Election polling is always wrong. Except when it’s not. And even then sometimes. But especially in New Hampshire this year. Confused yet? You should be.
Every single poll that’s come out in the last week has said something completely different is happening in the first primary, and not even in a linear trending way, with certain candidates ascending or descending. They’re literally all over the place.
The following is a list of all the polls currently included in the Real Clear Politics average for New Hampshire.
From this data we can tell only three things:
- Trump is leading in New Hampshire.
- There is a fight for second place between Rubio, Kasich, Cruz, and Bush.
- Christie, Fiorina, and Carson should probably have dropped out yesterday.
Second is where the real mystery is. Every poll shows a different person (or tie) in second place behind Trump [highlighted above in yellow]. Rubio comes in second place in one poll and ties for second in two others, but so does Bush in one, Cruz in one and ties in another, and Kasich in two and ties in another.
Some of these individual polls even show trends from earlier data, which usually signal a real change in support. But even here those trends are contradictory. In the CNN poll, Rubio has gained 6 points from the same poll taken a little over a week ago, while the other candidates have only gained 1 point. But in other polls, like the UMass daily tracking poll, Rubio has lost 2 points over the last week.
So which is it? Is Rubio surging after a surprise finish in Iowa? Or is Rubio faltering after a shaky debate performance on Saturday? It could be neither…or both consecutively! We have no way to tell before the results are counted tonight.
The more likely explanation at second place is that the four clumped in the low single digits are roughly tied. The margin of error for all these polls is between 3 and a little over 5%. Only a few of the results are outside that range (like Rubio at 17 and Kasich at 10 in one poll). At this point there really is no way to tell who will end up in second.
But do we even know who will end up in first? Every single poll taken since the month after Trump announced his campaign has shown him leading in New Hampshire. They started out as small leads of 5-7% but have settled around a 15-point lead since about September. That kind of unanimity in polling data is usually an easily predictable result.
Except even this week his lead is all over the place. As I highlighted in the RCP chart, Trump leads by 21 points at the most, or 11 points at the least, depending on which poll you use. But at least he leads. Or will he, when the votes are counted?
Trump also led pretty solidly in Iowa in the last week before the caucuses. Although that primary method is notoriously hard to predict by polling, there was a clear trend going in. Cruz had peaked, but faded a few weeks before the end. Trump had resumed his month-long lead, but Rubio was gaining steam in the last week.
As you can see from my l33t MS Paint skillz, the Trump and Rubio polling data predicted their trend lines but didn’t quite reach the end result. However, Cruz’s win came out of nowhere. If the polling had captured his campaign’s trajectory properly, his nosedive after the first week of January would have been just a minor fade instead.
The 8-point swing in Iowa from final polling average to end result was surprising, but not unprecedented. Bigger upsets have happened, and not just in Iowa. Three times in the 1980s, New Hampshire polling turned out to be 15+ points off from the final result, leading to two surprise winners.
— Bruce Mehlman (@bpmehlman) February 8, 2016
(Note: The main “average miss” point of this tweet is wrong, since it omits several elections with much smaller misses.)
Might Trump suffer the same fate in New Hampshire tonight? With one polls showing him only 11 points ahead, it’s not outside the realm of possibility.
But in this most farcical of election cycles, nothing is, and that’s the point. Anything could happen, and every candidate has at least one poll that they can use to spin their Surprise Finish!™ and Momentum Going Into South Carolina!™. Iowa really was a surprise, at least in magnitude. In New Hampshire tonight, nothing would be a surprise, since that’s all we know.
Here’s the race for 2nd-5th in NH, trendlines off.
When the polls tell you nothing, believe them. pic.twitter.com/8mX39LFOCY
— Ariel Edwards-Levy (@aedwardslevy) February 8, 2016