Sunday, August 21, 2005

British Journal of Medical Psychology (1980), 53, 181-186


By J. J. Ray and R. Bozek

It is suggested that 'freneticism' might be an appropriate name for the type of personality said by Rosenman, Friedman and Jenkins to be at risk from coronary heart disease. In study 1 the measure of this personality type (the JAS) was given a uniform answer format and, as such, was found to give improved reliability. It was administered to a random postal sample of Australians. A short form of 24 items was also produced. In study 2 a similar sample received the 24-item form and scales to measure achievement motivation, dominance and extraversion. 'A' types were found to be very much achievement motivated and dominant. A factor analysis showed these two traits as the main components of the A-B concept with only a third factor of 'freneticism'. This factor was measured in a third study which showed that the factor could be used as a scale in its own right.

Jenkins, Rosenman, Friedman and their associates (Jenkins et al., 1967, 1971; Zyzanski & Jenkins, 1970) have put forward the hypothesis that one of the important aetiological factors in coronary heart disease is the personality of the patient. They designate as 'A type' the personality most at risk.

The most important question to be asked about this hypothesis, of course, is whether such a personality does in fact provide the degree of prediction that the above authors claim. A prior question, however, has to do with elucidation of the hypothesis itself. What exactly does the A-B personality type consist of? What is it that the Jenkins Activity Survey (the proposed measuring instrument for the at-risk personality) is measuring? It is this question that is the concern of the present paper.

The label 'A type' (and its converse, 'B type') is, of course, deliberately uninformative. The underlying concept, however, would appear to concentrate on what one might call 'hard-driving-ness' -- particularly in relation to time. Friedman & Rosenman (1974, p. 67) define it as: 'Type A Behaviour Pattern is an action-emotion complex that can be observed in any person who is aggressively involved in a chronic, incessant struggle to achieve more and more in less and less time, and if required to do so, against the opposing efforts of other things or other persons.' Later (p. 70) the same authors say: 'Overwhelmingly, the most significant trait of the Type A man is his habitual sense of time urgency or 'hurry sickness'.' In general, then, if one had to find an ordinary English title for the A-B construct, one would perhaps choose 'freneticism' (from the adjective 'frenetic').

The one means so far used to measure this personality type is the Jenkins Activity Survey (JAS) in one of its various forms. Originally the instrument was designed to measure one continuous trait with high scorers termed type A and low scorers termed type B. Subsequently, the above-median scorers and below-median scorers were dichotomized further into A1, A2, B1 and B2 types. Also, factor analyses of the instrument showed that it could be conceived as measuring three fairly separate subfactors in addition to A-B type. The subfactors are: speed and impatience; job involvement and hard-driving-ness. In form the instrument is a behaviour inventory of generally around 60 questions. On some occasions, however, as few as 19 of these questions are used to score A-B type.

The instrument does have several features unusual in a personality scale: its items are a mixed set of Likert-type and forced-choice questions; it is very long and yet tends to have only minimal reliability (around 0.6); its conceptual identity is left vague and its possible relationship to standard psychological traits is not explored. Even in scoring the inventory, its authors took many years to rediscover to their 'chagrin' (Jenkins et al., 1971, p. 199) what Likert had discovered in the 1930s -- that weighted responses and unit-scored responses differ little in predictive power and reliability.

The purpose of the present paper, therefore, is to report the results of a program of research designed to: (1) shorten the JAS and improve its reliability; and (2) identify its relationship with conventional psychological trait measures. With regard to (1) above it should be noted that Shaw & Wright (1967) give 0.9 as the minimum reliability for a test designed to select out individuals for particular treatment and 0.75 as the minimum standard for a test used solely for research purposes. Although apparently used for both purposes, the JAS falls far short of even the lower standard. In Shaw & Wright's terms, it would appear to be only an instrument 'with possibilities for further development'.

