The Washington PostDemocracy Dies in Darkness

A deep dive into the news media’s role in the rise of Donald J. Trump

Donald Trump with Fox News anchor Megyn Kelly, who interviewed the presumed Republican presidential nominee for her Fox network special “Megyn Kelly Presents.” (Eric Liebowitz/Fox)
Placeholder while article actions load

The media’s coverage of Donald Trump has become one of the most debated topics of the 2016 campaign. The debate centers on how, and how much, media coverage helped Trump as he rose the top of the polls and became the presumptive Republican nominee.

This debate quickly (and typically) became polarized, with the usual starkly worded headlines: “Why is Trump surging? Blame the media” versus “Stop blaming the media for  Trump.” Unhelpfully, the first headline is from one of us (Sides). So we’re long past due for some nuance and fresh evidence.

Here’s the short version: Yes, media coverage helped Donald Trump clinch the Republican nomination. In the absence of media coverage, it is difficult to imagine that Trump could have done so. And because Trump often dominated news coverage, it was difficult for other Republican candidates to “break out.” Of course, the other candidates didn’t help themselves, either.

But the impact of Trump’s news coverage was contingent: It boosted Trump primarily among those who were predisposed find his message appealing. Among those who found him less appealing, this coverage would have the opposite impact — and that has proved to be a problem for Trump when it was time to pivot to the general election.

Here are key questions and answers about the news media’s role in Trump’s success.

Why would media coverage even matter in primary elections?

Citizens do not form political judgments in a vacuum. A seeming truism such as “Voters, not the press, decide elections” is more wrong than right. Citizens need information to make decisions, and in politics, a lot of that information comes from the news media.

Presidential primaries are a place where information and, therefore, news coverage is likely to matter more. There are often many candidates — 17 at one point in the 2016 Republican primary  — and voters begin the primary campaign with little knowledge of most of these candidates. Moreover, since all the candidates are in one party, voters cannot rely on the most useful political heuristic — their own partisan leanings — to make a decision.

It’s hardly a new idea that media coverage would be useful in this circumstance. In some ways, the idea dates to a 1948 paper by the eminent Paul Lazarsfeld and Robert Merton. They describe how the media can “confer status” on individuals:

The mass media bestow prestige and enhance the authority of individuals and groups by legitimizing their status. Recognition by the press or radio or magazines or newsreels testifies that one has arrived, that one is important enough to have been singled out from the large anonymous masses, that one’s behavior and opinions are significant enough to require public notice.

A 17-candidate field is pretty close to a “large anonymous mass,” which makes being “singled out” all the more important.

Unsurprisingly, previous research on primaries has shown that preferences change as information changes. See, for example, Larry Bartels’s seminal book on presidential primaries. In their book “The Gamble,” John Sides and Lynn Vavreck also found that news coverage catalyzed a surge for several Republican candidates in 2012, including Rick Perry and Herman Cain.

Did media coverage help Trump gain notoriety even before he ran for office?

To understand Trump, however, we have to start before the primary. Some journalists who weren’t persuaded that media coverage helped Trump tried to rebut the argument by saying that “The media didn’t create Trump.” The Washington Post’s Eugene Robinson wrote, “After decades in the public eye, he had more than 90 percent name recognition when he began his campaign.”

Actually, Trump’s notoriety before the primary did reflect media attention. After all, how does one get to spend “decades in the public eye” if not because of media coverage: first, from the New York City tabloids that covered Trump’s real estate deals and romantic exploits; and later, from the television network that gave him his own reality show? Nobody ends up in the public eye without a substantial assist from mass media.

This point is readily acknowledged by at least one of these New York tabloid writers. Susan Mulcahy, formerly of the New York Post and New York Newsday, began a long mea culpa in Politico with this: “I have a confession to make, and please don’t shoot when you hear it: I helped make the myth of Donald Trump.” She continued: “We didn’t see it at the time, but item by inky item we were turning him into a New York icon.” This is exactly the point: Celebrities, including Trump, are created in part by their own efforts and in part by news outlets’ willingness to write about them.

In essence, there is a symbiotic relationship, or what the New York Times’ Jim Rutenberg called a “mutual dependence”: Someone like Trump will say and do things that generate media interest, which generates audience and revenue for media outlets, which in turn leads someone like Trump to say and do more things to maintain that coverage. Trump is certainly not unique in this regard (see: the Kardashians et al.).

In the primary, how much coverage did Trump receive or earn?

Two sources of data shed light on media coverage during the Republican primaries. One source is Crimson Hexagon, a firm that collects data from multiple online media outlets. Based on their data, we can report the percentage of stories that they monitored during the primary that mentioned Trump.

