Instrumental variables is a popular approach for causal inference in education when randomization of treatment is not feasible. Using a first-year college program as a running example, this article reviews the five assumptions that must be met to successfully use instrumental variables to estimate a causal effect with observational data: SUTVA,…

The clustered observational study (COS) design is the observational study counterpart to the clustered randomized trial. In a COS, a treatment is assigned to intact groups, and all units within the group are exposed to the treatment. However, the treatment is non-randomly assigned. COSs are common in both education and health services research. In…

This case describes a current evaluation of an educational intervention to help disadvantaged children catch up in literacy at around the time they transfer to their senior school. The evaluation is based on a randomised controlled trial design, which the case explains is the most appropriate for testing or establishing a causal claim. The case…

An important target of education policy is to improve overall teacher effectiveness using evidence-based policies. Randomized control trials (RCTs), which randomly assign study participants or groups of participants to treatment and control conditions, are not always practical or possible and observational studies using rigorous quasi-experimental…

The authors thank Dr. Lindsay Page for providing a nice illustration of the use of the principal stratification framework to define causal effects, and a Bayesian model for effect estimation. They hope that her well-written article will help expose education researchers to these concepts and methods, and move the field of mediation analysis in…

Given the different possibilities of matching in the context of multilevel data and the lack of research on corresponding matching strategies, the author investigates two main research questions. The first research question investigates the advantages and disadvantages of different matching strategies that can be pursued with multilevel data…