It is important to direct resources to effective evidence- and value-based counter-radicalization strategies, especially within prisons. With the increasing threat ofviolence and terror from various violent extremist groups, such as Daesh, thefocus on prisons should be intensified. While radicalization within prisons was nota significant problem a decade ago, the new recruitment strategies from e.g.Daesh now pose a new challenge for prisons. This thesis identifies contemporaryliterature on radicalization, deradicalization, and violent extremism within aprison context. From this seven recurrent themes were identified: overcrowding,religious chaplains, sectioning, risk assessment, monitoring and supervision,rehabilitation programs and education of staff. This was then used to develop anassessment model, which was based on the Canadian Risk-Need-Responsivitymodel’s three core principles, for evaluating and conducting counterradicalizationstrategies. The assessment model was then used as a framework fora comparative analysis of the Danish and Swedish Prison and Probation Services’counter-radicalization strategies within prisons. The results show that bothcountries adhered to a degree to the assessment model and current literaturewithin the field. The results further reveal that there is a lack of empiricalevidence and data on radicalization and counter-radicalization within prisons, andthat the data available is somewhat outdated.

A presente tese buscou identificar os fatores proximais mais relevantes que afetam o desenvolvimento do comportamento violento na adolescência. Por meio de revisões sistemáticas e metanalíticas da literatura, o presente estudo identificou que os seguintes fatores de risco: exposição a eventos estressores, histórico de conduta violenta, fracasso em intervenções anteriores, baixo autocontrole, traços de psicopatia, emotividade negativa, transtornos de saúde mental específicos e de abuso de substâncias, pares infratores e envolvimento em gangues, isolamento social e inabilidade parental. Em relação aos fatores protetores, a revisão indicou que os fatores proximais mais relevantes consistem em: autocontrole, empatia, inteligência, atitudes positivas frente a figuras de autoridade e a intervenções, comprometimento com a escola ou ao trabalho, apoio social e vínculo com adultos pró-sociais. Na parte empírica dessa tese, objetivou-se identificar os principais fatores associados à conduta violenta em adolescentes em conflito com a lei brasileira e testar a capacidade preditiva desses fatores. Participaram 118 adolescentes do sexo masculino entre 15 e 19 anos que cumpriam medida socioeducativa no município de Ribeirão Preto SP, dos quais 78 foram entrevistados em um segundo momento (T2), 18 meses após a primeira entrevista (T1). Os adolescentes completaram o Questionário sobre Comportamentos Juvenis (QCJ) e o Inventário de Jesness Revisado Brasileiro (IJ-R-Br) no T1 e no T2, e foram testados nos testes de Stroop Victoria (TS-V) e das Torres de Londres (TOL) somente no T2. Adicionalmente, foram levantados os históricos de boletins de ocorrência dos adolescentes junto ao fórum da cidade. Os adolescentes foram divididos em grupos Violento (GV) e não violento (GNV) e comparados nas escalas psicossociais do QCJ, nas escalas de personalidade do IJ-R-Br e nos testes TS-V e TOL. Posteriormente, 70% da amostra foi utilizada para a construção de dois modelos para testar o poder preditivo para o envolvimento em condutas violentas de forma retrospectiva nos 30% dos adolescentes que não participaram da construção dos modelos. Adolescentes infratores com trajetória de conduta violenta apresentaram pior desempenho que adolescentes infratores sem trajetória de conduta violenta em variáveis dos domínios Pessoal (capacidade cognitiva, crenças, valores e aspectos de personalidade), Familiar (apego e supervisão parental), Escolar (desempenho acadêmico), Pares Infratores e Comunitário (percepção de violência). No T2, algumas dessas variáveis deixaram de produzir efeito. Em termos preditivos, o modelo 1 apresentou acurácia de 74% e o modelo 2 de 83%, destacando-se o poder preditivo das escalas do IJ-R-Br (Desadaptação Social e Transtorno de Conduta), do QCJ (Pares Infratores e Atraso Escolar) e especialmente do histórico de conduta violenta. Destaca-se que algumas variáveis impactam no desenvolvimento da conduta violenta especificamente na adolescência, reduzindo seu efeito no início da vida adulta, enquanto outras mantém seu efeito ao longo do tempo. E o fato de existirem indicadores que podem predizer o envolvimento em comportamentos violentos com boa acurácia traz contribuições para o conhecimento da realidade brasileira e também implicações para as políticas públicas e pesquisas futuras / The present thesis sought to identify the most relevant proximal factors that affect the development of violent behavior during adolescence. Through the systematic and meta-analytic reviews of the literature, the present study identified the following risk factors: exposure to stressful events, history of violent behavior, failure in previous interventions, low self-control, traits of psychopathy, negative emotionality, mental health disorders and substance abuse, peer offenders and gang involvement, social isolation and parental inability. Regarding the protective factors, the review indicated that the most relevant proximal factors consist in: self-control, empathy, intelligence, positive attitudes toward authority figures and interventions, commitment to school or work, social support and attachment to prosocial adults. In the empirical part of this thesis, the objective was to identify the main factors associated with violent behavior in adolescents in conflict with Brazilian law and to test the predictive validity of these factors. A total of 118 male adolescents between 15 and 19 years old in conflict with the lay in the city of Ribeirão Preto SP, of which 78 were interviewed in a second wave (T2), 18 months after the first interview (T1). The adolescents completed the Questionnaire about Youth Behavior (QCJ) and the Jesness Inventory - Revised - Brazilian Version (IJ-R-Br) on T1 and T2, and were tested in Stroop Test - Victoria (TS-V) and Tower of London (TOL) on T2 only. Additionally, the historical records of teenagers\' official offenses were collected at the city forum. Adolescents were divided into Violent (GV) and non-violent (NGV) groups and compared on the QCJ psychosocial scales, IJ-R-Br personality scales and TS-V and TOL tests. Subsequently, 70% of the sample was used to construct two models to test the predictive power to engage in violent behavior retrospectively in 30% of adolescents who did not participate in the construction of the models. Adolescents with violent behavior trajectories presented worse performance than adolescents with no trajectory of violent behavior in variables of the domains Personal (cognitive capacity, beliefs, values and personality aspects), Family (attachment and parental supervision), School (academic performance), Peers (offenders) and Community (perception of violence). On T2, some of these variables no longer have an effect. In predictive terms, model 1 presented an accuracy of 74% and model 2 of 83%, highlighting the predictive power of the IJ-R-Br scales (Social Disorder and Conduct Disorder), QCJ scales (Peer Offenders and School failure) and especially the history of violent conduct. It is noteworthy that some variables impact on the development of violent behavior specifically during adolescence, reducing its effect in early adult life, while others maintain their effect over time. And the fact that there are indicators that can predict the involvement in violent behaviors with good accuracy brings contributions to the knowledge of the Brazilian context and also has important implications for public policies and future research