IF "A-B" DOES NOT PREDICT HEART DISEASE, WHY BOTHER WITH IT? A COMMENT ON IVANCEVICH & MATTESON
University of N.S.W., Australia
A recent paper by Ivancevich & Matteson notes many problems with research into the "A-B" personality but recommends continued research into the construct. It is pointed out that these authors omit consideration of much of the relevant literature and that when the omitted work is included in consideration the "A-B" construct must be considered a false trail which should now be abandoned.
The "A-B" concept needs no overhaul?
A recent paper by Ivancevich & Matteson (1988) in this journal gives the appearance of assessing the current status of the findings with respect to the "Type A-B Behaviour Pattern" (TABP). The paper notes many problems with this field of research but concludes that "There is now growing acceptance of the proposition that a major overhaul in the TABP is not necessary or even feasible" (p. 37). I would like to submit that these authors were able to come to this conclusion only because they ignored a substantial part of the relevant literature. Sixteen adverse papers not cited.
I make no attempt to keep a complete file of references on "A-B". I would be photocopying a paper from almost every journal I pick up if I did. An examination of what was in even my limited files, however, turned up no less than 16 papers with conclusions adverse to the "A-B" hypothesis that Ivancevich & Matteson (1988) did not cite. Since Ivancevich & Matteson have some claim to being experts in this field (They have reviewed the "A-B" literature before. See Matteson & Ivancevich, 1980) this is rather surprising.
The "overlooked" papers that I have found are: Heilbrun, Palchanis & Friedberg (1986); Pickering (1985); Rime & Bonami (1979); Young, Barboriak, Hoffman & Anderson (1984); Katz & Toben (1986); Emara, El-Islam, Abu-Dagga & Moussa (1986); Hansson, Hogan, Johnson & Schroeder (1983); Ray & Bozek (1980); Ray & Simons (1982); Ray (1984 & 86); Lelouch & Kornitzer (1982); Davis & Cowles (1985); Appels, Mulder & Van Houtem (1985); Myrtek & Greenlee (1984) and Goldstein, Edelberg, Meier, Orzano & Blaufuss (1985).
There were also several similar papers that Ivancevich (1988) & Matteson would not have had available to them at the time they wrote: Linden (1987); Helmreich, Spence & Pred (1988); Nanjundappa, Friis & Taladrid (1987). Overall, this would suggest that a systematic review of the literature would yield many more such papers.
"A-B" a false trail
If the papers reviewed by Ivancevich & Matteson (1988) are supplemented by those listed above, I believe that we can only conclude that "A-B" was a classic false trail: There is a coronary-prone personality (i.e. aggression/hostility) but the most commonly used measure of "A-B" (the Jenkins, Zyzanski & Rosenman, 1979, "JAS") does not measure it. What it does measure is a confused mixture of traits that are best studied separately. Perhaps a brief look at what the "overlooked" references say will help substantiate this.
Goldstein, Edelberg, Meier, Orzano & Blaufuss (1985) studied two precursors of coronary heart disease (CHD) and found that one went with a high score on "A-B" while the other went with a low score. Myrtek & Greenlee (1984 p. 455) concluded: "No consistent differences on physiological measures could be found in our subjects based on behaviour type (according to SI or JAS criteria)". Appels, Mulder & Van Houtem (1985) reported that a 9.5 year prospective study of 2,750 older males found no association between JAS scores and CHD. Davis & Cowles (1985 p. 39) concluded: "A brief examination of methods of Type A personality assessment concludes that the use of the most popular technique (the student version of the Jenkins Activity Survey) and the majority of studies found in the recent literature will add little to our understanding of the aetiology of coronary heart disease". Lelouch & Kornitzer (1982) found a high prevalence of "A" types as measured by the Bortner scale in one population of CHD sufferers but a low proportion of them in two other populations of CHD sufferers. Emara, El-Islam, Abu Dagga & Moussa (1986) compared CHD sufferers with controls and found no differences on A-B. Katz & Toben (1986) found that the Novacco anger scale predicted CHD precursors while the JAS did not. Young, Barboriak, Hoffman & Anderson (1984) used a short parallel form of the JAS and found that it did not predict coronary artery occlusion. Rime & Bonami (1979) compared CHD sufferers with controls and found CHD suffers to be the more passive and dependant. This is very much the opposite of the description of the "A-Type".
