Personality Traits, Types, and Disorders: An Examination of the Relationship Between Three Self-Report Measures

Transcription

1 European Journal of Personality Eur. J. Pers. 19: (2005) Published online in Wiley InterScience ( DOI: /per.543 Personality Traits, Types, and Disorders: An Examination of the Relationship Between Three Self-Report Measures ADRIAN FURNHAM* and JOHN CRUMP Department of Psychology, University College London, UK Abstract As part of an assessment centre 431 candidates completed three self-report measures: one of personality disorders (Hogan Development Survey, HDS; Hogan & Hogan, 1997), one of personality traits (NEO-PI, Costa & McCrae, 1992), and one of personality type (MBTI; Briggs & Myers, 1987). Correlational and regressional analysis tested various hypotheses about the overlap between the different dimensions and confirmed previous research using different instruments (Saulsman & Page, 2004). Results revealed highest correlation between the HDS and NEO, showing neuroticism correlating (as predicted) with excitable (borderline) and cautious (avoidant); introversion correlating with avoidant (cautious), schizoid (detached), and (negatively) with colourful (histrionic); openness correlating with schizotypal (imaginative) and conscientiousness with diligent (obsessive compulsive). Many of the overlaps were suggested by Widiger, Trull, Clarkin, Sanderson, and Costa (2002). The overlap and lack of overlap is considered at the psychometric and conceptual level. Copyright # 2005 John Wiley & Sons, Ltd. Key words: Big Five; personality disorders; Jungian; MBTI Trait psychologists have put considerable research effort into establishing what the higher order super-factors of personality are, and the mechanics and processes underlying them (Matthews & Deary, 1998). There are also older theories and tests which conceptualize personality in terms of types rather than traits (Briggs & Myers, 1987). Psychiatrists talk about personality disorders that are typified by early onset (recognizable in children and adolescents), pervasive effects (on all aspects of life), and with relatively poor prognosis (that are difficult to cure) (Millon et al., 1996). This study focuses on the correlation of trait, type, and disorder measures in a normal adult population using a personality disorder instrument designed specifically to be used in the selection, development, and counselling of business leaders (Hogan & Hogan, 1997, 2001). The paper responds directly to the call by De Fruyt and Salgado (2003) for dark side studies to be used more in research in industrial, work, and organizational (IWO) psychology. *Correspondence to: Professor Adrian Furnham, Department of Psychology, University College London, 26 Bedford Way, London WC1H OAP, UK. a.furnham@ucl.ac.uk Received 18 December 2003 Copyright # 2005 John Wiley & Sons, Ltd. Accepted 15 October 2004

2 168 A. Furnham and J. Crump There remains debate, however, concerning whether personality disorders are qualitatively different from normal behaviour or whether the behaviours associated with each specific disorder should be considered simply as an exaggerated or extreme form of normal behaviour (Clark & Watson, 1999). Further it has been suggested that some forms of personality disorder (histrionic, narcissistic, compulsive) may even be socially acceptable and rewarded in business and society at large (Babiak, 1995; Millon, 1981). Thus successful business people maybe characterized by egocentrism, superficial charm, and manipulativeness (histrionic); exploitative, grandiose, and unempathic behaviour (narcissistic); and dictatorial, rigid, and stubborn (compulsive) characteristics. There are various self-report measures available to assess personality disorders (Kaye & Shea, 2000; Morey, Waugh, & Blashfield, 1985; Widiger & Coker, 2001). This study used the Hogan dark side measure now extensively used in IWO research and practice to measure personality disorders in the normal population (Hogan and Hogan, 1997). In the area of personality structure, many researchers (e.g. Busato, Prins, Elshout, & Hamaker, 1999; Digman, 1990) have agreed on the psychometrical advantages of the Five Factor Model (FFM) proposed by Costa and McCrae (1992). It is recognized as a viable model of general personality functioning, though some researchers cling on to older measures (Myers & McCaulley, 1985). According to the FFM, there are five higher-order personality traits (or factors), namely Neuroticism, Extraversion, Openness to Experience, Agreeableness, and Conscientiousness. Although the increasing use of the model and measure of Costa and McCrae s (1992) would not totally justify the omission of other personality models (or traits) (see Brand, 1994), employing the same assessment instrument is useful to establish comparisons between studies. This study uses the Costa and McCrae (1992) measure. Psychiatrists and psychologists share some simple assumptions with respect to personality. Both argue for the stability of personality over time and across situations. The DSM criteria talk of enduring pattern, inflexible and pervasive stable and of long duration. The pattern of behaviour is not a function of drug usage or some other medical condition. The personality pattern furthermore is not a manifestation or consequence of another mental disorder. Both argue that the personality factors are related to cognitive, affective and social aspects of functioning. The disorder or trait affect how people think, feel, and act. It is where a person s behaviour deviates, markedly from the expectations of an individual s culture that the disorder is manifest. The psychiatric manual is very clear that odd behaviour is not simply an expression of habits, customs, religious, or political values professed or shown by a people of a particular cultural origin. The DSM manuals note that personality disorders all have a long history and have an onset no later than early adulthood. Moreover, there are some gender differences: thus the anti-social disorder is more likely to be diagnosed in men while the borderline, histrionic, and dependent personality is more likely to be found in women. The manuals are at lengths to point out that some of the personality disorders look like other disorder, anxiety, mood, psychotic, substance-related, etc., but have unique features. The essence of the argument is Personality Disorders must be distinguished from personality traits that do not reach the threshold for a Personality Disorder. Personality traits are diagnosed as a Personality Disorder only when they are inflexible, maladaptive and persisting and cause significant functional impairment or subjective distress (DSM.IV, 1994, p. 633).

