The Role of Protective Factors and the Relationship With Recidivism for High-Risk Young People in Detention

Abstract

This study sought to ascertain the prevalence of protective factors and association with client risk level and future offending in a sample of 177 Australian youth in detention. The Protective Domain on the Structured Assessment of Violence Risk in Youth (SAVRY) instrument was utilized to identify protective items in the cohort. The mean number of protective factors for the entire sample was low (under two) with higher risk clients averaging less than one current protective item. Although the number of protective factors engendered criminal desistance, this effect did not extend to the highest risk young offenders. Clients who re-offended were significantly less likely to present with five out of the six SAVRY protective items. In addition, pro-social involvement and school engagement had the strongest associations with non re-offense. Clinical implications for client risk management are discussed.

Keywords

protective factors criminal desistance SAVRY violence risk assessment juvenile delinquency forensic psychology

It is well established that a concert of environmental and psychosocial risk markers increases the likelihood of an individual engaging in violence (see Borum, 2000; Borum & Verhaagen, 2006). The consideration of these items, often through structured violence risk instruments, is now a customary part of forensic risk assessment. The information derived from the instruments typically informs clinical and correctional decision making pertaining to offender management.

Traditionally, risk assessment has predominantly focused on evaluating the aspects that contribute to future antisocial behavior. Items that reduce the risk of offending have received comparatively less attention. These items are known collectively as protective factors and largely entail the positive attributes and strengths of an individual. Academic interest in protective factors has deepened after foundational criminal desistance literature and developmental longitudinal studies flagged commonalities among youth with reduced levels of antisocial behavior (Farrington & West, 1993; Laub & Sampson, 2001; Loeber, Farrington, Stouthamer-Loeber, Moffitt, & Caspi, 1998; Moffitt, Caspi, Dickson, Silva, & Stanton, 1996; Nagin & Tremblay, 1999; Stouthamer-Loeber, Wei, Loeber, & Masten, 2004). Resnick (2000) illustrated protective factors as the interaction of extra familial environmental processes (i.e., community and peer groups), familial processes (i.e., family and parenting dynamics), self-system processes (i.e., connectedness and social responsibility), and individual characteristics (i.e., psychosocial and cognitive development).

The disparity between risk and protective factor research has been described as therapeutically negative and may foster the inaccurate evaluation and prediction of risk (Rogers, 2000). Several analysts now agree that the consideration of both risk and protective factors enables a balanced appraisal of a client and assists in the development of treatment initiatives aimed at reducing risk (DeMatteo, Heilbrun, & Marczyk, 2005; Salekin & Lochman, 2008; de Vogel, de Vries Robbe, de Ruiter, & Bouman, 2011). These assertions are supported by a growing body of research demonstrating the role protective factors play in reducing the probability of future violence (de Vries Robbe, de Vogel, & Douglas, 2013; Lodewijks, de Ruiter, & Doreleijers, 2010; Rennie & Dolan, 2010; Wooditch, Tang, & Taxman, 2014). Moreover, the prominence of offender strengths and attainment of human goods has been emphasized in contemporary models of offender rehabilitation (Ward & Stewart, 2003).

Despite these advances, few risk assessment instruments to date have constitutionally incorporated protective items. Among those instruments are the Structured Assessment of Violence Risk in Youth (SAVRY; Borum, Bartel, & Forth, 2006), the Short-Term Assessment of Risk and Treatability (START; Webster, Martin, Brink, Nicholls, & Middleton, 2004), the Structured Assessment of Protective Factors for Violence Risk (SAPROF; de Vogel, de Ruiter, Bouman, & de Vries Robbe, 2009), and its youth extension the, SAPROF-YV (de Vries Robbe, Geers, Stapel, Hilterman, & de Vogel, 2015). The SAVRY and START instruments comprise both risk and protective items, whereas the SAPROF entirely encompasses protective items. In addition, Risk/Need instruments, the Level of Service/Risk, Needs & Responsivity (LS/RNR; Andrews, Bonta, & Wormith, 2009), the Level of Service/Case Management Inventory (LS/CMI; Andrews, Bonta, & Wormith, 2004), and the Youth Level of Service/Case Management Inventory (YLS/CMI; Hoge & Andrews, 2006), all permit raters to adjudge a risk domain as a strength.

The inclusion of dynamic (changeable) factors in risk approaches has provided opportunities to consider protective items for the mitigation of risk and reduction in the likelihood of re-offending. Yet, protective factor research is presently inadequate due to its contemporary introduction into the risk literature and the disproportionate focus on risk. Addressing the factors that may reduce offending through positive influences on an offender’s life has not been extensively explored. One possible explanation for the limited focus on protective factors is that the existing literature is marked by definitional inconsistency. Discrepancies remain over both the meaning and utility of protective items (see Walker, Bowen, & Brown, 2013). Some investigations have operationalized protective factors as representing the “weak” end of a risk item spectrum (Henry, Tolan, Gorman-Smith, & Schoeny, 2012; Herrenkohl, Lee, & Hawkins, 2012; Stouthamer-Loeber, Loeber, Wei, Farrington, & Wikstrom, 2002; Webster et al., 2004). For example, associating with deviant peers is a risk factor (Lipsey & Derzon, 1998), and so to not associate with deviant peer groups is directly “protective.” In contrast, protective factors are also viewed as potentially unique, stand-alone markers that may be separate from the existence (or non-existence) of a risk item (Borum et al., 2006; de Vogel et al., 2009). Under this rationale, the absence of delinquent peers is not the protective factor; rather, deliberate association with pro-social peers provides the protective aspect. In this regard, “positive” protective factors may interactively mitigate negative outcomes in the presence of risk (Jessor, Van Den Bos, Venderryn, Costa, & Turbin, 1995; Losel & Farrington, 2012; Hall et al., 2012; de Vogel et al., 2009; de Vries Robbe et al., 2013). Despite suggestion that some protective factors can be simultaneously promotive and mitigating, it is unclear to what extent this occurs. This is a disadvantage of the extant literature and poses a number of difficulties for their measurement and intended clinical use. As such, there is a need to conduct further research on the impact of protective items for individuals at risk of future violence.

