8:00:01pm
MSNBC calls IL for Obama and OK for Clinton.
MSNBC calls CT, IL and NJ for McCain and MA for Romney.
This is where the real fun begins. Isn't this so much more interesting than spending 5 hours on Ohio?
8:03pm
Looks like CNN made the same calls.
8:10pm
Since CT and NJ are winner take all, calling those states for McCain means that he has picked up 79 delegates. MA has a proportional allocation, so it remains unclear how many delegates Romney and McCain will each pick up there.
FOX has called AL for Huckabee.
8:13pm
MSNBC goes more into the exit polls...Clinton is winning Hispanics 61%-38%. What does that mean for the race, particularly in CA?
8:14pm
Keith Olberman is getting a little carried away now...getting way too excited about vote counts with less than 1% of precincts reporting. Calm down Keith, it is going to be a long night.
8:19pm
NYT is reporting that NBC is rescinding its call of OK for Clinton and that FOX has called TN for Clinton.
8:20pm
And NBC has now joined in on the TN call.
8:25pm
The broadcast news outlets and blogs appear to be abuzz with talk that the Obama camp is ramping up their expectations. One question: why do that?
8:29pm
Here comes Arkansas...immediate call for Clinton and Huckabee?
8:30pm
Answer: Yes.
8:34pm
There appears to be some discrepancy between the exit polls leaked earlier and the exit polls now being posted on the network websites as the polls close. I'll update the delegate estimate based on the exit poll numbers on the network sites as those go up.
8:47pm
I'm about to revise the delegate estimate based on the exit poll numbers. My revised numbers show Obama 727, Clinton 699, and 298 in states without exit polling.
8:52pm
The Clinton campaign is bragging about the red states that they've won...AR, TN, and OK.
8:55pm
Something interesting I found when going through the exit polls. About 6% in TN reported voting for John Edwards who, as you may have heard, left the race a week ago. Pretty strong support for a guy who is not in the race...why is that?
On MSNBC, Bill Richardson has taken advantage of his withdrawal of the race to grow a Gore-like beard.
8:58pm
NBC has called MA for Clinton. The exit polls showed Clinton and Obama virtually tied there.
9pm poll closings coming up. Any chance they'll be able to call NY for Clinton right away?
Tuesday, February 5, 2008
Democratic Delegate Estimates (Based on Exit Polls)
I've used the exit polls available online to create a new (and very rough) estimate of how the Democratic Delegates may be allocated after today. In addition to all the cautions I mentioned here about why these estimates are error-prone, add to that the fact that exit polls have had notorious issues in correctly predicting the vote in recent elections. Nevertheless, here are the estimates based on the leaked exit polls.
Remember that there is no entrance polling in caucus states and I also found no information for Utah. However, if this is at all accurate, Obama would get 724 delegates, Clinton would receive 725, and 275 are in states with no exit poll information. The big thing to note here is that if these polls are correct, Obama would have run up the score significantly in IL while Clinton would have won her own state by a much smaller margin. Obama would also have benefited from running up the score in states like AL and GA.
Stay tuned...
UPDATE: I've updated the sheet to reflect the slightly different numbers that the media outlets have made available for the states where polls have closed. (NEP* indicates an exit poll from a media website, NEP indicates that I'm using numbers from leaked polls).
UPDATE 2: The sheet now reflects only one leaked exit poll (CA).
Remember that there is no entrance polling in caucus states and I also found no information for Utah. However, if this is at all accurate, Obama would get 724 delegates, Clinton would receive 725, and 275 are in states with no exit poll information. The big thing to note here is that if these polls are correct, Obama would have run up the score significantly in IL while Clinton would have won her own state by a much smaller margin. Obama would also have benefited from running up the score in states like AL and GA.Stay tuned...
UPDATE: I've updated the sheet to reflect the slightly different numbers that the media outlets have made available for the states where polls have closed. (NEP* indicates an exit poll from a media website, NEP indicates that I'm using numbers from leaked polls).
UPDATE 2: The sheet now reflects only one leaked exit poll (CA).
UPDATE 3: I've added UT's exit poll to the mix. Now just waiting for CA's.
UPDATE 4: Just added CA's exit polls from the media outlet websites. With regard to the unallocated 275 delegates, Obama seems to be doing well in those unpolled caucuses states so far.