Hansson, Hogan, Johnson & Schroeder (1983) and Ray & Bozek (1980) both factor-analyzed the JAS and found it mostly to be resolvable into familiar psychological constructs. Ray & Bozek (1980) factor analyzed it along with items from the Achievement Motivation and Dominance scales from the Jackson (1967) "PRF". Most "A-B" items loaded on the two factors defined by the two Jackson scales.
It may also be noted that Ray & Bozek found the Bortner (1969) scale to have an internal reliability of only .53. This contrasts with the assurance by Ivancevich & Matteson (1988) to the effect that the Bortner scale is one of those that "appear to have acceptable internal reliability" (p. 49).
Ray & Simons (1982) found that "A-B" did give a tiny prediction of CHD but that a scale measuring aggressive dominance was a better predictor. Jenkins & Zyzanski (1982) criticized the Ray & Bozek (1980) study mainly on the grounds that Ray & Bozek did not use the usual complex weighting formula for scoring the JAS items. Ray (1984) replied, pointing out that Jenkins & Zyzanski and their collaborators had themselves found that a non-weighting scheme worked better.
Ray (1986) reported analyses showing that the small degree of prediction of CHD furnished by "A-B" was due solely to the degree to which "A-B" included a component of aggressive dominance. Pickering (1985) found the results of studies of the association between "A-B" and coronary artery disease to be "confusing". Heilbrun, Palchanis & Friedberg (1986) noted that the JAS is a poor predictor of CHD and proposed that only a small sub-set of the JAS items are likely to be useful predictors. Nanjundappa, Friis & Taladrid (1987) found no effect of "A-B" on CHD. Linden (1987) argued "that the Type A coronary-prone behaviour pattern when seen as a global construct has lost its usefulness". Helmreich, Spence & Pred (1988) factor analyzed the JAS and found two factors of achievement striving (AS) and impatience and irritability (II). They found that AS correlated with academic achievement but II did not.
In the light of this list of findings, does it now seem reasonable to conclude that: "There is now a growing acceptance of the proposition that a major overhaul or change in the TABP is not necessary or even feasible"? Surely not.
(What is true is that psychologists very often use scales uncritically and the JAS is no exception to this. The vast majority of studies with the JAS treat it as some sort of "black box" with no thought as to what the items actually say or whether the scale is valid or not. For instance, Yarnold and his colleagues have published a profusion of papers (e.g. Yarnold & Grimm, 1986 and Yarnold, Mueser & Grimm, 1985) which proclaim that "A-B" gives some prediction of dominance in various situations. Since so many of the JAS items measure dominance, this is not much of a discovery).
Is a non-predictor of CHD important for other reasons? Ivancevich & Matteson (1988) do of course show some awareness of the sort of conclusions that are to be found in the "overlooked" papers. Such conclusions are so common that it would be hard not to. They note, for instance, that "The three subscales and the global score of the JAS are not predictive of CHD (Jenkins, Rosenman & Zyzanski, 1974)" (p. 49) and that the "psychometric properties of Type A measures are questionable in terms of reliability, validity and concordance" (p.52).
So what is going on in the Ivancevich & Matteson (1988) paper? What they seem to be saying is: "Maybe "A-B" has got nothing much to do with heart disease but it does correlate with a lot of other things so we should go on to find out why". This is, however, remarkably uninsightful thinking. If I put together a bunch of dominance items and a bunch of achievement motivation items and threw in some other assorted items for good measure, I too would have a scale that correlated with a lot of things. But what would be the point? I would be obscuring the truth, not revealing it. Yet what I have just proposed is a reasonable description of the JAS.
The particular admixture of items in the JAS was of interest only because it was said to correspond to the "coronary-prone personality". Since it does not, it loses that interest. It is true that not everything in the JAS corresponds with already familiar psychological traits. The SI (speed and impatience) subfactor was indeed fairly original and continued study of it by itself might be of continued interest. Even here, however, it might be noted that a simple and short alternative measure of the same construct is also available (the Ray & Bozek, 1980, "Freneticism" scale). Ironically, this scale does predict CHD in its own right (Ray, 1986).
Possible criticisms of this critique
It might be objected that some of the papers which I cite as detrimental to the A-B hypothesis are themselves in various ways flawed or not decisive in the conclusions they have drawn. For example, the paper by Appels et al which I cite was acknowledged by its authors to suffer from a weak methodology so its failure to find in favour of the A-B hypothesis could simply be due to that. Further, although Pickering (1985) found the association between A-B and CHD "confusing", he does propose a rationale that would reduce this confusion.
