UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Diagnosis of psychopathy in a forensic psychiatric population Hart, Stephen David 1987

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
831-UBC_1987_A8 H38_7.pdf [ 4.33MB ]
Metadata
JSON: 831-1.0097159.json
JSON-LD: 831-1.0097159-ld.json
RDF/XML (Pretty): 831-1.0097159-rdf.xml
RDF/JSON: 831-1.0097159-rdf.json
Turtle: 831-1.0097159-turtle.txt
N-Triples: 831-1.0097159-rdf-ntriples.txt
Original Record: 831-1.0097159-source.json
Full Text
831-1.0097159-fulltext.txt
Citation
831-1.0097159.ris

Full Text

Diagnosis of Psychopathy in a Forensic Psychiatric Population B. A., The University of Bri t i s h Columbia, 1984 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF THE FACULTY OF GRADUATE STUDIES (Department of Psychology) We accept this thesis as conforming to the required standard THE UNIVERSITY OF BRITISH COLUMBIA August, 1987 (c) Stephen David Hart, 1987 By STEPHEN DAVID HART MASTER OF ARTS in In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. The University of British Columbia 1956 Main Mall Vancouver, Canada Department V6T 1Y3 DE-6(3/81) i i ABSTRACT Both researchers and clinicians, especially those working in criminal populations, have long suggested that psychopathy (or antisocial personality disorder) and schizophrenia are associated on an etiological or on some other level (e.g., Eysenck and Eysenck, 1976, 1978). Others (Hare, 1984; Hare and Harpur, 1986; Raine, 1985) argue that psychopathy i s not associated (or even negatively associated) with other psychiatric disorders, including schizophrenia. To evaluate these competing positions concerning the psychopathy-schizophrenia association, 80 male prisoners remanded to a forensic psychiatric institute for assessment of their fitness to stand t r i a l were diagnosed using both the Psychopathy Checklist (PCL; Hare, 1980, 1985a) and DSM-III Axis I and II c r i t e r i a . In addition, c l i n i c a l global ratings and self-report inventories were used to measure the strength of psychopathy- and schizophrenia-related t r a i t s . The results indicated that: a) although diagnoses of psychopathy (according to PCL criteria) did not have perfect specificity with respect to schizophrenia-related c l i n i c a l diagnoses, the overlap was small, and the PCL scales were either not associated or negatively associated with these disorders; b) diagnoses of antisocial personality disorder (APD, according to DSM-III criteria) were generally not associated with schizophrenia-related disorders, but had lower c l i n i c a l specificity than did the PCL c r i t e r i a with respect to both schizophrenia-related and other psychiatric i i i disorders; c) there was no association between psychopathy- and schizophrenia-related c l i n i c a l ratings; d) psychopathy and APD diagnoses and c l i n i c a l ratings were not related to scores on other standard rating scales of the severity of psychiatric symptomatology; and e) there was no difference between schizophrenic and non-schizophrenic subjects in the strength of psychopathy-related t r a i t s , and no difference between psychopaths and nonpsychopaths (or APD versus non-APD subjects) in the strength of schizophrenia-related t r a i t s . As well, self-report measures related to psychopathy and schizophrenia did not correlate with each other, or with c l i n i c a l ratings of the two disorders. The results are interpreted as supporting the view that psychopathy is not positively associated with schizophrenia or with psychiatric disorder in general. The practical u t i l i t y of various techniques for assessing psychopathy in forensic psychiatric populations i s also discussed. iv TABLE OF CONTENTS Abstract i i -List of Tables v i i i . Acknowledgements x i . I. Introduction 1 II. A Review of the Empirical Literature A. Hare (1982) 6 B. Raine (1985) 7 C. Boyd et a l . (1984) 8 D. Howard, Bailey, and Newman (1984) 10 E. Pfohl, Coryell, Zimmerman, and Stangl (1986) 12 F. Hare (1985c) 13 G. Wong (cited in Hare and Harpur, 1986) 13 H. Summary and discussion 14 III. Purpose of the Present Study 16 IV. Method A. Subjects 18 B. Procedure 1. F i l e information 18 2. Interview information 19 3. Criminal history information 20 4. DSM-III c l i n i c a l diagnoses and global ratings 20 V 5. Psychopathy Checklist diagnoses 22 6. Blind Antisocial Personality Disorder diagnoses .. 23 7. Schizophrenia-related self-report measures 24 8. Psychopathy-related self-report measures 24 9. Other scales 24 V. Results A. Characteristics of the sample 1. Race 26 2. Age 26 3. Marital status 26 4. Education 26 5. Social class 26 6. Psychiatric history 28 7. Criminal history 28 8. Instant offense 28 B. R e l i a b i l i t y of the measures 1. DSM-III Axis I major disorders 29 2. Substance abuse/dependence disorders 33 3. Axis II personality disorders 35 4. PCL scales 37 5. Blind APD diagnosis 40 6. Schizophrenia global rating (SZGR) 40 7. Personality disorder global ratings (PDGRs) 40 8. GAS and BPRS 42 v i C. The association between psychopathy and schizophrenia 1. Overlap between psychopathy-related measures and DSM-III Axis I major disorders 44 2. Overlap between psychopathy-related and alcohol and other substance abuse/dependence diagnoses — 50 3. Overlap between psychopathy-related and other Axis II PD diagnoses 50 4. Correlations among psychopathy-related dimensional measures 57 5. Correlations among schizophrenia-related dimensional measures 60 6. Correlations between psychopathy- and schizophrenia-related dimensional measures 63 7. Factor analysis of c l i n i c a l ratings 63 8. Correlations between psychopathy-related measures and standardized psychiatric rating scales 67 9. Schizophrenic traits in the PCL and Blind APD groups 70 10. Psychopathic traits in the Axis I diagnostic groups 70 VI. Discussion A. Summary of results and comments 74 B. Assessing psychopathy in a forensic psychiatric population v i i 1. The performance of the PCL scales 78 2. The performance of APD c r i t e r i a 81 3. The performance of self-report scales 82 C. Directions for research 83 D. Conclusion 84 References 85 Appendix A 90 v i i i LIST OF TABLES Table I. Most serious offense committed by subjects 30 Table II. Interrater agreement for the six superordinate DSM-III Axis I major diagnostic groups 32 Table III. Interrater agreement and base rate of Panel diagnoses of the Alcohol and Other Substances abuse/dependence disorders 34 Table IV. interrater agreement and base rate of Panel diagnoses of DSM-III personality disorders 36 Table V. Interrater agreement and base rate of Panel PCL diagnoses 39 Table VI. Agreement between Blind and Panel APD diagnoses 41 Table VII. Interrater agreement for the PDGRs 43 Table VIII. Overlap between a) the six superordinate DSM-III Axis I major diagnostic groups, and b) the PCL22, PCL20, and Blind APD groups 45 Table IX. Odds ratios (ORs) showing the association between a) the PCL22, PCL20, and Blind APD groups, and b) the Axis I major diagnostic groups 48 Table X. Overlap between a) Alcohol and Other Substances abuse/dependence diagnoses, and b) the PCL22, PCL20, and Blind APD groups 51 ix Table XI. Overlap between the PCL22, PCL20, and Blind APD groups and the Panel personality disorder diagnoses 53 Table XII. Overlap between the PCL22, PCL20, and Blind APD diagnostic groups and the three DSM-III personality disorder clusters 54 Table XIII. Odds ratios (ORs) showing the association between a) the PCL22, PCL20, and Blind APD diagnoses, and b) the three DSM-III personality disorder clusters 56 Table XIV. Correlations among the psychopathy-related dimensional measures 58 Table XV. Correlations between the factor scores on the PCL scales and APD diagnoses 61 Table XVI. Correlations among the schizophrenia-related dimensional measures 62 Table XVII. Correlations among the psychopathy- and schizophrenia-related dimensional measures 64 Table XVIII. Factor loadings (and communalities) of the 12 c l i n i c a l ratings 66 Table IXX. Means of the PCL22, PCL20, and Blind APD groups on two standardized psychiatric rating scales ... 69 Table XX. Means of the PCL22, PCL20, and Blind APD groups on the schizophrenia-related c l i n i c a l ratings ... 71 X Table XXI. Means of the Axis I major diagnostic groups on the PCL22, PCL20, and Panel APD PDGR 72 x i ACKNOWLEDGEMENTS Many people have helped this project along from the start, and I have several acknowledgements to make. First, I would like to express my appreciation to the staff and patients of the Forensic Psychiatric Institute, Port Coquitlam, B.C., for their cooperation; special thanks go to the medical records staff, psychiatrists, psychologists, and to the nurses of ward R3W. Second, thanks to Dr. Margaret Moreau and Adelle Forth for their patience and dedication in checking the r e l i a b i l i t y of my ratings and diagnoses under less-than-optimal circumstances. It was a privilege and a pleasure to work with both of them. Third, my committee members, Drs. Lynn Alden and Jerry Wiggins, provided me with sound advice and encouragement. Thanks also to Tammy, who kept me happy, healthy, and sane outside of academia. Finally, and most importantly, I would like to thank my advisor, Dr. Robert D. Hare. He gave me support and assistance beyond that required of an academic supervisor, and has nurtured both my interest in psychopathy and my progress in graduate school. 1 I. Introduction Those working with criminal and delinquent populations have often pondered the relation between "madness" and "badness" (e.g., Travin and Protter, 1982). For example, i t has frequently been observed that individuals suffering from psychopathy (the bad) and those suffering from schizophrenia (the mad) often exhibit some of the same symptoms, symptoms such as impulsivity, seemingly i l l o g i c a l crimes, f l a t affect, and chronic disturbances in interpersonal relations and occupational functioning (Bender, 1971; Cleckley, 1976; Geller, 1980; Heston, 1966; Kallman, 1938; Milburn and Goff, 1956; Travin and Protter, 1982). This has led some authors to suggest that psychopathy and schizophrenia are positively associated, and that they somehow 'overlap;' that i s , individuals suffering from one disorder also tend to suffer from the other, or that individuals with one disorder tend to have strong traits of the other (Claridge, 1985; Eysenck and Eysenck, 1976, 1978; Gunderson, 1979, 1984). To these authors, the overlap i s not due merely to a bias in diagnostic c r i t e r i a ; rather, i t reflects a conceptual link between the disorders on some etiological or other theoretical level. Kraepelin, as early as 1904, had noted that "eccentric personalities" such as "criminals, queer individuals, prostitutes, suicides, vagrants, [and] wrecked human beings" are frequently found in the families of schizophrenic patients (Kraepelin, quoted i n Kendler, 1985, p. 539). Other people suggesting that psychopathy f a l l s within the spectrum of schizophrenic disorders include Kety and his colleagues (Kety, Rosenthal, Wender, and Schulsinger, 1968) and Planansky (1972). Eysenck has speculated at length about the association between psychopathy and psychotic disorders (Eysenck and Eysenck, 1976, 1978). 2 To him, the problems some clinicians report in differentiating between psychopathy and schizophrenia "are almost inconceivable i f these two sets of disorders are quite separate, having no causal connection at a l l with each other" (Eysenck and Eysenck, 1976, p. 19). As further evidence of a link between psychopathy and psychosis, he cites a number of genetic, twin, and adoption studies reporting increased rate of psychopathy in the relatives of schizophrenic patients. Eysenck hypothesized that the disorders are both manifestations of a single personality t r a i t , namely psychoticism, which is posited to reflect a predisposition to psychosis. Psychoticism is assumed to be polygenic in nature, with a threshold effect, such that a large number of large- and small-effect genes related to the tr a i t summate to determine what form of psychopathology, i f any, w i l l be manifested (Eysenck and Eysenck, 1976). Thus, schizophrenia can result when the number of psychoticism-related genes reaches criterion, whereas the presence of a smaller number of genes might result in psychopathy; psychopathy here represents "a half-way house to psychosis" (Eysenck and Eysenck, 1978, p. 213). Psychoticism has become an integral part of Eysenckian personality theory, and the Psychoticism (P) scale of the Eysenck Personality Questionnaire (EPQ) was developed to measure i t (Eysenck and Eysenck, 1976). Claridge (1972) has suggested that a disturbance in the relation between arousal and attention may underlie psychoticism, whereby high-psychoticism subjects are hypersensitive to the environment when aroused, thus interfering with their cognitive processing. However, empirical support for this model has been mixed: for example, contrary to what one would expect on the basis of Eysenckian theory, psychopaths and schizophrenics do not exhibit 3 similar patterns of association between attention and arousal, and both these groups seem to differ from normal subjects who score high on the P scale (Robinson and Zahn, 1985). One problem with Eysenckian theorists and researchers in this area, however, is that their conception of psychopathy is extremely broad; i t is more or less synonymous with criminality, antisocial behavior, and delinquency. In contrast to the position held by Eysenck and others, some authors argue that psychopathy is not related to schizophrenia, or, for that matter, to any other form of psychopathology (Hare, 1984; Hare and Harpur, 1986). Indeed, Cleckley (1976) included "absence of delusions and other signs of psychotic thinking" and "absence of 'nervousness' or psychoneurotic manifestations" in his l i s t of diagnostic c r i t e r i a for psychopathy. As well, several of the most prominent symptoms of psychopathy (e.g., glibness, superficial charm, inflated self-esteem, etc.) seem to be inconsistent with the characteristics typical of those suffering from other psychiatric disorders, such as social anxiety, poor interpersonal s k i l l s , and poor self-image. Thus, Hare and Forth (1985) hypothesized that schizophrenic and other psychopathological traits may be more common in those subjects with psychopathic t r a i t s (e.g. ordinary criminals, "secondary psychopaths," etc.) than i n true ("full-blown") psychopaths. These authors would expect very few ( i f any) psychopaths to suffer from schizophrenia-related or any other psychiatric disorders. It i s important for both researchers and clinicians to clearly define the relation between psychopathy and schizophrenia. F i r s t , researchers are actively engaged in the search for genetic markers and experimental correlates of both disorders; a misdiagnosis, or a failure 4 on the part of diagnostic c r i t e r i a to isolate pure diagnostic groups, could bias data and obscure results (Howard, Bailey, and Newman, 1984). Second, forensic clinicians attempt to distinguish between the disorders because according to legal tradition in Britain and North America psychopaths, but not individuals whose antisocial behavior is the result of symptoms of a major mental ill n e s s , are held criminally responsible for their actions (Stuart, 1982; Void, 1979).. A misdiagnosis in this case may be very costly: i t could lead either to the wrongful punishment and denial of treatment to a person who i s , according to the law, not responsible for his actions, or to the failure to incarcerate a person who is not only responsible for his actions but is also unlikely to benefit from treatments offered him and like l y to victimize society in the future i f not properly supervised (Travin and Protter, 1982). Given the importance of determining pure diagnostic groups and the cost to both researchers and clinicians of making diagnostic errors, i t is surprising that so l i t t l e empirical research has been conducted on the intra-individual overlap between psychopathy and schizophrenia (Hare and Harpur, 1986; Raine and Venables, 1987). This lack of research may be due, in part, to a focus on definitional issues in the past, that i s , issues related to the description, validation, and diagnosis of psychopathy and schizophrenia as individual disorders. However, the development of psychometrically-sound diagnostic c r i t e r i a for both psychopathy and conceptually-related disorders (global ratings of psychopathy: Hare and Cox, 1978; the Psychopathy Checklist [PCL]: Hare, 1980, 1985a; antisocial personality disorder [APD]: American Psychiatric Association, 1980) and for the schizophrenia-spectrum 5 disorders (Feighner c r i t e r i a : Feighner, Robins, Guze, Woodruff, Winokur, and Munoz, 1972; Research Diagnostic Criteria [RDC]: Spitzer, Endicott, and Robins, 1974; Diagnostic and Statistical Manual of Mental  Disorders, 3rd edition [DSM-III] c r i t e r i a : American Psychiatric Association, 1980), the door has been opened for research on their overlap. The purpose of this paper i s to review the empirical literature concerning the psychopathy-schizophrenia overlap, and to present new data relevant to this issue. 