« April 2008 | Main | June 2008 »

A Puerto Rico Lookahead

May 30, 2008 4:53 PM

If Hillary Clinton does well in Puerto Rico on Sunday, it shouldn’t be a surprise: She’s been strong among Hispanic Democrats all year, a group that lifted her to victory notably in the California and Texas primaries; and likewise among Catholics. Put the two together – Hispanic Catholic voters – and you’ve got a powerful Clinton advantage.

There’s plenty of overlap between the two groups. In exit polls conducted in this year’s Democratic primaries to date, 63 percent of Hispanics have identified themselves as Catholics, far higher than the proportion of Catholics in other groups. Nearly all Puerto Ricans are Hispanics, and 85 percent are reported to be Catholics. The population also is lower-income and lower-education than the mainland’s, another two good groups for Clinton.

Among all Hispanics voting in Democratic primaries to date, Clinton’s beaten Barack Obama by 61-35 percent, compared with her 55-38 percent margin among whites (and his 82-15 percent among African-Americans). Among Catholics, Clinton’s also beaten Obama by 61-35 percent. And among Catholic Hispanics, by 69-29 percent, a 40-point margin.

Non-Catholic Hispanics are another matter; Clinton’s won them as well, but by a much narrower 51-46 percent. And there are differences by combined racial and ethnic self-identification. Among voters identifying themselves as black Hispanics (5 percent of all Hispanic voters), Obama’s beat Clinton by 69-30 percent. Among white Hispanics (12 percent of all Hispanics), Clinton’s held a 13-point advantage, 54-41 percent. And among those who’ve identified themselves as Hispanics racially and ethnically alike, Clinton’s lead has been largest – a 2-1 margin, 66-33 percent.

When Clinton won Texas and California, she did so with support from 66 and 67 percent of Hispanic voters, respectively. She won 68 percent of Hispanics in New Jersey and 73 percent in her home state, New York. Obama split Hispanics with Clinton in his home state, Illinois, while winning the state overall by more than 30 percentage points.

A new Field Poll in California shows Hispanics there are much more apt to say they’d support Clinton than Obama in a November matchup against John McCain. But they’re essentially no more likely to support McCain if Obama’s the candidate; rather, to be undecided or currently to favor someone else. If Hispanics in California aren’t thronging to Obama, neither are they ruling him out for November. The question in Puerto Rico, though, is whether they prefer him against Clinton. So far, relatively few have.

May 30, 2008 | Permalink | User Comments (20) | TrackBack (0)

What the Heck Happened

May 30, 2008 3:38 PM

The approaching end of the primary season prompts a terrific question: What the heck happened?

You’ll hear answers from a lot of sources. But a beautiful thing about the internet is that you can also dig out some of your own conclusions – or fact-check what you’re hearing from others – by going directly to the source. With that in mind we’ve posted full exit poll results from each state contest in which they were conducted. Click here to get there.

We’ve been dissecting these polls in detail all year; they’re an absolutely invaluable source of intelligence on the election outcomes. Our fundamental understanding of the contours of the Democratic and Republican nominating contests alike relies on them.

There’s some apparent misunderstanding of how these polls are done. Interviews are conducted in person at randomly selected polling places in each state; voters leaving the polling place are randomly selected to participate. Interviewers note the sex, race and estimated age of those who decline; these are used in a non-response adjustment. In states where it's estimated that about a quarter of the vote, more or less, will be by absentee ballot, telephone polls supplement the exit polls.

There can be differential non-response in exit polls; for example, in many of this year’s Democratic primaries Barack Obama's supporters have been disproportionately apt to participate. In addition to the sex, race and age adjustment, the exit poll is adjusted for non-response in two other ways: Preliminary results are partially weighted to an estimate derived from pre-election polls, and later results are weighted to the actual vote as it becomes available.

