Rapid Evidence Assessment of Interventions that Promote ...

21 downloads 10265 Views 365KB Size Report
Learning and Skills Unit (OLSU) at the Department for Education and Skills. (DfES). ... EPPI centre for technical support and advice with methodology and to the.
R ESEARCH

Rapid Evidence Assessment of Interventions that Promote Employment for Offenders

Jane Hurry, Laura Brazier, Mary Parker and Anita Wilson National Research and Development Centre for Adult Literacy and Numeracy (NRDC) Institute of Education (IoE)

Research Report RR747

Research Report No 747

Rapid Evidence Assessment of Interventions that Promote Employment for Offenders

Jane Hurry, Laura Brazier, Mary Parker and Anita Wilson National Research and Development Centre for Adult Literacy and Numeracy (NRDC) Institute of Education (IoE)

The views expressed in this report are the authors’ and do not necessarily reflect those of the Department for Education and Skills. © Department for Education and Skills 2006 ISBN 1 84478 732 X

1

PREFACE PAGES Authors This report was prepared by Jane Hurry, Laura Brazier, Mary Parker and Anita Wilson Review team membership Laura Brazier NRDC, Institute of Education Jane Hurry Psychology and Human Development, Institute of Education Mary Parker Psychology and Human Development, Institute of Education Katy Sutcliffe EPPI Centre, Institute of Education Anita Wilson NRDC, Literacy Research Centre, Lancaster University Advisory panel Paul Arnaldi Chris Barnham Meg Blumson Garry Broom Martin Copsey Sally Dench Kimmett Edgar David Faulkner Caroline Hudson Alison Ismail Del Jenkins Liz Lawson Robert Newman Sue O’Hara Malcolm Ramsey Bill Shepard Duncan Stewart Alan Taylor John Vorhaus Liz Walker Richard Ward Richard White Julie Wilkinson Rosie Zwart

Department for Education and Skills Offenders Learning and Skills Unit Home Office Learning and Skills Council Home Office Institute for Employment Studies Prison Reform Trust Centre for Criminology, University of Oxford Probation Service Directorate Department for Education and Skills Department for Work and Pensions Department for Education and Skills Youth Justice Board Learning and Skills Council Home Office Department for Work and Pensions Home Office Custody to Work Unit NRDC, Institute of Education Learning and Skills Development Agency Department for Education and Skills Department for Education and Skills Home Office Learning Skills Development Agency

The interim report was presented on Wednesday 22 June 2005 at the Institute of Education. The final report was presented on 28 September 2005 at Institute of Education. The Advisory panel is also listed in Appendix 1. 2

ACKNOWLEDGEMENTS The opportunity to undertake this review was provided by the Offenders Learning and Skills Unit (OLSU) at the Department for Education and Skills (DfES). The views expressed in this report are those of the authors and not necessarily those of the OLSU or policy. Thanks are due to colleagues at the EPPI centre for technical support and advice with methodology and to the NRDC Core team for administrative help. ABBREVIATIONS DfES Department for Education and Skills EPPI Evidence for Policy and Practice Information and Coordinating Centre NACRO National Association for Care and Resettlement of Offenders NRDC National Research and Development Centre for Adult Literacy and Numeracy NVQ National Vocational Qualifications OLSU Offenders Learning and Skills Unit ONS Office for National Statistics RCT Randomised Control Trial SEU Social Exclusion Unit

3

CONTENTS Page numbers

Preface pages Executive summary

6–10

1. BACKGROUND 1.1 Aims and rationale for current review 1.2 Definitional and conceptual issues 1.3 Policy and practice background 1.4 Research background 1.5 Authors, funders, and other users of the review 1.6 Review questions

11–15

2. METHODS USED IN THE REVIEW 16–20 2.1 User involvement 2.1.1 Approach and rationale 2.1.2 Methods used 2.2 Identifying and selecting studies 2.2.1 Defining relevant studies: inclusion and exclusion criteria 2.2.2 Identification of potential studies: search strategy 2.2.3 Characterising included studies 2.2.4 Identifying and describing studies: quality assurance process 2.2.4.1 Screening 2.2.4.2 Key wording 2.3 In-depth review 2.3.1 Selection of studies for in-depth review 2.3.2 Assessment of studies in the in-depth review 3. IDENTIFYING AND DESCRIBING STUDIES: RESULTS 21–29 3.1 Studies included from searching and screening 3.2 Characteristics of the included studies (systematic map) 3.2.1 Characteristics of the programmes 3.2.2 Characteristics of the participants 3.2.3 Employment focus of programmes and types of outcome reported 3.2.4 Types of studies 3.3 Speaking to practitioners 4. IN-DEPTH REVIEW RESULTS 4.1 Identifying studies for in-depth review 4.1.1 Criteria for inclusion in-depth review 4.2 Evaluation using the Maryland Scientific Methods scale Description of Maryland Scale 4.2.2 Evaluation of the studies 4.2.3 Discussion of the studies 4.3 What works? 4

30–47

4.3.1 Categories of interventions 4.3.2 Maryland criteria of ‘What works’ 5. THE WIDER REVIEW 48–62 5.1 What interventions promote employment in offender populations? 5.2 What are the implementation issues for interventions aiming to promote employment for offenders? 5.2.1 System-related barriers to employment 5.2.2 Employment-related barriers 5.2.3 Re-integration into the community 5.3 To what degree are programmes targeted according to assessment of needs? 5.3.1 Age of offender 5.3.2 Sex of offender 5.3.3 Ethnicity of offender 5.3.4 Individual needs of offender 5.3.5 Needs of the workplace 6. CONCLUSIONS AND IMPLICATIONS 6.1 Conclusions 6.2 Limitations of the review 6.3 Implications 6.3.1 Policy messages emerging from the evidence base 6.3.2 Practice emerging from the evidence base 6.4 Future research

63–67

7. REFERENCES 7.1 Studies included in map and synthesis 7.2 Other references used in the text of the report

68–75

APPENDIX 1–7 1 Advisory Group membership 2 Exclusion and inclusion criteria 3 Search strategy for electronic databases 4 Journals hand searched 5 EPPI Key word sheet including review-specific key words 6 Details of studies included in the systematic map 7 Details of studies included in the in-depth review

