Voter-group analysis in the Democratic contest has been flying thick and fast lately. Among the arguments: Barack Obama has a problem with white voters. And/or with Jewish voters. And/or with supporters of Hillary Clinton.
Each can use a second look.
Newsweek stressed the racial issue this weekend, saying Obama “is facing lingering problems winning the support of white voters,” and that his race “may well explain his difficulty” because he does less well among whites who express “racial resentment.”
Maybe, but there’s other evidence worth considering, including this: Obama’s currently doing no worse among whites – a 12-point deficit to John McCain in the Newsweek data – than Al Gore did; he lost them by an identical 12 points in 2004, yet won the popular vote. John Kerry lost whites, moreover, by 17 points; Mike Dukakis, by 19 points.
Each of these three Democrats lost their elections, and Bill Clinton, who won his, did better among whites, losing them by 3 points in 1996 and a scant point in 1992. Obama surely wants to do better among whites; after all they account for three-quarters of voters in presidential elections. But the fact that he’s currently even with Gore and outpacing Kerry and Dukakis among whites would seem to militate against racism as the prime agent.
My vote’s for the socioeconomic effect I’ve covered in previous blogs. Working-class whites are not a good group for Obama; he does much better among better-educated whites. That would seem to cut more to the politics of the man, not the color of his skin.
Then there are Jews; The New York Times headlined a piece last week, “As Obama Heads to Florida, Many of its Jews have Doubts.” The piece reported anecdotally that elderly Jews have particular concerns about Obama; said Jews were “important to his general election hopes,” especially in New York, California, New Jersey and Florida; and reported that “in recent presidential elections, Jews have drifted somewhat to the right.”
The Florida exit poll, however, found that Obama did no worse among Jews voting in the uncontested Florida primary – 26 percent support – than he did among other white voters, 23 percent. The Florida sample’s not big enough to look at seniors only, but across all primaries this year, Obama in fact has done slightly better with Jews over age 65 (35 percent support) than among non-Jewish white seniors (29 percent). (See here for Jan Crawford Greenberg’s blog on the subject last Thursday.)
As far as their importance, Jews are hardly a large group, even in the states listed. In the 2004 general election Jews accounted for 8 percent of voters in New York, 7 percent in New Jersey, 5 percent in Florida and just 4 percent in California. And Jews (second perhaps only to African-Americans) are among the most reliably Democratic voting groups. They voted more heavily Democratic in the last four previous elections than in the previous four – by more than a 3-1 margin in 2004. A rightward shift is tough to see; just 13 percent of Jews in the 2004 exit poll identified themselves as conservatives.
Finally there’s the ongoing brouhaha about polls in which Clinton supporters say they wouldn’t vote for Obama in November, and vice versa. I’ve argued against putting too much of a stake in these findings, simply because of the timing: Asking Democrats their November vote in the midst of their nominating contest is like asking a married couple in the middle of a knock-down, drag-out fight what they’ll be doing for Valentine’s Day. They need a little time to calm down and try to make up.
A run through our data from past elections reinforces the point: In primary polls since 1988, supporters of a losing nominee routinely have been loath to say they’d vote for the winner in November.
In an ABC/Post poll in January 1988, among Democrats and Democratic-leaning independents who supported someone other than Gary Hart (the front-runner at the time), just 54 percent said they’d support Hart against George H.W. Bush in November. Among Republicans who preferred someone other than Bush for the nomination, a less-than-monolithic 79 percent said they’d support him as the nominee in November.
In February 1992, among Democrats who did not support Bill Clinton, just 63 percent said they’d vote for him in the fall; 31 percent said they’d cross over to vote for Bush. In January 1996, among Republicans who did not support Bob Dole for the nomination, just 66 percent said they’d support him in November. In March 2000, among Democrats who supported Bill Bradley for the nomination, just 64 percent said they’d vote for Al Gore in November; and on the Republican side, among John McCain’s supporters that year, just 73 percent said they’d support George W. Bush as the nominee. Finally, in December 2003, among Democrats who did not support Howard Dean for president, just 67 percent said they’d support Dean in the fall.
In our last national poll, among Democrats who favor Clinton for the nomination, just 64 percent said they’d vote for Obama against John McCain in November. That looks a lot like most of the numbers above.
The history adds some context. It tells us the phenomenon we’re seeing now is not new. We don’t have data that let us clearly parse out how supporters of a losing candidate in the primaries voted in November. We do know that partisans by and large stick with their party, and independents make the difference.
Given the stickiness of partisanship, what may well matter more than crossover voting is voting in the first place: People disaffected with their party’s nominee probably are likeliest just not to vote, rather than to vote for the other side. That suggests the endgame matters. If the winning and losing candidates hold hands and make nice, that sends a message. If it ends ugly, that sends a different one.
There’s one clue in corresponding Republican data. In an ABC/Post poll in January, with the Republican contest still underway, 78 percent of leaned Republicans said they’d vote for McCain against Obama in November. This month, with the nomination in McCain’s hands, that had inched up to 84 percent. Time, it seems, heals at least some wounds.