شخصیت و رفتار ضد اجتماعی: مطالعه ابعاد خون گرم

Abstract Personality variables have been considered as major determinants of delinquent behaviour in various theoretical models and numerous empirical studies. Particular attention has been paid to “temperament” variables, which are considered to have a biological basis. In the present study, we examined relationships between self-reported antisocial behaviour and a number of temperament variables (extraversion, neuroticism, psychoticism, impulsivity, sensation seeking) in three subject groups: 435 school-attending male adolescents, 529 school-attending female adolescents, and 95 institutionalized delinquent male adolescents. This study design, unlike that of most previous studies of this type, allowed control for the factor of institutionalization, and included a longitudinal analysis, in that questionnaires were administered to the school-attending subjects twice with a 1-year interval. The results confirm that several temperament variables, characterized by high sensitivity to reward and/or weak response to punishment signals, are closely associated with antisocial behaviour. Our findings suggest that personality variables should be included in criminological models, and taken into account in intervention programs.

. Introduction A wide range of variables have been considered as risk factors for juvenile delinquency (see for example Rutter, Giller & Hagell, 1998). Of these, personality variables were for many years attributed scant importance (Farrington, 1992 and Romero et al., 1999): the powerful influence of sociology on criminological theory, a certain fear of biological reductionism, and the internal crises of personality psychology itself came together to create a climate in which personality factors were dismissed as unimportant (Stitt & Giacopassi, 1992). More recently, however, there has been what might be called a ‘rediscovery of the person’ in criminology (Andrews & Bonta, 1994). In the face of strong evidence that macrosocial variables cannot fully explain delinquency, and that there is great interindividual variability in the behaviour of subjects exposed to criminogenic conditions, increasing interest is turning to individual differences, even in schools of thought with a strong sociological influence (Gottfredson & Hirschi, 1990). Particular attention is being paid to what have been referred to as the ‘temperament’ variables, a group of characteristics assumed to depend on the individual's biological substrate, and showing a relatively high degree of stability over the lifespan (Bates and Wachs, 1994 and Strelau, 1998). In criminal psychology, Eysenck's three fundamental dimensions (extraversion, neuroticism, psychoticism), together with impulsivity and sensation seeking, have received particular attention.1 It has been hypothesized that delinquency is related to all three Eysenck dimensions. Extraversion, associated with a low level of corticoreticular arousal, hinders conditioning and thus the acquisition of social norms. As neuroticism tends to amplify acquired behavioural tendencies, Eysenck (1964) predicts that subjects with high extraversion and neuroticism scores are those at highest risk for antisocial behaviour. In 1976, the dimension psychoticism was introduced into Eysenck's system (Eysenck & Eysenck, 1976), and since that time has likewise been considered to be a predictor of criminal behaviour, in view of the hostility and emotional insensitivity characteristic of subjects with high scores on this dimension. Impulsivity has also been the subject of increasing attention as an explicator of antisocial behaviour, despite the conceptual and methodological confusion surrounding this construct (Gerbing et al., 1987, White et al., 1994 and Romero et al., 1994). The classical conceptions of psychopathic personality (Cleckley, 1976, Gough, 1948 and Hare, 1980) include as defining characteristics a lack of behavioural self-control, an orientation toward immediate gratification, and difficulty in carrying through long-term plans, all of which are implicit in the impulsivity construct. Approaches derived from Gray's model (for example, Newman, 1987) have suggested that reward sensitivity and difficulty in processing punishment signals make impulsive subjects prone to antisocial behaviour. The theoretical approach adopted by Wilson and Herrnstein (1985) proposes that impulsive subjects have a short time horizon, and are thus not adequately dissuaded by the typically delayed negative consequences of delinquent acts. Finally, in more recent models (e.g. Gottfredson and Hirschi, 1990 and Hirschi and Gottfredson, 1994), low self-control is considered to be a key factor underlying many types of deviant conduct, in association with situational