Cross-posted at Election Inspection
With the news of John Edwards affair all over the place, former Clinton communications director Howard Wolfson has made the bold assertion that, if John Edwards' affair had been exposed before the Iowa caucus, that Hillary Clinton would've ultimately won Iowa (and the nomination):
"I believe we would have won Iowa, and Clinton today would therefore have been the nominee," former Clinton Communications Director Howard Wolfson told ABCNews.com.
Nate Silver over at Fivethirtyeight.com takes Wolfson to task over his incredulous claims:
Iowa actually didn't turn out to be that close, with Obama defeating Edwards by 7.9 points and Hillary Clinton by 8.1 points. For Clinton to have beaten Obama, she would have needed (as Wolfson correctly points out) about two-thirds of those Edwards voters.
The thing about Iowa, however, is that unlike virtually any other electoral contest, second choices matter, since Democratic caucus rules dictate that a voter may caucus for her second-choice candidate if her first choice does not achieve the 15 percent of the vote required for viability. As such, Iowa pollsters did a lot of work in trying to determine voters' second choices. And in virtually every survey, Clinton did rather poorly as a second choice: an average of several surveys in December showed that she was the second choice of about 20 percent of voters, as compared with 25 percent for Obama and Edwards (an even later version I have sitting on my hard drive showed the second-choice breakdown as Edwards 30, Obama 28.5, Clinton 23.5)
So the odds are that, if John Edwards had dropped out on the morning before the Iowa caucus, Obama would have won by more points rather than fewer.
I'd actually like to elaborate on what Nate was talking about and disprove this particular claim made by Wolfson:
"Our voters and Edwards' voters were the same people," Wolfson said the Clinton polls showed. "They were older, pro-union. Not all, but maybe two-thirds of them would have been for us and we would have barely beaten Obama."
Unlike most caucus states, there were entrance polls taken to get a good look at the demographics and to find out whether or not Edwards voters' were mostly taken from Clinton's voting bloc. Now, one caveat to keep in mind is that thanks to the rules of the caucus, the final numbers won't sync with the calculations from the entrance polls (with, by calculations of the entrance poll data, accounts for about 7% of the vote going to Richardson and 8% going to all other minor candidates and uncommitted) even so, Edwards was really the greatest beneficiary of this which means that using Edwards numbers should also give us answers as to where many of these second-choice candidates would go.
Wolfson has made the claim that Senator Clinton and Senator Edwards were drawing from the same voting coalitions, so does this stand up to the test? Wolfson's argument has two specific points, that Edwards was garnering most of his support from the elderly and from those who are "pro-union" (which I'm going to assume means those who belong to a union). In addition to looking at these two groups, to get a clear picture of who intended to caucus for whom, I'm also going to take a look at the data on income and on the candidate qualities which are important to caucus goers.
First of all, on age Wolfson has a decent point to make; Edwards did get most of his support from older caucus goers, but the distinction here is that a strong plurality of Edwards' initial support came from those who were between the ages of 45 and 64 (accounting for slightly under 50% of his vote total). What does this mean? Well, if we look at the exit polls from Wisconsin (which is probably the only state which is demographically close to Iowa which has exit polls) we can see something interesting, Obama actually won those who were 45 years and older (which is the same group which gave Edwards roughly 70% of his voting coalition in Iowa). By the way, just so you don't think I'm arbitrarily using Wisconsin because the numbers are favorable to Obama, the two states are very similar demographically). Even if we take into account that Hillary Clinton didn't put very much effort into Wisconsin compared to the effort she put into Iowa (which is a reasonable objection) there is still not much reason to believe that older voters in Wisconsin would behave radically different from older caucus-goers in Iowa.