One use of the exit polls is to assist efforts by news organizations to project winners. Another use – the one I find most valuable – is to enable us to make an independent, valid and reliable assessment of who voted, how and why. You've seen our analyses. Have a look at the data.

May 30, 2008 | Permalink | User Comments (26) | TrackBack (0)

Gay Marriage: The California Questions

May 28, 2008 12:16 PM

In the wake of this month's state Supreme Court ruling, do most Californians support gay marriage?

It depends - and therein lies a cautionary tale in understanding poll results. A pair of polls in the state have had somewhat different results on the question, underscoring the fact that a single number rarely is sufficient to understand public attitudes fully – and that different approaches, compared and contrasted, can better inform our judgment.

Today’s headline, out of the latest Field Poll, is that 51 percent of Californians approve of allowing homosexuals to marry – up 7 points from when Field last asked in 2006, and over half for the first time.

So most Californians (barely) do support gay marriage. Right?

Sort of.

That result is in response to a question asking yea or nay on gay marriage. Field also asks another question, with three options, not a simple dichotomy – whether Californians prefer gay marriage, gay civil unions, or no legal recognition of a gay couple’s relationship. In this formulation support for gay marriage slips to 45 percent, with 32 percent preferring civil unions, 19 percent neither.

Here we have a difference from a Los Angeles Times poll a week ago. In that one, with a very similar question, fewer supported gay marriage, 35 percent (10 points fewer than in the Field Poll); civil unions drew 30 percent support (similar to Field); and neither got 29 percent (10 points more than Field).

There are other differences as well: In the L.A. Times poll just 41 percent approved of the court’s ruling; in the Field Poll it was 48 percent. And in the Times’ poll 51 percent said they’d vote to support a ban on same-sex marriage; in the Field Poll support was lower, 43 percent.

Sample differences can matter (the Times poll was among all adult Californians, the Field Poll among registered voters only, and both noted big differences among areas of the state and demographic groups). Timing can matter, too (the Field Poll was done May 17-26, an unusually long 10-day field period; the Times poll, May 20-21, a short one). So can the order of questions, and these are worth a look.

The Times poll’s first question on the subject was the one that asked about gay marriage vs. civil unions vs. neither; nothing ahead to influence answers. The Field Poll started differently, first asking its dichotomous “approve/disapprove of gay marriages” question. It would want a clean read on this, a long-term trend question. But having told Field that they approve or disapprove of gay marriages (without civil unions as an option) might have influenced some respondents' answers to subsequent questions - the three-parter including civil unions, as well as those on approval of the court’s ruling and views on an amendment banning gay marriage.

That's a guess (it'd take a split-sample test to prove it), and it doesn't mean one approach is intrinsically better. Both polls are high-quality, with clear, balanced questions. Instead they tell us that measurements on the subject can differ; a possible reason is that for some respondents attitudes on gay marriage are not firmly held, and thus more sensitive to options and order.

One thing that's clear from the Field poll is that support for gay marriages (in its dichotomous test) has grown sharply, from 28 percent in 1977 to today's 51 percent. Another is that on this, as on many issues, there’s more to public attitudes than a single number.

May 28, 2008 | Permalink | User Comments (56) | TrackBack (0)

Vote Count III

May 28, 2008 11:52 AM

A fundraising note from the Clinton campaign today declares her intention to win the "national primary vote." With all the votes in Oregon counted, that's a good enough excuse for us to update our total vote estimate in the Democratic primaries and caucuses. Our methodology for these computations is sketched here and detailed here. We’ll update after the remaining contests in Puerto Rico, South Dakota and Montana.

            With         Without      With FL,
          MI and FL     MI and FL    Without MI
Obama    18,191,417    17,615,203    18,191,417
Clinton  18,222,170    17,022,875    17,893,861
         
         Cl +30,753   Ob +592,328   Ob +297,556

May 28, 2008 | Permalink | User Comments (13) | TrackBack (0)

Democratic Voter Groups: A Second Look

May 27, 2008 11:56 AM

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.

