Você está na página 1de 20

CRIMINAL JUSTICE AND BEHAVIOR

Simourd, Olver / CRIMINAL ATTITUDES

THE FUTURE OF CRIMINAL


ATTITUDES RESEARCH
AND PRACTICE

DAVID J. SIMOURD
Queen’s University and Providence Continuing Care Center
MARK E. OLVER
University of Saskatchewan, Saskatoon, Canada

This study sought to explore the underlying dimension(s) of the criminal attitude construct.
Exploratory factor analyses using an oblique rotation method were conducted separately on the
subscales of the Criminal Sentiments Scale–Modified among a sample of 381 violent male
offenders. These procedures yielded four factors reflecting generic criminal attitudes, specific
attitudes about the law, generic rationalizations consistent with criminal subcultures, and crimi-
nally oriented self-views (i.e., a criminal self-concept). Confirmatory factor analysis using struc-
tural equation modeling found these factors to be relatively robust. Supplemental analyses
revealed the factors were linked to criminal conduct outcome criteria. These results are dis-
cussed in terms of potential future theory, research, and practice of the criminal attitude
construct.

T he attitude construct has a rich tradition in social psychology and


criminology. Criminal attitudes may be viewed as the construct at
the interface between these two disciplines. Within social psychology,
there is broad support indicating that attitudes are causally linked to
behavior. The process by which this occurs (e.g., Ajzen & Fishbein,

AUTHORS’ NOTE: The authors are grateful to the staff at Collins Bay Institution
for their cooperation with this research. Appreciation is also extended to several
pseudoanonymous reviewers for their helpful comments on an earlier version of this
article. Portions of this article were completed while the first author was on staff at
Collins Bay Institution. Address correspondence to Dr. David Simourd, Forensic
Unit, Providence Continuing Care Center, 752 King St. W., Kingston, Ontario, Can-
ada K7L 4X3; e-mail: Simourdd@pccc.kari.net.
CRIMINAL JUSTICE AND BEHAVIOR, Vol. 29 No. 4, August 2002 427-446
© 2002 American Association for Correctional Psychology

427
428 CRIMINAL JUSTICE AND BEHAVIOR

1980) seems to be mediated through a person’s intention to engage in a


particular behavior, which is influenced by the person’s attitude
toward the behavior and certain social influences. Criminal attitudes
are inherent in traditional criminological theories such as strain
(Cloward, 1959), control (Hirschi, 1969), and differential association
(Sutherland, 1947). This latter theory contends that criminal behavior
is learned through interactions with peers, which includes learning
both the techniques and attitudes fundamental to criminal conduct.
The general personality and social psychology perspective (Andrews
& Bonta, 1994) views criminal attitudes as one of the “big four” risk
factors that play a strong role during the “psychological moment” of
crime. Aside from these broad theoretical discussions, relatively little
attention has been devoted to the criminal attitude construct. What is
missing is a reasonably articulated understanding of the criminal atti-
tude construct(s), the most appropriate methods of assessing the con-
struct(s), and suggestions as to how best provide practical interven-
tions that offer the greatest prospects for altering the criminal attitudes
of offenders.
One reason for the perplexing status of the criminal attitude con-
struct within mainstream criminal behavior research and practice is
the lack of uniform terminology. References to the construct within
popular criminological texts range from “antisocial attitudes” (Andrews &
Bonta, 1994) to “social cognition” (Blackburn, 1993), although the
conceptual constructs appear to be similar. The term criminal atti-
tudes, which broadly constitutes the constellation of criminally ori-
ented attitudes, values, beliefs, and rationalizations, appears to be a
suitable label and will be used in the present article.
Although the link between criminal attitudes and criminological
theory is useful and interesting, it is perhaps more important to under-
stand the empirical link between criminal attitudes and deviant behav-
ior. Literature reviews, particularly meta-analytic reviews, provide
important insights into this issue. The advantage of meta-analysis over
traditional narrative reviews is that they provide a quantitative index
(effect size) of the phenomenon in question. At least four meta-analyses
of the correlates and/or predictors of criminal conduct provide infor-
mation on the importance of criminal attitudes. In an examination of
the predictors of adult recidivism, Gendreau, Little, and Goggin
(1996) found “criminogenic need” (which was a composite measure
Simourd, Olver / CRIMINAL ATTITUDES 429

of criminal peers and criminal attitudes) to be the best predictor of


recidivism, having a greater effect size than criminal history, social
achievement, family factors, intelligence, substance abuse, personal
distress, and socioeconomic status. In a meta-analysis on the predic-
tors of prison misconduct, Gendreau, Goggin, and Law (1997) found
that criminal attitudes and behavior fell into the “strongest predictor”
category, which included age and criminal history variables.
Two meta-analyses within the juvenile delinquent literature have
found a similar pattern of results. Simourd and Andrews (1994) exam-
ined the correlates of delinquency for males and females, grouping
risk factors into six specific domains (social class, personal distress,
educational/vocational achievement, parental/family factors, temper-
ament, and antisocial peers and/or attitudes). The antisocial peers and/
or attitude domain had the greatest effect size of the six domains. This
review also found no relative differences between males and females
on either the ranking or magnitude of the risk factors across domains.
In a recent meta-analysis of juvenile delinquent recidivism, Cottle,
Lee, and Heilbrun (2001) found “delinquent peers” (a specific crimi-
nal attitude variable was not examined) fell in what could be consid-
ered a second tier of predictors behind criminal history. Overall, these
meta-analyses show that criminal attitude is a dominant variable
inherent in a range of criminal conduct criteria.
In spite of its theoretical and empirical relevance to criminal con-
duct, the criminal attitude construct has been largely overlooked in the
mainstream assessment and treatment of offenders. One reason for
this may be the lack of suitable assessment instruments. Although
there is a relatively comprehensive array of criminal attitude measures
(see Brodsky & Smitherman, 1983), few have received appropriate
empirical verification. Furthermore, the majority of existing instru-
ments lack appropriate underlying theory, which can aid in advancing
research and practice. The Psychological Inventory of Criminal Thinking
Styles (PICTS) (Walters, 1995a) and the Criminal Sentiments Scale–
Modified (CSS-M) (Shields & Simourd, 1991) are two exceptions,
with the former measuring attitude process (i.e., how a person thinks)
and the latter measuring attitude content (i.e., what a person thinks).
The PICTS is broadly based on the lifestyle theory of criminal con-
duct (Walters, 1990) and specifically on criminal thinking errors pro-
posed by Yochelson and Samenow (1976). The PICTS itself is a 32-
430 CRIMINAL JUSTICE AND BEHAVIOR

