1 personality and criminal behavior
TRANSCRIPT
1
PERSONALITY AND CRIMINAL BEHAVIOR:
RECONSIDERING THE INDIVIDUAL
A dissertation submitted to the
Division of Research and Advanced Studies of the University of Cincinnati
In partial fulfillment of the requirements for the degree of
DOCTORATE OF PHILOSOPHY (Ph.D.)
in the Division of Criminal Justice
of the College of Education
2001
by
Shelley Johnson Listwan
B.S., Wright State University, 1995 M.S., University of Cincinnati, 1996
Committee Chair: Patricia Van Voorhis
2
TABLE OF CONTENTS
TABLE OF CONTENTS………………………………………………………………….1
TABLE OF TABLES………………….…………………………………………………..4
TABLE OF FIGURES………………….…………………………………………………6
CHAPTER 1: INTRODUCTION………..…………………………………………….…7
METHOD ......................................................................................................................... .9
CHAPTER 2: LITERATURE REVIEW.………………………………………………11
FINDING VALUE IN INDIVIDUAL THEORIES ................................................................... .11 Definition of an Elusive Concept ............................................................................. .11 Psychological Foundations of Personality .............................................................. .12 Personality Trait Measurement ................................................................................ 13 A Paradigm Crisis..................................................................................................... 15 State versus Trait ...................................................................................................... 17 Summary ................................................................................................................... 18
THEORETICAL AND EMPIRICAL CONTEXT ...................................................................... 19 OF THE CURRENT STUDY................................................................................................ 19
Grounds for Excluding the Individual ...................................................................... 20 PERSONALITY AND CRIME ............................................................................................. 22
The EPQ and Crime.................................................................................................. 22 The MPQ and Crime................................................................................................. 27 Summary ................................................................................................................... 31
DEVELOPMENTAL THEORY AND PERSONALITY.............................................................. 33 Self Control Theory................................................................................................... 34 Lifecourse Theory ..................................................................................................... 36 Developmental Theory .............................................................................................. 38 Summary ................................................................................................................... 42
PERSONALITY AND CORRECTIONS ................................................................................. 42 Effective Classification ............................................................................................. 43 Psychological Classification Systems ....................................................................... 46 Summary ................................................................................................................... 55
AN INTEGRATION OF PERSONALITY TYPES .................................................................... 55 Relationship Between Types ..................................................................................... 56
CONCLUSIONS................................................................................................................ 60
CHAPTER 3: METHOD…...……………………………………………………………61
METHOD………………………………………………………………………………….61
Design ....................................................................................................................... 61 Sample....................................................................................................................... 61
3
MEASUREMENT OF RESEARCH VARIABLES.................................................................... 69 Independent Variables .............................................................................................. 69 Dependent Variables................................................................................................. 74
DATA ANALYSIS ............................................................................................................ 76 Model #1: Personality and Any New Arrest ............................................................ 76 Model #2: Personality and Frequency of Arrest ..................................................... 82 Model #3: Personality and the Seriousness of the Offense....................................... 84
CHAPTER 4: RESULTS…...……………..…………………………………………….89
RESULTS. ....................................................................................................................... 89 Multivariate Analysis of Arrest: Model 1 ................................................................ 89 Multivariate Analysis of Multiple Arrests: Model 2 ................................................ 95 Multivariate Analysis of Offense Type: Model 3 ..................................................... 97
CHAPTER 5: CONCLUSIONS…………………………………………………….…109
DISCUSSION AND CONCLUSIONS. ................................................................................. 109 Limitations .............................................................................................................. 109
SUMMARY OF FINDINGS............................................................................................... 111 BROADER IMPLICATIONS OF THE FINDINGS.................................................................. 117 FUTURE RESEARCH DIRECTIONS.................................................................................. 126
REFERENCES………………………………………………………………………….128
APPENDIX A…………………………………………………………………………..140
APPENDIX B……………………………………………………………………………146
APPENDIX C…………………………………………………………………………..155
APPENDIX D……………………………………………………………………………162
APPENDIX E……………………………………………………………………………168
APPENDIX F……………………………………………………………………………171
4
TABLE OF TABLES
Table 1. Summary of key personality constructs for the Eysenck Personality Questionnaire ………………….………………………………………………………………………26
Table 2. Summary of key personality constructs for the Multidimensional Personality
Questionnaire……………………………………………………………..………….32 Table 3. Summary of key predictors of criminal behavior…..………………..………….45 Table 4. Summary of Interpersonal maturity levels ……………………………..………..51 Table 5. Summary of key personality constructs for the Jesness Inventory……..………52 Table 6. Summary of key personality constructs for the collapsed Jesness Inventory
types…………………………………………………………………………..……...54 Table 7. Relationship between EPQ, MPQ and Jesness I-level, personality, and collapsed
types…………………………………………………………………………..……...57 Table 8. Percentage and frequency distribution of participants’ social demographic
characteristics………………………………………………………………….…….64 Table 9. Percentage and frequency distribution of participants’ prior and current offense
record……….……...…………………………………………………………..…….66 Table 10. Percentage and frequency distribution of participants’ I-level and personality
types…………………………………………………………………………..……...72 Table 11. Percentage and frequency distribution of participants’ Salient Factor Scores.…74 Table 12. Percentage and frequency distribution of participants’ recidivism rates..….….77 Table 13. Distribution by Year of Parolees at Risk (Model 1)…………………..….……80 Table 14. Distribution by Year of Parolees at Risk (Model 3: Drug Offenses)…………..86 Table 15. Distribution by Year of Parolees at Risk (Model 3: Property Offenses)………87 Table 16. Distribution by Year of Parolees at Risk (Model 3: Violent Offenses)…….…..88 Table 17. Model 1 Logistic Regression: Probability of Rearrest by Personality & Control
Variables..………………………………………………………………………….…..92
5
Table 18. Model 2 Logistic Regression: Probability of Multiple Rearrest by Personality & Control Variables…………….………………………………………………….……..98
Table 19. Model 3 Logistic Regression: Probability of Rearrest for a Drug Offense by
Personality & Control Variables………………………………………………..……100 Table 20. Model 3 Logistic Regression: Probability of Rearrest for a Drug Offense by
Personality & Control Variables……………………………………………………..103 Table 21. Model 3 Logistic Regression: Probability of Rearrest by Personality & Control
Variables predicting Property Offenses…………..………………………………..105 Table 22. Model 3 Logistic Regression: Probability of Rearrest by Personality & Control
Variables predicting Violent Offenses………………………………………………108
6
TABLE OF FIGURES
Figure 1. Rearrest rates by personality type………………………….………………….….88 Figure 2. Rearrest rates by personality type: Low risk…..……………………………….91 Figure 3. Rearrest rates by personality type: High risk….....…………………………….92 Figure 4. Personality predicting multiple rearrest………..………………………………94 Figure 5. Rearrest rate by personality type: Drug offenses..……...……………………...97 Figure 6. Rearrest rate by I-level type: Drug offenses…….……………………………100 Figure 7. Rearrest rate by personality type: Property offenses.………………………...102 Figure 8. Rearrest rate by personality type: Violent offenses……..……………………...105
7
CHAPTER 1
Introduction
Understanding why crime occurs requires an appreciation for the complexity of
human behavior. Behavior is not determined by one factor, but rather influenced by a
host of interrelated factors (Bandura, Reese, & Adams, 1982). For example, an
individual’s reaction to losing his or her job may vary according to factors such as age,
coping skills, personality, future job opportunities, level of social support, or financial
status. Thus, both individual and structural level factors are essential to explaining
current behavior and to predicting future behavior. Related to this, the field of
criminology has intended to be interdisciplinary in nature and should logically support
the inclusion of literature from other disciplines. One such discipline that is particularly
relevant is psychology. However, some scholars assert that criminologists often fail to
consider psychological variables in their research (Andrews & Bonta, 1998; Andrews &
Wormith, 1988).
One glaring omission in this regard is criminology’s refusal to acknowledge well-
established findings showing a relationship between personality and crime (Andrews and
Wormith, 1989). Until the recent inclusion of personality as a key factor in social
learning (Andrews and Bonta, 1998) and several life course paradigms (see Caspi,
Moffitt, Silva, Stouthamer-Loeber, Kreuger, & Schmutte, 1994) interest in personality
and crime was primarily confined to corrections. Most notably during the 1960s and
1970s, Quay and Parsons (1972), Edwin Megargee (1983; 1984), Marguerite Warren
8
(1966) and Carl Jesness (1964) created a number of psychological typologies designed to
differentiate offenders for correctional treatment purposes. A number of studies find the
personality types identified by those typologies to be relevant to offender behavior (Van
Voorhis, 1994; Heide, 1982; and Jesness, 1988). More recently, correctional scholars
(see Andrews & Bonta, 1998) identify personality as a “responsivity” consideration,
which recognizes that individual characteristics may influence the success of correctional
programming.
Notwithstanding the important findings from the classification research, “few
studies have used classification models as measures of personality, developmental, or
behavioral contributors to models of crime causation” (Van Voorhis, 1994, p. 8). This is
unfortunate because a number of the types identified by the correctional typologies will
be shown to mirror those scales utilized by criminologists such as Eysenck (1996) and
Caspi et al. (1994).
Drawing from several paradigms, the present study explores the relationship
between one correctional typology, the Jesness Inventory, and long-term recidivism. In
doing so, the correctional paradigm is moved into the broader field of criminology to
assess its utility to explain, not simply correctional adjustment, but criminal behavior.
Moreover, a relationship between types identified by the Jesness Inventory and
recidivism supports its use as a measure of the constructs paralleled by the above-
mentioned criminologists. As outlined in the lifecourse and developmental research, the
study will also explore whether certain personality types are related to persistent criminal
behavior.
Finally, this study utilizes long term recidivism data to explore the full extent or
9
potential of the Jesness Inventory in predicting criminal behavior. It is theorized that if
the personality subtypes defined by the Jesness Inventory are useful in predicting
recidivism, the research will imply that personality should be added to existing risk
assessment models to increase their predictive power. As a result, the current research
will address both applied and theoretical issues. To date, no one has taken these two areas
of research and bridged the gap to develop a better understanding of criminal behavior
and to improve predictions of its likelihood.
Method
The current study extends a previous classification study conducted at a federal
penitentiary and prison camp in Terre Haute, Indiana between 1986 and 1988. The
National Institute of Justice (NIJ) funded study1, conducted by Patricia Van Voorhis,
accumulated extensive intake data on federal prison inmates. The original sample
includes 179 federal penitentiary inmates and 190 prison camp inmates. This intake
cohort is one of very few samples to contain personality-based classification measures of
adult male offenders. The research approach for the proposed study builds from the
design of the initial classification study. Given the sample is of federal inmates,
recidivism data includes nationwide arrest statistics by individuals. These data were
collected via the National Crime Information Center (NCIC) database by a pretrial
services center in a midwestern county. The following dependent measures will be
constructed: (a) time at risk to first offense, (b) time at risk to each offense (for those
arrested multiple times), and (c) nature or seriousness of the offense.
Preparing the data in this manner will facilitate a recidivism analysis that accounts
10
for time to first offense, the frequency and rate of offending, and the nature of the
offense. The data will afford the opportunity to employ bivariate and multivariate
analyses. Event history will be used to study the probability of arrest. By using this
technique, the study will distinguish whether those who fail (i.e., incur an arrest) differ by
personality type while controlling for time at risk.
In conclusion, using the Jesness Inventory and other data from this sample of
federal inmates, the proposed study endeavors to document the relationship between
personality and criminal behavior. If personality types set forth by the Jesness Inventory
are useful in predicting recidivism, they may also be useful in theory development, risk
assessment technology, and understanding the adult male offender. The following
research questions will be addressed:
(1) Are certain personality subtypes related to criminal recidivism? And if so,
which ones?
(2) Do certain personality subtypes predict persistence? And if so, which
ones?
(3) Can personality subtypes be used to predict the nature of the offense?
(4) Will the relationships found in the empirical literature on personality and
crime also be found in the current study limited to federal inmates?
(5) Can the Jesness Inventory be used to predict criminal behavior over time?
(6) Should personality be added to existing risk assessment models to further
increase the predictive power of more traditional predictors of recidivism
(i.e. age, prior record, and substance abuse)?
1 Grant Number 85-IJ-CX-0063
11
CHAPTER 2
Literature Review
This chapter reviews the theoretical groundwork and empirical findings regarding
personality and crime. A substantial amount of research is examined to discern the
following: (a) how the construct of personality has developed over time in the field of
psychology, (b) why psychology was neglected in criminology theory, discourse, and
research, (c) what conclusions cab be drawn from the current crime and personality
literature, (d) how personality fits into current theories such as self control, lifecourse,
and developmental criminology, (e) how personality is used in correctional typologies to
differentiate offenders for treatment, and (f) whether personality can be used to predict
future criminal behavior.
Finding Value in Individual Theories
Definition of an Elusive Concept
Gestalt psychologists maintain that in order to understand the whole, you must
take into consideration the sum of its parts. To understand individual behavior, a number
of factors such as personality must be considered. Personality researchers are concerned
with providing an account of how individuals differ from one another in their behavior
(Derlega, Winstead, & Jones, 1989). The study of personality focuses on the stable and
enduring characteristics influence emotion, motivation, and behavior. That is not to say
that all behavior is stable and not subject to external influences, rather that “human
behavior can change over time and that an individual’s behavior is often influenced by
situational factors. But it is still possible to speak of a set of personality characteristics
12
that organize behavior” (Derlega, et al, 1989, p. 4). Those sets of characteristics, then,
could also be used to explain criminal behavior.
Defining exactly what personality is, however, remains somewhat problematic.
Some psychologists theorize that individual definitions are a reflection of the researcher’s
bias rather than a true reflection of the construct (Allport, 1937; Hall & Lindzey, 1957
Murray, 1938; Eysenck, 1983). In fact, in the 1930s and 1940s, a time in which the study
of personality was gaining prominence, one author recorded over 50 definitions of the
construct (Allport, 1937). However, for the purpose of this study, personality will be
defined as follows: “the dynamic organization within the individual of those
psychophysical systems that determine his (or her) unique adjustments to his (her)
environments” (Allport, 1937, p.48). Personality is seen as evolving and dynamic,
determining actions; as every person is unique but also shares similar characteristics
(Allport, 1937).
Psychological Foundations of Personality
It is important to mention that although the early research on personality was not
concerned with criminal behavior, per se, its focus on explaining general behavior
remains applicable. A brief review of the historical development of the construct and the
varied ways in which it is measured are essential to defending its place in criminological
discourse.
Personality research originated in the 1930s with the work of Gordon Allport and
Henry Murray (Pervin, 1990). Although the study of personality developed throughout
the decade, Gordon Allport’s 1939 work, ‘Personality: A Psychological Interpretation,’
lent clear definition to this emerging field. Allport criticized psychologists for
13
consistently ignoring the individual for a more generalized study of human behavior. By
rejecting the view that all behavior is composed of simple innate habits, Allport defined
personality. The author introduced the concept that personality guides behavior and
decisions as people interact with the environment (Pervin, 1990).
Allport recognized that personality developed early in life and was influenced by
external factors in the environment. Personality, according to Allport, emerges around
four months of age, and develops continually while consistently influencing behavior.
Unlike criminologists such as Gottfredson and Hirschi (1990), however, Allport notes
that the consistency of personality at four months of age should not be taken to imply that
personality is fixed during the first year of life; rather later circumstances or interventions
can have a tremendous impact (Allport, 1937).
The importance of Allport’s book at this time was paralleled by the work of
Henry Murray (Pervin, 1990). Murray (1938) concurred with Allport and claimed that
psychologists have ignored the individual in their analyses. Murray advocated for
uncovering the determinants of single behavioral events. He viewed the individual as
consisting of an “infinitely complex series of temporally related activities extending from
birth to death” (p.39). Murray (1938) sought an understanding of the consistency of
personality while also recognizing that it is subject to maturation. That is, that individuals
display a tendency to react consistently but that fluctuations may occur.
Personality Trait Measurement
The study of personality gained prominence throughout the 1930s and 1940s,
however, one problem persisted: the measurement of personality traits. Strictly speaking,
no two individuals have an identical organization of personality traits, but rather develop
14
comparable characteristics that are manifested differently (Allport, 1938). Assuming
individual personality traits can be labeled, determining which labels to use was a
complex task. For example, Allport (1937) found there are “approximately 18,000 terms
(chiefly adjectives) in the English language designating distinctive and personal forms of
behavior” (p. 304). Within these groups’ of adjectives, it is theorized that some represent
major or cardinal traits and other minor or secondary traits (Allport, 1937). The field of
psychology needed a taxonomy that would permit researchers to study and replicate
personality traits across individuals instead of studying thousands of adjectives (John,
1990).
The work of Cattell (1950) set the stage for the measurement of personality
variables. Cattell criticized psychologists for jumping into personality research without
thoroughly dealing with the task of objectively describing and measuring its components.
The author argued that psychologists relied on the clinical guesswork in their efforts to
determine personality structure (Cattell, 1965). As a result, factor analysis gained
popularity as a way to develop a taxonomy of personality. The procedure involves
“present(ing) hundreds-in the long run thousands- of carefully chosen items to large
groups of normal persons. Each item is then correlated with every other to see which sets
‘go together’” (Cattell, 1965, p. 70). The factor analytic approach can be used to compile
traits with the recognition “that there are natural unitary structures in personality and that
it is these traits, rather than the endless labels in the dictionary, on which we should
concentrate” (Cattell, 1965, p. 55).
Once factor analytic procedures produced personality factors, psychologists began
constructing questionnaires. In addition, both World War I and World War II led to the
15
need for mass standardized testing of soldiers. This need combined with the inception of
graduate programs in clinical psychology acted as a catalyst to the development of the
standardized personality questionnaire. As such, the development and widespread use of
the MMPI (Hathaway & Meehl, 1951), a paper and pencil instrument consisting of 566
statements, would become the framework for many of the personality tests to emerge
(Pervin, 1990). Unfortunately the widespread expansion of personality tests also fueled a
paradigm crisis.
A Paradigm Crisis
This rapidly developing branch of psychology did not escape the 1950s and 60s
without difficulties. In fact, the study of personality all but disappeared during the 1960s
(Pervin, 1980). Negative research findings fueled the eventual shift away from
personality research. Specifically, studies testing the reliability and validity of personality
scales were not meeting with success. Moreover, there still was no agreed upon definition
of personality, researchers were using a variety of instruments that were plagued by
response bias, and many studies were breaking down under replication (Pervin, 1980).
These factors led many to question the utility of assessing such an abstract construct.
Furthering the crisis, Mischel critiqued traditional psychology in support of the
situational perspective (Pervin, 1990). Proponents of the situational perspective
advocated a shift from emphasizing traits to exploring how individuals cognitively
perceive and respond to the environment (Mischel, 1990).
The situational perspective was seen as important throughout the 1970s, with one
notable exception: the development of the interactional perspective. Researchers began
arguing that the trait and situational approaches need not be at odds with one another
16
(Epstein & O’Brien, 1985; Kendrick & Funder, 1988; Magnusson, 1990). For example,
Epstein and O’Brien (1985) outline the interaction focus as follows “interactionism
provides a reasonable resolution to the person-situation debate as behavior can never be
determined by the person or the situation alone, but is always the result of the interaction
between them” (p. 515). In addition, Kendrick and Funder (1988) reviewed the
arguments proposed by the situationists and advocated for a more inclusive model that
viewed traits as influencing behavior indirectly by causing the individual to respond to
different environmental settings and events.
Similarly, Eysenck & Eysenck (1985) argue that individuals choose the situations
in which they find themselves; therefore, situations are the consequence of an already
existing system of perceptions and attitudes influenced by personality traits. Magnusson
(1990) concurred and argued that past experiences influence current thought and shape
the individual; however, the individual’s personality influences and paints the situations
he or she will experience. Related to this was the development of Albert Bandura’s
social learning theory (Pervin, 1990).
Social learning theories embrace the complexity of this perspective insisting that
individual behavior is the result of the interaction between the individual’s prior learning
and the immediate environment. From this socio-cognitive perspective, learning and
behavior are driven by and individual’s perception of the world and their personality
experiences. Individual behavior and reactions emerge as the result of both mimicking
the behavior of others and from past behaviors that were reinforced (Bandura, 1989). As
such, features of the environment that include interactions with others both facilitate and
inhibit behavior (Andrews & Bonta, 1998). Social learning theories delineate the
17
importance of environmental stimuli while not discounting the importance of individual
traits2 (Bandura, 1989; Bierman & Furman, 1984; Farrington, 1994; and Moffitt, 1993).
In a similar vein, self-efficacy beliefs are theorized to affect thought patterns and
influence motivation. That is, those with high-perceived self-efficacy were more likely to
set higher goals and view themselves as competent and capable (Bandura, 1989). Self-
efficacy becomes increasingly important as we explore the literature with regard to
personality. Bandura (1989) theorizes that “people possess self-directive capabilities that
enable them to exercise some control over their thoughts, feelings, and actions by the
consequences they produce for themselves” (p. 1179).
State versus Trait
Throughout this discussion, personality has been portrayed as relatively stable and
enduring. However, some critics of personality theory often assert that traits are merely
states of mind or moods that constantly fluctuate (Mischel, 1968). Specifically, the
construct of neuroticism is questioned as a state of mind and not a stable or enduring
personality trait. Neuroticism has been defined as “a mood dispositional dimension that
reflects the pervasive individual differences in negative emotionality and self concept”
(Brief, Burke, George, Robinson, & Websert, 1988). Although many would agree that
anxiety can exist as a state of mind, others argue for its consistency across time and
situations (Tellegen, 1985).
Researchers have found that anxiety can exist both as a state of mind and as a
personality trait. Studies conclude that individuals higher in trait anxiety are consistently
more prone to perceive greater danger in their relationships and respond with greater
2 Yet early social learning theorists did not specifically focus on personality.
18
elevations of situational or state anxiety (Speilberger, 1985). Individuals with high trait
anxiety, often called negative affectivity, tend to be distressed and have a very negative
view of themselves, often worry, and tend to dwell on frustrations and disappointments
(Watson & Clark, 1984). Moreover, individuals high in neuroticism were shown to be
more distressed on average in comparison to low neuroticism subjects. This difference
could be attributed to a lower threshold for responding to stressful events among highly
neurotic subjects (Bolger & Schillings, 1991). Although some individuals may
experience these feelings as a state of mind during times of stress, those high in negative
affectivity manifest these feelings even in the absence of stress (Watson & Clark, 1984).
In sum, the research concludes that personality traits, including neuroticism, are enduring
and consistently affect behavior.
Summary
Personality research emerged in the 1930s and research continues to the present.
Although the feasibility of measuring personality has been questioned in the past, the
development of increasingly reliable tools now provides support for further research.
Moreover, factor analysis has proven to be a reliable tool in the construction of
personality scales and data support the validity and reliability of recent personality
inventories (Church, 1994; McCrae & John, 1992; & Tellegen, 1985). Although the
development of personality research did not escape the scrutiny of situationists, many
psychologists embrace a more tempered view of personality. That is, individual’s react
with the environment but also that individual’s choose the situations in which they find
themselves and the situations are the consequence of an already existing system of
perceptions and attitudes influenced by personality traits (Eysenck and Eysenck, 1985).
19
A review of modern personality texts will reveal that there are many areas
relevant to explaining individual behavior. Only a few have been discussed, however, the
hope is that this review will provide a historical framing, albeit brief, of how personality
has evolved over the last seventy years. Furthermore, in the field of criminology,
personality and crime causation studies are supporting the idea that personality is a
worthwhile area of inquiry. The focus of the following sections will now shift to the
review of how personality influenced the field of criminology.
Theoretical and Empirical Context of the Current Study
During the 1930s when the field of psychology was adopting personality as an
important factor in the explanation of individual behavior, criminology shifted away from
individual- centered explanations. As a result, the criminological literature has been
dominated by sociological theories for the majority of the twentieth century. Science
traditionally values knowledge based in theory that is supported by data and empirical
findings. However, with the exception of a few researchers such as Eysenck (1964),
studies since the 1950s regarding personality and crime have been limited. For example,
Stitt and Giacopassi (1992) concluded that one of the leading journals in the field,
Criminology, has only published four articles on the link between personality and crime
since its inception. Moreover, a review of Criminology journal issues from 1990 to 2000,
conducted by the author, found only three additional personality articles3. There are a
number of justifications cited as to why the field has been dominated by more
3 Self control articles were excluded given the authors argue that self-control is not a psychological construct.
20
environmental causes of crime.
Grounds for Excluding the Individual
For a number of reasons, explanations of crime have drawn heavily from
sociology and only tangentially from other disciplines such as psychology and biology.
The lack of individual-centered theories and research was justified by some on moral
grounds. That is, psychological research was discounted in an effort to protect the
individual from being labeled different and confined or segregated based on these
differences (Andrews & Bonta, 1998; Andrews & Wormith, 1989; and Hirschi &
Hindeland, 1977). Psychological theories were mentioned in the context of Lombroso’s
(1911) or Hooten’s (1939) work pertaining to physical inferiority. This type of research
did in fact arm some researchers in the early 1900s with evidence to support their own
personal biases (Gould, 1980); however, it did not itself negate the existence of
individual differences. As a result of the fear of abuse, criminological research moved
away from the individual and toward structural explanations of behavior (Andrews &
Bonta, 1998, & Lilly, Cullen, & Ball, 1994). However, as pointed out by Andrews and
Wormith (1989) “although this type of moral reasoning is appreciated in view of some
abuse in the early history of correctional treatment, it is obviously suspect on more
general grounds” (p. 290).
