| Summary/Abstract |
This paper establishes a new method of estimating public dissent that is both cost-effective and adaptable. Twitter allows users to post short messages that can be viewed and shared by other users, creating a network of freely and easily observable information. Drawing data directly from Twitter, we collect tweets containing specified words and phrases from citizens voicing dissatisfaction with their government. The collected tweets are processed using a regular expression based algorithm to estimate individual dissent; which is aggregated to an overall measure of public dissent. A comparative case study of Canada and Kenya during the summer of 2016 provides proof of concept. Controlling for user base differences, we find there is more public dissent in Kenya than Canada. This obvious, but necessary, result suggests that our measure of public dissent is a better representation of each country’s internal dynamics than other more sporadic measures. As a robustness check, we test our estimates against real-world civil unrest events. Results show our estimates of public dissent are significantly predictive of civil unrest events days before they occur in both countries.
|