From the conventions to the eve of the final presidential debate, Barack Obama and Mitt Romney have both received more negative than positive coverage from the news media, though overall Obama has had an edge, according to a new study.

That advantage for Obama, however, disappeared after the debates began in early October and news coverage shifted in Romney’s direction, mirroring the momentum change reflected in many public opinion polls, the study by the Pew Research Center’s Project for Excellence in Journalism found.

Overall from August 27 through October 21, 19% of stories about Obama studied in a cross section of mainstream media were clearly favorable in tone while 30% were unfavorable and 51% mixed. This is a differential of 11 percentage points between unfavorable and favorable stories.

For Romney, 15% of the stories studied were favorable, 38% were unfavorable and 47% were mixed-a differential toward negative stories of 23 points.

Most of the advantage in coverage for Obama, however, came in September in the form of highly negative coverage for Romney. This was a period when the GOP nominee was losing ground in the polls, he was criticized for his comments about Libya, and a video surfaced in which he effectively dismissed 47% of the American public.

All that changed almost overnight after the first debate on October 3. From that day through October 21, the coverage in effect reversed. In all, 20% of stories about Romney were favorable, 30% were unfavorable, and 50% were mixed-a differential of 10 points to the negative. For Obama, 13% of stories were favorable, 36% were unfavorable, and 50% were mixed-a differential of 23 points.

Throughout the eight-week period studied, a good deal of the difference in treatment of the two contenders is related to who was perceived to be ahead in the race. When horse-race stories-those focused on strategy, tactics and the polls-are taken out of the analysis, and one looks at those framed around the candidates’ policy ideas, biographies and records, the distinctions in the tone of media coverage between the two nominees vanish.

With horse-race stories removed, 15% of campaign stories about Obama were positive, 32% were negative and 53% were mixed. For Romney it was 14% positive, 32% negative and 55% mixed.

The portrayal of the two candidates this year in the mainstream press stands in marked contrast to what the Project found in 2008 when then Senator Barack Obama was running against Senator John McCain. In that race, Obama’s coverage was almost twice as positive as it has been this year (36% vs. 19%) and more positive than negative overall (36% positive vs. 29% negative that year). McCain’s coverage four years ago, by contrast, was much more negative than Romney’s this year. In 2008, nearly six in 10 stories about McCain were clearly negative in nature (57%), while only 14% were positive.

One other distinction between 2008 and 2012 is how much the narrative has changed with events. There have been three distinct periods in the coverage of Obama and Romney over the eight weeks studied, one of which favored Obama, the second of which favored Romney and a third that was closer with an advantage for the president.

From the conventions until the first debate, a period of improving polls for Obama, Romney suffered his period of the most negative coverage; just 4% of stories about him were positive while 52% were negative. Coverage of Obama during this period was fairly evenly split (20% positive vs. 24% negative). That narrative reversed sharply with the first debate. For the next two weeks, Romney saw the mixed treatment (23% positive vs. 23% negative) while Obama was caught in the critical loop, with 12% positive and 37% negative. After the second debate, coverage returned to its more general pattern, with an edge for Obama.

This treatment in the mainstream media also differs markedly from what the study finds in the newer realms of social media: Twitter, Facebook and blogs. There, the narrative about both men has been relentlessly negative and relatively unmoved by campaign events that have shifted the mainstream narrative-more a barometer of social media user mood than a reflection of candidate action.  On Twitter, for instance, the conversation about the campaign has consistently been harsher for Romney than for Obama. On Facebook, the tone improved for Obama in October with the debates, despite the sense that the president had stumbled in the first one. And in the blogosphere, neither candidate has seen a sustained edge in the narrative in the eight weeks studied.

The study also reveals the degree to which the two cable channels that have built themselves around ideological programming, MSNBC and Fox, stand out from other mainstream media outlets. And MSNBC stands out the most. On that channel, 71% of the segments studied about Romney were negative in nature, compared with just 3% that were positive-a ratio of roughly 23-to-1. On Fox, 46% of the segments about Obama were negative, compared with 6% that were positive-a ratio of about 8-to-1 negative. These made them unusual among channels or outlets that identified themselves as news organizations.

The study also found a difference between the three network evening newscasts and the morning shows. Obama also fared better in the evening, Romney in the morning.

An analysis of the coverage of the vice presidential candidates, meanwhile, found that Paul Ryan received roughly a third of the amount of coverage that Sarah Palin did in 2008. But of the two vice presidential candidates this year, Ryan and Joe Biden, Ryan received much more unfavorable coverage-28% unfavorable vs. 16% for Biden.