Other than protecting the offender, psychological considerations were ignored on
professional grounds. Specifically, Andrews & Wormith (1989) point out,
“professionally, many sociologists discounted psychological contributions in order to
promote the importance of their favored variables of political economy, social location,
social reaction, and inequality in the distribution of societal resources” (p. 291). This
21
trend is seen in three popular reviews of the psychological literature (See e.g.; Schuessler
& Cressey, 1950; Tennenbaum, 1977; and Waldo & Dinitz, 1967). The studies all found
differences in the personalities of offenders and non-offenders even when “only four of
the 30 different personality tests employed in the early studies were able to meet the
psychometric standards of the 1960s” (Andrews and Wormith, p. 300, 1989)
In another example of this trend, Vold and Bernard (1986) acknowledge a link
between personality and crime but as pointed out by Andrews and Wormith (1989),
dismissed the link as providing no theoretical relevance to understanding criminal
behavior. Interestingly, these reviews were met with resounding support and uncritical
acceptance (Hirschi and Hindelang, 1977). Hence, it appears the bias against theories of
personality and crime had less to do with protecting the disadvantaged and more to do
with protecting sociological criminology.
A third factor related to what has been termed “anti-individual bias,” (Andrews &
Wormith, 1989) is the effect of the political crisis during the 1960s and 1970s. Most
often the anti-individual bias is discussed in relation to the demise of rehabilitation (See
e.g., Andrews, Zinger, Hoge, Bonta, & Cullen, 1990; Cullen & Gendreau, 1989; Cullen
& Gilbert, 1982; Gendreau & Ross, 1987; and Martinson, 1974). During this period
conservative politics and the decline in support for treatment at the “hands of the state”
lent support for the development of social or environment-focused policies to reduce
crime. As a result, it became popular to discount the individual in discussions of
rehabilitation or theory development.
A final reason offered by some researchers as an explanation of anti-individual
bias, is the death of positivism and a rise in theoreticism (Andrews & Bonta, 1998;
22
Andrews & Wormith, 1989). Andrews and Bonta (1998) define theoreticism to refer to
theorists who “adopt and discard knowledge insofar as it is personally or politically
rewarding to do so” (Andrews & Bonta, 1998, p. 28.). One can argue, individual-
centered theories were discarded not due to evidence, but rather to support more popular
and often politically correct theories of crime.
It is apparent that early reviews and current thought among some criminologists
were influenced by anti-individual bias and knowledge destruction rather than an
unbiased reporting of the quality and significance of the findings (Andrews & Bonta,
1998 & Andrews & Wormith, 1989). This review was included to illustrate the
importance of exploring reasons why certain ideologies are attacked. The following
discussion details the research and possible reasons why personality should be integrated
into a broader discussion of criminological theory.
Personality and Crime
Given the focus of criminological research, studies exploring personality and
crime have been limited. However, the existing research is extremely informative. Some
of the earliest research on personality and crime comes from the work of Hans Eysenck.
A review of the research by Eysenck, outlining his proposal for why criminal behavior
occurs, will follow.
The EPQ and Crime
According to Eysenck, in order to study personality and the impact that it may
have on behavior, a theoretical system must be developed. The system should be able to
measure traits and make distinctions based on the similarities and differences among
individuals (Eysenck & Eysenck, 1985). As Eysenck and Eysenck (1985) claim “what
23
one must say, essentially, is that although human beings clearly do differ from each other,
they differ along certain dimensions, and their differences and similarities can therefore
be quantified and measured” (p. 8).
Many of the studies of personality and behavior rely on a descriptive model
containing: (a) traits and (b) types or superfactors. As stated previously, personality traits
are enduring characteristics within the individual that determine his or her behavior. A
type, however, is defined as “a group of correlated traits” (Eysenck & Eysenck, 1985).
The term type in modern personality theory is superordinate to that of a trait. The type
corresponds to what others, using factor analysis, have called second order types or
superfactors (Eysenck & Eysenck, 1985). Hence, the type or superfactor consists of
multiple individual personality traits.
Eysenck’s work emerged from his exploration of personality traits among 700
male service patients in the 1940s. Eysenck completed a large scale factorial study that
resulted in two main types: neuroticism and extroversion-introversion (Eysenck, 1947).
Later, Eysenck (1952) constructed the Eysenck Personality Questionnaire (EPQ) that
included psychoticism. The traits that compose the type neuroticism include: anxious,
depressed, guilt feelings, low self-esteem, tense, irrational, shy, moody, and emotional.
The traits comprising the type extraversion include: sociable, lively, active, assertive,
sensation seeking, carefree, dominant, surgent, and venturesome. Finally, the traits
included in the type psychoticism consist of: aggressive, cold, egocentric, impersonal,
impulsive, antisocial, unempathetic, creative, and tough-minded. The author notes,
however, that the majority of individuals are characterized by a balance between the
types (Eysenck & Eysenck, 1985).
24
Later in his career, Eysenck extended his work to crime causation studies.
Eysenck kept his original typology consisting of neuroticism, extraversion, and
psychoticism to explain crime. He theorized that psychoticism was always related to
crime, extroversion was related in younger samples, and neuroticism in older samples.
Eysenck postulated that neuroticism becomes more important in older samples and
contributes to stronger antisocial habits in adults (Eysenck, 1983). Eysenck viewed the
study of personality as having two interlocking aspects: the descriptive or taxonomic
aspect and the causal aspect including genetics and environment. These two aspects,
together, describe the dynamics of why personality exists (Eysenck & Eysenck, 1985).
Eysenck’s theory includes a chain of factors that begins with genetics and ends
with criminal behavior. He theorized that an individual’s DNA, the genetic structure
underlying individual differences, was the first link in the causal chain. But DNA is not
responsible for behavior, rather an individual’s DNA impacts what Eysenck terms
biological intermediaries such as cortical arousal and as a result conditionability and
conscience (Eysenck, 1996). Cortical arousal is a state in the brain that is marked by
being alert and attentive. Low cortical arousal is related to both extraversion and
psychoticism in that both are marked by poor arousal and therefore cause individuals to
act out in an effort to attain greater arousal. Specifically, he theorizes that individuals
with low cortical arousal seek out arousing and often risky activities that may include
criminal acts.
Eysenck sought an explanation not of antisocial behavior, but rather why people
behave in socially desirable ways. This is where the causal chain continues. Eysenck
states that criminals know right from wrong, but prefer the wrong to the right. He then
25
explains the reason some commit crimes and others do not is a matter of conscience
(Eysenck, 1996). Specifically, he theorized that individuals learn behaviors through
operant and classical conditioning. That is, the child who is punished repeatedly for an
antisocial act and does not develop the appropriate response (i.e., fear of repeated
punishment) will not learn from this punishment and correct behavior in the future by
developing a moral conscience (Eysenck, 1983). The likelihood of developing a moral
conscience is dependent on a number of factors, including whether conditioning
experiences are missing, whether the wrong experiences are reinforced, and whether the
person has low cortical arousal (Eysenck, 1996). He conceded that the learning process,
or the breakdown of this process, contributes to the likelihood of criminal behavior
(Eysenck, 1983). The causal chain ends here with the likelihood of criminal behavior
being predicted from the genetic make-up of the individual.
The personality tie becomes clearer as he explains that extroversion and
psychoticism are linked to low cortical arousal that influences conditionability,
conscience, and ultimately behavior. And with regard to neurotics, anxiety may act as
the drive or motivation for criminal behavior. In this circumstance, emotions may over-
ride reason leading to aggressive and impulsive behavior. In sum, according to Eysenck
(1983) “all three are involved in antisocial conduct, so that typically the person indulging
in such conduct would be extraverted rather than introverted, emotionally unstable rather
than stable, and high on psychoticism rather than on superego functioning” (p. 64). Table
1 lists the key personality constructs for Eysenck theory of personality.
26
Table 1. Summary of key personality constructs for the Eysenck Personality Questionnaire (Eysenck, 1952).
______________________________________________________________________________________________________ Construct Definition Relationship to Criminal Behavior Extroversion Tend to be outgoing, talkative Eysenck found extroverts tend to have low and friendly. But also tend to cortical arousal which influences conditionability be assertive, sensation seeking, Extroversion more likely in younger samples. and dominate in social situations Neuroticism Tend to be anxious in social Eysenck found neuroticism to be positively situations and often experience associated with criminal behavior in older feelings of low self esteem and samples. guilt. Also tend to be irrational tense, and emotional. Emotions may override reason. Can also exhibit aggressive and impulsive behavior. Psychoticism Tend to be very egocentric and are Eysenck found psychotics tend to have low cortical Unempathetic to others needs. Also tend arousal which influences conditionability. Eysenck To be impulsive, cold, and impersonal suggests psychoticism is positively associated with criminal behavior among all individuals exhibiting
the trait.
27
The MPQ and Crime
Types similar to those defined by Eysenck emerged from the Multidimensional
Personality Questionnaire (MPQ). The MPQ is a self-report personality instrument
designed to assess a broad range of individual differences in behavioral style (Silva,
1990). The ten scales produced by the MPQ can be combined to produce three types or
superfactors: constraint, negative emotionality, and positive emotionality (Silva, 1990).
As described by Caspi & Silva (1995) “constraint is the combination of the
traditionalism, harm avoidance, and control scales. Individuals high on this factor tend to
endorse social norms, act in a cautious and restrained manner, and avoid thrills. Negative
emotionality is a combination of the aggression, alienation, and stress reaction scales.
Individuals high on this dimension have a low general threshold for the experience of
negative emotions such as fear, anxiety, and anger and tend to be involved in antagonistic
relationships. Positive emotionality is a combination of the achievement, social potency,
well-being, and social closeness scales. Individuals high on positive emotionality have a
lower threshold for the experience of positive emotions and tend to view life as being
essentially a pleasurable experience” (p. 492). Researchers found that positive
emotionality, negative emotionality, and constraint emerged as three major dimensions
related to a variety of behavioral measures (Tellegen, 1985).
Utilizing a modified version of the Multidimensional Personality Questionnaire
(MPQ), Caspi et al. (1994) studied the relationship between personality and crime among
adolescents involved in both the Dunedin Multidisciplinary Health and Development
(DMHD) study and the Pittsburgh Youth Study. The DMHD study is a longitudinal
examination of the health, development, and behavior of 862 adolescent subjects. The
28
complete cohort of consecutive births between April 1, 1972 and March 31, 1973 in
Dunedin, New Zealand were traced at age 3, and every two years thereafter until the age
of 25 (Silva, 1990). Four measures of delinquency were examined: self-report, informant,
police contact, and court convictions. The study concluded that among both males and
females, constraint and negative emotionality emerged as predictors of delinquent
behavior across the self-report, informant, and official record measures (Caspi et al,
1994). In a second analysis of the data, the authors found that persons who engaged in
serious delinquent behavior score significantly lower on the superfactor constraint and
significantly higher on the superfactor negative emotionality (Caspi et al., 1994).
The DMHD study included only Caucasian adolescents living in a mid-sized city.
As a result of this sample limitation, a second study was conducted using data from the
Pittsburgh Youth Study, a longitudinal study examining the causes and correlates of
delinquency that included a more heterogeneous sample (Caspi et al., 1994). The sample
consisted of 508 10 year old boys, of which 54 percent were African-American, and 44
percent lived in a house where the main caregiver has been separated, divorced,
widowed, or never married (Caspi et al., 1994). The MPQ was not appropriate with this
age group, hence, the Child Q-sort (Block, 1961) was used as a replacement. The Child
Q-sort scales were used to create the three superfactors: constraint, negative emotionality,
and positive emotionality (Caspi et al., 1994). The delinquency measures included the
child’s self report, teacher’s self report, and parent’s self report. Among both Caucasians
and African American samples, low constraint and the presence of negative emotionality
emerged as correlates of delinquency.
Taken as a whole, the study revealed that individual differences in personality can
29
predict delinquency across age groups, geographic location, race groups, and gender.
The authors suggest that the importance of negative emotionality and constraint may
work in tandem. Specifically, negative emotionality refers to a tendency to construe life
in a very negative and anxious way and often may perceive benign situations as
threatening. Individuals with high trait anxiety, often called negative affectivity, tend to
be distressed and have a very negative view of themselves, often worry, and tend to dwell
on frustrations and disappointments (Watson & Clark, 1984). In addition, those
individuals who also score low on constraint tend to take risks and approach situations
with fewer worries of the consequences (Caspi et al., 1994). As a result, individuals with
a personality marked by negative emotionality paired with low constraint have a higher
likelihood of engaging in criminal acts (Caspi et al., 1994).
Given those scoring high on negative emotionality have a negative view of
themselves and perceive the world as hostile and threatening, it is also likely that they
will not undertake challenging situations. In fact, one study found that Italian children
who had high efficacy with regard to academic achievement performed better in school
and were able to resist negative peer pressure (Bandura, Barbaranelli, Vittorio Caprara, &
Pastorelli, 1996). In contrast, those who believe they cannot handle threatening or
challenging experiences are more likely to experience depression, anxiety, and stress
(Bandura, 1989).
A subsequent study explored whether personality could predict abstention from
delinquency (Kreuger, Schmutte, Caspi, Moffitt, Campbell, & Silva, 1994). Using the
same data as the previous study, the authors constructed a measure of delinquency that
consisted of three categories: those who reported they had never engaged in delinquency,
30
those who engaged in a wide variety of acts, and those who engaged in normative4
delinquency (Kreuger et al., 1994). The study revealed very different personality profiles
between those who engaged in delinquency on a persistent basis in comparison to those
who abstained from delinquency. Specifically, constraint and negative emotionality
emerged as combined correlates of delinquency among the persistent group. Just as
Moffitt (1993) suggests with her life course persisters typology, those high in negative
emotionality are likely to approach relationships with adversarial attitudes and have
relationships that are short lived. In contrast, the males and females who abstained from
delinquency were characterized as conventional, planful, non-aggressive, and non-
assertive (Kreuger et al., 1994).
Subsequent research has also explored the relationship between personality
characteristics, a child’s ability to delay gratification, and his or her behavior at home or
school. With a sample from the Pittsburgh Youth Study, 508 fourth grade boys were
asked to complete the California Q-sort (Kreuger, Caspi, Moffitt, White, and Stouthamer-
Loeber, 1996). Teachers and mothers were asked to complete the Child Behavior
Checklist (Achenbach and Edelbrock, 1983). The delay of gratification task had
participants choose between an immediately available 40 percent chance to win a nickel
or a delayed 80 percent chance to win a nickel. The authors found that youth who
persistently sought immediate gratification in the lab (e.g., those exhibiting poor impulse
control) were more likely to exhibit problem behaviors in school and at home (Krueger et
al., 1996). In contrast, those children who were able to delay gratification by diverting
attention away from the task at hand, experienced less frustration. The authors concluded
4 Normative delinquency refers to behavior such as truancy and curfew violations.
31
that an antisocial tendency, operationalized as low self-control, influenced the child’s
likelihood of seeking immediate gratification (Kreuger et al., 1996).
Finally, personality researchers have also attempted to document the relationship
between personality and violence. It is rare to find individuals who specialize in criminal
violence, although it is also rare to find complete generality (Farrington, 1982). In
addition, the reason why individuals are violent is the subject of much disagreement. A
study by Caspi, Harrington, Moffitt, Begg, Dickson, Langley, and Silva (1997)
investigated the relationship between personality and a variety of what they called health
risk behaviors that included violent crime. The study used the DMHD data at age 21
from 961 individuals. Personality was assessed at age 18 with the MPQ. The study not
only included violent crime but also alcohol dependence, sexual behavior, and driving
habits. The results concluded that all categories of health risk behavior were correlated
with negative emotionality and low constraint. With regard to violence, individuals with
at least one prior violent conviction were exhibiting personality types marked by a very
negative and hostile view of the world paired with an impulsive nature.
Summary
The purpose of this section is to review the literature on crime and personality.
The literature base is limited, however, it should be noted that each sample represents a
diverse group of individuals over a period of time. The results indicate that personality is
related to various types of criminal behavior with a variety of measures, in a variety of
situations, with a variety of individuals. Table 2 lists the key personality factors formed
by the MPQ and crime research. Given the research finding from the above-mentioned
studies, the current study will attempt to mirror those types and examine their relationship
32
Table 2. Summary of key personality constructs for the Multidimensional Personality Questionnaire (Caspi & Silva, 1990).
______________________________________________________________________________________________________ Construct Definition Relationship to Criminal Behavior Negative Emotionality Tend to experience adverse By itself may not be related to crime.
states such as anger, anxiety, and However, high negative emotionality related irritability. They perceive life to crime in tandem with low constraint. events and acts of others in very hostile and negative ways
Positive Emotionality Tend to view life in a very Not related; may act as insulator pleasurable way. They tend to engage in positive social and working interactions with others Constraint Tend to endorse convention and By itself may not be related to crime. normative behavior. Avoid risky Lack of constraint paired with negative situations and act cautious. emotionality more likely in criminal population.
33
to recidivism. The following section examines how personality fits into the
developmental theories of crime.
Developmental Theory and Personality
An emerging area of research in criminology is now focusing on the unique
circumstances of the individual in an effort to explain criminal behavior over time.
Lifecourse is not personality-crime research per se, but some aspects of the paradigm are
relevant. Lifecourse or developmental theorists explore how an individual develops over
time, the circumstances with which he or she is exposed, and his or her individual
position on the developmental continuum (Loeber & LeBlanc, 1990). It is commonly
thought that individuals differ with regard to their propensity to commit crime. Certain
individuals may commit one crime during their lifetime, whereas others commit multiple
crimes of varying severity throughout. Moreover, research indicates that most offenders’
start at the same level of seriousness, however, only certain individuals persist (Loeber &
LeBlanc, 1990). In an effort to explain these differences, lifecourse theorists rely on
longitudinal data. These studies allow the researcher to uncover how the patterns of
offending emerge over time and what factors may have an impact on behavior.
Within the discipline there is a debate between general theorists who assert a
predisposition established early in life explains the onset and persistence of criminal
behavior and lifecourse theorists who claim that age-related factors across the lifespan
can influence behavior. Both factions in this debate see personality as at least tangentially
important in the explanation of behavior. General theorists, although not directly viewing
their theory as involving personality, theorize that enduring characteristics are
responsible for behavior. In contrast, lifecourse researchers view behavior as influenced
34
by different pathways and turning points throughout the lifespan. A third group, the
developmental theorists, study the development of individual and structural factors and
their impact on behavior at different points in time. All three views will be discussed in
detail.
Self Control Theory
General theorists, Gottfredson and Hirschi (1990) view crime as a hedonistic
event. They assert that everyone has the same motivation to commit crime regardless of
their social situation or the existence of laws and sanctions designed to govern the
acceptability of behavior. Gottfredson and Hirschi developed a theory of crime that
focuses on a single determinant or cause. The determinant or cause of behavior,
according to Gottfredson and Hirschi (1990) is the existence of low self control. Low
self control is an enduring characteristic developed early in life that determines behavior.
Individuals with low self control tend to need immediate gratification, seek out exciting
or thrilling situations, and have very little control over their own behavior. Criminal acts
provide the excitement and gratification to satisfy their needs.
Gottfredson and Hirschi (1990) delineate a number of propositions to explain how
self control can explain criminal behavior over time. First, low self control is established
very early in life and cannot be altered. The causes of low self control, however, do not
fall on the individual, rather, from the absence of nurturance and discipline from families
(Gottfredson & Hirschi, 1990). They place the blame for this development on the
parents’ inability to socialize the child correctly. The authors mention a number of
factors that may inhibit the child’s socialization, including: inadequate punishments,
parental criminality, family size, and mothers who work outside of the home.
35
Second, the theory includes six components to explain behavior. Low self control
involves a “here and now” orientation in that the individual does not consider future
consequences. Individuals with low self control seek simple and gratifying tasks. The
individuals are adventuresome and enjoy thrilling situations. These individuals are more
likely to embrace physical rather than mental tasks. Individuals low in self control tend
to be selfish and self centered in their needs and desires. And finally, these individuals
appear to have a very low frustration tolerance and are more likely to become angered by
these events. Gottfredson and Hirschi (1990) “clearly assert that these six traits we have
sketched are not alternative ways of having low self control; rather, they form a single
one-dimensional latent trait” (Grasmick, Tittle, Bursik & Arkenklev, 1993, p. 9). This
trait then is able to predict involvement in crime as well as success in conventional
occupations and relationships.
Third, individuals with low self control are likely to commit crime or analogous
behaviors throughout their lifespan. The authors assert “these differences remain
reasonably stable with change in the social location of individuals and change in their
knowledge of the operation of the sanction system” (Gottfredson and Hirschi, 1990). In
other words, low self control remains stable over time regardless of environmental
changes or changes in the laws or punishments.
And finally, Gottfredson and Hirschi (1990) argue that age is not important in the
explanation of crime, but that the fluctuations in opportunity explain increases and
decreases in criminal behavior. Interestingly, the authors do not claim that low self
control is by itself the primary reason for crime. Rather, individuals with low self control
are unable to resist the opportunity to commit crimes. Hence, the condition (i.e. low self
36
control) interacts with the opportunity to increase the likelihood of the criminal act
(Grasmick, et al., 1993). More important, however, is the assertion that when
opportunity for crime decreases, as the offender ages, an individual with low self control
is likely to engage in analogous behaviors that are problematic but not necessarily
criminal. Analogous behaviors may include, smoking, gambling, drinking and driving,
and/or sexual promiscuity (Gottfredson and Hirschi, 1990).
As mentioned, Gottfredson and Hirschi (1990) assert that their theory is not a
theory of personality. However, the construct of low self-control is very similar to the
personality dimensions of impulsivity, and on some levels extraversion discussed by
Eysenck. In addition, low self control is directly related to a lack of constraint as
measured by the MPQ. This theory departs from traditional sociological explanations of
crime and for that should be applauded. However, Gottfredson and Hirschi (1990) depart
from many psychologists in their assertion that self-control does not change over time as
individuals react with the environment.
Lifecourse Theory
Sampson and Laub’s (1993) lifecourse perspective recognizes that factors in the
environment may influence an individual’s criminal career patterns. Although
personality is only tangentially discussed in Sampson and Laub’s lifecourse theory, its
relevance to the proposed study is two fold. First, the real value to the discipline is the
introduction of an individual-centered explanation. That is, they recognize that behavior
can change and that individual decision making influences behavior. Second, in contrast
to Gottfredson and Hirschi (1990), Sampson and Laub discuss how other established risk
factors such as employment, marriage, and changes in lifestyle could cause desistence
37
from crime. Although the authors fail to explore the relationship between personality and
criminal acts, the discussion of protective factors is relevant. This proposition will
become increasingly clear in preceding sections devoted to classification.
The authors reanalyzed the data originally utilized in the study by Sheldon and
Eleanor Glueck (1959), titled Unraveling Juvenile Delinquency. They examined why
some people commit crime throughout their lifespan and others desist or stop committing
crimes once they reached adulthood. They viewed individuals as active recipients to their
environment and concluded that the external influences could shape how individuals
make decisions. Each individual has experiences that may lead his or her life in different
directions or pathways (Sampson and Laub, 1993). Hence, the authors argued that
criminal behavior could change as pathways and turning points through life emerge.
Sampson and Laub (1993) address three main themes in their book Crime in the
Making. First, the authors contend that informal controls such as those exerted by
families, schools, and/or peers can mediate the structural contexts that may lead to
delinquency. Second, the authors recognize the existence of stability for some offending
careers. They maintain that early onset of delinquency is one of the best predictors of
future offending. Finally, they recognize that fluctuations in criminal behavior can also
result from what they term social capital. For example, job stability in a quality
occupation and the involvement in quality relationships or marriages can impact the
individual’s likelihood of remaining in crime. The authors, however, fail to include
personality as a factor that influences an individual’s environment. Specifically, McCrae
and Costa (1999) argue that an individual’s personality can influence their preference for
friends, situations, their ideologies about how the world works, and their own social
38
roles.
In sum, lifecourse perspective and subsequent debate is concerned with the
development and stability of criminal behavior. Any reference to personality is made
only as a recognition that individual decision making is influenced by many factors, one
of which may be personality. Although the studies provide valuable insight and much
needed support from individual theories of crime, lifecourse research in criminology has
failed to include personality as a causal or predictive variable. However, research
conducted by developmental theorists is able to contribute to this stability debate.
Developmental Theory
Developmental theories have been popular in psychology for many years.
However, the research on the developmental precursors to crime or deviance is more
limited. A review of the literature appearing in both psychological and criminological
journals will be discussed.
The full debate on the consistency of personality over time will not be addressed
here, however, the findings of a few studies will be highlighted. For example, there is a
growing consensus among psychologists that personality is very stable in adulthood
(McGue, Bacon, Lykken, 1993). However, others argue that personality can change
drastically during the transition from adolescence to adulthood. Researchers have found
evidence of both stability and change during this transition. For example, one study found
that subjects exhibited a decrease in negative emotionality, increase in constraint, but no
change in positive emotionality as measured by the MPQ (McGue, et al., 1993). Eysenck
(1996) also found that extroversion was more likely to decrease with age as did
Newcomb and McGee (1985) with regard to sensation-seeking behaviors. It may be that
39
socializing experiences influence risk taking behaviors in some individuals.