My reply to such an objection is twofold: Firstly, similar comments could be made about papers which appear to support the A-B hypothesis. For instance, Ivancevich & Matteson (1988) cite various studies which used the Bortner (1969) scale as a measure of A-B (e.g. the Belgian-French pooling project of 1984). Yet it was reported in this journal some time ago (Ray & Bozek, 1980) that this scale had an internal reliability of only .53 even in its original English-language form.
As some context to this, Shaw & Wright (1967) evaluated a large number of scales and concluded that .75 is the minimal level of reliability required in a research instrument and even regard the lower bound for the preliminary version of a research instrument as being .60. The Bortner scale, then, is even in English too multidimensional to be said to be a measure of anything. Since translated scales generally suffer a reliability decline that can be quite severe (Balakrishnan & Ray, 1986), the nature of the Bortner scale when translated into French can surely be imagined. It seems likely that it was little more than a random number generator.
One could go on in a similar vein at much length but I am endeavouring here only to present a pointed critique, not write a book. Detailed dissection of all the studies of A-B would in fact surely require a very large book indeed.
And my second point of reply is that, whatever the flaws in the papers I cite, a review that simply ignores adverse findings is surely a perilous one in which to place our trust. My aim is simply to point out that the peril exists.
The Structured Interview?
Another query that might be raised is whether I have ignored one of the main measures of A-B, the Structured Interview (SI). The JAS may not predict heart disease but the JAS is after all in some sense a secondary instrument. Might not the SI do a better predictive job? I can do little better than Yarnold & Bryant (1988) in commenting on this possibility. Yarnold & Bryant completely reject use of the SI on a variety of grounds. They point to the difficulty of training raters up to an adequate standard, the subjectivity of scoring and poor score reliability in various senses of that term. Yarnold & Bryant are particularly persuasive in noting that the SI can reasonably be administered only once to each subject because subjects tend to "wise up" after the first administration. This means that the reliability of the measure in what is arguably the most important sense of that term is in principle unknowable. From a psychometrician's point of view, therefore, it would be hard to imagine a worse measure of anything.
Additionally, in view of its weak relationship with the JAS, the small predictive success of the SI could even be due to its measuring things outside the A-B concept. The A-B concept is not therefore in any way saved by the SI.
Ivancevich & Matteson not alone
Perhaps it is worth noting at this stage, however, that Ivancevich & Matteson are not alone is appearing perversely to cling to the A-B concept. Houston & Snyder (1988) have recently edited a book which brings together papers by many of the leading protagonists of the A-B concept. Although these authors do in various ways acknowledge the predictive problems of A-B measures -- even showing an awareness that the long term results of the once much-relied on Western Collaborative Group Study do not show A-B as a significant predictor of CHD -- they offer only vague speculations for the predictive failure and continue on as if the concept were of undiminished importance. That it has now lost its whole raison d'etre seems to have passed them by completely.
Perhaps one could see something of a silver lining in the fact that the A-B concept did have a historical role in helping to establish the medical importance of psychological traits. This may well be so but it is surely hardly reassuring that the trait concerned ended up being found as not medically important.
Psychology relevant after all
The pity is that uncritical enthusiasm for the A-B concept seems to have caused psychologists almost completely to overlook the fact that there are in fact many other psychological traits that DO predict heart disease. Quite some time ago, Krug & Sherman (1977) pointed out that, in two large studies, 10 out of 16 scales in the 16PF differentiated CHD sufferers from non-sufferers. How much more useful for all of us it would have been if the medical importance of psychological traits had come to be accepted because of research such as this.
Appels, A., Mulder, P. & Van Houtem, J. (1985) De validteit van de Jenkins Activity Survey, een vragenlijst ter meting van het type A gedrag. Nederlands Tijdschrift voor de Psychologie en haar Grensgebieden 40, 474-487.
Bortner, R.W. (1969) A short rating scale as a potential measure of pattern A behavior. J. Chronic Diseases 22, 87-91.
Davis, C. & Cowles, M. (1985) Type A behaviour assessment: A critical comment. Canadian Psychology 26, 39-42.
Emara, M.K., El-Islam, M.F, Abu Dagga, S.I. & Moussa, M.A. (1986) Type A behaviour in Arab patients with myocardial infarction. J. Psychosomatic Res. 30, 553-558.