May 27, 2008 | Permalink | User Comments (143) | TrackBack (0)

McCain: Health and Age

May 23, 2008 11:10 AM

The release today of John McCain’s health records raises again the question of whether his age can hurt him in the 2008 campaign. The best answer: You bet.

This doesn’t mean it will hurt. Voter preferences are most likely to be determined by voters’ partisan affiliation and the candidates’ positions on issues. But – as we’ve seen in the primaries – the candidates’ personal attributes do matter. And McCain’s age, like Barack Obama’s relative inexperience, is an undeniable negative.

An AP report on McCain’s medical records today notes that in our ABC/Post poll in April, 70 percent of Americans said his age made no difference to them. This is true, but more to the point is the fact that 26 percent - one in four - said it made them less enthusiastic about his candidacy, while just 3 percent said it made them more enthusiastic.

Election politics is often a game of margins; when one in four expresses concern, that is well beyond the ignorable. So is the fact that concern about McCain’s age is far greater than the number who say they're less enthusiastic about Barack Obama because of his race or Hillary Clinton because of her sex. Indeed those are both net positives – twice as many people are drawn to those attributes than put off by them.

There is partisanship in views of McCain’s age; just 13 percent of Republicans see it as a concern, compared with 40 percent of Democrats. But it’s 24 percent among independents, the quintessential swing voters in national elections. And it’s 27 percent among seniors, who may be said to have a unique perspective on what 71 is like.

Perhaps the best news for McCain is that the “less enthusiastic” number in our poll ebbed slightly, from 31 percent in January to 26 percent in April.

But it’s still there. Our more recent poll, this month, used a different approach, asking not about McCain in particular, but about a president first taking office at age 72. Thirty-nine percent said they were uncomfortable with the idea, including 15 percent "entirely" uncomfortable. That again was more than twice the level of discomfort with either an African-American or a woman president.

In a Gallup poll in March, relatively few people raised McCain's age in an open-ended question asking their compunctions about him. But that was simply because other concerns were more salient – e.g., his closeness to President Bush and his support for the Iraq war. It doesn’t mean age is a non-issue.

We’ve been here before; as a 73-year-old candidate Ronald Reagan effectively defused questions about his age, and not just by his well-timed zinger in an October 1984 debate against Walter Mondale, then 56. “I will not make age an issue of this campaign,” Reagan said. “I am not going to exploit for political purposes my opponent's youth and inexperience.”

Reagan showed that a candidate can effectively address concerns about his age, not just with a snarky line but best by displaying vigor and acuity on the campaign trail. McCain – who, if elected, would be the oldest person sworn into office as a first-term president – has the opportunity, in the months ahead, to do the same.

May 23, 2008 | Permalink | User Comments (10) | TrackBack (0)

Updating the Vote Count

May 21, 2008 11:23 AM

(Check the bottom of this item for a 5 p.m. update.)

I blogged last week on the intricacies of counting the vote in the Democratic nominating contest, and proposed a better estimate – a count in which Barack Obama led, even with Michigan and Florida included. Hillary Clinton apparently didn’t get the memo.

Clinton’s continued to proclaim herself ahead in the popular vote; apparently she thinks it helps in her effort to win over superdelegates. (The recent direction of superdelegate preferences suggests otherwise, but that’s another blog.) So we've updated with the Kentucky and Oregon results to see if the picture's changed.

Answer: Yes, in one scenario. Given her net of 154,706 votes yesterday (in the count so far - it's not done in Oregon) our estimate now has Clinton ahead, by just over 52,000 votes out of more than 36 million cast, if we include both Florida (where neither candidate campaigned) and Michigan (where Obama wasn't on the ballot). Exclude Florida and Michigan, or just exclude Michigan, or give Obama a chunk of the Michigan uncommitted vote, and he's ahead.

