« Nothing Works’ – Bis repetita?

A meta-analysis of meta-analyses (an umbrella review) was recently published, casting doubt on RNR programmes:

Fazel, S., Hurton, C., Burghart, M., DeLisi, M., & Yu, R. (2024). An updated evidence synthesis on the Risk-Need-Responsivity (RNR) model: Umbrella review and commentary? Journal of Criminal Justice, 92, June. 102197: https://doi.org/10.1016/j.jcrimjus.2024.102197

This study has been wielded as “proof” that nothing works (e.g., https://www.cgtspip.org/rbr-une-science-exacte-pas-si-exacte/)—which, in turn, raises the question: if so, what now?

To contribute to the debate, I am sharing below excerpts from a book I am currently writing (but which, given my workload and health, will take two to three years to complete). These excerpts offer a commentary on this umbrella review, as well as on another similarly disappointing review concerning juveniles.

Note: typos included. This is a draft.

Umbrella review of meta-analyses on adult programmes: Methodological challenges.

The results of these [here, I am referring to previous paragraphs in my book] somewhat prolific meta-analyses[1], as observed across various research domains, are contradictory. In response, Fazel and colleagues[2] decided to undertake an umbrella review[3] of the existing meta-analyses.

To do so, they searched for meta-analyses in PubMed, PsycNET, Scopus, and the Cochrane Library, covering the period from 01/01/2002 to 15/12/2022. They also followed citation chains within the meta-analytic reviews. Additionally, grey literature was searched using Eldis, Google Scholar, and FindPolicy. The authors analysed research on the three core RNR principles, assessing both the quality[4] and methodological biases[5] of the studies using validated tools. The literature has highlighted self-referential biases among RNR scholars, and Fazel et alii, citing my work[6], suggest that I previously identified this issue. It does not, in fact, precisely reflect the content of my book chapter, which primarily advocated for the better integration of different models—each with its blind spots and shortcomings—and for the incorporation of insights from self-determination theory and criminal justice legitimacy, including procedural justice. In their umbrella review, Fazel et alii identified a total of 26 meta-analyses addressing one or more of the R-N-R principles, all of which appeared to affirm the validity of these principles.

However, the methodological and bias analyses, scored on a scale of 1 to 7, yielded disappointing results. Regarding the “Risk” principle (seven meta-analyses), scores were consistently below 2/7, making it challenging to demonstrate the effectiveness of the principle. For “Needs” (six meta-analyses), reviews directly focused on this principle were of poor quality (scoring 0–2/7). Reviews evaluating tools were of higher quality (scoring 3–7/7). Nevertheless, most reviews reported substantial heterogeneity in the studies. For the “general responsivity” principle (i.e., evidence-based practice delivery), the 15 reviews were classified as weak in eight cases, of moderate quality in two cases, and of high quality in five. For “specific responsivity” (i.e., intrinsic responsivity), the four meta-analyses ranged from low to moderate quality (scoring 0–3 to 6/7).

The authors concluded that these poor results raised “important and timely questions regarding the continued application and utility of RNR as a model informing criminal justice services”[7]. While RNR proponents attribute the variability in study results to the lack of fidelity to the principles in “real-world” settings—whereas fidelity tends to be better achieved when academics are present to assist in developing and evaluating programmes—Fazel and colleagues argue that this is, in fact, an authorial bias, with lower-quality studies systematically being produced. Fazel et alii go further, suggesting that there is an allegiance effect towards RNR and its associated tools, which are commercialised by the tools’ own developers (e.g., the Canadian LSI-R or LS-CMI). They note, “it is notable that many of the included reviews did not address or disclose potential conflicts of interest.”[8] It should be noted, however, that the methods used by Fazel et alii to score the quality and bias of studies explicitly deducted points for this lack of transparency, though these deductions were not directly linked to the research methodology itself. More problematically—albeit often due to space constraints in international journal formats—Fazel and colleagues also noted that many of the included reviews failed to adequately specify the sample size, its characteristics, the treatment administered to the control group, the features of the primary studies included, or the meta-analytic strategy employed. Furthermore, as others have noted previously, the authors observed that primary studies are rarely randomised controlled trials (RCTs), and even when such trials are present, their quality is often unclear or inadequate, particularly regarding the specification and implications of the various TAU conditions in the control groups. Lastly, the meta-analyses themselves frequently compared “apples and oranges,” as they combined primary studies addressing RNR treatments that varied significantly across institutions and countries.

