Monthly Archives: November 2008

Reminder About Leaked Exit Polls

Conor’s inquiry prompts me to remind all of you about leaked exit polls. Most of my regular viewers have heard this screed before, but for those who have not, remember that leaked exit polls aren’t necessarily “inaccurate” – they are incomplete, and therefore unreliable indicators of the vote.  I still recall the early exit polls that were leaked in 2004 – they had Kerry up – and my colleagues went delirious with joy despite my cautions.  Of course, when Bush won, they assumed vote fraud. Nothing of the sort happened.

In the recent primaries, of course, the early exit polls almost always overstated Obama’s support. I’ve talked extensively why this is the case, so I trust I need not revisit the issue.  Just remember that the raw exit numbers which come in waves are readjusted as the consortium running the polling adjusts for turnout.  So Drudge may leak a first wave as early as 1 p.m. – pay no heed!

The UnDecideds Decide – But for which candidate?


Three more national polls have come in overnight, from Marist, Reuters/Zogby and IBD/TPP.  Presumably all include some polling from as late as Sunday. None do much to change the scenario in the national popular vote scenario I painted yesterday (Obama up by roughly 5-7%), but it is interesting to see how IBD/TPP allocated its undecideds.  As I noted yesterday, most of the major pollsters, in line with what I had suggested a few days back, are allocating their undecideds roughly 60-40 for McCain.  They do so on the basis of demographics and by “pushing” the undecideds by asking them to make a choice.  Most break for McCain when pushed.  IBD’s final poll has Obama up 47.5%-42.4%, a 5.1% margin.  But they then allocate the undecideds to Obama by a 4-to-1 margin, 4.0-1.9, to give them a final poll result of 51.5-44.3%.  Had they allocated their undecided similarly to the other pollsters (say 3-2 for McCain), the final results would have 49.5%-45.4% – a different race on its face.

This doesn’t count the 4% “Other” that still remain in the IBD poll, which presumably included remaining undecideds.

Marist says it pushed its leaners, but it doesn’t reveal the breakdown. Its final poll has 2% undecided and 3% “Other”.  I have not been able to find a breakdown of Zogby’s undecideds.

I have no idea if IBD’s decision is the correct one. But one of these pollsters is allocating undecideds incorrectly.  We’ll know in less than 24 hours.

Who will win the popular vote? The polls speak one last time

Almost all the penultimate tracking results for the national polls are in, and they provide a rosy picture for Obama, and a bleak one for McCain.  For McCain diehards, however, there is still evidence that Obama has not yet clinched the deal.

First, the numbers (Obama first):

Gallup 53-42  Obama +11.  5% undecided (With undecideds allocated:  55-44.)

Hotline 50-45.  Obama +5   5% undecided.

Rasmussen 52-46 Obama +6   2% not accounted for.

Battleground 50-44   Obama +6  6% undecided.

Reuters/Zogby  50.9-43.8  Obama  +7.1.   5% undecided.

IBD/TPP 46.7-44.6  Obama +2.1 8.7% undecided

Average Tracking Poll Results:  50.4-44.2.  Obama +6.2.  5.3% undecided.

There are several things to note here. First, there’s no evidence of a trend toward either candidate. Indeed, the race has remained remarkably stable since the end of the media fixation on the fiscal meltdown in early October.  Most notably, Obama has hovered at about or slightly above the 50% barrier since then, but can’t seem to break that magic barrier decisively (beyond the margin of error in these polls) without some help from a decision to allocate the undecideds. By far the greater variation in the last month has  been in McCain’s support, which today ranges from 42 to 46%.

Second, the number of undecideds is simply not dwindling; depending on the pollsters, it remains mostly between 4-6%.  It may be that a significant portion of them will not vote. In my view, however, those who do will break in greater numbers for McCain.  But note that even if he gets 5 out of every 6 undecideds (unlikely based on past history, although he’s doing that in the last week in Pennsylvania), it won’t put him over the top.  Most pollsters who are allocating the undecideds are either splitting them evenly or giving McCain a slight plurality.  This means to draw even in this race, McCain will have to sweep the undecideds and see some slippage in Obama’s support.  And most of the polls are showing that Obama’s supporters are more committed than are McCain’s.

Now, could these polls be wrong?  Obviously there will be some sampling error. The only factor that might lead these polls to be systematically wrong, however, would be if the Sally Field/Bradley effect is somehow in play here.  And there’s no way to foresee that.

