Saturday, February 9, 2008

Convention Math: Obama wins three more UADs, narrows superdelegate gap

Loyal readers are already familiar with my effort to count the forgotten supderdelegates - the 76 Unpledged Add-On Delegates (UADs) to the Democratic Nationa Convention. To recap: 720 superdelegates get to go to Denver by virtue of the party or governmental offices they hold. The forgotten 76 are handed out to the states in proportion to the number of DNC members in each state, and (s)elected by state committees, conventions, or delegations. In this incredibly tight contest, it's naive to believe that any candidate whose supporters comprise a majority of the body that selects the UADs will select anyone not (unofficially, but no less firmly) pledged to support them.

Counting UADs is hard, because like delegates from caucus states, we don't know who they'll be, only how they're likely to vote in Denver. Further complicating the picture is the diverse array of selection processes. So in this count, I'm only including delegates who we know, pretty much for sure, are going to back a given candidate. Though February 5, the count stood at:
Obama: 11
Clinton: 7

Tonight's results bring a further advantage for Obama, whose lead is based upon his dominance of caucus states, on show again this evening.

Clinton Obama Undetermined
WA: 0 2 0
NE: 0 1 0
LA: 0 0 0

Your new totals:
Obama: 14
Clinton: 7

We'll keep a running tally here every election night, until one of the more reputable bodies tracking delegates notices that they're excluding more than twenty delegates who, by any right, ought to be counted.

If you find this worthwhile, please click the 'recommend this' link, so that other readers can share it. More election and polling analysis is available on my blog (click my name). And, as always, I welcome your comments and corrections.

Thursday, February 7, 2008

How Gallup Got It Wrong

I've seen a lot of indignant comments about TPM's pro-Obama bias, because it didn't flog the Gallup results yesterday that showed Clinton up, 52-39. I blogged about that poll, explaining that to the extent we look to tracking polls as predictors of voter sentiment, a poll that showed a thirteen point gap in the three days ending Tuesday at the same time voters divided their votes evenly was clearly flawed.

I also asked the guru of polling, Mark Blumenthal, for an explanation. He put forward a bunch of theories, and followed up with an extensive discussion this morning. Mark is on a righteous crusade to get pollsters to abide by the professional code of conduct to which they ostensibly subscribe, and to disclose their data and not just a summary of their results. So in addition to putting forward his own theories, he pressed Gallup for answers.

Gallup apparently delayed their usual early-afternoon release to crunch the numbers. And just now, Gallup has spoken. Their explanation boils down to this. They call 1,000 adults every night. About 200 of them say they're unlikely to vote, so they screen them out. The remaining 800 describe themselves as at least a little likely to vote (or that they have voted) in either the Democratic or Republican races. In other words, their sample includes 80% of American adults. But even in this seminal year, only about 30% are actually voting. And therein lies the problem.

Gallup went back and ran the numbers for the five days between John Edwards withdrawal and Super Tuesday. Among those who characterized themselves as highly likely to vote, about 50% of the total sample (thus still too high), Obama drew 48% and Clinton 45% - a statistical tie. Now there may have been more than that at work here - Mark's other questions remain unanswered, as does my biggest (what's the gender and racial composition of the sample, and does it vary night by night?) But it goes a long way to explaining the discrepancy.

I'd argue that it provides a good reason for Gallup to use a tighter screen in the future. Gallup says "a broad sample of over 80% of American adults would not be expected to match the actual voting patterns" which to me, is a very good argument for not using such a sample. But it should also provide us all with reason for caution. National tracking polls, it's worth remembering, are useful indicators of trends. Gallup picked up, succesfully, that Clinton had stopped the Obama surge by about Sunday. (And both Gallup and Rasmussen show Obama getting a post-Feb 5 bump). But they are absolutely lousy indicators, at least until they change their voter models, of how the electorate will actually vote.

Obama Closes the Superdelegate Gap?

Ben Smith of the Politico has some posts up today with major convention implications. He's put up a spreadsheet that Bloomberg News obtained, apparently an internal Obama campaign worksheet that attempts to project where the race is headed. It's of prurient interest, but most of its scenarios are fairly conservative, and there's not much news in it.

The big deal, though, is in a few boxes at the bottom. Obama's campaign lists the current superdelegate tally as 159-209. That's 40-50 more than any public tally I've seen. And it gets more interesting - Ben queried the Obama camp, and discovered that they're now claiming the backing of 170 superdelegates.

