I will be updating this post sporadically over the next hour or so, and then hopefully more often as the polls start to close. All times here are eastern.
8:31 [Mark] Ok, now I'm off to the live discussion on WashingtonPost.com. See you at 9:00. Brian will pick this up in a new entry.
8:29 [Mark] s.b. comments:
ok riddle me this again. CNN has Pennsylvania as a 15% spread and CBS has less than a 2 point spread 49-50. The CBS poll was for 1929 people. Can't find the CNN numbers but that's a big difference.
CNN's tabulation appears to be more recent than CBS-- based on 2,567 respondents (that number appears just below the "Pennsylvania" heading near the top
8:25 [Brian] It looks like late deciders broke evenly. According to the national exit poll data, 10% made up their mind at some point during the last week. Obama won this group by a margin of 51-46%.
8:22 [Mark] Major points (and thanks) to Thatcher:
To everyone - save bandwidth - if you want to see the updates on the Election Map ... don't refresh the whole page .... just go to the drop down menu and selection "Election Night 2008" again ... that way the only thing that gets refreshed is the flash application - save the bandwidth of Pollster - we don't want a crash tonight.
8:19 [Brian] The early Pennsylvania exit polls are showing that Obama is matching Kerry's support in Pittsburgh and exceeding his support in every other region of the state. No wonder the networks felt safe calling the state right away.
8:18 [Mark] The exit poll tabulations for Pennsylvania are weighted to a 15 point Obama lead and at two networks (NBC and ABC) called Obama the winner as the polls closed. That is the sort of margin the decision desk analysts require before they will make a projection on an exit poll alone, though on the other hand, the other networks have not called it yet.
8:07 [Brian] Early exit polls indicate that Hispanic voters in Florida are going for Obama 55-45%. In 2004, they went for Bush 56-44%.
8:03 [Mark] I really, really need a nap or a long vacation or both. Post live chat starts (for me) at 8:30. We'll do the switcheroo in 30 minutes. Meanwhile, networks call PA based on the exit polls. That requires a very, very large margin and overall confidence by the NEP analysts in their exit polls tonight.
7:57 [Mark] I'm going to be participating in a live online discussion at WashingtonPost.com for the next 30 minutes. Meanwhile, Brian is going to start up another entry and continue our live blog here. I'll be back at 8:30.
7:50 [Mark] Thatcher asks: "Will the Election Map auto update or will we need to refresh to see updates." You do need to hit refresh.
7:48 [Brian] The initial exit polls out of North Carolina are showing African Americans making up 22% of the electorate. That would actually be down from 26% in 2004. A decline in African American turnout would certainly run counter to the trend we've seen in some other states like Georgia. I wonder if the early vote may be causing some issues here?
7:35 [Brian] In Georgia, the exit polls are showing that African Americans make up 30% of the electorate, up from 25% in 2004. Obama is only pulling 25% of the white vote though. If that holds up, he would probably come up just short of winning the state.
7:31 [Mark] The map is now coloring correctly. Thanks for your patience and sorry about the confusing.
7:27 [Mark] Reader s.b. writes:
Ok riddle me this. CBS and CNN have very different exit poll results for Indiana. CNN shows obama winning and CBS shows McCAin winning??? Please explain.
As I write this, the CBS tabulation shows 1937 interviews in Indiana, while the CNN tabulation shows 2,336. The CNN numbers represent a more recent update. Keep in mind that these numbers constantly update and refresh, particularly in the 30 minutes or so after the polls closes, as exit poll interviewers call in their final batch of results.
7:23 [Brian] Here are some striking party id figures from Virginia. 41% are Democrats, 32% Republicans, and 21% independents. In 2004, 35% were Democrats, 39% Republicans, and 26% independents. If those numbers hold up, that is an amazing party id shift in just four years.
7:20 [Brian] Youth vote also looks significant in Virginia, at least in the initial exit polls. 18-29 years olds are 21% of the electorate and they are breaking 62-37% for Obama. In 2004, 17% were 18-29 year olds and they went 54-46% for Kerry.
7:15 [Mark] - Yes, we know Kentucky is not lighting up red as it should. Minor bug -- we're working on it.
7:13 [Brian] Youth vote in Indiana. In 2004, 14% of Indiana voters were 18-29 years old and they broke 52-47% for Bush. In the initial Indiana exit polls, 19% of the electorate are 18-29 years old and they are breaking 65-33% for Obama.
