Behavior Genetics, VoL 22, No. 5, 1992

An Adoption and a Cross-Fostering Study of the Minnesota Multiphasic Personality Inventory (MMPI) Psychopathic Deviate Scale Lee Willerman, 1,~ John C. Loehlin, x and Joseph M. Horn ~

Received 30 Dec. 199I--Final 9 Mar. 1992

The first of two complementary studies compared biological and adoptive parents of teenage adoptees with either high (n = 21) or low (n = 51) MMPI Psychopathic Deviate (Pd) scale scores. In comparison to biological mothers of the Iow-Pd adoptees, biological mothers of the highPd adoptees obtained significantly higher MMPI scores on six of eight clinical scales. Fewer differences existed between the corresponding groups of adoptive mothers, but adoptive mothers of the high Pd's did obtain significantly higher scores on the Pd and Hypomania scales. Substantial genetic correspondences also existed for Harris-Lingoes content subscales, with fewer correspondences between adoptees and their adoptive mothers. There were indications that adoptive mothers of the high-Pd children had personality traits which may have made them less effective in attenuating early signs of antisocial behavior. The second study employed a cross-fostering design dividing all biological and adoptive mothers (n = 138 each) by their respective median Pd raw scores to examine effects on offspring. Results confirmed the effect of biological mother Pd score, but only a trend suggested an adoptive mother effect, with no hint of an interaction. KEY WORDS: antisocial personality; inheritance; adoption; Minnesota Multiphasic Personality Inventory (MMPI); psychopathic deviate.

Research reported in this article was supported by Grants MH-24280 from the National Institute of Mental Health and BNS-7902918 and BNS-8209882 from the National Science Foundation. 1 Department of Psychology, University of Texas at Austin, Austin, Texas 78712. To whom correspondence should be addressed. 515 0001-8244/92/0900-0515506.50/0 9 1992 Plenum Publishing Corporation

516

Willerman, Loehlin, and Horn

INTRODUCTION Adoption studies have consistently pointed to heritable factors in adult antisocial behavior (Schulsinger, 1972; Crowe, 1974; Cloninger et al., 1982; Mednick et aL, 1984, t987). Twin studies have likewise suggested heritability among adults, but juvenile delinquency appears less heritable and affected to a much larger degree by shared environment (Cloninger and Gottesman, 1987). A reanalysis of a large twin study of adult criminality has recently called into question twin heritability, however, since monozygotic (MZ) twins may influence each other more than do dizygotic (DZ) twins, thus increasing the likelihood of MZ concordance (Carey, 1991). Other research focusing on possible social factors in antisocial behavior had also suggested mutual sibling influences because antisocial behavior in brothers of delinquent probands increased as a function of the number of brothers in a family (Jones et aL, 1980), and brothers closer in age are especially prone to offend together (Reiss and Farrington, 1991). This apparently strong juvenile environmental influence is somewhat puzzling in that shared family environment appears to have negligible effects on personality similarity during the adolescent years among pairs of adoptively related singletons (e.g., Loehlin et aL, 1987). A sample of adoptees ascertained independently from those in the present study had obtained lower scores than biological children of adoptive parents on the CPI Socialization scale, suggesting greater delinquency proneness, but unrelated children reared together correlated only rl = .06 on the scale (Loehlin et aL, 1985). In the Texas Adoption Project (TAP), the Psychopathic Deviate (Pd) scale of the Minnesota Multiphasic Personality Inventory (MMPI) yielded a genetically unrelated sibling correlation of only rl -- .05, though the majority of pairs in this correlation were of opposite sex and thus do not provide a strong test of same-sex sibling influences. The concept of antisocial personality disorder and what many clinicians believe to be its core personality feature, psychopathy, contains components tapping both personality and misbehavior. Cleckley's (1976) concept of psychopathy emphasized personality traits such as egocentricity, callousness, superficial charm, insincerity, and absence of guilt, while the Diagnostic and Statistical Manual of Mental Disorders of the American Psychiatric Association (1987) focuses more on misbehavior to diagnose antisocial personality disorder. Harpur et al. (1989) have empirically demonstrated the necessity for a two-factor conceptualization of psychopathy that incorporates both personality and deviant behavior, showing in several incarcerated offender and forensic patient samples

The M M P I Psychopathic Deviate Scale

517

that these two oblique factors characterize much of the psychopathy domain. The present study identifies genetic influences associated with antisocial personality using the Pd scale, one of the most heritable of the MMPI scales (Gottesman, 1963; Loehtin et aL, 1987). Although a high Pd score peak is not tantamount to psychopathy or antisocial personality disorder, it does indicate "angry disidentification with recognized conventions" (Carson, 1969, p. 288), presumably a personality component of the condition. Moreover, the scale shows modest discriminant validity among incarcerated prisoners who are otherwise judged to be high or low on psychopathy (Hare, 1985). The present study asks about the heritability of the Pd scale and associated traits, whether more specific content subscales front the MMPI can help clarify the patterns observed, and whether adoptive parent characteristics reflected in their MMPI scores influence Pd scores of the adoptees. METHOD

