1. The Risks and Needs of Juvenile Offenders with an Intellectual Disability
Professor Dianna Kenny, University of Sydney, Australia
Matt Frize, Statewide Behaviour Intervention Service, DADHC, Australia
Dr Chris Lennings, Sydney University, Australia

2. Prevalence
Is intellectual disability (ID) causal in offending?
Previous studies have highlighted many methodological issues:
Different measures used to define and code offending behaviour and ID (Keith & McCray, 2002; Kenny & Press, 2006; Kvarfordt, Purcell, & Shannon, 2005; McBrien, 2003)
Heterogeneity of offender and offending categories (Lindsay & Taylor, 2005);
Prevalence of ID at different points in the criminal justice system (e.g. court versus prison) (Hayes, 2004; Holland, Clare, & Mukhopadhyay, 2002);
Different sentencing options in different jurisdictions (McBrien, Hodgetts, & Gregory, 2003)
Intellectual disability (ID) has long been suggested to be a predictor of offending behaviour in young people (e.g. Herrnstein & Murray, 1994; Steinbach, 1934).
Methodological limitations in studies have precluded the emergence of clear trends or predictors of offending in people with an ID (Lindsay & Taylor, 2005). In particular, studies have been complicated by:
-different measures used to define and code offending behaviour and ID (Keith & McCray, 2002; Kenny & Press, 2006; Kvarfordt, Purcell, & Shannon, 2005; McBrien, 2003) and the resulting heterogeneity of offender and offending categories (Lindsay & Taylor, 2005);
-different prevalence rates of those with an ID at different points of the criminal justice system (e.g. court versus prison) (Hayes, 2004; Holland, Clare, & Mukhopadhyay, 2002);
-having different sentencing options available to people with an ID who offend in different jurisdictions (Lindsay, Steele, Smith, Quinn, & Allan, 2006; McBrien, Hodgetts, & Gregory, 2003)

3. Prevalence
Simpson & Hogg (2001) “…there is no convincing evidence that the prevalence of offending among people with an intellectual disability is higher than for the wider population” …. “…offending among those with an IQ less than 50 is rare.” (p. 394)
Lindsay & Taylor (2005) “[i]t is not clear … whether people with developmental disability commit more or less crime than those without developmental disability, or whether the type and frequency of crimes committed by offenders with developmental disability differ from those committed by the general populations of offenders” (p. 201).
Recent reviews (e.g. Lindsay & Taylor, 2005; Simpson & Hogg, 2001) have concluded that:

4. Prevalence Hayes (1997) Allerton, Kenny, Champion, and Butler (2003):
NSW
Hayes (1997)
9% of those in the juvenile justice system had an ID
Allerton, Kenny, Champion, and Butler (2003):
17% of juvenile offenders had an IQ below 70 in juvenile correctional centres in NSW
13% if WIAT-2 included as the measure of adaptive functioning
Chitsabesan, MacDonald, & Kenning (2008)
20% of juvenile offenders in the UK had an IQ below 70 when assessed using the Wechsler Abbreviated Scale of Intelligence (WASI)
However, there is strong evidence of an over-representation of juveniles in custody with an ID across jurisdictions. In the UK, a recent study by Chitsabesan, MacDonald, & Kenning, (2008) found that 20% of juvenile offenders in the UK had an IQ below 70 when assessed using the Wechsler Abbreviated Scale of Intelligence (WASI).
Hayes (1997), reporting on an Australian sample, estimated that 9% of those in the juvenile justice system had an ID.
Allerton, Kenny, Butler & Champion (2003) found 17% of juveniles detained in juvenile correctional centres in NSW had an IQ below 70. When academic functioning was used as a proxy measure for adaptive functioning, a conservative estimate of ID was found in 13% of those surveyed.
Culture fair testing methods (using performance IQ for those from culturally and linguistically diverse backgrounds) estimated 10% of incarcerated Aboriginal and Torres Straight Islanders and those from non-English speaking background had an ID.