item questionnaire that primarily measures eight thinking styles,


including mollification (justifications/rationalizations for norm-
violating behavior), cutoff (lack of psychological deterrents to crime),
entitlement (sense of ownership), power orientation (overt aggres-
sion), sentimentality (atonement for criminal violations), superoptimism
(overconfidence in abilities), cognitive indolence (lazy thinking), and
discontinuity (inadequate self-discipline). Several studies have reported
on the reliability and validity of the PICTS (Walters, 1995a, 1995b,
1996; Walters, Elliott, & Miscoll, 1998).
The Criminal Sentiments Scale–Modified (Shields & Simourd,
1991) is a modified version of the original Criminal Sentiments Scale
(CSS) (Gendreau, Grant, Leipciger, & Collins, 1979). It is a self-
report instrument consisting of 41 items grouped into five subscales:
Attitudes Toward the Law (Law; e.g., “Pretty well all laws deserve our
respect”), Court (Court; e.g., “Almost any jury can be fixed”), Police
(Police; e.g., “The police are honest”), Tolerance for Law Violations
(TLV; e.g., “A hungry man has the right to steal”), and Identification
With Criminal Others (ICO; e.g., “People who have broken the law
have the same sorts of ideas about life as me”). The first three sub-
scales are combined to form the Law-Court-Police (LCP) subscale,
which assesses respect for the law and criminal justice system. The
TLV subscale follows the concept of “neutralization” (Sykes & Matza,
1957), which reflects specific justifications for criminal behavior. The
ICO assesses personal evaluative judgments about law violators.
The CSS has been used in numerous studies with a wide range of sam-
ples, including probationers, provincial prisoners, young offenders,
forensic prisoners, and university students (Andrews, Wormith, &
Kiessling, 1985; Wormith & Andrews, 1984, 1995). Wormith and
Andrews (1995) found that the original CSS predicted recidivism
among relatively young (i.e., 21 years old), mainly first-time property
offenders, whereas Simourd and van de Ven (1999) found that the
CSS-M predicted recidivism among older, violent offenders.
In spite of the relative success of the PICTS and CSS-M, there is a
need to further explore dimensions of the criminal attitude construct.
The CSS-M may be ideally suited for this purpose for several reasons.
First, it is an assessment tool rich in relevant correctional theory. Sec-
ond, the CSS-M subscales reflect different theoretical components,
which speak to the breadth of the criminal attitude construct. Third,
Simourd, Olver / CRIMINAL ATTITUDES 431

the CSS-M measures attitude content rather than process, which may
be a more fruitful approach.
The purpose of the study, therefore, was to expand our knowledge
of the criminal attitude construct through an examination of the under-
lying factor structure of the CSS-M among a sample of high-risk (i.e.,
violent) offenders. It was expected that at least three factors would
emerge that correspond to the inherent nature of the CSS-M subscales
(i.e., reflecting attitudes toward the criminal justice system and per-
sonnel, criminal rationalizations, and criminal associates). It was fur-
ther expected that the factors would be empirically linked to recidi-
vism. Success in this endeavor has potential for renewed interest in the
theoretical, empirical, and practical attention to the criminal attitude
construct.

METHOD

PARTICIPANTS

Participants were 381 male inmates of a medium-security Cana-


dian federal institution in Ontario, Canada. They ranged in age from
19 to 60 years (M = 30.8, SD = 8.2). The racial composition of the sam-
ple was Caucasian (59%), Black (24%), Aboriginal (9%), Asian (5%),
and Other (3%). The distribution of index offense types were robbery
(34.8%), assault (23.3%), murder/manslaughter (12.8%), and other
(29.1%; various weapons offenses). Ninety-one percent had a convic-
tion for at least one previous offense, with an average of 13.1 (SD =
12.0) offenses; 79% had at least one prior incarceration, with a mean
of 2.5 (SD = 2.4) incarcerations.

MEASURE

The CSS-M (Shields & Simourd, 1991) is a self-report instrument


that measures antisocial attitudes, values, and beliefs directly related
to criminal activity. In scoring the CSS-M, each endorsement of an
antisocial statement (or rejection of a prosocial one) yields 2 points,
whereas rejection of an antisocial statement (or acceptance of a
prosocial one) yields a score of 0. Undecided responses are scored as 1.
432 CRIMINAL JUSTICE AND BEHAVIOR

Thus, higher scores reflect the presence of greater criminal attitudes.