Study 1

The JAS was included in a battery mailed out to a random sample of the population of the Australian State of New South Wales. The sample was drawn from the Australian electoral rolls. Electoral enrolment in Australia is compulsory not only for citizens but also for Australian resident citizens of other British Commonwealth countries. The sampling frame, therefore, is unusually comprehensive.

The questionnaire used included a two-paragraph preamble describing the purpose of the survey as: 'Research into how people see themselves and how they see their relationships with others'.

Of the 500 questionnaires sent out, a total of 122 were returned. A notable feature of the resulting sample was that the distribution of the four main demographic characteristics (age, sex, occupation and education) was not significantly different from that observed in contemporaneous random doorstep samples obtained in the Sydney metropolitan area. The postal interview technique did not therefore introduce any unusual biases as far as can be ascertained.

The 54-item version of the JAS was the one used but before inclusion, all items were recast into a uniform Likert format. Other scales included for the purpose of exploring possible relationships with the JAS were the Dominance scale from the Jackson (1967) PRF, the Ray (1976) 'Directiveness' scale (a scale of authoritarianism in behaviour inventory format), a short social desirability scale (Greenwald & Satow; 1970) and Eysenck's (1958) short scales of neuroticism and extraversion.

The reliability of the JAS and other scales was assessed by Cronbach's (1951) coefficient 'alpha'. This statistic is, of course, the mean of all possible split-half reliabilities and can be used as an estimate of test-retest reliability. Nunnally (1967, p. 210 ff.) however regards it as a superior statistic to the more easily understood reliability measures. Readers interested in a fuller discussion of the coefficient's meaning and use should refer to Nunnally's text.

The reliability observed for the JAS was, then, 0.76. This considerable rise over the levels reported earlier does hence suggest that the recasting of the scale into uniform Likert format was well worth while.

The correlations observed with other scales were: dominance 0.54, authoritarianism 0.44, extraversion 0.19, social desirability -0.18, neuroticism 0.02. All but the last are significant at the 0.05 level.

The JAS was then subjected to the automatic item analysis and deletion procedures of program ITRA (Ray, 1972). This produced a reduced 24-item version with a reliability of 0.77. The mean inter-item correlations of the shorter and longer versions of the scale were respectively 0.12 and 0.05. The two versions correlated 0.86. The correlations of the shorter version with other variables were: Dominance 0.42, authoritarianism 0.30, extraversion 0.20, social desirability -0.04 and neuroticism -0.02. The first three correlations are significant at the 0.05 level.

The implication of the correlations is that the 'A' type is dominant, authoritarian and (slightly) extraverted. He is not chronically anxious. These relationships are attenuated slightly with the shorter form of the scale even though it is more reliable. This indicates that the shorter scale has better claims to being a measure of a new and independent construct. The reduced openness of the shorter scale to dissimulation (social desirability responding) also seemed particularly worth while.
Study 2

This study was designed to cross-validate the new 24-item form of the scale. Would the 24 items work as well a second time around without the context of all the other items? Normally this is found to be so when a short form is produced (Ray, 1979), but it cannot of course be assumed.

Another question concerning the A-B construct that had not previously been examined was its relation to achievement motivation. Friedman & Rosenman (1974, p. 68) regarded achievement motivation as an unrelated construct on the grounds that one can be achievement motivated in a quiet as well as a hard-driving way, but the question would not appear to have been tested empirically. Many of the JAS items certainly appeared to embody the sort of sentiments normally associated with achievement motivation.

For the above reasons, then, a second battery was made up including the short A-B scale, the achievement motivation and dominance scales from the Jackson (1967) PRF, the two factors (impulsiveness and sociability) of extraversion according to Eysenck & Eysenck (1963) and the Bortner (1969) short A-B rating scale.

The Bortner scale was included because Jenkins at the time (personal communication) was recommending its use by researchers in lieu of the JAS until the JAS could be properly published.

Extraversion was measured by two scales on this occasion because the relationship observed in the first study appeared rather low given the nature of the A-B concept. It was thought that a break-up of extraversion into its main components might be more revealing. For this reason, the 10 highest loading items on each of Eysenck's two factors were used as scales.