5. Predictors of Recidivism in Juveniles by Predictive Strength (Cottle et al., 2001, p. 385)
Although several meta-analyses have shown that IQ, specifically ID, is predictive of offending in general (Cullen, Gendreau, Jarjoura, & Wright, (1997), and violent offending in particular (Lipsey & Derzon, 1998), the relationships between ID and offending are weak; for example, for 6-11 year olds, r = .12; for year olds, r = .11). In the most recent meta-analysis (Cottle, Lee, & Heilbrun, 2001), IQ was found to be a moderate predictor of offending for juveniles.
Although ID appears to not be a strong independent predictor of offending, Juveniles with an ID may still have a disproportionately higher likelihood of offending compared with juveniles who do not have an ID.
This may be because ID:
Directly influences offending behaviour
May be a risk factor for other factors that are more proximal risks, (as suggested in the Psychology of Criminal Conduct).
This paper explores whether those with an ID demonstrate a similar profile of risk factors to offenders without an ID.

6. Psychology of Criminal Conduct
Andrews & Bonta (1998; 2001; 2003; 2006)
Issue with previous theories were focus on either biology, society or psychology
Social learning and social cognition theory (Andrews & Bonta, 2006).
Behaviour is learned through complex interrelated biopsychosocial factors (Bonta, 2002).
Many routes to offending, with higher risk of offending occurring as the number of proximal factors increase.

7. Risks Needs Responsivity (RNR) Principles
Intensity of treatment matched to the level of risk assigned to the person
Needs
Programs directed towards changeable factors (dynamic variables or criminogenic needs)
Factors addressed are those that most influence risk of offending (proximal to the offence)
Responsivity
the treatment modes addressed by criminogenic needs services matched to the learning styles and abilities of the offender
The Psychology of Criminal Conduct model proposes three principles of effective offender assessment and treatment: the principles of Risk, Needs and Responsivity (RNR). The ‘risk principle’ states that the degree of treatment a person receives should be matched to their level of assessed risk. The ‘needs principle’ states that programs should be directed towards those criminogenic needs associated with recidivism. The ‘responsivity principle’ refers to the need for the styles and modes of service that address criminogenic needs to be matched to the learning styles and abilities of the offender (Andrews, Bonta, & Hoge, 1990; Andrews & Bonta, 2006)

8. RNR Based Tools Level of Supervision Inventory (LSI: Andrews, 1982)
Level of Service Inventory-Revised (LSI-R: Andrews & Bonta, 2001)
Level of Service / Case Management Inventory (LS/CMI: Andrews, Bonta, & Wormith, 2004)
the Youth Level of Service / Case Management Inventory (YLS/CMI: Hoge & Andrews, 2002)
Youth Level of Service / Case management Inventory: Australian Adaptation (YLS/CMI:AA: Hoge & Andrews, 1995)
A number of assessment tools have been developed that aim to put the principles of RNR into practise for the individual offender (Andrews & Bonta, 2001; Andrews, Bonta, & Wormith, 2004; Hoge & Andrews, 1994; 1996; 2002). RNR based assessment tools have demonstrated moderate predictive validity for risk of reoffending in a number of jurisdictions and across distinct demographic and offender populations
Although the role of ID as a responsivity factor has been addressed in the literature (Andrews & Bonta, 2006), the relationship between ID and risk and (particularly) needs principles has largely been ignored. Accordingly, this paper addresses the relationship between risk and needs and their associations with ID through the use of the YLS/CMI: AA with young offenders with and without an ID to assess relative risk of reoffending. Assessment of profile of domain scores from the YLSI/CMI: AA has the potential to provide a greater understanding of the differential needs of offenders with and without an ID (the needs principle) thus leading to an examination of those factors that may address the different needs of those with and without an ID (the responsivity principle).

9. Prior and Current Offences (scoring protocol)
1.1 Age at first court order:
14 years or less (2)
15 to 16 years (1)
17 years or older (0)
1.2 Outcome of first court order:
Control order or supervised order
1.3 Type of offence involved in first court order:
Common assault, break and enter, or motor vehicle theft
1.4 More than one court order in last 12 months
1.5 Three or more prior offences
1.6 Two or more failures to comply
(eg with bail conditions, court attendance, good behaviour bond, community service or supervision order and escape)
1.7 Prior committal
Has the young person ever served a control order for an offence? Do not consider remand as committal, even if long term.
1.8 Three or more current offences related to the current court order?

