This is to certify that the
thesis entitled

A LIFE HISTORY APPROACH TO THE TWO FACTOR
MODEL OF PSYCHOPATHY: FEARLESS DOMINANCE AND
IMPULSIVE ANTISOCIALITY AS SLOW AND FAST LIFE
STRATEGIES

presented by

 

Melissa Marie McDonald

has been accepted towards fulfillment
of the requirements for the

 

M.A degree In Psychology

£ng MAIL

fessor's Signature
14/6/2615

Date I

 

 

MSU is an Affirmative Action/Equal Opportunity Employer

‘fi—_._

LIBRARY
Michigan State
University

 

 

 

 

 

 

PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE

DATE DUE DATE DUE

 

APR t 0 2014

 

 

 

 

 

 

 

 

 

 

 

 

 

5/08 KlProj/Aoc&Pres/ClRC/Date0ue.indd

 

 

 

 

 

A LIFE HISTORY APPROACH TO THE TWO FACTOR MODEL OF
PSYCHOPATHY: FEARLESS DOMINANCE AND IMPULSIVE ANTISOCIALITY
AS SLOW AND FAST LIFE STRATEGIES

By

Melissa Marie McDonald

A THESIS

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

MASTERS OF ARTS

Psychology

2010

ABSTRACT

A LIFE HISTORY APPROACH TO THE TWO FACTOR MODEL OF
PSYCHOPATHY: FEARLESS DOMINANCE AND IMPULSIVE ANTISOCIALITY
AS SLOW AND FAST LIFE STRATEGIES

By
Melissa Marie McDonald

Although psychopathy has previously been conceptualized as a unitary construct
reflecting a frequency-dependent cheating strategy (Harpending & Sobus, 1987), here
preliminary evidence for a multifaceted model of psychopathy wherein two independent
dimensions represent distinct life strategies is presented. Using life history theory as a

' framework, the first factor, fearless dominance (FD), is characterized as a “slow” life
strategy where protective early rearing environments lead to greater energy allocation for
development. The second factor, impulsive antisociality (IA), is characterized as a “fast”
life strategy associated with harsh rearing environments and energy allocations primarily
directed toward reproduction. Results indicate that the quality of family relationships, but
not resource availability, predicts the differential expression of FD and IA. Each factor, in
turn, is associated with outcomes that complement distinct life strategies: PD is positively
associated with self-monitoring whereas IA is positively associated with mating effort,
aggression, and impulsivity, and negatively associated with educational achievement.
These findings highlight the importance of examining psychopathy as a multifaceted
psychological construct and also of considering the role of the environment in directing

the expression of psychopathic personality factors.

To my loving and supportive parents, Anne and Ron McDonald

iii

ACKNOWLEDGEMENTS
I would like to thank Brent Donnellan, Tim Pleskac, Deborah Kashy, Carlos Navarrete,
and Edward Witt for their helpful comments on previous versions of this thesis. I would
also like to thank my research assistants for their help with data collection and data
management, they include: Benjamin Asher, Leslie Saxon, Andrew Defever, Ariel
Walgraev, Hans Schroder, Kate Rowley, Brett TeSlaa, Chelsea Wise, Ryan McNamara,

Jessica Ceh, Karlie Witbrodt, and Stephanie Spencer.

iv

TABLE OF CONTENTS

LIST OF TABLES ................................................................................... vi
LIST OF FIGURES ................................................................................. vii
INTRODUCTION ................................................................................... 1
Etiology of Psychopathy ................................................................... 3
Deficit and Disorder Hypotheses ................................................ 3
Psychopathy as an Evolved Life Strategy ....................................... 3
A Two Factor Model of Psychopathy .................................................... 1 1
Life History Theory... 16
Psychopathy as Two Independent Evolved Life Strategies . ..........20
The Present Research ....................................................................... 23
METHOD ............................................................................................ 27
Participants ................................................................................. 27
Measures .................................................................................... 27
Procedure .................................................................................... 39
Assessment of Selection Bias ............................................................ 40
RESULTS ............................................................................................ 42
Preliminary Analyses ...................................................................... 42
Primary Analyses ........................................................................... 43
Measurement Model .............................................................. 44
Structural Regression Models ................................................... 51
Multiple Groups Analysis ........................................................ 54
Mediation Analyses ............................................................... 55
DISCUSSION ....................................................................................... 59
Strengths and Limitations .................................................................. 59
Conclusions and Future Directions ....................................................... 67
APPENDICES ....................................................................................... 69
Appendix A — Tables and Figures ......................................................................... 70
Appendix B - Experimental Instructions for Decision Making Games ................. 82
REFERENCES ....................................................................................... 85

LIST OF TABLES

Table 1. Gain and Loss Structure for the Gambling Task ................................................ 71

Table 2. Means, Standard Deviations, and Correlations for Predictor and Outcome
Variables ......................................................................................................... .72

Table 3. Means, Standard Deviations, and Correlations for Gendered Variables. ......... 73

Table 4. Factor Loadings and Residual Variances for the Original, Reduced, and
Final Measurement Models ............................................................ 74

Table 5. Measurement Model: Nested Comparisons and Model F1t75
Table 6. Final Measurement Model and Structural Regression Model Fit Indices. . .......76

Table 7. Structural Regression model Factor Correlations and Covariances (SE)

with Outcome Variables ................................................................. 77
Table 8. Structural Regression Model Path Coefficients ....................................... 78
Table 9. Multiple Groups Analysis for Gender: Nested model Comparisons .................. 79

vi

LIST OF FIGURES

Figure 1. Path Model for the Associations between Early Rearing Environment,
Psychopathy Factors, and Outcome Variables ............................................ 80

Figure 2. Structural Regression Model for the Association between Relationships,
Psychopathy Factors, and Outcome Variables .................................. 81

vii

Introduction

Psychopathy is a personality disorder that is typically characterized as a
constellation of interpersonal and affective traits coupled with a propensity to engage in
antisocial behaviors. Psychopaths are generally perceived as lacking a moral compass, as
evidenced by their ability to harm others for their own gain without remorse. More
specifically, the interpersonal and affective traits of psychopathy include a charming,
intelligent, fearless, and manipulative interpersonal style with affective deficits in social
emotions such as empathy, guilt, and shame. The behavioral component of psychopathy
is defined by a tendency to engage in impulsive, irresponsible, and repeated antisocial
behavior. In 1941, psychiatrist Hervey Cleckley published one of the most influential
clinical descriptions of psychopathy to date. Cleckley described psychopathy as a
“convincing mask of sanity,” such that psychopaths are able to charm others into
believing that they are trustworthy, moral, and good intentioned, only to reveal more
egocentric intentions when an opportunity for self-interested gains or a necessity to
protect oneself arises. It is this “two-faced” nature of psychopathy that makes it a
particularly compelling and dangerous personality disorder. Not surprisingly, prison
populations contain an abundance of individuals who would meet a formal diagnosis for
psychopathy (Hart & Hare, 1997). However, not all psychopaths find themselves behind
bars for their offenses. A subgroup of psychopaths embody many of the core features of
psychopathy but manifest them in such a way that results in successful careers in
medicine, academia, and business (Cleckley, 1941). In an attempt to reconcile these two
disparate expressions of psychopathy, Hall and Benning (2006) summarized three

potential conceptualizations for the etiology of the “successful” psychopath. They

suggested that this noncriminal expression could result from: (1) a subclinical
manifestation of the disorder, (2) attenuating factors (e. g. IQ, education, SES parenting,
etc), or a (3) primary expression of the interpersonal-affective component, rather than the
antisocial behavioral component, of psychopathy.

The present research will investigate the nature of successful psychopathy,
arguing that what is deemed as “successful” can be construed in multiple ways. From a
life history perspective (Kaplan & Gangestad, 2005), risky life strategies that appear to be
maladaptive may have evolved as adaptive responses to specific aspects of one’s
environment. In this sense, a risky strategy could be construed as successful to the extent
that it aids in one’s survival. Thus, I will argue that, contrary to the idea that only the
interpersonal/ affective component of psychopathy can function adaptively, that both
dimensions of psychopathy can be construed as adaptive in that they represent two
distinct life strategies that evolved to meet the challenges of different environments. As
such, it is posited that stressful environments favor riskier life strategies, and that the
antisocial behavior profile of psychopathy represents such a risky strategy. Alternatively,
in more harmonious environments, risky strategies are unnecessary, thus making a life
strategy characterized by the interpersonal/affective components of psychopathy more
adaptive. For a given life strategy to function adaptively, it must ‘fit’ with the
environmental context, as such, individuals likely evolved the ability to flexibly express
life strategies contingent upon available cues that indicate the level of stress in the
environment. Thus, this conceptualization predicts that early rearing context, serving as a

cue of environmental stress, will be associated with the expression of the two factors of

psychopathy, which will in turn be associated with differential behavioral outcomes that
function adaptively in the specific environment in which they arose.
Etiology of Psychopathy

Deficit and disorder hypotheses. Previous research examining the etiology of
psychopathy has targeted a deficit in morality, particularly the ability to feel empathy for
other individuals (Blair, 2008; Mealey, 1995; Soderstrom, 2003). Such a deficit would
result in an inability or unwillingness to predict or understand the emotions experienced
by victims. Others have proposed a broader deficit in the ability to feel any of the social
emotions, including shame, sympathy, guilt, and love (Mealey, 1995). Some explanations
have focused on neuroanatomical deficits, such as a hypoaroused nervous system that
interferes with one’s sensitivity to social influences generally and aversive stimuli
specifically (Benning, Patrick, & Iacono, 2005; Darnasio, 1996; Eliasz & Reykowski,
1986; Eysenck & Gudjonsson, 1989; Raine, 1997). This deficit is also related to a
lowered susceptibility to fear conditioning, and thus a resistance to the standard child
rearing techniques used to socialize children (Eliasz, 1987; Lykken, 1957; Schachter &
Latané, 1964; Schmauk, 1970). Additionally, more recent research points to reduced
grey matter in the prefrontal cortex (Raine, Lencz, Bihrle, LaCasse, & Colletti, 2000;
Yang, Raine, Lencz, Lacasse, & Colletti, 2005; but see Laakso, Gunning-Dixon,Vaurio,
Repo-Tiihonen, Soininen, & Tiihonen, 2002 for an exception) and deficits in amygdala
functioning (Blair, 2007; Rilling, Glenn, Jairam, Pagnoni, Goldsmith, Elfenbeing, et al.,
2007).

Psychopathy as an evolved life strategy. Despite the prevalence of deficit

hypotheses in regard to psychopathic functioning, other researchers have argued that

psychopathy should not be conceptualized as a true disorder, if one defines disorder as
referring only to “conditions in which harm to the organism is caused by... a failure of
some internal mechanism to operate as it was naturally design ” (Wakefield, 1992, p.
242). That is, to the extent that psychopathic individuals function in a way that has been
preferentially selected for over the course of our evolutionary history, then psychopathy
represents a reproductively viable life strategy that is counter to the idea of psychopathy
as a disorder (Harris, Rice, Lalumiére, 2001). Harris and colleagues argue that the suite of
risky behaviors associated with psychopathy likely led to successful reproduction among
our ancestors, thereby maintaining its presence in the population. The authors
acknowledge that the expression of psychopathy must be mediated by neuroanatomical
differences, but that these are not to be construed as defects if they exist to serve an
adaptive purpose in an evolutionary context. As such, a wide body of research has
focused on demonstrating that psychopaths do not display gross neuroanatorrrical
damage, strong associations with other psychopathology, or failure to perform well on
tasks designed to assess brain damage. Using structural equation modeling, Harris et al.
demonstrated that of the three most consistent predictors of criminal violence
(psychopathy, antisocial parenting, and neurodevelopmental insults) psychopathy and
neurodevelopmental insults were independently and directly related to criminal violence,
that is, neurodevelopmental insults and psychopathy were unrelated to one another.
Along similar lines, researchers have failed to find significant deficits among
psychopaths versus nonpsychopaths on a number of neuropsychological tests used to
asses brain damage and impaired cognitive functioning (Hare, 1984; Hart, Forth, Hare,

1990). Lalumiére, Harris, and Rice (2001) presented a series of studies demonstrating

that psychopathic violent offenders score lower than nonpscyhopathic violent offenders
on an obstetrical problems scale (associated with low IQ, mental retardation, epilepsy,
autism, schizophrenia, anxiety and behavior disorders, and other developmental
problems), thereby demonstrating that psychopaths actually experience fewer problems in
utero and during birth than nonpsychopathic offenders, which are typically a precursor to
a number of disorders that occur later in life. Additionally, if psychopathy was related to
developmental instability, one would expect psychopaths to score higher on a measure of
fluctuating asymmetry (an indicator of developmental instability and poor genetic
quality), however, psychopaths score very similarly to non-psychopaths on this
dimension. Furthermore, psychopaths were less likely to have received a diagnosis of
psychosis than nonpscyhopathic offenders, and psychopathy was positively correlated
with attractiveness ratings (a measure of high genetic quality). These results suggest that
the neuroanatomical differences associated with psychopathy are much more subtle and
finely tuned than the gross deficits associated with many other disorders. As such, this
evidence is more aligned with the hypothesis that psychopathy represents an evolved life
strategy rather than a “true” disorder.

Conceptualizing psychopathy as an adaptive life strategy is important in that it
provides a distal explanation for its emergence in the population and its continued
prevalence despite the seemingly negative outcomes associated with its expression. This
approach to psychopathy follows a trend in evolutionary psychology to explain
personality traits and disorders in terms of their ability to enhance an organism’s
reproductive fitness (Buss, 1991; Buss, 2009; F igueredo, Sefcek, Vasquez, Brumbach,

King, & Jacobs, 2005; Nettle, 2006). For example, researchers have posited evolutionary

explanations for a number of personality traits and disorders including the five factor
model of personality (Nettle, 2006), social dominance (Hawley, 1999), schizotypy (Nettle
& Clegg, 2006) and the collection of “Dark Triad” traits: Machiavellianism, narcissism,
and psychopathy (J onason, Li, Webster, & Schmitt, 2009). Nettle’s analysis of the five
factors of personality is interesting in that, much like psychopathy, some personality traits
are associated with a whole host of negative outcomes. However, Nettle argues that there
are a number of ways in which traits like low agreeableness and high neuroticism could
have evolved. The high levels of anxiety associated with neuroticism would have proved
beneficial in contexts where a vigilant and wary outlook is needed to avoid predation or
being cheated out of resources. Similarly, though agreeableness fosters more trusting and
cooperative interactions with others, an unconditional trust of others would most certainly
be taken advantage of by defectors, thus a more aggressive and untrusting interpersonal
style can be also be construed as adaptive. Nettle’s arguments help clarify the idea that,
though plagued with negative connotations today, a variety of personality traits and
disorders may have evolved because they increased reproductive fitness during our
evolutionary history.

Evolutionary explanations of psychopathy typically assert that it evolved as a
fiequency dependent cheating strategy. Throughout our evolutionary history, individuals
competed for the necessary resources to survive and reproduce. Strategies that cheated
others out of resources (including mates, food, and status) would have evolved only to the
extent that they were relatively rare in the population. That is, as the number of cheaters
in the population increase, adaptations to detect cheaters also increase and the likelihood

of a cheater encountering a cooperator decreases. This results in more failed interactions

where both individuals attempt to cheat the other out of a valued resource. As such,
Harpending and Sobus (1987) argued that a cheating strategy could only evolve under
conditions in which cheaters are rare, difficult to detect, able to easily move from one
group to another, verbally skilled, and adept at securing mating opportunities. With these
conditions present, it is not difficult to imagine the adaptive utility inherent in the
psychopathic profile. A cunning, manipulative, un-empathic interpersonal style
complemented with a risky, remorseless, and antisocial behavioral style would quickly
produce gains in fitness through acquisition of valued resources with minimal exertion of
physical effort.

If psychopathy is to be considered a frequency dependent, adaptive life strategy
that has been selected for over evolutionary time a demonstration of its heritability is
necessary. In a recent meta-analysis, Waldman & Rhee (2006) found evidence for
moderate additive genetic effects and non-shared environmental influences, along with
small non-additive genetic effects and shared environmental influences on antisocial
behavior. Although these assessments only focused on antisocial behavior, at the
exclusion of the interpersonal and affective factor of psychopathy, other studies that
include both factors of psychopathy have also demonstrated genetic heritability (Bloni gen
et al., 2005; Viding et al., 2005).

Additional evidence in support of psychopathy as an evolved life strategy is
apparent in the types of crimes they typically commit. Psychopaths have a tendency to
engage in more instrumental or goal directed violence that increases their reproductive
fitness, as opposed to the emotionally reactive violence that typifies the crimes

committed by nonpsychopaths. These reactive crimes are likely to also have reproductive

benefits, but are qualitatively different in that they lack the forethought that is inherent in
many crimes of psychopaths. In a study of 3 1 5 male inmates, Williamson, Hare, and
Wong (1987) found that psychopaths, as compared to nonpsychopaths, were significantly
less likely to have committed murder as the target offense responsible for their
imprisonment. Rather, the most prevalent target offenses for psychopaths were armed
robbery and property crimes (theft, breaking and entering, etc). An examination of the
motives for all crimes committed involving a victim revealed that psychopaths were
much more likely to have committed the offense out of a desire for material gain,
whereas nonpsychopaths cited strong emotional arousal (jealousy, rage, heated argument)
as the primary motive for their offenses. Though psychopaths commit more crimes on
average, their crimes tend to be less violent to the extent that they are less likely to result
in victim death, and more likely to result in no harm at all to the victim. Finally, the
victims of psychopaths were much more likely to be strangers, whereas the victims of
nonpsychopaths were more likely to be a family member or friend. Thus, it seems
possible that the crimes of psychopaths are violent only to the extent that the violence is
necessary to achieve a goal, whereas the crimes of nonpsychopaths are usually
emotionally overwhelming, directed at friends or family as a result of an altercation, and
thereby much more likely to be lethal. These results were replicated in a later study of
106 male inmates (Cornell, Warren, Hawk, Stafford, Oram & Pine, 1996) where the
authors found that psychopaths were more likely to exhibit patterns of violent offending
relating to both instrumental and emotionally reactive aggression, whereas the violence
committed by nonpsychopaths was characterized solely by reactive aggression. As such,

instrumental offenders were more likely to plan the crime in advance with an identifiable

end goal. Reactive offenders were more likely to have known their victim, report being
provoked by the victim, and to have acted out of anger. The results of these studies
suggest that the crimes committed by psychopaths are not typically the result of fleeting
moments of rage or jealousy, but rather that they are often premeditated, carefully
planned, and goal directed crimes committed in the service of increasing one’s
reproductive fitness. Furthermore, that most of the victims of their crimes are strangers
bolsters the assumption that psychopaths pursue a cheating strategy that requires high
mobility between groups so as to prevent their violent reputation fi'om preceding them.

