GOVT 702: Advanced Political Analysis

Georgetown University

Spring 2021
Mondays and Wednesdays 2 to 3:15 “in” Reiss 152

Professor Michael Bailey

[email protected]
Office hours: Reserve a time slot at https://professorbailey.acuityscheduling.com/schedule.php
Teaching Assistant: Ana Paula Pellegrino <[email protected]>
Labs: [Subject to change] Wednesday 7 to 7:50

Course Description

Being able to understand and analyze data is a part of the skill set for advanced research in
political science. The key is to get beyond merely calculating mathematical quantities and to be
able to understand conditions under which statistical relationships are more likely to represent
real phenomena rather than incidental correlation.

This course builds on the material covered in GOVT 701 to enable students to conduct and
evaluate statistical analysis using advanced tools such as time series analysis, panel data models,
instrumental variables, regression discontinuity and maximum likelihood.

By the end of the semester students should be able to

1) understand the challenge of isolating causal effects via random assignment,
experiments, instrumental variables and other methods;
2) understand and implement advanced statistical techniques such as limited dependent
variable models (such as ordered probit and selection models), panel data models and
instrumental variable models;
3) use R to conduct advanced statistical analysis; and
4) effectively communicate statistical analysis.

Texts and Materials

Bailey, Michael. Real Econometrics. 2nd edition. Oxford University Press (You can also use
Bailey, Michael. Real Statistics which is very similar; I can fill you in on the differences if need
be.)

I will provide lecture notes for lectures via the Canvas system in the course documents section. I
recommend that you print out the lecture notes (print as handouts with 4 to a page if you want to
save paper) before class, as this will free you from copying equations, output or figures.

1
Grading
Course grades will be based on the following:
Problem sets 25%
Participation/quizzes 15%
Presentation 15%
Midterm exam 25%
Final exam 20%

Problem sets
Problem set assignments are available in the assignments folder in Canvas. They are due in
hard copy at the beginning of class on the days indicated. You are strongly encouraged to
work on these problem sets in study groups; however, each student must prepare and
submit his or her own write-up for each problem set (including concise R programs and
output when appropriate). I will provide solutions for the problem sets after they are turned
in. No credit will be given for problem sets submitted after the stated due dates.

The final problem set will be oriented toward you doing a data project in an area of interest.
Please keep an eye out for interesting data/papers/questions over the course of the semester.

Participation and quizzes

Participation really helps you learn the material. It consists of being prepared and being
active. Participation grades are based engagement in class discussions and performance on
modest quizzes based on the readings. I will give you clear warning before each quiz; they
are based on the summary questions at the end of each chapter in the textbook.

The quizzes are based on the readings. I will let you know when they are ahead of time and
the questions will be based on the summary questions at the end of each chapter in the
textbook. The quizzes encourage reading ahead and to offer me a gauge of where you are.

Presentation
The presentation is a 10 minute in-class PowerPoint presentation on a paper related to the
material we are discussing. This presentation should summarize the substantive and
statistical issues addressed in the paper and provide context and a critique. This is a lot to
do and in most cases the presentation will have to isolate and discuss only one element of
the paper. I will provide a list of papers (and dates for presentation). Please email me a
copy of the presentation the day before your scheduled time.

Exams
There will be a midterm and final exam. The final exam will be cumulative but will
emphasize material after the midterm. I will provide practice exams with answer keys.
The dates of the exams are fixed. I cannot accommodate flight schedules for the final
exam; please be sure to plan accordingly.

Due dates
I will provide a grid with the dates for all quizzes, problem sets and labs.

How to succeed in this class

2
 1) Come prepared to class. Do the readings. Think about the readings on their own terms,
but also in terms of how the concepts apply to things you are interested in. This is a great
time to be greedy: ask “how will this help me?”

2) Ask questions.
a) Formulating a question helps you engage with the material much more deeply.
We all have experienced the thought “this doesn’t make sense.” Taking that
thought to the next level and formulating what doesn’t make sense pushes us to
deeper understanding.
b) If you have a question, it’s almost certain that others do too; asking a question
will not only help yourself, but you will help others.
c) Asking questions helps keep the class on track. If there are lots of questions,
we’ll slow down and get things figured out. If there are few questions, we’ll
charge ahead.
d) Asking questions helps you develop a voice in statistics. You’ll start to have
opinions about the right way to do things and will ask questions resisting other
ways. That’s great! That’s how we learn.
e) Questions make things interesting. We have to take a concept that may seem
straightforward and apply it on the fly to a new situation or probe what exactly is
going on. That’s fun.