6 II. A Review of the Empirical Literature If we limit a literature review to research using standard and reliable diagnoses of psychopathy-related disorders (i.e., the PCL and APD), only seven empirical studies relevant to the psychopathy-schizophrenia overlap can be found. These studies are discussed below. A. Hare (1982) In the f i r s t study, Hare (1982) administered the EPQ to a sample of 173 male inmates of a Canadian federal medium-security institution. The inmates were assessed using both global rating and the PCL. Correlations were computed between the four EPQ subscales on one hand and the PCL items, PCL factors, and a combination of the two diagnoses of psychopathy on the other hand. The inmates were also divided into three groups—high, medium, and low psychopathy—on the basis of the combined global ratings and PCL scores, and group differences on the EPQ subscales were tested for significance. Finally, zone (or octant) analyses were performed on the EPQ subscale scores to check between-group differences. The results, on the whole, were disappointing. Despite the variety of analyses, the only findings were: a) two small but significant correlations between the EPQ subscales P and L (Lie), and the combined psychopathy measures (r = .16 and r = .14, respectively); and b) significant correlations between 6 of 22 PCL items and 2 of 5 PCL factors, and the P scale ( a l l r < .31). The study has several strengths: the use of valid and reliable diagnoses of psychopathy, a large N, the use of a scale related to general psychoticism (the P scale) rather than just an all-or-none c l i n i c a l diagnosis, and a variety of data analysis techniques. The study certainly i s a fair test of the psychopathy-P scale relation. 7 However, i t was not intended to be a test of the relation between psychopathy and schizophrenia-spectrum disorders. The inmates constituted a nonpsychiatric population; inmates suffering from gross psychopathology were diverted to special hospitals. Second, no c l i n i c a l diagnoses of schizophrenia-spectrum disorders were used. As well, the interpretation and psychometric properties of the P scale have been questioned (e.g., Helmes, 1980). Finally, a reliance on self-report measures to investigate psychopathology in criminal populations is questionable, because of the tendency of inmates to dissimulate or employ unusual response sets (e.g., Hare, 1985b). B. Raine (1985) Raine (1985) examined psychotic traits in a sample of 34 inmates of a British maximum-security institution. A l l inmates were diagnosed using the PCL, rated along dimensions corresponding to the DSM-III categories of borderline and schizotypal personality disorder, and administered seven self-report questionnaires that yielded ten schizoid t r a i t scores. The PCL total score was not correlated with any of the schizoid, schizotypal, or borderline personality dimensional ratings. A single significant correlation was found between one of these latter measures and a PCL factor; even this correlation (r = -.36) would be non-significant i f Raine had controlled the family-wise error rate for the correlation matrix. Strengths of this study were the use of multiple measures of "borderline schizophrenia"—both c l i n i c a l ratings and self-report scales—to examine the psychopathy-schizophrenia overlap, and the fact that a l l c l i n i c a l ratings were made blind to the f i n a l diagnoses of psychopathy. However, Raine's sample was very small, and like Hare 8 (1982), he used a non-psychiatric population, thus reducing his chances of finding significant pathology. He also did not use well-known self-report inventories, making i t d i f f i c u l t for clinicians and researchers not familiar with his measures to interpret his results. Also, Raine did not present the means and standard deviations or r e l i a b i l i t i e s of his measures. Finally, Raine reported only correlations between PCL scores and the other measures, rather than presenting additional information indicating between-group differences (psychopaths versus nonpsychopaths) or range/percentage overlaps among the psychopathy and schizophrenia t r a i t measures. C. Boyd et a l . (1984) The third study comes from the NIMH Epidemiologic Catchment Area (ECA) program (see the Special Issue of the Archives of General  Psychiatry, Vol. 41, October 1984). The ECA program has investigated the prevalence of various psychiatric disorders in the general population using the Diagnostic Interview Schedule (DIS: Robins, Helzer, Croughan, Williams, and Spitzer, 1981). The DIS is a structured interview that can be used (even by lay interviewers) to make reliable diagnoses according to c r i t e r i a contained i n the DSM-III. Although the purpose of the ECA program was to determine the short-term and lifetime prevalence rates of various disorders, Boyd et a l . (1984) examined the co-occurrence of disorders whose simultaneous presence is precluded i n the DSM-III by exclusionary c r i t e r i a . In particular, we are interested in the case of APD, a diagnosis that i s conceptually related to psychopathy. APD criterion E i n DSM-III states that a person's antisocial behavior must not be "due to Severe Mental Retardation, Schizophrenia, or manic episodes" (APA, 1980, p. 321), and indicates 9 that a diagnosis of APD should never be made i f the individual also suffers from a schizophrenic disorder. With the ECA program data, however, i t is possible to ignore criterion E, and to calculate the APD-schizophrenia overlap. Boyd et a l . estimate the odds ratio for APD given schizophrenia to be about 19; that i s , the probability of APD/schizophrenia is about 19 times higher than the probability of APD/no schizophrenia. While APD and schizophrenia seem to be related according to the odds ratio, several factors must be considered. F i r s t , the APD/schizophrenia odds ratio was low compared to the odds ratios for other disorders. Second, the odds ratios were not calculated from exact prevalence statistics, but from prevalences estimated on the basis of sampling procedures. That i s , we are not told exactly how many subjects were actually diagnosed both schizophrenic and APD. As well, the odds ratio i n this sample may be an inaccurate estimate of the odds ratio in the general population due to the low prevalence of both disorders and, presumably, their overlap (this hypothesis is supported by the fact that the APD/schizophrenia odds ratio i s the least significant of a l l the odds ratios presented in th»2 Boyd et a l . study). Third, besides the s t a t i s t i c a l problems inherent in this study, i t did not examine the overlap among APD and schizophrenia-related disorders or schizophrenic tr a i t s , and (intentionally) did not sample criminal or psychiatric populations. Finally, although psychopathy and APD are conceptually and empirically related, they are not equivalent diagnoses: APD has been cr i t i c i z e d for focusing almost exclusively on criminal behaviors rather than on more general personality traits (Hare, 1983; Millon, 1981). Thus, the ECA program data are not well-suited for determination of a more general psychopathy-schizophrenia spectrum overlap. 10 D. Howard, Bailey, and Newman (1984) Howard et a l . (1984) provide some data on the psychopathy-schizophrenia overlap in a forensic psychiatric setting. They studied 50 consecutive male remands to Broadmoor, a forensic hospital in Britain. Each subject was given a rating of psychopathy using an abbreviated 15-item version of the PCL. Subjects were then separated into four groups on the basis of their past and present Mental Health Act (1959) diagnoses. Psychopathic group subjects, n = 9, had received only diagnoses of Psychopathic Disorder; Mixed group subjects, n = 13, had previously received diagnoses of both Psychopathic Disorder and schizophrenia; the Schizophrenic subjects, n = 21, had received only diagnoses of schizophrenia; and the Affective group, n = 7, had been diagnosed as either affective or schizo-affective psychosis. A l l subjects were also administered the Minnesota Multiphasic Personality Inventory (MMPI: Dahlstrom & Welsh, 1960). The results were as follows. Fi r s t , a between-groups Analysis of Variance (ANOVA) revealed significant differences in the mean PCL scores, with the Psychopathic and Mixed groups scoring higher than the Schizophrenia group, and a l l three groups scoring higher than the Affective group. Second, an item profile analysis was interpreted qualitatively to indicate similarity between Psychopathic and Schizophrenic subjects. Third, a discriminant function analysis was performed to try and classify subjects into two superordinate groups, one psychopathic (a combination of the original Psychopathic and Mixed groups) and one non-psychopathic (the Schizophrenic and Affective groups), on the basis of the PCL total scores and 5 MMPI-derived scales. Sixteen of 17 psychopathic subjects and 14 of 23 non-psychopathic subjects were correctly classified. 11 Howard et a l . then re-divided subjects into the four original groups, and determined that 50% of schizophrenics were misclassified as psychopaths. Based on these analyses, the authors inexplicably concluded that: a) psychopaths and schizophrenics have a common affective d e f i c i t ; and b) the PCL lacks c l i n i c a l specificity. A proper critique of this study would require lengthy discussion; the reader i s referred to Raine (1985) and Hare and Harpur (1986) for such critiques. Several problems with the Howard et a l . study are particularly relevant to the purposes of the present study. First, the Mental Health Act (1959) is a completely inadequate basis on which to make valid or reliable c l i n i c a l diagnoses; i t merely describes several medico-legal categories. Second, the authors seem unfamiliar with the PCL. It i s unclear why i t was scored i n a non-standard (and i l l o g i c a l ) fashion. As well, several of the items were incorrectly labelled in various analyses. Third, PCL ratings were not made blind with respect to MHA diagnoses. Fourth, the discriminant function analysis was performed incorrectly. In addition to the criticisms of Hare and Harpur and Raine, other factors that limit the u t i l i t y of these data in describing a general psychopathy-schizophrenia overlap include: a) the absence of schizophrenia-spectrum measures; b) failure to report the standard MMPI c l i n i c a l scale scores, despite the ava i l a b i l i t y of these results for a l l subjects; and c) inappropriate and inadequate data analysis techniques. (The ANOVA does not allow a determination of the actual overlap among groups in PCL scores, although i t appears that none of the subjects i n the study had scores high enough to warrant a diagnosis of psychopathy). As well, the item profile analysis should have been quantified. Finally, the proper way for the authors to test 12 specificity of the PCL would have been to perform a Bayesian analysis, and to report the number of schizophrenic subjects diagnosed as psychopathic by the PCL (e.g., Widiger, Hurt, Frances, Clarkin, and Gilmore, 1984). E. Pfohl, Coryell, Zimmerman, and Stangl (1986) Pfohl et a l . (1986) examined the overlap among APD and the other DSM-III personality disorders—including schizophrenia-spectrum personality disorders—in a sample of 131 noncriminal, nonpsychotic psychiatric inpatients. Each subject and a significant other was interviewed using the Structured Interview for DSM-III Personality (SIDP), a reliable structured interview for personality disorders (Stangl, Pfohl, Zimmerman, Bowers, and Corenthal, 1985). The results suggest that diagnoses of APD have low specificity with respect to other personality disorder diagnoses in general; of five APD subjects, only one did not receive some other personality disorder diagnosis. However, APD did seem to have greater specificity with respect to schizophrenia-spectrum personality disorders in particular: only one APD subject was diagnosed schizotypal, and none were diagnosed schizoid or paranoid. In comparison, three APD subjects were also borderline, three were histrionic, and one was avoidant. If the exclusion criterion for passive-aggressive personality (which rules out this disorder in the presence of another personality disorder) was ignored, three APD subjects would also have received this diagnosis. While this study used standard and reliable c l i n i c a l diagnoses, several factors limit i t s contribution to an understanding of the psychopathy-schizophrenia overlap. F i r s t , the number of APD subjects was extremely small, as a noncriminal population was used. Second, 13 although schizophrenia-spectrum personality disorders were studied, schizophrenics were specifically excluded from the sample. Third, no schizophrenic trait measures were taken. Finally, as discussed above, a diagnosis of APD may reflect criminality more than psychopathy. F. Hare (1985c) Hare (1985c) has some unpublished data relevant to the issue of the psychopathy-schizophrenia overlap. He had two clinicians make f u l l DSM-III diagnoses on the basis of f i l e information for 330 criminals. Among psychopathic subjects, 1.1% received a diagnosis of schizophrenia from at least one rater; for middle- and low-psychopathy groups, the numbers were 3.3% and 2.2% respectively. Diagnoses of personality disorders, however, were much more common in the high-psychopathy group: 1.1% were schizoid, 1.1% were schizotypal, 2.2% were borderline, 1.1% were paranoid, 17.6% were narcissistic, and 1.1% were avoidant, according to at least one rater. Clearly, however, limited non-psychiatric f i l e information alone i s an inadequate basis on which to make optimal c l i n i c a l diagnoses, and although some schizophrenia-spectrum diagnoses were examined, no schizophrenic t r a i t measures were used. Finally, the use of a non-psychiatric population reduces the likelihood of finding pathology. G. Wong (cited in Hare and Harpur, 1986) Wong also collected some preliminary data on the psychopathy-schizophrenia overlap. He rated 12 schizophrenic residents of a forensic psychiatric hospital on the PCL. A l l subjects had been diagnosed according to DSM-III c r i t e r i a by clinicians using f u l l f i l e and interview information. A l l the subjects had scores similar to those of schizophrenics in the Howard et a l . (1984) study, and none scored in 14 the psychopathic range. However, this sample was extremely small, and PCL ratings were not made blind to the c l i n i c a l diagnoses. As well, no schizophrenia-spectrum disorders or schizophrenic t r a i t s were measured. H. Summary and discussion Each of the above studies examined some aspect of the relation between schizophrenia and psychopathy. In those studies that used criminal or noncriminal psychiatric populations, psychopathy-related diagnoses (according to PCL and APD criteria) appeared to have at least adequate c l i n i c a l specificity, especially with respect to schizophrenia-spectrum disorders; that i s , psychopaths tended to receive fewer other psychiatric diagnoses than did nonpsychopaths, and there was no consistent association between psychopathic and schizophrenic t r a i t s . The one study that did report a significant overlap between diagnoses of psychopathy and schizophrenia (Howard et a l . , 1984) was conducted in a criminal psychiatric setting. This study was also the most inadequate among the formal, published studies with respect to methodological issues. It remains a possibility, however, that the positive association reported by Howard et a l . i s not due to methodological problems, but i s only apparent in populations with histories of both serious criminality and serious mental disorder. Overall, perhaps the most important criticism of the studies reviewed above i s that each looked only at some isolated aspect of the psychopathy-schizophrenia overlap, for example, the correlation between PCL and EPQ scores, or the association between DSM-III APD and schizophrenia diagnoses. If psychiatric disorders are conceived of as hypothetical constructs located within a nomological network (e.g., 15 Cronbach and Meehl, 1955), then i t must be recognized that a disorder may be operationalized in many different ways. The operationalizations of a psychiatric disorder typically are either diagnostic (categorical) or t r a i t (dimensional) measures. A strong test of the relation between psychopathy and schizophrenia w i l l therefore use a range of categorical and dimensional measures related to each disorder. These measures should be standard, valid, and reliable. A second problem with these studies (with the exception of Howard et a l . , 1984) i s that they sampled from populations with a low base rate of serious psychiatric disorder. A strong test of the psychopathy-schizophrenia overlap would use subjects from a population with high base rates of both psychopathy and schizophrenia-spectrum disorders. A f i n a l issue is that of data analysis. The strategies used to test the relation between psychopathy and schizophrenia should depend on the nature of the measure used (categorical versus dimensional). The overlap among categorical measures should be quantified using the appropriate epidemiological s t a t i s t i c s ; associations among the categorical measures, on the other hand, can be quantified using correlational methods. 