Person. indkid. D# Vol. I I. No. 9. pp. 923-929. 1990 Printed in Great Britain. All rights reserved

0191-8869/90 53.00 + 0.00 Copyright C 1990 Pergamon Press plc

THE DEVELOPMENT OF SINGLE TRAIT PERSONALITY THEORIES

ADRIAN FURNHAM

Department of Psychology, University College London, 26 Bedford Way, London WCI, England

(Received IO September 1989)

Summary-This paper describes the typical developmental pattern in the evolution of single and then multiple trait measures (and theories) in psychology. It is argued that eight stages typically exist: the identification through observation or experiment of behaviour pattern (phenomenon); tests for the robustness, validity and extrapolatory nature of the behaviour pattern; the development of a self-report single-trait measure; numerous construct and predictive validity studies on correlates of the measure; factor analytic work which suggests the behaviour pattern is multi- rather than uni-dimensional; the development and refinement of more than one, often sphere-specific and multi-dimensional measures of the same concept; doubts as to the actual validity of the original construct; acceptance into the canon of psychological measures. Different concepts and measures such as assertiveness, A-Type behaviour, and locus of control are used to illustrate these developmental stages. Problems associated with this proto-typical development are discussed along with difficulties associated with the notion of a trait.

INTRODUCTION

Textbooks on personality theory tend to fall into three types (Furnham, 1988). Some take the line of benevolent eclecticism describing, and sometimes comparing and contrasting, a long and venerable line of personality theorists whose ideas are offered with equal validity. Completely opposite to them are partizan zealots who are dismisive of all other approaches, and present only one theory as if it were the only, viable, robust and veridical theory of personality and individual differences. Thirdly some reviewers are enthusiastic taxonomists who attempt to classify theories according to such things as their epistemological origins, the form of their data base etc.

There are many different ways of taxonomizing current personality theorists but it is generally acknowledged that one of the most important and popular is the trait approach which assumes a limited number of specifiable broad predispositional variables that determine behaviour (Pervin, 1984). There are numerous arguments as to what these fundamental traits are, how they are related, their aetiology and how they ‘determine’ behaviour. Traits are generally regarded as dispositional factors that regularly and persistently determine conduct in a variety of everyday situations. The term type is usually superordinate to that of trait because traits are often inter-correlated which give rise to type. In the language of factor-analysis, type is a second-order or super factor while trait is a primary factor. The terms are used loosely and interchangeably which causes a good deal of difficulty and misunderstanding.

Hampson (1982) has distinguished between multi-trait theories which are designed to describe the entire personality and have the purpose of locating the constellation of traits, both higher and lower order, that make up the structure of personality. On the other hand single-trait theories are concerned with the role played by one single dimension of personality in the determination of behaviour. Hampson (1982) has argued that the most influential of these traits are those which have been demonstrated to describe, predict or explain a wide variety of behaviours. Recent research would suggest there are about five of these though they are variously labelled: extraversion, autonomy, neuroticism, conscientiousness and tender-mindedness (Brand & Egan, 1989).

Single trait theories are legion and include such trait dimensions named as authoritarianism, achievement motivation, A/B Type behaviour, field dependence-independence, conservatism, locus of control, just-world beliefs, assertiveness, Protestant Work Ethic beliefs, self-monitoring, etc. Three points need to be made about this list which may extend to nearly 50 primary traits or more. Firstly that although described as traits or dimensions, some researchers resist the term trait preferring type, style, behaviour pattern, cognitive belief pattern or need, because although they believe the disposition that they have isolated is stable over time and across situations, and highly

91-l ADRIAN FURNHAM

predictive of certain types of behaviour, the term trait seems to have certain implications that they would prefer not to be associated with their approach. For instance, some researchers might believe that the term trait suggests a biological, rather than a learned aetiology, or they may believe traits imply a continuous variable whereas their conception is discontinuous and dichotomous. Hence some authors who describe single dimension personality or individual difference variables resist the ascription of trait. This loose use of language has its problems and frequently leads to pseudo debates between theorists who use similar terms in different ways, or different terms in similar ways.

Secondly the origin of these traits is highly varied. Some arise out of cognitive psychology, others social and clinical, and yet others research from psychological research. Some traits, like authoritarianism might have first been articulated within a psycho-analytic framework while others like locus of control originated firmly within a behaviourist tradition. Hence there are very wide differences in how these traits are measured and the terminology employed in their use. These problems are compounded when traits described are measured in one epistemological tradition and translated or borrowed by another.

