Cognitive-Behavioural Interventions for Preventing Youth Gang Involvement for Children and Young People (7-16)

Abstract

Research indicates that youth who join gangs are more likely to be involved in delinquency and crime, particularly serious and violent offences, compared to non-gang youth and non-gang delinquent youth. Research also has found that both delinquent youth and youth who join gangs often show a range of negative thoughts, feelings and beliefs compared to non-delinquent peers. Cognitive-behavioural interventions, designed to address these deficits, have had a positive impact on a variety of behavioural and psychological disorders among children and youth.

This systematic review was designed to assess the effectiveness of such cognitive behavioural interventions for preventing youth gang involvement. A three-part search strategy found no randomised controlled trials or quasi-randomised controlled trials of the effectiveness of cognitive-behavioural interventions for gang prevention; four excluded studies examining the impact of Gang Resistance Education and Training (GREAT) were of too poor a quality to be included in analysis. The only possible conclusions from this review, therefore, are the urgent need for additional primary evaluations of cognitive-behavioural interventions for gang prevention and the importance of high standards required of the research conducted to provide meaningful findings that can guide future programmes and policies.

Abstract

Background

Many studies document a robust and consistent relationship between gang membership and elevated delinquency, with gang members disproportionately involved in crime compared to non-gang peers. Research also indicates that both delinquent youth and youth who join gangs often show a wide range of deficient or distorted social-cognitive processes compared to non-delinquent peers. Cognitive-behavioural interventions are designed to address cognitive deficits in order to reduce maladaptive or dysfunctional behaviour, and studies have documented their positive impact on a number of behavioural and psychological disorders among children and youth.

Objectives

To determine the effectiveness of cognitive-behavioural interventions for preventing youth gang involvement for children and young people (ages 7-16).

Search strategy

Electronic searches of ASSIA, CINAHL, CJA, Cochrane Library, Dissertations Abstracts A, EMBASE, ERIC, IBSS, LILACs, LexisNexis Butterworths, MEDLINE, NCJR Service Abstracts Database, PsycINFO, and Sociological Abstracts, to April 2007. Reviewers contacted relevant organisations, individuals, and list-servs and searched pertinent websites and reference lists.

Selection criteria

All randomised controlled trials or quasi-randomised controlled trials of interventions with a cognitive-behavioural intervention as the majority component, delivered to youth and children aged 7-16 not involved in a gang.

Data collection & analysis

Searching yielded 2,284 unduplicated citations, 2,271 of which were excluded as irrelevant based on title and abstract. One was excluded following personal communication with investigators. One citation, of a large randomised prevention trial, awaits assessment; personal communication with study authors yielded unpublished reports addressing gang outcomes, but insufficient detail precluded determining inclusion status. Seven remaining reports were excluded as irrelevant because they were narrative reviews or descriptions of programs without evaluations, did not address a gang prevention programme, or did not address a gang prevention program that included a cognitive-behavioural intervention. The remaining four full-text reports excluded because of study design, leading to 0 included studies.

Main results

No randomised controlled trials or quasi-randomised controlled trials were identified.

Reviewers’ conclusions

No evidence from randomised controlled trials or quasi-randomised controlled trials exists regarding the effectiveness of cognitive-behavioural interventions for gang prevention. Four evaluations of Gang Resistance Education and Training (GREAT) have been conducted, two of which were part of a US national evaluation, but all were excluded based on study design. Reviewers conclude there is an urgent need for rigorous primary evaluations of cognitive-behavioural interventions for gang prevention to develop this research field and guide future gang prevention programmes and policies.

