Psychosocial Interventions Preventing Gang-Related Crime Among Young People: A Systematic Review
Source:
Abstract
The objective was to assess the effectiveness of psychosocial interventions in preventing gang membership and gang-related crime among children and young adults under the age of 30. We performed a systematic review and synthesized interventions targeting universal, selective, and indicated populations published between January 2000 and April 2023. We included 42 (seven randomized, 12 nonrandomized, 23 controlled interrupted time series) studies evaluating 33 unique psychosocial interventions. Synthesis without meta-analysis found a preventive effect of psychosocial interventions in middle schools on gang membership. Furthermore, meta-analysis found that focused deterrence strategies prevented gang-involved violence, and that psychosocial support during probation decreased crime recidivism. This systematic review found significant effects of four psychosocial interventions compared to control in reducing future criminality, especially gun violence, among children and young adults. The findings are discussed regarding policy implications and ethical considerations.
Discussion and Applications to Practice
This systematic review identified four psychosocial interventions that prevent gang membership and gang-related crime, thus increasing the number of evidence-based interventions. Furthermore, two psychosocial interventions did not receive empirical support, which underlines the importance of controlled research to avoid ineffective or even harmful interventions (e.g., Petrosino et al., 2013; Welsh & Rocque, 2014).
The first of the effective interventions include universal interventions in middle school (G.R.E.A.T., Communities That Care, Keepin’ it REAL). The studies indicated that the three interventions decreased the risk of gang affiliation after 14, 48, and 72 months, respectively. It is worth noting that the two studies with the longest follow-up produced the same effect size (OR = 0.76), while the one with 14 months follow-up was somewhat stronger (OR = 0.66).
Of the four indicated interventions, three produced significant effects. The first, focused deterrence, based on 13 studies, reduced gun violence by 26% over an average of 37 months. This result parallels that of the meta-analyses of Braga and colleagues (Braga et al., 2018). Their review covered research up to 2015 with similar research questions as ours. Their meta-analysis demonstrated a significant effect of the intervention, but also with variation between different studies, like ours. It was not considered appropriate to compare ours and Braga and colleagues’ effect sizes because of different measures (Wilson, 2022).
The second indicated intervention is drug market intervention (DMI). Five studies have evaluated the effect of firearm violence, but without providing a statistically reliable result. This is consistent with a meta-analysis which showed that increased police efforts against drug markets lead to increased violence and more murders (Werb et al., 2011). One explanation for this paradoxical result is that removing leading people in the drug trade opens the way for other individuals and gangs to take over the void that has arisen. To maintain control against others who try to establish themselves requires violence. In the previously referenced review of focused deterrence (Braga et al., 2018), DMI had a weak effect bordering on being statistically significant (p < .046). The difference between this and our results might be because the previous review was based on a mixture of different types of crime (e.g., drug crimes, theft, assault, and homicide), while our review included only gun violence.
Although DMI did not affect gun violence, it reduced drug crime by 35% over an average of 18 months. According to Braga et al. (2018), research indicates no displacement and diffusion effects on criminal behavior to other locations. This is consistent with place-based interventions against crime more often leading to crime reductions in nontreated nearby locations rather than displacing crime there (Braga et al., 2019).
The third indicated intervention focused on correctional services for youth or young adults belonging to gangs or with a history of serious violent behavior. Five studies described the effect of psychosocial interventions during probation or during probation and prison. Three meta-analyses yielded the same result, indicating that psychosocial support during probation can reduce the number of criminal recidivisms by 44% during a follow-up time equaling 12 months. Two studies dealt solely with in-custody interventions. The meta-analysis did not result in a statistically significant difference.
The fourth indicated intervention is mediation offered to conflicting gangs. Taken together, the five evaluations did not provide a consistent picture of the impact of mediation between gangs with a follow-up time of at least 18 months: two reduced gun violence, one increased gun violence, and two reported no effect. Due to insufficient or contradictory statistical information, we refrained from a meta-analytic approach and concluded that it was not possible to determine the effect of mediation.
Gang affiliation, and the criminality by and against individuals in such groups, is challenging for many countries. This situation is reinforced by the lack of evidence-based interventions to prevent such negative outcomes. When young individuals are involved in gangs, the matter is morally pressing (UN, 1989), with high stakes involving significant physical and mental health consequences. In the absence of evidence, practitioners are faced with a situation where they risk implementing ineffective interventions or, in the worst case, interventions with negative effects (cf. the Scared Straight programs; Petrosino et al., 2013). The pressing nature of the problem, in combination with a legal obligation for the authorities to intervene and where the decision of how to act rests with the authorities, creates a situation where decisions must be made under great uncertainty.
When there is a lack of evidence, precautionary reasoning can be used. This kind of reasoning has been used to legitimize interventions with weak evidence that sometimes include infringements on people's liberty and rights (Ashworth & Zedner, 2014; Stacey et al., 2017). A sound application of the precautionary principle generally entails limiting, postponing, or refraining from implementing unproven interventions while awaiting assessment (Munthe, 2011). However, sometimes the pressing nature of the problem requires immediate action. In such cases, the interventions still need to be evaluated. There should also be a preparedness to act on the result of such evaluations, so that inefficient or, in the worst case, counterproductive interventions are abandoned. The assessment should also include potential social risks of the interventions, such as increased social tensions and distrust (Hardy, 2017).