Further evidence of an evolved psychopathic cheating strategy comes from
research investigating the mating styles of psychopaths. Thomhill and Thomhill (1992)
have suggested that sexual coercion by men could represent a psychological adaptation
such that ancestral men who engaged in forced matings experienced increased
reproductive fitness, thereby creating selection pressure for a rape specific adaptation.
The authors also proposed an alternative explanation that defines rape as a by product or
side effect of other more general psychological adaptations, such as the desire for sex and
a general coercive strategy used to acquire resources. Regardless of the adaptive status of
sexual coercion it is probable that psychopaths would be most likely to act on a
predisposition for rape as they perceive fewer social constraints and already exhibit
greater proclivities for risky and impulsive behavior. Thus, as a result of their geographic
mobility, one should expect the mating strategy of psychopaths to be characterized by a
series of uncommitted copulations. Psychopaths might also endorse a more coercive
sexual style to secure mates that capitalizes on their verbal manipulation skills and

willingness to engage in physical aggression when nonviolent tactics fail. Additionally,

for a psychopathic life strategy to be adaptive, a surplus of mating opportunities would be
required in order to offset the ephemeral life span often associated with a lifestyle of
excessively risky behavior.

Research investigating the sexual strategies of psychopaths has confirmed these
expectations. Psychopaths direct more effort than nonpsychopaths toward promiscuous
mating (Lalumiere & Quinsey, 1996), and endorse a more uncommitted approach to
romantic relationships (Seto, Lalumiere, & Quinsey, 1995). The sexual victims of
psychopaths are more likely to be of reproductive age (Harris, Rice, Hilton, Lalumiere, &
Quinsey, 2004; Quinsey, Rice, & Harris, 1995), and the sexual offense is more likely to
include genital to genital contact (Harris, et al., 2004). Psychopaths are less likely to
target children as sexual victims, and more likely to be rapists than molesters (Harris,
Rice, Hilton, Lalumiere & Quinsey, 2007; Porter, Fairweather, Drugge, Herve, Birt, &
Boer, 2000). Psychopaths are also more likely to use both violent and nonviolent sexually
coercive tactics than nonpsychopaths (Kosson, Kelly, & White, 1997; Lalumiere, et al.,
1996). This collection of findings strongly suggests that a coercive and promiscuous
mating style occupies a central role within psychopathic personality disorder. As such,
Harris and colleagues (2007) demonstrated that a three factor model of psychopathy that
included the two factors commonly associated with psychopathy (an
interpersonal/affective component and an antisocial behavioral component) and an
additional factor composed of variables indicating early, frequent and coercive sex
resulted in greater model fit than when the items comprising the coercive and precocious
sexuality factor were split among the other two factors that they correlated most highly

with. Thus, it seems clear that the cheating strategy engaged in by psychopaths contains a

10

propensity for sexually coercive and promiscuous behavior that may represent a core
aspect of psychopathy, and at the least, has proved important in the propagation of
psychopathic genes.
A Two Factor Model of Psychopathy

Though the conceptualization of psychopathy as an adaptive life strategy is not
new, all such previous research in this area has considered psychopathy as a unitary
construct (Harpending & Sobus, 1987; Harris et al., 2001; Harris et al., 2007; Jonason, et
al., Lalumiere et al., 2001). One exception to this was Mealey (1995) who argued that
there were two paths to sociopathy (psychopathy). The first was conceptually very
similar to the argument advanced by Harpending and Sobus, such that (primary)
sociopathy was described as being a normally distributed trait, where a relatively small
number of individuals at the extreme end of the distribution would fill a small, frequency
dependent, evolutionary niche. For the second path, Mealey argued that individuals
genetically disposed to (secondary) sociopathy who fall on the less extreme end of the
continuum might come to display sociopathic like traits under circumstances that make
its expression profitable, such as when resources are scarce. Though Mealey’s
conceptualization of primary and secondary sociopathy has many merits to which I will
seek to expand on, the research makes no distinction between the two dominant factors
that define psychopathy (the interpersonal/affective and antisocial/behavioral
components). These factors are typically collapsed into a total psychopathy score, or
alternatively, the interpersonal/affective dimension is ignored. This is evident in the
prevalence of researchers who consider a diagnosis of Antisocial Personality Disorder

(APD), a disorder predominantly characterized by behavioral symptomatology (DSM-IV,

ll

1994), as synonymous with psychopathy (Hare, 1996; Hare & Hart, 1995). However, the
interpersonal and emotional aspects of psychopathy, such as superficial charm,
remorselessness, and an absence of nervousness were central to Cleckley’s (1941)
original conceptualization of psychopathy. In fact, the sixteen criteria that Cleckley
outlined as being the defining features of psychopathy, many of which are interpersonal
and affective items, were the starting point for the design of Hare’s Psychopathy
Checklist—Revised (PCL-R; Hare, 1991, 2003), which is now widely considered to be
the preferred diagnostic tool for psychopathy, demonstrating good validity and reliability
(Hare, 1991). Interestingly, although the PCL was designed to assess a unitary construct,
it nevertheless exhibits a two-factor structure, one factor relating to the interpersonal and
affective features of psychopathy and the other relating to the antisocial deviance
features. Though other factor structures have been proposed for the PCL-R (Cooke &
Michie, 2001; Hare & Neurnann, 2006), the two factor structure is most common. Factor
one, or the emotional/interpersonal factor, includes the items: glib and superficial,
egocentric and grandiose, lack of remorse or guilt, lack of empathy, deceitful and
manipulative, and shallow emotions. Factor two, or social deviance, includes the items:
impulsive, poor behavioral controls, need for excitement, lack of responsibility, early
behavior problems, and adult antisocial behavior.

Though the PCL-R has demonstrated good psychometric properties, its primary
drawback is that it was designed to assess psychopathy among clinical samples. Thus,
because psychopathy is now largely considered to be a continuous trait (Cleckley, 1941;
Edens, Marcus, Lilienfeld, Poythress, 2006; Mealey, 1995), the PCL-R is not suited for

use among non-clinical samples. However, the Psychopathic Personality Inventory (PPI;

12

Lilienfeld, & Andrews, 1996) and its subsequent revision (PPI-R; Lilienfeld & Widows,
2005) was designed to detect psychopathic traits in noncriminal samples. The PPI breaks
down into 8 lower order facets that combine to create a total score representing global
psychopathy. However, Benning, Patrick, Hicks, Blonigen, and Krueger (2003)
demonstrated that the PPI also conforms to a two factor structure similar to that of the
PCL. The first factor consisted of the three sub-factors social influence (propensity to be
charming, engaging, influential), fearlessness (lack of anticipatory anxiety and an
eagerness to engage in risks) and stress immunity (tendency to remain calm under
pressure). The second factor consisted of the four sub-factors Machiavellian egocentricity
(willingness to manipulate others for social goals), rebellious nonconformity (propensity
for unconventionality, anti-authority attitudes, and defiance of social norms), blame
extemalization (the perception that the world is hostile and responsible for one’s
problems), and carefree nonplanfirlness (lack of forethought and a failure to consider
alternatives). Benning, Patrick, Blonigen, Hicks, and Iacono (2005) later labeled the first
factor “Fearless Dominance” (FD) and the second factor “Impulsive Antisociality” (IA);
this naming convention will be adopted throughout the rest of the paper unless referring
specifically to the factors of the PCL-R. Importantly, the total score of the PPI and its two
factors show good convergence with the PCL-R, with correlations ranging fi'om .40 - .54
(Poythress, Edens, & Lilienfeld, 1998). Interestingly, much like the PCL, the PPI was not
created with a two factor structure in mind. Rather, a broad, over inclusive, and
personality based definition of psychopathy was used to generate a large pool of items
which was then reduced down using factor analytic techniques. Thus, despite the fact that

the PPI was not created with a two factor structure in mind, it nevertheless conforms to

13

such a structure, and exhibits relatively strong correlations with the PCL-R’s two-factors
(however, see Neumann, Malterer, & Newman, 2008 for a critique of the PPI’s factor
structure within offender populations)

Interestingly, the factors of the PPI are largely orthogonal, whereas the factors of
the PCL-R are moderately correlated. The factors of the PPI also exhibit strikingly
different correlates. For example, across two different samples (a sample of university
undergraduates and a sample of male prison inmates), Benning et al. (2005) demonstrated
that FD exhibited significant negative correlations with measures of fearfulness and
anxiety, and significant positive correlations with measures of narcissism, sociability, and
sensation seeking. Alternatively, IA exhibited significant positive correlations with
fearfulness, anxiety, impulsivity, sensation seeking, and significant negative correlations
with sociability, and had a weak to non-existent relationship with narcissism. In other
research (Benning et a1. 2003) FD is positively associated with education level, high
school class rank, positive emotionality and negatively associated with negative
emotionality. In contrast, IA is negatively correlated with educational achievement,
income, verbal intelligence, constraint, and positively associated with negative
emotionality. The orthogonal and divergent factors of the PPI have led some researchers
to speculate that a reconceptualization of the construct of psychopathy may be warranted
(Lilienfeld & Fowler, 2006). Whereas typical syndromes are defined by a set of
covarying symptoms, psychopathy might instead be characterized by two independent
sets of symptoms. In accordance with this, Hicks, Markon, Patrick, Krueger, and
Newman (2004) used cluster analysis techniques to examine the different personality

profiles of male offenders who had scored highly on the PCL-R. Two clusters were

14

identified, an “aggressive” subgroup and a “stable” subgroup. The correlations of each
group with different personality measures suggested that the aggressive group was
conceptually similar to the second factor of the PCL-R, whereas the stable group was
similar to the first factor of the PCL. This analysis suggests that psychopathy is not a
unitary construct, but rather contains at least two distinct and independent facets.
Recently, researchers have proposed a dual-deficit approach to psychopathy
(F owles & Dindo, 2006; Fowles & Dindo, 2009; Patrick, 2007; Patrick & Bernat, 2009).
This approach suggests that the two factors of psychopathy are the result of distinct
etiological processes. Patrick and Bemat (2009) argue that [A represents a predisposition
towards externalizing problems, whereas FD actually protects against the experience of
internalizing problems. Thus, despite the fact that internalizing and externalizing
behaviors tend to exhibit moderate positive correlations with each other, psychopathy is
typically only associated with externalizing behaviors. Patrick thereby postulates that
psychopathy represents an integration of two distinct etiological processes, one involving
a deficit in fear reactivity that leads to an immunity towards internalizing problems, and a
deficit in impulse control that leads to a vulnerability towards externalizing problems.
Evidence in support of this distinction has come from studies demonstrating a fear
specific deficit among individuals high on Factor 1 of the PCL (Patrick, Bradley & Lang,
1993), and a cognitive processing deficit among individuals scoring high on factor 2 of
the PCL (Patrick, Cuthbert, & Lang, 1994). Thus, it seems that it is the combination of
both factors that leads to the expression of “true” psychopathy. However, because the
factors are orthogonal and seem to have distinct etiologies, it seems likely that each factor

evolved separately.

15

It is easy to see from the descriptions of the correlates of FD and IA that FD in
particular exhibits relationships with what are more typically considered positive traits,
such as educational achievement, low anxiety and fear, and a protection against
internalizing symptoms. Alternatively, IA seems to be associated with a whole host of
negative traits, including low educational achievement, low income, low intelligence,
high anxiety and fear, and a general tendency to engage in externalizing behaviors. Thus,
if one were to speculate as to the nature of the “successful” psychopath, one would
probably assume that it resulted from greater expression of traits related to FD as opposed
to those related to IA. Although, to make such a case, one must first define what
constitutes a successful life strategy. From an evolutionary perspective, success depends
on the environment in which the life strategy is employed.

Life History Theory

Life history theory is an evolutionary theory that focuses on the trade offs
individuals must make in energy allocation. Each individual has a finite amount of
available energy and therefore must “decide” what tasks to allocate that energy to in
order to reap the greatest benefits. Energy can be allocated towards bodily growth and
maintenance, reproduction, or parenting/kin investment (Kaplan & Gangestad, 2005).
Individuals who allocate more energy to their growth and development will likely incur a
number of future advantages including heightened intelligence, skillful acquisition of
resources, ascension through a status hierarchy, and greater ability to secure optimal
mating opportunities. However, early allocation of energy towards growth and
development requires that one delay reproduction, which can be a very risky strategy if

one does not expect to live long enough to reap the benefits of enhanced development.

16

Individuals who allocate more energy to reproduction face the trade off of quantity over
quality. Having many offspring means that the available resources must be divided
among them, thereby decreasing each child’s chance of survival. However, in very
uncertain environments with scarce resources, having many children maybe be the best
strategy, as it increases the probability that at least one child will survive. Similarly,
allocation of energy towards mating over parenting may result in fewer offspring
surviving. However, if one can assume that their mate will take care of their mutual
offspring, then abandoning the offspring in search of future mating opportunities can be
beneficial.

These different tradeoffs can be categorized into two distinct life history
strategies: fast and slow (Oli, 2004). A slow strategy is characterized by late sexual
maturation and subsequent delayed and reduced reproduction, which is then offset by
high levels of parental investment. A fast strategy is characterized by the opposite—early
sexual maturation, early reproduction, and many offspring with little investment in each.
Fast and slow life strategies are markedly different and most viable in specific
environments. Thus, the activation of one strategy over another should be strategic in that
its expression is sensitive to available cues from the environment in which it will be
employed. Environments that are harsh and unstable call for a fast life history strategy
because delaying reproduction might result in not reproducing before one’s death. This
life strategy is necessarily riskier because of the demands of the environment. When
scarce resources are competed over, some individuals are left with nothing, thus making a
risky, all or nothing strategy the most viable option. In contrast, stable environments with

predictable access to resources call for a slow life history strategy wherein individuals

17

can capitalize on the advantages of extended periods of growth and development later in
their lives.

In Belsky, Steinberg, and Draper’s 1991 theoretical paper, the authors proposed
that early environmental cues within the first 5-7 years of a child’s life direct the
expression of a slow or fast reproductive strategy. The relevant cues include the
availability and predictability of resources, the trustworthiness of others, and the
enduringness of interpersonal relationships. When the environmental cues suggest a
stressful environment, an insecure romantic attachment style develops that leads to a
short-term mating strategy characterized by earlier and more frequent reproduction.
Alternatively, when the environmental cues suggest a non-stressful
environment, a secure romantic attachment style develops that leads to a long-term
mating strategy characterized by delayed reproduction with fewer mates. These
environmental cues may be readily available in the environment, or they may be
implicitly expressed via one’s parents. That is, parental rearing strategies will reflect the
type of environment that parents expect their children to encounter. Support for this
model comes fiom a variety of sources, including a number of widely demonstrated
associations between rearing environment, parenting practices, attachment styles, and
developmental outcomes. For example, in a study of 54 nations, (Schmitt, Alcalay,
Allensworth, Allik, Ault, Austers, et al., 2004) found that nations with higher fertility
rates tended to have higher levels of insecure attachment among men and women (rs =
.32 - .38), and that two different measures of stressful environments correlated with
insecure attachment styles (rs = .34 - .48). Additionally, Schmitt (2004a) found that

short-terrn mating was linked to insecure attachment across cultures.

18

Also of interest are findings that demonstrate a relationship between
environmental context and reproductive outcomes across individuals. For example,
family conflict and father absence in childhood has been shown to predict early menarche
in women (Graber, Brooks-Gunn, & Warren, 1995; Moffitt, Caspi, Belsky, & Silva,
1992). Father absence is also associated with earlier sexual activity and adolescent
pregnancy (Ellis, Bates, Dodge, F ergusson, Horwood, Pettit, etal., 2003). Additionally,
earlier menarche is related to harsh maternal control and negative parenting strategies
(Belsky, Steinberg, Houts, Friedman, DeHart, Cauffrnan, et al., 2007). Divorce and
separation prior to the age of five has also been shown to predict a host of reproductive
outcomes in women, including earlier age at menarche and first sexual intercourse,
increased number of sexual partners, earlier age at first pregnancy, and shorter duration
of first marriage (Quinlan, 2003).

In a related area of research, Daly and Wilson (2001) have investigated the types
of environments that encourage a life strategy characterized by future discounting, that is,
the extent to which present rewards are valued over future rewards. A future discounter,
therefore, fails to plan for the future, and instead cashes in rewards as soon as they are
obtained. Such a strategy is usually construed as impulsive, impatient, or lacking in self
control, but Daly and Wilson argue that such a strategy is adaptive in unpredictable
environments. For example, a delay in gratification is most likely to occur in an
environment in which present effort can be converted into future reproductive success.
Thus, an individual in a resource rich environment might invest heavily in education in
order to increase one’s future status. In contrast, an individual in a resource scarce

environment might seek out a low paying job rather than going to college because the

19

unpredictability of resources may make planning for the future futile. As evidence of this,
Wilson and Daly (1997) found that male life expectancy at birth was the best available
predictor of homicide rates in Chicago neighborhoods. Similarly, Gartner (1990) found
that a measure of economic inequality was the best predictor of homicide rates when only
homicides of adult men were considered. These findings suggest that when the future is
unpredictable and competition over resources leaves some men with nothing, males
engage in riskier behavior, such as homicide, to gain access to valuable resources.
Additionally, local life expectancy also predicted truancy from school, particularly
primary school, suggesting that parents may prepare their children for an unpredictable
environment by focusing less on formal education and more on practical skills.

