Download as pdf or txt
Download as pdf or txt
You are on page 1of 21

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/6474084

The Psychometric Validation of the Social Problem-


Solving Inventory--Revised with UK Incarcerated Sexual
Offenders

Article  in  Sexual Abuse A Journal of Research and Treatment · September 2007


DOI: 10.1007/s11194-007-9038-3 · Source: PubMed

CITATIONS READS

23 1,368

1 author:

Helen Wakeling
Ministry of Justice, UK
34 PUBLICATIONS   720 CITATIONS   

SEE PROFILE

All content following this page was uploaded by Helen Wakeling on 29 January 2016.

The user has requested enhancement of the downloaded file.


Sex Abuse
DOI 10.1007/s11194-007-9038-3
ORIGINAL ARTICLE

The Psychometric Validation of the Social


Problem-Solving Inventory—Revised with UK
Incarcerated Sexual Offenders

Helen C. Wakeling


C Springer Science+Business Media, LLC 2007

Abstract This study examined the reliability and validity of the Social Problem-Solving
Inventory—Revised (SPSI-R; D’Zurilla, Nezu, & Maydeu-Olivares, 2002) with a population
of incarcerated sexual offenders. An availability sample of 499 adult male sexual offenders
was used. The SPSI-R had good reliability measured by internal consistency and test-retest
reliability, and adequate validity. Construct validity was determined via factor analysis. An
exploratory factor analysis extracted a two-factor model. This model was then tested against
the theory-driven five-factor model using confirmatory factor analysis. The five-factor model
was selected as the better fitting of the two, and confirmed the model according to social
problem-solving theory (D’Zurilla & Nezu, 1982). The SPSI-R had good convergent validity;
significant correlations were found between SPSI-R subscales and measures of self-esteem,
impulsivity, and locus of control. SPSI-R subscales were however found to significantly cor-
relate with a measure of socially desirable responding. This finding is discussed in relation to
recent research suggesting that impression management may not invalidate self-report mea-
sures (e.g. Mills & Kroner, 2005). The SPSI-R was sensitive to sexual offender intervention,
with problem-solving improving pre to post-treatment in both rapists and child molesters.
The study concludes that the SPSI-R is a reasonably internally valid and appropriate tool to
assess problem-solving in sexual offenders. However future research should cross-validate
the SPSI-R with other behavioural outcomes to examine the external validity of the mea-
sure. Furthermore, future research should utilise a control group to determine treatment
impact.

Keywords Sexual offender . Social problem-solving . Scale validation . Social


Problem-Solving Inventory—Revised

H. C. Wakeling ()
Offending Behaviour Programmes Unit, HM Prison Service, 108 Cleland House, Page Street, London,
SW1P 4LN, England
e-mail: helen.wakeling@hmps.gsi.gov.uk
Springer
Sex Abuse

Introduction

The term ‘social problem-solving’ refers to the process of problem-solving as it occurs in the
natural environment (D’Zurilla & Nezu, 1982). Thus, the study of social problem-solving
examines all types of problems that might affect a person’s functioning. The problem-solving
model utilised for this study originates from D’Zurilla and Goldfried (1971), and was later
revised by D’Zurilla and Nezu in 1982. Within this model, problem-solving is defined as:
‘the self directed cognitive behavioural process by which a person attempts to identify
or discover effective or adaptive solutions for specific problems encountered in everyday
living’ (D’Zurilla & Nezu, 2001, p. 212). This definition considers problem-solving to be a
conscious, logical, effortful and focused activity.
Within this model, problem-solving is regarded as comprising two partially autonomous
dimensions: problem orientation and problem-solving style (Nezu, 2004). A positive problem
orientation (PPO) is one that involves a tendency to regard problems as challenges, be
optimistic about solving problems, perceive one’s ability to solve problems as strong, and
believe that successful problem-solving involves time and effort. PPO approaches are found
to facilitate adaptive coping in stressful situations and to reduce emotional distress. A negative
orientation (NPO) involves the tendency to view problems as threats, expect problems to
be unsolvable, doubt one’s ability to solve problems well, and become frustrated and angry
when faced with problems. An NPO thus tends to increase avoidance behaviour, emotional
distress and inappropriate responses to problematic situations.
Problem-solving style refers to the activities people engage in when attempting to cope
with social problems. There are three styles. Rational Problem-Solving (RPS) refers to the
use of systematic and planned application of skills to a problem to provide an adaptive
solution. A person engaging the Impulsive/Careless style (ICS) will likely only consider a
few solution alternatives, and will often implement the first idea that comes to mind. The
strategies involved are unsystematic and considered carelessly. An Avoidance style (AS) is
characterised by passivity, and over-dependence on others to solve problems. Those using this
style will try to avoid problems rather than confront them, will wait for problems to resolve
themselves, and try to negate responsibility for problems. PPO and RPS are constructive
dimensions found to be related to adaptive functioning and positive psychological well-
being, whereas NPO, ICS and AS are dysfunctional dimensions found to be associated with
maladaptive functioning and psychological distress (D’Zurilla & Nezu, 1999; D’Zurilla,
Nezu, & Maydeu-Olivares, 2002).
Recent research has shown that the absence of effective problem-solving skills is as-
sociated with interpersonal difficulties, behavioural problems, negative affect and psycho-
logical adjustment (see Antonowicz & Ross, 2005; McGuire, 2001 for thorough reviews).
Particularly relevant for this study is that offenders have been found to have less effec-
tive problem-solving abilities than other more well-adjusted groups (e.g. McMurran, Egan,
Richardson, & Ahmadi, 1999; Zamble & Porporino, 1988; Zamble & Quinsey, 1997). In fact,
problem-solving is amongst the deficiencies empirically linked to greater risk of involvement
in criminal activity (see Andrews, 1995, 2001 for reviews).
Regarding sexual offenders, the most recent meta-analysis of risk factors for sexual re-
cidivism by Hanson and Morton-Bourgon (2004) found that poor problem-solving was not
significantly related to sexual and violent recidivism amongst sexual offenders. However
this finding was based on a relatively small sample size (n = 378), and only three stud-
ies. Furthermore, these three studies did not use the SPSI-R to assess problem-solving. It
is therefore possible that a structured measure like the SPSI-R identifies aspects of poor
coping that were not examined in these previous three studies. Thus, despite Hanson and
Springer
Sex Abuse

Morton-Bourgon’s finding, it may be premature to rule out poor problem-solving as a crim-