Georgia and the run-up to 8pm (7pm-8pm)
S7:05pm:
All the networks call Georgia for Obama. Evidently the networks are calling Georgia a 3-way race for Republicans--should be very interesting to see how that shakes out.
Question about Obama's quest for delegates in Georgia: Since 57 delegates are allocated at the congressional district level, will Obama be helped or hurt by the fact that Georgia has a couple of majority-minority districts?
7:12pm:
MSNBC is reporting that Obama got over 40% of the white vote. That is an impressive figure.
Of course, the margin for Obama is more important than winning the state, but the fact that they could call it right away suggests he will have a significant margin.
If there is nothing to report on the Republican side, Obama will get an hour's worth of favorable news coverage before the 8:00pm polls close. Could this possibly affect how people vote on the West Coast?
7:22pm:
Leaked exit polls...read with caution.
7:46pm:
Sorry to be MIA for a while there. As you can see above, I was creating a new delegate estimate based on leaked exit polls. Hey, the exit poll figures were out there, was I supposed to just ignore them?
7:55pm:
There has been nothing to say on the Republican side, so the networks have been focusing a great deal on Obama's victory and particularly with how well he did among whites.
Polls closing at 8pm are AL, CT, DE, IL, MA, MO, NJ, OK, and TN. Based on those leaked exit polls, I'd expect the networks to call AL and IL for Obama and OK for Clinton pretty quickly.
How many states will be called for McCain?
All the networks call Georgia for Obama. Evidently the networks are calling Georgia a 3-way race for Republicans--should be very interesting to see how that shakes out.
Question about Obama's quest for delegates in Georgia: Since 57 delegates are allocated at the congressional district level, will Obama be helped or hurt by the fact that Georgia has a couple of majority-minority districts?
7:12pm:
MSNBC is reporting that Obama got over 40% of the white vote. That is an impressive figure.
Of course, the margin for Obama is more important than winning the state, but the fact that they could call it right away suggests he will have a significant margin.
If there is nothing to report on the Republican side, Obama will get an hour's worth of favorable news coverage before the 8:00pm polls close. Could this possibly affect how people vote on the West Coast?
7:22pm:
Leaked exit polls...read with caution.
7:46pm:
Sorry to be MIA for a while there. As you can see above, I was creating a new delegate estimate based on leaked exit polls. Hey, the exit poll figures were out there, was I supposed to just ignore them?
7:55pm:
There has been nothing to say on the Republican side, so the networks have been focusing a great deal on Obama's victory and particularly with how well he did among whites.
Polls closing at 8pm are AL, CT, DE, IL, MA, MO, NJ, OK, and TN. Based on those leaked exit polls, I'd expect the networks to call AL and IL for Obama and OK for Clinton pretty quickly.
How many states will be called for McCain?
Super Tuesday Live Blog: Early Thoughts (5:45pm-7:00pm)
5:45pm:
There will be no entrance polls in caucus states tonight (NYT). That means it will be awfully late when we find out what happens in states like North Dakota and Idaho. We've had essentially no polling in those states, so we really don't what will happen there.
CNN has started telling us a bit about the exit polls. Evidently only 49% of those voting for McCain labeled themselves conservative (the numbers were much higher for Huckabee and Romney).
Questions I'm wondering about:
What does "victory" look like tonight? Given how close this seems to be on the Democratic side, can Obama or Clinton claim that they won if they come out with more than a 100 delegate lead (not including pledged delegates)?
Related to the above, if we don't know who won or lost, when do the candidates come out to address supporters? Do they come out after they win their home states? Wait for someone to call Calilfornia?
How much time will the networks spend obsessing over who wins and loses states like Missouri when the margin of victory in IL and NY is at least as important? (See the earlier post on this point).
Would McCain be more worried if only Huckabee leaves the race or if only Romney leaves the race?
This is the closest thing we've ever had to a national primary...what will we think of it when it is over? Big success? Big failure?
6:00pm:
My home state of Georgia is the first to close its polls one hour from now. Here is the site where you'll be able to track those returns as they come in.
6:20pm:
Is this a hint that things will go well for Obama? MSNBC releases more from the exit polls. Here is a potentially crucial nugget: 53% chose change at the most imporant candidate quality, 22% selected experience, 14% selected cares about people like me, and 9% selected electability. The high numbers for change seem good for Obama, which MSNBC confirms as they indicate that three-fourths of those selecting the change quality voted for Obama.
But the exit polls can't get to the earlier voters...how much will those help Clinton?
