Research Design I:
Research Design and Causal Inference
Peter Selb
Winter 2025/26
E-mail: [Link]@[Link] Web: [Link]/cdm/
Class hours: Tue 11:45-13:15 Office hours: by appointment
Class room: R-611 Office: D-335
Description
This course offers an advanced treatment of design issues in social science research that aims at
answering cause-and-effect questions of the general form: Is variable X a cause of outcome Y ? If so,
how large is the effect of X on Y ? And how do changes in X translate into changes in Y ? Starting
from an exposition of the potential outcomes framework and causal graphs, the course introduces
the assumptions necessary for identifying various causal effects and shows how these assumptions
are justified in different experimental and observational research designs. Regarding observational
studies, the course provides an overview of large-N methods for causal inference, such as matching,
regression and panel estimators, instrumental variable and regression discontinuity designs. The
course also discusses how the principles and methods introduced may be applied in small-N settings
and in studies that aim to parse the mechanisms underlying causal effects.
Goals
The course’s primary aim is to provide students with the methodological tools to critically evaluate
existing empirical work and to develop their own research designs that strengthen causal claims to
the greatest possible extent.
Organization
There is a course repository on ILIAS which contains all relevant materials ([Link]
[Link]/[Link]?target=crs_1950926.html).
The course covers 12 key topics in causal inference, one per week. Each weekly cycle includes:
1. lecture,
2. homework assignment,
3. reading task,
1
4. tutorial.
At the end of each lecture, I will set a homework assignment in which you will apply the
concepts and methods to problems and data. Collaboration on these exercises is encouraged. How-
ever, you will have to write up and submit your solutions individually by Sunday after the lecture
23:55 on ILIAS (folder “Homework assignments”).
The reading task requires you to read and annotate a research paper. We will use Perusall
([Link] a social annotation platform, for the reading tasks. “Social” means that
you can see the marks, questions, and comments your fellow students made. You are invited to
give feedback to their annotations. Annotations are also due on Sunday 23:55. The first reading
task is due on Nov. 16. Please create an individual account on Perusall by Oct. 31. Follow this
link to register: [Link] You will need your to use your
institutional email address (. . . @[Link]). Please also indicate your student ID number
(01/. . . ). The use of chatbots for homework assignments and reading tasks is not permitted!
A weekly tutorial serves to clarify your questions regarding the lectures, present individual
solutions to the homework assignment (be prepared!), and discuss the research paper from the
reading task as well as other applications. The tutorials will be instructed by Clemens Nollenberger
and Tyler Brown. You can choose your preferred slot (unless the distribution of students across
groups gets very skewed):
• Group 1: Mondays, 11:45-13:15, in M-701 (Tyler)
• Group 2: Wednesdays, 13:30-15:00, in L-602 (Tyler)
• Group 3: Thursday, 13:30-15:00, in BG-335 (Clemens)
• Group 4: Thursday, 17:00-18:30, in E-403 (Clemens)
Finally, we have set up a Forum on ILIAS where you can post both substantive and organiza-
tional questions. We encourage you to use the Forum (and not to send individual emails to me or
the tutors) for any concern which might be of interest for your fellow students as well. You may
post your contributions anonymously. Please mind the Forum’s structure!
Grading
There will be a written exam on Thursday, 12.02.26, 8:15-9:45. The exam will contribute
70% to your final grade. The exam will consist of multiple choice questions, arithmetic problems,
and one or two open ended questions. Examples will be given during the semester. The exam will
be written online on laptops provided by the university. Please check the department’s website
for updates on exam dates, rooms etc. ([Link]
service/forms/dates/).
Perusall has an integrated assessment tool which automatically evaluates student’s participation
in the reading tasks (though the evaluation settings are mine!). The summary score will enter
your final grade. Active engagement with the reading tasks and regular submission of homework
assignments will make for 30% of the final grade. There are 9 reading tasks on which you can
score up to 5 points each, the best 8 of which will be considered for grading (40 overall). Also,
there are 9 homework assignments. Each failure to submit a homework assignment in time will lead
to a two-point reduction of the overall reading score before grading (you will have one free shot).
Otherwise the rule will be strictly enforced unless you have a medical certificate.
2
If you have any special needs regarding the exam or the course in general, please email me early in
the term!
Prerequisites
While the course’s focus is on causal – not statistical – inference, some knowledge of statistics
(substantiated through successful completion of the BA class Introductory Statistics or equivalents)
is required. Those not yet familiar with basic probability, linear and generalized linear models are
highly recommended to attend Michael Herrmann’s Preparatory Course. A short and accessible
introduction to research design and causal inference in the social sciences is given in
King, Gary, Robert Keohane, und Sidney Verba. 1994. Designing Social Inquiry. Scientific
Inference in Qualitative Research. Princeton University Press. Chapter 3. (linked in the
“Textbooks” folder on ILIAS)
Textbooks
We do not use a single textbook. In developing the course, I relied heavily on the first edition of
Morgan, Stephen L., and Christopher Winship. 2015. Counterfactuals and Causal Inference.