16 III. Purpose of the Present Study A review of the c l i n i c a l and experimental literature indicates that there are conflicting views concerning the intra-individual association between psychopathy and schizophrenia. One the one hand, i t has been proposed on the basis of theory (e.g., Eysenck and Eysenck, 1978) and research (Howard et a l . , 1984) that there i s a positive association between psychopathy and schizophrenia. On the other hand, there i s substantial evidence, in terms of both c l i n i c a l observation (e.g., Cleckley, 1976) and empirical research (e.g., Hare, 1982, 1985c), that psychopathy i s a c l i n i c a l l y specific disorder, and that there i s no positive association between psychopathy and schizophrenia (or any other psychiatric disorder). These two views of the relation between psychopathy and schizophrenia-spectrum disorders make differential predictions about their overlap within individuals. The 'positive association' hypothesis predicts that: a) individuals suffering from one of the disorders are more li k e l y than those not a f f l i c t e d to suffer from the other disorder; b) individuals suffering from one of the disorders are more lik e l y than those not a f f l i c t e d to have strong traits of the other disorder; c) trai t s related to the two disorders covary; and, f i n a l l y , d) the association between psychopathy and schizophrenia should be stronger than the association between psychopathy and other psychiatric disorders. The 'no association' hypothesis predicts that: a) individuals suffering from one of the disorders are no more lik e l y than those not a f f l i c t e d to suffer from the other disorder; b) individuals suffering from one of the disorders l i k e l y do not have strong traits of the other disorder; c) traits related to the two disorders do not 17 covary; and, f i n a l l y , d) the association between psychopathy and schizophrenia i s weak, and no stronger than the association between psychopathy and other psychiatric disorders. The present study was conducted to examine the intra-individual association between psychopathy and schizophrenia, and to test the competing hypotheses concerning this association, by looking at the distribution of psychopathy-related t r a i t s and diagnoses in a sample of forensic psychiatric inpatients. While the results of this study are of theoretical interest, the study is also a test of the relative c l i n i c a l s pecificities of the PCL and APD diagnostic c r i t e r i a . Information regarding specificity should be of immediate use to both researchers and clinicians. 18 IV. Method A. Subjects Subjects were 80 males remanded by the courts under Sec. 543 (2) C.C.C. to the Forensic Psychiatric Institute (FPI) i n Port Coquitlam, BC, for inpatient psychiatric assessment of fitness (competency) to stand t r i a l . The sample consisted of consecutive admissions to Ward R3W of the FPI between May 1 and October 31, 1986. The remand population at the FPI appears to be well-suited for the needs of the present study: f i r s t , a l l patients have been charged with criminal offenses, and most have long criminal histories; second, an offender i s only remanded for inpatient evaluation of fitness i f there is evidence that he i s currently suffering from a major mental illness, and most offenders have long psychiatric histories. A description of the subjects i s given in the Results section. B. Procedure 1. F i l e information In order to assess their fitness to stand t r i a l , patients remanded to the FPI routinely undergo a number of procedures, including psychiatric interviews, nursing assessments, psychological testing, social history, and medical examination. As well, information relevant to the issue of fitness i s obtained from independent sources; this information includes police reports, criminal records, medical and psychiatric f i l e s from other hospitals, interviews with acquaintances or relatives, and parole/probation officer reports. To allow independent raters to make diagnoses blind to FPI f i n a l diagnoses and fitness recommendations, a l l relevant f i l e information was recorded on a standard, 18-page form under the following headings: demographics, 19 including age, race, marital status, current residential status, and height and weight; marital and dating history; occupational history; family history, including interviews with relatives; educational history; psychiatric and medical history; current medical and neurological status; juvenile criminal record; adult criminal record; description and narrative of the present charges; drug and alcohol abuse; nursing assessment, including mental status exam; psychological assessment, including personality and intellectual tests; and psychiatric interviews. The f i l e synopsis specifically excluded any mention of the attending psychiatrists' and psychologists' diagnostic opinions, and their conclusions with regard to the subject's fitness to stand t r i a l ; this information was kept hidden from the independent raters. As well, to insure confidentiality, a l l names and addresses were removed from the f i l e s . The independent raters made their diagnoses on the basis of these these f i l e synopses. 2. Interview information Where possible, subjects were also interviewed by the author. A subject was not interviewed unless: a) the attending psychiatrist gave his written consent, stating that he believed interviewing would not be detrimental to the well-being of the patient; and b) the patient gave his written consent to be interviewed, after being f u l l y informed as to the nature, purpose, and limited confidentiality of the interview. The interview consisted of the Depression and Mania sections of the Diagnostic Interview Schedule (DIS: Robins et a l . , 1981), the Hallucinations and Delusions sections of the Present State Exam (PSE: Wing, Cooper, and Sartorius, 1984), the Structured Interview for DSM-III Personality (SIDP: Stangl et a l . , 1985), and a semistructured interview for the Psychopathy Checklist. Not a l l subjects interviewed were willing or able to complete a l l the interviews, and the length of the interviews ranged from two to six hours. Each subject's responses were recorded verbatim and the (unscored) interview protocols were placed with his f i l e synopsis. 3. Criminal history information To supplement and update any criminal record found in the f i l e s , each subject's o f f i c i a l crime sheet—his Royal Canadian Mounted Police (RCMP) Finger Print Service (FPS) sheet—was obtained through RCMP Headquarters in Ottawa, Ontario. The FPS sheets contain details about a l l offenses the subject has been charged with and the disposition of those charges. 4. DSM-III c l i n i c a l diagnoses and global ratings The author (Rater A), an MA-level graduate student in c l i n i c a l psychology with training in assessment and diagnosis, diagnosed each subject according to the Axis I c r i t e r i a contained in the Diagnostic and Sta t i s t i c a l Manual of the American Psychiatric Association, 3rd edition (DSM-III) after collating the f i l e synopses. The Axis I diagnoses were made on a two-point scale: 0 (no diagnosis) versus 1 (definite or provisional diagnosis). DSM-III Axis II diagnoses were also made using the same two-point scale. However, personality disorder (PD) diagnoses were made using only inclusion c r i t e r i a ; a l l exclusion c r i t e r i a based on a patient's other Axis I or II diagnoses were ignored. This relaxing of exclusion c r i t e r i a was done for three main reasons. First, the purpose of this study i s to examine the theoretical and diagnostic overlap between psychopathy- and schizophrenia-related disorders. It i s impossible to look at the overlap between DSM-III APD and schizophrenia diagnoses unless the APD exclusion criterion i s ignored because, as noted above, this criterion specifies that a person suffering from schizophrenia must not receive an additional diagnosis of APD, even i f he meets a l l the inclusion c r i t e r i a . Second, besides excluding APD, a diagnosis of schizophrenia also excludes several other schizophrenia-spectrum disorders (schizoid, schizotypal, and paranoid personality disorders). While for some purposes i t may make sense to ignore "lesser" disorders once a hierarchically superior diagnosis is made (e.g. exclude a diagnosis of Schizoid Personality Disorder once the patient meets the cr i t e r i a for Schizophrenic Disorder), i f used in the present research this procedure might underestimate any association between psychopathy and the f u l l spectrum of schizophrenic disorders. The third reason to ignore exclusion c r i t e r i a i s that they were ar b i t r a r i l y constructed: no clear and comprehensive theoretical or logical framework guides their application in the DSM-III, and their va l i d i t y and u t i l i t y have not been examined empirically (Boyd et a l . , 1984). The use of exclusion c r i t e r i a (other than those excluding organic factors) i s not consistent with the stated principles of DSM-III, which allow—and even recommend—multiple diagnoses. Exclusion c r i t e r i a , especially in the diagnosis of personality disorders, are used less extensively in the revision of DSM-III. In addition to these standard, categorical diagnoses, c l i n i c a l ratings of prototypicality were made for each of the 11 specified Axis II personality disorders and for schizophrenia. For each rating, Rater A made a global judgement of how closely the subject matched the more detailed descriptions of the disorders given in the DSM-III (rather 22 than the actual diagnostic c r i t e r i a ) . The ratings were made on a scale from 1 to 10, where 1 indicated 'very unlike' and 10 represented 'very lik e ' the description of the disorder. To check the r e l i a b i l i t y of the author's ratings, a second rater (Rater B ) — b l i n d to subjects' FPI fi n a l diagnoses and fitness recommendations, Rater A's diagnoses, and to the psychopathy diagnoses (discussed below)—made the same set of DSM-III categorical diagnoses and prototypicality ratings described above for each subject on the basic of f i l e synopses and interview protocols. Rater B i s a Ph.D. level c l i n i c a l psychologist with extensive training and experience in making diagnoses for research purposes. Finally, after both raters had made their diagnoses, they reviewed each case, discussed diagnostic and rating disagreements, and produced a f i n a l set of categorical diagnoses and prototypicality ratings (referred to below as Panel diagnoses and ratings). 5. Psychopathy Checklist diagnoses Rater A, who has extensive training and experience i n the assessment of psychopathy, also rated each subject according to Hare's 22-item Psychopathy Checklist (PCL22: Hare, 1980) and 20-item Revised Psychopathy Checklist (PCL20: Hare, 1985a) after collating the f i l e synopses. Each item in the PCL22 i s scored on three-point scale (0 to 2, with 2 indicating the definite presence of a psychopathic t r a i t ) , and then the individual item scores are summed to yield a total score, which can range from 0 to 44. The PCL20 i s scored in a similar manner, except that a rater has the option of omitting an item i f he feels he does not have sufficient information to rate i t reliably. As with the PCL22, item scores are summed, in this case to yield a total score 23 between 0 and 40; i f items are omitted, the total score is prorated to yield a score out of 40. (Note that neither the PCL22 nor the PCL20 have exclusion criteria.) To check the r e l i a b i l i t y of Rater A's PCL ratings, a third rater (Rater C ) — b l i n d to a l l subjects' FPI, Rater A, Rater B, and Panel diagnoses, as well as to the FPI fitness recommendations and Rater A PCL ratings—scored the PCL22 and PCL20 for each inmate on the basis of f i l e synopses and interview protocols. Rater C is a Ph.D. level graduate student in psychology who also has extensive experience in the assessment of psychopathy. Panel PCL total scores were obtained for subjects by averaging Rater A and Rater C total scores on the PCL22 and PCL20. Research suggests that the PCL scales have two underlying factors: one reflecting a glib, superficial, and manipulative personality and interpersonal style; and the other reflecting a chronically unstable and criminal l i f e s t y l e (Harpur, Hakstian, and Hare, 1987; Templeman and Wong, 1987). Factor scores for the PCL were calculated by averaging the individual PCL Item scores across the two raters, and then summing the scores for the Items loading on each factor. (Items omitted by raters on the PCL20 were replaced with scores of 1, except for Items 17 and 19, which were replaced with scores of 0.) 6. Blind Antisocial Personality Disorder diagnoses In addition to making PCL diagnoses. Rater C also made categorical DSM-III Antisocial Personality Disorder (APD) diagnoses using the two-point scale described above. These diagnoses, like the c l i n i c a l diagnoses described above, were made solely on the basis of inclusion c r i t e r i a . 24 7. Schizophrenia-related self-report measures Self-report measures related to schizophrenia were also examined. As part of the standard psychological assessment procedure at the FPI, patients were asked to complete the MMPI and the Millon C l i n i c a l Multiaxial Inventory (MCMI: Millon, 1983). Two scales from the MMPI are used below as schizophrenia-related t r a i t measures: Schizophrenia, or Sc, and Paranoia, or Pa. On the MCMI, six scales are potential schizophrenia-related t r a i t measures: the Schizoid/Asocial basic personality pattern (scale 1), the Schizotypal and Paranoid pathological personality patterns (scales S and P, respectively), and the psychotic thinking, psychotic depression, and paranoid delusion c l i n i c a l scales (scales SS, CC, and PP, respectively). 8. Psychopathy-related self-report measures Several psychopathy-related t r a i t measures can be derived from the MMPI. Research (Hare, 1985b) has indicated that two of the best measures are the K-corrected T-scores on the Psychopathic Deviate (Pd) and Hypomania (Ma) scales, and these measures are used below. The MCMI has only one psychopathy-related t r a i t measure: the Antisocial/Aggressive personality pattern (scale 6). The f i n a l BR score for this scale is used for calculational purposes below. 9. Other scales Two other scales were used in this study as measures of general, nonspecific psychiatric impairment. The Global Assessment Scale (GAS: Endicott, Spitzer, Fleiss, et a l . , 1976) i s a global rating of the level of adaptive functioning achieved by an individual, scored on a scale from 0 (extremely poor functioning) to 100 (fully functioning); the scale has 10 major divisions (with a range of 10 points each), and 25 each division has an accompanying anchor description. Both Rater A and Rater B scored the GAS independently for each subject (with Rater B blind as described above) on the basis of his lowest level of functioning in the 30-day period immediately preceding the commission of the offense(s) with which he was charged. The second scale used was the Brief Psychiatric Rating Scale (BPRS: Overall and Gorham, 1962). The BPRS requires the cl i n i c i a n to give the subject a global severity rating on a scale from 1 (not present) to 7 (extremely severe) for each of 19 symptom dimensions (e.g., psychomotor retardation, emotional withdrawal). These ratings can they be summed to yield a global estimate of the severity of psychiatric symptomatology. Each subject was rated on the BPRS by his attending psychiatrist on the basis of the subject's presentation in psychiatric interviews during the f i r s t week after his admission to the FPI. To check the r e l i a b i l i t y of the psychiatrists' BPRS ratings, the author also rated each subject he interviewed. 26 V. Results A. Characteristics of the sample 1. Race The majority of subjects (90.1%) were White, with the rest being of Native Indian (5.0%), Asian (2.6%), or mixed (2.6%) descent. Most (91.1%) were raised in English-speaking homes. 2. Age The average age at admission was 32.6 (SD = 10.8) years. Subjects ranged in age from 18 to over 68 years. Most subjects had (allegedly) committed the offenses they were charged with about a week prior to their admission to the FPI. 3. Marital status The majority of subjects (57.5%) were single and had never been married or lived common-law. Only 8 subjects (10.0%) were married or in a common-law relationship at admission, while the rest (32.5%) were separated from their spouses. 4. Education In general, subjects in the present sample had poor educational records. 01: the 77 subjects whose educational level could be verified, 15 (19.5%) had received only special education, and had never attended regular academic classes. Of the remaining subjects, eleven (14.3%) completed some elementary school, and 36 (46.8%) completed some secondary school. Seven subjects (9.1%) had received some post-secondary training, and a further 8 (10.4%) had graduated from high school. 5. Social class The social class of subjects was coded on the basis of their 27 occupational record according to the scale developed by Hollingshead and Redlich (1958), which ranges from 1 (highest class) to 7 (lowest class). Subjects were scored on the basis of the best job they had held for a minimum of six months. Of the 79 subjects whose reported employment history could be at least partially verified, f u l l y 73.4% f a l l i n the lowest social class; indeed, many of these subjects have never really been in the labour force at a l l . The highest social class achieved by any subject was level 3, and i t was reached by only 3 subjects (3.8%), with the rest (22.9%) f a l l i n g between levels 4 and 6. Overall, the mean social class was 6.43 (SD = 1.09). Despite the low social class achieved by subjects, i t would be incorrect to assume that they were mainly from low social class families. If social class i s scored on the basis of the best job held by one of the subject's parents or caretakers during his childhood, a much different picture emerges. Of the 53 subjects whose familial social class could be verified, 43.3% come from level 3 or higher (with 7.5% coming from a level 1 background). Only 9.4% came from the lowest level background, with the rest (47.1%) coming from backgrounds between levels 4 and 6. The mean familial social class was 4.04 (SD = 1.74), significantly higher than the subjects' mean occupational social class: t (52) = 8.48, £ < .001. It could be argued that this mean familial social class i s a biased estimate, as i t could only be determined for more stable—and thus probably higher class—families; however, even i f a l l the remaining families were from class 7, the lowest class, the mean familial social class would be 5.01 (SD = 2.00), s t i l l significantly higher than subjects' mean occupational social class: t (78) = 5.94, £ < .001. It appears as though subjects in this sample have experienced some form of social d r i f t (Kohn, 1973). 