Course Instructor: Nan Zhang (Office hours by appointment)
Course Website: https://nanzhangresearch.github.io/Applied_Causal_Inference
Date and Time: Fridays, 13:45-15:15 in C 116 Seminarraum (A 5, 6 Bauteil C)
Course Aims: This course will provide students with hands‐on practice in using data from experimental and quasi‐experimental studies to estimate treatment effects. Specifically, the course will provide training in:
The course is designed for MA and PhD students with prior training in applied multivariate regression techniques.
Course Structure: The course is structured in two-week “units.” Typically, the first class meeting of each unit will include a lecture describing a specific method of analysis. Students then apply what they have learned by working together on problem sets outside of class. The second class session of each unit will then be devoted to student presentations of their analyses and Q&A.
Auditing: This is a hands‐on course in statistical analysis. Auditing a class like this without completing the assignments will not be productive for you. Consequently, auditors will not be permitted.
1. Required Readings. Every two weeks, students should complete a required reading before lecture. These readings provide an applied context for the lecture material, and the lectures themselves will assume that you have completed the reading.
2. Problem sets. Every two weeks, students will complete a problem set based on an analysis of a dataset that I provide. There will be 5 such problem sets in total. Students will work in groups on all the problem sets.
Students should complete their problem sets in R, and problem set answers with clearly annotated code should be written in RMarkdown.
Problem set answers should be submitted to me via email by 8am before the second class meeting in each unit. Each group will then present a solution to part of the problem set during this class meeting.
3. Active class participation. Attendance is expected at all course meetings. Of course, situations could arise where you need to miss class. As a courtesy, please let me (and others in your group) know beforehand if you cannot attend a class session.
In addition to attendance, students are expected to actively participate in class. We will be working intensively with examples in class, and active engagement with the material is essential. Your participation also signals what you find confusing or challenging, such that I can adjust the material accordingly.
4. Graded Final Assignment. The final assignment will consist of an independent original data analysis project applying one of the following methods we learned in class: IV, RDD, DiD, or Matching (or some combination thereof). You must complete this assignment on your own. The last two weeks of the semester will be devoted to workshopping your ideas. You should submit this assignment to me via email by 31 January 2025.
Note: the current version of the syllabus is a work in progress (so some links may not yet work).
Session 1 (6 Sept): Introduction: Why Causal Inference?
Session 2 (13 Sept): Potential Outcomes and
Uncertainty
Fundamental problem of causal inference. Randomization and average
treatment effects (ATE). Sampling distribution of the ATE. SUTVA.
Randomization inference.
Background Reading: Gerber and Green. 2012. Field Experiments: ch2, ch3.1-ch3.2
Problem Set 1, pre-post design data, corruption data and senators data.
Session 3 (20 Sept): Student Presentations
Session 4 (27 Sept): Instrumental Variables One- and two-sided non-compliance. Intent-to-treat effects, compliance rate, and complier average treatment effects. 2SLS estimation. IV assumptions.
Reading: Clingingsmith et al. 2009. Estimating the Impact of the Hajj: Religion and Tolerance in Islams Global Gathering
Problem Set 2 and data
Session 5 (4 Oct): Student Presentations
Session 6 (11 Oct): Regression Discontinuity
Reading: Myerson. 2014. Islamic Rule and the Empowerment of the Poor and Pious
Problem Set 3 and data
Session 7 (18 Oct): Student Presentations
Bonus Material on Unexpected Events During Surveys Designs
Session 8 (25 Oct): Diff-in-Diff and Two Way Fixed Effects (Intro)
Reading: Dinas et al. 2019. Waking Up the Golden Dawn
Problem Set 4 and data
No Class: 1 November (All Saints’ Day)
Session 9 (8 Nov): Student Presentations
Session 10 (15 Nov): Matching
Reading: Kam and Palmer. 2008. Reconsidering the Effects of Education on Political Participation
data, .Rmd file, and compiled pdf.
Session 11 (22 Nov): Student Presentations
Sessions 12 and 13 (29 Nov and 6 Dec): Workshop Final Paper Ideas
Although I try not to assign “methods” readings, you may find the following resources useful:
Alan Gerber and Donald Green: Field Experiments: Design, Analysis and Interpretation. Basically my go-to reference.
Nick Huntington-Klein: The Effect: An Introduction to Research Design and Causality. Available as ebook with accompanying Youtube videos.
Thad Dunning: Natural Experiments in the Social Sciences: A Design-Based Approach. Already a bit outdated, but gives a nice overview of our topics.
Scott Cunningham: Causal Inference: The Mixtape. A lot of people like this. I find the writing style a bit difficult.
Joshua Angrist and Jörn-Steffen Pischke: Mastering ’Metrics. A good easy introduction to many of the topics we cover. If you are feeling more ambitious, check out Mostly Harmless Econometrics.