Скачать или смотреть Replicability, Generalizability, and the ManyClasses Approach to Open Science

As in many fields, applied research in education is facing serious questions about the replicability of its findings — and further, some ponder whether findings should be expected to replicate in the first place. In education, a particularly vexing challenge is the variability of learners and learning settings, and it is a legitimate question whether any one finding might be expected to generalize across classes. To answer this question, the obvious solution is to conduct research that spans and incorporates this variability. Rather than embedding an experiment in one class, one discipline, or one domain, researchers should distribute an experiment across many such situations, customizing treatments and outcome measures to authentic local contexts, so that they are representative of a range of materials and implementations.

This talk will present ManyClasses, a research model that achieves this need, and it will also describe how the principles of open science are fundamental to the ManyClasses approach. The presenter’s work, which aims to test the generalizability of practical interventions, relies heavily on transparency both in our methods and in our findings.