28 6. Psychiatric history Almost a l l of the subjects had some history of psychiatric intervention: of the 78 subjects whose psychiatric history could be verified, only 12 (15.4%) had never been seen by a mental health professional. A small number of subjects had only been assessed or treated on an outpatient basis (10.2%), while the majority had received inpatient psychiatric treatment at a hospital (43.6%) or had previously been institutionalized (30.8%); several patients had been hospitalized more than ten times in the past. A considerable number of subjects (24, or 30%) had been admitted to the FPI previously; thirteen had previously either been found unfit to stand t r i a l on account of mental illness or had their charges stayed and were diverted into the mental health system. 7. Criminal history Twenty-five subjects (31.3%) had a confirmed history of juvenile delinquency; of those subjects for whom figures were available, the average age at f i r s t contact with juvenile court was 14.00 (SD = 1.50) year s. The majority of subjects (78.8%) had been previously charged with or convicted of an adult offense. The average age at f i r s t appearance in court was 20.52 (SD = 5.53) years. For these subjects, the mean number of months spent in adult correctional f a c i l i t i e s was 9.67 (SD = 17.91), and ranged from 0 to 83 months. On average, they had been convicted of .839 (SD = 1.200) and charged with a further .421 (SD = .600) offenses per year not incarcerated since age 16. Most of them had at least one known alias, and some had five to ten. 8. Instant offense There was great va r i a b i l i t y in the types of crimes subjects were 29 charged with. On average, each was charged with 2.48 (SD = 1.96) offenses, but the actual numbers ranged from 1 to 13. Table I groups offense types into 11 major categories, and indicates the most serious type of offense subjects were charged with. The offense types varied greatly, ranging from breach of probation to first-degree murder. From Table I, i t i s clear that most of the offenses were f a i r l y serious and involved contact with a victim. Indeed, according to police records, 36 subjects (45.0%) caused some degree of physical harm to their victims; four of these subjects i n f l i c t e d severe injuries, and two actually k i l l e d their victims (one of these latter subjects had two victims). Most subjects (allegedly) committed their offenses while sober; f i l e s indicated that only 10 subjects (12.5%) were definitely and seriously impaired by recent drug or alcohol use. A further 22 subjects (27.5%) appeared to have a psychosis of some type which impaired their social judgement, while in the case of 6 subjects the commission of the offense appeared to be directly related to a psychotic process (e.g. command hallucinations, paranoid delusions). B. R e l i a b i l i t y of the measures 1. DSM-III Axis I major disorders The Axis I primary diagnoses (excluding the Substance Abuse/Dependence disorders) were collapsed into six superordinate categories: NONE, including those subjects with no major Axis I diagnosis and those with diagnoses of malingering and borderline intellectual functioning; MR, including those subjects diagnosed mentally retarded; ORG, including subjects with substance-induced and other organic disorders; SZ, including a l l schizophrenic and related disorders (e.g. schizophreniform disorder); BP contains those subjects Table I Most serious offense committed by subjects Offense type Frequency 1. Breach of probation, breach of recognizance, etc. 3 (3.8%) 2. Willful damage, cause a disturbance 3 (3.8%) 3. Theft, break and enter 15 (18.8%) 4. Assault, aggravated assault 12 (15.0%) 5. Criminal negligence, dangerous driving, driving while intoxicated 2 (2.5%) 6. Robbery, armed robbery 2 (2.5%) 7. Possession of a dangerous weapon, assault with a weapon 19 (23.8%) 8. Sexual assault 14 (17.5%) 9. Arson 5 (6.3%) 10. Kidnapping, unlawful confinement 2 (2.5%) 11. Murder, attempted murder 3 (3.8%) 31 diagnosed bipolar disorder; and AAD, including those diagnosed with anxiety, dissociative, or affective (other than bipolar) disorders. Table II shows the agreement between Raters A and B for these Axis I disorders. As is evident from Table II, there was good agreement between the two raters; the overall kappa coefficient of agreement was .681 (the overall Axis I kappas for similar superordinate categories obtained during the DSM-III f i e l d t r i a l s were .68 [phase 1] and .72 [phase 2]; see APA, 1980: p.470). The individual kappa coefficients for the various disorders were as follows: NONE = .698; MR = .779; ORG = .614; SZ = .716; BP = .779; and AAD = .422. From Table II, i t i s obvious that most (50%) of the diagnostic disagreements between the two raters f a l l into two categories. First, the differential diagnosis between schizophrenia and organic mental disorders accounted for 5 disagreements. The major problem here i s that the present sample had a high rate of polydrug abuse (see below), and a number of subjects diagnosed substance abusers also had a history of psychosis. In some cases i t was unclear from the subject's history whether drug abuse could account for a l l his psychotic symptomatology, and i t appears that Rater A had a (nonsignificant) tendency to label these cases organic rather than schizophrenic. The second major problem area was the diagnosis of affective and dissociative disorders. From inspection of subjects' specific diagnoses, there was some disagreement over the number and severity of depressive and anxiety-related symptoms subjects reported, with Rater B having a (nonsignificant) tendency to label reported symptoms as c l i n i c a l l y significant and therefore to make a diagnosis. The Panel Axis I diagnoses were collapsed into the same six Table II Interrater agreement for the six superordinate DSM-III Axis I major diagnostic groups Rater B Rater A NONE MR ORG SZ BP AAD 1. NONE 12 0 1 2 0 4 2. MR 0 6 1 0 0 0 3. ORG 1 2 13 4 0 0 4. SZ 0 0 1 21 1 1 5. BP 0 0 1 1 6 0 6. AAD 0 0 0 0 0 2 Note. Overall kappa coefficient of agreement i s .681. For explanation of acronyms, see text or Appendix A. superordinate categories. Group NONE (n = 16) includes 12 subjects with no major DSM-III Axis I disorder, one with a diagnosis of borderline intellectual functioning, two with diagnoses of malingering, and one with a diagnosis of pedophilia. Group MR (n = 8) subjects a l l received a Panel diagnosis of mild retardation. Group ORG (n = 16) includes 7 subjects with substance-induced organic disorders and 9 subjects with organic disorders whose etiology was related to some brain trauma. Group SZ (n = 27) contains 23 subjects with acute schizophrenia (9 paranoid, 5 disorganized, and 9 undifferentiated), 1 subject with residual schizophrenia, 2 schizophreniform disorders, and 1 patient with a paranoid disorder. Group BP contains 8 patients with bipolar disorder, a l l of whom were in a manic phase at the time of admission. Group AAD (n = 5) contains 1 panic disorder, 2 depressed, and 2 atypical bipolar (Bipolar II) disorder subjects. 2. Substance abuse/dependence disorders Only 16 subjects (20%) did not receive some type of substance abuse or dependence disorder from Raters A and B, and there was a high incidence of polydrug abuse. It was therefore decided to divide the various substance abuse and dependence diagnoses into two main categories: Alcohol disorders versus Other Substance disorders. The abuse and dependence diagnoses of Raters A and B were then used to assign subjects to one of three levels within each category: nonusers, abusers, and those with a dependency. Table III shows the agreement between Raters A and B when these categorizations were used. The two raters agreed on 78.8% of the specific alcohol abuse/dependence diagnoses; the kappa coefficient of agreement in this case was .668. The raters also agreed on 80.0% of the Other Substance abuse/dependence diagnoses (kappa = .690). Table III Interrater agreement and base rate of Panel diagnoses of the Alcohol and Other Substances abuse/dependence disorders Rater A Rater B Alcohol Number of Panel Diagnoses None Abuse Dependence None 34 1 0 38 Abuse 6 12 4 17 Dependence 0 6 17 25 Other Substances None Abuse Dependence None 25 2 0 29 Abuse 1 14 2 32 Dependence 0 4 21 19 35 Table III also l i s t s the prevalence of Panel diagnoses for the Alcohol and Other Substances abuse/dependence dimensions. Twenty-five subjects (31.3%) received a diagnosis of alcohol dependence and 17 (21.3%) a diagnosis of alcohol abuse disorders; the other 38 received no alcohol-related diagnosis. With respect to Other Substance disorders, 19 subjects (23.8%) received a diagnosis of dependency and 32 (40%) a diagnosis of abuse; only 29 did not receive some Other Substance diagnosis. 3. Axis II personality disorders The categorical COM and PAG diagnoses are excluded from further analysis below, as they had Panel diagnosis base rates of 0. When diagnoses were made on the basis of inclusion c r i t e r i a only, there was a high prevalence of personality disorders in the present sample. Overall, 65 subjects (81.2%) received at least one PD diagnosis from either Rater A or Rater B, and 56 subjects (70.0%) received at least one PD diagnosis from the Panel. There was good agreement between Raters A and B as to whether or not a subject had a personality disorder: ignoring the specific diagnosis, kappa = .587. Table IV presents the interrater r e l i a b i l i t i e s for the individual PD diagnoses. Looking at the individual DSM-III PDs, the average level of interrater agreement for the specific PD diagnoses was f a i r . The mean kappa was .433, and the individual kappas ranged from a low of .068 (or chance level) for AVD to a high of .875 for APD. By far the most prevalent PD was APD, with half the subjects (50.0%) receiving a Panel diagnosis. The next most frequent PDs were HIS and SZT (n = 14 and 7, respectively, for Panel diagnoses). 36 Table IV Interrater agreement and base rate of Panel diagnoses of DSM-III personality disorders Agree Number of Personality Panel Disorder Present Absent Disagree Kappa diagnoses 1. PAR 4. 69 7 .491 5 2. SZD 3 70 7 .417 5 3. SZT 7 62 11 .480 11 4. HIS 7 61 12 .450 14 5. NAR 1 75 4 .319 3 6. APD 37 38 5 .875 40 7. BOR 3 67 10 .307 7 8. AVD 1 67 12 .068 5 9. DEP 1 77 2 .490 1 Note. A l l diagnoses made ignoring exclusion c r i t e r i a . For explanation of acronyms, see text or Appendix A. 37 The r e l i a b i l i t y coefficients of the PD diagnoses reported here are somewhat lower than those reported recently by other researchers (e.g., Loranger, Susman, Oldham, and Russakoff, 1987; Pfohl et a l . , 1986). However, those researchers typically use standardized assessment techniques (including corroborative assessment interviews with close relatives) and either outpatients or nonpsychotic inpatients. When researchers make personality disorder diagnoses on the basis on standard hospital assessments and f i l e information, as was the case for the majority of patients in the present study, they obtain interrater r e l i a b i l i t i e s very similar to those reported here (Mellsop, Varghese, Joshua, and Hicks, 1982). 4. PCL scales Interrater r e l i a b i l i t y (Pearson r) was .914 for the PCL22 and .912 for the PCL20, both £ < .001. These coefficients are similar in magnitude to those reported for the PCL22 by Schroeder, Schroeder, and Hare (1983) and by others (e.g., Raine, 1985). On the PCL22, the mean rating of Rater A was 26.83 (SD = 6.96) and the mean for Rater C was 26.38 (SD = 6.33), a nonsignificant difference: t (79) = 1.43, £ = .158. On the PCL20, the means for Raters A and C were 21.91 (SD = 6.94) and 22.03 (SD = 6.88) respectively, t (79) = -0.38, £ = .704. The distribution of Panel PCL scores was as follows: PCL22, M = 26.60 (SD = 6.49); and PCL20, M = 21.97 (SD = 6.76). The internal consistency of both scales was also examined. The individual item scores for both scales were summed across raters (for the PCL20, omissions by a rater on Items 17 and 19 were given a score of 0; on the remaining Items, a score of 1 was substituted). The resulting PCL22 and PCL20 scores had coefficient alphas of .844 and 38 .857, respectively. These coefficients are very similar to those obtained by Hare and his associates in studies with nonpsychiatric criminal populations (e.g., Schroeder, Schroeder, and Hare, 1983). To test the hypotheses about PCL group overlap with c l i n i c a l diagnoses, the dimensional ratings were used to form three groups: high scorers, or psychopaths (Group P); low scorers, or nonpsychopaths (Group NP); and a second group of nonpsychopaths with moderate scores (Group M). The cutoffs recommended by Hare (1985a, 1985b) were used to form these groups. The cutoffs for the PCL22 were: Group P, 34 or more; Group M, 25 to 33; and Group NP, 24 and less. For the PCL20, the cutoffs were: Group P, 30 or more; Group M, 21 to 29; and Group NP, 20 and less. The interrater r e l i a b i l i t y of the PCL categorical diagnoses is given in Table V. Several things are obvious from Table V. First, there was excellent agreement between raters on the PCL categorical diagnoses; for the PCL22, the kappa was .850, and for the PCL20 i t was .743. If we collapse these diagnoses into two groups—psychopaths versus others— the kappas are .945 for the PCL22 and .845 for the PCL20. Second, note that the base rate of psychopathy according to the Panel diagnoses was 12.5% for both the PCL22 and PCL20; this figure i s somewhat lower than that found in populations of provincial and federal prison inmates, where the base rate of psychopathy is typically 20% to 30% (Hare, 1982; Wong, 1985). In the present sample, the base rate of psychopathy was also much lower than that of APD. Finally, there were no extreme disagreements between the raters: no subject was placed in Group NP by one rater and in Group P by the other. Table V Interrater agreement and base rate of the Panel PCL categorical diagnoses Rater B Number of Panel PCL Rater A P M NP Kappa Diagnoses PCL22 .850 P 10 1 0 10 M 0 40 3 43 NP 0 3 23 27 PCL20 .743 P 10 2 0 10 M 1 38 3 44 NP 0 6 20 26 Note. Group P = psychopaths; Group NP = nonpsychopaths; Group M = middle-scoring nonpsychopaths. 1Kappa coefficient of agreement 40 The two PCL scales were also in good agreement with each other as to the classification of subjects into the three PCL groups: they were in exact agreement on 86% of subjects (three-group kappa = .758). If the groups are collapsed into psychopaths versus others (as above), they disagree on the classification of only four subjects (exact agreement = 95%, kappa = .771). 5. Blind APD diagnosis There was excellent agreement between the Panel APD diagnoses and the APD diagnoses made by Rater C, who was blind to subjects' c l i n i c a l diagnoses. Table VI shows the agreement between the two sets of ratings. There was exact agreement between the Blind and Panel APD diagnoses on 86.3% of the cases (kappa = .725). Note also that the base rates of APD for the two sets of ratings were high, and almost identical (blind = 48%, Panel = 50%). 6. Schizophrenia global rating (SZGR) The interrater r e l i a b i l i t y of the SZGR was very high: the Pearson product-moment correlation between Rater A and Rater B SZGRs was r = .899, £ < .001. There was no significant difference between the average ratings of Ratar A (M = 4.61, SD = 2.98) and B (M = 4.51, SD = 3.21): t (79) = 0.63, £ = .899. The Panel SZGRs were distributed with a mean of 4.43 (SD = 3.04). 7. Personality disorder global ratings (PDGRs) There were very low base rates for two of the PDGRs (COM and PAG) in the present sample. Over 95% of subjects received a rating of 1 from both raters and the Panel for these two PDs, and no subject received a rating higher than 4; as a consequence, the mean Panel, Rater A, and Rater B PDGRs for both disorders were very close to 1, with extremely Table VI The agreement between Blind and Panel APD diagnoses Blind APD diagnosis (Rater C) Panel APD diagnosis Not APD APD Not APD APD 35 6 5 34 Note. For explanation of acronyms, see text or Appendix A. 42 small standard deviations. Therefore, the PAG and COM PDGRs were not analyzed further. Table VII l i s t s the means and standard deviations of the PDGRs of both Rater A and Rater B for each of the 9 remaining specified DSM-III Axis II personality disorders. Table VII also contains a) t-tests of the differences between the mean PDGRs of Rater A and B, and b) interrater r e l i a b i l i t y coefficients (Pearson product-moment correlations between Rater A and B PDGRs) for each personality disorder. The mean PDGRs of Raters A and B were very similar; only in one case did the difference between the raters reach significance (for NAR). The interrater r e l i a b i l i t i e s were generally f a i r to good, and a l l highly significant; the average was .640. This relatively low mean r e l i a b i l i t y may be due in part to the skewness of the distribution of some PDGRs: note that the average rating on most disorders ranged between 1 and 3 out of 10. The most reliable rating was for APD (r = .776), the disorder which also had the highest mean GRs from both Raters. The means for the Panel PDGRs were as follows: PAR, M = 2.35 (SD = 1.42); SZD, M = 2.14 (SD = 1.46); SZT, M = 2.68 (SD = 1.83); HIS, M = 2.61 (SD = 1.97); NAR, M = 1.88 (SD = 1.21); APD, M = 5.20 (SD = 2.42); BOR, M = 2.49 (SD = 1.65); AVD, M = 1.78 (SD = 1.41); and DEP, M = 1.48 (SD = 0.95). 