We are live blogging tonight and posting the most recent estimates for each race derived from the official public exit poll tabulations as they are posted online. The table will update regularly (though you will need to reload the page to see updates), and live-blog updates will appear below in reverse chronological order below the table and its caveats below. All times are Eastern.
(If you do not see a chart, click here.)
CAVEATS: The "exit poll estimates" here are extrapolated from publicly available tabulations; they are not "insider information" about up-to-the-minute projections. These estimates are never based upon interview data alone. Tabulations posted as the polls close incorporate pre-election expectations (based on pre-election polls); during the night, the tabulations are updated (occasionally, not continuously) to reflect vote count data. The results posted here may not always reflect the very latest public information, although we hope to keep them close at least through midnight Eastern. Keep in mind that for the first few hours after the polls close these will be survey based estimates, and surveys are subject to error (some random, some from other sources). We encourage you to visit the complete tabulations available at the following links from CBS, CNN and MSNBC.
Candidates projected as winners by the networks are in bold.
1:19 - It is late, we are exhausted and the results seem reasonably settled, so we are going to call it a night. Sometime later today, the exit pollsters will update each cross-tabulation so that the overall result will match the actual vote count. As such, we will probably not do a final update of these tables.
Please join me in thanking Mark Lindeman (who crunched the numbers and did all of the updates) and Eric Dienstfrey (who set up the table so Mark could update it). And thank you Pollster.com readers for nearly 74,952 unique visits and 372,000 page views yesterday. Goodnight!
1:00 - Just to update the earlier discussion of delegates. On MSNBC, Chuck Todd just reported delegate "estimates" for tonight that include California and New Mexico: 841 for Obama, 837 for Clinton, with what he described as a "plus or minus" of 10 on the California delegates (and as Anon notes in a comment, "that's just the [estimated] number of delegates won tonight, not a running total"). I'm assuming other news organizations are or will be reporting similar estimates; please post those to the comments if you catch them.
12:56 - Back online. One commenter is reporting problems viewing the table with the Firefox browser. We tested it in several browsers before going live and I have been viewing on Firefox all night. If anyone is having trouble, please email us with your setup info (to questions at pollster dot com).
12:08 - I will be off the grid for about 30 minutes, although Mark Lindeman will continue to post updates.
12:05 - Anon posts some numbers gleaned from past exit polls:
- 2006 Dem gubernatorial primary - 12% Latino, 8% black (LAT)
- 2004 Dem presidential primary - 16% Latino, 8% black (via CNN)
- 2000 Dem presidential primary - 17% Latino, 11% black (via CNN)
- 1998 Dem gubernatorial primary - 12% Latino, 14% black (LAT)
12:00 - Mark just posted an update of the CA tabulations. One thing to keep in mind is that the exit poll analysts need hard vote counts to determine two very important things: The appropriate mix of early and polling place votes, and the correct regional distribution of the turnout. Adding hard data will either firm up the racial composition numbers reported below, or change them. My guess is that their level of confidence in those racial composition numbers will determine when they "call" a winner in CA.
11:55 - "Pragmatic Progressive" corrects me on the Chuck Todd delegate report I posted at 11:25. Todd projected a total of 1140 delegates, but (assuming I grabbed the right numbers) there are 1,681 delegates up for grabs tonight. Todd's estimate does not yet include roughly 546 delegates from California and some of the other Western states.
11:41 - As of the first report, the California exit poll shows Latinos contributing a larger percentage of the vote, and African Americans a smaller percentage of the vote, than indicated in the pre-election polls.
Latinos are currently showing as 29% of the Democratic vote in California on the exit poll tabulations. In the final pre-elections polls they were:
African Americans are currently showing as 6% of the Democratic vote on the exit poll tabulations. In the final pre-election polls they were:
- 12% on Field
- 9% on SurveyUSA
- 9% on Mason-Dixon
- 8% on Suffolk
As elsewhere, Clinton wins the vast majority of Latinos in California (65% as of now) and Obama wins a greater share of African-Americans (78%), so if those composition numbers hold, it bodes very well for Clinton. Update: Just noticed that SN made a similar observation in the comments as I was typing up these numbers.