From a policy perspective, this review imparts guidance. While some uncertainty regarding efficiency in, for instance, a European context remains, the available evidence provides reasons for cautious implementation with an appropriate evaluation process. If these measures turn out to be unsuitable or insufficient, other interventions can be considered. Any implementation of interventions for which evidence is lacking should be combined with a process of local evaluation (e.g., decide on the appropriate outcome measure, follow-up time, and responsible data collection) since the intervention may be turn out to be effective although no scientific study has yet demonstrated this. A particular aspect of importance is that the local evaluation includes all collaborative parties in the case of inter-agency collaboration (e.g., a collaborative gang intervention project between social service authorities, police, and prison and probation services).
There are further ethical considerations which may affect how, or even if, an intervention should be implemented. Interventions that target specific individuals or groups may thus treat them as, or signal to others that, these are criminals. When deciding which interventions to implement, a balance needs to be struck between targeting as many future criminals as possible, and minimizing targeting innocents so as to avoid undue burdens and intrusions of autonomy and privacy as well as unjustified stigma and loss of trust (Ferzan, 2014). There are also concerns about indirect discrimination if some groups are more burdened by the interventions than others, in particular groups that are, or have been, subject to injustices (Lippert-Rasmussen, 2022). Roman (Roman), for example, raises a concern regarding focused deterrence strategies and emphasizes that particular attention needs to be paid to how the burdens of the intervention are distributed and wonders whether “… GVI—a gang violence reduction intervention that relies on the heavy threat of punitive law enforcement action—has a place in cities today?”
Some interventions have, or could have, coercive elements. Standard requirements for when coercion is justified is that it should be necessary to prevent harm to others, that no less restrictive (but similarly effective) alternative is available, and that it should be proportional to the harm it seeks to prevent (Ashworth & Zedner, 2014). Since these interventions are, to a large extent, aimed at children, coercion may also be justified in the child's best interest but then a balance needs to be struck between the child's best interest, their autonomy, and parental authority.
Most results were graded as very low or low certainty of evidence. It stresses the importance of more controlled evaluations to reduce the risk of using interventions without significant effect. The two subgroup analyses of focused deterrence (short or long follow-up time and outcomes in the form of firearm violence with or without homicide) might guide the designing of new studies since they yield essentially the same outcome. If this result is correct, 21 months of follow-up is enough to secure a valid result on long-term effects, which decreases the timeframe of outcome research. The analysis also indicates that firearm violence without specification of outcome, and homicide, are equally valid as an outcome measure. Merging the two types of outcome research has the potential to produce more exact point estimates, thereby increasing statistical significance.
This review has some strengths and limitations. The strength of this review is that it has followed rigorous systematic review methods, including a comprehensive literature search, independent assessment of studies, assessment of risk of bias, and grading of evidence. The review also has limitations. The first deals with the complexity of identifying studies that include gang-related populations and avoiding the multitude of studies preventing juvenile delinquency in general. Our strategy was to include only those studies that either stated an aim of preventing gang crime or that included a typical gang-related outcome (e.g., carrying a gun). However, there might be other interventions against gang-related crime that our inclusion criteria missed.
A second limitation is the possibility of publication bias. Although we included seven databases and more than 18,000 references, it is still possible that there are unpublished trials that this review missed. This is supported by our analyses on focused deterrence, using a funnel plot. However, using trim and fill technique, the effect was somewhat lower but still statistically significant. We refrained from including “gray” literature such as dissertations and technical reports, because of the challenge to comprehensively identify all gray literature. We identified 20 studies in previous reviews that constitute gray literature (mainly technical reports). In most cases, 18 out of 20, the studies have also been published in peer-reviewed journals with a delay of one or more years. This indicates that time-lag bias is a larger problem than publication bias.
Third, a common challenge in intervention research is the lack of information on implementation quality. This also applies to the studies in this report. With flawed implementation, effective intervention risks appear ineffective. Another problem is the lack of description of the components of interventions; in which order they are given and with what dose. All indicated interventions studied here consist of several components but without a detailed description. Regarding the reliability of the reported results, 33 of 42 are based on criminal record data that is considered to have reasonably high reliability (eg, homicide, gun violence). The other nine studies concern universal and selective interventions and are based on self-report data of unknown reliability.
At last, of the 42 studies identified, 39 represented were from the United States. This raises the question of whether the results are valid for other nations and contexts. Although it is easy to find examples of seemingly important differences between the United States and other nations (e.g., the general level of gun violence), empirical differences do not necessarily tell us about transferability. One example of this is individual placement and support (IPS) that provides support for the competitive employment of patients with severe mental illness. A random-effects meta-regression, including 27 RCTs from different contexts, found that the efficacy of IPS was equal in Europe and the United States (Brinchmann et al., 2020).
Evidence from this review suggests that it is possible to reduce crime, but that the effect is limited. This suggests that the reviewed interventions do not contain all the components needed to create strong effects. Therefore, theory in combination with empirical research is needed to enhance our understanding of what drives change and what does not. For example, if there are components that can be excluded, it will save resources. This work is made difficult by the fact that necessary information on the components of experimental interventions and control conditions is often missing or vaguely described. It is important to document in a protocol or manual so that the content of the intervention can be used as intended. Also, many included studies lacked distinct descriptions of the comparison condition. Based on research examining common components across different intervention formats (van der Pol et al., 2019), intervention research should attempt to delineate components and methods most likely to affect outcome. Also, many of the reviewed studies lacked information on the scope and quality of program implementation as well as descriptions of the comparison condition. This reinforces the importance of increased transparency to allow analyses of how implementation and control characteristics may moderate client outcomes.
Keywords: gang-related crime, systematic review, meta-analysis
- ISSUP members can join Networks to comment – Sign in or become a member