The investigation of protective factors for young offenders is particularly important given that the offending rates of young people aged 10 to 17 years are often twice the rates of adults (Australian Institute of Criminology, 2014). Young offenders also comprise approximately one quarter of the total offender population in Australia (Australian Bureau of Statistics, 2015). Of this group of young offenders, many will desist from criminal activity entirely as they mature (Broidy et al., 2003; Laub & Sampson, 2003). However, a sizable minority will continue their offending (and offense severity) into adulthood. Given that almost 1,000 young people are in detention in Australia on any given day (Australian Institute of Health and Welfare [AIHW], 2014), it is necessary to understand the mechanisms that initiate desistance so that amenable youth can be averted from potential life course criminality. Thus, the identification of markers that significantly reduce the likelihood of violence and other crime becomes of great consequence to the quality of life for at-risk youth and the safety of the broader community.

The available protective factor literature comprises variegated cohorts of both adolescent and pre-adolescent school students (Australian Research Alliance for Children & Youth [ARACY], 2009; Henry et al., 2012; Herrenkohl et al., 2012; Jessor et al., 1995; Pardini, Loeber, Farrington, & Stouthamer-Loeber, 2012; Resnick, Ireland, & Borowsky, 2004) to young offenders under supervision (Gammelgard, Weizmann-Henelius, Koivisto, Eronen, & Kaltiala-Heino, 2012; Hoge, Andrews, & Leschied, 1996; Lodewijks et al., 2010; Rennie & Dolan, 2010). Despite sample variations and inconsistent outcome methodology, common protective themes have emerged in the literature. School engagement and achievement, family connection, non-delinquent peers, low propensity for risk taking, and pro-social attitudes have been regularly associated with a lower risk of both violence and other problem behaviors for young people (ARACY, 2009; Bernat, Oakes, Pettingell, & Resnick, 2012; Bushway, Krohn, Lizotte, Phillips, & Schmidt, 2013; Henry et al., 2012; Herrenkohl et al., 2012; Hoge et al., 1996; Jessor et al., 1995; Pardini et al., 2012; Resnick et al., 2004; Shlafer, McMorris, Sieving, & Gower, 2013; Stouthamer-Loeber et al., 2004).

Moreover, protective factor components on adolescent violence risk assessment instruments such as the SAVRY have demonstrated associations with decreased criminal involvement for young offenders (Dolan & Rennie, 2008; Lodewijks et al., 2010; Rennie & Dolan, 2010; Shepherd, Luebbers, Ferguson, Ogloff, & Dolan, 2014; Spice, Viljoen, Gretton, & Roesch, 2010). Despite these findings, there is limited research on the role of protective factors in understanding and managing adolescent offending. An implication of the RNR (Andrews & Bonta, 2010) risk management model is that different treatment approaches should be adopted for offenders with different levels of risk with a focus on criminogenic factors. As such, it is important to ascertain whether an increase in protective factors is associated with a reduced propensity to offend, irrespective of a client’s level of risk.

This study aims to investigate the prevalence of protective factors and their impact on recidivism in a sample of young offenders in custody. The SAVRY instrument was utilized to capture the presence of protective items. Existing SAVRY studies have found the magnitude of the protective domain to be inversely related to violent recidivism (Lodewijks et al., 2010; Rennie & Dolan, 2010); however, the paucity of this type of research has led to calls for replication. In addition, no Australian study has comprehensively investigated the influence of protective items on recidivism outcome and level of client risk. Our primary aims were to (a) identify the prevalence of each SAVRY protective item in a sample of young offenders in custody, (b) explore differences in the number of protective factors across youth with differing levels of risk, (c) determine whether the number of protective factors affect time to re-offend across levels of risk, (d) explore differences in the number of protective factors across recidivists and non-recidivists, and (e) establish which protective items have the strongest influence on recidivistic outcome. In light of the available research, we anticipated recidivists and high-risk youth to present with fewer protective items. Given the importance of school engagement and positive peer support in adolescence, we expected the SAVRY protective factor items Pro-social Involvement, Strong Social Support, and Strong Commitment to School to predict non re-offense. Last, we anticipate participants exhibiting greater numbers of protective factors to be more likely to desist from offending. We expect this result to hold across levels of risk.

Method

Participants

A sample of 177 young males was recruited from Youth Justice custody centers: Parkville Youth Justice Precinct (PYJP) and Malmsbury Youth Justice Centre (MYJC) in Victoria, Australia. PYJP accommodates adolescents aged 10 to 17 years, and MYJC accommodates young men aged 18 to 20. The final sample was reduced (N = 175), as two participants had incomplete data. The study capture rate over 12 months was very high ensuring a representative sample. For example, in the state of Victoria, 150 young people are in custody on any given night (AIHW, 2014). The mean age of the cohort was 16.9 years (SD = 1.8, range 13-21). The mean length of sentence that participants were servicing was 8.7 months (SD = 15.5). All participants had a self-reported history of violence, and 97% of the sample had previously received a police charge for a violent offense. Almost half the sample (48%, n = 84) self-identified as being from an English-Speaking Background (ESB), 34% self-identified as of a Culturally and Linguistically Diverse Background (n = 59), and 18% self-identified as Aboriginal and/or Torres Strait Islander (IND, 18%, n = 32).

The sample included 35 (20%) young people between 18 and 20 years of age. This reflects the composition of the Australian youth justice system of which 19% fall into this age group (AIHW, 2013). The state of Victoria is subject to Dual Track legislation, which allows for young offenders between 18 and 20 to be processed by either the youth or adult criminal justice systems. This system is intended to triage to the youth justice system, a subset of young adult offenders who are particularly immature and impressionable with strong prospects for rehabilitation (see Luebbers & Ogloff, 2011).

Materials

The SAVRY is a risk assessment protocol designed for gauging violence risk in young people aged 12 to 18 years. It comprises three risk domains (Historical, Socio/Contextual, Individual/Clinical) encompassing 24 items, and a Protective domain comprising of six items. Risk items are rated low, moderate, or high as per SAVRY operational guidelines, whereas Protective items (see Table 1) are scored as either absent or present. The instrument follows the Structured Professional Judgment model where an assessor arrives at an overall risk judgment (SAVRY Summary Risk Rating) of High, Moderate, or Low after considering a concert of risk items that have demonstrated empirical associations with violence, recidivism, or problem behaviors in the literature. For research purposes, SAVRY risk items are often tallied to produce a total score. Low, Moderate, and High ratings correspond to scores of 0, 1, and 2, respectively. The dichotomous SAVRY protective items (Absent = 0; Present = 1) can be calculated to generate a total Protective factor score.