10. Method
Data from the ‘NSW Young People on Community Order Health Survey ’
N =800 (42% of the sampling frame)
85% male, (proportion reflected total popn)
Mean age = 16 years 6 months (range: 12 to 21 years)
20% Indigenous
IQ score for 97.6% of participants
At least 1 YLS/CMI: AA was obtained for 94% of participants

11. Method Measures: ID defined as IQ < 70
IQ (WASI)
Criminogenic risk / need (YLS / CMI: AA)
Criminal justice involvement
ID defined as IQ < 70
‘Adaptive’ - AAMR definition (social deficit), which included criminal offending
Identify that other measures were used but not incorporated into this study

12. Youth Level of Service / Case Management Inventory: Australian Adaptation (YLS/CMI: AA)
Risk
Prior and current offences (8 items)
Education / Employment (7 items)
Family and living circumstances (7 items)
Peer relations (4 items)
Substance abuse (6 items)
Leisure / Recreation (3 items)
Personality / Behaviour (7 items)
Attitudes / Beliefs (5 items)
Strength
Individual level (1 item)
Family Level (1 item)
Social Level (1 item)
47 items
Based on the YLS/CMI
Thompson & Pope (2003; 2005) found acceptable validity / reliability (total scale Cronbach alpha = .91, n=290)
The YLS/CMI: AA is a 47 item instrument used to assess criminogenic risk factors assessed in eight subscales: offence history, family circumstances/parenting, education, peer relations, substance abuse, leisure/recreation, personality/ behaviour, and attitudes/ orientation and has three items that address individual strengths. The strength items were not used in this study because they are not used to calculate risk scores or inform criminogenic need, which was the focus of this study. The YLS/CMI: AA has been found as a valid instrument for assessing risk of reoffending of juvenile offenders (Thompson & Pope, 2003; Thompson & Pope, 2005). Thompson and Pope (2005) found internal consistency (cronbach alpha) high for the total score (0.91) and acceptable ( ) for 6 out of the eight sub-domain scores (N = 290) and Thompson and Pope (2003) found test-retest reliability of 84 youths over approximately 6 month was r = .75 for the total risk score. This indicates the YLS/CMI: AA also has acceptable reliability.

13. Youth Level of Service / Case management Inventory: Australian Adaptation (YLS/CMI:AA: Hoge & Andrews, 1995)
Level and Score (l=low; m=medium; h=high)
Prior and Current Offences: Low (2)   (l: 0-2, m: 3-4, h: 5-9)
Family/Living Circumstances: Low (0) (l: 0-1, m: 2-3, h: 4-7)
Education / Employment: Low (0) (l: 0-1, m: 2-3, h: 4-7)
Peer Relations: Low (0) (low: 0-1, medium: 2-2, high: 3-4)
Substance Abuse: Low (0)   (l: 0-0, m: 1-2, h: 3-6)
Leisure / Recreation: Low (0)   (l: 0-0, m: 1-2, h: 3)
Personality / Behaviour: Low (0)   (l: 0-0, medium: 1-2, h: 3-7)
Attitudes / Beliefs: Low (0)   (l: 0-0, m: 1-2, h: 3-5)  
Assessment of Risk of Re-offending: Low (2)   (l: 0-12, m: 13-23, h: 24-48)

14. Results Sample Characteristics
15.2% were found to have a Full Scale IQ score below 70 (M = 83.24, SD = 13.23)
On average, participants had a Performance IQ points higher than their Verbal IQ (SD = 12.82).
No significant difference in proportion of males vs females with an ID [² (1, N = 781) = .254, p = .61].
Mention indig if have time…?

15. Results Criminal Justice Involvement
No difference between ID vs nonID in:
type of offence (physical violence, traffic, sexual, other)
court outcome
Those with ID were more likely to
Have a larger number of attendances at court
Have more recorded offences
Have a greater frequency of bonds / probation
Have committed more property offences
CHECK FOR CONSISTENCY!!!

16. Level of Risk on YLS/CMI:AA for ID vs. non-ID
Mean YLS / CMI: AA total score was (SD = 9.35) for the total sample, placing participants, on average, in the ‘Medium Risk’ category of the YLS/CMI: AA. On average, those with an IQ below 70 scored in the ‘medium risk’ category and those with an IQ above 69 scored in the ‘low risk’ category.

