RAPID: Measuring the Intent of Leaders through Censorship Behavior

Harvard University, Cambridge MA

Investigators

Abstract

General Understanding the intent of foreign governments and their leaders is essential to a vast array of scholarly research and public policy analysis. Yet, measuring intent is usually only possible after the fact, by inferring it from government action or interviewing public officials after they leave office. The PIs develop and extend a computer-assisted text analysis method to predict government actions before they occur, thus offering a more direct measure of a leader's intent in a way that can be rigorously validated. The main input to these predictions will be measurable government action on the Internet. To accomplish this, the PIs leverage the infrastructure they have built from prior research and extend it for this purpose. Technical Understanding the intent of foreign governments and their leaders is essential to a vast array of scholarly research and public policy analysis. Yet, measuring intent is usually only possible after the fact, by inferring it from government action or interviewing public officials after they leave office. The PIs develop and extend methods to predict government actions before they occur, thus offering a more direct measure of a leader's intent in a way that can be rigorously validated. Methodologically, the PIs design computer-assisted methods of text analysis adapted to other languages to analyze the content of censored and uncensored social media posts. The research and resulting data may also reveal specific strategies employed by authoritarian regimes to consolidate power after major leadership transitions.