A number of studies have documented the relationship between early aggressive
or antisocial behavior and later delinquency. First, Statin and Magnusson (1989)
explored the relationship between early aggressive behavior and the frequency,
seriousness, and crime patterns occurring later in life. They found a clear and positive
relationship between early aggressive behavior and later crime among male subjects. The
authors also concluded that aggressive boys exhibited restless behavior, poor
concentration, and had poor peer relations as children (Stattin & Magnusson, 1989).
Second, research by Moffitt (1990) explored the relationship between juvenile
delinquency and attention deficit disorder of boys from three to fifteen years of age. The
study found that delinquent boys with attention deficit disorder were most likely to
engage in criminal activity, and exhibited the greatest family adversity, the worst
neuropsychological disorders, and were rated as most aggressive by parents and teachers
at age 13 (Moffitt, 1990). The overall findings indicated, however, that this group
exhibited significantly more antisocial behavior and non-disordered controls at every age.
Third, a study by Patterson, Crosby, and Vuchinich (1992) also found age of onset as a
significant predictor of childhood aggression. Finally, Farrington & West (1990) found
that future juvenile delinquents were more likely to have been rated as troublesome by
their teachers, tended to be hyperactive and lack concentration, and tended to be rated as
impulsive. Moreover, at age 18 they tended to engage in violence while drinking, tended
to be heavy gamblers, held low status jobs or were unemployed with erratic work
histories. All of these studies illustrate that the majority of eventual chronic offenders
were not only displaying antisocial and delinquent behaviors during childhood but
40
frequently exhibiting certain personality characteristics known to be related to criminal
behavior (Farrington, Loeber, Elliott, Hawkins, Kandel, Klein, McCord, Rowe, &
Trembley, 1990).
In addition to measures of aggressive or troublesome behavior, other studies have
explored whether specific personality types are predictive of future criminal behavior.
Specifically, one study explored the temperamental qualities, some say the origins of
personality, at age three and whether they predicted personality traits among young adults
up to age 18. The study found that children labeled as undercontrolled at age three had a
behavioral style similar to low constraint, described themselves as danger seeking,
impulsive, and were prone to negative emotions at age 18 (Silva, 1990). Another study
found a link between early temperamental style and later childhood problems. Among
both boys and girls, lack of control at age three and five was consistently and positively
related to reports of antisocial behavior at age nine and eleven and conduct disorder in
adolescence (Caspi, Henry, McGee, Moffitt, & Silva, 1995).
The research on the stability of behavior over time directly relates to the work of
Gottfredson and Hirschi. The above evidence illustrates that personality types remain
stable over time and can impact later behavior. However, are those with certain
personality characteristics destined to become future criminals? Terrie Moffitt claims
that the consistency of behavior exists only among certain individuals. Moffitt (1993)
proposes a dual taxonomy that contains two qualitatively different groups of offenders,
those who will persist in their behavior over the lifecourse and those who will limit their
criminal behavior to adolescence. The adolescent limited offender commits crime as a
rebellious act resulting from the maturing gap presented by today’s society. That is, youth
41
are physically and mentally capable of achieving adult status, however, are frustrated by
the fact that they are not given the opportunities until later in life. Their criminal behavior
is the result of this frustration. Adolescent limited offenders will stop committing crimes
once they have been given the adult role and have jobs and relationships. In contrast, the
lifecourse persisters’ antisocial behavior is stable over time, they tend to commit crimes
in a variety of situations, they tend to be aggressive, and their crimes tend to escalate over
time.
The relevance of this theory to personality rests with the antisocial persister. As
children they tend to be more impulsive and undercontrolled which may influence how
they conform to school and prosocial peers (Caspi & Silva, 1995). They tend to have
personality traits that have a negative orientation and have a difficult time controlling
their own behavior. The persisters, then, display the same personality constellation
discussed by Caspi and associates (1994). Personality could impact what Moffitt’s calls
cumulative continuity where the individual may have neurological deficits that influence
personality and their perceptions. These in turn influence cognitive ability that in turn
influences behavioral development (Moffitt, 1993). These problems interact to produce
an antisocial personality marked by impulsiveness, worry, and negativity. Or as
theorized by Eysenck (1996), the individual is less conditionable due to low cortical
arousal that interacts with personality to produce criminal behavior. Both Gottfredson and
Hirschi and Sampson and Laub recognize a group of persistent criminals exist, but they
both overlook this literature in their analysis.
Although personality may seem to “fit” with Gottfredson and Hirschi’s self
control theory, most tests of personality support Moffitt’s developmental theory. For
42
example, a study by Bartusch, Lynam, Moffitt, and Silva (1997) found that the
developmental model was supported as it represents a more comprehensive view of
behavior. Further, although self-control is similar to a lack of constraint, Caspi et al
(1994) argue that Gottfredson and Hirschi’s theory is simplistic and “crime proneness is
defined not by a single tendency (such as self control or impulsivity) but by multiple
psychological components” (p. 187).
Summary
The developmental tradition that encompasses both Gottfredson and Hirschi’s
theory of self-control and Sampson and Laub’s turning points for behavior are very
relevant to explaining delinquency and adult crime. However, both theories have not
embraced the impact of personality on behavior. Although it is unrealistic to assume that
personality is the only “cause” of behavior, the exclusion of multiple psychological
components leaves a theory of criminal behavior incomplete. This study will attempt to
determine if personality is predictive of criminal behavior over time among an adult
sample. A final group of studies pertinent to personality and criminal behavior appear in
the correctional classification literature.
Personality and Corrections
Criminological theory and corrections are distinct but inter-related areas in
criminal justice. Although many researchers may consider themselves as conducting
research in either theory or corrections, the two areas must co-exist as their contributions
are in fact reciprocal. Related to this, Andrews and Bonta (1998) assert “there are two
major empirical tests of the adequacy of a theoretical understanding of criminal behavior.
One involves the ability to predict accurately variation in criminal behavior. The second
43
involves the potential to influence criminal activity by way of deliberate interventions
that focus on the causal variables suggested by the theory” (p. 7). That is, the
development of a theory of crime allows for both: (a) an understanding of how
personality and crime are related (which then may be incorporated into risk assessments
of its likelihood of reoccurring) and (b) the formulation of correctional interventions to
address personality both as a risk and responsivity factor. The development and testing
of both provides a test of the adequacy of the theory (Andrews and Bonta, 1998).
Effective Classification
The prediction of criminal behavior is one of the most important issues in
criminal justice. Classification refers to arranging groups according to some principle
such as a risk of flight or amenability to treatment. Prior to the development of
assessment measures, classification was primarily based on intuition or gut level feelings
regarding placement decisions. Predicting who will reoffend offers guidance for
managing offenders as well as intervening to reduce the probability of future behavior.
Research indicates that if the principles of effective classification are followed, the
effectiveness of our correctional interventions is likely to increase (Andrews & Bonta,
1998). In this vein, much of the classification work in criminology has a predictive aim
and advocates assessing key factors based on the probability of an expected event
(Gottfredson, 1987).
Knowledge of the predictors of criminal behavior originates from research
utilizing longitudinal designs and meta-analysis techniques. The most common goal of
classification instruments rests with determining risk. The risk principle refers to
identifying personal attributes or circumstances predictive of future behavior (Andrews,
44
Bonta, & Hoge, 1990). Risk is important for two reasons. First, assessing an individual’s
risk for future behavior indicates to us what factors are important in explaining criminal
behavior. Second, risk can be used to match offenders with appropriate levels of service.
Research indicates that we should reserve our intensive services for our higher risk
offenders whereas minimal intervention is needed for our low risk group (Andrews &
Bonta, 1998).
There are a number of factors or correlates of criminal behavior that aid in the
prediction of future crime and in turn the explanation of behavior. The factors identified
in the literature include both factors that cannot change and those that can be influenced
through treatment. Static factors include such individual characteristics as gender, prior
record, and age. These factors, although static, are established in the literature as
important to the explanation of criminal behavior. Dynamic factors such as companions,
attitudes, personality, family relationships, education, and employment status are also
important in predicting future criminal behavior.
Often what is unknown, however, is the magnitude or importance each predictor
has in explaining future criminal behavior. In response to this problem, Andrews and
Bonta (1998) discuss the meta-analysis conducted by Gendreau, Andrews, Goggin, and
Chanteloupe (1992) to determine the predictors of recidivism. Table 3 illustrates the
factors in order of importance.
45
Table 3. Summary of key predictors of criminal behavior.
Factor Mean r # of studies examined Lower social class origin .06 97 Personal distress/ .08 226 Psychopathology Educational/vocational .12 129 Parental/family factors .18 334 Temperament/personality .21 621 Anti-social attitudes/associates .22 168 ________________________________________________________________________ The major predictors include: lower social class origin, personal
distress/psychopathology, educational/vocational achievement, parental/family factors,
temperament/personality, and antisocial attitudes and associates. The best predictors in
this study are personality and antisocial attitudes and associates. Collectively, research
indicates that the best predictors of criminal behavior, often referred to as the big four,
are antisocial associates, attitudes, personality, and criminal history (Andrews & Bonta,
1998; Gendreau, Little, and Goggin, 1996). This finding lends additional support to the
existence of a personality-crime link. Other dynamic factors considered to be major
predictors of criminal behavior include: familial factors and interpersonal relationships as
well as educational and vocational or financial achievement (Simourd and Andrews,
1994).
The second principle of effective classification refers to targeting the
criminogenic need presented by the offender in an effort to change behavior. The most
promising targets related directly to the most significant areas of risk: changing antisocial
attitudes, feelings and values, attending to skill deficiencies in the area of poor problem
46
solving skills, self management and self efficacy, and impulsivity, poor self control, and
irresponsibility (Van Voorhis, 1997).
The third principle of effective classification is responsivity. The responsivity
principle refers to delivering an intervention that is appropriate and matches “the specific
competencies, interests of learning styles that a client must posses in order to benefit
from particular types of programs” (Andrews & Bonta, 1998, p. 89). A number of
studies have found that the characteristics of the client may have an impact on treatment
(see, Andrews & Bonta, 1998). Overall, the effectiveness of correctional interventions is
dependent upon whether the program is varied based on risk, need, and responsivity
factors of the individual.
Psychological Classification Systems
In corrections, much of the work on personality and crime considers personality
as a responsivity factor. Specifically, since the 1950s, psychological classification
systems have been used to develop differential treatment and supervision plans for
offenders (Van Voorhis, 1994). This typology approach is based on the notion that
offenders are not all alike and different methods are needed to effectively deal with
different individuals (Levinson, 1988). This body of research relies on specific crime-
related personality dimensions (Andrews & Bonta, 1998). Previous research suggests
that offenders who are inappropriately placed are more likely to perform worse than those
who were appropriately placed (Palmer, 1974).
Interpersonal Maturity Level Classification. There are a number of classification
systems that could be used to distinguish personality. For the purpose of this study,
however, we will discuss Interpersonal Maturity level (I-level) classification theory and
47
its related systems. The theory was developed at University of California at Berkeley
during the 1950s (Warren, 1966). The system was originally developed to differentiate
treatment plans for juveniles under community supervision (Warren, 1966). As described
by Warren (1966) “the classification system focuses upon the ways in which the
delinquent is able to see himself and the world, especially in terms of emotions and
motivations; i.e. his ability to understand what is happening between himself and others
as well as between others” (p. 1). Specifically, the I-level theory maintains that cognitive
development involves changes in thought processes. An increase in I-level requires that
the thought processes must increase in complexity and remain stable over time. Seven
stages exist in the system and are hierarchical in nature (Warren, 1966). The levels of
interpersonal development range from least mature to a theoretically ideal stage of
interpersonal maturity rarely attained (Warren, 1966). Each level represents a stage at
which a crucial interpersonal problem must be mastered (Harris, 1988). The range of
maturity levels found in a delinquent population is from level 2 to level 5, although level
5 is infrequent.
In addition to the levels of interpersonal maturity, Warren defined nine different
personality subtypes. The types were defined for use in the Community Treatment
Project as a way to describe the juvenile population for treatment purposes. The types
included: unsocialized, aggressive/undersocialized, active (AA), unsocialized, passive
undersocialized, passive (AP), immature conformist/conformist (CFM), cultural
conformist/group-oriented (CFC), manipulator/pragmatist (MP), neurotic, acting
out/autonomy oriented (NA), neurotic, anxious introspective (NX), situational emotional
reaction/inhibited (SE), cultural identifier/adaptive (CI). According to Warren (1966) the
48
“nine subtypes then were described by lists of item definitions which characterized the
manner in which each group perceived the world, responded to the world, and were
perceived by others” (p. 3).
The Jesness Inventory. The Jesness Inventory (JI) classification system is a
revised version of Warren’s the I-level system (Jesness & Wedge, 1983; Sullivan, Grant,
& Grant, 1957; Warren et al., 1966). Jesness (1964) sought to revise earlier I-level
assessment process, which involved cumbersome and time consuming interviews. The
inventory was developed to provide a personality test that was “responsive to change,
comprehensible to youths as young as 8 years, predictive of delinquency, and
multidimensional with respect to a variety of attitudes, perceptions, and traits” (Jesness,
1964, p. 3).
The Jesness inventory consists of 155 true/false questions. An initial research
pool of 250 items were reduced to eliminate items that were nondiscriminating (Jesness,
1964). In 1972, revisions were made to make the instrument more relevant to adults and
females (Jesness, 1983) and further research illustrates it applicability to adult males
(Van Voorhis, 1994). “Initial normative and validation studies were based on a sample of
970 males delinquents and 1075 male non-delinquents between the ages of 8 and 18 and
on a sample of 450 female delinquents and non-delinquents ranging in age from 11
through 18” (Jesness, 1964, p. 5). The personality types are based on the types set forth
by Warren and associates (1961). The Jesness sample, however, did not contain any
level I-5 individuals; hence, the system contains 3 I-levels (e.g., I-2 to I-4).
The JI contains 11 personality scales and 9 subtypes scales. The nine subtypes
include the same labels as defined by Warren (1961), however, the cultural identifier type
49
differs between the two systems. The subtypes contained in the Jesness5 are as follows:
unsocialized, aggressive/undersocialized, active (AA), unsocialized, passive
undersocialized, passive (AP), immature conformist/conformist (CFM), cultural
conformist/group-oriented (CFC), manipulator/pragmatist (MP), neurotic, acting
out/autonomy oriented (NA), neurotic, anxious introspective (NX), situational emotional
reaction/inhibited (SE), cultural identifier/adaptive (CI) (Jesness, 1964). Tables 4 and 5
provide a description of each I-level and type.
Although the system was developed to differentiate treatment plans for offenders,
researchers have also explored the relationship to criminal behavior. Behavior. Palmer
(1974) grouped youths into three categories based on Warren’s subtypes that includes:
5 The eleven personality scales include: social maladjustment (e.g., exhibiting attitudes
associated with inadequate or disturbed socialization), valued orientation (e.g., exhibiting
tendency to share attitudes of persons in lower socioeconomic classes), immaturity (e.g., tendency
to display attitudes consistent with persons of younger age), autism (e.g., tendency to distort
reality according to personal desires), alienation (e.g., presence of distrust in others, especially
those representing authority), manifest aggression (e.g., tendency to react in response to
frustration or anger without control), withdrawal-depression (e.g., indicates extent of individual’s
dissatisfaction with him- or herself and others, and a tendency toward isolation), social anxiety
(e.g., exhibiting feelings of anxiety in interpersonal situations), repression (e.g., exclusion of
conscious awareness of feelings or failure to label emotions), denial (e.g., reluctance to
acknowledge unpleasant events or conditions of daily life), asocial (e.g., reflects a generalized
disposition to resolve problems in a way that shows disregard for social customs or rules).
50
passive conformist (Cfm’s), power oriented (comprised of Cfc’s and Mp’s), and
Neurotics (comprised of Na’s and Nx’s) to determine if juveniles with certain profiles
should remain in an intensive treatment program in the community. He found that all
groups performed well in the community if given intensive treatment services. However,
the likelihood of criminal behavior among the power oriented and passive conformist
group increased in direct relationship to the duration of the follow-up period (Palmer,
1974).
Finally, a study by Van Voorhis (1994) among adults found that the Jesness
Inventory types could be grouped into four categories: aggressives (comprised of Aa’s,
Cfc’s, Mp’s), neurotics (comprised of Na’s and Nx’s), dependents (comprised of Ap’s
and Cfm’s) and situationals (comprised of Se’s and Ci’s). Table 6 illustrates the
description of each collapsed type. The study was designed to determine, among other
things, whether the Jesness Inventory subtypes could predict disciplinary and treatment
related outcomes. The study concluded that aggressives were more likely to be cited for
disciplinary infractions in the prison camp, report aggressive behaviors in the prison
camp and penitentiary, and be evaluated unfavorably by staff in the penitentiary (Van
Voorhis, 1994). Moreover, the neurotic type in the penitentiary scored high on a self-
report aggression scale in both a penitentiary and prison camp samples. Van Voorhis
(1994) asserts that neurotic acting out inmates showed less stress but more problems with
emotional control and learning. Moreover, the neurotic type was more likely than the
aggressive or situational type to report higher scores on the Center for Epidemiological
Studies Depression Scale (CEDS) (Radloff, 1977). In addition, the study found that only
30 percent of the participants perceived others as willing to help them. The findings were
51
Table 4. Summary of Interpersonal maturity levels (Warren, 1966; Jesness, 1988).
______________________________________________________________________________________________________ Construct Definition Relationship to Criminal Behavior I-level 2 Tend to feel world should take Delinquency may result from poor care of him/her. Unable to impulse control or inability to cope understand or predict behavior with external pressures from others. Tends to be impulsive and unaware of effects of behavior on others; and sees others as barriers to
his or her satisfaction. I-level 3 Attempts to manipulate the Delinquency may result in an attempt environment in his/her favor. The to gain peer approval, gratification of manipulation may take the form of material needs, or an attempt to gain conforming behavior. They understand control in a situation via a “bad guy” how behavior is affected by role. relationships, but make no emotional investment. Unable to empathize and typically engage in only short term planning. I-level 4 Internalizes a set of standards by May respond to emotional conflict by which he/she judges one self acting out. This conflict may lead to internalizing and others. Feels guilt when not criminal standards and value systems. measuring up to standards or feels conflict when admiring others behavior
52
Table 5. Summary of key personality constructs for the Jesness Inventory (Jesness, 1962).
______________________________________________________________________________________________________ Construct Definition Relationship to Criminal Behavior Unsocialized aggressive (Aa) Displays negative attitudes toward High self reported criminal behavior convention. Behavior is unpredictable aggressive and antisocial Unsocialized passive (Ap) Displays negative attitudes toward Probability of criminal behavior average convention. Behaves in inappropriate and often bizarre ways. Negative self concept. Immature conformist (Cfm) Displays positive attitudes toward Low self reported criminal behavior. Peers convention. Behavior is conforming may be especially important explanation of and dependent (i.e. follower). criminal involvement. Cultural conformist (Cfc) Displays low motivation, negative High self reported criminal behavior. Above attitudes toward convention. Feels average violent activity alienated and hostile towards authority. Manipulator/pragmatist (Mp) Shows positive attitudes toward Low self reported criminal behavior but convention. Behaves in manipulative official records indicate high probability. way and obtrusive ways. Especially prone to property crimes
53
Table 5. Summary of key personality constructs for the Jesness Inventory, continued. ______________________________________________________________________________________________________ Construct Definition Relationship to Criminal Behavior Neurotic acting out (Na) Displays negative attitudes toward High self reported criminal behavior. convention. Experience conflicts in More apt to use drugs. relationships. See themselves as somewhat cynical and disenchanted, often exhibiting outspoken and non-conforming behavior Neurotic anxious (Nx) Mostly positive attitudes toward convention. Low self reported criminal behavior conforming but also dependent, anxious, and insecure. Do not have criminal orientation. Situational emotional reaction (Se) Positive attitudes toward convention. They are Low reports of criminal behavior, confidant, but naïve, rigid, and conforming. Expects to be caught if he/she breaks the law Cultural identified/adaptive (Ci) High verbal aptitude and positive toward Low reports of criminal behavior. convention. Maintains good interpersonal relationships. This type differs from Warren’s cultural identifier. Warren defines this type as having an internalized delinquent value system
54
Table 6. Summary of key personality constructs for the collapsed Jesness Inventory types (Van Voorhis, 1994).
______________________________________________________________________________________________________ Construct Definition Relationship to Criminal Behavior Aggressives Tend to be manipulative. Behavior High probability of criminal behavior. (Aa, Cfc, Mp) is unpredictable and negative. Feels alienated and hostile and has antisocial values. Not likely to possess prosocial values. Neurotics Tend to be anxious and insecure. High criminal probability when unable. (Na and Nx) Tend to be cynical, hostile, and to cope with anxiety. act inappropriately when anxious6. Dependents Although behavior may be Criminal behavior less likely than other types (Ap and Cfm) conforming, he or she tends When criminal behavior does occur it may be to follow others, including criminals the result of the influence of others and their This type is less clearly defined among limited cognitive functioning.
adults than among juveniles. Situationals Tend to view convention positively. Criminal behavior less likely, situational (Se and Ci) Tend to maintain good relationships if it does occur. but can be naïve and rigid
6 Some studies have found value in separating the Na and Nx categories (Van Voorhis, 1994).
55
expectedly mixed for the situational group, however, the dependent group was likely to
be evaluated unfavorably by staff. Finally, with regard to I-level, the results were often
not significant, however, when significant results were found it was the I-2 and I-3
inmates receiving more unfavorable results (Van Voorhis, 1994).
Summary
Classification in corrections has been evolving since the early 1900s. Recent
classification systems are increasingly comprehensive and have multiple goals. The
psychological systems designed to differentiate offenders for treatment have been in use
since the 1950s. It is widely accepted that differential treatment matching increases the
opportunity for interventions to be successful. Combining the importance of each
system, the proposed study will explore how well a psychological system such as the
Jesness Inventory predicts recidivism, but also how important personality is to predicting
risk. In order to accomplish these goals, a framework must be established.
An Integration of Personality Types
It is theorized that the Jesness Inventory can be used to mirror previous research
regarding the personality crime link. The extensive review of the relationship between
personality and crime as measured by the EPQ, MPQ, and Jesness Inventory have been
included to illustrate the importance of personality theory to criminology. As mentioned
in Chapter 1, the personality classifications produced by the Jesness Inventory will be
used to assess the relationship between personality and recidivism. The studies using the
EPQ and MPQ are of causation and the samples often include offenders and non-
offenders. However, the Jesness Inventory was developed to distinguish among a
criminal population and is more relevant to the exploration of recidivism. The types and
56
levels set forth by the JI and subsequent researchers (Van Voorhis, 1994) will be
analyzed in such as way as to mirror those types produced by both the EPQ and MPQ.
Table 7 details the relationships proposed between the various instruments.
Relationship Between Types
EPQ. The Jesness I-level types are more difficult to link to the types defined by
the EPQ and MPQ. However, I-2 and I-3 levels are more applicable than the I-4. The I-2
and I-3 three types most closely mirror the extraversion type set forth by Eysenck as they
show a tendency to act impulsively. It is important to note, however, that impulsivity is
not necessarily attributable to extraversion in all situations. There is no clear link
between the extraversion type and either the Jesness personality types or the collapsed
types set forth by Van Voorhis (1994). However, the two neurotic categories (i.e., Na’s
and Nx’s) correspond to his type neuroticism. Similarly, the collapsed neurotic type
discussed by Van Voorhis is also applicable. We see this in the descriptions set forth by
all three authors. Specifically, the type tends to be anxious and insecure, often irrational
and tense with emotions overriding reason. With regard to psychoticism, the Jesness
types Aa, Cfc, and Mp corresponds most closely. Moreover, the aggressive type set forth
by Van Voorhis (1994) best fits with the type of individual who is unempathetic, cold,
impersonal, and exhibits impulsive tendencies.
MPQ. With regard to the types defined by the MPQ, it appears the Se and the Ci
on the Jesness Inventory most closely resembles the type referred to as positive
emotionality given that Jesness found minimal occurrence of delinquency in these types.
The situational type defined by Van Voorhis (1994) mirrors this group. Again these
individuals tend to view convention positively and maintain good relationships but can
57
Table 7. Relationship between EPQ, MPQ and Jesness I-level, personality, and collapsed types.
EPQ/MPQ types Jesness I-level Jesness Personality Types Collapsed Jesness type Extroversion --- impulsivity not clearly tied to extraversion (Eysenck, 1952) Neuroticism --- Na & Nx Neurotic (Eysenck, 1952) Psychoticism --- Aa, Cfc, & Mp Aggressive (Eysenck, 1952) Positive Emotionality --- Not a criminal type Situational7 (Capsi and Silva, 1990) Negative Emotionality --- Na Neurotic (Caspi and Silva, 1990) Constraint --- Se & Ci Situational (Caspi and Siliva, 1990) Lack of Constraint I-2 & I-3 Ap, Cfm, Aa, Mp, & Cfc Aggressives/Dependents (Caspi and Silva, 1990) --- = no clear relationship 7 Although situational is defined as a criminal type, this type most closely reflects type positive emotionality.
