ASSESSMENT Stanford

10.1177/1073191103252064
ARTICLE et al. / AGGRESSIVE BEHAVIOR

Characterizing Aggressive Behavior

Matthew S. Stanford
University of New Orleans

Rebecca J. Houston
University of Connecticut School of Medicine

Charles W. Mathias
University of Texas at Houston

Nicole R. Villemarette-Pittman
Laura E. Helfritz
Sarah M. Conklin
University of New Orleans

In the research literature, aggressive behavior has traditionally been classified into two dis-
tinct subtypes, impulsive or premeditated. Impulsive aggression is defined as a hair-trigger
aggressive response to provocation with loss of behavioral control. Premeditated aggression
is defined as a planned or conscious aggressive act, not spontaneous or related to an agitated
state. The present study outlines the development of a clinically useful self-report instrument,
the Impulsive/Premeditated Aggression Scales (IPAS), designed to characterize aggressive
behavior as predominately impulsive or predominately premeditated in nature. The IPAS
showed strong reliability and validity. Analysis of the IPAS scores demonstrated the presence
of two types of aggressive behavior, impulsive and premeditated, in men referred for anger
problems. The aggression of most individuals in the present sample was characterized as
predominately impulsive in nature (90%).

Keywords: aggression; assessment; personality; violence; impulsivity

Although aggressive behavior is a potential correlate of incorporated an intent to do harm into their definition of
a number of psychiatric disorders and a significant clinical aggression (Eron, Walder, & Lefkowitz, 1971). Others
issue, a common and clinically relevant definition has re- consider this aspect essential in defining aggressive be-
mained elusive. The difficulty in defining aggression has havior (Barratt, 1991; Berkowitz, 1993; Geen, 1990). A
resulted in part because of the large number of biological, third issue has been the variety of behaviors that can and
cultural, environmental, and social forces that influence have been deemed aggressive. Overt aggression can vary
the expression of this problem behavior (Eron, 1987; from verbal aggression to homicide with a large number of
Lindsay & Anderson, 2000). A second issue in defining possible intermediate behaviors. For the sake of clinical
aggression has been the concept of intentionality or an in- relevance, aggression in the present study is defined as a
tent to do harm. Some researchers have argued that inten- behavioral display in which physical force is used with the
tionality is difficult or impossible to measure and have not intent to harm or damage another individual or object.

Correspondence concerning this article should be addressed to Matthew S. Stanford, Department of Psychology, University of New
Orleans, New Orleans, LA 70148; phone: (504) 280-5525; fax: (504) 280-6049; e-mail: mstanfor@uno.edu. This research was supported
by the Dreyfus Health Foundation, the Rogosin Institute, New York Hospital–Cornell Medical Center.
Assessment, Volume 10, No. 2, June 2003 183-190
DOI: 10.1177/1073191103252064
© 2003 Sage Publications

