I'm with the minority on this one. I think she is nearly finished, but not quite yet.

As those who know me in personal life can attest, I am a contrarian. For better or worse, when I see everybody looking right, the first thought in my head is, "What's over there on the left?" So, the following might just be a product of my contrarian instincts, but I have to say that I just can't get to where most everybody is on this race.

Two things are holding me back: West Virginia and Kentucky.

The conventional wisdom has it that Clinton did herself major damage Tuesday night by getting blown out in North Carolina. I completely agree. This hurt her with the pledged delegate count. Much more important, I think, is that it hurt her with the popular vote count, which she must win to press an argument with the superdelegates.

However, it is possible that she could counter Tuesday's blowout with two big blowouts of her own in the next two weeks. This could undo most of the damage done by her big loss in North Carolina, and put her back on track.

West Virginia is 95% white, and one of the poorest states in the nation. Demographically, Pennsylvania's twelfth congressional district is a decent proxy of it. Clinton won Pennsylvania's twelfth by 46 points. A recent Rasmussen survey put her up 29 points in the Mountaineer State, with 17% undecided. Another poll had her up 40 points, with Obama under 25%.

Kentucky is not as poor or as white as West Virginia, but it is nearly so. Demographically, Kentucky falls somewhere between Ohio's sixth congressional district, which went for Clinton by 45 points, and the seventeenth, which went for her by 28 points. A recent Survey USA poll of the Bluegrass State had her up 34 points - with a staggering 72 point lead in the east, where Obama was winning less than 20% of the vote. Rasmussen recently had her up 25 points with 13% undecided.

Courtesy of the perspicacious Sean Oxendine, here's a graphical representation of how Appalachia has performed. The deepest blue represents countywide Clinton victories of 30+.

As Oxendine says in his analysis of Indiana and North Carolina: "Appalachia didn't budge [on Tuesday]. She is going to absolutely blow him out of the water in West VA and KY."

So, here's my question. What happens to "It's Over" if Clinton pulls a 40-point victory in West Virginia on Tuesday, then follows it up a week later with a 30-point victory in Kentucky? If these states turn out in the same margins that states since March 4th have averaged, that would imply a net of about 290,000 votes for Clinton. That puts her within striking distance of a reasonable popular vote victory. "Over" will be over as we turn our attention to Puerto Rico.

There are good reasons not to take Puerto Rico lightly, even though the press has continued to do exactly that. I would note: (a) Puerto Ricans vote in large numbers (2 million in the last gubernatorial election); (b) Puerto Ricans have never had this important a role in United States presidential politics; (c) Puerto Rico's politics is focused at least partially on how (if at all) to adjust its relationship with the United States; (d) Puerto Rico's is an open primary, and the residents of the Commonwealth, who are United States citizens, do not see themselves as Republicans or Democrats.

The inference I draw is that Puerto Ricans could turn out in huge numbers. If they do, and they swing for Clinton in a sizeable way, the popular vote lead could swing, too. Add 290,000 votes from West Virginia and Kentucky to 250,000 votes from Puerto Rico, account for expected losses in Oregon, Montana, and South Dakota, and you get Clinton leading in many popular vote counts, some of which are really quite valid. If she has one of those leads when the final votes are counted on June 3rd, the race will go on to the convention.

Am I predicting that all of this will happen? No. That would be quite presumptuous. The problem is not that any of these incidents is individually unlikely. It is not unlikely that Clinton will get a huge victory in Kentucky, West Virginia, or Puerto Rico. Theoretically, I would wager at least one of the three will happen. The problem is that she has to do all three. What's more, she has to keep it competitive in Oregon (just how competitive depends on her margins in the other states). That's a tall order - four big things to do with no margin for error. I'd never predict that she could do all four. I may be a contrarian, but I am not an idiot!

Her biggest impediment might be the development (finally!) of some momentum. With the crush of stories touting the end of the race, will her vote be depressed in Kentucky and West Virginia? I doubt her voters would actually go for Obama - but they might stay home, thus diminishing both her overall margin of victory and/or her net vote score. Incidentally, I did find some good news for Clinton: finals week at WVU ends on Saturday.

My point is that those in the media who are declaring this race to be over are necessarily predicting that she can't do all of this. That's a conclusion I can't go along with. It's quite unlikely, but it is still possible - and it is more possible than the "Obama might have a meltdown" scenario.

Minimally, I will predict that West Virginia will be either her best or her second best finish, behind only Arkansas. Kentucky should come in right behind the two. This alone should be enough to induce some caution. I think it is too hasty to declare her finished just days before two of her three best states.

Am I on to something here, or is this merely my contrarian streak running amok? I'll let you decide. In fact, I'll help you make an informed decision! I have updated my vote spreadsheet to include all contests through Tuesday. I encourage you to play around with the numbers yourself. Follow this link to Predict the Race for Yourself, Version 2.0.

At the time of its initial publication, I had not put much thought into Clinton's vote margins in Kentucky and West Virginia. I merely used the results from Tennessee as a rough baseline. I think this was a bit naïve, given what we now know about the white vote in Appalachia. So, those numbers have been updated. I also updated the turnout projections, based on new data. Once again, you can adjust these figures however you like.

As you can see, Clinton did about as well in Indiana as she did in Pennsylvania and Ohio with white men, white Protestants, and seniors. However, beyond this, she suffered a decline among her best groups. Notice in particular her decline among white women, white Catholics, and union households. Basically, the core of her voting bloc was still with her, but Obama picked off a larger portion of it than he did in Ohio and Pennsylvania.

Why did this happen? It wasn't because of southern Indiana. All of the counties south of metro Indianapolis went for Clinton except Monroe County, where Indiana University is located. Clinton won almost all of these counties in the south by at least 10 points. In some instances, her margins were 30 and even 40 points. All in all, the region bore a resemblance to southern Ohio, where Clinton did really well.

One big problem for Clinton came in metropolitan Indianapolis. Obama beat her in metro Indy by 17 points. Much of this came from Marion County, where a large number of African Americans live. However, discounting Marion County, she only won about 51.6% of the vote. Factor that in with her losses in Fort Wayne, South Bend (home to Notre Dame, and thus a reason why she underperformed among white Catholics), and Gary, and we approach an answer to why the final result was so close.

What about North Carolina? It is instructive, I think, to compare the results from North Carolina to those of Tennessee and Virginia. For the sake of perspective, let's add some basic demographic features of the three states - namely median white income as of 2000 and the percentage of African Americans in the state. We'd expect a priori that as both figures fall, Clinton would do better.

As you can see, North Carolina performed roughly as we might expect, falling in between Virginia and Tennessee. Nevertheless, it is surprising that the results were closer to the Virginia end (i.e. Obama +29) than the Tennessee end (i.e. Clinton +13). What might explain the difference?

Unlike Indiana, it doesn't come from Clinton's core voting group. She did extremely well among white voters in North Carolina. Obviously, she didn't do as well with them as she did in Tennessee. However, she still trounced Obama among white men and white women, regardless of their religious affiliation.

Clinton's problem was with the African American vote, which came in at about 33%. Her trouble in North Carolina, as well as the South in general, is that white voters are more likely to be Republican than in decades past. This has given Obama a demographic edge in the region - one that has actually grown in the past few months. Note that African Americans in North Carolina went for Obama more strongly than they did in either Tennessee or Virginia. In fact, we can see a general trend in the African American vote toward Obama - not just in these states, but nationwide. It has not been much commented upon - most likely because African Americans have been supporting Obama more strongly than any other group. Nevertheless, as time has gone on, the African American vote has clustered around Obama much more tightly.

The following chart has the details. It delineates Obama's margin of victory among African Americans over time. The states are divided into the South and non-South, then arranged chronologically.

We have to be careful not to over-interpret these results because they are relatively small sub-samples of each exit poll. Nevertheless, there is a discernible trend in these numbers toward larger and larger Obama victories. The African American vote now goes much more heavily for Obama than it did at the beginning of the cycle. In fact, if we take Clinton's margin among African Americans in Tennessee and apply it to North Carolina, keeping all else equal, Obama would have barely defeated her.

So, we can conclude that Clinton's narrow victory in Indiana was largely because she didn't do as well with her strong groups as in Ohio and Pennsylvania. However, she did do just as well in the south. It was in metro Indy and in the north that she didn't do as well. On the other hand, Obama's extremely large victory in North Carolina was due to his strength among African Americans, a group with which he has improved over time.