3 Personality disorders, traits, and types 169 Various authors have written books for lay people trying to educate them in the concept of the disorders and differentiate them from normal styles/traits (Oldham & Morris, 1991). One of the most important ways to differentiate personal style from personality disorder is that behaviours associated with disorders are self-defeating. Personality disorders influence the sense of self the way people think and feel about themselves and how other people see them. The disorders often powerfully influence interpersonal relations at work. They reveal themselves in how people complete tasks, take and/or give orders, make decisions, plan, handle external and internal demands, take or give criticism, obey rules, take and delegate responsibility, and co-operate with people (Oldham & Morris, 1991, p. 24). The anti-social, obsessive, compulsive, passive aggressive, and dependent types are particularly problematic in the work-place. There have been numerous attempts to relate the two worlds of traits and disorders (Dyce, 1997) from the two disciplines of psychology and psychiatry. Indeed Wiggins and Pincus (1989) noted that the personality disorders are readily interpretable (p. 314) within the Big Five framework. Perhaps the most comprehensive comparative attempt has been made by Widiger et al. (Widiger, Costa, & McCrae, 2001; Widiger et al., 2002), who believed that having extreme (high or low) scores on personality traits renders individuals at risk for certain disorders. Further, a recent edited book examined the relationship between personality disorders and the FFM (Costa & Widiger, 2002). More recently various empirical studies have looked at personality disorder functioning from a FFM perspective. De Clercq and De Fruyt (2003) tested adolescents to see whether five factor facet variables (derived from the NEO-PI-R) could predict personality disorder symptoms. Their results were similar to those from adults, which showed that one could fairly easily distinguish between adaptive and maladaptive symptoms. Recently, Saulsman and Page (2004) meta-analysed 15 independent samples to investigate how personality disorders are respectively different from, and similar to, underlying personality traits. Predictably, Neuroticism correlated positively and Agreeableness negatively with many of the disorders. Further, Extraversion was strongly but directionally variably associated with many dimensions. Whilst nearly all these studies used the NEO to measure the Big Five personality traits, a wide variety of personality disorders was used, the most common of which was one of the Millon measures (MCMI-I, II, III). De Clercq and De Fruyt (2003), however, used the APDP-IV measure developed by Schotte and De Doncker (1994). This study uses a quite different measure of personality disorders to that used before and a central question is the amount of replication in the findings. It should be noted that traditionally personality traits are measured at two levels: the super-factor or domain level as well as primary or facet level. Thus in the Costa McCrae (1992) measures the big five superfactors can also be broken down and interpreted at the primary factor level. Each of the five factors has six facets. Analysing personality at the lower level offers a richer and more detailed description of personality functioning. Widiger et al. (2002) compared the personality disorders with the FFM traits at the primary factor facet level. Thus they specified that only one of the six facets of Agreeableness, namely low compliance, would be related to obsessive compulsive disorders. It is therefore quite possible that the total score on the superfactor Agreeableness will be influenced much more by the scores on the other five facets of Agreeableness rather than affected by the one stipulated primary, facet factor. Table 1, relating the personality disorders to the FFM at the primary factor or facet level, may serve as an excellent source of hypotheses. However, one problem with investigating

4 170 A. Furnham and J. Crump Table 1. Overlap between personality types and traits (Widiger et al., 2002) 1 Paranoid: They score low on agreeableness (particularly low trust) and straightforwardness. They also score high on facets of neuroticism, particularly angry hostility. They also are cold and anti-social (introverts) and rather closed rather than being open to experience. 2 Schizoid: They are strongly introverted: loners, isolated, withdrawn, with little interest in, or ability to initiate and maintain, social relationships. 3 Schizotypal: They too are introverted but can manifest fairly strong neuroticism traits. However, they tend to score high on openness, which reflects their association with creativity. This condition is particularly associated with self-consciousness, vulnerability, and a rich fantasy life. 4 Anti-Social: They are low on agreeableness and conscientiousness, being exploitative, vengeful and antagonistic. They have a mixed profile on neuroticism, being high on hostility but low on self-consciousness. 5 Borderline: They are essentially unstable, having high scores in most neuroticism facets, particularly hostility, impulsivity, vulnerability, depression, and anxiety. They are hot tempered, often apprehensive, and easily rattled. They are characterized by vulnerability to stress, impulsivity, dyscontrol, and negative emotionality. 6 Histrionic: There are extreme extraverts: convivial, assertive, energetic, flashy, and high spirited. They express emotions with inappropriate exaggeration and display inappropriate affection, intimacy, and seductiveness. They may also be low in self-discipline. 7 Narcissistic: They tend to score low on agreeableness and low on neuroticism. They are suspicious and manipulative, however, despite low self-consciousness, hostility, and depression. Their conscientiousness scores can be very low. 8 Avoidant: They are clearly introverted neurotics. They are anxious, timid, and insecure; easily rattled and panicked; apprehensive and prone to feelings of embarrassment and inferiority. They probably have low openness scores. 9 Dependent: They tend to score both high on agreeableness and neuroticism. Their pathological agreeableness makes them self-effacing, docile, submissive, and sacrificial. They may describe themselves as being low in competence and dutifulness, which makes them look low in conscientiousness. 10 Obsessive Compulsive: These perfectionistic, over-conscientious people tend to be preoccupied with details and order and often excessively devoted to productive work. But they can be very fearful of making mistakes. They can also be rather antagonistic: low on compliance and altruism, insisting that other follow orders, and stubborn. They are thus high on conscientiousness and neuroticism but low on agreeableness. 11 Passive Aggressive/Negativistic: They tend to be low on both agreeableness and conscientiousness. They can also said to be sullen, complaining, stubborn, irritable, and disgruntled. They may also be high on certain features of neuroticism like hostility. 12 Self-Defeating/Depressive: These are the neurotics with low conscientiousness scores. They feel inadequate, pessimistic, and worthless and are as a result self-blaming, self-critical, and brooding. They fail to finish tasks and choose situations that may lead to failure. May also have low agreeableness. 13 Sadistic: They score very low on agreeableness, but also high on extraversion and often low on conscientiousness. They are characterized by their tendency to harm, humiliate, intimidate, and act aggressively to others. They are ruthless, domineering, and brutal with few signs of warmth, gregariousness, or positive emotions. the overlap between traits and disorders lies in the measurement of the latter, which is traditionally done by psychiatric interview. However, the development of a psychometrically valid, self-report measure, the Hogan Development Survey (HDS: Hogan & Hogan, 1997), has made research much easier. Rolland and De Fruyt (2003), in a study of 130 French military, found that personality disorders (labelled maladaptive traits) did not predict negative affect beyond the Big Five personality traits, suggesting that they have no incremental validity in predicting affective

