Scientific Coordination
Alisa Remizova
Administrative Coordination
Claudia O'Donovan-Bellante
Please wait...
Foundations and Advances in Difference-in-Differences
About
Location:
Online via Zoom
Online via Zoom
General Topics:
Course Level:
Format:
Software used:
Duration:
Language:
Fees:
Students: 220 €
Academics: 330 €
Commercial: 660 €
Causal inference, Difference-in-differences, Synthetic control group
Additional links
Lecturer(s): Jan Marcus, Nicolas Frink (Teaching Assistant)
Course description
This two-day workshop offers a comprehensive introduction to Difference-in-Differences (DiD), a widely used method for causal inference in the social sciences, economics, and policy evaluation. We begin by introducing the foundations of causal inference through the lens of the potential outcomes framework, setting the stage for understanding the logic behind DiD. From there, we explore the core DiD design, focusing on its assumptions, intuitive appeal, and implementation using statistical software (Stata, R).
Building on this foundation, we look into several extensions and refinements of the DiD approach. Topics include Triple Differences (DDD), event study designs, and challenges in estimating standard errors.
The course further introduces the Synthetic Control Method, an influential extension of the DiD approach. In the final sessions, we focus on recent advances in DiD, especially in settings with staggered treatment adoption. We highlight key issues with the conventional two-way fixed effects estimator, introduce the Bacon decomposition, and present modern estimators (including the Callaway and Sant'Anna estimator) that provide more robust results.
The workshop will be taught using both Stata and R. You can choose the preferred software you want to use during the workshop.
Organizational structure of the course
Lectures and exercise sessions in Stata and R will be used alternately. During the exercise sessions, you will primarily engage in hands-on tasks using either Stata or R, applying the concepts introduced in the lectures.
Target group
You will find the course useful if:
- You conduct empirical research and are interested in causal inference, particularly in observational settings where randomized experiments are not feasible.
- You work with panel data or repeated cross-sectional data and want to evaluate the impact of policies, treatments, or interventions over time.
- You are a researcher or practitioner in fields such as economics, political science, public policy, sociology, health economics, education, or development studies, where difference-in-differences and related methods are widely applied.
- You are looking for a comprehensive hands-on introduction to DiD techniques.
- You want to understand and implement advanced DiD methods, including event studies, triple differences, synthetic controls, and recent estimators for staggered treatment adoption.
Learning objectives
By the end of the course, you will:
- Understand the fundamental concepts of causal inference, including the potential outcomes framework and the logic of Difference-in-Differences (DiD) designs.
- Be able to identify and assess the assumptions underlying DiD methods, including parallel trends.
- Gain practical skills to implement DiD analyses in statistical software (e.g., R or Stata), including how to structure data, estimate models, and interpret results.
- Learn to apply advanced extensions of DiD, such as Triple differences (DDD) and Event study designs for dynamic treatment effects.
- Understand and implement the Synthetic Control Method, including its conceptual foundations and appropriate use cases.
- Recognize the limitations of the traditional two-way fixed effects estimator in settings with staggered treatment timing.
- Acquire knowledge of recent methodological developments, including the Bacon decomposition and modern estimators (e.g., Callaway & Sant'Anna) for staggered adoption designs.
- Be able to critically evaluate empirical studies using DiD and related methods, identifying strengths, weaknesses, and appropriate design choices.
Prerequisites
- Basic knowledge in regression analysis.
- Basic knowledge of either R or Stata (e.g., importing and manipulating data, performing analyses).
Software requirements
You should have either Stata or R and RStudio installed prior to the workshop.
If you require access to Stata, please contact GESIS the latest two weeks before the workshop begins so we can arrange it in time.