76–103

5

LIST OF TABLES Table 3.2.1 Country of origin: where the research was carried out Table 3.2.2 Examples of names of programmes reported by studies Table 3.2.1.1 Type of Intervention reported by studies Table 3.2.1.2 Types of programmes not in a pre-designated category Table 3.2.1.3 Programme timing: the point at programmes took place Table 3.2.2.1 Type of offender targeted by programmes Table 3.2.2.2 Other type of offender specified by studies Table 3.2.3.1 Environment where programmes took place Table 3.2.3.2 Aspect of employment focused on by programmes Table 3.2.3.3 Specified employment focus not in a pre-designated category Table 3.2.3.4 Employment-related outcomes and re-offending Table 3.2.3.5 Employment-related outcomes specified Table 3.2.4.1 Design of studies Table 3.2.4.2 Sample size Table 4.2.1.1 Studies in in-depth review EXECUTIVE SUMMARY Background It is well documented that offenders have a high risk of unemployment and there is an association between unemployment and recidivism (Farrington et al, 1986; Finn, 1998b; Gendreau et al, 1998; May, 1999; Motiuk, 1996). It seems intuitively sensible that if offenders are in work, this will reduce their likelihood of offending and prospective studies tend to support this (Farrington et al, 1986; Gendreau et al, 1998). Offenders are also a legitimate concern for those involved in reducing unemployment and social exclusion. Offenders themselves consider that assistance in getting into employment is critical to help them stop re-offending (Gillis et al, 1996; Erez, 1987; NACRO, 1993 Gill, 1997). Reviews spanning the last 15 years have been optimistic that effective employment-focused interventions can reduce recidivism, though policy makers need to be cautious because not all interventions work equally well (Lipsey, 1995; Sherman et al, 1997). We need to gain a more specific understanding of what works and for whom in terms of employment rather than the more distal and thoroughly reviewed outcome of recidivism. Aims of the review This review aims to systematically draw together empirical evidence about interventions that focus on promoting employment for offenders. This provides a synthesised and sound evidence base that will better inform policy makers, practitioners and researchers. Methodological approaches and quality of the evidence within this field are documented and considered in depth for a subset of studies of particular relevance and quality. The review identifies gaps in our knowledge and highlights priority areas for further review and future research work. Review questions The key question this review seeks to address is: 6

What evidence is there about the types and levels of intervention that work best to promote employment for offenders? This is broken down into three sub-questions: 1. What interventions promote employment in offender populations? 2. What are the implementation issues for interventions aiming to promote offenders’ employment? 3. To what degree are programmes targeted according to assessment of needs? Methods Support and advice was received from the EPPI Centre for conducting this review. However, the final stage of extracting data for an in-depth analysis diverged from EPPI-Centre procedures. The principles of systematic review were applied but concessions were made to complete the review within a shorter time frame. A systematic review differs from a conventional literature review because it seeks to identify all the available evidence on a given theme. This evidence is then screened for relevance and quality, and synthesised into an overall summary. This review was conducted in three stages. The first stage used mainly electronic resources to identify studies, which were screened for relevance, with agreed inclusion criteria. Based on reading of abstracts, the relevant studies were mapped, and a preliminary analysis presented. In the second stage full papers of the selected studies were sought and these were rescreened for relevance. A second and wider systematic mapping also occurred at this stage. For the third stage a small number of studies were selected using narrowed inclusion criteria and these were analysed in greater depth for methodological rigour. Results of searches We searched for studies on nine electronic bibliographic databases. A total of 4,051 studies were identified and abstracts of all these studies were screened for relevance according to our agreed criteria. An initial list of 252 studies met our inclusion criteria and for these studies the full papers were sought. A second screening using the same inclusion criteria but this time based on the full papers resulted in a selected list of 53 studies and a descriptive mapping was carried out on these. Using criteria based on the design and content of the studies we then selected a final group of seven papers for further evaluation. Systematic map A systematic map based on the full papers of the 53 included studies has been produced (presented in Section 3.2). The majority came from the US. Only a small number (nine studies) came from the UK. Just over half (29 studies) of the studies specified names for 7

programmes. Vocational programmes were the most common, closely followed by educational programmes. Just over half of the studies reported types of intervention that did not fit into any of the pre-designated categories. It was difficult to map the duration of interventions with 35 of the studies not specifying how long participants attended. It was not possible to categorise the intensity of the programmes, as 45 of the studies did not specify how frequently participants attended. In other words, details of the precise content or nature of intervention was frequently not available. This is a well-known problem particularly associated with experimental methodologies. By a very small margin the largest identified type of offender was adults aged over 21. A quarter of the studies reported programmes that were aimed at juveniles or young offenders under age 21. It was difficult to discern from studies the precise age range of participants as many studies did not report this information. Over half the studies did not specify whether the intervention was aimed at men or women. In a similar way to age range and sex, precise information on ethnicity was limited. Many studies did not specify how or whether participants were categorised by race. Just over half the studies stated that programmes took place prior to release from custody. The second most common time for programmes to take place was both prior to and post release from custody. By a narrow margin the most common employment focus was on preparation for work. This was followed by full-time employment, then by part-time employment. The two most common outcomes indicated were gaining employment and not re-offending. In many cases studies measured a combination of employment and recidivism outcomes. The most common design for studies was looking at outcomes for a single group of participants who had attended a programme, a design which is easier to mount but frequently makes interpretation of findings difficult. A third of studies included a comparison group. Overall, the number of participants reported was fairly high. Just over a quarter of studies reported that they had a sample size of over 200. A quarter of studies provided contextual information about the labour market. Information about the other factors such as the relationship between offenders as supervisors was reported by a limited number of studies. Summary of main findings • Good quality studies of seven intervention programmes find that in six of the programmes, offenders in the treatment group are significantly more likely to be employed at least six months after completion than those in comparison groups. It would be preferable to have a larger number of studies on which to draw but, unfortunately, these were all we could find in our extensive review. However, the consistency of the positive results gives some confidence in the overall conclusion that employment interventions are effective. 8



• • •

• • •







Prison work and vocational training are identified as working (effective) and the one community employment programme identified is ‘promising’ according to the criteria developed by Sherman and his team (1997) (for definition of ‘working’ see page 47). Education does not emerge as having an impact on employment but the evidence base here is too small and we must conclude that this is still an open question, especially in view of the fact that evidence that education programmes can reduce recidivism (Lipsey, 1995). To complicate the interpretation of the impact of education programmes, there is very little description of what ‘education’ involves. Prison work involved conventional work, rather than training, in one case, equipment, management and training of offenders being conducted by the private sector employers. Vocational training in one case included formal vocational courses, experience of industrial work and apprenticeship. In the other study identified a description of the intervention was not provided. The sole community employment programme was broad-based, and extensive, including vocational and educational input in custody followed by career guidance, and help with job search and finding work for inmates post release. The focus in these types of programmes is to streamline the transition of the offender from custody to the community. Those aged under 26 years may be more difficult to engage in interventions and to help into employment than older people. There is no good quality evidence on programmes run entirely in the community. To learn more about the essential ingredients of effective programmes, it is necessary to employ qualitative methods as well as quantitative, giving us more information about how each dimension of programmes function, how they are implemented and how they are received by the client group. Various types of barriers emerge as the main implementation issues for interventions aiming to promote employment for offenders. Three categories of barriers were identified: - barriers inherent in the criminal justice system that are primarily related to sentencing policy and funding structures; - barriers related to employment include having a criminal record, a poor work history and facing negative attitudes of employers; - barriers that centre on the transition from custody back into the community. Programmes tended to be targeted according to a ‘generalised’ assessment of that particular group of offenders needs. This would include substance abuse, women, young fathers and ethnic minorities. It is unclear if programmes (specialised or otherwise) are targeted according to an assessment of an offender’s ‘individual’ needs. The research evidence on women and ethnic minorities is limited, with the majority of studies focusing on men only or including women as a minority group. Many studies do not report the ethnic make-up of their 9



• •

• •

participants and few studies focus specifically on majority ethnic groups such as Hispanics. Based on this limited information, Hispanic offenders were less likely to go on work release than black or white offenders. Middle-aged offenders and those convicted of property crime were the most likely to attend, older offenders were more successful than young offenders and white offenders were more successful than black and Hispanic offenders. Success was determined by the programme’s infractions or re-arrests. On the whole, the programmes examined by the studies in this review provided a good range of skills relevant to the realities of the workplace. Some prison-based employment has been criticised for employing outdated technology or equipping prisoners for work which is not available in the areas of their release. This limits its value for the job market outside prison. Programmes which address a wider range of issues (eg anger management, housing, job search, job skills, etc) are seen as more helpful than those with a narrow focus. Both the US and Canada have compiled a list of critical skills required in the workforce.