Goldstein, H.S., Edelberg, R., Meier, C.F., Orzano, J.A. & Blaufuss, L (1985) The paradoxical relation between diastolic blood pressure change under stress and the H factor of the Jenkins Activity Survey. J. Psychosomatic Research 29, 419-425.
Hansson, R.O., Hogan, R., Johnson, J.A. & Schroeder, D. (1983) Disentangling Type A behavior: The roles of ambition, insensitivity and anxiety. J. Res. Personality 17, 186-197.
Heilbrun, A.B., Palchanis, N. & Friedberg, E. (1986) Self-report measurement of Type A behavior: Toward refinement and improved prediction. J. Personality Assessment 50, 525-539.
Helmreich, R.L., Spence, J.T. & Pred, R.S. (1988) Making it without losing it: Type A, achievement motivation, and scientific attainment revisited. Personality & Social Psychology Bulletin 14, 495-504.
Ivancevich, J.M. & Matteson, M.T. (1988) Type A behaviour and the healthy individual. British J. Medical Psychology 61, 37-56
Jackson, D.N. (1967) Personality research form manual N.Y.: Research Psychologists Press.
Jenkins, C.D., Rosenman, R.H. & Zyzanski, S.J. (1974) Prediction of clinical coronary heart disease by a test for the coronary prone behavior pattern. New England J. Medicine 290, 1271-1275
Jenkins, C.D. & Zyzanski, S.J. (1982) The type A behaviour pattern is alive and well -- when not dissected: A reply. British J. Medical Psychology 55, 219.
Jenkins, C.D., Zyzanski, S.J. & Rosenman, R.H. (1979) Jenkins Activity Survey manual N.Y.: Psychological Corp.
Katz, R.C. & Toben, T. (1986) The Novacco anger scale and Jenkins Activity Survey as predictors of cardiovascular reactivity. J. Psychopathology & Behavioral Assessment 8, 149-155
Lelouch, M. & Kornitzer, M. (1982) Ischemic heart disease and psychological patterns. In: H. Denolin (Ed.) Psychological problems before and after myocardial infarction Basel: S. Karger
Linden, W. (1987) On the impending death of the type A construct: Or is there a phoenix rising from the ashes? Canadian J. Behavioural Science 19, 177-190.
Matteson, M.T. & Ivancevich, J.M. (1980) The coronary-prone behavior pattern: A review and appraisal. Social Science & Medicine 14a, 337-351
Myrtek, M. & Greenlee, M.W. (1984) Psychophysiology of Type A behavior pattern: A critical analysis. J. Psychosomatic Res., 28, 455-466.
Nanjandappa, G., Friis, R. & Taladrid, B. (1987) Type A personality: Hispanic and Anglo diabetes and cardiovascular disease. International Migration Rev. 21, 643-655.
Pickering, T.G. (1985) Should studies of patients undergoing coronary angiography be used to evaluate the role of behavioral risk factors for coronary heart disease. J. Behavioral Medicine 8,203-214
Ray, J.J. (1984) Confusions in defining A-B personality type: A rejoinder to Jenkins & Zyzanski. British Journal of Medical Psychology 57, 385
Ray, J.J. (1986) Alternatives to the A-B personality concept in predicting coronary heart disease. Personality Study & Group Behaviour 6(2), 1-8.
Ray, J.J. & Bozek, R.S. (1980) Dissecting the A-B personality type. British Journal of Medical Psychology 53, 181-186.
Ray, J.J. & Simons, L. (1982) Is authoritarianism the main element of the coronary-prone personality? British J. Medical Psychology 55, 215-218.
Rime, B. & Bonami, M. (1979) Overt and covert personality traits associated with coronary heart disease. British J. Medical Psychology 52, 77-84
Yarnold, P.R. & Grimm, L.G. (1986) Interpersonal dominance and coronary-prone behavior. J. Res. Personality 20, 420-433.
Yarnold, P.R., Mueser, K.T. & Grimm, L.G. (1985) Interpersonal dominance of Type As in group discussions. J. Abnormal Psychology 94, 233-236.
Young, L.D., Barboriak, J.J., Hoffman, R.G. & Anderson, A.J. (1984) Coronary-prone behavior attitudes in moderate to severe coronary artery occlusion J. Behavioral Medicine 7, 205-215