Downloaded from asm.sagepub.com at PENNSYLVANIA STATE UNIV on March 5, 2016

184 ASSESSMENT

Within the research literature, physical aggressive be- sity of aggressive acts, normalization of event-related po-
havior has traditionally been classified into two distinct tentials (P3), and improvement in mood state measures
subtypes (Barratt, 1991; Barratt, Stanford, Dowdy, during a 6-week trial of the anticonvulsant Phenytoin (Di-
Liebman, & Kent, 1999; Barratt, Stanford, Kent, & lantin) when compared to placebo. Inmates whose aggres-
Felthous, 1997; Dodge & Coie, 1987; Feshbach, 1964; sive behavior was classified as premeditated showed no
Vitiello et al., 1990; Weinshenker & Siegel, 2002): (a) an improvements during the same trial (Barratt, Stanford,
emotionally charged, uncontrolled type of aggressive dis- Felthous, et al., 1997).
play (impulsive, unintentional, affective, reactive) or (b) a Despite the evidence supporting the notion that persons
planned, controlled, unemotional aggressive act (premedi- who display impulsive aggressive behavior are distinct
tated, intentional, predatory, proactive). Impulsive aggres- from their premeditated counterparts, the clinical useful-
sive behavior is considered reactive in nature and is usually ness of these classification techniques is suspect because
accompanied by an agitated or irritated mood, poor modu- most were developed on incarcerated violent offenders or
lation of physiological arousal, and a loss of behavioral psychiatric inpatients. This makes interpretation difficult
control (Barratt, 1991). In humans, this type of aggressive given the demand characteristics (e.g., fear of future prose-
display is similar to descriptions of episodic dyscontrol cution, loss of privileges) that arise when testing incarcer-
syndrome (EDS) (Monroe, 1970) in the research literature ated or institutionalized individuals. In addition, it is
and intermittent explosive disorder (IED) (American Psy- uncertain how these results may generalize to members of
chiatric Association, 2000) in the clinical literature. In the general population where a majority of aggressive acts
contrast, premeditated aggression is goal oriented, re- take place. Finally, the development of many of these in-
quires forethought and planning, and is generally executed struments did not follow standard psychometric proce-
with low autonomic arousal (Stanford, Houston, dures, making the reliability and validity of these
Villemarette-Pittman, & Greve, 2003). Characteristically, measures questionable (Heilbrun et al., 1978; Linnoila
these two subtypes are distinguished by several features, et al., 1983).
but primarily by the amount of behavioral control exhib- In an attempt to overcome past methodological con-
ited during the incident. A number of research criteria, cerns, the present study outlines the development of a
semistructured interviews, and scales have been proposed clinically useful self-report instrument designed to charac-
to classify violent individuals or the aggressive acts they terize an individual’s aggressive acts as either predomi-
display as impulsive or premeditated (Barratt et al., 1999; nately impulsive or predominantly premeditated in nature.
Barratt, Stanford, Kent, et al., 1997; Heilbrun, Heilbrun, & It was hypothesized that two distinct sets of items (scales)
Heilbrun, 1978; Linnoila et al., 1983; Vitiello et al., 1990). would emerge that support the impulsive and premeditated
Application of these classification techniques has been characterization. It was further predicted that these scales
useful in demonstrating significant differences in neuro- would correlate differently with published instruments de-
chemistry, psychophysiology, cognitive functioning, and signed to assess anger and antisocial/aggressive behavior.
treatment/intervention outcomes between those catego-
rized as impulsive versus premeditatedly aggressive. Spe-
cifically, impulsive aggressive individuals have been
METHOD
found to have lower levels of cerebrospinal fluid 5-
hydroxyindolacetic acid (Linnoila et al., 1983), impaired
prefrontal function (Raine et al., 1998), lower P3 event- Participants
related potential amplitude (Barratt, Stanford, Kent, et al.,
1997), and lower scores on measures of verbal ability and Participants were 93 physically aggressive men (level
intelligence (Barratt, Stanford, Kent, et al., 1997; Vitiello of anger/aggression problem reported: physical assault n =
et al., 1990) when compared to premeditated aggressors. 80; throw objects/damage property n = 13; mean age =
Two studies have also investigated the differential out- 35.9, SD = 10.6; mean years of education = 12.9, SD = 2.9;
comes of intervention/treatment of these aggressive sub- marital status: married n = 52, divorced/separated n = 19,
types. Heilbrun et al. (1978) found that murderers whose never married n = 22; employment status: employed n =
aggression was classified as impulsive were more likely to 77, unemployed n = 16). Twenty-three of the participants
fail on parole than those whose murders were classified as (25%) were referred for assessment from local mental
premeditated. In a pharmacological treatment study com- health clinics because of problems with anger/aggression.
paring aggressive subtypes, incarcerated inmates whose The remaining 70 participants (75%) were self-referred
aggressive behavior was classified as impulsive in nature after hearing a radio advertisement asking for men who
showed significant reductions in the frequency and inten- have trouble controlling their anger. Axis I and Axis II di-

