Study debunks belief in media bias

Douglas Rojas

Throughout the 1996 presidential campaign, Republican candidate Bob Dole claimed that campaign media coverage had a strong liberal and pro-President Clinton bias. A University research study, however, found this not to be the case.
Two University faculty members, one from the biology department and one from the political science department, a visiting faculty member in the journalism school and two journalism graduate students conducted the study. The project looked at how positive and negative news coverage of each candidate during the 1996 presidential campaign influenced public-opinion polls.
The study, presented Friday at the Midwest Association of Public Opinion Research in Chicago, Ill., found that there was a balance between positive and negative information in the media for both candidates.
“We address questions like What’s the media bias against Dole? Was Dole cheated in some way?'” said Steven Smith, a political science professor. “But in fact, we do find a curiosity. That is, that the positive coverage of Dole didn’t seem to give him as large a boost in the polls as the positive coverage of Clinton, or the negative coverage of either of them.”
The study also found that it is possible to predict with a high level of accuracy how changes in the media coverage during the 1996 presidential campaign affected the polls.
“That’s what we are trying to estimate,” said Smith, “how bad does the media have to get to see a certain amount of change in the polls.”
More than 11,000 stories from major newspapers and television newscasts — such as The New York Times, ABC and CNN — from March 10 to Nov. 3 were processed for the study through a media content analysis program, InfoTrend. The computer program was designed and patented by David Fan, a cell and molecular biology professor who also participated in the research.
“The goal is not to predict the election result, but to predict polls’ results and show that, based on media content analysis, you can predict what the public opinion would be,” said Michael Fibison, a journalism doctoral student involved in the research.
The computer program follows a specific set of criteria; it looks for information, reads the information and determines which paragraphs are positive and which are negative. Then it can predict public opinion polls before the results are revealed with a small margin of error.
The study also found that factors such as peace and economic prosperity gave Clinton a comfortable lead. During the campaign polls showed Clinton averaging a 10 point lead over Dole. The computer analysis estimated that Clinton should have had a 12 point lead in the polls. The two point difference between the actual polls and the computer model was within the margin of error.
The effect of visual and audio factors that can’t be captured in the analyzed text causes this difference between actual and predicted poll results, Fibison said. “That’s where the impact of visual media plays a role,” he added.
“It was Dole’s visual and audio impression, his image and his voice that put him at a disadvantage against Clinton,” Smith said.
Another possibility is that issues emphasized by Dole, though widely covered by the media, didn’t resonate with the public, said Smith.
“He just made tactical errors; for example, people found his 15 percent tax cut not very credible,” Smith said.