A word of caution. They haven't substantiated that with lists of names - as of this morning, only 113 endorsements had been publicly announced. And its perfectly plausible that Hillary has her own internal tally which also shows her doing better than the major media organizations project. But given that this spreadsheet, on the whole, is fairly conservative, and it does the Obama camp no good to delude itself as to how many delegates are going for Hillary, I'd say its prima facie evidence that Obama has finally closed the superdelegate gap. If he's really trailing by just 40, and his Super Tuesday projections hold up, then the overall gap is now in the single digits. And that would be huge news.

(And, for readers just joining me, I discussed this morning why the gap is at least four delegates smaller than any current count.)

Convention Math: Adding up the Unpledged Add-On Delegates (UADs)

So by now, we're all familiar with the rudiments of the delegate selection process. We ordinary folk go to the polls or the precinct caucuses (or in Texas, both) and cast our votes. Those are tallied, and eventually used to determine all of the pledged delegates, of whom there are 3,253. Then there are the superdelegates, the party bigwigs who get their tickets to Detroit no matter what, of whom there are 796. And that's the end of the story.

But it turns out that not all superdelegates are the same. For 720 of them, the process works pretty much as you'd expect - they hold public office, a seat on the DNC, or have held an important position in the past. We know who they are, we're tracking their endorsements, and we're tallying their support.

But then there are the 76 Unpledged Add-on Delegates, the ones I'll call UADs for short. These aren't necessarily party bigwigs. In fact, we don't know who they are at all, because they haven't even been selected yet. But as a block with 76 votes, there as important as any midsized state. And if we look closely, we can already figure out how more than half of them are going to vote.

Here's how it works. Every state is eligible to receive one UAD for every four DNC members it possesses (except for DC, which is stuck with just 2). Since the party chair and vice-chair of every state serve on the DNC, and the rules stipulate that the total be rounded up to the nearest integer, that means every state gets at least one UAD. Some get more. Illinois gets three, for example, and California has five. And it's up to the state parties to decide how to select these delegates. In the past, they've typically been awarded as plums to political insiders. Since they're not distributed until late in the process, they've never had an impact on determining the nominee. But this year, that's going to change.

States have adopted a wide range of methods for selecting UADs. Those include a vote of the state party committee; a vote of the entire state convention; a vote of the entire delegation to the DNC; or, in some cases, a vote of one of those bodies that's a sham, because the state chair only presents the nominees he wants to serve. Because the methods vary widely, it's not always possible to figure out which way these delegates will lean. But in most cases, we can be fairly certain. Although these delegates are formally unpledged, I can guarantee that any candidate who controls a majority of the body which awards them will ensure that the slots are reserved for their own supporters. That's right - there's nothing proportional about it. For UADs, the system is winner take all. And that allows us to start to compile a tally.

Here's a quick rundown of states that have already voted:
Clinton Obama Undetermined
Iowa: 0 0 1
NH: 0 0 1
NV: 1 0 0
SC: 0 0 1
AL: 0 0 1
AK: 0 1 0
AR: 1 0 0
AZ: 0 0 1
CA: 5 0 0
CO: 0 1 0
CT: 0 0 1
DE: 0 1 0
GA: 0 0 2
ID: 0 1 0
IL: 0 3 0
KS: 0 1 0
MA: 0 0 2
MN: 0 2 0
MO: 0 0 2
ND: 0 0 1
NM: 0 0 1
NJ: 0 0 2
NY: 0 0 4
OK: 0 0 1
TN: 0 0 2
UT: 0 1 0

Total: 7 11 23

Those aren't enormous numbers, but I'll continue to track this, and I suspect the tallies will mount with time. For now, the result is a four delegate advantage for Obama that's not acknowledged in any tally of which I'm aware.

Now, a few words to forestall howls of outrage in the comments section. The most common method of selecting UADs in caucus states is a vote at the state convention; the most common method in primary states is a vote of the state committee. That confers certain advantages on a candidate who runs strongly in caucuses - in this case, Obama. He's already got the votes he needs to secure those UADs. Now it's likely that the state committees that meet in states that one candidate carried will choose delegates who will back that candidate - but we don't actually know that. A few examples. Hillary is almost certain to carry the four UADs from NY. But in Massachusetts, Obama still has the backing of a significant number of institutional players, even though he lost the primary. Similarly, Hillary has lots of endorsements in SC. So where the majority of people who will choose these delegates haven't publicly committed to supporting a particular candidate, I've left the delegates in the undecided column.

One other note: I'm sure I got things wrong! Please use the comments section to point out the error of my ways.