10

1.

BACKGROUND

1.1 Aims and rationale for current review This review was commissioned to provide an evidence base to inform British policy about employment programmes for offenders. There are a number of reasons why this is a legitimate concern for the criminal justice system and society more generally. Employment programmes may: • • • • •

reduce unemployment reduce crime advance social equality (people with a criminal record are overrepresented amongst the unemployed) facilitate prison management reduce prison costs (though this seems to be more commonly considered in the USA than in the UK)

The primary focus of the review is to explore whether or not employment programmes do promote employment for those currently serving a prison sentence, on probation or under a community order. Where possible the impact on recidivism is reported, which types of employment programme are most effective and whether programmes have a differential effect depending on the characteristics of the offender. Finally, it was anticipated that the review would leave certain questions unanswered. This review aims to identify the most fruitful and useful areas for further review or research. 1.2 Definitional and conceptual issues Four key terms are fundamental to this review: ‘offender’, ‘employment’, ‘outcomes’ and ‘intervention’. Offender is defined as a person convicted of a criminal offence in a court of law and given a community or custodial sentence. We further defined the offender group with ‘young offenders’ who are aged between 17 and 21 and ‘juveniles’ who are aged between 10 and 16. The juvenile offender category is necessary as it incorporates offenders aged 16 and studies may include mixed populations. The issue of how to deal with mixed populations (target and non-target group) is addressed in the exclusion criteria. In cases where offenders have recently completed their sentence they are defined as ‘exoffenders’. These distinctions may not necessarily be made explicit by studies. Employment is defined as broadly as possible as paid or unpaid regular work. This was to enable us to incorporate the various types of employment this particular population may be engaged in, for example cash-in-hand work with relatives or friends and voluntary work either full- or part-time. Intervention is again defined as broadly as possible to incorporate the wide variety of employment activities in a variety of criminal justice contexts. Offenders have multiple needs and are often involved in numerous 11

interventions. This means the employment aspects of an intervention may be embedded within a variety of other subjects. This may include preparation for release courses or substance misuse programmes. Four sub-types of employment related intervention can be identified. These are: 1. general offender programmes 2. education and/or vocational training programmes 3. career guidance, job search support, job centre plus 4. welfare to work programmes and in-employment support. In line with the aims of the review and the review questions, although the employment aspect of an intervention may be embedded in other subjects, we exclusively focused on interventions which have employment-related outcomes. Outcomes are defined and conceptualised in a two main ways. Direct employment outcomes report the numbers of offenders on a particular intervention gaining employment. Depending on the methodology, this may or may not be in comparison to a control group. Indirect employment outcomes report on various employment-related achievements. This would include measurable improvements in work-related skills such as gaining an NVQ. Other outcomes may include improving ‘soft skills’ such as interviewing skills or job searching techniques. 1.3 Policy and practice background There is increasing emphasis being placed on the links between offender resettlement, employment and lowered levels of recidivism. It is embedded at the broadest level within the government policy document ‘Reducing ReOffending: National Action Plan (2004) and reflected in policy documents in other parts of the world such as Australia (Borzycki, 2005) and the USA (Reentry Policy Council). Across Europe the scale of potential employees is aptly illustrated by ‘What Works with Offenders’ (EOEF, 2003), which notes that the current number of unemployed ex-offenders in Europe is likely to ‘run into millions’. The report also notes the dual efficacy of employment for offenders, suggesting that jobs not only help to cut crime but also to combat social exclusion, a key message in current UK government policy documents (SEU, 2002). There is also a growing awareness, however, that this is not only a current issue but also a complex problem (Webster et al, 2001). Employers’ attitudes towards ex-offenders may need to be re-appraised (Fletcher et al, 2001; Graffam et al, 2004). Changes in policy, such as the implementation of the Police Act (1997), may be having a detrimental effect on employers’ willingness to employ ex-offenders and there are wide-ranging implications for non-employability brought about by collateral sanctions, such as where an offender is allowed to live, who they may work alongside, their involvement with drugs and the restrictions of electronic monitoring or probationary restrictions. The new National Offender Management Strategy, introduced following the recommendations in the Carter Review for Correctional Services and the Government’s response, with its plan for an ‘end to end’ service for offenders, may well overcome the previous shortcomings of the Probation 12

Service’s limited involvement with offender employment identified by Bridges (1998), the limited success of Phase 1 of the Employment Pathfinder (Haslewood-Pocsik et al, 2003) and the noted omission within the Prison Service key performance indicators of impact on re-offending and getting people to work noted by Gill (1997). There has also been a move to address the needs of offenders within wider government policies on social inclusion both among young people (Bynner et al, 2004) and working age adults (Hasluk and Green, 2004) providing access to Connexions, New Deal (Sarno et al, 2001) and progress2work Linkup. Overall, policy generally appears to be moving away from the notion of single interventions. As noted by the Select Committee on Education and Skills’ recommendation on prison education (2005), any intervention needs to be seen as part of a wider approach to reduce recidivism through rehabilitation and this is reflected in a number of the interventions noted in the review. 1.4 Research background The overall association between employment and crime is well-established, having been consistently reported over the last 60 years. People who offend are more likely to be unemployed than their peers and are more likely to reoffend if they are unemployed (Andrews, 1995; Farrington et al, 1986; Finn, 1998b; Gendreau et al, 1998; Glueck and Glueck, 1930; Hawkins et al, 2000; May, 1999; Motiuk, 1996; Sampson and Laub, 1993; Wolfgang et al, 1972). For example, in the UK a study of over one thousand offenders under probation’s supervision found that only 21 per cent were in employment compared to around 60 per cent of the general population (Mair and May, 1997). However, this relationship is complex (Sherman et al, 1997). For example, Uggen (2001) reports that whilst unemployment is associated with higher levels of crime in adult offender populations, for juveniles, employment has actually been found to be positively associated with crime (Bachman and Schulenberg, 1993; Gottfredson and Hirschi, 1990; Milhalic and Elliott, 1997; Ploeger, 1997; Wright et al, 1997). Another demonstration of the complexity of this relationship is suggested by Field. In the UK, Field (1990, 1999) looked at the macro relationships between crime rates and the economy, finding that when consumption was low (and unemployment high), property crimes were high but when consumption was high, violent crimes were also high. It would seem that whilst it is generally true that employment and crime are negatively associated, this relationship does not hold across all ages or types of crime. To the extent that unemployment is associated with offending, it is logical to explore programmes which promote employment for offenders as a method of reducing crime. It is seems intuitively sensible that if offenders are in work, this will reduce their likelihood of offending and prospective studies tend to support this (Farrington et al, 1986; Gendreau et al, 1998). However, for such interventions to work there must be a causal relationship between employment and crime and this has been more difficult to establish. There are 13

a number of plausible explanations for the association between employment and crime: • • • •