6:30pm:
This is the first election for which Georgia has had its controversial photo identification requirement for voting. This may be causing long lines at the polls (story here).
6:40pm:
Another hint from the exit polls?
6:55pm:
Mark Blumenthal has a nice post discussing the large variance in the state polling leading up to Super Tuesday.
Five minutes till the polls close in Georgia. What will come first, the networks calling the state or my pizza out of the oven?
7:00pm:
My pizza beat MSNBC's call of Georgia for Obama by about 30 seconds.
There will be no entrance polls in caucus states tonight (NYT). That means it will be awfully late when we find out what happens in states like North Dakota and Idaho. We've had essentially no polling in those states, so we really don't what will happen there.
CNN has started telling us a bit about the exit polls. Evidently only 49% of those voting for McCain labeled themselves conservative (the numbers were much higher for Huckabee and Romney).
Questions I'm wondering about:
What does "victory" look like tonight? Given how close this seems to be on the Democratic side, can Obama or Clinton claim that they won if they come out with more than a 100 delegate lead (not including pledged delegates)?
Related to the above, if we don't know who won or lost, when do the candidates come out to address supporters? Do they come out after they win their home states? Wait for someone to call Calilfornia?
How much time will the networks spend obsessing over who wins and loses states like Missouri when the margin of victory in IL and NY is at least as important? (See the earlier post on this point).
Would McCain be more worried if only Huckabee leaves the race or if only Romney leaves the race?
This is the closest thing we've ever had to a national primary...what will we think of it when it is over? Big success? Big failure?
6:00pm:
My home state of Georgia is the first to close its polls one hour from now. Here is the site where you'll be able to track those returns as they come in.
6:20pm:
Is this a hint that things will go well for Obama? MSNBC releases more from the exit polls. Here is a potentially crucial nugget: 53% chose change at the most imporant candidate quality, 22% selected experience, 14% selected cares about people like me, and 9% selected electability. The high numbers for change seem good for Obama, which MSNBC confirms as they indicate that three-fourths of those selecting the change quality voted for Obama.
But the exit polls can't get to the earlier voters...how much will those help Clinton?
6:30pm:
This is the first election for which Georgia has had its controversial photo identification requirement for voting. This may be causing long lines at the polls (story here).
6:40pm:
Another hint from the exit polls?
6:55pm:
Mark Blumenthal has a nice post discussing the large variance in the state polling leading up to Super Tuesday.
Five minutes till the polls close in Georgia. What will come first, the networks calling the state or my pizza out of the oven?
7:00pm:
My pizza beat MSNBC's call of Georgia for Obama by about 30 seconds.
Huckabee Big Winner So Far
Yep, he won 18 of West Virginia's 30 delegates (the other 12 will be determined by their primary held in May) at the convention today. This gives Huckabee a great deal of momentum and positive news coverage for at least the next 2 hours. This afternoon belongs to Mike Huckabee.
See you in a couple of hours...
See you in a couple of hours...
Super Tuesday Democratic Delegate Cheat Sheet--How to Understand the Results
If either Obama or Clinton runs away with a state like Missouri or Alabama with a 10%-20% victory, the news media is likely to make a big deal of the size of the candidate's margin in that state. But keep in mind that we are in a delegate battle now, so a larger than expected victory is really only important inasmuch as it translates into delegates. Given that the Democrats are allocating delegates proportionally, it is possible to come up with an estimate of how many delegates a candidate is likely to receive for each additional percentage of the vote they win in an state. Of course, this is just an estimate as some states, like California, allocate some of their delegates by congressional district which means the statewide vote may not exactly match up with the shares of delegates won by each candidate.
To give you a sense of how much more important the vote is in the big states compared to the medium-sized states consider the following example. In California, each additional percentage of the vote won by the candidates is worth 3.7 more delegates. In comparison, every 1% of the vote in Missouri is worth .72 delegates. If either Obama or Clinton win California by 3% tonight, the victory will net them about 11 delegates more than the other candidate. Yet, since the polling has been neck-and-neck there, the 3% win probably wouldn't be a big deal to the news media. However, in Missouri, the candidates would need to win by about 15% of the vote to get the same 11 delegate advantage from that state. Despite the fact that a 3% win in CA is equivalent to a 15% win in MO, you can bet that the media would make a much bigger deal of the 15% victory.