Here are our estimates, with 88 percent of the estimated vote in Oregon reported:

            With        Without      With FL,          
          MI and FL    MI and FL    Without MI
Obama    18,148,061   17,571,847    18,148,061
Clinton  18,200,357   17,001,062    17,872,048

         Cl +52,296  Ob +570,785   Ob +276,013

As noted, the Oregon count isn't completed; we'll update when that happens. But our decision desk director, Dan Merkle, says it doesn't look like Obama will make up the more than 52,000 voters there that he'd need to surpass Clinton in the "all states" category.

You can see how we arrive at these calculations in the original blog. To recap briefly, the official counts you’ve seen elsewhere are based on delegate counts, not voter turnout, in five caucus states where voter preferences weren't tallied. We’ve adjusted as follows:

-In Iowa and Nevada we are using turnout reported by the state parties, allocated by vote preference in the entrance polls.

-In Maine we are using turnout reported by the state party, allocated by initial delegate proportions.

-In Washington we are using the beauty contest primary results.

-In Texas we are using both the primary results, and caucus turnout reported by the state party (given as “just under a million,” calculated at 900,000) allocated by initial delegate proportions.

Note, the estimate including Michigan gives Obama zero, since he wasn't on the ballot. There could be an argument for giving him all or some of the uncommitted vote there, since it included, for instance, 68 percent of African Americans, 50 percent of postgraduates and 50 percent in $100,000+ households, all among his core support groups. ("Uncommittted" beat Clinton in those groups, and a few others.) But we're not going there; this calculation sticks with our estimate of actual votes for actual candidates. Who knew they'd be this tricky to count.

5 p.m. update:

The AP today updated its vote totals in Idaho, Indiana and North Carolina, as well as in Oregon, where the count is now an estimated 96 percent complete. These have inched things in Obama’s favor:

            With        Without      With FL,
          MI and FL    MI and FL    Without MI
Obama    18,176,329   17,600,115   18,176,329
Clinton  18,214,506   17,015,211   17,886,197
         
         Cl +38,177  Ob +584,904  Ob +290,132

In the all-states count, Clinton’s lead, if you want to call it that, is a smidge over one-tenth of a percentage point. Obama’s is 1.7 percentage points without Michigan and Florida, eight-tenths of a point with Florida but not Michigan.

A Clinton staffer wrote in to ask how we can include the Texas caucuses, since that double-counts primary voters who also went to the caucuses. We discussed this in Friday's item. But it doesn’t make a big difference; take out our estimate for the Texas caucuses and Clinton gains 108,000 votes – still slightly ahead with Michigan and Florida, still a little less slightly behind without.

And as our political director, David Chalian, keeps saying - it's supposed to be about delegates.

May 21, 2008 | Permalink | User Comments (70) | TrackBack (0)

Counting the Vote

May 16, 2008 11:53 AM

In an interview with Charlie Gibson this week, Hillary Clinton contended that she's ahead in the popular vote – a critical claim in her last-ditch attempt to win over super delegates. The problem: It's arguably not so.

"Arguably" because there are myriad complications in trying to count votes in this nominating contest – and not just because of the disputed Michigan and Florida primaries. The other problem is that in a handful of Democratic caucuses votes simply weren't toted up. The "vote" totals being reported there aren’t votes at all, but initial delegate counts. Switch to an estimate of actual, human-being voters, and the story changes.

Here’s the deal: ABC and other news organizations are basing their totals on counts supplied by The Associated Press. Those numbers have Obama ahead (by 579,502 votes) if you leave out Michigan, where he wasn't on the ballot, and Florida, where neither candidate campaigned, in both cases to respect the national party in a dispute over those states' primary dates. Clinton's claim is that if you add in the votes from those two states, she's ahead – and she is, by a grand total of 43,579 votes out of more than 33 million cast.

"In fact," Clinton told Gibson, "I'm slightly ahead in the national vote right now."

Hold it.