Meta-analyses on juvenile programmes: Disappointment.

 Other meta-analyses have focused specifically on juveniles and yielded similarly disappointing results. For instance, Kim et alii[9] replicated Lipsey and Cullen’s influential 2007 review of meta-analyses[10] (see no. 08.262), but with a focus exclusively on meta-analyses concerning juveniles. In other words, they reviewed only meta-analyses. The broad conclusion of their study was that treatment methods effective for adults did not work for juveniles. Addiction treatment, training, vocational training, access to employment, and CBT interventions were ineffective, with control groups often achieving better outcomes. In contrast, juveniles appeared to benefit more from traditional probation with supervision.

In 2021, Pappas and Dent[11] conducted a review of 48 meta-analyses and systematic reviews examining the impact of treatment programmes on juveniles. Their findings were somewhat more positive and specific. The meta-analyses included in their review had to meet several criteria. They needed to: report on young people aged 10 to 25 receiving some form of treatment as part of a criminal sanction; include a control group not receiving treatment; allow for meta-analytic calculations; measure broad outcomes (e.g., reconviction, re-arrest, revocation or sanction in community-based supervision); and be written in English. Their conclusion was that these programmes significantly reduced recidivism compared to juveniles who did not participate, with an average differential impact of 17.4%. However, this reduction varied considerably depending on the type of justice system, characteristics of the juveniles, type of programme, and, of course, classically, the methodological quality of the studies (the higher the quality, the less pronounced the effect). The most substantial impact was observed among juvenile sexual offenders (a 30.54% reduction) and those committing violent or serious offences (a 25.48% reduction). The impact was weaker among general offenders, those committing less serious offences, or those involved with drug-related crimes.

The authors also found that programmes delivered in custodial settings had three times the impact of those delivered in community-based settings. This contrasts with findings for adults and may reflect the fact that juvenile institutions implement more structured and supportive programmes compared to those offered to adults. Probation and prison-based programmes were more effective than post-release social support programmes or alternative “diversion” initiatives (such as prosecutorial diversion programmes in France) designed to replace traditional interventions. From a French perspective, such diversion programmes are often shorter and less robust. Finally, the programme modalities that proved effective for juveniles were partly different from those typically effective for adults, with the exception of CBT, which was also effective. These included multisystemic therapy, family-focused treatments, restorative justice, and even wilderness therapy—often dismissed as quackery[12].

In 2022, another meta-analysis[13] arrived at similar conclusions regarding juveniles. However, the authors inferred a broader critique of the RNR model itself rather than its application to juveniles. Specifically, they found no evidence supporting the effectiveness of the risk and need principles compared to interventions that did not incorporate these principles. The only exceptions were antisocial attitudes and antisocial leisure activities.

In 2021, yet another meta-analytic review[14] demonstrated that none of the psychosocial interventions implemented with juveniles—including CBT, family therapies such as FFT (Functional Family Therapy), MDFT (Multidimensional Family Therapy), and MST (Multisystemic Therapy)—produced better outcomes in terms of recidivism than the control groups. This was a significant blow, given prior confidence in the efficacy of these approaches[15]. It is possible, as is often the case, that practical implementation failed to adhere faithfully to the original models or was hampered by institutional, legal, or practitioner competency issues.

Recent Meta-Analyses: Analysing Disappointment

Regarding adults, the programmes included generally focused solely on treatment without any psychosocial reintegration preparation. This is a crucial point: prison is not an “opportunity” for treatment but an institution that, as we have seen, inherently increases recidivism. To offset some of its criminogenic effects, at the very least, there must be a restoration of the individual’s social situation (housing, health, social rights, income, employment, etc.), if not a significant improvement in these areas. While this book emphasises the psycho-criminological aspect of treatment, this should not be interpreted as a denial of the essential, even vital, nature of active social work—much more proactive than the somewhat romanticised view of the past often suggests—serious and genuine, especially in prison settings.