Downloaded from asm.sagepub.com at PENNSYLVANIA STATE UNIV on March 5, 2016

 Stanford et al. / AGGRESSIVE BEHAVIOR 185

agnoses were assigned according to DSM-IV-TR (Ameri- Personality and Aggression Measures
can Psychiatric Association, 2000) criteria. Among the 93
participants in the present study, 28 met criteria for at least The following battery of personality and aggression
one Axis I diagnosis: major depression, n = 11; alcohol measures were administered in random order to each par-
abuse, n = 10; substance abuse, n = 10; obsessive-compul- ticipant.
sive disorder, n = 3; generalized anxiety disorder, n = 2;
Barratt Impulsiveness Scale (BIS-11) (Patton, Stan-
psychosis not otherwise specified, n = 1. Seventy-seven
ford, & Barratt, 1995). The BIS-11 is a 30-item self-report
participants met criteria for an Axis II personality disor-
questionnaire designed to assess general impulsiveness
der: antisocial, n = 44; obsessive-compulsive, n = 24;
taking into account the multifactorial nature of the con-
narcissistic, n = 6; borderline, n = 2; paranoid, n = 1.
struct. The items are scored on a 4-point scale (1 = rarely/
Twelve participants were comorbid for an Axis I and Axis
never, 2 = occasionally, 3 = often, 4 = almost always/
II disorder.
always). Published reliability coefficients for the BIS-11
(Cronbach’s alpha) range from .79 to .83 (Patton et al.,
Impulsive/Premeditated 1995).
Aggression Scales (IPAS)
Eysenck Personality Questionnaire (EPQ) (Eysenck &
A 30-item self-report instrument was developed to as- Eysenck, 1975). The EPQ is a 90-item self-report instru-
sess the impulsive and/or premeditated characteristics as- ment that assesses three broad dimensions of personality:
sociated with an individual’s aggressive acts (Table 1). extraversion (E), neuroticism (N), and psychoticism (P).
Impulsive aggression is defined as a hair-trigger aggres- The items are scored yes or no. Published reliability coeffi-
sive response to provocation with loss of behavioral con- cients for the EPQ (Cronbach’s alpha) range from .71 to
trol. Premeditated aggression is defined as a planned or .92 (Eysenck & Eysenck, 1975).
conscious aggressive act, not spontaneous or related to an Buss-Perry Aggression Questionnaire (BPAQ) (Buss &
agitated state. The content of the scales was based on pre- Perry, 1992). This 29-item self-report questionnaire is
vious scales and studies that have attempted to differenti- composed of four aggression subscales: physical aggres-
ate impulsive and premeditated aggressive participants sion, verbal aggression, anger, and hostility. A total ag-
(Barratt et al., 1999; Barratt, Stanford, Kent, et al., 1997; gression score is obtained by summing the subscales. The
Heilbrun et al., 1978; Heilbrun, Knopf, & Bruner, 1976; items are scored on a 5-point scale (1 = extremely unchar-
Linnoila et al., 1983; Stanford, Greve, & Dickens, 1995; acteristic of me to 5 = extremely characteristic of me). Pub-
Stanford, Greve, & Gerstle, 1997; Virkkunen, De Jong, lished reliability coefficients for the BPAQ (Cronbach’s
Bartko, & Linnoila, 1989; Vitiello et al., 1990). alpha) range from .72 to .89 (Buss & Perry, 1992).
Participants were asked to consider their aggressive
acts during the last 6 months and complete the IPAS in re- State-Trait Anger Expression Inventory (STAXI)
lation to those acts. Fifteen of the items focus on impulsive (Spielberger, 1996). This 44-item self-report scale as-
aggressive (IA) characteristics, and 15 items focus on pre- sesses both state and trait anger. Trait anger (T-Anger) is
meditated aggressive (PM) characteristics. The items are composed of two subscales, angry temperament (T-Anger/
scored on a 5-point Likert-type scale (strongly agree = 5, T), which measures a general propensity to experience and
agree = 4, neutral = 3, disagree = 2, strongly disagree = 1). express anger without provocation, and angry reaction
To characterize the participants’ aggressive behavior as (T-Anger/R), which measures the disposition to express
predominately impulsive or premeditated using the IPAS, anger when criticized or treated unfairly by others. In addi-
items for which the individual answered “strongly agree” tion, anger expression and its major components (anger in,
or “agree” were rated as positive. The total number of posi- anger out, anger control) are measured. The items are
tive items was then determined for both the IA and PM scored on a 4-point scale (1 = not at all, 2 = somewhat, 3 =
scales. The percentage of positive items for each aggres- moderately, 4 = very much so). Published reliability coeffi-
sion scale was then calculated. The aggressive behavior of cients for the STAXI (Cronbach’s alpha) range from .64 to
those individuals showing a higher percentage of positive .92 (Spielberger, 1996).
IA characteristics was characterized as predominately im- Lifetime History of Aggression (LHA) (Coccaro,
pulsive, whereas those showing a higher percentage of Berman, & Kavoussi, 1997). This 11-item semistructured
positive PM characteristics were characterized as predom- interview assesses three subcomponents of aggression his-
inately premeditated. Individuals showing the same per- tory: aggression, social consequences/antisocial behavior,
centage of positive IA and PM characteristics were not and self-directed aggression. A total score is obtained by
characterized. summing the subscale scores. The total number of occur-