Synopsis

Research indicates that youth who join gangs are more likely to be involved in delinquency and crime, particularly serious and violent offences, compared to non-gang youth and non-gang delinquent youth. Research also has found that both delinquent youth and youth who join gangs often show a range of negative thoughts, feelings and beliefs compared to non-delinquent peers. Cognitive-behavioural interventions, designed to address these deficits, have had a positive impact on a variety of behavioural and psychological disorders among children and youth. This systematic review was designed to assess the effectiveness of such cognitive-behavioural interventions for preventing youth gang involvement. A three-part search strategy found no randomised controlled trials or quasi-randomised controlled trials of the effectiveness of cognitive-behavioural interventions for gang prevention; four excluded studies examining the impact of Gang Resistance Education and Training (GREAT) were of too poor a quality to be included in analysis. The only possible conclusions from this review, therefore, are the urgent need for additional primary evaluations of cognitive-behavioural interventions for gang prevention and the importance of high standards required of the research conducted to provide meaningful findings that can guide future programmes and policies.

Background

Definition of a youth gang

There is no unanimously accepted definition for a youth gang, reflecting the reality that there is no universal model of a youth gang. Several characteristics, however, typically distinguish youth gangs from other youth groups or organized crime groups, primarily: participation in criminal activity, typically engaging in a range of criminal offences; and projection of a shared identity, through naming, symbols, colours, or association with physical or economic territory ( Huff 1993 ; Spergel 1993 ; Spergel 1994 ; Howell 1998 ; Esbensen 2000 ; White 2002 ; OJJDP 2004 ; Carlsson 2005 ). Most definitions of youth gangs refer to these two characteristics in some way, but also often include specific group characteristics or gang organizational structures that vary across regions. Most gang members in the United States and internationally are adolescents with the peak age of recruitment into gangs and increased criminal involvement between eleven and fifteen in the United States ( Kodluboy 1993 ; Huff 1998 ; Hill 2001 ; OJJDP 2004 ). Surveys conducted in the United States also find that most gang members are from ethnic or racial minorities, predominately Hispanic and black/African-American, and the vast majority are male, although the percentage of non-minority gang members and female gang members varies substantially with jurisdiction type and between self-report and law enforcement data ( Moore 1998 ; Moore 1999 ; Egley 2000 ; Egley 2006 ; Snyder 2006 ). The definition of a gang used in this review, based on those from the Eurogang Program of Research and the Office of Juvenile Justice and Delinquency Prevention ( OJJDP 2004 ), is intended to accommodate this diversity of gangs, whilst recognizing their common general attributes. A youth gang is defined as ‘any durable, street-oriented youth group whose involvement in illegal activity is part of their group identity’ ( Esbensen 2005 ), excluding prison gangs, ideological gangs, hate groups, and motorcycle gangs ( Huff 1993 ; Howell 1998 ).

International prevalence of youth gangs

Most of the research into youth gangs has been conducted in the United States, where the number of active gangs peaked in the mid-1990s with more than 30,000 gangs and 840,000 gang members nationwide ( Moore 1998 ; Snyder 2006 ). The most recent data estimate that there are about 24,000 gangs and 760,000 active gang members across the United States ( Snyder 2006 ). Internationally, gang research has only begun to occur, but street gangs or what are sometimes called ‘troublesome youth groups’ ( Decker 2005 ) have been identified in developed and developing countries in South America, Europe, Asia, and Africa ( Covey 2003 ; Carlsson 2005 ; Decker 2005 ; Papachristos 2005 ; Klein 2006 ). Preliminary examinations have concluded that the youth gangs in several of these countries mirror the nature, pattern of emergence, and behaviour of the more extensively studied American street gangs ( Klein 1995 ).

Cognitive-behavioural theories of delinquency

Research indicates that delinquent children and young people often show a range of deficient or distorted social-cognitive processes compared to non-delinquent peers, such as perception of social cues (encoding and representation), selection of solutions to social dilemmas, aggression management, self-control, locus of control, long-term planning, outcome expectations, self-perceptions, empathy and role-taking, and moral reasoning ( Gibbs 1993 ; Goldstein 1993 ; Hollin 1993 ; Lochman 1994 ; Lipsey 2006 ). Cognitive development may begin to influence the ability to control social behaviour as early as the first two years of life ( Wasserman 2003 ) and research has shown that different types of social-cognitive distortions can uniquely contribute to the variance among groups of violent and moderately aggressive children ( Lochman 1994 ). Several social-cognitive deficits or distortions also have been identified as risk-factors for youth gang involvement, such as lack of refusal skills, social disabilities, deviant attitudes, a fatalistic view of the world, and positive attitudes towards antisocial behaviour or gang membership ( Howell 1998 ; Maxson 1998 ; Hill 1999 ; OJJDP 2004 ). Some theories of youth gang involvement emphasize these cognitive mechanisms and other individual risk or protective factors, whilst others adopt a more macro approach, drawing on social disorganization, social control, or strain theory.

