School of Sociology, University of N.S. W., Kensington, N.S. W., Australia
(Received 4 March 1987)
Summary --- The results of 11 community-wide studies in which a short form of the Marlowe-Crowne Social Desirability scale was administered are summarized. In seven Australian studies significant correlations (ranging from 0.20 to 0.42) were found between the scale and age. In one study having mostly male respondents, however, no significant correlation was found. This led to the data from a ninth study being dissected into males and females. A significant correlation was then found for females only. Two further studies from West Germany and India gave results similar to the Australian findings. The results were interpreted to mean that older women seek to compensate for 'Loss of looks' by promoting themselves as having other virtues. These highly meaningful findings from use of a social desirability scale among the elderly tend to rebut attacks on the validity of such scales among the elderly.
Lie scales such as the Eysenck and Eysenck (1975) L scale or the Crowne and Marlowe (1964) Social Desirability scale have long been used to improve the validity of clinical and survey data. It was therefore a matter of some concern when Carstensen and Cone (1983) found that two commonly-used indices of well-being among the elderly correlated 0.70 and 0.58 with such a scale. It suggested that important findings on well-being among the elderly might be flawed by poor validity in the measuring instruments used.
McCrae (1986), however, has dealt with this problem by producing data which call into question the validity of lie scales when they are applied to elderly populations. They showed that the correlation between self-reported well-being and peer-rated Neuroticism falls when social desirability controls are applied. They conclude that for at least their own sample of elderly people (mean age 56 years) the social desirability scale used measured more than a tendency to fake good. They argued that it also measured real feelings of well-being. They imply that approval-seeking is not orthogonal to neuroticism.
A drop in validity when controls for artifacts are applied is however a not unexpected outcome. In the present case it may mean simply that people who fake good tend to get away with it to some extent. They even fool friends who are asked to rate them. Faking an appearance of well-being may simply be part of faking good. Clearly more data are needed before we can decide which explanation carries most weight.
One finding that would seem to lend support to the contentions of McCrae (1986) is the fact that scores on the Eysenck Lie scale have clearly been shown to increase with age (Eysenck and Eysenck, 1975). It could thus be argued that generally high scores among the elderly would tend to wash out individual differences and thus lead to lower validity. Social desirability could thus perhaps be seen as an example of a general tendency among the elderly to make more use of defence mechanisms. Shenfield's (1984-85) review of the evidence for and against the existence of such a tendency did not however produce very conclusive results either way. Older people may be prone only to greater use of some mechanisms rather than defence mechanisms in general. Social desirability responding, therefore, has to be considered in its own right.
At the theoretical level, there are cogent reasons why older people might be more prone to social desirability responding. The fact that old age tends to be negatively stereotyped in our society would certainly provide some impetus for old people to want to prop up their own self-image and indulging in propaganda about one's own virtues is an obvious way to do such propping up. On both empirical and theoretical grounds, therefore, the McCrae theory has some plausibility.
While the results with the Eysenck `L' scale may thus lend implicit support to the McCrae theory, however, it should not be overlooked that there are other social desirability scales in existence which are also widely used. Could it be that scales such as that by Crowne and Marlowe (1964) would be less prone to influence from the age of the respondents? Could it be that the Crowne and Marlowe scale might provide an alternative to the `L' scale that is superior in use with elderly respondents? Given the wide use of the Crowne and Marlowe scale, this does seem worth examination. If McCrae's theory can be shown to be inapplicable where the Crowne and Marlowe scale is used, many existing research results might yet be shown as meaningful. Some data on the age effects observable with the Crowne and Marlowe scale are therefore presented below.
METHOD AND RESULTS
As the present author has for many years used a short form of the Crowne and Marlow (1964) scale as a methodological control in general population surveys, a considerable archive of data on the question was readily available. Details of the scale and its reliability have been given elsewhere (Ray, 1984a) but its correlation with age has not so far been generally made available. This is remedied below. The mean age of the samples mentioned below was generally around 40 years but exact details and fuller methodological details can be obtained from the publications cited.
In a random doorstep sample survey of the Australian city of Sydney with an N of 95 (Ray, 1981) the product-moment correlation between the short Crowne and Marlowe scale and age in years was 0.25 (P < 0.05).