The questionnaire embodying the above items was again administered by mail in exactly the same way (random sample without replacements) as in study 1 above. The resulting sample of 119 again showed a distribution of demographic characteristics similar to that observed in doorstep studies.

The reliability (alpha) of the short A-B scale dropped to 0.61 - a level similar to or below that reported by the original authors for their form of the scale.

In spite of the reduced reliability, the correlation with the Jackson Dominance scale remained similar to that observed previously -- 0.42; other correlations observed were achievement motivation 0.66, sociability 0.23, impulsiveness 0.10 and the Bortner rating scale 0.33. The Bortner scale itself showed a reliability of only 0.53. This very low reliability may be one reason why the Bortner scale does not show the correlations it should. If its own items correlate poorly among one another, not much can be expected of its correlation with other variables. It seems a pity that this feature of the Bortner scale has not previously been noted.

To allow for correlations between the other scales used to predict A-B score, a two-step multiple regression analysis was then carried out. Using first all four predictors (achievement motivation, dominance, sociability and impulsiveness) a multiple R of 0.70 was obtained with beta weights respectively of 0.601, 0.177, 0.044 and 0.127. In the second step, the lowest correlation predictor (impulsiveness) was deleted to give a multiple R of 0.69 with beta weights of 0.584, 0-185 and 0-063. The program used was MULTR from Cooley & Lohnes (1962). It will then be seen that the predictors explain just less than 50

per cent of the A-B variance. It must be realized, however, that much (though certainly not all) of the unexplained A-B scale variance will be 'error' variance and, as such, is by definition not predictable by any method. In the context of what is normally observed with psychological data, the R observed is, then, very high. Three predictors only explain most of what the A-B scale systematically measures.

As a further method of getting at what A-B items measure, it was felt that some analysis of latent structure was called for. The method chosen for this was McQuitty's (1961) 'Elementary factor analysis'. In spite of its name, this method is in fact a form of cluster analysis. Several writers have reported it as giving a more interpretable set of factors than orthodox factor analysis does (see the references summarized in Ray, 1973, and also Rump, 1974).

Two cluster analyses were, then, carried out -- an analysis of the scale scores and an analysis of the individual items scores. On the first analysis, all the scale scores (including A-B) were found to come out on the one first-order cluster. In the second analysis, there were three second-order clusters. The first two were clearly achievement motivation and dominance respectively. Both loaded almost all of the respective Jackson scale items as well as A-B items. The third cluster was more interesting. It was also rather clearly a true 'freneticism' cluster -- with most of its items coming from the A-B scale or the Eysenck 'impulsiveness' scale. It comprised 17 items - all worded in the 'A' direction.

The fact that the 'freneticism' cluster came third, however, indicates that it was extracted only after all the really strong relationships (correlation coefficients) in the matrix had been used. It not only contained fewest items but they were more poorly correlated among one another. This cluster does, then, have least importance in summarizing what the A-B scale measures.

Study 3

In this study an attempt was made to explore the further usability of the newly discovered 'freneticisrn' cluster. An attempt was made to explore its usability as a new scale in its own right. Although only the third and weakest element of the JAS, it did seem to be the one that came closest to what the original authors aspired to measure.

A third postal survey similar to the previous two was then carried out. As well as the 17 freneticism items it included scales to measure dominance (the Ray, 1976, 'Directiveness' scale) and 'psychoticism' (Eysenck & Eysenck's, 1976, P scale). The former was included to preserve some point of contact with the previous studies and the latter was included because of the similarity of many of its items to how the A-B type had originally been conceived. The following items, for instance might be noted: 'Do you like to arrive at appointments in plenty of time?'; 'Do people who drive carefully annoy you?'; 'When you catch a train, do you often arrive at the last minute?'; 'Do you stop to think things over before doing anything?' All these items come, not, as might be imagined, from the freneticism or A-B scales but from the Eysenck P scale. The P scale does however contain many other items not so obviously related to the personality type or types being considered here. The implication, however, would be that these other items may correlate with the 'coronary' personality. In Eysenck's terms, then, we are exploring the possibility that the frenetic man may be 'psychotic'. The caveat must however be entered that Eysenck's use of the term 'psychotic' seems to differ somewhat from normal clinical parlance.