But barring systematic error of this type, they all tell the same story: Obama will win the popular vote.  By how much depends on the undecideds.  But at this point the forecast models from August look like they have hit Obama’s final popular vote total squarely on the nose.  We will see if McCain closes the gap at all in the next 48 hours.

What about the state-level polling?  Believe it or not, the picture is slightly less bleak for McCain.  But it will take a closing surge of historic proportions for him to squeak out an electoral college victory.  Because the situation is more complicated with state-level polling, I’ll need some time to put it in summary fashion. That comes next.

(PS – I’ll update these figues when the ABC tracker comes in later today, but I don’t expect it to change the basic picture.)

Waiting for the Youth Vote

The sun comes up, the sun goes down, and young voters stay home in presidential elections. Three observations that remain true today.

Here’s the data so far on the much ballyhooed “youth” vote.

Pew finds 19% of registered voters falling in the 18-30 years-old category, and 15% among likely voters – this is essentially unchanged from 2004 (19% and 14%).  Gallup has almost identical figures: it estimates that 18 to 29-year-olds constitute 12% of likely voters (14% in their expanded model).  That’s also almost identical to their final pre-election poll (13%) in 2004. They conclude that as a proportion of overall voters, we are unlikely to see an increase in the youth vote this year.

If Obama is basing his victory on higher than expected youth turnout, he’s never seen me try to get my teenage boys out of bed before noon on a Saturday.

Not going to happen.

Looking at Get-Out-The-Vote: Who is More Effective?

Contrary to the impression conveyed in many media reports, I have been arguing that I have seen no clear sign that either Obama or McCain has run the more effective campaign, strategically speaking.  Instead, media claims that Obama’s campaign is better organized is partly predicated on his massive advantage in money to spend.  One way in which that advantage ought to manifest itself most heavily is in the get-out-the-vote (GOTV) efforts.  We would expect. Obama’s campaign to be more effective than McCain’s at getting people to the polls.  This was a topic raised by Adam Lovell much earlier in the year, and we are now in a position to try to answer his question: which organization has the better GOTV?

Some 30 states have early voting, so we can use turnout in these states as an early indicator of the relative effectiveness of the GOTV for both candidates.  Of course, states cannot report actual vote totals as yet, but several pollsters have been surveying the early voters, so we can draw some conclusions.  CBS reports that about 25% of registered voters have already voted, and they report that early voters back Obama 56%-41%. Pew has 32% of likely voters already casting their ballots with Obama up 52%-39%.

However, how do we know that the Obama advantage in early voting is due to their GOTV superiority?  Maybe it simply reflects the broader support for Obama more generally, as well as the greater level of enthusiasm shown by Obama supporters?  One way to address the marginal impact of GOTV is to ask voters if they have been contacted by either the McCain or Obama campaign.  Several pollsters have asked that question.

The Wall St. Journal/NBC polls find almost no difference in the likelihood of voters getting contacted by either campaign: 24% of voters were contacted by both candidates’ organizations, 10% by the Republican only, and 12% by the Democrats only. (Btw, this proportion almost exactly mirrors the contact data from 2004, but the number of contacts has almost doubled).  So, in total, 36% of voters have been contacted by Obama, and 34% by McCain.

Gallup paints a somewhat similar picture; they find that 39% of registered voters have been contacted by the Obama campaign, but only 33% by the McCain campaign. . Interestingly, however, blacks report a much higher rate of contact – almost 4 times more – from Obama than from McCain.

Of course, not all contacts are the same: We know from previous studies that personal contact is the most effective way to get out the vote. Which organization is doing better at that? Fortunately, Pew breaks the contacts down by category: mailings, telephone calls (personal or automated) or personal visits. About 14% of registered voters have been personally contacted by someone from a campaign. Here we find that 18% of Obama supporters have received a personal visit, but only 10% of McCain supporters.  McCain has tended to rely more heavily than Obama on direct mailings.

Not surprisingly, then, based on data through Nov. 1, it appears that the disparity in funding has provided an edge to Obama in terms of GOTV, particularly in his campaign’s ability to target potential supporters through personal contacts.  (Truth in Advertising: I should acknowledge here that my oldest son did door-to-door canvassing for the Obama campaign this summer.  You should NOT take that as an expression of my partisan leanings or support for either candidate, however.)   Of course, we will know just how much of an edge tomorrow.  But this could prove crucial in the 5 battleground states on which hinge McCain’s very slight chance to pull out an Electoral College victory.