(1) Many people with whom I spoke at the conference were interested in Obama's performance among working class whites. Nobody seemed to have an answer as to whether his poor results are a sign of electoral trouble in the fall. I still do not. Nevertheless, the conversations I had inspired two thoughts.

First, I would intuit that, based on these primary results, voters in these places are at least going to give John McCain more of an audience than they otherwise would in a year like this. That's not to say they will ultimately go for him, just that they will consider him more than usual for a year when the macro conditions favor the Democrats so heavily. I think Obama is not yet resonating with them. So, even if he wins the strong Democrats among them who have voted for Clinton - those soft Democrats, Independents, and soft Republicans might give McCain a harder look.

Second, if Obama were to carry a state like Ohio - there is a possibility he will do so in a new way. To appreciate what I mean by this, open the following links. From David Leip's Election Atlas, here are geographical representations of the Ohio results from 1976, 1992, and 1996 - the three elections in the last 40 years the Democratic Party has won. Every time, the nominee carried Ohio. Counties the Democrat won are in red; counties the Republican won are in blue. Note the similarity between the three. The counties that Carter and Clinton won form an inverted "C." Both Democrats did strongly along Lake Erie, strongly down the Pennsylvania-Ohio border, and then they carried many of the counties in the Ohio River Valley. Now, compare these Democratic victories to the party's losses in 2000 and 2004. What is missing? Gore and Kerry won fewer counties around Lake Erie, and they did notably worse in the Ohio River Valley. Bush won almost all of those counties.

It is in the Ohio River Valley where Clinton beat Obama on the order of 30, 35, even 40 points. These are the kinds of voters I expect to give McCain more of an audience than he would otherwise receive. If they ultimately back McCain (again, a big "if" in a year like this!), and Obama wins the state anyway, he will have won with a voting coalition we haven't seen before. Perhaps he will have pulled in upscale, suburban Republican-leaners around Columbus, Cleveland, and Cincinnati.

(2) I am very interested in next week's election in West Virginia. Everybody expects Hillary Clinton to win, but I can't help but wonder if they'll be surprised by the size of the margin.

We can reasonably expect it to be enormous. From a socioeconomic standpoint, West Virginia is almost entirely comprised of the sort of counties that Obama has done poorly in. The median white income in West Virginia is about $30,000 per year. African Americans comprise roughly 3% of the state's population. This puts it somewhere between Belmont County, Ohio and Greene County, Pennsylvania. Clinton won 72% of the vote in Belmont and 75% in Greene. From another angle, we see a similar situation. If we take the counties of Ohio, Pennsylvania, Maryland, and Virginia that border West Virginia, we see that Clinton won on average 63.5% of the countywide vote. However, if we exempt the counties in Maryland and Virginia that border the eastern panhandle of West Virginia, her share of the vote jumps to 70.1%.

So, how would this be surprising? To date, you'd have to look closely to see Obama's poor results among working class whites in Appalachia. If you only take the 55-45 margins in Ohio and Pennsylvania, without drilling down a bit deeper, you might think, "That's not so bad." That probably will not be possible next week. The whole state of West Virginia should mimic what we saw in southern Ohio, southwestern Pennsylvania, and southwestern Virginia.

How will the press and the superdelegates react if Clinton wins by 30 points, 40 points, even 50 points? Remember that John Kennedy essentially wrapped up the 1960 nomination with a win in West Virginia because it proved that he had crossover appeal. What happens if Obama "proves" the opposite? Psychologically speaking, are people prepared for a loss of this magnitude, having fully absorbed the countywide details of previous results, or will this come as a shock to them?

Obama's impending loss in West Virginia might reinforce the previous point - an Obama electoral college victory might look different than anything a Democrat has ever put together. A Democrat has won the White House having lost West Virginia just once. Woodrow Wilson did it in 1916. Again, this is not a sign of any impending electoral doom for Obama should he win the nomination. States can and do move into and out of a party's voting coalition. Take Delaware and New Jersey, for instance. Both usually supported the GOP when it won the White House, but this is no longer the case. The point here is simply that an Obama victory might look like something we've never seen before.

(3) Despite his loss in Pennsylvania and the reemergence of Reverend Wright, Obama continues to close Clinton's lead in superdelegates. Why?

I think it has to do with his pledged delegate lead. The last few weeks have exposed some weaknesses in Obama. However, from the superdelegates' perspective, this does not mean that Clinton is the superior choice. She has her own problems, of course. More important than this is the fact that Clinton's path to the nomination is necessarily "dirty." She must out-muscle Obama at the convention. That's the only way. After all, Obama's lead in pledged delegates is 154. Even if the superdelegates rally around Clinton, Obama would not disappear, nor should he! He could and surely would take the fight to the convention, and try to win there.

Let's try a counterfactual thought experiment. Reduce Obama's pledged delegate lead by 139, so that he currently has a lead in pledged delegates of just 15. What would the superdelegates be doing now in response to the Wright controversy? How would they have reacted to his losses in Ohio and Pennsylvania? I imagine they'd be moving to Clinton, possibly in large numbers.

But why 139 delegates? That is precisely the number of delegates he has netted from the caucuses, which have all been low-turnout affairs. That's a key point. Even in states where Obama held a demographic advantage - there were presumably enough Clinton supporters in the state to level the playing field. Kansas is a good example. Obama won Kansas by about 18,000 caucus votes - out of only 37,000 or so cast. This is a state that gave Bill Clinton 388,000 votes in 1996. Surely, she could have found another 18,000 Kansans to support her.

So, why didn't she?

The answer is simple. He prepared for the caucuses. She didn't. He was organized. She wasn't. This enabled Obama to rack up huge delegate victories, all of which occurred (at the time) under the radar. We were looking at California, Massachusetts, New Jersey, Virginia, Wisconsin, Texas, Ohio, and Pennsylvania. Meanwhile, he netted 24 delegates in Minnesota, 26 delegates in Washington, 15 delegates in Colorado, 15 delegates in Idaho, 14 delegates in Kansas.

I think his foresight to organize the caucus states has served him doubly well. Not only has it given him a large delegate lead compared to a modest popular vote lead - it has served as protection against political peril. My sense is that with Ohio, Pennsylvania, and then Wright - superdelegates would be flocking to Clinton if it were not for his caucus victories.

Above all, this highlights a stark contrast between the Obama and Clinton campaigns. The Clinton campaign formulated a poor nomination strategy. When it fell to pieces, the campaign essentially began improvising. To this day, it lives week-to-week, one "do-or-die" primary after another. This has diminished its capacity to take advantage of political opportunities. The Obama campaign, on the other hand, formulated a superb nomination strategy, which it is still following even after 14 months of campaigning, and which has minimized the damage from a major political controversy.

I am tentatively schedule to participate in the opening panel of the conference, with Kathleen Hall Jamieson and Thomas Mann, entitled "Overview 2008: Where We've Been and Where We're Going."

The details of the conference can be found here. The program schedule looks very promising, and there are some superb scholars participating.

If you live in the area and find this of interest, I encourage you to check it out.

So, who is right? I do not know. In fact, I don't think the question can be answered.

Before we get into this, we should probably be more specific. After all, it is wrong to assert that Obama cannot win the white working class. He did exactly that in Wisconsin. His problem has been the white working class in certain geographical reasons. The following graph, provided by Sean Oxendine, tells the tale.

Obama's weakest performances among whites have been in Appalachia, which is traced in solid black. Oxendine has put counties that Obama won in green, counties that Clinton won in blue. Note the expanse of deep, dark blue that moves from Mississippi to New York. This is where Obama has had his greatest problems. This is why Clinton will not drop out next week, even if she loses Indiana. West Virginia comes the week after, and Kentucky the week after that. She's bound to win both, and candidates do not drop out immediately prior to impending victories.

In all likelihood, weak general election performances in Kentucky, Tennessee, and West Virginia will not cost Obama the presidency - though he would be the first Democrat ever to win the White House having lost all three. The trouble comes with southern Ohio and western Pennsylvania.

To appreciate Obama's potential dilemma, consider the following chart. It examines Clinton's primary victories in the counties of metropolitan Pittsburgh compared to the general election performance of Republican candidates dating back to 1972.

