The Dramatic Erosion of the Republican Brand

There has been a lot of discussion recently about how the Republican brand is in trouble. There are a variety of ways of quantifying what this means--fewer are identifying themselves with the Republican Party and Bush's approval is at historic lows--but one important thing that brands are meant to engender is trust. In politics, a party brand is strongest when citizens trust that party to handle a wide array of issues facing the country. The brand is weakest when there is little trust in the brand.

Generally, citizens trust one party to handle some issues while they have more confidence in the other party on another set of issues. For example, over the last few decades, Republicans have been more trusted on issues involving taxes and national security while Democrats have fared better on domestic policy issues such as health care, social security, and education. However, as the figure below indicates, Republicans are no longer trusted more than Democrats on any of these issues. (This figure was based on a national ABC News/Washington Post survey conducted in early February.)


As the figure shows, Democrats are far more trusted by the public on issues such as the economy, the budget deficit, and health care. But Democrats also now have a narrow edge on issues Republicans have traditionally owned, including taxes, the conflict in Iraq, and even the campaign against terrorism.

The decreasing trust in the Republican Party (and increasing trust in Democrats) on Iraq and terrorism has been one of the most notable changes over the past six years. The figures below track the public's trust of the parties on these two issues since 2002. Note that at the beginning of the period, Republicans held about a 20% advantage over Democrats on these issues. However, in the course of just a few years, Democrats are now significantly more trusted on handling the situation in Iraq and slightly more trusted on handling terrorism.


Last week, the Bush and McCain went up against Obama on national security issues, something that Republicans have consistently used to their advantage over Democrats in the past. But the shifts I've outlined below suggest that these themes may not be as successful as they once were. The public no longer trusts Republicans substantially more than Democrats on these issues. On the other hand, Republicans have to choose from a bad lot when it comes to the issues they stress. After all, they no longer seem to have a definitive advantage on any issues. And, as the figure below shows, the other major issue in this campaign, the economy, strongly favors Democrats at an even greater rate than it did in 1992 when George H.W. Bush lost his bid for a second term.


Ultimately, the erosion of the Republican brand has been dramatic. In the current climate, there are no strong issue areas for Republican candidates (presidential or otherwise) to focus on. Rather, there are simply some issues that are not as bad as others. Currently, it appears as though terrorism may remain a target issue for Republicans if only because their brand isn't as weak on this issue as it is on most others now.

Comparing the Obama Campaign's Predictions to Reality

As I noted in an earlier post, we can gain some interesting insight from comparing how the Obama campaign thought they were going to perform in states to how they actually did. It is not all that often that you get such a clear sense of how a campaign thinks it is going to do in a series of primaries, but thanks to a memo the Obama campaign accidentally sent a reporter for Bloomberg, we can do precisely that.

The figure below plots the percentage of the vote that the Obama expected to get in each state (based on their memo released just after Super Tuesday) against the percentage of the vote Obama actually received. States falling along the line would be those in where Obama did just as his campaign expected him to do. States falling above the line are those in which Obama's actual voted total exceeded his campaign's expectations, and those below are where he failed to meet those expectations.
Several interesting patterns are evident from this plot. The most obvious is that Obama over-performed in every primary and caucus held between Super Tuesday and March 4th. This is the period during which Obama won the nomination by accumulating a much bigger delegate lead than even they expected, one that became so insurmountable that the Clinton campaign was never able to recover. It has often been repeated by reporters and pundits that the Clinton campaign was not prepared with a post-Super Tuesday strategy, and Obama's over-performance in these is strong evidence of this dynamic.

The second point that stands out from the figure is that Obama has under-performed (relative to his campaign's expectations) in seven of the 13 states with caucuses or primaries on or after March 4th. He did just as his campaign expected in TX and MS, and better than expected in VT, WY, NC, and OR. Once again, this very much reflects the dynamics of the race, with Obama suffering from the Jeremiah Wright remarks and the general scrutiny involved with being the nominee while Clinton regained her footing on March 4th and began a relatively effective two months of campaigning as the front-runner turned underdog.

Finally, on the "under-performing" side of the line, note WV and KY as the real outliers. Why did the Obama campaign think that they would do about 10-15% better in these states than they actually did? Did the underestimate the role that race would play in these primaries? Did they not anticipate Clinton's increased appeal to working class whites? Did the Jeremiah Wright hurt them more in those states than in other places?