Research (Simourd, 1997; Simourd & van de Ven, 1999) has found
that the CSS-M possesses acceptable psychometric properties and
construct validity and can be used to predict criminal conduct among
serious violent offenders.

PROCEDURE

All federal offenders with current convictions for crimes against


persons are required by the National Parole Board and Correctional
Service of Canada policy to undergo a psychological assessment prior
to conditional release (i.e., parole). Case management staff determine
which offenders meet the criteria for a psychological assessment
based on the current convictions. As part of the assessment, the pres-
ent participants completed a battery of self-report psychometric tests,
including the CSS-M, either alone or in small groups (e.g., five or
less). Additional demographic and offense/incarceration history was
collected from available case file information.
Recidivism information was collected through the Offender Man-
agement System (OMS), a computerized database operated by the
Correctional Service of Canada. The OMS contains detailed case file
information on offenders’ criminal history, incarceration, and post-
release behavior. All offense-based and follow-up data were coded
from the OMS. The postrelease outcome measures included supervi-
sion violations, rearrests, violent rearrests, reconvictions, and rein-
carcerations. For the purposes of this study, recidivism was defined as
the occurrence (dichotomously scored as yes/no) of any of these crite-
ria during the follow-up period.
Two hundred and seven participants of the original sample were eli-
gible for the predictive criteria analysis (150 were still incarcerated at
the time of follow-up, 32 had been deported, and 2 were deceased). Of
these, 102 were rearrested during the follow-up period for a base rate
of 49.3%. The time in which participants were at risk to recidivate fol-
lowing release ranged from 1 to 2,100 days (M = 266.4, SD = 274.8).
The mean follow-up time of the recidivists (M = 140.1, SD = 145.6)
and nonrecidivists (M = 382.8, SD = 310.1) was statistically different,
t(205) = 7.2, p < .001. The mean length of time between assessment
Simourd, Olver / CRIMINAL ATTITUDES 433

and community release was 800.7 (SD = 543.0) days, ranging between
17 and 2,442 days.

RESULTS

DESCRIPTIVE STATISTICS OF THE CSS-M

The mean CSS-M score was 18.2 (SD = 12.0). The mean subscale
scores were as follows: Law = 2.9 (SD = 2.8), Court = 4.7 (SD = 3.8),
Police = 3.5 (SD = 2.9), LCP = 11.1 (SD = 8.2), TLV = 3.9 (SD = 3.6),
and ICO = 3.2 (SD = 1.9). Cronbach’s alpha was computed for the
aggregate scale and individual subscales. The aggregate scale demon-
strated good internal consistency (alpha = .91), whereas the individual
subscales ranged from moderate to high: Law (alpha = .72), Courts
(alpha = .76), Police (alpha = .72), LCP (alpha = .87), TLV (alpha =
.76), and ICO (alpha = .51).

EXPLORATORY FACTOR ANALYSIS

The individual subscales of the CSS-M were subjected to an


exploratory factor analysis using the following set of procedures. For
each subscale, a principle components analysis was initially con-
ducted using varimax rotation to arrive at a preliminary solution. Prin-
ciple axis factoring was then used to extract factors for the potential
remaining solutions on those subscales that had more than one factor.
In the case where a single-factor solution seemed to best fit the data
set, the unrotated factor matrix was examined. Where appropriate,
however, potential factor solutions were then rotated using varimax
and then promax until a satisfactory solution emerged.
Five criteria were used to determine whether factor solutions were
acceptable. First, the solution was as parsimonious as possible. A
good guideline to follow is that the final solution contains as few fac-
tors as possible. Second, the solution accounted for a reasonably large
amount of the variance. Third, each factor was homogeneous in con-
tent and thus had high internal consistency. Fourth, a good factor solu-
tion will possess the property of simple structure. That is, each vari-
434 CRIMINAL JUSTICE AND BEHAVIOR

able will load highly on a single factor, with remaining loadings on


other factors close to zero (Reise, Waller, & Comrey, 2000). Finally,
the solution should be readily interpretable and “make sense.” A factor
solution will be of little value if one is incapable of deciphering the
factors derived.
An oblique rotation method was used in addition to varimax for
several reasons. According to Reise et al. (2000), oblique rotations
will produce orthogonal solutions when the factors are truly uncorrelated;
however, orthogonal analyses will not yield oblique solutions because
the technique does not allow the factors to correlate. Moreover, oblique
solutions tend to be more replicable and are more likely to pass the cri-
terion of simple structure and “may represent a more realistic model-
ing of psychological phenomena” (Reise et al., 2000, p. 292).
Analysis of the LCP subscale yielded a two-factor solution, which
is presented in Table 1. Examination of the pattern matrix indicates
that a preponderance of normally keyed items loaded on the first factor
and reverse keyed items on the second factor. The correlation between
the two factors was r = .58. The final solution accounted for 31.86% of
the total variance. Factor 1 is labeled General Criminal Sentiments
because it reflects a more global and heterogeneous construct of crim-
inal attitudes and negative sentiments toward criminal justice person-
nel. Factor 2 is labeled Adversarial Beliefs Toward the Law because it
reflects a constellation of strongly adversarial beliefs toward laws in
general.
Separate factor analyses on the TLV and ICO subscales yielded single-
factor solutions, with the majority of items loading on one factor. The
factor pattern matrix of the TLV and ICO are presented in Table 2 and
Table 3, respectively. Using stringent loading criteria (i.e., .40), 7 of
the 10 original items loaded on the TLV factor, with the resulting mod-
ified factor retaining internal consistency (alpha = .75). This factor
was labeled Criminal Subcultural Beliefs because it reflects a general
nonconformist and criminal rationalization orientation. For the ICO
subscale, Table 3 shows that 2 items did not load, with item homoge-
neity increasing slightly by removing these items (alpha = .55). This
factor was labeled Criminal Self-Concept because it reflects personal-
ization of a criminally oriented nature.
Simourd, Olver / CRIMINAL ATTITUDES 435