Downloaded from asm.sagepub.com at PENNSYLVANIA STATE UNIV on March 5, 2016

186 ASSESSMENT

rences of each item since the age of 13 are scored as fol- differences between correlations were determined using
lows: none = 0, 1 event = 1, 2 or 3 events = 2, 4 to 9 events focused comparisons of effect size (Rosenthal, 1984).
= 3, 10 or more events = 4, more events than can be Concurrent validity was assessed by determining the level
counted = 5. Published reliability coefficients for the of agreement between the IPAS and the aggression inter-
LHA (Cronbach’s alpha) range from .48 to .88 (Coccaro view (Baldessarini, Finklestein, & Arana, 1983).
et al., 1997).
Aggression Interview (Stanford & Barratt, 2001). This
RESULTS
semistructured interview is designed to characterize an in-
dividual’s aggressive acts as either predominately impul-
sive or predominately premeditated in nature. The Item Analysis
interview is composed of open-ended questions related to
Two IPAS items did not meet the criteria for continued
specific aggressive acts the individual has identified dur-
analysis (significant item-total correlation and significant
ing the last 6 months (e.g., Discuss your motives for the
discrimination of extreme groups). The following items
[aggressive act] as you remember them; How did you feel
were removed from further analysis: I was concerned for
after you committed the [aggressive act]?). Thus, there is
my personal safety during the acts; The acts were a “re-
little overlap with the specific items on the IPAS. Interrater
lease,” and I felt better afterward.
reliability for the interview has been shown to be high (κ =
.79). Previous studies using the aggression interview have
shown significant personality, neuropsychological, and Principal Components Analysis
psychophysiological differences between impulsive and
premeditated aggressive individuals (Barratt, Stanford, The remaining 28 items were submitted to an explor-
Felthous, et al., 1997; Barratt, Stanford, Kent, et al., 1997; atory PCA. Using tables provided by Lautenschlager
Stanford et al., in press). Subsequent to the aggression in- (1989), three factors were retained for varimax rotation
terview, participants completed the IPAS. (Table 1). The factor labels were the following: Factor 1,
Premeditated Aggression; Factor 2, Impulsive Aggres-
sion; and Factor 3, Familiarity With Target/Remorse/Agi-
Data Analyses
tation. Two items (I was confused during the acts; I was in
a bad mood the day of the incident) did not load on a factor
IPAS development. An item analysis (Nunnally &
at or above .40 and were excluded from further analysis.
Bernstein, 1994) was performed on the IPAS scales.
Given that the goal of the study was to develop scales to as-
Pearson’s product-moment item-total correlations were
sess specifically impulsive and premeditated aggression,
computed between each item and its respective scale with-
Factor 3 was not included in further analyses.
out the item of interest included in the scale score. To de-
The present sample size is small compared to that tradi-
termine whether items differentiated extreme groups,
tionally suggested for PCA (5-10 participants per item).
t tests were computed between item scores in the top and
Sample size has long been considered the most important
bottom quartiles of the respective scale. Items that did not
influence on the reliability of components. However,
have significant corrected item-total correlations (p < .05)
Guadagnoli and Velicer (1988) have demonstrated that
and did not differentiate extreme groups were not included
component saturation (the absolute magnitude of the load-
in further analyses. The items that met criteria were sub-
ings) is the most important factor related to component re-
mitted to an exploratory principal components analysis
liability. Interpretation of the present data followed
(PCA). The PCA was conducted with no assumptions re-
Gudagnoli and Velicer’s recommendation that compo-
garding the number of potential factors. The number of
nents with four or more loadings above .60 in absolute
factors to be rotated was determined by the use of tables
value are reliable, regardless of sample size.
provided in Lautenschlager (1989). A varimax rotation
with .40 as the minimum loading criterion was used to ob-
tain an orthogonal factor solution. Cronbach’s alpha Internal Consistency (Reliability)
(Nunnally & Bernstein, 1994) was used to determine the and Intercorrelation
internal consistency (reliability) of the IPAS scales.
Cronbach’s alpha for the Premeditated Aggression
IPAS validation. Construct validity was assessed by ex- (PM) scale was .82. The alpha coefficient for the Impul-
amining the correlations between scores on the IPAS and sive Aggression (IA) scale was .77. The IA and PM scales
standard personality and aggression measures. Significant were not significantly intercorrelated (r = –.02).

Downloaded from asm.sagepub.com at PENNSYLVANIA STATE UNIV on March 5, 2016

 Stanford et al. / AGGRESSIVE BEHAVIOR 187

TABLE 1
Principal Components Analysis of Impulsive/Premeditated Aggression Scales (IPAS)
IPAS Item 1 2 3

I think the other person deserved what happened to them during some of the incidents. .75 –.04 –.18
I am glad some of the incidents occurred. .68 –.30 –.20
I wanted some of the incidents to occur. .68 –.07 .00
The act led to power over others or improved social status for me. .66 –.04 .20
Some of the acts were an attempt at revenge. .60 .19 .16
I feel my actions were necessary to get what I wanted. .58 –.22 .00
I felt my outbursts were justified. .56 .04 –.07
I planned when and where my anger was expressed. .52 .48 .34
I was under the influence of alcohol or other drugs during the acts. .46 .08 .15
Sometimes I purposely delayed the acts until a later time. .45 –.42 .30
Anything could have set me off prior to the incident. .43 .25 –.04
I felt pressure from others to commit the acts. .43 .08 .03
I consider the acts to have been impulsive. –.06 .77 .09
I feel I lost control of my temper during the acts. .05 .68 .19
I feel I acted out aggressively more than the average person during the last 6 months. .17 .68 .23
a
I was in control during the aggressive acts. .12 –.67 –.09
When angry, I reacted without thinking. .19 .57 .36
My behavior was too extreme for the level of provocation. –.00 .54 .38
a
I understood the consequences of the acts before I acted. .07 –.50 .26
I usually can’t recall the details of the incidents well. .33 .43 –.25
I knew most of the persons involved in the incidents. –.05 .05 .67
I typically felt guilty after the aggressive acts. -.12 .23 .66
I feel some of the incidents went too far. –.17 .23 .56
Prior to the incidents, I knew an altercation was going to occur. .40 –.18 .49
My aggressive outbursts were usually directed at a specific person. .24 –.08 .46
I became agitated or emotionally upset prior to the acts. .16 .26 .41

% total variance 16.56 14.03 9.72

NOTE: Factor loadings above .40 are in italics.

a. Reverse-scored item.

