Skip to main content

How the Media Fails the Public in Reporting Election Polling

The horse race national polls don’t matter.

By Daniel J. McGraw


Barack Obama shakes hands with Mitt Romney following a debate in Boca Raton, Florida. (Photo: Michael Reynolds-Pool/Getty Images)

It was about one week before the November 2012 election and I was sure I had missed something. The national media had declared the race between Mitt Romney and Barack Obama “too close to call.” So close, in fact, that we might be staring at a re-count. But, after spending hours staring at state polling figures — and adding up the Electoral College numbers — I’d long ago come to the conclusion that this was going to be a landslide victory for Obama.

And as I sat there in front of my screen, looking at the latest polls on RealClearPolitics, and adding up those state numbers in every permutation possible — Obama had at least 300 electoral votes locked up, well above the 270 needed to win — I could only come to one conclusion: Obama had won his re-election. But the media — outlets like the Boston Globe, National Public Radio, USA Today, and countless others — continued to headline the race as being a nail-biter. (The Los Angeles Times was comparatively tempered, admitting that, while the race was very close, it “might not drag out for days and weeks of wrangling over disputed ballots, as some feared.”)

How, I kept wondering, was I so off base on this? I checked in with some moderately intelligent friends and family who follow politics; they all claimed the race was too close to call. After the numbers were finally tallied, Obama won handily, 332 to 206.

How did the media get this so wrong? Well, they didn’t, really. It was a close race, if the only numbers you looked at were the national polls, rather than adding up the Electoral College numbers from the states. A week before the election, the national popular vote polls had the 2012 race in a virtual dead heat. So when members of the media called the election “too close to call,” they were right in some respects — if you were only talking about the national popular vote. Obama did win the popular vote, but just barely, 51.1 percent to 47.2 percent. So, indeed, it was close.

The problem here is that the national popular vote does not count, at least not when deciding a winner. And yet most of the media, with every possible data resource available, uses national polling as its primary news data when reporting who is winning and who is losing the race. There is a good reason for this. In the media’s current business model, where news is cheap programming and the presidential race is prime time, the national polls are important: They almost always make the race seem closer than it really is, and a closer race keeps the readers and viewers tuned in.

This is not to say that national polls are meaningless. Measuring the public’s view of the candidates after a big news event, such as a debate or a party convention, can give the media and the electorate a gauge on how these events affect the race. To be sure, after Obama bungled things in the first debate against Romney in 2012, the national polls showed a dip in his numbers.

Where news is cheap programming and the presidential race is prime time, the national polls are important.

But the media coverage of the “horse race” rarely mentions the Electoral College vote. This is because national polling always makes the races look closer, in the process creating a better narrative. Take the 1984 race. Ronald Reagan beat incumbent president Jimmy Carter by 10 percentage points in the popular vote, 51 to 41 (with John Anderson, an Independent, taking seven percent). But Reagan won the Electoral College by a 489 to 49 margin. So 49 percent of the public voted against Reagan, yet he took 90 percent of the Electoral College numbers.

Part of this emphasis on national polling comes because the political punditry tends to put more states in the “toss-up” category, either to err on the side of caution or to keep the interest up. In the 2012 race, polling aggregators like RealClearPolitics (which averages out all of the Independent polls), identified 11 toss-up states until the very end (where the candidates were within five percentage points of each other) with a total of 146 electoral votes. When you took those 146 electoral college votes out, Obama was leading 201 to 191. Seems too close to call, right?

The mainstream media went with those state averages put forth by RealClearPolitics — which included polls by Gallup, Quinnipiac, universities, media companies, think tanks, etc. — even when the standard for what states were in play was broadly defined with a very low bar. It’s not that the media was misreporting the data, but very few in the media were questioning which states were no longer “toss-ups” as the election drew closer.

Specifically, the news that Obama was trending upward in the last six weeks. In Pennsylvania, for example, Obama had won six of the last seven polls in the state by an average of four percentage points. Yet it was still considered in play (Obama won Pennsylvania by six). Other Midwestern states like Iowa, Michigan, Ohio, and Wisconsin were all also considered “too close to call,” but the latest polling numbers showed Obama was comfortably ahead in all of them.

In the end, of the 11 states considered toss-up states under this very loose criteria, only Colorado, Florida, and Virginia really were. If Romney would have taken all three, he still would have been 13 short of the 270. He lost all three.

Many years ago, before network news and the big newspapers poured most of their resources into garnering clicks, reporters and writers tended to focus on what was happening in the presidential race, and not what they hoped would happen. All that has changed in the past 10 years or so, as the presidential race has morphed into near-constant news programming.

In order to keep the election cycle producing its maximum click potential, making the race seem closer than it really is — by emphasizing national polls over the more important state polling numbers — is a business decision first and foremost. The national polling also keeps the interest higher in states that are not in play — populous states like New York, California, Texas, and Illinois.

Does this mean we should get rid of the national polls? Of course not. But they shouldn’t be a primary news source.

This time around, the media is once again going full bore on the national numbers. In a piece on RealClearPolitics last week, “How Clinton vs. Trump Could Play Out,” the author notes that Donald Trump needs “to win more than 60 percent of Independents” to beat Hillary Clinton. My question is: “In what state?” Taking 60 percent of independent votes in California probably won’t matter much, but taking that same number in Florida might.

CNN’s story (headlined “Will Trump vs. Clinton Be a Nailbiter?”) spends the first 18 paragraphs quoting experts claiming the race will be very close in November. It isn’t until the 19th paragraph that the writer admits: “[T]he latest electoral map compiled by the University of Virginia Center for Politics assumes 190 electoral votes are safe for Democrats and 142 are assured for Republicans. Once 57 likely Democratic votes are added, Clinton theoretically comes within 23 electoral votes — or a win in Florida, or in a trio of swing states Virginia, Colorado and Nevada — of claiming the presidency. Trump has a much thinner margin for error.”

So the answer to the headline, at this point in time, given that Hillary is likely one state away from victory, is that this race will not be a nail-biter. But then again, recent national polls do show Trump to be gaining ground on Clinton. If, a month before the election, it’s looking like Trump will lose to Clinton in the Electoral College by a large margin, expect to see the media emphasize Trump’s slim trailing in the national polls. If it’s the other way around, with Trump crushing Hillary in the swing state polls and with more than 270 Electoral College votes likely in his column, they’ll claim Clinton is only a few points behind.

That’s how they keep us clicking.