Context and Sources in Broadcast Television Coverage of the 2004 Democratic Primary

Abstract

This study examined context variables (reporter speculation, multiple viewpoints, and story emphasis) and source variables (anonymous sources and source transparency) in broadcast television coverage of the 2004 Democratic presidential primaries. Primary coverage was compared with coverage of other major stories. Primary coverage was no more focused on conflict than were other major stories. Primary coverage was, however, more focused on winners and losers, and primary reporting was more likely to include reporter speculation. Primary stories were just as likely to include anonymous sources as were other major stories, but primary stories also were more likely than other major stories to fully identify sources.

Acknowledgement

The authors thank Tom Rosenstiel and Amy Mitchell of the Project for Excellence in Journalism for use of the PEJ data.

Notes

¹The following dates made up the 28-day sample: January 13, 16, 23; February 2, 13, 23, 29; March 8, 12, 13, 14, 19, 24; April 8, 15; May 1, 4, 20; June 8, 9, 16; July 19, 25; August 10, 12; and September 4, 22, 26.

²The PEJ study used the term frames for story qualities that this study characterizes as “emphases.” The PEJ definition of story “frames” does not correspond exactly to conceptual and operational definitions used in much framing literature. Our research avoids possible confusion by changing the term to one that better describes the story qualities relevant for this study.

³The options included conflict, consensus, winners and losers, problem needs solving, good yarn, how audience members do something, reality check, and underlying principles in play. Some stories were coded no emphasis or multiple emphases.

⁴A Scott's pi of .8 or higher is acceptable and values as low as .667 can be acceptable for tentative solutions. See Riffe, Lacy, and Fico (1998, p. 151). The reporter speculation variable has a larger difference between the percentage of agreement and Scott's pi than for the other variables. This reflected the large number of stories in the reliability sample that received zeros (absence of speculation) for a dichotomous variable. Scott pi scores tend to be much lower than simple agreement when the cases of a dichotomous variable are dominated by one of the choices. However, Scott's pi does not measure what was determined by chance but rather what “could” have been decided by chance. Unlike a flip of a coin, which is totally independent, classification of content is based on a protocol that affects probability.

Note. χ²(3) = 160.22, p < .001.

Note. χ²(1) = 6.99, p = .01.

Note. χ²(1) = 43.49, p < .001.

Note. χ²(3) = 47.59, p < .001.

Note. χ²(1) = .595, p = .440.

Additional information

Notes on contributors

Geri Alumit Zeldes

Geri Alumit Zeldes is an assistant professor in the School of Journalism at Michigan State University. Her research includes studying how the news media covers issues involving race and gender during presidential elections.

Frederick Fico

Frederick Fico is a professor in the School of Journalism at Michigan State University. His research interests include content analysis of media coverage of conflict and elections.

Stephen Lacy

Stephen Lacy is a professor in the School of Journalism and Department of Communications of Michigan State University. His research interests include media economics, content analysis methods, and newspaper reporting.