By the way, just for the record, the Obama campaign predicted that they would have accumulated 1,605 pledged delegates by today compared to 1,536 for Clinton, a lead of less than 80 delegates. Instead, Obama's lead in pledged delegates is nearly twice as large.

Assessing the Accuracy of the Oregon/Kentucky Predictions: The Big Winner was Pollster.com

While there are still votes to be counted in Oregon, it looks like Obama's margin in that state will be 18%, while Clinton won Kentucky by 35%. So, who came closest to getting the margins correct in both states? The figure below plots the predictions from three survey firms that released polls from both states during the last week of the campaign, along with Poblano's predictions based on demographic models and Pollster.com's poll averages.


While each of these organizations underestimated the size of the Obama margin in Oregon, Pollster.com came closest to getting the results correct in both states. The pollster.com averages had Clinton winning by 35% in Kentucky (which she did) and Obama winning by 12% in Oregon. Among the individual pollsters, it appears as though Survey USA came closest on the pair of margins. While American Research Group was closets on Oregon (they predicted a 36% margin for Clinton), they greatly underestimated Obama's margin of victory in Oregon. Survey USA came closest on the Oregon margin and were only off on the Kentucky margin by 4%.

Unlike two weeks ago, when he nailed the NC and IN outcomes, Poblano's model was the farthest off last night. To his credit, Poblano did state that he thought his predictions for Kentucky were likely to be understating Clinton's support. He was right, he missed that margin by 16%. Poblano's model did come much closer on Oregon.

UPDATED: Post was updated at 2pm on May 21st to reflect the fact that Obama's lead in Oregon is now 18%.

Kentucky/Oregon Primaries Live Blog

12:46am: Well, as promised, that was a watered-down version of the traditional primary night live blog. Frankly, I'm not sure what else there is to say at this point. This is getting a bit redundant. Obama struggles in the Appalachia region (WV, KY, central PA, southern OH, etc.) but there is no way he isn't going to be the nominee at this point. Frankly, anyone who was really doing the math has known since the end of February that Clinton was facing a very difficult climb; it is no surprise that she fell short.

The only truly interesting thing is to see how Clinton exits. My guess is that FL and MI get half their delegates back when the Rules and Bylaws Committee meets on May 30th. Even when that happens, Obama will still have a majority of the pledged delegates and will not be too far off of the overall delegate majority. In the meantime, I'd expect to see a fair number of superdelegates coming Obama's way in the next, many of whom may make the the point that he was the pledged delegate winner so he should be the nominee. (Not to mention that Chuck Todd just made the point that a majority of the uncommitted superdelegates are from states Obama won). Then, on the night of June 3rd, after the final votes are cast, I'd expect a gracious concession speech from Clinton and an equally gracious victory speech from Obama. But that is just a guess. The only thing I know for sure is that this has been a crazy ride (and a political scientist's dream, in many ways). It is kinda sad that it is winding down.

11:13pm: Democratic Convention Watch points out that not only has Obama won the pledged delegate majority if you do not include FL and MI, but he has also clinched a majority of pledged delegates under almost any compromise that is likely to pass out of the Rules and Bylaws Committee at the end of the month. The only way he hasn't clinched it is if you seat MI and FL delegates "as is." There is no way that the DNC is going to do that and allow MI and FL to send delegates without any penalty.

11:07pm: The "exit poll" out of Oregon is showing a 56-42% Obama margin. If that is correct, then the pollster.com average will have nailed the margins in both states almost perfectly. But let's see if that is really the Oregon margin.

11:00pm: Nevermind on that whole "gun shy" thing. They called for Obama right away.

10:57pm: By the way, there will be six add-on superdelegates chosen between today and June 3rd. All six come from states Obama won and will, therefore, likely be Obama supporters. If that is the case, then Obama need only about 20 more superdelegates endorsements to be able to claim an overall delegate majority on June 3rd (though the DNC's Rules and Bylaws Committee may change the magic number when it rules on the FL and MI challenge on May 30th).

10:50pm: Ten minutes from the Oregon "polls closing." Will Obama get an immediate call there? The difficult part of calling Oregon is that there isn't really exit polling. Rather, the "exit poll" is just another poll that has been conducted over several days leading up to the election. May mean that the networks are slightly more gun shy on the call.

10:45pm: Interestingly, CNN and MSNBC seem to have slightly different pledged delegate counts. CNN says he has already captured a majority of pledged delegates while MSNBC says he is one shy and will have to wait until the Oregon delegates are allocated.