Wednesday, February 6, 2008

Clinton: Loans $5M, Staff Working for Free. Obama: On Pace for another $30M in February

That's the big news at this hour. A roundup of the headlines:

The Clintons have loaned Hillary's campaign $5 million.

Senior staffers, including campaign manager Patti Solis Doyle, are working for free to help the campaign save cash.

And Obama is apparently on pace to raise $30 million in February, on top of the $31.5 he raised in January. And you've gotta believe that to be a conservative estimate, given that his campaign is going to be held to hit, and hit if it fails to meet the target.

This is not the story that the Clinton campaign wants dominating the news cycle today.

So Much for Tracking Polls!

There's been a lot of ink spilled about daily tracking polls over the past few weeks, some of it (at least metaphorically speaking) by me. And today, I'm heartily sorry for it.

Let's cut to the chase. Both tracking polls purport to offer a snapshot of the preferences of likely Democratic voters. Here are the numbers from today's reports:
Rasmussen: Clinton 46, Obama 42
Gallup: Clinton 52, Obama 39

The problem is that we ran a controlled experiment yesterday, and discovered that voters are split evenly - really damn evenly. The popular vote aggregate totals for yesterday, apparently, were 50.2% to 49.8%. So Rasmussen skates by, at the very edge of its +/-4% margin of error. Conveniently, I might add, as their margin shrunk dramatically from 47-40% yesterday, but I'll assume they're being honest about it. But Gallup, the supposed gold standard for polling organizations, totally flubs. They're just incredibly far from telling us how voters actually feel.

Wait, I can almost hear you say. What if it's voters in the other 27 states who overwhelmingly back Clinton, and in the 23 who voted yesterday they're evenly split? Funny you should ask. Gallup broke out the Feb. 5 states a couple days ago, and found them, if anything, more supportive of Clinton than the rest.

So Rasmussen, as I've explained before, is erratic and error-prone, even if they barely skate by this time around. (See, for example, their results from CA). But until we get an explanation out of the Gallup folks, anything coming out of that tracking poll - be it an Obama surge or Clinton domination - should be fairly suspect.

Taking Stock: Tallying up Advantages for the Race Ahead

The Money: This race has already consumed record amounts of cash, and we're scarcely halfway through. Campaigning takes ever-increasing amounts of money, for television ads and polling, for paid staff and for volunteers, for office space and for lawn signs.
Obama: The Obama camp raised an astonishing $32 million in January, $25 million of that from mostly small-dollr donors via the internet. That enabled them to go on the air in all but two Feb 5 states - and also to begin airing ads in states with contests through Feb 12. Obama reportedly spent $12 million on TV in the past two weeks alone. His dependence on small-dollar internet fundraising now provides a critical advantage, as he can raise funds more quickly, and return to those who have already donated for another round. Look for a press release within 24 hours trumpeting the dollars flowing in after Super Tuesday.
Clinton: In one of the strangest twists to the primary season, Hillary Clinton is now facing some significant fundraising challenges. She outraised her rivals in 2007, and fundraising has always been one of her key strengths. She's done much better at persuading her donors to give the maximum allowable contribution, and many have gone a step further, writing checks for her general election fund. But the river of cash is slowing down. In January, she raised $13.5 million - impressive in any other year, but not enough to keep pace. She spent $9 million, to Obama's $12 million, on television in the last two weeks, and has yet to go on the air in the remaining states. So there's no doubt that she's hurting here. But at the same time, she's likely to raise enough money to remain competitive. It doesn't take tens of millions to win a race (although that rarely hurts), it takes a better message and a more compelling candidate. And Clinton should be able to get her hands on enough cash to stay competitive.

