So, What Happened in NH?
The results from last night's New Hampshire primary have the political world grasping for explanations, and, in many cases stretching for them.
I've been doing some number crunching, looking at exit polls for NH this cycle and last, the returns, the Iowa entrance poll and Pollingreport's standardized rolling average of polls taken before the NH primary.
Here are the main take-aways:
- Pro-Hillary women surged
- Obama's support remained relatively stable
- John Edwards lost support among women
- Clinton's ground game turned out a large number of Democrats, especially in areas that were her strongest
- Most of Biden's supporters went to Clinton
Here are some theories that I find little support for in the data:
- There was a "Bradley Effect," in which people told pollsters they'd support a black man and then voted a different way in the privacy of the voting booth
- Hillary's emotional moment caused people to "break" for Clinton late
I'm agnostic on this one:
- Obama seemed like such a sure thing that independents decided to go for McCain on the GOP ticket
Women and Hillary's "softer side"
The exit polls showed that 57 percent of those voting on the Dem ballot were women. There was a huge turnout overall -- far higher than projected -- and among those who were not expected to come out and vote, Hillary's support was very high.
Many of those who turned out appear to have been women who already tended to support Hillary but were inspired to get to the polls in larger than expected numbers. There's little evidence of a late break for Clinton -- Obama beat her among voters who decided their votes within three days of the primary. That suggests that Clinton's display of emotion and the perception that she was being ganged up on was not the determining factor for these late-deciding voters.
Clinton's support among women doesn't appear to have come out of Obama's hide -- he won the support of 35 percent of women in Iowa and 34 percent of those in New Hampshire. It does appear that some women who may have been inclined to support Edwards went for Clinton; Edwards had the support of 23 percent of the women caucusing in Iowa, but his support in that demo dropped to 15% in NH -- a pretty significant hit.
Speculation: When Clinton had a tearful moment, Obama responded gracefully, saying that the process is a long grind and all the candidates were exhausted. Edwards, on the other hand, took the opportunity to fire a shot at Hillary, suggesting that America needed a tough Commander-in-Chief. If there was a general sense among women that Edwards and Obama were piling on at the debate and in the days leading up to the vote, it may be that because of Edwards' reaction, he bore the brunt of their anger. Caveat: Edwards overall support was within the pre-election polls' margin of error.
Obama didn't lose much ground
The pre-election polls didn't get it all wrong; they nailed everyone's support except for Hillary's. Obama's average support in the pre-primary polls was 36.7 percent, and he scored 36 percent of the vote (rounded to the nearest whole number), which is basically on the money. Obama lost ground among voters aged 25-29, and among ideological liberals compared to Iowa. There's a little bit of trickiness here, because they didn't track the same age groups in Iowa and NH. In Iowa, Obama won the 17-29 year-olds by a big 57-14 margin over Edwards. Hillary got 11. In NH, the exit poll split that demographic into two: Obama did just as well, slightly better even, among the 18-24 group, while Clinton doubled her support and Edwards slipped behind. But in the 25-29 group, Hillary won by 2 points. In Iowa, Edwards got the mom and dad voters -- aged 45-64, while in NH Clinton won every group over 40 by wide margins.
Why? How should I know? They're different states. The weather was especially nice, so maybe some of those younger people played Frisbee instead of going to the polls. Speculation: these micro-groups breaking differently than in Iowa suggest a very strong ground game by Team Clinton -- they turned out their supporters. More on that in a second.
The fact that Obama had a similar percentage of the women's vote in NH and Iowa, which also had the same gender split overall, really shows that although he might have lost some ground, there was no major shift away from him among women.
Biden gave Hillary 1-2 percent
All of the factors above don't quite add up to Hillary's turnaround from the polls. She needs another couple of points, and I imagine they came from Biden, who the pre-poll average still had at 2.6 percent and only got 600 votes. Unfortunately, they didn't release the numbers on how people who liked Biden voted, but given that A) Biden did better than twice as well among those who said "experience" was important than he did among others in Iowa, and B) that's Hillary's central pitch, and C) given that they're similar in their policy prescriptions, it makes sense that Biden's supporters would go to Clinton.
Let me turn to Ari Melber for some insight into the workings of the campaigns:
The Clintons shared another political asset in New Hampshire, though farther offstage. Michael Whouley Ã¢â‚¬Â¦ [a ground specialist] was dispatched to overhaul the mobilization program in the state. Clinton aides had debated whether to deploy him in Iowa, where he had helped engineer John Kerry's huge comeback in 2004, or task him with fortifying the famous "firewall." Some feared that his efforts would simply be wasted in New Hampshire if Clinton lost Iowa, but the "Plan B" advocates won, and now they look pretty shrewd.Sounds right, given the kinds of movements we saw, including a very large turnout in the Western part of the state, where Clinton expected to do well. Also, the state Democratic establishment was largely in Clinton's corner, and it appeared to have paid off: Obama won Iowa among registered Dems, 32-30; in NH, Clinton beat Obama 43-32.
Over at MYDD, Senate2008guru offers a theory:
New Hampshire's independent voters ostensibly preferred Obama among the Democrats and McCain among the Republicans. These independents were waiting until the day of the primary election to decide which candidate to turn out for.
