Birds of the same feather tweet together: Bayesian ideal point estimation using Twitter data
Citation: Barberá Pablo (2015) Birds of the same feather tweet together: Bayesian ideal point estimation using Twitter data. Political Analysis (RSS)
DOI (original publisher): 10.1093/pan/mpu011
Semantic Scholar (metadata): 10.1093/pan/mpu011
Sci-Hub (fulltext): 10.1093/pan/mpu011
Internet Archive Scholar (search for fulltext): Birds of the same feather tweet together: Bayesian ideal point estimation using Twitter data
Tagged: Twitter (RSS), media polarization (RSS)
Rather than using surveys and political contributions to determine voter ideology, the author used twitter networks to create a better understanding of the exact ideals of particular voters. Because many citizens and virtually every legislator are on Twitter, their user habits create a more complete picture of political ideologies. However, the main drawback is that this is not a representative sample of the voting population. Barbera estimated the political ideology of users based on the political views of who they follow. The political views of the pundits and politicians being followed were determined by more traditional public ideology methods such as voting records in Congress. When comparing their results with a small subset of the group which has public registration and contribution records. This method was given strong evidence to suggest that it’s a mostly reliable model for determining ideology. It also found evidence that suggests the existence of echo chambers by using retweet interactions. They found that users were more likely to retweet a tweet about Obama or Romney if the user tweeting it had a similar ideology.
Theoretical and Practical Relevance
This could be used for our studies to determine the implicit bias of some journalists who might not outright state their political beliefs, especially local ones, who tend to be viewed as more objective. A similar method could be tested on the Twitter accounts of media companies as another way to determine bias, especially using the tweet interaction method described above. Lastly, the idea of echo chambers could be carried out further from this study.