The Delegates: Anything I choose to write here is likely to be out of date by the time it gets posted. But indulge me in a little math. It takes 2,025 delegates to secure the nomination. Of those, 722 are superdelegates, not bound by the results of any primary or caucus. Just over half the pledged delegates have been awarded, and just under half the superdelegates have announced their commitments. This race remains wide open.
Clinton: Going into Super Tuesday, Clinton held an edge on the strength of her endorsements from superdelegates. By the only transparent count, she had 201 to Obama's 110. That tally will grow this morning - some (like Sen Barbara Boxer) were holding off, but promised to follow the will of voters in their states. It certainly appears that teh candidates split the available pledged delegates yesterday; any edge enjoyed by either candidate will be small. Hillary believes that she can capture most of the remaining superdelegates, keeping the remaining contests close enough, when she doesn't win outright, to prevent Obama from overcoming her current lead. She's hoping that party insiders will ultimately announce in favor of the candidate of experience, and is counting on members of the Democratic National Committee, who constitute the bulk of the current crop the unpledged superdelegates, to back her.
Obama: Going into Super Tuesday, Obama held an edge among the pledged delegates who had been awarded. It's not clear where things stand this morning, but by most counts, he probably retains that edge. Winning a majority of the delegates awarded by voters gives him a powerful moral claim, but moral claims don't win conventions. He has enjoyed increasing success in winning superdelegates, particularly elected officials from red states, but still lags significantly with DNC members and overall. Those are not promising trends. He's certainly trailing this morning in the overall count, and faces an uphill struggle building a large enough margin in the remaining states to overcome Hillary's edge in superdelegates. He's banking on three things: that many superdelegates will be reluctant to tip the convention away from the candidate backed by a majority of pledged delegates, that most elected officials would rather he topped the November ticket than Hillary, and that a good number of the 411 superdelegates who have yet to endorse have been holding off because they dislike Hillary, but are afraid to cross the Clintons until they know they can be defeated. Those are persuasive theories, but there's been scant evidence to back them thus far.

The Calendar: Suddenly, all of the states that resisted the rush to advance their primaries to Super Tuesday are looking pretty clever. They were awarded bonus delegates by the DNC, in proportion to how late their primaries will be held, and will now enjoy the sort of attention to their issues and constituencies that few states lost in the rush of Super Tuesday managed to garner.
Obama: It's fair to say that the junior senator from Illinois is looking forward to the rest of February. Four of the ten remaining contests are caucuses; Obama won six of seven caucuses yesterday (losing only American Samoa). They play to his strengths in enthusiasm and organization, and to his red-state edge. The heavily-black Democratic primary electorates in Louisiana and DC should deliver both of those contests by healthy margins. The remaining primaries shape up as more contentious battles: Maryland and Virginia are fairly even matches, Wisconsin may tip toward Obama, like the rest of the midwest, and Hillary will likely carry the Virgin Islands. Still, for the next month, Obama can look forward to a steady drumbeat of victories, and a mounting edge in the pledged delegate count. That sets the stage for the six states in March, when Texas and Ohio are the big contests. Obama needs to build and broaden support in February to keep those big states close, or to have a chance at winning them himself, although he'll likely sweep the four smaller contests. And April is, for the moment, too far to contemplate.
Clinton: Hillary is acutely conscious of the dangers of February. She already has her surrogates pressing for a large number of debates, to ensure that Obama's coming success won't hijack the campaign narrative. She had recaptured momentum in the two leading tracking polls even prior to yesterday's election, and she knows better than to believe polls that show her trailing Obama in some of the remaining states. She wants to remain in the public eye, return the focus to the two of them on stage, and keep the races too close for him to emerge with a significant advantage. March brings more favorable terrain. Actually, the slow pace delivers certain advantages for Team Clinton. After February 19, there are no contests until March 4, a span of two weeks. That will give Clinton a chance to kick her relatively ponderous fundraising operation into high gear; she relies on personal networks and events to bring in cash, and that takes longer than internet appeals. It also gives her a chance to replicate Obama's more substantial field operations. And she can be fairly confident that she'll retain her overall edge in delegates even after the remainder of the February contests. That gives her two weeks, without the distractions of Obama's wins, to remind voters why they have favored her for most of the past year.

The only guarantee about these elections is that they'll be close, and unpredictable. And of course, there's always the chance that a scandal or a gaffe could upset all of these prognostications. But to a bleary-eyed observer this Wednesday morning, this is how things look.

(If you've found this summary useful, please recommend it, so that others can enjoy it, too. And as always, I welcome your comments.)

Tuesday, February 5, 2008

How to Read the Early Exit Polls

Josh has just posted the second wave of the exit polls, and a handful of first wave results besides. Here are three quick points to take away:

1) SurveyUSA bet wrong: The optimistic projections for Clinton were based on SurveyUSA, which either boldly bucked the conventional wisdom or was off in a world of its own. Now, we know the answer. These polls look much more like the consensus averages.

2) Blacks voted for Obama in a big way: We don't even have the demographic breakdowns yet, but we can see that Obama's margins in Georgia and Alabama are much larger, and his loss in Tennessee much smaller, than almost any polls had predicted.