Lack of legitimate means of getting money may lead to crime. Unemployment may lead to boredom which may lead to involvement in drugs or fights. Employment may encourage social integration and therefore act as a protective factor. Being unemployed and offending may be associated with similar risk factors, such as problems with anger management, living in a poor community where unemployment and crime are high, having problems with drugs, etc. Having a criminal record may cause unemployment.

There is evidence for each of these explanations in the literature on the link between employment and crime (eg, May, 1999; Nacro, 1993a, 1993b; Bridges, 1998; Gillis et al; 1998; Bouffard et al, 2000). Sherman and his colleagues (1997) also outline a range of theoretical explanations for the link between employment and crime: economic choice theory (choice between work and crime on the economic attractiveness of the two options); control theory (being at work exerts social control); anomie (crudely speaking, social exclusion); labelling theory (people will not want to employ offenders). If all of these dimensions play a part in the association between employment and crime we may expect different programmes to be effective for different groups. For example, a person with drug problems who has difficulty getting or keeping a job and commits drug-related offences will need help with their drug problem, someone with serious literacy and numeracy problems will need a different programme and someone just out of jail may have different needs again. There are four major barriers to offenders being in employment. 1. Offenders tend to have lower levels of education, qualifications and vocational skills than other members of the community and this may act as a barrier to employment (eg Parsons, 2002). 2. Offenders may have psychological problems, including drug abuse, which make keeping a job difficult (May, 1999; Gendreau and Ross, 1980). 3. Living circumstances, in particular insecure housing, but also living in areas of high unemployment, can act as a barrier both to gaining work and keeping it (Nelson et al, 1999; Webster et al 2001). 4. Employers are reluctant to hire people with a criminal record (NACRO, 1997; Hamlyn, 2000). As a consequence of this complex picture, it is increasingly being recommended that interventions aimed at reducing re-offending should address as many potential barriers as possible (McGuire, 1995; Nuttall et al, 1998), though, as Bouffard et al (2000) point out, this makes it difficult to determine what aspects of interventions are responsible for positive outcomes. 14

The uncertainty of these inter-relationships also casts doubt on the role of employment programmes in crime reduction, although offenders themselves consider that assistance in getting into employment is critical in helping them stop re-offending (Gillis et al, 1996; Erez 1987; NACRO 1993a; Gill, 1997). It is to clarify these relationships that interventions are evaluated to determine ‘what works’. The review has systematically identified such studies for indepth examinations and synthesis. 1.5 Authors, funders, and other users of the review The review team has considerable expertise of research with both offenders and education, but is relatively new to interventions aiming to promote employment for offenders. The team members come from psychological and sociological disciplines and from a variety of quantitative and qualitative methodological backgrounds. The Department for Education and Skills funding this report have an interest in a comprehensive evidence base to contribute to the policy-making process. Various departments within the Department for Education and Skills, such as the Offenders Learning and Skills Unit (OLSU), are potential users of this review, as are other government departments, such as the Home Office. Other users of the review include research colleagues who are interested in both methodology and further work in this field. It is hoped that the review would be of interest to practitioners working with offenders and developing interventions aiming to promote employment for offenders. The Advisory Panel was designed to represent policy, practice and research interests. It is hoped that by bringing together different types of users, the final review will provide a joined-up analysis of how evidence can be synthesised into a coherent body of work, with multiple uses. 1.6 Review questions The primary question this review seeks to address is: What evidence is there about the types and levels of intervention that work best to promote employment for offenders? This has been sub-divided into three more specific questions: 1. What interventions promote employment in offender populations? 2. What are the implementation issues for interventions aiming to promote offenders’ employment? 3. To what degree are programmes targeted according to assessment of needs?

15

2.

METHODS USED IN THE REVIEW

Support and advice was received from the EPPI Centre for conducting this review. However, the final stage of extracting data for an in-depth analysis diverged from EPPI-Centre procedures. This project was conducted in three stages. The first stage was a rapid assessment of the research evidence from the titles and abstracts of studies obtained using a variety of bibliographic databases. The titles and abstracts were mapped using a review specific key wording tool. At the end of the first stage, a consultation with DfES, research and practitioners took place to further clarify scope of the project and direct more in-depth analysis of the identified literature. The second stage of the review involved locating the full texts of the titles and abstracts used in the first stage. The full texts of the papers obtained within the time frame were re-screened. This re-screening process resulted in a smaller group of studies which were key worded using a slightly modified version of the review specific key wording tool used in the first stage. A more comprehensive map of the included studies is provided in Section 3. In the third stage, a narrower set of inclusion criteria was agreed for the indepth review. Selected studies were analysed and evaluated for methodological quality. This evaluation is provided in Section 4. 2.1 User involvement 2.1.1 Approach and rationale For systematic reviews to be relevant to policy and practice, potential users of the review must be involved in key stages of the review process. User involvement has been sought through an Advisory Group with policy-maker, practitioners and research representation. 2.1.2 Methods used In early April 2004 an initial proposal for the review was submitted to DfES. A meeting was convened to discuss the scope and methodology of the review. After collaboration with DfES, the key issues were clarified and the revised proposal formed a working document to guide the project. Two reports were planned. The first was an unpublished interim report of the first stage of the project. This final report builds upon the interim report incorporating user feedback and will be submitted for peer review publication. Two seminars were planned with DfES representatives and research colleagues. The first seminar considered the scope of the evidence identified so far, noted further literature to incorporate into the review and considered initial interpretation of findings. Messages for policy and practice were discussed and the seminar aimed to identify and refine the focus of questions or issues for the review to consider in its second stage. The second seminar 16

will consider the development of the findings and messages for policy and practice and discuss future programmes of work. 2.2 Identifying and selecting studies 2.2.1 Defining relevant studies: inclusion and exclusion criteria To address the research questions identified for this review, we searched for reports of studies which evaluated the effect of interventions on post release employment outcomes for offenders over the age of 16. After discussion within the research team a decision was made to exclude studies if they: • • • •

• •

are not published in the English language; were published in or before 1984; do not concern offenders and employment – the direction of the relationship must be that the intervention is aimed at having an impact on promoting employment; concern offenders under 16 years old – in cases of mixed populations, ie non-offenders and offenders and under 16s as well as over 16s, the target group (offenders over 16) must comprise a majority of the sample and be distinguishable in reported outcomes; do not concern interventions that aim to have an impact on promoting offender employment – this may include opinion papers; do not evaluate the degree to which interventions promote offender employment.

