Personality Traits in terms of Social-Role Probabilities; an innovative theoretical essay on the possibility of overcoming the chaotic diversity in personality theories

PERSONALITY TRAITS IN TERMS OF SOCIAL-ROLE PROBABILITIES (An innovative theoretical essay on the possibility of overcoming the chaotic diversity in personality theories)

by

Popko P. van der Molen

presented at the Third European Conference of Personality Gdansk, Poland, Sept. 23-26, 1986.

(Registered as Heymans Bulletin HB-86-815-EX, State University of Groningen (RUG), Jan. 1987)

Deze front-pagina nog van betere lay out voorzien ****************

Keywords : abilities and skills adaption vs. innovation anchoring dimensions dominance (styles) energy level genetic vs. environmental factors personality traits person- vs. thing-orientedness reactivity rotations (oblique + orthogonal) self-will vs. compliance stability vs. over-sensitivity social roles subordinacy (styles of)

ABSTRACT

Social Psychology and Personality Psychology have developed into separate disciplines. Personality psychologists seldom ask questions about the way social roles are distributed and come into existence and social psychologists rarely explore the influence of temperamental differences and other basic personality traits on the way social roles develop.

Focussing on the interface between these two research traditions, we can find anchoring points for factor rotations which can bring order to the unmanageable multiformity of classification systems, categorizations and theoretical approaches in personality- and social psychology. To that end some personality dimensions are defined on the level of congenital predispositions as well as in terms of social role distribution. They are, in fact, biologically based personality traits, defined in terms of the probability of obtaining specific social roles (-scores).

The use of such doubly defined personality dimensions as reference points can produce considerable conceptual and theoretical advantages for both the social-role and the trait domain. It can help us to integrate both disciplines without leading to desultory complexities.

PERSONALITY: A MULTI-PERSPECTIVE ISSUE

Social Psychology and Personality Psychology have developed into separate disciplines. Personality psychologists seldom ask questions about the way social roles come into existence and are distributed and social psychologists rarely explore the influence of temperamental differences and other basic personality traits on the way social roles develop. However, since these disciplines proper are strongly interrelated, both dealing with individual differences, there is a considerable overlap in the dimensions, studied in each domain. As a result, there is, in the field of personality theories, an almost unmanageable multifor¬mity of classification systems, categorizations, and theoretical approaches. Some of them supplement each other, or may be considered as partly overlap¬ping and redundant. Others seem to compete, or even to be mutually exclusive.

For the present disquisition we need to distinguish aspects of social interactive behaviour as far as they are primarily dependent on time- and place-specific role distributions, from behavioural aspects with strong conge¬nital roots and from behavioural aspects which primarily depend on processes of learning and development. To that end we shall utilize a preliminary classification of personality dimensions in the following categories:

1. ) Social-role dimensions, refering to differences in the way people interact socially; personality dimensions stemming from social psychological research can mostly be considered as belonging to this category; differences between individuals with regard to these dimensions are often found to be strongly situation-dependent. Reviews of this category of personality research can for instance be found in Leary (1957),. Foa (1961), Wiggins (1979,1982) and Kiesler (1982,1983).

2. ) Traits of temperament, refering to the most stable differences between persons, often shown to have strong congenital roots. Personality-trait psychologists traditionally used to be interested in these aspects of personality in particular. This area of research is represented by a considerable number of "schools" and traditions. Other than in the realm of social interactive beha¬viour, there is little consensus here with regard to which axes of reference are to be considered most important. Moreover, most of the traditions in this realm cherish at least some personality dimensions which are evidently a mix of the presently applied three conceptual categories of personality differences. In the research traditions of Cattell, Eysenck, Strelau, Buss and Plomin, to mention a few, a major part of the personality dimensions under investigation is representative for the category "traits of temperament".

3.) Skills and abilities, sensitive to learning, training and education. Dimensions in this category emerge from research on learning and education as well as from research on personality differences in general. They are the most changeable and malleable category of personality differences.

These three categories primarily differ in stability over time and in the ease with which the personality characteristics in question can be influenced from the environment. In most research groups or university departments the research efforts are focussed exclusively on one of the major broad categories. That is to say, all dimensions under investigation are often either interpreted as traits of temperament, or as social-role aspects, or as skills and abilities, depending solely on the perspective from which the research group in question operates. When scrutinized more closely however, it is often rather difficult - especially in the case of orthogonal personality dimensions - to determine unambiguously for each dimension under investigation, whether it is indeed primarily to be considered a "trait of temperament", a "social role dimension", a "skill", merely a dimension of "judgment", unrelated to actual behaviour, or as a combination of these.