In the TAP, 300 white unwed mothers were administered the MMPI and the Revised Beta intelligence test during their stay in a home for unwed pregnant women. The women usually entered the home during the middle trimester of their pregnancy. Conscientious efforts were made to ensure good prenatal nutrition and health care from time of entry in the home. Their offspring, adopted away at birth, were subsequently located and administered age-appropriate intelligence tests, as were other adopted or natural children in the adoptive home. A parent (usually the mother) also rated the child on bipolar adjectives meant to characterize the child's personality (Loehlin et al., 1990). The adoptive parents also were administered intelligence tests and completed the MMPI. Results from this first study are reported by Horn et al. (1979) and Loehlin et al. (1981). A follow-up study was undertaken nearly i0 years later. Fifty-five percent of the adoptees from the original study participated, as did 57% of the biological children of the adoptive parents (Loehlin et al., 1987). Parents participating in the follow-up were slightly younger and of higher socioeconomic status (SES) and IQ than those who refused. Only 1 of 56 personality measures revealed a difference between the two groups of mothers; those in the follow-up had higher MMPI Psychasthenia scores. Parental ratings of the adoptees on the 24 traits meant to cover the range of Cattell's personality constructs in the initial study revealed only one significant difference; those in the follow-up had been rated as more deliberate and inactive versus unrestrained and changeable. In the follow-

518

Willerman, Loehlin, and Horn

up the children were administered the Revised Beta examination, an ageappropriate version of a Wechsler intelligence test, Cattell's Sixteen Personality Factor Questionnaire (16PF), and the MMPI. Parents were not retested, though they completed personality ratings on their children. The present studies employ both the adoptees family method and the adoptees study method (Rosenthal, 1970) to evaluate convergences in the findings. The family method begins with adoptees manifesting the trait of interest and examines their biological and adoptive relatives. The adoptees study method begins with parents of adoptees and looks at their offspring. This approach provides a check on the results obtained from the family method, as well as making possible a cross-fostering study in which high and low Pd scores of biological and adoptive mothers can be crossed to examine Pd effects on a larger sample of adoptees. From the pool of 184 adoptees 13 years or older who completed the MMPI in the follow-up study, potential high-Pd probands were required to have MMPI K-corrected T scores of ---70 on the Pd scale, this elevation being the highest or second highest in the clinical profile. A K-corrected T score __.70 is conventionally regarded as clinically significant, the K correction being an upward adjustment to the Pd raw score, based on evidence that people often underreport psychopathology. The nominal Pd K-corrected T score mean is 50 (SD = 10), but teenagers and young adults obtain higher scores (Hathaway and Monachesi, 1963). An exception to the T score ->70 inclusion rule was permitted if the Hypomania (Ma) scale was T -> 70 and produced the highest or second highest T score in the profile. Low-Pd probands were required to have a Pd scale not exceeding a K-corrected T score of 55, with other scale scores allowed to vary freely. We also required that MMPI validity scales not be grossly deviant. Thus, the F scale had to be less than a T score of 80, and the Lie scale not more than a T score of 70. Thirty-seven adoptees met the high-Pd criteria, but biological mother MMPIs were lacking on 9, and 7 of the probands had invalid profiles, resulting in a final group of 21 high-Pd probands (10 female). Eighty-three adoptees satisfied the criteria for low Pd, but biological mother MMPIs were lacking for 27, and an additional 5 had invalid profiles, resulting in 51 low-Pd adoptee probands (27 female). Not all adoptee MMPIs excluded for invalidity were necessarily untrustworthy, but it seemed prudent to eliminate those whose scores exceeded these standards. The biological mothers and adoptive parents had completed an abbreviated version of the MMPI which included only the first 366 items plus 7 others that allowed the standard clinical and validity scales to be completely scored. No validity restrictions were imposed on the biological mothers or adoptive parents because selection was on the adoptees.

T h e M M P I Psychopathic Deviate Scale

519

Whether or not invalidity criteria were applied to the parents, however, had little effect on the results to be reported. After the two groups of probands were identified, comparisons were made using MMPI raw scores, Harris-Lingoes content subscales derived from the MMPI (Harris and Lingoes, 1955, 1968), Sixteen Personality Factor Questionnaire, parental ratings of their child's personality on 24 bipolar dimensions meant to describe the traits measured by Cattell's series of personality questionnaires, and IQ. Outcomes of primary interest in the adoptees' families study were the MMPI raw scale scores and Harris-Lingoes subscale scores of biological mothers and adoptive parents. In the cross-fostering study, adoptee Pd scores were the outcome of interest. RESULTS Table I provides descriptive statistics on the two groups of probands and their parents. Age differences between the two groups were not significant, nor were differences in IQ or adoptive parent SES [an arbitrarily scaled combination of adoptive parent education and paternal occupation (Loehlin, 1979)]. Proband Differences. Figure 1 presents mean K-corrected MMPI profiles for the high- and low-Pd adoptees. Univariate statistical analyses were done on raw scale scores without K correction as recommended by Butcher and Tellegen (1978). The high-Pd group scored significantly higher on the F and K validity scales and on all eight clinical scales (all p ' s < .05). The Mf scale (Opposite Sex Interests) results are not shown because probands were of both sexes, but there were no significant group differences. Reflecting selection criteria, the mean profile of the high-