Taken together, the results outlined above provide evidence that environmental
cues early in one’s rearing, such as the availability and predictability of resources, the
degree of competition over those resources, one’s likely life expectancy, the enduringness
of close interpersonal relationships, and the trustworthiness of others, may contribute to
the life strategy adopted by an individual. As such, a risky and high mating effort strategy
is adopted when the environment is unpredictable and harsh, whereas a more deliberate
and low mating effort strategy is undertaken when the environment is stable and rich with
resources. Thus, the behavior of criminals, though typically considered to be
maladaptive, may actually be a legitimate response to the environment in which they
were reared.

Psychopathy as Two Independent Evolved Life Strategies
Previous work has considered psychopathy as a constellation of covarying

antisocial behaviors and interpersonal/emotional traits. As such, it was argued that the

20

combination of these two factors represented a fi'equency dependent life strategy
characterized by a manipulative, charming, and fearless interpersonal style, along with a
propensity to engage in risky, impulsive, and antisocial behaviors. However, recent work
in the measurement of psychopathy suggests that these factors are independent of one
another. This begs the question then, if these factors did not evolve together as a cohesive
cheating strategy, how did they evolve? The present research attempts to answer this
question using life history theory as an evolutionary framework from which to work.

As discussed previously, fearless dominance (FD) and impulsive antisociality
(IA) have markedly different correlates. These differences provide clues as to how each
factor may have evolved. That FD is associated with educational achievement, higher
socioeconomic status, and higher intelligence suggests that it may have evolved as a slow
life strategy where greater energy is allocated towards growth and development.
Alternatively, the associations of IA with lower intelligence, socioeconomic status, and
educational achievement are suggestive of a fast life strategy wherein less effort is
allocated toward growth and development and more effort is allocated to mating.

Additional evidence for this characterization of F D and IA comes from studies
that have examined the factor score correlations of psychopathy with outcomes relevant
to reproductive success. For example, Blair, Mitchell, & Blair (2005) found that FD was
positively associated with instrumental aggression, whereas IA was positively associated
with reactive aggression. This finding supports the life strategy approach to the two
factors of psychopathy in that both types of aggression, reactive and instrumental, are
likely adaptive, but are characteristic of different life strategies. Reactive aggression

should be more typical of a risky, present oriented and impulsive life strategy, whereas

21

instrumental aggression should be more strongly associated with a careful, future oriented
life strategy. Additionally, in a review of the relevant literature, Knight & Guay (2006)
concluded that the relationship between sexual coercion and psychopathy is most likely
driven by the impulsive-antisocial deviance component of psychopathy. This finding is
also in accordance with the idea that IA is expressed as a fast, or risky and impulsive, life
strategy. Also of interest, in a recent exploration of psychopathic behavior in a prisoner’s
dilemma game Mokros, Menner, Eisenbarth, Alpers, Lange, & Osterheider (2008) found
that individuals who scored highly on psychopathy (as measured by the PPI-R) were
more likely to choose to defect, i.e. not cooperate, than nonpsychopaths, which resulted
in greater monetary rewards. The subfactors of psychopathy that were most predictive of
these gains were two subscales of the IA factor, Machiavellian egocentricity and
rebellious nonconforrnity (along with the total score). The other two subscales that
comprise IA also correlated positively with monetary reward (although not significantly),
whereas the components of FD did not. These results suggest that a risky strategy, such
as IA, can sometimes be beneficial, and provide some evidence demonstrating how such
a strategy could evolve.

Using the model provided by Belsky et al. (1991) wherein early environmental
indicators of resource availability, enduringness of close interpersonal relationships, and
the trustworthiness of others adaptively channel children down one of two reproductive
pathways, it is predicted that an adverse rearing environment characterized by a lack of
economic resources and unstable close relationships will be positively associated with the
expression of the IA facet of psychopathy. Alternatively, an adverse rearing environment

is predicted to be negatively associated with the expression of FD, that is, FD will be

22

associated with a less stressful rearing environment. Subsequently, FD and IA should
show differential associations with outcome variables that are relevant to their respective
slow and fast life history strategies. These outcome variables should function to increase
reproductive success, given the environment to which the life history strategy evolved in
response. More specifically, FD should be associated with outcomes that require
forethought and careful or strategic planning, whereas IA should be associated with
outcomes related to immediate gratification and a lack of consideration of future
consequences.

The Present Research

The present research attempts to test the prediction that the two factors of
psychopathy, FD and IA, represent slow and fast life strategies, respectively. Using life
history theory as the guiding framework, this prediction is explored by examining the
associations between one’s early rearing environment and the expression of the two
factors of psychopathy. In addition, the outcomes associated with the expression of each
factor will be examined in order to demonstrate that FD and IA are differentially
associated with outcome variables that are specifically adaptive for the type of rearing
environment in which the life history strategy evolved.

Specifically, this research seeks to explore how the availability of economic
resources and the quality of family relationships in the early rearing environment predict
the expression of FD and IA. These environmental characteristics are thought to provide
cues for individuals about the quality of the environment in which they will develop and
mature. If the environment provides cues that resources are scarce, family members are

untrustworthy, and romantic relationships are unstable, then it is increasingly likely that

23

the 1A factor of psychopathy (a fast life strategy) will be expressed. As a fast life strategy,
IA optimally responds to a harsh and unpredictable environment by increasing behaviors
that capitalize on access to immediate rewards, rather than waiting for and exerting a
great amount of effort to obtain larger, but much more uncertain, future rewards.

For the present research, a variety of reward seeking outcome variables are used
to test the prediction that IA represents a fast life strategy; these variables include
impulsivity, reactive aggression, risky behavior, educational achievement, and mating
effort. It is predicted that, given the proclivity for immediate gratification that
characterizes fast life strategies, IA will be associated with greater impulsivity, reactive
aggression, and riskier behavior during an economic decision making game.
Additionally, because energy in‘a fast life strategy is primarily allocated toward
reproductive ends, it is expected that IA will predict the exhibition of greater mating
effort. Mating effort will be defined in multiple ways including behavioral measures of
previous mating success, attitudes that favor the acquisition of multiple mates, the desire
to obtain many mating opportunities, and the endorsement of coercive behavior in order
to obtain access to mating opportunities. In addition, biological changes in the timing of
menarche among women will be investigated, as menarche signals the body’s readiness
for reproduction. Finally, it is expected that, in accordance with the reduced amount of
energy allocated toward cognitive growth and development, IA will be associated with
poor academic performance.

In contrast, the expression of PD is expected to emerge when the environment
provides cues that the availability of resources is abundant or adequate, family members

are trustworthy, and romantic relationships are stable and loving. As a slow life strategy,

24

FD optimally responds to a safe and predictable environment by capitalizing on the
likelihood that the environment is sufficiently safe that one will live long enough to reap
the benefits of delayed gratification rather than taking advantage of smaller more
immediate rewards. To test the prediction that FD represents a slow life strategy, it is first
expected that FD will not exhibit significant associations with mating effort, impulsivity,
reactive aggression, or risky behavior. In contrast, FD should be associated with
behaviors that utilize the enhanced cognitive development associated with a slow life
strategy. This should include a calculated, strategic, and proactive implementation of
aggressive behavior that is goal directed rather than reactive and impulsive. F D should
also be associated with greater educational achievement given the cognitive resources
available for such pursuits. Finally, FD should be associated with a greater ability to
monitor one’s own and other’s behavior, and subsequently adjust one’s own behavior in a
manner that permits the manipulation of social interactions to one’s benefit.
A model of the predicted relationships among variables is provided in Figure 1.
To summarize, the hypotheses for the study are as follows:
0 An early rearing context characterized by negative family relationships and a
lack of resources will predict the expression of IA.
0 An early rearing context characterized by positive family relationships and
abundant or adequate resources will predict the expression of PD.
0 IA will positively predict impulsivity, reactive aggression, risky behavior in an
economic decision making game, exertion of mating effort, and will negatively

predict educational achievement.

25

o FD will positively predict proactive aggression, educational achievement, and

self-monitoring behavior.

Although the assertion that risky and impulsive behavior can serve an adaptive
purpose is counter-intuitive, it is important to consider the specific environment fiom
which that suite of risky behaviors emerged. The outcome variables associated with each
factor of psychopathy are adaptive in that they generate opportunities to gain access to
valuable resources, which ultimately increase one’s reproductive fitness. It should be
rather straight forward that the type of strategy employed to obtain those resources should
vary as a function of the environment in which it is utilized. As such, it is expected that
the quality of one’s rearing environment will predict the expression of a particular life
strategy, which will in turn predict suites of behavior that function adaptively in that
specific type of environment.

These relationships will be analyzed using structural equation modeling. Separate
latent variables will be constructed for both family relationships and the availability of
resources. Each of these latent variables is expected to predict the expression of the two
factors of psychopathy. In turn, FD and IA are expected to differentially predict the
expression of the outcome variables previously described. The mediating role of FD and
IA in this process will be explored to determine the extent to which the two factors of
psychopathy can account for the association between early rearing environment and later
outcomes. This approach allows for the evaluation of the overall pattern of relations (i.e.

model fit) as well as an evaluation of specific associations between variables.

26

Method
Participants,

Participants were recruited through Michigan State University’s psychology
participant pool and were compensated for their participation via course credit. The study
was comprised of two parts, a laboratory session and a follow-up online survey. A total
of 440 participants attended the laboratory session; of these participants 282 (64%) also
completed the follow-up online survey. Participants who completed both parts of the
study comprise the primary sample for analyses. Given the gender distribution of the
participant pool at Michigan State University (60-7 5% women), a selection procedure
was used in which only males were recruited to participate in the study for a period of 7
days, every 3 weeks. This was done in order to recruit more equal numbers of males and
females. The resulting sample was comprised of 135 men (48%) and 147 women (51%).
The racial breakdown of the sample was as follows: 228 White (80.9%), 22 Asian/Pacific
Islander (7.8%), 21 Black (7.4%), 4 Hispanic/Latino (1.4%), 4 Middle Eastern (1.4%),
and 2 Other (.7%). The mean age of the sample was 19.5 (SD = 2.11), with a range from
18 — 45.

Measures

Given the assumption of multivariate normality in structural equation modeling
(Kline, 2005), all of the measures for this study underwent an analysis of their
distribution. When the ratio of skewness or kurtosis to standard error exceeded a value of
4, a transformation of the measure was performed. The specific transformation used was
determined by performing a series of transformations in order of least to most powerful

(square root, logarithmic, inverse, inverse squared) until the distribution could not be

27

normalized further. The resulting transformed variable was used in all subsequent
analyses. However, the original mean and standard deviation for the variables are
reported in text; the transformed values can be found in Table 2. When correlations or
regression coefficients are reported in text, effect sizes are interpreted using Cohen’s
recommendations: small = .10, medium = .30, large = .50 (Cohen, 2002).

Psychopathy. Participants completed three self-report measures of psychopathy.
The first measure, the Psychopathic Personality Inventory — Revised (PPI-R; Lilienfeld &
Widows, 2005) consists of 154 items that form 2 primary factors]: fearless dominance
(PPI FD; M = 3.02, SD = .40, alpha = .89) and impulsive antisociality (PPI IA; M = 2.28,
SD = .32, alpha = .91). The response scale was structured as follows: 1 = “False,” 2 =
“Mostly False,” 3 = “Mostly True,” and 4 = “True.” The two factors, FD and IA,
exhibited a small correlation of .15 (p < .05), suggesting that although the scales are
largely orthogonal, there is some overlapping content. Inspection of the sub-factors that
comprise FD and IA suggest that this overlap is likely a result of two moderate cross-
correlations, that is, rebellious non-conformity (a sub-factor of IA) correlates .36 (p <
.001) with FD, and fearlessness (a sub-factor of FD) correlates .30 (p < .001) with IA.

Participants also completed a 40 item measure of psychopathy constructed from
items measuring standard personality traits (Witt, Donnellan, & Blonigen, 2009) from the
International Personality Item Pool (IPIP; Johnson, 2000). Item responses were assessed
on a five-point scale ranging from 1 = “Strongly Disagree” to 5 = “Strongly Agree.” The
scale includes two 20 item subscales that serve as approximations of PD (IPIP F D; M =

3.47, SD = .56, alpha = .87) and IA (IPIP IA; M: 2.11, SD = .66, alpha = .91). Each

 

l . . . .
There rs a thrrd factor that comprises 16 items of the PPI-R, coldheartedness, which is largely distinct
from FD and IA, however, the factor tends to be more strongly related to IA (r = .25) than to FD (r = .17).

28

subscale demonstrated good convergence with the two factors of the PPI-R (r ’s = .65 -
.70). Additionally, the two factors exhibited a small correlation with one another (r = -
.12, p = .046).

The third revision of the Self-Report Psychopathy Scale (SRP-III; Paulhus,
Hemphill, & Hare) is a 64 item measure of psychopathy originally developed to be an
analogue to the clinician-administered PCL-R (Hare, 1991 , 2003). Responses are
recorded on a five—point scale ranging from 1 = “Strongly Disagree” to 5 = “Strongly
Agree.” Unlike the PPI-R and the [PIP measures of psychopathy, the SRP-III does not
produce a two factor solution. Instead, the scale is comprised of four subscales:
interpersonal manipulation, callous affect, erratic lifestyle, and criminal tendencies.
Consistent with previous findings (Witt & Donnellan, 2008), all subscales were strongly
correlated with other measures of IA. Using the PPI-R, average correlations between the
SRP subscales and IA range from .45 - .66 (mean r = .54) compared with a range of .10 -
.44 (mean r = .24) for FD. Using the [PIP measure of psychopathy, average correlations
between the SRP subscales and IA range fiom .62 - .73 (mean r = .67), compared with a
range of -.11 to .19 (mean r = .00) for FD. Given the strong associations between the SRP
subscales and IA measures, this scale was used only as a measure of IA. Additionally, as
these subscales are highly correlated with each other (r ’s = .47 - .68), the average of all
items was used in analyses (M = 2.25, SD = .50, alpha= .93).

Developmental History. Using the Belsky et al. (1991) model as a guide, two
sets of variables were assessed as indicators of the quality of the early rearing
environment. This model asserts that the development of a fast versus a slow life strategy

should be associated with the available cues in the child’s environment. These cues

29

include the availability and predictability of resources, the trustworthiness of others, and
the enduringness of interpersonal relationships. For this study, variables were included
that assessed the availability of resources and the quality of family relationships during
the participants’ childhood.

Resource Availability. To assess the availability of resources among participants
during their childhood, participants were asked to provide information about their
parents’ level of education and income. Although these items asked for current
information about the participants’ parents rather than retrospective accounts, it was
presumed that the rank-ordering of the data points would be relatively stable across time.
That is, parents with a current earned income in the top 10% of all parents are also likely
to have been in the top 10% of parents when the participant was a child. Some empirical
support for this comes from large scale panel studies in Germany and Switzerland in
which stability of income over the course of 6 years was approximately .48 in the Swiss
sample and .34 over the course of 15 years in the German sample (R.E. Lucas, personal
communication, April 15, 2010). It also seemed unlikely that participants would be able
to accurately recall information about their parents’ resources fiom a time when the
participant was very young.

Parental education was assessed via a seven-point categorical scale ranging fiom
l = “Elementary/Middle School” to 7 = “M.A. / Professional degree, Ph.D / M.D.”
(Mother: M = 4.52, SD = 1.69; Father: M = 4.56, SD = 1.67). The median and mode for
both mothers and fathers was 5 (Bachelor’s Degree). Combined annual parental income

was assessed with a ten point categorical scale ranging from 1 = “Less than $10,000” to

30

10 = “Greater than $200,000” (M = 7.06, SD = 2.09). The median and mode for
combined parental income was 7 ($70,000 - $100,000).

Family Relationships. To assess the quality of the participants’ relationship with
their mother and father as a child, the Mother/Father Relationship Quality scale was
administered (M/FRQ; Brim, Baltes, Burnpass, Cleary, Featherrnan, Hazzard, et al.,
2000). The scale consists of 13 items relating to the parent child relationship (e. g. “How
much love and affection did he/she give you?”; “How consistent was he/she about
rules?”). Each item is scored separately for the participants’ relationship with his/her
mother and father on a four-point response scale ranging from 1= “not at all” to 4 = “a
lot” (MRQ: M = 3.31, SD = .42, alpha = .80; FRQ: M = 3.11, SD = .56, alpha = .86).
These two scales were strongly correlated, r = .51. Additionally, because both scales
were negatively skewed, an inverse transformation was performed.

To assess the relationship quality between the participants’ parents, the Children’s
Perceptions of Interparental Conflict scale was completed by participants (CPIC; Grych,
Seid, & Fincham, 1992). The phrasing of the items was re-worded so that all items were
in the past-tense. That is, participants were reporting about conflict between their parents
when they were a child. Responses were made on a seven-point scale ranging from 1 =
“Strongly Disagree” to 7 = “Strongly Agree.” Only the Frequency (e. g. “I often saw my
parents arguing”), Intensity (e. g. “My parents got really mad when they argued”), and
Resolution (e. g. “My parents still acted mean after they had an argument”) subscales of
the CPIC were used. The other subscales were excluded because they focus on the
subjective experience of conflict and the specific content of the arguments, which

deviates from the intended construct, which is simply the level of conflict present in the

31

home. The subscales were averaged to create a total score (M = 3.19, SD = 1.45, alpha =
.96). Because the distribution of the CPIC was positively skewed, a square-root
transformation was performed. The scale was also reverse scored so that higher scores
indicate a more positive relationship (less conflict) among one’s parents.

