Since charter school research on student achievement is mixed, many researchers and policy makers advocate looking inside the "black box" of schools to better understand the conditions under which schools of choice may be effective. We begin to address this issue with data from charter schools and a comparison group of traditional public schools. We also conduct propensity score matching at the student level to further understand achievement gains. In our analyses of these data, we find no charter school effects on students' achievement gains. Instructional conditions, such as teachers' focus on academic achievement, are related to mathematics gains. However, we find that our innovation measure is negatively associated with gains (when other conditions are controlled for), which suggests that innovation for innovation's sake should not be the sole focus of schools, whether charter or not.