Australian Journal of Psychology, Vol. 31, No. 2, 1979, pp. 73-78


J. J. RAY and G. J. PRATT

University of New South Wales

It is argued that the inherent ambiguity of the one-word items in the Wilson C-scale might make them particularly subject to acquiescent responding. When administered in classroom settings there is already some evidence that the Wilson scale is so affected. In a study of 110 Australian Army conscripts, the Wilson scale was found to show a correlation between its radical and conservative items of only -.199. Scored without reversals, it made an acquiescence scale with a reliability of .61. In a second study with 126 Australian Army officers, a revised C-scale was used but the equivalent statistics were still -.226 and .60. It is concluded that the Wilson scale may be valid and acquiescence-free only when personal help is given with the answering of each item.

Of the very many self-report measures of attitudes and personality that are devised and reported on in the literature every year, very few seem to see repeated use. One notable exception to this that has emerged in recent years is the Wilson and Patterson (1968) C-scale. Designed as a measure of conservatism, the attraction of this scale is its unusual format. Instead of items in full sentence form, it uses single words or phrases as stimuli for subject response. Instead of being asked to agree or disagree with (for instance) "The death penalty should be retained for certain crimes", the subject is asked simply to respond to the simple expression "Death penalty". Although this format might at first seem impossibly ambiguous, Wilson (1973) has found that it does in fact elicit reliable and valid responses. Since one recent incomplete review lists over 70 subsequent published applications of the scale (Boyce, Note 1), it is apparent that Wilson is not alone in finding his "catchphrase" format useful.

The C-scale has, however, received several rather critical notices (e.g., Ray, 1972, 1974; Schneider, 1973) and one recent American review (Pedhazur, 1978) is particularly scathing - on the general ground that Wilson is unduly expansive in his claims for what the C-scale measures. As a general measure of social conservatism, however, the C-scale would not seem to be seriously impugned by Pedhazur's attack.

One issue affecting the C-scale that has never been resolved, however, is the role of acquiescence in responses to such short items. Wilson (1973) claims that the short format is particularly immune to the influence of acquiescence while Ray (1972) notes that the C-scale in fact appeared to attract an inordinate number of "Yes" responses. It may also be noted that a strong acquiescent set would well explain the noticeable phenomenon exhibited in Ray (1971) where alternating items get scores either above or below the item midpoint. All odd-numbered items show the sample as being very conservative whereas all even-numbered items show the sample as being very radical. This can be understood if it is realized that all odd-numbered items in Wilson's original schema are keyed conservative while all even-numbered items are keyed radical. Reverse-scoring has then the effect of leaving the odd-numbered items unchanged with their original, acquiescence-generated high proportion of high scores ("Yeses") whereas reversal turns the high proportion of "Yes" responses into a low mean score for the even-numbered items. A high level of acquiescent responding does then explain the high means, the low means and their regular alternation.

The balanced format of the C-scale should however control for any effects due to acquiescence and Cloud and Vaughan (1970) have supported such a view. This is true, however, only as far as the level of the scale mean is concerned. Regardless of any other property of that scale, any balanced scale will ensure that acquiescers do not get high scores. What acquiescence may affect is the scale's validity. If acquiescence causes items to be responded to similarly that should be responded to differently, how can we say that the items are measuring what we intend them to measure?

It was then as an indication of validity that the rPN statistic (correlation between "positive" and "negative" halves of a balanced scale before reverse-scoring) came into wide use as a test of the adequacy of balanced versions of the California F scale (Christie, Havel, & Seidenberg, 1956). How then does the C-scale fare against this test?

Generally (Wilson, 1974) the scale shows correlations of around -.7 -- which is eminently satisfactory. On one student sample, however, Ray (1972) found a correlation of .288 - which is in the wrong direction altogether. Using the children's version of the C-scale, Heaven (in press) also found a level of rPN (-.26) that must be regarded as unsatisfactory. There are then some reasons to believe that the C-scale may be seriously affected by acquiescence with certain samples only and it would hence seem important to explore the extent of this phenomenon.

The present author's experience in administering the C-scale door-to-door would suggest that the ambiguity of the items may be the source of the problem. Many items (e.g. "Socialism") are very prone to cause respondents to say "What do you mean by that?". A refusal to paraphrase in such circumstances is not generally well received and if one does paraphrase, the C-scale gets back to something very much like the old format it was designed to replace. In either case the presence of a person-to-person interaction does however ensure that the answering task is not totally rejected and some attempt to give meaningful answers will still be made. In a classroom situation this may not be at all true unless the task is particularly well "sold" to the respondents. Without the motivation provided by a person-to-person interaction, the response to ambiguity may well be indifference expressed in careless acquiescence.

