Another Few Rounds for the Democrats

Clinton's surprising victory shows the nomination is very much up for grabs.


The Democratic race remains very much alive, thanks to Hillary Clinton's surprising 39-to-37 percent victory in New Hampshire. While relatively downscale Iowa favored the upscale voters' candidate, Barack Obama, relatively upscale New Hampshire favored the downscale voters' candidate, Hillary Clinton. How to explain this? One way is to note that the addition of a substantially disproportionate number of young and upscale voters to the relatively small universe of Iowa caucusgoers created an electorate considerably more upscale and younger than the larger body of regular Democratic primary voters. The New Hampshire primary, in contrast, is a high turnout contest by American standards. It has always had a large upscale component. The key to leverage here is to turn out marginal downscale voters, which the Clinton campaign seems to have done. But before we get to that, let's consider a couple of other points.

The polls were wrong. The summary of immediate preprimary polls showed Obama ahead, with an average of 38 percent for him and 30 percent for Clinton. In other words, they got the Obama percentage just right but underestimated the Clinton percentage. In post-Iowa polls, only the Suffolk/WHDH poll had Clinton's portion near the 39 percent she finally got, and they had it at 34 percent in their final track.

Could all, or almost all, the polls be wrong? I kind of doubt it. I'm inclined to think that we're seeing what Mickey Kaus calls the "Bradley effect." It's named for former Los Angeles Mayor Tom Bradley, who in his 1982 race for governor ran ahead of Republican George Deukmejian in the polls but lost to him on Election Day (actually, when the absentees were all counted). It appears that a lot more voters told pollsters that they were voting for Bradley, who is black, than actually voted for him on the secret ballot. There was a similar discrepancy in the 1989 race for governor of Virginia, which the black candidate, Democrat Douglas Wilder, won by a much more narrow margin than the pre-election polls indicated.

I have been inclined to believe that the Bradley effect was a thing of the past and have noted (as Kaus does) that it doesn't seem to have been operating in the 2006 Tennessee Senate race, in which the black candidate, Democratic Rep. Harold Ford, ran a little behind on Election Day as he had in most pre-election polls. Nor does it seem to have been operating in the 2006 Massachusetts governor race, which the black candidate, Democrat Deval Patrick, won. I thought it would not be operating in New Hampshire this year. But the data push me toward the opposite conclusion. The folks at Fox News headquarters in New York, my companions on election night, told me that they found discrepancies between the exit poll results and the actual tabulated vote in exit poll precincts in the Concord/West segment of the state, a liberal area that has been leaning Vermontward in its politics and was the strongest area for Howard Dean in 2004. This is a bit puzzling, since the exit poll, like the actual vote, is a secret ballot: The voter gets a ballot that he marks up and then deposits in a box. But there may have been a nontrivial number of voters who wanted the exit pollster to think they'd voted for Obama when they'd actually voted for Clinton.

In any case, it's clear that the upscale towns didn't deliver big margins for Obama. The Manchester area was Hillary Clinton's strongest region in the state, and she carried next-door upscale Bedford (by just a few votes) and most of the other surrounding towns. The initial exit poll precincts showed Clinton carrying the region denoted "South" (these definitions were made back in the 1970s and reflect demographic conditions then, not now), about which I was skeptical. But as the votes came in, it became clear that she indeed was carrying most of the towns in these areas.

But what really won for Clinton was organization. Patrick Ruffini stresses the role of legendary organizer Michael Whouley, who was quietly wheeled into New Hampshire. I would add that the Clinton state director, native New Hampshireman Nick Clemons, struck me as a first-rate political operative, whom Ruffini quotes. Manchester cast about 15,000 votes in 2000 and about 16,000 in 2004 (you can click on Hillsborough County to get the Manchester figures); this time there were 15,000 votes cast in the first nine (of 12) wards that reported. Clinton carried Manchester 45 to 31 percent and Nashua, seemingly less favorable ground, 45 to 33 percent. She racked up big margins in the former mill towns of Dover, Rochester, and Somersworth on the Maine border, where turnout was almost double that in 2000. Organization in these cities was responsible for Clinton's entire statewide margin and more.

As I was following the incoming city and town results, I found a substantial congruity between the 2000 and 2008 contests, between the race between Al Gore and Bill Bradley on the one hand and Hillary Clinton and Barack Obama on the other. Gore and Clinton had similar demographic and geographic bases; ditto Bradley and Obama, with the latter attracting more young and upscale voters and the former the opposite. (These maps illustrate the point.) Interestingly, there is some congruence here between the historical partisan differences in New Hampshire (and New England generally) between Roman Catholics and Protestants. The 2008 exit poll shows the following on religion:

  % of voters Clinton Obama
Protestants 23 40 40
Catholics 36 44 28
No religion 22 29 46

Historically, from circa 1900 up through the middle-1960s anyway, Catholics in New England voted heavily Democratic and Protestants heavily Republican. There was a very high correlation between religion (that is, Catholic or Protestant) and partisan preference—it looked something like a continuation of the 17th-century religious wars.

In general elections in New Hampshire and New England overall, that is a thing of the past. Heavily Catholic Manchester, for example, typically votes Republican in general elections; Nashua, which I believe is more Protestant, is more Democratic. Places with lots of secular residents, like yuppie-restorer-infested Portsmouth, are heavily Democratic. But in the 2000 and 2008 Democratic primaries, we still see these divisions. Catholics (a category highly correlated with "ancestral Democrats") are heavily for Clinton and Gore. Protestants (a category that includes many "ancestral Republicans") are evenly split between Clinton/Gore and Obama/Bradley. Those with no religion (including presumably many from outside the state and most presumably with fewer communal ties than they grew up with) are heavily for Obama/Bradley over Clinton/Gore. So we see echoes of past general election divisions in these Democratic primaries.

New Hampshire has left both parties with nominations that are undetermined and open for grabs. The Democratic race is obviously a two-candidate contest between Barack Obama and Hillary Clinton that either candidate could win. (John Edwards's speech Tuesday night sounded more like a trial lawyer's pitch to the jury than a candidate's reasons he should be elected to the presidency.) The next state to vote is Michigan, on January 15, but the national Democratic Party has ruled it illegal; Obama and Edwards removed their names from the ballot, while Clinton left hers on. Will there be a significant vote for Clinton or a write-in effort for Obama? Who knows?

The next officially sanctioned Democratic contest will be the Nevada caucuses January 19, where the endorsement of the giant Culinary Workers Union is considered crucial. (The Las Vegas hospitality and gaming industries are the only fast-growing businesses in the country whose workers are unionized.) But no one really knows: Nevada has not had a consequential caucus before and is hard to organize; Las Vegas and Clark County, with more than 70 percent of the state's population and the nation's fastest-growing metro area, has got to be hard to organize.

Then Democrats compete in South Carolina on January 26, a week after the Republican contest there. About half the voters here will be black. Polls up through November showed Clinton ahead of Obama among blacks (and whites as well); polls in December showed blacks equally divided and the race close; January polls show Obama leading among blacks and leading statewide. All of which suggests that Clinton will in effect concede South Carolina to Obama and concentrate on the January 29 Florida primary, where blacks are likely to be fewer than one quarter of Democratic primary voters. And then there are the 22 states that have contests February 5.

You may ask, what's my prediction? You may ask that as often as you want. But you won't get any answer. This nomination is very much up for grabs.