Interpretation of Personality Factors; Ambiguity and Confusion

One of the reasons for the apparent ambiguity of the classificatory status of many personality factors can be found in the fact that the various main sources of personality differences are strongly inter(cor)related.

Suppose for instance that certain innate properties account for an important part of personality-variance (e.g., genetical factors influencing physical strength). Then these genetic factors for strength will also influence:

a) the individual's chances of taking up certain social roles,

b) the likelihood of profiting by learning versus suffering irrepairably from certain environmental influences,

c) the probability of evoking certain judgments from other people, etc.,etc.

When a strong resourceful individual is competing with weak individuals for a dominating position, chances of success are clearly unequal. And after a social role distribution has been established and consolidated, a dominant position will in itself determine part of the individual's behaviour and thus also the experiences and skills to be acquired later on. Apart from social role positions, strong and resourceful individuals will be more able to learn from harsh environmental influences, to avoid the suffering of irreparable damage, and thus to better their chances of becoming more skillful. This in turn influences the individual's behaviour and its chances for success. In this way correlations between genetical factors and other sources of personality-differences will prevail. And the same is mutatis mutandis likely to hold for other combinations of personality sources.

All this means that it will be essentially impossible to find purely genetical personality factors, purely developmental personality factors, purely social role factors, etc., which are all mutually orthogonal to one another. The various main sources of personality differences being basically inter- (cor)related, we may still - staying within one realm, say, within the realm of trait-differences or the realm of social-role differences - choose to work with orthogonal factors within that area. But even then - limiting ourselves to one of the main realms of personality and also limiting ourselves to orthogonal factors another problem arises. Having decided that a particular n-dimensional euclidian space is the most proper representation of the data under consideration, there is still the problem of determining which position of personality factors is most convenient. Guilford & Hoepfner (1969) remark:

"It should be agreed that the aim of those who apply factor analysis for the purpose of discovering scientific constructs in psychology should be to achieve psychologically significant factors, which can be replicated, which fit into systematic psychological theory, and which can be investigated meaningfully by other methods. Only in this way can there be general agreement upon factorially discovered constructs and thus the unambiguous communicability that science requires."

When dealing with a general and rather exhaustive trait space, generally agreed upon external criteria to determine where to put our labels of reference are generally not available. Therefore, internal criteria are used such as e.g., a varimax rotation of principal components. But, as Guilford (1975) states:

"It is probably commonly known that the most disturbing deficiency of factor analysis is its indeterminacy - the lack of any completely dependable criterion with regard to where to place the reference axes. And when mathematical specifications for simple structure are written in the form of analytical rotation models (e.g., varimax or promax), the model may not fit psychological reality."

Were generally agreed upon external criteria available, a criterium rotation might render factors (and labels) to span the resulting multi-dimensional personality space in a way, more convenient for experimentation and manipula¬tion (Guilford et al., 1969; Elshout et ah, 1975).

Below we will attempt to deal with this general problem of searching for hard and generally acceptable external rotation criteria. But first let us rehearse the methodical problems encountered up to this point.

Note:

• For a clear demonstration of these principles on relatively simple data, i.e. measures of books or coffee cups, refer to Overall (1964), Cattell & Dickman (1962) and Cattell & Sullivan (1962).

manipulative viewpoint.

Oblique versus Orthogonal Axes of Reference

In attempting to summarize huge and complex data on personality differences, it seems attractive, for reasons of sparsity and mathematical elegance, to work with orthogonal factor rotations of principal component solutions.

If we choose for orthogonality, we are still burdened with the problem of where to put the axes of reference. Moreover, paradoxically, mathematical independence appears not to guarantee independence in terms of (functional) interpretation. Orthogonal personality factors are in fact often interpreted as and labeled with concepts that can theoretically and experimentally be shown to be clearly functionally interconnected and correlated. This seems to imply that these conceptual factors have been "forced" into orthogonal positions by the rotation methods applied, and that their orthogonality is not warranted by psychological reality. This problem is especially relevant between personality factors stemming from different main sources of personality variance, for instance between genetically based trait factors and social-role dimensions, or between these two types of personality factors and factors of skills and abilities. In other words, having found an n-dimensional data-space describing the correlations in a rather exhaustive set of personality variables, and having found an efficient description of the n-dimensional space by an "orthogonal simple structure treatment" of its principal components (e.g., "varimax"- rotation), we may almost be sure that each of these resulting orthogonal factors contains elements of many classes of personality-sources at the same time, thereby rendering the model sub-optimal from an experimental and Eysenck's system of higher-order personality factors may serve as an example. His Neuroticism and his Extraversion dimension both appear to be a blend of very different personality aspects. Extraversion for instance, is associated with the social-role aspect dominance as well as with the tempera¬mental attitude of person-orientedness (sociability) (Eysenck, 1953; Buss & Plomin, 1975). Extraversion, as a term in daily personality descriptive language, also appears to be correlated with levels of skill (Reeder et al., 1977). Furthermore, Eysenck's operationalization of Extraversion appears to be correlated with high activity levels (Feij, 1978; 1979,et al), while a high activity level may in itself be regarded as a basic trait which is important enough to be maintained as a separate dimension. The latter consideration has indeed been taken into account in the factorial personality models of many investigators. (See e.g., Heymans, 1932; Cattell, 1950; Thurstone, 1951; Guilford, 1959, 1975; Mehrabian, 1972,(ch.8); Strelau, 1974a and Buss & Plomin, 1975; For a review refer to e.g., Mischel, 1976 or Feij, 1978).

