Combining natural language processing and network analysis to examine how advocacy organizations stimulate conversation on social media

Significance Social media provide potent opportunities for advocacy organizations to shape public debates because of the rapidly increasing number of people who frequent such forums each day. However, social scientists have not yet explained why some advocacy organizations create large-scale public debate whereas most others do not. Using automated text analysis and a Facebook application, I found that advocacy organizations are more likely to stimulate conversation if they produce messages that link discursive themes that are usually discussed in isolation from each other. Such messages not only resonate with multiple audiences, but also put such audiences in conversation with each other. This manuscript thereby contributes a theory of public deliberation on social media for the emerging field of computational social science.