The results in New
Hampshire last night were stunning. Everyone expected
Barack Obama to win, even the Hillary Clinton campaign. The RCP average had
Obama up 8.3%. The RCP average does what it says, it averages out the polls.
Every poll taken within the last 3 days had Obama up by 5 or more points with
some even in double digits. The Clinton
camp was just hoping that it would be respectable.
Well, they got more than respectable, they won. While no
means a blowout for Clinton,
it was a solid 3 point win. This leads me, an aspiring pollster to wonder, what
the heck happened with the polls. I am not alone in this pursuit, but let me
take a stab at what happened.
I think the biggest thing that happened is that the polls
themselves changed the election. Obama was expected to win by 5-12 points, and
everyone knew that. Obama’s support is based off of young voters and
Independants. The young voters heard this news, and, unlike the Iowa young voters,
figured they could stay home and Obama would win anyway.
A bigger factor is that a lot of people love John McCain and
Barack Obama in NH. These people were going to vote for Obama, but when
everyone was saying the Democratic race was over, the Republican race was a
dogfight between Romney and McCain. These voters cast ballots for McCain
instead of Obama, and it helped McCain and Clinton, and hurt Romney and Obama.
Another factor is the weather and the old people (esp.
women) , which are related. The old people were Clinton supports, many of which were not
going to be able to make it out to the polls. They heard that Obama was going
to clean house and it was a very nice day and they were able to drive out to
the polls in large numbers for Clinton.
The pollsters clearly got this one wrong. I would suggest
that we figure out why this happened, and I think that we need to get back to
more using the data raw and less weighting.