Sources used for data collection.

Irrespective of discipline, the publication of null or non-significant findings is rare in the social sciences. For burgeoning fields like terrorism research, this is particularly problematic. As well as increasing the likelihood of Type II errors, the selective reporting of significant findings ultimately impedes progression, hindering comprehensive syntheses of evidence and enabling ill-supported lines of scientific enquiry to persist. This manuscript discusses several structural and individual-level variables which failed to produce significant, linear associations with involvement in terrorist violence in a dataset (N = 206) of right-wing and jihadist extremists active in Europe and North America. After considering methodological factors such as non-random distributions of missing data, we illustrate how certain variables are significantly associated with involvement in terrorist violence at particular periods in a radicalizing individual’s lifespan, but not others (i.e., pre- or post-radicalization onset). Moreover, we demonstrate that while some static, binary constructs (such as whether or not a radicalizing individual was exposed to diverse viewpoints) are not associated with terrorist violence, their influence over time produces different associations. We conclude that radicalization may be less about individuals having pre-disposing risk factors, such as biographical stressors, and more about cognitive changes that allow individuals to re-evaluate their lives through the lens of an extremist ideology. We also underline the importance of taking a temporal, rather than static, perspective to better understand the variables associated with the outcomes of radicalization trajectories.