Media Partisanship During Election: Indonesian Cases

摘要

Analysis of media partisanship during election requires an objective measurement of political bias that frames the content of information conveyed to the audience. In this study, we propose a method for political stance detection of online news outlets based on the behavior of their audience in social media. The method consists of 3 processing stages, namely hashtag-based user labeling, network-based user labeling, and media classification. Evaluation results show that the proposed method is very effective in detecting the political affiliation of Twitter users as well as predicting the political stance of news media. Overall, the stance of media in the spectrum of political valence confirms the general allegations of media partisanship during the 2019 Indonesian election. Further elaboration regarding news consumption behavior shows that low-credibility news outlets tend to have extreme political positions, while partisan readers tend not to question the credibility of the news sources they share.