Table 1:

SAVRY Protective Domain

SAVRY Protective Items
1. Pro-social involvement
2. Strong social support
3. Strong attachments and bonds
4. Positive attitude toward intervention and authority
5. Strong commitment to school
6. Resilient personality traits

Source: SAVRY (Structured Assessment of Violence Risk in Youth; Borum et al., 2006).

Procedure

Detention center clients were approached by researchers and asked whether they were interested in hearing about the study. If they demonstrated interest in participating, the study was explained to them in detail by the researcher. Written informed consent was then obtained from clients interested in partaking. If a participant was unable to understand the nature of the project or give informed consent, he or she was not invited to take part in the study.

Participants were individually interviewed in private rooms in the youth justice centers. Each interview was approximately 90 min in length and was conducted by master’s-level clinician-researchers who were trained in the use of the SAVRY instrument. The SAVRY was coded using information from a semi-structured interview and collateral information from the Department of Human Services Client Information System for Service Providers Database (CRISSP) and the Victorian Police Law Enforcement Assistance Program (LEAP) database. The CRISSP database included client demographics, case management notes, and court order information. The LEAP database comprised the consenting participants’ criminal histories. Inter-rater reliability was obtained for 11 (6.3%) participants rated independently. Strong concordance was achieved for the Protective Domain: intraclass correlation coefficient (ICC) = .996 (α = .996).

Recidivism

Follow-up recidivism data (police charges) were obtained for participants who were released during the study and who consented to researchers accessing their criminal records. Follow-up data were collected for at least 6 months for all consenting participants. The maximum follow-up time was 18 months. General recidivism was defined as any future incident that resulted in a police charge (excluding technical breaches of orders). Violent recidivism was defined as any act intended to cause physical harm to another person.

Data Analyses

Data were analyzed using IBM SPSS statistics Version 22. The occurrence of individual SAVRY protective factors for the entire sample was established. Next, chi-square analyses were conducted to explore group differences in protective item frequency across categories of recidivistic outcome and SAVRY Summary Risk Ratings. The sample was also divided into groups to represent differing levels of protective factors (Low = 0 Protective Factors; Moderate = 1-2 Protective Factors; High = 3-6 Protective Factors). High and Low risk levels were determined by isolating the top and bottom SAVRY total score quartiles, respectively (the Summary Risk Rating was not used in this specific analysis as it is not independent from SAVRY Protective Factor scores). Two series of Kaplan–Meier survival analyses were then conducted across protective factor groups. The first analysis compared time with re-offense across protective factor groups for the entire cohort. The second analysis compared time with re-offense across protective factor groups for solely high-risk summary rating and then solely low-risk summary rating clients. Log-rank tests were used to establish protective factor group differences for the survival analyses. To ascertain which protective factors account for the most variance in re-offense, a forward stepwise logistic regression analysis combining all six protective factors was conducted. Last, receiver operator characteristic (ROC) analysis was used to determine the capacity of each protective item to distinguish between general and violent non-re-offenders. ROC analysis provides an area under the curve (AUC) value by plotting true positives against false positives. AUC scores of .75 and above are considered large effect sizes in the violence risk literature (Dolan & Doyle, 2000).

Results

Item Prevalence

The number and percentage of young people in the cohort presenting with the individual SAVRY protective items are displayed in Table 2. The absence of each protective item was the most common response for the cohort. The most common protective item was “pro-social involvement.” Having no protective factors at all was the most common result among the sample (n = 59, 33.7%) followed by having one protective factor (n = 37, 21.1%).

Table 2:

Proportion of Participants Presenting With Each SAVRY Protective Item (N = 175)

Protective Factors	Present—n (%)	Absent—n (%)
Pro-social involvement	70 (40.0)	105 (60.0)
Strong social support	42 (24.0)	133 (76.0)
Strong attachments and bonds	65 (37.1)	110 (62.9)
Positive attitude toward intervention and authority	51 (29.1)	124 (70.9)
Strong commitment to school	32 (18.3)	143 (81.7)
Resilient personality traits	58 (33.1)	117 (66.9)

Note. SAVRY = Structured Assessment of Violence Risk in Youth.

Group Differences

Table 3 displays the mean number of protective factors for each SAVRY Summary Risk Rating category and for the overall cohort. The mean number of protective factors for the cohort was 1.82 (SD = 1.90). An analysis of variance discovered significant group differences across risk rating category, F(2, 175) = 56.47, p < .001. Post hoc analyses indicated that low-risk participants had a significantly higher mean number of protective items compared with both moderate- and high-risk participants (p < .001). Similarly, moderate-risk participants had a significantly higher mean number of protective items compared with high-risk participants (p < .001). The mean number of risk factors was also found to significantly differ across risk categories (p < .001).

Table 3:

Differences in Mean Protective Factor Domain Scores and Total Number of Risk Factors Across SAVRY Risk Rating

Risk Rating	N	Protective Factor Domain M (SD)	95% CI	Number of Risk Factors M (SD)	95% CI
Low	28	3.86 (1.94)_a	[3.11, 4.61]	9.36 (2.60)_a	[8.35, 10.37]
Moderate	63	2.43 (1.80)_b	[1.97, 2.88]	14.60 (3.03)_b	[13.84, 15.37]
High	84	0.68 (0.96)_c	[0.47, 0.89]	19.51 (2.38)_c	[19.00, 20.03]
Total	175	1.82 (1.90)	[1.53, 2.10]	16.12 (4.56)	[15.44, 16.80]

Note. Means with different subscripts differ significantly between columns. SAVRY = Structured Assessment of Violence Risk in Youth; CI = confidence interval.