3) It's going to be a good delegate night for Obama: Earlier today, Howard Wolfson attempted to lower the bar for Hillary, suggesting that instead of winning a majority of the available delegates tonight, she should be credited with a win if she still holds the overall lead (that is, loses by fewer than 70-100 delegates). Now, we know why. If these numbers hold up, Obama should win slightly more delegates than Hillary will, when it all shakes out. It'll be close - that in itself is the Obama camp's definition of victory - but it'll probably tilt toward Obama - and that was yesterday's definition of defeat from Camp Clinton.

More on reading the tea leaves from my earlier post. And if you enjoyed this, please recommend it, so others can read it, too.

Five Things to Watch for Tonight

5) The Black Vote: Most polls going in to South Carolina and Florida dramatically underestimated the extent of black support for Obama. It's what TNR's Noam Schieber has christened the "Reverse-Bradley/Wilder Effect," and he seems to be on to something. SurveyUSA, for example, puts Obama's black support at 76% in AL, 71% in MA, 74% in MO, and only 57% in NY. In Georgia, Zogby spots Obama a 20 point lead, but projects just 68% of the black vote, ten points lower than he drew in SC. (But there's at least one state in which it's unlikely that polls are underprojecting: SurveyUSA puts Obama's black support in Illinois at a stunning 93%. If that figure holds up, he'll sweep all 24 delegates from Chicago's 3 majority-black districts.) This is a great early indicator, because demographic breakdowns are among the first exit poll results to leak - Marc Ambinder generally posts them in the early evening. Also, watch the results out of Georgia, where polls close at 7pm.
The Indicators: If Obama can top 70% of the black vote in NY, 80% elsewhere, or 90% in Illinois, he's going to do very well in the delegate tally. If Hillary can hold him under 60% in NY, 70% elsewhere, or 80% in IL, it'll be a major coup.

4) Gender Patterns: This is a complicated issue, to be sure. Hillary's core support comes from women - the older, whiter, and poorer they are, the more likely they are to support her. In five of the six Democratic primaries held to date, women have composed 57-59% of the voters. The exception is South Carolina, where 61% were women. In SC, as in other the Southern states, a large female turnout is a sign of a high black turnout, and may well be good news for Obama. But elsewhere, if women compose a greater-than-expected percentage of the electorate, that's probably good news for Hillary. (It's worth noting that, contra Linda Hirshman, there's no real evidence that Hillary has succeeded in drawing women to the polls in disproportionately large numbers so far this cycle. That's something she'd like to change today.)
The Indicators: Non-Southern states in which women compose 60% or more of the voters may be harbingers of a good night for Hillary.

3) Show Me the Garden State: Here's as stark a contrast as you'll find in the annals of polling.
Missouri: SurveyUSA, Clinton +11; Rasmussen, Clinton +9; Zogby, Obama +3
New Jersey: SurveyUSA, Clinton +11; Rasmussen, Clinton +6; Zogby, Clinton +5
Someone, obviously, is using a model that's ill-suited to this year's actual electorate. If you were to array all the results of these three polling operations along the spectrum of published polls, you'd find SurveyUSA consistently among the most favorable to Hillary, Rasmussen tilting slightly toward Obama, and Zogby among the most favorable to Obama. I don't mean to suggest that there's bias at work, just that in this groundbreaking cycle, it's been awfully tough for pollsters to guess who's actually going to show up to vote. But here's why this argument matters. All three of these polls have recent numbers out of the Golden State, far and away the day's biggest prize:
California: SurveyUSA, Clinton +10; Rasmussen, Obama +1; Zogby, Obama +13.
Wow. That's an absolutely insane 23-point spread. Someone is very, very wrong. But who?
The Indicators: If Hillary captures the Garden State by double digits, or wins Missouri by a decent margin, then No, Obama Can't. But if the race is close in Jersey, and Obama can win Missouri, then Yes, He Can.

2) Independent Voters: Among the Super Tuesday states, most are using modified or open primaries, meaning that voters unregistered with any party have a chance to vote. In general, a high turnout among political independents is very good news for Obama. Hillary continues to outpoll him, in almost every state, among registered Democrats. So if few independents show up, or if they decide (where they're able) to vote in the Republican contest for McCain or Ron Paul instead of in the Democratic race, that's good news for her. In California, for example, Suffolk projects independents at an eye-popping 25% of the total, the highly-respected Field Poll projects independent turnout at 13%, while SurveyUSA puts it at just 10%. In Massachusetts, Suffolk projects 16%, but SurveyUSA finds 25%.
The Indictator: If independent turnout in the Bay State tops 20% of voters, Obama's going to have a good day. If it tops 15% in California, it's going to turn into a great day. But if bad weather or apathy keep these folks at home, Clinton will pad her margins substantially.