At this stage, however, it still seems necessary to see whether or not the unsatisfactory levels of rPN that have so far been reported are at all common. Is it a phenomenon confined to students?

Since the results reported in Ray (1971) were derived from a sample of Australian Army conscripts, it would appear likely that low levels of rPN might be expected in such samples also. Since conscripts in Australia at the time were selected in a particularly fair way (by a random birth-date ballot) and with few opportunities for "dodging", they must be seen as representing a very good sampling of the different levels of society. If, then, the C-scale is inapplicable to them, its applicability to society in general must be seen as being in some doubt.

It is the aim of the present research, then, to examine the role of acquiescence in responses to Wilson-type items across further samples.


This study was carried out by reprocessing the data from the sample of 110 Australian Army conscripts reported on briefly in Ray (1972). This sample was gathered at the Kapooka training centre in 1969 but is in fact different to the other sample gathered in the same year at the same centre and reported on in Ray (1971).

The reliability and rPN of a revised version of the C-scale for this sample have already been reported in Ray (1972) but equivalent statistics have not been reported for the original C-scale administered at the same time. The original C-scale, then, showed a reliability of .74 and an rPN of -.199. Although both the present sample and the sample of Ray (1971) were given the scale in a classroom situation by Army psychologists, it may be noted that the reliability was slightly higher on the present occasion (.63 earlier). Whatever the level of the reliability may be, however, it is clear that the rPN was far from satisfactory.

Although rPN has been widely accepted as reflecting the prevalence of acquiescence in a particular set of responses, it is in fact not a direct measure of acquiescence. Low rPNs could equally arise, as Kerlinger (1978) argues, from a genuine non-opposition in the implications of the items concerned. In view of the high rPNs that have on other occasions been reported for the C-scale, this seems an unlikely explanation for the present results. Nonetheless, to preclude such claims, it does seem important to attempt a somewhat more direct measurement of the role of acquiescence.

One simple way to do this (Martin, 1964) is to score a scale for acquiescence. Given that one has a balanced scale to start with, one simply adds the item scores without doing any reverse-scoring. In effect, one simply counts the number of "yeses" regardless of meaning and thus uses each item as a minitest of a person's propensity to say "Yes". Reliability (split-half etc.) can then be calculated in the usual way. There appears to have been no previous instance of the C-scale being evaluated in this way.

The reliability of the C-scale so scored on the present occasion was .61 (Cronbach's (1951) coefficient "alpha"). This indicates that with the present sample the C-scale was almost as good as a measure of acquiescence as it was of conservatism.


One possible explanation of the above results is to see them as evidence of rebelliousness against the task. Conscripts are not after all there of their own accord and some resentment against any task they are given while in the Army might well be expected. Acquiescence could be one way of expressing non-co-operation.

The group of Army personnel who are most obviously in a position opposite to that of the conscripts are the career officers. It was decided therefore to carry out a second study using an officer sample. It was hoped thus to rule out subject characteristics as far as possible as an explanation of the above results.

The sample used was simply those officers who could be contacted while doing Army courses at various training centres. The sample is not thus random but it did at least cover a wide variety of corps, ranks and geographical origins. The sample size was 126 and the modal rank was Major. Subjects were given the questionnaire in a classroom situation by one of their peers. A careful introduction to the questionnaire answering task was given in an attempt to overcome in advance any objections they may have had to its ambiguity.

The C-scale version used was not the Wilson (1973) version but rather the Ray (1972) revision. This version had shown a better rPN on the previous sample than the original and was hence to be preferred. The rPN on the previous sample had been - .373.

Sampling was carried out in mid-1978. In the period that had elapsed since the publication of the revised C-scale (Ray, 1972), some of the items appeared to have dated somewhat -- a common problem with conservatism scales. One or two changes of wording had to be made to cope with this. "Beatniks", for example, was replaced by "Hippies".

On the resulting scale, then, the reliability ("alpha") observed was -75, the rPN was -.226 and the reliability of the scale scored for acquiescence was .60. All three results are remarkably similar to those found 9 years previously with a different scale and a different sample.

The means and SDs observed on the two occasions were also similar: 102.93 (9.97) in 1978 and 105.95 (11.37) in 1969.


The above results bring up to four the number of occasions when classroom administration of the C-scale has demonstrably produced unsatisfactory results. Although Wilson's (1973) data would appear to show that classroom administration is not always unsatisfactory, the conditions required are not at all clear. At the very least, any results obtained from classroom administration of the C-scale must in future be evaluated with extreme caution. At the very least, information on the levels of rPN should always be produced.

Given the origin of the above two studies in fears held for the ambiguity of catchphrase format items, the above results should represent some confirmation of their ambiguity. Scale items in the more traditional format do not normally require person-to-person administration before they can be guaranteed to function in the expected way so the catchphrase items must at least differ in some important respect.