Thirdly it is rare that any single-trait measure is entirely new, in terms of the way it is conceived, described or measured. Although there are exceptions it is frequently the case that researchers having done extensive reading noticed a consistent pattern in previous reports which made sense of their data or results and which could be explained in terms of ‘new’ trait. Frequently then the origin of the trait term can be ascribed to a particular source or research team, but the ideas that are articulated can be traced back to many other authors including those who were not psychologists or never actively themselves did any research. For instance H. J. Eysenck has argued that the origin of his theory can be found in the work of Hippocrates, Galen, Kant, and Wundt as well as European and American researchers in this century (Eysenck & Eysenck, 1985).

Despite both the enormous number of single ‘trait’ theories in psychology with variously different origins, terminologies and measurement techniques they share very frequently similar histories. That is, the developmental history from the first published study on the new trait to world-wide research efforts, investigating it seems to follow a remarkably similar pattern. It is the aim of this paper to describe the very common sequential stages in the developmental of these ‘traits’ and then illustrate these stages by examples from different areas of research.

The development of single trait theories or models appears to go through most of the following stages sequentially. There are, of course, many problems associated with any stage-wise theory- how long each stage lasts; what determines movement from the one stage to the next; whether one can skip a stage or not; whether one might return to an earlier stage; and whether all phenomena pass through all stages (Furnham & Bochner, 1986). Despite these obvious and important short-comings eight stages will be described in the development.

Eight stages of a trait measure

(I) IdentiJication of the phenomena. This may occur as a result of laboratory experimentation or observation in a clinic, at work or through critical reading. The first stage can occur when a researcher operationalizes that which is well-known in literature into a psychological measure such as was the case with the Protestant Work Ethic (Furnham, 1990). But what is more normally the case is that a researcher observes a psychological phenomenon which he or she gives a name. Examples are legion; Seligman (1975) noticed the behaviour in dogs (learned helplessness) which later became translated into an attribution style questionnaire; Lerner (1980) noticed how people tended to blame victims of misfortune for their own fate which was developed into self-report measures of the just world; while Rosenman, Brand, Jenkins, Friedman, Strauss and Wurm (1975) studies on coronary-prone behaviour led to the development of the original type A behaviour. A number of points need to be made about this original stage. Firstly the person or persons who originally make the observations need not necessarily be the ones who develop the single-trait theory or the self-report measure. Secondly that often the phenomenon is only new in the sense that it has not been recorded before in quite the same way. Thirdly that nearly always this stage occurs in the laboratory from observational studies or occasionally from the systematic recordings of clinicians who note consistent relationships in the behaviour of their patients. Very rarely, if ever, are the researchers aiming to develop a trait measure or theory. The theory and measure ‘develop’ because it proves a succinct, robust and efficient descriptive and/or explanatory tool for behaviour.

Development of single trait personality theories 925

(2) Replication of the effect. The second stage is characterized by replications and considerably more experimental work on the nature of the effect observed. An excellent illustration of this can be found in Lerner’s (1980) book on Just World Beliefs which reports on numerous experiments using the concept. The idea of this phase is to test the robustness of the findings often by subtle yet simple changes in the dependent and independent variable. To some extent this is simply a case of data gathering attempting to find support for observations made while in others a series of studies attempt to test the various hypotheses that make up the nascent theory. Another good example is the development of the locus of control or personal responsibility concept described by Weiner (1980). These studies are usually reported in the first paper or book, to mention or describe the behaviour pattern/phenomenon.

(3) The development of a self-report measure. Despite the fact that the original researchers may not be personality, clinical or social psychologists and in fact may have little faith in self-report measures, the next stage involves the development of a self-report measure. These questionnaires may be of highly variable psychometric quality and the research that goes into establishing them somewhat inadequate. Reliability, validity and normative statitics may be fairly minimal to begin with and it is unlikely that first published versions will have gone through what psychometricians would consider acceptable testing. Indeed it is precisely because the originators of the concept are not psychometricians (being clinicians or experimentalists) that they do not always know the minimum criteria required of a good self-report measure. Weiner (1980) has pqinted out an interesting paradox in the shift from laboratory (experimental) research to self-report (correlational) research on the same concept. Whereas experimental manipulations systematically examine little else but the behaviour (e.g. perceived control), self-report studies probe nearly everything else rather than manipulate it. “Thus, there is an imbalance between studies manipu- lating chance and skill orientation and those that measure individual differences in the perceptions of these environments” (Weiner, 1980, p. 253).

