Post

Personality Traits

Posted Updated
By Yue Lin 26 min read

This post is an exerpt of this book: Matthews, G. Personality traits. Cambridge University Press, 2003.

History

  • Allport and Odbert (1936) identified almost 18,000 English personality-relevant terms.

The trait concept and personality theory

Introduction: conceptions of traits

Everyday conceptions of traits

Everyday conceptions of personality traits make two key assumptions.

  • Traits are stable over time.
    • Most people would accept that there is a core of consistency which defines the individual’s “true” nature.
    • In other words, there are differences between individuals that are apparent across a variety of situations.
    • Stability distinguishes traits from more transient properties of the person, such as temporary mood states.
  • Trait directly influence behavior.
    • If a person spontaneously breaks into chearful song, we might “explain” the behaviour by saying that he or she has a happy disposition (a person’s natural tendency or temperament; 性情/秉性).

One of the major tasks for a scientific psychology of traits is to distinguish mental properties of the person from overt (openly observable, not hidden; 外显的) behaviours, and to investigate the causal relationship between them.

Scientific conceptions of traits

If there is to be a speciality called personality, its unique and therefore defining characteristic is traits.

Steps of developing a science of traits:

  1. The measurement and classification of traits.
    1. Verbal report
      1. self-report
      2. ask the respondent
    2. Behavior tasks in the laboratory
      1. In practice, however, personality measures based on objective behavior tests have had only limited success, and few have been validated.
      2. (Verbal report has been the preferred method of trait assessment used by personality researchers.)
    3. (What is the number of broad dimensions needed to describe the main elements of any individual personality?)
      1. Many of these (personality traits) words have rather similar meanings.
      2. Such overlapping traits can be grouped together as a broad aspect or dimension of personality.
  2. To test whether and how traits relate to behaviors.
    1. There is no guarantee that people’s self-descriptions are accurate.
    2. Traits may also be useful in applied settings, in predicting a person’s job performance, or the response of a patient to therapy, for example.
  3. Development of a satisfactory theory of personality traits.
    1. (We may be able to access people’s levels of extraversion and other traits, and show that our assessment predicts some aspects of their behavior)
    2. But in themselves these observations tell us nothing about why the personality dimension predicts behaviour.
    3. One difficulty is that personality may be represented at a variety of levels of psychological description.
    4. For example, extraversion might be associated
      1. with simple properties of the central nervous system, such as the excitability of individual neurones
      2. or with style of information processing
      3. or with acquired social knowledge and beliefs.
      4. We can only distinguish these broad possibilities by the normal, somewhat laborious scientific methods of formulating specific hypotheses and testing them rigorously against experimental and observational evidence.

Whether we can ever develop a general scientific theory of traits at all?

  • The idiographic (focusing on the uniqueness of each individual; 个案式的) approach to personality (e,g,m Lamiell, 1981) considers that all aspects of personality are fundamentally unique and idiosyncratic (peculiar or specific to an individual; 独特的/特异的) to each individual, so that no generalised theoretical statements are possible.
  • In this book, we adopt the alternative nomothetic (seeking general laws that apply across people; 普遍法则式的) apporach, wich assumes that we can arrive at general hypotheses concerning stable individual differences through the normal scientific method.
    • We cannot, ofcourse, expect such hypotheses to predict all or even most of the person’s behaviour; the uniqueness of individuals seems secure.

Causal primacy (the claim that traits are genuine causes of behaviour, not merely summary labels; 因果首要性). There is uncertainty too over the causal status of traits.

A brief history of traits

The history of traits unfolds along three threads: classical thinking, the earliest scientific work on traits, and the emergence of current models of personality.