Recidivism

Survival Analysis

The follow-up sample was reduced (n = 139) from 175 as 17 participants had not been released from custody and 21 participants did not consent to researchers accessing their criminal histories. Two participants were neither released from custody during the follow-up period nor consented to researchers accessing their criminal histories. From the revised sample, approximately 75% of the sample (n = 104) generally re-offended. Of those who generally re-offended, 82 (79%) were charged with a new violent offense. Cumulative protective item frequencies were assembled into three groups denoting level of protective strength, with similar participant distribution across groupings: Low (n = 59, 33.7%), Moderate (n = 62, 35.4%), and High (n = 54, 30.9%). Figures 1 and 2 exhibit the time to event by protective factor group. Significantly different survival times across protective factor groups were found for general recidivism, χ²(2) = 9.67, p = .008. Post hoc analyses found that participants in the low protective factor group had a lower mean survival time compared with participants in the high protective factor group, χ²(1) = 8.82, p = .003. Similar main effect results were obtained for violent re-offense, χ²(2) = 8.71, p = .013. Again, the low protective factor group differed significantly from the high protective factor group for time to violent re-offense, χ²(1) = 8.32, p = .004.

Figure 1:

Survival Curve for General Offending by Level of Protection

Figure 2:

Survival Curve for Violent Offending by Level of Protection

Next, differences in survival time of protection were analyzed with consideration to risk level. For the sub-group of 40 high-risk participants (SAVRY Total Score ≥ 33), no mean survival time differences were found across protective factor groups, General Recidivism: χ²(2) = 1.42, p = .492; Violent Recidivism: χ²(2) = 1.68, p = .431.

For the sub-group of 29 low-risk participants (SAVRY Total Score ≤ 19), significant differences were found across different levels of protection for general recidivism, χ²(2) = 9.63, p = .008. Both the low protective factor group, χ²(1) = 7.62, p = .006, and moderate protective factor group, χ²(1) = 3.98, p = .046, had significantly lower mean survival times compared with the high protective factor group. No low-risk group by protective group differences were found for time to violent re-offense, χ²(2) = 1.27, p = .530.

Outcome Group Differences

The proportion of general offending recidivist and non-recidivists presenting with individual protective items is displayed in Table 4. Chi-square analyses demonstrate that non-recidivists were significantly more likely to exhibit five of the six protective items compared with recidivists: Pro-social Involvement, Strong Social Support, Positive Attitude toward Intervention and Authority, Strong Commitment to School, and Resilient Personality traits. When exploring group differences across general and violent recidivistic outcome, violent recidivists were found to be significantly more likely to have a strong commitment to school compared with general recidivists, χ²(1) = 4.26, p < .05, Φ = .20.

Table 4:

Differences in Recidivistic Outcome Across Protective Items (n = 139)

Protective Factors	GR (%)	NR (%)	χ²	Φ
Pro-social involvement	26.0	71.4	23.12^***	.41
Strong social support	13.5	37.1	9.38^**	.26
Strong attachments and bonds	30.8	48.6	3.64	.16
Positive attitude toward intervention and authority	18.3	51.4	14.74^***	.33
Strong commitment to school	6.7	37.1	19.66^***	.38
Resilient personality traits	25.0	48.6	6.81^**	.22

Note. GR = General Recidivism; NR = No Recidivism.

p < .05. **p < .01. ***p < .001.

Prediction

The SAVRY protective items were entered into a stepwise binary logistic model to determine item contribution to explaining variance in re-offense. The models accounting for the most variance for both general and violence recidivism are presented in Table 5. Pro-social Involvement and Positive Attitude toward Intervention and Authority comprised the strongest predictive model for general recidivism. The two-item model for violent recidivism comprised protective items Pro-social involvement and Strong Commitment to School.

Table 5:

Logistic Regression Model—Contribution of Protective Items in Prediction Non Re-offense

	B	SE	Wald	p	Exp(B)
General
Pro-social involvement	−1.67	.46	13.74	.000	.19
Positive attitude toward intervention and authority	−1.08	.46	5.52	.019	.34
Violent
Pro-social involvement	−0.91	.43	4.55	.033	.40
Strong commitment to school	−1.39	.66	4.39	.036	.25

ROC analyses were carried out to determine which protective items were able to discriminate between re-offenders and non re-offenders (see Table 6). Predictive results were generally stronger for general recidivism. Low to Moderate AUC indices were commonly obtained across the individual items. The item Pro-social Involvement produced the strongest predictive validity for both general and violent recidivism. Items Positive Attitude toward Intervention and Authority and Strong Commitment to School also demonstrated significant moderate predictive validity for general recidivism. Only Strong Attachments and Bonds and Resilient Personality Traits failed to reach significance for specific types of recidivism.

Table 6:

AUC Values for Protective Items for Non Re-Offense (n = 139)

Protective Factors	General Recidivism		Violent Recidivism
Protective Factors	AUC (SE)	95% CI	AUC (SE)	95% CI
Pro-social involvement	0.73 (.05)***	[.63, .83]	0.66 (.05)**	[.57, .75]
Strong social support	0.62 (.06)*	[.50, .73]	0.62 (.05)*	[.52, .72]
Strong attachments and bonds	0.59 (.06)	[.48, .70]	0.60 (.05)*	[.51, .70]
Positive attitude toward intervention and authority	0.67 (.06)**	[.56, .78]	0.60 (.05)*	[.50, .70]
Strong commitment to school	0.65 (.06)**	[.54, .77]	0.62 (.05)*	[.52, .71]
Resilient personality traits	0.62 (.06)*	[.51, .73]	0.57 (.05)	[.47, .66]

Note. AUC = area under the curve; CI = confidence interval.

p < .05. **p < .01. ***p < .001.

Discussion

This study investigated the presence of protective factors and their association with recidivism in a judicially determined high-risk sample of young people in detention. The common absence of protective items was a notable finding and typical of young offender samples (Chapman, Desai, Falzer, & Borum, 2006; Gammelgard et al., 2012; Rennie & Dolan, 2010). Findings indicated a relationship between level of client risk and number of protective factors. As expected, the higher the assessed risk of the clients, the less likely they were to present with protective influences. Relatedly, Cuervo & Villanueva (2015) found re-offending juveniles to have fewer protective factors compared with non re-offenders. Although the presence of protective factors during rating would influence the final assessed risk, suggesting this association may be spurious, there was a corresponding trend with the number of total risk factors present. This leaves the possibility that clients who present with numerous areas of risk (risk factors) may also struggle to develop more protective domains in their life through pro-social initiatives, if the areas of risk are not addressed at least concurrently. The mean number of protective factors for low-risk clients in this sample was similar to mean scores of community-based young offenders in other research (Hilterman, Nicholls, & van Nieuwenhuizen, 2013), perhaps underscoring their comparatively lower level of overall severity within the sample.