Another critical point is that studies have shown that the so-called CBT programmes implemented in criminal justice settings barely employed actual CBT techniques due to probation officers and even some therapists lacking the necessary training and clinical experience[16]. Proper initial training within the LMD (Degree (Bachelor) Master’s (post-grad)-Doctorate) university framework, rather than quick institutional training programmes, is indispensable to significantly enhance performance in this area. Probation officers must simultaneously be social workers, criminologists, and “super-psychologists,” or the treatment must be programmatic and delivered by teams of specialists appropriately trained in EBPs in social work, criminology, and psychology.

However, another critique comes from opponents of the RNR (Risk-Need-Responsivity) model: it is based on the premise that we truly understand what causes delinquency. While the RNR variables are useful for predicting recidivism, they are not sufficiently anchored in the actual causes of crime[17]. This critique is valid: the RNR model refers to a single criminological theoretical framework—social learning theory and its predecessor, differential association. It – deliberately? – neglects other criminological models (e.g., strain theory, social bond theory), numerous psychological dimensions (e.g., social psychology, self-determination, and human motivation), validated theoretical models related to specific offenses (e.g., violence or sexual delinquency), and even gender-specific considerations. This is not merely an academic quibble: effective programmes with sufficient breadth and depth will only emerge when we have a comprehensive and integrated understanding of what causes delinquency. This cannot solely be defined within the RNR context, even if it remains the sole approach that produces real—but modest—results in addressing delinquency.

As for the 2024 meta-analysis pertaining to juveniles, it is important to emphasise that the authors did not target conventional treatment, but family-based interventions grounded in these principles. It is therefore entirely plausible that, for minors, family interventions constitute a distinct treatment unit within the RNR model, supplementing it and adhering to its own principles. Furthermore, the 2024 meta-analysis reaffirms that the antisocial dimension remains a relevant target, which is central to the RNR model as well as many of the most validated macro-criminological theories (notably self-control and social learning). Finally, the authors examined leisure activities not from the problematic perspective of structured activities but through the lens of antisocial tendencies. Minors have far more leisure time than adults, which creates opportunities for interaction with antisocial peers. It is therefore unsurprising that this factor is significant in their specific context.

Nonetheless, and getting back to adults, the 2024 umbrella review (Fazel et alii) highlights presumed methodological shortcomings of older studies, especially as research standards have become increasingly stringent. In the field, it is rarely possible to conduct high-level studies capable of providing conclusive evidence. The settings are complex, institutions are reluctant if not resistant, operations must continue uninterrupted, and legal and ethical constraints often impede randomisation. Samples are typically too small, and in France, bureaucratic hurdles—especially in universities—severely hinder effective research. Furthermore, academic journals impose stringent limits on word and page counts, often preventing sufficient detail about methodology and treatment descriptions. It would thus be excessive to declare a new “Nothing Works,” but it is clear that institutional and international reflection on these matters is ongoing.

More importantly, let us also remember that “RNR” encompasses more than just the R, the N, and the R. These three principles are part of about fifteen overarching principles, including those cherished by certain critics, such as a focus on: community programmes and intervention; prevention; human services (e.g., education and social work); ethics; and optimal leadership. Most notably, the RNR framework introduced the “Core Correctional Practices” (CCPs), which some critics unfairly oppose to RNR. The CCPs are part of the overarching principles and include, inter alia, the therapeutic alliance. It bears mentioning that this is where RNR advocates, including Andrews in 1975, began their work. Conducting meta-analyses of only three principles—or implementing only those three principles—is both incorrect and unfair.

What then should we make of these results? Clearly, we need to fund large-scale, rigorous, and entirely independent research—something that is sorely lacking in France—where institutions are required to operate with absolute transparency. This includes mandating that institutions adhere to protocols involving randomisation and allocating funds for long-term longitudinal studies, the absence of which in France is shameful. Finally, we must establish forensic psycho-criminology faculties (or forensic social sciences) to produce practitioners genuinely capable of conducting clinical interviews, applying EBP in psychology, criminology, and related forensic disciplines, and implementing them effectively. From this pool of talent, researchers and laboratories can emerge to experiment with innovative therapeutic methods (e.g., schema therapy) in areas where we are currently failing to deliver results (domestic violence, Cluster B personality disorders, high-risk recidivists, and other LCP offenders), thus limiting the prospects for drastically reducing recidivism.