5 Personality disorders, traits, and types 171 (military) work outcomes. The Hogan Development Survey (HDS) was explicitly based on the DSM Axis II personality disorder descriptions, but it was not developed for the assessment of all DSM disorders. The HDS focuses only on the core construct of each disorder from a dimensional perspective (Hogan & Hogan, 2001, p. 41). An overview of the item selection guidelines can be found in Hogan and Hogan (2001). The HDS has been cross-validated with the MMPI personality disorder scales. Correlations (n ¼ 140) range from 0.45 for Antisocial to 0.67 for Borderline (Hogan & Hogan, 2001). Fico, Hogan, and Hogan (2000) report coefficient alphas between 0.50 and 0.70, with an average of 0.64 and test retest reliabilities (n ¼ 60) over a three-month interval ranging from 0.50 to 0.80, with an average of There were no mean-level differences between sexes, racial/ethnic groups, or younger versus older persons (Hogan & Hogan, 2001). This study will use the HDS to measure the personality disorders. Whilst it is less well known and used in psychiatric circles, it is now widely used in IWO research and practice. This study focuses on the relationship between the psychiatric personality disorders and the measures of normal traits. The major focus is on the relationships between the FFM dimensions and the disorders as discussed by Widiger et al. (2001, 2002). In addition, the Myers Briggs Type Indicator (MBTI) (Myers & McCaulley, 1985) will be used, which is probably the most famous and frequently used test of personality in applied psychology today (Pittenger, 1993). It has recently been described as the most widely used non-clinical measure of personality in the world (Bayne, 2003). It is popularly used for assessment, counselling, and development despite obvious theoretical and psychometric shortcomings (McCrae & Costa, 1989). According to McCrae and Costa (1989), the MBTI is unusual among personality assessment devices for three reasons: it is based on a classic theory, it purports to measure types rather than traits or continuous variables, and it is widely used to explain individuals personality characteristics not only to professionals but also to the individuals themselves and their co-workers, friends, and families. However, they also point out its limitations: the original Jungian concepts are distorted and even contradicted; there is no bi-modal distribution of preference scores; and studies using the MBTI have not always confirmed either the theory or the measure s validity. Various criticisms of the typology theory are also cogently put by Hicks (1984). A consistent criticism of the MBTI, which is an ipsative measure, is that this response format distorts data, which causes serious psychometric shortcomings. Yet DeVito (1985) has described the MBTI as probably the most widely used instrument for nonpsychiatric populations in the area of clinical, counselling and personality testing (p. 1030). Recent studies have shown considerable overlap with the FFM (Furnham, 1996). However, the MBTI does not directly measure neuroticism (a factor implicated in its popularity) and hence it is expected to show less overlap with the personality disorders. The aim of this study was to examine the inter-correlations and conceptual overlap between these three measures frequently used in assessment exercises. Table 1 may serve as a list of testable hypotheses for overlap between the measures of traits and disorders. METHOD Participants There were 858 participants, of whom 743 were male. The total population completed the HDS and the NEO but only 431 participants completed all three measures, of whom 365

6 172 A. Furnham and J. Crump were males and 66 were females. Most (95.9 per cent) were British but all had a good command of English. In all, 67.7 per cent were senior managers, 25.3 per cent manager/ supervisors, and 8.1 per cent specialists. They all participated in an assessment centre sponsored by their organization. Table 2. Overlapping themes from HDS and DSM-IV Axis 2 personality disorders DSM-IV personality disorder HDS themes Borderline Paranoid Avoidant Schizoid Passive Aggressive* Narcissistic Antisocial Histrionic Schizotypal Obsessive Compulsive Dependent Inappropriate anger; unstable and intense relationships alternating between idealization and devaluation. Distrustful and suspicious of others; motives are interpreted as malevolent. Social inhibition; feelings of inadequacy and hypersensitivity to criticism or rejection. Emotional coldness and detachment from social relationships; indifferent to praise and criticism. Passive resistance to adequate social and occupational performance; irritated when asked to do something he/she does not want to. Arrogant and haughty behaviours or attitudes; grandiose sense of self-importance and entitlement. Disregard for the truth; impulsivity and failure to plan ahead; failure to conform with social norms. Excitable Sceptical Cautious Reserved Leisurely Bold Mischievous Excessive emotionality and Colourful attention seeking; self-dramatizing, theatrical and exaggerated emotional expression. Odd beliefs or magical thinking; Imaginative behaviour or speech that is odd, eccentric, or peculiar. Preoccupations with orderliness; Diligent rules, perfectionism and control; over-conscientious and inflexible. Difficulty making everyday Dutiful decisions without excessive advice and reassurance; difficulty expressing disagreement out of fear of loss of support or approval. Moody and hard to please; intense but short-lived enthusiasm for people, projects, or things. Cynical, distrustful, and doubting others true intentions. Reluctant to take risks for fear of being rejected or negatively evaluated. Aloof, detached and uncommunicative; lacking interest in or awareness of the feelings of others. Independent; ignoring people s requests and becoming irritated or argumentative if they persist. Unusually self-confident; feelings of grandiosity and entitlement; over-valuation of one s capabilities. Enjoying risk taking and testing the limits; needing excitement; manipulative, deceitful, cunning and exploitive. Expressive, animated and dramatic; wanting to be noticed and needing to be the centre of attention. Acting and thinking in creative and sometimes odd or unusual ways. Meticulous, precise and perfectionistic, inflexible about rules and procedures; critical of others performance. Eager to please and reliant on others for support and guidance; reluctant to take independent action or to go against popular opinion. *From DSM.III.R (American Psychiatric Association, 1987).