Given that the polls looked so much stronger for Obama's victory, many of these independents (who were polling in support of Obama) decided that, with Obama's New Hampshire victory looking so secure, they might as well use their vote to make sure McCain bested Romney. Only, so many of these independent voters thought the same thing that it shifted Obama's comfortable margin of victory to McCain.The strength of this theory is that it helps explain McCain's larger-than-expected margin-of-victory. I haven't crunched the GOP side in any serious way, so that helps balance out this analysis.
My quibble is this: the Secretary of State projected that 60% of independents would vote on the Dem ticket. The numbers overall were far greater than expected, but the final break-down, according to my back-of-the-napkin calculations, is that about 60.5 percent of independents did in fact vote on the Dem ticket. Indies made up 52 percent of Dem voters in 2008, compared to 48 percent in 2004. That tracks generally with the larger decline in partisan loyalty, but doesn't suggest a significant and late indie shift to the GOP.
Let's put those things together: it may have been a surge in independent voters overall that provided McCain his cushion -- almost certainly -- but the polls should have had samples that conformed more or less with the projections, so it's hard to see how that explains the polling being so far off on Hillary.
Little evidence to support "Bradley Effect"
Last night on MSNBC, Chuck Todd was almost obsessive in his desire to explain the pollsters' failure to nail the results on the "Bradley Effect." Named after former L.A. Mayor Tom Bradley, who lost a narrow race for the California governorship after leading by double-digits in the polls, the idea is that when an educated-sounding pollster with a "Midlantic" accent calls, people, subconsciously wanting to please the pollster, say they'll vote for a black candidate when in fact, once inside the confines of the voting booth, they don't.
One should never discount the role race plays in American politics, but I see little evidence to support the theory. If the Bradley Effect were in play, one would expect A) Obama to fall significantly, and B) his support to be divided among the white candidates. You'd certainly expect a boost to the prominent white guy, Edwards. But none of that happened -- Obama got just what the polls suggested he would overall, he won 36% of the white vote, according to the exit polls, and, instead of getting a bump, all of the other candidates also lost a small amount of ground to Hillary -- all except for Richardson, the only other non-white whose vote total came in just as it had been expected to.
How were the polls so wrong
They weren't, really. Hillary's been polling way up in NH for an entire year, and Obama led for four days after Iowa. Looking at the big picture, it's clear that the media got caught up in Obamamania, or else more analysts would have noted this fact and the predictions may have been tempered a bit.
And, again, they nailed everyone else's results, more or less, and simply missed some groups that turned out at greater-than-expected numbers for Hillary.
Journalist Matt Yglesias offers us a graph:
Polls are not only just a snapshot in time; because pollsters adjust their samples based on the results of past elections and changes in the populations, they also have some history factored into them. In 2004, 54 percent of NH primary voters were women. The polling average doesn't give you any sense of the individual polls' samples, but it's a good bet that they ended up undersampling women relative to the final vote. Given that Hillary beat Obama firmly among that group, 46-34, that must have played a significant roll in under-estimating Hillary's support.
Of course, I don't discount other factors, like NH's infamous desire to go against the conventional wisdom, or perhaps wanting to avoid anointing a candidate. But I can't determine those factors' influence based on the numbers, so I'll leave that speculation to others.
A few words more about the women's vote
After a big upset like this, it's inevitable that people will run around telling the world that 'This Proves What I've Been Saying All Along!' -- whatever that may be. There's lots of speculative stuff out there today.
So I really want to stress that there's no evidence that people in large numbers, especially women, were moved to change their choice of candidate after Hillary's much-publicized tearful moment, or because Obama and Edwards were mean to her in the debate.
The narrative is sexist and infantilizing -- it says that those little ladies are so emotional and lacking in reason that they'd vote for Hillary just because she shed a few tears. It's Ann Coulter territory. It's also wrong; again, Obama beat Clinton among those who decided on their candidate in the last three days -- meaning those whose decision took place in the time-frame of those two stories.
It's possible, likely even, that her emotional moment served to humanize her, which the campaign has been trying to do with little success for a year. And the women did come out for Hillary and decide the vote. But, again, they were women who, by and large, already supported Hillary and were motivated, likely by Hillary's show of emotion and the perception she was being ganged-up on by the men and the media, to get out to the polls. So the tears didn't change their minds about whom to support, but, rather, whether to vote.
That's a more nuanced and accurate view. I also think it's likely that it wasn't the tears, but the reaction of the media to the incident, that motivated a lot of these women. After days of incredibly sexist drivel in the mainstream press -- as Megan Garber details on AlterNet today -- a large number of women, especially older women who can identify with Hillary on a personal level, and who already viewed her favorably, said, 'screw this crap,' and headed to the polls in unexpected numbers to stick up for their sister.
There isn't an irrational thing in the world about that. In fact, as someone who is often driven half-mad by our political discourse, I find it perfectly rational.
One final factoid: among New Hampshire primary voters, Clinton won, handily, beating Obama by 12 points, on the question of who would make the best "Commander-in-Chief." I can't help but wonder if that high number didn't have something to do, specifically, with Edwards' crack that Hillary may lack the "strength" to be a good C-in-C.
Just a thought.