Educational Achievement Participants reported their cumulative high school
grade point average (GPA) on a four-point scale (M = 3.59, SD = .33). High school GPA
was chosen as a marker of educational achievement (as opposed to college GPA) because
it reflects approximately four years of educational testing over a wide breadth of courses.
Given that a majority of this sample were freshman or sophomore undergraduates (60%
of the sample was aged 18-19), high school GPA was presumed to be an easily accessible
number to recall. In addition, because the data were collected during the Fall semester,
freshman in the sample did not yet have a college GPA on record.

Self-Monitoring The revision of the Self-Monitoring scale (R-SM; Lennox &
Wolfe, 1984) was used to assess the extent to which individuals are sensitive to the
expressive behavior of others and the extent to which they are able to modify their self-
presentation in response. The scale consists of 13 items and two subscales: (1) ability to
modify self-presentation (e. g. “I have the ability to control the way I come across to
people, depending on the impression I wish to give them.”), and (2) sensitivity to
expressive behavior of others (e. g. “I am often able to read people’s true emotions
correctly through their eyes.”). Responses were recorded on a two-point true or false
scale where l = “False” and 2 = “True.” The subscales were averaged to create a total
score (M = 1.80, SD = .18, alpha = .70). A square-root transformation was performed on

the total scores because of a negatively skewed distribution.

32

Aggression. The Reactive-Proactive Aggression Questionnaire (RPQ: Raine,
Dodge, Loeber, Gatzki-Kopp, Lynam, & Reynolds et al., 2006) was used to assess
instrumental and reactive aggression. Proactive aggression is very similar to instrumental
aggression in that it is seen as being goal directed, organized, and carried out without
autonomic arousal. In contrast, reactive aggression is thought to be a fear-induced
response characterized by a hostile and/or irritable defensive response to some perceived
or actual provocation. The RPQ contains 23 items, 11 of which assess reactive aggression
(e. g. “How often have you reacted angrily when provoked by others?”) and 12 which
assess proactive aggression (e. g. “How often have you hurt others to win a game?”). The
response scale consists of three points, 1 = “Never,” 2 = “Sometimes,” and 3 = “Often.”
The mean score for reactive aggression was 1.77 (SD = .32; alpha = .77) and the mean
score for proactive aggression was 1.21 (SD = .27; alpha = .84). These two subscales
were highly correlated, r = .59. The distribution of the proactive aggression subscale was
positively skewed and was therefore transformed using an inverse squared
transformation.

Reproductive Strategy. To assess the extent to which individuals display a fast
reproductive strategy (i.e. one characterized by increased effort directed towards mating)
the Revised Sociosexual Orientation Inventory (SOI-R/Sociosexuality, Penke &
Asendorpf, 2008) and the Sexual Experiences Survey (Koss & Oros, 1982) were
administered. The sociosexuality scale consists of 9 items that are summed to form three
subscales: behavioral experiences (e. g. “With how many different partners have you had
sex within the past 12 months?”), attitudes toward uncommitted sex (e. g. “Sex without

love is OK”), and sociosexual desire (e. g. “In everyday life, how often do you have

33

spontaneous fantasies about having sex with someone you have just met?”). Each
subscale uses a different response scale but all scales range from 1 to 9 and are summed
to create a total score (M = 31.21, SD = 14.86, alpha = .86).

The sexual experiences survey assesses the extent to which a person has engaged
in sexually coercive behavior. Given the item content and the low incidence of female
initiated sexually coercive behavior (Greenfeld, 1997), this scale was administered only
to men in the sample. The scale consists of 13 questions that ask participants to report
whether they have committed various acts of sexual coercion, threat, and force (e. g.
“Have you ever had sexual intercourse with a woman even though she didn’t really want
to because you threatened to end your relationship otherwise?”). The first item of the
questionnaire was dropped from analyses because it was not conceptually related to
sexual coercion; the item asked participants whether they have ever engaged in
consensual sex with a woman. A response of “no” to this item would not necessarily
indicate that sexual encounters were non-consensual, but could also mean that the
participant has never had sex. The original response scale requires a dichotomous yes/no
response. Given the population being sampled in this study, it was presumed that low
variability in responses would be an issue. That is, it was expected that very few
participants would report having engaged in the described acts of sexually coercive
behavior. This expectation was confirmed in that the percentage of “no” responses for
each item (excluding item 1) was greater than 95% for all but two items (“Have you ever
had a woman misinterpret the level of sexual intimacy you desired?” and “Have you ever
obtained sexual intercourse by saying things you didn't really mean?”). Additionally, one

item had no variability at all, such that all participants reported not ever having

34

committed the behavior in question, that is “Have you ever been in a situation where you
obtained sexual acts with a woman, such as anal or oral intercourse, when she didn't want
to by using threats or physical force (twisting her arm, holding her down, etc.)?

A second version of the sexual experiences survey was created (Likelihood of
Sexual Coercion Scale; LSC) and administered to participants. This scale fi'amed the
items as asking for the likelihood of engaging in a particular sexually coercive act (e. g.
“How likely is it that you will ever have sexual intercourse with a woman even though
she doesn't really want to because you threaten to end your relationship otherwise?”).
Responses were recorded on a seven-point scale such that 1 = “Very Unlikely,” and 7 =
“Very Likely.” Given the issues with variability and low reliability of the original scale
(alpha = .55), only the revised version of the scale was used in analyses (M = 1.75, SD =
.91 , alpha= .91). The resulting distribution of this scale was positively skewed; an inverse
transformation was performed to normalize the data.

As a final measure of reproductive strategy, female participants were asked to
retrospectively report their age (in years) when they experienced their first menstrual
period, that is, when they experienced menarche (M = 12.70, SD = 1.44). This measure
was included in response to the research discussed previously demonstrating that the
timing of menarche can change as a function of factors in the early rearing environment,
such as father absence and family conflict (Belsky et al., 2007 ; Graber et al., 1995;
Moffitt et al., 1992).

Impulsivity. The Barratt Irnpulsiveness Scale - version 11 (Barratt, 1985; Patton,
Stanford, & Barratt, 1995) was used to assess impulsivity. The measure consists of 30

items and three different dimensions: motor impulsivity (e. g. “I do things without

35

thinking”), attentional impulsivity (e. g. “I don’t pay attention”), and future-planning
impulsivity (e. g. “I am more interested in the present than the future”). Responses were
recorded on a four-point scale ranging fiom l = “Rarely/N ever” to 4 = “Almost
Always/Always.” The items were averaged to create a composite total score (M = 2.14,
SD = .34, alpha = .83).

Risky Behavior. Participants completed two behavioral measures of risk taking,
the Balloon Analogue Risk Task (BART; Lejuez, Read, Kahler, Richards, Ramsey,
Stuart, et al., 2002) and a gambling task based on the Iowa Gambling Task (Bechara,
Damasio, Damasio, & Anderson, 1994). For each'task, participants are instructed that the
goal of the game is to earn as many points as possible. As an incentive for participants to
take the tasks seriously, they were provided with the opportunity to win one raffle ticket
for every 1,000 points earned over the course of both tasks. Tickets were entered into a
drawing to win one of ten $25 gift certificates to Amazon.com.

The BART is a computer-simulated assessment of risk taking behavior. The task
requires that the participant incrementally inflate a simulated balloon on a computer
screen. Each pump of the balloon earns the participant one point that is placed in a
temporary reserve. At any time the participant can choose to stop inflating the balloon
and transfer the points earned to a permanent bank. If, however, the balloon overinflates
(signified by a simulated balloon popping), all of the money in the temporary reserve
accrued for that balloon is lost. The participant repeats this task for 30 trials. The average
breaking point for each balloon was set via a variable ratio schedule, where on average
the balloons would pop at the 64th pump. The average number of balloon pumps for trials

on which the balloon did not pop (aijART) was calculated as the outcome measure

36

(including trials where the balloon popped artificially reduces the average because it is
unknown how many pumps the participant would have reached had the balloon not
popped). For this sample, the average aijART score was 40.76 (SD =14.24).

The Iowa gambling task is a computer simulated assessment of decision making
and sensitivity to probabilities of reward and punishment. In this altered version of the
task, participants are given 2,000 points to begin the task and given 100 trials to
maximize the number of points earned. For each trial participants are asked to select a
card from one of four decks. Each card selected provides the participant with information
about how many points were gained and lost with the selection of that card, and also
provides a running total of the points they have earned (or lost). For example, selecting a
card from Deck D on the first trial might result in the following feedback: Gain = 50,
Loss = 0, Total = 2050. Decks A and B are considered “good decks” because on average
they produce a gain of 250 points every ten cards. Decks C and D are considered “bad
decks” because on average they produce a loss of 250 points every ten cards.

The original Iowa gambling task was designed such that the gain/loss structure
was non-random. That is, two participants selecting the same cards would earn the same
number of points. For the purposes of this study, the gain/loss structure was partially
randomized. For the good decks, participants always gained some number of points
randomly selected between 40 and 60 with the constraint that, on average, the gain must
be equal to 50 points. For bad decks, participants always gained some number of points
randomly selected between 90 and 110, with the constraint that, on average, the gain
must be equal to 100 points. The loss structure varied by deck, such that Deck A yielded

a loss of 250 points on 10% of trials, Deck B Yielded a loss of 50 points on 50% of trials,

37

Deck C yielded a loss of 1250 points on 10% trials, and Deck D yielded a loss of 250
points on 50% of trials. The gain and loss structure of the GT is summarized in Table 1.
The total points accrued (M = 1961.93, SD = 2201.23) and the proportion of good decks
chosen (M = .47, SD = 16.04) were used as outcome measures. However, these two
indices of performance were completely uncorrelated with one another (r = .004, p =
.94), thereby suggesting that performance was not improved via the selection of good
decks. These findings suggest that the introduction of randomization to the task may have
significantly changed the nature of the task. To investigate this issue further, the 100
trials of the task were broken down into ten blocks of ten trials, and the proportion of
good decks chosen in each block was examined. If the task is being administered
properly, the proportion of good decks chosen should increase across blocks. Such a
pattern of findings would indicate that participants were learning to distinguish between
good decks and bad decks. However, in our sample, participants showed little evidence of
learning. For block one good decks were chosen 47.5% of the time; at block ten good
decks were chosen 48.4% of the time. Between these two blocks, the proportion of good
decks chosen fluctuated with a max of 49.2% and a minimum of 43.3% (range = 5.9%).
No consistent pattern of learning was revealed, suggesting that the task was not
measuring the desired construct. In addition, the number of good decks chosen did not
correlate with any of the other variables in the study, and the total score only exhibited a
small correlation with one other variable in the study, reactive aggression (r = .13, p <

.05). In light of this, the task was removed from all analyses.

38

Procedure

Participants signed up to participate in the study via an online participant pool
management system. Participation required attending a lab session and completing an
online follow-up survey. During the lab session, participants completed the PPI-R along
with the BART and the gambling task. Each lab session accommodated up to four
participants, seated at semi-private computer stations. Participants were greeted by an
experimenter who explained the details of the study, provided an informed consent for
each participant, and then had them complete the PPI-R. Upon completion, the
experimenter provided brief instructions for the BART and the gambling task. For each
task, detailed instructions were provided on the computer. The content of the
experimenter and computer instructions is provided in Appendix B.

The order of the tasks alternated across adjacent computer assignments. That is,
participants at the first and third computers completed the BART first, whereas
participants at the second and fourth computers completed the gambling task first. A set
of independent sample t-tests revealed no effect of this order difference for any of the
outcome measures (t ’s = .068 — 1.09). Upon completion of the first task, participants
notified the experimenter so that the experimenter could record their score (in order to
determine the number of raffle tickets earned), and then start the next task for them. After
completion of the second task, the experimenter recorded the number of points earned,
added it to the total from the first game, divided by 1000, and then tore off that number of
raffle tickets for the participant. The number associated with each ticket earned was
entered into a spreadsheet where the ticket numbers could be linked to the participant’s e-

mail address. Participants were then thanked for their participation, reminded to complete

39

the online follow-up survey, and provided with a debriefing. The raffle ticket stubs were
saved in a sealed bucket until the end of the semester when 10 tickets were drawn. The
winners were notified by e-mail so that they could retrieve their gift card.

Upon completion of the lab session, participants were instructed to go online
within the next seven days to a secure website to complete the follow-up survey.
Participants used an anonymous ID created during the lab session to log in to the survey;
this ID was used to match both sets of data. All remaining measures described above that
were not included in the lab session were administered online. Participants were
permitted to skip questions they felt uncomfortable answering. After completing the
survey, participants were directed to a debriefing webpage and provided with contact
information for the researchers.

Assessment of Selection Bias

Given that only 64% of participants completed the follow-up survey, it was
important to assess if participant attrition between the lab session and the online-follow
up survey introduced a selection bias. An independent samples t-test was computed to
compare participants’ scores on the measures completed by all participants (PPI-R,
BART, and the gambling task), as a function of whether or not they completed the online
follow-up survey. No differences were found for FD, the BART, or the gambling task.
However, a near significant difference was found for IA, t(438) = 1.74, p = .083, such
that participants who completed the follow-up survey scored lower on IA (M = 2.28, SD
= .32) than those who did not complete the follow-up survey (M = 2.33, SD = .30).
However, the effect size for this mean difference is small (Cohen’s d = .16). Given that

IA was the only measure for which the mean differences approached significance, the

40

selection bias present in this sample is likely to be small, but should nevertheless be taken

into consideration when interpreting results.

41

Results
Preliminary Analyses

An examination of the intercorrelations among variables (Table 2) revealed a
pattern of associations that was largely consistent with hypotheses. The correlations
between variables used to assess availability of resources in the early rearing environment
were positive in direction and moderate to large in size (r 's = .36 to .48), as was the case
with the variables assessing family relationships in the early rearing environment (r 's =
.44 to .51). However, the correlations across these two domains, that is, correlations
between variables indicating family resources and those indicating family relationships,
exhibited small correlations (r ’s = .00 to .22). This pattern suggests that latent variables
should be constructed separately for each of these constructs.

The PPI and IPIP measures of PD were strongly correlated (r = .65), as were the
PPI, IPIP and SRP measures of IA (r ’s = .67 to .83). In turn, these measures of FD and
IA were associated with a number of the outcome variables, such that IA predicted higher
scores on the measures of impulsivity, aggression, sociosexuality, and lower scores on
educational achievement. In addition, FD was positively associated with self-monitoring,
but not associated with educational achievement—despite the prediction that the two
variables would be positively correlated.

Although gender was not associated with any of the early rearing environment
Variables, it did exhibit a number of relations with the measures of psychopathy and the
outcome measures. These associations, consistent with previous literature, indicated that
men scored higher on PPI IA (d = .59), IPIP IA (d = .79), SRP (d = 1.12), PPI FD (d =

-50), impulsivity (d = .33), proactive aggression (d = .64), reactive aggression (d = .52),

42

and sociosexuality (d = 1.22), but scored lower on GPA (d = -.27). Considering these
differences, the moderating effects of gender will be explored later in the overall model
to determine if there is evidence of differences in the associations among variables for
men and women, in addition to the mean level differences observed here.

Table 3 provides correlational data for LSC (likelihood of sexual coercion)
reports from men and menarche reports from women. LSC was positively associated with
IA, impulsivity, and proactive aggression. Additionally, LSC was negatively associated
with the [PIP measure of FD, GPA, and self-monitoring. Age at menarche was positively
correlated with mother and father education, suggesting that menarche is increasingly
delayed among women whose mothers and fathers are more highly educated. However,
menarche was unrelated to the factors of psychopathy and the family relationship
variables.

Primary Analyses

Structural Equation Modeling (SEM) was used to conduct the primary analyses.
All analyses were performed in AMOS (Analysis of Moment Structures; Arbuckle, 2003)
which uses a maximum likelihood estimation procedure. The issue of missing data was
handled using full information maximum likelihood estimation. This procedure does not
actually irnpute missing values, but instead uses an algorithm that takes into account the
observed data, missing data, the associations among the observed data, and an underlying
statistical assumption of normality in order to calculate estimates of the model
parameters. This procedure was used because it has been shown to produce robust

parameter estimates (MckNight, Mcknight, Sidani, & Figueredo, 2007). For this sample,

43

excluding those participants that did not complete the follow-up survey, less than 1% of
values were missing.

Determination of model fit will be evaluated using the chi-square test statistic
(x2), the comparative fit index (CPI), and the root mean square error of approximation
(RMSEA). Good fit is indicated by a non-significant chi-square test, a CFI of .95 or
higher, and a RMSEA of .05 or lower. Reporting of the RMSEA will include the 90%
confidence interval (CI) along with the test of ‘close fit’ that indicates whether the value
of the RMSEA is significantly different from .05. If the p-value from this test is greater
than .05, or if the confidence interval includes .05, then close fit is indicated. It is
important to note that the chi-square test is sample size dependent, thereby causing
models with relatively minor misfit to be rejected when sample sizes are large (Bentler &
Bonett, 1980). As such, this statistic will primarily be used to compare fit between
models. In addition, the discussion of indicator loadings and path coefficients will be
reported in standardized units because many of the variables included in the model have
undergone a transformation of metric that increases the difficulty with which estimates
can be interpreted. Unstandardized results will be reported in accompanying tables.
Measurement Model

A two-step modeling approach (Anderson & Gerbing, 1988) was used in order to
identify areas of weakness in the model. This approach requires that a confirmatory factor
analysis (CFA) measurement model first be specified and model fit evaluated. If model
fit is poor, the model is re-specified until model fit is good (or the model is rejected). If
good fit is obtained, the structural regression model is then specified. This procedure

allows a researcher to know whether the measurement or structural portion of the model

is the source of poor fit. For this analysis, the measurement model was constructed in a
series of steps. The first step included examining the fit of the measurement model
without the inclusion of the outcome variables. In the second step, each outcome variable
was added in iteratively and run as a new model to determine if the fit of the
measurement model was good when each outcome variable was included.