8. GAS and BPRS Only 79 ratings could be made on the GAS; there was insufficient information for one subject to rate his level of adaptive functioning in the month prior to the commission of the offense. Based on these 79 43 Table VII Interrater agreement for the PDGRs Rater A Rater B Personality Disorder Mean (SD) Mean (SD) t a r 1. PAR 2.55 (1.40) 2.48 (1.79) .52 .696* 2. SZD 2.11 (1.27) 2.31 (1.79) -1.22 .584* 3. SZT 2.94 (1.97) 2.55 (2.13) 1.91 .613* 4. HIS 2.85 (2.01) 2.58 (2.12) 1.45 .667* 5. NAR 1.75 (1.09) 2.18 (1.64) -3.09* .659* 6. APD 5.55 (2.26) 5.39 (2.70) 0.85 .776* 7. BOR 2.81 (1.92) 2.39 (1.70) 2.07 .490* 8. AVD 1.76 (1.40) 1.79 (1.61) -0.15 .524* 9. DEP 1.70 (1.22) 1.46 (1.03) 2.10 .607* Note. For explanation of acronyms, see text or Appendix A. a t - t e s t s on the difference between the means of Raters A and B, a l l df = 79; the familywise Type I error rate was held at alpha p w = .05 by setting the testwise Type I error rate at alpha T W = alphas/9 = .006, according to the Dunn-Bonferroni procedure. Pearson product-moment correlations between PDGRs of Raters A and B, a l l N = 80; the familywise Type I error rate was held at alpha p w = .05 by setting the testwise Type I error rate at a l p h a ^ = alphas/9 = .006, according to the Dunn-Bonferroni procedure. *2™, < '05; p^, < .006 epw ' CTW 44 ratings, the interrater r e l i a b i l i t y of the GAS was f a i r : the Pearson product-moment correlation between Rater A and Rater B was r = .694, p_ < .001. There was no significant difference between the average ratings of Rater A (M = 36.69, SD = 10.42) and B (M = 36.77, SD = 8.93): t (79) = -0.09, p = .930. The Final (panel) GAS scores have M = 36.20 (SD = 9.10). Attending psychiatrists' BPRS scores were obtained for 79 subjects. These scores had a mean of 49.97 (SD = 16.37). Rater A was able to score the BPRS for 31 subjects. For the BPRS Total scores of these 31 subjects, the mean of Rater A was 47.71 (SD = 15.12) and the mean for the attending psychiatrists i s 47.39 (SD = 14.97), a nonsignificant difference: t (30) = 0.17, p_ = .869. The interrater r e l i a b i l i t y (Pearson r) of the BPRS total score was .742, p_ < .001. C. The association between psychopathy and schizophrenia 1. Overlap between psychopathy-related measures and DSM-III Axis I major disorders Table VIII presents the overlap between between the six superordinate Axis I major diagnostic groups and the PCL22 and PCL20 groups. Looking at specific diagnoses, the PCL22 and PCL20 psychopathy diagnoses had good c l i n i c a l specificity; few Group P subjects also received a major DSM-III Axis I diagnosis. For both versions of the PCL, the c l i n i c a l specificity coefficient, or p (no major Axis I diagnosis/Group P), was .60. None of the Group P subjects received a diagnosis of mental retardation, bipolar (manic) disorder, or an anxiety, dissociative, or non-bipolar affective disorder. With respect to the psychopathy-schizophrenia association i n particular, note that the PCL scales did not have perfect specificity: Table VIII The overlap between a) the six superordinate Axis I major diagnostic groups, and b) the PCL22, PCL20, and Blind APD groups DSM-III Diagnostic Group PCL22 PCL20 APD NP M P NP M P Absent Present 1. NONE 5 5 6 4 6 6 5 11 2. MR 3 5 0 4 4 0 4 4 3. ORG 6 8 2 7 6 3 8 8 4. SZ 10 15 2 9 17 1 17 10 5. BP 2 6 0 1 7 0 4 4 6. AAD 1 4 0 1 4 0 3 2 Note. APD diagnoses ignore the exclusion criterion. For explanation of acronyms, see text or Appendix A. 46 two of the PCL22 and one of the PCL20 psychopaths f e l l in the SZ diagnostic group (these subjects are examined in greater detail below). Depending on which scale is used, then, either 10% or 20% of psychopaths were given a diagnosis of schizophrenia; however, only 3.7% to 7.4% of schizophrenics were diagnosed psychopaths. To test the two hypotheses concerning the psychopathy-schizophrenia association, the observed and expected proportions of subjects within each PCL group who received a specific diagnosis were compared. According to the hypothesis of no association between psychopathy and schizophrenia, we would expect that the same proportion of subjects within each PCL group should f a l l into the various diagnostic categories. In the case of schizophrenia, there was a trend for both PCL22 and PCL20 psychopaths to receive fewer diagnoses of schizophrenia than expected on the basis of chance; however, these 2 trends were nonsignificant, X (2, N = 27) = .663 and 1.671 respectively, both p_ > .30. Among the other diagnostic groups, the only 2 significant difference that emerged was for NONE: X (2, N = 16) = 7.244 and 6.820 for PCL22 and PCL20 respectively, both p < .05. In both cases, Group P had more, and Group M fewer, subjects with no major Axis I diagnosis than would be expected on the basis of chance. A second way to quantify the association between psychopathy and Axis I groups i s to calculate cross-product ratios, or odds ratios (ORs). ORs are a measure of association that are easily calculated from 2x2 contingency tables, and have several positive characteristics (Bishop, Fienberg, and Holland, 1975; Fienberg, 1980). Perhaps their most appealing characteristic i s that they are readily interpretable: in this study, the OR is the probability of a diagnosis given the 47 subject is a psychopath divided by the probability of a diagnosis given the subject i s a nonpsychopath. Therefore, ORs less than 1 indicate a negative association between the the diagnosis in question and psychopathy and ORs greater than 1 indicate a positive association; values close to 1 indicate no association. Table IX shows the ORs between PCL psychopathy diagnoses and the DSM-III major diagnostic groups. From Table IX, the strongest finding is that PCL22 and PCL20 diagnoses of psychopathy were positively associated with NONE, that i s , no DSM-III Axis I diagnosis (both ORs = 9.000, SD = 6.573). The second finding i s that these diagnoses were significantly and negatively associated with SZ, or Axis I schizophrenia-related disorders; for PCL22, OR = .398, and for PCL20, OR = .188. This suggests that psychopaths were less than half as l i k e l y as nonpsychopaths (those in Group NP or M) to receive a schizophrenia-related diagnosis. Third, PCL psychopathy appears to be more or less unrelated to ORG diagnoses, suggesting the disorders are independent (for PCL22, OR = 1.000, and for PCL20, OR = 1.879). Finally, ORs for the other disorders were a l l 0, as no PCL22 or PCL20 psychopaths received these diagnoses. Table VIII also shows the overlap among APD and the Axis I major diagnostic superordinate categories. APD had somewhat lower c l i n i c a l specificity than did the PCL scales. About 71% of subjects who received a diagnosis of APD (blind with respect to their present Axis I status) received a major Axis I diagnosis; the c l i n i c a l specificity coefficient for the blind APD diagnoses was therefore .29. Subjects diagnosed APD were included in every one of the six DSM-III superordinate diagnostic groups, in rates about the same as for those subjects diagnosed non-APD. 48 Table IX Odds ratios (ORs) showing the association between a) the PCL22, PCL20, and Blind APD groups, and b) the Axis I major diagnostic groups DSM-III Psychopathy-related diagnosis Diagnostic Group PCL22 PCL20 APD 1. NONE 9.000 (6.573) 9.000 (6.573) 2.829 (1.684) 2. MR 0 0 1.057 (.788) 3. ORG 1.000 (.845) 1.879 (1.420) 1.065 (.595) 4. SZ .398 (.330) .188 (.204) .487 (.236) 5. BP 0 0 .771 (.616) 6. AAD 0 0 .685 (.645) Note. Figures in brackets are SD. For explanation of acronyms, see text or Appendix A. « 49 The APD diagnoses also had lower specificity than did the PCL scales with respect to schizophrenia in particular. Note that 26% of APD subjects received a diagnosis of schizophrenia (or, looking at i t another way, 37% of schizophrenic subjects also received a diagnosis of APD). Despite their lower specificity, the APD diagnoses (like the PCL 2 diagnoses) were not positively associated with schizophrenia. The X goodness-of-fit statistics for the diagnostic groups indicate that there were no differences between the APD and non-APD groups i n the proportion of subjects receiving any of the various diagnoses, or in 2 the number of subjects in each group with no diagnosis ( a l l X < 1.824, £ > .30). As Table IX shows, the ORs for APD versus the diagnostic groups also indicate that APD was less specific than were the PCL psychopathy diagnoses: f i r s t , a l l of the ORs were greater than 0, because subjects in every diagnostic group also met the inclusion c r i t e r i a for APD; second, most of the ORs were very close to 1; and f i n a l l y , the OR for APD versus NONE (2.829) was less than a third the size of the PCL psychopathy versus NONE ORs. However, once again this lower specificity does not imply a positive association between APD and schizophrenia. In fact, APD was significantly and negatively associated with SZ (OR = .487), although the negative association was weaker than was the case with the PCL scales. An examination of the specificity of APD diagnoses with the exclusion criterion l e f t intact might be useful to researchers and clinicians. We can make a (liberal) estimate of the specificity of APD diagnoses with the exclusion criterion intact by recategorizing a l l subjects in the SZ/APD and BP/APD cells into their respective non-APD c e l l s . (MR 50 subjects are not recategorized here, as the exclusion criterion rules out only severe or profound retardation and a l l MR subjects were only mildly retarded.) Using this procedure, the base rate of APD dropped to .313, and the specificity of the diagnosis increases by about 50% to .44; however, i t was s t i l l considerably below that of the PCL scales. That i s , even i f the DSM-III exclusion criterion for APD had been used, over half of these APD subjects (14 of 25) would have received some other major Axis I diagnosis. 2. Overlap between psychopathy-related and alcohol and other substance abuse/dependence diagnoses No specific predictions were made with respect to alcohol or other substance abuse and dependence diagnoses. However, i t is well established by previous research that both psychopathy (as diagnosed by the PCL) and APD are associated with high rates of a l l types of substance abuse and dependence (e.g., Goodwin and Guze, 1984). As this information may prove interesting or useful to other researchers, i t is presented below. Table X shows the relation between: a) the PCL22, PCL20, and blind APD groups; and b) the Alcohol and Other Substances abuse/dependence groups. There was a high degree of alcohol and substance abuse in the current sample, and this abuse cut across a l l diagnostic categories. Two features of this table are worth noting: f i r s t , as expected, Group P had the highest rates of abuse and dependence among the PCL22 and PCL20 groups; and second, also as expected, subjects diagnosed APD had a greater rate of abuse/dependence than did non-APD subjects. 3. Overlap between psychopathy-related and other Axis II PD diagnoses It was noted above that, ignoring exclusion c r i t e r i a , 70.0% of Table X The overlap between a) Alcohol and Other Substances abuse/dependence diagnoses, and b) the PCL22, PCL20, and Blind APD groups Alcohol Other Substances Diagnosis None Abuse Dep. None Abuse Dep. PCL 2 2 NP 16 4 7 17 4 6 M 18 11 14 11 22 10 P 4 2 4 1 6 3 PCL 20 APD NP 16 5 5 16 4 6 M 19 11 14 11 24 9 P 3 1 6 2 4 4 Not APD 27 6 8 20 13 8 APD 11 11 17 9 19 11 Note. For explanation of acronyms, see text or Appendix A. 52 subjects in this sample received some Panel PD diagnosis. As one might expect, these personality-disordered subjects were represented in a l l the diagnostic groups discussed above. Analysis of overlap with PDs i s complicated by the fact that a subject could receive multiple PD diagnoses. Table XI shows the overlap between a) the PCL22, PCL20, and blind APD groups, and b) the Panel Axis II PD diagnoses made ignoring the exclusion c r i t e r i a . The Table also shows the number of subjects in each group who received any PD diagnosis. The most obvious feature of Table XI i s the large (and expected) overlap between a) the PCL scales and Blind APD diagnoses, and b) the Panel APD diagnoses. Among schizophrenia-spectrum PD diagnoses, note that the PCL scales had perfect specificity with respect to PAR and SZT, and lower (but s t i l l good) specificity with respect to SZD. APD had lower specificity with respect to a l l three disorders; indeed, over half of the SZD subjects also received a Blind APD diagnoses. The biggest overlap, however, was between the psychopathy-related diagnoses and HIS: 30% of PCL22 psychopaths, 40% of PCL20 psychopaths, and 23% of Blind APD subjects also received this diagnosis. With respect to the remaining PD diagnoses, the PCL scales had perfect specificity; APD, however, overlapped with every one. Table XII presents the overlap in a different way, excluding Panel APD diagnoses and grouping the remaining diagnoses into the three clusters described in DSM-III: O-E, or odd-eccentric, comprised of the paranoid, schizoid, and schizotypal disorders (in fact, a l l the schizophrenia-spectrum PDs); D-E-E, or dramatic-emotional-erratic, comprised of the histrionic, narcissistic, and borderline disorders (recall APD i s excluded here); and A-F, or anxious-fearful, comprised Table XI The overlap between the PCL22, PCL20, and Blind APD groups and the Panel personality disorder diagnoses Panel Personality Disorder diagnoses Any PD Diagnosis PAR SZD SZT HIS NAR APD BOR AVD DEP diag-nosis PCL22 NP 3 2 2 2 0 8 2 3 . 0 16 M 2 2 9 9 3 24 5 2 1 32 P 0 1 0 3 0 8 0 0 0 8 PCL20 NP 4 2 3 1 0 7 2 3 0 15 M 1 1 8 9 3 24 5 2 1 32 P 0 2 0 4 0 9 0 0 0 9 Blind APD Not APD 4 2 7 5 2 6 3 3 0 21 APD 1 3 4 9 1 34 4 2 1 35 Note. Panel PD diagnoses made ignoring exclusion c r i t e r i a . For explanation of acronyms, see text or Appendix A. Table XII The overlap between the PCL22, PCL20, and Blind APD groups and the three DSM-III personality disorder clusters Psychopathy- DSM- •III PD Cluster Any related non-APD Diagnosis 0-E D-E-E A-F diagnosis PCL22 NP 4 3 3 10 M 10 13 2 23 P 1 3 0 3 PCL 20 NP 4 2 3 9 M 9 13 2 22 P 2 4 0 5 Blind APD Not APD 8 6 3 16 APD 7 13 2 20 Note. A l l Panel PD diagnoses made ignoring exclusion c r i t e r i a . For explanation of acronyms, see text or Appendix A. 1Not including Panel APD diagnoses. 55 of the avoidant and dependent disorders (no Panel compulsive or passive-aggressive diagnoses were made). The number of subjects in each of the PCL22, PCL20, and Blind APD groups with at least one diagnosis in a cluster i s given in this Table. Finally, the Table l i s t s the number of subjects i n each group with any non-APD Panel PD diagnosis. For the PCL22, the overall specificity of a psychopathy diagnosis with respect to non-APD personality disorders was .70. The overlap with the 0-E (schizophrenia-related) cluster was 10%; in comparison, the PCL22 overlap with the D-E-E cluster was 30%, and there was no overlap with the A-F cluster. PCL20 had somewhat lower specificity (.50). Again, there was some overlap with 0-E (20%), but most of the overlap was with the D-E-E cluster (50%); there was none with A-F. The blind APD diagnoses had slightly lower specificity than either of the PCL scales (.487). However, both had the pattern of overlap with the DSM-III PD clusters: 17.9%, 33.3%, and 5.1% with 0-E, D-E-E, and A-F, respectively. The association between the DSM-III PD clusters and psychopathy-related diagnoses can be quantified by calculating ORs; these statistics are presented in Table XIII. F i r s t , i t should be noted that none of the ORs was significantly different from 1, indicating that the null hypothesis of the independence of diagnoses cannot be rejected. However, some trends are apparent in the Table. First, psychopathy-related diagnoses were associated with an increased likelihood of non-APD diagnoses within the D-E-E cluster; in the case of APD, this increase was almost threefold. Second, the psychopathy-related measures were negatively associated with diagnoses in the A-F cluster. Finally, with respect to the psychopathy-schizophrenia association, psychopathy-56 Table XIII Odds ratios (ORs) showing the association between a) the PCL22, PCL20, and Blind APD diagnoses, and b) the three DSM-III personality disorder clusters Psychopathy-related Diagnosis DSM-III PD Cluster 0-E D-E-E1 A-F PCL22 .444 (.487) 1.446 (1.080) 0 PCL 20 1.096 (.930) 2.444 (1.731) 0 Blind APD .902 (.518) 2.917 (1.626) .685 (.645) Note. Figures in brackets are SD. For explanation of acronyms, see text or Appendix A. ^Not including Panel APD diagnoses. 57 related measures appear to be independent of diagnoses in the 0-E cluster, although the PCL22 tended to be negatively associated with this cluster. 4. Correlations among psychopathy-related dimensional measures Two types of dimensional measures were collected in this study: clinical.