TABLE 1: Pattern Matrix of Law-Court-Police Factor Solution

Factor
General Criminal Adversarial
Criminal Sentiments Scale–Modified Itema Sentiments Law Beliefs

The police are honest .793 –.165


Judges are honest and kind .642 –.109
The police are crooked* .641 .181
Lawyers are honest .614 –.181
Court decisions are fair .597 .00
Money can “fix” court decisions* .482 .108
The Crown produces fake witnesses* .468 .159
A judge is a good person .461 .00
A cop is a friend to people in need .445 .145
The police should be paid more .429 .00
The law is good .360 .260
Almost any jury can be fixed* .331 .297
Society needs more police .244 .132
Laws deserve our respect .226 .106
Law and justice are the same .185 .00
The law is rotten to the core* –.150 .809
The law only helps a few people* .00 .785
Laws are usually bad* –.151 .729
The police almost never help people* .00 .671
The law doesn’t help people* .00 .619
The law makes slaves of most people* .220 .586
Life would better with fewer cops* .129 .438
You cannot get justice in court* .278 .414
It’s our duty to obey all laws .123 .215
All laws should be obeyed .164 .168

Eigenvalue 5.45 5.39


Variance accounted for (%) 15.97 15.89
Alpha .84 .84

NOTE: Items comprising each factor are in bold. *Denotes reverse scored.
a. Some items abbreviated for simplicity.

CONFIRMATORY FACTOR ANALYSIS

In a confirmatory factor analysis, a factor structure is hypothesized


a priori and then tested for its degree of fit with the observed pattern of
covariance of the measured variables (Floyd & Widaman, 1995). Fac-
tor models can then be trimmed or modified and instruments can be
436 CRIMINAL JUSTICE AND BEHAVIOR

TABLE 2 Factor Loading Matrix of Tolerance for Law Violations Factor Solution

Factor: Criminal
Criminal Sentiments Scale–Modified Itema Subcultural Beliefs

Successful people break the law to get ahead* .710


It’s OK to break the law, but don’t get caught* .707
People like me must break the law to get ahead* .693
It’s best to earn an easy living, even by breaking the law* .610
Only obey laws that seem reasonable* .588
People commit crimes when they think they won’t get caught* .457
A hungry man has the right to steal* .417
You shouldn’t break the law to try to get ahead in life .395
There is never a good reason to break the law .352
It’s OK to get around the law, as long as you don’t break it* .352

Eigenvalue 2.99
Variance accounted for (%) 29.88
Alpha .75

NOTE: Items comprising each factor are in bold. *Denotes reverse scored.
a. Some items abbreviated for simplicity.

TABLE 3 Factor Loading Matrix of Identification With Criminal Others Factor


Solution

Factor: Criminal
Criminal Sentiments Scale–Modified Itema Self-Concept

I’m more like a professional criminal than other people* .488


I’m like other people who have trouble with the law* .473
I prefer to be with people who obey rather than break the law .464
I have little in common with people who don’t break the law* .459
I have the same ideas as people who break the law* .396
No one who breaks the law can be my friend .135

Eigenvalue 1.06
Variance accounted for (%) 17.71
Alpha .55

NOTE: Items comprising each factor are in bold. *Denotes reverse scored.
a. Some items abbreviated for simplicity.

refined based on the results of the analysis (Floyd & Widaman, 1995).
As an additional means of testing the robustness of the derived mod-
els, a confirmatory factor analysis was conducted using structural
equation modeling. EQS for Windows 5.6 (Bentler, 1997) was used to
Simourd, Olver / CRIMINAL ATTITUDES 437

perform this analysis. Parameters were estimated using the maximum


likelihood technique, and the goodness-of-fit model was evaluated
using Bentler’s (1990) comparative fit index (CFI). Generally, CFI
values of approximately .90 or greater are indicative of good-fitting
models (Tabachnick & Fidell, 1996).
As a preliminary test, an initial model was conducted based on the
original item arrangement of the LCP scale. This analysis yielded a
model comprising three oblique factors with simple structure (e.g.,
Law items loading exclusively on the Law factor). This model pro-
duced a relatively poor fit to the data (CFI = .71), χ2(269) = 1138.337,
p < .001, suggesting that the item organization of the rationally
derived LCP subscale was not empirically replicated and that an alter-
native model would provide a better fit.
A second model based on the factor solution obtained from the
exploratory analysis was tested. Only items loading above the cutoff
criterion (.40) were included, thus generating an oblique two-factor
model containing 18 of the original 25 LCP items. This model pro-
vided an acceptable fit to the data (CFI = .89), χ2(132) = 368.117, p <
.001, and no further modifications to the model were attempted to
maximize parsimony. A chi-square difference test was conducted to
test the changes in fit after making these modifications, indicating a
highly significant improvement in model fit, χ2(137) = 770.22, p <
.001.
As with the LCP subscale, preliminary models using the original
item arrangement of the TLV and ICO subscales were conducted. The
initial model based on the original structure of the TLV yielded a rea-
sonable fit (CFI = .84), χ2(34) = 179.296, p < .001, but that neverthe-
less could withstand further improvement. The exploratorily modified
TLV demonstrated acceptable fit (CFI = .89), χ2(19) = 107.031, p <
.001, which was a significant improvement over the original model,
χ2(15) = 72.265, p < .001.
The a priori model based on the ICO yielded a strong fit to the data
(CFI = .90), χ2(8) = 22.295, p < .005. Further attempts to improve fit
by removing an item (i.e., “No one who breaks the law can be my
friend”) that loaded well below our loading criterion did not produce
any meaningful increment in fit over the original model (CFI = .90),
χ2(4) = 16.680, p < .005, as assessed using the chi-square difference
test, χ2(4) = 5.615, ns.
438 CRIMINAL JUSTICE AND BEHAVIOR