The other reason why this is important is because the candidates' home states of NY and IL are the 2nd and 3rd largest voting today. The media may not focus much attention on those states because they are not expected to be close and likely will not be. Yet, it makes a much bigger difference whether the candidates carry their own states by 70% rather than 66% than it does whether they win Connecticut by 2% or lose Connecticut by 2%. In the latter case, winning Connecticut by 2% rather than losing it by 2% will only translate into about a 2 delegate difference. On the other hand, if Obama wins IL by 70% rather than 66%, that would mean 6 additional delegates. If Clinton wins NY by 70% rather than 66%, that would mean 9 or 10 additional delegates.
Keep all of this in mind while you watch the returns tonight. And don't forget, check in to the blog as you watch those returns come in...I'll be posting early and often.
Monday, February 4, 2008
Live Blogging Super Tuesday
I will be live blogging (perhaps with some assistance from my CCPS colleagues) tomorrow night with news, commentary, and other thoughts as the results come in all through the night. I'll be armed with my laptop, a lot of caffeine, and a quick trigger on my remote control.
Republican Super Tuesday Delegate Estimate
I've spent a few hours going through the delegate allocation rules for each state posted on National Journal's site. There are about as many variations in delegate selection rules as there are states voting on Tuesday (that is a slight exaggeration, of course) which makes it nearly impossible to come up with any reasonable way of figuring out how many delegates each candidate will win on Tuesday. But I'll take a shot at a prediction anyway...
Let's start with the states with the most simple allocation plans. AZ, CT, DE, MO, NJ, NY, and UT each give all of their delegates to the winner of the statewide vote. McCain seems to have a safe lead in AZ, CT, NJ and NY, which would give him 230 delegates (50 from AZ, 27 from CT, 52 from NJ, and 101 from NY). UT is safely in the Romney column, which will give him 36 delegates. That leaves two winner-take-all states. In DE (18 delegates), the only poll taken recently indicates a 41% to 35% lead for McCain over Romney. In MO, the polls seem to indicate that either McCain, Huckabee, or Romney could win the state's 58 delegates, but McCain has a small lead in the pollster.com average, so let's give those to him.
MA allocates its 40 delegates proportionally according to the statewide vote. Based on current polls, Romney would receive about 25 of those delegates and McCain would receive 15.
TN allocates its delegates proportionally statewide and by congressional district unless someone receives 2/3 of the vote. Right now, polls put McCain at about 32%, Huckabee at about 25%, and Romney at about 22%. The number for Romney is significant since the threshold for qualifying for delegates is 20%. If we assume all three qualify and get a share of the vote proportional to what the polls show, then McCain would receive 21 delegates, Huckabee 16, and Romney 15.
Ok, so that was the relatively simple part, now come the complicated states. First, we have no polling for AK, AR, MT, ND, and WV (which is a convention), so we don't know who will win those 123 delegates. Second, CO, MN, and IL all appear to have events that don't really lead to any pledged delegates being selected, so I won't make any guesses to how that will turn out. That leaves us with AL, CA, GA, and OK. Each of these states allocates part of their delegates to the winner of the statewide vote, and the rest according to who wins the vote in each congressional district. But this mix varies significantly from state to state. While AL, GA, and OK, all award about half of their delegates to the winner of the state, CA only awards 11 delegates to the state winner and the rest are allocated by district. Since we don't have congressional district level polling, the estimates we create for these states are going to be even more subject to error. But I'll give it a try:
AL gives 24 delegates to the winner of the state and the remaining 21 go to winners of each congressional district. The polling shows McCain and Huckabee essentially tied (37%-35%) with Romney farther back (19%). If McCain wins, that gives him 24 delegates. Then let's assume that McCain wins 4 of the congressional districts and Huckabee wins three. That gives McCain another 12 delegates and Huckabee 9. The total then would be: McCain 36, Huckabee 9, Romney 0.
The polling in GA (30 statewide, 39 among districts) shows a three way split: McCain 32%, Romney 29%, and Huckabee 27%. Let's assume McCain wins the state and 5 of the congressional districts, with Romney and Huckabee each winning 4 congressional districts. The total would then be: McCain 45 delegates, Romney and Huckabee with 12 each.
The polling in OK (23 statewide, 15 among districts) shows McCain with 40%, Huckabee with 28%, and Romney with 21%. Given the way OK's districts are divided, I'd say that McCain would win the state and 3 districts, while Huckabee might pick up 2 districts. The total would then be: McCain 32, Huckabee 6, Romney 0.
