Absolutely, correlation does not imply causation. While two variables may be correlated, it does not necessarily mean that changes in one variable cause changes in the other. This principle underscores the importance of employing robust causal inference methods in impact evaluation to accurately determine the effects of interventions or policies.

Understanding the difference between correlation and causal inference is vital in impact evaluation to ensure accurate assessment of interventions’ effectiveness, informing evidence-based decision-making and policy formulation.

This workshop focuses on quasi-experimental methods, including Difference-in-Differences (DID), Instrumental Variables (IV), and Regression Discontinuity Design (RDD). Quasi-experimental designs are powerful tools for evaluating the impact of interventions when randomized controlled trials are not feasible. Throughout this workshop, participants will gain insights into the theory behind these methods and practical applications in evaluating program effectiveness. Join us as we delve into innovative approaches to assessing causal relationships and informing evidence-based decision-making. This workshop will cover deeply on:

The difference-in-differences (DD) is a quasi-experimental approach to obtain the causal impact of government policy. The technique compares changes in outcomes of the exposed group before and after policy relative to the changes in outcomes of the unexposed group. The workshop will start with the basic canonical DD model, its basic assumptions, and some recent examples. If time permits, the workshop will touch on recent developments in the DD model.

The Instrumental variables (IVs) as a causal inference tool are used to control for selection-on-unobservable in observational studies and correcting the non-successful delivery of treatment to every unit in a randomized experiment. The core of the technique based on four LATE theorem elements: exogeneity, exclusion, relevance, and monotonicity.

The Regression Discontinuity Design (RDD) is widely used to evaluate programs or policies that have cut-off points, which determine who is eligible to participate and become the treatment group and who is ineligible (control group). RDD methods can be used to compare groups above and below the cut-off point and estimate the average treatment effect or the program’s impact. In this topic, we will cover two methods in the RDD design: sharp and Fuzzy RDD.