Computational methods applied to mass communication research: the case of press release content in news media

Simple item page
dc.contributor.advisorHanson, Christopheren_US
dc.contributor.authorGolitsynskiy, Sergeyen_US
dc.contributor.departmentJournalismen_US
dc.contributor.publisherDigital Repository at the University of Marylanden_US
dc.contributor.publisherUniversity of Maryland (College Park, Md.)en_US
dc.date.accessioned2013-10-10T05:33:32Z
dc.date.available2013-10-10T05:33:32Z
dc.date.issued2013en_US
dc.description.abstractIn this dissertation, I apply a variety of computational methods to explore new approaches to investigate the problem of news media's use of press release content. Being used by the public relations industry in an effort to influence the media agenda, press releases often promote the organization's viewpoint on issues. Journalism scholars have expressed numerous concerns over news media using such content as a source, often without attribution. A review of previous research has revealed a number of shortcomings, with the main problem being the lack of a reliable methodology to establish a connection between a press release and an article, which is essential for such research. This deficiency is explained by the need for in-depth textual analysis on the one hand, and the requirement of large representative samples on the other - which is near impossible to achieve using traditional methodological approaches used in journalism research. I propose using computational methods to address this problem. I use computation to extract large amounts of text from web sites, transform loosely structured text into well-formatted data, and reduce a data set consisting of 6,171 press releases and 48,664 related news articles to a sample of 1,643 press release/news article pairs, showing reasonable evidence that each of the press releases has been used as a source by a corresponding news article. Such evidence is established through verbatim text matches of sufficient length. I use the constructed data sample to investigate the extent to which press release content is used by news media verbatim, how such content is used and whether proper attribution is made identifying the true source of the news. Although my findings suggest that the problem of press release content might be not as severe as presented in previous research, due to the limitations of verbatim text matching, it might be also possible that such practice remains undetected, with all content borrowed from press releases appearing in news media in paraphrased form. Finally, my investigation leads to a discovery of a "smoking gun" - a striking example of PR influence in the form of a corporation "manufacturing" statements, getting elected officials to repeat them, and the media reporting them as a regular news story.en_US
dc.identifier.urihttp://hdl.handle.net/1903/14638
dc.subject.pqcontrolledMass communicationen_US
dc.subject.pqcontrolledJournalismen_US
dc.subject.pqcontrolledComputer scienceen_US
dc.subject.pquncontrolledcomputational text analysisen_US
dc.subject.pquncontrollednews mediaen_US
dc.subject.pquncontrolledpress releaseen_US
dc.subject.pquncontrolledpublic relationsen_US
dc.titleComputational methods applied to mass communication research: the case of press release content in news mediaen_US
dc.typeDissertationen_US

Files

Original bundle
Now showing 1 - 1 of 1
Thumbnail Image
Name:
Golitsynskiy_umd_0117E_14480.pdf
Size:
2.68 MB
Format:
Adobe Portable Document Format
Download

Collections

UMD Theses and Dissertations
Journalism Theses and Dissertations