58
Table 7. Relationship between EPQ, MPQ and Jesness I-level, personality, and collapsed types, continued. EPQ/MPQ types Jesness I-level Jesness Personality Types Collapsed Jesness type Negative Emotionality/ I-4 Na Neurotic Lack of Constraint Extraversion/ I-4 Na Neurotic Neuroticism Extraversion/ --- Aa, Cfc, Mp Aggressive Psychoticism Neuroticism/ I-4 Na Neurotic Psychoticism Extraversion/Neuroticism/ --- Aa, Cfc, Mp Aggressive Psychoticism
59
exhibit the tendency to be naïve. The negative emotionality type defined by the MPQ
most closely resembles the Na type defined by Jesness and the neurotic group defined by
Van Voorhis (1994). Negative emotionality seems to represent a slightly different type
than the neurotic category defined by Eysenck. The type negative emotionality tends to
display anger, irritability, and perceives life in a very threatening and hostile way. These
traits are more similar to the Na than the Nx category set forth by the Jesness. The Na
type also displays negative and hostile attitudes toward convention and relationship.
Finally, individuals low on the type constraint most closely resemble the I-2 and I-3 level
and the Se and Ci type set forth by the Jesness. Again individuals with high levels of
constraint are less likely to be involved in crime, tend to endorse social norms and be
cautious. This also most closely resembles the situational type set forth by Van Voorhis.
The individual who exhibits a lack of constraint however, would more closely resemble
the Ap, Aa, Cfm, Mp, and Cfc types set forth by Jesness. These are individuals who
typically display negative attitudes toward convention. These individuals are both
aggressive and feel alienated and can become hostile and manipulative. The types
aggressive and dependent most closely mirror this type as well.
Combined Types. The combined types set forth by the EPQ and the MPQ
represent high-risk cases. That is, those individuals who are high on negative
emotionality and low on constraint are more likely to be involved in criminal behavior.
Likewise, it can be theorized that those individuals who exhibit extraversion paired with
neuroticism may also be more likely to commit crimes. The analysis will center on each
type separately and in tandem to determine the relationship between personality and
recidivism.
60
Conclusions
The purpose of this review is not to disparage sociological theories of crime.
Rather, to formulate an argument that personality deserves recognition within
criminological discourse. Research has not progressed to the point of identifying and
organizing how personality interacts with many other social variables. Moreover,
although personality is discussed tangentially in lifecourse theories and is actually used in
correctional typologies as a consideration for treatment and sometimes placement,
research has not progressed to detailing how personality might act as a predictor of
criminal behavior. Hence, the current study hopes to address two areas. First, this study
is primarily concerned with adding to the current literature on personality and crime.
Second, although personality is recognized as a risk factor (Andrews & Bonta, 1998),
whether personality can predict future criminal acts of adult males over an extended
period of time remains to be explored.
61
CHAPTER 3
Method
This study rests on both theoretical and empirical assumptions. As indicated in the
literature review the research pertaining to personality and criminal behavior is limited.
This study extends previous research by using a corrections-based personality measure
and a long-term follow-up period. This chapter will discuss the design, sample
characteristics, data sources, the various independent and dependent variables, and the
data analysis plan.
Design
In order to assess the relationship between personality and recidivism and the
degree of importance personality has in conjunction with traditional risk factors, the
current study utilizes a longitudinal design. Specifically, the study reports outcomes of a
sample of federal inmates over a 10 to 12 year follow-up period. A number of research
questions will be addressed. First, the study will explore whether certain personality types
are related to criminal behavior. Second, the relationship between personality and
persistence as outlined in the lifecourse and developmental research will be explored.
Finally, the study will explore whether individuals with certain personality profiles have
a propensity to offend in specific ways.
Sample
The original (Time 1) sample is comprised of inmates admitted to either a Federal
Penitentiary or a Federal Prison Camp between September 1986 and July 1988. The
individuals were selected as part of an NIJ funded classification study conducted in Terre
62
Haute, Indiana (Van Voorhis, 1994). The penitentiary is designated level 4/5 on the
Federal Bureau of Prisons security continuum. The facility is considered a low-
maximum security or high medium security facility. The prison camp is a minimum
security or Level 1 facility in the federal system. The groups will be combined in the
analysis to expand the sample size and represent a more diverse group of offenders8.
Given that inmates’ psychological states are often affected by imprisonment
(Bukstel & Kilman, 1980), the Time 1 sample was limited to those who recently began
their incarceration experience. The sample included only new inmates received at the
facility immediately after sentencing. Additional selection criteria excluded those who
were non-English speaking (n=80) or could not read or were expecting to be released
within four months after admission (n=26). Participation in the original study was
voluntary. In addition to those excluded above, it should be noted that some new intakes
did not participate in the Time 1 study for a variety of reasons, including: (a) refusal to
participate (n=97), (b) a repeated failure to respond to “call outs” (n=12), (c) being
unavailable during the first month of their sentence (e.g., out of writ, in lock-up) (n=13),
and (d) not being contacted by the research staff (n=168). The last situation occurred
when there were too many admissions for program staff to contact all potential
participants. In this situation, staff members were instructed to select inmates at random
in order to avoid biasing the selection process. The overall response rate was 76 percent
for the penitentiary group and 85 percent for the prison camp group. This resulted in an
overall sample of 369 individuals (179 from the penitentiary and 190 from the prison
8 Group membership (institution vs. prison camp) will be included as a control variable in the analysis
63
camp).
Attrition Analysis. Both data from Time 1 and Time 2 were utilized in the
current study. Demographic and Jesness Inventory data from Time 1 comprise the
independent variables used in the multivariate analysis. Jesness Inventory data were
available for approximately 88 percent of the Time 1 sample. Time 2 recidivism data
were used to create the dependent variables used in the analysis. However, some attrition
occurred when social security numbers did not produce an NCIC record. Specifically,
recidivism data were available for approximately 85 percent of the Time 1 sample. Once
the sample was limited to only those with a NCIC record and Jesness Inventory data, the
overall attrition rate for the Time 2 sample was 25 percent. Tables 1 and 2 in Appendix
A illustrate that there are no significant differences between the Time 1 sample (N=369)
and the current Time 2 sample limited to those with recidivism and Jesness Inventory
(N=277) data.
Tables 8 and 9 describe the current sample. With regard to race, 66 percent were
Caucasian and 28 percent were African American. The remaining 6 percent were
members of American Indian, Hispanic, or Asian ethnic groups. The mean age of the
sample is 34 years of age (median age 32). Marital status was mixed among the group as
40 percent classified themselves as married, 23 percent single, and 23 percent divorced.
Sixty-four percent of the males in the sample had a least one child. With regard to
education, 33 percent of the sample did not complete high school and 47 percent reported
they had experienced school failure. Finally, 47 percent were unemployed at the time of
the arrest that led to their Time 1 incarceration.
The majority of the participants had a prior record. Specifically, 87 percent of
64
Table 8. Percentage and frequency distribution of participants’ social demographic characteristics. ________________________________________________________________________ Characteristic N % ________________________________________________________________________ Race White 184 66.4 Black 77 27.8 American Indian 6 2.2 Hispanic 8 2.9 Asian 2 0.7 Age at admission 19 to 29 90 32.7 30 to 45 154 56.0 46 and older 31 11.3 Mean 33.5 Median 32 Marital Status Married 109 39.9 Never married 66 23.1 Divorced 66 23.1 Separated 16 5.9 Widowed 1 0.4 Common-law 21 7.7 Number of Dependents None 98 36.3 One 57 27.1 Two 44 16.3 Three 42 15.6 Four 17 6.3 Five 6 2.2 Six or more 6 2.8
65
Table 8. Percentage and frequency distribution of participants’ social demographic characteristics. ________________________________________________________________________ Characteristic N % ________________________________________________________________________ Education 6 to 11 years 91 33.3 High school 81 29.7 GED 46 16.8 Some post high school 31 11.4 College graduate 20 7.3 Some post college 4 1.5 Evidence of school failure Yes 125 47.2 No 140 52.8 Primary Occupation No occupation 26 9.6 Professional 19 7.0 Manager/Admin 27 10.0 Sales 13 4.8 Clerical 2 0.7 Craftsman 29 10.7 Equipment Operator 9 0.4 Laborer 105 38.9 Farmer 6 2.2 Service worker 2 0.7 Armed Services 3 1.1 Student 2 0.7 Househusband 2 0.8 Criminal occupation 24 8.9 Employment Status Not working 127 47.0 Full time 100 37.0 Occasionally 33 12.2 Status Unknown 10 3.7
66
Table 9. Percentage and frequency distribution of participants’ prior and current offense record.
________________________________________________________________________ Characteristic N % ________________________________________________________________________ Prior adult or juvenile record Yes 242 87.4 No 35 12.6 Prior adult arrests Yes 240 86.6 No 37 13.4 Prior adult convictions Yes 220 79.7 No 55 19.9 Conflicting evidence 1 0.4 Prior prison sentence Yes 108 44.8 No 133 55.2 Amount of prior time served Two years or less 42 30.4 25 mo. to 5 years 33 23.9 61 mo. to 10 years 13 8.7 More than 10 years 7 5.1 Time served unknown 44 31.9 Prior arrest dealing drugs Yes 66 27.6 No 173 72.4 Prior professional criminal activity Yes 65 27.2 No 174 72.8 Prior violence Yes 104 43.7 No 134 56.3
67
Table 9. Percentage and frequency distribution of participants’ prior and current offense record, continued. ________________________________________________________________________ Characteristic N % ________________________________________________________________________ Prior arrest for same offense As present offense Yes 60 25.0 No 180 75.0 Prior or current drug related offense Yes, prior 26 9.7 Yes, current 42 15.7 Yes, both 47 17.5 No 153 57.1 Prior or current alcohol offense Yes, prior 54 19.8 Yes, current 11 4.0 Yes, both 21 7.7 No 187 68.5 Prior prison escapes Yes 22 16.2 No 114 83.8 Prior probation revocations Yes 50 35.0 No 89 65.0 Prior parole revocations Yes 36 10.5 No 45 55.6 Prior record as a juvenile Yes 81 30.5 No 185 69.5 ________________________________________________________________________
68
Table 9. Percentage and frequency distribution of participants’ prior and current offense record, continued. ________________________________________________________________________ Characteristic N % ________________________________________________________________________ Prior incarceration as a juvenile Yes 43 58.1 No 31 41.9 Prior violent record as juvenile Yes 19 25.7
No 55 74.3
________________________________________________________________________
69
the participants reported a prior record (either adult or juvenile) and 45 percent served at
least one prior prison sentence. Twenty-eight percent had a prior record that involved
drug dealing and 44 percent a record of violence. Prior records also involved: (a) prison
escapes (16%), (b) probation revocations (35%), (c) parole revocations (11%), (d) a
juvenile record (31%), (e) incarceration history as a juvenile (58%), and (f) violent
behavior as a juvenile (26%).
Data Collection
The data for the present study were obtained from three sources. First, social,
demographic, and criminal record background data were obtained through interviews and
the review of PSI’s by the research staff working on the original project. For the current
study, Time 1 data were utilized to create the Salient Factor Score (SFS81) to account for
risk of reoffending. Second, the Jesness Inventory was administered by research staff at
Time 1. Test results were sent to Consulting Psychologists Press in Palo Alto, California
for computerized scoring. Third, recidivism data were collected in November 1998
(Time 2) by a pretrial services agency through the use of the National Crime Information
Center (NCIC) database. As of 1998, 36 states provided criminal record data to this
national database designed to provide states with criminal history information. As of
February 1, 1998, there were 27,368,119 subjects on file.
Measurement of Research Variables
Independent Variables
A variety of independent variables or factors are explored in relationship to
recidivism. This analysis provides a measure of the overall importance of each of the
variables with specific attention paid to the relative importance of personality as a
70
predictor. All of the variables and items used in the analysis have been previously
discussed in the literature as factors related to recidivism. Four independent variables are
included in the multivariate analysis.
Personality. Personality is considered one of the four main risk factors discussed
in the classification literature (see Andrews and Bonta, 1998) and as important in the
emerging criminological literature (Caspi et al., 1994; Krueger et al., 1994; 1996). The
Jesness Inventory is a paper-and-pencil test that contains 155 true-false items that yield
scores on eleven personality scales and 9 subtype scales (see description in Chapter 2).
Research for the Jesness Inventory spans over 50 years. The original sample of 1,862
used by Carl Jesness included approximately equal numbers of delinquent and non-
delinquent males ranging in age from 10 to 18. The mean odd-even item correlations
found was .73 with correlations ranging from .62 to .88 (Jesness, 1962). Moreover,
according to Jesness (1962), the “research shows the JI leads to correct classification and
prediction a large percentage of the time. This research supports the discriminate validity
of the JI” (p. 27). More recent research adds considerable support for the construct
validity of the JI (Jesness, 1988). Studies both with delinquents and non-delinquents
found that the subtypes could predict different types of behavior (Jesness & Wedge,
1982; Jesness, 1986). Finally, one study found differences across subtypes in the average
number of probation referrals, and the highest proportion of delinquency was found
among those classified as CFC, NA, and AP (Jesness, 1988). The four types used in the
present study were successfully tested for construct and predictive validity in the original
study (Van Voorhis, 1994).
Besides the established validity and reliability of the Jesness Inventory, its use in
71
the current study has several additional advantages. First, the Jesness Inventory was
developed out of the work of Warren on I-level theory (Warren, 1966). The Jesness
Inventory, however, is shorter and more reliable than the lengthier I-level assessment.
Moreover, Van Voorhis (1994) found that the Jesness Inventory subtypes, with one
exception9, are in fact comparable to the types produced by the more complex I-level
assessment developed by Warren and associates. Second, the Jesness Inventory was
developed to provide a personality classification that was relevant to criminal
populations. Unlike the MMPI or the California Psychological Inventory that were not
developed with the purpose of distinguishing types of delinquents, the JI was originally
developed to evaluate treatment for young male delinquents (Jesness, 1964). Third, as
previously discussed, the correctional classification findings on personality types are
discussed in relationship to treatment amenability but not criminological theory. By using
a classification system traditionally used in corrections and testing its applicability in the
area of prediction, it becomes relevant for both criminological theory and corrections.
Table 10 illustrates the distribution of the sample by the Jesness Inventory personality
types and I-levels. This study employs the four collapsed types shown at the bottom of
Table 10. As noted in Chapter 2, the nine types are grouped into four categories:
aggressives (comprised of Aa’s, Cfc’s, Mp’s), neurotics (comprised of Na’s and Nx’s),
dependents (comprised of Ap’s and Cfm’s) and situationals (comprised of Se’s and Ci’s).
Salient Factor Score (81). Prior record and the current offense variables have been
used in risk assessment measures since the late 1920s with the advent of the Burgess
score (Burgess, 1928). To measure risk, the present study utilizes a modified version of
9 The Ci and level 5
72
Table 10. Percentage and frequency distribution of participants’ I-level and personality types.
_______________________________________________________________________ Characteristic N % ________________________________________________________________________ Jesness I-Level Types I – 2 12 4.3 I – 3 167 60.3 I – 4 98 35.4 Jesness Personality Types Unsocialized Aggressive (AA) 5 1.8 Unsocialized Passive (AP) 7 2.5 Immature Conformist (CFM) 37 13.4 Cultural Conformist (CFC) 67 24.3 Manipulator/Pragmatist (MP) 63 22.8 Neurotic Acting Out (NA) 30 10.9 Neurotic Anxious (NX) 21 7.6 Situational/Inhibited (SE) 38 13.8 Cultural Identifier (CI) 8 2.9 Collapsed Jesness Types Aggressive 135 48.9 (AA, CFC, MP) Neurotic 51 18.5 (NA & NX) Dependent 44 15.9 (AP & CFM) Situational 46 16.7 (SE & CI) ________________________________________________________________________
73
the Salient Factor Score (SFS). The SFS was designed as an assessment device used to
predict the likelihood of recidivism among federal parolees (Hoffman and Beck, 1985).
The most recent version of the SFS, the SFS(81), has been in use by the United States
Parole Commission since 1981 (Hoffman, 1994). This assessment instrument “contains
six items that, applied together, produce a score from zero to ten points. The higher the
score the lower the likelihood of recidivism” (Hoffman & Beck, 1985, p. 502). Research
indicates that the SFS was predictive of recidivism in a 5-year follow-up study (Hoffman
and Beck, 1985). In addition, using three different samples ofparolees, another study
found that the assessment instrument retained its predictive accuracy over a 17-year
follow-up period (Hoffman, 1994). Finally, the SFS was also used successfully to predict
offending in a recent evaluation study of parolees in Georgia (Van Voorhis, Spruance,
Johnson Listwan, Ritchey, Pealer, & Seabrook, 2001).
The six items contained in the SFS (81) scale are: prior
convictions/adjudications; prior commitments of more than thirty days (adult or juvenile),
age at current offense; recent commitment free period; whether the offender is currently a
probation and parole violator; and history of heroin or opiate dependence. The risk
categories are defined as: 10-8 very good risk, 7-6 good risk, 5-4 fair risk, and 3-0 as poor
risk. The SFS risk calculation was modified for the current study given data on recent
commitment free periods and a history of opiate or heroin use were not available.
Commitment free periods were replaced with a calculation of whether the offender had a
prior record of probation or parole violations. History of heroin or opiate use was
replaced with a history of drug dependence. In the current study the SFS categories were
significantly correlated with the arrest at the .01 level (r = -.342).
74
Table 11 illustrates the risk scores for the current sample. Thirty-three percent of
the sample scored as very good, 34 percent as good, 27 percent as fair, and 7 percent as
poor. Collapsed, it appears that 66 percent of the sample would be considered low risk
and 34 as high risk.
Race. Race is prominent in many criminological theories of crime ranging from
individual level explanations to macro-level theories. However, the impact of race on
criminal behavior is multifaceted. Race by itself does not provide us with an explanation
of crime, but rather is highly related to other risk factors of crime such as social class,
age, and gender (Hindelang, 1981). It is important to note, given the sample was selected
in the Federal system, the majority of the participants are Caucasian (66%).
Group Membership. The two groups, federal penitentiary and prison camp, were
combined to create the sample used in this analysis. Group is entered as a dummy
variable into the multivariate analysis to determine whether membership in the group
influences outcome.
Dependent Variables
The impact of personality on crime is assessed by examining recidivism rates
among the sample of federal penitentiary and prison camp inmates. The data collected
through NCIC reveal how many individuals in the sample recidivate, the frequency by
which they do so, and the type of offending behavior. Table 12 frequency distribution of
the participants arrest record for each of the dependent variables. By providing a long-
term follow up period, the current study allows for an in-depth examination of these
factors.
75
Table 11. Percentage and frequency distribution of participants’ Salient Factor Scores.
________________________________________________________________________
Characteristic N % ________________________________________________________________________ Risk Score (SFS) Very Good 91 32.9 Good 93 33.6 Fair 74 26.7 Poor 19 6.9 Risk Score Collapsed Low 184 66.4 High 93 33.6
________________________________________________________________________
76
The first measure of recidivism is any arrest10. The data collected includes both
whether the defendant was arrested as well as the arrest date. Maltz (1984) asserts that
arrest is more useful than conviction data because it is closer temporally to the crime and
is less subject to court processing biases. The second measure of recidivism is based on
the number of times the individual was arrested. The third measure is based on the type
of offense that the offender was charged with after release from either the penitentiary or
prison camp.
Data Analysis
The analysis will produce three models predicting different aspects of offending.
All three models will examine the relationship between the Jesness Inventory I-level and
personality classifications and outcome. The three outcome events are: (1) arrest, (2)
frequency of arrest, and (3) nature of the offense.
Model #1: Personality and Any New Arrest
In the first model, discrete time event history analysis was utilized to explore the
relationship between personality, I-level, and any new arrest. Event history analysis is
used as a means to study events by modeling the probability or likelihood of the event
occurring (Allison, 1995). Event history analysis is an analysis of changes over time. In
traditional statistical technique such as logistic regression, the analysis of change with
time simply inserted as a control variable will yield biased estimates. Specifically, “they
can severely misrepresent the structure of the process since they provide descriptions of
the associations over time among variables but offer no possibility of identifying the
parameters of the underlying processes that produce these associations” (Palloni and
10 The only contact that is excluded is non-DUI traffic offenses.
77
Table 12. Percentage and frequency distribution of participants’ recidivism rates (Time 2 data).
________________________________________________________________________
Characteristic N % ________________________________________________________________________ Arrested new offense Yes 145 52.3 No 132 47.7 Arrested Multiple Times Yes 85 30.0 No 191 69.0 Arrest Charge Drugs 34 23.5
Property 42 29.0 Violence 25 17.2 Other 44 30.3
78
Sorensen, 1990, p. 292). By using this technique, this analysis will distinguish whether
those who fail (i.e. incur an arrest) differ by personality or I-level type while controlling
for time at risk.
To accurately estimate the probabilities of an event, in this case arrest, the
analysis must deal with the fact that subjects are at risk for differing time periods of time.
Given the sample includes inmates entering prison at different periods and the
individuals are arrested at differing rates, it would be inaccurate to treat the length of their
histories at risk as equivalent. Specifically, incarceration dates for the adults in the study
varied from September 1986 to July 1988. Record checks were conducted in November
of 1998; hence the follow-up period varied from 10 to 12 years. Because inmates can be
arrested while in prison, the follow-up period began at the beginning of their
incarceration in the original study11. The record check data revealed that 25 inmates,
17.4% of those rearrested at least once, were charged with offenses while in prison. The
charges incurred while in prison include: drugs (40%), attempted escape (24%), property
(12%), weapons (8%), unspecified violations (8%), fraud (4%), and sexual misconduct
(4%).
Concepts such as the risk set, person periods, and the hazard rate are key to event
history analysis (Allison, 1984). The risk set refers to the number of individuals who are
at risk at each time period. The follow-up period is collapsed into years at risk after
analyses indicated that no one month was significantly different from another in
predicting outcome. In other words, overall there does not appear to be a tendency for
11 Data on the exact release date were not available on all participants. Projected release dates were available for 64% of the sample.
79
the arrest rate to significantly increase or decrease in any one month. Everyone is at risk
in year one (n=277). In each subsequent year, the number arrested were taken out of the
sample that remain at risk. As can be seen in Table 13, 16 individuals were arrested in
year one and are then no longer at risk in the remaining 11 years. Logically, the number
at risk declines with each year. One person died while in prison during year two and was
subsequently removed from the model. In addition, the follow-up period varied for those
incarcerated in 1987 and 1988 (i.e., they were at risk for shorter periods of time), hence
three additional individuals not arrested by the end of their follow-up period were taken
out of the at-risk pool in year 10 and 69 were taken out in year 11.
The second important concept in event history analysis is the creation of the
person period. For each period of time at risk (e.g., each year), a record is created for that
case or individual. This is known as the person period and it corresponds to the number
of years the person is at risk. The person arrested in the fourth year of the follow-up
would be given four person periods. The person arrested in the eight year would have
eight person periods. Those never arrested are considered right censored cases. In other
words, “right censoring occurs when the waiting time for the occurrence of an event is
longer than the period of observation” (Palloni and Sorenson, 1990, p. 304). They are
assigned the number of person periods that corresponds with the highest number of years
at risk. The follow-up period varies between ten and twelve years depending on when the
case entered prison, hence if the person was not arrested at all during our follow-up
period they were assigned either 10, 11, or 12 person periods. We do not want to
conclude that the censored cases will never incur another arrest, rather all we
80
Table 13. Distribution by Year of Parolees at Risk (Model 1)
_______________________________________________________________________ Number Number Estimated Year Arrested at Risk Hazard Rate 1 18 277 .0649 2* 19 259 .0733 3 12 239 .0502 4 19 227 .0837 5 17 208 .0817 6 16 191 .0837 7 12 175 .0686 8 8 163 .0491 9 5 155 .0323 10** 6 150 .0400 11** 10 141 .0709 12 3 64 .0468 Total 145 2249 * One person died during the follow-up period and was excluded after year 2 **The follow-up period varied from 10 to 12 years. Of the 277, 3 only had a 10 year follow up and were taken out of the number possibly at risk and 67 had a 11 year follow up and were also censored at that time.
81
can say is that during our follow-up period an arrest was not incurred by these
individuals. Yet the need to “analyze simultaneously the data from individuals with
censored and noncensored events times is apparent because the former are a key group of
people: those least likely to experience the event” (Willett and Singer, 1997, p. 273). As
a result, for the sample of 277, there were 2268 person periods created. As seen in Table
13, this is just the sum of the number at risk in each of the 12 years. For each person
period, the dependent variable is coded a 1 if they are arrested and a 0 if not arrested.
The third concept that is important in event history analysis is the hazard rate. As
indicated by Allison (1984) “the hazard rate is the probability that an event will happen at
a particular time to a particular individual, given that individual is at risk” (p. 16). In the
present study, the hazard rate is the probability of an arrest within a particular year for
those who have not been arrested. To get the hazard rate, we divide the number arrested
by the total number still at risk during a particular year. Table 13 lists the hazard rates by
year. In addition to the hazard rate, it is useful to display the proportion of the sample
that fail through each period. Because there are individuals who never experience an
arrest during the follow-up period, the curves will never reach 100 percent. The rate of
failure over time will be displayed in the results section through failure curves.
Multivariate analyses were conducted by estimating a logistic regression model in
which arrest was regressed simultaneously on personality while controlling for the time
invariant variables: group (penitentiary vs. prison camp), race, time, and risk score.
Participants who do not recidivate are considered censored but their survival times are
included in the denominator, which is used to evaluate the failure rate. The multivariate
approach has two advantages in this case: it isolates the effects of personality on
82
recidivism while controlling for other variables and illustrates the significant predictors
of recidivism.