premeditated, F(1, 91) = 32.2, p < .001; IA M = 6.2, SD =

Construct Validity 1.6; PM M = 3.0, SD = 1.7. Those characterized as premed-
itated endorsed a significantly higher number of PM items
Table 2 presents correlations between the IPAS scales compared with those characterized as impulsive aggres-
and the personality and aggression measures. The IA scale sive, F(1, 91) = 25.3, p < .001; PM M = 7.7, SD = 2.2;
correlated significantly higher than the PM scale with IA M = 3.3, SD = 2.5.
BPAQ Anger and STAXI Anger Control. The PM scale Finally, an instrument’s ability to accurately character-
correlated significantly higher than the IA scale with EPQ ize an individual is indicated by the sensitivity, specificity,
Psychoticism, EPQ Extraversion, EPQ Neuroticism, and predictive values (Baldessarini et al., 1983) of said in-
BPAQ Physical, BPAQ Hostility, LHA Social Conse- strument. Therefore, sensitivity (probability of a positive
quences/Antisocial Behavior, LHA Self-Directed Aggres- test result among individuals with the condition); specific-
sion, and LHA Total Score. ity (probability of a negative test result among individuals
without the condition); and the predictive values, positive
Concurrent Validity predictive power (PPP) and negative predictive power
(NPP), were calculated for the IA and PM scales by com-
Using the IPAS, 84 (90%) individuals were character- paring the agreement between the semistructured aggres-
ized as predominately impulsive aggressive, and 9 (10%) sion interview and the IPAS characterization. PPP
were characterized as predominately premeditatedly ag- indicates the probability that an individual does have the
gressive. Individuals characterized as impulsive aggres- indicated condition given a positive test result, whereas
sive using the IPAS endorsed a significantly higher NPP indicates the probability that an individual does not
number of IA items compared with those characterized as have the condition given a negative test result. Using the

Downloaded from asm.sagepub.com at PENNSYLVANIA STATE UNIV on March 5, 2016

188 ASSESSMENT

TABLE 2 corrected PPP and NPP (cPPP and cNPP) estimates range
Relationship Between Personality and from –1 to +1, with 0 being chance predictive power and
Aggression Measures and the IPAS Scales positive numbers indicating predictive power greater than
IPAS Scale chance. The corrected predictive values for the IPAS were
Measure IA PM Z
a cPPP .45, cNPP .62 for the IA scale and cPPP .63, cNPP
.56 for the PM scale.
BIS-11 .21* .38** 1.80
EPQ
Psychoticism .03 .57** 5.94 DISCUSSION
Extraversion –.01 –.22* 2.06
Neuroticism .24* .40** 1.72
Lie –.04 –.02 0.19 The psychometric properties of the IPAS appear to be
BPAQ strong with only a few weak items and good internal con-
Physical .30** .45** 1.68 sistency (Cronbach’s α). The analyses suggest strong con-
Verbal .20 .32** 1.24 struct and concurrent validity for the IPAS scales. The
Anger .53** .16 4.13
Hostility .11 .34** 2.35
IPAS data clearly demonstrate the presence of two types of
Total .38** .51** 1.57 aggressive behavior, impulsive and premeditated, in the
STAXI population assessed. The aggression of most individuals in
T-Anger .40** .41** 0.10 the present sample was characterized as predominately
T-Anger/T .37** .37** 0.00 impulsive in nature (90%). This is consistent with previous
T-Anger/R .31** .31** 0.00
Anger In .26* .39** 1.16
research using the same semistructured aggression inter-
Anger Out .26* .29* 0.26 view. In that study, 81% (n = 65) of the sample (aggressive
Anger Control –.42** –.09 4.27 psychiatric outpatients) were characterized as predomi-
Anger Expression .25* .38** 1.15 nately impulsive, whereas 19% (n = 15) were character-
LHA ized as predominately premeditated (Stanford et al.,
Aggression .24* .30** 0.63
Social Consequences/
2003).
Antisocial Behavior –.09 .36** 4.49 Although not originally developed to characterize ag-
Self-Directed Aggression –.01 .27** 2.69 gressive behavior, the self-report instruments used in the
Total .08 .41** 3.42 present study do appear to have distinct impulsive and pre-
NOTE: IPAS = Impulsive/Premeditated Aggression Scales; IA = Impul- meditated subcomponents. The IA scale correlated signifi-
sive Aggression scale; PM = Premeditated Aggression scale; BIS-11 = cantly higher than the PM scale with measures that assess
Barratt Impulsiveness Scale; EPQ = Eysenck Personality Questionnaire; irritability (BPAQ Anger) and anger control (STAXI An-
BPAQ = Buss-Perry Aggression Questionnaire; STAXI = State-Trait An-
ger Expression Inventory (n = 63); LHA = Lifetime History of Aggres- ger Control). This result is in agreement with previous re-
sion. search that has demonstrated high levels of irritable
a. Absolute value reported, numbers in italics indicate a significant differ- impulsiveness and physiological reactivity in individuals
ence between correlations (p < .05).
*p < .05. **p < .01. who display impulsive aggressive outbursts (Coccaro
et al., 1989; Demaree & Harrison, 1997; Mawson & Maw-
son, 1977; Stanford et al., 1995). The PM scale correlated
significantly higher than the IA scale with measures of an-
semistructured aggression interview, 81 (87%) individuals tisocial/impulsive behavior (BIS-11, EPQ Psychoticism,
were characterized as predominately impulsive, 10 (11%) LHA Social Consequences/Antisocial Behavior), extra-
were characterized as predominately premeditated, and 2 version (EPQ Extraversion), neuroticism (EPQ
(2%) could not be characterized. Sensitivity for the IA Neuroticism), physical aggression (BPAQ Physical), hos-
scale was .96 (i.e., only 3.8% of individuals characterized tility (BPAQ Hostility), self-harm (LHA Self-Directed
as IA by the interview were not characterized as IA on the Aggression), and overall aggression (LHA Total). This
IPAS), and specificity was .50. For the PM scale, sensitiv- profile of results is consistent with previous studies that
ity and specificity were .60 and .96, respectively. have shown antisocial/psychopathic individuals engage in
It has been demonstrated that PPP and NPP are highly more premeditated aggressive behavior (Williamson et al.,
influenced by the base rate of the condition in the popula- 1987; Woodworth & Porter, 2002) and lack emotional
tion or sample (Baldessarini et al., 1983). To incorporate awareness (Meloy, 1997). The BIS-11 has been shown to
base rate and achieve more accurate estimates, the PPP and correlate highly with antisocial behavior (Stanford &
NPP were corrected for base rate by subtracting the num- Barratt, 1992), which may explain the present relationship
ber of agreements expected by chance from the number of with the PM scale. Anger (BPAQ Total, STAXI T-Anger,
observed agreements (Frick et al., 1994). The resulting STAXI Anger Expression) and general aggression scores