All studies identified by the initial search process were screened, using title and abstract, on the basis of relevance to the topic area and the other inclusion/exclusion criteria outlined above. Studies that met the criteria were included in the review. 2.2.2 Identification of potential studies: search strategy Studies were identified using a number of electronic databases. A wide range of terms to capture the type of population such as ‘offender’, ‘prisoner’, etc, were combined with terms to capture the possible types of intervention to be considered, and also the topic focus of employment, so that only studies that contained all three types of terms in the title or abstract were retrieved. Nine bibliographic databases were used: the British Education Index; Web of Knowledge (Social Science Citation Index); ASSIA; ERIC; Australian ERIC; PsychINFO; Education On Line; CERUK; National Criminal Justice Reference Service Abstracts; C2-SPECTR (Campbell Collaboration). Please refer to Appendix 3 for the search strategy for electronic databases. Internet searches were conducted using Google and Google Scholar. Specific websites were searched for references. These include the Home Office, DfES and Youth Justice Board. 17

A reference list of key papers based on expertise of the review group was also used to identify additional potential studies that should be included. Searching of nine electronic bibliographical databases identified 4,051 studies. These studies were imported into the EPPI Reviewer database and 86 duplicates were removed. The exclusion criteria were applied to the remaining 3,965 studies based on the details provided in the database entry such as publication date and language, the title and abstract. This screening process resulted in the identification of 252 studies meeting the inclusion criteria. Of these 252 studies, we were able to obtain the full papers for 201 studies. Most studies were obtained either electronically or through library searches. These 201 studies were re-screened and 53 were included in the review. The review specific coding frame was applied to the included studies. 2.2.3 Characterising included studies After initial screening, all the included studies were characterised using a coding framework devised specifically for this purpose. (See Appendix 5 for review specific coding frame.) The types of codes applied were categorised into the following sections: • • • • • • • • • • • • • • • • • •

Country of origin: where was the report carried out Programme name Type of intervention Duration of programme Intensity of programme Point of intervention Type of offender Age Gender Ethnicity Environment Employment focus Employment-related outcomes Type of report Design of study Sample size Contextual information about the labour market Contextual information about the relationship between offender and the supervisor.

2.2.4 Identifying and describing studies: quality assurance process To assure reliability and consistency of results, quality assurance exercises between researchers were carried out at the screening stage (applying the exclusion criteria) and at the key wording stage. After devising a draft of the 18

exclusion criteria, the review team met to code a number of studies separately and compare results as a group. 2.2.4.1 Screening • Abstracts – A moderation exercise was carried out on 25 studies to ensure consistency of understanding of exclusion criteria. This was then tested on the abstracts of a further 60 studies which were screened in pairs. There was a 96% agreement between researchers on whether a study was included or excluded. • Full-document – Approximately 10% of the full documents was doublescreened and agreement to include or exclude studies was achieved in 80% of the cases. This was followed by discussion on ways of clarifying criteria to achieve a higher rate of consistency between researchers. 2.2.4.2 Key wording • Abstracts – The quality assurance process for the key wording tool was conducted in a similar way. Firstly, a draft key wording tool was developed and tested within the group. Then ten studies were chosen which were key worded by four members of the review team with full agreement between the four on 74% of questions. The results were compared and put into a table. The key wording tool was then revised and tighter guidelines for how to apply codes were agreed. • Full document – A random selection of five of the 53 studies were double key worded using the revised key wording tool and achieved a consistency between researchers of just under 90%. 2.3 In-depth review 2.3.1 Selection of studies for in-depth review Studies to be included in the in-depth review had to satisfy two additional criteria, which were discussed and agreed at the seminar with DfES representatives: • •

The design of the study must include a control group so that a comparison can be made between outcomes with and without the intervention programme. Data on employment outcomes must be tracked over time to demonstrate that offenders maintain employment for a minimum of three months post release.

Only seven of the studies satisfied both these criteria. A table of these studies is included in the Appendix 7. 2.3.2 Assessment of studies in the in-depth review To assess the quality of the studies at this stage of the project we moved away from the EPPI reviewer standard template for educational research and agreed instead to analyse the studies on the basis of the Maryland Scientific Methods Scale, which is designed specifically to communicate how studies 19

that evaluate the effects of criminological interventions differ in methodological quality. The Maryland Scale of Scientific Methods was developed by a research group at the University of Maryland’s Department of Criminology and Criminal Justice when they were selected by the National Institute of Justice to determine what juvenile crime prevention programmes work (Sherman et al, 1997). Studies are ranked from 1 (weakest) to 5 (strongest) on methodological quality. As an example, on this scale the cross-sectional design would score 1, the simple before-and-after design would score 2, and the randomised experiment would score 5 (the Maryland scale is explained in more detail in Section 4). An evaluation of the validity of the studies, taking into account factors such as selection bias, sample size and operational definitions was also included in analysis at the in-depth stage.

20

3.

IDENTIFYING AND DESCRIBING STUDIES: RESULTS

3.1 Studies included from searching and screening Searching of nine electronic bibliographical databases identified 4,051 studies. These studies were imported into the EPPI Reviewer database and 86 duplicates were removed. The exclusion criteria were applied to the remaining 3,965 studies based on the details provided in the database entry such as publication date and language, the title and abstract. This screening process resulted in the identification of 252 studies meeting the inclusion criteria. Of these 252 studies, we were able to obtain the full papers for 201 studies. Most studies were obtained either electronically or through library searches. These 201 studies were re-screened and 53 were included in the review. The review specific coding frame was applied to the included studies. Please refer to Appendix 6 for details of studies included in the systematic map. 3.2 Characteristics of the included studies (systematic map) This section of the report describes the characteristics of the 53 studies included in the review. The full papers of the studies were mapped using the review specific key wording tool (described in Section 2.2.4). The results of the key wording process were analysed using the EPPI Reviewer database reports function and are set out below in tables. The mapping is a largely descriptive exercise designed to characterise the included studies. The majority of the studies came from the United States, followed by the United Kingdom. Only two studies came from Australia and Canada. See Table 3.2.1 below for details. Table 3.2.1 Country of origin: where the research was carried out Country

Number of studies

United States

41

United Kingdom

9

Australia

1

Canada

1

Country not known

1

Based on N=53 mutually exclusive categories

The names of the programmes varied. No programme names were specified for just under half of the studies, leaving names specified for 29 of the studies. See Table 3.2.2 for some examples of the names of programmes.