Model #2: Personality and Frequency of Arrest
In the second model, repeated event history analysis was utilized to explore the
relationship between personality, I-level, and persistence. This second model explored
the relationship between personality and persistence. In other words, whether those
individuals arrested multiple times are likely to have certain personality profiles. This
analysis also included race, risk, time, and group as covariates with personality; however,
risk now becomes a time varying instead of a time-invariant predictor. It was mentioned
in earlier chapters that the salient factor score was chosen because it consisted of static
factors or those that cannot change. However, the overall risk score can increase over
time as individuals incur additional arrests. This analysis takes into account this
possibility and recalculates the risk score as needed.
In the first model once the person was arrested they were removed from the
analysis of subsequent years. Now the occurrence of an arrest does not remove
individuals from the risk set given they can experience multiple arrests during the
remaining follow-up periods. The only qualitative difference between this second model
and the first is in the definition of time; now the probability of an arrest refers to the
“duration since the prior occurrence rather than to chronological time elapsed since the
origin of the process” (Palloni and Sorensen, 1990, p. 303). Each arrest is treated
separately and the duration is measured since the date of the last arrest. The question
becomes whether the occurrence of one event or arrest predicts the occurrence of another
event or arrest (Willett and Singer, 1997). For example, if an individual is arrested in
83
year one and then again in year three and finally in year eight, all three of those arrests
are accounted for in the calculation of the regression model predicting outcome.
The formulation of the person period database remained the same as in model 1,
however, instead of creating person periods based on the amount of time to elapse (i.e,
years) until first arrest, the calculation was based on whether a person was arrested in
each year. Years with an arrest were under study instead of the raw number of arrests
over the 12 year period. Considering that an arrest does not remove a case from the
analysis of subsequent years, each person had 10, 11, or 12 periods depending on the
length of his follow-up period.
Naturally, race and group membership do not change, however, the overall risk
score will for some cases. Hence, the risk score was recalculated each time an arrest
occurred in a given year. This model allows the risk score to increase over time (e.g.,
low to high). In each person period the individual is still given a risk score, however, in
some cases that value changes based on the calculation.
Multivariate analyses using logistic regression were conducted in which arrest
was regressed on personality, group, race, time, and risk score. Analyzing repeated
events with logistic regression might yield biased standard error terms. In response to
this possible problem the analysis was also conducted in Stata, which compensates and
corrects for the repeated measures effect by adjusting the standard error terms. However,
the analysis between the traditional logistic regression in SPSS and the regression
analysis conducted in Stata produced similar results. The results presented will be those
conducted in SPSS.
84
Model #3: Personality and the Seriousness of the Offense
In the third model, discrete time event history analysis was utilized to explore the
relationship between personality, I-level, and offense type. There are a variety of ways to
explore offense seriousness. Offense seriousness scales have been used in sentencing
decisions for a number of years. For example, “since 1973 the United States Parole
Commission has used explicit guidelines to structure its discretion in deciding the
duration of prison terms for federal prisoners under its jurisdiction. The guidelines
matrix is in the form of two dimensional charts. The seriousness of the prisoner’s current
offense is graded on an index that forms the vertical axis of the chart. The empirically
developed scale known as the Salient Factor Score, designed to assist in assessing the
prisoner’s risk of recidivism forms the horizontal axis of the guideline chart” (Hoffman
and Hardyman, 1986, p. 414). However, in order to calculate a severity scale such as this
one, the researcher must know not only the offense type, but also whether the offense
involved a weapon, whether co-defendants were involved, whether the crime resulted in
physical injury, or, if a theft, the amount and type of goods stolen.
The data on offense details in the NCIC system is very limited. Data on charge
type is available, however, other characteristics such as weapon and co-defendant
involvement, victim impact, and theft characteristics are not collected. As a result, this
study is limited to an analysis of the impact of personality on type of offense. The
categories chosen are: drug offenses, property offenses, and violent offenses. In effect to
answer whether individuals with certain personality profiles are more likely to be arrested
for specific offense types. While these categories may not reflect a detailed assessment
of seriousness, the types do reflect important distinctions in offending behavior.
85
The creation of the risk set, person periods, and hazard rates are the same as found
in Model 1. However, the current models are limited to those arrested for specific
offenses. In all three models, those not arrested for the offense in question are all treated
the same. That is, if someone was not arrested for a drug related offense but was arrested
for a theft, that person would be combined in the first model with those never arrested
during the follow-up period. As seen in Table 14, there were 34 individuals arrested for
drug related offenses. As a result, for the sample of 277, there were 2948 person periods
created. For each person, the dependent variable is coded a 1 if they were arrested for a
drug related offense and 0 if they were not arrested for a drug related offense. Similarly,
Table 15 indicates that 42 individuals were rearrested for a property related offense. For
the sample of 277, there were 2891 person periods created. Again, the dependent
variable is coded a 1 if the person is arrested for a property related offense and 0 if not.
Finally, as seen in Table 16, 25 individuals were arrested for a violent offense. For the
sample of 277, there were 3037 person periods created. The dependent variable was
coded
Multivariate analyses were conducted by estimating each model separately to
determine whether personality was related to offense type. Only the first offense was
analyzed in this model. That is, the analysis determined whether personality was related
to the offense type of the first arrest during the follow up period. The regression model
includes group, race, risk, time, and personality as predictors of arrest for the three
offense categories. As with the previous models, participants who are not arrested for the
specific offense in question are included in the analysis and considered right censored.
86
Table 14. Distribution by Year of Parolees at Risk (Model 3: Drug Offenses)
________________________________________________________________________ Number Number Estimated Year Arrested at risk Hazard Rate For Drugs 1 7 277 .0253 2* 7 270 .0259 3 2 262 .0076 4 3 260 .0115 5 3 257 .0116 6 1 254 .0039 7 3 253 .0118 8 1 250 .0040 9 1 249 .0004 10** 3 248 .0121 11** 2 240 .0083 12 1 128 .0078 Total 34 2948 * One person died during the follow-up period and was excluded after year 2 **The follow-up period varied from 10 to 12 years. Of the 277, 5 of those not arrested only had a 10 year follow up and were taken out of the number possibly at risk and 110 of those not arrested had a 11 year follow up and were also censored at that time.
87
Table 15. Distribution by Year of Parolees at Risk (Model 3: Property Offenses)
________________________________________________________________________ Number Number Estimated Year Arrested at risk Hazard Rate For Property 1 4 277 .0144 2* 6 273 .0220 3 4 266 .0150 4 8 262 .0305 5 6 254 .0236 6 5 248 .0202 7 1 243 .0041 8 4 242 .0165 9 1 238 .0042 10** 1 237 .0042 11** 1 231 .0043 12 1 120 .0083 Total 42 2891 * One person died during the follow-up period and was excluded after year 2 **The follow-up period varied from 10 to 12 years. Of the 277, 5 of those not arrested only had a 10 year follow up and were taken out of the number possibly at risk in the subsequent years and 110 of those not arrested had a 11 year follow up and were also censored in that year.
88
Table 16. Distribution by Year of Parolees at Risk (Model 3: Violent Offenses)
________________________________________________________________________ Number Number Estimated Year Arrested at risk Hazard Rate For Violence 1 2 277 .0072 2* 2 275 .0073 3 2 272 .0074 4 2 270 .0074 5 1 268 .0037 6 4 267 .0150 7 5 263 .0190 8 1 258 .0039 9 2 257 .0078 10** 1 255 .0039 11** 3 249 .0120 12 0 126 .0000 Total 25 3037 * One person died during the follow-up period and was excluded after year 2 **The follow-up period varied from 10 to 12 years. Of the 277, 5 of those not arrested only had a 10 year follow up and were taken out of the number possibly at risk in the subsequent years. In addition, 120 of those not arrested had a 11 year follow up and were also censored after that time.
89
CHAPTER 4
Results
Multivariate Analysis of Arrest: Model 1
Of the 277 participants in the study, 145 (52.3%) were arrested at least once
during the follow-up period. By using event history analysis, it is possible to estimate a
model of the probability of an arrest during the follow-up period and distinguish the
relative influence of three variables assumed to be constant over time: personality, race,
and risk (as measured by the SFS which includes only static risk factors). Results of the
event history analysis by personality type are shown in Figure 1. Figure 1 represents the
independent effects of personality with race and risk controlled. For ease of
interpretation, the figures will be presented as failure curves instead of survival curves.
The failure curves shown are based on the probability of arrest at each year based on the
proportion that failed or were arrested. The results are presented as curves and “the
smoothness results from the constraints inherent in the population hazard model
stipulated in (1), which forces the vertical separation between the two hazard functions to
be identical in every time period. Just as we do not expect a fitted regression line to
touch every data point in a scatterplot, we do not expect a fitted hazard function in
survival analysis to match every sample value of hazard” (Willett and Singer, 1997, 281).
A number of regression equations were estimated to determine the impact of a
variety of independent variables. After exhausting several models that included race,
risk, group, personality, and time (see appendix B) the final model is limited to the
90
Figure 1
Personality Type by Rearrest Rate
Statistically significant differences
)
)
))
) ) ) ) ) ) ) )
∋
∋
∋ ∋ ∋ ∋ ∋ ∋ ∋ ∋ ∋ ∋
&
&
&&
&&
&& & & & &
#
#
##
##
# # # # # #
Year1 Year2 Year3 Year4 Year5 Year6 Year7 Year8 Year9 Year10 Year11 Year120
20
40
60
80
100Aggressives Neurotics Dependents Situationals# & ∋ )
Aggressives 6.4 12.4 18.1 23.3 28.2 32.9 37.2 41.2 45 48.5 51.8 54.9Neurotics 8.5 16.2 23.3 29.8 35.8 41.2 46.2 50.7 54.9 58.7 62.2 65.4
Dependents 3.6 7.1 10.4 13.6 16.7 19.7 22.6 25.4 28.1 30.7 33.2 35.6Situationals 5.7 11.1 16.2 21 25.6 29.8 33.9 37.7 41.3 44.6 47.8 50.8
91
significant variables which include: a dichotomous measure of risk (1= low/good risk; 0=
high/poor risk), race (1=white; 0= non-white), and the four personality types (aggressive
1=yes; 0 = other; neurotic 1=yes; 0= other; etc) originally discussed in Van Voorhis
(1994). The analysis indicated that all three variables were significantly related to the
probability of failure among the group of federal inmates (see Table 17). The curves
shown represent the predicted failure rates by each personality group by length of time
(years) from the beginning of the inmate’s incarceration. Throughout the follow-up
period, the highest probabilities of arrest was among the neurotics followed by the
aggressives, situationals, and finally the dependents (Appendix B also illustrates the
models with rotated personality types). The analysis indicated that aggressives and the
neurotics were significantly different than the dependents but not significantly different
from each other or the situationals. In fact, the situationals were not significantly different
from any of the other types. In addition to the importance of the difference between
types was the finding that personality still remains a significant predictor of recidivism
even while controlling for race and risk12. Finally, the same models were estimated using
Interpersonal level maturity categories, however, as seen in Appendix B the relationship
between I-level and outcome was not significant.
12 Regression models with and without personality were estimated to distinguish whether personality contributed to the model. It is noted that exploring R-square differences is not an ideal method for assessing the magnitude of the relationship, however, the differences are informative. The model including only race and risk had an Nagelkerke R-square is .136 and with the addition of the personality variables the R-square valued increased to .179.
92
Table 17. Model 1 Logistic Regression: Probability of Rearrest by Personality & Control Variables
________________________________________________________________________ Parameter Standardized Odds Ratio Estimate Estimate SFS Riska .559** .188 1.749 Raceb .636*** .185 1.888 Personalityc
Aggressive .610** .281 1.841 Neurotic .909** .322 2.482 Situational .483 .354 1.621 Intercept -3.690 Model Chi Square 33.520 * p <.05 ** p <.01 *** p <.001 a Categories: 0= low; 1= high b Categories: 0= white; 1= non-white c Dependent type is the omitted variable
93
Figure 2
Rearrest Rate by Personality Type:
Low Risk
Statistically significant differences
)
)
) ) ) ) ) ) ) ) ) )
∋
∋
∋ ∋ ∋ ∋ ∋ ∋ ∋ ∋ ∋ ∋
&
&
&&
& & & & & & & &
#
#
# # # # # # # # # #
Year Year Year Year Year Year Year Year Year Year1 Year1 Year10
20
40
60
80
100 percentage
Aggressive Neurotic Dependent Situational # & ∋ )
Aggressive 4.4 8.6 12.6 16.5 20.1 23.6 27 30.2 33.3 36.2 39 41.7 Neurotic 5.8 11.3 16.5 21.4 26 30.3 34.4 38.2 41.8 45.2 48.4 51.4
Dependent 2.4 4.8 7.1 9.4 11.6 13.8 15.9 17.9 19.9 21.9 23.8 25.6 Situational 3.9 7.6 11.2 14.7 18 21.2 24.3 27.2 30 32.8 35.4 37.9
94
Figure 3
Rearrest Rate by Personality Type:
High Risk
Statistically signficant differences
)
)
))
))
)) ) ) ) )
∋
∋
∋∋
∋ ∋ ∋ ∋ ∋ ∋ ∋ ∋
&
&
&&
&&
&&
& & & &
#
#
##
##
##
# # # #
Year Year Year Year Year Year Year Year Year Year1 Year1 Year10
20
40
60
80
100 Aggressive Neurotic Dependent Situational # & ∋ )
Aggressive 9.5 18.1 25.8 32.9 39.2 45 50.2 54.9 59.2 63.1 66.6 69.7 Neurotic 12.4 23.2 32.7 41 48.3 54.7 60.3 65.2 69.5 73.3 76.6 79.5
Dependent 5.4 10.5 15.3 19.9 24.2 28.3 32.1 35.8 39.2 42.5 45.6 48.5 Situational 8.4 16.2 23.3 29.7 35.7 41.1 46.1 50.6 54.8 58.6 62.1 65.3
95
In the model described above, the analysis explored the independent contributions
of personality while controlling for risk and race. Both race and risk were significantly
related to outcome. Specifically, those individuals classified as high risk had a higher
likelihood of recidivism than the lower risk group. However, Figure 1 does not show the
probabilities separated by those classified as high risk and low risk. Hence, Figures 2 and
3 illustrate the probabilities of arrest by personality for high risk participants and low risk
participants.
It should be noted that an analysis was conducted to determine whether an
interaction existed between personality and race. Although a significant difference did
exist between the additive and interaction models, we could not localize the significant
differences due to the concerns for the power of the interaction models.
Multivariate Analysis of Multiple Arrests: Model 2
Of the 277 individuals under study, 85 (31.0%) were arrested multiple times
during the follow-up period. By using repeated events event history analysis, it is
possible to estimate a model of the probability of multiple arrests during the follow-up
period and distinguish the relative influence of time varying and time invariant
predictors. The results of the repeated measures event history analysis, which takes into
account multiple offenses, are presented in Figure 4. In this model, personality was
significant even after controlling for race, time, and risk score.
As with the first model, a number of regression equations were estimated to
determine the impact of a variety of independent variables. After exhausting several
models (see Appendix C) the final model included: a dichotomous measure of risk (1=
low/good risk; 0= high/poor risk), race (1=white; 0= non-white), and the four personality
96
Figure 4
Personality Predicting Multiple Rearrest
Statistically Significant Differences
∋
∋
∋∋ ∋ ∋ ∋ ∋ ∋ ∋ ∋ ∋
)
)
)) ) ) ) ) ) ) ) )
&
&
&&
&&
& & & & & &
#
#
##
## # # # # # #
Time1 Time2 Time3 Time4 Time5 Time6 Time7 Time8 Time9 Time10 Time11 Time120
20
40
60
80
100Aggressives Neurotics Situationals Dependents# & ) ∋
Aggressives 15.1 24.9 30.7 36 41 45.5 48.3 51 53.5 55.8 58.1 60.2Neurotics 18 29.3 35.9 41.9 47.3 52.3 55.2 58 60.6 63.1 65.4 67.5
Situationals 11.9 19.8 24.6 29.1 33.4 37.4 39.8 42.2 44.5 46.7 48.8 50.8Dependents 12.3 20.4 25.4 30 34.3 38.4 40.9 43.3 45.6 47.8 49.9 52
97
types (aggressive 1=yes; 0 = other; neurotic 1=yes; 0= other; etc) and four time variables
(year 1, year 2, year 3-6, year 7-12). Time was collapsed into four different variables:
Year 1, Year 2, Year 3-6 and Year 7-12 in order to create a more parsimonious model.
The model with all twelve time variables was not needed after the analysis indicated that
the differences among the parameter estimates were not significant between years 3-6 and
years 7-12 (see Appendix C, Table 1). The analysis indicates that risk, race, personality
and time were significantly related to the probability of multiple failures (see Table 18).
The curves shown represent the predicted failure rates by each personality group by
length of time (years) from the beginning of the inmate’s incarceration. Throughout the
follow-up period, the highest probabilities of multiple arrests were among the neurotics
followed by the aggressives, situationals, and finally the dependents. The analysis
indicated that neurotics were significantly different from the dependents and situationals
but not significantly different from the aggressives. In fact, the aggressives were not
significantly different from any of the other types (see Appendix C). Finally, I-level was
not significantly different in any of the models.
Multivariate Analysis of Offense Type: Model 3
Drug Offenses. Of the 277 participants in the study, 34 (12% of the total sample)
were arrested for a drug related offense during the follow-up period. The influence of
personality, race, risk, and group were assessed using discrete time event history analysis.
After exhausting a more inclusive model (see Appendix D), the final model included a
dichotomous measure of risk (1= low/good risk; 0= high/poor risk), race (1=white; 0=
non-white), and the four personality types (aggressive 1=yes; 0 = other; neurotic 1=yes;
0= other; etc). Results are presented in a failure curve in Figure 5. As indicated in Table
98
Table 18. Model 2 Logistic Regression: Probability of Multiple Rearrest by Personality & Control Variables
________________________________________________________________________ Parameter Standardized Odds Ratio Estimate Estimate SFS Riska .826*** .125 2.284 Raceb .649*** .126 1.914 Personalityc Aggressive -.208 .160 .812 Dependent -.483* .216 .617 Situational -.447* .220 .639 Time Time 2 -.319 .175 .727 Time 3 – 6 -.756*** .148 .470 Time 7 – 12 -1.197*** .179 .302 Intercept -2.0109 Model Chi Square * p <.05 ** p <.01 *** p <.001 a Categories: 0= low; 1= high b Categories: 0= white; 1= non-white c Neurotic type is the omitted variable Year one is the omitted variable
99
Figure 5
Rearrest Rate by Personality Type: Drug Offenses
Statistically significant differences
)
)
) ) ) ) ) ) ) ) ) )
∋
∋∋ ∋ ∋ ∋ ∋ ∋ ∋ ∋ ∋ ∋
&
&
&&
& & & & & & & &
#
#
# # # # # # # # # #
Year1 Year2 Year3 Year4 Year5 Year6 Year7 Year8 Year9 Year10 Year11 Year120
10
20
30
40
50Aggressives Neurotics Dependents Situationals# & ∋ )
Aggressives 0.9 1.9 2.8 3.7 4.6 5.5 6.4 7.3 8.2 9.1 9.9 10.8Neurotics 2.5 4.9 7.3 9.6 11.9 14.1 16.2 18.3 20.3 22.3 24.2 26.1
Dependents 0.6 1.2 1.7 2.3 2.9 3.4 4 4.5 5.1 5.6 6.2 6.7Situationals 0.7 1.5 2.2 2.9 3.6 4.3 5 5.7 6.4 7.1 7.8 8.5
100
Table 19. Model 3 Logistic Regression: Probability of Rearrest for a Drug Offense by Personality & Control Variables
________________________________________________________________________ Parameter Standardized Odds Ratio Estimate Estimate SFS Riska .179*** .377 1.116 Raceb .109*** .387 1.197 Personalityc Aggressive -.987* .408 .372 Dependent -1.479* .662 .228 Situational -1.124* .582 .325 Intercept -3.7633*** Model Chi Square 8.486 * p <.05 ** p <.01 *** p <.001 a Categories: 0= low; 1= high b Categories: 0= white; 1= non-white c Neurotic type is the omitted variable
101
19, all three variables were significantly related to the probability of failure for a drug
related offense. Throughout the follow-up period the neurotics had a much greater
likelihood of being arrested for a drug related crime in comparison to the aggressives,
dependents, and situationals. The neurotics were significantly different from the other
personality categories; however, the other three (i.e., aggressives, dependents, and
situationals) were not significantly different from each other (see Appendix D).
The same models were estimated using the I-level categories. The model
included a dichotomous measure of risk (1= low/good risk; 0= high/poor risk), race
(1=white; 0= non-white), and the three I-levels (I-2 1=yes; 0 = other; I-3 1=yes; 0= other;
etc). Results are presented in a failure curve in Figure 6. As seen in Table 20, the
relationship between I-level and arrest for a drug related offense were significant. The I-
4 category had the highest probability of arrest followed by the I-2s and I-3s. The I-4
category is significantly different from the I-3 category, however, not significant from the
I-2 category. In fact, the I-2 category is not significantly different from either the I-4 or
the I-3 categories.
Property Offenses. Of the 277 individuals in the study, 42 were arrested for a
property offense. Appendix E contains the results of several runs, however, the final
model included a dichotomous measure of risk (1= low/good risk; 0= high/poor risk),
race (1=white; 0= non-white), and the four personality types (aggressive 1=yes; 0 =
other; neurotic 1=yes; 0= other; etc). Results are presented in a failure curve in Figure 7.
As indicated in Table 21, the only variable found to be significant was risk. Risk (higher
risk individuals) was significantly related to the probability of failure for a property
related offense. Neither personality or the levels were significantly related to the
102
Figure 6
Rearrest Rate by I-Level Type: Drug Offenses
Statistically significant differences
+
+
+ + + + + + + + + +
∗
∗
∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗
∃
∃
∃ ∃ ∃ ∃ ∃ ∃ ∃ ∃ ∃ ∃
Year1 Year2 Year3 Year4 Year5 Year6 Year7 Year8 Year9 Year10 Year11 Year120
10
20
30
40
50I - 2 I - 3 I - 4∃ ∗ +
I - 2 1.4 2.8 4.1 5.4 6.7 8 9.3 10.6 11.8 13 14.2 15.4I - 3 0.8 1.6 2.4 3.2 4 4.8 5.5 6.3 7.1 7.8 8.6 9.3I - 4 1.7 3.3 4.9 6.4 8 9.5 11 12.5 13.9 15.3 16.7 18.1
103
Table 20. Model 3 Logistic Regression: Probability of Rearrest for a Drug Offense by Personality & Control Variables
________________________________________________________________________ Parameter Standardized Odds Ratio Estimate Estimate SFS Riska .224 .389 1.251 Raceb .054 .389 1.055 Interpersonal Leveld I – 2 -.177 .786 .838 I – 3* -.722* .375 .486 Intercept -4.180*** Model Chi Square 4.014 * p <.05 ** p <.01 *** p <.001 a Categories: 0= low; 1= high b Categories: 0= white; 1= non-white c I-4 is the omitted variable
104
Figure 7
Rearrest Rate by Personality Type: Property
Offenses
)
)
) ) ) ) ) ) ) ) ) )
∋
∋
∋ ∋ ∋ ∋ ∋ ∋ ∋ ∋ ∋ ∋
&
&
& & & & & & & & & &
#
#
# # # # # # # # # #
Year Year Year Year Year Year Year Year Year Year1 Year1 Year10
10
20
30
40
50 percentage
Aggressive Neurotic Dependent Situational # & ∋ )
Aggressive 1.5 2.9 4.4 5.8 7.2 8.5 9.9 11.2 12.5 13.8 15.1 16.3 Neurotic 1.6 3.2 4.8 6.3 7.8 9.3 10.8 12.2 13.7 15.1 16.4 17.8
Dependent 1 2 3 4 5 6 7 7.9 8.9 9.8 10.7 11.6 Situational 0.9 1.8 2.7 3.5 4.4 5.3 6.1 7 7.8 8.6 9.5 10.3
105
Table 21. Model 3 Logistic Regression: Probability of Rearrest by Personality & Control Variables predicting Property Offenses
________________________________________________________________________ Parameter Standardized Odds Ratio Estimate Estimate SFS Riska .157*** .332 1.170 Raceb .814 .333 2.256 Personalityc Aggressive -.095 .425 .910 Dependent -.463 .559 .629 Situational -.595 .616 .552 Intercept -4.4328*** Model Chi Square 8.987 * p <.05 ** p <.01 *** p <.001 a Categories: 0= low; 1= high b Categories: 0= white; 1= non-white c Neurotic type is the omitted variable
106
probability of failure for a property related offense (see Appendix E).
Violent Offenses. Of the 277 participants in the study, 25 were arrested for violent
offense. The same procedure and models listed above were used to predict the probability
of violence by personality types. The more inclusive model can be seen in Appendix F,
however, the final model included a dichotomous measure of risk (1= low/good risk; 0=
high/poor risk), race (1=white; 0= non-white), and the four personality types (aggressive
1=yes; 0 = other; neurotic 1=yes; 0= other; etc). The results are presented in Figure 8.
As indicated in Table 22, race was the only significant variable found in the model. It
appears that African Americans have a higher probability of violence than whites with
risk and personality included in the model. The I-level types were not significantly
related to the probability of failure for a violent offense (see Appendix F). Similar to the
property offense model, the personality characteristics were not significantly different
from each other in their probability of being arrested for violent offenses.