10:07pm: Democratic Convention Watch lists superdelegates who are members of the "Pelosi Club." This refers to a group of superdelegates who have said that they will vote for whichever candidate wins the majority of pledged delegates. The question is whether the "Pelosi Club" will make their endorsements tomorrow, now that Obama has clinched this metric. My guess is that, at the very least, its namesake will not endorse until after June 3rd.

9:58pm: Votes won't be reported out of Oregon for another hour still, but the Obama campaign has sent out an email with the following:

"The polls are closed in Kentucky and votes are being counted in Oregon, and it's clear that tonight we have reached a major milestone on this journey.

We have won an absolute majority of all the delegates chosen by the people in this Democratic primary process."

9:55pm: Clinton's victory in Kentucky is certainly large in terms of sheer numbers. Although the state is much smaller and has fewer delegates, the number of votes she will win by is larger than her margin in Pennsylvania and Ohio.

7:47pm: This post is about right. I would add that the Clinton campaign could have said similar things at any point over the last few months and been very wrong. The crystal clear pattern throughout has been that the later a superdelegate has made an endorsement, the more likely that endorsement was for Obama.

7:39pm: Check out the Clinton margins in the counties in eastern Kentucky. She is winning a lot of those counties by a margin of 85-10% or more.

7:30pm: As expected, Clinton will win Kentucky by a significant margin. I'd imagine that her margin will continue to grow as the western part of the state comes in.

6:52pm: I probably won't be blogging as much as usual tonight for two reasons. First, we pretty much know who the nominee is going to be, so the only thing that is really left to determine is how this is going to end, not what the ending will be. Second, there are four hours between when the Kentucky polls close (7pm) and when the Oregon "polls close" at 11pm (I put that in quotes since this is an all mail election). I doubt there will be an abundance of news to discuss during these four hours.

I assume that the networks will be calling Kentucky right at 7pm. If not, then that will be the first surprise of the evening.

Oregon and Kentucky Delegate Estimates and an Update on the End Game

Another Tuesday brings another pair of primaries. And it appears as though it will also bring another split, with Obama holding a strong lead in the polls out of Oregon while Clinton holds a commanding lead in the polling out of Kentucky. Three pollsters (Survey USA, ARG, and Suffolk) have conducted surveys in both Kentucky and Oregon over the past several days. The figure below plots the margins these polls found for Obama in both states along with the Pollster.com averages (the margin is just the percent for Obama minus the percent for Clinton). (UPDATE: I have now added Poblano's predictions to the figure). ARG has the best case scenario for Clinton--a margin of greater than 35% in Kentucky and only a narrow win for Obama in Oregon. Survey USA predicts a bigger (double-digit) margin for Obama in Oregon, as does the Pollster.com average.


As always, I will use the pollster.com averages to generate predictions for today's primaries. Because of the enormous margin that Clinton has in Kentucky, she is poised to pick up a net gain of more than 19 pledged delegates in that state. Obama will only be able to partially off-set Clinton's Kentucky gains with his win in Oregon, meaning that Clinton should gain about 13 delegates on Obama tonight.

Despite the fact that Clinton will win more delegates on Tuesday night, the 45 delegates that Obama is poised to win should be more than enough to allow him to clinch a majority of elected delegates. According to the Democratic Convention Watch site, Obama currently holds 1,612.5 pledged delegates, just 15 shy of a pledged delegate majority (a benchmark I predicted Obama would reach on this very date back in March).

But what about the magic number of 2,025 total delegates? Well, Obama is not too far from that benchmark either. If you add the 302.5 superdelegates that Democratic Convention Watch tallies for Obama, then he is just 110 delegates short of 2,025. If he captures 45 delegates in tomorrow night's primaries, that will put him within 65 of a majority. However, as the figure below shows, there aren't quite enough pledged delegates left to get him there.


Based on the only poll out of Puerto Rico (showing Clinton with a 50-37% lead), Obama would pick up 23 delegates in that primary on June 1st. I have not seen a poll out of Montana, so let's split those 16 delegates evenly, giving Obama 8. In South Dakota, the only poll I've seen puts Obama up 46-34%, which would give him 9 more delegates for a total of 17 on June 3rd.
If not a single superdelegate endorses Obama between now and June 3rd, he would fall 25 delegates shy of the magic number when all the voting ended. The question remains, can the Obama campaign line up enough superdelegates so that the pledged delegates from the June 3rd primaries are the ones that put him over the top for the nomination? Even though we know who is going to win, it should be interesting to see how it ends.

