Scientific Coordination

This course will introduce you to the concepts and methods of causal inference and causal modeling in the social sciences. It will highlight the relevance of research design, analytical methods, and their systematic combination to optimize the validity of causal inferences drawn from empirical studies. Building on existing knowledge concerning linear regression modelling and research design, the course will then cover key methods to estimate causal effects, including fixed effects estimations with various addons, event study analyses, matching, difference-in-differences, regression discontinuity, and instrumental variables. Throughout the course, you will apply these concepts and methods in hands-on sessions to real-world examples in the social sciences. The application will be conducted with the statistical software package Stata. A solid background in Stata is expected. By the end of the course, you will have the skills and knowledge to design, conduct, and interpret causal inference studies in the social sciences. You will be able to engage with the contemporary literature of causal inference and identify state-of-the-art methods which might be most relevant to your specific research question.

For additional details on the course and a day-to-day schedule, please download the full-length syllabus.

Organizational structure of the course

The course will be split into three-hour morning and three-hour afternoon sessions, including coffee breaks. To secure a close link between the learning and the application of contents, we will switch between lecture format (~50%) and hands-on exercises, tutorials, or lab sessions (~50%) in a flexible way. The exercises include the application of causal inference methods to estimate effects based on existing datasets using Stata. Lecturers will be available for individual consultations to support work on group assignments and to facilitate discussions within groups.

Target group

You will find the course useful if you:

have a background in the social, behavioral or economic sciences (economists, political scientists, sociologists, criminologists, psychologists, etc.),

are interested in methods for causal inference based on observational data, especially panel data,

have a firm knowledge in linear regression modelling,

are motivated to apply the concepts and statistical approaches in hands-on sessions.

Learning objectives

By the end of the course, you will:

have acquired an understanding of up-to-date causal estimation approaches to understand and critically reflect on current studies,

be able to conceptually design your own studies to identify causal effects,

bring the conceptual considerations into practice and conduct the relevant estimations.

Prerequisites

Knowledge of basic statistical concepts and their formal background, including the principles of (multiple) linear regression and binary logistic regression

Solid background in Stata

Basic understanding of designing quantitative studies

Software and hardware requirements

You will need to bring a laptop computer with a recent version of Stata (15 or higher) installed to successfully participate in this course.

GESIS will provide you with short term licenses for Stata for the duration of the course if needed.

To familiarize yourself with the statistical software Stata, you can use the following textbooks:

Kohler, U./Kreuter, F. (2008). Data Analysis Using Stata. College Station: Stata Press.

Cameron, A. C., and Trivedi, P.K., 2022. Microeconometrics Using Stata, Stata Press.

For an introduction or refresher in Stata programming, you might also consider enrolling in GESIS's two-day hybrid (onsite in Cologne/online via Zoom) course, “ Introduction to Stata for Data Management and Analysis ” held in the first week of the Summer School.

Schedule

Day 1 (10:00-17:30)

A primer on regression analysis and an introduction to causal identification

Quick update on Ordinary Least Squares (OLS) estimation

Counterfactual model and the fundamental problem of causal inference

Different types of experiments and their link to the counterfactual model

OLS with control variables

Stata application: OLS with controls

Good controls, bad controls

Day 2 (09:00-16:30)

Cross sectional Data

What is your estimand?

ATE, ATT, LATE

Matching: (Coarsened) Exact Matching, Propensity Score Matching

Stata application: Matching

Literature:

Iacus, S. M., King, G., & Porro, G. (2012). Causal inference without balance checking: Coarsened exact matching. Political analysis, 20(1), 1-24. https://doi.org/10.1093/pan/mpr013

Lundberg, I., Johnson, R., and Stewart, B.M. (2021). What Is Your Estimand? Defining the Target Quantity Connects Statistical Evidence to Theory. American Sociological Review, 86, 532-565. https://doi.org/10.1177/00031224211004187

Morgan, S.L., and Winship, C. (2015). Counterfactuals and Causal Inference: Methods and Principles for Social Research. New York: Cambridge University Press (2nd Edition). Chapters 4, 5 & 6. https://doi.org/10.1017/CBO9781107587991

Day 3 (09:00-16:30)

Panel Data

Longitudinal (in particular panel) designs and data quality

Panel regression models: Pooled OLS, first-differences, and fixed-effects regression

Fixed-effects regression with individual slopes

Stata Application: Panel regression models

Event-study designs and impact dummies

Literature:

Allison, P.D. (2009). Fixed Effects Regression Models. Thousand Oaks: Sage.

Brüderl, J., and Ludwig, V. (2014). Fixed-effects Panel Regression. In: Best, H., and Wolf, C. (eds.): The Sage Handbook of Regression Analysis and Causal Inference. London: Sage, pp. 327-357.

Collischon, M., & Eberl, A. (2021). The link between relative pay and job satisfaction revisited. European Sociological Review, 37(2), 238-252.

Ludwig, V., and Brüderl, J. (2021). What You Need to Know When Estimating Impact Functions with Panel Data for Demographic Research. Comparative Population Studies, 46, DOI: 10.12765/CPoS-2021-16. https://doi.org/10.12765/CPoS-2021-16

Day 4 (09:00-16:30)

Natural Experiments: RDD and DiD

Stata application: Event-study designs and impact dummies

Natural experiments I: Regression discontinuity

Natural experiments II: Diff-in-diff-estimation

Stata application: Diff-in-diff-estimation

Literature:

Cunningham, S. (2021). Causal Inference. The Mixtape. New Haven/London: Yale University Press. Chapters 6 & 9. https://mixtape.scunning.com/

Dunning, T. (2012). Natural Experiments in the Social Sciences. A Design-Based Approach. Cambridge, UK: Cambridge University Press. Chapters 1, 2 & 3. https://doi.org/10.1017/CBO9781139084444

Day 5 (09:00-16:00)

Instrumental Variables & Outlook

Natural experiments III: Instrumental variables

Stata application: Instrumental variables

Instrumental variables and fuzzy regression discontinuity designs

Wrap-up and outlook

Literature:

Angrist, J.D., and Pischke, J.-S. (2009). Mostly Harmless Econometrics. An Empiricist's Companion. Princeton: Princeton University Press. Chapter 4.

Scientific Coordination

Administrative Coordination

Course 4: Causal Inference in the Social Sciences

About the lecturer - Matthias Collischon

About the lecturer - Matthias Collischon

About the lecturer - Florian Zimmermann

About the lecturer - Florian Zimmermann

Course description

Target group

Learning objectives

Prerequisites

Schedule

Schedule

Recommended readings

Recommended readings