11:25 - Daniel T made this point about an hour ago in the comments:
"[W]inning" states in the primary means nothing since no democratic states are winner take all. The real key is the delegate count and while Clinton is still leading there, the night is still young.
He is right, of course, which is why I will pass on something that Chuck Todd just reported on MSNBC. Their estimates are that when all votes are counted, the delegates at stake tonight will split 595 for Obama and 546 for Clintons, so "we're looking at an even split" in the total delegate count tomorrow. [Correction: I heard that wrong. Todd's totals apparently do not include CA and other western states -- see my 11:55 update].
11:12 - A quick summary table of the current tabulations among white voters as reported on the exit polls (and these are not extrapolations but the report among whites in each state):
11:05 - Mark L just added the CA estimates.
10:50 - Oops. Looks like we overlooked New Mexico for the Democrats. Updating shortly. Thanks to Anon and Daniel for the edit.
10:40 - Mark L just posted more recent updates from most of the states. The additional actual vote counts have changed some of the estimates considerably. For example, Clinton's margin in Massachusetts the margin is now roughly 15 points.
9:32 - Another apparently slow-to-update estimate is New Jersey. MSNBC has called the state for Clinton, but the update has not yet updated since we last checked at 8:30 and found Clinton receiving less than a percentage point more than Obama.
9:12 - For what it's worth, regarding the spin and counter-spin now underway over Massachusetts: We reported only a handful of pre-election polls there in recent weeks. The Suffolk University/ News Poll had the two candidates essentially deadlocked (Obama 46%, Clinton 44%), Rasmussen Reports (a week ago) had Clinton leading by six points (43% to 37%) and Sur veyUSA had Clinton leading by 17 percentage points (56% to 39%). The current estimate is Clinton by 5 (50% to 40%). Of course, that estimate will likely change as more hard count comes in. [UPDATE - Not so fast...the 10:36 update increases the Clinton lead to roughly 15 points (56% to 41%)].
9:05 - Josh asks:
MSNBC just called MA for Clinton... any idea what they believe that you don't at this stage?
What Josh noticed (that MSNBC called Massachusetts while the estimate we had here still showed Obama slightly ahead) underscores the point I made at 7:30. The exit pollsters update the cross-tabulations less frequently than the estimates they use to make projections. They updated the MA cross-tabulations online just as the call was going on the air. And the updated estimate in Massachusetts -- which is now undoubtedly based on a lot of actual vote count in the sampled precincts -- has Clinton ahead by roughly five points.
8:56 - Two things: First, as should be obvious, we are not bothering with information that you can get faster and better from the networks. Second, I have been watching MSNBC, and just heard our friend Chuck Todd reading projected delegate splits from various states. Needless to say, those counts are the most important numbers tonight.
8:51 - Just saw this comment posted earlier by our colleague Marc Ambinder about those leaked early exit polls (that appear, not here, but on sites like the one who's name rhymes with "budge"):
Fellow journalists and pundits. I have the same data you have... and I would just remind all of you that the first wave of exit poll data is not reportable or reported for a simple reason: the sample sizes are not large enough to accurately tell us much of anything, unless one candidate is getting, like, 80% of the vote.
That's about right.
8:45 - One note: An informed source advised earlier today that the polling place "exit" interviews will be supplemented tonight with telephone surveys conducted among early and absentee voters by the network consortium in California, Arizona and Tennessee.
8:42 - In the comments, Daniel T suggests:
it would be helpful if you but in italics or bold the number of the "projected" winner for that state
Great suggestion - consider it done (literally). We have highlighted candidates projected as winners by the networks (or at least one of them) in bold.
Would it be possible for you to highlight the cells in the spreadsheet according to who is leading in each state? It would make it easier to follow as the estimates shift throughout the
Possible, obviously, but we would rather not, as we're really not sure that someone "leading" is really leading (we do not have access to the confidence level assigned to each margin). Daniel T also asks:
Could you add a column that shows the number of delegates atstake in each state. I think there is room. It would be helpful.
