**************************************************************************************
The Journal of Social Psychology, Volume 123, Second Half, August 1984, pp. 189-193.
Measuring Trait Anxiety in General Population Samples
JOHN J. RAY
School of Sociology, University of New South Wales, Australia
ABSTRACT
The Eysenck "N" scale has been tested on general population samples but contains no balancing against acquiescence. The Spielberger STAI and the Taylor MAS contain some balancing against acquiescence but appear to have been tested on students or clinical groups only. As the Spielberger scale was the more nearly balanced, it was administered in Study I to a random doorstep sample of 113 Australians; and it showed a reliability of .84. It was then administered in Study II to a random mail-out survey of 115 Australians; the reliability was .91. It was concluded that the Spielberger scale is roughly as suitable for general population samples as it is for students. In a third study, a new balanced scale to measure chronic anxiety was constructed on the basis of a random mail-out sample of 95 Australians using items from both the Taylor and the Spielberger scale; its reliability (alpha) was .90. Its control against acquiescence and the availability of general population norms for it should make it preferable in most research to either of the two scales from which it was derived.
THE THREE MOST WIDELY USED measures of chronic anxiety would appear to be the Taylor (15) Manifest Anxiety Scale (MAS), the Spielberger et al. (14) Trait Anxiety Inventory (STAI), and the Eysenck (2) Neuroticism ('N") scale. Of these, only the Eysenck scale would appear to have been tested on general population samples. The other two appear to have been evaluated with the use of only student or clinical groups. The major problem with the Eysenck scale, however, is that it is all one-way worded. There are no items to which an anxious person would say "No." This means that scores on the scale could be influenced by acquiescent response bias. Not only true neurotics but also "yeasayers" would tend to get high scores. That this is no mere hypothetical problem was shown recently in a study where the Eysenck scale correlated .36 with an independent measure of meaningless acquiescence (11). Any correlations with the scale could therefore have a potential second interpretation; they may be due not to the effect of anxiety but to the effect of acquiescence. Heaven (4) has shown that acquiescence alone can produce high correlations with other variables. Rorer (13) claimed that the confounding effects of acquiescence are a "myth" but Rorer's work has been dismissed by Jackson as "inconsistent, arbitrary and misleading" (5, p. 82). See also Ray (7, 10). Clearly, then, an alternative to the Eysenck scale must be found for studies of chronic anxiety on general population samples.
Of the two widely used but student-based measures mentioned above, the one most nearly balanced is the Spielberger scale. It has seven out of 20 items worded in an anti-anxiety direction. This does give at least some potential for controlling against or examining the effects of meaningless acquiescence. One could, for instance, adopt the practice of scoring two forms of the scale: the usual form and a second form with six "anxious" items randomly deleted. Any correlation that remained significant with both forms of the scale could then not be explained away as the effect of acquiescence. Some test of whether the Spielberger scale is transferable to random samples of the general population seems therefore desirable. Scales constructed on one type of sample often undergo a disastrous collapse of internal consistency when applied to another type of sample (6).
METHOD and RESULTS
Study I
As part of a longer questionnaire concerned with attitudes toward crime, the Spielberger scale was administered to a random cluster sample of 113 people living in the Sydney metropolitan area of Australia by trained and supervised student interviewers. The mean age of the sample was 36 years, there was a slight preponderance of females, mean education fell between Junior School and Senior School, and 62% worked in nonmanual occupations. Other details of the study can be found elsewhere (9).
The scale showed a reliability (alpha) of .84, which indicates a level of internal consistency only little inferior to that usually reported for this scale when applied to students. All items correlated highly significantly with the scale total (after overlap correction). With four response options per item and an "anxious" item earning a score of 4 for "Almost always," the scale mean was 35.38 (SD = 8.65). This, then, would appear to represent a first approach in determining general population norms for this scale. Also, the scale showed no significant correlations with age, sex, education, or occupation, which suggests that it perhaps can be used without fear of demographic biases.
Study II
A defect of Study I is that the sample was entirely urban. Country people could conceivably show different levels of anxiety. Study II, therefore, used postal sampling methodology. The sampling frame was the voter registration roll for the entire State of New South Wales. (It should be noted that Australia is the world's most urbanized country and the city of Sydney alone accounts for three million of the state's five million population. In geographical area, New South Wales is roughly the size of Alaska.) Four hundred questionnaires were sent out with 115 returns forming the final sample. The distribution of demographic characteristics was very similar to that of Study I. Other details of the study can be found elsewhere (8).
On this occasion, the alpha was even more satisfactory: .91. The mean was 37.33 (SD = 9.53). Again there were no significant correlations with demographic variables. The scale did, however, correlate .67 with the MMPI "Pd" scale. It appears to have been demonstrated again, then, that the Spielberger scale functions very well on general population samples. In Study I the items were presented in the "Do you behave?" format, whereas in Study II the original "I behave" format was used. For example, the original item "I feel pleasant" had to be reworded as "Do you feel pleasant?" for the purposes of interviewer administration in Study I. As the scale appears to have been designed for self-administration, the results of this second study are, then, the ones more closely comparable with earlier work.
Study III
Although the transferability of the Spielberger scale to general population samples was found to be remarkably good, it still remains something of a problem that there are roughly twice as many pro-anxiety as anti-anxiety items in the scale. This means that careful research would require tedious controls for acquiescence whenever the scale is used. While this is not an insuperable difficulty, a scale that was completely balanced would certainly be more convenient. Doubts about acquiescence effects could thus be dispelled from the outset. It was therefore the aim of this third study to provide such a scale.
