A worker sweeps the Des Moines Register Soapbox at the Iowa State Fair on Thursday in Des Moines. (John Taggart/Bloomberg News)

Imagine watching a horse race in slow motion. Maybe you’re a racing aficionado or maybe you have a lot of money on one of the ponies but, for whatever reason, you’re very invested in knowing the outcome. As the horses round the first corner — slowly — you’re examining the positioning closely. You’re poring over high-definition video footage of how the horses are running, where the jockeys are leaning, how their footsteps are landing in the dirt. There’s a lot of time left, but given the stakes you can’t take your eyes off what’s happening.

Welcome to presidential primary polling.

We are, as of writing, about half a year away from the first voting in the 2020 Democratic presidential primary. There’s an enormous amount of time left in which any number of things can happen. And yet political obsessives and candidates rip apart new polls like the handicappers in our imagined scenario, looking for any indication of where and how the candidates are moving against one another. This is not meant derogatorily: We do it, too.

(In fact, we even have an animated version of the 2020 contest as a horse race, just to get that out of the way.)

It’s all interesting and the stakes are obviously enormous. It’s just that … we might want to be wary about reading too much into things.

Since 2008, the states of Iowa, New Hampshire and South Carolina have held 15 party primary contests. Of those 15, only five times were the eventual winners leading in state polls taken as far out as we are now from those contests.

Consider Iowa. Only in 2016 did the eventual winner, Hillary Clinton, lead this far out. But this was actually one of the least predictive results: Her lead was about 30 points, but Sen. Bernie Sanders (I-Vt.) essentially ended up closing the race to a near-tie.


(Philip Bump/The Washington Post)

In New Hampshire, polls this distant from the 2020 primaries were better. Three times, the eventual winner led in the polls this far out. One exception was Hillary Clinton, who led in New Hampshire but ended up losing the state. The other exception was in 2008, when John McCain trailed for most of the primary season.


(Philip Bump/The Washington Post)

In South Carolina, Clinton was again the only eventual leader who was winning at this point. Donald Trump, who won the state in 2016, surged into a lead that he’d end up maintaining soon after this point, but the other three candidates who won the state only gained ground later on.


(Philip Bump/The Washington Post)

You’ll notice that we’ve also included markers indicating where the contests ended up. In three of those 15 contests, even the final polling averages had the eventual winner trailing. The final averages in the five Iowa contests were an average of 4.6 points off. In New Hampshire, 4.1 points. In South Carolina, 8.8 points.

New Hampshire tended to have the most consistently accurate polling for the eventual winners since 2008. Iowa and South Carolina were generally more accurate now than they would be in a few months.


(Philip Bump/The Washington Post)

Notice on that chart all of the volatility at the end of the graphs. This is always worth keeping in mind: The results of one contest shape the results of other contests. Winners in Iowa tend to get a boost in New Hampshire and so on. People drop out. Things shift quickly once people start voting.

It’s like watching that slow-motion horse race knowing that, once the first horse crosses the finish line, there’s a brief tremor which causes the horses to stumble and shift. But nonetheless, here we are, watching each quivering muscle and obsessing over each jockey’s every move.

Because it’s interesting. Just be wary of placing bets midrace.