Modeling Political Activism around Gun Debate via Social Media

TL;DR

This paper demonstrates that social media signals, including network and linguistic features, can effectively predict individual and state-level political activism regarding gun control, complementing traditional demographic data.

Contribution

It introduces a comprehensive model combining social media, network, and psycho-linguistic features to predict political activism on gun regulation.

Findings

Network information improves stance classification accuracy.

Social media features predict real-life political actions.

State-level debate dynamics relate to social media signals.

Abstract

The United States have some of the highest rates of gun violence among developed countries. Yet, there is a disagreement about the extent to which firearms should be regulated. In this study, we employ social media signals to examine the predictors of offline political activism, at both population and individual level. We show that it is possible to classify the stance of users on the gun issue, especially accurately when network information is available. Alongside socioeconomic variables, network information such as the relative size of the two sides of the debate is also predictive of state-level gun policy. On individual level, we build a statistical model using network, content, and psycho-linguistic features that predicts real-life political action, and explore the most predictive linguistic features. Thus, we argue that, alongside demographics and socioeconomic indicators, social media provides useful signals in the holistic modeling of political engagement around the gun debate.