(PCL scores and global ratings) and self-report. C l i n i c a l ratings were made for a l l subjects. Scores on the self-report scales, however, were not available for a l l subjects for a variety of reasons. First, a considerable number of subjects either refused to complete one or both inventories. Second, some subjects were inaccessible to psychological assessment due to f l o r i d psychosis. Third, i l l i t e r a t e and intellectually retarded subjects were not administered the inventories. Finally, a few subjects relied on a gross response set (e.g., a l l true, a l l false) to complete the inventories. As a result of these factors, the psychological f i l e s of the subjects contained only 50 MCMI and 35 MMPI profiles, and only 32 subjects completed both inventories. However, some profiles in this group were produced by patients with severe psychopathology and are technically invalid. Of the completed MCMI inventories, 12 can be discarded due to extreme scores on the Validity scale (V > 0) or the symptom complaint/denial index (Sum 1-8 < 95 or > 165). Ten MMPI profiles can be discarded due to extreme scores on the F-K index (F-K < -11 or > 9: e.g., Greene, 1980). Calculations involving valid MCMI profiles are based on n = 38, those involving valid MMPI profiles on n = 25, and those with valid profiles on both inventories on n = 22. The correlations among the ten psychopathy-related dimensional measures are presented in Table XIV. The patterns that emerge in the 58 Table XIV Correlations among the psychopathy-related dimensional measures Psychopathy-related Measure PCL 2 2 PCL 20 PDGR MMPI MCMI a a b c a b c a a b PCL 2 2 a) Total — 77* 91* 95* 77* 88* 76* 10 27 11 b) Factor 1 80 — 46* 78* 94* 48* 40* -18 -09 -04 c) Factor 2 80- 80 — 83* 48* 94* 83* 24 45 23 PCL 20 a) Total 80 80 80 — 85* 87* 71* 21 26 06 b) Factor 1 80 80 80 80 — 53* 44* 00 00 -11 c) Factor 2 80 80 80 80 80 — 80* 34 52 20 Panel PDGR a) APD 80 80 80 80 80 80 — 34 48 23 MMPI a) Pd 25 25 25 25 25 25 25 — 60 -13 b) Ma 25 25 25 25 25 25 25 25 — 32 MCMI a) Scale 6 38 38 38 38 38 38 33 22 22 — Note. Numbers above the diagonal are Pearson product-moment correlations (decimals omitted); numbers below the diagonal are the Ns for each calculation. The familywise Type I error rate for the correlation matrix i s held at alpha_ r 7 = .05 by setting the testwise Type I error rate at alpha^ = alpha p w/45 = .001, according to the Dunn-Bonferroni procedure. For explanation of acronyms, see text or Appendix A. *P_„r, < «05; p m i = .001 CFW ' CTW 59 Table are very similar to those noted by Hare (1985b) in his comparison of procedures for the assessment of psychopathy. F i r s t , c l i n i c a l measures related to psychopathy were moderately to highly intercorrelated. Total scores on two PCL scales were highly correlated with each other (r = .95) and with Panel APD GRs (rs = .76 and .71 for the PCL222 and PCL20, respectively). Second, the various self-report measures related to psychopathy were only weakly intercorrelated; indeed, although similar in magnitude to those reported by Hare (1985b), none of these correlations reached significance in the present study. The clinical-behavioral measures were weakly correlated with self-report measures; once again, none of the correlations reached significance. One thing Hare (1985b) was not able to examine was the relation between the PCL Factor scores and other assessment procedures. As noted above, Factor 1 reflects the glib, superficial, and manipulative personality characteristics of psychopathy, while Factor 2 reflects an impulsive, unstable, and criminal l i f e s t y l e (Harpur, Hakstian, and Hare, 1987). While the two Factors are correlated, Factor 1 may be a relatively pure measure of the core personality characteristics of psychopathy, while Factor 2 may be related more to general criminality. It i s apparent that both the PCL22 and PCL20 Factor 2 scores correlated more highly with the Panel APD GR than did the Factor 1 scores, consistent with the view that APD i s more a measure of criminality than of personality per se (Hare, 1983; Millon, 1981). Similarly, Factor 2 also correlated more highly with the various self-report measures related to psychopathy than did Factor 1, suggesting that the self-report measures also reflect general criminality or delinquency more 60 than they do personality characteristics (e.g., Hawk and Peterson, 1974). The trend for APD to correlate more highly with the general criminality component of the PCL scales than with their personality component is also apparent i f we examine the point-biserial correlations between Factors 1 and 2 and categorical APD diagnoses; the trend holds true whether Blind or Panel APD diagnoses are.used as a criterion. These correlations are presented in Table XV. The correlations between Factor 2 and APD ranged from a low of .59 to a high of .67, and averaged about .63. For Factor 1, the same correlations ranged from .16 to .42, and averaged about .28. In each case, the correlation between Factor 1 and Blind or Panel APD diagnosesis significantly higher than the corresponding correlation between Factor 2 and APD diagnoses ( a l l p < .003). 5. Correlations among the schizophrenia-related dimensional measures The correlations among the twelve schizophrenia-related dimensional measures are presented in Table XVI. Note that the c l i n i c a l global ratings were only moderately intercorrelated, while the self-report measures were generally moderately to highly intercorrelated. Some high correlations among the schizophrenia-related t r a i t measures within the two self-report inventories were to be expected, as the scales share a large number of items (Dahlstrom and Welsh, 1960; Millon, 1983). As was the case with the psychopathy-related measures, however, there was not a single significant correlation between the c l i n i c a l ratings and the self-report measures; the largest correlation was only .22 (between the SZD PDGR and MCMI scale PP). 61 Table XV Correlations between the factor scores on the PCL scales and APD diagnoses APD diagnoses PCL • Factor Blind Panel PCL22 a) Factor 1 .34 (p = .002) .16 (p_ = .153) b) Factor 2 .66 (p_ < .001) .60 (p < .001) PCL 20 a) Factor 1 .42 (p_ < .001) .21 (p_ = .067) b) Factor 2 .67 (p < .001) .59 (p_ < .001) Note. N = 80. For explanation of acronyms, see text or Appendix A. 0 62 Table XVI Correlations among the schizophrenia-related dimensional measures Panel GRs MMPI MCMI a b c d a b a b c d e f Panel GRs a) SZGR — 39* 26 61* -06 -22 -08 04 -06 00 -09 08 b) PAR 80 — 36* 55* -02 -14 -16 -17 -12 -02 00 07 c) SZD 80 80 — 42* -09 21 -02 10 04 22 18 05 d) SZT 80 80 80 — 13 -10 -05 11 -19 -03 05 -07 MMPI a) Pa 25 25 25 25 — 53 20 55 34 39 58 13 b) Sc 25 25 25 25 25 — 58 54 05 58 62 -17 MCMI a) 1 38 38 38 38 22 22 — 77* 11 61* 41 01 b) S 38 38 38 38 22 22 38 — 14 49 50 -15 c) P 38 38 38 38 22 22 38 38 — 48 04 83* d) SS 38 38 38 38 22 22 38 38 38 — 67* 37 e) CC 38 38 38 38 22 22 38 38 38 38 — -06 f) PP 38 38 38 38 22 22 38 38 38 38 38 — Note. Numbers above the diagonal are Pearson product-moment correlations (decimals omitted); numbers below the diagonal are Ns for each calculation. The familywise Type I error rate for the correlation matrix i s held at approximately alpha p w = .05 by setting the testwise Type I error rate at alpha_, = alpha_.7/66 < .001 (to three decimal places), according to the Dunn-Bonferroni procedure. * E F W < -05; < .001 63 6. Correlations between psychopathy-related and schizophrenia-related dimensional measures The correlations among the psychopathy-related and schizophrenia-related dimensional measures are presented in Table XVII. Several trends are apparent in this Table. First, there was no significant relation between the c l i n i c a l measures of psychopathy-related and schizophrenia-related disorders; many of the correlations were negative, and a l l were very close to 0 in magnitude. Second, there was no relation between c l i n i c a l ratings of schizophrenia-related disorders and self-report measures of psychopathy. The highest correlation among this latter group was r = -.40 (n.s.), between the MMPI Pd scale and the Panel SZGR. Third, although the correlations among the psychopathy-related c l i n i c a l and schizophrenia-related self-report measures were in the moderate range (typically around .30), a l l were nonsignificant. Finally, the highest correlations in Table XVIII were those between the psychopathy-related and schizophrenia-related self-report scales. The MMPI Sc scale was significantly correlated with the MMPI Pd and Ma scales (r = .66 and .72, respectively), and the MMPI Pa scale was also moderately correlated with these latter scales (r = .54 ard .40, n.s.). Among the MCMI scales, CC correlated most highly with Pa and Sc (r = .57 and .50, n.s.). Interestingly, scale 6 from the MCMI did not correlate with Pa or Sc from the MMPI; i t s highest correlations were with MCMI scales P and PP (r = .51 and .38, n.s.). Because the MCMI and MMPI scales correlated primarily with other scales from the same inventory, the most parsimonious explanation for these associations i s that there was item overlap among the inventories' various scales. 7. Factor analysis of c l i n i c a l ratings Another way of using dimensional measures to examine the Table XVII Correlations among the psychopathy- and schizophrenia-related dimensional measures Schizophrenia- PCL22 PCL20 PDGR MMPI MCMI related Measure a b c a b c a a b a Panel GRs a) SZGR -07 -03 -06 -10 -05 -06 -18 -40 -03 -05 b) PAR -03 05 -08 -01 05 -11 -15 -05 00 -29 c) SZD 14 09 12 11 13 12 06 -14 16 10 d) SZT -03 -06 00 -06 -07 -02 -17 -13 04 -25 MMPI a) Pa 28 07 33 32 17 40 36 54 40 05 b) Sc 32 10 40 40 26 51 42 66* 72* 13 MCMI a) 1 40 32 36 38 31 30 29 17 29 25 b) S 39 26 36 45 34 39 27 31 30 15 O P -14 -04 -15 -11 -02 -12 -14 -26 14 51 d) SS 38 22 40 37 22 41 37 20 45 35 e) CC 39 22 41 42 26 45 39 57 50 13 f) pp -12 -01 -12 -14 -02 -16 -12 -41 07 38 Note. Ns vary from 22 to 80. PCL22 and PCL20: a = Total score, b = Factor 1, c = Factor 2. PDGR: a = Panel PDGR. MMPI: a = Pd, b = Ma. MCMI: a = 6 (Antisocial/Aggressive). For an explanation of the other acronyms, see text or Appendix A. The familywise Type I error rate for the correlation matrix i s held at approximately alpha p w = .05 by setting the testwise Type I error rate at alpha,^ = alphas/120 = .001 (to three decimal places), according to the Dunn-Bonferroni procedure. * E F W < -05; E t w < .001 65 association between psychopathy and schizophrenia i s to perform a factor analysis of the various clinical-behavioral ratings and see what factor(s) the psychopathy-related ratings load on. A l l the Panel PDGRs (excluding COM and PAG), Panel SZGR, and PCL scales—12 scales in total—were factor analyzed. A principal components analysis was used to extract four factors according to both Scree and Kaiser-Guttman c r i t e r i a ; factors 1 through 4 account for 27.2%, 24.2%, 12.4%, and 11.0% of the common variance, respectively (74.8% of the variance in total). A l l variables had good communality (greater than .50). The factors were then subjected to a number of different oblique and orthogonal rotations until a simple structure was obtained. A varimax rotation produced the best solution, with a l l variables loading on at least one factor, only two factorially-complex variables, and 48.6% of the nonsignificant loadings f a l l i n g on the factor hyperplanes. The factor loadings and communalities of the variables are presented in Table XVIII. The four factors appear to be easily interpretable. Factor 1, defined exclusively by the PCL scales and APD, reflects psychopathy/antisociality. Factor 2 i s a schizophrenia-spectrum factor, and i s defined by the SZT, PAR, SZD, and SZGR variables. Factor 3 seems to reflect neuroticism/high anxiety: DEP, AVD, and BOR a l l load highly on i t . (Interestingly, the SZGR variable i s factorially complex, and loads significantly but negatively on this factor; this may be due to the flattened and blunted affect exhibited by many schizophrenics in this sample.) Finally, Factor 4, defined by the BOR, NAR, and HIS variables, seems to reflect unstable emotionality. There i s a remarkable similarity between these factors and the Table XVIII Factor loadings (and communalities) of the 12 c l i n i c a l ratings Factor Communality Panel „ Rating 1 . 2 3 4 (h ) 1. PCL 2 2 .92 -.05 .07 -.06 .92 2. PCL 20 .89 .01 -.20 .21 .91 3. APD .92 .00 -.20 .27 .86 4. SZT -.08 .85 -.10 -.16 .76 5. PAR .02 .78 -.14 .12 .64 6. SZD .14 .67 .10 -.20 .53 7. SZGR -.27 .65 -.40 -.01 .65 8. DEP -.11 -.20 .83 .01 .75 9. AVD -.23 -.02 .77 -.07 .66 10. BOR .13 -.07 .62 .54 .70 11. NAR .09 .00 -.18 .86 .78 12. HIS .19 -.23 .22 .83 .82 Note. For explanation of acronyms, see text or Appendix A. 67 three DSM-III Axis II personality disorder clusters described above: Factor 2 corresponds to the O-E-E cluster, Factor 3 to the A-F cluster, and Factor 4 to the D-E-E cluster. The factor analysis and DSM-II clusters are not identical, however. First, BOR i s factori a l l y complex, with significant and positive loadings on Factors related to both the D-E-E and A-F clusters, while DSM-III includes BOR in the D-E-E cluster only. Second, and most importantly, note that APD and the PCL scales define a unique factor that i s orthogonal to the others; DSM-III, on the other hand, places APD within the D-E-E cluster. These results lend some empirical support for the clustering of personality disorders described i n DSM-III, although they suggest that there may be problems including the Borderline and Antisocial disorders within the existing clusters. With respect to the c l i n i c a l specificity of psychopathy-related diagnoses, these results suggest that PCL and APD ratings are best conceived of as independent of (uncorrelated with) other major dimensions of psychopathology in the present sample. In particular, with respect to schizophrenia-related disorders, there i s no evidence for a correlation between psychopathic and schizophrenic t r a i t dimensions. 8. Correlation between psychopathy-related measures and standardized psychiatric rating scales The analyses reported thus far have only examined the relation between psychopathy-related measures and specific psychiatric diagnoses, c l i n i c a l ratings, and self-report measures. It i s conceivable that psychopathy-related measures could be unrelated to specific indicators of psychiatric illness (such as those described 68 above) while s t i l l being associated with global, nonspecific, or undifferentiated psychiatric symptomatology. Therefore, the association between psychopathy-related measures and two standardized rating scales of the severity of psychiatric impairment—the GAS and BPRS—were examined. There was no significant correlation between subjects' best level of functioning in the month before the commission of the offense for which they were charged, as measured by the GAS, and the PCL22, PCL20, or Panel APD GRs; these latter scores were also uncorrelated with the attending psychiatrist's assessment of the severity of psychiatric symptomatology, as measured by total scores on the BPRS ( a l l rs between -.07 to -.20, p > .07). One problem with these correlational analyses is that there may be a curvilinear relation between psychopathy-related measures and scores on the GAS and BPRS. More specifically, i f Group M of the PCL scales i s associated with greater psychiatric impairment than Groups NP and P, as predicted by Hare and Forth (1985), then correlational analyses w i l l not reflect this relation. Therefore, a series of one-way ANOVAs was performed using the PCL and Blind APD groups as independent variables and the GAS and BPRS scores as dependent variables. The means of the PCL and Blind APD groups on the GAS and BPRS are presented in Table IXX. (Note that low score on the GAS represent poor overall functioning, while high scores on the BPRS reflect the presence of severe psychiatric symptomatology). Once again, there is no support for a positive association between psychopathy and global psychiatric impairment. Looking at the PCL scales, there was a consistent trend for Group M to be more impaired than Groups NP and P, according to scores 69 Table IXX Means of the PCL22, PCL20, and Blind APD groups on two standardized psychiatric rating scales Psychiatric Rating Scale Psychopathy-related Measure Panel GAS Total BPRS PCL22 a) NP 38.7 (12.5) 48.5 (14.4) b) M 33.9 (6.1) 52.1 (17.9) c) P 39.1 (6.7) 43.2 (16.4) PCL 20 a) NP 38.5 (12.7) 48.8 (15.5) b) M 34.4 (6.7) 52.0 (17.2) c) P 37.9 (9.1) 42.5 (13.5) APD (Blind) a) Not APD 35.9 (10.5) 52.1 (15.8) b) APD 36.6 (7.5) 47.2 (16.8) Note. N = 79. BPRS = Brief Psychiatric Rating Scale (Overall and Gorham, 1962); GAS = Global Assessment Scale (Endicott, Spitzer, Fleiss, et a l . , 1976). For an explanation of the other acronyms, see text or Appendix A. 70 on the GAS and BPRS, and Group P appears to be the least impaired on both rating scales. Only one of the group differences reached significance, however: PCL20 Group M had lower scores on the GAS than did the other two groups, F (1, 77) = 3.08, p = .05. The differences between subjects with and without a Blind APD diagnosis were much smaller than the group differences on the PCL scales, and a l l the differences were nonsignificant. However, there was a trend for APD subjects to be less impaired than non-APD subjects. 