Katherine Ognyanova, U.S. News Outlets - Network of Shared Topics

Doctoral Candidate at University of Southern California, Annenberg



News Ties: A Study of Media-Level Agenda-Setting




The research presented here proposes an exploration of inter-media agenda-setting. One of the big questions in the field is whether proliferation of channels and information overload have fragmented news content, rendering the idea of a consistent mainstream media agenda obsolete. The study employs network analysis methods to test this proposition. News outlets are seen as a set of interconnected nodes with ties based on overlap of issue coverage. Two news sources are linked if there is an above average similarity in their selection of topics over a period of time. The research demonstrates that mainstream media outlets not only have overlapping agendas at a given point in time, but also exhibit patterns of shared coverage that persist over time. A QAP analysis comparing networks of shared topics compiled for January and December of 2007 shows a strong and significant correlation (r=0.87, p<0.001) between the two. Furthermore, both networks are dense, well-connected and have a relatively high centralization index - something that goes against the idea of fragmentation and disintegration of media agenda.


The study goes on to explore the set of properties that are likely to influence the overlap of media agendas. Factors taken into consideration include format (newspaper, radio, network TV, cable TV, Internet), common ownership, and volume of the outlet production. An ERG model including parameters for those factors as well as basic structural signatures is fitted to the December 2007 news outlet network. Results of the estimation indicate that both production volume and common ownership affect the topic overlap of news outlets. In order to investigate the patterns of shared coverage within and across media sectors, a blockmodel analysis is performed (results presented on the figure above). Network TV, Internet and radio outlets are likely to produce coverage similar to that of other media of the same type. This does not hold true for newspapers and cable TV. The Network TV outlets have dense ties with all of the other four sectors a finding consistent with previous research on the influence of television news.



This submission shows the topic similarity between news outlets, where a tie exists if two organizations cover a similar array of topics. The pairing of the block-model results and the detail graph is very helpful, as the detailed layout is so dense as to make identifying positions challenging. I wonder if the image would be improved by forcing the nodes into a positioning more similar to the block layout or somehow incorporating a weighting that differentiates within and between-block ties?



The visualization uses a block-modeling approach to show that despite a proliferation of media outlets, there is much convergence on news topics. While the block model image portrays a general sense of the interconnectedness of the media system, the network image itself is somewhat swamped by density of ties, with little information aside from the size and centrality of the media outlet evident from the detailed image. The rendering itself is nicely done, with a color scheme that is easy to view and differentiate.



This visualization has excellent theoretical framing, providing an interesting question, and an easily understood visual answer. The blockmodel and adjacency plots compliment each other well, and give us both a sense of connectedness and structure. The use of loops in the blockmodel is especially effective at giving us a sense of internal connectedness compared to connectedness between. I'd be interested to see a hybrid of the blockmodel and adjacency views.