The focus in other reviews has been on the relative impact of different intervention methods. Reviews have dealt with educational and vocational programmes; with the impact of specially designed socio-therapeutic prison regimes in Germany; and with the effects of therapeutic communities defined in broader terms. There is a meta-analysis on the importance of addressing cognition as a mediating variable in behavioural change, several on the effectiveness of structured cognitive-behavioural programmes, and others of family-based interventions and school-based interventions respectively. Meta-analysis has also been used to synthesise findings from evaluations of restorative justice and victim-offender mediation. Moderator variables, including (as we saw earlier) the impact of age, gender and ethnicity, and the importance of staff skills and other aspects of organisational practices, have also been reviewed in this way.
In meta-analysis, the principal outcome of interest is known as the effect size. This can be represented in various ways, but the function in each case is the same: to provide a measure of the extent of any difference between “experimental” and “control” conditions in an intervention study. There are different methods of calculating effect sizes, and given the complexities of meta-analysis, some researchers have been critical of it. Ellis (2010) provides an overview of these issues, and of how effect sizes can be interpreted and compared.
The overall finding from meta-analysis of offender treatment sharply contradicts the once commonly repeated assertion that “nothing works” (Hollin, 2001; McGuire, 2013). The impact of psychosocial interventions on criminal recidivism is on average positive, that is, it is associated with a net reduction in reoffending rates in experimental relative to comparison samples. However, the average effect taken across a broad spectrum of different types of treatment or intervention is relatively modest. Expressed as a correlation coefficient, it is estimated on average to be approximately 0.10. This average finding obtained from the meta-analyses corresponds to recidivism rates of 45% for experimental groups, and 55% for control groups, respectively. Though this figure is low, it is statistically significant, and it compares reasonably well with effects found in many other fields. Some healthcare interventions that are generally regarded as producing worthwhile benefits have lower mean treatment effects. But crucially, when different subgroups of studies are compared, the variations between them show some consistent patterns that prove to be much more informative in illuminating aspects of intervention associated with higher rates of success.
There are some notable trends within the meta-analytic findings, but the following can be highlighted as emerging most clearly. On average, it has been found that effect sizes are larger for adolescent and for adult offenders than for those in what is usually called the “young adult” age range (15–18). On balance, with other factors held equal, community-based interventions have larger effect sizes than those delivered in institutions: the ratio of relative effect sizes obtained has ranged from approximately 1.33/1 to as high as 1.75/1. There are some complex interaction effects between criminal justice settings, types of interventions, and their quality of delivery. The best-designed services have their optimal benefit when provided in a non-custodial location. But even well designed interventions can have zero, possibly even negative effects if the quality of delivery is poor.
With reference to methods of intervention there are some markedly divergent outcomes. As discussed in the earlier part of this chapter, deterrent sanctions overall have nil or even negative effects. Hence, counter to what might widely be expected, they are associated with net increases in recidivism. Some interventions have positive but rather weak effects, whereas others have effects that are both statistically and practically meaningful in terms of reductions in rates of reoffending.
There is now a widespread consensus that it is possible to maximise effect sizes by combining a number of elements in offender programmes (Andrews & Bonta, 2010). Effective interventions are thought to possess certain common features that Andrews, Zinger, Hoge, Bonta, Gendreau, & Cullen (1990), in an early but highly influential review, called “principles of human service”, subsequently formulated as the Risk-Needs-Responsivity (RNR) model (Andrews, Bonta, & Wormith, 2006). When Andrews and his associates (1990) pinpointed those features that contributed separately to enhancing effect size, they found that the combination of them produced an additive effect. Interventions that possessed these features yielded an average reduction in recidivism rates of 53%. So although the mean effect size across all studies is not especially large, that is partly as a result of the negative effects of punishment. When interventions are appropriately designed and delivered it is possible to secure positive and much larger effects. When interventions based on RNR principles and treatment philosophy are compared with a response to criminal offending based on deterrence, the resultant outcome effects are visibly, indeed dramatically different (Lowenkamp, Flores, Holsinger, Makarios, & Latessa, 2010).
Probably the most widely disseminated innovation flowing from the above findings has been the synthesis of methods and materials into a number of prearranged formats known as programmes. This word sounds sinister to some people as it conjures up images of a rigid and insensitive method of working. But strictly defined, a programme consists simply of a planned sequence of learning opportunities (McGuire, 2001). Used in criminal justice settings, its general objective is to reduce participants’ subsequent criminal recidivism. Within that context, the typical programme is a prearranged set of activities, with clearly stated objectives, entailing a number of elements interconnected according to a planned design, which can be reproduced on successive occasions.
The largest single part played by psychologists in translating research findings into practice has been in designing structured programmes for use in this way (see Hollin & Palmer, 2006). Most such programmes currently employed in the criminal justice system use methods derived from cognitive social learning theory and are known as cognitive-behavioural interventions. While this is far from being the only theoretical option available, to date it has been the most consistent in yielding positive outcomes. Programmes of this type are usually supported by a specially prepared manual and other accompanying material.