CRITERION-RELATED VALIDITY ANALYSES

The next step was to explore the link between the CSS-M factors
and criminal conduct criteria. This was accomplished by correlating
the factors with two postdictive (total number of criminal convictions
and number of different offenses) and five predictive (supervision vio-
lation, rearrest, violent rearrest, reconviction, and reincarceration) cri-
teria. Table 4 presents these results. The factors were largely unrelated
to previous conviction information, with the exception of criminal
self-concept.
The magnitude of the correlations between the factors and predic-
tive criteria was generally modest but statistically significant. The
criminal self-concept factor had the broadest links to outcome criteria,
being significantly related to all criteria except violent rearrest. The
general criminal sentiments factor was unrelated to postdictive crite-
ria but significantly correlated with most predictive criteria. The
subcultural beliefs factor was linked to rearrest and violent rearrest,
whereas the adversarial law beliefs factor was unrelated to any
criteria.

DISCUSSION

The primary purpose of this study was to uncover latent dimensions


of criminal attitudes as measured by the CSS-M. Items comprising the
individual subscales of the CSS-M were separately subjected to an
exploratory factor analysis and then a confirmatory factor analysis
using structural equation modeling. These analyses yielded a total of
four factors, two from the LCP subscale and one each from the TLV
and ICO subscales. Factor 1 of the LCP reflected a global set of nega-
tive sentiments directed largely toward authority figures in the crimi-
nal justice system and was labeled General Criminal Sentiments. Fac-
tor 2 reflected a set of extremely pessimistic beliefs about the integrity
of the law and administration of justice and was labeled Adversarial
Beliefs Toward the Law. The Tolerance of Law Violation factor was
labeled Criminal Subcultural Beliefs because it reflected a general
nonconformist and criminal rationalization orientation. The ICO fac-
tor reflected personalization of a criminally oriented nature and was
Simourd, Olver / CRIMINAL ATTITUDES 439

TABLE 4 Correlations Between Criminal Sentiments Scale–Modified Factors


and Offense-Based Criteria

Postdictive Criteria Predictive Criteria


Measure TCON DOFF SUV RAR VAR REC REI

General Criminal
Sentiments .01 .04 .20* .22* .21* .09 .24*
Adversarial Law Beliefs –.05 –.08 .06 .04 .11 .02 .09
Criminal Subcultural
Beliefs .02 .02 .09 .18* .22* .00 .17*
Criminal Self-Concept .21* .20* .18* .16* .11 .17* .18*

NOTE: TCON = total number of convictions; DOFF = total number of different offenses;
SUV = supervision violations (yes/no); RAR = rearrest (yes/no); VAR = violent rearrest;
REC = reconvictions (yes/no); REI = reincarceration (yes/no).
*p < .05.

labeled Criminal Self-Concept. The CFA of the four factors provided


a strong fit to the data, with comparative fit indices hovering in the .89
to .90 range, which is at or slightly below the conventional standard for
acceptable fit (i.e., .90).
There are few studies for which to compare the results of the pres-
ent study, and those that do exist have employed different methodolo-
gies and/or are based on a different criminal attitude assessment
instrument. Kroner and Mills (1998) conducted a principal compo-
nents analysis with a varimax rotation on the entire scale of the origi-
nal version of the Criminal Sentiments Scale among a sample of vio-
lent and sexual offenders. A two-factor solution emerged that essen-
tially reflected item response format, with Factor 1 being mostly
true-keyed items and Factor 2 mostly false-keyed items. The factors
were unrelated to recidivism, which was defined as arrest and convic-
tion of a new offense following release. The results of this study are
more encouraging than those of Kroner and Mills (1998) and may be
due to the use of more comprehensive factor analytic techniques, sep-
arate analysis of the CSS-M subscales rather than the aggregate scale,
and use of less stringent predictive criteria. These differences notwith-
standing, there is considerable overlap in item content between the
factors derived by Kroner and Mills (1998) and those of the present
study providing some evidence to the strength of the underlying crimi-
nal attitude dimensions.
440 CRIMINAL JUSTICE AND BEHAVIOR