Finally, we need to improve the RNR model itself, which is a broad framework that must connect with all validated macro-criminological theories (e.g., social bond, self-control) currently excluded from this framework. This book has also demonstrated that RNR has “Gruyère cheese” holes, notably regarding gender, diversity, specific offenses and their unique factors (e.g., identity, need for closure in terrorism), and both institutional, public policy and legal challenges in programme implementation. While RNR acknowledges the importance of the therapeutic alliance, it has not sufficiently leveraged the relevant literature on alliance conditions. It would be absurd to demand that a model created by a handful of individuals has anticipated every challenge and addressed every gap. Critics are right to critique, but they have yet to propose an alternative model—or better still, to improve RNR itself.

To conclude, it would be a grave error to follow the disastrous post-Martinson path in the USA and declare that science is ineffective or non-existent, and that, as a result, one should abandon it altogether. In this age of rising populism, we risk repeating the conclusion once drawn by Republicans: “If nothing works, lock everyone up and throw away the key.” Some might advocate retreating solely to social work — non-existent or nearly so in reality — which Martinson’s era demonstrated was incapable of reducing recidivism. Social work is essential for addressing responsivity and is imperative for human rights reasons. Therefore, it would be equally irresponsible to naïvely conclude—against the Republicans but just as deaf to current realities—that “if nothing works, let’s focus on access to rights and social work.”

Jim Bonta’s response.

The meta-analysis by Fazel et alii, which shocked many with its particularly negative results, prompted The International Association for Correctional and Forensic Psychology Bulletin to dedicate an issue to this subject[1]. The journal first presented Fazel et alii’s findings in an honest and precise manner. Subsequently, the journal invited Jim Bonta to provide commentary. Bonta identified a series of novice-level errors in the understanding of the RNR model, as well as biases and questionable interpretations.

First, Fazel et al. analysed the principle of risk solely by examining outcomes for low- and high-risk users. This approach misinterprets the risk principle, which never meant “prediction”, but in fact, means adequate dosage, as this book has emphasised repeatedly. Let us remind our readers that RNR is not a Minority Report model; it is a treatment model. Instead, the proper approach would have been to examine dosage levels across all risk categories and assess the relationship between appropriate dosage and these categories.

Second, the authors also demonstrated a misunderstanding of the needs principle by including only studies that tested the predictive validity of needs. While it is true that needs can contribute to prediction, this is not the core meaning of the needs principle, as we have repeatedly clarified in this book. The needs principle refers to targeting the relevant treatment goals. What should have been done here was to verify whether criminogenic needs, when appropriately targeted, led to a reduction in recidivism.

Finally, as Bonta rightly points out, Fazel and colleagues also misinterpreted the principle of specific responsivity (referred to in this book as intrinsic responsivity). They focused their attention on the satisfactory completion of treatment as a measure of positive outcomes. Bonta argues that this conflates intrinsic responsivity with risk. In our view, while satisfactory completion could initially appear to be an element of intrinsic responsivity, it may also conceal over-adaptation or feigned compliance. Moreover, and more importantly, completing a programme without issues often owes as much, if not more, to the quality of the programme itself, the therapeutic relationship, the institution’s nature and functioning (e.g., the ability to find partners using EBPs, the existence of transformational leadership – i.e. extrinsic responsivity as understood in this book), as it does to risk (dosage) or needs (targeting). Additionally, this book has shown above that attrition is largely caused by institutional factors, penal policies, and, in the French context, judicial practices such as recall and other sanctions.