Cognitive-behavioural interventions

Cognitive-behavioural interventions are designed to address these cognitive deficits and learning patterns in order to reduce maladaptive or dysfunctional behaviour ( Connor 2002 ; DOH 2001 ; Farrington 2002 ; Lipsey 2006 ; Turner 2007 ). They are skill-based, combining cognitive and behavioural therapy approaches from cognitive and learning theories of delinquency and behaviour change ( DOH 2001 ). Common techniques include anger management, empathy, social perspective taking, lateral thinking, critical thinking, problem solving, self-control, self-instruction, life skills development, goal setting, moral reasoning, social-information processing, and social skills training ( Gibbs 1993 ; Goldstein 1993 ; Hollin 1993 ; Ribisl 1993 ; Stephens 1993 ; McGuire 2000 ; Connor 2002 ; Farrington 2002 ; Andreassen 2006 ; Lipsey 2006 ). Additionally, cognitive-behavioural interventions often include role-play or rehearsal in real situations to consolidate new skills and cognitive processes and may be combined with other intervention strategies, such as recreational interventions or mentoring, in multi-component intervention or prevention programmes ( Connor 2002 ; Lipsey 2006 ).

Research indicates that cognitive-behavioural interventions can reduce delinquent and antisocial behaviour among children and youth ( Connor 2002 ; Farrington 2002 ). Several meta-analyses have found that cognitive-behavioural programmes are effective for reducing recidivism of juvenile and adult offenders ( Lipsey 2001 ; Pearson 2002 ; Landenberger 2005 ; Wilson 2005 ). Additionally, research over the past twenty years has shown that cognitive-behavioural programmes can produce positive outcomes for a variety of behavioural and psychological disorders ( DOH 2001 ; Andreassen 2006 ; Turner 2007 ).

Potential of cognitive-behavioural interventions for gang prevention

Cognitive-behavioural interventions also may be effective in preventing youth gang involvement. Such an adaptation of a delinquency-prevention strategy to gang prevention is supported by the overlap between several identified risk-factors for delinquency and for gang involvement, specifically social-cognitive attributes ( Howell 1998 ; Maxson 1998 ; Hill 1999 ; OJJDP 2004 ). Previous involvement in delinquency, problem behaviour, or violence also has been identified as a strong predictor of later youth gang involvement in longitudinal analyses (ibid); it is possible that by reducing delinquent behaviour, cognitive-behavioural interventions may have a carry-over effect of reducing gang involvement or may impact youth gang involvement in unique ways, independent of any influence on delinquency. Finally, the demonstrated effectiveness of cognitive-behavioural interventions for reducing entrenched problem behaviour indicates the potential potency of the cognitive-behavioural model of behaviour change that may be applicable to youth gang prevention.

Studies of the developmental stages of youth and gang involvement suggest that cognitive-behavioural interventions for gang prevention may have the greatest potential for effectiveness when administered within late childhood and early adolescence, approximately between ages seven and sixteen. This is the period when young people demonstrate decreasing levels of supervision by parents and increasing independence in the community ( Dishion 1999 ). It includes the period of middle childhood when peer influence is increasingly salient and peer deviant behaviour begins to develop, which appears to be a precursor of involvement in youth gangs ( Howell 1998 ; Maxson 1998 ; Hill 1999 ; OJJDP 2004 ). It is in these early stages of peer group and gang development that young people may be most responsive to prevention programmes ( Kodluboy 1993 ; Huff 1998 ; Hill 2001 ; Connor 2002 ; OJJDP 2004 ).