Frequently the self-report measure is developed some years after the concept/behaviour pattern has been described in the literature. In this case, what might occur is that over the space of a few years a number of synonymous (but not highly correlated) measures will be developed. For instance over a 10 yr period at least half a dozen assertiveness inventions were developed by different researchers in the social skills literature (Furnham & Henderson, 1981). Similarly there are about the same number of measures all attempting to measure the Type A behaviour pattern (Linden, 1987). More popular concepts like locus of control might have been at least a dozen self-report questionnaires all supposedly providing a reliable and valid measure of that concept (Furnham, 1986).

(4) Validation of the measure. The fourth phase may continue for some time and involves numerous experimental and correlational studies of various sorts all aiming to validate the measure and concept underlying it. Studies are often the sort that make-up a PhD and include a programmatic series of studies aimed to test logical corollaries of the theory. What links the studies is the uni-dimensional trait measure used to assess the independent variable. A large number of these studies are essentially attempts to establish the concurrent, construct and predictive validity of the self-report scale by correlating it with other well-known measures or behaviours. The danger, however, of this sort of approach as Kline (1985) has pointed out is that correlating a new measure with an established, but itself poorly psychometrized measure does not provide good evidence of the validity of the theory or research. Whilst some studies provide nice evidence of the construct validity of the measure, it is far less common to find any evidence of the predictive validity of the measure possibly because of the difficulty and expense associated with longitudinal work. A glance at the citation index of any well-known self-report measure shows the extent to which validational studies are done, some by the original author and his/her acolytes but more commonly researchers from quite different laboratories. Paradoxically it is not lack of validity that prevents research into a measure or concept, but more likely the extent to which the measure taps the zeitgeist of (North American) psychology. Thus just as poverty is a popular sociological research topic during periods of affluence (Furnham & Lewis, 1986), so conformity and authoritarianism is studied in periods of liberalism.

(5.. Factor analytic work and multi-dimensionality. Although researchers may identify what they believe to be a single, albeit complex, dimension of phenomenon, and thence develop a

926 ADRIAN FLXNHAM

uni-dimensional scale, subsequent multivariate statistics (cluster analysis, factor analysis, multi- dimensional scaling) nearly always show the measure to be multi-dimensional with specific interpretable primary factors which may be orthogonal or oblique. Assertiveness questionnaires for instance have been shown to tap four quite different types of assertiveness depending on whether the behaviour is positive or negative; initiating or responsive (Furnham & Henderson, 1984; Henderson & Furnham, 1983).

Similarly, many distinctions have been made in the A-type literature such as healthy/adaptive-unhealthy/unadaptive A and B behaviours but factor analysis of the widely used Jenkins Activity survey (Jenkins, Zyzanski & Rosenman, 1979) revealed four factors: the original type A, speed and impatience, job involvement and hard-driving and competitive. Considerable debate still surrounds the psychometric properties of these scales (Boyd & Begley, 1987; Shipper, Kreitner, Reif & Lewis (1986). Locus of control scales have also proved to be multi-dimensional though there is predictably some debate as to the number of dimensions, their relationship and how they should be labelled (Collins, 1974; Fumham & Henry, 1980; Zuckerman & Gerbasi, 1977).

Factor analytic work usually poses problems for the original author as the ‘theory’ upon which the measure is founded usually assumes a uni-dimensional concept. At least three responses are common. One is to maintain that the concept, measure and trait are unified at a higher order (i.e. superfactors) and that although it may have various components, these are second order (primary) distinctions/factors that do not threaten the theory. A second approach is the attempt to revise the scale either by attempting to eradicate items that load on irrelevant factors or build a truly multi-dimensional instrument. An example of the former is Snyder’s (1988) attempt to revise the self-monitoring scale after its multi-dimensional nature had been clearly established (Furnham & Capon, 1983). A third approach is to do a meta-analysis of factor-analytic studies, decide on the factor structure and accept the original scale as multi-dimensional. This phase may last many years but may help resolve equivocal findings when they can be attributed to the multi-dimensional structure of the trait measure.

