Académique Documents
Professionnel Documents
Culture Documents
: : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : :
Research has demonstrated that risk of violent behavior is high in individuals who display psychopathic traits. However, prediction
of general aggression, and in particular unprovoked aggression, in nonforensic men who possess such traits has received little
experimental attention to date. This study examined the role of psychopathic traits in the prediction of unprovoked aggression in a
nonforensic sample. One hundred and ve men completed the Levenson Self-Report Psychopathy Scales and competed in a sham
aggression paradigm. These men were identied as unprovoked aggressors, provoked aggressors, and provoked nonaggressors
contingent on their responses to the experimental paradigm. Findings indicated that men with high levels of psychopathic traits had
a 30% greater probability of becoming aggressive in the absence of provocation relative to those with lower psychopathic traits.
Results also indicated that unprovoked aggressors evinced more general aggression when provocation was initiated. Findings are
discussed in terms of the importance of trait-based risk factors in the prediction of seemingly random, unprovoked aggression.
r 2007 Wiley-Liss, Inc.
Aggr. Behav. 34:319328, 2008.
: : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : :
Keywords: unprovoked aggression; psychopathy; GAM; provocation
INTRODUCTION
METHODS
Levenson et al. [1995] labeled the two LSRP factors primary and
secondary psychopathy because they argued that this factor
structure, derived from the Psychopathy Checklist-Revised, ts
Karpmans [1948] distinction between primary and secondary
psychopathy. However, there is no clear consensus regarding the
existence of these subtypes or whether they are congruent with
various psychopathy factors identied in empirical study. As such,
for the limited scope of this study, we refer to the LSRP factors as F1
and F2.
SI
SD
SDa
SF
.41
.36
.58
RESULTS
underlying continuous variable [Anath and Kleinbaum, 1997; Scott et al., 1997]. This model is based
on probability estimates of inclusion in category j
conditional on inclusion in a category greater than j.
Hence, this model can be formulated as follows:
log PrY yj x=PrY4yj x
Psychopathy scores were treated as continuous
predictors of group assignment. The ordered
arrangement of the groups was unprovoked aggressors, provoked aggressors, and nonaggressors.
According to West et al. [1996], psychopathy scores
were standardized to allow for meaningful interpretation of effects. That is, as it is not possible to
attain a score of zero for any of the scales on the
LSRP, the interpretation of regression coefcients
would be rendered meaningless. In contrast, standardization of these variables produces a mean,
which has a value of zero and, thus, allows for the
interpretation of coefcients at the average presentation of these traits.
In the rst equation, total psychopathy scores
were entered as the independent variable to establish
whether
psychopathy
predicts
engagement
in unprovoked aggression. The model proved
to be signicant, w2(1, N 5 105) 5 8.03; Po.005,
R2N 5 .09. The negative coefcient for psychopathy,
Wald w2(1, N 5 105) 5 8.51; b 5 .36, Po.005,
indicated that as self-reported psychopathy trait
scores increase the probability of being an unprovoked aggressor (i.e., lower ordinal grouping) also
increases. By exponentiating the regression coefcient for the model, an odds ratio (OR) can be
derived. The OR for the independent variables was
0.70, indicating that a one standard deviation
increase in psychopathy increases the probability
of being an unprovoked aggressor by 30%.
Next, a simultaneous ordinal regression equation
was computed with F1 and F2 of the LSRP to
determine the relative contribution of a given
variable when the other is held constant. For
the full model containing both factors, t statistics
indicated that the model containing both coefcients
signicantly predicted unprovoked aggression,
w2(1, N 5 105) 5 8.40, Po.05, R2N 5 .09. However,
although the coefcient for F1 was signicant,
Wald w2(1, N 5 105) 5 6.22; b 5 .32, P 5 .01, the
coefcient for F2 did not appear to contribute to the
models
predictive
power,
Wald
w2(1,
N 5 105) 5 0.55; b 5 .08, P4.50. The OR for F1
was 0.72, indicating that a one standard deviation
increase in F1 increases the probability of being an
unprovoked aggressor by 28%. A summary of
F1
F2
Total P
Wald w2
OR
.32
6.22
0.55
8.24
0.72
0.91
0.70
28
9
30
.09
.36
DISCUSSION
Block 1
Block 2
Unprovoked
aggressors
Provoked
aggressors
Nonaggressors
0.017 (0.72)
0.316 (0.54)
0 (0)
0.128 (0.57)
0 (0)
0 (0)
3
Bushman and Anderson [2001] have put forth a thoughtful critique
on the hostile versus instrumental dichotomy arguing that the
distinction is problematic because it fails to take into account
aggressive acts in which mixed motives are present and makes
assumptions that often are not met, regarding the level of
automaticity versus control that is present in the two forms of
aggression.
Aggr. Behav.
REFERENCES
Anath CV, Kleinbaum DG. 1997. Regression models for ordinal
responses: A review of methods and applications. Int J Epidemiol
26:13231333.
Anderson CA, Bushman BJ. 2002. Human aggression. Ann Rev
Psychol 53:2751.
Barry TD, Thompson AH, Barry CT, Lochman JE, Adler K, Hill K.
2007. The importance of narcissism in predicting proactive and
reactive aggression in moderately to highly aggressive children.
Aggr Behav 33:185280.
Benning SD, Patrick CJ, Blonigen D, Hicks BM, Iacono WG. 2005.
Estimating facets of psychopathy from normal personality traits:
A step toward community epidemiological investigations. Assessment 12:318.
Berkowitz L. 1988. Frustrations, appraisals, and aversively stimulated aggression. Aggr Behav 14:311.
Berkowitz L. 1993. Pain and aggression: Some ndings and
implications. Motiv Emotion 17:277293.
Bettencourt BA, Miller N. 1996. Gender differences in aggression as
a function of provocation: A meta-analysis. Psychol Bull 119:
422447.
Aggr. Behav.
Aggr. Behav.