Although narrative summaries of gang prevention programmes have emerged over the past fifteen years and meta-analyses of cognitive-behavioural interventions for reducing recidivism and other behavioural problems have been conducted, the effectiveness of cognitive-behavioural interventions for preventing youth gang involvement had never been systematically assessed. This review sought to address this important gap in the gang prevention research base and thereby enable practitioners and policy-makers to develop evidence-based interventions in response to a youth gang presence in their community.

Objectives

To assess the effectiveness of cognitive-behavioural interventions for preventing youth gang involvement for children and young people (7-16).

Criteria for considering studies for this review

Types of studies

Studies were eligible for inclusion if allocation to group was by random allocation or quasi-random allocation (for instance, by alphabetical order, by alternating sequence, or by day of the week).

Types of participants

Children and young people aged 7-16 not involved in a gang.

Types of interventions

Cognitive-behavioural interventions, as defined in Background.

Programmes combining cognitive-behavioural interventions with other interventions, such as recreational intervention or opportunities provision, were included only if cognitive-behavioural interventions were the majority intervention, i.e. more than 50% of total programming, based on frequency and duration as determined independently by all reviewers (HF, FG and PM). Study authors would have been contacted for more information if there had been any discrepancy between the review authors or if either had estimated that the proportion of cognitive-behavioural intervention programming was between 40% and 60%.

Multi-component intervention programmes that included cognitive-behavioural interventions but had opportunities provision as the majority intervention would have been excluded from this review and considered for inclusion in a separate review (Opportunities provision for preventing youth gang involvement for children and young people (7-16)).

Studies with any other intervention as the majority component were excluded.

The primary control comparison for cognitive-behavioural interventions was no intervention. Comparisons against other interventions, specifically designed for gang or delinquency prevention or other social services or support interventions being delivered to the control group, were included but would have been discussed separately.

Types of outcome measures

Primary outcomes include:

Primary outcomes included:

• 1)

Gang membership status (dichotomous); and

2)

Conviction for gang-related delinquent behaviour and criminal offences, including homicide, assault, robbery, burglary, and drug trafficking.

Secondary Outcomes included:

• 1)

Measures of behavioural, cognitive, or social skills, i.e. anger management, empathy, social perspective taking, problem solving, self-control, self-instruction, goal setting, moral reasoning, or social-information processing;

2)

Delinquent behaviour and criminal offences external to gang activities or committed by an individual not involved in a gang;

3)

Association with delinquent peers (measured through a peer delinquency scale, as a dichotomous variable, as a percentage of time spent with delinquent peers, or as a percentage of friends who are identified as delinquent);

4)

Objective and subjective measures of illegal drug abuse;

5)

Hospitalisation or injury due to a) gang-related activities, or b) delinquent activities, as determined by self-report or hospital record;

6)

Firearm possession (both conviction and self-report);

7)

School-reported truancy;

8)

Achievement of scholastic benchmarks for youth eighteen and under at outcome measurement; and

9)

Employment status for youth sixteen and older at outcome measurement.

Many of these measures have been selected as predictors of future criminality. As such, it is important that they are both reliable and valid, and therefore, instruments must meet minimum standards: i) the psychometric properties of the instrument should have been described in a book or peer-reviewed journal; and ii) the instrument should be either (a) a self report, or (b) completed by an independent rater or relative.

Instruments used to measure outcomes could have included self-report or official records, such as school, police, probation, or court data. When applicable, self- and other-reported outcome measures would have been analysed separately due to possible divergence, but would not have been ranked in terms of reliability ( Dishion 2005 ).

When available, behavioural and attitudinal measures of problem behaviour and related constructs, such as those in the National Evaluation of GREAT Student Questionnaire ( Esbensen 1999 ), would have been included.

Outcomes had to be reported in quantitative terms and include end point (post-intervention) data for both experimental and control groups.

Outcome intervals

Outcomes would have been measured post-intervention, after a short-term follow-up period up to 6 months, after a medium-term follow-up period up to 18 months, and after a long-term follow-up period up to 5 years, as data were available, to assess the durability of the intervention.

