A Field-Wide Systematic Review and Meta-analysis of Putative Risk and Protective Factors for Radicalization Outcomes

This systematic review sought to collate and synthesize the risk and protective factors for different outcomes of radicalization. We aimed to firstly quantify the effects of all factors for which rigorous empirical data exists, and secondly, to differentiate between factors related to radical attitudes, intention, and behaviors. The goal was to develop a rank-order of factors based on their pooled estimates in order to gain a better understanding of which factors may be most important, and the differential effects on the different outcomes.

Random effects meta-analysis pooled primarily bivariate effect sizes to calculate pooled estimates for each factor. Meta-regression was used to examine the effects of a range of study-level characteristics, including the effects of using partial effects sizes as supplementary effect sizes where bivariate estimates were unavailable. Subgroup analysis was used to further analyze the extent to which the combining of effect sizes from different sources contributed to heterogeneity and estimate inflation. Leave-one-out sensitivity analysis was used to identify cases where a single study was a significant source of heterogeneity.

Extensive searches in English, German and Dutch resulted in the screening of more than 10,000 items, and a final inclusion of 57 publications published between 2007 and 2018 from which 62 individual level factors were identified across three radicalization outcomes: attitudes, intentions, and actions. Effect sizes ranged from z − 0.621 to 0.572. The smallest estimates were found for sociodemographic factors, while the largest effect sizes were found for traditional criminogenic and criminotrophic factors such as low self-control, thrill-seeking, and attitudinal factors, with radical attitudes having the largest effect on radical intentions and behaviors.

The most commonly researched factors, sociodemographic factors, have exceptionally small effects, even when effect sizes are derived from bivariate relationships. The finding regarding the effects of radical attitudes on intentions and actions provide empirical support for existing theoretical frameworks. The consistency among the clustering of familiar criminogenic factors within the rank-order could have implications for the development of a more evidence based approach to risk assessment and counter violent extremism policies.