To determine the fit of the measurement model without the outcome variables,
confirmatory factor analysis (CFA) was used to construct a total of four latent variables
representing early family relationships, family resources, FD, and IA. Indicators of the
relationships latent variable included MRQ, FRQ, and the CPIC. Indicators of the
resources latent variable included mother education, father education, and parental
income. FD was indicated by IPIP FD and PPI FD. The IA latent factor was indicated by
IPIP IA, PPI IA, and the SRP total score. The variance of each latent variable was
constrained to be equal to one (rather than constraining an indicator loading to be equal to
one) in order to provide a standardized scaling metric for the latent variable and so that
the magnitude of the factor loadings could be observed for all indicators. All possible
correlations among latent variables were modeled. Results of the measurement model
(Table 4) indicated good fit, )8 (38) = 76.49, p < .001; CFI = .96; RMSEA = .06 (Cl: .04
to .08; p = .187). Most indicators loaded strongly onto their indicated factors, with
standardized loadings ranging from .62 to .74 for the relationships latent variable, .55 -.75
for the resources latent variable, and .75 to .92 for the IA latent variable. However, the
factor loadings and error variances for the FD latent factor suggested that there was a

specification problem.

45

The model produced a negative variance associated with the residual of the
manifest indicator PPI FD (-2.29) and a standardized loading for PPI FD that exceeded
one (3.94). It is difficult to know the exact cause of this type of specification problem,
and it may simply be that a two-indicator factor is relatively unstable as compared to a
three indicator factor. Given that a third indicator could not be added to the latent factor,
it was necessary to remove one of the indicators of FD. The decision of which indicator
to remove was conceptual, such that the indicator that seemed conceptually most aligned
with the FD construct was maintained for the model. Of primary importance here, was
the independence among the two factors of psychopathy. With this in mind, the indicator
with the strongest correlations with the measures of IA was removed. PPI F D was
significantly associated with all three measure of IA, exhibiting small to moderate
correlations (r’s = .14 to .31), whereas IPIP FD exhibited non-significant to small
correlations with the measures of IA (r’s = .00 to -. 12), as such, PPI FD was dropped
fi'om the model. This revised model in which IPIP FD served as the sole indicator for the
FD latent factor resulted in good overall fit, )6 (30) = 52.52, p = .007, CFI = .97, RMSEA
= .05 (CI: .03 to .07; p = .425), and no negative variances.

Investigation of the correlations within the revised model revealed a positive
association between relationships and resources (r = .21, p = .012), and a non-significant
association between FD and IA. In addition, contrary to predictions, the resources latent
variable was not significantly associated with the two factors of psychopathy. However,
the correlations between the relationships latent variable and the two factors of
psychopathy were in the predicted direction and larger in magnitude (IA = -. 12, p = .086;

FD = .20, p = .004).

46

For the second step of the measurement model, each observed outcome variable
was added to the model in an iterative process, where each outcome variable constituted a
separate model. All possible correlations among the latent factors and the outcome
variable were modeled. The first outcome variable added to the model was impulsivity.
This addition resulted in poor model fit, x2 (36) = 100.76, p < .001; CFI = .94; RMSEA =
.08 (CI: .06 - .10; p = .004). To determine what aspect of the model was fitting poorly,
modification indices were examined.2 This analysis suggested that model fit would be
substantially improved (> 15 x2 units) by correlating the residuals of either of two
manifest indicators of IA (SRP and PPI IA) with the residual variance of the impulsivity
manifest variable. When each of the other outcome variables were added to the model a
similar pattern of modification indices was found for sociosexuality (all other models fit
the data well). This may suggest that there is substantial predictor criterion overlap
among the measurement of IA and the outcome variables impulsivity and sociosexuality.
Given these findings, it seemed important to either model this overlapping variance or
remove it fi'om the model.

In order to maintain a recursive model and to avoid sample-specific solutions, the
decision was made to drop either SRP or PPI IA. In light of the fact that previous
research has raised concerns with respect to issues of predictor criterion overlap using the
SRP (Mahmut, Homewood, & Stevenson, in press; Witt & Donnellan, 2008), it was
dropped from the model. To ensure that the removal of the SRP indicator significantly
improved model fit, a nested model comparison was made between the model where the

IA latent factor was indicated by all three IA measures, and the reduced model where

 

2 AMOS will not provide modification indices if there is missing data. To circumvent this, a new data set
was created that did not contain any missing values. The new data set retained 247 participants and was
only used to examine modification indices.

47

SRP is dropped as an indicator. The chi-square difference test was performed for this
comparison for every model, that is, with every non-gendered outcome variable. Results
indicated that the nested model fit the data significantly better than the original model for
every outcome variable. In addition, the fit indices for each model suggested good fit,
that is, all CFI values exceeded .95 and all RMSEA values were equal to or less than .05.
Nested model comparisons were also run for the gendered outcome variables, LSC and
menarche. Although the male sample with LSC as the outcome variable followed the
same basic pattern as the other models, when the nested model was run for the female
sample with menarche as the outcome, a negative error variance was generated. This
problem occurred regardless of which indicator of IA was dropped from the model, as
such, SRP was kept in the model for this and all future analyses with menarche as the
outcome variable. The path loadings and residual variances for this reduced model, in
which PPI FD and SRP have been dropped as indicators for the latent variables FD and
IA, respectively, are provided in Table 4. Additionally, a full reporting of the nested
model comparisons (comparing the model before and after SRP is dropped as an
indicator of IA) and model fit statistics is provided in Table 5.

When the intercorrelations between the factors were examined for the model with
each outcome variable the pattern of findings remained largely unchanged from the
original model, and echoed the earlier finding that the latent factor for resources exhibited
no significant associations with the PD and IA latent factors. Thus, even though the
measurement model indicates good fit with the inclusion of the resources latent factor, it
may be a non-essential component of the model. Given that these null relationships are

unlikely to change in the structural regression model, the resources latent variable was

48

dropped from analyses. The resulting model, without the inclusion of outcome variables,
was suggestive of very close fit, 752 (7) = 7.69, p = .361; CFI = 1.00; RMSEA = .02 (CI:
.00 to .08; p = .748). When each outcome variable was included sequentially, model fit
was also good, with all CFIs equal to or exceeding .95, RMSEAs less than or equal to
.08, and .05 in every confidence interval for RMSEA. These fit statistics are provided in
Table 6.

The correlations between resources, FD, and IA did not vary much across the
different non- gendered outcome variables and were very similar to the correlations
obtained prior to the changes to the model. The correlations between relationships and IA
ranged from -.11 to -.14 (p’s = .077 to .124), whereas the correlations between
relationships and FD were all equal to .20 (p’s < .01). The relationship between FD and
IA ranged from -.12 to -.13 (p’s = .047 to .062). These correlations were somewhat
stronger when the male sample was examined with LSC as the outcome variable: family
relationships with IA, r = -.28 (p = .007); family relationships with FD, r = .21 (p = .027);
IA with FD, r = -.24 (p = .012). When the model was run for women with menarche as
the outcome variable all correlations were in the predicted direction, but were not
statistically significant. The weakness of these relationships is not surprising given the
small zero-order correlations among menarche and the other variables in the model.

In accordance with predictions, when the correlations between FD, IA, and each
outcome variable were examined (Table 7), the results indicated that the IA latent factor
correlated positively and strongly with impulsivity, aggression, sociosexuality, and LSC,
and was negatively correlated with GPA. In addition, FD was positively related to self-

monitoring and negatively related to LSC. However, contrary to predictions, IA was

49

positively associated with proactive aggression, not associated with performance on the
BART, and FD was not associated with GPA or proactive aggression.

In summary, the analysis of the measurement model indicated multiple areas of
misfit. To address these issues, PPI FD was removed as an indicator of FD and SRP was
removed as an indicator of IA. Although the resulting model fit the data well, it was
evident that the latent factor for resources was unrelated to both factors of psychopathy
and the outcome variables, and was therefore dropped from the model. The final model
exhibited good model fit for all outcome variables. The correlations among the factors in
the model were largely consistent with the overall predictions. FD and IA exhibited only
a small negative correlation with one another, which is consistent with the notion of
independent expression of the two factors. Positive family relationships predicted the
expression of FD, whereas negative family relationships predicted the expression of IA.
These results are consistent with the life history approach to psychopathy in which FD
represents a slow life strategy that is associated with a safe and predictable early rearing
environment, and where IA represents a fast life strategy associated with a harsh and
uncertain early rearing enviromnent. In turn, FD and IA were differentially associated
with the outcome variables that would serve an adaptive function in the corresponding
environment in which the life history strategy arose. Given that this measurement model
fits the data well and corresponds well to predications, the model will now be converted
to a structural regression model in order to complete the second step of the two-step SEM

modeling approach.

50

Structural Regression Models

Using the final measurement model as the guide, a structural regression model
was analyzed for each outcome variable (see Figure 2). Direct paths were drawn from the
relationships latent variable to each factor of psychopathy, and from the psychopathy
factors to the outcome variable of interest. One indicator path for each latent variable was
constrained to be equal to one in order to provide a scaling metric for the latent variable.
Factors with a single indicator (F D and all outcome variables) were modeled as latent
factors, as opposed to manifest variables, with their path loadings constrained to one and
the residual error variance of the manifest indicator constrained to zero. This was done
because, conceptually, a manifest endogenous variable cannot be distinguished from an
indicator of the latent variable fi'om which it is being predicted.

For all models, excluding those for the gendered outcome variables, model fit was
good with CFIs greater than or equal to .96, and RMSEAs ranging from .00 to .06, with
.05 within the bounds of each confidence interval. In addition, with the exception of the
model that included the BART as the outcome variable, all models produced non-
significant chi-square values. A full reporting of model fit statistics for the structural
regression models is provided in Table 6.

An examination of the standardized path coefficients for each model revealed a
pattern that was largely consistent with the measurement model correlations. A statistical
summary of these coefficients is provided in Table 8. For the non-gendered outcome
variables, the standardized path coefficients from the relationships latent factor to FD
were very stable, ranging only from .205 to .206 (p’s < .05). This suggests that more

positive family relationships are associated with an increased expression of FD. The path

51

coefficients fiom the relationships latent factor to IA were small in magnitude (,8 = -.11 to
-.15), and only significant at the p < .05 level for the outcome variable impulsivity.
However, a trend toward significance (p < .10) arose for models with proactive
aggression, the BART, and GPA as outcome variables. These results suggest that,
although the association is small, negative family relationships are associated with the
increased expression of IA.

The associations between the two factors of psychopathy and each outcome
variable were generally in the predicted direction (see Table 8). IA positively predicted
scores on impulsivity (fl = .77, p < .001), reactive aggression (,B = .44, p < .001), and
sociosexuality (B = .46, p < .001). In addition, IA negatively predicted GPA (fl = -.20, p =
.004), and did not predict self-monitoring. IA was also expected to predict performance
on the BART, but this relationship fell short of significance. Interestingly, IA positively
predicted proactive aggression (B = .69, p < .001), which was an unanticipated effect. FD
was expected to positively predict self-monitoring, GPA, and proactive aggression, but
only significantly predicted self-monitoring (,8 = .38, p < .001). Additionally, as
hypothesized, FD was unrelated to reactive aggression, impulsivity, and the BART. A
small and unpredicted positive association arose between FD and sociosexuality (fl = .13,
p = .013), however, this association was not as strong as the predicted association
between IA and sociosexuality.

With respect to the gendered outcome variables, model fit was good for the male
sample with LSC as the outcome variable, x2 (12) = 14.56, p = .266; CFI = .99; RMSEA
= .04 (CI: .00 to .10; p = .543), and acceptable for women with menarche as the outcome

variable, x2 (18) = 33.63, p = .014; CFI = .95; RMSEA = .08 (c1: .03 to .12; p = .128).

52

For the women’s model with menarche, none of the estimated path coefficients were
statistically significant. With respect to the model for men with LSC as the outcome
variable, all path coefficients were in the predicted direction and statistically significant.
The relationships latent factor positively predicted the expression of FD (13 = .23, p
=.021) and negatively predicted the expression of IA (,8 = -.29, p =.009). Additionally, IA
predicted a greater likelihood of endorsing sexually coercive behaviors ()6 = .47, p <
.001), whereas FD predicted a decreased likelihood of endorsing sexually coercive
behaviors (,B = -.18, p =.022).

These findings largely replicate the results obtained fiom the measurement model.
Positive family relationships predict the expression of PD, whereas negative family
relationships predict the expression of IA. In general, the association between family
relationships and FD was stronger than that between family relationships and IA, which
was small and rarely met criteria for statistical significance. As demonstrated in the
measurement model, the associations between the psychopathy factors and the outcome
variables were largely consistent with a life history fi'amework, with a few exceptions. IA
did not predict performance on the BART, which was expected given that the BART is
intended to measure risky behavior. Additionally, IA was positively related to proactive
aggression, which was actually expected to be related to FD given that proactive
aggression is thought to require careful, unernotional, and strategic planning. However no
association between FD and proactive aggression was revealed. These effects may be a
function of the large overlap in content between the reactive and proactive measures of
aggression. FD was also expected to be related to GPA as a function of devoting more

energy to cognitive development, however no association was found. Similarly, FD was

53

expected to be unrelated or negatively related to sociosexuality given that energy is
primarily allocated toward development rather than reproduction, however, FD exhibited
a small positive correlation with sociosexuality. Despite these inconsistencies, many of
the predicted correlations were confirmed and explanations for the exceptions will be
explored in the discussion.

Multiple Groups Analysis

The mean level gender differences that were observed for the psychopathy and
outcome variables during the preliminary analyses raised a concern with respect to
whether gender moderates the relationships exhibited in the model. Although mean level
gender differences are expected, gender differences in process, that is the strength of
associations for each gender, are not predicted. As such, a multiple groups analysis was
performed in order to rule out this possibility.

To conduct a groups analysis the structural regression model was run separately
for men and women with most parameters free to vary3 . This process was then repeated
with the exception that the path loadings for each factor and the path coefficients were
constrained to be equal across groups. A chi-square difference test was then conducted to
determine if the constrained model fit the data significantly worse than the unrestrained
model. This was repeated for every non-gendered outcome variable. Results of these
analyses revealed that the constrained model did not fit the data significantly worse (i.e.
all chi-square difference tests resulted in p—values that exceeded .05). Table 9 provides
the chi-square values for the unconstrained and constrained models, along with the values

for the chi-square difference test. These results suggest that although there are mean level

 

3 The variance of the relationships latent variable was constrained to 1, and one path loading for the IA
latent varrable was constrained to 1. For the one-indicator latent variables (FD and the outcome) the path
loading was constrained to one, and the residual error variance was constrained to zero.

54

gender differences, there are no significant gender differences in the strength of the
associations in the model.
Mediation Analyses

To examine whether the two factors of psychopathy are responsible for the
associations between early family relationships and the outcome variables, a mediation
analysis was performed. In accordance with Baron and Kenny’s (1986) approach to
mediation, it must first be demonstrated that the initial variable predicts the mediating
variable (path a), that the mediating variable predicts the outcome variable (path b), and
that the initial variable predicts the outcome variable (path c). However, some
controversy exists over whether or not path c must necessarily be significant, and
methodologists have argued that when effect sizes are thought to be small, this constraint
should be removed (e. g. Shrout & Bolger, 2002). For this analysis, given that the effect
sizes of the relationships among variables are small, particularly for path a, this pre-
requisite will be dropped. F or most outcome variables, it was predicted that either FD or
IA would mediate the relationship between the latent factor for early relationships and the
outcome variable. Given this, it was necessary to examine the significance of the a and b
paths when only the relevant mediating variable was included in the model. For example,
when testing the extent to which IA mediates the relationship between early relationships
and LSC, FD was dropped from the model in order to ensure that the mediation is
occurring via the anticipated factor of psychopathy. When these adjustments are made,
only three models meet the criteria for mediation. These include the mediation of the
relationship between early relationships and LSC by both FD and IA, and the mediation

of the relationship between early relationships and self-monitoring by FD.

55

Mediation of the relationship between early relationships and LSC via IA was
evaluated by comparing model fit between a mediated model that included both a direct
and indirect effect, and a nested model that included only the indirect effect. If the
addition of the direct effect does not significantly improve model fit then there is
evidence of mediation. For this analysis, the nested model with an indirect effect from
early relationships to LSC via IA, x2 (8) = 4.97, was compared to the mediated model
with both an indirect and direct effect, )8 (7) = 4.73. The chi-square difference test was
non-significant, x2 (1) = .24, p = .624, suggesting that the addition of the direct path from
early relationships to LSC did not significantly improve model fit. The addition of the
mediator, IA, reduced the relationship between early relationships and LSC from -.18 (p
= .073) to -.05 (p = .618). Multiplying the a path (-.28) and b path (.49) generates the
estimate of the indirect effect, -.14. To test the significance of this indirect path, Bayesian
estimation was used within AMOS as an alternative to bootstrapping, which cannot be
used when the sample contains missing data. Similar to bootstrapping, Bayesian
estimation does not presume a normal distribution, and can therefore allow for an
asymmetric distribution. The estimated 95% confidence interval ranged from a lower
bound of -.248 to an upper bound of -.026, and because the confidence interval does not
include zero, this indirect effect is statistically significant.