7 Personality disorders, traits, and types 173 Measures Hogan Development Survey (Hogan & Hogan, 1997) An overview of the item selection guidelines can be found in the work of Hogan and Hogan (2001). The survey includes 154 items, scored for 11 scales, each grouping 14 items. Respondents are requested to agree or disagree with the items. The HDS has been cross-validated with the MMPI personality disorder scales. Table 2 shows the overlap of the HDS and DSM-IV disorder terminology. NEO Personality Inventory Form S (NEO-PI; Costa & McCrae, 1992) The NEO Personality Inventory is a paper-and-pencil, self-report measure based upon the currently popular five-factor model of trait personality (McCrae & Costa, 1989). The domains measured by this inventory are Neuroticism, Extraversion, Openness to Experience, Agreeableness, and Conscientiousness. The inventory is composed of 240 self-descriptive statements to which respondents use a five-point scale in Likert format and anchored by strongly agree and strongly disagree to rate the extent to which each statement describes the person s self. Item responses are numerically coded and summed to obtain 30 facet and five domain scores. Myers Briggs Type Indicator Form G (MTBI; Briggs & Myers, 1987) The Myers Briggs indicator is a Jungian-based inventory that uses a paper-and-pencil self-report format. It is composed of 94 forced-choice items that constitute the four bipolar discontinuous scales that are implied in Jung s (1971) theory i.e. Introversion Extraversion, Sensation Intuition, Thinking Feeling, and Judging Perceiving. Respondents are classified into one of 16 personality types based on the largest score obtained for each bipolar scale (e.g. a person scoring higher on Introversion than Extraversion, Intuition than Sensation, Feeling than Thinking, and Judging than Perceiving would be classified as Introverted Intuitive Feeling Judging ). The Myers Briggs Indicator has been the focus of extensive research and substantial evidence has accumulated suggesting the inventory has satisfactory validity and reliability (Carlson, 1985; Furnham & Stringfield, 1993). Procedure Subjects were required (mandated) to attend a middle management assessment centre, where they completed these questionnaires amongst other exercises. A psychological consultancy ran the day long course and attendance was compulsory. Each manager was given feedback on the results of the questionnaires by a trained and certificated assessor. RESULTS Disorders and traits Correlations Table 3 shows the pattern of correlations (Pearson s) between the Big Five personality traits and the 11 personality disorders. All but two of the correlations with Neuroticism were significant, particularly those associated with Excitable (Borderline) and Cautious (Avoidant). Correlations with Leisurely (Passive Aggressive) and Dutiful (Dependent) types were also significant at r > Two were significantly negative, indicating that Stability, rather than Neuroticism, was associated with Bold (Narcissistic) and Colourful (Histrionic) personality types.