21

Table 3.2.2 Examples of names of programmes reported by studies Programme names Model Employment Training MET Offender Job Linkage Programme STEEP Student Transition Education Employment Programme TRACS – Transitional Research on Adjudicated Youth in Community Settings A Youth Employment and Training Initiative (YETI) Fresh Start – a career development project CETA Comprehensive Employment Training Act programme APEX specialist employment agency Options for Learning Scheme DTAP – Drug Treatment Alternative Programme

3.2.1 Characteristics of the programmes The types of programmes reported by the studies were categorised. See Table 3.2.1.1 below for details. Vocational programmes were the most common, closely followed by educational programmes. Some studies reported programmes that came into more than one category. Just over half of the studies reported types of intervention that did not fit into any of the predesignated categories. See Table 3.2.1.2 for examples of programmes that did not fit into an existing category. Table 3.2.1.1 Type of intervention reported by studies Type of programme Number of studies Vocational programme

30

Other (please specify)

27

Educational programme

27

Career guidance programme

18

Substance misuse awareness programme 9 General offender programme

6

Cognitive programmes

3

Statutory programme

2

Not specified

2

Unclear whether the study evaluates an intervention programme

1

Based on N=53 non mutually exclusive categories

22

Table 3.2.1.2 Types of programme not in a pre-designated category Types of intervention specified Parenting fatherhood and employment Work assignments in House of Correction Pre-professional social work education Industrial therapy – simulated work environment Specialist employment agency – APEX Industrial work within prison apprenticeship Job fairs – teleconferencing between offenders and prospective employers Sex offender programmes (accredited) A resource centre plus ‘on the job’ training Intensive pre-release support, including counselling, and training in creative problem solving It was difficult to map the duration of interventions with 35 of the studies not specifying how long participants attended. Eleven of the studies were longterm (more than 12 weeks) and one was mid-term (between one week and up to 12 weeks). Six studies specified a different duration. In two cases the duration was presented in average hours: 180 and 50 hours. For the other four studies the duration of the programme varied but an average was not specified. This was because many programmes were made up of a variety of components and participants may have attended what was relevant to them. It was not possible to categorise the intensity of the programmes as 45 of the studies did not specify how frequently participants attended. The one study that did specify intensity specified five days a week attendance. The remaining seven studies were categorised as ‘Other’. In a similar way to duration, three studies reported that the intensity varied. Of these three studies one went on to say that the programme was designed to be high intensity. Two studies were very specific: 500 hours of treatment was reported over a nine-month period and an average of 604 hours during the incarceration period. Just over half the studies stated that programmes took place prior to release from custody. The second most common time for programmes to take place (just under a quarter) was both prior to and post-release from custody. Six studies reported programmes that took place post-release only. In four cases, programmes took place at other points; this included when an offender had been given a community sentence only, at pre-trial or within other services such as substances misuse provisions or in a psychiatric hospital. See Table 3.2.1.3 for details of programme timing.

23

Table 3.2.1.3 Programme timing: the point at which programmes took place Programme timing Number of studies Prior to release

26

Prior to and post release

12

Post release

6

Other – please specify

5

Not specified

4

Based on N=53 mutually exclusive categories

3.2.2 Characteristics of the participants There are various types of offender targeted by programmes. Please see Table 3.2.21 below for details. By a very small margin the largest identified type of offender was adults over 21. A quarter of the studies reported programmes that were aimed at juveniles or young offenders under age 21 followed. Other identified categories included substances misusing offenders, ex-offenders, female offenders, sex offenders and mentally challenged offenders. Programmes could be targeted at more than one type of offender, ie young substance misusing offenders. Just over a quarter of studies specified programmes aimed at other types of offender. Please see Table 3.2.2.1 for details of other types of offenders targeted by programmes. Table 3.2.2.1 Type of offender targeted by programmes Type of offender Number of studies Adult: over 21

15

Other

14

Juvenile/young: under 21s

13

Substance misuse related (drug or alcohol)

7

Ex-offender

5

Female offender

3

Sex offender

3

Mentally challenged

1

Not specified

16

Based on N=53 non mutually exclusive categories

24

Table 3.2.2.2 Other types of offender specified by studies Specified types of offender Specified types of offender In maximum security prison

With custodial sentences

Long v short criminal career

White collar criminals

High risk and in need of intervention

Violent offenders

Probationers/parolees

Sentences greater than 30 days

Awaiting trial – they are pre-trial.

Non-violent offenders

Although it was possible to examine the type of offender and this gives an indication of the age of participants, it was difficult to discern from studies the precise age range of participants. Thirty-two studies did not specify the age of the participants. A third of studies reported programmes focusing on over 21s, with eight studies focusing on young offenders aged 18–21 and four studies aimed at juveniles under 18. Again, there was a lack of information included about sex. Over half the studies did not specify whether the intervention was aimed at men or women. Twelve of the studies focused exclusively on men and five on women. Eleven studies reported programmes that looked at both men and women. In a similar way to age range and sex, precise information on ethnicity was limited. Thirty-six studies did not specify how or if participants were categorised by race. Eleven studies reported participants from mixed ethnic groups. Three studies reported white participants and five studies black participants. Four studies reported participants that formed a majority ethnic group such as Hispanic and four studies reported minority ethnic groups. 3.2.3 Employment focus of programmes and types of outcome reported Custody was where most of the employment programmes took place (see Table 3.2.3.1). Community was the second most common place for employment programmes to take place. Other settings included a therapeutic community, a halfway house and an inpatients forensic psychiatric institution. Programmes could take place in more than one environment, ie both custody and the community. Three studies did not report where the employment programme took place. Table 3.2.3.1 Environment where programmes took place Environment

Number of studies

Custody

34

Community

23

Other setting (please specify)

5

Not specified

3

Based on N=53 non mutually exclusive categories

25

The programmes focused on various aspects of employment (see Table 3.2.3.2). By a narrow margin the most common employment focus was on preparation for work (20 studies). Eighteen studies focused directly on fulltime employment and a quarter focused on part-time employment. Three studies reported programmes that were linked to supplemented work and only one study reported a programme that focused on apprenticeships. A quarter of studies did not specify a categorical employment focus. A further quarter of studies specified an employment focus that did not fit into a pre-designated category (see Table 3.2.3.3 for examples). Table 3.2.3.2 Aspect of employment focused on by programmes Employment focus

Number of studies

Preparation for work: interviewing techniques, job 20 applications, programmes that aim to do this Full-time employment

18

Part-time employment

13

Other

13

Supplemented work: e.g. linked to benefits such as working while on benefits, supported while working

3

Apprenticeship

1

Not specified

21

Based on N=53 non mutually exclusive categories

Table 3.2.3.3 Specified employment focus not in a pre-designated category Employment focus specified Charted young men's attitude towards the programme and their perceptions of where they would be in 12 months Make women offenders more aware of the benefits of working in maledominated occupations Work maturity skills development, job retention skills, career and vocational training, basic life skills, academic skills, career counselling Transition into role of paraprofessional social workers Some programmes focused on other issues such as housing or tackling poverty Help for up to one year after starting employment Necessary skills to get a job Examined accredited studies with a variety of focuses Related to post release community adjustment Voluntary work in the community