It is important to recognize that online media can different substantially from print and broadcast coverage and that accurately monitoring media is challenging. Even two different online media monitors may yield different results depending on their respective collection strategies. Thus, we do not claim that these results reflect the entirety of the media but, rather, offer an important window into a portion of the online media space.

The second source is from GDELT and the Internet Archive. Their data focus on a set of cable news outlets and allow us to tabulate the percentage of mentions of the Republican presidential candidates that were mentions of Trump. Focusing on mentions, rather than just stories, gives us a more granular sense of how much coverage Trump received. More importantly, television has fixed airtime — in contrast to the infinite space of online media — meaning more coverage of Trump must displace other coverage. Television coverage can therefore offer a clearer signal of media attention. These data are based on the raw closed-captioning stream, which captures all of the news programming on each channel.

Of course, just because news is published doesn’t mean anyone is consuming it. For example, 62 percent of American adults report getting at least some of their news from social media, which we have not measured here. Those platforms clearly affect which media coverage is seen.

With all that in mind, here is a graph with both sets of data, plotted from May 2015 through April 2016, just before Ted Cruz and John Kasich withdrew.

It would continue this way for the duration of the campaign, except for a few brief periods. One was in late October and early November, when Ben Carson began to receive comparable amounts of coverage. Another came after Trump placed third in the Iowa caucus.

But in general, Trump received more coverage than any other candidate, and sometimes — at least on cable news outlets —more than all the other candidates combined.

Other analyses reach a similar conclusion. For example, Trump received 78 percent of all coverage on CNN between Aug. 24 and Sept. 4, 2015. He also dominated evening network news coverage in the first half of 2015, despite announcing his candidacy late in this period. By November 2015, Trump had received more evening network news coverage — 234 minutes — than the entire Democratic field. By contrast, Ted Cruz had received seven minutes.

Trump also received the majority of newspaper coverage. One analysis of American newspaper coverage between July and December 2015 found that Trump had received 54 percent of the coverage — defined as articles that mentioned a candidate’s name in the first paragraph, but no other candidate.

In sum, Trump dominated coverage across different media types.

Why did Trump receive this much coverage?

Campaign news coverage is the product of four facets: the decisions of the candidates, the values of news organizations, the economic incentives of news organizations, and the decisions of readers and viewers.

Start with Trump himself. He is, of course, famously concerned about how much and how he is covered in the news. And so he makes decisions that give news organizations an incentive to cover him. As Roger Ailes noted, Trump is more available than the other candidates, for one.

For another, Trump knows how to align his behavior with the values of news organizations. News organizations have implicit values that shape what they consider newsworthy. Trump specializes in several of these values: novelty (he’s different from many other candidates), controversy (he makes statements that many condemn as wrong or outrageous), conflict (he constantly instigates feuds), and strong personalities (he has one).

You can see this reflected in the graph above. Coverage of Trump tended to spike after certain controversial remarks. An initial example was his attack on Republican Sen. John McCain’s record as a prisoner of war. In November, Trump regained the spotlight after his call for a database of Muslims living in this country. His coverage would spike again after his call for a ban on Muslims traveling to the United States. Another spike came after Trump canceled a Chicago rally on March 11 because of the presence of protesters.

This brings us to economic incentives. News organizations must compete in the market, obviously, and they need readers and viewers to generate income.

And Trump has been good for the news business. Cable news ratings set records. New executives have made no secret of their pleasure. As New York Times media columnist Jim Rutenberg reported:

Understandably, Jeff Zucker, the president of CNN Worldwide, was beaming when I saw him at a lunch with other reporters last week. “These numbers are crazy — crazy,” he said, referring to the ratings.

Meanwhile, CBS President and CEO Leslie Moonves said: “It may not be good for America, but it’s damn good for CBS.”

News organizations today also have increasingly refined data on what gets ratings and clicks, as Dave Karpf notes. It is easier to know what gets clicks when Chartbeat will track it for you in real time.

But who are news organizations tracking with Chartbeat, of course? Readers. They are the ones who are clicking, watching, and reading coverage of Trump. It’s not as if most people turn on CNN hoping for a good in-depth discussion of entitlement reform, and then are chagrined to find that they have to watch a story about Trump’s latest Twitter spat.

Often missed in many critiques of horserace journalism is this fact: Readers like it. One study of the 2000 election showed that, when given the choice, citizens gravitated more to news stories about campaign strategy and the biographies of the candidates, and less to stories about “the issues.”

But the public isn’t the only factor. You simply can’t say, as Politico’s Jack Shafer did, “If you want to blame anybody, blame the public.” You have to put all of these factors together.