8 174 A. Furnham and J. Crump Table 3. Correlation between the 11 personality disorders and the five personality traits N E O A C 1. Enthusiastic Volatile EXCITABLE 0.58*** 0.27*** *** 0.27*** 2. Shrewd Mistrustful SCEPTICAL 0.18** *** Careful Cautious CAUTIOUS 0.54*** 0.50*** 0.19*** ** 4. Independent Detached RESERVED 0.15** 0.53*** 0.26*** 0.15** 0.14*** 5. Focused Passive LEISURELY 0.27** 0.19** 0.14** * Aggressive 6. Confident Arrogant BOLD 0.17** 0.34*** 0.18** 0.26** 0.23*** 7. Charming Manipulative MISCHIEVOUS 0.09** 0.40*** 0.34*** 0.24*** Vivacious Dramatic COLOURFUL 0.16** 0.55*** 0.34*** 0.23*** Imaginative Eccentric IMAGINATIVE *** 0.43*** 0.13** 0.09** 10. Diligent Perfectionistic DILIGENT ** *** 11. Dutiful Dependent DUTIFUL 0.24** ** 0.08* 12. Social Desirability 0.08* 0.03* 0.10** ** ***p < 0.001; **p < 0.01; *p < Seven of the correlations with Extraversion were significant. Extraverts were more likely to be Bold (Narcissistic), Mischievous (Anti-Social), and Colourful (Histrionic) while Introverts were more likely to be Excitable (Borderline), Cautious (Avoidant), Reserved (Schizoid), and Leisurely (Passive Aggressive). Eight correlations with Openness were significant, four weakly negative and four more strongly positive. Those participants with high openness score tended to score higher on Mischievous, Imaginative, Colourful, and Bold, but lower on Independent, Careful, Focused, and Diligent. Predictably perhaps, seven of the 11 disorders were negatively correlated with Agreeableness. The only significant positive correlation was Dutiful. Two disorders showed positive significant correlations with Conscientious (Diligent, Confident), while six significant correlations were negative (Enthusiastic, Careful, Independent, Focused, Imaginative, and Dutiful). It is possible to compare the results from Table 3 with the meta-analytic results from Saulsman and Page (2004), based on an N ¼ 1158 and using different measures of personality disorders, as well as the more usual DSM criteria terminology. Although 9 of the 11 disorders were significantly associated with neuroticism, five stood out with effect sizes greater than 0.20: Borderline (Excitable), Avoidant (Cautious), Dependent (Dutiful), Schizotypal (Imaginative), and Paranoid (Sceptical). The results from Table 3 show also that Borderline (Excitable) and Avoidant (Cautious) were most strongly correlated with Neuroticism, as was Dependent (Dutiful). In this study Histrionic (Colourful) was negatively associated with Neuroticism, whereas in the Saulsman Page (2004) analysis there was no significant relationship. Of the ten correlations between personality disorders and Extraversion as shown by Saulsman and Page (2004), nine were significant, but five with effect sizes over Three were positive, Histrionic (Colourful), Schizotypal (Imaginative), and Narcissistic (Bold), and two negative, Avoidant (Cautious) and Schizoid (Reserved). Results from this study are very similar but with two exceptions. In this study Antisocial (Mischievous) was highly positively correlated with Extraversion, whereas in the Saulsman Page (2004) review there was no significant association. Second, in this study there was no correlation between Paranoid (Sceptical) and Extraversion, while in that by Saulsman and Page (2004) there was a significant negative association. Saulsman and Page (2004) found 8 of the 10 associations between Openness and the personality disorders significant but only modestly so. Five were negative, Schizoid

9 Personality disorders, traits, and types 175 (Reserved) and Dependent (Dutiful) being the highest, and three positive, with Histrionic (Colourful) and Narcissistic (Bold) being the highest. Results of this study were broadly similar but with two notable exceptions. Schizotypal (Imaginative) was strongly positively correlated with Openness but there was no significant association in the Saulsman Page (2004) meta-analysis. In this study there was no significant association between Openness and Dependent (Dutiful) whereas there was a modest significant negative correlation in the Saulsman Page (2004) review. With respect to Agreeableness, the Saulsman Page (2004) review showed eight significant negative correlations, of which the biggest (in order) were Antisocial (Mischievous), Paranoid (Sceptical), Narcissistic (Bold), Borderline (Excitable), and Schizotypal (Imaginative). Results from this study were very similar indeed in both size and direction of association. For the final dimension, Conscientiousness, Saulsman and Page (2004) found all associations were negative, particularly Antisocial (Mischievous) and Borderline (Excitable), except Obsessive Compulsive (Diligence), which was strongly positive. Results from this study confirmed the latter association but not the association with Antisocial (Mischievous). Furthermore, in this study Narcissistic (Bold) was fairly strongly positively correlated with Conscientiousness, but in the Saulsman Page (2004) study the correlation was small and negative. Three further analyses were made. First, the correlation corrections were computed at the subscale level to test some of the hypotheses set out in Table 1 (details of these analyses from the author). The results were also examined at the primary factor level. Many of the subscale correlations with the Introversion Extraversion facets were high, especially E3 Assertiveness. Thus all correlations with the Reserved/Schizoid types were large, negative, and significant. However, contrary to the hypotheses, the Schizotypal/ Imaginative types are Extraverted, not Introverted. The Avoidant/Cautious types were consistently introverted across all facets and the Histrionic/Colourful types Extraverted across all facets. Interestingly, only three facets of Openness (fantasy, feelings, and actions) seemed relatively strongly related to three of the personality disorders (Anti-Social/Mischievous; Histrionic/Colourful; and Schizotypal/Imaginative). Few of the correlations with Agreeableness were greater than r > As predicted the Paranoid/Sceptical were low on trust (r ¼ 0.27) and straightforwardness (r ¼ 0.13); the Anti-Social/Mischievous were low on straightforwardness. However, the Narcissistic/Bold were only particularly low on modesty (AS) and the Dependent/Dutiful were not high on four of the six facets but they were high on compliance (A4) as predicted. Finally, the conscientious facet correlations tended to support the Widiger et al. (2002) hypotheses. Obsessive Compulsive were high on all six facets. However, Narcissistic/ Bold tended to score high on two facets (C1, C4) while the Anti-Social/Mischievous only scored low on C6 (Deliberation). Regressions Next, two sets of regressions were computed. In the first set the 11 disorders were each regressed onto the five trait scores. Table 4 shows this data. Table 4 shows that all were significant and that with only one exception (H6: leisurely) neuroticism was a significant predictor. The results show that disagreeable neurotics are excitable/borderline; disagreeable, conscientious, neurotics sceptical/paranoid; agreeable,