26

All the studies mapped reported some form of employment-related outcome (see Table 3.2.3.4). The two most common outcomes indicated were gaining employment (three quarters of studies) and not re-offending (37 studies). In many cases studies will have measured a combination of employment and recidivism outcomes. Eleven studies reported promoting positive attitudes to employment. Nine studies reported offenders holding employment for six months or more and seven studies examined offenders holding employment for up to 6 months. Levels of pay were reported by seven studies. In three cases, gaining unpaid full or part-time employment was reported. Eighteen studies reported other employment outcomes, examples of which are listed in Table 3.2.3.5. Table 3.2.3.4 Employment-related outcomes and re-offending Employment outcome reported Number of studies Gained employment: full or part-time 40 Not re-offended 37 Other 18 Promoted positive attitudes to employment 11 Held employment for six months or more 9 Held employment for up to six months 7 Levels of pay 7 Gained unpaid employment: full or part-time 3 Not specified 2 Job satisfaction 1 Based on N=53 non mutually exclusive categories

Table 3.2.3.5 Employment-related outcomes specified Employment-related outcomes specified Accessed other community services Achieved programme targets Less rebellious and more 'responsible' Reduction of drug/alcohol use: effect on depression, tension, and socialcompetence Better re-socialisation and enhanced social skills plus better reading and maths levels No longer homeless Looked at monetary cost of the Programme in relation to effectiveness Additionally showed higher rates of self-esteem and committed less inhouse infractions plus higher educational levels Earned higher wages than control group; successful completion of community supervision

27

3.2.4 Types of studies Three quarters of the reports included in the review were of single studies, but just over a quarter of the papers reviewed more than one study. Single studies were conducted in various ways. The most common design (20 studies) was looking at outcomes for a single group of participants who had attended a programme. In this case no comparisons are made. A third of studies included a comparison group. This means that the outcomes for participants attending an intervention are compared with a similar group of offenders who did not attend the intervention. Studies with a comparison group have a higher degree of certainty that it is the intervention that has caused the outcomes reported. One study reported that they had used a Randomised Control Trial which not only uses a comparison group, but also randomly allocates participants to the groups. This increases the reliability of the findings. It was not possible to discern the design for six studies. Nine studies specified different designs which included qualitative methods such as case studies. A quarter of studies included qualitative data as well as using a quantitative design. Table 3.2.4.1 Design of studies Design of the study Number of studies Single group (intervention outcomes for one group 20 only are measured, no control group) Group plus control comparison (there is a comparison between outcomes for the targeted group and another 18 group which did not have the same intervention) Studies including qualitative data 18 Other design (please specify) 9 Design not specified 6 Randomised control trial (participants for target and 1 control groups are randomly selected) Based on N=53 non mutually exclusive categories

In addition to the type of methodology used, the number of participants taking part in the studies was mapped (see Table 3.2.4.2). Overall, the number of participants reported was fairly high. Just over a quarter of studies reported that they had a sample size of over 201. Eleven studies reported that they had a sample size of over 1,000. Nine studies had between 11 and 100 participants. Two studies had up to ten participants and one study had between 101 and 200 participants. Sixteen studies did not specify numbers of offenders taking part.

28

Table 3.2.4.2 Sample size Sample size 201–999 1000+ 11–100 0–10 101–200 Not specified

Number of studies 14 11 9 2 1 16

Based on N=53 mutually exclusive categories

A quarter of studies also provided contextual information about the labour market. Information about the other factors such as the relationship between offenders as supervisors was reported by 12 studies. Finding from these studies are discussed in the wider review (section 5). 3.3 Speaking to practitioners The research team drew upon informal chats with colleagues working with offenders in practice. This is not systematic data and was compiled from adhoc conversations. Three themes were identified: Q1 What helps offenders get a job and stay in it? Lessons in interview techniques and obtaining skills that are in demand from employers and that are at a level where pay equals the skills. These jobs are usually vocational skills with IT and communication skills. Q2 What stops prisoners getting and keeping employment? Not being able to earn a living that pays above the level of benefits or not gaining skills that never go out of fashion such as plastering, bricklaying. There is also the question of rehabilitation: 'keeping a job means that in addition to being skilled, you have to be honest, reliable and able to get on with people'. Offenders may also lack a work ethic as many offenders are in low pay jobs and with relatively poor prospects. Q3 What programmes do you think would help people get and keep a job? Practical courses like fork lift truck courses, painting and bricklaying, and communication skills which are important as people with skills lose their jobs 'cos they think they 'don’t fit in'.

29

4.

IN-DEPTH REVIEW RESULTS

4.1 Identifying studies for in-depth review For this programme of the review we used a narrower set of inclusion criteria, identifying a group of studies to review in-depth. Our focus at this stage was on the design of the study. In order to answer our research question ‘What interventions promote employment in offender populations?’, we wanted to look at studies which investigate a causal connection between interventions and post release employment levels. The randomised control experiment is usually considered to be the strongest experimental design for testing causal hypotheses because it makes possible the exclusion of the largest number of plausible alternative hypotheses. However criminal justice research often precludes, for legal and ethical reasons, randomisation for selection of experimental and control groups. Many of the studies in our review describe why, in the author’s opinion, a randomised control experiment in the context of interventions for offenders would be unsuitable. Only one study of the 53 in the main review employed a random assignment of Section and control conditions. The largest group of studies in our main review (20) describe the effects of intervention on a single group, reporting outcomes in percentages of that group who found employment after the intervention, but do not compare these outcomes with those of offenders who do not receive the intervention. Nevertheless, in investigating the question ‘Does participation in a particular intervention lead to increased prospects of employment?’ some kind of control is needed in order to estimate what would have happened to the population if the intervention had not been applied to them. Criteria for inclusion in the in-depth review We agreed that the minimum research design considered adequate for drawing conclusions about impact evaluation should include a comparison of outcomes from at least two groups, one with and one without the intervention programme. Therefore only studies which included more than one group were considered. Our second criterion for inclusion in the in-depth review concerns the postintervention measurement period. In discussion with DfES representatives and research colleagues, concern was expressed that any measure of employment should continue for a reasonable length of time, and not merely be a record of a one-off job offer. We agreed on a minimum follow-up period of three months, and therefore excluded studies which did not provide employment data for at least this length of time. Implementing these criteria resulted in a short list of evaluations of seven intervention programmes (Appendix 7) whose design includes both a comparison group of inmates who did not participate in the intervention, and 30