Sigh. Yes, it would, but that one will have to wait for Super Tuesday 2009 (or perhaps next week). The table ain't broke, so we are not going to try to fix it. But thanks for all of the suggestions.
8:27 - So I've alluded to the fact that these estimates improve over the course of the evening. What does that mean? It means that as the polls close, the estimates are based on some combination of results from the exit poll interviews and (believe it or not) pre-election poll averages. Once the polls close, the interviewers attempt to obtain actual results for their sampled precinct (or another "reporter" attempts to get the results from the county or state registrar). The exit poll analysts use these numbers to do two things: First, they gradually replace the exit poll results in their estimate models with the actual count precinct-by-precinct. They also calculate the "within precinct error" statistic for each state (or regions with each state - not sure) that are used to adjust exit poll results from the other precincts where actual count is not yet available.
The exit pollsters also have "reporters" who gather hard vote counts from a much larger random sample of precincts. All of this data goes into the computer models and is used to create various estimates of the vote. The "decision desk" analysts look at all of those estimates in deciding whether to "call" a race.
A separate operation within exit-poll-central takes whatever estimate they deem most trustworthy and uses it to weight the subgroup tabulations that we can read online. And we take those tabulations as they appear online extrapolate the estimates above from those tabulations. They typically do one updates 30 to 60 minutes after the polls close and another two or three over the course of the evening. All of this is a long-winded way of saying that what you are seeing here is not as current as the information the network analysts are using to make their calls.
8:13 - Mark L has just posted more estimates, and it is worth stressing something important: Most of these margins of very close, well within the "margin of sampling error." Needless to say, the networks are not yet ready to project and will make their calls as hard returns become available. A difference of a point or two is really meaningless at this point.
8:00 - Polls just closed in Alabama, Connecticut, Delaware, Illinois, Massachusetts, Missouri, New Jersey, Oklahoma and Tennessee. We should have updates for these soon.
7:51 - "Anon" comments:
you'll get slightly different numbers depending on which breakdowns (age,
race, geography, etc) you plug into the equation. your numbers are prettyclose to mark's, though i don't know which one(s) he's using.
Thanks for asking. You remind of two things I neglected to mention. First, Anon knows that the brains behind these tabulations is Mark Lindeman, an assistant professor of political studies at Bard College in Annandale-on-Hudson. Mark is the one doing the extrapolating. The table that runs with a Google Docs spreadsheet, was put up by the indefatigable Eric Dienstfrey. Thank you Mark and Eric!
Second, Mark has put together a program that does not one, but more than 20 extrapolations and then calculates the average of all of them. The point is that the extrapolation introduces rounding error, so averaging a bunch of calculations minimizes that potential.
7:44 - So what's the point of posting the estimates extrapolated from the exit poll tabulations? One reason is that those estimates, imperfect as they are, provide what is arguably the best projections available of the final result. Again, and we cannot stress this enough, those estimates are subject to change and should improve over the course of the night. But in watching the extrapolated estimates after the polls closed since in all of the primaries so far this year, I can say that they provide a far more accurate data on the outcome than the first waves of actual vote that crawl across the bottom of the network broadcasts.
7:31- A certain web site (rhymes with "Grudge") has a bunch of leaked exit poll estimates in states whose polls have not yet closed. I'd recommend to those curious about those leaked numbers -- and everyone else, for that matter -- to review my post from earlier today. Remember, it's just a survey, the networks rarely "call" an election on the exit poll alone unless the leader's margin is *very* large (i.e. a "lead" of 3-4 points isn't enough to assure a victory) and the early estimates (as well as the tabulations we're extrapolating from above) are usually weighted by something called the "composite estimate" which includes an average of pre-election polls.
One more thing to keep in mind: The leaked numbers from the West coast states are extremely preliminary. Remember, it's only 3:30 in the afternoon there. These are mostly from the first wave of reported exit poll data (there are three). Experience has shown that the first two waves can be unreliable, but the numbers usually settle down in the third wave that reports just before the polls close.