Mediation of the relationship between early relationships and LSC via FD was
examined next. The nested model with an indirect effect from early relationships to LSC
via FD, 78 (5) = 7.34, was compared to the mediated model with both an indirect and
direct effect, )8 (4) = 5.50. The chi-square difference test was non-significant, x2 (l) =

1.84, p = .175, suggesting that the addition of the direct path from early relationships to

56

LSC did not significantly improve model fit. Addition of the mediator, FD, reduced the
relationship between early relationships and LSC from -.18 (p = .073) to -.13 (p = .174).
Multiplying the a path (.22) and b path (-.24) resulted in an indirect effect of —.05. Using
Bayesian estimation, the estimated 95% confidence interval ranged from a lower bound
of -.l 18 to an upper bound of -.002, thereby suggesting that the indirect effect is
statistically significant.

Finally, mediation of the relationship between early relationships and self-
monitoring by FD was examined. The nested model with an indirect effect from early
relationships to self-monitoring via FD, )6 (5) = 4.57, was compared to the mediated
model, )8 (4) = 4.56. The chi-square difference test was non-significant, x2 (1) = .01, p =
.920, suggesting that the addition of the direct path from early relationships to self-
monitoring did not significantly improve model fit. The addition of the mediator, FD,
reduced the relationship between early relationships and LSC fi'om .076 (p = .281) to
.006 (p = .929). Multiplying the a path (.20) and b path (.37) generates the estimate of the
indirect effect, .07. Using Bayesian estimation, the estimated 95% confidence interval
ranged from a lower bound of .020 to an upper bound of .13], thereby suggesting that the
indirect effect is statistically significant.

These results provide some preliminary but limited evidence for a mediated
model wherein the association between early family relationships and the outcome
variables is mediated via the expression of the two factors of psychopathy. However,
mediation could only be tested for a subset of the models because the association between

early family relationships and IA was generally non-significant. Furthermore, when

57

mediation could be tested, the overall effect that was to be tested for mediation was

generally quite small.

58

Discussion

Overall, these results provide some evidence for the validity of a life history
approach to the two factor model of psychopathy. There was some support for the notion
that aspects of one’s early rearing environment, primarily the quality of family
relationships, is associated with the expression of FD and IA. These associations were in
the direction hypothesized, such that family conflict was associated with the expression
of IA whereas family harmony was associated with the expression of PD. In turn, FD and
IA predicted the expression of a number of outcome variables in accordance with
expectations derived from life history theory. That is, IA was associated with outcome
variables that would be adaptive in an uncertain and unsafe environment in which a fast
life strategy develops, whereas FD was associated with outcomes that would be most
adaptive in an environment characterized by relative certainty and safety that is
associated with the expression of a slow life strategy. There were, however, a number of
inconsistencies with the proposed model that should be addressed.
Strengths and Limitations

The early rearing environment variables that were selected to map onto the
environmental cues that Belsky et al. (1991) argued were important in the development of
different life strategies, held together well as latent factors representing the availability of
resources and the quality of family relationships. However, only family relationships
exhibited significant associations with the expression of the two factors of psychopathy.
This leaves open the question as to the importance of the availability of resources in the
development of life history strategies. Rather than dismiss resources as a relevant

predictor, a number of alternative explanations for the null results should be examined.

59

One potential explanation centers around the measurement of the latent factor. The
indicators that comprised the latent factor included mother and father education and
combined parental income. College aged participants may not be able to provide accurate
estimates of their parents’ income. Additionally, the assumption was made that current
level of parental income and education would correlate strongly with the level of
education and income during the participants’ childhood. That is, mean differences were
expected for income and education between the two time points, but the relative standing
of parents in the sample was expected to stay the same. This assumption may not
accurately represent reality. Had access to parental retrospective reports of these variables
been available, this certainly would have been preferable.

It may also have been beneficial to ask participants to report on more subjective
measures of resource availability as a child, for example, questions concerning the extent
to which parents were able to provide them with what they wanted and not just what they
needed. Given that this sample was comprised entirely of college students whose parents
tended to be highly educated and financially secure (median and mode education and
income was a bachelor’s degree and $70-100,000, respectively), it is possible that there
was insufficient variation in income and education, at least at the lower end of the
distribution, to produce the anticipated effects. As such, it may be the case that resources
only play a role in the development of life strategies when they are below a particular
threshold. Alternatively, the relatively positive relationships among families in the
sample may serve as a buffer against the potential negative effects of resource
deficiencies. In any case, more precise measures of these variables are needed before any

strong arguments can be made for their removal in the model.

60

In addition to the null relationships between resources and the psychopathy
factors, the associations between the quality of early family relationships and
psychopathy were quite small, particularly for the IA dimension of psychopathy. This
result is not particularly surprising given that the expression of psychopathy (and life
history strategies) is certainly multiply determined, thereby reducing the impact of any
one contributing factor. As such, it is possible that the composition of the latent variable
for family relationships did not exhaust all of the possible family variables that may be
relevant cues for the development of different life strategies. In addition, the effects of
early rearing environment relationships on one’s expression of psychopathy should be
expected to be small given the time delay between the presumable cause and the much
later effect. Also, as mentioned with regard to the relationship between resources and the
expression of psychopathy, a restriction in the range of the quality of family
relationships, such that this sample was skewed toward positive relationships, may have
resulted in attenuated associations.

It should also be noted that the latent variables for FD and IA were not easily
specified, which resulted in having to remove an indicator fiom both factors. It is often
difficult to determine the cause of such specification errors, but these issues may reflect
the general confusion among researchers with regard to the definition and structure of
psychopathy. Although purportedly independent constructs, measures of FD and IA do
contain some overlapping content that may contribute to the specification problems.
Additionally, researchers have debated over how many factors form the construct of
psychopathy (Cooke & Michie, 2001) and whether FD is relevant at all for the definition

of psychopathy (Williams & Paulhus, 2004; Williams, Paulhus, & Hare, 2007). Although

61

the two factor structure of psychopathy maps well on to fast and slow life strategies, more
work is needed to verify that these are the only, or at least the primary facets, of
psychopathy.

In contrast to the associations between the early rearing environment variables
and the expression of FD and IA, the associations between the two factors of
psychopathy and the outcome variables were relatively large and were generally
supportive of the hypotheses. IA was positively related to measures of impulsivity,
aggression, and mating effort. These associations can be collectively framed as the output
of a fast life strategy in which an individual, faced with an uncertain and harsh
environment, makes decisions that generate immediate benefits, rather than saving for an
unpredictable future. In contrast, FD was unrelated or negatively related to almost all of
these measures, and instead displayed positive associations with self-monitoring. These
relationships are consistent with the notion of a slow life strategy in which the
environment is sufficiently safe and predictable that an individual need not act on
impulse to obtain immediate rewards. Instead, individuals who adopt this strategy use the
enhanced time provided for cognitive development to their advantage by engaging in
tasks, such as self-monitoring, that require enhanced cognitive effort and skill, and which
can have important benefits in social situations. Although much support was garnered for
a life history approach to understanding the behavioral strategies associated with the two
factors of psychopathy, a number of inconsistencies with the predictions arose.

Contrary to expectations, FD did not demonstrate any significant associations
with the either educational achievement (GPA) or proactive aggression. With respect to

GPA, one potential problem with the use of this variable was the fact that it generates a

62

ceiling for achievement, that is, GPA cannot exceed a 4.0. Given the relatively high mean
for this variable, 3.59, this seems to be a serious problem, such that moderate achievers
cannot readily be distinguished from high and very high achievers. The model would
likely have benefitted from the use of a measure without an upper ceiling, that was less
skewed, and which generated more variability in scores.

The null relationships between FD and proactive aggression prompted a closer
examination of the items comprising the proactive aggression scale. Although the
proactive aggression subscale is purportedly a measure of goal directed aggression, a
number of the items on the scale lack a well-defined goal. For example, the items
“vandalized something for fun,” “had a gang fight to be cool,” and “made obscene phone
calls for fun” lack an obvious or coherent goal that would fulfill an adaptive need.
Additionally, many of the items refer specifically to physical aggression, which may not
be as relevant to a sample of college students as it is to a sample of prison inmates, and
specifically may be less relevant to the expression of FD in which aggressive tendencies
may be more likely to be expressed via verbal or passive aggression, or other generally
manipulative behaviors. Many of the items comprising the proactive aggression scale
actually seem more reflective of reactive aggression than proactive, and may indeed
contribute to the substantial correlation among the two dimensions of the scale (r = .59).
Given this overlap, it is not surprising that IA exhibited a positive correlation with
proactive aggression, as many of the items can be construed as relatively impulsive rather
than goal directed. The overlapping content of the scale dimensions may suggest that the
two constructs are not distinct, however, it seems more likely that it reflects design flaws

of the scale.

63

As a measure of behavioral risk taking, it was surprising that IA did not exhibit
any significant relationships with performance on the BART. In addition, the BART was
also uncorrelated with the BIS-II measure of impulsivity along with reactive and
proactive aggression. Considering the expected overlap among these domains, this is a
surprising finding. It may be the case that this young college sample lacks sufficient
variation on risky behavior. Many of the participants were in their freshman and
sophomore years of college, a time in which the newly gained freedom from one’s
parents may result in more risky behavior among all students, thereby decreasing the
amount of variability in the sample. Interestingly, although Hunt, Hopko, Bare, Lejuez, &
Robinson (2005) found a relationship between psychopathy and performance on the
BART in a college sample, the average age of participants in that sample was
approximately one year older than the participants in this sample, and participants were
recruited via flyers that read “Are you a risk taker?” with the intent of increasing the
variability of riskiness in the sample. It may also be the case that mechanisms underlying
the type of risky behavior that the BART measures are not specifically relevant to IA.
Given that the rewards obtained for each pump on a balloon are quite small (1 point),
sensitivity to reward may not be the driving force behind risky behavior in this task.
Alternatively, watching one’s store of hard earned points disappear after one’s balloon
pops, may be tapping into the domain of sensitivity to punishment. If this is the case, IA
may be unrelated to performance on the BART because the riskiness underlying IA may
have more to do with increased sensitivity to reward than insensitivity to punishment.
This seems to coincide with a fast-life strategy approach to the conceptualization of IA,

in that immediate rewards are valued very highly, even at a cost, because waiting for a

64

reward that can be obtained more safely, may result in no rewards at all. Empirical
support for this notion has come fi'om research demonstrating that the IA facet of
psychopathy is more strongly associated with the Behavioral Activation System (BAS)
than the Behavioral Inhibition System (BIS; Wallace, Malterer, & Newman, 2009).

With respect to the assessment of the relationship between mating effort and the
two factors of psychopathy, results were slightly mixed. Although IA was associated with
higher scores on sociosexuality and a greater likelihood of endorsing sexually coercive
acts among men, neither 1A nor FD were related to age at menarche. However, the
associations were in the predicted direction such that IA was related to an earlier age at
menarche and FD was related to an older age at menarche. These null relationships may
simply reflect the fact that the effect size for the relationship with a biologically oriented
outcome variable, generally perceived to be resistant to external influence, is likely to be
quite small. In addition, the measurement of the variable was relatively imprecise.
Ideally, female participants would have recalled both the month and year of their first
period, but this level of recall is unlikely to yield accurate information. Koo and Rohan
(1997) found that after a lapse of only an average of 430 days, only 60% of participants
accurately recalled the month and year of their first period. Accurate recall also declined
as the interval increased. Thus, considering that the average delay in recall for this sample
is 6.8 years, the estimates are likely far fiom perfect. Asking participants to report
whether they were early, normal, or late when starting their period leads to more accurate
recall (Cooper et al., 2006), but necessarily reduces variance. As such, it may be
necessary to use longitudinal designs that begin prior to the onset of puberty in females in

order to adequately test this hypothesis.

65

In contrast to IA, FD was expected to be unrelated or negatively related to mating
effort variables; however FD was positively associated with sociosexuality. Although this
relationship was small and not as strong as that between IA and sociosexuality, it was
statistically significant. A closer examination of the relationship revealed that it was only
the behavioral facet of sociosexuality that correlated with FD, specifically two of the
three items, one assessing the number of sexual partners in the past 12 months, and the
other assessing the number of sexual partners in which there was no interest in a long
term committed relationship. Although these items indicate that individuals high on FD
are likely to have more sexual partners, they may not necessarily indicate that these
individuals are expending greater effort to obtain these partners. It may be the case that
the charming, dominant, and manipulative style associated with FD lends itself to
increased access to mating opportunities without necessarily exerting much effort. This
explanation would also shed some light on the negative relationship that was observed
between FD and the endorsement of engaging in sexually coercive acts. If individuals
high on FD have a predisposition for charming interpersonal interactions then it would be
unnecessary to engage in sexually coercive behavior.

With respect to the overall pattern of relationships in the model, fit was generally
good. Although models are not intended to produce exact representations of reality, good
fit of a model does indicate a general similarity among the patterns of relationships
generated by the model and those observed in the data. In addition, the model generalized
across gender, which is important given that there was no a priori reason to expect gender
differences in the relationships among variables. There was also some limited evidence

for mediation in the model, suggesting that psychopathy partially mediates the

66

relationship between early rearing environment relationships and later outcomes in life.
However, only a few of the models met the conditions to test for mediation. In light of
the fact that the a path in most of the models failed to reach significance, it seems
possible that this analysis could be extended to the other outcome variables if the
measurement of early rearing environment and psychopathy variables was improved. It
should be noted, however, that because all measures involved in the mediation analysis
were measured at the same time point, evidence for causality is quite weak. Attempts
were made to set temporal precedence by having participants retrospectively report
information from their childhood, and by using relatively stable personality variables as
mediators, however, an ideal model of mediation would examine the expression of the
variables via a longitudinal design.
Conclusions and Future Directions

Although there is much room for improvement in the model, as a first attempt at
using evolutionary theory to establish a developmental trajectory for the two factor model
of psychopathy, the results are promising. It is certainly worth exploring the applicability
of the model in more diverse samples, where greater variability in both early rearing
environment variables and outcome variables may generate stronger relationships among
the variables. Additionally, replications and extensions of the model should seek out
measures that address the concerns raised here, such as restriction of range (6. g. GPA),
ability of participants to provide accurate self report and/or retrospective report of
variables (e. g. age at menarche) and overlap of content between scales (e. g. reactive and

proactive aggression).

67

As it stands, these results provide some support for two relatively distinct
pathways to the expression of the two factors of psychopathy. Although previous
research has examined psychopathy from an evolutionary perspective, little thought has
been given to the possibility that the two factors of psychopathy represent distinct life
strategies. This likely reflects the fact that only recently have researchers begun to treat
the two factors as independent constructs. Such an approach has important ramifications
- for how the etiology of psychopathy is to be understood, and how such disorders are to be
treated. At the least, this research demonstrates that the conception of a “successful” life
strategy requires an understanding of the specific environmental context in which that
strategy is being employed. Although risky and impulsive behavior stray far from a lay
definition of adaptive or successful, if these strategies are employed in unsafe and
uncertain environments then such a strategy is likely to be the most well suited for
survival. In contrast, a risky and impulsive strategy is generally not adaptive in a safe and
predictable environment, but a slow life strategy characterized by long-terrn planning for
future rewards is well suited for survival in such an environment. This research also
highlights the importance of examining the two factors of psychopathy as independent
constructs given their remarkably different associations with a variety of outcomes, as

well as the potential for distinct etiological pathways.

68

APPENDICES

69

APPENDIX A

Tables and Figures

70

Table I

Gain and Loss Structure for the Gambling Task

 

 

Deck Average Points Average Gain Probability Loss
(10 trials) (1 trial) of a Loss Amount
A 250 50 10% 250
B 250 50 50% 50
C -250 100 10% 1250
D -250 100 50% 250

 

71

mo.vm...

63282“ :23» $32.35 go mos—S, BBB-«28.: 05 mam: “.08—=28 05>» 28:22.80 28 222833 33.88 .2802 .202-

 

Nfl. 3.. 3.3 o~.3 NM. mm. x. on. we. cm. 3. am. :0. E. 2. SN 8; on; om. Qu-
mw. cod 3. _ m Snow EL mm; 3 .N nmfi 2N Sam wN.N N06 :4 R. S. 00.5 mné Nmé wv. 53:
~ co. 2. No.- No. mar 3r no. no: .wm. No.- .5. vo- No. _ _. vo- :.- g _.- 5.- Em-M .m—
_ .m _ .- wo- .w _ .- .3... .mm: .m _.- .2.- mo. .2.- mo: :- o_. co. No. oo. wo- .m~.- <m0 .w_

3 8. .8. .8. .8 .8. .8. 8. .8. .8. :. 8. 8. .3. 8. 8. .8. 2-52:
2 8.- 8.- 8. 2. 2. 8. 8. 8. 8. 8.- .2.- .2. 8. 8. .2. bad .2

_ .8 .8. .:V. .8. 8.- ._ m. 2 3.- 8.- 8. 2. 8.- 8. 8.- .8. 8.32 .2

_ .3. .8. .8. 8.- .8. 8. 8.- 8. 8.- 8.- 8. 8. .2. 84.3 .3

_ Ln. .v0. 3.. no. we. .m—.- 3.- :6 no. no.- nor 2. =-m_m.m~

_ .8. 8. .8. .8. 8.- 8.- 2.- 8.- 8.- 8.- .3. 54222
_ .2.- .E. .3. 2.- 8.- 8.- 8.- 8. 8.- .2.. 5.82.:

2 8.- .8. .3. 8. .2. 8. 8. 8. 8.- 8.28.2

_ .2. .2.- 8.- .2.- 8. 8. 8. .8. 5.73.8

_ .3 8. 8. 8. 8. 8. 8. 9:3 .8

 

_ .8. .8. .8. .2. 8. 8. can:
2 .8. .2 .2. 8. 8. 93.8

_ 8. 8. 8. 8. 9.38

_ .8. .8. 8. 0885 .4

2 .2.. :. 8mm .m

_ 8. .8828

_ c8802

2 2 2 2 2 3 2 2 : 2 a w e 8 n v m N _ 288$

 

833232 2:830 38 aSoEo-i nae-22332280 35» 8:23.2st @3355. .2282

N 2an

72

Table 3

Means, Standard Deviations, and Correlations for Gendered Variables

 

 

Likelihood of
Variable Sexual Menarche
Coercion
Resources
1. Mother Education .11 .24“
2. Father Education .10 .19‘
3. Parental Income .12 -.01
Family
4. Mother Relationship Quality -.16 .01
5. Father Relationship Quality -.07 .03
6. Children’s Perception of Interparental Conflict .15 .01
Psychopathy
7. PPI Fearless Dominance -.07 -.02
8. PPI Impulsive Antisociality .39" .03
9. IPIP Fearless Dominance -.26" .06
10. IPIP Impulsive Antisociality .41 ” -.03
11. Self-Report Psychopathy Scale .33" -.O6
Outcomes
12. Barratt Impulsiveness Scale .31M .01
13. Proactive Aggression .39M -.08
14. Reactive Aggression .17 -.10
15. Adjusted BART Score -.05 -.16
l 6. Sociosexuality Inventory-Revised . 13 -.O7
17. High School GPA -.18‘ .07
18. Self-Monitoring Scale-Revised -.26M -.11
Mean 1.32 12.70
SD .22 1.44

 

Note. Means, standard deviations, and correlations in the table were calculated using the
transformed values of variables when available.