Downloaded from asm.sagepub.com at PENNSYLVANIA STATE UNIV on March 5, 2016

 Stanford et al. / AGGRESSIVE BEHAVIOR 189

(LHA Aggression) were significantly, but moderately, individuals fall into a “mixed” group showing both impul-
correlated with both the IA and PM scales. sive and premeditated aggressive characteristics (Barratt
Of the instruments used in the present study, the BPAQ et al., 1999; Vitiello et al., 1990). Consequently, it is sug-
seems to characterize aggressive behavior most similarly gested that aggressive behavior should be characterized as
to the IPAS, with the Anger and Hostility subscales mea- predominately impulsive or predominately premeditated
suring impulsive and premeditated aggression, respec- in nature. The IPAS was developed to provide a simple
tively. The STAXI, on the other hand, treats aggression as a technique by which this characterization might be accom-
more unitary construct. With the exception of Anger Con- plished. This approach should work well with the present
trol, the STAXI subscales did not differentially correlate diagnostic system, which in most instances subsumes ag-
with the IA and PM scales. The mixing of irritability and gression under some broader psychopathology (e.g.,
hostility items on the T-Anger scale and the significant mood disorder, personality disorder). Within this scheme,
overlap between Anger Control and Anger Out the characterization of aggressive behavior would be used
(Spielberger, 1996) likely results in nonspecific subscales by the clinician in addition to other relevant information to
that correlate highly with both impulsive and premeditated make treatment decisions (e.g., pharmacotherapy vs. in-
characteristics. sight-oriented therapy). It is anticipated that this approach
Sensitivity, specificity, and the predictive values (cPPP to characterizing aggression will facilitate the successful
and cNPP) for the IPAS scales were in the moderate to high treatment of this problem patient group.
range. The IA scale of the IPAS accurately identified 96%
of those individuals characterized as predominantly im-
pulsive via the interview process. The PM scale was accu- REFERENCES
rate for 60% of those individuals identified as pre-
dominantly premeditated by the interview. Therefore, the American Psychiatric Association. (2000). Diagnostic and statistical
manual of mental disorders–text revision (4th ed.). Washington, DC:
sensitivity of both IPAS scales appears to be very high. The American Psychiatric Press.
discrepancy between the two scales (96% vs. 60%) is Baldessarini, R. J., Finklestein, S., & Arana, G. W. (1983). The predictive
likely due to the high incidence of predominantly IA indi- power of diagnostic tests and the effect of prevalence of illness. Ar-
viduals in the present sample. Specificity for the IA scale chives of General Psychiatry, 40, 569-573.
Barratt, E. S. (1991). Measuring and predicting aggression within the
was low, .50. This figure suggests that the IA scale may context of a personality theory. Journal of Neuropsychiatry and Clin-
misidentify approximately 50% of individuals as IA when ical Neurosciences, 3, S35-S53.
they might be otherwise characterized through the inter- Barratt, E. S., Stanford, M. S., Dowdy, L., Liebman, M. J., & Kent, T. A.
view process. The PM scale had excellent specificity of (1999). Impulsive and premeditated aggression: A factor analysis of
self-reported acts. Psychiatry Research, 86, 163-173.
.96. It is important to note that the low specificity of the IA Barratt, E. S., Stanford, M. S., Felthous, A. R., & Kent, T. A. (1997). The
scale, as well as the high specificity of the PM scale, may effects of phenytoin on impulsive and premeditated aggression: A
be due in part to the disproportionate number of predomi- controlled study. Journal of Clinical Psychopharmacology, 17, 341-
349.
nantly IA versus predominantly PM participants in this Barratt, E. S., Stanford, M. S., Kent, T. A., & Felthous, A. R. (1997).
sample. However, regardless of the base rate of impulsive Neuropsychological and cognitive psychophysiological substrates of
or premeditated aggressive behavior in the sample, the impulsive aggression. Biological Psychiatry, 41, 1045-1061.
corrected predictive values (cPPP and cNPP) suggest that Berkowitz, L. (1993). Aggression: Its causes, consequences and control.
New York: McGraw-Hill.
the accuracy of both IPAS scales is very good. Buss, A. H., & Perry, M. (1992). The aggression questionnaire. Journal
Despite the strong reliability and validity of the IPAS, of Personality and Social Psychology, 63, 452-459.
several limitations must be considered when interpreting Coccaro, E. F., Berman, M. E., & Kavoussi, R. J. (1997). Assessment of
life history of aggression: Development and psychometric character-
the present data. First, the sample included only men; istics. Psychiatry Research, 73, 147-157.
therefore, these results may not generalize to aggressive Coccaro, E. F., Siever, L. J., Klar, H. M., Maurer, G., Cochrane, K., Coo-
women. Second, the recruiting procedures may not have per, T. B., et al. (1989). Serotonergic studies in patients with affective
generated a representative aggressive sample given that and personality disorders: Correlates with suicidal and impulsive ag-
gressive behavior. Archives of General Psychiatry, 46, 587-599.
75% of the individuals were self-referred from radio ad- Demaree, H. A., & Harrison, D. W. (1997). Physiological and
vertisements. Finally, the potential social and legal conse- neuropsychological correlates of hostility. Neuropsychologia, 35,
quences of reporting calculated, premeditated aggressive 1405-1411.
behavior may influence how a participant responds on the Dodge, K. A., & Coie, J. D. (1987). Social information processing factors
in reactive and proactive aggression in children’s peer groups. Jour-
IPAS. nal of Personality and Social Psychology, 53, 1146-1158.
Although past studies of aggression subtypes have Eron, L. D. (1987). The development of aggressive behavior from the
tended to categorize aggressive individuals into two mutu- perspective of a developing behaviorism. American Psychologist, 42,
435-442.
ally exclusive categories, it appears that a dimensional ap- Eron, L. D., Walder, L. O., & Lefkowitz, M. M. (1971). Learning of ag-
proach is more useful. A majority of aggressive gression in children. Boston: Little, Brown.