The first thing you'll notice is that the term "Reagan Democrat" is a bit of a misnomer here. In fact, Reagan barely improved over Ford, and his landslide reelection was nothing compared to Nixon's. Aside from Nixon, George W. Bush has performed best in metropolitan Pittsburgh.

Next, notice the wide variability in vote returns. Outside Allegheny County, where the African American vote provides a solid base for the Democrats, we see twenty-point swings in these counties.

Finally, turn to the primary results from last week. Obama did very poorly across the entire metropolitan area. Allegheny County was the sole exception. Interestingly, Fayette County is the poorest county in all of Pennsylvania, and no Pennsylvania county supported Clinton more strongly.

This is what John Judis is worried about. Even if you allocate most of Clinton's voters to Obama in November, he still might face trouble in metropolitan Pittsburgh. It might be that metro Pittsburgh Democrats, by swinging so heavily to Clinton, have indicated that the region as a whole is unimpressed by Obama and will therefore back McCain. Even if strong Democrats support Obama - weak Democrats, Independents, and persuadable Republicans might not.

What happens if counties like Fayette swing against Obama? It depends on how big the swing is. Would it be on the order of 1984, 2004, or (worst case scenario) 1972? Depending on the size of the swing, Pennsylvania could become trouble for him. After all, Obama had problems not just in metro Pittsburgh, but also in the northeast. He lost Lackawanna County by 48 points, and Luzerne County by 50 points. Kerry won both counties in 2004. Obama should pull a big victory out of Philadelphia, but trouble in the northeast and southwest would put more pressure on metro Philly to perform.

Ohio could also be trouble. Bill Clinton won the state in 1996 largely because he won the same counties that gave Obama just 30% of the primary vote. These are the counties that hug the Ohio River Valley in the second, sixth, and eighteenth congressional districts. Again, Obama might be able to improve on Kerry in the suburbs of Cincinnati, Cleveland, and Columbus - but trouble in the southeast would put more pressure on these areas.

Obama might also have trouble in Missouri. Oxendine's map does not indicate it, but he was very weak in southern Missouri, especially in the seventh and eighth districts. Many of these counties split their votes in 1996, which is one reason why Bill Clinton won the state twelve years ago.

In all, these results have been so lopsided that people have begun to wonder about Obama. Can he hold the line in these places? If he can't, will that cost him these states, and therefore the presidency itself? Some people see this happening. Some do not. This is what has brought about this discussion. His poor showings in these places is a bit of a puzzle, and has induced a debate.

What I find interesting about the conversation over places like Fayette County is that people from there are not really participating in it, at least on their own terms. Their role in the discussion is mediated by the public opinion survey. They merely answer the questions the pollsters ask them.

So, if you think about it, the voters' ability to influence the conversation is really quite constrained. In a statewide survey of Pennsylvania, you might get three or four residents from Fayette who answer "yes" or "no" to a handful of questions. That's it. No amplification permitted.

This makes me wonder about these competing theories. Just how much do they depend on the theorists, and how much on the voters?

What is more, the candidate in question happens to be black. This further limits the utility of the polls. There has been a heated debate in political science for a quarter century over whether white America has become less racist, or whether its racism has become less transparent. Researchers have found that whites now tend to answer straightforward questions about race in a much less racist way. However, some researchers claim to detect racism in responses to seemingly unrelated questions. Others vehemently disagree. Putting aside the merits of the arguments pro and con, the point is that polls have failed to arbitrate the dispute. Researchers look at the same answers to the same questions, yet see two different opinions.

Those outside the academy have been struggling with similar difficulties as regards the "Bradley Effect." Like so-called "latent racism," the issue hinges on whether public opinion surveys give us a clear read on the thinking of the American voter.

Unfortunately, they probably do not. They are limited, imperfect tools for understanding him - especially on the issue of race. Above all, they tightly restrict his role - and therefore limit his ability to influence our conclusions about him. It is very possible for a researcher or pundit to approach the polls with erroneous preconceived notions, and have the data's ambiguity "validate" the false theory.

The problem is that we are exclusively dependent upon the polls for our understanding of public opinion. Nobody is really going "out there" to interact with average people, to find out what they are thinking. We have the polls, and that is all we have. So, we only know as much as they tell us. When they are silent, ambiguous, or misleading, we are ignorant.

They did not always monopolize our knowledge base. While studying for my qualifying exam in American politics several years ago, one of my favorite reads was a dusty old book called Political Ideology by Robert Lane. Unfortunately, it is no longer part of the canon. I only read it by mistake. My department gave me an outdated reading list to study from, and I read it before the error was corrected. I'm glad I did. It is a fascinating read, not so much for its conclusions, but for its methodology. Lane engaged in detailed, dialectical conversations with a group of fifteen men from "Eastport, USA." His study was certainly not wide, but it was very deep.

For better or worse, public opinion researchers have not embraced Lane's method. Today, research is dependent upon survey data. Academics develop abstract theories of how voters think, and test them via these surveys. While some academics interact with their subjects, this is by no means required for publication. The same goes for analysts outside the academy. Extended, dialectical conversations are not required for opining on the average voter. Instead, the latest readings from Gallup are.

There are good reasons to go with the public opinion survey over Lane's methodology. If I were only going to select one method, I'd select the former without thinking twice. Lane's methodology is simply too subjective. However, the objectivity of the survey comes at a great cost - namely, distance from our subjects. This distance gives the answers to the survey questions ambiguity, and thus an opportunity for subjectivity to weasel its way back in. Ideally, both methods should be employed. The public opinion survey, with its objective and quantifiable answers to specific questions, should be supplemented with extensive conversations with our subjects.

Practically speaking, this kind of methodological pluralism rarely happens - though there are exceptions. Instead, most researchers rely exclusively on the polls. This is a very unfortunate development. It is particularly debilitating when it comes to understanding Obama and Appalachian whites. As should be clear from the discussion of "latent racism" and the "Bradley Effect," polls are particularly unhelpful here. Race is a hard topic to explore from thirty thousand feet, which is essentially what the polls do.

What we need, then, is somebody like Robert Lane. We need an expert who is up-to-date on the latest scholarly research, and who has spent time soaking-and-poking in places like Fayette County to see whether people there are willing to vote for Obama. As far as I know, there is no such expert.

And so, we have no answers, just questions.

Yesterday, I was corresponding with a friend of mine who noted that Clinton's performance among certain groups worsened relative to Ohio, and that she made up the difference because her best groups were more populous.

The most striking instance of this was Clinton's victory among the elderly. Clinton won the elderly by 46 points in Ohio, but by just 26 in Pennsylvania. According to this hypothesis, what made up the gap is that the elderly constituted 14% of the electorate in Ohio, compared to 22% in Pennsylvania. The upshot of this is that if you take Clinton's vote margins in Pennsylvania, apply them to the demographics of Ohio, the latter would have been much closer.

However, there is a catch. Can we take the Pennsylvania results and place them with the Ohio demographics? The validity of that action depends upon how similar the two states are. I argued in March that Ohio could give us a rough estimate of what to expect in Pennsylvania. In a situation such as that, where there is not much data and we have to use what we can find, bringing Ohio into a discussion of Pennsylvania was very useful. However, as noted at the time, there are real limits to this line of analysis. Pennsylvania is a very diverse state. Some places have a lot in common with Ohio. Some places do not.

So, this offers us an interesting analytical question. Did geography play a factor in the Pennsylvania race? More specifically, how close were the results in certain parts of the state to the results in Ohio?

An easy way to test this would be to carve up the exit polls by region to create more detailed cross-tabulations. We'd look not only at how Clinton did among the elderly statewide, but the elderly in the southeast, the southwest, etc. Unfortunately, we cannot do this. We do not have access to this kind of data.

We can approach this in another way, using the countywide vote results. In March, we used linear regression to build a predictive model for countywide Ohio results based on median white income, the percentage of African Americans in a county, and the percentage of residents aged 20 to 24. We can tweak this model to work for Pennsylvania. In fact, we can build a model to explain Pennsylvania and Ohio at the same time. We'll use the three variables mentioned above, plus the percentage of senior citizens among all whites in a county.