7:08 [Mark] Numbers Guy Carl Bialik has a blog item out (quoting me among others) on what pollsters do on Election Day. It includes news that SurveyUSA is conducting telephone polls tonight in 30 states for a voter-fraud watchdog group
7:04 - [Brian] The initial national exit poll Mark just posted shows Obama winning women by 16%. This would match the size of Clinton's advantage among women in 1996--the largest gender gap a Democrat has enjoyed in a recent presidential contest.
6:52 - [Mark] MSNBC (and presumably the other networks) has posted initial tabulations from the national exit poll sample. This one deserves extra cautions: It is based on very preliminary results in the western half of the U.S. and may not include any early vote telephone interviews. And keep in mind that the national exit poll tabulations at about this time four years ago showed President Kerry with a three point lead. Having said that, the estimate that weighted those results indicates a roughly ten point Obama lead (54-44).
6:43 - [Brian] Based on early results, the electorate appears be more favorably disposed towards an activist government this year than in 2004. In 2004, respondents were 3% more likely to think the government was doing too much. So far this year, exit poll respondents are 8% more likely to think that the government should do more.
6:42 - [Mark] Starting now, Brian Schaffner will be live blogging as well.
6:38 - A couple of notes on the map. Unfortunately, it appears on the front page only - we didn't have time or budget to get it to work on the little map (and the tool tip is way too big). Also, the map does not automatically refresh. You may need to occasionally reload the front page. Just keep an eye on the update time in the lower right corner. And thanks to those who emailed or commented about that odd glitch that mde the analysis column disappear. We're working on it.
6:31 - Time's Sean Gregory has a helpful piece up on how the networks use exit polls and vote returns to make their projections. Here are two especially relevant paragraphs:
At that time, the network decision makers start running the Edison/Mitofsky data through their models. Besides exit polls, these numbers also include telephone surveys of absentee and early voters from 18 key states. As the polls start closing at 7 p.m., Edison/Mitofsky also start providing the networks actual results from precincts in every state. The stations also toss the Associated Press's proprietary counts of the actual vote into the mix. Each network employs a team of statisticians and experts to analyze the numbers. Of the 14 staffers on the CBS desk, five have political science Ph.D.'s. NBC has a Google employee and physics Ph.D. on staff.
The math geeks crunch the numbers and present their recommendations to a senior staffer, who makes the final call. That decision, of course, is based on some key guidelines. NBC, for example, won't declare a winner until its models show less than 1 in 200 chance of error. And the networks won't declare a state's winner until polls in that state close. Since 2000, the networks have revamped their projection process, beefing up the qualifications of its decision-day staffers, adding more sophisticated statistical tools and instituting more dry runs. Gawiser, who was NBC's projection chief in 2000 (and still lives with the shame of that night), says the network had just three or four models in place for Bush-Gore. Now, NBC has "several hundred."
6:15 - By the way, if you are just tuning in, make sure to see my post on exit polls earlier today and why you would do best to ignore whatever leaks in the next hour. Here's another reason: Right now Drudge says "EXIT POLLS SHOW OBAMA BIG" while Gawker is showing results that say, not so much. Moral: We have no idea what we are looking at, where it comes from or what it means.
6:02 - Republican pollster Chris Wilson is blogging a post election telephone survey they are conducting and reporting as interviews come in. I do not vouch for the methodology nor claim I have even read it closely, but Wilson's writes that his initial data "looks grim for McCain/Palin."
5:49 - And while I am pointing to web sites worth checking out, Swing State Project has a terrific, color coded map up that displays the closing times. A great guide to what we'll see when.
5:47 - 538's Sean Quinn reminds me to point out that Michael McDonald's invaluable early voting site has the most current and most thorough reporting on early voting.
5:40 - The Page has posted some initial preliminary exit poll results broadcast by the networks that do not include any estimates of candidates. Surprisingly enough, the economy is the issues voters care about most. Marc Ambinder also has a nice summary of same.
5:20 p.m. -- Welcome to live blogging here at Pollster.com. If you haven't noticed yet, be sure to open up our front page in a separate browser tab or window. And if you have noticed, no the map isn't broken. We have added a revised Flash map to track the results and network calls. All the details are in this post. The old maps are still there (and the small version in the right column will remain in its standard format for the rest of the night).