9. Schizophrenic traits in the PCL and Blind APD groups Table XX presents the mean scores of the PCL and Blind APD groups on the schizophrenia-related c l i n i c a l measures (SZGR and the PAR, SZD, and SZT PDGRs). The results are generally consistent with the hypothesis of no association between psychopathy and schizophrenia. Looking at the PCL scales, there was a trend for Group P to score the lowest, and Group M the highest, on the schizophrenia-related measures. The one exception was for the SZD PDGR: on the PCL22, Group NP was higher than Groups M and P (which were equal), and on the PCL20, Group P was higher than Groups NP and M. However, MANOVAs indicate that nei.ther the PCL22 nor the PCL20 group differences on these four schizophrenia-related measures were significant: F (8, 148) = 1.18 and 1.69, respectively (Wilk's method). Similarly; looking at the Blind APD groups, those subjects with no APD diagnosis had higher SZGR and SZT PDGRs than did those diagnosed APD; the scores of the two groups on the PAR and SZD PDGRs were equal. Again, a MANOVA indicated that these between-group differences were nonsignificant: F (4, 75) = 1.13, p = .348 (Wilk's method). 10. Psychopathic traits in the Axis I diagnostic groups Table XXI presents the mean scores for the DSM-III Axis I Table XX Means of the PCL22, PCL20, and Blind APD groups on the schizophrenia-related c l i n i c a l ratings Group SZGR PAR SZD SZT M (SD) M (SD) M (SD) M (SD) PCL22 a) NP 4.4 (3.1) 2.3 (1.7) 2.2 (1.5) 2.5 (1.6) b) M 4.8 (3.0) 2.4 (1.3) 2.1 (1.4) 3.0 (2.0) c) P 2.8 (2.6) 2.5 (1.5) 2.1 (1.6) 1.7 (1.2) PCL 20 a) NP 4.4 (3.3) 2.3 (1.9) 2.1 (1.5) 2.6 (1.8) b) M 5.0 (2.9) 2.4 (1.0) 2.1 (1.4) 3.0 (1.9) c) P 2.1 (1.9) 2.2 (1.6) 2.5 (1.8) 1.5 (1.1) APD (Blind) a) Not APD 5.0 (3.0) 2.3 (1.4) 2.1 (1.4) 3.0 (1.9) b) APD 3.8 (3.0) 2.3 (1.5) 2.1 (1.5) 2.4 (1.7) Note. N = 80. For explanation of acronyms, see text or Appendix A. Table XXI Means of the Axis I major diagnostic groups on the PCL22, PCL20, and Panel APD PDGR Axis I PCL22 PCL20 APD PDGR Diagnostic Group M (SD) M (SD) M (SD) 1. NONE 29.9 (7.8) 25.4 (8.1) 6.2 (2.5) 2. MR 25.6 (3.8) 19.6 (3.7) 5.6 (1.9) 3. ORG 25.3 (7.3) 21.1 (8.2) 5.6 (2.4) 4. SZ 25.6 (5.9) 20.7 (5.9) 4.3 (2.4) 5. BP 27.8 (4.4) 23.9 (3.4) 5.3 (2.0) 6. AAD 25.4 (7.3) 21.0 (7.3) 5.0 (3.0) Note. N = 80. For explanation of acronyms, see text or Appendix A. 73 diagnostic groups on the PCL22, PCL20, and the Panel APD PDGR. Contrary to the predictions of the hypothesis of a positive association between psychopathy and schizophrenia, subjects in the SZ group did not have higher scores than did subjects in other diagnostic groups on the psychopathy-related measures. Oneway ANOVAs revealed no significant differences among the groups on the various dimensional measures: F (5, 74) = 1.21, 1.45, and 1.46 for the PCL22, PCL20, and APD PDGR, respectively ( a l l p_ > .20). There was a clear trend for subjects in group NONE to receive the highest scores on each measure, followed by subjects in the BP group (on both PCL scales) and those in groups MR and ORG (on the APD PDGR). On average, subjects in group SZ had the lowest scores on the APD PDGR. Overall, the range of PCL scores in the present sample was 9.5 to 40.5 for the PCL22, and 8.16 to 38.5 for the PCL20. For subjects in the SZ group, the ranges were 12.5 to 34.0 and 8.5 to 31.5 for the PCL22 and PCL20, respectively. That i s , even the most psychopathic schizophrenics in the present sample had scores that were only marginally higher than the standard PCL cutoffs. These results suggest that i t may be possible for researchers studying forensic psychiatric populations to use the PCL to select pure psychopathic groups merely by raising the PCL cutoff scores. On the basis of the data presented here, the following cutoffs might be recommended for determining membership in Group H (psychopaths) in forensic psychiatric populations: PCL22, 37 and above; PCL20, 34 and above. 74 V. Discussion A. Summary of results and comments This study was designed to examine the association between psychopathy and schizophrenia. The results presented here f a i l to support the arguments (e.g., Eysenck and Eysenck, 1976, 1978; Howard et a l . 1984) that there i s a positive association between psychopathy and schizophrenia. First, although diagnoses of psychopathy (according to PCL criteria) did not have perfect specificity with respect to schizophrenia-related c l i n i c a l diagnoses, the overlap was small, and st a t i s t i c a l analyses indicated that the PCL scales were in fact either not associated or negatively associated with these disorders. Similarly, diagnoses of antisocial personality disorder (according to DSM-III criteria) were generally not associated with schizophrenia-related disorders; however, the DSM-III c r i t e r i a had lower c l i n i c a l specificity than did the PCL cr i t e r i a with respect to both schizophrenia-related and other psychiatric disorders. Second, there was no correlation between psychopathy- and schizophrenia-related c l i n i c a l ratings, and a factor analysis indicated that ratings related to the two disorders load on different personality dimensions. Third, self-report scales from the MMPI and MCMI related to psychopathy and schizophrenia were not correlated with any c l i n i c a l ratings; the only significant correlations here were between psychopathy- and schizophrenia-related scales of the MMPI, correlations that are li k e l y attributable in part to item overlap in the construction of the scales. Fourth, psychopathy and antisocial personality disorder diagnoses and c l i n i c a l ratings were not related to scores on standard rating scales of the severity of psychiatric symptomatology (GAS and BPRS). Finally, 75 there was no difference between schizophrenic and non-schizophrenic subjects in the strength of psychopathy-related traits, and no difference between psychopathic and nonpsychopathic (or antisocial personality disorder subjects versus others) in the, strength of schizophrenia-related t r a i t s . The present study improved on previous studies relevant to the association between psychopathy and schizophrenia in several ways. First, both c l i n i c a l constructs were operationalized in several ways, using categorical and dimensional (including self-report) measures. Diagnoses were made using standard and reliable c r i t e r i a , and experimental bias was minimized by using blind raters to check the r e l i a b i l i t y of a l l diagnoses and c l i n i c a l ratings. Second, subjects were sampled from a forensic psychiatric population, a population which intuitively should have a high base rate of both psychopathy-related disorders and serious psychopathology. Intuition was correct in this case: As a group, subjects had long criminal and psychiatric histories. While the base rates of psychopathy and antisocial personality disorder observed were somewhat lower than those of normal Canadian prisons, the rate of serious mental disorder was much higher (e.g., over 33% diagnosed schizophrenic). Finally, a wide range of epidemiological and traditional parametric statistics was used to examine the association between psychopathy and schizophrenia. The results presented here suggest that psychopathy may not be a "half-way house to psychosis" (Eysenck and Eysenck, 1978, p. 213). As well, despite sampling from a forensic psychiatric population, the present study does not replicate the findings of Howard et a l . (1984). There was no consistent or significant association between psychopathy 76 and schizophrenia-related disorders. It seems that, consistent with the views of Hare and his colleagues (Hare, 1984; Hare and Forth, 1985; Hare and Harpur, 1986), the PCL measures a relatively distinct and c l i n i c a l l y specific disorder that i s generally unrelated to other forms of psychopathology. Although PCL diagnoses of psychopathy are not perfectly specific, they seem to be at least independent of, and possibly negatively associated with, most psychiatric disorders. Two points need to be made here. Fi r s t , this study focuses on symptomatic and diagnostic issues. It i s possible that research may yet reveal some deeper (e.g., etiological) link between psychopathy and schizophrenia on a cognitive, psychophysiological, or genetic level, consistent with Eysenckian theory. Second, even i f Eysenck's prediction that psychopathy and criminality are related to psychosis proves to be incorrect, his three-dimensional model may s t i l l be useful in the investigation of the personality in criminal populations. For example, in the factor analysis of c l i n i c a l ratings reported above, four orthogonal factors were identified: one was interpreted as a psychopathy-antisocial personality factor, and the other three factors were interpreted to reflect the DSM-III personality disorder clusters. However, the latter three factors can also be interpreted as the Eysenckian personality dimensions of psychoticism, neuroticism, and extraversion. To help c l a r i f y the interpretation of the psychopathy-antisocial personality factor, I performed the factor analysis again, this time using the scores on the two PCL factors instead of the PCL total scores. Again, four orthogonal factors emerged, almost identical to those in the f i r s t analysis. This time, however, the psychopathy-antisocial personality factor, was defined by the Panel APD global 77 ratings and scores on Factor 2 of both PCL scales; i t was interpreted to reflect criminality (i.e., a tendency to engage in criminal behaviors or lead a criminal lifestyle) rather than a personality factor per se. PCL Factor 1, the factor reflecting psychopathic personality characteristics, loaded moderately and positively on this criminality factor. In addition, Factor 1 loaded moderately and positively on the extraversion-related factor, moderately and negatively on the neuroticism-related factor, and was orthogonal to the psychoticism factor (that i s , in Eysenckian terms, PCL psychopaths could be described as somewhat sociable and impulsive, with l i t t l e anxiety and no predisposition to psychosis). This analysis suggests that Eysenck's three personality dimensions may be useful in the description of the major dimensions of abnormal personality in the present sample. Although the PCL scales did demonstrate good overall c l i n i c a l specificity and were not positively associated with schizophrenia-related disorders, there i s some indication that they (and the APD criteria) may be positively associated with diagnoses of histrionic personality disorder. This finding would not surprise some authors, who posit that psychopathy i s male equivalent of "hysteria" and that the disorders may share an etiological mechanism (Goodwin and Guze, 1984). The relatively low base rates of both psychopathy and histrionic personality disorder in the present sample reduced the pos s i b i l i t y of finding any significant association. 78 B. Assessing psychopathy in a forensic psychiatric population 1. The performance of the PCL scales The PCL scales proved to be extremely reliable in the present sample. The presence of psychopathology in subjects did not appear to adversely affect the interrater r e l i a b i l i t y of individual Item or Total scores or the internal consistency of the scales. The distribution of PCL22 and PCL20 scores was also acceptable; they had approximately the same mean and standard deviation in the present sample as they do in normal prison populations. The PCL scales exhibited a high degree of c l i n i c a l specificity. Half of the psychopathy overlap with Axis I was with organic disorders, diagnoses that are theoretically not inconsistent with psychopathy. Three subjects received a diagnosis of psychopathy on the PCL22 or PCL20 and a definite diagnosis of organic disorder. Of these three subjects, one had a serious closed head injury and the other two had substance organic disorders. The former subject was clearly psychopathic/antisocial since at least his mid-teens. At age 25 he f e l l into a construction pit while walking home drunk, and required extensive neurosurgery. Unfortunately, according to the attending psychiatrist and psychologist, this accident merely seemed to disinhibit him further. He was charged with attempted murder after he stabbed a cab driver in the face; apparently, the sole motivation for the crime was racial prejudice. The two other subjects differed greatly in age; one was 20, while the other was 55. However, both were admitted to the FPI suffering from transient organic mental disorders related to substance use. The younger subject had been sniffing gasoline and glue almost continually since age 13, while the older subject was a long-time alcoholic who had suffered from an episode of alcohol hallucinosis. If one was to ignore these Axis I organic disorders, the c l i n i c a l specificity coefficients of the PCL22 and PCL20 would rise to .80 and .90, respectively. Two subjects were diagnosed as psychopathic by one of the two PCL scales. The f i r s t subject, age 28, was diagnosed definitely paranoid schizophrenic by his attending FPI psychiatrist and by both raters in the present study. He was admitted to the FPI after being charged with indecent assault; most of his crimes were very similar and apparently related to a bizarre religious delusion. Interestingly, although his f i l e revealed that he had been reporting psychotic symptoms since age 15 and that he had multiple inpatient admissions to psychiatric f a c i l i t i e s before age 20, mental health professionals saw this patient as primarily personality disordered due to his history of heavy substance abuse and delinquency (also, some professionals doubted the validity of his reports of psychotic symptoms). His f i r s t schizophrenia-related diagnosis was not until age 22, but his primary diagnosis was schizophrenia at every hospitalization from that time on. He received a score of 34 on the PCL22, and a score of 29 on the PCL20. The second subject was admitted to the FPI at age 42. He was charged with possession of a dangerous weapon after an incident where he threatened a woman with a knife while she was walking her dog and forced her to l e t the dog loose. This subject has an extremely long and varied criminal record, having been convicted of multiple offenses including theft, possession of drugs, robbery, dangerous driving, and fraud. He has at least 11 aliases. He also has a long history of drug abuse (starting in his teens) and psychiatric hospitalization (since 80 age 27). There is considerable disagreement in this man's f i l e as to his psychiatric diagnosis: Although he frequently presents with apparently psychotic symptoms and has been living a transient, 'bag man' l i f e s t y l e for some years, some professionals feel that his symptoms are attributable entirely to either drugs or malingering. Diagnosis is hampered by the lack of f i l e information and his tendency to give a new history upon every hospital admission. This uncertainty was reflected in his research diagnoses. His Axis I Panel diagnosis was paranoid schizophrenia, chronic with acute exacerbation (possible); Rater A's diagnosis was unspecified substance mixed organic mental disorder, while Rater B's was paranoid schizophrenia, chronic with acute exacerbation. This subject received a score of 34 on the PCL22 and 31.5 on the PCL20. Because certain aspects of their histories (especially the criminal and other antisocial behaviors) were so clearly documented, i t appears that these two subjects were not diagnostic errors of the PCL scales—that i s , schizophrenics misdiagnosed as psychopaths; rather, they may have been actual cases of the co-occurrence of psychopathic and schizophrenic syndromes. Indeed, due to the unreliable nature of self-reports of symptomatology (especially in criminal and psychopathic populations), i t is more lik e l y that any diagnostic errors, i f they occurred, involved the misdiagnosis of psychopaths as schizophrenics. Finally, a possible strength of the PCL scales for use in forensic psychiatric settings i s their flexible nature. As noted above, the cutoff scores used to classify subjects into various groups can be altered according to the base rate of psychopathic traits in the 'local' population (Finn, 1982). 81 2. The performance of APD c r i t e r i a The DSM-III APD c r i t e r i a have several characteristics that make them problematic for use in a forensic psychiatric settings. For one, they have an arbitrary exclusion criterion that does not allow those with schizophrenia to receive the diagnosis. This means that even those schizophrenic patients with serious criminal histories and psychopathic features cannot receive the diagnosis, thus denying other professionals of what may be valuable descriptive or predictive information. Of course, one can always ignore the exclusion criterion, as was done in the present study. However, i t appears that the APD inclusion c r i t e r i a do not have high c l i n i c a l specificity. At least one subject in every Axis I and Axis II diagnostic group also received a diagnosis of APD. Although the overlap was not large, and although APD was not positively and significantly associated with very many disorders, APD diagnoses may have less descriptive val i d i t y than PCL diagnoses in this sample. For example, some subjects diagnosed APD were also diagnosed with dependent, schizotypal, and borderline personality disorder, and with bipolar disorder, panic disorder, and mental retardation. No subjects with a diagnosis of psychopathy-' also received these diagnoses. From this perspective, a label of APD has less heuristic value than a label of psychopathy; psychopathy i s probably not typically associated with psychiatric descriptors such as "intellectually deficient," "anxious," or "dependent," while APD may be. Another undesirable characteristic of APD diagnoses i s that they are inflexible. Unlike the PCL, the APD c r i t e r i a cannot be adjusted for fluctuations in local base rates. This means that researchers and clinicians cannot easily improve the natural specificity of the APD 82 c r i t e r i a just by adjusting a cutoff, as may be the case with the PCL c r i t e r i a . Rather, i t is necessary to do further research to determine i f specificity can be boosted by merely increasing the number of symptoms required to receive a diagnosis or i f the l i s t of symptoms themselves must be altered. Note that these problems with the DSM-III c r i t e r i a are lik e l y to be just as bad— i f not worse—with the DSM-III-R c r i t e r i a for APD. First, the inclusion c r i t e r i a s t i l l consist almost entirely of checklists of antisocial behaviors, with the addition of several childhood c r i t e r i a and one adult criterion and removal of the criterion requiring continuous antisocial behavior. These changes appear likely to increase the number of people who can receive the diagnosis. Second, the revised exclusion criterion allows the diagnosis to be made in the case of schizophrenia i f there is "a clear pattern of antisocial behavior" and requires only that the "Occurrence of antisocial behavior [is] not exclusively during the course of Schizophrenia or Manic Episodes" (APA, 1987, p.344-346). To sum, the inclusion c r i t e r i a have been expanded and may be even more inclusive than in the DSM-III, while at the same time the exclusion criterion has been relaxed, almost certainly lowering c l i n i c a l specificity. Further research w i l l be necessary to determine the c l i n i c a l specificity of the new c r i t e r i a . 