Remember that our analytical question is whether voters in certain parts of Pennsylvania behaved like Ohio voters. Accordingly, we'll divide Pennsylvania into five segments: southwest, northwest, central, southeast, and northeast. Our predictive model will include a factor for each of them. The idea behind this is that if Obama did better in a given Pennsylvania region relative to Ohio - controlling for race, income, and age - it will be picked up by one of these variables.

We might expect Obama to have improved relative to Ohio in the southeast. However, this does not appear to have been the case. When we control for race, income, and age, we get roughly the same results in Ohio and southeast Pennsylvania. The same goes for southwest Pennsylvania.

What is significant is the variable that captures counties in central Pennsylvania. This was surprising. The model indicates that, controlling for race, income, and age, Obama performed better in central Pennsylvania than he did in Ohio. Additionally, there is a modest statistical significance to the variables for the northeast and northwest segments of the state. However, when we use a more expansive definition of central Pennsylvania, re-classifying the counties in the northeast and northwest segments that abut the center segment as part of the center, this significance washes away.

What is the upshot of this? Obama did not improve relative to Ohio in Erie, Pittsburgh, Scranton/Wilkes-Barre, or even Philadelphia. However, he did improve in the "Middle T" of the state. This improvement was not puny. If we compare a county in Ohio to one in central Pennsylvania with similar racial, income, and age demographics, we should find Clinton's margin to be 7 to 17 points smaller in the Pennsylvania county.

Let's enliven this with a graphical illustration.

First, let's build a simple predictive model of Ohio countywide returns based upon median white income. We know, of course, that other variables are important factors. We just finished building a comprehensive model, after all. However, median white income is the best predictor, and our task here is just to illustrate the point.

This model gives us a line to graph. It looks like this.

The idea here is that we plug in the value of median white income for an Ohio county, and we get a prediction for Clinton's margin of victory in the county.

Next, we place on top of this graph a scatter plot of the counties in each segment of Pennsylvania.* What we are looking for is whether the Pennsylvania observations systematically fall above or below the line. We expect that there will be no systematic pattern for the counties of the southwest, southeast, northwest, or northeast. They will fall above or below the line randomly because each segment of the state behaved roughly similar to Ohio. However, we do expect a systematic difference between this line and central Pennsylvania counties. In particular, we expect the observations to fall systematically below this line because Clinton's margins should be smaller in central Pennsylvania.

Let's check the northeast, northwest, and southeast first.

There seems to be no pattern here. The counties in these segments of Pennsylvania do not fall systematically above or below the line. Next, let's check the southwest and the center.

Notice how the counties of southwestern Pennsylvania fall very tightly along the line. No part of the state mimicked Ohio more closely than southwest Pennsylvania.

Next, notice the two counties toward the bottom. One of them is Centre County, home to Penn State. The other is Union County, home to Bucknell University. So, the fact that Clinton "underperformed" here should come as no surprise.

Placing them aside, we can notice that the remaining central counties fall systematically below the Ohio prediction line. This means that Clinton's margins in central Pennsylvania were smaller than they "should" have been. This is exactly what we found above. Controlling for race, income, and age, Obama did better in central Pennsylvania than he did in Ohio. We can't say that about any other part of Pennsylvania.

This is not to imply that he did particularly well in central PA. Clinton still won the counties by an average of 25 points. The point is that, if this area were behaving like Ohio or the rest of Pennsylvania, she would have won them by something closer to 33 points.

What might explain this result? It is hard to say, though it is noteworthy that central Pennsylvania is the most Republican part of the state. We have found again and again in this primary season that, outside of the South, white Democrats in heavily Republican areas tend to prefer Obama more than other areas. It is unclear what has caused this trend, but the observations in central Pennsylvania are consistent with it.

Finally, we should note the irony of central Pennsylvania's support of Obama. These are the locations where you can find many of the "small towns" about which Obama was speaking in San Francisco - and yet they seemed to be tilted in his favor. In a certain sense, small town Pennsylvanians preferred Obama more than the rest of the state!

***
Endnotes

^[*] We'll display only the counties where the African American population is less than or equal to 10%. The reason for this is that, as the African American population increases, the tightness of the dispersion of the data points decreases. Remember that this is only for the purposes of illustration. Our linear regression model accounted for this perfectly well.

The following chart details this by comparing Clinton's performance among the select demographic groups in Ohio and Pennsylvania.

By and large, we can see why Pennsylvania and Ohio produced similar results in the aggregate. Clinton did roughly as well these groups in both states. Obama, for his part, improved here and there on her best Ohio groups. For instance, he trimmed her lead among white men. However, Clinton minimized this by doing slightly better with some of Obama's best groups - like, for instance, the college educated. Overall, it added up to a roughly similar result. Clinton won Ohio by 10.3%. She won Pennsylvania by 9.4%.

What we see, then, is what we have seen again and again in this contest. Clinton continues to do well with "downscale" whites. Obama does well with "upscale" whites and African Americans. What is intriguing about this result is not just that it is similar to Ohio - but also that it is similar after seven weeks and millions of dollars in campaign expenditures. Clearly, these voting groups are entrenched.

Let's amplify this analysis with a look at how both candidates did in each region of this diverse state.

These results are as we might have expected. Clinton dominated the western portion of the state - winning Erie in the northwest and Pittsburgh in the southwest. What is more, if you look at the counties in the far southwest corner of the state, you'll see that Clinton's margins were quite lopsided. For instance, Fayette County gave Clinton 78.9% of the vote. This, I think, is an indication of how West Virginia will go. Expect Clinton to win the state overwhelmingly. A 40-point victory does not seem unimaginable to me.

She ran strong in the northeast - winning Scranton and its surroundings. Furthermore, while the chart does not indicate it, Clinton performed extremely well in the "Middle T" of Pennsylvania - the great rural expanse that stretches across the center of the state. Obama only broke her winning streak in Centre County (home to Penn State), neighboring Union County, and Dauphin County (home to Harrisburg).

In the east, Obama ran strong in Philadelphia County, but the two split the five suburban counties. Clinton also won the Lehigh Valley. Additionally, the two split fast-growing York and Lancaster counties in the southeast.

All in all, it was a strong, and generally predictable win for Hillary Clinton. But what does it mean?

We can say the following. If the superdelegates had grown concerned after Ohio about Obama's ability to win lower income whites in the general election - these results will not alleviate their worries. Pittsburgh, Scranton, and Erie all swung decisively for Clinton. If Ohio had them worried, so will these results.

As for whether it will serve Clinton's short-term goal of spinning herself as still having a chance to capture the nomination - that remains to be seen. That will be a matter up to the media, potential Clinton donors, and superdelegates. As of this writing, she was up by 9.4% in the Pennsylvania vote. Whether this is "enough" is not a question I can answer.

Nevertheless, I can say the following. As of this writing, Clinton has netted 216,000 votes from the Keystone State. Last month, when working out my "Predict the Race for Yourself" spreadsheet, I inserted values that seemed to be favorable numbers for Clinton that were also doable. These were not meant as predictions, but rather as an illustration of how Clinton might come back to win at least one valid popular vote count. The number of votes inserted for Pennsylvania was 211,000. I took that to be a reasonably optimistic outcome for her. She basically met that goal last night. This implies that she is "on track" to catch Obama in one of the more valid vote counts. It won't be easy, but she can still do it. If she does, this could be a compelling argument for the superdelegates.

I hope to return later in the week with a more comprehensive analysis of how Pennsylvania voted. Stay tuned!

For the good of the party, the Democratic primary battle needs to end. It is providing no benefit whatsoever.

I think the primary battle has actually been quite helpful for the Democrats. It has exposed weaknesses in both campaigns that might not have been identified until October. This has given both an opportunity to strengthen themselves.

Consider a few examples. We have learned that the Clinton organization was plagued by pro-Clinton myopia. Operating under the assumption that she could not lose, it failed to do everything it could to ensure victory. This included small things like mismanaging Bill, to big things like leaving caucus states unorganized. If Clinton had won Iowa and New Hampshire, knocking Obama out, it might not have discovered its myopia until it was too late. Learning in October that its basic assumptions were fundamentally flawed would have been disastrous.