On methodological issues, Bonta also levels criticisms. He argues that Fazel and colleagues demand the impossible from RNR research, expecting unattainable perfection—that is, that studies should encompass all principles, all methodological aspects, and all theoretical elements. No study, nor any literature review for that matter, could achieve such comprehensiveness, particularly given the limited space typically allotted in scientific journals. Interestingly, Fazel et alii lament reliance on a single principle, yet this is precisely what they do by examining each principle in isolation, despite RNR clearly stating that individual principles alone yield insufficient results. It is, in fact, the application of all three principles, as well as the additional overarching principles (a total of fifteen), that is required to achieve meaningful outcomes.

Lastly, the accusation of bias recurs in Fazel et alii’s critique of all RNR research as biased. Yet, ironically, as James Bonta notes, Figure 4 of their own study exclusively cites Fazel and colleagues’ work, thereby exhibiting a bias favourable to their own work. Furthermore, these cited studies are all predictive in nature—far removed from the topic they claim to address.

---

[1] The International Association for Correctional and Forensic Psychology Bulletin, 2024, July/August.

Other experts’ responses

The International Association for Correctional and Forensic Psychology Bulletin subsequently invited a panel of experts to analyse Fazel and alii’s results. Like Bonta, the authors highlight both methodological issues and scientific misunderstandings.

Paul Gendreau, an RNR proponent, who writes with Bonta, mainly observes that Fazel and colleagues limit themselves to « cheap shots » regarding economic conflicts and conclude in a very Martinsonian fashion that nothing works anymore. Clearly, for these authors, only the methodology of the research matters, without any attention being paid to the quality of treatment implementation. This is indeed the comment I made earlier. I know from my own immersion studies in France, that RNR is indeed rarely implemented, even when it is claimed to be.

Furthermore, Gendreau also criticises Fazel and colleagues on the methodological front. First, the latter ignore the numerous and independent replication studies of Bonta and colleagues’ initial research, assuming these to be biased without much evidence. However, replication is the cornerstone of scientific testing. Moreover, adds Gendreau, the presentation of Fazel and colleagues’ results is incomprehensible to non-mathematicians, whereas RNR has always been careful to present its findings in a clear manner for decision-makers and practitioners.

Mark Olver, in turn, critiques both the methodology and the attitude of Fazel and colleagues.

Concerning the methodology, their list of key words is lacking. A verification reveals that they are not actually present in the article, but rather in an « appendix ». Consulting the appendix, which can be downloaded from the journal’s website, confirms this impression. Olver then argues that the interpretations of the study results are « idiosyncratic », meaning highly personal, if not unique and odd. In other words, these conclusions are not what any other person would arrive at. The conclusions of Fazel et alii are therefore exaggerated. Like this book, Olver observes that Fazel et alii offer no viable alternative. Methodologically, Fazel and colleagues ignore a multitude of studies, including meta-analyses involving hundreds of studies showing positive effects, on the grounds that the authors would all be biased. He states, “There are probably close to 1,000 replications of RNR in some form over the last 35 years from around the world.”[1] I would also point out that one might also suggest that Fazel and colleagues are equally biased against RNR. Should armies of new authors start producing literature from scratch, in order to avoid both pro-RNR and anti-RNR biases?! With such demands, we would make no progress. In fact, Olver deems Fazel and colleagues’ conclusions “irresponsible[2].”

Unsurprisingly, Franck Porporino, although in full agreement with Mark Olver, believes that there is indeed an alternative: the GLM. This book shall address this later. It will show that, unfortunately, this is not the case, either theoretically or empirically, even though I acknowledge that GLM supporters make a series of very useful points regarding some theoretical gaps in RNR. Porporino advocates for “modern pluralism”, which is quite close to what I have proposed[3]. Joël Dvoskin supports the same idea in the special issue under review here.

Finally, Devon Polaschek notes that Fazel and colleagues’ article presents “certain conceptual aspects that are, disturbingly, quite wonky.”[4] Essentially, she identifies the same elementary errors as the other commentators. She specifies another mistake: Fazel et alii claim to test the principle of needs (a targeting principle), not only confusing it with risk but also using studies that verify the predictive validity of risk-need tools. This is indeed a striking mix-up. I can only concur. At this stage, it goes beyond the proverbial mixing of apples and oranges to a mix of apples and napkins. Polaschek concludes soberly, but firmly, that in her view, Fazel et alii are not merely attempting to destroy the state of knowledge; they simply do not understand what RNR is.