In Study III, the Taylor (15) MAS was administered together with the seven anti-anxiety items of the Spielberger scale. Additionally, two normally pro-anxiety items from the MAS (numbers 15 and 23) were reworded in an anti-anxiety direction. In a previous South African study (12) the reworded items had been found to show satisfactory item-total correlations. The reworded items follow: "I am in general as happy as anyone else" and "I am not a very nervous person." All items were rated "True," "?," or "False" by the respondents.
The sample was a random mail-out survey of the whole of Australia -- again drawn from the national voter registration lists. It should be noted that voter registration is universal and compulsory in Australia for all Australian-resident citizens and even for many noncitizen residents. The sampling frame is thus unusually comprehensive. Five hundred questionnaires were sent out with 95 returns.
All items were combined into a single scale and items correlating poorly with the scale total were deleted. This left a balanced 24-item scale comprising all seven Spielberger items plus items 3, 6, 8, 10, 13, 14, 15 (reworded as noted above), 16, 18, 20, 21, 22, 23 (reworded), 24, 25, 26, 28 of the MAS. In order to make the administration instructions of the Spielberger items compatible with those of the MAS, the word "generally" was transferred from the Spielberger introduction to each individual Spielberger item. Thus, for example, "I feel pleasant" became "I generally feel pleasant."
The reliability (alpha) of the new scale was .90 and its positive and negative halves correlated - .72 before reverse-scoring. Both results are highly satisfactory. With "True" to an anxious item scored 3, "?" scored 2, and "False" scored 1 (nonanxious items being scored in reverse to this), the mean score was 38.69 (SD = 10.34). These statistics should be useful as interim general population norms for the scale.
Some correlations of the new scale with other scales included in the battery may be of interest. It correlated .555 with the Rosenberg (1) Self-esteem scale and .413 with the Templer (16) Death Anxiety scale. These correla tions would appear to demonstrate convergent validity. The correlation observed with a short form (3) of the Marlowe-Crowne Social Desirability scale was -.226. This demonstrates a slight tendency of anxious people to answer more frankly than usual. The correlation is, however, very low in absolute terms and might, as such, be seen as demonstrating discriminant validity. The scale showed no significant correlations with demographic variables.
DISCUSSION
The present studies have not only shown that the Spielberger scale is suitable for general population use but they have also provided an improved alternative to that scale. As the new scale is derived entirely from items of the Spielberger STAI and the Taylor MAS, it should partake of any validity and general appropriateness that characterize these two widely used scales. It represents an innovation only inasmuch as it controls for any possible influence of meaningless acquiescence.
The general population norms provided in the present paper not only for the new scale but also for two forms of the Spielberger scale should also be invaluable in enabling comparison of any particular group with the community at large. To date, comparisons of particular groups with one another have been possible but any inferences about what is "normal" have been very poorly founded.
REFERENCES
1. Bahr, H. M, & Caplow, T. Old Men Drunk and Sober. New York: New York University Press, 1973.
2. Eysenck, H. J., & Eysenck, S. B. G. Personality Structure and Measurement. London: Routledge, 1969.
3. Greenwald, H. J., & Satow, Y. A short social desirability scale. Psychol. Rep., 1970, 27, 131-135:
4. Heaven, P. C. L. Authoritarianism or acquiescence? South African findings. J. Soc. Psychol., 1983, 119, 11-15.
5. Jackson, D. N. Acquiescence response styles: Problems of indentification and control. Pp. 71-114 in I. A. Berg (Ed.), Response Set in Personality Assessment. Chicago: Aldine, 1967.
6. O'Gorman, J. G. Limits to the general izability of the Marlowe-Crowne measure of social desirability. J. Clin. Psychol., 1974, 30, 81.
7. Ray, J.J. (1979) Is the acquiescent response style not so mythical after all? Some results from a successful balanced F scale. Journal of Personality Assessment 43, 638-643.
8. Ray, J.J. (1983) Psychopathy, anxiety and malingering. Personality & Individual Differences, 4, 351-353.
9. Ray, J.J. (1982) Prison sentences and public opinion. Australian Quarterly 54, 435-443.
10. Ray, J.J. (1983) Reviving the problem of acquiescent response bias. Journal of Social Psychology 121, 81-96.
11. Ray, J.J. & Lovejoy, F.H. (1984) The great androgyny myth: Sex roles and mental health in the community at large. J. Social Psychology 124, 237-246.
12. Ray, J.J. & Heaven, P.C. L. (1984) Conservatism and authoritarianism among urban Afrikaners. Journal of Social Psychology, 122, 163-170.
13. Rorer, L. G. The great response style myth. Psychol. Bull., 1965, 63, 129-156.
14. Spielberger, C. D., Gorsuch, R. L., &. Lushene, R. E. Manual for the State-Trait Anxiety Inventory. Palo Alto, CA: Consult. Psychologists Press, 1970.
15. Taylor, J. A. A personality scale of manifest anxiety. J. Abn. Soc. Psychol., 1953, 48, 285-290.
16. Templer, D. I. The construction and validation of a death anxiety scale. J. Gen. Psychol., 1970, 82, 165-177.
Go to Index page for this site
Go to John Ray's "Dissecting Leftism" blog
Go to John Ray's "Australian Politics" blog
Go to John Ray's "Gun Watch" blog
Go to John Ray's "Education Watch" blog
Go to John Ray's "Socialized Medicine" blog
Go to John Ray's "Political Correctness Watch" blog
Go to John Ray's "Greenie Watch" blog
Go to John Ray's "Leftists as Elitists" blog (Not now regularly updated)
Go to John Ray's "Marx & Engels in their own words" blog (Not now regularly updated)
Go to John Ray's "A scripture blog" (Not now regularly updated)
Go to John Ray's recipe blog (Not now regularly updated)
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