7 Terrible Presidential Polls You Should Know

Some Democratic voters might be sweating a new poll that showed Donald Trump defeating Hillary Clinton by two points in a general election. A couple of polls have shown Trump leads over the last few weeks, in fact, and while the anxiety they produce is understandable for those anti-Trump folks, it’s not warranted just yet. At this point, a few polls here and there don’t mean much. In fact, the polls are likely to be all over the place, and a look back at some of the most inaccurate presidential polls of recent elections makes that clear.

While polls are generally pretty accurate, there is a right and wrong way to read them. The surveys that show Trump winning should be viewed with skepticism for two reasons: First, because it’s only May and, second, because the general election campaign hasn’t even started yet. And once the campaign does start, polls in swing states will be much more valuable than national polls, since the United States elects its presidents through the electoral college, not the national popular vote.

Of course, Trump could still win. The point, though, is that this one poll shouldn’t convince us that he will. There are many more ways in which it’s possible to misinterpret polls, so let's see what we can learn from some more woefully inaccurate presidential polls:

John McCain Wins 48 States

Back in 2006, SurveyUSA conducted an enormous poll (of 30,000 people) across all 50 states. The survey found that if Barack Obama and John McCain were the two major party nominees, McCain would win every state except for Hawaii and Illinois, coming away with 510 electoral votes to Obama’s 28 and easily winning the presidency.

Of course, the opposite happened, as Obama won more than half of the states in the country and soundly defeated McCain by almost 200 electoral votes. This is just one example of why you shouldn’t put too much faith in polls that are taken two years in advance.

Ross Perot Wins Three-Way Race

The 1992 election featured incumbent President George H.W. Bush going up against Arkansas Governor Bill Clinton and billionaire Ross Perot. Although we all know how the election turned out, there was a period during the summer in which independent candidate Perot was ahead of both Clinton and Bush in the polls. In fact, Clinton came in dead last in several three-way polls.

Maybe it was Clinton’s successful debate performances that pushed him into the lead; maybe it was Perot’s surprise decision to suspend, then restart, his campaign that tipped the scales. Whatever the reason, it’s clear that June 1992 was too early for the polls to be reliable, especially in a three-way race.

Mitt Romney Leads Obama On Eve Of Election

Even polls taken within weeks or days of an election can be wrong. During the 2012 race, Gallup’s national polls consistently showed Mitt Romney defeating President Obama, including one in mid-October that showed Romney up by six points. Romney did not win that election.

In a sense, this poll can be retroactively discounted on the grounds that it was a national poll, and the popular vote isn’t how presidents are determined. But even then, the poll was wrong — Obama won the popular vote by four points. Gallup’s polls were so consistently wrong during the 2012 cycle that the firm launched an internal review to identify all of the systemic problems in their methodology, of which there were several.

McCain Gets A Convention Bounce

Shortly after the 2008 Republican National Convention, polls began showing McCain taking the lead over Obama for the first time in months. In fact, McCain’s numbers jumped up by six points on average after the convention.

However, that wasn’t a good reason for Republicans to get excited. Throughout history, candidates of both parties have received a “convention bounce” in the polls shortly after they’re officially nominated. (This usually fades after a couple of weeks and doesn’t actually indicate any change in the state of the race.) As an interesting aside, only two presidential candidates have actually received negative convention bounces: John Kerry and Romney.

Clinton Wins Michigan In Landslide

Some of the worst presidential polls in history were taken shortly before the Democratic primary in Michigan earlier this year. All surveys showed Hillary Clinton winning the state in a landslide, including one showing Clinton winning by 37 points. To everybody's surprise, Bernie Sanders managed to eek out a two-point lead and an overall victory.

This fiasco was one of the worst in polling history, and it was widespread — Clinton had an average lead of 10.5 points across many different polling firms. The only takeaway here is that every so often, there’s a system-wide error in polling that completely fails to capture the mood of the electorate. Sometimes, polls are just wrong.

Half Of Clinton Supporters Won’t Back Obama

The question of whether Sanders’ supporters will vote for Clinton if she wins the nomination has recent precedent: Democrats were asking themselves a similar question during the 2008 primary, and they had reason to be worried. One exit poll from May found that 50 percent of Clinton’s supporters wouldn’t vote for Obama if he won the nomination.

It may be a tad unfair to call this a “bad poll,” because maybe those Clinton voters truly didn’t plan to vote for Obama at the time. Nevertheless, the point is that general election campaigns really do matter, and the bitter feelings that result from an intense primary often fade as the reality of the opposition sinks in.

Rudy Giuliani Becomes Republican Front-Runner

Remember when Rudy Giuliani was president? Nope, you don’t. But in the early stages of the 2008 race, polls showed that Giuliani was far and away the favorite to win the Republican nomination. His success was considered such a given that Democrats felt compelled to attack him during their own debates.

But Giuliani unwisely bet his entire campaign on winning the Florida primary, so when he lost Florida, he ended his campaign. This shows that even the best polls can’t account for bad campaign strategies.

We’ll be seeing countless polls between now and November, and it’s important to remember that a whole lot of them, when all is said and done, will turn out to be wrong.