Finally, while Fazel et alii are right to emphasise the importance of randomised controlled trials, Polaschek reminds us that these too are not without bias, particularly in that, in penal settings, they cannot be double-blind, unlike in medical contexts.

I could not help but smile when I discovered that Devon Polaschek uses the same cancer metaphor as I repeatedly do. She remarks that she would love to see the reaction of doctors at a cancer conference if someone were to present conclusions similar to those of Fazel et alii and then suggest that we should stop implementing current cancer treatment until we know more.

---

[1] Ibid. p. 10.

[2] Ibid, p. 10.

[3] Herzog-Evans, (2018). RNR, Evidence diversity and integrative theory. In P. Ugwudike, P. Raynor & J. Annison (Eds.) Evidence-Based Skills in Criminal Justice. International Research Supporting Rehabilitation and Desistance [pp. 99–126.] Policy Press.

[4] Ibid. p. 11.

Post scriptum (not in the book)

What are we to do? Since nature hates a vacuum, already, three non EBP-simplistic solutions are being promoted, depending on one’s ideology or faith:

1. Lock’em up and throw away the key
2. Let’s get back to pre-Martinson social work
3. let’s get back (or remain) to Crystal Ball’ – homeopathy – Zodiac astrology, aka psychoanalysis

The three temptations are already heard in France as in « I told you so »!

In short, the common discourse is: “Since science is complex, too much for my liking, and takes time to advance, and since I prefer, like populists, the binary and the simplistic, it is therefore that solutions focused on only one type of factor (look for the bad mother only; restore housing or employment only; lock everyone up, only) – which have been widely invalidated for decades, but shh, we are not supposed to say that – should be implemented.”

In the post-truth world we live in, where anything goes, where magic is deemed superior to science, and where scientific culture is either non-existent or dismissed, everyone prefers the magical/religious solution they “believe in”, because it fits their ideology (whether right-wing, left-wing, or the centre), or their professional interest (old-fashioned, non-evidence-based social work or psychoanalysis).

None of the Great Simplifiers from all sides participate in real, difficult, and complex applied research.

Meta-analyses are essential, but the quantitative authors, admirable though they may be, often have fairly simplistic conclusions themselves, because they usually do not deal with institutions, practitioners, or the messy nature of real-world environments, unlike many qualitative researchers or quantitative researchers invested in field experiments. Faye Taxman is right: “Nothing is implemented” is the real problem. My modest research and programme development endeavours continuously throw this in my face in a shocking, sardonic, and always concerning way.

The solution for practitioners lies in strong, comprehensive, multidisciplinary initial training, and, for France, in renewed faculties, which would stop making us spend half our time in bureaucratic Kafkaesque processes and instead give us the time for research and teaching. Teaching human sciences with an applied dimension (law, psychology, criminology, as well as social work, which, incidentally, is scandalously not taught at universities in France, and therefore, is totally detached from empirical (EBP) evidence) should include repeated clinical/practical exercises, which would be drilled and adequately supervised and coached. Practice most definitely is what makes perfect. We can no longer teach only theory and then send students on internships, expecting them to practice something they have not been prepared for. It is like putting a co-pilot in a plane who has only had theoretical hours, and then handing them the controls once airborne ‘to teach them practice.’

When will we finally implement ambitious systemic policies?

---

[1] Ioannidis J. P. (2016). The Mass Production of Redundant, Misleading, and Conflicted Systematic Reviews and Meta-analyses. The Milbank quarterly, 94(3), 485–514.

[2] Fazel, S., Hurton, C., Burghart, M., DeLisi, M., & Yu, R. (2024). An updated evidence synthesis on the Risk-Need-Responsivity (RNR) model: Umbrella review and commentary? Journal of Criminal Justice, 92, Juin. 102197: https://doi.org/10.1016/j.jcrimjus.2024.102197.

[3] Higgins, J.P.T., Thomas, J., Chandler, J., Cumpston, M., Li, T., Page, M.J., & Welch, V.A. (editors). (2023). Cochrane Handbook for Systematic Reviews of Interventions version 6.4 (updated August). Cochrane, Available from www.training.cochrane.org/handbook.