Again 500 questionnaires were sent out randomly. This time, however, 140 were returned. Again the demographic structure of the resulting sample was indistinguishable from that observed in contemporaneous doorstep samples.

Item analyses indicated that seven of the 17 freneticism items were not correlating significantly with the total score on the scale. They were therefore discarded to leave a 10-item scale with a coefficient 'alpha' (Cronbach. 1951) reliability of 0.71. A list of these items is given in the Appendix. A reliability of 0.71 is not sufficient for use in selecting out individuals for treatment but it does provide a preliminary research instrument. It may be noted that although it is much shorter than the JAS, it is apparently more reliable.

The new freneticism scale, then, correlated 0.230 with psychoticism and 0.112 with directiveness. Its mean was 27.35 (SD 5.73). It was thus shown to be independent of the dominance component that had so permeated the JAS but was somewhat similar to what the Eysenck P scale measured. With the given n, the critical level of the correlation coefficient for significance at the 0.05 level was 0.165.

Study 4*

In this study , it was desired to examine how the 10-item freneticism scale might function in routine clinical use. For this purpose, it was included in a questionnaire administered to several weeks' intake of patients at the Sydney Coronary Heart Disease Prevention Programme. Based on the responses of 201 people, the reliability of the scale was found to be 0.65. This indicates that the third 'component' of the A-B scale is no more satisfactory than the A-B scale itself as far as reliability is concerned.


It has been shown that Friedman & Rosenman (1974) erred badly in believing their
construct to be unrelated to ambition. The correlation between A-B and achievement motivation is in fact among the higher ones to be found in the psychological literature.

In fact, one must ask whether the A-B scale is measuring anything other than achievement motivation and dominance. Certainly we cannot tell just what component of the A-B scale is providing the prediction of cardiac risk. Perhaps it is the achievement motivation component alone. If this is so, the whole effort of producing a new scale was rather uneconomical.

Aside from its conceptual impurity, there are, of course, also good psychometric reasons why the A-B scale should not continue to be used for predictive tasks. Even in its 54-item Likert form, the JAS shows a quite poor degree of internal consistency. Although a reliability of above 0.7 is an improvement on a reliability of above 0.6, it still corresponds to an average correlation between all the items of 0.05. This again suggests strongly that the A-B scale is not, as its authors claim, measuring a single trait or construct. Nor does the 24-item version of the scale when administered independently of the original much improve on this. The mean r in that case was only 0.06. Even the Bortner scale was only a slight improvement over this -- a mean r of 0.08.

Future research directed towards cardiac risk prediction, then, can surely no longer use the single confounded measure provided by the JAS. Instead, at least two scales should be used to measure its two main components separately. For both these components -- achievement motivation and dominance -- there are already good scales available from the Jackson (1967) PRF.

Whether the third freneticism component is worth further examination must be a matter for individual judgement. It does seem that low reliability is an ineluctable aspect of anything uniquely associated with the A-B scale.

*The authors would like to thank Dr L. Simons, Senior Lecturer in Medicine, University of New South Wales, and director of the Sydney Coronary Heart Disease Prevention Programme, who made this study possible. Further details of the programme can be found in Simons & Jones (1978).


{Articles below by J.J. Ray can generally be accessed simply by clicking on the name of the article. I am however also gradually putting online a lot of abstracts, extracts and summaries from older articles by other authors so if an article not highlighted below seems of particular interest, clicking here or here might just save you a trip to the library}

BORTNER, R. W. (1969). A short rating scale as a potential measure of pattern A behaviour. Journal of Chronic Diseases, 22, 87-91.