Once you do that, you can also see why these patterns don’t require some media “cabal” (again, Shafer) — or, if you prefer, a “behind-the-scenes cabal,” which is clearly the worst kind of cabal. News outlets don’t produce similar patterns of coverage through explicit coordination. They do so because they all operate by many of the same values, they all face similar economic incentives, and they are all seeking to appeal to many of the same readers and viewers.

The implications of Trump’s news coverage for the other Republican candidates were grim. In 2012, Vavreck and I documented how surges for many Republican candidates were short-lived because another candidate would do something deemed newsworthy and then experience his own surge in news coverage. But it was difficult for any candidate besides Trump to accomplish this when Trump was so often dominating coverage.

In part, this was a function of candidate personality and strategy. Someone like Jeb Bush really could not provide the controversy and conflict that Trump could. But in part, it was a function of how much media oxygen Trump was sucking up.

Did he receive “too much” coverage?

Many have suggested that Trump received “too much” coverage or more coverage than he “should have.” Certainly Trump’s opponents thought so, including Jeb Bush, Ted Cruz, John Kasich and Hillary Clinton. So do at least some in the media:

  • Campbell Brown: “Early on, even before he was the front-runner, TV news was giving Trump far more attention than other candidates and far more than he deserved.”
  • Brian Stelter: “Trump is the media’s addiction. When he speaks, he is given something no other candidate gets. That’s wall-to-wall coverage here on cable news. He sucks up all the oxygen.”
  • Walter Shapiro: “Since I began covering presidential campaigns in 1980, I can think of nothing as unfair as the disproportionate media attention that has been lavished on Trump from the beginning.”
  • Ann Curry: She is “embarrassed by the unfairness to other Republican candidates, who didn’t get nearly the same airtime.”
  • Roger Ailes: “Did he get too much coverage? Yes.”
  • A New York Times headline: “Television Networks Struggle to Provide Equal Airtime in the Era of Trump”

If we want to know whether Trump got “too much” coverage, we have to ask the question, “Compared to what?” You could frame the comparison in terms of other candidates. For example, the New York Times’ Jonathan Martin asks whether coverage of Hillary Clinton rallies would feature a video stream of the empty stage, as coverage of Trump rallies sometimes have.

One analysis from December found that Trump’s news coverage — the 54 percent of American newspaper coverage mentioned earlier — exceeded his polling numbers by over 20 points. This put Trump near the top of the list of all the 1984-2012 primary candidates in terms of how much their share of news coverage exceeded their share in the polls.

Poll numbers might not be the only relevant metric. But it strikes us as persuasive that Trump’s coverage exceeded any defensible reflection of his poll numbers — again, leaving aside the fact that those poll numbers were in part a function of news coverage.

If Trump was covered “more” than poll numbers alone would suggest, we can again chalk it up to some combination of Trump’s behavior, news values, economic incentives and the public’s interests. Whether, then, you think that Trump got too much coverage will likely depend on how reasonable you think any of those factors is, or perhaps on whether you think other candidates “deserved” more coverage in their own right.

Ultimately, this is not an easy question to answer, because there is no shared standard. But we can say that historically speaking, Trump was unusual in the volume of coverage he received, relative to his standing in the electorate.

How favorable or unfavorable to Trump was this coverage?

Defenders of the news media are quick to argue that, even if Trump got a lot of news coverage, much of it was negative. Thus, the argument goes, the media did its job in scrutinizing Trump.

Certainly one can point to many examples of scrutiny of Trump or of coverage that seemed unfavorable. But at the same time, it is easy to find other examples — including from journalists — suggesting that, in fact, the media failed to scrutinize Trump enough and instead let him dictate the terms of coverage. See, for example, Dana Milbank, Marc Ambinder, Ezra Klein, Ben Smith, Nicholas Kristof, Campbell Brown and David Folkenflik.

Warring anecdotes aren’t going to get us very far. Here are two pieces of hard data that reflect analysis of hundreds if not thousands of news stories.

First, Crimson Hexagon uses computer algorithms to estimate whether the language in each campaign article is positive, negative or neutral. This gives us a sense of the tone of the articles mentioning each candidate. Of course, since any article can mention more than one candidate, the resulting data is not as precise as it could be. Nevertheless, you might expect differences among the candidates. But there is little:

Second, the firm Media Tenor uses humans, not computers, to code individual articles and determine whether they are positive. Their conclusion was reported in a recent report from the Shorenstein Center, and it is different than what Crimson Hexagon’s data suggest — but not in a way that suggests Trump received distinctly unfavorable coverage.

Media Tenor finds that, across eight different outlets, Trump’s coverage throughout 2015 was mostly positive. Why? Because a substantial fraction of the coverage — about 55 percent — focused on horse-race metrics like the polls or on campaign activities and events. This coverage was positive because Trump tended perform well by horse-race metrics. Nate Silver reached the same conclusion analyzing a different set of news coverage. Good poll numbers make for better coverage.