This confusing of various behavioural dimensions with individual differences in sheer basic activity level is characteristic for the realm. For one thing, a dominant social-role position may boost the level of overt activity, just as being trapped in an outcast- or scapegoat-like social position may considerably reduce it. Basic activity levels are therefore strongly mixed with secondary role-determined activity levels (Strelau, 1974a, p. 121). Another difficulty with the activity aspect in personality research stems from the above mentioned habit of focussing on varimax solutions when applying factor analysis. The varimax criterium pushes the axes of reference towards the most dense clusters of behavioural labels, whereas behavioural labels tend to be attached to certain activities rather than to certain "flavours" of activities. Therefore varimax factors tend to represent certain types of styles and of activities, each mixed with a high activity and energy level, rather than one pure activity factor plus factors of activity flavours. As a consequence, a separate dimension "general activity level" often appears to be missing in factor analytic studies on personality in which no further a-priori rotation criteria are applied than varimax- or similar statistical criteria.

This is the more disturbing, because energy is as it were the all-important exchange currency in nature. Life is a continuous struggle for materials and energy which are needed for maintenance, growth and propagation. Within every species there is a continuous selection going on, the "fittest" individuals propagating more successfully than the "unfit". One of the bottlenecks in the race for fitness is the ability to generate enough energy, even with a minimum of resources. Lack of energy produces losers and the process of losing in its turn causes more lack of energy. Therefore we must assume that a factorial model of personality is incomplete without a dimension which purely describes differences in the level of energy and/or activity.

Questionnaires versus Dictionaries

Another critique on approaches as discussed above, refers to the use of questionnaires and the way they are developed and statistically "refined". Questionnaire batteries tend to suffer from the "Baron von Munchhausen syndrome". The more they are used and refined, the more the test items are made to fit together statistically, the more "one-pointed" is their measuring capacity, and the more doubt there can be raised with regard to their being representative for the whole area of personality differences in general. This has led many investigators to the "lexicographic" approach in which virtually the complete realm of personality descriptive terms is included in empirical research for the dimensions underlying their utilization in daily life (See e.g., Baumgarten, 1933; Allport & Odbert, 1936; Norman, 1963; Brokken, 1978; John et al., 1984, 1986; Hampson et al., 1986). Norman's "Big Five" are a well known result of such an approach (they are commonly labeled "Surgency or Extra version", "Agreeableness", "Conscientiuousness or Dependability", "Emotional Stability" and "Culture"). These five dimensions represent an orthogonal rotation of the first five principal components of the correlations in the daily use of all personality descriptive adjectives (Norman, 1963). These factors have also in part been replicated in similar research outside of the English speaking world (Brokken, 1978, on dutch personality descriptive adjectives). The "big five" can, to a greater degree than eclectically chosen questionnaire batteries, be considered representative for the whole realm of personality differences, at least as sedimented in daily language. But here the disadvantages of orthogonality also apply. The adjectives loading high on each factor are at best blends of conceptually very diverse personality aspects. Brokken (1978, pp.51 -52) comments:

"It should be noted that the labels for the factors are only very general ones, capturing the tendency of meaning of each factor at most. Frequently, rather unexpected adjectives appear which suggests that other, aesthetically more appealing positions of the factors may be found ....... As three of the five Norman factors are presented in the factors in their current position [orthogonal varimax rotation of principal components] while another factor (Conscientiousness) is represented by the factors Dominance and Orderliness, it was decided to postpone new rotations to future research."

We shall return to the interpretational vagueness of these Norman-factors below, and discuss ways to avoid their ambiguity.