Finally comes CA (11 statewide, 159 among districts) where polling shows a tight two-way race between McCain and Romney. Let's assume that McCain wins the state and one more district than Romney. That would give McCain 92 delegates and Romney 67.
So, before I add this all up, keep in mind that this estimate is likely to be much farther off than the Democratic estimate I calculated, largely because of how many of these delegates are distributed by winner-take-all rules at the congressional district level. But here is the very rough estimate of how things might turn out on the Republican side:
McCain would win 547 delegates, Romney would win 155, and Huckabee would capture 43 delegates, with another 261 delegates being determined in states where there is no polling or where the process should keep us from having a clear idea of how the delegate count might turn out. Even without the 261 delegates that we aren't sure about, that would get McCain about halfway to what he needs. Would that be enough to force Romney and Huckabee from the race?
Let's start with the states with the most simple allocation plans. AZ, CT, DE, MO, NJ, NY, and UT each give all of their delegates to the winner of the statewide vote. McCain seems to have a safe lead in AZ, CT, NJ and NY, which would give him 230 delegates (50 from AZ, 27 from CT, 52 from NJ, and 101 from NY). UT is safely in the Romney column, which will give him 36 delegates. That leaves two winner-take-all states. In DE (18 delegates), the only poll taken recently indicates a 41% to 35% lead for McCain over Romney. In MO, the polls seem to indicate that either McCain, Huckabee, or Romney could win the state's 58 delegates, but McCain has a small lead in the pollster.com average, so let's give those to him.
MA allocates its 40 delegates proportionally according to the statewide vote. Based on current polls, Romney would receive about 25 of those delegates and McCain would receive 15.
TN allocates its delegates proportionally statewide and by congressional district unless someone receives 2/3 of the vote. Right now, polls put McCain at about 32%, Huckabee at about 25%, and Romney at about 22%. The number for Romney is significant since the threshold for qualifying for delegates is 20%. If we assume all three qualify and get a share of the vote proportional to what the polls show, then McCain would receive 21 delegates, Huckabee 16, and Romney 15.
Ok, so that was the relatively simple part, now come the complicated states. First, we have no polling for AK, AR, MT, ND, and WV (which is a convention), so we don't know who will win those 123 delegates. Second, CO, MN, and IL all appear to have events that don't really lead to any pledged delegates being selected, so I won't make any guesses to how that will turn out. That leaves us with AL, CA, GA, and OK. Each of these states allocates part of their delegates to the winner of the statewide vote, and the rest according to who wins the vote in each congressional district. But this mix varies significantly from state to state. While AL, GA, and OK, all award about half of their delegates to the winner of the state, CA only awards 11 delegates to the state winner and the rest are allocated by district. Since we don't have congressional district level polling, the estimates we create for these states are going to be even more subject to error. But I'll give it a try:
AL gives 24 delegates to the winner of the state and the remaining 21 go to winners of each congressional district. The polling shows McCain and Huckabee essentially tied (37%-35%) with Romney farther back (19%). If McCain wins, that gives him 24 delegates. Then let's assume that McCain wins 4 of the congressional districts and Huckabee wins three. That gives McCain another 12 delegates and Huckabee 9. The total then would be: McCain 36, Huckabee 9, Romney 0.
The polling in GA (30 statewide, 39 among districts) shows a three way split: McCain 32%, Romney 29%, and Huckabee 27%. Let's assume McCain wins the state and 5 of the congressional districts, with Romney and Huckabee each winning 4 congressional districts. The total would then be: McCain 45 delegates, Romney and Huckabee with 12 each.
The polling in OK (23 statewide, 15 among districts) shows McCain with 40%, Huckabee with 28%, and Romney with 21%. Given the way OK's districts are divided, I'd say that McCain would win the state and 3 districts, while Huckabee might pick up 2 districts. The total would then be: McCain 32, Huckabee 6, Romney 0.
Finally comes CA (11 statewide, 159 among districts) where polling shows a tight two-way race between McCain and Romney. Let's assume that McCain wins the state and one more district than Romney. That would give McCain 92 delegates and Romney 67.
So, before I add this all up, keep in mind that this estimate is likely to be much farther off than the Democratic estimate I calculated, largely because of how many of these delegates are distributed by winner-take-all rules at the congressional district level. But here is the very rough estimate of how things might turn out on the Republican side:
McCain would win 547 delegates, Romney would win 155, and Huckabee would capture 43 delegates, with another 261 delegates being determined in states where there is no polling or where the process should keep us from having a clear idea of how the delegate count might turn out. Even without the 261 delegates that we aren't sure about, that would get McCain about halfway to what he needs. Would that be enough to force Romney and Huckabee from the race?
