Skip to main content
ABC News
Trump Boom Or Trump Bubble?

The Republican debate on Tuesday night in Las Vegas will offer Republican presidential contenders a rare opportunity: the chance, at least for a couple of hours, to compete on a level stage with Donald Trump, who has dominated coverage of the campaign for months on end.

Understanding the dynamics of the modern media environment is an important skill for a candidate, and it’s a skill that Trump has mastered. But it’s also important to understand the effects that media coverage can have on the campaign and on the polls. By one measure we’ll get to in a moment, Trump has received about the most disproportionate media coverage ever for a primary candidate. The risk to Trump and candidates like him is that polling built on a foundation of media coverage can be subject to a correction when the news environment changes.

The data I’ll cite in this article comes from searches of, an online archive of American newspapers from the 1980s onward. For each competitive primary since 1984, I’ve counted the number of articles about each candidate in the last six months of the pre-primary year (for example, July 2003 through December 2003 in the case of the 2004 Democratic nomination). I count it as a “hit” whenever a candidate’s full name1 appears in the lead paragraph of an article, but no other candidate’s name2 does.3 I limited the search to candidates who were still officially running for president as of Dec. 31 of the pre-primary year, excluding those who dropped out of the race early or who never officially entered.

Which candidates dominated the media primary?
1984 Democratic Jesse Jackson (33%) Walter Mondale (22%) 1.5
1988 Democratic Gary Hart (36%) Jesse Jackson (21%) 1.8
1988 Republican George H.W. Bush (41%) Bob Dole (23%) 1.8
1992 Democratic Douglas Wilder (40%) Bob Kerrey (17%) 2.3
1996 Republican Bob Dole (69%) Phil Gramm (10%) 6.7
2000 Democratic Al Gore (70%) Bill Bradley (30%) 2.3
2000 Republican George W. Bush (72%) John McCain (16%) 4.5
2004 Democratic Howard Dean (40%) Wesley Clark (19%) 2.1
2008 Democratic Hillary Clinton (45%) Barack Obama (27%) 1.7
2008 Republican Rudy Giuliani (24%) Mitt Romney (24%) 1.0
2012 Republican Rick Perry (37%) Mitt Romney (22%) 1.7
2016 Democratic Hillary Clinton (77%) Bernie Sanders (20%) 3.8
2016 Republican Donald Trump (54%) Jeb Bush (8%) 6.3

Since July, Trump has received 54 percent of the media coverage of the GOP primary4 — about six times more than Jeb Bush, who’s in second place with just 8 percent of coverage.

Trump isn’t the only candidate to receive such a large fraction of coverage in his primary — Hillary Clinton is getting 77 percent of the media coverage in hers, far exceeding Bernie Sanders’s 20 percent. Bob Dole dominated media coverage in the 1996 GOP race, and George W. Bush and Al Gore did in their respective races in 2000.

But Dole, Bush and Gore were much more dominant than Trump in the polls (not to mention other measures like the “invisible primary5). So is Clinton this year. Trump has spent most of the year in the high 20s or low 30s in national polls (with some higher numbers recently — we’ll get back to those in a moment). That’s better than most observers, us included, would have expected. The other candidates I mentioned, however, routinely had — or have in Clinton’s case — numbers in the 40s, 50s or 60s in national polls, about twice Trump’s level of support.6

Take a listen to our election podcast pilot

We’re launching our election podcast in January. For now, take a listen to the pilot in the feed for our podcast What’s The Point.

Historically, in fact,7 there has been nearly a one-to-one correspondence between a candidate’s share of media coverage and his share of the vote in the polls. That is, other things held equal, a candidate earning 30 percent in national polls tends to get about 30 percent of the media coverage, while one polling at 10 percent will get 10 percent of it instead. It’s just that simple.8

Thus, we can readily compare a candidate’s share of media coverage to his polling average. Trump, for example, has received an average of 28 percent of the Republican vote in national polls since July, according to HuffPost Pollster. Prorate that number upward to exclude undecided voters and candidates who have exited the race, and you get him up to 32 percent. By comparison, Trump has received 54 percent of the media coverage of the GOP race, so his media coverage has exceeded his share in the polls by 22 percentage points.