Again, all these problems and arguments point toward the replacement of orthogonal factors from a simple structure procedure by (often oblique) canonical factors, each relating to some major (set of) personality source(s) which can be experimentally manipulated and verified. The problem with this single-line recipe however, is that not enough definite, well established, external criteria can be found which render clear criterium rotations and canonical factors, and which are at the same time sufficiently acceptable as reference points for most investigators in the field of personality research. After all, having relinquished the demand for orthogonality, the above men¬tioned problem of determining enough rotation criteria has increased to power n! Therefore we would need such enormous amounts of generally agreed-upon and experimentally verifiably rotation criteria as to be virtually impossible.

Cattell's 16 PF set (Cattell et al., 1957,1970), while ultimately based also on a rather exhaustive set of trait names, is exemplary in this respect. Orthogonality was relinquished in favour of interpretational and conceptual clarity, aiming at a reduction of ambiguity of the factors itself. The wide spread tendency to stick to a limited number of orthogonal personality dimensions is criticized by Cattell as "indulging in a simple man's paradise of over-simplification". However, relinquishing orthogonality increases the degrees of freedom dramatically and puts a heavy burden on the investigator in terms of arguments about which particular axes of reference to choose from the multitude of possibilities. In his case too, the international scientific commu¬nity as a whole has in no way been convinced that the particular anchoring points in the space of personality data, as eclectically chosen by Cattell and coworkers, and resulting in their 16 PF system, are most desirable in terms of interpretative, manipulative, and experimental advantages. At present, the status of the 16 PF system is therefore "just one arbitrary personality system amongst many others" (Buss & Poley, 1976).

Because of all these ambiguities, many authors have pointed out that an approach in which external rotation criteria are obtained from experimental data seems more attractive (Armstrong, 1976; Gorsuch, 1974; Guilford & Hoepfner, 1969). Only then would we be able to avoid the Scylla and Charybdis of on the one hand the ambiguity of blended orthogonal higher-order factors and on the other the indeterminacy of too many degrees of freedom of oblique factors. To be of maximum use such experimental data should result from genetic experiments, from experiments with social roles, etc., etc., while at the same time being related to the set of personality(-descriptive) variables under consideration. In addition, such experiments should of course deal with sufficiently important and representative aspects of personality.

RECIPE: FEWER DEGREES OF FREEDOM IN SPITE OF SOME OBLIQUENESS OF FACTORS

If we could find temperamental trait factors which do have implications for what happens at the social role level, i.e. which determine biases in the distribution of social roles, then such temperamental personality factors would not only tell us something about the temperamental personality aspects, but would also, and at the same time, tell us something about the relative likeli¬hood of individuals to drift into certain social role positions. Since individual differences in social role behaviour are very much influenced by external variables, biologically based temperamental aspects can never of course fully determine what will manifest at the level of social interactive behaviour. However, the distribution of social roles is very much a complex stochastic process, depending on accidental, changing circumstances and on myriads of minor moments of choice. And it is in these moments of choice that tempera¬mental biases exert their systematic influence.

If we could find such temperamental trait factors which determine biases in social interactive behaviour, then they would represent an interface between the realm of social interactive (role-)behaviours and the realm of tempera¬mental traits. They would therefore specify functional and causal relations between specific aspects of temperament and certain specific aspects of social role differences, thus specifying oblique relations between dimensions from these two realms of personality differences. This would help us to avoid the serious interpretational confusion which is to be expected with factors that are orthogonal to one another and that are stemming from different source areas of personality variance. We would know precisely where to allow for obliqueness without having to give up the sparsity of orthogonal factors elsewhere. Using such personality factors as anchoring points for rotations would there¬fore reduce the degrees of freedom, while at the same time allowing for correlations and functional relations between specific personality factors. This strategy can be summed up in the following way:

a) Allow in principle for obliqueness if necessary, but only between personality factors from different main sources of personality variance, and particularly where experimental evidence with respect to their functional relations is available;

b) For the sake of sparsity insist on orthogonality between factors within each separate domain of personality differences (congenital traits of temperament; skills and abilities; social interactive and role behaviour); and

c) Choose the anchoring points for rotations not within the various realms of personality differences, but between them, at their interfaces.

Differences with Previous Attempts at Integration

Before continuing our search along these lines, it seems proper to point out how and why this approach differs from some other, previous, attempts to integrate the overwhelming multitude of factorial personality models and theories.