But wait a minute, what about Wolfson's claim about Edwards' fabled strength among the unions? Would it surprise you to learn that a staggering 75% of Edwards' support came from those who did not come from Union households? According to the entrance poll data, Edwards got 24% of the union vote (making up 22% of caucus goers) while he got 23% from those who were not part of a union (making up 78% of all caucus-goers). Even if everyone from a union household who voted for John Edwards switched over to Hillary Clinton (which is an absurd claim) Clinton still would have to take at least 55% of Edwards' non-union caucus goers to beat out Obama. In fact, giving Clinton 75% of Edwards union caucus-goers and 60% of Edwards non-union caucus-goers still leaves Obama ahead by a point and a half (this is slightly under two-thirds of Edwards initial support, by the way).
Now that we've taken care of those two points, let's look at things which obviously disprove Wolfson's claims. John Edwards base of support in the Iowa caucus was not among poorer voters, in fact, Edwards only won 19% of voters making under $50K per year while he won 25% of those making over $50K per year (63% of Edwards voters made over $50K per year, and 23% of Edwards voters made over $100K per year). Now, Clinton has tended to do better among poor whites in almost every state than other groups, but Obama has, outside of Appalachia and the deep south, done well among more affluent white voters. In other words, Edwards voting coalition seems to come from a group which would normally break for Obama and not Clinton.
On top of that, looking at candidate qualities, it becomes increasingly obvious that Edwards' voters aren't necessarily the ones who would've chosen Clinton as a second choice. Edwards did best among voters whose top qualities were "Electability" and "Cares about me" (neither is all that telling as both Obama and Clinton have had states where one has done better on these qualities than their performance in the state overall) but 45% of Edwards' coalition wanted someone who could "bring change" (compared to only 7% of Edwards supporters who cited "Experience). Since we do know that the "Change" question is one that Obama has always won by a wide margin (the only exceptions being West Virginia and Arkansas). This is quite telling since voters who seemed to want "Change" a lot more than "Experience" would be a lot more likely to break for Obama rather than Clinton.
In fact, looking at some data from New Hampshire, it seems to me that more likely than not, Edwards being in the campaign was favorable to Clinton (and, that she might not have won New Hampshire if Edwards wasn't taking votes from Obama). The exit polls for New Hampshire was the only one where favorability ratings were gathered for each candidate, so I'm going to use that to get a little bit of insight into what Edwards voters really thought of Obama and Clinton (also to see what Obama and Clinton voters thought of Edwards, to see just which voters had more in common with each other). Thanks to the magic of excel, I was able to determine that among Edwards supporters, 65% of them have a favorable opinion of Hillary Clinton (with 12% having a very favorable opinion) while 38% of Edwards supporters have an unfavorable of Clinton (with 15% having a very unfavorable opinion) (by the way, because of some rounding issues, the overall favorable/unfavorable numbers do not add up to 100%, for those who want to point that out). On the other hand, 89% of Edwards voters have a favorable opinion of Obama (with 34% having a very favorable opinion) with 14% of Edwards voters having an unfavorable opinion of Obama (and only 3% having a very unfavorable opinion). Looking at how Clinton supporters saw Edwards, we see that 67% of Clinton voters had a favorable opinion of John Edwards (15% having a very favorable opinion) while 29% viewed Edwards as unfavorable (there wasn't a big enough sample to get the very unfavorable ratings). Now looking at Obama supporters, 75% of them viewed Edwards favorably (with 25% viewing him very favorably) and only 17% seeing Edwards as unfavorable. Not only were Edwards votes substantially more likely to view Obama as being very favorable (and a lot more likely to see Clinton as very unfavorable) Obama supporters were more likely to have a favorable opinion of Edwards than Clinton supporters were. Edwards's supporters were a lot more likely to break for Obama in New Hampshire than Clinton (and by a decent margin) while in Iowa, the evidence also seems to suggest that Obama would've actually had a stronger win in Iowa without Edwards than with.
UPDATE: The point of this diary is not to say that Obama would've definitely would've won New Hampshire if Edwards wasn't in the race (because even with my data, it's difficult to know what would've happened if things were different) the point is to show why this notion of Clinton being hurt by Edwards is as ridiculous as all the other idiotic memes which have been pushed (Obama can't win Latinos in the general, etc.)
UPDATE II: Looks like the Washington Post has even more data to show why Howard Wolfson is a liar (h/t brainwave