Parkin, 'Working Class Conservatism'.
17. Nordlinger, 'The Working Class Tories'.
18. McKenzie and Silver, 'Angels in Marble'.
19. Parkin, 'Working Class Conservatism'.


POST-PUBLICATION ADDENDUM

It may be of interest to note that Fox et al. (1977) also found -- using a large random U.S. community sample -- that "workers without authority (the obey class) tend to perceive the authority structure as less legitimate" .

Reference
Fox, W.S., Payne, D.E., Priest, T.B. & Philliber, W.W. (1977) Authority position, legitimacy of authority structure, and acquiescence to authority. Social Forces, 55 (4), 966-973.



Replication is one of the cornerstones of science. A new research result will normally require replication by later researchers before the truth and accuracy of the observation concerned is generally accepted. If a result is to be replicated, however, careful specification of the original research procedure is important.

In questionnaire research it has been my observation that the results are fairly robust as to questionnaire format. It is the content of the question that matters rather than how the question is presented (But see here and here). It is nonetheless obviously desirable for an attempted replication to follow the original procedure as closely as possible so I have given here samples of how I presented my questionnaires in most of the research I did. On all occasions, respondents were asked to circle a number to indicate their response.




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 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)