The Obama campaign has learned several important lessons about "elitism." It has learned that Republicans are quite attracted to this idea. This is a good thing. Now it knows how the Republicans will come after him. Furthermore, thanks to last week's debate, it also knows it must have a better response ready for the GOP. Suppose Obama had won Texas and Ohio, knocking Hillary out. Flash forward to the fall debates, when Obama is asked about William Ayers. Not having the benefit of having been asked in April, he gives a tepid answer like the one he actually gave last week. This time, his debate opponent is not Hillary Clinton, whose spouse pardoned members of the Weather Underground, but John McCain, who was in the Hanoi Hilton when they were engaging in terrorism. Obama would have been in much more jeopardy.

The problem is not that the campaign has gone on this long. Rather, it is that there is no obvious terminal point. There will be a point at which the benefits to the campaign are outweighed by the costs. I do not think we are there yet, but we are getting close. The trouble is that there is nothing to stop the race when that point is reached.

Put another way, when does Clinton meet her Waterloo? It probably won't happen today. It probably won't happen on May 7. Even if she loses Indiana and North Carolina, she can still limp to West Virginia the next week and Kentucky the week after that. What's to stop her then? If she can limp to West Virginia and Kentucky, can't she limp to Puerto Rico, and then to Denver? Remember - no candidate who has won as many votes and delegates as Clinton hasn't taken the fight to the convention.

Of course, it is easy to overestimate the likelihood that Denver will be a mess. There are two distinct ways I could see the nomination battle ending, even if Clinton doesn't get knocked out in a specific contest. First, the superdelegates could grow tired of the race and swing Obama's way. In response, Clinton could work to flip them. However, if her endeavors to do so are met with strong assurances that no, in fact, they are not interested in changing their minds - there will be strong pressure on her to drop out.

Second, she could run out of money. This is why many presidential candidates drop out. They can no longer afford to put fuel in the plane. Clinton is not yet at this point. Pundits who frequently mistake the race for cash as a proxy for the race for votes have been hung up on the fact that Obama has outraised her so far this year. This emphasis misses the point. At least as of March, Clinton was raising enough cash to subsist. In fact, she could still put forward a real campaign. Nothing on the level that Obama could, but she could still advertise and do mailers and get-out-the-vote activities.

If this changes, that could be the end for Clinton. And it might change. In the last six weeks, we have had just two primaries. In the next six weeks, we will have eight. It will become more expensive to subsist, let alone put on a real campaign.

This prolonged campaign is damaging the party's prospects by dividing the base.

Any poll you look at will indicate that the Democrats are divided. However, I think these numbers belie the relative ease the nominee will have in stitching the core coalition back together.

According to the American National Elections Study, the last time "strong Democrats" defected in significant numbers was 1984, when 11% went for Reagan. In an election like this one, where the Democrats face better-than-even odds, it is quite unlikely that this will occur. Barring extraordinary circumstances, the two parties are going to pull 95% of their strong partisans. In that case, this election will be determined by the question that has decided all recent ones: who wins that middle chunk of the electorate, the weak partisans and the Independents?

Here's how I think it could work out. Suppose that the nomination battle continues to Denver. It will be messy and divisive. However, when it is over, the general election campaign will begin. McCain will attack Obama or Clinton, and Obama or Clinton will attack McCain. This should unify the Democratic base. Disappointed Democrats will begin to perceive the vast differences between their side's nominee and McCain, and they'll feel affinity for the candidate they currently oppose. The rhetoric will re-activate their partisanship. By November, they'll be ready to go.

That being said, I think the Democrats could suffer some damage if the nomination battle continues to Denver. The problem will involve the nominee's campaign organization, rather than his or her core voting group. I see three potential difficulties.

First, the nominee will have spent the summer angling for the nomination, rather than preparing a general election strategy. A good strategy is going to be harder to develop than it might first appear, given that the Republicans have put forth John McCain. He needs to be tied to the Bush administration - but this will not be as easy as Democrats might think. The guy has a reputation for being a thorn in Bush's side. So, the strategy that links him to Bush has to be a clever one.

Small example. I was watching Hardball last week. Chris Matthews had Joe Biden on. He was critical of McCain, but he couldn't resist complimenting him at several points. As an individual incident, it was pretty trivial - but if Democratic surrogates can't help but say nice things about McCain amidst their attacks, the forcefulness of those attacks is going to be muted.

Second, there are organizational tasks in the swing states that might be delayed because the nominee's campaign is distracted. I'm talking about the little stuff like hiring staff, getting office space and supplies, preparing a get-out-the-vote strategy, etc. Unfortunately for the Democrats, the DNC does not have the funds to make up the difference while Obama and Clinton are wrapping up their nomination fight.

Third, there is the simple matter of fatigue. I'd wager that John McCain is getting more and better sleep than Barack Obama or Hillary Clinton. Over the course of four more months, this could make a major difference - for the candidates and the staffs. Obama seemed a bit drained in last week's debate. How drained will he be if he has to fight Clinton all the way to Labor Day, only to turn around to face McCain?

The only reason this pointless, self-destructive race persists is Hillary Clinton. She's a hyper-ambitious pol who won't do the high-minded thing and drop out.

I have three problems with this argument.

First, while it might be that Clinton is more ambitious than Obama, both of them get up in the morning, look in the mirror, and think to themselves, "I should be the next President of the United States." This makes them more ambitious than 99.999999% of the nation. The difference between them, if there is any at all, is in the barest of degrees. It is certainly not in kind. If the shoe was on the other foot, Obama would probably remain in the race, too.

Second, what exactly are we talking about here? Is it hyper-ambition that is driving Clinton, or is it simple scrappiness? I would submit that it is scrappiness, which is actually kind of admirable.

The Clintons are a scrappy crew, and most Democrats have appreciated this at one point or another in the past. Most were grateful for their scrappiness when the bottom fell out in 1994. In just two years, Bill Clinton went from defending his relevancy to trouncing Bob Dole. That's the thing about the Clintons. They play until the buzzer sounds. If you watch C-Span late at night, as I do, you'll see people crowd around Bill and Hillary after their speeches are over. At least half of the intelligible phrases I hear are, "Thank you!" I think this is what they're thanking them for. The Clintons hang in there. They don't quit.

Third, blaming Clinton just obscures the real problem, which is that the Democrats' nomination rules are socially inefficient. They are not designed to secure the collective good of the party, i.e. maximizing the chances of electoral victory. Instead, they are a hodge-podge of rules designed to satisfy the personal interests of politicians, state governments and parties, and interest groups. Each group gets a slice of the pie; all the while, the party's collective good remains unsecured. Why? Nobody is really in charge of the party.

Take a simple example. Super Tuesday saw 51% of all pledged delegates allocated. This was a bad idea. It greatly enhanced the likelihood that the Democrats would face the problem they currently do: no decisive tiebreaker. It would have been better to hold back several more significant states - California, Massachusetts, etc. - to serve as tiebreakers. In fact, the only reason Pennsylvania can serve as a modest tiebreaker is that the Pennsylvania legislature refused to go along with Rendell's idea to move the primary forward.

This is thanks to the state governments. They saw all of the attention and money that campaigns and media organizations devoted to Iowa and New Hampshire in 2004 - and they wanted a piece of the action. Never mind that these individual choices added up to collective peril for the party. That wasn't their concern, nor should we expect it to have been. It is unreasonable to expect California to submit willingly to a diminished role so that there might be a tiebreaker in case one is needed. Instead, what is necessary is some central agent who could force states to behave in a way consistent with the party's collective good.

Nobody like that exists. Nobody has the power to make sure that the rules are made to give the Democrats the best chance of winning in November. You can thank the party reforms of the 20th century for his. When progressives reformed the plutocratic party system of the 19th century, they chose not to make them wield power responsibly. Instead, they chose to disempower them. The power of the parties, which was once concentrated among the state organizations, was disseminated downwards - to state governments, interests groups, and politicians. The parties retain many of their old functions - like nominating presidential candidates - but they lack the power to ensure that those functions are performed efficiently.

There's nobody approaching a "party boss" these days, and the Democrats are paying the price for his absence.

To confirm this, I graphed Clinton's lead in the RCP averages of the Pennsylvania and Ohio polls over the 21 days prior to each state's primary. The horizontal axis represents the number of days before the primary. The vertical axis represents Clinton's lead. The blue line is for Pennsylvania; the red line is for Ohio.

Clearly, the polls in both states have behaved similarly. The only significant difference is that Obama closed Clinton's lead much earlier in Pennsylvania than he did in Ohio. He moved to within 10 points 19 days before the Pennsylvania primary. This happened about 12 days before the Ohio primary. Other than this, Clinton's lead in one state has fluctuated almost identically with her lead in the other state.

