Comparing a pre-election poll in New Hampshire (which showed Obama winning), the exit polls in New Hampshire, as presented in the LA Times (which showed Clinton winning) and the actual results of the New Hampshire primary, I think the main story of Clinton’s surprising victory yesterday is apparent.
You’ve probably heard the story, but I think the punditocracy continues to blow the nuance (partly from a self-serving agenda; judge for yourself based on the data below).
In sum, I would argue:
- Obama basically hit his numbers (no "Bradley" effect)
- Clinton wildly exceeded her numbers – and did so on the basis of a surge of women to her candidacy (46% support in NH vs. 30% in Iowa)
- The ~7,000 votes Edwards lost between the last pre-election poll and the actual result were probably mostly women, and mostly defectors to Clinton – maybe even providing her margin of victory - and many probably shifted because of his crass attempt to capitalize on "the tear."
(If Clinton goes on to win the nomination, maybe someday Obama and Edwards will be able to laugh about this over a beer. Well, maybe not.)
Please follow on the other side of the flip.
Now the narrative that goes with this shift in voters can be told in two ways.
In one version, the key event was Clinton choking up and revealing desperately needed evidence of her humanity, thereby drawing people to her campaign.
In the other version, it wasn’t so much "the tear" as the outrageously misogynistic coverage of "the tear" that made a lot of women angry and brought them around to her candidacy.
I strongly believe it was the latter. This also fits with the hypothesis that John Edwards probably gave Clinton the margin of victory when he argued that "the tear" showed Clinton was too big a sissy to be President. I was personally appalled, and will show below that Edwards did much worse among women in New Hampshire than in Iowa, overall.
Now, my reference pre-election poll is the one accessible through the TPM Election Central.
The "Current Sensitive" estimate shows:
Obama 39
Clinton 30
Edwards 19
Other 12
The final results of the Primary (per the NY Times) were:
Obama 37
Clinton 39
Edwards 17
Other 7
Now let’s start by observing what happened with voters who were supporting other candidates prior to the event. To recall last week’s news, the ingoing "other" vote in Iowa polling was about 15%, and the "other" vote at the end was about 3%. So about 12% got reallocated to second choice candidates through the caucus, and it seemed to split about 6% to Edwards, 4% to Obama, and 2% to Clinton.
Edwards was the clear "second-choice candidate" in Iowa. He sure wasn’t in New Hampshire.
In New Hampshire, there were about 13% supporting "other" candidates (non-big-3) in the pre-election poll, and there were about 7% in the "other" column in the actual primary count. Clinton appeared to essentially pick up all of the 6% difference. The rest of her 9% jump (the remaining 3%) came from Edwards and Obama, more of it from Edwards if you get inside the rounding.
Overall, Clinton did 16 points better among women in New Hampshire than she did among women in Iowa -- 46% vs. 30% (based on the Iowa entrance poll results in the NY Times).
That’s a lot. Its equivalent to about ~26,000 more votes in New Hampshire than she would have gotten if Iowa percentages applied.
Clinton apparently did about 6 points or ~6,000 votes better among men, as well. (Note: proportionally fewer men voted – the male/female split was 43/57, the same as Iowa – so a "point" of male vote is worth less than a "point" of female vote. Note also that women did not "turn out" for Clinton in exceptional numbers -- the ratio was also 43/57 in Iowa; it was just that many more of those women who did turn out voted for Clinton).
Now Obama also did better among men in New Hampshire, 5 points better than in Iowa. The 11 point improvement for Clinton and Obama came at the expense of Edwards (who had about 5 points less among men than in his Iowa entrance poll result), and "other" candidates who had 6 points less male vote than going into Iowa. I would infer from this that voters forced away from second-tier candidates probably fall in the Clinton and Obama camps in about equal numbers.
In the New Hampshire exit poll, Edwards was 8 points lower among female voters than he was in Iowa. To be sure, he didn’t "lose" all these points in the last 48 hours – he never had most of them in New Hampshire – but the ~2 points of the overall vote he clearly did lose (and the polling was pretty consistent ~19 vs. ~17 in the actual) were probably about 70% or more women, based on comparison with Iowa.
Which brings us back to the 26,000 vote advantage among women Clinton had in New Hampshire versus her Iowa finish, based on pulling a an estimated 46% share in NH vs. the 30% share she pulled in Iowa (I believe 30% was was probably close to what she had among women before "the tear" in New Hampshire).
You see, 26,000 votes is also almost exactly how much better Clinton did in the actual New Hampshire Primary than she would have been projected to do based on the pre-election New Hampshire poll. It accounts for the 9 point surge
Someone with detailed tracking poll data from prior to the election could say whether Clinton’s advantage among women was the exact same people as those comprising her surge overall. Its probably not quite this simple (she may well have had more female support before the last 48 hours). But the overall math and the comparison with Iowa make a compelling case. And there is a compelling narrative that would appear to explain a last-minute shift among women to Clinton.
Some have argued that our leading media lights are idiots because they "got it wrong" in New Hampshire.
I would argue, instead, that the media through its atrocious misogynistic coverage may have actually (once again) profoundly influenced the overall outcome of the Presidential election (for better or worse).