10 176 A. Furnham and J. Crump Table 4. Results of the 11 regressions where the disorders were each regressed onto the Big Five H1 H2 H3 H4 H5 H6 Beta t Beta t Beta t Beta t Beta t Beta t N *** *** *** * *** E *** *** *** O * ** A *** *** *** ** *** C *** *** F(5, 853) 50.85*** 21.68*** *** 73.87*** 16.26*** 50.85*** Adj. R H7 H8 H9 H10 H11 Beta t Beta t Beta t Beta t Beta t N * * * *** *** E *** *** *** O *** *** *** A *** *** *** *** C ** *** * *** F(5, 853) 63.77*** *** 53.15*** *** 21.58*** Adj. R introverted, neurotics cautious/avoidant; disagreeable, stable individuals reserved/ schizoid; closed-to-experience neurotics leisurely/passive-aggressive; disagreeable, extraverted, conscientious people bold/narcissistic; extraverted, disagreeable, stable, open, low conscientious people mischievous/anti-social; extraverted, open, disagreeable, stable, low conscientious people colourful/histrionic; open, stable, extraverted, disagreeable, low conscientious people imaginative/schizotypal; conscientious neurotics diligent/obsessive compulsive; and agreeable neurotics dutiful. The 30 subscale scores were then used in 11 further regressions (details from the authors). Three things were apparent from these analyses. First, as might have been expected, the amount of variance accounted for dropped dramatically, often by two-thirds. Second, even when the higher-order trait score was highly significant, with a large beta, usually only two or three of the primary lower-order factors were significant predictors. Third, within each of the five super-trait scores it was often the same primary score that was significant. In the Neuroticism factor it was N3 (depression) and to a lesser extent N1 (anxiety). In Extraversion it was E2 (gregariousness) that was by far the most significant correlate. It was O4 (actions) in Openness and A2 (straightforwardness), A3 (altruism), and A4 (compliance) that were the most often significant predictors in Agreeableness. No one Conscientiousness subscale score seemed particularly frequently a significant predictor. Factor analysis The 11 personality disorders were then subjected to a Varimax rotated factor analysis (see Table 5). Four factors emerged similar to those reported in the Hogan manual (p. 1), though not the same as found in the analysis by Schroeder, Wormworth, and Livesley (1994). Their oblique rotation of the 16 dimensions of the DAPP-BQ yielded four factors labelled Neuroticism, Disagreeableness, Introversion, and Compulsivity. Using the terminology of the Hogan manual, the first factor in this study was labelled moving against people, the second moving away from people, the third critical, and the fourth

11 Personality disorders, traits, and types 177 Table 5. Varimax rotation of the 11 personality disorders Vivacious Colourful 0.78 Charming Mischievous 0.77 Imaginative Mischievous 0.71 Confident Bold 0.62 Enthusiastic Excitable 0.74 Careful Cautious 0.74 Focused Leisurely 0.57 Independent Reserved 0.51 Diligent Diligent 0.80 Shrewd Sceptical 0.56 Dutiful Dutiful 0.82 Eigenvalue Variance 24.13% 17.01% 10.65% 9.92% Table 6. Regressing the four factors onto the five traits N E 0 A C Beta t Beta t Beta t Beta t Beta t Factor * *** *** *** ** Factor *** *** *** *** *** Factor ** *** *** *** Factor *** *** *** *** F(4, 853) *** *** 66.63*** 55.31*** *** Adj. R diligent. These four factors were then regressed onto each of the Big Five personality traits (Table 6). All were significant. Factors 2 (moving away from people) and 4 (dutiful) were best predictors of Neuroticism while factor 1 (moving against people) was a positive predictor and factor 2 (moving away from people) a negative predictor of Extraversion. All factors were significant predictors of openness but factor 1 was the most powerful predictor. Agreeableness was negatively predicted by factors 1, 2, and 3 but positively by factor 4. The two best predictors of conscientiousness were factor 2 and factor 3 positively. For Neuroticism, Extraversion, and Conscientiousness over 40 per cent of the variance was accounted for. Disorders and types Table 7 indicates the overlap between scores on the eight-bipolar-type indicator and the personality disorders. Given the scoring the correlations are near mirror images for the four opposite types (EI; SN; TF; JP). Three-quarters of the correlations with EI were significant. Nor surprisingly, in size, direction, and significance level they were similar to the correlations for extraversion in Table 3. As before highest correlations were with Reserved and Colourful. Many of the correlations with the SN (Sensing Intuitive) dimension were significant, though only three above r > Intuitive people were more Mischievous, Colourful, and Imaginative but less Diligent. Half of the correlations were significant for the TF dimension but none greater than r > Again half of the

12 178 A. Furnham and J. Crump Table 7. Correlations between the 11 personality disorders and the eight Jungian types E I S N T F J P 1. Enthusiastic Volatile EXCITABLE 0.22*** 0.24** ** 0.10** 2. Shrewd Mistrustful SCEPTICAL ** 0.12** 0.17*** Careful Cautious CAUTIOUS 0.31*** 0.34*** 0.19** 0.16** 0.19*** 0.15** Independent Detached RESERVED 0.57*** 0.58*** 0.11*** 0.09** Focused Passive Aggressive LEISURELY 0.20** 0.20** 0.12** 0.10** Confident Arrogant BOLD 0.27*** 0.28*** 0.15** 0.16*** 0.20*** 0.13** Charming Manipulative MISCHIEVOUS 0.28*** 0.29*** 0.29*** 0.30*** 0.14** *** 0.36*** 8. Vivacious Dramatic COLOURFUL 0.52*** 0.53*** 0.34*** 0.34*** *** 0.24*** 9. Imaginative Eccentric IMAGINATIVE 0.26** 0.23** 0.34*** 0.33*** * 0.25*** 0.28*** 10. Diligent Perfectionistic DILIGENT *** 0.29*** 0.12** 0.16** 0.50*** 0.51*** 11. Dutiful Dependent DUTIFUL * 0.13** 0.16** 0.08* 0.12** 0.12** 12. Social Desirability ***p < 0.001; **p < 0.01; *p < 0.05.