Is Obama Relying on Late Ballots in Oregon?

Oregon's vote-by-mail system may make polling a little more tricky close to election day. After all, a substantial portion of the electorate have already cast their ballots, while there are many who wait until just before the election to vote.

On this front, one interesting finding stands out when it comes to the surveys out of Oregon over the past week or so. According to polls from Survey USA and American Research Group, Obama has a significant lead in Oregon among those who intend to vote, but he is in a dead heat with Clinton among those who say they have already voted.

In the Survey USA poll (conducted 5/9-11), 43% reported that they had already mailed in their ballots and Obama's lead in this group was just 49-48% (well within the margin of error). On the other hand, among those who intended to vote but had not yet done so, Obama a 58-38% lead. Similarly, the American Research Group survey (conducted 5/14-15) indicated that Obama was tied with Clinton at 49% among the 58% of Oregon Democrats who already mailed in their ballots (the higher figure for those who had already voted makes sense given that the poll was conducted a few days later than the Survey USA poll). Among those who intended to vote but had not yet done so, Obama led 52-40%. Thus, if these two polls are correct, Obama and Clinton are in a dead heat among the votes that are already in and Obama is relying on those who have not yet cast their ballots to generate the big margin he is expected to win by.

Two things are notable about these findings from those who have already voted. First, Obama has generally performed better among early voters in other states, which might have led us to expect him to over-perform rather than under-perform among those who had already mailed in their ballots. Second, Clinton is the candidate who has generally done better among Democrats who made their decisions in the last few days of the campaign, yet she appears to be doing better among those who have already cast their ballots.

Survey USA is set to release a new poll out of Oregon on Monday. It will be interesting to see whether these patterns persist in their last poll before election day. Stay tuned...

UPDATE (5/19, 3:00pm): Survey USA has now released their final Oregon poll (interviews conducted 5/16-5/18). It shows that 77% of Oregon Democrats have now cast their ballots and Obama now has a significant 53-44% lead among this group. Of the 23% who had not yet cast their ballots but intended to, Obama led 62-34%. The size of Obama's victory in Oregon will likely depend on how many of those 23% actually do get their ballots in on time.

The figure below shows the amount of support Obama and Clinton have in both polls from those who had already voted and those who intended to vote:


UPDATE 2 (5/19, 11:55pm): Public Policy Polling also released a survey today showing Obama with a comfortable lead among both those who had already voted and those who had not yet mailed in their ballots but intended to do so. Thanks to "x curmudgeon" for pointing out the poll.

The Mississippi Special Election: Outlier or Continuation of a Trend?

Republicans certainly seemed a bit rattled after they lost their bid to retain a seat that they had previously held in Mississippi's first congressional district. Part of the reason for being so worried is that the district they lost on Tuesday was not just one that they had previously held, but also one that Bush had carried with over 62% of the vote in 2004. But was this just an outlier, or a continuation of what happened with Republicans in 2006?

The figure below plots each congressional district contested in 2006 based on the percentage of the vote Bush won in the district in 2004 and the percentage of the vote the Republican candidate won in the district in 2006. The plot is divided into quadrants. Districts in the top left quadrant are those where the Republican House candidate won a majority despite the fact that Bush did not carry the district in 2004; districts in the bottom right are those where Republican candidates did not win a majority of the vote despite the fact that Bush did carry the district in 2004.

Note that very few districts fall in the top quadrant as very few Republicans over-performed in 2006. Far more of the districts fell in the lower right quadrant with many Republican candidates losing districts that Bush carried in 2006. I added the Mississippi special election outcome to this plot to give some perspective of the extent to which this result followed what happened in 2006. The district also falls in the lower right hand quadrant, in close proximity to dozens of other districts where Republican candidates under-performed.

Of course, many of those districts in the lower right-hand quadrant are those where Democratic incumbents were running, therefore making it more difficult for the Republican candidates. Therefore, it is also instructive to look just at open seat contests, which the figure below does.