opportunity. Sensation seeking likewise has been considered a correlate of antisocial behaviour (Zuckerman, 1979 and Zuckerman, 1994a). It is widely considered that a desire for stimulating, intense, and novel experiences is an essential component of delinquent conduct (Farley and Farley, 1972, Haapasalo, 1990 and Zuckerman, 1978). Many researchers have examined the relationship between these personality dimensions and antisocial behaviour (see for example Berman and Paisley, 1984, Daderman, 1999 and Mak, 1991). Despite the vigour of these lines of research, however, most previous studies have suffered two limitations. The first frequent limitation is one of design, in that many studies have been based on comparison between institutionalized delinquents and individuals in the general population, on the implicit assumption that all non-institutionalized subjects are non-delinquent. As pointed out by other authors (Farrington, 1987 and Krueger et al., 1994), such ‘known-groups’ designs have a number of clear drawbacks. First, only a small proportion of subjects who commit crimes are subsequently institutionalized, so that known-groups designs cannot guarantee that the putative controls are non-delinquent subjects. In fact, the two groups probably overlap along a behavioural continuum, so that the delinquency/non-delinquency dichotomy is of little value (Eysenck & Gundjonsson, 1989). Second, institutionalization is in itself an important variable affecting the relationships between personality and delinquency. Certain personality traits may favour official processing and institutionalization (Feldman, 1977), giving rise to ‘differential apprehension’: for example, impulsive and emotionally unstable individuals may be less capable of escaping detention by the police. Similarly, it should not be forgotten that internment may itself influence personality variables, as suggested by various authors (Banister et al., 1973, Landau, 1976 and Zamble and Porporino, 1988). As a result, it may be difficult to determine whether a given trait is associated with delinquent behaviour or simply with institutionalization. The second frequent limitation of studies of this type is that they are cross-sectional in approach. Despite the clear need for a longitudinal approach in studies of delinquent behaviour (Farrington, 1997, Heaven, 1993 and Loeber, 1990), most studies have evaluated personality/delinquency relationships as existing at a given moment in time.2 However, the identification of risk factors implies a need for prospective studies, allowing control of the temporal order of variables, and thus assessment of the extent to which personality variables can be considered genuine predictors of antisocial behaviour. In the present study we investigated the relationship between temperament variables and delinquency, using a design that attempted to overcome these limitations. First, delinquency was considered as a behavioural continuum rather than an all-or-nothing variable: subjects were asked to supply self-reports of antisocial behaviour, allowing direct evaluation of personality–delinquency relationships. In order to cover a wide range of levels of antisocial behaviour, we studied both school-attending adolescents and institutionalized delinquents. In addition, the inclusion of officially recognized delinquents allowed us to perform specific analyses of the effects of the factor institutionalization.

. Results 3.1. Personality and antisocial behaviour: descriptive statistics Table 1 shows mean values for ABQ score and each of the personality variables (T1 data) in each subject group. Also shown for each variable are the results of F tests to compare means between school-attending boys and girls, and between school-attending and institutionalized boys, after partialling out the effects of age. In multivariate analysis of variance considering all personality variables, significant between-gender differences were observed (Wilks' lambda=0.86, P< 0.001). Girls showed significantly higher mean neuroticism and significantly lower mean psychoticism than boys, with no significant difference in extraversion. In addition, girls showed significantly lower scores on all components of sensation seeking except experience seeking. Impulsivity did not differ significantly between boys and girls. In general, these findings are in agreement with those of previous studies of Spanish subjects ( Eysenck et al., 1994, Pérez and Torrubia, 1985 and Silva et al., 1987). In addition, and in accordance with numerous previous studies (e.g. Junger-Tas et al., 1994 and Huizinga and Elliott, 1987), the mean antisocial behaviour score for girls was much lower than that for boys. Table 1. Mean scores in each group on the different personality scales