It is at this stage in the development of a trait that some of the most interesting and dilhcult questions have to be answered, namely “When is a trait not a trait?’ Take for instance the concept/trait of impulsiveness or sensation seeking. Eysenck and Eysenck (1977) used factor- analysis to examine impulsiveness and found four factors: narrow impulsiveness; risk taking; non planning; and liveliness, all of which correlated with each other between 0.18 to 0.50. Similarly Zuckerman (1979) found his sensation seeking concept factored into four factors: thrill and adventure seeking; experience seeking; disinhibition and boredom susceptibility again which correlated between 0.10 and 0.48. If subfactors correlate together at about r = 0.3, does this mean they can be grouped together as a single trait, or should they be regarded as single traits and the group or collective term dropped? In other words what sort of level does one have to set in order to consider a trait a genuine and meaningful concept? There is no agreed solution to this problem. Research has consistently demonstrated that superfactors (like extraversion) are relatively invariant, replicable in their determinants and factor-analytically stable while smaller primary factors (themselves probably divisible) cause problems in identification and naming. Eysenck and Eysenck (1985) recommend concentration on superfactors and those primary factors which are relatively pure conceptually and in terms of the items loading upon them.

(6) Multiple, multi-dimensional measures. The malaise following repeated psychometric investi- gations into an established uni-trait measure often leads scholars to despair because as has been noted it is uncertain at which level analysis should proceed. A common response, however, is for a team of psychometrically oriented researchers to develop a new better scale or self-report device. These new ‘improved’ measures often have various specific features. Firstly they are nearly always multi-dimensional in the sense that they provide subscale scores which may be combined or not to a single score depending on the needs of the researchers. Hence, Levenson (1974) developed a three dimensional locus of control scale (internal, chance and powerful others). Secondly, many researchers develop sphere-specific scales to measure the trait, belief or behaviour system within a very restricted range of behaviours as this has been shown to improve the predictive validity considerably. Thus there are measures of health locus of control (Lau & Ware, 1981), weight locus of control (Saltzer, 1982), mental health locus of control (Calhoun, Pierce & Dawes, 1973), dental health locus of control (Ludenia & Denham, 1983), political locus of control (Davis, 1983) and

Development of single trait personality theories 927

economic locus of control (Furnham, 1986). There are of course problems with this proliferation of measures because studies using different measures are not strictly comparable. Also it is possible that a person may be high on one measure (internal, economic locus of control), but low on a related measure (external mental health locus of control). Some authors have attempted not so much a multi-dimensional measure but sphere-specific measures which set out to measure the same beliefs (i.e. locus of control or just world) in different contexts (intra, inter-personal, socio-political) (Paulus, 1983; Furnham & Procter, 1989). Thus Fumham and Procter (1989) argue one can have just world beliefs regarding inter-personal relations and simultaneously unjust world beliefs concerning the socio-political sphere.

(7) Douim about the original concept. It is not infrequent that after a decade or so of intensive psychometric work on a measure/concept, authors begin to cast doubts about its conceptual and psychometric status. Compare researchers concerned with the measurement of assertiveness and A-type behaviour. Assertiveness: “. . . first and foremost, the assertion construct is outmoded and should be relinquished. The construction has proven to be vague, difficult to define, and to be laden with assumptions reflecting traditional rather than more contemporary views of personality and behaviour change. In the future, we need to concentrate more on response-, and situation-specific behaviour falling under the rubric of social skills-social competence and retire the assertiveness (assertion) construct” (Galassi, Galassi & Vedder, 198 1, p. 330). Type A : “The present paper argues that the Type A coronary-prone behaviour pattern, when seen as a global construct has lost its usefulness. Recent empirical evidence is reviewed to justify this claim. It is clear that much initial promise of the Type A construct has not been borne out, and it is argued that only the hostility component of the Type A construct continues to possess predictive and concurrent validity and therefore, usefulness” (Linden, 1987, p. 177).

In other words, the complexity of measurement and the equivocal nature of the findings leads reviewers to conclude that the original concept/phenomenon/behaviour pattern, and all question- naires that attempt to measure it should be abandoned either in favour of a new concept, usually a subscale of the former, or else that the original behaviour pattern is too unstable to be considred as a trait. This stage is characterized, not like the last two by increased empirical work, but theoretical reconceptualization. Naturally the commitment of various researchers to the concept and scale, mean that they are loath to relinquish it but happy to make further attempts to refine it. Thus while some laboratories and teams continue happily using the concept and measure of it, others abandon it in favour of a new revised concept and measure. However, it is only when the latter number exceed the former that the concept is unlikely to reach the next phase.