Note that when it came to open seats, not a single Republican candidate carried a district that Bush did not win in 2004. On the other hand, more than a half-dozen districts were won by Democrats in 2006 despite the fact that Bush had carried those districts two years earlier. The Mississippi special election, which was also an open seat contest, stands out as a bit more of an outlier here. There were only three open seat contests in 2006 where the Republican candidate under-performed by as much as Greg Davis did in the Mississippi 1st district this week.

Thus, the special election isn't a dramatic outlier when compared to other House elections in 2006. However, the result does suggest that prospects are not improving for Republicans; indeed, they may very well be getting worse. This is particularly true when you consider that Republicans are going to be defending a lot more open seats in 2008 than they were in 2004.

For the most recent ranking of House races by the Cook Political Report, check out this helpful chart.

Are Superdelegates Afraid of Obama at the Top of the Ticket?

If you watch too much cable news coverage of the campaign (as I do) then you'll notice that one theme the pundits like to return to repeatedly is the question of whether Democratic office holders are worried that Obama will be a drag at the top of the ticket. Yesterday's special election victory for Travis Childers (D) in Mississippi seemed to temper these arguments somewhat, but it is surely something that the Clinton campaign hopes is weighing on the minds of superdelegates. After all, most of the superdelegates have to win reelection in the fall and who the Democratic nominee is may affect their prospects. Superdelegates from states and districts that tend to be more Republican may be particularly concerned with which Democratic presidential candidate their opponent will attempt to tie them to during the general election campaign.

So who do these red state/district superdelegates see as the safer choice? Well, based on those who have already declared their support, it appears as though Obama's superdelegates actually come from slightly more Republican areas than those who have endorsed Clinton. The figure below shows the percentage of the vote that Bush won in the state (or district for House members) of superdelegates who have already endorsed. In each category, Obama's superdelegates, on average, come from areas where Bush fared better in 2004 compared to Clinton's supers.

The same general pattern holds when you look at who superdelegates from strongly Republican states or districts are endorsing. The figure below presents this information. Of the senators and governors who come from states that Bush carried with more than 55% of the vote in 2004, Obama has received 17 endorsements from this group while Clinton has received the support of 12 of these superdelegates. Obama also leads among House members from districts that went 55% or more for Bush in 2004.

However, what might be most notable is that of the 28* House member superdelegates who come from districts where Bush won at least 55% of the vote in 2004, only 11 have endorsed either Obama or Clinton. The remaining 17 in this group are still undeclared. Perhaps they have decided that the safest thing for their reelection prospects is to avoid being tied to either candidate for as long as possible.

* This does not include the new House member from Mississippi, Travis Childers.

Pledged Democratic Delegate Projections: West Virginia and Beyond

Even though the race for the Democratic Nomination is winding down, there are still 6 primaries left to contest. While there hasn't been the same abundance of polling in the remaining states as we've seen in places like Pennsylvania, Ohio, Texas, and North Carolina, there has been at least one recent poll in 5 of those states. Based on the polls, today's primary in West Virginia looks to be a blowout in Clinton's favor. If you average the four surveys released in May from that state, you get a 61-24% advantage for Clinton. If you allocated the pledged delegates accordingly, Clinton would receive 20 delegates compared to 8 for Obama (a net gain of 12).

You can see the estimates for West Virginia and the other 5 states here:

On May 20th, the polls estimate that Obama and Clinton will split Oregon and Kentucky, with Clinton taking 3 more pledged delegates than Obama. The one state without a poll is Montana, where Obama is expected to fare well. Overall, Clinton should pick up between 122 and 124 of the remaining pledged delegates while Obama should accumlate between 93 and 95.

So, how close is Obama to actually hitting that magic number? The figure below starts with Obama's delegate (both elected and superdelegates) support as of this morning and then adds to that total the projected elected delegates he would pick up in each state (assuming an 8-8 split in Montana). As I estimated back in March, May 20th is the magic date when Obama will have a majority of the elected delegates. If he doesn't pick up one more superdelegate between now and May 20th (which is obviously very unlikely), he will be about 100 delegates shy of the 2025 figure he needs to clinch the nomination (assuming you don't include Michigan and Florida). If he didn't pick up a single superdelegate between now and June 3rd, he would end up just 66 delegates shy of the magic number.


Even though Obama has picked up at least 25 superdelegates in the past week, it seems unlikely that another 100 would endorse between now and next Tuesday. Nevertheless, I'm sure his campaign is working hard to accumulate enough superdelegates so that he will clinch the nomination after winning a state's primary rather than doing so when some unknown DNC official pledges his/her support on some random Thursday in June. It will be interesting to see how this all plays out, even if we do already know who the winner is going to be.

