| Keywords [eng] |
CNN, Fox News, Wordfish, Wordshoal, polarization, text mining, text analysis, sentiment analysis, news articles, television shows, US. |
| Abstract [eng] |
Political polarization among Americans has grown rapidly in the last decades. It is well-known that the rise of 24-hours partisan cable news is one of the explanations for the increase. This thesis aims to determine the main polarization factors between two US primary partisan media sources – CNN and Fox News – and evaluate their central persons’ influence. First, two core news sources are investigated – CNN and Fox News television shows and news articles presented on their news websites and analyzed by text properties – length, simplicity, richness, and negative sentiment level. Second, using the Poisson scaling methods Wordfish and Wordshoal, each anchors polarization is evaluated, which is their position variation on different subjects from the 2018 to 2021 period. Parallelly, by nine distinguished topics (for example, Senate, immigration, education and more), news websites articles are evaluated in the period 2016-2021. Third, the thesis contributes to polarization and text mining literature by introducing the novel data preprocessing approach and augments the polarization estimation methods incorporating anchors’ status and his television shows’ popularity indicators. The study suggests that CNN and Fox News are highly polarized in articles and television shows. On the one hand, there is a widening gap between the media, especially after Robert Mueller’s study of Russia’s influence in 2016. On the other hand, the polarization of television shows remains relatively stable. There is a tendency for CNN and Fox News anchors’ polarization positions to be broadly in line with their respective media platforms. In addition, in recent years, it has become clear that formerly more neutral is getting closer to the CNN and Fox News clusters. Finally, the results obtained show that negativity level is also relevant and should be taken into consideration in the analysis of media polarization. |