*p<.05,**p<.01

73

8.va.

602882 83 oBatg 8:22 moo-588 05 20:3 $38 05 9 2&8 .258 Ram of. 823%.: mm vogue-6

803 Q.”— Hmm fig aw £253 5 ~08?— 05 OH ”898 ~0UOE 002:5.— OFH. .0—DQE> afloau— (J 08 MOM muoawom—ufim 00:: U5 0-DQE> «:32 e 05 how

80822: 25 58> 388 05 8 2&2 6608 859.8 2:- 23:26 583 mos—2, 38885.: was»: 3.2328 0203 032 05 E mamaumm $82

 

 

8. 2: -- 8. 82 -- 28.8 on. 88 8. 2. 888.com ease-.3 22

-- - - -- -- - E: c 88- Ed 82 88 855.89 3.2.8.3 :8

a: 2. c 3 8. .8 83 2. 83 8. .8. .23 8. A88 8. .8. 32885. 22.325 22

28.8 8. 28.8 8. . _ 8. A88 8. 28.8 8. .8. .83 8. 28.8 8. .8. 23885.. 2.382 r:

- - -- - - -- .23 8. 28.8 2.. .8. 2.258228 cocoa-row
3:85:08.—

8580 358985

.23 :. 28.8 8. .8. .23 2. A88 8. .8. .23 2. A88 8. .8. 8 8888.2 .8225

.83 8. :3 2. .2. .83 8. 28.8 2. . _ 2.. .83 8. A 8.8 :. .8. 8:30 8282.23— .652

.83 8. :3 8. .8. .83 8. 83 8. .2. .83 8. A 8.8 8. .8. 2.30 8382.233 .2822
$5..—

- -- -- .83 88 23.8 2.. .8. .88 88 23.8 22 .8. .885 3:28

-- -- - .38 82 a: 82 .8. 28.8 82 a: 82 .8. 888:3 358

- - - .288 :2 a: 8.. .8. .28 o: a: 8._ .8. 8288 see:
woo-BOmOM

$2 $2 22
85.52 28 c n 8583 a. c n oo§E> 22 c n .2885
Eco: :88 .352 8882 252 :8ch

 

“echo: Emfimémcm: EEK ~23 .hwcznmm .3535 m5 xc\mmo=u.:§ 333.6% ~23 nMEhQQ-N 228k

v 033-

74

Table 5

Measurement Model: Nested Comparisons and Model Fit

 

 

Comparison Nested Difference
Model Model Test
Outcome {(36) x207) 35(9) C“ RMSEA (CI)
Barratt Impulsiveness 100.76 38.41 62.35“ .98 .04 (.00 - .07)
Proactive Aggression 58.60 35.66 22.94" .99 .03 (.00 - .06)
Reactive Aggression 64.27 43.41 20.86* .97 .05 (.02 - .07)
Adjusted BART Score 67.41 48.61 18.80* .96 .05 (.03 - .08)
Sociosexuality Inventory 79.78 41.42 38.36" .97 .04 (.01 - .07)
High School GPA 56.31 36.37 19.94* .98 .04 (.00 - .06)
Self-Monitoring Scale 62.67 39.89 22.69" .98 .04 (.00 - .07)
Sexual Coerciona 39.24 22.74 16.501' 1.00 .00 (.00 - .05)

 

Note. The comparison model refers to the measurement model with one indicator for FD

and three indicators for IA. The nested model refers to the model in which SRP has been

dropped from the comparison model. CFI and RMSEA values are for the nested model.

Menarche was not included in the table because SRP was not dropped as an indicator for

that model.

3 Indicates that a gendered sample was used for analyses; LSC was men only.

Tp<.06,*p<.05,**p<.01

75

Table 6

Final Measurement Model and Structural Regression Model Fit Indices

 

 

 

 

Measurement Model Structural Regression Model
Outcome x2 (10) CFI RMSEA (on x2 (12) CFI RMSEA (CI)
Barratt Impulsiveness 7.91 1.00 .00 (.00 - .05) 10.28 1.00 .00 (.00 - .05)
Proactive Aggression 9.08 1.00 .00 (.00 - .06) 12.43 1.00 .01 (.00 - .06)
Reactive Aggression 15.04 .99 .04 (.00 - .08) 22.08 .98 .06 (.01 - .09)
Adjusted BART Score 19.75 .97 .06 (.02 - .10) 24.58 .96 .06 (.03 - .10)
Sociosexuality Inventory 13.17 .99 .03 (.00 - .08) 17.56 .99 .04 (.00 — .08)
High School GPA 8.81 1.00 .00 (.00 - .06) 13.16 1.00 .02 (.00 - .07)
Self-Monitoring Scale 9.35 1.00 .00 (.00 - .06) 11.15 1.00 .00 (.00 - .06)
Sexual Coerciona 10.70 1.00 .02 (.00 - .10) 14.56 .99 .04 (.00 - .10)
Menarchea 32.67 .95 .08 (.04 - .13) 33.63 .95 .08 (.03 - .12)

 

Note. a Indicates that a gendered sample was used for analyses; LSC was men only, Menarche was women

only. Degrees of freedom for the model with menarche as the outcome variable are 16 for the measurement

model and 18 for the structural regression model because SRP was not dropped as an indicator.

76

Table 7

Structural Regression Model Factor Correlations and Covariances (SE) with Outcome Variables

 

 

 

Fearless Impulsive
Relationships Dominance Antisociality

Outcome r Cov (SE) r Cov (SE) r Cov (SE)

Barratt Impulsiveness -.13 -.04 (.02) -.05 -.01 (.01) .76" .26 (.02)
Proactive Aggression -.04 -.01 (.02) -.07 -.01 (.01) .69" .16 (.01)
Reactive Aggression .08 .03 (.02) -.06 -.01 (.01) .44“ .14 (.02)
Adjusted BART Score -.12 -l.76 (.99) .02 .18 (.47) .11 1.52 (.93)
Sociosexuality Inventory .06 .85 (1.06) .08 .69 (.50) .44" 6.57 (.94)
High School GPA .13 .05 (.02) .03 .01 (.01) -.2o" -.07 (.02)
Self-Monitoring Scale .08 .01 (.01) .38" .03 (.00) -.05 -.01 (.01)
Sexual Comma -. 19‘ —.04 (.02) -.27" -.03 (.01) .51" .11 (.02)
Mmhea .03 .05 (.15) .06 .05 (.07) -.02 -.03 (.24)

 

Note. a Indicates that a gendered sample was used for analyses; LSC was men only, menarche

was women only.

*p<.05,“p<.01

77

~o.VQ**.mo.VQ*

3:0 :08?» we? one-85.2 5:0 :08 83 Umq $83.28 com com: me? 038% 353% a 35 8339.: e .202

 

 

 

:S 2. 8. 33 :.- 3.- GS 22 2. $5 3.- 2.- senses:
9.3 3.- .2.- G3 .8. :2. $3 :2 .8. a: @3- :emc .. 8658 333
:3 we. :2. :3 8. S. Ami 22 :z. 33 2.- m 2 .- 28m greases-rem
$3 8. 8. $3 2.- :8.- QE 2? :5. 63 Ma.- 2.- $6 62% emf
61:8.»- .2. Ge: :2: :8. a3 :2 :5. a: «K.- 2; bangs bzeiemeaem
an: mm. 8. and N3 :. $2 :2 :5. a: $.- 2.- 28m 52m 333.
$3 8. 3.- $3 mm. :3. a3 22 :a. 83 3.- :.- seesaw-«s Essex
33 co. 8. 33 R. :23. a: 3.2 :5. 33 e5- m 2 .- Sesame. 2,885
$3 No. 8. $3 2.. :R. ai 5.2 .-_N. 5: a.- .2.- 3802233 gee-am
ma» e n «.3 e n «mu-we e. «a e m eeeeso
080330 AI Dn— DEOOSO A1 (A Om 4. mQEmcotfloM S $- mmmsmcomuflom

 

flammogmeb fink 33: 283.33% 3erth

w 2an

78

Table 9

Multiple Groups Analysis for Gender: Nested Model Comparisons

 

Unconstrained Constrained Difference Difference

 

Model Model Test Test
Outcome )6 (24) x2 (32) x2 (8) p value
Barratt Impulsiveness Scale 32.75 38.54 5.79 .671
Proactive Aggression 37.60 45.23 7.63 .470
Reactive Aggression 42.90 52.74 9.84 .276
Adjusted BART Score 44.21 52.18 7.97 .436
Sociosexuality Inventory 40.24 46.55 6.3 1 .61 3
High School GPA 31.17 38.37 7.20 .515
Self-Monitoring Scale 29.85 35.87 6.02 .645

 

Note. Unconstrained model refers to the model in which most indicator loadings and

path coefficients were freely estimated; the constrained model refers to the model in

which these parameters were constrained to be equal across gender.

79

 

+ P ti A +
Relationships # 01:20“? gg( )

 

Self-Monitoring

 

 

Reactive A g (+)
Impulsivity (+)
Risky Behavior (+)
GPA (-)

Mating Effort (+)

l/-

Resources

 

 

 

 

Figure 1. Path Model for the Associations between Early Rearing Environment,

Psychopathy Factors, and Outcome Variables

80

.8523.» 392%: 53898 8:: 3202 05 use dozen-"32¢ u u vote 3358 u o .ocswc 2: 5 $82

$33.82 2:830 238 .EcBuk ASS-395$ .nmcxacwmx 2333 30.238qu mi .5333: 222,-?me 325.93% .N 953K

.n H

 

 

 

088:5

 

 

 

 

388:0

 

S Em

Senna

 

 

 

 

 

3390:23—

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

81

APPENDIX B

Experimental Instructions for Decision Making Games

82

Experimenter Instructions:

You are now going to play two decision making games. For each game you will try to
win as many points as possible. I will record your performance after each game. For
every 1000 points you win, you will earn one raffle ticket. At the end of the semester we
will draw ten raffle tickets, and the winners will receive a $25 gift certificate to Amazon.
So, the more points you earn, the better your chances of winning one of the $25 gift
certificates. After you have finished each game, raise your hand to get my attention, and I
will start the next game for you. Instructions for each game will be provided on the

computer.

Gambling Task Instructions:
You will see four decks displayed on the screen. Each deck has both rewards and
penalties associated with it. Choose a deck by pressing its key and you will be either
rewarded or penalized.
[. . .on the following screen]

Win: 50

Lose: 50

Total: 2150

When you make a choice, you will be shown how many points you won and possibly lost
by choosing that deck (see above).You will also be shown a running total. You start with
2000 total points. You can choose any deck at any time and switch decks at any time. The
goal is to earn as many points as possible. Remember that the more points you earn the

better your chances of winning the raffle!

83

BART Instructiog

Welcome to the Balloon Game!

In this game you will see some pictures of balloons. Your goal is to pump up each
balloon until it is as big as possible without popping the balloon. Every time you pump
the balloon you get 1 point. But be careful!! If the balloon pops, you will lose all your
points for that balloon. If you stop pumping before the balloon pops then your points will
be saved. It is up to you to decide how much to pump up each balloon. Some of the
balloons might pop afier just one pump. Others might not pop until they fill the whole
screen. You can press the LEFT arrow key to pump up a balloon and make it bigger.
When you want to stop pumping, press the RIGHT arrow key to save your points.
Remember, the more points you earn, the better our chances of winning the raffle.

[. . .on the following screen]

Get Ready to play! Remember: You will have 30 balloons to inflate. Press the LEFT
arrow to PUMP up the balloon for more points. Each pump is worth 1 point. If the
balloon pops you lose all the points for that balloon. Press the RIGHT arrow to stop and
SAVE your points. The more points you earn the better your chances of winning the

raffle! Please notify the experimenter if you have any questions.

84

REFERENCES

Arbuckle, J.L. (2003). Amos 17 [Computer Software]. Chicago: SmallWaters.

American Psychiatric Association. (1994). Diagnostic and statistical manual of mental
disorders (4‘h ed.). Washington, DC: Author.

Anderson, J.C., & Gerbing, D.W. (1988). Structural equation modeling in practice: A
review and recommended two-step approach. Psychological Bulletin, 103, 411 —
423.

Barratt, E. S. (1985). Impulsiveness subtraits: Arousal and information processing. In J.
T. Spence and C. E. Izard (Eds), Motivation, Emotion and Personality (pp. 137-
146). Elsevier Science, North Holland.

Belsky, 1., Steinberg, L., & Draper, P. (1991). Childhood experience, interpersonal
development, and reproductive strategy: An evolutionary theory of socialization.
Child Development, 62, 647-670.

Belsky, J., Steinberg, L., Houts, R.M., Friedman, S.L., DeHart, G., Cauffman, E., et al.
(2007). Family rearing antecedents of pubertal timing. Child Development, 78(4),
1302-3121.

Bechara, A., Damasio, A.R., Damasio, H., & Anderson, SW. (1994). Insensitivity to
future consequences following damage to human prefi'ontal cortex. Cognition,
50(1-3), 7-15.

Benning, S.D., Patrick, C.J., Blonigen, D.M., Hicks, B.M., & Iacono, W.G. (2005).
Estimating facets of psychopathy from normal personality traits: A step toward
community-epidemiological investigations. Assessment, 12, 3-18.

Benning, S.D., Patrick, C.J., Hicks, B.M., Blonigen, D.M., & Krueger, RF. (2003).
Factor structure of the Psychopathic Personality Inventory: Validity and
implications for clinical assessment. Psychological Assessment, 15, 340-350.

Benning, S.D., Patrick, C.J., & Iacono, W.G. (2005). Psychopathy, startle blink
modulation, and electrodermal reactivity in twin men. Psychophysiology, 42, 753-
762.

Bentler, P.M. & Bonett, D.G. (1980). Significance tests and goodness of fit in the
analysis of covariance structures. Psychological Bulletin, 88(3), 588-606.

Blair, R.J.R. (2007). The amygdala and ventromedial prefrontal cortex in morality and
psychopathy. Trends in Cognitive Sciences, 11(9), 387-392.

85

Blair, 1. (2008). Empathic dysfunction in psychopathy, In C. Sharp, P. F onagy, I.
Goodyer (Eds.), Social cognition and developmental psychopathology (pp. 1 75-
197). New York: Oxford University Press.

Blair, R.J.R., Mitchell, D., & Blair, K. (2005). The psychopath. Oxford, Blackwell.

Blonigen, D.M. et al. (2005). Psychopathic personality traits: Heritability and genetic
overlap with internalizing and externalizing psychopathology. Psychological
Medicine, 35, 637-648.

Brim, O.G., Baltes, PB, Bumpass, L.L., Cleary, P.D., Featherman, D.L., Hazzard, W.R.,
et al. (2000). National survey of midlife development in the United States
(MIDUS), 1995-1996. Ann Arbor, MI: DataStat, Inc/Boston, MA: Harvard
Medical School, Dept. of Health Care Policy, 1996. Ann Arbor, MI: Inter-
university Consortium for Political and Social Research, 2000.

Buss, D.M. (1991). Evolutionary personality psychology. Annual Review of Psychology.
42, 459-491.

Buss, D.M. (2009). How can evolutionary psychology successfully explain personality
and individual differences? Perspectives on Psychological Science, 4(4), 359-366.

Cleckley, H. (1941). The mask of sanity (lst edn). St. Louis, Mosby Medical Library.
Cohen, J . (2002). A power primer. Psychological Bulletin, 112(1), 155-159.

Cooke, D.J., & Michie, C. (2001). Refining the construct of psychopathy: Towards a
hierarchical model. Psychological Assessment, 13, 171-188.

Cooper, R., Blell, M., Hardy, R., Black, S., Pollard, T.M., Wadsworth, M.E.J, Pearce,
M.S., & Kuh, D. (2006). Validity of age at menarche self-reported in adulthood.
Journal of Epidemiological Community Health, 60, 993—997.

Cornell, D.G., Warren, 1., Hawk, G., Stafford, E., Orarn, G., & Pine, D. (1996).
Psychopathy in instrumental and reactive violent offenders. Journal of Consulting
and Clinical Psychology, 64, 783-790.

Damasio, AR. (1996). The somatic marker hypothesis and the possible functions of the
prefrontal cortex. Philosophical Transactions: Biological Sciences, 351, 1412-
1420.

