Could a shift in newspaper endorsements have changed the outcome of the 2000 or 2004 election? What could it do this year?
A study released this month by Brian Knight, associate professor of economics, and Chun-Fang Chiang AM'04 PhD'08, a former graduate student in the economics department, may be able to shed light on these questions with their study on how endorsements from different newspapers can affect voter decisions.
Their analysis indicates that endorsements for a Democratic candidate coming from a more right-leaning or neutral newspaper will convince more voters than the same endorsement coming from a left-leaning newspaper, according to Knight. The reverse is also true: Endorsements from left-leaning newspapers are more likely to convince voters if they endorse a Republican candidate, he said. These types of crossover endorsements are also referred to as "surprising" endorsements and are considered more credible than a newspaper endorsing a candidate with the same ideological leanings as the paper, according to the survey. The study found that "unsurprising" endorsements lead to a change of less than 1 percent of voter preference whereas surprising endorsements saw up to a 3 percent change in voter response.
"These results suggest that voters are sophisticated and attempt to filter out any bias in media coverage of politics," the researchers wrote in the conclusion of their study.
For the study, Knight and Chiang analyzed data collected by the National Annenberg Election Surveys from 2000 and 2004, which asked people which newspaper they read and whom they planned to vote for. One set of these surveys asked people those questions before a newspaper made an endorsement which was compared to another survey, which asked a separate sample of readers for the same information after a given newspaper had endorsed a candidate. Knight and Chiang studied whether biased media outlets are able help candidates. For example, Knight said they wanted to see if "Fox News can help the Republican."
"What was novel and interesting - and maybe surprising - was that voters think about if this is a surprising endorsement," Knight said. "Voters actively filter out media bias."
Assistant Professor of Political Science Jennifer Lawless and Associate Professor of Political Science Wendy Schiller said more factors about the papers and their readers might need to be taken into account.
"Keep in mind that it is part of a much broader set of information," Lawless said. Factors such as an individual's trust in a newspaper are important to consider, as is how often an individual reads the paper, she said. If people get their news from "15 different sources" on a regular basis, then an endorsement from one of them is going to matter less, she said.
Schiller said it is important to know about a reader's political knowledge before predicting how much of an effect the endorsement will have.
"If you have strong opinions, it might not matter as much," she said.
So far this year Sen. Barack Obama has about twice as many newspaper endorsements as Sen. John McCain, Knight said, and "in the past it has been 50-50" between the Republican and Democratic candidates. Knight said the endorsements "will matter" in this upcoming election and "Obama is getting more endorsements from right-leaning newspapers ... that have never endorsed a Democrat before," such as the Chicago Tribune.
"Our analysis indicates that given how close the elections have been in the past two cycles, the Democrat may have been able to win if he had secured more endorsements," Knight said.