Super Tuesday Democratic Delegate Count Estimate (2/4 Update)
I've updated my estimates of how the Democratic delegates will be divided by Clinton and Obama on Super Tuesday. Please refer to my original post for the many reasons why this is a VERY rough estimate. In the last version, I used the most recent poll I could find in each state. Since several polling firms have released very recent polls in many states (like CA, MO, and GA), I'm using Pollster.com's current average for 11 states, rather than choose one polling firm over the other (keep in mind, however, that if there is a strong surge toward one candidate or the other in the closing days, it may not show up in the average as clearly). In a handful of states where there has been little polling, I just use the most recent survey. As with the last estimates I produced, if there has been no polling in a state over the past few weeks, I simply put the delegates from that state in the "unclear" column.
One other thing to keep in mind, polling error in large states like CA, IL, or NY would affect the delegate total significantly more than polling error in AL or OK, simply because of the number of delegates at stake in the larger states. For example, the pollster.com average for CA shows a 44-37 Clinton lead, but some polls taken in the last few days have shown the race even. Depending on which estimate is correct, it could mean a significant difference in delegates.
Based on the estimates from the polls, Clinton would win 821 delegates on Super Tuesday, Obama would win 746, and 122 additional delegates would be decided in states where no polling is available. Thus, if the survey results are correct, Clinton would come out of Super Tuesday with about a 50 delegate lead (not including super delegates).
I will try to post an estimate/prediction for Republicans later this evening.
CORRECTION: I mistakenly swapped Obama and Clinton's numbers in IL in my first posting and have now corrected the error (thanks to Dan for spotting the mistake).
One other thing to keep in mind, polling error in large states like CA, IL, or NY would affect the delegate total significantly more than polling error in AL or OK, simply because of the number of delegates at stake in the larger states. For example, the pollster.com average for CA shows a 44-37 Clinton lead, but some polls taken in the last few days have shown the race even. Depending on which estimate is correct, it could mean a significant difference in delegates.
Based on the estimates from the polls, Clinton would win 821 delegates on Super Tuesday, Obama would win 746, and 122 additional delegates would be decided in states where no polling is available. Thus, if the survey results are correct, Clinton would come out of Super Tuesday with about a 50 delegate lead (not including super delegates).
I will try to post an estimate/prediction for Republicans later this evening.
CORRECTION: I mistakenly swapped Obama and Clinton's numbers in IL in my first posting and have now corrected the error (thanks to Dan for spotting the mistake).
Sunday, February 3, 2008
Democratic Super Tuesday Delegate Count Estimate
Ok, so this is really a fool's errand. First, we don't have any polls for a handful of states. Second, in the states where we do have polls, they are often relatively inconsistent. For example, of a handful of polls taken in NJ over the past few days, pollsters are reporting anything from a 43-42 (Zogby) Clinton lead to a 51-39 Clinton advantage (Survey USA). Third, and perhaps most importantly, my methodology really over-simplifies how delegates are allocated in each state. Nevertheless, I took some time during the slow first half of the Super Bowl (but, wow, what an ending!) to come up with a rough calculation of how Clinton and Obama would divide the delegates if the polling is correct.
I began with Rasmussen polls, mostly because they covered the most states (11). Then I tried to fill in other other polls where they were available and had been conducted within the last week. (The only exception is Colorado, where I used a poll taken 1/20-1/21). There were six states (and Democrats Abroad) where I could not locate any 2008 polling.
I used the percentage each candidate received in each state's poll to then allocate that state's delegates. I did this by calculating each candidate's share of the two-candidate vote. So, in the latest Alabama survey, 46% supported Clinton, 41% supported Obama, and 13% went to undecided or other candidates. However, that 13% will likely end up with Clinton and Obama on election day, so my transformation gives Clinton 53% of the two-candidate vote and Obama receives 47%. Based on these percentages, I simply allocate each state's delegates proportionally. Once again, this is not exactly how it is done in most states, but this is just one of many sources of error with my estimate and all the more reason to not take this estimate all that seriously. That said, my calculations produce the following estimate:
Based on these calculations, Clinton would win 784 delegates on Super Tuesday, Obama would win 757, and there are 148 delegates in states where we have no polling whatsoever. If you allocate the 148 evenly between the candidates, it would mean that Clinton would come out of Super Tuesday with a very small delegate lead.