107
Figure 8
Rearrest Rate by Personality Type: Violent Offenses
))) ) ) ) ) ) ) ) ) )∋∋
∋ ∋ ∋ ∋ ∋ ∋ ∋ ∋ ∋ ∋&
&& & & & & & & & & &
##
# # # # # # # # # #Year1 Year2 Year3 Year4 Year5 Year6 Year7 Year8 Year9 Year10 Year11 Year12
0
10
20
30
40
50Aggressive Neurotic Dependent Situational# & ∋ )
Aggressive 0.3 0.7 1 1.4 1.7 2.1 2.4 2.7 3.1 3.4 3.7 4.1Neurotic 0.4 0.8 1.2 1.6 1.9 2.3 2.7 3.1 3.5 3.8 4.2 4.6
Dependent 0.5 1 1.5 2.1 2.6 3.1 3.6 4.1 4.6 5.1 5.6 6Situational 0.2 0.5 0.7 0.9 1.2 1.4 1.6 1.8 2.1 2.3 2.5 2.7
108
Table 22. Model 3 Logistic Regression: Probability of Rearrest by Personality & Control Variables predicting Violent Offenses.
________________________________________________________________________ Parameter Standardized Odds Ratio Estimate Estimate SFS Riska 1.173 .438 3.233 Raceb .597*** .448 1.816 Personalityc Aggressive -.125 .596 .883 Dependent .282 .696 1.326 Situational -.526 .875 .591 Intercept -5.540*** Model Chi Square 14.702* * p <.05 ** p <.01 *** p <.001 a Categories: 0= low; 1= high b Categories: 0= white; 1= non-white c Neurotic type is the omitted variable
109
CHAPTER 5
Discussion and Conclusions
As depicted in Chapter 2, the research on personality and crime is fairly limited.
More importantly, comparatively fewer studies have isolated the likelihood that
personality predicts recidivism over time. The current study builds upon the research in
criminology and in corrections to develop an understanding of whether certain
personality types are more likely to be arrested and whether a measure of personality
should be added to traditional risk assessment instruments. The implications of this
research impact theory development, risk assessment technology, and correctional
programming. The purpose of this chapter is to summarize the limitations, findings,
theoretical and policy implications, and suggestions for future research.
Limitations
Although this study does advance the current literature base, there are several
limitations worth noting. First, the sample is limited to adult males. The findings, while
relevant for female populations, are based on male offenders. Second, the recidivism
data were collected via official records. The NCIC records are based on reports of
criminal activity by state. As of 1998, 36 states provided records to this national arrest
data source. The fourteen states that did not report data to NCIC include: Nebraska,
Kansas, Wisconsin, Louisiana, Mississippi, Tennessee, Kentucky, West Virginia, District
of Columbia, Rhode Island, Massachusetts, Maine, New Hampshire, and Vermont. This
limitation paired with the recognition that official records fail to capture crimes not
reported to police (Schneider and Wiersemer, 1990) may lead some to question whether
110
the study has failed to capture all of the crimes committed by these research subjects.
Although the extent of this problem for the current study cannot be assessed, a few points
should be taken into consideration. First, given many of the participants in the study
have lengthy prior records and were on parole supervision, one may speculate that the
detection of their criminal behavior was more likely. Second, the overall recidivism rate
of 52 percent is within a typical range of recidivism for adult male offenders. Finally, the
findings are consistent with the research by Caspi et al. (1994), which included four
measures of delinquency: self-report, informant, police contact, and court convictions.
Even with these considerations in mind, future efforts to understand the relationship
between personality and criminal behavior among an adult sample would inevitably
benefit from incorporating multiple outcome measures.
Second, and of equal concern, the data reported to NCIC are often limited to the
arrest date and charge and do not adequately capture crime seriousness. Having access
to data limited to charge type can bias an analysis of crime seriousness in two ways.
First, arrest charge is subject to decision bias, that is, the decision to charge with a
specific offense. Second, data including weapons or co-defendant involvement were not
available. Data of this nature would allow for a more accurate assessment of crime
seriousness. Given this circumstance, there is no way to examine seriousness. However,
it should be noted that the models pertaining to offense type do contribute to our
understanding of the relationship between personality and criminal behavior.
Third, the data were collected at only two points in time. The personality and
demographic data were collected at Time 1 (1986-1988) and the recidivism data at Time
2 (1998). This method did not allow for an assessment of important dynamic risk factors
111
(i.e., marriage, stability, employment status, educational attainments) that may be related
to outcome (Sampson, & Laub, 1993; Andrews & Bonta, 1998). Moreover, data
pertaining to service exposure or treatment post-release were not available. As such, it
cannot be discerned whether the effects found would have been mediated by other
variables.
Finally, and in a similar vein, it is recognized that this study examines the
individual’s offending career in late adulthood. Hence, the investigation of personality
and crime across developmental periods could not be examined. Although research
supports the idea that personality is relatively stable throughout adulthood, it is also
recognized that personality and the situation are mutually reinforcing (Caspi & Roberts,
1999). Future research would benefit from studying the personality of an offender
during developmental stages to further explore the impact of the environment on
personality among a criminal population.
Summary of Findings
As observed by Winter and Barenbaum (1999), the attempt to study personality
rests on the assumption that personality is worth studying. Although the development of
personality research did not escape the scrutiny of situationists, the literature review in
Chapter 2 illustrates that personality is able to contribute to our understanding of
behavior. In criminology, the view that personality is correlated with criminal behavior
has developed with some resistance (Andrews & Bonta, 1998; Andrews & Wormith,
1989). Aiding this cause, current research conducted with personality inventories such as
the EPQ and the MPQ finds that personality is related to various types of criminal
behavior and can predict criminal behavior over time (Caspi et al., 1994; Caspi et al.,
112
1994; Eysenck, 1952, 1983, 1996; Krueger et al., 1994; & Krueger et al., 1996).
Moreover, research in corrections with psychological typologies supports the notion that
treatment should be matched with individual differences in mind. As such, this study was
designed to address several research questions.
The main focus of the current research identified whether certain personality
types were significantly related to recidivism among adult male offenders. The data were
used to explore three main research questions in an effort to explore whether individuals
with certain personality characteristics were more likely to be (a) arrested, (b) arrested
multiple times, or (c) arrested for a certain type of offense. The findings from the
questions above have implications for answering whether the Jesness Inventory is useful
in predicting criminal behavior and whether personality should be added to existing
traditional risk assessment models. Overall, the findings from this study support the
notion that personality is an important predictor of criminal behavior.
The findings from the first model indicate that personality contributes to the
prediction of criminal behavior while controlling for race and risk. In other words,
personality is significantly related to the probability of criminal behavior above and
beyond the variation explained by the risk variable and race. The findings from this
study support the research conducted by Eysenck and Caspi et al. (1994) and the research
on psychopaths conducted by Hemphill and Hare (1995) and Hemphill, Hare, and Wong
(1998), which finds a relationship between psychopathology and recidivism. The current
study found that the highest probabilities of arrest were among the neurotics followed by
the aggressives, situationals, and dependents. The neurotics and aggressives were
significantly different from dependents in their probability of criminal behavior. Simply
113
put, neurotics and aggressives had a significantly higher probability of experiencing an
arrest in comparison to the dependents. Interpersonal maturity levels were also explored,
however, the levels were not significant in the model predicting arrest.
Second, having established that personality was significantly related to outcome,
the analysis shifted to exploring whether personality could predict persistence.
Specifically, whether certain personality types could characterize the individuals who
were arrested multiple times. Much of the previous work in this area has been with
juveniles and delineates the relationship between temperament styles or early aggressive
behavior and later delinquent behavior (Caspi, et al., 1995; Moffitt, 1990; Patterson, et
al., 1992; & Statin & Magnusson, 1989). While many of the previous studies followed
the youth into their early twenties, the average participant in the current study was 30
when the follow-up period began. This study provided an opportunity to study the adult
offender during the later stages of their criminal career.
The findings from the second model are similar to those found in previous
research. The results indicate that the neurotic type had a significantly higher probability
of being arrested multiple times in comparison to the other types even when controlling
for time, race, and risk. Although previous research finds that the Jesness aggressive
type, which is argued to mirror the EPQ psychoticism type, may be more likely to be
arrested over the lifespan (Eysenck, 1983), neuroticism is also important. Specifically,
Eysenck (1983) argues that neuroticism is more important in older samples and
contributes to stronger antisocial habits in adults. By using repeated event history
analysis, it became clear that neurotics had the highest probability of arrest and were
significantly different from both the dependents and situationals. The significant
114
differences in this model are slightly different from the first in that the aggressives were
not significantly different from any of the other types. It should be noted that while not
significant, the aggressive type did have slightly higher probabilities of multiple arrests in
comparison to the dependents and situationals. Finally, the differences between
personality types holds over time. That is, the neurotic type predicted the probability of
multiple offenses over an extended period of time.
Finally, the analysis focused on offense type to explore whether personality could
predict specific types of offending behavior. Personality was significant in one of the
models. Specifically, neurotics were more likely than the other types to be arrested for a
drug related offense. This is consistent with previous research on the neurotic type;
specifically, research that finds drug and alcohol use among those with high levels of trait
anxiety (Chambless, Cherney, Caputo, and Rhienstein, 1987; Smail, Stockwell, Canter,
and Hodgeson, 1984). Further, research indicates alcohol may be used to reduce the
negative affect that drinkers feel when they are anxious or overaroused (Cloninger,
Sigvardsson, Prybeck, and Svrakic, 1995). Related, a study by Roy (1999) found that
depressed alcoholics had significantly higher neuroticism scores as measured by the EPQ.
In addition, I-level was also significant in relationship to drug charges. The I-4s have an
internalized value system and are typically more likely to speak out against criminal
values; however, the use of drugs and alcohol may inhibit this control.
The evidence suggests that personality types were not significantly different from
one another in the analysis of violent or property offense types. In other words, the
analysis was unable to distinguish any differences between the types when predicting the
probability of being arrested for either a property or violent offense. This finding is
115
counterintuitive to the research which finds that psychopathy is related to violent
recidivism (Forth, Hart, & Hare, 1990; Hare, & McPherson, 1984; & Harris, Rice, &
Cormier, 1991). However, there may be a number of explanations for these findings.
First, it is very difficult to predict violent behavior (Andrews & Bonta, 1998). The
difficulty stems from the low base rate considerations. Specifically, the results with
regard to violence may be related to the occurrence of violence in the current sample. Of
the 277 individuals at risk, only 25 individuals were arrested for a violent offense during
the follow-up period. It may be that the behavior did not occur frequently enough within
the personality types to distinguish any differences. Second, the previous research that
finds psychopathy is related to violence classifies psychopathy with the Psychopathy
Checklist Revised (PCL-R) (Hare, 1991). The aggressive type measures by the Jesness
Inventory includes offenders who have antisocial values and lifestyles but may not be
psychopaths. In fact, Hare (1982) questions the comparability between the psychoticism
type as measured by the EPQ and the construct of psychopathy measures by the PCL-R.
Third, we may speculate that by the time individuals reach this stage of their criminal
career, they may be less likely to engage in violent acts or are more likely to be serving
time in institutions. Finally, with regard to property offenses, while research supports the
relationship between personality and drug related behavior, there is no evidence that
personality should distinguish the likelihood of an offender committing a property
offense. This analysis was explorative and the lack of evidence is not necessarily
inconsistent with previous research. Future research, however, may benefit from
exploring the relationship between offense type and personality across developmental
stages.
116
The findings regarding the situational type appear somewhat inconsistent with
previous research. Specifically, situationals were not significantly different from the
aggressives or the dependents in any of the models. While it would be expected that the
dependents and situationals might have similar arrest patterns, the fact that the situtionals
had a rate similar to the aggressives was surprising. Although it is difficult to pinpoint
why this may be the case, a few points of speculation may be offered.
First, the Case Management Classification (CMC) system, which is another
personality classification system that measures the situational type, divides the type into
situationals with a substance abuse problem and those without such a problem. The data
for the current study do not allow for this distinction. In other words, it may be that the
situational type used in the current study is too broad and does not accurately capture the
personality traits that comprise positive emotionality or constraint as outlined by the
MPQ. Second, this group is assumed to have a lower risk of recidivism over time;
however, no long-term recidivism studies with Jesness Inventory have been conducted. It
could be that the older more entrenched criminals that are classified as situationals
perform differently. Finally, the effects of prisonization (see Thomas, 1977) may have
influenced individuals characterized as situationals more than the other types. The
personality data obtained for this research were collected at prison intake for Time 1
(1986-1988). The impact of prison and exposure to criminal peers may have influenced
the future behavior of these individuals.
Finally, taken as a whole, I-level did not contribute to our understanding of
criminal behavior in this sample of adult males. Theoretically we should expect the
highest probability of arrest among the I-2s followed by the I-3s and I-4s. In fact,
117
although the findings pertaining to I-level in the original study were minimal, the inmates
classified as I-2 and I-3 were significantly more likely to be cited for disciplinary
infractions in comparison to the I-4s (Van Voorhis, 1994). These maturity levels
represent successive stages from least mature to a higher level of maturity. While the
analysis indicated that there were significant relationships between I-level and drug use
(e.g., I-4s had a significantly higher probability of arrest from the I-3s), I-level was not
significant in any of the other models. There are a number of possibilities why I-level
was not significant. The first is sample size; of the 277 individuals under study only 12
(4.3%) were classified as I-2. The second possibility is that the I-level types do not
perform as expected in a prediction model with an adult sample. Given this study is the
first of its kind with an adult sample, the findings may illustrate the limitations of using
maturity levels to predict future criminal behavior.
Overall, the data appear to suggest that the relationship between personality and
crime is significant. The aggressive type performed as expected in the first model
predicting arrest and the neurotic type consistently emerged as important in each of the
models predicting offending behavior. By successfully using the a correctional typology
scale in predicting recidivism, this study supports the notion that personality is an
important risk factor and can assist our understanding of offenders both theoretically as
an explanation for behavior and practically for the application of treatment. As such,
these findings have implications for both theory and policy development.
Broader Implications of the Findings
This research set out to answer several questions about personality and crime. As
a result, the study advances the knowledge reviewed in Chapter 2 and provides support
118
for a theory of personality in criminology and the formulation of policies in corrections
designed to assess and treat the offender.
Overall these findings are important because they provide support for the work of
criminologists researching individual-centered theories of crime. The findings are in
contrast to the earlier reviews of personality and crime (e.g., Schuessler & Cressey, 1950;
Tennenbaum, 1977; and Waldo & Dinitz, 1967) and claims by researchers such as Vold
and Bernard (1986) who argue that personality provides no theoretical relevance to
understanding criminal behavior. As researchers begin to take seriously the individual-
centered theories of crime, the relationship between personality and criminal behavior,
and as additional longitudinal studies assessing this relationship are conducted, better
estimates of how personality influences behavior over the lifecourse will come available.
While accumulating studies are finding a relationship between personality and
criminal behavior, many of these studies are conducted with adolescents and young
adults (Caspi et al, 1994; Caspi et al., 1997; Krueger et al., 1996). This study adds to the
current research by delineated the importance of personality among adult male offenders.
Investigation of this group is important because some may argue that personality acts
differently across age groups (e.g., extroversion in adolescent groups in contrast to
neuroticism in older adults; see Eysenck, 1983). As mentioned previously, many of the
previous studies only follow the juvenile into their early 20s. The average age of the
participant in this study was 30 years old and by the end of the follow up period the
typical person was in their early 40s. This exploration of older adults sheds light on the
importance of personality in predicting offending behavior in what is considered the late
stages of an individual’s criminal career.
119
From a policy perspective, researchers advancing risk assessment technology
should also consider the effects of personality. Few researchers have used personality or
correctional typologies as predictors of risk. Moreover, although the research by
Andrews and Bonta (1998) find that personality is among the strongest predictors of
behavior, risk assessment instruments currently in use such as the Salient Factor Score
(Hoffman & Beck, 1985), the Level of Service Inventory-Revised (Andrews & Bonta,
1995), and Wisconsin Risk Assessment System (Baird, Heinz & Bemus, 1979) typically
do not include measures of personality. Further, while researchers such as Andrews and
Bonta (1998) identify a variety of individuals characteristics such as impulsivity,
restless/aggressive energy, and weak problem solving skills that are descriptors of
personality types (e.g., the neurotic and aggressive), they do not specifically outline
which personality type(s) is most important. The finding that neurotics and aggressives
are more likely than the other groups indicates that future assessment research should
explore these types in particular. Notwithstanding, the finding that personality was
significantly related to outcome while controlling for risk implies that personality should
be added to existing risk assessment models.
In addition to risk assessment, the results of this research could also impact
treatment strategies. From this perspective, the results would suggest that strategies to
intervene with the adult male offender should recognize personality, specifically the
characteristics of the neurotic type. Given that individuals high in negative affectivity
construe their environment in ways that are consistent with their personality traits (Costa
and McCrae, 1999), teaching individuals how to change thought patterns and subsequent
reactions becomes central to their probability of change. Similarly, behavior therapy is
120
based on the notion that environment affects learning and behavior. Cognitive-behavioral
therapy targets thinking and problem solving through a system of reinforcement, pro-
social modeling, and role-playing. Individuals with antisocial attitudes that are high on
neuroticism need intensive treatment towards increasing their self-regulation. Similar
strategies would work with the aggressive type. For example, as described by Van
Voorhis (1994), the aggressive type tends to be manipulative, often feel alienated and
hostile and has antisocial attitudes. Cognitive based curriculums designed to challenge
these attitudes would be an effective option. Similarly, Caspi and Roberts (1999) argue
that behavioral models can influence change in a person’s behavior and over time can
impact change in personality.
While many of these strategies (i.e., cognitive behavioral driven models) are
designed to treat the aggressive criminal with antisocial attitudes supportive of criminal
behavior, fewer strategies have been developed for the aggressive psychopath. Treatment
strategies should also be matched to this type, however, the research on the effectiveness
of treatment with psychopaths is relatively mixed. Research indicates that psychopaths
suffer from an information-processing disorder and this inhibits their ability to pay
attention (Wallace, Schmitt, Vitale, & Newman, 2000). One study done by Hitchcock
(1995) found psychopaths who participated in a cognitive thinking errors group appear to
have decreased their the frequency of thinking errors as measured by a post-test.
Interestingly, another study by Seto and Barbaree (1999) found that subjects
characterized as psychopaths did well in treatment but were more likely to reoffend than
non-psychopaths. This may imply that psychopaths are more likely to appear to make
progress in treatment but in reality do not internalize the appropriate skills.
121
In addition to considering personality type and the impact on treatment strategies,
personality should also be further explored as a responsivity factor. As argued by Van
Voorhis (1987), by failing to recognize that individuals respond to treatment differently,
we routinely mask a treatment effect. If personality types are not considered when
matching offenders to treatment groups, the result could be a decrease in the overall
effectiveness of the intervention. It could be argued, for example, that highly anxious
individuals would not perform well in a therapeutic community setting that utilizes
shaming and highly confrontive techniques. Similarly, research reviewed by Hobson,
Shine, and Roberts (2000) found that psychopaths who participate in therapeutic
communities tended have high attrition rates and low levels of motivation. Taken
together, these findings suggest that the effectiveness of treatment strategies will
inevitably be compromised if personality type is not taken into consideration.
Prison officials should recognize that the neurotic type might respond negatively
to incarceration. In fact, in the original study, Van Voorhis (1994) found evidence of this
as the neurotic types experienced difficulties with stress and fear and expressed a greater
need for programmatic support. The neurotic type was more likely than the aggressive or
situational type to report higher scores on the Center for Epidemiological Studies
Depression Scale (CEDS) (Radloff, 1977). In addition, the study found that only 30
percent of the participants perceived others as willing to help them. Research by Warren
(1983) also finds the neurotic category showing evidence of emotional disturbance such
as depression. Moreover, she argues that these individuals often evidence tension and
fear due to their feelings of failure and guilt. This is in line with the idea that the
neurotic type tends to construe his or her life and interactions negatively. This negative
122
orientation and self-defeating thought process should be recognized before deciding
treatment strategy and placement.
From a broader policy perspective, future research should explore why the
neurotic type has a higher probability of criminal behavior given that comparatively less
research has been devoted to the neurotic type in comparison to the aggressive type in
corrections. Strategies to intervene earlier in the offender’s life could have a later impact.
For example, one reason why the neurotic type has an increased probability of criminal
behavior could lie with Moffitt’s theory of cumulative continuity. As reviewed in
Chapter 2, Moffitt theorizes that later adult criminals who are high on negative
emotionality and low on constraint tend to be more impulsive and undercontrolled as
children, which may influence how they conformed to school and prosocial peers (Caspi
& Silva, 1995).
In a similar vein, Caspi and Roberts (1999) outline three types of transactions and
discuss how they “play particularly important roles both in promoting the continuity of
personality across the life course and in controlling the trajectory of the life course itself”
(p. 313). The three types of interactions are as follows (Caspi & Roberts, 1999, p. 314):
Reactive transactions occur when different individuals exposed to the same environment experience it, interpret it, and react to it differently. Evocative transactions occur when an individual’s personality evokes distinctive responses from others. Proactive transactions occur when individuals select or create environments of this own.
The basic assumption in all of these models is that individuals play a role in creating their
own environment. Researchers argue that people have a hand in promoting the
continuities in their life by the environments they select themselves into (Bandura, 1999;
Caspi & Roberts, 1999; Eysenck & Eysenck, 1985; Sampson & Laub, 1990). Moreover,
123
Bandura (1999) theorizes, “through selection and construction of environments,
personality patterns can become self-perpetuating” (p 164). McCrae and Costa (1999)
also discuss how interactions with the environment could serve to strengthen certain
personality traits. For example, they argue that an individual’s personality can influence
their preferences both for friends and situations, their ideologies about how the world
works, and their own social roles. In effect, these traits influence the situations they
choose to become involved in, the friends they seek out, and their subsequent behaviors.
The personality and environment are mutually reinforcing.
Research, however, also points to the impact that parents can have in childhood
toward reducing the later impact of personality on behavior. Caspi and Roberts (1999)
argue this point and review an example regarding behaviorally inhibited children:
Behaviorally inhibited children experience greater levels of distress at lower thresholds when confronted with novel situations….. Parents who expose their inhibited children to novelty and provide firm and consistent limits and do not overprotect their children from novel situations may help children overcome behavioral inhibition. In contrast, many parents respond to their child’s distress in novel situations by rewarding the child for avoiding these situations in the future. The reinforcement of these avoidance behaviors inadvertently promotes continued behavioral inhibition and may increase the likelihood that the child grows up to become an inhibited adult. (p. 316)
Drawing from this same perspective, research indicates that individuals high in negative
affectivity avoid challenging situations, dwell on frustration, and have a lower threshold
for experiencing negative events (Watson & Clark, 1984; Bolger & Schillings, 1991;
Bolger & Zuckerman, 1995). It could be argued that neurotics often act out as they have
difficulty controlling their emotional state, establishing boundaries, and so on. Similarly,
according to Andrews and Bonta (1998), “the aggressive personality has learned to
124
interpret a wide variety of persons and situations as threatening and frustrating, and has
learned habits of aggression to these cues” (p. 123).
Moreover, the experience of a negative event can strengthen the expectation that
the world is a negative place. This is echoed by the work of Warren (1983) as she argues
that the neurotics often do not communicate well with other family members and feel
they are not meeting expectations. Related to this, McHale (undated) finds that the
neurotic type tends to perceive their family as chaotic and believes that they play a
significant role in the development of this chaos. Moreover, he theorizes that the
neurosis begins as a result of feelings of rejection and insecurity. The parents, according
to McHale, tend to reinforce the negative thoughts and feelings of the individual, which
fosters feelings of guilt and inadequacy. As a result, the neurotic become more proficient
at being negative over time and also becomes more proficient at making excuses, getting
drugs, committing crime, etc. In addition, Eysenck (1992) argues that “with anxious
individuals, their attentional and interpretive biases may lead them to regard other
people’s social behavior towards them as more threatening than it actually is. As a
consequence they may behave aggressively or defensively or in other ways which
actually elicits threatening behavior from others” (p. 162).
However, if the environment of a neurotic person has clear expectations, good
role models, and teaches the individual strategies for control, the change in personality is
possible. In fact, research by Warren (1983) found certain factors seemed to predict
whether neurotic youth did well in the Community Treatment Project. They include:
disengagement from family problems, the availability of a strong and caring figure, and
the improvement of the individual’s self esteem. In sum, it is clear that environmental
125
contingencies on behavior could serve to teach the individual the coping skills or sense of
coping efficacy needed to avoid negative situations or deal with them more effectively
when they do occur.
Related to the discussion of the interaction of personality and the environment, is
the relationship between personality and social support. Cullen (1994) theorizes that
social support reduces the probability of criminal involvement. Related to this,
individuals who perceive that others will be there for them should they need it experience
less stress and better psychological health. Many researchers assert that our perceived
network size and satisfaction is related to personality factors (Sarason, Pierce, & Sarason,
1990). Additionally, Vinokur, Schul and Caplan (1987) assert, “people who are disposed
to view themselves and their personal experiences in a negative light are thus
hypothesized to misperceive or underrate the social support provided to them” (p. 1138).
Hoboll and Wallfisch in examining the role of personality as a predictor of social support
ask “whether the mediating effect attributed to social support is a byproduct of
personality characteristics of persons who see themselves as successful, in control, or
who react and utilize social support in a manner different than persons who see
themselves as failures and helpless” (1984, p. 89).