Methods and Principles for Social Research. 2nd ed. Cambridge University Press,
but that changed in later iterations of the course. However, the legacy remains and I am still
referring to relevant chapters in the schedule below. Equivalent materials from different disciplinary
perspectives can be found in other textbooks on causal inference, some of which I have linked in
the folder “Textbooks” on ILIAS. If you want a textbook, take a look and decide for yourself which
one suits you best.
Preliminary schedule
Text categories:
A: Application to be annotated on Perusall
B: Background reading for further study, not mandatory
I. INTRODUCTION
21.10. Overview, organizational matters
B Morgan and Winship, chapter 1.
28.10. Potential outcomes
B Holland, Paul W. 1986. Statistics and causal inference. Journal of the American Statistical
Association 81: 945-960.
B Morgan and Winship, chapter 2.
3
04.11. Causal graphs
B Rohrer, Julia M. 2018. Thinking clearly about correlations and causation: Graphical causal
models for observational data. Advances in Methods and Practices in Psychological Science 1:
27-42.
B Morgan and Winship, chapter 3.
II. EXPERIMENTS
11.11. Randomization designs
A Chen, Jidong, Jennifer Pan, and Yiqing Xu. 2016. Sources of authoritarian responsiveness:
A field experiment in China. American Journal of Political Science 60: 383-400.
B Hernán, Miguel A. and James M. Robins. 2023. Causal Inference: What If ? Chapman and
Hall/CRC, chapter 2 (see folder “Textbooks” on ILIAS)
18.11. Noncompliance and attrition
A Imai, Kosuke. 2005. Do get-out-the-vote calls reduce turnout? The importance of statistical
methods for field experiments. American Political Science Review 99: 283-300.
B Morgan and Winship, chapter 9.
III. CONDITIONING ON OBSERVABLES
25.11. Matching estimators
B Morgan and Winship, chapter 5
A Fjelde, Hanne, and Hannah M. Smidt. 2022. Protecting the vote? Peacekeeping presence and
the risk of electoral violence. British Journal of Political Science 52: 1113-1132.
B Morgan and Winship, chapter 5
02.12. Regression estimators
A Keele, Luke, Randolph T. Stevenson, and Felix Elwert. 2020. The causal interpretation of
estimated associations in regression models. Political Science Research & Methods 8: 1-13.
B Morgan and Winship, chapter 6
09.12. Beyond simple conditioning: interrupted time-series, panel and difference-in-
differences estimators
A Selb, Peter, and Simon Munzert. 2018. Examining a most likely case for strong campaign
effects: Hitler’s speeches and the rise of the Nazi party, 1927-1933. American Political Science
Review 112: 1050-1066.
B Morgan and Winship, chapter 11.
4
IV. NATURAL EXPERIMENTS
16.12. Regression discontinuity designs
A Becker, Sascha O., Peter H. Egger, and Maximilian von Ehrlich. 2010. Going NUTS: The
Effect of EU Structural Funds on Regional Performance. Journal of Public Economics 94:
578-590.
B Dunning, Thad. 2012. Natural Experiments in the Social Sciences. A Design-Based Approach.
Cambridge University Press, chapter 3. (see folder “Textbooks” on ILIAS)
23.12. Holiday, no class
30.12. Holiday, no class
06.01. Holiday, no class
13.01. Instrumental variable estimators
A Kern, Holger L., and Jens Hainmüller. 2009. Opium for the Masses: How Foreign Media Can
Stabilize Authoritarian Regimes. Political Analysis 17: 377-399.
B Dunning, Thad. 2012. Natural Experiments in the Social Sciences. A Design-Based Approach.
Cambridge University Press, chapter 4. (see folder “Textbooks” on ILIAS)
V. CURRENT TOPICS
20.01. Causal mechanisms
A Didden, Christiane. 2025. Targeting mediating mechanisms of social disparities with an
interventional effects framework, applied to the gender pay gap in Western Germany. Journal
of Causal Inference 13: 1-23.
B Morgan and Winship, chapter 10.
27.01. Synthetic controls for case studies
A Masi, Tania, and Roberto Ricciuti. 2019 The heterogeneous effect of oil discoveries on democ-
racy. Economics & Politics 31: 374-402.
B Abadie, Alberto, Alexis Diamond, and Jens Hainmüller. 2015. Comparative Politics and the
Synthetic Control Method. American Journal of Political Science 59: 495-510.
03.02. Generalization of study results, Q&A
B Lesko, Catherine R., Buchanan, Ashley. L., Westreich, Daniel, Edwards, Jessie K., Hudgens,
Michael G., and Cole, Stephen R. 2017. Generalizing study results. Epidemiology 28: 553-561.