Simourd (1997) examined the psychometric properties and con-


struct validity of the Pride in Delinquency Scale (Shields & Whitehall,
1991), an alternate measure of criminal attitudes. This scale consists
of 10 items that evaluate the degree of pride or shame in committing
various criminal offenses, which is a different assessment orientation
than the CSS-M. A preliminary principal components analysis with
varimax rotation among violent offenders yielded two factors, one
representing attitudes toward criminal offenses, the other reflecting
attitudes characteristic of a criminal subculture. The relationship
between the factors and various criminal conduct criteria were mixed
but generally positively related. The content of the two factors of the
Pride in Delinquency Scale is similar in nature to that of the present
study as is the link between the measures and criminal conduct crite-
ria. These results crudely suggest multimethod convergent validity of
the underlying criminal attitude dimensions found in this study.
Traditional social psychology attitude theory distinguishes between
attitudes and beliefs, with beliefs serving as a cognitive building block
upon which attitudes are formed. With respect to criminally oriented
beliefs and attitudes, it is possible that criminal beliefs are the founda-
tion on which criminal attitudes are formed. This hypothesis was not
tested directly in the present study, although the generally weaker cor-
relations between the law beliefs and subcultural beliefs factors rela-
tive to the general criminal sentiments factor suggests general crimi-
nal sentiments may be a more robust criminal attitude construct.
Insight into this issue is important because it has implications regard-
ing which constructs, criminal beliefs, or criminal attitudes may be
more fruitful in terms of assessment and treatment. Further research is
needed to shed light on this issue.
Assessment is fundamental to the treatment and management of
offenders. The assessment of criminal attitudes would appear to be an
important and relevant assessment area among offenders, given its
theoretical and empirical relationship to criminal conduct. To a large
degree, however, the criminal attitude construct is virtually ignored by
people in the field. Even more disheartening is the fact that very few
correctional psychologists, those who represent the profession that is
supposed to lead the charge on applying contemporary assessment
technology, employ contemporary broad-based assessment technol-
ogy in their practice (Boothby & Clements, 2000). It is not surprising
Simourd, Olver / CRIMINAL ATTITUDES 441

that specialized domains such as criminal attitudes have been ignored


in actual practice given the apparent lack of attention to modern cor-
rectional assessment procedures. On the other hand, there are virtually
no criminal attitude–specific measures commercially available,
which means people must become aware of such measurement tools
primarily through professional upgrading either by way of attending
conferences or perusing the correctional literature.
One benefit of this study is that it provides direction as to the more
potent dimensions of the criminal attitude construct, which should aid
the development and refinement of measurement devices. One cannot
go too far astray in assessing criminal attitudes if he or she attends to
offenders’opinions about the justice system, negative views regarding
the law, criminal rationalizations, and a personalized criminal orienta-
tion. We and our colleagues are pursuing the development of addi-
tional structured assessment devices in these areas and will report on
them once they become available. Until this time, practitioners are
encouraged to consider using the CSS-M in their assessment and
treatment efforts.
It is important to note that structured criminal attitude measurement
devices—in fact, all correctional assessment devices for that matter—
must be driven by appropriate and relevant criminological theory.
Wormith (2001) has commented on an increasing trend in offender
risk assessment toward purely empirically derived scales over theory-
driven approaches. Although the empirical derived approach has
some advantages, a major shortcoming is that it is an atheoretical
approach that offers limited insights that may advance research and
practice.
The results of the present study provide further insight into offender
rehabilitation research. Andrews, Zinger, et al. (1990) have shown that
cognitive behavioral interventions (those that focus on the range of
cognitions that give rise to behavior) produce the best outcomes
among offenders as indicated by reduced recidivism. The modifica-
tion of criminal attitudes makes for a logical therapeutic target given
its cognitive nature. Unfortunately, the treatment of criminal attitudes
is virtually absent in corrections. One reason for this is the lack of
appropriate programs. Those interventions that focus on criminal atti-
tudes differ in the degree to which they address core issues and use
various therapeutic orientations. A distinction can be made between
442 CRIMINAL JUSTICE AND BEHAVIOR

indirect and direct criminal attitude interventions. Indirect criminal


attitude interventions are those in which discussion of criminal atti-
tudes are tangential to other therapeutic targets. Cognitive Living
Skills (Ross & Fabiano, 1985) and Cognitive Self-Change (Bush &
Bilodeau, 1993) are examples of two indirect programs. The major
thrust of these interventions is on thinking/decision-making skills,
with criminal attitudes being somewhat pedestrian. Direct criminal
attitude interventions, on the other hand, are those programs in which
identification, debate, and restructuring of criminal attitudes is the
central therapeutic activity. Andrews and his associates (Andrews,
Wormith, Kennedy, & Daigle-Zinn, 1977; Andrews, Young, Wormith,
Searle, & Kouri, 1973) developed one of the earliest direct criminal
attitude programs, which consisted of structured personal, social, and
moral discussions about the law. Recently developed programs, includ-
ing the Criminal Attitude Program (Simourd, 2001) and Counter-
Point (Van Dieten & Graham, 1998) programs, follow in this tradition
but include contemporary correctional theory, research, and practice
into the treatment curriculum.
Another limitation with indirect programs is that they tend to be
process (i.e., how a person thinks) rather than content oriented (i.e.,
what a person thinks). The difficulty with process-oriented interven-
tions is that they focus more on cognitive style than cognitive mean-
ing. Although both process and content are important contributors to
offending behavior, attitude-behavior research within social psychol-
ogy (e.g., Ajzen & Fishbein, 1980) suggests that attitude content
assumes a more antecedent role in behavior. In practical terms, a typi-
cal bank robber who successfully completes a process-oriented inter-
vention may indeed slow his or her mental speed, consider more
behavioral options, and improve on his or her overall decision-making
ability. However, his or her fundamental desire to rob a bank will be
unaffected unless his or her positive attitudes, values, or beliefs about
robbing a bank are altered. In fact, a worst case scenario is that the
bank robber develops better cognitive skills such that he or she slowly,
carefully, and ingeniously consider more novel, less impulsive, and
more successful methods of robbing a bank.
In a study of the effectiveness of a direct criminal attitude program,
Wormith (1984) found that the optimum link between attitude change
and improved behavior occurred when direct criminal attitude pro-
Simourd, Olver / CRIMINAL ATTITUDES 443