We can check this via another direction. Let's graph Clinton's performance in Ohio and Pennsylvania against Obama's performance in both states. We'll keep red for Ohio and blue for Pennsylvania. We'll use triangles for Clinton and squares for Obama.

This confirms what the initial graph indicated. Here we see that perhaps Pennsylvanians have been a bit more undecided than Ohioans. However, as of yesterday in Pennsylvania, Clinton and Obama were essentially where they were at that point in the Ohio cycle.

What does this mean moving forward? I have no idea! Clinton could improve relative to Ohio. She could worsen. Yesterday gives us no assured indication of today or tomorrow. But it does put the Pennsylvania race to date in context.

Of course, it should not come as a big surprise that Pennsylvania and Ohio have behaved similarly to date. There are reasons to expect them to move in tandem. For those interested in an in-depth review of the Keystone State primary, I think this analysis I wrote last month is still largely valid. If you haven't read it already, you might find it worth your while.

That being said, I disagree with the suggestion that this is a new feature to these debates. Hardly. This is par for the course. I recall the debate on October 30, 2007 - hosted by MSNBC - that focused relentlessly on Clinton. When the focus was off Clinton, the questions were weak.

What connects these debates? Both were focused on the frontrunner. In October, Clinton was in the lead. So, she took the heat. Today, Obama is the frontrunner. So, it's his turn. The media being the media, it asks hot questions not about policy - but about the silly campaign soap opera. That's what it does.

Unfortunately, Obama did not help himself. In fact, the hard time he received was partially his fault. There were two critical instances of this.

The first came near the beginning of the debate. After the moderators asked each candidate about the veep slot, Gibson gave Obama an opportunity to clarify his "bitter" comments. He did a good job. He said:

Well, I think there's no doubt that I can see how people were offended. It's not the first time that I've made, you know, a statement that was mangled up. It's not going to be the last.

But let me be very clear about what I meant, because it's something that I've said in public, it's something that I've said in television, which is that people are going through very difficult times right now and we are seeing it all across the country. And that was true even before the current economic hardships that are stemming from the housing crisis. This is the first economic expansion that we just completed in which ordinary people's incomes actually went down, when adjusted for inflation, at the same time as their costs of everything from health care to gas at the pump have skyrocketed.

Good answer.

Gibson then gave it over to Clinton for a response. She hit Obama, but not terribly hard. She said:

I don't believe that my grandfather or my father, or the many people whom I have had the privilege of knowing and meeting across Pennsylvania over many years, cling to religion when Washington is not listening to them. I think that is a fundamental, sort of, misunderstanding of the role of religion and faith in times that are good and times that are bad.

And I similarly don't think that people cling to their traditions, like hunting and guns, either when they are frustrated with the government. I just don't believe that's how people live their lives.

Now, that doesn't mean that people are not frustrated with the government. We have every reason to be frustrated, particularly with this administration.

But I can see why people would be taken aback and offended by the remarks. And I think what's important is that we all listen to one another and we respect one another and we understand the different decisions that people make in life, because we're a stronger country because of that.

Now, let's get real. That's not too hard. I mean, Obama really stepped in it with those "bitter" comments, and he should expect to pay. He should have let this be the last word. Let his people do the serious pushback, but get the debate off this subject. Take this hit - it wasn't that bad! - and just move on.

For their part, Gibson and Stephanopoulos seemed satisfied. They moved on to a tough question for Clinton. Stephanopoulos asked: "Senator Clinton, when Bill Richardson called you to say he was endorsing Barack Obama, you told him that Senator Obama can't win. I'm not going to ask you about that conversation. I know you don't want to talk about it. But a simple yes-or-no question: Do you think Senator Obama can beat John McCain or not?"

Stephanopoulos boxed Clinton in here. Either she says something that she can't say in public, or she looks two-faced. She chose the latter, thus yielding a free point for Obama.

They then turn it over to Obama for a rebuttal, which would have been a great opportunity to talk up his electability. What did he do? He returned to the "bitter" comments. He said:

Let me just pick up on a couple of things that Senator Clinton said, though, because during the course of the last few days, you know, she's said I'm elitist, out of touch, condescending. Let me be absolutely clear. It would be pretty hard for me to be condescending towards people of faith, since I'm a person of faith and have done more than most other campaigns in reaching out specifically to people of faith, and have written about how Democrats make an error when they don't show up and speak directly to people's faith, because I think we can get those votes, and I have in the past. [SNIP]

So the problem that we have in our politics, which is fairly typical, is that you take one person's statement, if it's not properly phrased, and you just beat it to death. And that's what Senator Clinton's been doing over the last four days. And I understand that.

This wasn't a convincing hit. He certainly didn't say anything that he hasn't said many times already. Of course, since he attacked Clinton, Gibson gave her a chance to respond. This was when she really nailed him:

Well, first of all, I want to be very clear. My comments were about your remarks.

And I think that's important, because it wasn't just me responding to them, it was people who heard them, people who felt as though they were aimed at their values, their quality of life, the decisions that they have made.

Ouch.

Obama was hit later on when Gibson went after him for his staff's petty emails about Bosnia. So, his response actually induced two clean counter-punches. The frustrating thing watching this is that his initial response was great. He should have let that be the end of it. Instead, Obama chose not to let this go. Rather than take the hit and move on, he hit back - and he was worse off for it.

This happened again, immediately after this exchange ended. Gibson moved on to ask Obama about Jeremiah Wright.

Senator Obama, since you last debated, you made a significant speech in this building on the subject of race and your former pastor, the Reverend Jeremiah Wright. And you said subsequent to giving that speech that you never heard him say from the pulpit the kinds of things that so have offended people.

But more than a year ago, you rescinded the invitation to him to attend the event when you announced your candidacy. He was to give the invocation. And according to the reverend, I'm quoting him, you said to him, "You can get kind of rough in sermons. So what we've decided is that it's best for you not to be out there in public." I'm quoting the reverend. But what did you know about his statements that caused you to rescind that invitation?

This was a fair question. Obama, for his part, gave a good response. He didn't answer Gibson's question directly. Instead, he pivoted to a nice discussion of how his campaign is a people's movement. So, two times he did well with the initial response. His problem, once again, came in his follow up.

As per usual, Gibson gave Clinton a chance to respond. But he asked her a tough question about what she said:

There are 8,000 members of Senator Obama's church. And we have heard the inflammatory remarks of Reverend Wright, but so too have we heard testament to many great things that he did. Do you honestly believe that 8,000 people should have gotten up and walked out of that church?

Clinton being Clinton, she wriggled out of this in a single sentence. "I was asked a personal question, Charlie, and I gave a personal answer." Then, Clinton being Clinton, she hit Obama for the Wright thing. It was a good hit - about Wright's post-9/11 comments. It was harder than her first hit on him about "bitter."

Once again, Obama could have responded or moved on. Once again, he chose to respond. I would have chosen to move on. The Wright thing has not done any appreciable damage yet, so why belabor it? For some reason, that's what he chose to do. Gibson, who apparently didn't like that Obama had not answered his initial question, would have none of it. This was the exchange:

SENATOR OBAMA: Well, let me just respond to -- to two things. Absolutely many of these remarks were objectionable. I've already said that I didn't hear them, because I wasn't in church that day. I didn't learn about those statements until much later.

But --

MR. GIBSON: But you did rescind the invitation to him --

SENATOR OBAMA: But that was on -- that was on something entirely different, Charlie. That -- that was on a different statement. And I think that what Senator Clinton referred to was extremely offensive, to me and a lot of people.

This response was poor. It inclines one to ask, "Well - what 'entirely different' remark were you worried enough about to uninvite him?" Not a good exchange for Obama, who continued with a broad point about Wright. Stephanapoulus then asked him a tough-but-trivial question about whether his pastor loves America. By this point Obama was a little dazed. He went on to say that he had "disowned" Wright. Of course, he had done precisely the opposite, but he knows that. He was simply off his game by now.

