Maybe Twitter Actually Is Real Life: An In-Depth Examination of the Connections Between Twitter and Politics
Shackelford, Andrew W.
MetadataShow full item record
CitationShackelford, Andrew W. 2020. Maybe Twitter Actually Is Real Life: An In-Depth Examination of the Connections Between Twitter and Politics. Bachelor's thesis, Harvard College.
AbstractDespite its third-place rank in total users, Twitter represents the most popular social media platform for politicians to communicate with their constituents. In addition, journalists, pundits, and politicians alike have formed a so-called “Political Twitter” bubble that has an outsized influence on policy, media, and politics itself. In this thesis, we examine the relationships between politics and Twitter. We analyzed how partisanship affected politicians’ willingness to tweet about politically toxic events such as impeachment, and found a strong correlation between the partisanship of a given politician and their willingness to tweet about impeachment. We also found that politicians in so-called “crossover” districts, where the representa- tive’s political party does not match the party that their district chose for the presidency, are much less likely to tweet about impeachment than counterparts in non-crossover districts with the same level of partisan lean. Having examined one side of the relation, we then at- tempted to use Twitter as a predictor for Democratic primary polling. After finding limited results for national primary polling, we focused on the more limited domain of Democratic debate performances. We found strong evidence that the number of times a candidate’s name and username are mentioned during a debate are powerful predictors for polls of that candidate’s debate performance. We conclude by ex- amining the implications of our findings, suggesting that Twitter can in fact be a valid predictor for certain more high-engagement domains, and that despite Twitter’s reputation as hyper-partisan, politicians perform a remarkable amount of self-mediation on the platform.
Citable link to this pagehttps://nrs.harvard.edu/URN-3:HUL.INSTREPOS:37364696
- FAS Theses and Dissertations