Raudenbush, S. W. (2008). Advancing educational policy by advancing research on instruction. American Educational Research Journal, 45(1), 206-230.
Understanding the impact of "instructional regimes" on student learning is central to advancing educational policy. Research on instructional regimes has parallels with clinical trials in medicine yet poses unique challenges because of the social nature of instruction: A child’s potential outcome under a given regime depends on peers and teachers, requiring the need for multilevel methods of causal inference. The author considers studies of the impact of intended versus experienced instructional regimes. Both are important; however, intended regimes are well measured and accessible to randomized trials, whereas experienced instruction is measured with error and not amenable to randomization. Multiyear sequences of experienced instruction are of central interest but pose special methodological challenges. A 2-year study of intensive mathematics instruction illustrates these ideas.