Above and beyond the Michigan/Florida issue are the challenges in counting votes in the Iowa, Nevada, Maine, Washington and Texas caucuses. There the AP counted initial delegates, not votes, simply because votes weren't tabulated. ("Initial" delegates because final delegate selection is a more drawn-out process. Don’t ask.) It makes sense if what you’re really after is a delegate count. But if the vote count is what you care about – as Clinton clearly does – well, it doesn't.

Take Iowa. The AP count there gives Obama 940 votes, Clinton 737. That seems bizarre in a state where the state Democratic Party reports that 236,000 caucus-goers turned out. Caucuses are party-run affairs; they make their own rules, and vote-counting wasn't on the agenda. AP had no choice. Its "vote" count tallies initial delegates, because that's all it had to tally.

But there is a way to estimate actual voters – an estimate to be sure, but an entirely plausible one. In Iowa we can multiply total caucus-goers by the candidate preferences measured in the entrance poll – 34 percent for Obama, 27 percent for Clinton. That produces a vote estimate of 80,240 votes for Obama, 63,720 for Clinton – an Obama margin of 16,520 votes, rather than the 196-"vote" margin in the delegate-based count.

Doing the same in Nevada helps Clinton, giving her an 8,229-vote margin, rather than her 582-delegate margin in the AP delegate count.

In Maine we don't have an entrance poll, but we do have delegate percentages – 60 percent of initial delegates went to Obama, 40 percent to Clinton. Applying those shares to the state party's count of 44,866 caucus-goers produces an 8,773-vote Obama margin, compared to his 683-delegate margin.

Washington's a tougher nut. The AP count based on initial delegates is 21,629 for Obama, 9,992 for Clinton – a 68-31 percent Obama margin. But we can’t use that margin to produce a vote estimate, because unlike in Iowa, Nevada and Maine, the state party in Washington didn't produce an overall turnout figure for its caucuses.

However there was another event – a "beauty contest" primary in Washington, held in addition to its caucuses. The primary did not elect delegates, so it's not included in most tallies. But it does represent people who got up on their hind legs and voted – 354,112 for Obama, 315,744 for Clinton – an Obama margin of 38,368 votes. This seems more than fair to Clinton, since Obama won delegates by a much wider margin – but with no total caucus-goer count, it's the only vote-based data we've got.

Texas is the big kahuna, with its own complications. There was both a primary and a caucus there, and both awarded delegates. The initial delegate count for the Texas caucuses has 23,918 for Obama and 18,620 for Clinton. This is complicated by the fact that only 41 percent of the caucus precincts were included in the AP count, but again it's what we've got. That's 56-44 percent for Obama.

How many people attended Texas caucuses? The state party tells us it was “a little under a million.” That's an awfully round number, and some anecdotal reporting suggests the Texas caucuses weren't, shall we say, supremely well-organized. Let’s call it 900,000. That produces a margin for Obama of 108,000 votes.

We can debate whether it's fair to include both the Texas primary and caucus results, since that double-counts people who participated in both. But people who voted in both Texas events were playing by the rules, and including them seems at least as fair as including Michigan, where Obama voluntarily stayed off the ballot, thus netting exactly zero votes to Clinton's 328,309. (Pushing it a bit, one could argue to give Obama all or some of the "undeclared" vote in Michigan, 40 percent or 238,168 voters, including disproportionate numbers of supporters that elsewhere have been strong for him – including young voters, African-Americans, independents and better-educated whites.)

We're leaving aside one other state, Nebraska – it had a caucus for which we do have a vote count. It also had a beauty-contest primary. But since the primary didn’t award delegates, and the caucus votes were counted, it doesn’t quite fit the mold. If we did count its beauty contest, though, it'd produce another 2,665 votes for Obama – not enough to change our basic conclusion.

And that conclusion? Using these estimates of actual voters in the Iowa, Nevada, Maine, Washington and Texas caucuses, rather than the initial delegate counts, we get a net total Democratic vote to date of 17,607,152 for Obama and 17,504,742 for Clinton, an Obama lead of 102,410 votes – even with Michigan and Florida included.