Downloaded from asm.sagepub.com at PENNSYLVANIA STATE UNIV on March 5, 2016

190 ASSESSMENT

Eysenck, H. J., & Eysenck, S. B. G. (1975). Eysenck Personality Ques- Stanford, M. S., Houston, R. J., Villemarette-Pittman, N. R., & Greve,
tionnaire (Junior and Adult), manual. San Diego, CA: Educational K. W. (2003). Premeditated aggression: Clinical assessment and cog-
and Industrial Testing Services. nitive psychophysiology. Personality and Individual Differences, 34,
Feshbach, S. (1964). The function of aggression and the regulation of ag- 773-781.
gressive drive. Psychological Review, 71, 257-272. Virkkunen, M., De Jong, J., Bartko, J., & Linnoila, M. (1989).
Frick, P. J., Lahey, B. B., Applegate, B., Kerdyck, L., Ollendick, T., Hynd, Psychobiological concomitants of history of suicide attempts among
G. W., et al. (1994). DSM-IV field trials for the disruptive behavior violent offenders and impulsive fire setters. Archives of General Psy-
disorders: Symptom utility estimates. Journal of the American Acad- chiatry, 46, 604-606.
emy of Child and Adolescent Psychiatry, 33, 529-539. Vitiello, B., Behar, D., Hunt, J., Stoff, D., & Ricciuti, A. (1990).
Geen, R. G. (1990). Human aggression. Pacific Grove, CA: Brooks/Cole. Subtyping aggression in children and adolescents. Journal of
Guadagnoli, E., & Velicer, S. (1988). Relation of sample size to the stabil- Neuropsychiatry and Clinical Neurosciences, 2, 189-192.
ity of component patterns. Psychological Bulletin, 103, 265-275. Weinshenker, N. J., & Siegel, A. (2002). Bimodal classification of ag-
Heilbrun, A. B., Heilbrun, L. C., & Heilbrun, K. L. (1978). Impulsive and gression: Affective defense and predatory attack. Aggression and Vi-
premeditated homicide: An analysis of subsequent parole risk of the olent Behavior, 7, 237-250.
murderer. Journal of Criminal Law and Criminology, 69, 108-114. Williamson, S., Hare, R. D., & Wong, S. (1987). Violence: Criminal psy-
Heilbrun, A. B., Knopf, I. J., & Bruner, P. (1976). Criminal impulsivity chopaths and their victims. Canadian Journal of Behavioural Sci-
and violence and subsequent parole outcome. British Journal of ence, 19, 454-462.
Criminology, 16, 367-377. Woodworth, M., & Porter, S. (2002). In cold blood: Characteristics of
Lautenschlager, G. J. (1989). A comparison of alternatives to conducting criminal homicides as a function of psychopathy. Journal of Abnor-
Monte Carlo analyses for determining parallel analysis criteria. mal Psychology, 111, 436-445.
Multivariate Behavioral Research, 24, 365-395.
Lindsay, J. J., & Anderson, C. A. (2000). From antecedent conditions to
Matthew S. Stanford, Ph.D., is currently associate professor
violent actions: A general affective aggression model. Personality
and Social Psychology Bulletin, 26, 533-547. and chair of the Department of Psychology at the University of
Linnoila, M., Virkkunen, M., Scheinin, M., Nuutila, A., Rimon, R., & New Orleans, New Orleans, Louisiana. His research focuses on
Goodwin, F. K. (1983). Low cerebrospinal fluid 5-hydroxy- the biological bases and treatment of impulsive and aggressive
indolacetic acid concentration differentiates impulsive from non- behavior.
impulsive violent behavior. Life Sciences, 33, 2609-2614.
Mawson, A. R., & Mawson, C. D. (1977). Psychopathy and arousal: A Rebecca J. Houston, Ph.D., received her doctoral degree in ap-