The four humours

  • The earliest progenitors (originators/forerunners who gave rise to a line of thought; 先驱/鼻祖) of trait theory include Hippocrates (ca. 460–377 BC) and Galen of Pergamum (AD 130–200).
  • Galen’s four temperaments (still surviving in today’s English):
    • melancholic (sad/depressive type, historically linked to excess black bile; 忧郁质) — tending towards low mood
    • choleric (irritable/quick-tempered type, historically linked to yellow bile; 胆汁质) — tending toward anger
    • phlegmatic (calm/unexcitable type, historically linked to phlegm; 黏液质) — tending towards stolid (unemotional, showing little reaction; 木讷的/沉稳的) calmness
    • sanguine (cheerful/optimistic type, historically linked to blood; 多血质) — tending towards optimism and confidence
  • A balanced blend of humours produced an optimal temperament; imbalance led to physical illness or mental disturbance.
  • Kant recast the four humoral temperaments along the dimensions of feeling and activity.
  • Wundt described the four temperamental types in terms of two dimensions: strong–weak emotions versus changeable–unchangeable activity.
    • These dimensions broadly resemble the present-day dimensions of neuroticism and extraversion.
  • The aetiological (concerning the causes of diseases or conditions; 病因学的) significance of the humours did not long outlast the Middle Ages; their later veracity (truthfulness/accuracy; 真实性) is owed to shrewd (astute, sharp-witted; 精明的) psychological observation, not the classical theory.

Beginnings of the science of traits

Three ingredients were required for scientific research on traits:

  1. systematic data collection,
  2. statistical techniques for data analysis (especially correlation, and later factor analysis),
  3. development of testable theories.
  • These prerequisites became available around the beginning of the twentieth century.
  • Thurstone’s (1947) introduction of multiple factor analysis was particularly influential and began the modern research era in personality.

The first empirical studies

  • Sir Francis Galton (1884) hypothesised that individual differences in personality might be represented in natural language terms.
    • Trawled (searched thoroughly/dragged through; 翻查) Roget’s Thesaurus for character-descriptive terms.
    • Later dubbed (given the name of; 称为/冠名) the lexical hypothesis.
  • Heymans and Wiersma (1906–1909) reduced ratings of character to three factors using a method conceptually related to (but cruder than) factor analysis.
    • Eysenck (1970) identified the first dimension with emotionality, and the other two with introversion–extraversion.
  • Webb (1915) collected detailed ratings on 194 students at a teacher training college and 140 schoolboys.
    • After general intelligence was extracted, a second general factor of character emerged: persistence of motives or will.
    • Re-analyses suggest five or six factors existed in Webb’s data, similar to modern dimensions of personality (Deary, 1996).

The beginnings of trait theory

  • Allport (1937) commented:

    To use trait terms, but to use them cautiously, is, then, our lot. Nor need we fear them simply because they bear the age-long sanction (authoritative approval/endorsement; 认可) of common sense.

  • Carr and Kingsbury (1938) emphasised:
    • The predictive nature of traits.
    • Traits are not directly observable — they may only be inferred from behaviour.
    • The need for trait scales to compare individuals on a given characteristic.
  • McCrae et al. (2000) reaffirmed:

    Traits cannot be directly observed, but rather must be inferred from patterns of behaviour and experience that are known to be valid trait indicators.

  • Allport’s Personality: a Psychological Interpretation (1937) opened with the famous sentence:

    In everyday life, no one, not even a psychologist, doubts that underlying the conduct of a mature person there are characteristic dispositions or traits.

  • Allport accommodated both:
    • Common traits (the nomothetic approach emphasised in this book).
    • More idiographic traits specific to individuals.

Psychometric approaches to identifying personality dimensions

Contemporary views of traits are intimately related to the processes of measurement and assessment.

Questionnaire construction and psychometrics

  • Trait researchers begin with a hypothesis about the number and nature of principal dimensions, then design a questionnaire.
  • Items must be:
    • easily understood and unambiguous,
    • applicable to all respondents,
    • unlikely to cause offence,
    • resistant to response sets and biases (social desirability, yea-saying (tendency to agree regardless of content; 盲目附和), extreme responding).
  • Psychometrics is the science of psychological measurement; it tells us how good a measuring tool a particular questionnaire is.

Psychometrics of single scales

A trait scale must satisfy three essential, complementary criteria: reliability, stability, and validity. Reliability sets a ceiling on validity — a noisy scale cannot correlate strongly with anything — but a highly reliable scale can still measure the wrong construct. The three criteria therefore have to be established jointly.