13 Personality disorders, traits, and types 179 Table 8. Correlations between the 11 personality types and the four personality traits EI SN TF JP 1. Enthusiastic Volatile EXCITABLE 0.13** Shrewd Mistrustful SCEPTICAL 0.08* ** Careful Cautious CAUTIOUS 0.19*** ** Independent Detached RESERVED 0.28*** Focused Passive Aggressive LEISURELY 0.14** Confident Arrogant BOLD 0.13** *** Charming Manipulative MISCHIEVOUS * 0.13** 8. Vivacious Dramatic COLOURFUL 0.22*** * 9. Imaginative Eccentric IMAGINATIVE * * 10. Diligent Perfectionistic DILIGENT * Dutiful Dependent DUTIFUL * 0.10* Social Desirability ***p < 0.001; **p < 0.01; *p < Table 9. Correlations between HDI and MBTI factors MBTI scores Hogan factors EI SN TF JP 1. Moving against 0.24** * 0.10* 2. Moving away 0.18** Moving toward ** 0.13** 4. Dutiful ***p < 0.001; **p < 0.01; *p < dimensions were significant for the final factor (JP) and two of note (r > 0.30). Judging people were less Mischievous and more Diligent than perceiving people. Table 8 shows that when the personality disorders are correlated with the computed four-factor scores the correlations are by-and-large more modest. Table 9 shows the higher-order correlations between both factors. Extraverts move against, Introverts away from, people. People with high feeling versus thinking type scores tend to move against and toward others but have low dutiful scores. Those with high judging type scores tend to move towards and not against people. DISCUSSION This study demonstrated considerable overlap between the measures of personality traits and types which were hypothesized by Widiger and Costa (2001) and Widiger et al. (2002). Further, the results by and large concur with the meta-analytic findings of Saulsman and Page (2004). Table 3 provides evidence for the hypothesized relationships in Table 1. Assuming that Table 1 provides a list of hypotheses, the data provide support for most of the hypotheses. However, this was not always the case. Thus Widiger et al. (2002) assumed a relationship between Schizotypal (Imaginative) and Introversion and Neuroticism, while the data showed no relationship with Neuroticism and a significant relationship with Extraversion.

14 180 A. Furnham and J. Crump One other noticeable difference between this study and others was the number of significant correlations between the personality disorders and Openness. Considering the meta-analytic results of Saulsman and Page (2004) in Table 2 (pp ), it seems that few other studies found many correlations between Openness and the disorders, perhaps with the exception of the Schizoid, Histrionic, and Narcissistic disorders. This is also the case with the Widiger et al. (2002) hypotheses shown in Table 1, where, compared with the other Big Five traits, Openness is relatively rarely mentioned. It is not immediately clear why the pattern of results for Openness in this study should be different. A close examination of Table 1 shows that Widiger et al. (2002) have a number of specific facet level hypotheses for each personality disorder. Thus for Borderline (Excitable in Hogan s nomeclature) the only hypothesized correlation was with Neuroticism. This was confirmed (r ¼ 0.58, p < 0.001), but Borderlines were also significantly Introverted and low on Agreeableness and Conscientiousness. Similarly, Narcissistic (Bold) types were hypothesized to be low on Agreeableness, Neuroticism, and Conscientiousness. This was confirmed as was the fact that Avoidants (Cautious) were Neurotics with low Openness. Not all hypotheses were confirmed. Thus Obsessive Compulsives (Diligent) were thought to be high on Conscientiousness and Neuroticism but low on Agreeableness. Results confirmed only the high on Conscientiousness and Neuroticism part of the proposed pattern. A number of the correlations were highly significant, i.e. over r ¼ Thus Borderline/ Excitable types were characterized primarily by Neuroticism; the Obsessive Compulsive/ Diligent by Conscientiousness; the Histrionic/Colourful by Extraversion; the Schizoid by Introversion; and the Avoidant/Cautious by both Neuroticism and Introversion (Choleric). These confirm both the size and direction of the correlations reported by Saulsman and Page (2004) in their meta-analysis of 15 different studies. Correlations between the two measures were lowest and fewest for Paranoid/Sceptical, Passive Aggressive/Leisurely, and Dependent/Dutiful. In the Saulsman and Page (2004) meta-analysis five of the mean effect size estimates were over They indicated that Borderline (Excitable), Avoidant (Cautious), and Dependent (Dutiful) individuals were neurotic while Histrionic (Mischievous) but low Avoidant (Cautious) people were Extraverts. The regressional analysis showed much clearer findings both at the super- and primary factor level. Looking at Table 1 and considering the superfactor level results the following (A M) overlap hypotheses were confirmed: A, B, C, D, E, F, H, I, and J. At the primary factor level again most of the hypotheses were confirmed. Certainly, there was almost no evidence of the correlations or betas going the wrong way in the sense that the opposite to the hypothesis was the case. However, the factor analytic results are perhaps the most interesting, showing again considerable amounts of overlap. For instance, the four factors accounted for over 50 per cent of the variance attributable to Extraversion (Table 6). The results show that it is the Borderline/Excitable, Avoidant/Cautious, Passive Aggressive/ Leisurely, and Schizoid/Reserved that are the most Neurotic. Interestingly, the beta weights show both factors 1 and 2 are closely related to Extraversion and Introversion respectively. Note too that the beta weights for Agreeableness indicated that for the three major factors there is a negative relationship. In short, most personality disorders are associated with disagreeableness. Equally, Conscientiousness is positively associated with the third factor (critical) but negatively correlated with the second factor (moving away from people).