In a quota sample of 87 Sydney people (Ray, 1982a) the r was 0.39 (P < 0.01).
In a random doorstep sample of 200 Sydney people (Ray, 1984b) the r was 0.36 (P < 0.01).
In a random doorstep sample of 84 people from the Australian city of Brisbane (Ray, 1985) the r was 0.42 (P < 0.01).
In a random doorstep sample of 100 Sydney residents (Ray, 1984c) the r was 0.20 (P < 0.05).
In a random mail-out survey of 122 residents of the Australian State of New South Wales (Ray and Bozek, 1980, Study I) the r was 0.26 (P < 0.01).
In a random mail out sample of 95 people from throughout Australia (Ray, 1984d, Study III) the r was 0.30 (P < 0.01).
In a postal sample of 128 Australians designed for males only and with mostly male respondents (Ray and Ray, 1982) the r was 0.17 (n.s.).
The fact that the last-mentioned sample was alone in showing no significant relationship between social desirability and age strongly suggested that the relationship might be peculiar to female respondents. Perhaps it is the female component of the various random samples that gives rise to the relationship between age and social desirability response. Such a restriction was not foreseen when the analyses were planned so there is only one study that enables a clear test of the hypothesis. This was a postal sample of 88 Australian males and 126 Australian females (Ray and Lovejoy 1984). In this sample the r was 0.15 (n.s.) for males and 0.33 (P < 0.01) for females.
All the data given so far have been obtained in Australia so it is fair to ask how similar Australia is in this respect to other countries. Two studies are available which do suggest that the Australian situation may be fairly generalizable. In a random doorstep sample of people from the West German city of Munich (N = 136), the r between a German version of the short Marlowe--Crowne scale and age was 0.28 across the sample as a whole (Ray and Kiefl, 1984) and in a sample of 305 residents of the Indian city of Bombay (Ray, 1982b) the r was 0.19 (P < 0.01).
The need for careful distinctions that flows from a consideration of Shenfield's (1984-5) work was well borne out by the present results. What looked like a clear finding that old people are particularly prone to make use of a particular defence mechanism turns out to look much more like an effect that concerns older women only. Age appears to have no effect on social desirability amongst men but does increase social desirability responding amongst women.
A fairly obvious explanation for this phenomenon is the well-known emphasis in our society on the particular importance for women of physical attractiveness. Since such attractiveness tends to be defined in terms of a very youthful ideal, women face considerable pressure on self-esteem as they grow older. Where men may even be seen as becoming 'distinguished' in looks as they grow older, aging brings only 'loss of looks' for women. Faced with such a loss, it is hardly surprising that women might seek something to compensate for their loss and promotion of themselves as having other virtues is an obvious recourse. Promoting one's virtues is, however, approval seeking behaviour or social desirability responding so the high level of social desirability responding among older women can be seen as a fairly direct outcome of the problems women have in measuring up to the demands of a culture that is so heavily oriented towards youth and attractiveness.
The fact that highly intelligible results (such as those outlined above) are obtained from use of a social desirability scale among older people does, however, also have the effect of supporting the validity of such scales in examining problems of the elderly. In such settings, lie scales give good results qua lie scales. They do not need reinterpretation as measuring something else. Thus the attack by McCrae (1986) on the validity of such scales among the elderly is not sustained. Old people do alter their responses in a way that approval-motivation theory would lead us to expect. In deciding between competing explanations for the Carstensen and Cone (1983) and McCrae (1986) findings, therefore, we might conclude that an explanation in terms of deficient validity for lie scales among the elderly is not the most plausible.
A final incidental implication of the present findings is that future retrospective studies of childrearing practices that use the mother as an informant might have to be particularly aware of the unreliability of such informants. Fathers may be a more truthful source of information than are mothers.
Carstensen L. L. and Cone J. D. (1983) Social desirability and the measurement of psychological well-being in elderly persons. J. Geront. 38, 713-715.
Crowne D. P. and Marlowe D. (1964) The Approval Motive. John Wiley, New York.
Eysenck H. J. and Eysenck S. B. G. (1975) Manual of the Eysenck Personality Questionnaire. Hodder & Stoughton Educational, London.
McCrae R. R. (1986) Well-being scales do not measure social desirability. J. Geront. 41, 390-392.