I began with Rasmussen polls, mostly because they covered the most states (11). Then I tried to fill in other other polls where they were available and had been conducted within the last week. (The only exception is Colorado, where I used a poll taken 1/20-1/21). There were six states (and Democrats Abroad) where I could not locate any 2008 polling.
I used the percentage each candidate received in each state's poll to then allocate that state's delegates. I did this by calculating each candidate's share of the two-candidate vote. So, in the latest Alabama survey, 46% supported Clinton, 41% supported Obama, and 13% went to undecided or other candidates. However, that 13% will likely end up with Clinton and Obama on election day, so my transformation gives Clinton 53% of the two-candidate vote and Obama receives 47%. Based on these percentages, I simply allocate each state's delegates proportionally. Once again, this is not exactly how it is done in most states, but this is just one of many sources of error with my estimate and all the more reason to not take this estimate all that seriously. That said, my calculations produce the following estimate:
Based on these calculations, Clinton would win 784 delegates on Super Tuesday, Obama would win 757, and there are 148 delegates in states where we have no polling whatsoever. If you allocate the 148 evenly between the candidates, it would mean that Clinton would come out of Super Tuesday with a very small delegate lead.
NOTE: I will update these estimates with the most recent polling on Monday night.
NOTE 2: The estimates are a bit more difficult to create for Republicans since the rules differ so much from state to state. However, if I can, I'll try to produce a Republican estimate as well this evening.
NOTE 3: In the original post, I mistakenly transposed Clinton and Obama's numbers in Illinois, but have fixed it here and in the updated post here.
48 Hours: Last Minute Blitz from Obama and Clinton
The Obama campaign is apparently airing an ad in 24 states during tonight's Super Bowl. You can see that ad here.
There is also an interesting video that first popped up yesterday and is now furiously circulating the web. It puts Obama's New Hampshire concession speech to song and features several celebrities. If you are interested, check it out here:
In the meantime, Clinton is holding a virtual town hall meeting on Monday evening, an hour of which will be aired on the Hallmark channel.
The New York Times is reporting that the candidates have spent at least $19 million combined on these ad blitzes. How much will this work for either candidate?
Could there by any surprises in store for the last 48 hours before the polls close?
Will the Super Bowl keep our minds off of politics for a few hours?
There is also an interesting video that first popped up yesterday and is now furiously circulating the web. It puts Obama's New Hampshire concession speech to song and features several celebrities. If you are interested, check it out here:
In the meantime, Clinton is holding a virtual town hall meeting on Monday evening, an hour of which will be aired on the Hallmark channel.
The New York Times is reporting that the candidates have spent at least $19 million combined on these ad blitzes. How much will this work for either candidate?
Could there by any surprises in store for the last 48 hours before the polls close?
Will the Super Bowl keep our minds off of politics for a few hours?
Dems in a Dead Heat: Gallup Tracking Poll
Gallup has released its latest national daily tracking poll. As always, the results include interviews conducted over the previous three days. Yesterday, I posted about a bump in Clinton's numbers, but that has largely disappeared today with Obama up 3% and Clinton down 2%, which means that the two candidates are in a statistical dead heat nationwide.
There are a ton of new state polls out and a lot of them lead to different conclusions. It looks like Clinton and Obama will each win their home states fairly easily. There are several that are clearly leaning one way or another (GA for Obama, OK for Clinton, for example), several that are neck-and-neck (such as CA, MO, CT), and then several states where we really have no polling to look at (ID, MT, and AK, for instance).
I'll try to post more on the state-by-state polls later, particularly if the Pats are winning in a blowout.
Saturday, February 2, 2008
Who Benefits from Edwards' Withdrawal? (Volume II)
The newest Gallup tracking poll is out. This is the first of their tracking polls including only interviews conducted after the Edwards announcement (interviews conducted Jan. 30-Feb. 1). I've added a line to show the last day that Edwards was in the race.
Before Edwards exited, his support stood at 12%, Clinton was at 42%, and Obama was at 36%. Since then, Obama has gained 5% and Clinton has gained 6%. Given that the margin of error for the survey was 3%, we can't be confident that either candidate did better since Edwards left. The other problem is that the movement in the candidates' numbers could be chalked up to anything else that has happened since Edwards left, such as Thursday's debate. In fact, given that Clinton picked up only 1% on Wednesday and Thursday but 4% on Friday, does this actually reflect that her bump can be attributed to her performance in the debate? If her numbers continue to go up over the next two days, then that may provide more evidence for the latter explanation.