Search strategy for identification of studies

A three-part search strategy was undertaken in order to maximise chances of capturing all relevant literature.

I. Electronic search

Databases were searched for published and unpublished studies. No language restrictions were imposed on any results from any search attempts, although most databases were searched in English. No filters based on methodology were applied because test searches indicated that such filters might eliminate relevant studies. A highly sensitive search strategy (a search that was likely to capture all relevant reports) was used rather than a more specific one (a search that would have identified fewer irrelevant papers).

The following databases were searched electronically:

The Cochrane Library (Issue 2, 2007)

MEDLINE (1950 to April Week 3 2007)

ASSIA (1987 to April 2007)

CINAHL (1982 to April Week 4 2007)

Criminal Justice Abstracts (1968 to November 2007)

Dissertation Abstracts (1861 to April 2007)

EMBASE (1980 to 2007 Week 17)

ERIC (1966 to May 2007)

International Bibliography of Social Sciences (IBSS)(1951 to April Week 04 2007)

LexisNexis Butterworth Services (up to April 2007)

LILACS (up to April 2007)

National Criminal Justice Reference Service (up to October 2007)

PsycINFO (1806 to April Week 1 2007)

Sociological Abstracts (Earliest to 2007)

The search strategy used for the Cochrane Library (including The Cochrane Database of Systematic Reviews, Database of Abstracts of Reviews of Effects, The Cochrane Central Register of Controlled Trials, The Cochrane Methodology Register, Health Technology Assessment Database, NHS Economic Evaluation Database, and About The Cochrane Collaboration) was as follows:

[(MeSH descriptor Adolescent explode all trees) OR (youth OR adolescen* OR juvenile OR child OR schoolchild OR boy OR girl OR teen OR (young person*) OR (young people*)):ti,ab,kw]

AND

[(MeSH descriptor Juvenile Delinquency explode all trees) OR (gang OR delinquen* OR devian* OR (anti NEXT social) OR (youth* NEAR group)):ti,ab,kw]

AND

[(MeSH descriptor Cognitive Therapy explode all trees) OR ((cognitive NEAR/3 (therapy OR training)) OR (behavio*r NEAR/3 (therapy OR training OR modif*)) OR (skill NEAR/3 training)):ti,ab,kw]

Terms were modified as necessary for all other databases. See additional Table 01, Table 02, Table 03, Table 04, Table 05, Table 06, Table 06, Table 07, Table 08, Table 09, Table 10, and Table 11.

II. Personal communications

Appropriate government departments, non-governmental organisations, non-profit groups, advocacy groups, user groups, and experts in the field were contacted. Additionally, delinquency prevention and gang oriented email lists (list-servs) were sent a letter requesting assistance in locating studies.

The primary reviewer contacted authors of all included and excluded studies to request details of ongoing and unpublished studies.

III. Hand searching

Relevant websites, including those maintained by users, governments, other agencies, and academics and reference lists from previous reviews and all included and excluded studies were searched by the primary reviewer.

Methods of the review

No trials met inclusion criteria for this review. For information on methods planned in the protocol and archived for use in future updates of this review, please see Table 14.

Description of studies

The search strategy generated 2,284 unduplicated citations. HF and FG checked titles and abstracts for relevance and excluded 2,271 citations as clearly irrelevant. One citation appeared potentially relevant but personal communication with study authors indicated that it was excluded from analysis because it did not address a gang prevention programme with cognitive-behavioural component. Another citation, of a large randomised prevention trial with published reports on other delinquency-related outcomes, also appeared potentially relevant. Personal communication with study authors yielded unpublished reports addressing gang outcomes (conference, slide presentation; Lacourse 2003 ), but insufficient detail regarding the study and outcomes prevented determining the study's inclusion or exclusion. Consequently, this citation remains awaiting assessment.

The remaining 11 citations, which one or both reviewers felt might be relevant, were retrieved in full-text.