and on the Antisocial Behaviour Questionnairea Mean F School-attending males School-attending females Institutionalized males School-attending males vs. school attending-females School-attending males vs. institutionalized males Extraversion 17.77 17.78 16.37 0.00 (ns)b 8.83** Neuroticism 11.46 12.67 12.70 20.06*** 5.47* Psychoticism 4.18 2.79 6.38 64.38*** 36.23*** Impulsivity 11.64 11.22 14.63 1.42 (ns) 21.66*** Thrill-and-adventure seeking 5.66 5.23 5.63 6.05* 0.01 (ns) Experience seeking 4.20 4.13 4.74 0.23 (ns) 4.23* Disinhibition 5.93 5.09 6.18 28.54*** 0.65 (ns) Boredom susceptibility 3.52 2.97 4.02 22.49*** 4.84* Antisocial behaviour 15.06 7.70 90.53 33.15*** 229.26*** a Results of comparisons by analysis of variance ns. b Not significant (P>0.05) * P<0.05 ** P<0.01 *** P<0.001 Table options Personality variables likewise differed significantly between school-attending and institutionalized boys (Wilks' lambda=0.89, P< 0.001). Note that this comparison corresponds to that performed in studies with ‘known groups’ designs. Institutionalized boys also showed significantly higher mean scores for neuroticism and psychoticism, and significantly lower mean scores for extraversion. They also showed significantly higher mean scores for impulsivity and on two components of sensation seeking, namely experience seeking and boredom susceptibility. 3.2. Cross-sectional analyses: personality-antisocial behaviour correlations Table 2 shows coefficients of correlation (after partialling out the effects of age) between personality variables and antisocial behaviour in each group. Significant correlations are observed in all three groups. Among school-attending boys, the strongest correlations were with psychoticism, impulsivity, disinhibition, and experience seeking. Neuroticism and the other two dimensions of sensation seeking (thrill-and-adventure seeking, and boredom susceptibility) likewise showed significant positive correlations with antisocial behaviour, though with r<0.30. Among school-attending girls, the strongest correlations were with impulsivity and disinhibition. However, all variables considered showed significant correlations. Table 2. Correlations between personality variables and self-reported antisocial behaviour (Antisocial Behaviour Questionnaire score) in the three subject groups, after partialling out the effects of agea School-attending males (n=435) School-attending females (n=529) Institutionalized males (n=95) Extraversion 0.01 0.11** −0.09 (0.01) (0.12**) (0.11) Neuroticism 0.13** 0.18*** 0.00 (0.15**) (0.19***) (0.01) Psychoticism 0.37*** 0.26*** 0.35*** (0.47***) (0.34***) (0.44***) Impulsivity 0.37*** 0.36*** 0.35*** (0.42***) (41***) (0.40***) Thrill-and-adventure seeking 0.16*** 0.17*** 0.14 (0.18***) (0.20***) (0.16) Experience seeking 0.30*** 0.28*** 0.35*** (0.38***) (0.36***) (0.44***) Disinhibition 0.35*** 0.33*** 0.20* (0.42***) (0.42***) (0.25*) Boredom susceptibility 0.18*** 0.26*** 0.19* (0.22***) (0.33***) (0.23*) a Values in brackets are correlations after correction for attenuation. * P<0.05 ** P<0.01 *** P<0.001 Table options Girls showed significantly higher correlations than boys in extraversion (P< 0.001), neuroticism (P< 0.01) and boredom susceptibility (P< 0.01). By contrast, the correlation between ABQ and psychoticism was higher among boys (P< 0.001). Among institutionalized boys, the strongest correlations (r> 0.30) were with psychoticism, impulsivity, and experience seeking. Disinhibition and boredom susceptibility likewise showed significant positive correlations with antisocial behaviour, though with r values in all cases less than 0.30. In the criminological literature, research into personality–delinquency relationships has often been criticized on the grounds that personality items and antisocial behaviour items may overlap (see for example Tennenbaum, 1977). In the present study, the psychoticism scale of the EPQ contained two items that overlap in content with the aggression scale of the ABQ (“Do you think you get involved in more fights than other people?”, and ”In general, do you enjoy bothering other people?”). In order to assess the extent to which the observed correlations between psychoticism and ABQ score might be attributable to this item overlap, we calculated correlations after elimination of these two items from the psychoticism scale. The correlations obtained did not differ significantly from those obtained previously: specifically, r values were 0.34 for school-attending boys (0.44 after correction for attenuation), 0.23 for school-attending girls (0.30 after correction for attenuation), and 0.29 for institutionalized boys (0.38 after correction for attenuation). In addition, we determined correlations between psychoticism and the individual subscales of the ABQ (vandalism, theft, aggression, rule breaking, drug involvement). We found that the correlations remained high even when there was no relationship between the psychoticism items and the items of the subscale in question: for example, the correlation between psychoticism and vandalism was 0.34 in school-attending boys (0.44 after correction for attenuation), 0.26 in school-attending girls (0.32 after correction for attenuation), and 0.36 in institutionalized boys (0.45 after correction for attenuation). These results thus allow us to rule out the possibility that the observed correlation between psychoticism and ABQ is an artifact of item overlap. 