new interpretation of the psychophysiological literature. Biological plied biopsychology from the University of New Orleans. She is
Psychiatry, 12, 49-74.
currently completing a postdoctoral fellowship in the Depart-
Meloy, J. R. (1997). Predatory violence during mass murder. Journal of
Forensic Science, 42, 326-329. ment of Psychiatry at the University of Connecticut Medical
Monroe, R. R. (1970). Episodic behavioral disorders: A psychodynamic School. Her research interests include the physiological bases of
and neurophysiologic analysis. Cambridge, MA: Harvard University impulsive, aggressive, and antisocial behavior particularly in the
Press. context of substance abuse.
Nunnally, J. C., & Bernstein, I. H. (1994). Psychometric theory (3rd ed.).
New York: McGraw-Hill. Charles W. Mathias, Ph.D., received his doctoral degree in ap-
Patton, J. H., Stanford, M. S., & Barratt, E. S. (1995). Factor structure of plied biopsychology from the University of New Orleans. He is
the Barratt Impulsiveness Scale. Journal of Clinical Psychology, 51,
currently completing a postdoctoral fellowship in the Depart-
768-774.
Raine, A., Meloy, J. R., Bihrle, S., Stoddard, J., LaCasse, L., & Buchs-
ment of Psychiatry and Behavioral Sciences at the University of
baum, M. S. (1998). Reduced prefrontal and increased subcortical Texas Health Science Center. His area of research is in the
brain functioning assessed using positron emission tomography in psychophysiological and behavioral assessment of impulsivity
predatory and affective murderers. Behavioral Sciences and the Law, and aggression.
16, 319-332.
Rosenthal, R. (1984). Meta-analytic procedures for social research. Nicole R. Villemarette-Pittman, M.S., is currently a doctoral
Beverly Hills, CA: Sage. student in applied biopsychology at the University of New Or-
Spielberger, C. D. (1996). State-Trait Anger Expression Inventory man-
leans. Her research interests include the psychophysiological,
ual. Odessa, FL: Psychological Assessment Resources.
Stanford, M. S., & Barratt, E. S. (1992). Impulsivity and the multi-impul- neuropsychological, and personality substrates of aggression.
sive personality disorder. Personality and Individual Differences, 13,
831-834. Laura E. Helfritz, M.S., is currently a doctoral student in ap-
Stanford, M. S., & Barratt, E. S. (2001). Procedures for the classification plied biopsychology at the University of New Orleans. Her re-
of aggressive/violent acts, manual. New Orleans, LA: University of search interests include psychopharmacology and the
New Orleans. psychophysiological bases of aggression.
Stanford, M. S., Greve, K. W., & Dickens, T. J. (1995). Irritability and im-
pulsiveness: Relationship to self-reported impulsive aggression. Per- Sarah M. Conklin, M.S., is currently a doctoral student in ap-
sonality and Individual Differences, 19, 757-760. plied biopsychology at the University of New Orleans. Her re-
Stanford, M. S., Greve, K. W., & Gerstle, J. E. (1997).
search interests include the psychophysiology of aggression and
Neuropsychological correlates of self-reported impulsive aggression
in a college sample. Personality and Individual Differences, 23, 961- emotional processing.
966.

Downloaded from asm.sagepub.com at PENNSYLVANIA STATE UNIV on March 5, 2016