gramming was paired with training in self-control. That is, recidi-


vism was reduced most under conditions of more prosocial attitudes
and improved self-management skills. Moreover, the long-term suc-
cess of offenders was best determined by pretreatment-
posttreatment change scores from a test battery rather than test bat-
tery information from an isolated assessment. In other words, demon-
strated attitude change was a better indicator of recidivism than an iso-
lated attitude measurement.
The risk-need-responsivity model (Andrews, Bonta, & Hoge,
1990) is the dominant perspective in correctional rehabilitation. The
essence of this is that (a) services should be delivered proportionally to
the level of criminal risk (high-risk cases should receive more ser-
vices), (b) services should target the factors most closely related to
offending behavior, and (c) interventions should be delivered in such a
way that clients can best understand program material. The tentative
finding in the present study of a criminal self-concept relates particu-
larly to this latter principle. Wells (1978) viewed self-concept as a
causal antecedent to deviant behavior primarily because it serves as
motivational function. Someone with a “criminal” self-concept may
be motivated toward engaging in behavior (e.g., unconventional, non-
conformist, norm violation) that is consistent with this view. More-
over, he or she may become insensitive to the negative consequences,
both socially and psychologically, of such deviant behavior and may
even take pride in such consequences. Self-talk such as “I am a crimi-
nal and I believe criminals act this way” may dominate his or her
cognitions and ultimately behavior. The responsivity principle would
predict that interventions delivered that are insensitive to this orienta-
tion would produce substandard results.
Simourd (2001) conducted a preliminary test of this hypothesis by
examining the pretreatment and posttreatment criminal attitude change
scores of participants who completed a direct criminal attitude pro-
gram. Participants were categorized as having either a high or low
criminal self-concept using the six-item ICO subscale of the CSS-M
representing the Criminal Self-Concept factor obtained in the present
study. Results showed that participants with high criminal self-concepts
had significant pre- and postreductions in measured criminal atti-
tudes, whereas those with low criminal self-concepts had no change
(in fact, they had a slight increase, albeit not significant) in measured
444 CRIMINAL JUSTICE AND BEHAVIOR

criminal attitudes. Although these results are clearly preliminary and


await verification, they provide preliminary support for the possibility
that challenging a client’s reasons for and comfort with his or her
criminal self-views may be a fruitful therapeutic approach. Cognitive
restructuring and irrational beliefs techniques common in general
psychotherapy may be particularly useful in this regard as long as the
underlying focus is on criminal content.
In summary, offending behavior is the result of a dynamic interplay
between environmental and psychological factors. Criminal attitude
is but one of a host of important criminogenic variables that must be
considered in the assessment and treatment of offenders. Simourd
(1996) once referred to the criminal attitude construct as the “silent
partner” in crime because of its strong link to offending behavior but
rather clandestine appearance in research and practice. Academic
debates about the causal mechanisms involved in the criminal attitude/
criminal behavior relationship will and should continue. In the mean-
time, it is important that practitioners integrate this important con-
struct into their assessment and treatment efforts.

REFERENCES

Ajzen, I., & Fishbein, M. (1980). Understanding attitude and predicting social behavior.
Englewood Cliffs, NJ: Prentice Hall.
Andrews, D. A., & Bonta, J. (1994). The psychology of criminal conduct. Cincinnati, OH:
Anderson.
Andrews, D. A., Bonta, J., & Hoge, R. D. (1990). Classification for effective rehabilitation:
Rediscovering psychology. Criminal Justice and Behavior, 17, 19-52.
Andrews, D. A., Wormith, J. S., Kennedy, D. J., & Daigle-Zinn, W. J. (1977). The attitudinal
effects of structured discussions and recreational association between young criminal
offenders and undergraduate volunteers. Journal of Community Psychology, 5, 63-71.
Andrews, D. A., Wormith, J. S., & Kiessling, J. J. (1985). Self-reported criminal propensity and
criminal behavior: Threats to the validity of assessments of attitudes and personality.
Ottawa: Solicitor General Canada.
Andrews, D. A., Young, J. G., Wormith, J. S., Searle, C. A., & Kouri, M. (1973). The attitudinal
effects of group discussions between young criminal offenders and community volunteers.
Journal of Community Psychology, 1, 417-422.
Andrews, D. A., Zinger, I., Hoge, R. D., Bonta, J., Gendreau, P., & Cullen, F. T. (1990). Does cor-
rectional treatment work? A psychologically informed meta-analysis. Criminology, 28,
369-404.
Bentler, P. M. (1990). Comparative fit indexes in structural models. Psychological Bulletin, 107,
238-246.
Simourd, Olver / CRIMINAL ATTITUDES 445

Bentler, P. M. (1997). EQS for Windows 5.6. Multivariate Software Inc.