1) Delegates: Here's a guarantee. Neither candidate will gain an insurmountable lead in delegates by the end of the night. Going into Tuesday, Clinton leads in the delegate tally, 249-172, including those superdelegates who have made a public declaration of their support. A candidate needs the votes of 2,025 delegates to secure the nomination. There are only 1,681 delegates available today, so we're not going to see this settled, one way or the other. But both the campaigns have tried to set artificially-low expectations for their delegate hauls, so that they can proclaim victory. David Plouffe, on behalf of Obama, writes that if his man's take is within 100 delegates of Clinton's, that's a win. The Clinton camp, meanwhile, says that gaining more delegates than Obama ought to count as a win. So we can read that as an acknowledgement, on both sides, that the delegates are likely to be split remarkably evenly tonight.
The Indicator: If Obama wins a majority of the delegates who are awarded (at least 841) then even Hillary's people should acknowledge his win. If Hillary wins over 100 more than Obama (at least 891), then Axelrod and Plouffe should stand an applaud her. And if, as is likely, the totals lie between those extremes, I'll leave the call at your discretion.

So that's what I'd watch for as news starts to trickle in today. If you missed it before, you can catch my take on the Five Myths of Super Tuesday, early voting in California, or yesterday's polls elsewhere on the site. Enjoy! And as always, I welcome your feedback in the comments section.

Monday, February 4, 2008

More on the Myth of Early Voting

Josh posted this morning on the subject of early voting in California, and though I discussed it here yesterday, it’s probably worth tackling the subject at full length.

The conventional wisdom is that early voting may have given Hillary an insurmountable lead. Let’s get the big news out of the way up-front: there’s no good reason to believe that those who vote absentee in California will break much differently than the electorate at large. And if they do, it’s unlikely to hurt Obama.

There are, to the best of my knowledge, just three polls in the last month that have looked closely at this issue. Unsurprisingly, they were all sponsored by in-state organizations that are familiar with the vagaries of California voting. Two were done for the LA Times/CNN/Politico consortium, and one by the highly-respected, nonpartisan Field Poll. We’ll take them one at a time.

First, the Field Poll. In polling done between January 25 and February 1, it found Hillary leading 36-34% over Obama. There was, in fact, some variation by method of voting, but it wasn’t what the pundits expected: “Precinct voters narrowly favor Clinton 40% to 36%, while those who already have or intend to vote by mail are evenly divided (Obama 32% vs. Clinton 31%).” Those are, however, modest effects. In every sample, the electorate appears fairly evenly divided.

The LA Times consortium conducted two polls. The first, from January 11-13, found Clinton leading 47-31 among likely voters, 49-30 among early voters, and 48-32 among election day voters. The differences among those results are not statistically significant. The second, taken from Jan 23-27, had Clinton’s advantage virtually unchanged at 49-32%. But it found evidence of a big split: early voters backed her 53-30, but her lead among precinct voters was down to just eight points. In their write-up, the pollsters theorized that this was an artifact of Obama’s late surge – that precinct voters were backing him in larger numbers, but early voters had cast their ballots when Hillary still had a substantial lead, and so couldn’t change their minds.

As best I can tell, it’s on this second poll that pundits who predict a big early-voting advantage for Hillary are hanging their hats. But here are three reasons they’re likely wrong:

(1) It’s just one poll: There were three polls taken in January, and this is the only one to show such a split. The sub-sample sizes for these questions allow for a very large margin of error.

(2) Early voters hadn’t actually voted yet: The pollsters asked whether voters had or intended to vote by mail – they didn’t break those two categories apart. In California, the ballots have to arrive at the county election offices by mail, or be hand-delivered to the local precinct, before the end of voting on Tuesday. The Field Poll (Jan 25-Feb-1) found that scarcely six percent of Democratic voters had cast their ballots when they called them, even though 30-40% of the votes are likely to be absentee or vote-by-mail. I’d encourage the LA Times pollsters to visit a post office on April 15 – it might permanently disabuse them of the notion that Americans ever mail things in before the deadline.