The underlying framework is classical test theory (CTT; 经典测验理论):

\[X = T + e\]

where the observed score $X$ is the sum of a true score $T$ and a random error $e$. Reliability is the proportion of observed-score variance attributable to true score:

\[r_{xx} = \frac{\mathrm{Var}(T)}{\mathrm{Var}(X)}\]
  1. Reliability — accuracy/consistency of measurement (how much of the score is signal versus error).
    • Internal consistency — do the items within a scale measure the same thing?
      • Cronbach’s alpha (α; 克隆巴赫α系数) — roughly the average correlation across all possible split-halves; a function of item intercorrelations and scale length.
      • Conventional thresholds: α ≥ 0.70 adequate for research, ≥ 0.80 good, ≥ 0.90 excellent — but α > 0.95 may indicate item redundancy (items asking essentially the same question; 题目冗余).
      • Caveats: α is inflated by scale length and does not guarantee unidimensionality (all items loading on a single factor; 单维性); it is best interpreted alongside a factor analysis.
    • Split-half reliability — correlation between two halves of the scale, corrected by the Spearman–Brown prophecy formula (predicts reliability when scale length changes; 斯皮尔曼–布朗校正公式).
    • Parallel-forms (alternate-forms) reliability — correlation between two independently constructed versions of the same trait measure; demanding, but the cleanest index.
    • Inter-rater reliability — agreement between different observers on rating scales, indexed by Cohen’s kappa (κ; 科恩卡帕) for categorical ratings or the intraclass correlation coefficient (ICC; 组内相关系数) for continuous ratings.
    • Standard error of measurement (SEM; 测量标准误) = $\sigma_X \sqrt{1 - r_{xx}}$; defines the confidence band around an individual’s observed score and is the practical face of reliability.
  2. Stability — test–retest correlation (同一批被试在两个时点作答的相关; 重测相关) over a specified time interval.
    • Distinguishes a genuine trait from a transient mood state or situational response.
    • Short-term stability (days to weeks) largely reflects measurement precision and is often hard to separate from internal consistency.
    • Long-term stability (months to years) reflects the persistence of the underlying disposition; Big Five dimensions typically show stability coefficients of roughly 0.60–0.80 across several years in adulthood, with rank-order stability (relative ordering of individuals; 等级稳定性) peaking in midlife.
    • A low stability coefficient is ambiguous — it may indicate (a) that the construct is genuinely state-like, (b) real developmental change, or (c) poor measurement.
    • Distinguish differential / rank-order stability (do people keep their relative positions?) from mean-level stability (does the group average stay the same?; 均值稳定性); a trait can be rank-order stable while its mean still shifts with age.
  3. Validity — whether the measure actually assesses what it purports (claims/professes; 声称/宣称) to assess.
    • Face validity (表面效度) — does the scale look as if it measures the intended trait? Useful for respondent acceptance, but weak scientific evidence and easily gamed.
    • Content validity (内容效度) — adequate coverage of the theoretical domain; assessed by expert judgement of item sampling, not by statistics.
    • Criterion validity (效标效度) — correlations with external criteria.
      • Concurrent validity (同时效度) — criterion measured at the same time (e.g., clinical diagnosis, peer ratings, group membership).
      • Predictive validity (预测效度) — criterion measured later (e.g., job performance, therapeutic response, academic attainment).
    • Construct validity (构念效度) — whether the scale genuinely measures the intended theoretical construct; external correlations are predicted in advance from theory, not merely observed. The ultimate goal of trait research.
      • Convergent validity (聚合效度) — high correlation with other measures of the same construct (ideally via different methods).
      • Discriminant validity (区分效度) — low correlation with measures of unrelated constructs.
      • Convergent and discriminant evidence are classically organised in the multitrait–multimethod (MTMM) matrix (Campbell and Fiske, 1959; 多特质-多方法矩阵).
      • Construct validity arises from the nomological network (the web of theoretical and empirical links that give a construct its scientific meaning; 法则关系网) of empirical and theoretical analysis around a trait (Cronbach and Meehl, 1955; Eysenck, 1981).
      • It is always somewhat provisional (temporary, subject to revision; 暂定的/临时的) and may be reduced or enhanced by fresh research.
    • Incremental validity (增量效度) — does the scale predict the criterion beyond existing, established measures? Particularly important for defending a new scale against the charge of mere relabelling.
Threats to psychometric quality
  • Response biases (作答偏差) can simultaneously inflate reliability and distort validity:
    • Social desirability (社会赞许性) — responding in a way that is seen favourably by others; the EPQ-R Lie scale was designed to detect this.
    • Acquiescence / yea-saying (盲目附和, tendency to agree with items regardless of content) — controlled by including reverse-keyed items.
    • Extreme responding (极端反应偏差) — overuse of the endpoints of a Likert scale; varies by culture.
    • Midpoint bias (中间点偏差) — overuse of the neutral category, often reflecting disengagement or ambiguous items.
  • Item-level quality indicators:
    • Item–total correlation (题总相关) — correlation between an item and the remainder of the scale (corrected for overlap); items below about 0.30 are candidates for removal.
    • Item difficulty / endorsement rate (题目难度 / 通过率) — mean endorsement; items endorsed by near 0 or 100 per cent of respondents carry little discriminating information.
    • Item discrimination (题目区分度) — how well an item separates high-scorers from low-scorers on the underlying trait.
  • Reliability caps validity: the observed correlation between two measures cannot exceed $\sqrt{r_{xx} \cdot r_{yy}}$, a relation known as the correction for attenuation (测量误差衰减校正). Two scales each with reliability 0.70 can correlate at most about 0.70 — so poor reliability alone can make a real construct relation look weak.
  • Contemporary Item Response Theory (IRT; 项目反应理论) and Generalizability Theory (G-theory; 概化理论) extend CTT by modelling item-level probabilities and multiple sources of error (raters, occasions, items) simultaneously; they are increasingly standard in modern personality scale construction but do not overturn the three core criteria above.

