Text as Data

This course is an introduction to the quantitative analysis of text as data – a rapidly growing field within the social sciences. The availability of textual data has grown massively in recent years, and so has the demand for skills to analyze it. Vast amounts of digital content are becoming increasingly relevant to various policy-relevant questions. For example, social media data are now commonly used to understand public opinion, engagement with politics, behavior during natural disasters, and even pathways to extremism; candidates’ statements and rhetoric during elections are useful for estimating policy positions; and large amounts of text from news sources are used to document and understand world events.

While the wealth of information in text data is incredible, its sheer size makes it challenging to summarize and interpret without quantitative methods. In this course, we will learn how to quantitatively analyze text from a social-science perspective. Throughout the course, students will learn different methods to acquire text, how to transform it to data, and how to analyze it to shed light on important research questions. Each week we will cover different methods, including dictionary construction and application, sentiment analysis, scaling and topic models, and machine learning classification of text. Lectures will be accompanied by hands-on exercises that will give students practical experience while working with real-world texts. By the end of the course, students will develop and write their own research projects using text as data.