3. The performance of self-report scales The results presented here suggest that two of the most popular self-report inventories presently in use in North America—the MMPI and the MCMI—may not be helpful to clinicians working in forensic psychiatric settings. First , the psychopathy-related scales from those inventories did not correlate with any psychopathy-related c l i n i c a l 83 ratings, consistent with the findings of Hare (1985b). Second, the schizphrenia-related self-report scales did not correlate with any schizophrenia-related c l i n i c a l ratings. These scales appear to have poor descriptive validity, and poor predictive validity with respect to predicting c l i n i c a l diagnosis, at least i n the present sample. The poor performance of the self-report scales is not surprising, considering the known tendency of criminals to engage in dissimulation and to use response sets (e.g., Hare, 1985b). Although the psychopathy- and schizophrenia-related self-report scales were not linearly related to c l i n i c a l measures, i t i s s t i l l possible that interpretation of the entire inventory profile may s t i l l yield valid and reliable information. C. Directions for research Future research on the psychopathy-schizophrenia association can addressed several questions. First , a replication of the present study would be useful. That study could use a methodology similar to the one used here, but could sample from a different population; for example, patients found not guilty by reason of insanity and committed to a forensic psychiatric f a c i l i t y , a remand (pretrial) f a c i l i t y , or, in Canada, a psychiatric hospital within the federal prison system. A successful replication would increase the likelihood that the present results were not due merely to sampling from an unusual population. Second, i t would be useful to follow up subjects from the present study for a number of years. Although the PCL scales have demonstrated predictive validity in normal prison populations, i t may be that psychiatric symptomatology alters the predictability of behavior in some way. A related issue i s what, i f any, i s the hierarchical relation 84 between psychopathy and schizophrenia; for example, in those "schizophrenic psychopaths," which disorder has more predictive validity? A third possible area for investigation is the association between psychopathy and the new DSM-III-R c r i t e r i a , especially the revised personality disorder c r i t e r i a . The personality disorder section i n the revised manual has undergone some major changes, including a switch from a mixture of monothetic and polythetic c r i t e r i a in the DSM-III to s t r i c t l y polythetic c r i t e r i a and a general relaxation of exclusion c r i t e r i a . As well, a number of structured assessment instruments for the diagnosis of these personality disorders has been developed. These revised c r i t e r i a may lead to more reliable diagnoses and a greater possibility of finding any diagnostic overlap that may exist. Finally, i t could prove useful to look at the relation between the PCL, personality disorder diagnoses, and the behavioral and psychophysiological correlates of psychopathy. If some association does emerge—for example a positive association between psychopathy and histrionic personality disorder—then i t can be determined i f this diagnostic overlap i s accompanied by a similarity on some other level. D. Conclusion This results presented above offer no support for the hypothesis that psychopathy and schizophrenia are positively associated. Psychopathy, as diagnosed by the PCL scales, appears to be a c l i n i c a l l y specific disorder, generally unrelated to other forms of psychopathology. However, further research using revised and more reliable diagnostic c r i t e r i a for psychiatric disorders, especially the personality disorders, w i l l be necessary to determine the placement of psychopathy among the major dimensions of psychopathology. 85 References American Psychiatric Association. (1980). Diagnostic and s t a t i s t i c a l  manual of mental disorders (3rd ed.). Washington, D.C.: Author. American Psychiatric Association. (1987). Diagnostic and s t a t i s t i c a l  manual of mental disorders (3rd ed., rev.). Washington, D.C.: Author. Bender, L. (1971). Psychopathic personality or schizophrenia in adolescence. Biological Psychiatry, 3, 197-204. Bishop, Y. M. M., Fienberg, S. E., and Holland, P. W. (1975). Discrete  multivariate analysis: Theory and practice. Cambridge, Mass: MIT Press. Boyd, J., Burke, J., Gruenberg, E., Holzer, C , Rae, D., George, L., Karno, M., Stoltzman, R., McEvoy, L., and Nestadt, G. (1984). Exclusion c r i t e r i a of DSM-III. Archives of General Psychiatry, 41, 983-989. Claridge, G. S. (1972). The schizophrenias as nervous types. British  Journal of Psychiatry, 121, 1-17. Claridge, G. S. (1985). Origins of mental i l l n e s s . Oxford: Blackwell. Cleckley, H. (1976). The mask of sanity. St. Louis: Mosby. Cronbach, L. J., and Meehl, P. E. (1955). Construct validity i n psychological tests. Psychological Bulletin, 52, 281-302. Dahlstrom, W., and Welsh, G. (1960). An MMPI handbook: A guide to use  in c l i n i c a l practice and research. Minneapolis: University of Minnesota Press. Endicott, J., Spitzer, R. L., Fleiss, J. L., et a l . (1976). The Global Assessment Scale - A procedure for measuring the overall severity of psychiatric disturbance. Archives of General Psychiatry, 33, 766-771. Eysenck, H., and Eysenck, S. (1976). Psychoticism as a dimension of  personality. London: Hodder and Stoughon. Eysenck, H., and Eysenck, S. (1978). Psychopathy, personality, and genetics. In R. Hare and D. Schalling (Eds.), Psychopathic  behaviour: Approaches to research. Chichester, England: Wiley. Feighner, J., Robins, E., Guze, S., Woodruff, R., Winokur, G., and Munoz, R. (1972). Diagnostic c r i t e r i a for use in psychiatric research. Archives of General Psychiatry, 26, 57-63. Fienberg, S. E. (1980). The analysis of cross-classified categorical 86 data (2nd ed.). Cambridge, Mass: MIT Press. Finn, S. E. (1982). Base rates, u t i l i t i e s , and the DSM-III: Shortcomings of fixed-rule systems of psychodiagnosis. Journal of  Abnormal Psychology, 91, 294-302. Geller, M. (1980). Sociopathic adaptations in psychotic patients. Hospital and Community Psychiatry, 31, 108-112. Greene, R. L. (1980). The MMPI: An interpretive manual. Orlando, FL: Grune and Stratton. Goodwin, D. W., and Guze, S. B. (1984). Psychiatric diagnosis. New York: Oxford University Press. Gunderson, J. (1979). The relatedness of borderline and schizophrenic disorders. Schizophrenia Bulletin, 5 (1), 17-22. Gunderson, J. (1984). Borderline personality disorder. Washington: American Psychiatric Association. Hare, R. D. (1980). A research scale for the assessment of psychopathy in criminal populations. Personality and Individual Differences, 1, 111-119. Hare, R. D. (1982). Psychopathy and the personality dimensions of psychoticism, extraversion, and neuroticism. Personality and  Individual Differences, 3, 35-42. Hare, R. D. (1983). Diagnosis of antisocial personality disorder in two prison populations. American Journal of Psychiatry, 140, 887-890. Hare, R. D. (1984). Performance of psychopaths on cognitive tasks related to frontal lobe functions. Journal of Abnormal Psychology, 93, 133-140. Hare, R. D. (1985a). The Psychopathy Checklist. Unpublished manuscript, University of British Columbia, Vancouver, Canada. Hare, R. D. (1985b). Comparison of procedures for the assessment of psychopathy. Journal of Consulting and C l i n i c a l Psychology, 53, 7-16. Hare, R. D. (1985c). The relation between Psychopathy Checklist and  DSM-III diagnoses. Unpublished manuscript. Hare, R. D., and Cox, D. N. (1978). C l i n i c a l and empirical conceptions of psychopathy. In R. D. Hare and D. Schalling (Eds), Psychopathic  behavior: Approaches to research. Chichester, England: Wiley. Hare, R. D., and Forth, A. E. (1985). Psychopathy and lateral preference. Journal of Abnormal Psychology, 94, 541-546. 87 Hare, R., and Harpur, T. (1986). Weak data, strong conclusions: Some comments on Howard, Bailey, and Newman's use of the Psychopathy Checklist. Personality and Individual Differences, 7, 147-151. Harpur, T. J., Hakstian, A. R., and Hare, R. D. (1987, June). The factor structure of the Psychopathy Checklist. Paper presented at the I l l r d Meeting of the International Society for the Study of Individual Differences, Toronto, Canada. Hawk, S. S., and Peterson, R. A. (1974). Do MMPI Psychopathic Deviate scores reflect psychopathic deviancy or just deviancy? Journal of  Personality Assessment, 38, 462-468. Helmes, E. (1980). A psychometric investigation of the Eysenck Personality Questionnaire. Applied Psychological Measurement, 4, 43-55. Heston, L. L. (1966). Psychiatric disorders in foster-reared children of schizophrenic mothers. British Journal of Psychiatry, 112, 819-825. Hollingshead, A. B., and Redlich, F. C. (1958). Social class and mental  il l n e s s . New York: Wiley. Howard, R., Bailey, R., and Newman, A. (1984). A preliminary study of Hare's "Research Scale for the Assessment of Psychopathy" in mentally-abnormal offenders. Personality and Individual  Differences, 5, 389-396. Kallman, F. J. (1938). The genetics of schizophrenia. New York: Locust Valley. Kendler, K. (1985). Diagnostic approaches to schizotypal personality disorder: A historical perspective. Schizophrenia Bulletin, 11 (4), 538-553. Kety, S. S., Rosenthal, D., Wender, P. H., and Schulsinger, F. (1968). The types and prevalence of mental illness in the biological and adoptive families of adopted schizophrenics. In D. Rosenthal and S. S. Kety (Eds.), The transmission of schizophrenia. Oxford: Pergamon. Kohn, M. L. (1973). Social class and schizophrenia: A c r i t i c a l review and a reformulation. Schizophrenia Bulletin, Issue No. 7, 60-79. Loranger, A. W., Susman, V. L., Oldham, J. M., and Russakoff, L. M. (1987). The Personality Disorder Examination: A preliminary report. Journal of Personality Disorders, 1, 1-13. Mellsop, G., Varghese, F., Joshua, S., and Hicks, A. (1982). The r e l i a b i l i t y of Axis II of DSM-III. American Journal of Psychiatry, 139, 1360-1361. 88 Milburn, A., and Goff, M. (1956). Pseudosociopathic schizophrenia. West  Virginia Medical Journal, 52, 374-378. Millon, T. (1981). Disorders of personality: DSM-III Axis II. New York: Wiley. Millon, T. (1983). Millon C l i n i c a l Multiaxial Inventory manual, (3rd ed.). Minneapolis, MN: National Computer Systems. Overall, J. E., and Gorham, D. R. (1962). The Brief Psychiatric Rating Scale. Psychological Reports, 10, 799-812. Pfohl, B., Coryell, W., Zimmerman, M., and Stangl, D. (1986). DSM-III personality disorders: Diagnostic overlap and internal consistency of individual DSM-III c r i t e r i a . Comprehensive Psychiatry, 27, 21-34. Planansky, K. (1972). Phenotypic boundaries and genetic specificity in schizophrenia. In R. A. Kaplan (Ed.), Genetic factors in  'Schizophrenia'. Springfield: Thomas. Raine, A. (1985). Psychopathy, schizoid personality, and borderline/ schizotypal personality disorders. Personality and Individual  Differences, 7, 493-501. Raine, A., and Venables, P. H. (1987, June). Schizopathy: Conjoint  psychopathic and schizoid traits and their relationship with  smooth-pursuit eye movements, skin conductance orienting, and P3  evoked potentials. Paper presented at the I l l r d Meeting of the International Society for the Study of Individual Differences, Toronto, Canada. Robins, L., Helzer, J., Croughan, J., Williams, J., and Spitzer, R. (1981). NIMH Diagnostic Interview Schedule, Version III. St. Louis: Washington University School of Medicine. Robinson, T. N., and Zahn, T. P. (1985). Psychoticism and arousal: possible evidence for a linkage between P and psychopathy. Personality and Individual Differences, 1, 47-66. Schroeder, M. L., Schroeder, K. G., and Hare, R. D. (1983). Generalizability of a checklist for the assessment of psychopathy. Journal of Consulting and C l i n i c a l Psychology, 51, 511-516. Spitzer, R., Endicott, J., and Robins, E. (1974). Research diagnostic  c r i t e r i a . New York: Biometrics Research, New York State Department of Mental Hygiene. Stangl, D., Pfohl, B., Zimmerman, M., Bowers, W., and Corenthal, C. (1985). A structured interview for the DSM-III personality disorders. Archives of General Psychiatry, 42, 591-596. Stuart, D. (1982). Canadian criminal law: A treatise. Toronto: 89 Carswell. Templeman, R., and Wong, S. (1987, June). A proposed factor structure  of the Psychopathy Checklist. Paper presented at the Annual General Meeting of the Canadian Psychological Association, Vancouver, Canada. Travin, S., and Protter, B. (1982). Mad or bad? Some c l i n i c a l considerations in the misdiagnosis of schizophrenia as antisocial personality disorder. American Journal of Psychiatry, 139, 1335-1338. Void, G. B. (1979). Theoretical criminology (2nd ed.). New York: Oxford University Press. Widiger, T., Hurt, S., Frances, A., Clarkin, J., and Gilmore, M. (1984). Diagnostic efficiency and the DSM-III. Archives of General  Psychiatry, 41, 1005-1012. Wilson, G. (1981). Personality and social behaviour. In H. J. Eysenck (Ed.), A model for personality (pp. 210-245). New York: Springer-Verlag. Wing, J., Cooper, J., and Sartorius, N. (1984). Measurement and  classification of psychiatric symptoms. London: Cambridge University Press. Wong, S. (1985). The criminal and institutional behaviors of psychopaths. Program Branch Users Report, Ministry of the Solicitor General of Canada, Ottawa, Canada. 90 APPENDIX A An explanation of the acronyms used in the text and tables 1 (scale) = Scale 1 (Schizoid/Avoidant) of the MCMI 6 (scale) = Scale 6 (Antisocial/Aggressive) of the MCMI AAD = DSM-III Axis I diagnosis of an anxiety, dissociative, or non-bipolar affective disorder A-F = anxious-fearful personality disorder cluster (DSM-III; APA, 1980) APD = antisocial personality disorder (DSM-III; APA, 1980) AVD = avoidant personality disorder (DSM-III; APA, 1980) BOR = borderline personality disorder (DSM-III; APA, 1980) BP = DSM-III Axis I diagnosis of bipolar disorder, manic COM = compulsive personality disorder (DSM-III; APA, 1980) D-E-E = dramatic-erratic-emotional personality disorder cluster (DSM-III; APA, 1980) DEP = dependent personality disorder (DSM-III; APA, 1980) GR = global (prototypicality) rating HIS = histrionic personality disorder (DSM-III; APA, 1980) M (group) = nonpsychopathic group; those subjects with moderate scores on a PCL scale Ma = Scale 9 (Hypomania) of the MMPI MCMI = Millon Clinical Multiaxial Inventory (Millon, 1983) MMPI = Minnesota Multiphasic Personality Inventory (Dahlstrom and Welsh, 1960) MR = DSM-III Axis I diagnosis of mental retardation (mild) NAR = narcissistic personality disorder (DSM-III; APA, 1980) NONE = No DSM-III Axis I major diagnosis NP (group) = nonpsychopathic group; those subjects with low scores on a PCL scale 0-E = odd-eccentric personality disorder cluster (DSM-III; APA, 1980) 91 ORG = DSM-III Axis I of organic mental disorder P (group) = psychopathic group; those subjects with high scores on a PCL scale P (scale) = Paranoid Personality scale of the MCMI Pa = Scale 6 (Paranoia) of the MMPI PAG = passive-aggressive personality disorder (DSM-III; APA, 1980) PAR = paranoid personality disorder (DSM-III; APA, 1980) PCL = Psychopathy Checklist (Hare, 1980, 1985a) PCL20 = 20-item Revised Psychopathy Checklist (Hare, 1985a) PCL22 = 22-item Psychopathy Checklist (Hare, 1980) Pd = Scale 4 (Psychopathic Deviate) of the MMPI PD = personality disorder PDGR = personality disorder global (prototypicality) rating PP = Psychotic Delusions scale of the MCMI S = Schizotypal scale of the MCMI Sc = Scale 8 (Schizophrenia) scale of the MMPI SS = Psychotic Thinking scale of the MCMI SZ = DSM-III Axis I of schizophrenic (or conceptually related) disorder SZD = schizoid personality disorder (DSM-III; APA, 1980) SZGR = schizophrenia global (prototypicality) rating SZT = schizotypal personality disorder (DSM-III; APA, 1980) 

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.831.1-0097159/manifest

Comment

Related Items