Psychometrics of multiple traits: factor analysis

  • A satisfactory model of personality cannot be obtained simply by accumulating different traits — many will be correlated.
  • Factor analysis identifies underlying dimensions or factors that account for most of the variation in individuals’ item scores.
  • Factors are defined by the items which correlate with or “load” on them.
  • Two stages:
    1. Initial extraction — the first factor explains as much variance as possible, then the next, etc.
    2. Rotation — re-computes the factor matrix to give the most interpretable solution (the principle of simple structure (each variable loads clearly on one factor only; 简单结构)).
  • (In personality research, grabbing the most variance for each successive factor does not usually give psychologically meaningful results — unlike ability research where the first factor approximates g.)

Limitations of factor analysis

Three questions should always be asked:

  1. Are the data actually suitable for factor analysis?
    • Correlation does not represent non-linear relationships validly.
    • Sample sizes need to be large.
  2. How much do results depend on the methods used?
    • Choice of orthogonal (factors kept uncorrelated/independent; 正交的) vs oblique (factors allowed to correlate; 斜交的) rotation.
    • Number of factors extracted (no definitive rule).
  3. What do the results actually mean?
    • Factor analysis indicates structural relationships only — construct validity must be established by relating factorial measures to external criteria.

Further techniques of factor analysis

  • Confirmatory factor analysis (Jöreskog, 1973): tests whether data fit a hypothesised factor structure; part of structural modelling.
  • If oblique rotation is used and factors are correlated, a further factor analysis can identify second-order or higher-order factors.
    • In personality research this yields broader, secondary traits such as extraversion and neuroticism.

Primary factors of personality: the 16PF and other questionnaires

The Sixteen Personality Factor Questionnaire (16PF)

  • Cattell’s project sought to explain individual differences in every area of life — psychometrically sound measures of ability, motivation, personality and mood.
  • Cattell began with the lexicon (the vocabulary/word-stock of a language; 词库) of trait-descriptive words, then shifted to questionnaire items, eventually identifying twenty-three fundamental primary factors (one being general intelligence).
  • The sixteen most robust dimensions are measured by the 16PF (Cattell, Eber, and Tatsuoka, 1970).
  • Concerns:
    • Internal consistencies of some scales were low.
    • Several investigators were unable to recover the Cattellian primary factors from factor analysis of the 16PF.
  • The latest version (16PF5: Conn and Rieke, 1994) features improved internal consistency (mean alpha 0.74), but at the cost of comparability with previous versions.
  • The 16PF has a hierarchical factor structure — secondary factors may be derived from intercorrelations of the sixteen primary factors, with some correspondence to the Big Five.
  • Predictive validity is well demonstrated; construct validity is more doubtful.

