Government Research Intern: Karl Bock

Title: Print Media Tone and Chance of Winning in the 2012 Presidential Election

Author: Karl Bock


The media and its tone are always perceived as playing an important role in politics, with pundits on the right constantly decrying the liberal biases of the “Main Stream Media” and their counterparts on the left characterizing Fox News and it’s brethren as the mouthpieces of billionaires and corporations. And in what is referred to as “horse race politics” by The Washington Post, the media is constantly concerned with the small fractions of a percent that polling data changes from day to day, and tries to connect these changes to the overall narrative of the race. But is the positive or negative atmosphere in the media towards the respective candidates shaped by the polling data or does the media drive polling? And when Election Day finally arrives, do either of these make a difference in who the American people choose to be inaugurated in January? I became interested in this topic in the fall after Mitt Romney received a large amount of negative press about his remarks on the Benghazi attack. Throughout the campaign, Romney seemed to have a knack for putting his foot in his mouth, from insulting the British when talking about the Olympics, to calling 47% of Americans “takers”. I was surprised that a candidate who had so many public gaffes was still being perceived as doing relatively well in the race. The purpose of my research then was to explore the relationship between print media tone and a candidate’s chance of winning in the 2012 presidential election. My working hypothesis was that there would be some kind of relationship between media tone and chance of winning, with one likely preceding the other.

I elected to use print media as opposed to broadcast or web-based media because published news articles would be easier to work with from a coding perspective. I coded articles published in The New York Times, The Washington Post, and The Wall Street Journal from September 7, 2012, the day after the conclusion of the democratic national convention, until November 5, 2012, the day before Election Day.  I selected those three papers because they have strong journalistic reputations, have wide circulation, and the library had access to databases of their articles. Moreover, they are perceived as being pretty well distributed across the ideological spectrum, with The New York Times being more left leaning, and The Wall Street Journal having more of a conservative bent.  I only coded articles that were not editorials, were in the main section of the print edition, and that were substantially about Obama or Romney. I coded articles as either “Positive Incumbent” (Pos) or “Negative Incumbent” (Neg) based on how the Obama campaign would perceive the article. So, if an article was positive or neutral in tone and focused on Obama or on both candidates, it was coded Pos. If an article was negative and focused on Romney it would be coded as Pos as well. Articles that were negative in tone and focused on Obama or both candidates were coded Neg.  And lastly, if an article was positive or neutral and focused on Romney, it would be coded Neg.  I then graphed the percentage of positive articles with a seven-day moving average and compared that to the chance of winning measure from the 538 blog, a poll aggregator run by Nate Silver and published by The New York Times. The resulting correlation between the two measures was found to be .094, so it may be that the measures do not interact in a manner as straightforward as I hypothesized.