The degree of protective influence for participants in the study was determined and examined in relation to time at risk. The first key finding showed that participants who have three or more protective factors take longer to re-offend on average. Conversely, participants who have no protective factors re-offend significantly earlier. The pattern was similar for violent recidivists. These results are consistent with previous studies investigating the association between protective factors and re-arrest rates (Lodewijks et al., 2010; Pearl, Ashcraft, & Geis, 2009; Rennie & Dolan, 2010; Turner & Fain, 2006), and underscore the importance of positive stimuli in a young person’s life, which, if sustained, may deviate them away from deleterious influences or at least delay criminal behavior (i.e., reduced intensity).

The findings of the survival analyses are consistent with previous SAVRY Protective Factor research that has shown that protective factors can lessen the likelihood of adolescent recidivism or other problem behaviors (Losel & Farrington, 2012; Rennie & Dolan, 2010; Lodewijks et al., 2010; Shepherd et al., 2014). However, the current findings extend this knowledge to consider how the presence and intensity of protective factors effect criminal re-engagement for youth across different risk levels. Importantly, the survival analyses split by quartile-determined high and low risk indicate that the association that protective factors have with time to re-offense is only observed for participants assessed as being of low risk of re-offending. For lower risk young offenders, the presence of several protective items significantly reduced time to recidivism. In contrast, the number of protective factors was not associated with increased survival time for young offenders assessed as a high risk for re-offense. In other words, patterns of re-offense for high-risk offenders are unaffected by how many protective items they possess. This result differs to Lodewijks et al. (2010) who found that protective items demonstrated a buffering effect for both low- and high-risk adolescent offenders. Similarly Ullrich and Coid (2011) discovered specific protective items to affect violent outcome irrespective of risk level for adult male offenders. However, Bushway et al. (2013) identified that there are a limited number of factors that protect against ongoing violent behavior (i.e., reduce recidivism) among young adults whose offending is entrenched.

High-risk young people often have numerous and co-existing psychological and environmental risk factors, which could either prevent or subdue the initiation of mitigating influences. For such youth, multi-systemic treatment initiatives including cognitive behavioral therapy may be required to address criminogenic needs before the fostering factors that increase protection and resilience. This does not mean that cultivating and sustaining pro-social experiences and involvement for high-risk youth should be curtailed. Rather, it may be that protective influences are blunted for high-risk youth until more immediate criminogenic needs are addressed first or at least concurrently.

The potential differential influence of protective factors at differing risk levels has important implications for young offender management and treatment. Most youth justice jurisdictions adopt a greater well-being and strengths-based approach (highlighted in the “Beijing Rules”) to offender treatment compared with their respective adult correctional systems (see Luebbers & Ogloff, 2011; United Nations [UN] General Assembly, 1985). A contrast that arguably parallels the debate about the Good Lives Model (GLM; Ward & Brown, 2004) and RNR (Andrews & Bonta, 2010) approaches of offender treatment and management. The current findings would support an approach with young offenders that both considers the presence of risk and protective factors, and targets treatments accordingly, with a greater focus on protective and well-being factors for those assessed as low risk and an emphasis on addressing criminogenic factors for those assessed as high risk. In this regard, risk assessment may entail a two-step process. First, a client’s risk and protective factors should initially be identified. Second, the weight allocated to protective items would then depend on the client’s propensity for future violence based on their presenting risk factors. If a client is deemed to be of a higher risk, then less clinical weight should be afforded to protective factors. If a client is a low to moderate risk, then protective items should be taken into account.

Another key finding was the greater presence of protective items for non-recidivists compared with recidivists, and reflects prior research with male adolescent offenders (Lodewijks et al., 2010). Participants who did not re-offend were more likely to be engaging in pro-social involvement, harbor positive attitudes toward intervention and authority, and display a strong commitment to schooling at the time of assessment. A plethora of research has outlined the importance of educational engagement and attainment during adolescence and the increased likelihood of negative outcomes if schooling is prematurely terminated (e.g., Nagin, Pagani, Tremblay, & Vitaro, 2003; Piquero, 2008). Indeed, engagement in school is one of the primary sources of pro-social involvement during adolescence, both peer and pro-social adult, and is fertile ground for the development of positive relationships with authority figures and a self-belief in one’s capacity for change and growth. Perhaps unexpectedly, violent re-offenders were more likely to possess an attachment to school compared with general re-offenders. This finding may necessitate further inquiry; however, it is possible that the violent re-offender cohort comprised more youth who possess short yet violent criminal histories as opposed to chronic and persistent offenders who exhibit an extensive and diverse range of criminal activity, which affects their ability to engage regularly at school.

Existing studies have primarily focused on global measures of protective factors. For example, the SAVRY protective domain has demonstrated significant negative relationships with future recidivism (Shepherd, Luebbers, Ogloff, Fullam, & Dolan, 2014), in addition to adding incremental validity to the SAVRY risk total (Dolan & Rennie, 2008), and total scores on other youth risk instruments (Shepherd, Luebbers, & Ogloff, 2014). To address gaps in the literature, this study conducted an analytic breakdown of the individual SAVRY protective factors. It was identified that pro-social involvement demonstrated the strongest contribution to both forms of re-offense. This is the antithesis of one of the strongest predictors of re-offending—negative peer associations. Maintaining pro-social peer relations has been previously associated with criminal desistance (Hoge, Andrews, & Leschied, 1996) and disengagement from gang involvement (Sweeten, Pyrooz, & Piquero, 2013). Although associating with a deviant peer group is a well-established risk marker for delinquency, desistance from criminal behavior may require more than simply a passive withdrawal from deviant peers but rather an active engagement with pro-social peers.

Other strong predictors of non-recidivism included having a positive attitude toward intervention and authority and a strong commitment to school. A positive attitude toward remediation indicates a willingness to comply with risk reduction initiatives and address criminogenic needs. Attitudinal factors are a key element in the desistance of crime (Hoge et al., 1996). Similarly, Spice, Viljoen, Gretton, & Roesch (2010) found that the SAVRY protective domain score strongly correlated with treatment amenability. As anticipated, young offenders who embrace schooling are more likely to desist from criminal activity. Educational achievement, particularly motivation to succeed at school, is a prominent buffering variable for at-risk adolescents. ROC analyses confirmed the results of the logistic regression underscoring the influence of key protective items. Again, pro-social involvement, positive attitude toward intervention and authority, and strong commitment to school were salient predictors of recidivism. Findings suggest that young offenders who possess these particular positive attributes are much less likely to continue re-offending. It may be that participation in educational pursuits and organized pro-social activities imparts a sense of structure and edifying cooperation, facilitating a sense of purpose and of self-worth.