Other systems of primary factors

  • California Psychological Inventory (CPI): assesses eighteen traits using criterion-keying (selecting items by their ability to discriminate specific criterion groups; 效标关键法) rather than factor analysis; risks poor construct validity.
  • Occupational Personality Questionnaire (OPQ): measures thirty-one traits relevant to personnel selection; re-analysis suggests about twenty dimensions are identifiable, with good correspondence to Saville et al.’s (1984) hypothesised traits.

Higher-order factors: the ‘Big Five’ or the ‘Gigantic Three’?

Two prominent personality schemes advocate higher-order secondary factors describing personality in broad, abstract terms. Within these schemes, each dimension may relate to hundreds of basic trait terms.

H. J. Eysenck’s three factor model

  • Three broad personality factors: neuroticism (N), extraversion–introversion (E), and psychoticism (P).
  • Assessed via the Eysenck Personality Questionnaire-Revised (EPQ-R: Eysenck and Eysenck, 1991), which also contains a Lie scale.
  • Distributions:
    • N and E approximate a normal curve.
    • P is markedly skewed (asymmetric, leaning to one side of the distribution; 偏斜的) towards low scores.
  • Trait portraits:
    • Extravert: sociable, craves excitement, takes chances, fond of practical jokes, not always reliable, can lose temper.
    • Introvert: quiet, retiring (reserved, keeping to oneself; 内敛的), fond of books rather than people, serious, reliable, has high ethical standards.
    • High N: tends towards anxiety and depression, worries, has bad sleep and psychosomatic (bodily symptoms caused or worsened by psychological factors; 心身的) disorders, is preoccupied with things that might go wrong.
    • High P: solitary, often troublesome, sometimes cruel, unempathetic, aggressive, has unusual tastes; overlaps with schizoid (socially detached, emotionally restricted; 分裂样的) and antisocial personality disorders.
  • Eysenck emphasised the importance of the nomological network in which a dimension is embedded — psychometric, biological, cultural, behavioural, and clinical.
  • Major contributions include theories of the biological bases of personality dimensions.
  • Ironically, despite his antipathy (strong dislike/opposition; 反感) to psychoanalysis, Eysenck’s scheme contains terms partly attributable to Jung (introversion–extraversion) and Freud (superego).

Costa and McCrae’s five factor model

  • Five broad domains, each composed of six lower-level facets (narrower sub-traits that together make up a broad domain; 侧面/小面), measured by the NEO-PI-R (240 items, 48 per dimension):
    • Neuroticism (N): anxiety, angry hostility, depression, self-consciousness, impulsiveness, vulnerability.
    • Extraversion (E): warmth, gregariousness (liking to be with others; 合群性), assertiveness, activity, excitement seeking, positive emotions.
    • Openness (O): fantasy, aesthetics, feelings, actions, ideas, values.
    • Agreeableness (A): trust, straightforwardness, altruism (selfless concern for others’ welfare; 利他), compliance, modesty, tender-mindedness (emotional sensitivity to others’ suffering; 多愁善感/心软).
    • Conscientiousness (C): competence, order, dutifulness (strong sense of moral obligation; 尽责), achievement striving, self-discipline, deliberation (carefulness before acting; 审慎).
  • Development was driven partly by rational and partly by statistical concerns.
    • Block (1995): N and E arose from Cattellian analyses; O was built from embryonic (in an early, undeveloped form; 萌芽的) status; C and A were “grafted (attached/inserted from outside, like joining a plant cutting; 嫁接)” on from lexical results.
  • About half of the common variance in most personality inventories can be accounted for by the five factor model.
  • McCrae et al. (2000): “personality traits are more expressions of human biology than products of life experience.”
  • Caveat (important warning/qualification; 警示) — personality inventories are not personality theories. The tests should be treated as best attempts at the three- and five-factor models, not as the dimensions themselves.