These are among the findings of the content analysis of 2,457 stories from 49 outlets from August 27, the week of the Republican convention, through October 21, five days after the second presidential debate. For mainstream media, the study included the three broadcast networks, the three major cable news networks, the 12 most popular news websites, 11 newspaper front pages and news programming from PBS and NPR along with radio headlines from ABC and CBS news services. From these outlets, PEJ researchers watched, listened or read every story in the sample and counted each assertion for whether it was positive in nature about a candidate, negative in nature or neutral. For a story to be deemed to have a distinct tone, positive or negative assertions had to outnumber the other by a factor of three to two. Any story in which that was not case was coded as mixed.

For social media, the researchers combined a mix of traditional human coding with technology from the firm Crimson Hexagon. Researchers trained the computer “monitors” to replicate their human coding according to PEJ rules. For Twitter, the sample includes the full “fire hose” of public tweets. For Facebook, the study includes a large sampling of public posts about the campaign. The study included a sample of several million blogs as well.

The study of the tone in news coverage is not an examination of media bias. Rather, it measures the overall impression the public is receiving in media about each candidate, whether the assertion is a quote from a source, a fact presented in the narrative that is determined to be favorable or unfavorable, including poll results, or is part of a journalistic analysis.

Among the findings of the study:

  • Horse race coverage is down from 2008. Overall, 38% of the coverage coded during these two months was framed around what is typically called horse-race coverage, stories substantially concerned with the strategy and tactics of the campaign and the question of who was winning. That is down from four years ago, when 53% of the coverage studied during a similar period was focused on the horse race. Coverage of the candidate policy positions comprised the second-largest category of coverage, 22%, similar to 2008. Coverage of voter fraud laws and other political topics that largely did not involve the candidates was tied for the third-largest category at 9%, and was a subject that was almost nonexistent in the narrative four years earlier.
  • Debate coverage was more about who won than what candidates said. During the three-week debate period studied, October 4-21, horserace coverage grew, filling nearly half of all the stories about the campaign (47%), the largest of any period in the study. In other words, rather than a window to examine the candidates’ ideas at more length, the debates became a frame about campaign momentum to a greater degree than the rest of the campaign. Coverage of foreign policy during this time, by contrast, fell by roughly half to 7%, as did coverage of the personal topics about the candidates, which fell to 1%.
  • The two candidates received similar amounts of coverage. Overall Obama was a significant figure in 69% of the stories studied during the eight weeks, while Romney was a significant figure in 61%.The difference is explained almost entirely by coverage of the Obama presidency. Roughly 9% of all stories studied involved Obama functioning as president outside the realm of the campaign. Four years ago, during a similar period in the campaign, the volume of coverage was evenly balanced between Obama and his Republican rival John McCain, who were each significant newsmakers in 62% of the campaign stories.
  • Among the issues, the economy dominated but less so than in 2008. The economy accounted for 10% of all campaign coverage studied, down from 15% four years earlier. It still overshadowed all other policy issues. Turmoil in the Middle East, particularly the attack on the U.S. consulate in Libya, was next-at 5%. Some subjects were notable for less attention. Health care accounted for 1% of campaign coverage studied, a sign perhaps that while Romney was opposed to Obamacare, it was not an issue he pressed as much as others because of comparisons to his experience in Massachusetts. Social issues were also notable for their absence. Together, abortion and gay rights, for instance, accounted for less than 1% of the coverage. So did the war in Afghanistan and the situation in Iraq.
  • Of all the platforms studied, the tone of conversation was the most negative on Twitter. Every week on Twitter resembled the worst week for each candidate in the mainstream press. Negative Twitter conversation about Romney exceeded positive by 42 points in the eight weeks studied.[1] For Obama, negative assertions outnumbered positive by 20 points, though it demonstrated somewhat more fluidity from week to week.
  • Network news viewers received a different narrative about the candidates depending on when they watched. Romney fared better than Obama on the network morning shows on ABC, CBS and NBC. During the 7 a.m. half hour, negative segments outnumbered positive ones by 9 points for Romney vs. 17 for Obama. In the evening, Obama fared better. His narrative was fairly evenly mixed, with positive segments outnumbering negative ones by 2 points. For Romney, negative exceeded positive by 17 points.



1. [1] For social media, tone is measured as a percentage of the total conversation. A tweet or blog post might have multiple assertions and each is tallied as percentage of the whole. For mainstream media, the Project tallies tone by stories.