As mentioned, individuals high in neuroticism were shown to be more distressed
on average than low neuroticism subjects which could be attributed to a lower threshold
for responding to stressful events (Bolger & Schilling, 1991). An individual predisposed
to negatively construe his or her environment may require much different levels of
support (Sarason et al., 1990). In one study, a negative correlation was found between
neuroticism and perceived availability of support with regards to reported or perceived
126
stress (Sarason et al., 1983). Moreover, another study suggested not only does a general
negative outlook influence our perceptions of reality but also may influence how much
support is actually obtained (Vinokur et al., 1987). Finally, a study by Russell, Booth,
Reed, and Laughlin (1997) found that the personality characteristics of extraversion and
neuroticism were related to perceived social support in both the cross sectional and
longitudinal analyses. In sum, certain personality types are likely to impact treatment
certain treatment strategies, reactions to sanctions such as incarceration, and interactions
with others. This impact should be taken into consideration when designing strategies
designed to change offending behavior.
Future Research Directions
Many of the implications outlined above should also serve as directions for future
research. A number of the questions related to personality and criminal behavior have
yet to be addressed. First, the research should investigate the process that underlines the
development and stability of personality. For example, the research should assess the
impact of cumulative continuity and personality into late adulthood. In addition, the
future studies should explore whether criminal-prone personalities interact with negative
social environments differently than types predicted to behave more prosocially.
Second, the impact of personality on assessment results and treatment strategies
should be further explored. Specifically, research should explore personality and its
relationship to predicting criminal behavior with different age, gender, and racial groups.
Arguably, empirical literature on assessment for these groups would provide the
foundation for the development of risk assessment technology. Similarly, research
should be conducted to develop and test treatment strategies to deal with highly neurotic
127
subjects. Testing the effectiveness of interventions with a variety of populations will
begin to unravel which treatment strategies work best with different individuals.
Third, race was significantly related to outcome in the current study. However,
sample size precluded an interaction analysis of race and personality. As with any
research involving race, the research is limited and highly controversial. Currently, there
is very little research to support the idea that racial differences in personality traits exist.
Even so, future research should explore the impact of race and personality as it relates to
criminal behavior. For example, research is still needed to determine whether we can
expect to find differences in personality by race within criminal populations.
In conclusion, it is anticipated that this study will provide a framework for further
research in an area relatively untapped. Hopefully, by exploring the impact of
personality on behavior and the effectiveness of correctional interventions, we will move
the field of corrections one step closer to becoming an effective agent in the rehabilitation
of offenders. As poignantly suggested by Costa and McCrae (1999), “people are neither
passive victims of their life circumstances nor empty organisms programmed by histories
of reinforcement. Personality is actively involved in shaping people’s lives” (p. 142).
Without exploring the underlying causes of behavior and understanding how those
factors impact treatment, the field will likely fall victim to another swing in the pendulum
towards more punitive and retributive polices.
128
REFERENCES
Achenbrach, T. M., & Edelbrock, C. S. (1983). Manual for the child behavior checklist and revised child behavior profile. Burlington: University of Vermont. Allison, P. D. (1984). Event history analysis: Regression for longitudinal event data. Beverly Hills, C.A.: Sage. Allport, G. W. (1937). Personality: A psychological interpretation. New York: Hold, Rinehart, & Winston.
Andrews, D. A. (1989). Recidivism is predictable and can be influenced: Using risk assessments to reduce recidivism. Forum on Corrections Research, 1(2), 11-18. Andrews, D. A. & Bonta, J. (1995). The Level of Supervision Inventory- Revised (LSI-R). North Tonawanda, NY: Multi-Health Systems. Andrews, D. A., & Bonta, J. (1998). The psychology of criminal conduct. Cincinnati: Anderson. Andrews, D., Bonta, J., & Hoge, R.,. (1990). Classification for effective rehabilitation: Rediscovering psychology. Criminal Justice and Behavior, 17, 19-52. Andrews, D. A., & Wormith, J. S. (1989). Personality and crime: Knowledge destruction and construction in criminology. Justice Quarterly, 6(3), 289-309.
Andrews, D., Zinger, I., Hoge, R., Bonta, J., Gendreau, P., & Cullen, F. (1990). Does correctional treatment work? A clinically relevant and psychologically informed meta-analysis. Criminology, 28(3), 369-404.
Baird, C. Heinz, R. & Bemus, B. (1979). The Wisconsin Case
Classification/Staff Deployment Project. Project Report No. 14. Madison, WI: National Council on Crime and Delinquency. Bandura, A. (1989). Human agency in social cognitive theory. American Psychologist, 44(9), 1175-1184. Bandura, A. (1995). Exercise of personal and collective efficacy in changing societies. In A. Bandura (Ed.), Self-efficacy in changing societies (pp. 1-45). New York: Cambridge University Press.
129
Bandura, A. (1999). Social cognitive theory of personality. In L. A. Pervin & O. P. John (Eds.), Handbook of personality: Theory and research (pp. 154-196). New York: The Guilford Press. Bandura, A., Reese, L., & Adams, N. E. (1982). Microanlysis of action and fear arousal as a function of differential levels of perceived self efficacy. Journal of Personality and Social Psychology, 43(1), 5-21.
Bandura, A., Barbaranelli, C., Caprara, G. V., & Pastorelli, C. (1996). Multifaceted impact of self efficacy beliefs on academic functioning. Child Development, 67, 1206-1222. Bierman, K. L., & Furman, W. (1984). The effects of social skills training and peer involvement on the social adjustment of preadolescents. Child Development, 55, 151-162. Block, J. (1961). The California child q-set. Palo Alto: Consulting Psychologists Press. Bolger, N., & Schillings, E. A. (1991). Personality and the problems of everyday life: The role of neuroticism in exposure and reactivity to daily stressors. Journal of Personality, 59, 355-386.
Bolger, N. & Zuckerman, A. (1995). A framework for studying personality in the stress process. Journal of Personality and Social Psychology, 69, 890-902. Burgess, E. W. (1928). Factors determining success or failure on parole. In A. J. H. A. A. Bruce, E. W. Burgess, & J. Landesco (Ed.), The workings of the indeterminate-sentence law and the parole system in Illinois (pp. 221-234). Springfield: IL: State Board of Parole. Buskstel, L. & Kilmann, P. (1980). Psychological effects of imprisonment on confined individuals. Psychological Bulletin, 88(3), 469-493. Capaldi, D. M., & Patterson, G. R. (1996). Can violent offenders be distinguished from frequent offenders: Prediction from childhood to adolescence. Journal of Research in Crime and Delinquency, 33(2), 206-231. Caspi, A., Moffitt, T. E., Silva, P. A., Stouthamer-Loeber, M., Krueger, R. F., & Schmutte, P. S. (1994). Are some people crime prone? Replications of the personality-crime relationship across countries, genders, races, and methods. Criminology, 32(2), 163-195.
130
Caspi, A., Henry, B., McGee, R. O., Moffitt, T. E., Silva, P. A. (1995). Temperamental origins of child and adolescent behavior problems: From age three to age fifteen. Child Development, 66, 55-68. Caspi, A., & Silva, P. A. (1995). Temperamental qualities at age three predict personality traits in young adulthood: Longitudinal evidence from a birth cohort. Child Development, 66, 486-498. Caspi, A. & Roberts, B. W. (1999). Personality continuity and change across the life course. In L. A. Pervin & O. P. John (Eds.), Handbook of personality: Theory and research (pp. 300-326). New York: The Guilford Press.
Caspi, A., Harrington, H. L., Moffitt, T. E., Begg, D., Dickson, N., Langley, J., and Silva, P. A. (1997). Personality differences predict health-risk behaviors in young adulthood: Evidence from a longitudinal study. Journal of Personality and Social Psychology, 73(5), 1052-1063. Cattell, R. B. (1946). The description and measurement of personality. New York: World. Cattell, R. B. (1950). Personality, a systematic theoretical and factual study. New York: McGraw-Hill. Cattell, R. B. (1965). The scientific analysis of personality. Baltimore: MA: Penguin Books.
Chambless, D. L., Cherney, J., Caputo, G. C. & Rheinstein, B. J. (1987). Anxiety disorders and alcoholism: A study with inpatient alcoholics. Journal of Anxiety Disorders, 1, 29-40. Church, A. T. (1994). Relating the tellegen and five factor models of personality structure. Journal of Personality and Social Psychology, 67(5), 898-909.
Cloninger, C. R., Sigvardsson, S., Prybeck, T. R., & Svrakic, D. M. Personality antecedents of alcoholism in a national area probability sample. European Archives of Psychiatry and Neuroscience. 245(4-5), 239-244. Cullen, F. T. (1994). Social support as an organizing concept for criminology: Presidential address to the academy of criminal justice sciences. Justice Quarterly, 11, 4, 527-559. Cullen, F. T. & Gendreau, P. (1989). The effectiveness of correctional rehabilitation: Reconsidering the ‘nothing works’ debate. In L. Goodstein and D. MacKenzie (Eds.) The American Prison: Issues in Research and Policy (pp. 23-44. New York: Plenum.
131
Cullen, F. T. & Gilbert, K. E. (1982). Reaffirming rehabilitation. Cincinnati: Anderson. Derlega, V. J., Winstead, B. A., & Jones, W. H. (1991). Personality: Contemporary theory and research. Chicago: Nelson-Hall Publishers. Epstein, S., & O'Brien E. J. (1985). The person-situation debate in historical and current perspective. Psychological Bulletin, 98(3), 513-537. Eysenck, H., J. (1947). Dimensions of personality. New Brunswick: Transaction Publishers. Eysenck, H., J. (1952). Personality. Annual Review of Psychology, 3, 151-174. Eysenck, H., J. (1964). Crime and personality. Boston: Houghton Mifflin.
Eysenck, H. J. (1983). Personality, conditioning, and antisocial behavior. In W. a. D. Lauffer, J. (Ed.), Personality theory, moral development, and criminal behavior (pp. 51-81). Lexington, MA: Lexington Books. Eysenck, H. J. (1992). Anxiety: The Cognitive Perspective. Hillsdale, N.J.: Lawrence Erlbaum Associates.
Eysenck, H., J. (1996). Personality and crime: Where do we stand. Psychology, Crime & Law, 2, 143-152. Eysenck, H., J., & Eysenck, M. W. (1985). Personality and individual differences: A natural science approach. New York: Plenum Press. Farrington, D. P. (1982). Longitudinal analyses of criminal violence. In M. E. W. N. A. Weiner (Ed.), Criminal violence (pp. 171-200). Beverly Hills: Sage.
Farrington, D. P. (1994). Human development and criminal careers. In R. M. Mike Maguire, & Robert Reiner (Ed.), The oxford handbook of criminology (pp. 511-584). New York: Oxford University Press. Farrington, D. P. (1999). A criminological research agenda for the next millennium. International Journal of Offender Therapy and Comparative Criminology, 43(2), 154-167. Farrington, D. P., & West, D. J. (1990). The Cambridge study in delinquent development: A long term follow up of 411 London males. In H. J. K. G. Kaiser (Ed.), Kriminalitat (pp. 115-138). New York: Springer-Verlag.
132
Farrington, D. P., Loeber, R., Elliott, D. S., Hawkins, D., Kandel, D. B., Klein, M. W., McCord, J., Rowe, D. C., & Tremblay, R. E. (1990). Advancing knowledge about the onset of delinquency and crime. In B. L. A. E. Kazdin (Ed.), Clinical child psychology (Vol. 13, pp. 283-342). New York: Plenum Press. Gendreau, P., & Ross, R. (1987). Revivification of rehabilitation: Evidence from the 1980s. Justice Quarterly, 4(3), 349-407. Gendreau, P., Andrews, D. A., Goggin, C., & Chanteloupe, F. (1992). The development of clinical and policy guidelines for the prediction of criminal behavior in criminal justice settings (Unpublished manuscript ). St. John: University of New Brunswick, Department of Psychology. Gendreau, P., Little, T., & Goggin, C. (1996). A meta-analysis of the predictors of adult offender recidivism: What works! Criminology, 34, 575-607. Gerstley, L. J., Alterman, A. I., McLellan, A. T., & Woody, G. E. (1990). Antisocial personality disorder in patients with substance abuse disorders: A problematic disorder? American Journal of Psychiatry, 147(2), 173-178. Glueck, S. & Glueck, E. (1959). Predicting delinquency and crime. Cambridge: Harvard University Press. Gottfredson, S. (1987). Prediction: An overview of selected methodological issues. In D. Gottfredson & M. Tonry (Eds). Predication and Classification: Criminal Justice Decision Making (pp. 21-51). Chicago: University of Chicago Press. Gottfredson, M. R., & Hirschi, T. (1990). A general theory of crime. Stanford: Stanford University Press. Gould, S. J. (1981). The mismeasure of man. New York: Norton. Grasmick, H. G, Tittle, C. R., Bursik, R. & Arneklev, B. (1993). Testing the core empirical implications of Gottfredson and Hirschi’s general theory of crime. Journal of Research in Crime and Delinquency, 30, 5-29. Hall, C. S., & Lindzey, G. (1957). Theories of personality. New York: Wiley. Hare, R. D. (1982). Psychopathy and the personality dimensions of psychoticism, extraversion, and neuroticism. Personality and Individual Differences, 3(1), 35-42. Hare, R. D. (1991). The psychopathy checklist-revised. Toronto, Canada: Multi-Health Systems.
133
Hare, R. D. & McPherson, L. M. (1984). Violent and aggressive behavior by criminal psychopaths. International Journal of Law and Psychiatry, 7(1), p. 35-50. Harris, G.T., Rice, M. E., & Cormier, C. A. (1991). Psychopathy and violent recidivism. Law & Human Behavior, 15(6), 625-637. Harris, P. W. (1988). The interpersonal maturity level classification system. Criminal Justice and Behavior, 15(1), 58-77. Hathaway, S. R., & Meehl, P. E. (1951). An atlas for the clinical use of the MMPI. Minneapolis: MN: University of Minnesota Press. Heide, K. (1982). Classification of offenders ordered to make restitution by interpersonal maturity level and by specific personality dimensions. Unpublished Dissertation, State University of New York at Albany, Albany. Hemphill, J. F. & Hare, R.D. (1995). Psychopathy checklist factor scores and recidivism.” Issues in Criminological and Legal Psychology, 24, 68-73. Hemphill, J.F., Hare, R.D., & Wong, S. (1998). Psychopathy and recidivism: A review.” Legal and Criminological Psychology, 3(1), 139-170. Hindelang, M. J. (1981). Variations in sex-race-age-specific incidence rates of offending. American Sociological Review, 46, 461-474. Hirschi, T., & Hindelang, M. (1977). Intelligence and delinquency: A revisionists review. American Sociological Review, 42(4), 571-587. Hitchcock, G. L. (1995). The efficacy of cognitive group therapy with incarcerated psychopaths. Dissertation Abstracts International Section B: The Sciences and Engineering, 56(1B). Hobson, J., Shine, J., & Roberts, R. (2000). How do psychopaths behave in a prison therapeutic community? Psychology, Crime, & Law, 6(2), 139-154. Hoffman, P. B. (1994). Twenty years of operational use of a risk prediction instrument: The United States parole commission’s salient factor score. Journal of Criminal Justice, 22(6), 477-494. Hoffman, P. B. & Beck, J. L. (1980). Revalidating the salient factor score: A research note. Journal of Criminal Justice, 8, 185-188. Hoffman, P. B. & Beck, J. L. (1985). Recidivism among released federal prisoners: Salient factor score and five year follow-up. Criminal Justice and Behavior, 12(4), 501-507.
134
Hoffman, P. B. & Hardyman, P. L. (1986) Crime seriousness scales: Public perception and feedback to criminal justice policymakers. Journal of Criminal Justice, 14(5), 413-431.
Hooten, E. A. (1939). The American criminal. Cambridge: Harvard University Press. Forth, A.E., Hart, S. D., & Hare, R. D. (1990). Assessment of psychopathy in male young offenders. Psychological Assessment, 2(3), 342-344. Jeglum Bartusch, D. R., Lynam, D. R., Moffitt, T. E., & Silva, P. A. (1997). Is age important? Testing a general vs. a developmental theory of antisocial behavior. Criminology, 35(1), 13-48. Jesness, C. F. (1964). The Jesness Inventory. North Tonawanda N.Y: MultiHealth Systems, Inc. Jesness, C. F. (1983). The Jesness Inventory (rev. Ed). Palo Alto: CA: Consulting Psychologists Press.
Jesness, C. F. (1988). The jesness inventory classification system. Criminal Justice and Behavior, 15(1), 78-91. Jesness, C. F., & Wedge, R. (1983). Classifying offenders: The jesness inventory classification system. Sacramento: CA: California Youth Authority. John, O. P. (1990). The "big five" factor taxonomy: Dimensions of personality in the natural language and in questionnaires. In L. Pervin (Ed.), Handbook of personality theory and research (pp. 66-100). New York: Guilford. Kendrick, D. T., & Funder, D. C. (1988). Profiting from controversy: Lessons from the person-situation debate. American Psychologist, 43(1), 23-34. Kreuger, R. F. C., Avshalom, Moffitt, T. E., White, J. & Stouthamer-Loeber, M. (1996). Delay of gratification, psychopathology, and personality: Is low self-control specific to externalizing problems? Journal of Personality and Social Psychology, 64, 107-129. Krueger, R. F., Schmutte, P. S., Caspi, A., Moffitt, T. E., Campbell, K., and Silva, P. A. (1994). Personality traits are linked to crime among men and women: Evidence from a birth cohort. Journal of Abnormal Psychology, 103(2), 328-338. Krueger, R. F., Caspi, A., Moffitt, T. E., Silva, P. A. & McGee, R.. (1997). Personality traits are differentially linked to mental disorders: A multitrait-multidiagnosis study of an adolescent birth cohort. Journal of Abnormal Psychology, 105(3), 299-312.
135
Lee, C., & Bobko, P. (1994). Self efficacy beliefs: Comparison of five measures. Journal of Applied Psychology, 79(3), 364-369. Levinson, R. B. (1988). Developments in the classification process: Quay's aims approach. Criminal Justice and Behavior, 15(1), 24-38. Lilly, J. R., Cullen, F. T., & Ball, R. A. (1995). Criminological theory: Context and consequences. Thousand Oaks: Sage. Loeber, R., & Stouthamer-Loeber, M. (1990). Prediction. In H. C. Quay (Ed.), Handbook of juvenile delinquency (pp. 325-382). New York: John Wiley & Sons. Lombroso-Ferrero, G. (1911). Criminal man according to the classification of Cesar Lombroso. New York: Putnam.
Magnusson, D. (1990). Personality development from an interactional
perspective. In L. A. Pervin (Ed.), Handbook of personality: Theory and research (pp. 193-222). New York: The Guilford Press.
Maltz, M. D. (1984). Recidivism. Orlando: FL: Academic Press. Martinson, R. (1974). What works? Questions and answers about prison reform. Public Interest, 35(1), 22-54. McAvay, G. J., Seeman, T. E., & Rodin, J. (1996). A longitudinal study of change in domain-specific self efficacy among older adults. Journal of Gerontology, 51B(5), 243-253. McCrae, R. R. & Costa, P. T. (1999). A five-factor theory of personality. In L. A. Pervin & O. P. John (Eds.), Handbook of personality: Theory and research (pp. 139-153). New York: The Guilford Press. McCrae, R. R., & John, O. P. (1992). An introduction to the five factor model and its applications. Journal of Personality, 60, 329-361. McGue, M., Bacon, S. & Lykken, D. T. (1993). Personality stability and change in early adulthood: A behavioral genetic analysis. Developmental Psychology, 29, 96-109. McHale, J. (undated). I-4 delinquent subtypes: Definitions, similarities, and differences. Unpublished internal office document. Mischel, W. (1968). Personality assessment. New York: Wiley.
136
Mischel, W. (1990). Personality dispositions revisited and revised: A view after three decades. In L. A. Pervin (Ed.), Handbook of personality: Theory and research (pp. 111-134). New York: Guilford Press. Mischel, W. (1993). Introduction to personality (5th edition). Orlando: Fl: Holt, Rinehart, & Winston. Moffit, T. E. (1990). Juvenile delinquency and attention deficit disorder: Boys' developmental trajectories from age 3 to age 15. Child Development, 61, 893-910. Moffit, T. E. (1993). Life-course-persistent and adolescence-limited antisocial behavior: A developmental taxonomy. Psychological Review, 100, 674-107. Moffitt, T. E. (1993). The neuropsychology of conduct disorder. Development and Psychopathology, 5, 135-151. Murray, H. A. (1938). Explorations in personality. New York: Oxford University Press. Newcomb, M. D., & McGee, L. (1991). Influence of sensation seeking on general deviance and specific problem behaviors from adolescence to young adulthood. Journal of Personality and Social Psychology, 61(4), 614-628. Palloni, A. & Sorensen, A. B. (1990). Methods for the analysis of event history data: A didactic overview. In P. B. Baltes, D. L. Featherman, R. M. Lerner (Eds.), Life-Span Development and Behavior (291-323). Hillsdale, N.J.: Lawrence Erlbaum Associates. Palmer, T. (1974). The youth authority's community treatment project. Federal Probation, 38(1), 3-13. Patterson, G. R., Debarysbe, B. D., & Ramsey. (1989). A developmental perspective on antisocial personality. American Psychologist, 44, 329-335. Patterson, G. R., Crosby, L., & Vuchinich, S. (1992). Predicting risk for early police arrest. Journal of Quantitative Criminology, 8(4), 335-355. Pervin, L. A. (1990). A brief history of modern personality theory. In L. A. Pervin (Ed.), Handbook of personality: Theory and research (pp. 3-18). New York: Guilford Press. Quay, H. C., & Parsons L. (1972). The differential behavioral classification of the juvenile offender. Washington D.C.: Department of Justice.
137
Quay, H. C. (1987). Patterns of delinquent behavior. In H. C. Quay (Ed.), Handbook of juvenile delinquency (pp. 118-138). New York: A Wiley-Interscience Publication. Radloff, L. (1977). CESD scale: A self-report depression scale for research in the general population. Applied Psychological Measurement, 1, 385-401. Roy, A. (1999). Neuroticism and depression in alcoholics. Journal of Affective Disorders, 52(1-3), 243-245. Russell, D. W., Booth, B., Reed, D. & Laughlin, P. R. (1997). Personality, social networks, and perceived social support among alcoholics: A structural equation analysis. Journal of Personality, 65(3), 649-692. Sampson, R. J., & Laub, J. H. (1993). Crime in the making: Pathways and turning points through life. Cambridge: Harvard University Press. Sarason, P. I., Pierce, G. R., & Sarason, B. R. (1990). Social support and interactional processes: A triadic hypothesis. Journal of Social and Personal Relationships, 7(4), 495-506. Scheussler, K. F., & Cressey, D. R. (1950). Personality characteristics of criminals. American Journal of Sociology, 55, 476-484. Seto, M. C. & Barbaree, H. E. (1999). Psychopathy, treatment behavior, and sex offender recidivism. Journal of Interpersonal Violence, 14(12), 1235-1248. Silva, P. A. (1990). The Dunedin multidisciplinary health and development study: A fifteen year longitudinal study. Pediatric & Prenatal Epidemiology, 4, 96-127. Simourd, L., & Andrews, D. A. (1994). Correlates of delinquency: A look at gender differences. Forum on Corrections Research, 6, 26-31.
Smail, P., Stockwell, T., Canter, S. & Hodgson, R. (1984). Alcohol dependence and phobic anxiety states: A prevalence study. British Journal of Psychiatry, 144, 53-57. Spielberger, C. D. (1985). Anxiety, cognition, and affect: A state-trait perspective. In A. H. T. J. Maser (Ed.), Anxiety and the anxiety disorders (pp. 171-182). Hillsdale, NJ: Lawrence Erlbaum Associates Publishers. Stattin, H., & Magnusson, D. (1989). The role of early aggressive behavior in the frequency, seriousness, and types of later crime. Journal of Consulting and Clinical Psychology, 57(6), 710-718.
138
Sullivan, C., Grant, M. Q., & Grant, D. (1957). The development of interpersonal maturity: An application to delinquency. Psychiatry, 20, 373-386. Tellegen, A. (1985). Structures of mood and personality and their relevance to assessing anxiety, with an emphasis on self report. In A. H. T. J. Maser (Ed.), Anxiety and the anxiety disorders (pp. 681-706). Hillsdale: Lawrence Erlbaum Associates Publishers. Tellegen, A. (1991). Personality traits: Issues of definition, evidence, and assessment. In W. M. G. D. Cicchetti (Ed.), Thinking clearly about psychology: Vol 2: Personality and psychopathology (pp. 10-35). Minneapolis: Minnesota Press. Tennenbaum, D. J. (1977). Personality and criminality: A summary and implications of the literature. Journal of Criminal Justice, 5, 225-235. Thomas, D. (1977). Theoretical perspectives on prisonization: A comparison of importation and deprivation models. Journal of Criminal Law and Criminology, 68, 135-145. Tupes, E. C., & Christal, R. E. (1961). Recurrent personality factors based on trait rating (USAF ASD Technical Report). Texas: Lackland Air Force Base. Van Voorhis, P. (1987). Correctional effectiveness: The high cost of ignoring success. Federal Probation, 51(1), 56-62. Van Voorhis, P. (1988). A cross classification of five offender typologies: Issues of construct and predictive validity. Criminal Justice and Behavior, 15(1), 109-124. Van Voorhis, P. (1994). Psychological classification of the adult male prison inmate. New York: State of New York Press. Van Voorhis, P. (1997). An overview of offender classification. In M. B. P. Van Voorhis, & D. Lester (Ed.), Correctional counseling and rehabilitation (pp. 81-108). Cincinnati: Anderson.