I didn't have time to download the Republican figure, but McCain's lead is up to 20% over Romney (44%-24%, with Huckabee receiving 16%).
Before Edwards exited, his support stood at 12%, Clinton was at 42%, and Obama was at 36%. Since then, Obama has gained 5% and Clinton has gained 6%. Given that the margin of error for the survey was 3%, we can't be confident that either candidate did better since Edwards left. The other problem is that the movement in the candidates' numbers could be chalked up to anything else that has happened since Edwards left, such as Thursday's debate. In fact, given that Clinton picked up only 1% on Wednesday and Thursday but 4% on Friday, does this actually reflect that her bump can be attributed to her performance in the debate? If her numbers continue to go up over the next two days, then that may provide more evidence for the latter explanation.
I didn't have time to download the Republican figure, but McCain's lead is up to 20% over Romney (44%-24%, with Huckabee receiving 16%).
Friday, February 1, 2008
New Gallup Tracking Poll: Democratic Race Continues to Tighten and Some Evidence of the Effect of Edwards Withdrawal
Today's national Gallup tracking polls are out and they show the race continuing to tighten on the Democratic side and continuing to favor McCain among Republicans. Clinton's lead is now down to 3% nationwide, which is the margin of error for the survey. The most recent survey included interviews conducted the day before, day of, and day after the Edwards' announcement. Before his announcement, Edwards support stood at 12%, it is now down to 4% and will likely be gone altogether by tomorrow's poll (which will only include interviews conducted after his withdrawal). Neither Clinton nor Obama seems to have gotten all of this support, with both seeing relatively similar upticks in their percentage of support since his withdrawal (the changes are so small that it is impossible to say with any statistical confidence that one candidate has done better than the other since Edwards withdrew).
McCain continues to hold a 15% lead over Romney in the tracking polls. McCain's support has also increased signifcantly more than Romney's since his victory in Florida and subsequent endorsement from Giuliani.
McCain continues to hold a 15% lead over Romney in the tracking polls. McCain's support has also increased signifcantly more than Romney's since his victory in Florida and subsequent endorsement from Giuliani.
Pew Report: Evidence on the Effects of Cell Phone Use on Polling Accuracy
The Pew Research Center for the People and the Press has released an interesting new report addressing the effects of increasing cell phone use on the quality of telephone surveys. The concern has been that as more Americans rely more exclusively on cell phones (which are unreachable by traditional telephone survey methods), the representativeness of survey responses are declining significantly.
Pew used a survey of landlines compared with a survey they conducted of cell phone numbers. While they found some demographic differences between the two groups, particularly with regard to age, they noted that "When data from both samples are combined and weighted to match the U.S. population on key demographic measures, the results are virtually identical to those from the landline survey alone." Pollsters using traditional dialing methods have often argued that they can minimize the concerns about cell-phone-only households by weighting their samples accordingly. The Pew report provides support for this assertion:
"On key political measures such as presidential approval, Iraq policy, presidential primary voter preference, and party affiliation, respondents reached on cell phones hold attitudes that are very similar to those reached on landline telephones. Analysis of two separate nationwide studies shows that including interviews conducted by cell phone does not substantially change any key survey findings."
There is a lot of information in the Pew report, and it is a much read for those interested in survey research. Check out the full report here.
Pew used a survey of landlines compared with a survey they conducted of cell phone numbers. While they found some demographic differences between the two groups, particularly with regard to age, they noted that "When data from both samples are combined and weighted to match the U.S. population on key demographic measures, the results are virtually identical to those from the landline survey alone." Pollsters using traditional dialing methods have often argued that they can minimize the concerns about cell-phone-only households by weighting their samples accordingly. The Pew report provides support for this assertion:
"On key political measures such as presidential approval, Iraq policy, presidential primary voter preference, and party affiliation, respondents reached on cell phones hold attitudes that are very similar to those reached on landline telephones. Analysis of two separate nationwide studies shows that including interviews conducted by cell phone does not substantially change any key survey findings."
There is a lot of information in the Pew report, and it is a much read for those interested in survey research. Check out the full report here.
Subscribe to:
Posts (Atom)