Moreover, it is likely that the presence of one SAVRY protective item may eventually lead to the attainment of further protective items. For example, involvement in pro-social activities may lead to the development of strong social support and caring relationships with positive role models. These supportive relationships, although not identified as significant predictors of non-recidivism in the current study, are factors that are likely relevant to general well-being that may have distal influences on risk of offending. The findings of this study support the notion that protective factors/items play an important role in mitigating future re-offense. However, for high-risk individuals, these effects may be limited as evidenced in this analysis. Addressing criminogenic needs should be the primary focus for young offenders, particularly those with violent and/or chronic criminal histories.

Results should be considered in light of several limitations. Due to the severe nature of the cohort, findings may not generalize to community-based young offender samples. However, outcomes should extend to samples of high-risk youth. Second, the study did not determine ethnic differences in the prevalence of protective factors. Although this was not the direct focus of the study, we believe this inquiry is worthy of future investigation. Third, the analysis of risk level by level of protection engendered small sample sizes. For the high-risk group, there may have been insufficient variability to observe an effect. Nonetheless, differences were still found at lower levels of risk despite the limited size of the sub-samples. The severe nature of the overall sample may have rendered the delineation of distinct risk levels unusually challenging. However, statistical differences were still observed despite the extremity of the clientele. Stronger differences across risk levels may have been identified in a cohort with greater risk variability. We encourage further research with larger samples utilizing factor analysis statistical procedures. Last, the sample included a small number of young adult offenders aged 18 to 20 years. Although clients aged 18 years and above are beyond the recommended age range of the SAVRY, their inclusion was grounded on their status as “dual track” young offenders who are processed by, and under the supervision of, the Victorian youth justice system.

Footnotes

This research was supported in part by a grant from the Australian Research Council (DP1095697).

Stephane M. Shepherd is a research fellow and lecturer at the Centre for Forensic Behavioural Science, Swinburne University of Technology. His research interests include adolescent violence risk assessment with a focus on cross-cultural validity and risk factors for violence across ethnicity and gender.

Stefan Luebbers is a lecturer at the Centre for Forensic Behavioural Science, Swinburne University of Technology and a clinical forensic psychologist. His research interests include youth mental health and violence, neurodevelopment and high-risk behaviors, assessment and intervention with young offenders, and adolescent service delivery.

James R. P. Ogloff is the director of the Centre for Forensic Behavioural Science, Swinburne University of Technology and the director of Psychological Services at the Victorian Institute of Forensic Mental Health (Forensicare). He has more than 30 years of expertise in forensic psychology, forensic mental health, mental health and law, and the assessment and management of offenders.

References

Andrews

D. A.

Bonta

(Eds.). (2010). The psychology of criminal conduct (5th ed.). New Providence, NJ: Matthew Bender.

Andrews

D. A.

Bonta

Wormith

J. S.

(2009). The Level of Service Inventory/Risk-Need-Responsivity (LSI/RNI). Toronto, Ontario, Canada: Multi-Health Systems.

Andrews

D. A.

Bonta

J. L.

Wormith

J. S.

(2004). LSI/CMI: Level of Service/Case Management Inventory: An offender assessment system user’s manual. North Tonawanda, NY: Multi-Health Systems.

Australian Bureau of Statistics. (2015). 4519.0 Recorded Crime—Offenders, 2013-2014. Retrieved from http://www.abs.gov.au/ausstats/abs@.nsf/Lookup/by%20Subject/4519.0~2013-14~Main%20Features~Youth%20Offenders~5

Australian Institute of Criminology. (2014). Australian crime: Facts & figures: 2013. Canberra: Australian Institute of Criminology.

Australian Institute of Health and Welfare (AIHW). (2013). Youth justice in Australia 2011–12: An overview (Bulletin no. 115). Canberra, Australia: Author.

Australian Institute of Health and Welfare (AIHW). (2014). Youth detention population in Australia 2014 (Juvenile Justice series no.16. Cat. no. JUV 53). Canberra: Author.

Australian Research Alliance for Children & Youth (ARACY). (2009, October). Violent and antisocial behaviours among young adolescents in Australian communities: An analysis of risk and protective factors. Retrieved from http://www.aracy.org.au/publications-resources/command/download_file/id/161/filename/Violent_and_antisocial_behaviours_among_young_adolescents_in_Australian_communities_-_An_analysis_of_risk_and_protective_factors.pdf

Bernat

D. H.

Oakes

J. M.

Pettingell

S. L.

Resnick

(2012). Risk and direct protective factors for youth violence: Results from the National Longitudinal Study of Adolescent Health. American Journal of Preventive Medicine, 43(2, Suppl. 1), S57-S66. doi:10.1016/j.amepre.2012.04.023

10.

Borum

(2000). Assessing violence risk among youth. Journal of Clinical Psychology, 56, 1263-1288.

11.

Borum

Bartel

Forth

(2006). Manual for the Structured Assessment of Violence Risk in Youth. Tampa: University of South Florida.

12.

Borum

Verhaagen

(2006). Assessing and managing violence risk in juveniles. New York, NY: The Guilford Press.

13.

Broidy

L. M.

Nagin

D. S.

Tremblay

R. E.

Bates

J. E.

Brame

Fergusson

. . . Vitaro

(2003). Developmental trajectories of childhood disruptive behaviors and adolescent delinquency: A six-site, cross-national study. Developmental Psychology, 39, 222-245. doi:10.1037/0012-1649.39.2.222

14.

Bushway

S. D.

Krohn

M. D.

Lizotte

A. J.

Phillips

M. D.

Schmidt

N. M.

(2013). Are risky youth less protectable as they age? The dynamics of protection during adolescent and young adulthood. Justice Quarterly, 30, 84-116. doi:10.1080/07418825.2011.592507

15.