Van Voorhis, P., Spruance, L. M., Listwan, S. J., Ritchey, P. N., Pealer, J., & Seabrook, R. (2001). The Georgia Cognitive Skills Experiment Outcome Evaluation Phase One. Technical Report submitted to the Georgia Board of Pardons and Parole.
Vinokur, A., Schul, Y. & Caplan, R.D. (1987). Determinants of perceived social
support: Interpersonal transactions, personal outlook, and transient affective states. Journal of Personality and Social Psychology, 53(6), 1137-1145. Vold, G. B., & Bernard, T. J. (1986). Theoretical criminology, 3rd edition. New York: Oxford University Press.
139
Waldo, G. P., & Dinitz, S. (1967). Personality attributes of the criminal: An analysis of research studies 1950-1965. Journal of Research in Crime and Delinquency, 4, 185-202. Wallace, J. F., Schmitt, W. A., Vitale, J. E., & Newman, J. (2000). Experimental Investigations of information processing deficiencies in psychopaths: Implications for diagnosis and treatment. In C. B. Gacono (Ed.), The Clinical and Forensic Assessment of Psychopathy: A Practitioner’s Guide (pp. 87-109). Mahwah, N.J: Lawrence Erlbaum Associates. Warren, M. (1983). Application of interpersonal maturity theory to offender populations. In W. Laufer & J. Day (Eds.), Personality Theory, Moral Development, and Criminal Behavior. Lexington, M.A.: Lexington Books. Warren, M. (1966). Interpersonal maturity level classification: Diagnosis and treatment of low, middle, and high maturity delinquents. Sacramento: California Youth Authority. Watson, D., & Clark, L. (1984). Negative affectivity: The disposition to experience aversive emotional states. Psychological Bulletin, 96(3), 465-490. Willett, J. B. & Singer, J. D. (1993). Investigating onset, cessation, relapse, and recovery: Why you should, and how you can, use discrete-time survival analysis to examine event occurrence. Journal of Consulting and Clinical Psychology, 61(6), 952-965. Winter, D. G. & Barenbaum, N. B. (1999). History of modern personality theory and research. In L. A. Pervin & O. P. John (Eds.), Handbook of personality: Theory and research (pp. 3-30). New York: The Guilford Press.
Zuckerman, M., Eysenck, S., & Eysenck, H. J. (1978). Sensation seeking in England and America: Cross cultural, age, and sex comparisons. Journal of Consulting and Clinical Psychology, 46(1), 139-149.
140
APPENDIX A
141
Table 1. Percentage and frequency distribution of inmates’ social demographic characteristics _______________________________________________________________________ Original Current
Sample Sample Characteristic N % N % (N = 369) (N = 277) ______________________________________________________________________ Race White 240 64.5 184 66.4 Black 105 28.6 77 27.8 American Indian 6 1.6 6 2.2 Hispanic 13 3.5 8 2.9 Asian 3 0.8 2 0.7 Age at admission 19 to 29 111 30.6 90 32.7 30 to 45 198 54.5 154 56.0 46 and older 54 14.9 31 11.3 Marital Status Married 145 40.2 109 39.9 Never married 85 23.5 66 23.1 Divorced 80 22.2 66 23.1 Separated 21 5.8 16 5.9 Widowed 4 1.1 1 0.4 Common-law 25 6.9 21 7.7 Other 1 0.3 0 0.0 Number of Dependents None 136 37.8 98 36.3 One 78 21.7 57 27.1 Two 52 14.4 44 16.3 Three 57 15.8 42 15.6 Four 23 6.4 17 6.3 Five 7 1.9 6 2.2 Six or more 6 1.6 6 2.8 Education 6 to 11 years 128 35.3 91 33.3 High school 108 29.8 81 29.7 GED 52 14.3 46 16.8 Some post high school 43 11.8 31 11.4 College graduate 26 7.2 20 7.3 Some post college 6 1.7 4 1.5
142
Table 1. Percentage and frequency distribution of inmates’ social demographic characteristics ________________________________________________________________________ Original Current
Sample Sample Characteristic N % N % (N = 369) (N = 277) ________________________________________________________________________ Evidence of school failure Yes 165 46.6 125 47.2 No 189 53.4 140 52.8 Primary Occupation No occupation 43 12.0 26 9.6 Professional 30 8.4 19 7.0 Manager/Admin 39 10.9 27 10.0 Sales 18 5.0 13 4.8 Clerical 4 1.1 2 0.7 Craftsman 37 10.3 29 10.7 Equipment Operator 12 3.4 9 0.4 Laborer 127 35.4 105 38.9 Farmer 6 1.7 6 2.2 Service worker 3 0.8 2 0.7 Armed Services 3 0.8 3 1.1 Student 2 0.6 2 0.7 Househusband 2 0.6 2 0.8 Criminal occupation 31 8.6 24 8.9 Employment Status Not working 170 47.8 127 47.0 Full time 133 37.4 100 37.0 Occasionally 40 11.2 33 12.2 Status Unknown 13 3.7 10 3.7
143
Table 2. Percentage and frequency distribution of inmates’ prior and current offense record. ________________________________________________________________________ Original Current
Sample Sample Characteristic N % N % (N = 369) (N = 277) ________________________________________________________________________ Prior adult or juvenile record Yes 311 84.7 242 87.4 No 56 15.3 35 12.6 Prior adult arrests Yes 309 84.2 240 86.6 No 58 15.8 37 13.4 Prior adult convictions Yes 280 76.5 220 79.7 No 85 23.2 55 19.9 Conflicting evidence 1 0.3 1 0.4 Prior prison sentence Yes 146 47.4 108 44.8 No 162 52.6 133 55.2 Amount of prior time served Two years or less 52 29.1 42 30.4 25 mo. to 5 years 42 23.5 33 23.9 61 mo. to 10 years 23 12.8 13 8.7 More than 10 years 10 5.6 7 5.1 Time served unknown 52 29.1 44 31.9 Prior arrest dealing drugs Yes 86 28.0 66 27.6 No 221 72.0 173 72.4 Prior professional criminal activity Yes 85 27.8 65 27.2 No 221 72.2 174 72.8 Prior violence Yes 130 42.8 104 43.7 No 174 57.2 134 56.3
144
Table 2. Percentage and frequency distribution of inmates’ prior and current offense record. ________________________________________________________________________ Original Current
Sample Sample Characteristic N % N % (N = 369) (N = 277) ________________________________________________________________________Prior arrest for same offense As present offense Yes 77 24.9 60 25.0 No 232 75.1 180 75.0 Prior or current drug related offense Yes, prior 31 8.7 26 9.7 Yes, current 54 15.2 42 15.7 Yes, both 61 17.1 47 17.5 No 210 56.9 153 57.1 Prior or current alcohol offense Yes, prior 67 18.7 54 19.8 Yes, current 14 3.9 11 4.0 Yes, both 30 8.4 21 7.7 No 248 69.1 187 68.5 Prior prison escapes Yes 30 16.8 22 16.2 No 149 83.2 114 83.8 Prior probation revocations Yes 65 36.7 50 35.0 No 110 63.3 89 65.0 Prior parole revocations Yes 46 41.1 36 10.5 No 66 58.9 45 55.6 Prior record as a juvenile Yes 101 29.3 81 30.5 No 244 70.7 185 69.5 Prior incarceration as a juvenile Yes 55 59.1 43 58.1 No 38 40.9 31 41.9
145
Table 2. Percentage and frequency distribution of inmates’ prior and current offense record. ________________________________________________________________________ Original Current
Sample Sample Characteristic N % N % (N = 369) (N = 277) ________________________________________________________________________ Prior violent record as juvenile Yes 28 30.4 19 25.7 No 64 69.6 55 74.3
146
APPENDIX B
147
Table 1. Model 1 Logistic Regression: Probability of Rearrest by Personality & Control Variables Parameter Standardized Odds Ratio Estimate Estimate Group .221 .216 1.247 Race .489** .197 1.631
SFS Risk Poor 1.452 .373 4.273
Fair .673 .292 1.959 Good .495 .259 1.641 Time Year 2 .024 .354 1.024 Year 3 -.240 .388 .787 Year 4 .328 .346 1.388 Year 5 .337 .356 1.401 Year 6 .410 .363 1.506 Year 7 .185 .391 1.203 Year 8 -.148 .442 .863 Year 9 -1.080 .635 .340 Year 10 .319 .488 .727 Year 11 .324 .416 1.383 Year 12 -.029 .647 .971 Personality Aggressive .630* .291 1.877 Neurotic .879* .328 2.409 Situational .439 .361 1.551 Intercept -4.081*** Model Chi Square 59.077*** * p <.05 ** p <.01 *** p <.001
148
Table 2. Model 1 Logistic Regression: Probability of Rearrest by Personality & Control Variables Parameter Standardized Odds Ratio Estimate Estimate SFS Riska .559** .188 1.749 Raceb .636*** .185 1.888 Personalityc Dependent -.610** .281 .543 Neurotic .229 .230 1.348 Situational -.127 .273 .881 Intercept -3.080 Model Chi Square 33.520 * p <.05 ** p <.01 *** p <.001 a Categories: 0= low; 1= high b Categories: 0= white; 1= non-white c Aggressive type is the omitted variable
149
Table 3. Model 1 Logistic Regression: Probability of Rearrest by Personality & Control Variables Parameter Standardized Odds Ratio Estimate Estimate SFS Riska .559* .188 1.749 Raceb .636*** .185 1.888 Personalityc Aggressive -.229 .230 .742 Dependent .909* .322 .403 Situational -.426 .305 .653 Intercept -2.781 Model Chi Square 33.520 * p <.05 ** p <.01 *** p <.001 a Categories: 0= low; 1= high b Categories: 0= white; 1= non-white c Neurotic type is the omitted variable
150
Table 4. Model 1 Logistic Regression: Probability of Rearrest by Personality & Control Variables Parameter Standardized Odds Ratio Estimate Estimate SFS Riska .559** .188 1.749 Raceb .636*** .185 1.888 Personalityc Aggressive .127 .273 1.135 Neurotic .426 .305 1.531 Dependent -.483 .354 .617 Intercept -3.207 Model Chi Square 33.520 * p <.05 ** p <.01 *** p <.001 a Categories: 0= low; 1= high b Categories: 0= white; 1= non-white c Situational type is the omitted variable
151
Table 5. Model 1 Logistic Regression: Probability of Rearrest by I-Level & Control Variables Parameter Standardized Odds Ratio Estimate Estimate Groupa .385 .209 1.470 Raceb .418** .194 1.520
SFS Riskc .505* .208 1.657 Time Year 2 .022 .352 1.022 Year 3 -.251 .386 .778 Year 4 .316 .345 1.371 Year 5 .306 .355 1.357 Year 6 .354 .361 1.425 Year 7 .118 .389 1.125 Year 8 -.231 .440 .794 Year 9 -1.171 .634 .310 Year 10 -.402 .486 .669 Year 11 .219 .413 1.245 Year 12 -.196 .644 .822 Interpersonal Leveld I – 3 -.308 .378 .735 I – 4 -.054 .398 .948 Intercept -3.053*** Model Chi Square 43.795*** * p <.05 ** p <.01 *** p <.001 a Categories: 0= low; 1= high b Categories: 0= white; 1= non-white c I-2 is the omitted variable
152
Table 6. Model 1 Logistic Regression: Probability of Rearrest by I-Level & Control Variables Parameter Standardized Odds Ratio Estimate Estimate SFS Riska .494** .188 1.639 Raceb .684** .187 1.981 Interpersonal Leveld I – 3 -.280 .375 .756 I – 4 -.008 .394 .992 Intercept -2.958*** Model Chi Square 26.853*** * p <.05 ** p <.01 *** p <.001 a Categories: 0= low; 1= high b Categories: 0= white; 1= non-white c I-2 is the omitted variable
153
Table 7. Model 1 Logistic Regression: Probability of Rearrest by I-Level & Control Variables Parameter Standardized Odds Ratio Estimate Estimate SFS Riska .494** .188 1.639 Raceb .684** .187 1.981 Interpersonal Leveld I – 2 .008 .394 1.008 I – 3 -.272 .195 .762 Intercept -2.966*** Model Chi Square 26.853*** * p <.05 ** p <.01 *** p <.001 a Categories: 0= low; 1= high b Categories: 0= white; 1= non-white c I-4 is the omitted variable
154
Table 8. Model 1 Logistic Regression: Probability of Rearrest by I-Level & Control Variables Parameter Standardized Odds Ratio Estimate Estimate SFS Riska .494** .188 1.639 Raceb .684** .187 1.981 Interpersonal Leveld I – 2 .280 .375 1.323 I – 4 .272 .195 1.313 Intercept -3.238*** Model Chi Square 26.853*** * p <.05 ** p <.01 *** p <.001 a Categories: 0= low; 1= high b Categories: 0= white; 1= non-white c I-3 is the omitted variable
155
APPENDIX C
156
Table 1. Model 2 Logistic Regression: Probability of Multiple Rearrest by Personality & Control Variables Parameter Standardized Odds Ratio Estimate Estimate Group .260 .139 1.297 Race .633*** .129 1.884
SFS Risk
Poor .674** .246 1.963 Fair .597** .185 1.816 Good .214 .178 1.239 Time Year 2 -.335 .174 .715 Year 3 -.892*** .218 .410 Year 4 -.695*** .217 .499 Year 5 -.726** .231 .484 Year 6 -.859*** .259 .424 Year 7 -1.083*** .299 .339 Year 8 -1.395*** .361 .248 Year 9 -1.863*** .469 .155 Year 10 -1.619*** .433 .198 Year 11 -.722** .279 .462 Year 12 -1.248* .607 .287 Personality Aggressive -.232 .160 .793 Dependent .-.548* .217 .578 Situational -.565* .219 .568 Intercept -1.989*** Model Chi Square 172.89*** * p <.05 ** p <.01 *** p <.001
157
Table 2. Model 2 Logistic Regression: Probability of Rearrest by Personality & Control Variables Parameter Standardized Odds Ratio Estimate Estimate SFS Riska .826*** .125 2.284 Raceb .649*** .126 1.914 Personalityc Aggressive .274 .181 1.316 Neurotic .483* .216 1.620 Situational .035 .241 1.036 Timed Time 2 -.319 .175 .727 Time 3 – 6 -.756*** .148 .470 Time 7 – 12 -1.197*** .179 .302 Intercept -2.493*** Model Chi Square 181.246*** * p <.05 ** p <.01 *** p <.001 a Categories: 0= low; 1= high b Categories: 0= white; 1= non-white c Dependent type is the omitted variable d Time one is the omitted variable
158
Table 3. Model 2 Logistic Regression: Probability of Rearrest by Personality & Control Variables Parameter Standardized Odds Ratio Estimate Estimate SFS Riska .826*** .125 2.284 Raceb .649*** .126 1.914 Personalityc Dependent -.274 .181 .760 Neurotic .208 .160 1.231 Situational -.239 .195 .787 Timed
Time 2 -.319 .175 .727 Time 3 – 6 -.756*** .148 .470 Time 7 – 12 -1.197*** .179 .302 Intercept -2.219*** Model Chi Square 181.246*** * p <.05 ** p <.01 *** p <.001 a Categories: 0= low; 1= high b Categories: 0= white; 1= non-white c Aggressive type is the omitted variable d Time one is the omitted variable
159
Table 4. Model 2 Logistic Regression: Probability of Rearrest by Personality & Control Variables Parameter Standardized Odds Ratio Estimate Estimate SFS Riska .826*** .125 2.284 Raceb .649*** .126 1.914 Personalityc Aggressive .239 .195 1.270 Neurotic .447* .220 1.564 Dependent -.035 .241 .965 Timed Time 2 -.319 .175 .727 Time 3 – 6 -.756*** .148 .470 Time 7 – 12 -1.197*** .179 .302 Intercept -2.458*** Model Chi Square 181.246*** * p <.05 ** p <.01 *** p <.001 a Categories: 0= low; 1= high b Categories: 0= white; 1= non-white c Situational type is the omitted variable d Time one is the omitted variable
160
Table 5. Model 2 Logistic Regression: Probability of Multiple Rearrest by I-level & Control Variables Parameter Standardized Odds Ratio Estimate Estimate Group .238 .140 1.268 Race .580*** .128 1.785
SFS Risk
Poor .697** .250 2.008 Fair .681** .186 1.977 Good .294 .176 1.342 Time Year 2 -.345* .174 .708 Year 3 -.893*** .218 .409 Year 4 -.700*** .217 .496 Year 5 -.730** .231 .482 Year 6 -.874*** .259 .417 Year 7 -1.103*** .299 .332 Year 8 -1.423*** .361 .241 Year 9 -1.891*** .469 .151 Year 10 -1.635*** .433 .195 Year 11 -.807** .279 .446 Year 12 -1.290* .607 .275 Interpersonal Level I- 2 -.102 .295 .903 I-3 -.057 .135 .945 * p <.05 ** p <.01 *** p <.001
161
Table 6. Model 2 Logistic Regression: Probability of Rearrest by Personality & Control Variables Parameter Standardized Odds Ratio Estimate Estimate SFS Riska .878*** .125 2.407 Raceb .595*** .125 1.818 Interpersonal Levelc I- 2 -.213 .290 .809 I-3 -.069 .134 .933 Timed Time 2 -.329 .175 .727
Time 3 – 6 -.759*** .148 .470 Time 7 – 12 -1.215*** .179 .302 Intercept -2.196*** Model Chi Square 181.246*** * p <.05 ** p <.01 *** p <.001 a Categories: 0= low; 1= high b Categories: 0= white; 1= non-white c I-4 is the omitted variable d Time one is the omitted variable
162
APPENDIX D
163
Table 1. Model 3 Logistic Regression: Probability of Rearrest for a Drug Offense by Personality & Control Variables Parameter Standardized Odds Ratio Estimate Estimate Group .160 .408 1.173 Race .086 .310 1.090
SFS Risk Poor -.973 1.087 .378 Fair .141 .502 1.151 Good -.278 .471 .757 Time Year 2 .051 .544 1.052 Year 3 -1.193 .808 .303 Year 4 -.774 .698 .461 Year 5 -.763 .698 .466 Year 6 -1.860 1.074 .156 Year 7 -.741 .697 .477 Year 8 -1.839 1.074 .159 Year 9 -1.838 1.074 .159 Year 10 -.724 .698 .485 Year 11 -1.086 .809 .338 Year 12 -6.497 14.272 .002 Personality Aggressive -.997* .414 .369 Dependent -1.455* .665 .233 Situational -1.129* .589 .323 Intercept -2.870*** Model Chi Square 25.658 * p <.05 ** p <.01 *** p <.001
164
Table 2. Model 3 Logistic Regression: Probability of Rearrest for a Drug Offense by Personality & Control Variables Parameter Standardized Odds Ratio Estimate Estimate SFS Riska .168 .403 1.181 Raceb .101 .397 1.106 Personalityc Dependent -.491 .643 .612 Neurotic .984* .409 2.675 Situational -.138 .581 .871 Intercept -4.760*** Model Chi Square 8.494 * p <.05 ** p <.01 *** p <.001 a Categories: 0= low; 1= high b Categories: 0= white; 1= non-white c Aggressive type is the omitted variable
165
Table 3. Model 3 Logistic Regression: Probability of Rearrest for a Drug Offense by Personality & Control Variables Parameter Standardized Odds Ratio Estimate Estimate SFS Riska .179 .377 1.196 Raceb .109 .387 1.116 Personalityc Aggressive .138 .581 1.147 Neurotic 1.124* .582 3.077 Dependent -.355 .777 .701 Intercept -4.887 Model Chi Square 8.486 * p <.05 ** p <.01 *** p <.001 a Categories: 0= low; 1= high b Categories: 0= white; 1= non-white c Situational type is the omitted variable
166
Table 4. Model 3 Logistic Regression: Probability of Rearrest for a Drug Offense by Personality & Control Variables Parameter Standardized Odds Ratio Estimate Estimate SFS Riska .179 .377 1.196 Raceb .109 .387 1.116 Personalityc Aggressive .492 .642 1.636 Neurotic 1.479* .662 4.388 Situational .355 .777 1.426 Intercept -5.242*** Model Chi Square 8.486 * p <.05 ** p <.01 *** p <.001 a Categories: 0= low; 1= high b Categories: 0= white; 1= non-white c Dependentl type is the omitted variable
167
Table 5. Model 3 Logistic Regression: Probability of Rearrest for a Drug Offense by Personality & Control Variables Parameter Standardized Odds Ratio Estimate Estimate Group .082 .393 1.086 Race .050 .397 1.051
SFS Risk .183 .420 1.201 Time Year 2 .032 .542 1.033 Year 3 -1.208 .807 .299 Year 4 -.788 .696 .455 Year 5 -.777 .696 .460 Year 6 -1.874 1.073 .154 Year 7 -.761 .696 .274 Year 8 -1.856 1.073 .156 Year 9 -1.849 1.073 .157 Year 10 -.734 .696 .480 Year 11 -1.106 .807 .331 Year 12 -6.521 14.424 .001 Interpersonal Level I – 2 -.129 .790 .879 I – 3 -.704 .376 .495 Intercept -3.398*** Model Chi Square 19.566 * p <.05 ** p <.01 *** p <.001
168
APPENDIX E
169
Table 1. Model 3 Logistic Regression: Probability of Rearrest of a Property Offense by Personality & Control Variables Parameter Standardized Odds Ratio Estimate Estimate Group -.127 .373 .881 Race .842 .346 2.322
SFS Risk Poor 1.598** .591 4.944
Fair .149 .539 1.161 Good .548 .462 1.729 Time Year 2 .434 .655 1.544 Year 3 .073 .715 1.075 Year 4 .808 .622 2.244 Year 5 .555 .655 1.741 Year 6 .415 .680 1.514 Year 7 -1.207 1.124 .299 Year 8 .204 .716 1.226 Year 9 -1.187 1.124 .305 Year 10 -1.183 1.124 .307 Year 11 -1.141 1.124 .319 Year 12 -5.828 14.671 .003 Personality Aggressive -.016 .429 .984 Dependent -.470 .569 .625 Situational -.536 .621 .585 Intercept -4.803*** Model Chi Square 36.716** * p <.05 ** p <.01 *** p <.001
170
Table 2. Logistic Regression: Probability of Rearrest of a Property Offense by I-Level & Control Variables Parameter Standardized Odds Ratio Estimate Estimate Group .086 .359 1.089 Race .760 .342 2.137
SFS Risk .155 .366 1.168 Time Year 2 .438 .652 1.549 Year 3 .061 .713 1.063 Year 4 .790 .620 2.203 Year 5 .524 .653 1.689 Year 6 .370 .678 1.448 Year 7 -1.245 1.122 .288 Year 8 .159 .714 1.172 Year 9 -1.222 1.122 .295 Year 10 -1.219 1.122 .295 Year 11 -1.185 1.122 .306 Year 12 -5.931 14.851 .003 Interpersonal Level I – 2 .055 .795 1.057 I – 3 .067 .365 1.070 Intercept -4.700*** Model Chi Square 26.829* * p <.05 ** p <.01 *** p <.001
171
APPENDIX F
172
Table 1. Model 3 Logistic Regression: Probability of Rearrest of Violent Offense by Personality & Control Variables Parameter Standardized Odds Ratio Estimate Estimate Group .578 .529 1.783 Race .393 4.65 1.482
SFS Risk Poor 1.649 .971 5.110 Fair 1.701 .840 5.479 Good 1.035 .837 2.814 Time Year 2 .019 1.006 1.020 Year 3 .031 1.006 1.031 Year 4 .039 1.006 1.039 Year 5 -.642 1.230 .526 Year 6 .775 .874 2.170 Year 7 1.022 .845 2.779 Year 8 -.581 1.231 .559 Year 9 .135 1.007 1.144 Year 10 -.537 1.231 .585 Year 11 .601 .921 1.824 Year 12 -5.159 13.958 .006 Personality Aggressive -.128 .599 .879 Dependent .409 .709 1.505 Situational -.469 .876 .626 Intercept -6.606*** Model Chi Square 27.586 * p <.05 ** p <.01 *** p <.001
173
Table 2. Logistic Regression: Probability of Rearrest of a Violent Offense by I-Level & Control Variables Parameter Standardized Odds Ratio Estimate Estimate Group .755 .525 2.127 Race .519 .438 1.680
SFS Risk .864 .477 2.372 Time Year 2 .015 1.006 1.015 Year 3 .026 1.006 1.026 Year 4 .037 1.006 1.037 Year 5 -.648 1.230 .523 Year 6 .769 .874 2.158 Year 7 1.014 .845 2.757 Year 8 -.585 1.230 .557 Year 9 .127 1.007 1.135 Year 10 -.541 1.230 .582 Year 11 .590 .920 1.804 Year 12 -5.184 14.096 .006 Interpersonal Level I – 2 .508 .859 1.662 I – 3 .166 .503 1.181 Intercept -6.201*** Model Chi Square 24.817 * p <.05 ** p <.01 *** p <.001