(3) It’s all about the demographics: In the unlikely event that the LA Times is actually documenting a real effect, it probably has much more to do with the demographics of early voters than it does with their being “locked in” to their choices. A Field Poll this spring found that those who vote by mail (VBM) are older, whiter, more female, more partisan, and more Republican than the general electorate. By contrast, Hispanics, young voters, the LA and San Diego areas, men, and independents are all overrepresented at the precinct level. So if Obama is surging among independents, Hispanics, and young voters, but Hillary is holding on to her share of voters over the age of 65 (30% of VBM, 34% of Democratic VBM) and women (56%/58%), you’d expect to see a bit of a split. But here’s the crucial point – it’s not because these voters mailed their ballots too soon, it’s because different kinds of people vote different ways.

We actually have a precedent on early voting this cycle. In Florida, we heard much the same blather before the Democratic election there. When all was said in done, exit polls showed that early voters had backed Hillary 50-31, and precinct voters went for her 50-33. Those results are well within the margin of error, meaning there was no statistically significant difference. What a shock!

Monday Morning Polling Wrap-Up

With less than 24 hours until the polls open, media outlets have dumped a new batch of data in our laps this morning. Let's try to make sense of the numbers.

First up is a bunch of tracking polls from the Reuters/C-Span/Zogby consortium, that look at four key states. Zogby updates his three-day tracking sample by dropping January 31 and adding in results from yesterday. So what we're looking for here, most of all, is momentum. Any gains are likely to be about 1/3 of the difference in support between the day that was dropped and the day that got added. Without further ado:

California: Obama 46 (+1), Clinton 40 (-1)
Missouri: Obama 47 (+4), Clinton 42 (-2)
New Jersey: Obama 43 (+1), Clinton 43 (+0)
Georgia: Obama 48 (+0), Clinton 31 (+3)

That's some pretty astonishing stuff. Of course, it's Zogby, so it's worth taking it with an entire salt cellar.Unlike most pollsters, for example, who are loath to disclose their individual single-day small-sample results, Zogby often trumpets them, the better to capture media attention. Here, according to Zogby, is what he found yesterday:

California: Obama 49, Clinton 32
Missouri: Obama 49, Clinton 39

You may now pick your jaw back up off the floor. The odds that Obama actually has a 17 point lead over Hillary in California are vanishingly small. But as a gauge of raw momentum, Zogby's not too bad. And he also claims that Obama has been maintaining his traditional strengths whilest wittling away at Hillary's lead among Hispanics, and that's downright fascinating.

The other Golden State poll out this morning comes from Suffolk, better known for tracking the NH primaries. The pollsters there have a good reputation, and they actually publish all of their questions, results, and sample information, so give them added points for transparency:

California: Obama 40, Clinton 39

Suffolk also supplies some explanation. About half their sample had watched their debate, but they were evenly split on who had won. Asked to select which of three endorsements was most influential, 23% went for Bill, 19% for Oprah, and 34% for Camelot's heir.

So the bottom line seems to be that Obama's still gaining ground, both in the national surveys and in the state-level samples, almost across the board. He's got Georgia locked up by a large enough margin to win a bunch of extra delegates, and has erased Hillary's leads in New Jersey and Missouri. Those latter two states are likely to split fairly evenly, meaning that neither will be able to gain much of a delegate edge. That's bad news for the Clinton campaign. But the real story here is California, where the race remains too close to call, but where Obama continues to pick up steam.

Sunday, February 3, 2008

Five Myths About Super Tuesday

Come Tuesday, a seemingly-interminable succession of pundits will mouth mindless

(1) A majority of delegates will be selected: Almost every major media outlet has trumpeted this one. But it’s bunk. In fact, only 42% of the delegates will be selected on Super Tuesday (or in subsequent conventions that reflect the results of February 5 polls and caucuses).
So what’s up? The media is reporting the number of delegates who reside in Super Tuesday states – in fact, more than half. But 400 of these are superdelegates, who get a ticket to Denver irrespective of the outcome of the voting, and are not bound to reflect the views of voters in their states. More than one media outlet is already counting superdelegates from these states in its running tallies, while simultaneously reporting that a majority of delegates will be selected. Go figure.

(2) Early voting may make the difference: This is one we hear every couple of years. Pundits scrutinize demographic breakdowns, polling numbers, and tea leaves, looking for some indication of how the early voting will go in states that allow it. In fact, there’s only rarely a substantial difference between early voters and those who cast their ballots on primary day. In Florida, for example, early and absentee voters supported Clinton 50-31, and regular voters backed her 50-33. In the Republican Primary, Rudy Giuliani learned to his chagrin that his substantial lead with two weeks left hadn’t left him with a substantial cache of votes.
So what’s up? It turns out that the overwhelming majority of “early” voters cast their ballots just before the election, even if the polls have been open for weeks. So unless there’s a major development in the last day or two, there’s not much variation.