The End of Our Superdelegate Predictions?

I got an email over the weekend that made me laugh:

"You have done great work with your superdelegate predictions. However, I think I have a model that I would bet a buck will outperform your model going forward: Model definition: if (superdelegate), then prediction = Obama"

The email was funny, particularly since I had already been mulling over whether to continue generating predictions for this very reason. Since Obama's big win in North Carolina, 21 previously unpledged superdelegates have made endorsements (not including add-ons). You can see the names along with our predictions here:


As you can see, we've gotten 17 out of these 21 endorsers (81%) correct. This would generally be very good for the model, which has usually hovered around a 70% accuracy rate since we started generating the predictions in early February. However, 81% isn't so good once you realize that if somebody had guessed that everyone was going to endorse Obama, he/she would have gotten 18 right (86%). This clearly reflects the status of the Democratic nomination race at this point: it doesn't take a statistical model to estimate what the remaining undeclared superdelegates will do at this point.

Just for fun, I did add the newly committed superdelegates to our dataset along with a variable capturing the change in dynamics after the May 6th primaries. The predictions generated by the model appear to largely match reality. There are only 31 superdelegates left on the Clinton side of the distribution with most siding with Obama:Of course, it is hard to say that even the few superdelegates falling on the right side of the distribution will actually endorse Clinton now. In short, unless something changes, from here on out I'd also take our reader's model over the one we have been using.

We began generating these predictions on February 8th and 9th. The idea behind the model at the time was to examine whether one candidate was likely to pick up more superdelegate support than the other over the coming weeks and months. Our underlying assumption was that superdelegates were politicians who were accountable to various constituencies. As such, we believed that their behavior would be relatively predictable. Based on those first predictions, I wrote:

"The pattern in the figure is pretty clear: our model estimates that there are a lot more unpledged superdelegates who are likely to support Obama over Clinton than vice versa. In fact, the Obama advantage is about 2 to 1. This is significant since Obama presently trails Clinton by nearly 100 superdelegates."

Indeed, three months later, it appears that our predictions were borne out. Obama has now passed Clinton in superdelegates just as we have always predicted he eventually would.

Despite the fact that the model seems unnecessary at this point, I will continue to update the data and check how we are doing from time to time. And if something changes in the contest and superdelegate endorsements are again up for grabs, we will be ready to jump right back into business. But, for now, I will be spending much less time focusing on the model.

I want to thank the Democratic Convention Watch website for supporting our efforts to generate these predictions and for maintaining such helpful information on the superdelegates. I also want to thank Alicia Prevost and Caitlin Zook at CCPS for working hard to collect the data we used in the model. And I definitely want to show my appreciation for all of the readers who posted comments that helped us refine and tweak the model.

For those of you who missed out on all the superdelegate prediction fun over the past few months, check out the archives of this blog and see Carl Bialik's post on it here.

Mr. Super Endorses, and We Predicted Him Correctly!

UPDATED (5/10)

The superdelegate behind the blog Mrsuper.org was revealed in the past few days to be Ed Espinoza, a member of the DNC from California. Several weeks ago, when he was still anonymous, Mr. Super expressed interest in this web site's predictions, even noting that he had found his own name in our list of estimates.

Well, Mr. Super has evidently declared his support for Barack Obama and guess what? The model predicted his endorsement correctly! He was actually in the "unclear" range for the model, but the estimate was that his probability of supporting Obama was .56.

By my count, 15 (updated from 10) previously undeclared superdelegates have announced their support since May 6th and the model has gotten 12 (updated from 8) of them correct so far.* Here is the list:


* Two other superdelegates who had already declared for Clinton switched their endorsement to Obama. However, the model is not set up to predict who will switch sides, just to predict who undeclared superdelegates will support. There is also some debate about whether Ellsworth had endorsed Clinton or not. For now, I'll punt on this one.

Recommended Reading: Fivethirtyeight.com's Primary Predictions

Mark Blumenthal has a nice column on the statistical models generated by Poblano that out-performed the pollsters. It is worth reading.

Also, if you want to see me speaking with Australian Broadcasting about the NC and IN primary results, you can visit this site and click on the relevant story. The interview is about 9 minutes in length. If I look tired it is because I did the interview at 7:30am after staying up until 2am waiting for Lake County the night before.