inogenic need for sexual offenders. Certainly there is other evidence to suggest that sexual
offenders may have deficient problem-solving skills. For example, an early study by Grier
(1988) found that on the Means-End Problem-Solving Procedure (Spivack & Platt, 1980),
sexual offenders exhibited deficient problem-solving skills in interpersonal relationships,
compared to a non-offender control group. Barbaree, Marshall, and Conner (unpublished,
1988, cited in Fisher & Howells, 1993) reported that child molesters had deficiencies in
problem-solving, so that although they could recognise a problem and generate different op-
tions for responding, they chose the wrong option and failed to see the negative consequences
of their choice.
A recent study by Nezu, Nezu, Dudek, Peacock, and Stoll (2005) examined problem-
solving skills amongst child molesters. Their sample of 124 male incarcerated child mo-
lesters displayed higher levels of NPO and lower PPO problem-solving compared to the
general population, as measured by the Social Problem-Solving Inventory—Revised (SPSI-
R; D’Zurilla et al., 2002). In addition, the child molesters displayed higher levels of avoidant
and impulsive styles of problem-solving and lower levels of rational problem-solving. Social
problem-solving deficits were associated with sexual aggression, as measured by the Sexual
Aggression Severity Index (Nezu et al., 2005), and sexual deviance, as measured by the
Multiphasic Sex Inventory (Nichols & Molinder, 1984). The authors concluded that social
problem-solving factors appeared to be related to sexual offending dimensions amongst child
molesters.
The research outlined above provides evidence that sexual offenders may have problem-
solving deficits. Problem-solving skills have in fact been specifically proposed as impor-
tant causal pathways to sexual offence behaviour (Hudson, Ward, & McCormack, 1999;
Polaschek, Hudson, Ward, & Siegert, 2001), although the exact mechanisms of this relation-
ship are unclear. Nezu, Nezu, and Dudek (1998) suggested that problem-solving deficits may
lead offenders to rely more heavily on denial, avoidance and sexually deviant fantasies as ha-
bituated means of coping with life stress. Sexual offending behaviour would therefore reflect
a limited and destructive solution to the perpetrator’s life problems. Despite the Hanson and
Morton-Bourgon findings, social problem-solving may still be regarded as a dynamic risk
factor for sexual offending (Thornton, 2002), which needs to be targeted during treatment
by developing problem-solving skills, helping offenders to recognise previous poor coping
and practicing new coping strategies.
Evidence also exists to suggest that individuals who are deficient in problem-solving
skills can be given training to improve their abilities (McGuire, 2005). Problem-solving is
an important component within treatment programmes currently used within HM Prison
Service to target offending behaviour. The Sex Offender Treatment Programme (SOTP) is a
cognitive behavioural programme designed to reduce reoffending in sexual offenders, cur-
rently offered in over twenty five prisons across England and Wales (see Mann & Thornton,
1998 for a comprehensive review). One of the treatment targets within SOTP is enhanc-
ing problem-solving skills. Thus SOTP includes problem-solving interventions, comprising
specific treatment blocks on coping strategies. The aim of these treatment blocks is to in-
crease awareness of deficits in emotional coping, enhance understanding of functional and
dysfunctional coping strategies and practice effective problem-solving strategies, with the
aim of reducing reoffending.
There are two types of social problem-solving measures, process and outcome measures.
Process measures assess the general cognitive and behavioural activities which facilitate or
inhibit the discovery of effective or adaptive problem-solving solutions. They provide infor-
mation as to how people might approach and solve problems, and are typically self-report.
Springer
Sex Abuse

The two most widely used process measures are the Social Problem-Solving Inventory—
Revised (SPSI-R: D’Zurilla et al., 2002), and the Problem-Solving Inventory (PSI: Heppner
& Peterson, 1982). Outcome measures, on the other hand, assess the quality or performance
of specific solutions to problems, and are judged by the products of a problem-solving ac-
tivity. Examples of outcome measures include the Alternative Thinking Test (Spivack &
Platt, 1980), the Means-End Problem-solving Procedure (Spivack & Platt, 1980), and the
Adolescent Problems Inventory (Freedman, Rosenthal, Donahoe, Schlundt, & McFall, 1978;
Palmer & Hollin, 1996). The tools used to measure problem-solving have often been regarded
as inadequate or problematic (Butler & Meichenbaum, 1981; D’Zurilla, 1986). Most often
noted is the lack of evidence for the construct validity of many problem-solving measures.
The SPSI-R was employed by HM Prison Service in 2004 to measure problem-solving
within sexual offenders attending a treatment programme. The decision to use the SPSI-
R stemmed from the relative ease of using a self-report process measure for a National
based treatment programme rather than an outcome measure, in terms of both time and
standardisation. In addition to this, the author found greater support for the validity of the
SPSI-R tool in the literature compared to the PSI. The PSI has also not been linked to a specific
theory whereas the SPSI-R is rooted within a long-standing theory of problem-solving.
A wealth of research supports the use of the SPSI-R in terms of validity, test design,
development, and theoretical foundation. The SPSI-R is a 52-item self-report measure
which assesses the major components of the D’Zurilla and Nezu (1982) model of social
problem-solving outlined earlier. The SPSI-R was developed through a series of exploratory
and confirmatory factor analyses of the original 70-item theory driven SPSI, conducted
by Maydeu-Olivares and D’Zurilla (1995, 1996). The tool assesses the two constructive
problem-solving dimensions (PPO and RPS), and three dysfunctional dimensions (NPO,
ICS and AS). The SPSI-R scales have been found to have good psychometric properties
with non-offender populations (D’Zurilla et al., 1998, 2002; Maydeu-Olivares & D’Zurilla,
1996). Good internal consistency and test retest reliability are reported for all SPSI-R sub-
scales within the manual (D’Zurilla et al., 2002). Structural validity of the SPSI-R has been
verified in two independent samples using confirmatory factor analysis (Maydeu-Olivares
& D’Zurilla, 1996). Concurrent validity has been examined by correlating SPSI-R sub-
scales with the PSI (D’Zurilla & Nezu, 1990). Correlations between the PSI total score and
SPSI-R subscales were all significant (ranging from .54 and .69), indicating that the two
constructs tapped by the SPSI-R and PSI overlap. Predictive validity of the SPSI-R has
been demonstrated by correlating scores with several measures of psychological distress,
including depression, anxiety, worry, hopelessness and suicidal ideation (D’Zurilla et al.,
2002). NPO is consistently the best predictor of distress, with correlations between the NPO
subscale and psychological distress constructs ranging from .42 to .68, all significant. The
SPSI-R has also been shown to be related to positive psychological well-being and social
competence (D’Zurilla et al., 2002). Areas of self-esteem and locus of control are constructs
that overlap with social problem-solving processes (D’Zurilla & Maydeu-Olivares, 1995).
Good convergent validity for the SPSI-R subscales has been found with these two constructs
(D’Zurilla & Maydeu-Olivares, 1995; D’Zurilla & Nezu, 1990). Significant correlations be-
tween SPSI-R subscales and a measure of self esteem were found (ranging from .27 to .51),
whilst correlations with a measure of locus of control ranged from .23 to .30, all significant.
Discriminant validity of the SPSI-R subscales has also been demonstrated by examining the
relationship between the SPSI-R subscales and the Constructive Thinking Inventory (CTI;
Epstein & Meier, 1989), a measure of automatic thinking believed to be largely under the
control of experiential processing (Epstein, 1994). Rational problem solving was found to

Springer
Sex Abuse

have very little variance in common with automatic thinking, and non significant correlations
were found between SPSI-R subscales and CTI measures.
Despite the extensive validation of the SPSI-R tool with various populations, including
college students, adolescents, young adults, middle-aged adults, elderly adults, and psychi-
atric adults (D’Zurilla et al., 2002), to date there has been no validation of the psychometric
properties of the tool with sexual offender populations. When examining responses on a
measure within a specific population, such as sexual offenders, it is important that the re-
liability and validity of the measures has been established for that population (Kroner &
Weekes, 1996). Furthermore, issues relating to incarceration such as loss of freedom, adapt-
ing to structure, guilt and anxiety may all influence the relevance of test items (Zamble &
Porporino, 1988). Therefore, the content of instruments developed in non-prison settings,
such as the SPSI-R, may be inappropriate for use with prison populations. Previous research
has also shown that offenders may have a desire and ability to manipulate scale scores over
and above the normal population (Schretlen & Arkowitz, 1990). Thus, it is important to
consider the validity of self-report inventories with forensic populations, and in particular
the examination of potential response bias and socially desirable responding is critical.
The hypotheses for this study are therefore as follows. Firstly it is hypothesised that the
SPSI-R will be a reliable tool to use with sexual offenders as shown via internal consistency
and test-retest reliability. Secondly, it is hypothesised that the SPSI-R will be a valid tool
to use with sexual offenders, as shown via examination of construct validity through factor
analyses, convergent validity via correlations with measures of similar constructs, and by
low susceptibility to social desirable responding. Finally, it is hypothesised that if shown to
be a reliable and valid tool, there will be significant improvements in social problem-solving
in sexual offenders following an intervention to target problem-solving skills. To expand
on previous research using the SPSI-R which focused on child molesters only (Nezu et al.,
2005), both rapists and child molesters will be examined in the present study. The author
has found no reason to hypothesise that one type of sexual offender will be more deficient in
problem-solving than the other, and it is expected that problem-solving skills in both types
of sexual offenders will increase following cognitive-behavioural treatment. However due to
emergent differences between offence types in previous research examining psychometric
self report measures (e.g. Mann, Webster, Wakeling, & Marshall, in press; Webster, Mann,
Thornton, & Wakeling, in press), differences between rapists and child molesters will be
investigated in the present study.

Method

Participants

A total of 499 sexual offenders formed the non-random availability sample for this study.
All were incarcerated adult (aged over 18) male offenders who had undertaken a cognitive-
behavioural Sex Offender Treatment Programme (SOTP) between April 2004 and April
2005. SOTP is currently offered to men who have been imprisoned for a sexual offence or
an offence with a sexual element. Participation on the programme is voluntary, but requires
admission of the offence committed. Readers are directed to Mann and Thornton (1998)
for a detailed description of SOTP. The sample comprised all of the sexual offenders who
had undertaken either the Core programme designed for medium/high-risk offenders, or
the rolling programme designed for low risk offenders during this one year time period.
All offenders had an IQ of greater than 80 as tested by the Wechsler Abbreviated Scale
Springer
Sex Abuse

Table 1 Sample Demographic and Offence Specific Characteristics

Total Sample N = 499


Mean SD

Age 43.51 (12.73)


Number of Victims 1.96 (2.22)
Number of Preconvictions 5.06 (9.93)

Number %
Offence type
Child molester 387 (77.6)
Rapist 112 (22.4)
Victim gender
Male 74 (15.2)
Female 381 (78.1)
Male & Female 33 (6.8)
Relationship to victim
Intra-familial 182 (37.9)
Extra-familial 255 (53.1)
Intra and extra-familial 43 (9.0)
Ethnicitya
White 430 (91.3)
Black 34 (6.9)
Not stated 7 (1.5)
a The minority-ethnic participants have been defined as ‘black’ using the
term in its political sense to apply to all those non-Caucasians who are
either Black-African, Black-Caribbean, Black-Other or from the Asian
Subcontinent.

of Intelligence (WASI: Wechsler, 1981). There were two known dropouts included within
the sample for whom pre treatment psychometrics were collected and used for all analyses
apart from treatment impact. It is likely that there were more treatment dropouts for whom
psychometric data was not collected centrally and therefore not utilised for the present study.
Table 1 displays the demographic and offence characteristics of the sample. Cell size
discrepancies within the tables are the result of missing data.
The sample comprised 387 child molesters, defined as having a victim aged 15 years or
younger, and 112 rapists, defined as having a victim aged 16 years or older. The majority of
the sample had female victims (78.1%) as opposed to male victims (15.2%), or victims of
both genders (6.8%).
A further sample of 30 sexual offenders was used for the test-retest reliability analysis.
These offenders had all been assessed for their suitability to attend an SOTP but had not begun
treatment. The mean age of this group was 45.3, and examination of further characteristics
suggested they were a comparable sample of offenders to the main sample described above.
The offenders selected had had no treatment intervention targeted at problem-solving between
the two administration time points.

Measures

The Social Problem-solving Inventory Revised (SPSI-R; D’Zurilla et al., 2002) is a 52-
item self-report measure that assesses people’s ability to solve problems in everyday living.
Respondents rate each item on a 5-point Likert scale, with the following scale anchors:
Springer
Sex Abuse

0 = Not at all true of me, 1 = Slightly true of me, 2 = Moderately true of me, 3 = Very
true of me, 4 = Extremely true of me. Missing items are given a score of 2. The SPSI-R
contains five empirically derived subscales: Positive Problem Orientation (PPO: 5 items),
Negative Problem Orientation (NPO: 10 items), Rational Problem-solving Skills (RPS: 20
items), Avoidance Style (AS: 7 items), and Impulsivity/Carelessness Style (ICS: 10 items).
Convergent validity of the SPSI-R was examined using the following three measures
selected because of their similarity in construct to problem-solving. The Self-esteem ques-
tionnaire (Thornton, Beech, & Marshall, 2004) is an eight-item scale measuring general
levels of self-esteem. Webster et al. (in press) found the scale to have good internal consis-
tency (α = .84), excellent test-retest reliability, r(29) = .90, and good discriminant validity
tested on a sample of sexual offenders.
The Impulsivity Scale (Eysenck & Eysenck, 1978) is a 13-item scale measuring a tendency
to act without thinking about long-term consequences. Item responses are ‘yes’ or ‘no’, and
scale scores range from 0 to 26. With sexual offenders, the scale has been shown to have
adequate internal consistency (α = .84), and test-retest reliability, r(25) = .79 (Rallings &
Webster, 2001).
The Locus of Control Questionnaire (Levenson, 1972) is an 18-item questionnaire mea-
suring the extent to which a participant believes his experiences are determined by internal
or external factors. Items are scored on a 5-point Likert scale, with higher scores ( > 40)
indicating someone who attributes his experience to internal factors and lower scores indi-
cating a tendency to blame external factors. With sexual offenders, the internal consistency
of the scale is α = .79 (N = 1,883), and the test re-test reliability of the scale is r(32) = .87
(Rallings & Webster, 2001).
The issue of social desirability was examined by correlating scores on the SPSI-R with
scores on the Balanced Inventory of Desirable Responding (BIDR: Paulhus, 1984), a 40-item
self-report questionnaire, which measures the tendency to give socially desirable responses.
The BIDR can be administered alongside other measures and is useful in identifying indi-
viduals who may distort their responses. The BIDR contains two subscales, Self-Deception
and Impression Management. Scale items are scored dichotomously, with higher scores
indicating exaggerated desirable responses to questions. Paulhus (1984) reported excellent
psychometric properties for the BIDR.

Procedure

The sexual offenders completed all of the measures as part of a wider psychometric test battery
designed to assess treatment need in sexual offenders. Questionnaires were administered by
treatment staff within each prison. A standard set of instructions was given to participants
prior to completing assessments, and participants filled them out under supervision within a
group setting.
Following completion of the treatment programme, all participants were re-tested with
the same psychometric assessments, including the SPSI-R. Re-testing took place six weeks
post-treatment, and conformed to pre-treatment test conditions.
Test-retest reliability data were collected from an additional group of 30 sexual offenders.
These participants were located in five different prisons, ranging from higher security to
medium security. Participants for this section of the study completed the SPSI-R twice
before entering treatment. The mean length of time between testing sessions was 28.7 days
(SD = 12.8, range 13 to 51 days).
Springer
Sex Abuse

Analyses

Prior to analyses four outliers with SPSI-R scores on the original five factor model that were
greater than three Standard Deviations from the group mean were removed from the sample,
a process advised by Tabachnick and Fidell (1989) as a means of preventing very extreme
scores from affecting the mean. In addition, if more than 10% of an individual’s SPSI-R
data was missing, the individual was omitted from the sample in order to ensure scale scores
were meaningful. For cases where there was less than 10% missing data, the missing items
were filled with the median scale value, a technique commonly used with psychological
psychometric testing (Levesque, Gelles, & Velicer, 2000).

Results

Reliability

Internal consistency of SPSI-R items

Item-to-total correlations were examined on all 52 SPSI-R items. Five items had correlations
less than .20, the remainder being between .20 and .49. The Cronbach alpha of the total scale
was .86, indicating very high internal consistency. The alpha was not increased by deleting
any of the items.
Internal consistency for each of the SPSI-R subscales was also examined. The Cronbach
alphas ranged from good to excellent: PPO = .73; NPO = .92, RPS = .95, ICS = .91;
and AS = .82. On all of the subscales apart from AS deleting items did not increase the
subscale alpha. On AS, deleting item 42 increased the alpha from .82 to .87, suggesting that
the AS scale is more consistent when this item omitted: ‘When I am faced with a difficult
problem, I go to someone else for help in solving it’.
These alpha coefficients are comparable to four normative samples described in the SPSI-
R manual (adolescents, young adults, middle-aged adults and elderly adults; D’Zurilla et al.,
2002).

Test-retest reliability of SPSI-R items

Test-retest reliability of the total scale score was excellent, r(30) = .94, p < .001. Test-retest
reliabilities of all the subscales were very high: PPO = .75, NPO = .88, RPS = .85,
ICS = .91, and AS = .80. All correlations were significant (p < .001). These results
are comparable to test-retest reliabilities provided in the SPSI-R manual for a sample of
young adults and a sample of nursing students (D’Zurilla et al., 2002). In fact the ICS
scale is more reliable with the current population than previously reported samples (sexual
offenders = .91; young adults = .78; nursing students = .79).

Construct validity

To examine construct validity a two-stage procedure was utilised conducting both exploratory
and confirmatory factor analysis. The sample was split in half using SPSS random sample
generation tools (sample 1, N = 253, sample 2, N = 246). Exploratory factor analysis
was performed on the first sample to explore the emerging factor structure of the data, and
confirmatory factor analysis on the second sample to compare the theory-driven a-priori
Springer
Sex Abuse

five-factor model with any models which emerged through EFA. Using separate samples for
this purpose is the preferred method (James, Hunsley, & Hemsworth, 2002).

Exploratory factor analysis

Exploratory factor analysis (EFA) was performed on the data for sample 1 (N = 253)
using Maximum Likelihood Extraction (MLE) techniques. Oblique rotation was used, as the
emerging factors were expected to correlate with each other, as found by Maydeu-Olivares
and D’Zurilla (1995, 1996) in their factor analyses of the SPSI-R. A range of criteria
were used to decide the number of factors to retain in this analysis including: Guttman’s
(1954) Eigenvalue over 1 rule; Cattell’s (1966) scree test; examination of the number of
variables which have significant factor loadings on factors; and consideration of whether the
factors can be interpreted in a meaningful way with respect to the social problem-solving
theory.
On examination of the resulting EFA models, it became apparent that one item in the
scale (item 42) was not able to meaningfully integrate into factors, did not load onto any
extracted factor (factor loading < .3), and had low correlations with the other items. On
examination of the item meaning, there was sufficient indication to suggest that with this
particular population the interpretation of the item may differ. The item refers to going to
someone else for help with solving problems: ‘When I am faced with a difficult problem, I
go to someone else for help in solving it’. According to social problem-solving theory, this
would be an avoidant strategy. However people who are incarcerated in prison lose much
of their freedom to act as they wish, and are forced to ask permission and obtain help with
daily tasks. Furthermore, advice frequently given to offenders is to go to officers/prison staff
when they have any issues. Thus, this problem-solving strategy may be a consequence of
incarceration, which makes this question less useful with an imprisoned population. This
was also the item that negatively affected the internal consistency of the AS scale. Item 42
was therefore taken out of the scale for the purpose of further analyses.
EFA was thus conducted on 51 of the SPSI-R items. According to the
Eigenvalues greater than 1 rule, eight factors were extracted. However on examination
of the rotated loadings, the factors were un-interpretable, and many of the factors did not
have any high loadings. In addition the scree plot suggested that fewer factors would be most
appropriate. A three-factor and a two-factor solution were thus examined, and the two-factor
solution was selected due to greater interpretability and fewer cross loadings.
The two-factor solution explained 44.3% of the variance, and converged in four iterations.
Table 2 presents the rotated factor loadings (structure matrix was utilised as this was more
easily interpreted than the factor matrix), along with Eigenvalues and percentage variance
explained by each factor.
Factor 1 consists of high positive loadings from 26 items, which seem to relate to unpro-
ductive problem-solving techniques and helpless attitudes towards problem-solving. Factor 2
consists of high positive loadings from 25 items, which seem to concern effective and sys-
tematic problem-solving techniques.
One of the items within factor 2 loaded very slightly more onto factor 1 negatively, but
examination of the item content found that it fit better within factor 2.

‘When my first efforts to solve a problem fail, I know that if I persist and do not give up
too easily, I will eventually find a good solution’ (item 7; .380 loading onto factor 2;
− .397 loading onto factor 1).

Springer
Sex Abuse

Table 2 Rotated factor loadings for the two-factor solution

Item Factor 1 Factor 2

Unproductive problem-solving techniques and helpless attitudes towards


problem-solving (26 items)
I become depressed and immobilized when I have an important problem to solve. .780
When my first efforts to solve a problem fail, I get discouraged and depressed. .777 − .311
I go out of my way to avoid having to deal with problems in my life. .756 − .375
When a problem occurs in my life, I put off trying to solve it for as long as .733 − .442
possible.
When I am trying to solve a problem, I get so upset that I cannot think clearly. .726
I put off solving problems until it is too late to do anything about them. .722 − .323
I spend too much time worrying about my problems instead of trying to solve .703 − .395
them.
I feel nervous and unsure of myself when I have an important decision to make. .699
I prefer to avoid thinking about the problem in my life instead of trying to solve .690 − .365
them.
I feel threatened and afraid when I have an important problem to solve. .686
When my first efforts to solve a problem fail, I get very frustrated. .671 − .303
Difficult problems make me very upset. .670
I spend more time avoiding problems than solving them. .659
When making decisions, I go with my “gut feeling” without thinking too much .656 − .472
about the consequences of each option.
When I try to think of different possible solutions to a problem, I cannot come up .651
with many ideas.
When making decisions, I do not evaluate all my options carefully enough. .647 − .428
When I am attempting to solve a problem, I act on the first idea that comes to me. .645 − .449
I hate having to solve the problems that occur in life. .632 − .337
When a solution that I have carried out does not solve my problem satisfactorily, .629
I do not take the time to examine carefully why it did not work.
I am too impulsive when it comes to making decisions. .628 − .441
When I am trying to solve a problem, I go with the first idea that comes to mind. .624 − .458
When I am faced with a difficult problem, I doubt that I will be able to solve it on .621
my own no matter how hard I try.
After carrying out a solution to a problem, I do not take the time to evaluate all of .605 − .415
the results carefully.
When I have a decision to make, I do not take the time to consider the pros and .600 − .334
cons of each option.
When I have a decision to make, I fail to consider the effects that each option is .571 − .426
likely to have on the well-being of other people.
I wait to see if a problem will resolve itself first, before trying to solve it myself. .527
Effective and systematic problem-solving techniques (25 items)
When my first efforts to solve a problem fail, I know that if I persist and do not − .397 .380
give up too easily, I will eventually find a good solution.
When I have decisions to make, I weigh the consequences of each option and − .410 .815
compare them against each other.
When I have a problem to solve, one of the first things I do is get as many facts − .449 .812
about the problem as possible.
When making decision, I consider both the immediate consequences and the − .372 .794
long-term consequences of each option.

Springer
Sex Abuse

Table 2 Continued

Item Factor 1 Factor 2

After carrying out my solutions to a problem, I analyze what went right and − .328 .752
what went wrong.
When I am trying to solve a problem, I think of as many options as possible − .336 .743
until I cannot come up with any more ideas.
When I am having trouble understanding a problem, I try to get more specific − .436 .740
and concrete information about the problem to help clarify it.
When I have a problem to solve, I examined what factors or circumstances in .706
my environment might be contributing to the problem.
After carrying out a solution to a problem, I try to evaluate as carefully as .700
possible how much the situation has changed me for the better.
After carrying out my solution to a problem, I analyze what went right and .694
what went wrong.
When I am trying to solve a problem, I keep in mind what my goal is at all − .318 .693
times.
When I have a decision to make, I try to predict the positive and negative − .326 .682
consequences of each option.
Before I try to solve a problem, I set a specific goal so that I know exactly what .657
I want to accomplish.
Before carrying out my solution to a problem, I practice the solution in order .638
to increase my chances of success.
When I have a decision to make, I consider the effects that each option is likely .633
to have on my personal feelings.
When the outcome of my solution to a problem is not satisfactory, I try to find − .398 .630
out what went wrong and then I try again.
When I am attempting to solve a problem, I try to be creative and think of new .614
or original solutions.
When problems occur in my life, I like to deal with them as soon as possible. − .384 .610
When making decisions, I use a systematic method for judging and comparing .596
alternatives.
When I have a problem, I try to see it as a challenge, or opportunity to benefit − .325 .575
in some positive way from having the problem.
When I have a problem to solve, one of the things I do is analyze the situation − .325 .547
and try to identify what obstacles are keeping me from getting what I want.
When I am trying to solve a problem, I often think of different solutions and .531
then try to combine some of them to make a better solution.
When I am attempting to solve a problem, I approach it from as many different .446
angles as possible.
Whenever I have a problem, I believe that it can be solved. − .378 .398
When I am faced with a difficult problem, I believe that I will be able to solve .344
it on my own if I try hard enough.
Percentage Variance Explained 32.7% 11.6%
Eigenvalue 17.2 6.4

The two factors are negatively correlated with each other ( − .46), indicating that scoring
high on one factor relates to scoring low on the other factor. This makes sense considering the
two factors relate in general terms to positive and negative problem-solving. This two-factor
solution is similar to the two-factor model found by Maydeu-Olivares and D’Zurilla (1996)
in their factor analytic investigation of the SPSI scale. In their exploration of the original
Springer
Sex Abuse

scale (70 items), they found a solution which could be defined similarly in terms of positive
and negative problem-solving.

Confirmatory Factor Analysis

Maximum-Likelihood CFA using LISREL 8.52 (Jöreskog & Sörbom, 2001) was conducted
using the second half of the sample (N = 246). The theory driven five-factor model was
compared with the two-factor model derived through EFA. The two models were as follows:
Model 1: Theory driven five-factor model.

Factor 1: Positive Problem Orientation, 5 items


Factor 2: Negative Problem Orientation, 10 items
Factor 3: Rational Problem-solving, 20 items
Factor 4: Impulsive/carelessness Style, 10 items
Factor 5: Avoidant Style, 6 items

Model 2: EFA two-factor model.

Factor 1: Unproductive problem-solving techniques and helpless attitudes towards


problem-solving, 26 items
Factor 2: Effective and systematic problem-solving techniques, 25 items

For both of these models, the factors were allowed to correlate. The models were obtained
by estimating all factor loadings as specified above in the model descriptions, but fixing all
other loadings at zero. Contrary to recommendations for using Likert-scaled items with EFA,
it is agreed that data obtained from Likert rating scales should not be treated as continuous in
CFA (Floyd & Widaman, 1995). Polychoric correlations were therefore examined for these
analyses (Jöreskog & Sörbom, 1989).
The fit of the models was evaluated according to several criteria. Firstly the chi-square
statistic and the ratio of chi-square to degrees of freedom were examined. In addition
to this, two comparative fit indices were examined: the Expected Cross-Validation Index
(ECVI), and the consistent Akaike’s information criterion (CAIC). These comparative fit
indices allow comparisons between different models, with smaller values indicative of better
fitting models. The following goodness-of-fit indices were also used: the goodness-of-fit
(GFI) and the adjusted goodness-of-fit indices (AGFI); the normed fit index (NFI); and the
normed comparative fit index (CFI) (Bollen, 1989). These fit indices all range from 0 to 1,
with higher values indicating better fit. Bentler and Bonett (1980) suggest that a minimum
acceptable level of .90 is required to deem a model adequate. The root mean square error of
approximation (RMSEA) was also examined, for which values of less than 0.05 indicates a
good fit, between 0.05 and under 0.08 a reasonable fit, between 0.08 and 0.1 a mediocre fit,
and greater than 0.1 a poor fit. The root mean square residual (RMSR) can also be used to
compare the fit of competing models; the closer RMSR is to zero, the better the fit of the
model.
The criteria for the fit of the two models are presented in Table 3. It can be seen that the
theory-driven five-factor model appears to be the better fitting model of the two. The chi-
square statistics for both models were significant indicative of poor fitting models. However
this is often the case in CFA when using models with many observed variables (Thompson,
2004). The ratio of chi-square to degrees of freedom was lower for the five-factor model.
In addition, the ECVI and the CAIC were smaller for the five-factor model. Referring to
the other goodness-of-fit indices, the five-factor model appears to be the better fitting. The
Springer
Sex Abuse

Table 3 Goodness of fit indices of SPSI-R models using CFA

Goodness of fit index Theory driven five-factor model EFA Derived two-factor model

χ2 3680.12 4995.49
df 1214 1223
p-value .000 .000
χ /df 3.03 4.08
ECVIa 15.94 21.23
CAICb 4408.72 5665.54
GFIc .63 .56
AGFId .60 .52
NFIe .92 .91
CFIf .94 .93
RMSEAg .091 .11
RMSRh .068 .079
a Expected cross-validation index.
b Consistent Akaike’s information criterion.
c Goodness-of-fit.

d Adjusted goodness-of-fit.
e Normed fit index.
f Normed Comparative Fit.
g Root mean square error of approximation.
h Root mean square residual.

GFI, AGFI, NFI and CFI values were all higher for the five-factor model in comparison to
the other model. Although the GFI (.63) and AGFI (.60) did not meet the criteria for good
fit, the remaining two goodness-of-fit indices values suggest that the five-factor model is
adequate. The RMSEA was .09 suggesting a mediocre fit, and the RMSR was .07 suggesting
an adequate fit.
Within the five-factor model, standard errors for each of the factor loadings were also
low, ranging from .05 to .06, indicating good estimations. The reliability of each item with
respect to its underlying latent construct were moderate to high (R2 ranging from 0.26 to
0.80), which suggests that the manifest variables are reasonably successful as measures of
the latent variables. The correlations between factors were all significant (p < .005), and in
the expected direction.
The five-factor model was thus selected as the better model. To examine whether the
EFA results were affected by sample size, a further EFA was conducted on the full sample
(N = 499). Principal Components Analysis using oblique rotation (promax) was conducted,
which extracted six factors with eigenvalues over 1.0 explaining 59.1% of the variance. The
first five factors were almost an exact replicate of the theory-driven five-factor model. Only
two items loaded onto different factors. The sixth factor only had three loadings > .3, and
each of these items loaded to a much greater extent on to another factor. Thus, the five-factor
solution appeared to be the most parsimonious solution using EFA on the whole sample. The
scree plot supported extracting five factors.

Convergent validity

To examine convergent validity, series of correlations were calculated between each of the
five SPSI-R pre-treatment factor scores and scores on a measure of self-esteem, a measure
Springer
Sex Abuse

Table 4 Convergent Validity and Susceptibility to Social Desirable Responding (∗∗ p < .001)

SPSI Total
PPO NPO RPS ICS AS score

Self esteem .342∗∗ − .500∗∗ .301∗∗ − .426∗∗ − .380∗∗ .488∗∗


Impulsivity − .369∗∗ .525∗∗ − .462∗∗ .645∗∗ .461∗∗ − .614∗∗
Locus of control .461∗∗ − .623∗∗ .438∗∗ − .604∗∗ − .554∗∗ .671∗∗
Self deception .430∗∗ − .476∗∗ .373∗∗ − .432∗∗ − .421∗∗ .533∗∗
Impression .259∗∗ − .258∗∗ .309∗∗ − .344∗∗ − .271∗∗ .359∗∗
management
Total BIDR score .392∗∗ − .415∗∗ .395∗∗ − .448∗∗ − .395∗∗ .511∗∗

of impulsivity, and a measure of locus of control, all of which are constructs similar to
problem-solving. Table 4 presents these correlations.
Self-esteem is significantly positively correlated with the two positive problem-solving
styles, PPO and RPS, as well as the SPSI total score, and significantly negatively correlated
with the dysfunctional problem-solving dimensions, NPO, ICS and AS. The same pattern of
significant correlations is seen for the locus of control and impulsivity measures.