In the realm of social-role behaviour or, more generally, in the realm of interpersonal behaviour characteristics, there is a long tradition of attempts to integrate the various empirical and experimental findings (Leary, 1957; Foa, 1961,1965; Lorr & McNair, 1965; Smith-Benjamin, 1974,1979,1982; Wiggins, 1979,1982; Kiesler, 1983). One of the most interesting results of this tradition is the so called "circumplex structure". This is a two-dimensional summary model of social interactive behaviour in which dominance (or ascendancy) and love/hate (or mutual acceptedness) are generally found to be the principal axes.

The present study has a broader perspective in that it tries to create an integrated picture of these dimensions of interpersonal behavioural styles together with their relation to basic temperamental traits, processes of learning and concomitant ability-dimensions. The results of the above mentioned research tradition can however, as shall be shown below, very well be incorporated in our approach.

The emphasis in our approach on the issue of orthogonal versus oblique factors and the essential impossibility of finding satisfactory interpretations for orthogonal final factors resulting from mathematical rotation criteria such as varimax etc., is similar to the approach of Guilford (e.g., 1959, 1975, 1969 et al.). Guilford tries to overcome the oversimplified way in which factorana-lysis is ordinarily applied, by discriminating between different levels of personality factors which are hierarchically organized (like in Pawlik's (1968) integrative approach), and by developing target-criteria for procrustes rotations through which meaning and unambiguity of the final factors is maximized.

This study is different in that we first tried to explain why orthogonality of factors thwarts any functional and conceptual clarity automatically, and second, try to indicate precisely which factors should a-priori be allowed to correlate and also because of which causal relations, and which sets of factors had better be kept orthogonal.

Apart from these traditions of integrative research, there have frequently been efforts to escape from these dilemmas by introducing a completely novel classification system for personality factors. In general, the idea is to elucidate the functional relations between the various personality dimensions, having determined the - up to that moment "hidden" - qualities of each personality dimension in terms of the novel classification system. Royce (1983; see also Royce & Powell, 1983 and Powell & Royce, 1982) for instance, uses a categorization in terms of functional units (sensory-, motor-, cognitive-, affective-, value- and style-) in order to re-assess and further specify the status of established first-, second- and third-order factors of personality. The result is a model in which most personality aspects are redefined and reformulated in terms of their novel concepts, producing considerable conceptual complications in which one easily loses track of the original problems.

Our approach, in contrast, has no need of novel concepts. Instead, it first analyses why conceptual and functional clarity of factors in integrative personality models is in general so low. This is ascribed primarily to failing to take into account the effects of conditional relations between the different major classical domains of personality, congenital basic traits, abilities and skill dimensions, and the social-interactive domain. The solution is not sought in the abolishment of established categories, but in specifying how we should handle them creatively without going astray.

AN ANCHORING POINT IN THE REALM OF SOCIAL-INTERACTIVE BEHAVIOUR

In summarizing studies and reviews of social psychological research two more or less orthogonal dimensions generally emerge as the most important points of reference (Wiggins, 1979,1982; Kiesler, 1982,1983). One of these may be labeled "Ascendancy" or "Dominant versus Submissive" and the other dimension "Acceptance versus Rejection", "Love versus Hate" or "Positive Affiliation versus Hostility" (dimensions [9] and [10] in fig.l). Specifically, the following interpretations emerge in factoranalytic studies: "Dominance versus Submission" and "Love/Positive versus Hate/Negative/Hostility" (Leary, 1957; Foa, 1961; Lorr & McNair, 1965; Hare, 1972); "Assertiveness" and "Sociableness" (Borgatta, 1963); "Authority" and "Solidarity" (Gouman, Hofstee & de Raad, 1973); "Authority/Control" and "Affection/Intimacy" (Sampson, 1971); "Aggressive Dominance" and "Affiliation/Sociability" (Golding & Knudson, 1975). And these factoranalytic dimensions of social behaviour may be found on the verbal level as well as on the non-verbal level of behaviour: "Positiveness" (affiliative behaviour) and "Dominance vs. Submission" (relaxation) are two of the most conspicuous dimensions which Mehrabian (1972) found in his R-type **** factor-analytic studies on non-verbal social behaviour in man.

Peabody (1970) points at a very basic distinction between the two axes spanning this two-dimensional domain. One of them represents "asymmetrical" interactions, whereas the other describes "symmetrical" interactions. Relations involving "love/hate" or "affiliation" (dimension [10]) tend to be symmetrical - i.e., involving similar characteristics for the two parties - and relations involving "power" (dimension [9]) tend to be asymmetrical - i.e., involving dissimilar characteristics for the two parties - (see also Wiggins, 1982 and Kiesler, 1983 for recent reviews of research on this aspect).