That is a big gap, although not the largest on record. Instead, the record belongs to Jesse Jackson, who received 33 percent of the media coverage in the run-up to the 1984 Democratic primaries despite usually polling only in the high single digits.

Candidates with more support in media than in polls
Jesse Jackson 1984 D 33% 9% +24%
Donald Trump 2016 R 54% 32% +22%
Douglas Wilder 1992 D 40% 19% +21%
Rick Perry 2012 R 37% 18% +19%
Howard Dean 2004 D 40% 21% +19%
Pat Robertson 1988 R 23% 7% +16%
Hillary Clinton 2016 D 77% 65% +12%
Mitt Romney 2008 R 24% 13% +11%

It’s odd to compare Jackson and Trump, but their candidacies have some similarities: Both were nationally renowned (and controversial) figures before embarking on their campaigns, and their candidacies were strongly opposed by most members of their party establishment. Eventually, Jackson fared reasonably well, winning two states and 18 percent of the Democratic vote in the 1984 primaries and advancing political participation in the black community, although he never came close to winning the nomination.

Other candidates whose media coverage exceeded their polls did not perform so well, however. The media excitement surrounding Virginia Gov. Douglas Wilder’s campaign in 1991 — understandable given that he’d recently become the first African-American governor elected in any state since Reconstruction — never translated into much momentum, and Wilder exited the race in January.9 Rick Perry’s highly anticipated campaign during the 2012 cycle went boom-and-bust quickly.

A potentially more important precedent for Trump is Howard Dean, who overcame Democrats’ initial skepticism to build up quite a lot of momentum in the polls in late 2003. Juxtaposed against a bland field of candidates such as Dick Gephardt and John Kerry, Dean was the most exciting thing going. But he became the central figure in the narrative in a way that exceeded his good-but-not-great level of popular support, with media coverage that roughly doubled his share of the vote in polls. Dean’s numbers faded in the last few days before the Iowa caucuses, and the media turned on him after his third-place finish there, contributing to a crash in his national standing.

This is not to say that candidates whose media coverage exceeds their polls are necessarily doomed; in a regression analysis, the effect of media coverage on a candidate’s eventual share of the national primary popular vote is neutral, controlling for his share of the vote in polls. But media-dependent candidates have considerably more volatility and uncertainty in their results once the voting takes place; a higher share of media coverage is correlated with a higher error in predicting a candidate’s eventual vote share.

Here’s how that could be a problem for Trump. Despite how contentious the Republican race has become, the overwhelming majority of Republican voters are still considering multiple candidates and have a favorable impression of several Republicans. Trump has consistently led in polls when voters are asked for their first choice, but his net favorability ratings are only in the middle of the pack. In other words, he’s converting an unusually high percentage of potential supporters into people who list him as their first choice.

How is he managing to pull that off? Some of it may be because he excites voters more than other candidates do — and differentiates himself more from the pack. But some of it may also be because he comes up first in voters’ minds because of the way he’s so dominated news coverage of the race.

It’s possible to develop a model of this. Below, I’ve listed each Republican candidate’s share of media coverage since July, along with the candidate’s average net favorability rating10 among Republican voters. In a linear regression, these two variables predict a candidate’s polling average almost perfectly11 — usually within a percentage point or two.

Polls can be predicted based on favorability ratings and media coverage
Carson +61% 7% 14% 10%
Rubio +61% 4% 8% 9%
Cruz +49% 5% 8% 8%
Fiorina +42% 4% 4% 6%
Huckabee +42% 2% 4% 6%
Bush +29% 8% 9% 7%
Trump +28% 54% 28% 28%
Santorum +23% 1% 1% 3%
Kasich +21% 5% 3% 4%
Christie +18% 5% 3% 4%
Paul +9% 3% 4% 2%
Gilmore -7% 0% 0% 0%
Graham -8% 2% 1% 0%
Pataki -11% 0% 0% 0%

What this means is that a candidate with medium favorability ratings may nevertheless come up as the plurality first choice if he cleans everyone’s clock in terms of media coverage. (Conversely, if they’re not top-of-mind, candidates — like Ben Carson or Marco Rubio — may be everyone’s second or third choice despite being broadly acceptable to Republican voters.)

