Includes bibliographical references (pages 120-138).
1. Introduction -- 2. Causality: forming an evidence base -- 3. Estimating causal effects using observational data -- 4. Analysis of large-scale datasets: examples of NSF-supported research -- 5. Conclusions and recommendations.
0
"The ... report is designed to help researchers, policymakers and funders understand the capacities and limits of examining the causes of educational outcomes with large-scale databases. It is intended to help researchers consider a range of methods designed to establish causality, from experimental to non-experimental designs. For example, the ordinary question "Did A cause B" takes on great significance when applied to education intervention programs. Can small class size, or highly qualified teachers, be shown to have caused increased or decreased learning? With what degree of confidence? ... The report specifically: Considers key issues involved in selecting research designs that allow researchers to draw valid causal inferences about treatment effects using large-scale observational datasets; Addresses why key issues of establishing causal inference are of particular interest to education researchers, briefly explains how causality is commonly defined in scholarly literature, and describes tools that analysts use to approximate randomized experiments with observational data; Reviews four National Science Foundation-funded studies which illustrate the difficulties of and possibilities for making causal inferences when conducting studies focused on significant education issues"--Publisher's website.