Low-Cost Randomized Controlled Trials in Education

摘要

Despite the increased frequency of randomized field experiments in the social sciences, they are relatively uncommon in educational research. For instance, Alpert, Couch, and Harmon (2016) conducted a randomized controlled trial (RCT) to estimate the impact of online education in a college-length course and referenced only three previous studies that did the same.A likely explanation for the lack of RCTs in education is their high cost. A study sponsored by the US Department of Education used an RCT to compare the relative efficacy of four elementary school math curricula, recruiting 110 schools for participation in the study (Agodini et al. 2010). While such large-scale studies may be justified from the high value of robust evidence on established, scalable, and highly standardized educational interventions, education researchers often evaluate techniques that are not well-established or fully reproducible, modifications of existing practices, or practices implemented in different disciplines or settings than have been evaluated previously. Bowen et al. (2013) conduct a large-scale RCT of online learning but also acknowledge that "online learning" refers to a vast array of interventions, requiring any one study to narrow the scope of the programsconsidered. Indeed, critics of large-scale randomized trials of educational interventions highlight their high costs and limited ability to generalize findings (Thomas 2016).