17. Level of Risk on YLS/CMI:AA
YLS/CMI: AA Risk Category
IQ < 70 (n = 102)
IQ > 69 (n = 526)
Offence Type
Low
Medium
High
TOTAL
Robbery %
(n)
26.7
(4)
40.0
(6)
33.3
(5)
(15)
51.6
(64)
27.4
(34 )
21.0
(26)
(124)
Break and Enter
(0)
53.8
(7)
46.2
(13)
41.7
(20)
43.8
(21)
14.6
(48)
Other assault
24.3
(9)
40.5
35.1
(37)
25.9
(38)
36.7
(54)
37.4
(55)
(147)
Car & other theft
20.0
(3)
53.3
(8)
38.9
40.7
(22)
20.4
(11)
Aggravated assault
14.3
(1)
71.4
50.6
(44)
28.7
(25)
20.7
(18)
(87)
AVO
80.0
30.3
(10)
42.4
(14)
27.3
(33)
Other
(2)
48.5
(16)
39.4
12.1
20.6
45.1
(46)
34.3
(35)
(102)
(213)
34.8
(183)
24.7
(130)
(526)
However upon inspection, risk categorisation appeared to be associated with offence type. As, although proportions in some cells were small, for offences like AVO and break and enter, ID offenders were in higher risk categories on the YLS / CMI: AA compared with non ID offenders.

18. Criminogenic Needs Prior and current offences
YLS/CMI: AA Domain
Total
IQ<70
IQ>69
Mann- Whitney U
Mdn
(Range)
Prior and current offences 3
(0-9) 4
30196**
Family & living circumstances 2
(0-7)
(0-6)
Education/employment **
Peer relations
(0-4) **
Substance abuse
Leisure and recreation 1
(0-3) **
Personality/behaviour
31408
Attitudes and orientation
(0-5) *
The Mann-Whitney U statistic was used to compare the profile of YLS/CMI: AA domain scores for those with an IQ above 69 and below 70. Those with an IQ below 70 scored significantly higher (p <.05, two tailed) on: ‘Prior and current offences’; ‘Peer Relations’; ‘Education and Employment’; ‘Leisure and Recreation’; and ‘Attitudes and Orientation’. Only in the ‘Substance Use’ domain did those with an IQ above 69 score higher than those with an IQ below 70. However, this was not significant (p = .97, two tailed).

19. Prior and current offences Family & living circumstances
YLS/CMI: AA Domain
Total
IQ<70
IQ>69
Mann-Whitney U
Mdn
(Range)
Prior and current offences 3
(0-9) 4
30196**
Family & living circumstances 2
(0-7)
(0-6)
Education/employment **
Peer relations
(0-4) **
Substance abuse
Leisure and recreation 1
(0-3) **
Personality/behaviour
31408
Attitudes and orientation
(0-5) *

20. Criminogenic Needs Prior and current offences Education/employment
YLS/CMI: AA Domain
Total
IQ<70
IQ>69
Mann-Whitney U
Mdn
(Range)
Prior and current offences 3
(0-9) 4
30196**
Education/employment 2
(0-7) **
Peer relations
(0-4) **
Leisure and recreation 1
(0-3) **
Attitudes and orientation
(0-5) *

21. Implications - Risk ID is a risk factor for reoffending
Higher rates of ID in higher risk services
Impacted by offence type that requires further analysis (though no greater expression of sexual or physical violence)
Clear need for awareness of ID when entering criminal justice system.
ID offenders have higher average YLS/CMI: AA scores and more frequent placement in the higher risk category, indicating that young offenders with an ID may have a greater risk of reoffending. This is consistent with the PCC and recent meta-analytic studies that indicate that ID is a risk factor for recidivism in juveniles (Cottle et al., 2001; Lipsey & Derzon, 1998). With a very high over-representation of those with an ID in the ‘high risk’ category, it is possible that services delivering to ‘high risk’ offenders are likely to have greater over-representation of offenders with an ID.
The percentage of those with and without an ID in categories of ‘Low’, ‘Medium’ and ‘High’ risk appeared to differ for different offences. This suggests a possible interaction between offence type, ID status and risk categorisation that requires further exploration.