COOLEY, W. W. & LOHNES, P. R. (1962). Multivariate Procedures for the Behavioural Sciences. New York: Wiley.

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

EYSENCK, H. J. (1958). A short questionnaire for the measurement of two dimensions of personality. Journal of Applied Psychology, 42, 14-17.

EYSENCK, H. J. & EYSENCK, S. B. G. (1963). On the dual nature of extraversion. British Journal of Social and Clinical Psychology, 2, 46-55.

EYSENCK, H. J. & EYSENCK, S. B. G. (1976). Psychoticism as a Dimension of Personality. London: Hodder & Stoughton.

FRIEDMAN, M. & ROSENMAN, R. H. (1974). Type A Behaviour and your Heart. New York: Knopf.

GREENWALD, H. J. & SATOW, Y. (1970). A short social desirability scale. Psychological Reports, 27, 131-135.

JACKSON, D. N. (1967). Personality Research Form Manual. New York: Research Psychologists Press.

JENKINS, C. D., ROSENMAN, R. H. & FRIEDMAN, M. (1967). Development of an objective psychological test for the determination of the coronary-prone behaviour pattern in employed men. Journal of Chronic Diseases, 20, 371-379.

JENKINS, C. D., ZYZANSKI, S. J. & ROSENMAN, R. H. (1971). Progress towards validation of a computer-scored test for the type A coronary-prone
behaviour pattern. Psychosomatic Medicine, 33, 193-202.

MCQUITTY, L. C. (1961). Elementary factor analysis. Psychological Reports, 9, 71-78.

NUNNALLY, J. C. (1967). Psychometric Theory. New York: McGraw-Hill.

RAY, J.J. (1972) A new reliability maximization procedure for Likert scales. Australian Psychologist 7, 40-46.

RAY, J.J. (1973) Factor analysis and attitude scales. Australian & New Zealand Journal of Sociology 9(3), 11-13.

RAY, J.J. (1976) Do authoritarians hold authoritarian attitudes? Human Relations, 29, 307-325.

RAY, J.J. (1979) A quick measure of achievement motivation -- validated in Australia and reliable in Britain and South Africa. Australian Psychologist 14, 337-344.

RUMP, E. E. (1974). Cluster analysis of personal questionnaires compared with principal components analysis. British Journal of Social and Clinical Psychology, 13, 283-292.

SHAW, M. E. & WRIGHT, J. M. (1967). Scales for the measurement of attitudes. New York: McGraw-Hill.

SIMONS, L. A. & JONES, A. S. (1978). Coronary risk factor screening and long-term follow-up. Year one of the Sydney Coronary Heart Disease Prevention Programme. Medical Journal of Australia, 2, 455-458.

ZYZANSKI, S. J. & JENKINS, C. D. (1970). Basic dimensions within the coronary-prone behaviour pattern. Journal of Chronic Diseases, 22, 781-795.

Received 17 April 1979; revised version received 3 September 1979.

Requests for reprints should be addressed to Dr J. J. Ray, School of Sociology, University of New South Wales PO Box 1, Kensington, NSW, Australia 2033.

R. Bozek is at the same address.


The items of the new freneticism scale. All items are answered 'Yes' (scored 3), '?' (scored 2) or 'No' (scored 1).

(1) Do you often long for excitement?
(2) Would you rate yourself as an impulsive individual?
(3) Would you be unhappy if you were prevented from making numerous social contacts?
(4) Do you often act on the spur of the moment without stopping to think?
(5) Do you often have trouble finding time to get your hair cut or styled?
(6) Do you often find yourself facing things such as: Unexpected changes, frequent interruptions, inconveniences and 'things going wrong'?
(7) Do you like to have many social engagements?
(8) Would people who know you well agree that you tend to do most things in a hurry?
(9) Would you do almost anything for a dare?
(10) Are you given to acting on impulses of the moment which later land you in difficulty?


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. 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.


No comments: