Psychometric Evaluation of the Structured Assessment of Violence Risk in Youth (SAVRY) in Forensic Psychiatric Youth
Master Thesis Forensische Orthopedagogiek
Pedagogische en Onderwijskundige Wetenschappen Universiteit van Amsterdam
Juli 2014
Student: Nienke Smulders
Studentnummer: 10578706
Begeleiders: drs. E. Kornelis1, dr. I.L. Bongers2, drs. E.A.W. Janssen-De Ruijter2 1
Universiteit van Amsterdam 2
Table of contents Abstract p.3 Introduction p.4 Method p.6 Participants p.6 Procedure p.7 Measures p.7 Statistics p.8 Results p.10 Discussion p.14 References p.17 Appendix p.21
Abstract
Background: Given the minimal research that has been conducted on the domains of the Structured Assessment of Violence Risk in Youth (SAVRY), the aim of this study was to investigate whether the original SAVRY domains were reliable and valid domains for scoring the SAVRY by using file data. Method: Participants were 187 male forensic psychiatric patients with a mean age of 16.8 years (SD = 1.8, range = 14-23). After admission to the hospital, (electronic) patient files were used to score the SAVRY and Juvenile Forensic Profile (FPJ). Results: Factor analyses were performed using Mplus. Examination of the confirmative and exploratory factor analyses resulted in a two-factor structure of the SAVRY, consisting of a Social/Contextual domain and an Individual domain. Good construct validity of the modified SAVRY domains with the FPJ domains Family and Environment and Behavior during Stay in the Institution was demonstrated. Conclusions: A two-factor structure was found for the SAVRY scored by using file data. Future research should investigate this two-factor structure in larger samples and should examine the predictive validity for violent recidivism.
Introduction
The possible serious consequences of adolescent’s violent delinquent behavior (Cohen & Piquero, 2009; Moffit, 1993) make it essential to reliably assess the risk of violent recidivism (Childs et al., 2013; Vincent, Guy, Fusco, & Gershenson, 2012). Historically, risk assessment was based on unstructured clinical judgment where clinicians’ intuition was central to risk assessment (Childs et al., 2013). Since unstructured assessment is described as inadequate and unreliable (Douglas & Kropp, 2002), more structured methods of risk assessment were developed (Childs et al., 2013). Nowadays, a large number of structured instruments are available (Childs et al., 2013). For adolescents, the Structured Assessment of Violence Risk in Youth (SAVRY; Borum et al., 2006), the Psychopathy Checklist Youth Version (PCL-YV; Forth, Kosson, & Hare, 2003) and the Youth Level of Service/Case Management Inventory (YLS/CMI; Hoge & Andrews, 2002) are often-used risk assessment instruments. In earlier studies, the predictive validity of all three instruments is established (Fazel, Singh, Doll, & Grann, 2012; Olver, Stockdale, & Wormith, 2009) with the strongest predictive validity for the SAVRY (Singh, Grann, & Fazel, 2011; Vincent, Chapman, & Cook, 2011; Welsh, Schmidt, McKinnon, Chattha, & Meyers, 2008). Therefore, it is understandable that the SAVRY is a commonly used risk assessment instrument in clinical practice as well as for research purposes.
The SAVRY is a risk assessment instrument which is used to estimate the risk of violent behavior in adolescents (Borum et al., 2006). According to the developers of the SAVRY, violent behavior can be defined as ‘each form of abuse or physical violence that is severely enough to cause injuries regardless of the fact whether injuries are truly present, each form of sexual assault, weapon threat and/or the possibility to prosecution given the seriousness of the offense’ (Borum et al., 2006, p.21). The SAVRY consists of four domains containing 24 risk and six protective factors. This structure of domains was based on empirical studies in which the risk and protective factors were found to be connected with violent delinquency (Borum et al., 2006). There are ten items in the Historical domain, six in the Social/Contextual domain and eight in the Individual domain. The six protective factors form the Protective domain. After scoring all risk and protective factors, the final level of risk, ‘the Summary Risk Rating’ (Vincent, 2012), is determined by the overall interpretation of risk and protective factors in relation with the adolescent’s unique circumstances (Gammelgard, Koivisto, Eronen, & Kaltiala-Heino, 2008). The SAVRY was originally designed for individual risk classification and case-management
planning (Borum et al., 2006). However, in research studies SAVRY scores of individual adolescents are frequently grouped together into a research sample in order to investigate the predictive validity (Childs et al., 2013; Lodewijks, Doreleijers, De Ruiter, & Borum, 2008).
Given the aim of the SAVRY, it is reasonable that in most studies the focus was on predictive validity. In earlier research, the predictive validity of the SAVRY Summary Risk Rating varied from fair to good (e.g., Catchpole & Gretton, 2003; Lodewijks, Doreleijers, & De Ruiter, 2008; Schmidt, Campbell, & Houlding, 2011). However, as the developers of the SAVRY (Borum et al., 2006) already mentioned themselves, the final score alone does not determine the risk of violent behavior. For example, an adolescent whose school career and family are relatively good, but who has a history of serious violent behavior can still be rated as a high risk adolescent. By the sole use of a total score, the impact of a specific domain on violent recidivism may disappear. Therefore, the predictive validity of the SAVRY domains is equally important as the predictive validity of the total score. Results of the predictive validity of the domains were not always satisfactory with poor to fair predictive validity for the Historical and the Social/Contextual domain, poor to good predictive validity for the Individual domain and insufficient to good predictive validity for the Protective domain (Dolan & Rennie, 2008; Gammelgard et al., 2008; Lodewijks, Doreleijers, & De Ruiter, 2008; Lodewijks, Doreleijers, De Ruiter, et al., 2008; McGowan, Horn, & Mellott, 2011; Viljoen et al., 2008). Given the inconsistent results about the predictive validity of the domains, more insight into the psychometric properties of the SAVRY domains is needed.
To our knowledge, there is only one study in which the internal consistency of the SAVRY domains was evaluated (Hilterman & Van Nieuwenhuizen, 2013). For this study, scoring of the SAVRY was done by trained clinicians who interviewed the involved adolescents. The internal consistency was found to be good for the Historical, Individual and Protective domain but minimally acceptable for the Social/Contextual domain. Since file data are more frequently used in research on the SAVRY (Childs et al., 2013; Meyers & Schmidt, 2008; Welsh et al., 2008), research into the internal consistency of the SAVRY domains scored by using file data is essential. A further investigation of the SAVRY domains could lead to a more reliable and valid prediction of violent behavior. Therefore, the aim of this study was to examine the psychometric properties of the SAVRY. More specifically, this study aimed to investigate:
1. Whether the current factor structure of the SAVRY is satisfactory in terms of the model fit and internal consistency of the domains for data collected by file research. 2. Whether another factor structure of the SAVRY can be identified in a sample of
forensic psychiatric patients that fits the data better than the original factor structure of the SAVRY in terms of model fit and internal consistency of the domains, when the original factor structure turns out to be insufficient.
3. Whether this new factor structure is valid in terms of the construct validity with the Juvenile Forensic Profile (FPJ; Brand & Van Heerde, 2010).
Method Participants
Participants included in this study were male forensic psychiatric patients (n = 187) with a mean age of 16.8 years (SD = 1.8; range = 14-23). These patients were staying at youth forensic psychiatric hospital ‘De Catamaran’ or, after discharge from De Catamaran, at clinical training house ‘De Schakel’ in the Netherlands. Both De Catamaran and De Schakel are part of GGzE Center for Child and Adolescent Psychiatry. De Catamaran offers adolescents between 14 and 23 years, who have a psychiatric or a disruptive behavior disorder and/or who are a risk for themselves or others, psychological and psychiatric treatment. All male patients who were admitted to De Catamaran between April 2009 and October 2013 with a minimal stay of three months were included in this study. The majority of patients were detained under juvenile criminal law (49.2%) or juvenile civil law (40.6%). The other 10.2% of participants stayed on a voluntary basis. Most patients committed three or more violent acts before admission (44.1%). The most common psychopathology based on the Axis-I classification of the DSM-IV (APA, 2004) were pervasive developmental disorders (47.6%) and disruptive behavior disorders (44.9%). Participants characteristics are displayed more detailed in Table 1.
Table 1
Demographic characteristics
Characteristic M (SD) range n %
Age at admission in years (n = 187) 16.8 (1.8) 14-23
Judicial measure (n = 187)
criminal law 92 49.2
civil law 76 40.6
voluntary 19 10.2
Previous violent behavior (n = 177)1,2
none 29 16.4
one or two violent acts 70 39.5
three or more violent acts 78 44.1
DSM-IV classifications(n = 187)3,4
Pervasive Developmental Disorder 89 47.6
Disruptive Behavior Disorder 84 44.9
Attention Deficit/Hyperactivity Disorder 50 26.7
Reactive Attachment Disorder 26 13.9
Substance Disorder 43 23.0
Schizophrenia or other psychotic disorder 19 10.2
Other disorders5 45 24.1
Note: 1 Previous violent behavior was measured with item 1 of the SAVRY. 2 For 10 participants, previous violent behavior was unknown. 3 Only DSM-IV classifications with a prevalence of >10% are displayed. 4 Due to comorbidity, percentages of DSM-IV classifications do not sum up to 100. 5Other disorders are Mood Disorder, Anxiety Disorder, Gender Identity Disorder, Somatic Symptom Disorder, Sleep Disorder, Impulse Control Disorder and Adjustment Disorder.
Procedure
Three months after admission the historical items of the SAVRY and FPJ were scored based on medical and criminal files. Six months after admission the dynamic items of the SAVRY and FPJ were scored and the historical items of the SAVRY were complemented based on information from the electronic patient records including treatment plans and reports by staff members. When length of stay was less than six months the SAVRY and FPJ were scored at the time of release (for example after four months instead of six months). Scoring of the SAVRY and FPJ was done by qualified psychologists and trainees under supervision. Individual scoring under supervision took place after an interrater reliability of at least 80 percent with a qualified psychologist was reached. Anonymity of the patients was guaranteed by using a research number for each patient.
Measures
Structured Assessment of Violence Risk in Youth (SAVRY). The SAVRY is a risk
assessment instrument consisting of 24 risk factors and six protective factors (Borum et al., 2006). Risk items are scored on a three-point scale with low risk (0), moderate risk (1) and high
risk (2), whereas protective factors are scored on a two-point scale with present (0) and absent (2). For research purposes, the SAVRY Risk Total can be used. The Risk Total is calculated by summing all risk and protective factors into a numerical total score (Lodewijks, Doreleijers, & De Ruiter, 2008).
In earlier studies, the psychometric properties of the SAVRY were demonstrated. The interrater reliability of the SAVRY Risk Total was good to excellent and the predictive validity for violent recidivism varied between fair to excellent (Catchpole & Gretton, 2003; Dolan & Rennie, 2008; Gammelgard, Koivisto, Eronen, & Kaltiala-Heino, 2008; Lodewijks, Doreleijers, De Ruiter, & Borum, 2008; McGowan, Horn, & Mellott, 2011; Meyers & Schmidt, 2008; Welsh, Schmidt, McKinnon, Chattha, & Meyers, 2008). In addition, the convergent validity of the SAVRY with the PCL-YV (Forth, Kosson, & Hare, 2003), the YLS/CMI (Hoge & Andrews, 2002), the Juvenile Sex Offender Assessment Protocol-II (J-SOAP-II; Prentky & Righthand, 2003) and the Juvenile Sexual Offense Recidivism Risk Assessment Tool (J-SORRAT-II; Epperson, Ralston, Fowers, DeWitt, & Gore, 2006) was confirmed (Borum et al., 2006; Viljoen et al., 2008; Welsh et al., 2008).
Juvenile Forensic Profile (FPJ). The FPJ is a standardized list which was specially
developed to give insight into risk factors of an adolescent by using file data (Brand & Van Heerde, 2010). The FPJ consists of 70 risk items which are divided in seven risk domains: History of Criminal Behavior (five items), Family and Environment (22 items), Offense-Related Risk Factors and Substance Use (six items), Psychological Factors (six items), Psychopathology (13 items), Social Behavior/Interpersonal Relationships (four items) and Behavior during Stay in the Institution (14 items). Risk items are scored on a three point scale with no problems (0), mild problems (1) and severe problems (2).
In earlier research, the psychometric properties of the FPJ were evaluated. The interrater reliability of the FPJ can be described as sufficient (0.61; Brand & Van Heerde, 2010). Furthermore, the convergent validity with the SAVRY and the predictive validity were satisfactory (Brand & Van Heerde, 2010).
Statistics
SPSS 19.0 was used for the descriptive, reliability and validity analyses. Factor analyses were performed using Mplus Version 5.2. In this study, two items of the SAVRY were not
incorporated in the statistical analyses due to zero variance: item sixteen (Community Disorganization) of the Social/Contextual domain and item six (Resilient Personality Traits) of the Protective domain. The item Community Disorganization is about a youth’s residence in the past six months. Since all participants stayed at the Catamaran, which is not a disorganized community based on the zip code, all participants scored low (0) on this item. In addition, all participants scored absent (2) on the item Resilient Personality Traits, which means that no participant showed clear traits of a resilient personality.
The internal consistency of the original SAVRY domains was first examined. Then, a confirmative factor analysis (CFA) of the original SAVRY domains was performed with the robust weighted least square estimator (WLSMV). Subsequently, an exploratory factor analysis (EFA) was performed with the WLSMV. To estimate potential correlation among factors, an oblique rotation was first performed (Tabachnick & Fidell, 2007). Since the correlation between the optimal number of factors was apparently zero (r = 0.001), it was decided to perform a VARIMAX rotation. The optimal number of factors was based on the eigenvalues, a scree plot and interpretation of the factors. Factor loadings of <0.300 were excluded (Kline, 1994). The most appropriate factor structure was further investigated by performing a CFA. Since correlated errors were expected due to reversed worded items (risk factors and protective factors), correlated errors of item I23 (Poor Compliance) with item P4 (Positive Attitude towards Authority/Intervention) and item I24 (Low Interest/Commitment to School) with item P5 (Strong Commitment to School) were specified (Brown, 2006). For the CFA’s as well as for the EFA, the Bentler’s comparative fit index (CFI; Bentler, 1990) and the root-mean-square error of approximation (RMSEA; Steiger, 1998) were used to examine the model fit. When the CFI was greater than 0.90 or when the RMSEA was 0.05 or below, the model was indicated as a good fit (Hu & Bentler, 1999). Afterwards, the internal consistency of the modified SAVRY domains was tested. Cronbach’s alpha was used to examine the internal consistency of the original and the modified SAVRY domains. Cronbach’s alpha values of 0.70 or above were defined as acceptable (Nunnally, 1978). Item homogeneity was measured by calculating the mean inter-item correlations.
Finally, scores on two FPJ domains were used to examine the construct validity in terms of the convergent and divergent validity by computing bivariate Pearson correlation coefficients. The FPJ domains used in this study were selected by investigating which FPJ domains were most
likely to correlate with the modified SAVRY domains identified in this study. As a result, two FPJ domains were selected: Family and Environment and Behavior during Stay in the Institution. The Family and Environment domain consists of 22 items about family, peer, and school characteristics, for example ‘Criminality of Family Members’. The Behavior during Stay in the Institution domain includes 14 items about the individual characteristics of the adolescent during stay in an institution, for example ‘Negative Coping Manners’. The criteria of Cohen (1988) were applied for the interpretation of the correlation coefficients with correlations of 0.50 or above interpreted as a large effect.
Table 2
Descriptive statistics of the SAVRY data
N of items M SD Range Historical 10 7.61 3.44 0-17 Social/Contextual 6 4.35 1.67 0-8 Individual 8 6.28 2.92 0-14 Protective 6 9.57 1.76 4-12 Risk Total 30 27.82 5.80 11-41 Results
Internal consistency and model fit of the original SAVRY domains
The descriptive statistics of the SAVRY are shown in Table 2. The Cronbach’s alpha’s of the original factor structure were not satisfactory (see Table 3). The lowest Cronbach’s alpha was found for the Social/Contextual domain ( 0.05) and the highest, but still not satisfactory, Cronbach’s alpha was found for the Individual domain ( = 0.56).
Table 3
Internal consistency of the original SAVRY domains and the modified SAVRY domains
Domains N of items Cronbach’s alpha Item homogeneity Original domains Historical 10 0.53 0.10 Social/Contextual 6 0.05 0.01 Individual 8 0.56 0.14 Protective 6 0.24 0.06 Modified domains Social/Contextual 7 0.65 0.21 Individual 12 0.74 0.19
Table 4
Fit indices of the CFA of the original four-factor structure of the SAVRY
Fit index1 Value
CFI 0.62
RMSEA 0.10
R2 43.77
Note: 1 due to zero variance, item 16 (Community Disorganization) and item P6 (Resilient Personality Traits) were not incorporated into the CFA for the original four-factor structure of the SAVRY.
The CFA of the original SAVRY four-factor structure did not meet the cut-off values (see Table 4). The values of the CFI (0.62) and RMSEA (0.10) indicated that the model, in particular the Protective domain, did not fit the data well. However, testing the CFA without the Protective domain did not improve the fit of the data (CFI = 0.57, RMSEA = 0.10).
The EFA two-factor structure of the SAVRY
The EFA showed that five factors had eigenvalues greater than 2 and factors six to eleven had eigenvalues greater than 1 (see Appendix). The examination of eigenvalues and the scree plot indicated a five-factor structure as most appropriate. However, the five-factor solution resulted in uninterpretable factors with the fifth factor containing only one item. The three- and four-factor solution also consisted of uninterpretable factors with respectively three and four cross loadings. Therefore, interpretability of the factors resulted in the two-factor structure as most relevant to examine in a CFA (see Table 5).
Due to cross loadings above 0.300 or low factor loadings, eight items were not included in the CFA of the two-factor structure (see Table 6): item 1 (History of Violence), item 5 (History of Self-Harm/Suicide Attempts), item 10 (Poor School Achievement), item 15 (Lack of Personal/Social Support), item 18 (Risk Taking/Impulsivity), item 22 (Attention Deficit/Hyperactivity), item P1 (Prosocial Involvement) and item P3 (Strong Attachments and Bonds). With specification of correlated errors, values of the CFI (0.88) and the RMSEA (0.07) indicated a fair model fit. The Cronbach’s alpha of the Individual domain was satisfactory ( = 0.74; see Table 3), yet the Cronbach’s alpha of the Social/Contextual domain was minimally acceptable ( = 0.65).
Table 5
Varimax Exploratory factor analysis of the SAVRY
Factor 1: Social/Contextual Factor 2: Individual Historical factors 1 History of Violence 0.145 0.104
2 History of Non-Violent Offending 0.026 0.321
3 Early Initiation of Violence 0.441 0.065
4 Past Supervision/Intervention Failures 0.025 0.539
5 History of Self-Harm/Suicide Attempts 0.150 -0.067
6 Exposure to Violence in the Home 0.743 0.053
7 Childhood History of Maltreatment 0.882 -0.072
8 Parental/Caregiver Criminality 0.911 -0.033
9 Early Caregiver Disruption 0.501 0.216
10 Poor School Achievement -0.022 0.013
Social/Contextual factors1
11 Peer Delinquency -0.285 0.355
12 Peer Rejection 0.007 -0.074
13 Stress and Poor Coping 0.021 0.364
14 Poor Parental Management 0.330 0.101
15 Lack of Personal/Social support 0.046 0.097
Individual factors
17 Negative attitudes -0.004 0.479
18 Risk Taking/Impulsivity -0.183 0.244
19 Substance use Difficulties -0.174 0.354
20 Anger Management Problems 0.086 0.459
21 Low Empathy/Remorse -0.064 0.469
22 Attention Deficit/Hyperactivity -0.136 0.087
23 Poor Compliance -0.028 0.803
24 Low Interest in/Commitment to School 0.014 0.665
Protective factors1
1 Prosocial Involvement 0.130 0.286
2 Strong Social Support 0.322 0.084
3 Strong Attachments and Bonds -0.005 0.119
4 Positive Attitude towards Intervention/Authority 0.001 0.982
5 Strong Commitment to School 0.149 0.660
Note: 1due to zero variance, item 16 (Community Disorganization) and item P6 (Resilient Personality Traits) were not incorporated into the Varimax EFA.
Table 6
Confirmatory factor analysis of the two-factor structure of the SAVRY
Factor loading Factor 1: Contextual
3 Early Initiation of Violence 0.373
6 Exposure to Violence in the Home 0.771
7 Childhood History of Maltreatment 0.866
8 Parental/Caregiver Criminality 0.926
9 Early Caregiver Disruption 0.502
14 Poor Parental Management 0.359
P2 Strong Social Support 0.330
Factor 2: Individual
2 History of Non-Violent Offending 0.354
4 Past Supervision/Intervention Failures 0.582
11 Peer Delinquency 0.346
13 Stress and Poor Coping 0.364
17 Negative Attitudes 0.499
19 Substance use Difficulties 0.359
20 Anger Management Problems 0.492
21 Low Empathy/Remorse 0.524
23 Poor Compliance 0.692
24 Low Interest/Commitment to School 0.595
P4 Positive Attitude towards Intervention/Authority 0.836
P5 Strong Commitment to School1 0.543
Fit indices
CFI 0.88
RMSEA 0.07
% of explained variance 33.01
Note: 1due to zero variance, item 16 (Community Disorganization) and item P6 (Resilient Personality Traits) were not incorporated into the CFA.
Construct validity of the modified SAVRY domains with the FPJ
Concerning the convergent validity, there was a significant relation between the Social/Contextual domain of the SAVRY and the Family and Environment domain of the FPJ (see Table 7; r = 0.70, p <0.01). Likewise, a significant relation was found between the Individual domain of the SAVRY and the Behavior during Stay in the Institution domain of the FPJ (r = 0.68, p <0.01). Concerning the divergent validity, no relation was found between the Social/Contextual domain of the SAVRY and the Behavior during Stay in the Institution domain of the FPJ (r = 0.07, p >0.05) nor between the Individual domain of the SAVRY and the Family
and Environment domain of the FPJ (r = 0.30, p >0.05). These results indicate a good construct validity of the modified SAVRY domains with the FPJ.
Table 7
Construct validity of the modified SAVRY domains with the FPJ
Note: * p <0.01
Discussion
To our knowledge, this was the first study in which the factor structure of the SAVRY was examined. Specifically, the aim was to investigate whether the original SAVRY domains were reliable and valid domains for scoring the SAVRY by using file data. The original four-factor structure of the SAVRY could not be replicated. This may be due to the original structure of the SAVRY being not empirically defined, but theory-driven by studies in which the risk and protective factors were found to be connected with violent delinquency (Borum et al., 2006; Childs et al., 2013). Since the original structure of the SAVRY did not fit the data well, investigation for another factor structure was justified.
As a result, a two-factor structure was found with a Social/Contextual domain and an Individual domain. The combination of static and dynamic items in these domains corresponds with previous studies in which early abnormal child and parenting behaviors were found to be related to later child and parenting problems (Fergusson, 2007; Herrenkohl et al., 2000; Loeber & Farrington, 2000; Moffit, 1993; Moffitt & Caspi, 2001; Patterson, DeBaryshe, & Ramsey, 1989; Timmermans, Van Lier, & Koot, 2009). In the Social/Contextual domain the emphasis is on parental characteristics. However, the item Early Initiation of Violence included in this factor may be more of an individual risk factor. Yet, the inclusion of this item in the Social/Contextual domain is supported by studies in which the impact of negative parenting behaviors on child’s early onset of antisocial behavior was demonstrated (Aguilar et al., 1985; Eckenrode et al., 2001; Moffitt & Caspi, 2001; Thornberry, Freeman-Gallant, & Lovegrove, 2009). Instead of parental characteristics, in the Individual domain the adolescent’s individual risk and protective factors for
FPJ
Family and Environment Behavior during Stay in the Institution
SAVRY Social/Contextual domain
0.70* 0.07
violent delinquency are central. The inclusion of both risk and protective factors in this domain suggests a counter-balancing effect of risk and protective factors. In other words, the more risk factors are present and the more protective factors are absent, the more likelihood of delinquent behavior (Stouthamer-Loeber, Loeber, & Wei, 2002; Van der Laan, Veenstra, Bogaerts, Verhulst, & Ormel, 2010; Vanderbilt-Adriance & Shaw, 2008). Several times a reciprocal interaction between individual and parental characteristics was indicated with initial risk factors for delinquency being situated in the child domain, whereas parental risk factors as physical abuse play an important role in the maintenance and continuity of delinquency (Loeber & Farrington, 2000; Moffit, 1993; Patterson et al., 1989; Stouthamer-Loeber et al., 1993; Stouthamer-Loeber et al., 2002). This reciprocal interaction between individual and parental characteristics indicates that the Social/Contextual domain and the Individual domain are both needed in order to gain an accurate overall picture of the adolescent’s risk and protective factors for delinquent behavior.
Concerning the validity of the two-factor structure, it was demonstrated that the Social/Contextual and the Individual domain estimate the same constructs as respectively the Family and Environment domain and the Behavior during Stay in the Institution domain of the FPJ. The SAVRY was not officially designed for scoring based on file data (Borum et al., 2006). Yet, the high coherency between the modified SAVRY domains with the domains of the FPJ, an instrument specially designed for scoring risk factors by using file data, affirms earlier findings that risk assessment instruments can be reliably scored based on file data (Catchpole & Gretton, 2003; Grann, Långström, Tengström, & Kullgren, 1999; Gretton, McBride, Hare, O’Shaughnessy, & Kumka, 2001). In these studies the reliability of the SAVRY domains scored by using file data was not formally tested, whereas our results suggest that the two-factor structure is a promising structure when scoring the SAVRY based on file data.
In this study, the protective item Resilient Personality was always scored absent. Individuals with a resilient personality are defined as ‘exhibiting positive emotionality, being socially adept and being able to regulate their emotions effectively’ (Atkins, 2007). Personality characteristics like these are hard to determine on the basis of file data, which might be the cause of scoring absent for all participants. This indicates that probably not every item of the SAVRY can be reliably scored by using file data, which could possibly explain the item loss in this study. Another reason for the item loss is the research population (Kline, 1994). Given all participants being male forensic psychiatric patients who were staying at De Catamaran, for instance the item
Community Disorganization was scored low for all patients. Therefore, the found structure has to be replicated.
The generalizability is one of the limitations which have to be taken in mind when drawing conclusions from this study. Hence, more studies with larger samples including female forensic psychiatric patients examining our findings are needed. Another limitation is that the predictive validity of the modified SAVRY domains was not investigated. Since an incorrect prediction of youths can have serious negative consequences for both youths and wider society (Catchpole & Gretton, 2003), for future research it would be valuable to investigate the predictive validity of the modified SAVRY domains for violent recidivism. For example, this could be done with a Chi-squared Automatic Interaction Detecter (CHAID) analysis in order to examine interaction effects between risk factors, which can then be used to identify groups with low or high risk of recidivism (Kass, 1980; Van der Put et al., 2010).
Given the minimal research that has been conducted on the reliability of the SAVRY domains when using file data, this study is important since a two-factor structure instead of the original four-factor structure was identified for scoring the SAVRY based on file data. More research with larger samples should investigate whether the two-factor structure has predictive validity for violent recidivism in order to further support the usefulness of this two-factor structure.
References
Aguilar, B., Sroufe, L. A., Egeland, B., Carlson, E. , Donovan, J. E., & Jessor, R. (1985). Distinguishing the early-onset/persistent and adolescent-onset antisocial behavior types: From birth to 16 years. Development and Psychopathology, 12, 109-132.
APA. (2004). Diagnostic and statistical manual of mental disorders, fourth edition-text revision (DSM-IV-TR). Washington: American Psychiatric Association.
Atkins, R. (2007). The association of personality type in childhood with violence in adolescence. Research in Nursing & Health, 30, 308-319. doi: 10.1002/nur.20214
Bentler, P.M. (1990). Comparative fix indexes in structural models. Psychological Bulletin, 107, 238-246.
Borum, R., Bartel, P. , Forth, A. , Lodewijks, H. , Doreleijers, T. , De Ruiter, C., & De Wit-Grouls, H. (2006). SAVRY: Handleiding voor de Gestructureerde Risicotaxatie van Gewelddadig Gedrag bij Jongeren. Zutphen: Rentray.
Brand, E.F.J.M., & Van Heerde, W.K. (2010). Handleiding FPJ. Forensisch Profiel Justitiële Jeugdigen. Editie 2010. Den Haag: Dienst Justitiële Inrichtingen.
Brown, T.A. (2006). Confirmatory Factor Analysis for Applied Research. New York: The Guilford Press.
Catchpole, R., & Gretton, H. (2003). The predictive validity of risk assessment with violent young offenders: A 1-year examination of criminal outcome. Criminal Justice & Behavior, 30, 688-708. doi: 10.1177/0093854803256455
Childs, K., Ryals, J., Frick, P.J., Lawing, K., Phillipi, S.W., & Deprato, D.K. (2013). Examining the validity of the Structured Assessment of Violence Risk in Youth (SAVRY) for predicting probation outcomes among adjudicated juvenile offenders. . Behavioral Sciences and the Law, 31, 256-270. doi: 10.1002/bsl.2060
Cohen, J. (1988). Statistical Power Analysis (2nd edition). Hillsdale: Lawrence Erlbaum. Cohen, M., & Piquero, A. (2009). New evidence on the monetary value of saving a high risk
youth. Journal of Quantitative Criminology, 25, 25-49. doi: 10.1007/s10940-0089057-3 Dolan, M.C., & Rennie, C.E. (2008). The Structured Assessment of Violence Risk in Youth
(SAVRY) as a predictor of recidivism in a UK cohort of adolescent offenders with conduct disorder. Psychological Assessment, 20, 35-46. doi: 10.1037/1040-3590.20.1.35 Douglas, K.S., & Kropp, P.R. (2002). A prevention-based paradigm for violence risk assessment:
Clinical and reserach applications. Criminal Justice and Behavior, 29, 617-658. doi: 10.1177/009385402236735
Eckenrode, J., Zielinski, D., Smith, E., Marcynyszyn, L.A., Henderson, C.R., Kitzman, H., . . . Olds, D.L. (2001). Child maltreatment and the early onset of problem behaviors: Can a program of nurse home visitation break the link? Development and Psychopathology, 13, 873-890.
Epperson, D.L., Ralston, C.A., Fowers, D., DeWitt, J., & Gore, K.S. (2006). Actuarial risk assessment with juveniles who offend sexually. Development of the Juvenile Sexual Offense Recidivism Risk Assessment Tool II (J-SORRAT-II). Oklahoma City: Woods N Barnes.
Fazel, S., Singh, J. P., Doll, H., & Grann, M. (2012). Use of risk assessment instruments to predict violence and antisocial behavior in 73 samples involving 24,827 people: Systematic review and meta-analysis. British Medical Journal, 345, 1-12. doi: 10.1136/bmj.e4692
Fergusson, D.M., Horwood, L.J., & Ridder, E.M. (2007). Conduct and attentional problems in childhood and adolescence and later substance use, abuse and dependence: Results of a 25-year longitudinal study. Drug and Alcohol Dependence, 88, 14-26. doi:
10.1016/j.drugalcdep.2006.12.011
Forth, A.E., Kosson, D.S., & Hare, R.D. (2003). The Psychopathy Checklist: Youth Version. Toronto: Multi-Health Systems.
Gammelgard, M., Koivisto, A.M., Eronen, M., & Kaltiala-Heino, R. (2008). The predictive validity of the Structured Assessment of Violence Risk in Youth (SAVRY) among institutionalised adolescents. The Journal of Forensic Psychiatry & Psychology, 19, 352-370. doi: 10.1080/14789940802114475
Grann, M., Långström, L., Tengström, A., & Kullgren, G. (1999). Psychopathy (PCL-R) predicts violent recidivism among criminal offenders with personality disorders in Sweden. Law and Human Behavior, 23, 205-217.
Gretton, H. M., McBride, M., Hare, R. D., O’Shaughnessy, R., & Kumka, G. (2001).
Psychopathy and recidivism in adolescent sex offenders. Criminal Justice & Behavior, 28, 427-449. doi: 10.1177/009385480102800403
Herrenkohl, T.I., Maguin, E., Hill, K.G., Hawkins, J.D., Abbott, R.D., & Catalano, R.F. (2000). Developmental risk factors for youth violence. Journal of Adolescent Health, 26, 176-186. doi: 10.1016/S1054-139X(99)00065-8
Hilterman, E.L.B., & Van Nieuwenhuizen, Ch. (2013). Predictive validity of risk assessments in juvenile offenders: Comparing the SAVRY, PCL:YV and YLS/CMI with unstructured clinical assessments. Assessment, 1-16. doi: 10.1177/1073191113498113
Hoge, R., & Andrews, D. (2002). The Youth Level of Service/Case Managament Inventory manual and scoring key. Toronto: Multi-Health Systems.
Hu, L.T., & Bentler, P.M. (1999). Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Structural Equation Modeling 6, 1-55. Kass, G.V. (1980). An exploratory technique for investigating large quantities of categorical data.
Journal of Applied Statistics, 29(2), 119-127.
Kline, P. (1994). An easy guide to factor analysis. London: Routledge.
Lodewijks, H.P.B., Doreleijers, T.A.H., & De Ruiter, C. (2008). Savry risk assessment in violent dutch adolescents: Relation to sentencing and recidivism. Criminal Justice and Behavior, 35, 696-709. doi: 10.1177/0093854808316146
Lodewijks, H.P.B., Doreleijers, T.A.H., De Ruiter, C., & Borum, R. (2008). Predictive validity of the Structured Assessment of Violence Risk in Youth (SAVRY) during residential
treatment. International Journal of Law and Psychiatry, 31, 263-271. doi: 10.1016/j.ijlp.2008.04.009
Loeber, R., & Farrington, D.P. (2000). Young children who commit crime: Epidemiology, developmental origins, risk factors, early interventions, and policy implications. Developmental and Psychopathology, 12, 737-762.
McGowan, M.R., Horn, R.A., & Mellott, R.N. (2011). The predictive validity of the Structured Assessment of Violence Risk in Youth in secondary educational settings. Psychological Assessment, 23, 478-486. doi: 10.1037/a0022304
Meyers, J., & Schmidt, F. (2008). Predictive valdity of the Structured Assessment of Violence Risk in Youth (SAVRY) with juvenile offenders. Criminal Justice & Behavior, 35, 344-355. doi: 10.1177/0093854807311972
Moffit, T.E. (1993). Adolescence-limited and life-course-persistent antisocial behavior: A developmental taxonomy. Psychological Review, 100, 674-701. doi: 10.1037/0033-295X.100.4.674
Moffitt, T. E., & Caspi, A. (2001). Childhood predictors differentiate life-course persistent and adolescent-limited antisocial pathways among males and females. Developmental and Psychopathology, 13, 355-375.
Nunnally, J.C. (1978). Psychometric Theory. New York: McGraw Hill.
Olver, M. E., Stockdale, K. C., & Wormith, J. S. (2009). Risk assessment with young offenders: A meta analysis of three assessment measures. Criminal Justice & Behavior, 36, 329-353. Patterson, G.R., DeBaryshe, D., & Ramsey, E. (1989). A developmental perspective on antisocial
behavior. American Psychologist, 44, 329-335.
Prentky, R.A., & Righthand, S. (2003). Juvenile Sex Offender Assessment Protocol II (J-SOAP-II) manual. Washington, DC: U.S. Department of Justice, Office of Justice Programs, Office of Juvenile Justice and Delinquency Prevention.
Schmidt, F., Campbell, M.A., & Houlding, C. (2011). Comparative analsyis of the YLS/CMI, SAVRY and PCL:YV in adolescent offenders: A 10-year follow-up into adulthood. Youth Violence and Juvenile Justice, 9, 23-42. doi: 10.1177/1541204010371793
Singh, J.P., Grann, M., & Fazel, S. (2011). A comparative study of violence risk assessment tools: A systematic review and metaregression analysis of 68 studies involving 25.980 participants. Clinical Psychology Review, 31, 499-513. doi: 10.1016/j.cpr.2010.11.009 Steiger, J.H. (1998). A nota on multiple sample extensions of the RMSEA fit index. Structural
Equation Modeling, 5, 411-419.
Stouthamer-Loeber, M., Loeber, R., Farrington, D. P., Zhang, Q., Kammen, W. van, & Maguin, E. (1993). The double edge of protective and risk factors for delinquency: Interrelations and developmental patterns. Developmental and Psychopathology, 5, 683-701.
Stouthamer-Loeber, M., Loeber, R., & Wei, E. (2002). Risk and promotive effects in the
explanation of persistent serious delinquency in boys. Journal of Consulting and Clinical Psychology, 70, 111-123. doi: 10.1037//0022-006X.70.1.111
Tabachnick, B. G., & Fidell, L. S. (2007). Using multivariate statistics. New Jersey: Pearson Allyn & Bacon.
Thornberry, T.P., Freeman-Gallant, A., & Lovegrove, P.J. (2009). The impact of parental stressors on the intergenerational transmission of antisocial behavior. Journal of Youth and Adolescence, 38, 312-322. doi: 10.1007/s10964-008-9337-0
Timmermans, M., Van Lier, P.A.C., & Koot, H.M. (2009). Pathways of behavior problems from childhood to late adolescence leading to delinquency and academic underachievement. Journal of Clinical Child and Adolescent Psychiatry, 38(5), 630-638. doi:
10.1080/15374410903103502
Van der Laan, A.M., Veenstra, R., Bogaerts, S., Verhulst, F.C., & Ormel, J. (2010). Serious, minor, and non-delinquents in early adolescence: The impact of cumulative risk and protective factors. The TRAILS study. Journal of Abnormal Child Psychology, 38, 339-351. doi: 10.1007/s10802-009-9368-3
Van der Put, C.E., Dekovic, M., Hoeve, M., Stams, G.J.J.M., Van der Laan, P. H., &
Langewouters, F.E.M. (2010). Risk assessment of girls: Are there any sex differences in risk factors for re-offending and in risk profiles? . Crime & Delinquency, xx(x), 1-24. doi: 10.1177/0011128710384776
Vanderbilt-Adriance, E., & Shaw, D. S. (2008). Conceptualizing and re-evaluating resilience across levels of risk, time, and domains of competence. Clinical Child and Family Psychology Review, 11, 30-58.
Viljoen, J.L., Scalora, M.J., Cuadra, L., Bader, S., Chavez, V., Ullman, D., & Lawrence, L. (2008). Assessing risk for violence in adolescents who have sexually offended: A
comparison of the J-SOAP-II, J-SORRAT-II, and SAVRY. Criminal Justice & Behavior, 35, 5-23. doi: 10.1177/0093854807307521
Vincent, G.M. (2012). Does risk assessment make a difference? Results of implementing the SAVRY in juvenile probation. Behavioral Sciences and the Law, 30, 384-405. doi: 10.1002/bsl.2014
Vincent, G.M., Chapman, J., & Cook, N.E. (2011). Risk-needs assessment in juvenile justice: Predictive validity of the SAVRY, racial differences, and the contribution of needs factors. Criminal Justice & Behavior, 38, 42-62. doi: 10.1177/0093854810386000 Vincent, G.M., Guy, L.S., Fusco, S.L., & Gershenson, B.G. (2012). Field reliability of the
SAVRY with juvenile probation officers: Implications for training. Law and Human Behavior, 36, 225-236. doi: 10.1037/h0093974
Welsh, J.L., Schmidt, F., McKinnon, L., Chattha, H.K., & Meyers, J.R. (2008). A comparative study of adolescent risk assessment instruments: Predictive and incremental validity. Assessment, 15, 104-115. doi: 10.1177/1073191107307966
Appendix
Screeplot and Eigenvalues of the Varimax Exploratory Factor Analysis of the SAVRY
Figure 1
Scree plot of the Varimax Exploratory factor analysis of the SAVRY 0 1 2 3 4 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 E igenval u e Factor Number
Table 8
Eigenvalues of the Varimax Exploratory factor analysis of the SAVRY
Number of factors Eigenvalues
1 3.601 2 3.141 3 2.578 4 2.307 5 2.071 6 1.853 7 1.546 8 1.411 9 1.253 10 1.192 11 1.080 12 0.915