17.5.3 Implications for Effective Practice
Expert reviewers agree that there are certain features of criminal justice interventions that maximise the likelihood of securing a practical, meaningful impact in terms of reduced reoffending. The major findings that arise from this include the following.
Theory and evidence base. Intervention efforts are more likely to succeed if they are based on a theory of criminal behaviour that is conceptually sound and has good empirical support. This provides a rationale for the methods that are used and it should identify the “vehicle of change” considered to be at work when an individual participates in and benefits from the programme. This might be accomplished, for example, by learning new skills, changing attitudes, improving ability to communicate, increasing self-knowledge, solving problems or overcoming bad feelings.
Risk level. It is generally regarded as good practice to assess risk levels and allocate individuals to different levels of services accordingly. Risk assessment is usually based on information about an individual’s criminal history, such as the age at which he or she was first convicted of a crime, and the total number of convictions to date. The most intensive types of intervention should be reserved for those offenders assessed as posing the highest risk of reoffending; individuals estimated as posing a low risk should not be exposed to such interventions. This has been called the risk principle (Andrews & Bonta, 2010) and it has acquired substantial support from review of the relevant literature (Lowenkamp, Latessa, & Holsinger, 2006).
Risk-need factors as targets of change. Research on the emergence of delinquency suggests that certain patterns of social interaction, social or cognitive skills, antisocial attitudes, the influence of delinquent peer-groups, and other factors are associated with its onset and maintenance. If work
ing with offenders is to make a difference to their prospects of reoffending, those variables should be its targets of change; they are factors that need to be addressed to alter offending and are therefore entitled criminogenic needs. Given their susceptibility to change they are sometimes called dynamic risk factors and there are strong reasons for prioritising them in rehabilitation services.
Multiple targets. Given the multiple factors known to contribute to criminal activity, there is virtual unanimity amongst researchers that more effective interventions will comprise a number of ingredients, addressed at a range of the aforementioned risks. Interventions that successfully do this are called multimodal. For example, working with a group of persistent young offenders might involve training them in social skills, learning self-control of impulses, and providing support for these changes through a mentoring scheme.
Responsivity. There are certain methods or approaches that have a superior record in engaging, motivating and helping participants in criminal justice interventions to change (Andrews, 2001; Gendreau & Andrews, 1990; McMurran, 2002). There are two aspects of this. First, rehabilitative efforts will work better if they have clear, concrete objectives, their contents are structured, and there is a focus on activity and the acquisition of skills. Personnel involved in providing this should possess high quality interpersonal skills and foster supportive, collaborative relationships within clearly explained boundaries (general responsivity). Second, it is vital to adapt intervention strategies to accommodate diversity amongst participants with respect to age, gender, ethnicity, language and learning styles (specific responsivity). With reference to ethnicity, in a review of structured interventions provided by Correctional Services Canada, Usher and Stewart (2014) found that programmes designed according to RNR principles produced similarly large, positive effect sizes across Caucasian, Aboriginal, Black, and a range of other ethnic groups.
Integrity. Lipsey (1995) and other meta-analysts have noted that intervention services appear to work better when they are being actively researched. Regular collection of data on how an intervention is delivered sustains its clarity of purpose, and its adherence to the methods it was intended to deploy. This feature is called the integrity or fidelity of an intervention (Bernfeld, 2001; Hollin, 1995) and in the best intervention services it is regularly monitored and checked. When the RNR model is applied with high integrity and adherence to the principles of effective services, outcome sizes are much higher than when that is not the case (Lowenkamp, Latessa, & Smith, 2006).
To provide the most favourable conditions for the delivery of the above kinds of services, many other ingredients should be in place, for example to ensure the validity of the assessment methods used, and the thoroughness of monitoring and evaluation procedures, alongside larger-scale strategies for managing the implementation of services within the provider agency (for fuller discussion, see Andrews, 2001).
PHOTO 17.3 Special programmes have been designed for offenders.
Source: © Mark Bowden/iStockphoto
17.5.4 Examples of Effective Interventions
There are some variations, but also important overlaps, between the kinds of interventions shown to have the largest impact on young and adult offenders respectively. For young offenders who have committed more serious offences, applying results from his own meta-analyses, Lipsey (1995, 2009) recommends that intervention programmes generally need to be provided with a duration of not less than six months, and with a minimum of two contacts per week. On the basis of the large-scale reviews, several methods have emerged as more likely to be effective for working with this group (Dowden & Andrews, 1999; Lipsey & Wilson, 1998):
Interpersonal skills training. This consists of a series of exercises designed to improve participants’ skills in interacting with others. Working in a small group, individuals identify situations in which they are not sure how to act, or which they sometimes mishandle, for example giving into pressure applied by others. Suitable ways of managing the situation are discussed, then practised using role-play, plus practice and feedback.