Edens, J.F., Marcus, D.K., Lilienfeld, S.O., & Poythress, N.G. (2006). Psychopathic, not

psychopath: Taxometric evidence for the dimensional structure of psychopathy.
Journal of A bnormal Psychology, 1 15(1), 13 l -144.

86

Eliasz, A. (1987). Temperament-contingent cognitive orientation toward various aspects
of reality . In J. Strelau & H.J. Eysenck (Eds), Personality dimensions and
arousal. Plenum.

Eliasz, H., & Reykowski, J. (1986). Reactivity and empathic control of aggression. In J.
Strelau, F.H. Farley & A. Gale (Eds), The biological bases of personality and
behavior: Psychophysiology, performance, and application. Hemisphere.

Ellis, 8.1., Bates, J .E., Dodge, K.A., Fergusson, D.M., Horwood, L.J., Pettit, G.S. et al.
(2003). Does father absence place daughters at special risk for early sexual
activity and teenage pregnancy? Child Development, 74(3), 801 -821.

Eysenck, H.J., & Gudjonsson, G. (1989). The causes and cures of criminality. Plenum.

Figueredo, A.J., Sefcek, J .A., Vasquez, G., Brumbach, B.H., King, J .E., & Jacobs, W.J.
(2005). Evolutionary personality psychology. In D.M. Buss (Ed.), The handbook
of evolutionary psychology (pp.851-877). New York: Wiley.

Fowles, D.C., & Dindo, L. (2006). A dual-deficit model of psychopathy. In C.J. Patrick
(Ed.), Handbook of Psychopathy. New York: Guilford Press.

Fowles, D.C. & Dindo, L. (2009). Temperament and psychopathy: A dual-pathway
model. Current Directions in Psychological Science, 18(3), 179-183.

Gartner, R. (1990). The victims of homicide: A temporal and cross-national review.
American Sociological Review, 55, 92-106.

Graber, J.A., Brooks-Gunn, 1., & Warren, MP. (1995). The antecedents of menarcheal
age: Heredity, family environment, and stressful life events. Child Development,
66, 346-359.

Greenfeld, LA. (1997). Sex offenses and offenders: An analysis of data on rape and
sexual assault. US. Department of Justice: Bureau of Justice Statistics.

Grych, J. H., Seid, M., & Fincham, F. D. (1992). Assessing marital conflict from the
child’s perspective: The Children’s Perception of Interparental Conflict Scale.
Child Development, 63, 558—572.

Hall, J .R., & Benning, SD. (2006). The “successful” psychopath: Adaptive and
subclinical manifestations of psychopathy in the general population. In C.J .
Patrick (Ed.), Handbook of Psychopathy. New York: Guilford Press.

Hare, RD. (1984). Performance of psychopaths on cognitive tasks related to frontal lobe
function. Journal of Abnormal Psychology, 93, 133-140.

87

Hare, R.D. (1991b). The Self-report psychopathy scale, version 2. (SRP-II). Unpublished
manuscript, University of British Columbia, Vancouver, Canada.

Hare, RD. (1991). The Hare Psychopathy Checklist—Revised. Toronto, ON, Canada:
Multi-Health Systems.

Hare, RD. (2003). The Hare Psychopathy Checklist—Revised, 2““| edition. Toronto, ON,
Canada: Multi-Health Systems.

Hare, RD. (1996). Psychopathy and antisocial personality disorder: A case of diagnostic
confusion. Psychiatric Times, 13, 39-40.

Hare, R.D., & Hart, SD. (1995). Commentary on antisocial personality disorder: The
DSM-IV field trial. In W.J. Livesley (Ed.), The DSM-I V personality disorders
(pp.127-134). New York: Guilford Press.

Hare, R.D., & Neumann, CS. (2006). The PCL-R assessment of psychopathy:
Development, structural properties, and new directions. In C.J. Patrick (Ed.),
Handbook of Psychopathy. New York: Guilford Press.

Harpending, H.C., & Sobus, J. (1987). Sociopathy as an adaptation. Ethology and
Sociobiology, 8, 63S-72S.

Harris, G.T., Rice, M.E., Hilton, Z., Lalumiere, M.L., & Quinsey, V.L. (2004).
Psychopathic sexuality: Implications for the assessment and understanding of
psychopathy. Paper presented at the annual conference of the Association for the
Treatment of Sexual Abusers. Albuquerque, NM.

Harris, G.T., Rice, M.E., Hilton, Z., Lalumiere, M.L., & Quinsey, V.L. (2007). Coercive
and precocious sexuality as a fundamental aspect of psychopathy. Journal of
Personality Disorders, 21(1), 1-27.

Harris, G.T., Rice, M.E., & Lalumiere, M. (2001). Criminal Violence: The roles of
psychopathy, neurodevelopmental insults, and antisocial parenting. Criminal
Justice and Behavior, 28(4), 402-426.

Hart, S.D., Forth, A.E., & Hare, RD. (1990). Performance of criminal psychopaths on
selected neuropsychological tests. Journal of Abnormal Psychology, 99, 374-3 79.

Hart, S.D., & Hare, R.D., (1997). Psychopathy: Assessment and association with criminal
conduct. In D.M. Stoff, J . Breiling, & J.D. Maser (Eds), Handbook of antisocial
behavior (pp. 22-35). New York: John Wiley.

Hawley, RH. (1999). The ontogenesis of social dominance: A strategy-based
evolutionary perspective. Developmental Review, 19(1), 97-132.

88

Hicks, B.M., Markon, K.E., Patrick, C.J., Krueger, R.F., & Newman, JP. (2004).
Identifying psychopathy subtypes on the basis of personality structure.
Psychological Assessment, 16, 276-288.

Hunt, M.K., Hopko, D.R., Bare, R., Lejuez, C.W., & Robinson, E.V. (2005). Construct
validity of the Balloon Analogue Risk Task (BART): Associations with
psychopathy and impulsivity. Assessment, 12(4), 416-428.

Johnson, J. A. (2000). Developing a short form of the IPIP-NEO: A report to HGW
Consulting. Unpublished manuscript. Department of Psychology, University of
Pennsylvania, DuBois, PA.

Jonason, P.K., Li, N.P., Webster, G.D., & Schmitt, DP. (2009). The dark triad:
Facilitating a short-term mating strategy in men. European Journal of
Personality, 23(1), 5-18.

Kaplan, H.S. & Gangestad, S.W., (2005). Life history theory and evolutionary
psychology. In D.M. Buss (Ed.), The handbook of evolutionary psychology
(pp.68-96). New York: Wiley.

Kline, RB. (2005). Principles and Practice of structural equation modeling (2"d edition).
New York, NY: Guilford Press.

Knight, R.A., & Guay, J -P. (2006). The role of psychopathy in sexual coercion against
women. In C.J. Patrick (Ed), Handbook of Psychopathy. New York: Guilford
Press.

Koo, M.M., & Rohan, TE. (1997). Accuracy of short-term recall of age at menarche.
Annals of Human Biology, 24, 61—4.

Koss, M.P., & Oros, C.J. (1982). Sexual experiences survey: A research instrument
investigating sexual aggression and victimization. Journal of Consulting and
Clinical Psychology, 50(3), 455-457.

Kosson, D.S., Kelly, J .C., & White, J .W. (1997). Psychopathy-related traits predict self-
reported sexual aggression among college men. Journal of Interpersonal
Violence, 12, 241-254.

Laakso, M.P., Gunning-Dixon, F., Vaurio, O., Repo-Tiihonen, E., Soininen, H., &
Tiihonen, J ., (2002). Prefrontal volumes in habitually violent subjects with
antisocial personality disorder and type 2 alcoholism. Psychiatry Research:
Neuroimaging, 114, 95-102.

Lalumiere, M.L., Harris, G.T., & Rice, ME. (2001). Developmental instability and
psychopathy. Evolution and Human Behavior, 22, 75-92.

89

Lalumiere, M.L., & Quinsey, V.L. (1996). Sexual deviance, antisociality, mating effort
and the use of sexually coercive behaviors. Personality and Individual
Differences, 21, -33-48.

Lejuez, C.W., Read, J .P., Kahler, C.W., Richards, J.B., Ramsey, S.E., Stuart, G.L., et al.
(2002). Evaluation of a behavioral measure of risk taking: The balloon analogue
risk task (BART). Journal of Experimental Psychology: Applied, 8(2), 75-84.

Lennox, R.D., & Wolfe, RN. (1984). Revision of the self-monitoring scale. Journal of
Personality and Social Psychology, 46(6), 1349-1364.

Lilienfeld, S.O., & Andrews, RP. (1996). Development and preliminary validation of a
self report measure of psychopathic personality traits in noncriminal populations.
Journal of Personality Assessment, 66, 488-524.

Lilienfeld, S.O., & Fowler, K.A. (2006). The self-report assessment of psychopathy:
Problems, pitfalls, and promises. In C.J. Patrick (Ed.), Handbook of Psychopathy.
New York: Guilford Press.

Lilienfeld, S.O., & Widows, M. (2005). Manual for the Psychopathic Personality
Inventory—Revised (PPI-R). Lutz, FL: Psychological Assessment Resources.

Lykken, D.T. (1957). A study of anxiety in the sociopathic personality. Journal of
Abnormal and Social Psychology, 55 , 6-10.

Mahrnut, M. K., Homewood, J ., & Stevenson, R. J. (2008). The characteristics of non-
criminals with high psychopathy traits: Are they similar to criminal psychopaths?
Journal of Research in Personality, 42(3), 679-692.

MckNight, P.E., Mcknight, K.M., Sidani, S., & Figueredo, AJ. (2007). Missing data: A
gentle introduction. New York: Guilford Press.

Mealey, L. (1995). The sociobiology of sociopathy: An integrated evolutionary model.
Brain and behavioral Sciences, 18, 523-599.

Moffitt, T.E., Caspi, A., Belsky, J., & Silva, RA. (1992). Childhood experience and the
onset of menarche: A test of a sociobiological model. Child Development, 63, 47 -
58.

Mokros, A., Menner, B., Eisenbarth, I-I., Alpers, G.W., Lange, K.W., & Osterheider, M.
(2008). Diminished cooperativeness of psychopaths in a prisoner’s dilemma game
yields higher rewards. Journal of Abnormal Psychology, 1 1 7(2), 406-413.

Nettle, D. (2006). The evolution of personality variation in humans and other animals.
American Psychologist, 61(6), 622-631.

90

Nettle, D., & Clegg, H. (2006). Schizotypy, creativity and mating success in humans.
Proceedings of the Royal Society of London, Series B, 273, 61 1—615.

Neumann, C.S., Malterer, M.B., & Newman, J .P. (2008). Factor structure of the
Psychopathic Personality Inventory (PPI): Findings from a large incarcerated
sample. Psychological Assessment, 20(2), 169-174.

Oli, MK. (2004). The fast-slow continuum and mammalian life-history patterns: an
empirical evaluation. Basic and Applied Ecology, 5, 449—463

Patrick, C.J., Curtin, J.J., & Tellegen, A. (2002). Development and validation of a brief
form of the Multidimensional Personality Questionnaire. Psychological
Assessment, 14(2), 150-163.

Patrick, C.J . (2007). Antisocial personality disorder and psychopathy. In W. O’Donohue,
K.A. Fowler, & S.O. Lilienfeld (Eds.), Handbook of Personality Disorders
(pp. 1 09-166). New York: Sage.

Patrick, C.J., & Bemat, EM. (2009). Neurobiology of psychopathy: A two-process
theory. In G. G. Bemtson & J. T. Cacioppo (Eds.), Handbook of Neuroscience for
the Behavioral Sciences (1 l 10 — 1131). New York: John Wiley and Sons.

Patrick, C.J., Bradley, M.M., & Lang, P.J. (1993). Emotion in the criminal psychopath:
Startle reflex modulation. Journal of Abnormal Psychology, 102, 82-92.

Patrick, C.J., Cuthbert, B.N., & Lang, P.J. (1994). Emotion in the criminal psychopath:
Fear image processing. Journal of Abnormal Psychology, 103, 523-534.

Patton, J.H., Stanford, M.S., & Barratt, ES. (1995). Factor structure of the Barratt
Impulsiveness Scale. Journal of Clinical Psychology, 51, 768—774.

Paulhus, D.L. (1991). Measurement and control of response bias. In Robinson, J .P.,
Shaver, P.R., & Wrightsman, L.S. (Eds.). Measurement of Personality and Social
Psychological Attitudes (pp.17-59). New York: Harcourt, Brace, Jovanovich.

Paulhus, D.L., Hemphill, J .D., & Hare, R.D. (in press). Manual for the Self-Report
Psychopathy scale. Toronto: Multi-Health Systems.

Penke, L., & Asendorpf, J .B. (2008). Beyond global sociosexual orientations: A more
differentiated look at sociosexuality and its effects on courtship and romantic
relationships. Journal of Personality and Social Psychology, 95(5), 1113—1 135.

Porter, S., Fairweather, D., Drugge, J ., Herve, H., Birt, A., & Boer, DP. (2000). Profiles

of psychopathy in incarcerated sexual offenders. Criminal Justice and Behavior,
27(2), 216-233.

91

Quinlan, R]. (2003). Father absence, parental care, and female reproductive
development. Evolution and Human Behavior, 24 , 376-390.

Quinsey, V.L., Rice, M.E., & Harris, GT. (1995). Evolutionary perspectives on sexual
offending. Sexual Abuse: A journal of Research and Treatment, 7, 301-315.

Raine, A., Dodge, K., Loeber, R., Gatzke-Kopp, L., Lynam, D., Reynolds, C.,
Stouthamer-Loeber, M., & Liu, J. (2006). The reactive—proactive aggression
questionnaire: Differential correlates of reactive and proactive aggression in
adolescent boys. Aggressive Behavior, 32, 159-171.

Raine, A., Lencz, T., Bihrle, S., LaCasse, L., & Colletti, P. (2000). Reduced prefrontal
gray matter volume and reduced autonomic activity in antisocial personality
disorder. Archives of general Psychiatry, 5 7, 119-127.

Rilling, J .K, Glenn, A.L., Jairam, M.R., Pagnoni, G., Goldsmith, D.R., Elfenbein, H.A.,
& Lilienfeld, 8.0. (2007). Neural correlates of social cooperation and non-
cooperation as a function of psychopathy. Biological Psychiatry, 61, 1260-1271.

Schachter, S., & Latané, B. (1964). Crime, cognition, and the autonomic nervous system.
Nebraska symposium on Motivation, 12, 221-273.

Schmauk, F .J . (1970). Punishment, arousal, and avoidance learning in sociopaths.
Journal of Abnormal Psychology, 76, 325-335.

Schmitt, D.P. (2004a). Sociosexuality, self-esteem, and romantic attachment.
Unpublished manuscript.

Schmitt, D.P., Alcalay, L., Allik, J ., Angleitner, A., Ault, L., Austers, I., et al. (2004).
Patterns and universals of adult romantic attachment across 62 cultural regions:
Are models of self and other pancultural constructs? Journal of Cross-Cultural
Psychology, 35, 367-402.

Seto, M.C., Lalumiere, M.L., & Quinsey, V.L. (1995). Sensation seeking and males’
sexual strategy. Personality and Individual Differences, 19, 669-676.

Soderstrom, H. (2003). Psychopathy as a disorder of empathy. European Child &
Adolescent Psychiatry, 12(5), 249-252.

Thomhill, R., & Thomhill, NW. (1992). The evolutionary psychology of men’s coercive
sexuality. Behavioral and Brain Sciences, 15, 363 -421.

Viding, E. et al. (2005). Evidence for substantial genetic risk for psychopathy in 7-year
olds. Journal of Child Psychological Psychiatry, 46, 592-597.

92

Wakefield, J.C. (1992). Disorder as harmful dysfunction: A conceptual critique of DSM-
III-R’s definition of mental disorder. Psychological Review, 99(2), 232-247.

Waldman, I.D., Rhee, SH. (2006). Genetic and environmental influences on psychopathy
and antisocial behavior. In C.J. Patrick (Ed.), Handbook of Psychopathy. New
York: Guilford Press.

Wallace, J .F., Malterer, M.B., & Newman, JP. (2009). Mapping Gray’s BIS and BAS
constructs onto factor 1 and factor 2 of Hare’s Psychopathy Checklist—Revised.
Personality and Individual Dtflerences, 4 7(8), 812-816.

Williams, K. M., & Paulhus, D. L. (2004). Factor structure of the Self-Report
Psychopathy Scale (SRP-II) in non-forensic samples. Personality and Individual
Dflerences, 37, 765—778.

Williams, K.M., Paulhus, D.L. & Hare, RD. (2007). Capturing the four-factor structure
of psychopathy in college students via self-report. Journal of Personality
Assessment, 88(2), 205-219.

Williamson, 3., Hare, R.D., & Wong, S. (1987). Violence: Criminal psychopaths and
their victims. Canadian Journal of Behavioural Science, I 9, 454-462.

Wilson, M., & Daly, M. (1997). Life expectancy, economic inequality, homicide and
reproductive timing in Chicago neighbourhoods. British Medical Journal, 314,
1271-1274.

Witt, E.A. & Donnellan, MB. (2008). Furthering the case for the MPQ-based measures
of psychopathy. Personality and Individual Dificrences, 45(3), 219-225.

Witt, E.A., Donnellan, M.E., & Blonigen, D.M. (2009). Using existing self-report
inventories to measure the psychopathic personality traits of fearless dominance

and impulsive antisociality. Journal of Research in Personality, 43(6), 1006-
1016.

Yang, Y.L., Raine, A., Lencz, T., Lacasse, L., & Colletti, P. (2005). Volume reduction in

prefrontal gray matter in unsuccessful criminal psychopaths. Biological
Psychiatry, 5 7, 1 109-1 1 16.

93