Susceptibility to socially desirable responding

The susceptibility of the SPSI-R to socially desirable responding was examined by correlating
SPSI-R subscale scores with the BIDR subscales. These correlations are also shown in Table 4
above. The correlations are all significant, indicating that that as socially desirable responding
increases, so do scores on the SPSI-R subscales.

Sexual offender treatment impact

To examine whether the SPSI-R is able to detect treatment change, a repeated measures
Analysis of Variance (ANOVA) was conducted. This examined impact of treatment by
offender type (rapist or child molester). Table 5 shows the pre and post-treatment scores on
the five subscales and SPSI-R total score by offence type. The scores change in the desired
direction pre to post-treatment.
Significant differences were found on the PPO (F(1, 469) = 58.47, p < .001), and RPS
(F(1, 469) = 125.52, p < .001) subscales, with scores increasing pre to post-treatment.
Significant differences were also found on the NPO (F(1, 469) = 122.17, p < .001), ICS
(F(1, 469) = 116.13, p < .001), and AS (F(1, 469) = 93.90, p < .001) subscales, with
scores decreasing pre to post-treatment. Finally, there was a significant increase in total SPSI-
R scores pre to post-treatment, F(1, 469) = 161.18, p < .001. No interactions between
treatment effect and offence type were found for any of these analyses.

Discussion

The results provide some support for the use of the SPSI-R with sexual offender popula-
tions. First, the SPSI-R had very good internal consistency, and test-retest reliability. These
reliabilities are comparable to those generated with samples of adolescents, young adults,
middle-aged adults and elderly adults reported by D’Zurilla and colleagues (2002). Second,
Springer
Sex Abuse

Table 5 Pre and post-treatment SPSI-R scores by offence type

Pre-treatment score Post-treatment score


Mean SD Mean SD F Value P Value

PPO ↑
Whole sample 12.05 3.80 13.70 3.77 58.47 .001
Child molesters 12.04 3.83 13.57 3.72
Rapists 12.10 3.70 14.14 3.92
NPO ↓
Whole sample 12.43 8.72 7.82 7.53 122.17 .001
Child molesters 12.77 8.85 8.03 7.64
Rapists 11.22 8.20 7.07 7.12
RPS ↑
Whole sample 44.12 15.84 52.93 16.31 125.52 .001
Child molesters 44.28 15.94 52.49 15.76
Rapists 43.53 15.84 54.49 18.14
ICS ↓
Whole sample 12.65 8.16 7.96 7.87 116.13 .001
Child molesters 13.00 8.06 8.18 7.86
Rapists 11.37 8.42 7.16 7.89
AS ↓
Whole sample 7.51 5.55 4.52 4.79 161.18 .001
Child molesters 7.85 5.57 4.65 4.86
Rapists 6.29 5.34 4.05 4.41
SPSI Total Score ↑
Whole sample 12.86 3.33 15.06 3.23 161.18 .001
Child molesters 12.74 3.39 14.94 3.21
Rapists 13.29 3.10 15.46 3.29

concerning internal validity, the SPSI-R was found to be a reasonably valid tool with this
population.
Construct validity was confirmed by factor analysis. EFA found a two-factor model as the
best fitting solution with the first half of the sample. This model was similar in structure to
a two-factor model found by Maydeu-Olivares and D’Zurilla (1996) in their factor-analytic
investigation of the SPSI. Comparing this two-factor model with the original theory-driven
five-factor model proposed by D’Zurilla et al. (2002) using CFA found that on most of the fit
indices the five-factor model was a more reasonable, though not perfect, fit of the data. All
the examined criteria for goodness of fit suggested that this model was a better fit of the data
than the two-factor model. Thus the five-factor model was selected as the most appropriate
solution of the two. This decision was confirmed when a further EFA was conducted to
determine whether sample size had affected the interpretability of the factors. EFA using the
whole sample of 499 sexual offenders, found a five-factor model almost identical in structure
to the theory-driven five-factor model. These results therefore support the previous validation
studies of the SPSI-R (Maydeu-Olivares & D’Zurilla, 1996) with samples of undergraduate
students. The SPS model outlined in the introduction is supported, indicating that with sexual
offenders problem-solving can be reduced to five dimensions.
The data also supported convergent validity of the SPSI-R. Significant correlations were
found between all five SPSI-R subscales and measures of self-esteem, locus of control,
and impulsivity. These are constructs that are similar to social problem-solving processes
(D’Zurilla & Maydeu-Olivares, 1995). Previous research has found SPSI-R scores to corre-
Springer
Sex Abuse

late with internal locus of control (D’Zurilla & Nezu, 1990), as well as self-esteem (D’Zurilla
& Maydeu-Olivares 1995). The present study supports these findings with a sexual offender
population. In addition, this research found a relationship between impulsivity and problem-
solving, which to the author’s knowledge has not been examined previously using the SPSI-R.
This suggests that impulsivity may be a further area to focus on when examining convergent
validity of the SPSI-R.
The SPSI-R subscales were found to be significantly correlated with the BIDR social
desirability scale. This response bias is not unusual with offenders (e.g. Schretlen & Arkowitz,
1990), and could be taken to support the notion that self-report measures have a low validity. It
is often assumed that sexual offenders who are being assessed are likely to attempt to respond
in a way that will make them appear in the best possible light perhaps in an attempt to affect
progression through the prison system, thus dissimulating within all self-report measures.
Despite this assumption, however, recent research has found that self-report measures can be
used to predict violent recidivism (Mills, Kroner, & Hemmati, 2005) and general recidivism
(Motuik, Motiuk, & Bonta, 1992). This suggests that socially desirable responding may not in
fact undermine the relationship between self-report and predictive criteria. Furthermore, it has
been shown that offenders high on impression management report fewer antisocial attitudes
and are at less risk to commit a criminal offense (Mills & Kroner, 2005). Thus, instead of
confounding the validity of the SPSI-R measure, the significant correlations found between
problem-solving and socially desirable responding may occur not because the validity of
the measure has been compromised by social desirability but because both are related to
risk (Mills & Kroner, 2005, 2006). A further explanation could be that socially desirable
responding should be viewed as a characteristic or trait, rather than being situation-dependent
(Mills, Loza, & Kroner, 2003). Thus the present findings may reflect the fact that problem-
solving and BIDR scores reflect similar constructs or similar character traits. Further research
is required to clarify the relationship between socially desirable responding and problem-
solving as measured by the SPSI-R before any firm conclusions can be made regarding the
impact of social desirability on the SPSI-R, and indeed other self-report measures.
Treatment impact was also examined by examining pre and post intervention SPSI-R
scores. It was revealed that participation in a cognitive-behavioural treatment programme
impacts positively on social problem-solving, for all sexual offenders, of different offence
types. This research has expanded on previous research by examining both child molesters
and rapists. The present study has confirmed that there are few differences between these two
types of sexual offenders in terms of problem-solving deficits and the impact of treatment
in this area. In comparison to the mean scores for normative samples provided in the SPSI-
R manual (D’Zurilla et al., 2002), the pre-treatment social problem-solving scores of the
sexual offenders in this sample were similar to young adults, and the sexual offenders’ post-
treatment scores were similar to those of middle-age adults. The present sample however,
had better problem-solving skills than the psychiatric or distressed populations reported by
D’Zurilla and colleagues (2002), and slightly better problem-solving skills than the sample of
child molesters examined by Nezu et al. (2005), although the PPO and RPS subscale scores
were similar. It seems thus that the sexual offenders in this study do not differ markedly
from young adult non-offenders in problem-solving. Referring back to Hanson and Morton-
Bourgon’s (2004) recent meta analysis, these results may indicate that problem-solving is
indeed not a particularly problematic area for sexual offenders. Alternatively it could be
that young adults (age 17–39) are a generally impulsive and poor problem solving subgroup
of adults, similar to sexual offenders. Future research would need to clarify this, and also
confirm the association between problem solving and sexual offending by examining the
relationship between SPSI-R scores and reconviction.
Springer
Sex Abuse