Current conceptions of personality structure

The differences between the three- and five-factor models are probably the most significant disagreement in trait psychology — but important differences between schemes are often more apparent than real.

  • The Cattellian sixteen dimensions are not relevant to overview discussions, since they reduce to a smaller number of orthogonal higher-order dimensions.
  • Costa and McCrae (1993):

    The five factor model has provided a unified framework for trait research; it is the Christmas tree on which the findings of stability, heritability (proportion of trait variance attributable to genes; 遗传度), consensual validation, cross-cultural invariance (remaining the same across cultures; 不变性) and predictive utility are hung like ornaments.

  • De Raad and Perugini (2002):

    The Big Five model has acquired the status of a reference model … its five main constructs capture so much of the subject matter of personality psychology.

Similar five factor solutions have arrived from a number of disparate (very different, unrelated in kind; 迥异的/各自不同的) sources.

The consensus from the lexical approach

Key premises of the lexical perspective (Saucier and Goldberg, 2001):

  1. Personality language refers to phenotypes (observable characteristics; 表型), not genotypes (underlying genetic makeup; 基因型).
  2. Important phenotypic attributes become encoded in natural language.
  3. The degree of representation of an attribute in language corresponds to its general importance.
  4. The lexical perspective provides a strong rationale for selecting personality variables.
  5. Person-description and language sedimentation (the gradual depositing of important distinctions into vocabulary; 积淀) work primarily through the adjective function.
  6. Phrase- and sentence-based descriptions closely match single-word descriptions.
  7. The lexical perspective is particularly germane (relevant, closely related; 贴切的) to personality, yet not in others.
  8. The most important dimensions in aggregated personality judgements are the most invariant and universal — replicating across samples, raters, analytic procedures, and languages.
  • Tupes and Christal (1961) found five robust factors across eight different samples — little affected by sample, situation, rater, or rater knowledge.
  • Re-analyses of Cattell’s data (Fiske, 1949; Digman and Takemoto-Chock, 1981) and peer ratings (Norman, 1963) confirmed five similar factors.
  • Goldberg (1990): English trait adjectives in self- or peer descriptions consistently elicit (evoke/draw out a response or result; 引出) a variant of the Big Five — broadly relating to Power, Love, Work, Affect, and Intellect.
  • Cross-language replication: similar factors emerge in Italian, Polish, Hungarian, German, etc.
  • Emic (insider/native-culture perspective; 主位的) vs etic (outsider/cross-cultural perspective; 客位的) approaches (Saucier and Goldberg, 2001):
    • Emic: native descriptors found in each language → a “big three” (agreeableness, extraversion, conscientiousness) emerges from a wider range of languages.
    • Etic: imports structures via translations of questionnaires → a “big five” tends to emerge in Anglo-Germanic studies.
  • Webb’s (1915) pioneering data, when re-analysed, also yields six factors that match present-day schemes (Deary, 1996).
  • Goldberg’s team provides public domain items in the International Personality Item Pool (http://www.ipip.ori.org/ipip).

The consensus from questionnaire studies

  • Joint factor analyses of two or more questionnaires on the same sample show large overlap with the five factor model.
  • The NEO-PI-R correlates with the Guildford-Zimmerman Temperament Survey, MMPI, Revised CPI, and many others.
  • Krug and Johns (1986): a large 16PF study found five second-order factors — Extraversion, Neuroticism, Tough Poise, Independence, and Control.
  • Hofer and Eber (2002):

    Global factors extracted at the second-order level of the 16PF Questionnaire are highly similar to factors known as the Big Five.

  • 16PF–NEO-PI-R correspondences (large correlations):
    • Extraversion vs Introversion = 0.65
    • Anxiety vs Neuroticism = 0.75
    • Tough-mindedness vs Openness = 0.56
    • Self-control vs Conscientiousness = 0.66
  • Independence and Agreeableness do not map cleanly onto each other.

Remaining doubts: psychometric and theoretical issues

Costa and McCrae’s “four ways the five factors are basic”:

  1. Longitudinal (tracking the same people over time; 纵向的) and cross-sectional (comparing different groups at one time point; 横断的) studies show five robust enduring behavioural dispositions.
  2. Traits associated with the five factors emerge from different personality systems and natural language.
  3. The five factors are found across age, sex, race, and language groups.
  4. Heritability studies demonstrate biological bases for each factor.