3.3. Cross-sectional analyses: analyses of variance To investigate the extent to which personality variables differed between subjects at different positions on the antisocial behaviour continuum, the three subject groups were each subdivided on the basis of antisocial behaviour scores. For male subjects (in both the school-attending and institutionalized groups), the ABQ-score cutoffs for the subgroupings were 5, 34, and 89, corresponding to the 25th, 90th and 100th percentiles for the school-attending group. As in previous studies (Carrillo et al., 1994 and Luengo et al., 1994), these cutoffs were used to define the groups denominated “nondelinquent” (ND, score ⩽5), “moderately delinquent” (MD, score 6–34), “severely delinquent” (SD, score 35–89), and “very severely delinquent” (SD+, score >89). For female subjects, the ABQ-score cutoffs were 1 and 17, corresponding to the 25th and 90th percentiles for this group (ND, score ⩽1; MD, score 2–17; SD, score >17). Table 3 shows the distribution of subjects among the subgroups thus defined. As can be seen from the table, none of the institutionalized subjects had an ABQ score below 6 (i.e. nondelinquent). Table 3. Distribution of subjects in each subgroup by Antisocial Behaviour Questionnaire (ABQ) score categoriesa Subgroup ABQ score School-attending males Institutionalized males Subgroup ABQ score School-attending females ND (0–5) 121 0 ND (0–1) 130 MD (6–34) 262 17 MD (2–17) 352 SD (35–89) 52 35 SD (> 17) 47 SD+ (> 89) – 43 a ND, Non delinquent; MD, Moderately delinquent; SD, Severely delinquent; SD+, Very severely delinquent. Table options Within each group (school-attending female, school-attending male, institutionalized male), we then compared personality variables between the ABQ-score subgroups. These comparisons allowed exploration of the extent to which the different variables are related to different degrees of antisocial behaviour. For each group, we used multivariate analysis of variance with personality variables as response variables, ABQ-score subgroup as the grouping factor, and age as covariate (Table 4, Table 5 and Table 6). Note that analysis of variance, unlike correlation analysis, is capable of detecting non-monotonic relationships. Table 4. Comparisons of Antisocial Behaviour Questionnaire score subgroups (non delinquent [ND], moderately delinquent [MD], severely delinquent [SD]) in the school-attending males Mean F Significant differences ND MD SD Extraversion 17.17 18.15 16.67 5.05* ND-MD MD-SD Neuroticism 10.53 11.73 12.45 4.03* ND-MD ND-SD Psychoticism 3.22 4.19 6.40 20.91*** ND-MD ND-SD MD-SD Impulsivity 9.29 11.83 14.64 23.52*** ND-MD ND-SD MD-SD Thrill-and-adventure seeking 4.99 5.67 6.02 4.91** ND-MD ND-SD Experience seeking 3.69 4.02 5.23 17.20*** ND-MD ND-SD MD-SD Disinhibition 4.87 5.92 7.21 25.85*** ND-MD ND-SD MD-SD Boredom susceptibility 2.92 3.59 4.23 10.76*** ND-MD ND-SD Wilks' lambda= 0.73*** * P<0.05 ** P<0.01 *** P<0.001 Table options Table 5. Comparisons of Antisocial Behaviour Questionnaire score subgroups (non delinquent [ND], moderately delinquent [MD], severely delinquent [SD]) in the school-attending females Mean F Significant differences ND MD SD Extraversion 16.44 18.06 18.75 10.32*** ND-MD ND-SD Neuroticism 11.23 12.93 14.08 10.68*** ND-MD ND-SD Psychoticism 2.20 2.80 4.14 11.72*** ND-MD ND-SD MD-SD Impulsivity 8.46 11.62 15.18 42.63*** ND-MD ND-SD MD-SD Thrill-and-adventure seeking 4.52 5.33 6.16 11.78*** ND-MD ND-SD MD-SD Experience seeking 3.57 4.18 5.16 22.73*** ND-MD ND-SD MD-SD Disinhibition 4.10 5.21 6.65 40.41*** ND-MD ND-SD MD-SD Boredom susceptibility 2.56 2.97 4.02 17.87*** ND-MD ND-SD MD-SD Wilks' lambda=0.72*** *** P<0.001 Table options Table 6. Comparisons of Antisocial Behaviour Questionnaire score subgroups (moderately delinquent [MD], severely delinquent [SD], very severely delinquent [SD+] in the institutionalized subjects Mean F a Significant differences MD SD SD+ Extraversion 16.92 16.88 16.00 0.17 (ns) – Neuroticism 12.15 13.70 12.39 0.58 (ns) – Psychoticism 4.15 5.61 6.94 6.54** MD-SD MD-SD+ Impulsivity 10.26 14.29 15.78 