In recent years scholars have increasingly been interested in the effects of filter bubbles, echo chambers, and the potential for algorithmic deviancy amplification. The hypothesis generated by these new perspectives is that personalization algorithms contribute to the creation of online echo chambers, and that echo chambers contribute to the development of deviant beliefs, including radical beliefs. Despite the interest in this topic, and its relevance to current policy, there is little in the way of quantitative study. In this study we conducted a randomized control experiment on new Twitter users in which we tested the hypothesis with respect to justification of suicide bombings. We find statistically significant interactive effects between the treatment of algorithm suppression, and certain network structure characteristics representative of echo chambers, as predictors of radical beliefs.