15 Personality disorders, traits, and types 181 The correlations (tables) between the personality disorders and personality types show three important things. First, they are on average lower than shown in Table 3, showing less overlap. Second, they are, however, quite consistent. Thus extraversion from the Big Five and EI from the MBTI all measure essentially the same variable. The pattern is very similar. Take for example Excitable (r ¼ 0.27, r ¼ 13) or Reserved (r ¼ 0.53, r ¼ 28). Third, some of the dimensions appear to relate very weakly to the personality disorders. Thus the Sensing/Intuitive and Judging/Perceiving dimensions are very weakly related to all 11 personality disorders. The fact that the overlap between some of these measures is weaker may be due to two factors. First, the MBTI is supposedly measuring normal personality, not personality disorders. Second, the test has unique constructs such as Sensing Intuition, which are unique to it and unrelated to pathological behaviour. Third, there remains some doubt about the psychometric properties of the MBTI, despite its extensive use in consulting and training circles. Those using the MBTI consistently argue that the test does not measure pathology and that all types are equally healthy or adaptable. These results show this not to be the case. Thus in MBTI terminology the ENTI may be prone to anti-social personality syndrome behaviours. Equally, IS types may be prone to Schizoid behaviour. The overlap between these measures needs to be recognized by those who use the MBTI in selection and training. What are the implications of these results for an understanding of the personality disorders? It should be pointed out that whilst the size of the N was large and representative enough to have confidence in the results, there may be correlation inflation due to item overlap. It is important to note that this study examined the relationship between traits and disorders on a large normal adult population using a relatively new but well psychometrized measure of disorders that had acceptable face validity to ensure little dissimulation. There are essentially three implications of these results: the first psychometric, the second theoretical, and the third practical. There are several self-report measures of both personality disorders (i.e. Hogan & Hogan, 1997; Oldham & Morris, 1991; Morey et al., 1985; Widiger & Coker, 2001) and personality traits (Costa & McCrae, 1992; Eysenck & Eysenck, 1985). Whilst studies on overlap between measures of similar concepts show high agreement not all questionnaire measures have undergone rigorous psychometric assessment. Certainly, from a psychometric point of view there has been considerably more work done on measures of personality traits than types. Moreover, there remains relatively little psychometric evaluation of self-report measures of personality disorders, in particular evidence of internal reliability; concurrent, discriminant, and predictive validity; and dimensionality. Clearly, the psychiatric researchers interested in self-report measures of personality disorders would do well to employ the services of a good psychometrician to examine the properties of their measures (Kaye & Shea, 2000). Further, the face validity of these measures is important in assessment and counselling to ensure evidence of construct and predictive validity. One obviously important psychometric issue is the discriminant validity of personality disorders measures over personality trait measures. Equally, there is the issue of incremental validity: of what adding one or more measures of normal traits or types adds to a (self-report) assessment of the personality disorders (or vice versa). Inevitably, this is an empirical question, which depends on what one is trying to predict. The answer from the perspective of Hogan and Hogan (2001) is that trait measures assess the bright side and disorder measures the dark side of personality, and that the former are more useful from a select in perspective and the latter from a select out perspective.

16 182 A. Furnham and J. Crump More importantly, researchers in personality disorders may benefit considerably from understanding the research by trait psychologists on the mechanisms and processes that account for the behaviour associated with identified traits (Eysenck & Eysenck, 1985). Not only are the processes well identified and understood, but there is now considerable evidence of the biological and genetic foundations. Over the past decade there has been something of a renaissance in personality trait theory and measurement (Matthews & Deary, 1998). Thus Cooper (1998) describes recent empirical results on the biological basis of personality using positron emission tomography as well as recent impressive studies on the genetics of personality. Trait theory offers insight then into the structure, determinants, and functions of personality traits. More importantly, it moves from the relatively simple descriptive and categorical level where agreement has been reached to attempt to understand the mechanisms and processes that explain how they function. Given the clear overlap between the descriptive aims of both personality disorder and personality trait researchers, it would seem beneficial for the two to collaborate more closely to increase their theoretical understanding on the mechanisms involved in the aetiology of both traits and disorders. Third, psychiatric researchers may benefit from studies on the psychotherapeutic utility of the FFM (Miller, 1991). Equally, IWO and applied psychologists have much to learn from how personality disorders affect work related behaviour (Babiak, 1995). At an applied level, psychologists are frequently asked to recommend personality measures to provide comprehensive, relevant, and predictive profiles that help in business decision making. This may involve both selecting-in and selecting-out criteria. It is important to have measures with acceptable face validity as well as necessary predictive and construct validity. Further, it is thought of as both expensive and pointless to use different tests, which, though using different terms, essentially measure the same thing. Hence we have practical questions of incremental validity, in the sense of the question of what adding another test to the battery adds the quality of the information on the psychological makeup of the candidate. This study has shown that trait measures can be used to detect personality disorders, which are themselves predictive of business failure and derailment (Furnham, 2003). Thus the anti-social and narcissistic personalities, which seem most related to business derailment (Babiak, 1995), are systematically related to the Big Five traits, though the latter only account for a quarter of the variance (see Table 4). Certainly, the results do indicate that measures of personality disorders may serve selectout functions well and personality traits select-in functions. Further, despite their popularity it seems that older personality type measures have such poor psychometric qualities that it remains difficult to justify their use on grounds of reliability or predictive validity (Furnham, Moutafi, & Crump, 2003). This was essentially a correlational study looking at concurrent validity. Whilst this is important, it is also perhaps more important and difficult to demonstrate predictive and construct validity of questionnaires, showing that they predict behaviour over time. REFERENCES Babiak, P. (1995). When psychopaths go to work: a case study of an industrial psychopath. Applied Psychology, 44, Bayne, R. (2003). Love, money and studying. The Psychologist, 16, Brand, C. (1994). Open to experience closed to intelligence. European Journal of Personality, 8,