[Update, 3:40 PM. Upon prompting from an emailer, I took a closer look at the relevant passage, and found that Obama does not actually say he had "disowned" Wright. I thought he had accidentally said that, and had just flubbed it. He didn't. Instead, his sentence was just awkward and his word choice poor. The relevant sentence is, "And, you know, the notion that somehow that the American people are going to be distracted once again by comments not made by me but by somebody who is associated with me, that I have disowned, I think doesn't give the American people enough credit." This is what induced the request for clarification from Gibson, after which Obama stated that he meant the remarks, not the man. The first problem is the choice of word, "disown." The second problem is placing the "I have disowned" clause after "somebody," not after "comments." Both injected some ambiguity, and were thus signs that he was off.]

I would have been, too. By this point, he had taken some serious blows from Clinton and the moderators. But who is to blame? We can, and perhaps should, give Gibson and Stephanopoulos a rough time for hitting him, though it is silly to say that this is unique. The only unique thing about this is that Obama, not Clinton, was taking the shots from the moderators.

What's more, Obama clearly committed some tactical errors here. He chose to get into it with Clinton on both subjects. He had an opportunity to take his lumps and move on - but he refused. And what happened both times? He was hit even harder.

This is Clinton's territory. She's completely comfortable down there in the muck. She even seems to like it. Journalists like it, too. It gives them an opportunity to seem smart and tough. So, unsurprisingly, Clinton, Gibson, and Stephanopoulos were all trying to draw Obama into it. Obama is not comfortable down there, and it showed. His mistake was being tricked into going down there.

Clinton often reminds me a bit of George Foreman. She has one mode in these debates: pound the crap out of everything in her path. You'd think that this would give Obama an opportunity. Like Ali in the Rumble in the Jungle, he could bide his time until she over-extends herself, then nail her. But that didn't happen. Last night, Clinton managed to draw Obama into a slugfest.

For her part, Clinton knew she had gotten the better of Obama, and chose to back off. This was the most incredible moment in the debate. It came after the Ayers exchange. Obama had just responded with the comment that her husband had pardoned some members of the Weather Underground. Then:

MR. GIBSON: And Senator Clinton, I'm getting out of balance in terms of time.

SENATOR CLINTON: I've noticed. (Laughs.)

MR. GIBSON: And you're getting shortchanged here. And so if you want to reply here, fine. If you want to wait, we'll do it in the next half hour.

SENATOR CLINTON: We can wait.

My feeling is that we don't know what he meant. All of this analysis is based on brief, off-the-cuff remarks made behind "closed" doors. It is difficult to tease out a man's inner philosophy from such a slender data set. At best, we can only hope to have a vague sense of the thoughts that inspired the words. So, the quotation has been a bit of a Rorschach test. Commentators have seen what they are inclined to see.

This is one reason why, politically, it was a stupid thing to say. Candidates should not say vague things unless there is an identifiable benefit, like evading a journalist's direct question. Otherwise, clear and concise is the way to go. Vagueness implies interpretation. Interpretation implies discussion. Discussion eats up precious news cycles a week before the Pennsylvania primary.

And, of course, vagueness invites opponents to interpret, and therefore define. This is what we've seen. Clinton smartly chose to put her interpretation in the mouths of ordinary Pennsylvanians - and Obama was forced to respond by using the "boo's" at the Alliance for American Manufacturing meeting as the alternative interpretation. In one sense, this has already been a victory for Clinton. What is Obama spending money and airtime on? His comments in San Francisco! The Obama campaign seems to have learned one lesson from the Kerry/Edwards debacle. It doesn't let accusations go unanswered. However, it has not yet learned the other one. It needs to be more proactive in managing the definition of its candidate.

The other big problem with his comment is that Obama presumed to explain the behavior of the voters he is courting. We might not know for sure exactly how he was explaining them, but we know that he was trying to. This is something that is best left to political scientists, not candidates. They should never speak of voters in any but the most flattering terms. Otherwise, there is a risk of alienating them. When you analyze people, you are signaling that you are separate from them. You are an "other." What is more, nobody likes to feel that they are being analyzed. The analyst can come across as haughty. "Who the hell does he think he is to explain me?"

This is not the first time Obama has done this. His Wright speech sought to explain the behavior of the voters - black and white - he was courting. He really needs to knock this off. It is not the job of the candidate to analyze the voters. His job is to court them, to form a bond with them. He must have them believe that he understands them on their terms, not on some set of abstract principles derived from a book they've never heard of.

This is one reason "Bubba" and "Dubya" have won the last four presidential elections. Nobody ever tied those two to Theodore Adorno.

Of course, Obama did not analyze just any group of voters. He analyzed the ones Democrats need: whites who don't make a lot of money. In 1992, Bill Clinton and Herbert Walker essentially split the white vote. Clinton got 39%, Herbert Walker got 40%. This is all Democrats need. They don't need to win white voters outright. They just need to split them. Flash forward to 2004. Bush beat Kerry among white voters, 58% to 41%, and won a solid victory.

You can tell the same story again and again. When Democrats break even with white voters, they win, as in '60, '64, '76, '92, and '96. When Republicans win them decisively, Democrats lose. This happened in '52, '56, '80, '84, '88, '00, and '04.

So, what Obama really did last week was analyze the group that will swing this election.

I'm beginning to wonder if analysis is a problem for candidate Obama. All candidates have quirky "ticks" that impede them from being perfect campaigners. George W. Bush has a habit of mangling words. John Kerry has a habit of going off script. Al Gore has a slightly condescending tone to his voice. These are all basically knee jerk responses that candidates do without thinking. They don't mean anything by them; they're just "ticks." But they still distract people. Obama might have a knee jerk inclination to analyze. Maybe I'm wrong, but we have seen this happen enough to make me wonder. After all, he did publish an autobiography when he was just 34. Maybe he is an analyst by nature.

In a lot of other contexts, this is a highly desirable trait. But not in this one. He needs to stop this. So does his wife, who should never again tell us that we have a "hole in our souls." If this kind of stuff continues, Clinton and the Republicans might just get that "elite" label to stick.

There is some good news for Obama in all of this. The Pennsylvanians whom Obama was analyzing were mostly going to vote for Clinton, anyway. So, it's unlikely that the comment will damage him on Tuesday. It might cost him a point or two, but that's probably it. Of course, the reason this nomination battle is continuing through Pennsylvania is Obama has failed to woo lower income whites, the same voters he'll need in the fall.

When not questioning her memory of Bosnia or her chief strategist's conflicts of interest, people have been asking how Clinton can actually win. Few think she is likely to. Those who give her a chance, such as myself, can only imagine her winning "dirty" in Denver, muscling her way to half-plus-one via the super delegates. No Democrats, aside from Clinton's most ardent supporters, want that. It implies a nominating campaign through the end of August and a debacle on national television.

Of course, there is a group who can stop this from happening - the uncommitted super delegates. If they swung to Obama in large enough numbers, they could effectively kill Clinton's campaign. If 50 or more of them endorsed Obama in a short span of time - Clinton would have a very serious viability problem.

Perhaps unsurprisingly, we've read a bunch of stories suggesting that Clinton has a "super delegate problem." But by and large the super delegates haven't budged. Most of those who were undecided in early March are undecided as of today. According to Dem Convention Watch, Clinton had a 97-delegate lead on February 10th. By March 9th, Obama had cut that lead to 39. But since then, despite all of these stories about Clinton having no real chance, Obama has netted just 13 super delegates. As a group, the super delegates have not moved. More than 40% remain uncommitted.

I think this is curious. They surely do not want a bitter convention battle, so why haven't they brought an end to this? I think their reticence has to do with Obama's terrible performance in Ohio. He not only lost, he was roundly defeated - even after his great victories in Wisconsin, Virginia, and Maryland. The nature of his defeat might be giving the super delegates pause.

Specifically, Obama's problem in Ohio was with white voters. Consider the following chart:

As you can see, Obama did worse in Ohio among whites than in these other major states. Again, what is so intriguing about the Ohio result is that it came amidst stories of how Clinton was finished. That curiosity continues. Analysts give Clinton very long odds - but Pennsylvania Democrats haven't hopped aboard Obama's bandwagon.

Unfortunately, the exit polls only tell us so much. Nevertheless, we've seen enough data to know which socioeconomic groups he's having trouble with: rural/small town whites who do not make a lot of money. We can confirm this by looking at the counties in Ohio's sixth congressional district, which makes up most of the Ohio River Valley. This is the premier swing district of the 21st century. Bush won it by 5,000 votes in 2000 and 2004. Obama did horribly there last month, as the following chart details.