Both reviewers examined these full-text articles to determine eligibility and excluded 7 as clearly irrelevant, because: they were descriptions of programs or narrative reviews without evaluations (n=3), did not address a gang prevention programme (n=3), or did not address a gang prevention program that included a cognitive-behavioural intervention (n=1). The remaining 4 studies, all of which were evaluations of Gang Resistance Education and Training (GREAT) in the United States, were assessed for inclusion criteria. None qualified as a randomised or quasi-randomised study, and therefore all were excluded from analysis. Their methodology and findings are presented in the Excluded Studies table, the Description of Studies, and the Discussion.

There were zero included studies.

There were no disagreements between reviewers regarding study inclusion or exclusion. However, study authors would have been contacted if further information could have resolved initial disagreements about inclusion and the Review Group Coordinator of the CDPLPG would have been consulted if consensus could not have been reached.

A flowchart of the process of trial selection was made in accordance with the QUORUM statement ( Moher 1999 ). See Figure 01 .

Four studies were assessed for inclusion criteria. None qualified as a randomised or quasi-randomised study, and therefore all were excluded from analysis (please see Table of Excluded Studies).

All excluded studies were evaluations of Gang Resistance Education and Training (GREAT) in the United States. GREAT is a gang prevention programme based on a cognitive approach of behaviour change ( Esbensen 1999 ; Esbensen 2002 ; OJJDP 2004 ). Delivered in schools by uniformed law enforcement officers over nine weeks, the GREAT curriculum includes thirteen lessons designed to teach life skills and give students a sense of empowerment, competency, and usefulness to help them avoid involvement in youth violence and youth gangs ( Esbensen 1999 ; Lundman 2001 ; Esbensen 2002 ; OJJDP 2004 ).

The first excluded study ( Palumbo 1995 ) was a multi-site pre/post-test evaluation of the impact of GREAT among 2,029 students who received the intervention. It was excluded from analysis because the study did not have a comparison group, precluding assessment of potential effects on outcome measures from factors other than the intervention, such as a maturation effect. The study also had no follow-up period, which prevented assessment of the durability of programme effects or the presence of any lag-effects, and the study's exclusively attitudinal outcome measures may not have assessed what they intended to, i.e. students’ reported desire to become a gang member may not reflect their ultimate gang membership.

The second excluded study ( Ramsey 2003 ) used a non-randomised cohort design to assess the impact of GREAT for 7th grade (n=274) students who had received the intervention, compared to 8th grade students (n=148) from the same school who had not received GREAT. The study was excluded because students were not randomised to condition but allocated as a group based on year in school. Additionally, there was no baseline assessment of group characteristics, impeding appraisal of potential group comparability and selection bias. Other methodological limitations include: minimal geographic diversity due to all students coming from one school, which compromised the generalizability of findings; no measure of implementation fidelity or intervention components actually received by participants; and outcomes focusing exclusively on attitudinal change with no behavioural measures.

The two other excluded studies were part of the National Evaluation of GREAT, funded by the United States Bureau of Alcohol, Tobacco and Firearms and the National Institute of Justice. One study ( Esbensen 1999 ) was a cross-sectional study across eleven cities of 5,935 8th grade students. The study created two ex-post facto comparison groups (students who had participated in GREAT the year before and students who had not). Strengths of this study included its geographic diversity and use of dummy variable sets and progressively restricted samples to control for potential differences among individuals and schools that might bias assessment of programme effect. A fundamental limitation of this study and its reason for exclusion from analysis was that students were not randomised to condition but predetermined based on classroom membership. As the study authors discussed, the tracking system common in United States’ schools, in which students are divided into classrooms based on aptitude, can produce substantial heterogeneity across classes that could have created statistically significant differences between comparison groups if only a few classrooms from each school were sampled. This potential for selection bias was compounded by the lack of a baseline assessment of group comparability. This study also suffered from the standard limitations associated with public school surveys-i.e. exclusion of private school, truant, or absent students that could create a population bias-and it did not include a behavioural outcome measure to directly assess the intervention's intended long-term impact.

