Core Subject: Advanced quantitative methods in the study of political behavior

Overview

This course is part of the specialization in Political Behavior and Applied Quantitative Methods, and provides a broad overview of quantitative methods as they are commonly used to study political behavior. Its primary focus is on methods for causal inference: randomized controlled trials, instrumental variables, difference-in-differences, and regression discontinuity.

The course is designed to interact with Political Behavior—the specialization’s substantive course—and will incorporate its texts and themes. The course will examine some of the best-known recent work on political behavior, and ask: How did the authors produce their results? Can we re-produce them? And how sensitive are the results to other methodological strategies? The course will place particular emphasis on students’ ability to apply the methods introduced in the course to real-world data.

In particular, the course revolves around three sections, as described below.

1. Statistical programming in R

The first classes (classes 1-4) introduce the statistical programming language R, the most widely used statistical software among political scientists who use advanced quantitative methods. Like Python, R’s data science counterpart, R is also widely used in government and the private sector for data analysis. For applied academic, government, and private-sector research, knowing R (and Python) will likely be the most valuable technical tools you learn from your time in university. Accordingly, you will use R in every class throughout the term. You will learn R in class, and also through exercises on DataCamp, an intuitive platform for learning a variety of statistical programming languages and methods.

The first section of the course will also re-introduce linear regression analysis (OLS), which you should be familiar with from your undergraduate methods sequence, and its implementation in R. Finally, I will introduce you to LaTeX, a type-setting system used by the majority of researchers who do advanced quantitative research in political science, and which may be very useful when you write your assignments and, later, your Master’s thesis.

2. Causal inference

The second section of the course will focus on causal inference (classes 5-11), and recent methods developed for estimating causal effects. The social sciences (particularly economics and political science) have recently undergone a “ credibility revolution” that places methods for estimating causal effects at the center of research design. This is because theoretically driven empirical research often centers on questions of a causal nature. As a result, much of the course will emphasize how to think about causality conceptually, and the statistical techniques for causal inference.

3. Data collection

The third section of the course will focus on data collection and the analysis of unstructured data (classes 12-14). The variety and scale of data have increased immensely during the past decade. This is primarily due to the growth in social media and the vast arrays of online texts and other forms of online data. In the last three classes of the term, we will therefore examine tools for scraping data, using application programming interfaces (APIs) (e.g. the Twitter API), and analyzing the resulting data. Depending on the time available, I will also provide an introduction to text analysis and machine learning.