MEDIA BEHAVIOR DURING 2008 US PRESIDENTIAL ELECTION: A WEB CONTENT ANALYSIS

This paper presents an empirical analysis of web news content during the 2008 USA Presidential Campaign. The analysis is based on a representative sample of online news sources, monitored on a daily basis, during and immediately after the campaign. Data was gathered from 38 websites of the highest-rated USA news sources across print and broadcast media, and several popular electronic media news sources. The sample mostly drew from the online affiliates and content sources of the highest-circulated newspapers and magazines in the USA and the top-rated USA-based network and cable television stations. The web-based content of these media sources were assumed to mirror the content and tone of their corresponding traditional print and broadcast media. Keywords were identified from Obama's and McCain's nomination acceptance speeches and the three presidential debates. Keywords were the most frequent content-bearing words found in the transcripts of these events. One goal of the analysis was to understand how different media sources presented each of the candidates and their key issues. We also analyzed and visualized the network that was created by keywords within news sources' web pages. Network analysis tools were used to explore media behavior and trends during the campaign. The network's nodes are keywords and candidates, while the frequency of co-occurrence of nodes is graphically represented by relative proximity. A set of centrality measures (both Freeman's and Bonacich's approach) was applied to the network of keywords. Further research involves forecasting modelling, trend-setter analysis, applying a weighted taxonomy to keywords, linking keywords, and calculating distances between keywords within web pages.