[4] Shea, B. J., Reeves, B. C., Wells, G., Thuku, M., Hamel, C., Moran, J., Moher, D., Tugwell, P., Welch, V., Kristjansson, E., & Henry, D. A. (2017). AMSTAR 2: a critical appraisal tool for systematic reviews that include randomised or non-randomised studies of healthcare interventions, or both. BMJ (Clinical research ed.), 358, j4008. https://doi.org/10.1136/bmj.j4008.

[5] Whiting, P., Savović, J., Higgins, J. P., Caldwell, D. M., Reeves, B. C., Shea, B., Davies, P., Kleijnen, J., Churchill, R., & ROBIS group (2016). ROBIS: A new tool to assess risk of bias in systematic reviews was developed. Journal of Clinical Epidemiology, 69, 225–234

[6] Herzog-Evans, M. (2017). The risk-need-responsivity model: Evidence diversity and integrative theory. In P. Ugwudike, P. Raynor, & J. Annison (Eds.), Evidence-based skills in criminal justice (pp. 99–126). Policy Press.

[7] Fazel, S., Hurton, C., Burghart, M., DeLisi, M., & Yu, R. (2024). An updated evidence synthesis on the Risk-Need-Responsivity (RNR) model: Umbrella review and commentary? Journal of Criminal Justice, 92, Juin. 102197: https://doi.org/10.1016/j.jcrimjus.2024.102197, spe. p. 5.

[8] Ibid. p. 7.

[9] Kim, B., Merlo, A. V., & Benekos, P. J. (2013). Effective Correctional Intervention Programmes for Juveniles: Review and Synthesis of Meta-Analytic Evidence. International Journal of Police Science & Management, 15(3), 169–189.

[10] Lipsey, M. W., & Cullen, F. T. (2007). The effectiveness of correctional rehabilitation: A review of systematic reviews. Annual Review of Law and Social Science, 3, 297–320.

[11] Pappas, L.N., Dent, A.L. (2021). The 40-year debate: a meta-review on what works for juvenile offenders. Journal of Experimental Criminology, 12:1-30. doi: 10.1007/s11292-021-09472-z.

[12] Gendreau, P., Smith, P. & Theriault, Y. (2009). Chaos theory and correctional treatment. Common sense, correctional quackery, and the law of fartcatchers. Journal of Contemporary Criminal Justice, 25 (4), 384-396.

[13] Bijlsma, A. M. E., Assink, M., Stams, G. J. J. M., & van der Put, C. E. (2022). A Critical Evaluation of the Risk, Need, and Responsivity Principles in Family Interventions for Delinquent Youth: A Meta-Analysis. Criminal Justice Review, 0(0). https://doi.org/10.1177/07340168221140830.

[14] Olsson, T. M., Långström, N.,Skoog, T., Löfholm, A., Leander, C., Brolund, L., Ringborg, A., Nykänen, A., Syversson, A., &  Sundell, K. (2021). Systematic review and meta-analysis of noninstitutional psychosocial interventions to prevent juvenile criminal recidivism. Journal of Consulting and Clinical Psychology, 89(6), 514-527

[15] Trotter, C. (2013). Collaborative family work : A practical guide to working with families in the human services. Allen & Uwin; Bonta, J. & Andrews, D.A. (2023). The psychology of crim inal conduct. Sage, 7e ed

[16] Gannon, T.A., Olver, M.E., Mallion, J.S. & James, M. (2019). Does specialized psychological treatment for offending reduce recidivism? A meta-analysis examining staff and program variables as predictors of treatment effectiveness? Clinical Psychology Review, 73, Nov., 101752, https://doi.org/10.1016/j.cpr.2019.101752.

[17] Ward, T., Arrigo, B., Barnao, M., Beech, A., Brown, D.A., Cording, J., Day, A., Durrant, R., Gannon, T.A., Hart, S.D., Prescott, D., Straus-Hughes, A., Tamatea, A. & Taxman, F. (2022). Urgent issues and prospects in correctional rehabilitation practice and research. Legal & Criminological Psychology, 00, 1-26. »