Chapman

J. F.

Desai

R. A.

Falzer

P. R.

Borum

(2006). Violence risk and race in a sample of youth in juvenile detention: The potential to reduce disproportionate minority confinement. Youth Violence and Juvenile Justice, 4, 170-184. doi:10.1177/1541204006286316

16.

Cuervo

Villanueva

(2015). Analysis of risk and protective factors for recidivism in Spanish youth offenders. International Journal of Offender Therapy and Comparative Criminology, 59, 1149-1165. doi:10.1177/0306624X14557917

17.

DeMatteo

Heilbrun

Marczyk

(2005). Psychopathy, risk of violence, and protective factors in a noninstitutionalized and noncriminal sample. International Journal of Forensic Mental Health, 4, 147-157. doi:10/1080/14999013.2005.10471220

18.

de Vogel

de Ruiter

Bouman

de Vries Robbe

. (2009). SAPROF. Guidelines for the assessment of protective factors for violence risk [English version]. Utrecht, The Netherlands: Forum Educatief.

19.

de Vogel

de Vries Robbe

de Ruiter

Bouman

Y. H. A

. (2011). Assessing protective factors in forensic psychiatric practice: Introducing the SAPROF. International Journal of Forensic Mental Health, 10, 171-177. doi:10.1080/14999013.2011.600230

20.

de Vries Robbe

de Vogel

Douglas

K. S

. (2013). Risk factors and protective factors: A two-sided dynamic approach to violence risk assessment. The Journal of Forensic Psychiatry & Psychology, 24, 440-457. doi:10.1080/14789949.2013.818162

21.

de Vries Robbe

Geers

Stapel

Hilterman

de Vogel

. (2015). SAPROF Youth Version: Guidelines for the assessment of protective factors for violence risk in juveniles. Utrecht, The Netherlands: Van der Hoeven Kliniek.

22.

Dolan

M. C.

Doyle

(2000). Violence risk prediction. The British Journal of Psychiatry, 177, 303-311. doi:10.1192/bjp.177.4.303

23.

Dolan

M. C.

Rennie

C. E.

(2008). The Structured Assessment of Violence Risk in Youth as a predictor of recidivism in a United Kingdom cohort of adolescent offenders with conduct disorder. Psychological Assessment, 20, 35-46. doi:10.1037/1040-3590.20.1.35

24.

Farrington

D. P.

West

D. J.

(1993). Criminal, penal and life histories of chronic offenders: Risk and protective factors and early identification. Criminal Behaviour and Mental Health, 3, 492-532.

25.

Gammelgard

Weizmann-Henelius

Koivisto

A.-M.

Eronen

Kaltiala-Heino

(2012). Gender differences in violence risk profiles. The Journal of Forensic Psychiatry & Psychology, 23, 76-94. doi:10.1080/14789949.2011.639898

26.

Hall

J. E.

Simon

T. R.

Mercy

J. A.

Loeber

Farrington

D. P.

Lee

R. D.

(2012). Centers for Disease Control and Prevention’s expert panel on protective factors for youth violence perpetration. American Journal of Preventive Medicine, 43(2, Suppl. 1), S1-S7. doi:10.1016/j.amepre.2012.04.026

27.

Henry

D. B.

Tolan

P. H.

Gorman-Smith

Schoeny

M. E.

(2012). Risk and direct protective factors for youth violence. American Journal of Preventive Medicine, 43(2, Suppl. 1), S67-S75. doi:10.1016/j.amepre.2012.04.025

28.

Herrenkohl

T. I.

Lee

Hawkins

J. D.

(2012). Risk versus direct protective factors and youth violence. American Journal of Preventive Medicine, 43(2, Suppl. 1), S41-S56. doi:10.1016/j.amepre.2012.04.030

29.

Hilterman

E. L.

Nicholls

T. L.

van Nieuwenhuizen

(2013). Predictive validity of risk assessments in juvenile offenders: Comparing the SAVRY, PCL: YV, and YLS/CMI with unstructured clinical assessments. Assessment, 21, 324-339. doi:10.1177/1073191113498113

30.

Hoge

R. D.

Andrews

D. A.

(2006). Youth Level of Service/Case Management Inventory user’s manual. New York, NY: Multi Health Systems.

31.

Hoge

R. D.

Andrews

D. A.

Leschied

A. W.

(1996). An investigation of risk and protective factors in a sample of youthful offenders. Journal of Child Psychology and Psychiatry, 37, 419-424. doi:10.1111/j.1469-7610.1996.tb01422.x

32.

Jessor

Van Den Bos

Vanderryn

Costa

F. M.

Turbin

M. S.

(1995). Protective factors in adolescent problem behavior: Moderator effects and developmental change. Developmental Psychology, 31, 923-933.

33.

Laub

J. H.

Sampson

R. J.

(2001). Understanding desistance from crime. In Tonry

(Ed.), Crime and justice: A review of research (Vol. 28, pp. 1-76). Chicago, IL: The University of Chicago.

34.

Laub

J. H.

Sampson

R. J.

(2003). Shared beginnings, divergent lives: Delinquent boys to age 70. Cambridge, MA: Harvard University Press.

35.

Lipsey

M. W.

Derzon

J. H.

(1998). Predictors of violent or serious delinquency in adolescence and early adulthood: A synthesis of longitudinal research. In Loeber

Farrington

D. P.

(Eds.), Serious and violent juvenile offenders: Risk factors and successful interventions (pp. 85-105). Thousand Oaks, CA: SAGE.

36.

Lodewijks

H. P. B.

de Ruiter

Doreleijers

T. A. H.

(2010). The impact of protective factors in desistance from violent reoffending: A study in three samples of adolescent offenders. Journal of Interpersonal Violence, 25, 568-587. doi:10.1177/0886260509334403

37.

Loeber

Farrington

D. P.

Stouthamer-Loeber

Moffitt

T. E.

Caspi

(1998). The development of male offending: Key findings from the first decade of the Pittsburgh Youth Study. Studies on Crime & Crime Prevention, 7, 141-171.

38.

Losel

Farrington

(2012). Direct protective and buffering protective factors in the development of youth violence. American Journal of Preventive Medicine, 43(2, Suppl. 1), S8-S23. doi:10.1016/j.amepre.2012.04.029

39.

Luebbers

Ogloff

J. R. P.

(2011). “Boys will be boys” or budding criminal: Differentiating youthful offenders. Psychiatry, Psychology & Law, 18, 557-572. doi:10.1080/13218719.2010.501779

40.

