Our baseline sample is drawn from high-school seniors, from 20 schools in Yaoundé, Cameroon’s capital. The schools will be randomly drawn from the full roster of the city’s 120 secondary schools, after stratification by school quality (as indicated by tuition rate) and between public and private institutions. Within each school, we will randomly select 4 to 8 classes, focusing of terminal year students. This sampling strategy (___________), will provide us with ~5000 students and allow for robust data analysis.
PROCESS & PURPOSE
A school-based sample is not representative of the larger population because it excludes adolescents who drop out of school fairly early, or those who never enroll in school at all. However, this sampling strategy is well suited for the purpose of this study, for several reasons. First and most fundamentally, we are less interested in a population-level estimate of the risk of terrorism, as we are in the effects of leaving school and being subjected to the frustrations of a difficult job market and weak social supports. From this perspective, our sampling approach usefully starts all students from a common baseline to see how paths may diverge, depending upon subsequent circumstances and students’ first encounter with the world of work and adulthood. Furthermore, the collection of baseline information during the first round of interviews helps understand prior differences among students, and how these initial differences –coupled with later events—affect the radicalization process. There are also practical and policy reasons why a school-based sample makes sense. First, it helps investigate what schools –a key institution for youth—can preemptively do to support youth before they leave the system. Further, engagement with a large group of youth in a school environment during the first year makes it possible to build rapport in ways that can reduce future attrition and improve the quality of data collected.
 Finally, even if one wanted to generalize the findings to all urban youth, it is worth noting that the selection bias associated with school participation is less severe in urban Cameroon, where education levels are much higher than the national average. Second, much previous work suggests that middle and upper classes are more likely to embrace terrorism, as opposed to those in poverty (Kreuger 2007). For instance, 60% of Palestinian suicide bombers had more than a high school degree, as opposed to only 15 % of the general population. A similar pattern is seen in Pakistan, where analyses of 141 killed militants shows they were nearly all recruited from middle class and well-educated families (Fair 2008; Blair et al. 2013).