T a b l e I. Descriptive Data on the High- and Low-Pd Adoptive Probands, Biological Mothers, and Adoptive Parents at the Time of Testing Variable Proband age (yr) Biomother age (yr) Adoptive SES

High Pd (n = I9-21)

Low Pd (n = 45-51)

Significance

15.9 19.9 176.7

16.5 19.3 169.7

ns ns ns

112.6 111.4 109.6 114.9 116.6

ns ns ns ns ns

W e c h s l e r Full-Scale I Q Proband--first study Proband--second study Biological mother Adoptive mother Adoptive father

111.4 108.4 107.2 112.1 116.5

520

W i l l e r m a n , Loehlin, and H o r n

80 112

7o

Proband Adoptees

-ll- HighP d -c- LowPds

s

~

0

pO 02

,./ 4(?

i

'

I

'

i

9

i

'

[

'

!

'

t

'

I

'

i

,

i

'

i

L F K Hs D Hy Pd Pa Pt Sc Ma Scale

Fig. 1. MMPIK-corrected profiles of high-Pd (n = 21) and low-Pd (n = 51) adoptees.

Pd children in Fig. 1 is of the classic "psychopathic" type in which clinical scales 4 and 9 mark the peaks; the tertiary elevation on Sc suggests that acting-out behavior may occasionally take a bizarre twist (Friedman et aL, 1989), unaccompanied by the superficial charm that many clinicians believe to be an important component of psychopathy. The childrens' 16PF supported the MMPI results in most, but not all respects. High-Pd children were significantly more dominant (E), suspicious (L), and nonconforming (Q1), less conscientious (G), and less controlled (Q3) than the low-Pd children (all p ' s < .05). Contrary to prediction, the high-Pd adoptees were not low on the 16 PF traits of warmheartedness and guilt proneness, implying little callousness or lack of conscience. Adoptive parent personality ratings of the children at about 6 years of age indicated no group differences, but the teenage follow-up ratings revealed significant differences on 10 of the 24 bipolar dimensions. In comparison to the low Pd's, the high Pd's were rated as emotionally less stable, less conscientious, less patient, less controlled, more tense, socially bolder, more apprehensive, more careless about following social rules, more frivolous, and more affected by feelings. To examine whether high- and low-Pd adoptees differed from their unrelated siblings, all high- and low-Pd probands were compared to all other children in their families who were not in the same Pd classification as the probands. These comparisons could be made for only probands with siblings (either other adopted or natural children of the adoptive parents). To increase statistical power, adoptees for whom biological mother MMPIs were lacking, but who otherwise met high- or low-Pd criteria were also included. The within-family personality ratings of the

The M M P I Psychopathic Deviate Scale

521

high-Pd adoptees and their siblings (n = 21 pairs) confirmed the patterns described above. At 6 years of age, only one within-family difference between high-Pd adoptees and sibs was significant; high-Pd children were rated as being less controlled. In the teen years, however, the highPd adoptees were rated as more tense, more frivolous, more stubborn, and less conscientious than their siblings. The low Pd's, in comparison to their siblings (n = 6%68 pairs), were judged significantly more favorably on 9 of the 24 ratings, including emotional stability, conscientiousness, patience, and control. In short, many differences existed between high and low Pd's on the MMPI, 16 PF, and personality ratings during the teen years, but a difference in the early school years was observed only for the within-family rating of undercontrol. Biological Mothers. The profiles of the two groups of biological mothers are quite similar in shape but substantially different in elevation (Fig. 2). Using raw scale scores, the mothers of the high-Pd children scored significantly higher (all univariatep's < .05) on F and on six of the eight MMPI clinical scales, the exceptions being Hysteria (Hy) and Hypomania (Ma). In comparison to the biological mothers of the low Pd's, mothers of the high-Pd children also were significantly more variable by the Fm~, test on all clinical scales, except Depression and Psychasthenia. Separate variance estimates were used for calculating t values between the two groups when this occurred, but whether variances were pooled or separated had little statistical effect. Ten of the 21 biological mothers of the high Pd's themselves had Pd T scores of more than 70, compared to I0 of the 51 biological mothers of the low Pd's (• _ 5.81,

80

Biological Mothers of:

-a-Hg i hPd 70 ~~ LowPd ~ o' I..--) 0

~ 6o O

o. 5o

40 L

F

K

Hs D

Hy Pd Mf Pa Pt Sc Ma

Scale Fig. 2. MMPI K-corrected profiles of birth mothers of high- and low-Pd adoptees.

522

Willerman, Loehlin, and Horn

p < .02). This group difference was even more marked at Pd T-score elevations > 80, with six and one, respectively, for the two groups (X2 = 12.0, p