Why the Polls were Wrong in New Hampshire

Five days after absolutely nailing the Iowa Democratic Caucuses in the Evans-Novak Political Report, Bob Novak and I wrongly predicted that Obama would win New Hampshire in a blowout. We weren’t the only ones who were wrong — all of the published polls showed the same thing: CBS and Rasmussen had Obama winning by 7, CNN had him up 9, and Zogby’s polls showed him thrashing Hillary by 13 points.

What happened? Why were the polls — all on conducted January 5, 6, and 7 — so horribly wrong? Three likely causes come to mind — and they all matter going forward.

First, we’re talking about polls — they are a lot less reliable than many people think. Second, at play is the fact that Hillary is the candidate Democrats are embarrassed to like. Finally — she cried.

Polling is an Art, not a Science

It’s a strange forgetfulness and a misplaced faith that causes us to be so surprised when polls are wrong. To disabuse yourself of this myth, go onto Amazon and order Mobocracy, Matt Robinson’s enjoyable evisceration of the polling industry and the media that lives on it.

Some of the history in Mobocracy makes Hillary’s win less of a surprise. In 1988, Bob Dole had an 8 point lead in the final New Hampshire Gallup poll. Dole lost by 9 points to George Bush the elder. In 2000, CBS News had McCain up by 4 points, but he actually won by 19 points. On the Democratic side, Gallup had Al Gore beating Bill Bradley by 12 points. The final margin was 4 points.

Pollsters don’t just call a random, representative sample of the electorate and tally up their responses. They make many assumptions: how many independents will vote, who counts as a “likely voter,” how many seniors will vote, and so on. If the pollster’s sample has 15% senior citizens, but he estimates that seniors will comprise 20% of the electorate, he will weight his results. There’s tons of guessing and tweaking. A couple of tweaks in the wrong direction, and you’ve got the wrong guy winning.

The Douglas Wilder Effect — Kind of

One popular explanation for Obama’s underperformance on Election Day is the Douglas Wilder Effect (also known as the Bradley Effect or the Dinkins Effect). The idea is that people tell pollsters they will vote for the black guy, but then they don’t do it.

The media love this explanation because it allows them to posit widespread racism hidden in the souls of white Americans. They imagine someone who wants to vote for Obama, but once the curtain is pulled shut on the voting booth, just can’t bring himself to put a minority in the Oval Office.

While the secret-racism interpretation is probably a complete fantasy, a variation of the Wilder Effect is probably at play in the Hillary-vs.-Obama contest. Sure, it makes white people feel good to support a black guy for President. (Did you see how pleased John Kerry was with himself on Thursday? Did you see how uneasy Obama was receiving this potential kiss of death?) But race is not the central factor here. It’s the Ice Queen factor at play.

To understand the dynamic, contrast New Hampshire to Iowa, where Hillary underperformed the polls. In the Iowa Democratic caucuses, there are no secret ballots. If you vote for Obama, all your neighbors see that. That’s a plus for Obama. If you vote for Clinton, on the other hand, you must, before your friends and neighbors, stand in the Hillary corner, under the Hillary signs. They’ll probably even smack a Hillary sticker on your shirt.

For many Democrats, put off by her personality and blind ambition, admitting to Hillary support is a tough thing to do. For these same Democrats, though, she’s the safer bet–a third Clinton term.

So the Wilder Effect in New Hampshire was not so much a matter of people pretending to back Obama who didn’t. Instead, people wouldn’t admit to backing Hillary, but did.

Cry Me a Victory

And yes, the crying helped.

In travelling around the country for the past two election cycles, it’s become crystal clear that regular voters — who don’t have time to study the details of everyone’s health care policy and conduct an independent economic analysis — vote largely based on likability.

In Dubuque four years ago, I met a pro-choice, pro-gay-marriage, not particularly anti-government man who didn’t see a difference between Kerry’s foreign policy and Bush’s. He voted for Bush basically because of the Botox and Teresa. In 2000, on a similar score, I’m convinced Bush would not have won had he not kissed Oprah.

So, when Hillary cried, it made her seem less cold and more human — diminishing some of her biggest negatives. But also, as a civilized society with decent remnants of chivalry, when a woman cried, we sympathized.

Going Forward

These three explanations shouldn’t be considered merely explanations for a past event. They are important to keep in mind going forward. As you’re following the primary contest and then the general election, recall these three points:

1) Basically, don’t trust polls. They are a good starting point, but that’s all.

2) From here on out, votes will be cast behind closed curtains. This helps Hillary–the candidate Democrats hate to support–immensely. In fact, it makes her the favorite.

3) The New Hampshire waterworks are a foretaste of the feminist two-step we’re going to see through November, and possibly through 2012. Hillary is a strong woman, and treating her differently — or expecting anything different from her — based on her sex is rank chauvinism. But if you’re mean to her — well, you’re just a monster.

Happy election season.