The study provides some support for the use of the SPSI-R as a valid and reliable tool for
examining social problem-solving in sexual offenders, and for detecting treatment impact,
and thus further verifies the value and integrity of the SPSI-R as a useful measure of SPS.
Sexual offenders were not found to have dissimilar problem-solving skills to non-offending
young adults. However, sexual offenders’ skills were able to be enhanced through cognitive-
behavioural treatment. The study supports the use of the SPSI-R as a clinical assessment for
identifying problem-solving deficits and facilitating the development of effective treatment.
The study also indicates the usefulness of social problem-solving therapy as a clinical
intervention method with sexual offenders.
This research goes some way to support the importance of scale revalidation when a
measure is being applied to a new population (Kroner & Weekes, 1996). Although the
research found very similar results to studies with non offenders (in terms of reliability,
factor structure and convergent validity), it is still important to investigate this rather than
assume it is the case. This is made clear by the fact that one of the SPSI-R items was less
useful with this population, possibly due to the context of incarceration. Incarceration may
lead to loss of freedom to solve problems in ways which might be more available for those
not in prison.
It is important to note the limitations of this study. Firstly, the research could have been
strengthened by using an untreated control group. Without a non-treatment comparison
group we cannot say with certainty that the change observed pre to post treatment in the
SPSI-R scores is due to participation in treatment. It could be for example that change occurs
naturally over time, via practice effects or it could be something within the prison setting that
is causing the changes. Future research should attempt to overcome this problem by utilising
a comparison group.
Secondly, the limitation of using self-report data to measure problem-solving needs to be
mentioned. There is no evidence that SPSI-R scores reflect real-life problem-solving. Thus
it would be useful to cross-validate the SPSI-R with a behavioural or external measure not
subject to faking or lack of insight as with self-report measures. This would enhance the
current findings by providing evidence for the external validity of the SPSI-R. Similarly, this
study only examined problem-solving, which refers to the process of finding solutions to
specific problems, and not solution implementation, the process of carrying out the solutions
and applying skills to specific problems. Research has shown that the two are not always
correlated (D’Zurilla, Nezu, & Maydeu-Olivares, 2004). Thus, further research should ex-
amine the ability of sexual offenders in carrying out solutions they produce. This could be
directly targeted by examination of the relationship between problem-solving ability and
recidivism (used as a measure of ability to carry out solutions as laid out by offenders in
their relapse prevention plans). Validity should be further examined with sexual offenders by
correlating SPSI-R scores with other measures of problem-solving, preferably both process
and outcome measures. Additionally, the disriminant validity of the SPSI-R with sexual of-
fenders should be investigated in future research by correlating the measure with an unrelated
construct.
The study would also have benefited from a larger sample, particularly for the factor
analyses. Using a larger sample size may have resulted in emergence of clearer factors, and
less cross-loadings, providing greater confidence in the results of the EFA. Furthermore a
greater sample size in the CFA may have produced better fitting models according to the
fit indices. The five-factor model produced from CFA was adequate but not perfect. Thus,
ideally CFA should be repeated on a sexual offender population using a larger sample.
This study has for the most part confirmed the hypotheses set out in the introduction. The
SPSI-R has been shown to be a reliable and reasonably internally valid tool to use with sexual
Springer
Sex Abuse

offenders. The results provide support for the use of the SPSI-R with sexual offenders, and
for the social problem-solving theory as proposed by D’Zurilla et al. (2002).

References

Antonowicz, D. H., & Ross, R. R. (2005). Social problem-solving deficits in offenders. In M. McMurran &
J. McGuire (Eds.), Social problem-solving and offending (pp. 91–102). Chichester: Wiley.
Andrews, D. A. (1995). The psychology of criminal conduct and effective treatment. In J. McGuire (Ed.),
What works: Reducing re-offending: Guidelines from research and practice (pp. 35–62). Chichester:
Wiley.
Andrews, D. A. (2001). Principles of effective correctional programs. In L. L. Motiuk & R. C. Serin
(Eds.), Compendium 2000 on effective correctional programming(pp. 9–17). Ottawa: Correctional Service
Canada.
Barbaree, H. E., Marshall, W. L., & Conner, J. (1988). The social problem-solving of child molesters,
unpublished manuscript. As cited in Fisher, D., & Howells, K. (1993). Social Relationships in Sexual
Offenders. Sexual and Marital Therapy, 8(2), 123–136.
Bentler, P. M., & Bonett, D. G. (1980). Significance tests and goodness-of-fit in the analysis of covariance
structures. Psychological Bulletin, 88, 588–606.
Bollen, K. A. (1989). Structural equations with latent variables. New York: Wiley.
Butler, L., & Meichenbaum, D. (1981). The assessment of interpersonal problem-solving skills. In P. C. Kendall
& S. D. Hollen (Eds.), Assessment strategies for cognitive-behavioral interventions (pp. 197–225). New
York: Academic Press.
Cattell, R. B. (1966). The scree test for the number of factors. Multivariate Behavioral Research, 1, 245–276.
D’Zurilla, T. J. (1986). Problem-solving therapy: A social competence approach to clinical intervention. New
York: Springer.
D’Zurilla, T. J., & Goldfried, M. R. (1971). Problem-solving and behaviour modification. Journal of Abnormal
Psychology, 78, 104–126.
D’Zurilla, T. J., & Maydeu-Olivares, A. (1995). Conceptual and methodological issues in social problem-
solving assessment. Behavior Therapy, 26, 409–432.
D’Zurilla, T. J., & Nezu, A. M. (1982). Social problem-solving in adults. In P. C. Kendall (Ed.), Advances in
cognitive-behavioural research and therapy (pp. 201–274). New York: Academic Press.
D’Zurilla, T. J., & Nezu, A.M. (1990). Development and preliminary evaluation of the Social Problem-
Solving Inventory. Psychological Assessment: A Journal of Consulting and Clinical Psychology, 2, 156–
163.
D’Zurilla, T. J., Chang, E. C., Nottingham IV, E. J., & Faccini, L. (1998). Social problem-solving deficits
and hopelessness, depression, and suicidal risk in college students and psychiatric patients. Journal of
Clinical Psychology, 54, 1–17.
D’Zurilla, T. J., & Nezu, A. M. (1999). Problem-solving therapy: A social competence approach to clinical
intervention. New York: Springer.
D’Zurilla, T. J., & Nezu, A. M. (2001). Problem-solving therapies. In K. S. Dobson (Ed.), Handbook of
Cognitive Behavioural Therapies (pp. 211–245). New York, London: The Guildford Press.
D’Zurilla, T. J., Nezu, A. M., & Maydeu-Olivares, A. (2002). Social Problem-Solving Inventory—Revised.
Technical Manual. New York: Multi-Health Systems Inc.
D’Zurilla, T. J., Nezu, A. M., & Maydeu-Olivares, A. (2004). What is social problem-solving?: Meaning,
models, and measures. In E. C. Chang, T. J. D’Zurilla, & L. Sanna (Eds.), Social problem-solving: Theory,
research, and training (pp. 11–27). Washington, DC: American Psychological Association.
Epstein, S. (1994). Integration of the cognitive and the psychodynamic unconscious. American Psychologist,
49, 709–724.
Epstein, S., & Meier, P. (1989). Constructive thinking: A broad coping variable with specific components.
Journal of Personality and Social Psychology, 57, 332–350.
Eysenck, S. B. G., & Eysenck, H. J. (1978). Impulsivity and venturesomeness: Their place in a dimensional
system of personality description. Psychological Reports, 43, 1247–1255.
Fisher, D., & Howells, K. (1993). Social relationships in sexual offenders. Sexual and Marital Therapy, 8(2),
123–136.
Floyd, F. J., & Widaman, K. F. (1995). Factor analysis in the development and refinement of clinical assessment
instructions. Psychological Assessment, 7(3), 286–299.
Freedman, B. J., Rosenthal, L., Donahoe, C. P., Schlundt, D. G., & McFall, R. M. (1978). A social-behavioral
analysis of skill deficits in delinquent and non-delinquent adolescent boys. Journal of Consulting and
Clinical Psychology, 46, 1448–1462.
Springer
Sex Abuse