One suspicion that the present results must give rise to also is the extent to which the subject is normally "helped" by the interviewer in door-to-door surveys. If this "help" at interpreting ambiguities is extensive, not only is the point of the new format (brevity) substantially lost but so also is the element of standardization for which closed-ended questions are normally valued. If door-to-door administration of the scale does produce vastly different internal consistency characteristics from classroom administration, one must suspect that the amount of "help" given in deciding responses door-to-door is in fact substantial.

If the C-scale is to continue in door-to-door use, then, explicit and workable policies on the amount of "help" to be given in respect of each item must be laid down and declared in advance for each particular instance of such use. This might well make the C-scale more difficult to apply than normal scales.

The above findings also would seem to have implications of a broader sort for attitude scales generally. Since the influential paper by Rorer (1965) declared "acquiescent response style" to be a "myth", almost all interest in the phenomenon appears to have died out. Perhaps understandably, scale users have welcomed the apparent exemption thus given them and have abandoned attempts to examine the possible influence of acquiescence on their measures.

That the exemption offered is only apparent, however, can be shown by Peabody's (1966) rejoinder to Rorer (1965). Peabody pointed out that although acquiescence may not generally be a problem, it will still be a problem with ambiguous scales. Although Rorer himself also acknowledged this point, its major implication appears to have escaped widespread attention: How can we know in advance whether or not our particular scale is one of the ambiguous ones? Thus if all the demons that Rorer appears to have exorcized are not to descend upon us we must still produce the talisman of evidence that our scales are not affected by acquiescence.

The present results show in fact that the possibility of ambiguity artifact is not a remote one. Rorer (1965) founded his entire thesis on the often repeated empirical finding that different measures of acquiescence fail to correlate with one another. In such circumstances, the concept of a general trait of acquiescence must be abandoned. Yet the present results show exactly the opposite of what Rorer's thesis depends on. Both studies showed that acquiescence was consistent from item to item. Otherwise the coefficient alpha would have approximated zero when the scale was scored for acquiescence. The unreversed C-scale was shown in fact to be a quite good measure of the tendency to acquiesce. Acquiescence does exist as a consistent tendency to respond - though it is apparently elicited only in the presence of ambiguous items.

Future work with scales generally, then, should perhaps return to a pre-Rorer level of concern with the potential effects of acquiescence.


1. BOYCE, M. W. The measurement of conservatism. Unpublished paper, 1976. (Available from S. C. V. Toorak, 336 Glenferrie Road, Malvern, Victoria 3144, Australia).


CHRISTIE, R., HAVEL, J., & SEIDENBERG, B. Is the F scale irreversible? Journal of Abnormal Social Psychology, 1956, 56, 141-158.

CLOUD, J., 8L VAUGHAN, G. M. Using balanced scales to control acquiescence. Sociometry, 1970, 33, 193-202.

CRONBACH, L. J. Coefficient alpha and the internal structure of tests. Psychometrika, 1951, 16, 297-334.

HEAVEN, P. C. L. The internal consistency of Wilson's conservatism scale. Journal of Social Psychology, 1979, 109, 143-144.

KERLINGER, F. N. Similarities and differences in social attitudes in four western countries. International Journal of Psychology, 1978, 13, 25-37.

MARTIN, J. Acquiescence - Measurement and theory. British Journal of Social and Clinical Psychology, 1964, 3, 216-225.

PEABODY, D. Authoritarianism scales and response bias. Psychological Bulletin, 1966, 65, 11-23.

PEDHAZUR, E. J. Review of the Wilson scale. Mental Measurements Yearbook, 1978.

RAY, J.J. (1971) "A new measure of conservatism" -- Its limitations. British Journal of Social & Clinical Psychology, 10, 79-80.

RAY, J.J. (1972) Are conservatism scales irreversible? British J. Social & Clinical Psychology 11, 346-352.

RAY, J.J. (1974) How good is the Wilson & Patterson Conservatism scale? New Zealand Psychologist 3, 21-26.

RORER, L. G. The great response style myth. Psychological Bulletin, 1965, 63, 129-156.

SCHNEIDER, J. F. The conservatism scale: Independent of S.D.? British Journal of Social and Clinical Psychology, 1973, 12, 90-91.

WILSON, G. D. The psychology of conservatism. London: Academic Press, 1973.

WILSON, G. D. Evaluation of the conservatism scale: A reply to Ray. New Zealand Psychologist, 1974, 3, 27.

WILSON, G. D., & PATTERSON, J. R. A new measure of conservatism. British Journal of Social and Clinical Psychology, 1968, 7, 264-269.

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)