(8) Acceptance and ‘text -bookization ‘. Once having gone through the above 7 stages and having survived particularly the last one, the concept and its measures is usually accepted into the canon of individual difference measures. Sure sign of this process is the inclusion into the numerous, synonymous, benevolent eclectic text-books on personality. But at this stage there is probably a sizeable literature on the concept and the measure as citation counts show. However, one should not assume that because a test and concept has gone through the above baptismal and confirmatory process, that it is therefore necessarily a psychometrically valid, theoretically important or diagnostically useful measure. Small bands of zealots wedded to the original ideas in the scale can propel a measure of dubious theoretical and psychometrical validity into the text-books and research consciousness. Equally, extremely good measures based on sound theory and careful psychometric work can get ‘lost’ and never make it to the laboratories of the world. The zeitgeist in psychology seems to determine measure popularity most.

CONCLUSION

This paper has proposed an eight-stage model in the development of self-report, trait-like measures. However, like all stage-wise theories it has its limitations and unanswered questions. Do theories/measures have to go through all stages sequentially? Can some stages be skipped? Does development have to be linear or can it be cyclical? What prompts movement from one stage to another?

Despite these unresolved difficulties it may be useful to adapt the above stage-wise model to evaluate the progress of a trait-like measure or indeed predict further developments. Thus for

928 ADRIAN FURNHAM

instance one might predict that recent work showing that the Kirton (1976) adaptor-innovator measure is multi-dimensional (with three clear interpretable factors) would mean that it has reached the sixth stage, and that a multi-dimensional measure will soon be developed. Similarly, despite doubts about the validity of both measures, it is probable that sphere-specific measures (i.e. work-related, non-work related) of both assertiveness and A type is constructed.

However, it should not be assumed that a robust, valid and pretty universally accepted measure undergoes no further development and refinement, Despite exhaustive and extensive work on the factor structure, construct validity and theoretical parsimony, Eysenck, Eysenck and Barrett (1985) have revised the EPQ-R in order that the measure is as psychometrically sound as possible.

At the very heart of the development of single trait personality theories lies a number of difficult and as yet unanswered questions. If all traits (primary factors) can be shown to be multifactoral (and most factor analysis shows this) what should be done about this? If superfactors like extraversion account for practically all the real variance in an alleged trait, should it be retained? Should we always eliminate the portion of variance due to the agreed superfactors (like extracting g in the case of intelligence) and only then consider a ‘trait’ if it accounts for sufficient remaining variance?

There are theoretical, methodological and ‘political’ factors to be taken into consideration. Theoretically it would seem that parsimony would require the minimum number of super- and primary factors but there remains disagreement as to what both are and indeed the best way to measure them. Methodologically there is also agreement as to whether superfactors should be orthogonal or oblique; or for instance whether factor-analysis is an appropriate tool. Perhaps it is the ‘political’ reasons which seem to give one most problems. Discovering, naming and providing a measure for a unique trait is for some psychologists like having a plant named after one for botanists, and a star for astronomers. A trait-term and measure ensures attention (perhaps notoriety), increased citation counts and longevity, and it is naturally the case that some researchers pursue that goal to the detriment of the real theoretical and methodological issues.

REFERENCES

Boyd, D. & Begley, T. (1987). Assessing the Type A behaviour pattern with the Jenkins Activity Survey. Bri~irh Journal of Medical Psychology, 60, 155-161.

Brand, C. & Egan, V. (1989). The ‘Big Five’ dimension of personality? Evidence from ipsative, adjectival, self-attributions. Personality and Individual D@rences, 10, I 165-I 171.

Calhoun. A., Pierce, J. & Dawes, A. (1973). Attribution theory concepts and outpatients’ perceptions of the causal locus of their psychological problems. Journal of Communiry Psychology, I, 52-58.

Collins, B. (1974). Four components of the Rotter internal external scale. Journal of Personality and Social Psychology, 29, 381-391.