Psychometric criticisms:

  1. Big-Five-like factors from different investigators are not necessarily equivalent.
    • Goldberg (1992): correlations between supposedly equivalent measures range 0.46–0.69 — markedly lower than parallel-form standards.
    • The lowest (0.46) involves Openness, the most contested factor (also called intellect, culture, or imagination).
  2. Five broad trait factors may be insufficient.
    • Zuckerman’s Alternate Five: replaces openness with activity, adds aggression–hostility (low A) and impulsive sensation seeking (low C).
    • Hogan’s Hogan Personality Inventory replaces extraversion with sociability and ambition.
    • Brand: with intelligence included, there should be six factors — Neuroticism, Energy (like E), Conscientiousness, Affection, Will.
    • Big Seven models include factors of positive and negative valence (emotional value/charge, positive vs negative; 效价).
  3. Five factors may be too many.
    • Eysenck (1991): Agreeableness and Conscientiousness are facets of his higher-order Psychoticism.
    • Openness may form part of Extraversion; low Conscientiousness part of Neuroticism.

Theoretical criticisms:

  • Block (1995): pre-structuring of data sets may artificially yield five factors.
    • But Webb’s data and Saucier and Ostendorf’s studies, which avoided prestructuring, still produced Big-Five-like solutions.
  • Eysenck argued that the five factor model lacks a nomological or theoretical network and is therefore arbitrary (based on choice rather than principle or reason; 随意/武断的); his psychoticism dimension is by contrast rooted in mental illness phenomena.
  • The deeper contrast:
    • Eysenck’s reductionist (explaining complex phenomena by reducing them to simpler/lower-level components; 还原论的) scheme treats traits as expressions of partly heritable nervous system variance.
    • Five-factor proponents (advocates/supporters of a view; 支持者/倡导者) treat them as descriptive phenotypes (with growing acknowledgement that they may also be genetically influenced indicators).
  • Cloninger (1987) proposed brain systems — novelty seeking, harm avoidance, reward dependence — measured by the Tridimensional Personality Questionnaire.
  • Pervin (1994): traits may be merely descriptive rather than explanatory; this is a fundamental challenge addressed only by genetic, physiological, and process-level evidence.

Conclusions

  1. Trait terms abound (exist in great quantity, are plentiful; 比比皆是/充斥) in everyday language. People use them to differentiate styles of behaviour. There is, however, a difference between lay (non-expert/by ordinary people; 外行的/通俗的)/pre-science conceptions and a science of traits.
  2. The science of personality traits is contained mostly within the twentieth century, marked by:
    • the growth of psychometric techniques that support deriving and validating traits;
    • the survival of trait and cognitive-behavioural approaches as the viable scientific routes to study personality;
    • the converging consensus around a relatively small number of broad personality domains.
  3. Understanding traits requires psychometrics — correlation and factor analyses are the everyday tools of the trait-oriented personality psychologist.
    • Correlation was available at the start of the twentieth century, multiple factor analysis emerged in the first half, and confirmatory factor analytic techniques emerged in the later decades.
  4. Trait systems exist at the primary and broader levels — broader traits are often called dimensions or domains.
    • The most influential model of the last two decades is the five factor model: neuroticism, extraversion, openness/intellect, agreeableness, conscientiousness.
    • There is no single five factor model; lexical and questionnaire-based versions vary somewhat.
    • Most personality theories and instruments overlap substantially with the Big Five.
  5. Personality trait systems are descriptions of phenotypes. Validating these systems requires finding out the causes and consequences of personality traits.

Further reading

  • De Raad, B. and Perugini, M. (2002). Big Five Assessment. Seattle, WA: Hogrefe and Huber.
  • Saucier, G. and Goldberg, L. R. (2001). Lexical studies of indigenous personality factors: premises, products and prospects. Journal of Personality, 69, 847–79.
Interdisciplinarity, Psychology
Life Interdisciplinarity Psychology Personality
This post is licensed under CC BY 4.0 by the author.