Grier, P. E. (1988). Cognitive problem-solving skills in antisocial rapists. Criminal Justice and Behavior,
15(4), 501–514.
Guttman, L. (1954). A new approach to factor analysis: The circle. In P. E. Lazarfeld (Ed.), Mathematical
thinking in the social sciences (pp. 258–348). Glencoe, IL: Free Press.
Hanson, R. K., & Morton-Bourgon, K. (2004). Predictors of sexual recidivism: An updated meta-analysis.
Retrieved March 20, 2006, from Public Safety and Emergency Preparedness Canada website.
Heppner, P. P., & Peterson, C. H. (1982). The development and implications of a personal problem-solving
inventory. Journal of Counselling Psychology, 29, 66–75.
Hudson, S. M., Ward, T., & McCormack, J. C. (1999). Offence pathways in sexual offenders. Journal of
Interpersonal Violence, 14(8), 779–798.
James, S., Hunsley, J., & Hemsworth, D. (2002). Factor structure of the relationship belief inventory. Cognitive
Therapy and Research, 26(6), 729–744.
Jöreskog, K. G., & Sörbom, D. (1989). LISREL 7: User’s reference guide. Mooresville, IN: Scientific Software.
Jöreskog, K. G., & Sörbom, D. (2001). LISREL 8.52. Mooresville, IN: Scientific Software.
Kroner, D. G., & Weekes, J. R. (1996). Balanced inventory of desirable responding: Factor structure, reliability,
and validity with an offender sample. Personality and Individual Differences, 21(3), 323–333.
Levenson, H. (1972). Distinctions within the concept of internal-external control: Development of a new scale.
Proceedings of the 80th Annual Convention of American Psychological Association, 7, 259–260.
Levesque, D. A., Gelles, R. J., & Velicer, W. F. (2000). Development and Validation of a stages of change
measure for men in batterer treatment. Cognitive Therapy and Research, 24(2), 175–179.
Mann, R. E., & Thornton, D. (1998). The evolution of a multi-site sex offender treatment programme. In
W. L. Marshall, Y. M. Fernandex, S. H. Hudson, & T. Ward (Eds.), Sourcebook of treatment programs
for sexual offenders (pp. 47–58). New York: Plenum Press.
Mann, R. E., Webster, S. D., Wakeling, H. C., & Marshall, W. L. (in press). The Measurement and Influence
of child abuse supportive beliefs. Psychology, Crime and Law.
Maydeu-Olivares, A., & D’Zurilla, T. J. (1995). A factor analysis of the Social Problem-Solving Inventory
using polychoric correlations. European Journal of Psychological Assessment, 11, 98–107.
Maydeu-Olivares, A., & D’Zurilla, T. J. (1996). A factor-analytic study of the Social Problem-Solving
Inventory: An integration of theory and data. Cognitive Therapy and Research, 20(2), 115–133.
McGuire, J. (2001). What is problem-solving? A review of theory, research and applications. Criminal
Behaviour and Mental Health, 11, 210–235.
McGuire, J. (2005). Social Problem-solving: Basic Concepts, Research and Applications. In M. McMurran &
J. McGuire (Eds.), Social problem-solving and offending: Evidence, evaluation and evolution (pp. 3–30).
Chichester: Wiley.
McMurran, M., Egan, V., Richardson, C., & Ahmadi, S. (1999). Social problem-solving in mentally disordered
offenders: a brief report. Criminal Behaviour and Mental Health, 9, 315–322.
Mills, J. F., & Kroner, D. G. (2005). An investigation into the relationship between socially desirable responding
and offender self-report. Psychological Services, 2(1), 70–80.
Mills, J. F., & Kroner, D. G. (2006). Impression management and self-report among violent offenders. Journal
of Interpersonal Violence, 21(2), 178–192.
Mills, J. F., Kroner, D. G., & Hemmati, T. (2005). The measures of criminal attitudes and associates (MCAA):
The prediction of general and violent recidivism. Criminal Justice and Behavior, 32, 565–585.
Mills, J. F., Loza, W., & Kroner, D. G. (2003). Predictive validity despite social desirability: evidence for the
robustness of self-report among offenders. Criminal Behaviour and Mental Health, 13, 140–150.
Motuik, M., Motiuk, L., & Bonta, J. (1992). A comparison between self-report and interview-based inventories
in offender classification. Criminal Justice and Behavior, 19, 143–159.
Nezu, C. M., Nezu, A. M., & Dudek, J. A. (1998). A cognitive-behavioural model of assessment and treatment
for intellectually disabled sexual offenders. Cognitive and Behavioral Practice, 5, 25–64.
Nezu, C. M., Nezu, A. M., Dudek, J. A., Peacock, M. A., & Stoll, J. G. (2005). Social problem-solving
correlates of sexual deviancy and aggression among adult child molesters. Journal of Sexual Aggression,
11(1), 27–36.
Nezu, A. M. (2004). Problem-solving and Behavior Therapy Revisited. Behavior Therapy, 35, 1–33.
Nichols, H. R., & Molinder, I. (1984). Multiphasic sex inventory. A test to assess the characteristics of the
sexual offender. Tacoma, WA: Nichols and Molinder Assessments.
Palmer, E. J., & Hollin, C. R. (1996). Assessing adolescent problems: An overview of the Adolescent Problems
Inventory. Journal of Adolescence, 19, 347–354.
Paulhus, D. L. (1984). Two component models of socially desirable responding. Journal of Personality and
Social Psychology, 46(3), 598–609.
Platt, J. J., Spivack, G., Altman, N., Altman, D., & Peizer, S. B. (1974). Adolescent problem-solving thinking.
Journal of Consulting and Clinical Psychology, 42, 787–793.

Springer
Sex Abuse

Polaschek, D. L. L., Hudson, S. M., Ward, T., & Siegert, R. J. (2001). Rapists’ offense processes: A preliminary
descriptive model. Journal of Interpersonal Violence, 16, 523–544.
Rallings, M., & Webster, S. D. (2001). The psychometric properties of HM Prison Service SOTP assessment
battery. Offending Behaviour Programmes Unit, HM Prison Service, unpublished document.
Schretlen, D., & Arkowitz, H. (1990). A psychological test battery to detect prison inmates who fake insanity
or mental retardation. Behavioural Sciences and the Law, 8, 75–84.
Spivack, P., & Platt, J. J. (1980). Measures of social problem-solving for adolescents and adults. Philadelphia,
PA: Preventive Intervention Research Center, Hahnemann University.
Tabachnick, B. G., & Fidell, L. S. (1989). Using multiple statistics (2nd ed.). California, USA: Harper-Collins.
Thompson, B. (2004). Exploratory and confirmatory factor analysis. Understanding concepts and applica-
tions. Washington, DC: American Psychological Association.
Thornton, D. (2002). Constructing and testing a framework for dynamic risk assessment. Sexual Abuse: A
Journal of Research and Treatment, 14, 139–153.
Thornton, D., Beech, A., & Marshall, W. L. (2004). Pretreatment self-esteem and posttreatment sexual
recidivism. International Journal of Offender Therapy and Comparative Criminology, 48, 567–599.
Webster, S. D., Mann, R. E., Thornton, D., & Wakeling, H. C. (in press). Further validation of the Short
Self-esteem Scale with sexual offenders. Legal and Criminological Psychology.
Wechsler, D. (1981). Wechsler adult intelligence scale-revised. New York: Psychological Corporation.
Zamble, E., & Porporino, F. J. (1988). Coping, behavior, and adaptation in prison inmates. New York:
Springer-Verlag.
Zamble, E., & Quinsey, V. L. (1997). The criminal recidivism process. Cambridge: Cambridge University
Press.

Springer

View publication stats

You might also like