This is a familiar pattern. In the 2008 general election campaign, for example, the front-runner, Barack Obama, tended to receive more favorable coverage than John McCain — except right after the Republican National Convention, when the polls briefly shifted in McCain’s favor.

The Shorenstein Center report also finds that coverage of Trump’s personality and issue positions wasn’t always negative — even if his most controversial statements are the easiest to call to mind. Writes Thomas Patterson:

Moreover, not everything that was said about Trump’s personal characteristics and issue positions was negative in tone. Over 40 percent of it was positive in tone, often in the form of statements by voters who agreed with his policy positions or liked his personal style.

Of course, it’s perfectly legitimate to quote from Trump’s supporters in news articles. But this coverage — combined coverage of horse-race metrics showing that Trump was, yes, winning — may not been more positive than negative.

Obviously, these two sources of data — which are based in different methods for coding media content — don’t tell the same story. But taken together, they suggest two conclusions. First, even though Trump has been routinely characterized as a candidate who goes beyond accepted standards of political discourse, he wasn’t covered less favorably than the other candidates. Second, he might not have been covered that unfavorably at all.

If that is true, one partial reason is the behavior of the other candidates. They were loath to attack Trump — even though attacks are often what helps generate unfavorable coverage as the media quotes these attacks, investigates the claims and uncovers more dirt. Trump’s opponents were slow to attack him in debates and slow to attack him in television ads. If the other GOP candidates feel as if Trump wasn’t scrutinized enough by the news media, they have themselves to blame, at least in part.

But how do we know media coverage actually affected voters?

It shouldn’t be difficult to think that coverage of Trump — valued at about $2 billion as of March — did affect voters. The logic here is simple: People do not change their opinions without new information. People do not suddenly decide, for no reason, to support a candidate they hadn’t previously supported. For Trump, a lot of the relevant information came from the news media.

It is why Republicans came to see him more favorably in the initial months after he announced his candidacy.

And it is why the volume of coverage that Trump and the other Republican candidates received was so strongly correlated with their poll numbers. Here are two graphs to show this.

The first graph compares each candidate’s average share of cable news coverage to his average share in national polls, beginning when Trump announced his candidacy and ending on April 30:

We can also track news coverage and Trump’s polling numbers over time. This helps show how Trump’s poll numbers increased after the initial surge of media coverage. It also shows that Trump’s poll numbers tended to decline or plateau when he received less coverage — as during the fall of 2015.

The two people who have put the most work into sorting out correlation and causation in primary news coverage are Kevin Reuning and Nick Dietrich, both PhD  students in political science at Penn State. Their analysis involves careful statistical modeling to uncover the relationships among media coverage, poll numbers and public interest (as manifest in Google searches). Based on analyzing both the 2012 and 2016 primaries, they conclude:

Our analysis demonstrates that the effects of media coverage are long lasting, leading to substantively large increases in public interest and support for candidates.

They find that the effects of media coverage on polling are much larger than the effects of polling on media coverage. Media coverage is the driver.

Which voters were most affected?

One of the classic books on public opinion starts with a simple formulation: Opinions are the product of both information and predispositions. Information is what citizens receive from media coverage and other sources. Predispositions, or people’s prior beliefs, filter that information. Unsurprisingly, people tend to believe that which accords with their prior beliefs, and discount what does not.

So we would expect the impact of Trump’s news coverage to be more pronounced among those who are inclined to believe his message. And, indeed that is what the data shows over and over. In the primary, Trump’s support was strongest among those who were not orthodox conservatives, who expressed economic anxiety, who had a stronger white identity and who had less favorable views of immigrants, Muslims, African Americans and other minority groups. Meanwhile, views of Trump became less favorable among Latinos and women.

In other words, both support for Trump and opposition to Trump are the result, in part, of media coverage.

What about the general election?

The dynamics of media coverage and support for the candidates change once we move from the primary to the general election campaign.

Now that it is a two-person race, not a 17-person race as in the primary, Trump’s advantage in the quantity of coverage is likely to shrink and perhaps disappear.

Now that it is a two-person race, Trump is experiencing more scrutiny. Part of this is of his own making. Part of this is because some in his party are balking or at least visibly uneasy. Part of this is because there are more media resources available to do the investigations now that there aren’t 17 candidates to cover.

And now that we are in the general election campaign, the relevant electorate is broader, and key groups of potentially persuadable voters will have political values different than Trump’s supporters in the Republican primary. Trump now needs to appeal to a different group of voters, and he may find that media coverage will not help him do this — at least if he persists with a similar message.

This all adds up to a potentially different impact of media coverage in the general election than we saw during the primary. It is perhaps why Trump has seen his poll numbers slump.