The national vote count, of course, has nothing to do with winning the Democratic nomination under party rules - that's done by delegate counts. Clinton nonetheless has found her claim of an advantage in total vote a useful talking point. The problem: It doesn't quite add up.

(With thanks to Peyton Craighill, Pat Moynihan, Scott Clement and Dick Sheffield for help with the math.)

May 16, 2008 | Permalink | User Comments (29) | TrackBack (0)

Obama and Working-Class Whites

May 13, 2008 12:42 PM

The anticipated outcome of today’s contest in West Virginia is prompting a fresh review of Barack Obama’s difficulties winning support from working-class white voters in this year’s Democratic primaries. One question: The extent to which it does or doesn’t predict problems for Obama if he’s the party’s nominee in November.

The effect, which we started reporting back in February, has been clear in the primaries: Whites who don’t have a college degree have voted for Hillary Clinton by a 2-1 margin, 62-31 percent, while those who’ve gone through college have divided evenly, 48-47 percent.

It seems that the effect stems in part from the thematic positioning of the two, with Clinton’s more nuts-and-bolts approach better attuned to the working class, Obama’s inspirational pitch for a new politics resounding better among more upscale Democrats. Obama’s been helped by the fact that better-educated voters are disproportionately likely to turn out – especially in primaries.

But primaries only tell us so much about general elections. In our latest ABC/Post poll, testing each of the Democrats against John McCain, there’s a shortfall among less-educated whites for both: McCain leads Obama by 12 points in this group, Clinton by 8.

Obama, with his upscale appeal, does better among better-educated whites: McCain’s just +3 vs. Obama, compared with McCain’s 12-point advantage against Clinton among college-educated whites. That accounts for Obama’s better showing against McCain overall, 51-44 percent in our poll, vs. 49-46 percent in a Clinton-McCain matchup.

                         Obama  McCain    Clinton  McCain
    Whites, no college     40%    52        44%     52    
    Whites, college grads  47     50        42      54

There are other potential impacts of race and socioeconomic status. As we noted in our poll analysis yesterday, 17 percent of less-educated whites say they’re at least somewhat uncomfortable with the idea of an African-American president; among better-educated whites that declines to 4 percent. As noted, there’s a similar effect on comfort with a woman president – and McCain’s age is a far bigger negative than either of these. Each of the candidates has room for some consciousness-raising on these concerns.

It’s also worth noting that the latte-vs.-lunch bucket effect has not been entirely consistent in all primaries this year. Obama won less-educated whites in the Vermont and Wisconsin primaries, was +2 in Utah and came within 4 points in his home state of Illinois (although in each he again did better with upscale whites).

It’s fair for the Obama camp to point out that he doesn’t do significantly worse against McCain among working-class whites than Clinton does, and that he does better with their upscale counterparts. And Obama’s numbers are nothing like John Kerry’s and Al Gore’s; they lost working-class whites to George W. Bush by 24 points and 17 points, respectively.

But working-class whites nonetheless are a group with which Obama might well like to improve. If he loses today’s primary, and next week’s in Kentucky, they’ll be the first place to look. And less-educated voters account for a greater share of the turnout in general elections than in primaries. While Obama could win a general election without them – just as he leads McCain today – it’s also true that the last Democrat to capture the White House, Bill Clinton, ran evenly among working-class whites as he did so.

May 13, 2008 | Permalink | User Comments (125) | TrackBack (0)

Follow the Lines

May 09, 2008 2:44 PM

This morning our Political Unit broke news when Barack Obama for the first time inched into the lead in its tally of Democratic super delegates. That inspired the attached charts – showing Obama and Hillary Clinton’s accumulation of pledged delegates, super delegates, total delegates and total votes this year.