The interesting thing, as we’ve pointed out before, is that a candidate can potentially gain in the polls in the short term by increasing his media coverage, even if he potentially hurts his favorability rating. Trump seems to realize this. So far in December — a month in which, among other things, he’s proposed to ban Muslims from entering the United States — he’s been the subject of 70 percent of media coverage of the Republican race, even higher than his long-term average of 54 percent. According to the regression, that extra media coverage is worth about 8 percentage points in the polls: almost exactly how much Trump has gained in national polls since the month began and enough to put him in the mid-30s in the polling average instead of the high 20s.

Is such a strategy sustainable? Can Trump continue to dominate media coverage as much as he has already?

The news cycle over the past month has been favorable for Trump. The terror attacks in Paris and San Bernardino played into his core message about the dangers posed by Islamic State. Meanwhile, there hasn’t been much going on to compete with this news, with no Republican debate since Nov. 10 and no other major shake-ups in the race such as a leading candidate dropping out.

Meanwhile, as Trump probably realizes, the media’s obsession with polls can become a self-perpetuating cycle: Trump’s being in the media spotlight tends to help him in the polls, which in turn keeps him in the spotlight, which in turn helps in the polls, and so forth.

But that will change as debates occur more frequently (there are three scheduled between now and Iowa) and other candidates begin to drop out of the race. Ted Cruz is also emerging as a potential foil to Trump, when one may have been lacking before. Most importantly, Republicans will begin going to the polls, starting Feb. 1 in the Iowa caucuses.

My guess is that most of these eventualities represent more downside than upside for Trump, simply because his dominance of the news cycle is so complete right now that other candidates almost can’t help but catch up. One of the usual rewards for winning Iowa or New Hampshire is a massive increase in media coverage, but Trump already has plenty of it. If Cruz or Rubio were to win one of those states, conversely, the newly won attention could help them convert their broad acceptability across the Republican electorate into first place in the polls and in future states.

So far, however, Trump has exploited every opportunity to keep his momentum going. And even if his candidacy is a bubble, there’s a chance that it won’t burst until after he’s started racking up delegates and primary wins.

Check our our live coverage of the Republican debate.


  1. In cases where candidates are commonly referred to by more than one full name — for example, Hillary Clinton and Hillary Rodham Clinton or Richard Gephardt and Dick Gephardt — I counted either name toward the total.

  2. More specifically, no other candidate who was running for the same nomination in the same year.

  3. This serves to exclude cases where there is a long, undifferentiated list of names — for example, an article that listed polling results for several candidates.

  4. Excluding coverage of candidates who have since dropped out.

  5. Dole, Bush and Gore had overwhelming leads in endorsements, similar to what Clinton has now; Trump, by contrast, has not yet been endorsed by a single governor or member of Congress. There’s historically been no correlation between endorsements and media coverage, however, once polls are accounted for.

  6. Maybe the polling front-runner always wins the lion’s share of media coverage, whatever his or her absolute level of support? (In other words, maybe it matters that you’re ahead — not whether you have 52 percent of the vote or 27 percent.) Not quite. Rudy Giuliani had highly similar numbers to Trump’s throughout the second half of 2007, almost always leading national polls with support in the high 20s or low 30s. But he got just 24 percent of the media coverage of his race, compared with Trump’s 54 percent. Likewise, Mitt Romney got only 22 percent of the media coverage four years ago despite having fairly similar polling numbers to Trump’s, although Romney periodically lost his lead to “surging” candidates like Newt Gingrich while Trump’s advantage has been steadier.

  7. According to a regression analysis.

  8. Well, almost that simple. In some of the data you’ll see below, I adjust polling averages by excluding undecided voters and supporters of candidates who dropped out or never officially entered the race.

  9. Wilder’s polling support, 19 percent, in the table above is slightly deceptive because there were large numbers of undecided voters in the 1992 cycle, along with supporters of candidates like Mario Cuomo who never officially entered the race; Wilder’s unadjusted polling average was 8 percent.

  10. Net favorability ratings are based on an average of Quinnipiac polls since July, since they regularly test favorability ratings even for minor candidates. Three candidates (Lindsey Graham, Jim Gilmore, George Pataki) have a negative net favorability rating, which are adjusted upward to zero for purposes of running the regression analysis.

  11. The r-squared in the regression is .95.

Nate Silver founded and was the editor in chief of FiveThirtyEight.