Davis, J. (1983). Does authority generalize? Political Psychology, 4, 101-120. Eysenck, H. & Eysenck, M. (1985). Personaliry and individual differences: A natural science approach. New York: Plenum

Press. Eysenck, S. & Eysenck, H. (1977). The place of impulsiveness in a dimensional system of personality description. British

Journal of Social and Clinical Psychology, 16, 57-68. Eysenck, S., Eysenck, H. & Barrett, P. (1985). A revised version of the Psychoticism scale. Personaliry and Individual

Differences, 6, 2 I-30. Furnham, A. (1986). Economic locus of control. Human Relations, 39, 2943. Furnham, A. (1988). Lay theories. Oxford: Pergamon Press. Furnham, A. (1990). The Proresram pork ethic. London: Routledge. Furnham, A. & Bochner, S. (1986). Culrure shock. London: Methuen. Furnham, A. & Capon, M. (1983). Social skills and self monitoring processes. Personalify and Indiuidual Djfirences, 4,

171-178. Furnham, A. & Henderson, M. (1981). Sex differences in self-reported assertiveness in Britain. British Journal of Clinical

Psychology, 20, 50-62. Furnham, A. & Henderson, M. (1984). Assessing assertiveness: A content and correlational analysis of five assertiveness

inventories. Behavioural Assessmem, 6, 79-88. Furnham, A. & Henry, J. (1980). Cross-cultural locus of control studies. Psychological Reporrs, 47, 23-29. Furnham, A. & Lewis, A. (1986). The economic mind. Brighton: Wheatsheaf. Furnham. A. & Procter. E. (1989). Belief in a just world: Review and critique of the individual difference literature. British

Journal of Social Psychdlogyi 20, 365-384. Galassi. J., Galassi, J. & Vedder, M. (1981). Perspectives on assertion as a social skills model. In Wine, J. Jr Smye, H. (Eds).

Social comaerence. New York: Guildford Press. Hampson, S. (1982). The construction of personality. London: Routledge. Henderson, M. & Furnham, A. (1983). Dimensions of assertiveness. Journal of Behaviour Therapy and Experimenral

Ps.ychiarry. 14. 223-23 I.

Development of single trait personality theories 929

Jenkins, C., Zyzanski, S. & Rosenman, R. (1979). Jenkins Acticily Sumey Manual: Form C. New York: Psychological Corporation.

Kirton, M. (1976). Adaptors and innovators: A description and measure. Journal of Applied Psychology. 61. 622429. Kline, P. (1985). Personality. London: Hutchinson. Lau, R. & Ware, J. (1981). Refinements in the measurement of health specific locus of control beliefs. Medical Care. 19.

1147-l 158. Lemer, M. (1980). The Belief in a Just World: A fundamental delusion. New York: Plenum Press. Levenson. M. (1974). Activism and powerful others. Journal of Personalifv Assessmenf. 38. 377-383. Linden, W. (1987). dn the impending-death of the Type A construct: Or is there a phoenix arising from the ashes. Canadian

Journal of Behavioural Science, 19, 178-190. Ludenia, K. & Denham, G. (1983). Dental outpatients: Health locus of control correlates. Journal of Clinical Psxchologx.

39, 854-858. Paulus, D. (1983). Sphere specific measures of perceived control. Journal of Personality and Social Psychology, 4-t.

1253-1265. Pervin, L. (1984). Personality: Theory and Research. New York: Wiley. Rosenman, R.. Brand, R., Jenkins, C., Friedman, M., Strauss, R. & Wurm, M. (1975). Coronary heart disease in the

Western Collaboration Group Study. Journal of the American Medical Association, 233, 872-877. Saltzer, E. (1982). The weight locus of control (WLOC) scale: A specific measure for obesity research. Journal of Personalit]

Assessment, 46, 620-628. Seligman, M. (1975). Helplessness: On depression, development and death. San Francisco: Freeman. Shipper, F., Kreitner, R., Reif, W. & Lewis, K. (1986). A study of four psychometric properties of the Jenkins Activity

Survey Type A scale with suggested modifications and validation. Educational and Psychological Measurement. 46. 551-564.

Snyder, M. (1988). Public appearances/private realifies. New York: Freeman. Weiner, B. (1980). Human motivation. New York: Holt. Rinehart & Winston. Zuckerman,‘ M. (1979). Sensation seeking: Beyond the optimal level of arousal. London: Wiley. Zuckerman, M. & Gerbasi, K. (1977). Dimensions of the 1-E scale and their relationship to other personality measures.

Educational and Psychological Measurement, 37, 159-175.