Moffitt

T. E.

Caspi

Dickson

Silva

Stanton

(1996). Childhood-onset versus adolescent-onset antisocial conduct problems in males: Natural history from ages 3 to 18 years. Development and Psychopathology, 8, 399-424.

41.

Nagin

D. S.

Pagani

Tremblay

R. E.

Vitaro

(2003). Life course turning points: A case study of the effect of school failure on interpersonal violence. Development and Psychopathology, 15, 343-361.

42.

Nagin

Tremblay

R. E.

(1999). Trajectories of boys’ physical aggression, opposition, and hyperactivity on the path to physically violent and nonviolent juvenile delinquency. Child Development, 70, 1181-1196.

43.

Pardini

D. A.

Loeber

Farrington

D. P.

Stouthamer-Loeber

(2012). Identifying direct protective factors for nonviolence. American Journal of Preventive Medicine, 43(2, Suppl. 1), S28-S40. doi:10.1016/j.amepre.2012.04.024

44.

Pearl

Ashcraft

R. G. P.

Geis

K. A.

(2009). Predicting juvenile recidivism using the San Diego Regional Resiliency Check-Up. Federal Probation, 73(3), 46-49.

45.

Piquero

A. R.

(2008). Taking stock of developmental trajectories of criminal activity over the life course. In Liberman

A. M.

(Ed.), The long view of crime: A synthesis of longitudinal research (pp. 23-78). New York, NY: Springer.

46.

Rennie

Dolan

(2010). Predictive validity of the Youth Level of Service/Case Management Inventory in custody sample in England. The Journal of Forensic Psychiatry & Psychology, 21, 407-425. doi:10.1080/14789940903452311

47.

Resnick

M. D.

(2000). Protective factors, resiliency and health youth development. Adolescent Medicine, 11, 157-165.

48.

Resnick

M. D.

Ireland

Borowsky

(2004). Youth violence perpetration: What protects? What predicts? Findings from the National Longitudinal Study of Adolescent Health. Journal of Adolescent Health, 35, 424.e1-424.e10. doi:10.1016/j.jadohealth.2004.01.011

49.

Rogers

(2000). The uncritical acceptance of risk assessment in forensic practice. Law and Human Behavior, 24(5), 595-605. doi:10.1023/A:1005575113507

50.

Salekin

R. T.

Lochman

J. E.

(2008). Child and adolescent psychopathy: The search for protective factors. Criminal Justice and Behavior, 35, 159-172. doi:10.1177/0093854807311330

51.

Shepherd

S. M.

Luebbers

Ferguson

Ogloff

J. R. P.

Dolan

(2014). The utility of the SAVRY across ethnicity in Australian young offenders. Psychology, Public Policy, & Law, 20, 31-45. doi:10.1037/a0033972

52.

Shepherd

S. M.

Luebbers

Ogloff

J. R. P.

(2014). Are youth violence risk Instruments interchangeable? Evaluating instrument convergence in a sample of incarcerated adolescent offenders. Journal of Forensic Psychology Practice, 14, 317-341. doi:10.1080/15228932.2014.954871

53.

Shepherd

S. M.

Luebbers

Ogloff

J. R. P.

Fullam

Dolan

(2014). The predictive validity of risk assessment approaches for young Australian offenders. Psychiatry, Psychology & Law, 21, 801-817. doi:10.1080/13218719.2014.904262

54.

Shlafer

R. J.

McMorris

B. J.

Sieving

R. E.

Gower

A. L.

(2013). The impact of family and peer protective factors on girls’ violence perpetration and victimization. Journal of Adolescent Health, 52, 365-371. doi:10.1016/j.jadoheatlh.2012.07.015

55.

Spice

Viljoen

J. L.

Gretton

H. M.

Roesch

(2010). Psychological assessment for adult sentencing of juvenile offenders: An evaluation of the RSTI and the SAVRY. International Journal of Forensic Mental Health, 9, 124-137. doi:10.1080/14999013.2010.501846

56.

Stouthamer-Loeber

Loeber

Wei

Farrington

D. P.

Wikstrom

P. H.

(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

57.

Stouthamer-Loeber

Wei

Loeber

Masten

A. S.

(2004). Desistance from persistent serious delinquency in the transition to adulthood. Development and Psychopathology, 16, 897-918. doi:10.1017/S0954579404040064

58.

Sweeten

Pyrooz

D. C.

Piquero

A. R.

(2013). Disengaging from gangs and desistance from crime. Justice Quarterly, 30, 469-500. doi:10.1080/07418825.2012.723033

59.

Turner

Fain

(2006). Validation of the risk and resiliency assessment tool for juveniles in the Los Angeles county probation system. Federal Probation, 70(2), 49-55.

60.

Ullrich

Coid

(2011). Protective factors for violence among released prisoners—Effects over time and interactions with static risk. Journal of Consulting Clinical Psychology, 79, 381-390. doi:10.1037/a0023613

61.

United Nations General Assembly. (1985, November 29). United Nations standard minimum rules for the administration of juvenile justice (“The Beijing Rules”): Resolution/adopted by the General Assembly (A/RES/40/33). Retrieved from http://www.refworld.org/docid/3b00f2203c.html

62.

Walker

Bowen

Brown

(2013). Psychological and criminological factors associated with desistance from violence: A review of the literature. Aggression and Violent Behavior, 18, 286-299. doi:10.1016/j.avb.2012.11.021

63.

Ward

Brown

(2004). The Good Lives Model and conceptual issues in offender rehabilitation. Psychology, Crime & Law, 10, 243-257. doi:10.1080/10683160410001662744

64.

Ward

Stewart

C. A.

(2003). The treatment of sex offenders: Risk management and good lives. Professional Psychology: Research and Practice, 34, 353-360. doi:10.1037/0735-7028.34.3.353

65.

Webster

Martin

Brink

Nicholls

Middleton

(2004). The short term assessment of risk and treatability (START). Vancouver, British Columbia, Canada: Forensic Psychiatric Services Commission.

66.

Wooditch

Tang

L. L.

Taxman

F. S.

(2014). Which criminogenic need changes are most important in promoting desistance from crime and substance use? Criminal Justice and Behavior, 41, 276-299. doi:10.1177/0093854813503543