Instructional Approaches


The proposed workshop addresses design and analytic issues for four main quasi-experimental approaches—the regression-discontinuity (RD) design, the interrupted time series (ITS) design, the nonequivalent control group (matching) design, and the instrumental variable ( IV) design. For each of the four quasi-experimental designs, the workshop discusses the theory of the design, required assumptions for producing valid results, and recent developments in the analysis and implementation of the method.

To help link theory to practice, participants will have opportunities to conduct hands-on analyses of quasi-experimental data. The applied sessions consist of a series of standalone modules that address introductory to advanced analytic topics for each research design. Examples of introductory modules for the RD design include: preparing the data for an RD analysis; graphical inspection of the data; checking assumptions in an RD design; and parametric and non-parametric methods for estimating treatment effects. Advanced modules for the RD design include calculating power in an RDD; extrapolating treatment effects away from the cutoff; and analyzing RD designs with multiple sites, cutoffs, and assignment variables. For participants with little or no experience analyzing a particular research design, we provide a recommended order for working through the modules. Advanced participants may select module topics that address their own areas of interest. 

Data work will be conducted in R and STATA. We will provide sample code for analyses as well as data, but attendees should have access to either R or STATA on their own laptops that they will bring to the training. Familiarity with R or STATA is not required, but it is helpful if participants have experience with similar software, such as SAS or writing syntax files in SPSS.