8.28** MD-SD MD-SD+ Thrill-and-adventure seeking 4.93 5.55 5.97 1.11 (ns) – Experience seeking 3.53 4.62 5.17 6.34** MD-SD+ SD-SD+ Disinhibition 5.40 6.00 6.63 4.35* MD-SD+ SD-SD+ Boredom susceptibility 3.00 3.85 4.17 1.97 (ns) – 3.00 3.85 4.17 1.97 (ns) – Wilks' lambda=0.49** a ns, not significant (P>0.05) * P<0.05 ** P<0.01 Table options In all three subject groups, the overall multivariate test statistic (Wilks' lambda) was significant at the 1% level. Considering the F values for the individual univariate analyses, all personality variables discriminated among ABQ-score subgroups, in both the school-attending male and school-attending female groups. In the school-attending male group, pairwise comparisons indicated that mean psychoticism, impulsivity, experience seeking, and disinhibition differed significantly between all three ABQ subgroups. Mean neuroticism, thrill-and-adventure seeking, and boredom susceptibility differed significantly between the ND group and the other two groups (MD and SD), but not between MD and SD. Extraversion showed an apparently non-monotonic trend, being significantly higher in MD than in both ND and SD. In the school-attending female group, pairwise comparisons indicated that all personality variables differed significantly between all three ABQ subgroups, except extraversion and neuroticism, which differed significantly between ND and the other two groups but not between MD and SD. In the institutionalized male group, pairwise comparisons revealed significant differences in psychoticism, impulsivity, experience seeking, and boredom susceptibility. Psychoticism and impulsivity were significantly lower in MD than in SD and SD+. Experience seeking and boredom susceptibility were significantly lower in MD and SD than in SD+. 3.4. Cross-sectional analyses: detected versus undetected delinquents The above analyses indicated that the observed relationships between personality variables and delinquency are not exclusively attributable to institutionalization, since relationships were observed in the non-institutionalized groups. However, to investigate if institutionalization may affect these relationships, we compared the personality variables between the SD subgroups of school-attending and institutionalized boys (i.e. between “undetected” and “detected” delinquents).3 As shown in Table 7, the overall multivariate test statistic (Wilks' lambda) for this analysis was not significant at the 5% level. The F values for the individual analyses were likewise not significant, except for disinhibition, which was significantly lower among institutionalized delinquents than among non-institutionalized delinquents. Since the two groups differed significantly in socioeconomic class (as assessed by the index of Hollingshead, 1975), the analysis was repeated with this variable as covariate; the results were practically identical. Table 7. Comparisons between the severely delinquent (SD) subgroups of the school-attending male group and the institutionalized male group Mean Fa School-attending severe delinquents Institutionalized severe delinquents Extraversion 16.67 16.88 0.21 (ns) Neuroticism 12.45 13.70 1.65 (ns) Psychoticism 6.40 5.61 0.00 (ns) Impulsivity 16.64 14.29 0.00 (ns) Thrill-and-adventure seeking 6.02 5.55 0.15 (ns) Experience seeking 5.23 4.62 2.67 (ns) Disinhibition 7.21 6.00 8.30** Boredom susceptibility 4.23 3.85 0.92 (ns) Wilks' lambda=0.77 (n.s.) a ns: Not significant (P>0.05) ** P<0.01 Table options 3.5. Longitudinal analyses To identify personality variables that were effective predictors of increasing antisocial behaviour in the male and female school-attending groups, we performed multiple regression analyses with T2 ABQ score as response variable and personality variables and T1 ABQ score as candidate predictors (Table 8). Table 8. Multiple regression analysis to identify predictors of change in antisocial behaviour Step Variable Beta R2 F for ΔR2 School-attending females 1 T1 Antisocial behaviour 0.54 0.29 165.48*** 2 Disinhibition 0.16 0.31 12.66*** 3 Psychoticism 0.12 0.33 8.25** 4 Experience Seeking 0.10 0.34 5.76* School-attending females 1 T1 Antisocial behaviour 0.57 0.33 240.62*** 2 Impulsivity 0.17 0.35 18.87*** 3 Extraversion 0.09 0.36 6.71* * P<0.05 ** P<0.01 *** P<0.001 Table options In the male group, disinhibition, psychoticism, and experience seeking were selected as significant predictors. In the female group, impulsivity and extraversion were selected. The regression coefficients for personality variables were rather low, but it should be borne in mind that the time period under consideration was short, and that a conservative analysis was used (i.e. T1 ABQ score was included as a predictor, in order to partial out the strong autoregressive effects).