Behavioural interventions. In work with offenders this has included contingency contracts, where individual offenders and their supervisors compose a list of problem behaviours and a system of rewards for progress in modifying them. Behavioural training procedures such as modelling and graduated practice form part of many other types of interventions.
Cognitive skills training. There are several programmes of this type. Most consist of a series of structured sessions, each containing exercises designed to help participants acquire or develop their abilities in the domain of thinking about and solving everyday (usually interpersonal) problems. Typical material includes work on putting an everyday problem into words, gathering information about it, generating possible solutions, linking means and ends, anticipating consequences, making decisions, and allied skills.
Structured individual counselling. In its most widely used format, counselling is a relatively unstructured activity, in which the counsellor acts in a person-centred, non-directive manner, allowing the client to take the lead. This can be valuable for a number of purposes, but it has not emerged as an effective means of reducing offender recidivism. In order for it to work in that context, research suggests it needs to be more directive and structured, and based on a “reality therapy” or “problem- solving” framework.
Teaching family homes. These are residential units or group homes in which specially trained adults work in pairs as “teaching parents”. Their role is to develop positive working alliances with residents, impart a range of social and self-management skills, clarify boundaries, and provide counselling and advocacy services. Young people can continue to attend school and return to their homes-of-origin at weekends.
Some of the largest effect sizes published so far have been obtained from functional family therapy, parenting wisely, family empowerment, and similar therapeutic approaches that involve working with young offenders and their families. The most elaborate approach is multi-systemic therapy (MST), which comprises work with the young person, his or her family, and school staff (Henggeler, Schoenwald, Borduin, Rowland, & Cunningham, 1998).
For adult offenders, as mentioned earlier, average effect sizes are generally lower than for those at the younger age range. Nevertheless results are meaningful in practice, and comparable patterns emerge in respect of the types of intervention most likely to work. With this age group, with the exception of domestic violence programmes, family-based work is uncommon and interventions are almost exclusively conducted with individuals themselves, though often employing a group format.
The best-validated and most widely disseminated approaches involve “manualised” cognitive-behavioural programmes focused on risk factors for criminal recidivism (Lipsey, Landenberger, & Wilson 2007; Lipton, Pearson, Cleland, & Yee 2002aa; Tong & Farrington, 2006; Wilson, Bouffard, & MacKenzie, 2005). Variants of the approach have been well supported primarily for individuals with mixed patterns of offending that may include property, violent, and substance-related crimes (Hollin, 2001; McGuire, 2006; Motiuk & Serin, 2001). These types of programme have also yielded positive results when disseminated on a large scale in probation service settings (Hollin, McGuire, Hatcher, Bilby, Hounsome, & Palmer, 2008; McGuire, Bilby, Hatcher, Hollin, Hounsome, & Palmer, 2008; Palmer, McGuire, Hounsome, Hatcher, Bilby, & Hollin, 2007). Despite initially high drop-out rates, after the delivery process became better established the level of completion showed substantial improvement (National Offender Management Service, 2008). There is evidence that on two-year follow-up, the actual rate of reconviction of those attending such programmes is significantly below the expected rate (Hollis, 2007).
Specially designed programmes with additional components have been developed for adults who have committed violent offences. This may include a focus on anger control, modulation of moods, and recognition and self-management of risk (Bush, 1995; Henning and Frueh, 1996). While there have been some very strong findings fro
m anger management programmes, such as those of Dowden, Blanchette, and Serin (1999) who reported an 86% reduction in violent reoffending over a three-year follow-up with high-risk prisoners in Canada, in other instances treatment gains have been very marginal and more research is need on the appropriateness of allocation to this type of programme and the related issue of “readiness to change”.
For men who have committed domestic violence offences, to achieve successful outcomes further sessions need to be included examining perceptions of male and female roles, beliefs concerning responsibility for actions, or concepts of masculinity (Dobash & Dobash, 2000). While claims regarding successful intervention in this area have so far been only tentative (Babcock, Green, & Robie, 2004), there are positive findings through use of behavioural and cognitive-behavioural couple therapy for reducing both low (Simpson, Atkins, Gattis, & Christensen, 2008) and high (Easton, Mandel, Hunkele, Nich, Rounsaville, & Carroll, 2007; O’Farrell, Murphy, Stephan, Fals-Stewart, & Murphy, 2004) levels of marital discord and violence.
Still further ingredients are added in work with individuals who have committed sexual offences, the vast majority of whom are male. In addition to cognitive and social skills training and similar activities, these interventions usually also include a focus on deviant sexual arousal, cognitive distortions, impulsivity, and other sessions designed to address the established risk factors for this kind of offence. The precise contents may vary further according to specific types of offence, differentiated mainly by whether victims are children or adult women (Marshall, Fernandez, Marshall, & Serran, 2006). The content of such programmes alters over time, as knowledge accumulates on the relative importance of different categories of risk variables (Mann, Hanson, & Thornton, 2010).
Forensic Psychology Page 79