3) Use the discussion questions in the readings. We’ll do some as exercises in class. Work
through the others on your own or with your study groups. Statistics needs to be learned
by experience. You won’t really know if you understand something until you are able to
work with it, apply it to new context or do something active. Simply reading and
highlighting is not enough.

4) Work in groups, but do so wisely. Working with classmates is a great way to learn from
each other. Often, classmates will have some way of explaining things that clicks for
you. And, more often, the act of explaining something to someone else will make things
click for you. This only works, though, if you prepare by yourself first. If you show up
and wait for classmates to do the work, you can probably muddle through the exercises,
but you’ll have trouble participating in classes and may fall behind as the material we
cover cumulates and needs to be understood at each step.

5) Start assignments early. Sometimes the data doesn’t cooperate; you don’t want to find
this out at 11pm the night before the problem set is due. Also, the more you are doing
exercises, the more you will be able to follow the lectures.

Technology
For in person classes (if applicable!): Laptops, iPads, and iPhones are not allowed during lectures or exams
except when we are doing lab or group work. Assume that computers/phones are not allowed until I tell you
otherwise.

Academic honesty
Plagiarism or other acts of academic dishonesty will not be tolerated. Cases of suspected academic dishonesty
will be handled according to the university’s honor code. Please review Georgetown’s guide to understanding
and avoiding plagiarism.

COURSE OUTLINE BY WEEK

3
January 25 Course introduction
Real Econometrics/Stats foreword for students and chapters 1-2

OLS review
Topics: Review of OLS estimation and interpretation, omitted variables,
multicollinearity, dummy variables
Reading: Real Econometrics/Stats chapters 5- 7
Arronow and Samii (2016) Does Regression Produce Representative
Estimates of Causal Effects?

February 1 OLS review continued and post-treatment variables (Section 7.3)

Post-treatment
Topics: Mediator and collider bias
Readings: “PostTreatment_Dec2019.pdf” in Canvas
Dablander (2019) An Introduction to Causal Inference
Supplemental: Montgomery, Nyhan, Torres (2018) How Conditioning on
Posttreatment Variables Can Ruin Your Experiment and What to
Do about It

February 8 Panel data

Topics: Pooled models, one-way and two-way fixed effects models,
difference-in-difference approaches
Reading: Real Econometrics/Stats chapter 8
Optional: Erikson, Robert S., Pablo M. Pinto and Kelly T. Rader. 2014.
Dyadic Analysis in International Relations: A Cautionary Tale.
Political Analysis 22:457–463.

February 15 Panel data (continued)

February 22 Instrumental variables/ 2 stage least squares

Topics: Instrumental variables, 2SLS (“two stage least squares”),
overidentification, weak instruments, simultaneous equations
Reading: Real Econometrics/Stats chapter 9

March 1 Instrumental variables/ 2 stage least squares (continued)

March 8 Experiments
Topics: Randomization, balance, compliance, 2SLS for experiments,
attrition, natural experiments
Reading: Real Econometrics/Stats chapter 10

March 15 Midterm exam

March 17 (Wed). Regression discontinuity

Topics: Basic model, flexible models, diagnostics
Reading: Real Econometrics/Stats chapter 11

March 22 Regression discontinuity (continued)

4
March 29 SPRING BREAK

April 5 Time series

Topics: Autocorrelation, dynamic models
Reading: Real Econometrics/Stats chapter 13

April 12 Time series (continued)

April 19 MLE: probit and logit

Topics: Probit
Reading: Real Econometrics/Stats chapter 12

April 26, May 3 We will decide together on topics for the last 2 weeks
Option 1: Selection models
Topics: Missing data, endogenous selection, Heckman selection models
Reading: Available in Canvas

Option 2: Text as data

Topics: Organizing and cleaning text data, similarity measures, ideology
Reading: Available in Canvas

Final exam
Saturday May 15, 2021 from 12:30pm to 2:30pm, location TBD