The final excluded study and the second part of the National Evaluation of GREAT ( Esbensen 2001 ) was a four-year longitudinal analysis of 3,568 students from 153 classrooms, 22 schools, and 6 cities across the United States. Allocation to condition occurred on the classroom level and was conducted through an unspecified ‘random process’ in 15 out of 22 schools. In the other 7 schools, allocation was purposive, due to officer availability or school district restrictions. Results were not available for individual schools, nor could study authors recall or produce documentation as to which schools were randomised and what specific processes were used (communication with study authors via e-mail; message sent by primary reviewer (HF) on 1 June 2007; response from Finn A. Esbensen on 11 June 2007). As such, the study's unclear and at best only partially-randomised allocation process, excluded it from analysis.

Methodological quality of included studies

No randomised controlled trials or quasi-randomised controlled trials were found that fulfilled the inclusion criteria.

Results

No randomised controlled trials or quasi-randomised controlled trials were found that fulfilled the inclusion criteria.

Discussion

This review found no evidence from randomised controlled trials or quasi-randomised controlled trials regarding the effectiveness or ineffectiveness of cognitive-behavioural interventions for gang prevention. Four excluded studies examining Gang Resistance Education and Training (GREAT), found mixed but generally weak indications of programme effect: one case study of students receiving the intervention found a slight, non-significant impact from pre-test to post-test on attitudinal measures, generally in the direction of positive programme effect ( Palumbo 1995 ); a cohort study found increased gang resistance attitudes from pre-test to post-test among both the intervention and the control group with no differential effect of GREAT participation ( Ramsey 2003 ); a cross-sectional survey comparing youth who had received GREAT to those who had not found statistically significant differences favouring GREAT participation in nine out of twenty-three peer-group and attitudinal measures and two out of eight behavioural measures (drug use and minor delinquent offences) ( Esbensen 1999 ); and a longitudinal cohort study similarly comparing youth who had received GREAT to those who had not found statistically significant differences favouring GREAT participation on five out of thirty-two attitudinal and behavioural measures, including risk-seeking behaviour and victimization ( Esbensen 2001 ) Only two of the studies measured gang membership ( Esbensen 1999 ; Esbensen 2001 ) and neither found a statistically significant effect from GREAT participation. However, as discussed in the Description of Studies, study design excluded all of these studies from analysis. Therefore, based on the findings of this systematic review, it is impossible to reach any conclusions regarding the effectiveness or ineffectiveness of cognitive-behavioural interventions for preventing youth gang involvement.

Reviewers’ conclusions

Implications for practice

The absence of any randomised controlled trials or quasi-randomised controlled trials of cognitive-behavioural interventions for gang prevention found by this extremely sensitive search strategy makes it difficult to advise practitioners as to future intervention and policy efforts. Four excluded studies evaluating Gang Resistance Education and Training (GREAT), a cognitive-behavioural school-based prevention programme, suggest that there may be only a marginal potential positive impact of such a cognitive-behavioural intervention. However, methodological weaknesses in these excluded studies’ designs preclude any definitive conclusions regarding programme effectiveness, ineffectiveness, or harm. As such, the main recommendation for future practice is to demand rigorous primary evaluations that include gang-related outcomes for any existing or developing cognitive-behavioural prevention programmes. Such rigorous evaluations are urgently needed to develop this research field and guide future funding and intervention profiles.

Implications for research

Had any of the four excluded studies been conducted with a more rigorous study design that included randomised or quasi-randomised allocation to condition, their results would have been eligible for inclusion in this systematic review. Had authors of the longitudinal study ( Esbensen 2001 ) specified allocation processes for each city and/or school, portions of their study also may have been eligible for inclusion. The potential impact of these evaluations was inherently restricted because of their study design. The paucity of research and the insufficient attention to methodological rigour in conducting and funding these evaluations, two of which were executed under the auspices of the United States Bureau of Alcohol, Tobacco, and Firearms and the National Institute of Justice, are unacceptable. Researchers, funding institutions, and policy makers all must demand higher standards of social research to succeed in reducing youth gang involvement and the associated crime and delinquency.