22. Implications - Needs
ID have the same profile of needs as non-ID offenders
ID often have higher needs than non-ID offenders
ID have significant social needs
Juvenile offenders with an ID have clear anti-social attitudes
Those with an ID had a significantly different profile of needs to those without an ID. Higher needs for those with an ID included domains relating to peers, leisure, education, employment and attitudes. This domain profile strongly reflected that those with an ID had greater social needs than offenders without an ID. These findings are consistent with the hypothesis that a social skills deficit is an important factor for those with an ID (Dodge & Pettit, 2003; Timms & Goreczny, 2002). It was anticipated that much offending by those with an ID was due to being implicated with anti-social peers who had antisocial attitudes. However, the significantly higher score on the ‘Attitudes and Beliefs’ domain suggest those juveniles with an ID have already adopted antisocial attitudes. It was also unexpected that those with an ID did not have a significantly higher score on the ‘Family and Living Circumstances’ domain as higher rates of neglect and abuse among children with an ID have been previously reported (Sullivan & Knutson, 2000; Verdugo, Bermejo, & Fuertes, 1995; White, Holland, Marsland, & Oakes, 2003).
The previously reported difference between Verbal IQ and Performance IQ of juvenile offenders (Katsiyannis, Zhang, Barrett, & Flaska, 2004; Teichner & Golden, 2000; Vermeiren, De Clippele, Schwab-Stone, Ruchkin, & Deboutte, 2002) was also replicated. This finding is consistent with the above social deficits as previous research has highlighted the link between low Verbal IQ, poor communication and poor social skills (Timms & Goreczny, 2002). However, since the YLS/CMI: AA has not been validated for young offenders with an ID, further research is needed before needs profiles of young offenders with an ID can be assessed reliably with this tool and used in treatment planning.

23. Implications - Responsivity
ID offenders are younger
High risk offenders had a higher risk of being ID
Offenders have lower Verbal IQ
Indigenous status

24. Issues Definition and diagnosis of ID in custodial settings
Under representation of ID in the sample?
Need to look at a comparative non-offending sample
Need to target criminogenic needs of juvenile offenders with an ID
Highlights that ID is not just a responsivity issue
While this study addressed common methodological flaws and considered a population often ignored in criminology research (those on community orders), there were a number of limitations of the study.
The WASI, although it has excellent psychometric properties, may not be a suitable substitute for a full psychometric assessment (Sattler, 2001). Adaptive functioning for the sample was not specifically assessed. However, we argued that all participants had a social adaptive functioning deficit due to offending that would sufficiently fulfil the adaptive behaviour deficit criteria required for a diagnosis of ID (Luckasson et al., 2002).
The number of participants with an ID in the sample may be an under-representation of young offenders on community orders. Within NSW law, there are a number of provisions that direct an offender away from court appearances and orders managed by correctional departments. There are specific provisions that target those with an ID, but it is not known whether those with an ID compared to those without an ID are more likely to be diverted from the criminal justice system. Regardless, the number of participants with an ID is likely to have been under-represented due to them being less likely than those without an ID to complete the survey due to challenging behaviour, mental health issues or limited capacity to answer questions.
The study did not consider a number of individual factors that may further moderate the role of ID in offending. In Australia, indigenous people are known to have a very high risk of recidivism (Snowball & Weatherburn, 2006; Weatherburn, Snowball, & Hunter, 2006) and indigenous juveniles have recently been found to be 23 times more likely to be in custody than non-indigenous juveniles (Taylor, 2006; Weatherburn et al., 2006). There is a need for further investigation into the relationship between offending, risk profiles, ID status and Aboriginality.
Finally, this study only considered those juveniles who had already offended. The question regarding the extent to which the profile obtained for this subgroup of young people with an ID pertains to ID offenders or to young people with an ID in the general population with no history of offending. If these needs are not exclusive to offenders, their role in offending behaviour and recidivism of offending would need to be reconsidered. Therefore, future research requires the inclusion of groups of young people with and without an ID who have not offended.
Conclusions
This study showed that those with an ID have a greater risk of reoffending based on the YLS/CMI: AA. It also showed that juvenile offenders with an ID have many of the same needs that have been previously shown to be present in juvenile offenders without an ID. In many domains, those with an ID presented with higher levels of need suggest that offending by those with an ID is not due simply to the individual’s inability to distinguish right from wrong. However, the different risk and needs profile and responsivity factors of those with an ID emphasise the need for juvenile justice departments to assess IQ and adaptive functioning of those entering the criminal justice system to ensure effective risk management and rehabilitation for this group. Services are likely to be most effectively delivered by a number of government departments and community services, highlighting the need for a whole of government approach to address the multiplicity of issues of young offenders with an ID. Such approaches would also be effective with those young offenders with borderline IQ functioning.