Blackburn, R. (1993). The psychology of criminal conduct: Theory research and practice.
Chichester, UK: Wiley.
Boothby, J. L., & Clements, C. B. (2000). A national survey of correctional psychologists. Crim-
inal Justice and Behavior, 27, 715-731.
Brodsky, S. L., & Smitherman, H. O. (1983). Handbook of scales for research in crime and
delinquency. New York: Plenum.
Bush, J., & Bilodeau, B. (1993). Options: A cognitive change program. Longmount, CO:
National Institute of Corrections.
Cloward, R. A. (1959). Illegitimate means, anomie, and deviant behavior. American Sociology
Review, 24, 164-176.
Cottle, C. C., Lee, R. J., & Heilbrun, K. (2001). The prediction of criminal recidivism in juve-
niles: A meta-analysis. Criminal Justice and Behavior, 28, 367-394.
Floyd, F. J., & Widaman, K. F. (1995). Factor analysis in the development and refinement of clin-
ical assessment instruments. Psychological Assessment, 7, 286-299.
Gendreau, P., Goggin, C., & Law, M. A. (1997). Predicting prison misconducts. Criminal Justice
and Behavior, 24, 414-431.
Gendreau, P., Grant, B. A., Leipciger, M., & Collins, C. (1979). Norms and recidivism rates for
the MMPI and selected experimental scales on a Canadian delinquent sample. Canadian
Journal of Behavioral Science, 11, 21-31.
Gendreau, P., Little, T., & Goggin, C. (1996). A meta-analysis of predictors of adult recidivism:
What works! Criminology, 34, 401-433.
Hirschi, T. (1969). Causes of delinquency. Los Angeles: University of California Press.
Kroner, D. G., & Mills, J. F. (1998). The structure of violent and antisocial attitudes among vio-
lent and sexual offenders. International Journal of Offender Therapy and Comparative
Criminology, 42, 246-257.
Reise, S. P., Waller, N. G., & Comrey, A. L. (2000). Factor analysis and scale revision. Psycho-
logical Assessment, 12, 287-297.
Ross, R. R., & Fabiano, E. A. (1985). Time to think: A cognitive model of delinquency prevention
and offender rehabilitation. Johnson City, TN: Institute of Social Science and Arts.
Shields, I. W., & Simourd, D. J. (1991). Predicting predatory behavior in a population of young
offenders. Criminal Justice and Behavior, 18, 180-194.
Shields, I. W., & Whitehall, G. C. (1991, December). The Pride in Delinquency Scale. Paper pre-
sented at the eastern Ontario correctional psychologists’ winter conference, Burritts Rapids,
Canada.
Simourd, D. J. (1996). Criminal attitudes: The silent partner in crime. Symposium on Violence
and Aggression (pp. 29-34). Saskatoon, Canada: University of Saskatchewan.
Simourd, D. J. (1997). The Criminal Sentiments Scale–Modified and Pride in Delinquency
Scale: Psychometric properties and construct validity of two measures of criminal attitudes.
Criminal Justice and Behavior, 24, 52-70.
Simourd, D. J. (2001, March). Criminal attitudes: A background of theory, assessment, and
intervention. Workshop presented at Diversion Services, Tulsa, Oklahoma.
Simourd, D. J., & van de Ven, J. (1999). Assessment of criminal attitudes: Criterion-related
validity of the Criminal Sentiments Scale–Modified and Pride in Delinquency Scale. Crimi-
nal Justice and Behavior, 26, 90-106.
Simourd, L., & Andrews, D. A. (1994). Correlates of delinquency: A look at gender differences.
Forum on Correctional Research, 6, 26-31.
Sutherland, E. H. (1947). Principles of criminology (6th ed.). Philadelphia: Lippincott.
446 CRIMINAL JUSTICE AND BEHAVIOR

Sykes, G. M., & Matza, D. (1957). Techniques of neutralization: A theory of delinquency. Amer-
ican Sociological Review, 22, 664-670.
Tabachnick, B. G., & Fidell, L. S. (1996). Using multivariate statistics (3rd ed.) New York:
HarperCollins.
Van Dieten, M., & Graham, I. (1998). Counter-Point: A program for attitude and behavior
change. Ottawa: John Howard Society of Ottawa-Carleton.
Walters, G. D. (1990). The criminal lifestyle: Patterns of serious criminal conduct. Newbury
Park, CA: Sage.
Walters, G. D. (1995a). The Psychological Inventory of Criminal Thinking Styles: Part I. Reli-
ability and preliminary validity. Criminal Justice and Behavior, 22, 307-325.
Walters, G. D. (1995b). The Psychological Inventory of Criminal Thinking Styles: Part II. Iden-
tifying simulated response sets. Criminal Justice and Behavior, 22, 437-455.
Walters, G. D. (1996). The Psychological Inventory of Criminal Thinking Styles: Part III. Pre-
dictive validity. International Journal of Offender Therapy and Comparative Criminology,
40, 105-112.
Walters, G. D., Elliott, W. N., & Miscoll, D. (1998). Use of the Psychological Inventory of Crimi-
nal Thinking Styles in a group of female offenders. Criminal Justice and Behavior, 25,
125-134.
Wells, L. E. (1978). Theories of deviance and the self-concept. Social Psychology, 41, 189-204.
Wormith, J. S. (1984). Attitude and behavior change of correctional clientele: A three year fol-
low-up. Criminology, 22, 595-618.
Wormith, J. S. (2001, July). Assessing offender assessment: Contributing to effective correc-
tional treatment. The ICCA Journal, pp. 12-23.
Wormith, J. S., & Andrews, D. A. (1984). Criminal sentiments and criminal behavior: A con-
struct validation. Ottawa: Solicitor General Canada.
Wormith, J. S., & Andrews, D. A. (1995, June). The development and validation of three mea-
sures of criminal sentiments and their role in the assessment of offender attitudes. Paper pre-
sented at the annual convention of the Canadian Psychological Association, Charlottetown,
Canada.
Yochelson, S., & Samenow, S. E. (1976). The criminal personality: Vol. 1. A profile of change.
New York: Jason Aronson.

Você também pode gostar