POLITICS

Why It’s Ridiculous To Report On Every Poll Coming Out Of New Hampshire

There are a lot of polls. And they might not reveal much about what will happen on Tuesday.

Pollsters have been busy in New Hampshire the last few days.

As of this post's publication, HuffPost Pollster’s charts have added 15 polls of the New Hampshire primaries that were conducted after last Monday's Iowa caucuses -- nine of the Republican primary and six of the Democratic primary. If you include all of the daily tracking polls, which are only added to the charts every other day because of their overlapping dates, there have been 22 poll releases since Wednesday. 

If that seems excessive or ridiculous, that’s because it is. And there will be several more -- those numbers above will be out of date very quickly. Even as a professional poll follower and analyst, I understand this sentiment: 

The phenomenon of heavy post-Iowa caucus polling in New Hampshire isn’t new. Even though there was only a race for the Republican nomination in 2012, there were still 13 polls released in the seven days between the Iowa caucuses and New Hampshire primary.

Part of the reason for so many polls right before the New Hampshire primary is that the state is known for its volatility. Voters there often make up their minds at the last minute, partly in response to results from Iowa. Indecisive voters can easily skew Election Day results away from what polls had indicated would happen. Pollsters try to poll as late as possible to capture 11th-hour momentum and decisions, but it doesn’t always work.

In 2008, pollsters missed the Democratic vote outcome in New Hampshire despite lots of late polling. Hillary Clinton’s surprise win on primary night over then-Sen. Barack Obama stunned the polling community so much that the American Association for Public Opinion Research put out a report about why the polls missed the outcome. But most of the major issues the report pointed to remain unresolved -- polls still end before voting begins, because they have to in order to get out before the vote results. Likely voter models are often flawed in their predictions of who will vote, and hard-to-reach voters are still hard to reach.

That’s why pollsters, even as they put out their numbers and vie for media attention, are warning us that their final estimates of where the race stands might not be indicative of how people will vote on Tuesday.

So why are we inundated with polls if they might not mean anything for the outcome?

Keep in mind, polls are meant to show the state of the race at the time they are conducted. It’s interesting just to see how and why opinions may have changed so quickly, or what went wrong when polls don’t anticipate the outcome. Polls are not predictive.

That said, there are still a lot of polls. One major thing many of the New Hampshire polls have in common is media sponsorship. Nine media outlets have partnered with polling organizations to conduct polls in New Hampshire in the last few days. Only two of those 15 polls conducted since the Iowa caucuses -- and 1 of 8 pollsters -- are unsponsored.

The utility for media organizations of having so many different polls is something of a mystery. Many of the New Hampshire polls show similar trends, although with some notable differences -- Sen. Bernie Sanders (I-Vt.) might have a commanding 30-point lead over Clinton, or he might be up by only 9 points. Ohio Gov. John Kasich might be in second place for the Republicans, but it's possible either Sen. Marco Rubio (R-Fla.) or Sen. Ted Cruz (R-Texas) holds that spot. The best option for understanding where things stand is probably to look at polling averages

It’s impossible to know which poll is more accurate at this point -- and we will likely never know if voters' opinions change as quickly as they usually do leading up to the New Hampshire primary. Just don’t be surprised if the polls don’t nail the outcomes. New Hampshire likes to be surprising.

CONVERSATIONS