Measuring the Thresholds of Extremism: Testing for Measurement Invariance Between Muslim Converts and Muslim Non-Converts of Radicalism With an Ordinal Model

Assessment of the risk of radicalism remains underdeveloped by terrorism researchers in part due to difficulties in standardized measurement. The Activism and Radicalism Intention Scales (ARIS) by Moskalenko and McCauley (2009)— an already robust and widely applicable tool—will be further improved upon and standardized in this proposed project using data previously collected via a grant-funded research project. Specifically, the Measurement Equivalence/Invariance of the latent factor Radicalism is tested for using two radicalism-relevant groups—Muslim converts and Muslim non-converts. Ordinal logistic regression, rather than a typical linear regression, is used since it is a more accurate statistical estimation of Likert survey data, like that found in the ARIS. These methods are the first of many steps to set a standard for more accurate measurement for surveying intervention outcomes and group comparisons in Countering Violent Extremism (e.g., at risk versus control or pre- versus post-treatment)