Which Superdelegates Are Most Likely to Desert Clinton?

Yesterday, Jennifer McClellan, a superdelegate from Virginia, switched her support from Clinton to Obama. A week earlier, Joe Andrew did the same. With Clinton's chances of winning the nomination shrinking even more after Tuesday, we may begin seeing even more superdelegates fleeing the Clinton campaign to support Obama. But which of Clinton's supporters are most likely to switch their support?

My model for predicting who superdelegates will endorse is not really built to estimate which superdelegates would change their endorsements. After all, making an endorsement is a very different process from choosing to switch an endorsement. However, what the model can do is provide a list of superdelegates who appear to be most mismatched. In other words, I can use the model to identify which superdelegates have endorsed Clinton when the model would've predicted that they would endorse Obama.

Here is the list of superdelegates who have endorsed Clinton despite the fact that the model suggests that they would be better matched to Obama (these are the 25 most mis-matched Clinton endorsers):

Two interesting things stand out on this list. First, most of these superdelegates come from states that Obama won. It makes sense that these superdelegates would be under some pressure to desert Clinton first given that their state went with Obama. It will be interesting to see if any of them do abandon her campaign in the coming days/weeks.

Second, several names on this list include superdelegates that have endorsed Clinton very recently. For example, Governor Mike Easley (NC), Rep. John Tanner (TN) and Rep. Tim Ryan (OH) all endorsed Clinton in the last few weeks. In the case of Tanner and Ryan, the decision to endorse Clinton is understandable given that Clinton did win their states. Easley's state went definitively for Obama, but he is term-limited and will be leaving office at the end of this year, so his endorsement of Clinton will probably not cost him much.

The (Very) High Turnout in North Carolina and Indiana

One thing about yesterday's primaries that deserves further attention is the high turnout for the Democratic primaries in both states. Generally, primary turnout tends to lag far behind general election turnout. Therefore, one statistic really jumped out at me about yesterday's primaries: in both contests, turnout in the Democratic primary exceeded the number of votes Kerry won in the states in the general election. In North Carolina, over 67,000 more voters voted in the Democratic primary than cast votes for Kerry in 2004. This is particularly impressive given that North Carolina has a semi-closed primary, where Republican registrants could not participate in the Democratic primary. In Indiana, which has an open primary, over one-quarter of a million more Hoosiers voted in the Democratic primary than voted for Kerry in 2004. And Rush Limbaugh can't take credit for this gap. The exit polls indicated that 10% of Democratic primary voters identified themselves as Republicans. Even if you subtract this group from the turnout figure, Democratic participation still would've out-paced Kerry's vote by well over 100,000.

How did this play out in other primaries? The figure below plots these numbers. The diagonal line shows the point at which Democratic primary turnout and the 2004 vote for Kerry are equal. As you can see, Texas is the only other state where Democratic turnout exceeded Kerry's 2004 vote, though many other states had totals that approached the 2004 Kerry vote.

Comparing Early Voters to the Overall Electorate in North Carolina

I blogged several times over the past week about the composition of early voters in North Carolina. I speculated that the demographics of early voters strong favored Obama and had likely put him firmly in the lead going into election day. The only question was whether the composition of the complete electorate would differ substantially from that of early voters. Well, the answer to that question appears to be no. Here is the comparison of early voters and the full electorate (including early voters) as provided by the exit polls. (The exit poll figures may be adjusted over the next few days).

	% of Early Voters	Exit Poll
Men	38.7%	43%
Women	60.8%	57%
White	56.5%	62%
Black	39.9%	34%
White Women	33.2%	34%
White Men	23.1%	27%

African Americans voted early in larger numbers than they did overall, suggesting an impressive organizational effort to bank their votes. White women, the core Clinton constituency in most states, made up one-third of early voters and not much more of the electorate that voted on the day of the contest. White men appeared to be the least likely to take advantage of early voting, as they made up only 23% of early voters but 27% of the electorate overall.

All told, 404,599 North Carolinians voted early in the Democratic primary compared to 1,188,370 who cast their ballots on the day of the election. Thus, early voters made up about one-fourth of the electorate in North Carolina.

On a related note, I have already said this several times, but kudos go out to the North Carolina Board of Elections for having one of the best websites out there. The early voting figures on the site were updated constantly and on election night the site reporting of the results was particularly well done with tons of great features. Whoever their web designer is deserves a big raise.