(3) The race may not be settled on February 5: Actually, this one’s a rare media understatement. It turns out that Super Tuesday simply can’t settle the race. It has nothing to do with how close the polls are, and everything to do with the fact that even if one candidate were to win every available delegate, they would still fall well short of the 2,025 needed to secure the nomination. So what’s passing as a prognostication is in fact a statement of the glaringly obvious.
So what's up? After building up Super Tuesday for months, the media suddenly realized that it was not the decisive contest that they’d made it out to be. Most years, one candidate becomes the prohibitive frontrunner – a self-fulfilling prophecy. This year, we’re actually going to count the votes.

(4) It’s all about momentum: If there’s one thing we’ve learned this cycle, it’s that momentum matters very little. Obama had huge momentum entering New Hampshire, and lost to Hillary, whose momentum mounted in Nevada, setting the stage for her historic rout in South Carolina. It turns out that voters actually pay attention to what candidates say and do on the trail. Winning in Iowa gave Obama a boost, but not enough to overcome Hillary’s sudden willingness to display her essential humanity. Similarly, all the momentum in the world wasn’t enough to compensate for the Clintons’ willingness to launch slimy attacks, and the voter disgust it engendered.
So what’s up? Momentum gives pundits something to discuss in the long dry spells between primaries. It also freights the early contests with added weight, driving audience interest. Without momentum, Iowa and New Hampshire would just be early contests that award a small number of delegates. Which, come to think of it, is what they are. Now, can we ditch our addiction to subsidizing corn and milk?

(5) It’s a few big states that matter most: In fact, the states that will matter most come Tuesday are those in which one candidate manages to amass a considerable margin. If Hillary and Obama divide California’s 370 delegates evenly between them, then picking up 10 extra Clinton delegates in Oklahoma or 20 for Obama in Georgia might make all the difference.
So what’s up? Actually, there’s a grain of truth in this one. Winning 55% of 370 delegates is obviously better than winning the same percentage of 38. But for all the media attention focused on a handful of large states, it seems more likely that a larger number of smaller states may tip the scales decisively. Polls in the past few days have shown the races neck-and-neck in California, New Jersey, and Massachusetts, making it improbable that either will confer a decisive advantage. The home states – Illinois and New York – seem poised to largely offset each other, as Obama’s larger margin in the Land of Lincoln compensates for the higher delegate tally in the Empire State. Come Tuesday night, don’t be surprised to find yourself waiting with bated breath to learn if Obama has carried Kansas and Idaho, or if he’s won all fourteen half-votes from Democrats Abroad.

On Tracking Polls: Gallup vs. Rasmussen

We've seen a wide range of results in national polls released over the past twenty-four hours. The most interesting results come from the two daily tracking polls: the robocalls from Rasmussen show Hillary's lead widening, while the Gallup Organization's three-day tracking poll, just posted, shows Clinton 46%, Obama 44%. That’s a statistical dead heat.
From the write-up:
“Clinton appeared to reverse the trend toward Obama with a strong day on Feb. 1 -- the first day of interviewing after the Thursday night debate between the two candidates in Hollywood. But Saturday's polling showed a strong day for Obama, bringing the candidates back closer together in the latest three-day rolling average.”
That's more than a little interesting. Rasmussen and Gallup both showed Clinton widening the gap right after the debate. That means that most pundits (myself included) got the debate wrong. Once again, America's women found themselves drawn to Hillary when she wasn't spending her time launching attacks.

But in yesterday's sample, Rasmussen and Gallup produced very different results. Why would that be? The obvious answer is that it was Saturday, the toughest day of the week for polling. (Mark Blumenthal has eloquently detailed the challenges associated with Saturday polling at, for the curious.) In general this cycle, Saturday polling has hurt Obama. It tends to undersample those who are out and about, and has a particularly large effect in robopolls, where many respondents hang up when they're busy. Gallup has a much more robust methodology, and though the poll is fairly new, we've seen much less variation between its Saturday and weekday samples that with Rasmussen.

So if I were a betting man, I'd wager that Gallup is closer to the mark here, and Rasmussen, too, will show a tightening race tomorrow. Worth noting: Sunday is the best day of the week for pollsters. Look for the final day of tracking polls to show this race tightening back up (Rasmussen) or evenly split (Gallup). And, because we're talking about trends, look for Obama to ride that surge into a better than expected performance on Tuesday.