We’re fond of numbers here in the Polling Unit, but sometimes pictures sure help tell the story. Click here to see our slides on how the Democratic primary campaign has unfolded so far.

(Note, we've time-stamped the charts that include superdelegates; they're a moving target that our Political Unit continues to track. And we've run popular vote three ways - without Florida or Michigan; with Florida but not Michigan, where Obama wasn't on the ballot; and with both. Take your pick.)

May 9, 2008 | Permalink | User Comments (20) | TrackBack (0)

Battling Data: What Gives?

May 05, 2008 12:37 PM

There were at least a few crossed eyes today over conflicting data and analysis in the latest New York Times/CBS and USA Today/Gallup polls. We share your pain.

Briefly: Times/CBS has Barack Obama +12 vs. Hillary Clinton, with a headline saying Obama “survives furor” over the Rev. Jeremiah Wright. USAT/Gallup has Clinton +7, saying the flap over Wright “pulls Obama down.” Adding to the mix is Gallup’s daily poll, which has Obama +4.

These polls also differ in their general election match-ups: Times/CBS has Obama +11 and Clinton +12 vs. John McCain, while USAT/Gallup has them basically tied. Gallup daily has Clinton-McCain tied, McCain +5 vs. Obama.

Before we get into what gives, we’ll use this as an opportunity to repeat our long-standing advice to de-emphasize the horse race in pre-election polls. It is lowest-common-denominator reporting. And in poll-to-poll comparisons it’s the single most unstable measure we see. (Just a few weeks ago Newsweek had Obama +19 and Gallup daily had him +3 on the same day. Aagh.)

As to possible causes: Times/CBS ask Democrats (and Dem-leaning independents) whom they’d like to see the party nominate; USAT/Gallup asks whom “you would be most likely to support.” Since most Democrats who wanted to vote already have done so, these can be understood differently (potentially confusing past vote with vs. current support), possibly contributing to different results. (In any case it’s maybe not the best question to emphasize, since all those who already have voted don’t actually get a do-over.)

Also, Times/CBS asks the Democratic race first, then asks favorability ratings, then asks general election matchups. USAT/Gallup does it opposite: First the general election, then the favorables, then the Democratic preference. That can create differences (and asking favorability directly before vote preference is something we try to avoid).

USAT/Gallup gave its Obama-Clinton result among all Democrats and Democratic-leaning independents; Times/CBS, among “Democratic primary voters.” (It also reported this race among registered leaned Dems, Obama +8.) On the general election trial heats, USAT/Gallup reported results among likely voters, Times/CBS, among registereds. But it’s hard to blame the differences on this; we haven’t seen a lot of variation in gen pop vs. reg. vs. LV results. (Sample size can be a factor as well; USAT/Gallup interviewed 1,019 adults, including 516 leaned Democrats; Times/CBS, a smaller sample of 671 adults, including 283 “Democratic primary voters.” Both called Thursday through Saturday.)

In reporting such polls it's perhaps a natural reflex to take the horse-race result and search for explanations. USAT/Gallup had Obama doing less well, so it seems to have reached out to the Wright issue for support. Times/CBS did not have this change, so it went the other way, though the Times story is laden with caveats.

USAT notes that a third of likely voters say they’re less likely to support Obama because of Wright. That’s like the 30 percent in our own April poll, on the first go-around of the Wright story, who said Obama hadn’t done enough to distance himself from Wright. At least in our data, though, it didn’t appear to hurt Obama either vs. Clinton or McCain; critics on Wright fit the non-Obama profile.

None of this resolves the matter. At best it’s a reminder that all polls – even good-quality ones – are done differently, and don’t always get the same results or engender the same analysis. And that horse-race results, in the midst of a close and unsettled campaign, may be particularly vulnerable to these kinds of influences. Said it before, saying it again: Easing back on the horse race, and looking instead at the underlying dynamics, is always a better way to go.

May 5, 2008 | Permalink | User Comments (14) | TrackBack (0)