As you can see, Obama got blown out in the sixth. The only exception is Athens County, where Ohio University is located.

There are several Ohio districts that tell a similar tale. In the west, Obama did quite poorly in the fifth and the eighth. In the northeast, he did poorly in the fourteenth and the seventeenth. He actually did worse in the second and eighteenth than he did in the sixth. What's more, in geographically large congressional districts, you can always find at least a county or two where Clinton beat him by 25 points.

Beyond Ohio, Obama seems to have had this problem again and again, as Sean Oxendine illustrates in this incisive essay. Big wins in places like Virginia and Mississippi often belie a weakness with the same types of voters.

This could be a potential problem for Obama come November, but the reason is not obvious. Democrats should not worry about whether the primary voters who supported Clinton last month will support Obama in November. They probably will. Voting in a primary election is a sign that the voter is a strong partisan, and therefore unlikely to support the opposition in the fall. Nor, for that matter, are they likely to abstain from voting.

Rather, the concern for Democrats is whether Obama's poor performance among white, strongly partisan Democrats is a sign he will be weak among white, persuadable voters. We're talking about weak partisans and Independents. They're the ones who swing elections in Ohio. Obviously, they differ from strong Democrats in terms of partisanship - but they still have many socioeconomic characteristics in common with them. The weak partisans and Indies are the relatives, friends, neighbors and coworkers of the strong Democrats who voted so overwhelmingly for Clinton last month. While the persuadables do not share their strong partisan orientation, they might share the same disinclination to Obama. The strong partisans expressed it in March by voting for Clinton; the weak partisans and Independents might express it in November by voting for McCain.

The operative word here is might. This is only a possibility. Nobody knows whether these primary results are indicative of the general election. To argue they definitely are requires an inferential leap that we simply cannot take. These primary results could signal trouble for Obama in November, but they could just as easily signal nothing at all.

In other words, these primary results only raise the questions. They don't provide the answers. But when we're examining the super delegates, that's exactly the point. These questions might be holding them back. Perhaps they want the nomination battle to continue so they can get some answers. Perhaps what Obama needs to do is simply improve his showing with these voters in other states. That would show the super delegates that, when it comes time for the general election, he can compete in places like Ohio's sixth district.

He'll have plenty of chances - what with Pennsylvania, Indiana, Kentucky, and West Virginia all coming up. There are lots of these types of voters for him to win over, and thus lots of chances to show that the Buckeye State was an outlier, that it just takes him longer to catch on with these folks.

This is why his comments in San Francisco were so unfortunate. If they are going to turn off anybody, it's the people we've been discussing. Will they? It is too early to say. We should know in a day or so. If they do, they'll impede his cause in Pennsylvania. He needs to do just one thing in the Keystone State, and it isn't win. He just needs to pull in some respectable numbers among white voters. They don't need to be as good as they were in Wisconsin - just something closer to Maryland and Texas. Failing to do that in Ohio meant that he failed to deliver the knockout blow to Clinton then and there. If he fails again in Pennsylvania, the race will go on. And the longer it goes on, the better chance she has of coming from behind and taking this nomination.

There is no doubt that it has been a poorly run campaign. But what has been so bad about it? We could point to a lackluster message, or Bill's various gaffes over the last three months, or the staff that couldn't stop watching soap operas long enough to pay the bills. There's something to all of these things, but I think they are symptoms of an underlying malady.

As you well know, Obama has a huge lead in pledged delegates. But you might not know that nearly 90% of this lead comes from caucuses. Obama has netted 147 delegates via the caucuses alone.

It need not have been this way. Caucuses have exceedingly low turnout - and so victory depends upon organizational prowess. Clinton was poorly organized in the caucus states, and it cost her. For every caucus state she has lost, Clinton could have found enough supporters in those states to at least tied Obama. This is the case even in states where Obama would win a broad-based primary. In Kansas, for instance, Obama had about 18,000 more caucus supporters than Clinton. Grant that Kansas is a state Obama would win in a primary. Shouldn't Clinton have been able to find 18,000 more people? She received less than 10,000 votes for goodness sake!

Even more amazingly, Obama crushed her in states where Clinton probably would have won or barely lost a primary. Obama netted 15 delegates on her in Colorado. He netted 6 in Maine. I'd put the odds of Clinton winning primaries in those states at no less than 50-50. Clinton won the Texas primary outright, but Obama walked away with 9 more delegates in the caucus. Obama won 26 more delegates than Clinton in Washington state. One week later, he beat Clinton in the Washington state beauty contest by just 4 points. Clinton won a clear plurality of voters in the Nevada entrance poll, but Obama walked away with a net of one delegate.

This is an organizational failure of monumental proportions. There is no other way to put it. The question is why did it happen?

There is no great skill that the Obama campaign possesses that the Clinton campaign lacks. Organizing caucus states still has a lot in common with 19th century politicking. You need a friendly smile, a good handshake, and a sturdy pair of shoes. Obama didn't develop a new way to organize. He just chose to organize while the Clinton campaign chose not to.

The only reason it would choose not to organize is if it did not think it was worth the cost. More than 400 pledged delegates have been allocated through caucuses. So, it wasn't worth it because it was insignificant. Then why didn't her campaign organize? I believe it is because it never thought Obama would mount this kind of challenge. It never thought it would have to scrap for every spare delegate. Instead, it thought the race would be over before Super Tuesday.

In other words, the Clinton campaign did not see Obama coming. It underestimated him.

Of course, much of the Washington press corps didn't see him coming, either. But that's not terribly surprising. Washington is their beat, and many of them don't have a great read on how politics outside the Beltway works. But politicians are different, or at least they should be. They should be in touch with life outside the Beltway, and they should know better than journalists. The Clinton campaign should not have underestimated Obama. There were warning signs that it should have picked up on.

First and foremost, Obama raised a gazillion dollars last year, none of which came from PACs. This was an early warning of many things. First, his campaign operation was going to be awesome. It could basically match Clinton dollar-for-dollar without the benefit of a former President. If it could fundraise that well, the Clinton people should have expected it would campaign that well, too. Second, this was an early indication that Obama was resonating with people out there. Political donors are a miniscule subset of the electorate, but in terms of demographic and socio-economic characteristics they have a lot in common with a broader set of people - namely, upscale voters who can figure prominently in primaries. Unsurprisingly, Obama has been winning upscale voters coast-to-coast.

The media wrongly turned the race for money into a proxy for the race for votes. But the Clinton people should have known better. They should have said, "Holy crap! This guy raised a gazillion dollars! We better get things locked down!" I don't think the Clinton people ever said that. They certainly never got things locked down.

Her campaign also overestimated its own position. Once again, the media did this, too. Journalists looked at the summer and fall polls and bought into the inevitability argument. Again, this is par for the course - many journalists do not know the difference between good polling data and bad. Candidates should know the difference. The Clinton campaign should have known. It should have suspected that those eye-popping leads were merely a consequence of her superior name recognition, which would not hold after Obama unloaded his gazillion dollars. I don't think it suspected this. I don't know how else to explain Penn's snide memos touting Hillary's inevitability.

Why did it make these mistakes? Is it because it doesn't understand electoral politics? Unlikely. After all, Bill won two national campaigns.

I think its mistake was its starting point. It bought the same inevitability line it sold to the press. It began with the assumption that Clinton could not lose the nomination. If you assume this a priori, you will inevitably interpret all of the evidence in a way that reinforces your preconceived notions. It's like adding epicycles. If she cannot lose, there is no reason to worry about Obama's money, no need to anticipate that this might be an early indication of his appeal. If she cannot lose, those summer polls are not mere artifacts of her name recognition; they are critical pieces of evidence that demonstrate how the race is over before it begins. If she cannot lose, there is no need to organize in the caucus states because the race will be over by then.

What we are talking about here is plain old arrogance. I think this is the central mistake of the Clinton campaign. It presumed that the nomination was Clinton's. Not Clinton's to lose. Just Clinton's. Period. As a consequence, it behaved in an unduly confident manner. Mark Penn is to be blamed, for sure. So is Patti Solis Doyle. But so also is the entire upper-echelon of the campaign. Above all, it's Hillary's fault. She's the candidate. She sets the tone.