longitudinal data, which examines whether offenders maintain employment for a period of at least three months post release. In some cases more than one study evaluates the same programme. Menon et al (1995) and Finn (1998a) both provide an evaluation of Project Rio and Finn discusses the statistics from Menon’s study. Robinson (2000) uses statistics from Fowles et al (1997) in the evaluation of Project Horizon. Saylor and Gaes write about PREP in two separate articles (1992 and 1997). In all these cases we have drawn on information from both publications. 4.2 Evaluation using the Maryland Scientific Methods Scale The Maryland Scale of Scientific Methods was developed by a research group at the University of Maryland’s Department of Criminology and Criminal Justice when they were selected by the National Institute of Justice to determine what juvenile crime prevention programmes work. The studies in our review were ranked from 1 (weakest) to 5 (strongest) on overall internal validity, according to the criteria described by Sherman et al (1997). 4.2.1 Description of Maryland Scale (Sherman et al, 1997) Level 1: Correlation between a crime prevention programme and a measure of crime or crime risk factors at a single point in time Level 2: Temporal sequence between the programme and the crime or risk outcome clearly observed; the presence of a comparison group without demonstrated comparability to the treatment group Level 3: A comparison between two or more comparable units of analysis, one with and one without the programme Level 4: Comparison between multiple units with and without the programme, controlling for other factors, or using comparison units that evidence only minor differences Level 5: Random assignment and analysis of comparable units to programme and comparison groups The differences between levels 2, 3, and 4 can be further clarified by analysing the amount of information available in the studies about the comparison group(s). In a level 2 study there will be very little or no information on the pre-treatment equivalence of the groups and little attempt to control for obvious differences between the groups (for example, a comparison between those who completed and those who dropped out of a programme). In a level 3 study there will be extensive information provided on pretreatment equivalence of groups, and groups may be matched on a variety of 31

variables such as race, gender, education, etc. However some obvious group differences on important variables which are likely to affect the outcome (such as motivation, for example) may remain. In a level 4 study there will be extensive information provided on pretreatment equivalence of groups and only minor group differences will be evident. Table 4.2.1.1 Studies in in-depth review Study Programme Maryland Description of intervention name Scale Rating Drake et al Class 1 3 Employment programme – private (2003) Employment businesses provide equipment, supervision and training For offenders employed during incarceration Hiller et al Step-down 2 Substance misuse awareness (1996) programme programme – life skills training (educational, recreational, and vocational therapy) For high risk probationers Menon et al Project Rio 4 Employment preparation programme (1995); (Reintegration – career guidance and employment Finn of exservices (1998a) Offenders) Provided to releasees Robinson Project Horizon Not ranked Employment preparation programme (2000); – educational classes and career Fowles guidance, counselling, support with (1997) finding and attending interviews For pre-parole offenders Saylor and Post release 4 Work experience programme – Gaes (1992 employment training for and experience of and 1997) project (PREP) industrial work Within prison Schumacker No programme 3 Vocational and academic training et al (1990) name For prison inmates Steurer and No programme 4 Correctional Education Programme – Smith name may have included many (2001) programmes but little detail provided For prison inmates

4.2.2 Evaluation of the studies 1. Drake et al (2003) Class 1 Impacts: Work during incarceration and its effects on post-prison employment patterns and recidivism Type of publication 32

This is a published report for Washington State Department of Correction Planning and Research Programme. Description of intervention This study focuses on offenders who had participated in Class 1 employment during their incarceration. Class 1 jobs are those in which a contract exists between the Department of Correction and a private sector business. The contracted business provides the equipment and manages, supervises and trains offenders who are employed by the business. Sample The test group was drawn from a population of offenders who were involved in Class 1 employment at some point between 1992 and 1995, and were released from prison between 1992 and 1997. This group included 366 males and 88 females. The comparison group included 323 males and 80 females. Presence of comparison groups The comparison group consisted of offenders released at the same time as the test group who were not involved in a Class 1 industry during incarceration. This comparison group was drawn to match the test group on a range of variables: sex, offence type, age, race and number of years incarcerated. However, although matched on several variables, important differences remained between the test group and the comparison group in this study due to the basis of initial selection for Class 1 employment. The study describes how in order to be selected for Class 1 industry the offender would typically have to meet the following criteria: ‘the offender is of the appropriate classification level, has the mental and physical health status required for the position, has been serious-infraction-free for the past six months, has the general skills necessary for the job they are applying for and has worked in an institutional support job for at least three months.’ It seems likely that this list of criteria would distinguish the Class 1 group from the non-participating comparison group in ways which might also contribute to employment outcomes. Indeed, in spite of attempts to match the groups on demographic variables, difference in education levels were shown to be significant between the two groups. Post-treatment measurement period The employment records were tracked for one year after release. Recidivism (a return to a Washington State adult correctional facility) was tracked for three years after release. Outcomes Outcomes measured in the study include infractions while incarcerated, postprison employment rates, post-prison wages earned, and recidivism. Offenders who participated in Class 1 industries had statistically higher rates of employment than offenders who did not participate. Approximately 69% of 33

the test group obtained employment within one year of release compared to 58% of the comparison group. Class 1 participants also earned higher wages within one year of release. There was no significant difference between groups in the number of infractions while incarcerated. Class 1 participants reoffended at a lower rate (17%) than the control group (24%). Class 1 participants who were employed upon release had a lower recidivism rate than Class 1 participants who were unemployed upon release. Other findings from this study indicated that offenders with an education (in both groups) were more likely to be employed and that there is a relationship between age and post-prison outcomes in both the test and comparison groups, with lower employment rates for older offenders. Rating This study was given a Maryland rating of 3. Although there was an attempt to control for other factors in the comparison group, obvious group differences on important variables exist between the two groups. 2. Hiller et al (1996) Post-treatment outcomes for substance-abusing probationers mandated to residential treatment Type of publication The article by Hiller et al (1996) was published in the Journal of Psychoactive Drugs, vol. 28. Description of the intervention The study is an evaluation of a community-based substance abuse treatment programme for probationers. It describes a specialised after-care treatment called Step Down, which is part of the Dallas County Jails Judicial Treatment Centre residential treatment. The programme provides up to six months of primary substance abuse treatment as an alternative to incarceration for highrisk probationers. Treatment provides life-skills training (educational, recreational and vocational therapy) as well as drug education and group therapy. Sample The sample consisted of 492 probationers who were mandated by the courts to complete residential treatment programmes. Presence of comparison groups Two sets of comparison were conducted. Firstly, graduates of the programme were compared with expulsion and transfers from the programme on a measure of arrests during the six months following treatment. Secondly, probationers who completed the Step Down programme were compared with those who only attended the outpatient aftercare on measures of six-month post-treatment substance use, employment and arrests. Post-treatment measurement period Six months after treatment 34

Outcomes The study found a significantly larger proportion of the transfer (27%) and expulsion (13%) groups were re-arrested during the six-month follow-up interval than programme graduates (6%). Those who completed the Step Down programme had higher rates of employment (72%) compared with 60% employment for those who completed the primary treatment programme but who had less service-oriented aftercare. Rating As this study does not establish equivalence between the comparison groups it is difficult to infer that the lower recidivism levels of those who completed the programme are due to the effect of the intervention. It is likely that programme completers would have a lower recidivism rate than those expelled or transferred from the programme regardless of treatment. Both groups in the Step Down programme had graduated from the initial programme so might be more equivalent. However the sample size is small for the Step Down programme (a study group of 76% of the 351 graduates compared to a control group of 24% of the graduates). Comparison of employment levels between these two groups is reported as significant at p