Monday, April 07, 2008


April 03, 2008
No Really. Hillary Has a Decent Shot
By Sean Oxendine

It has become something of a pastime among polling geeks like myself to use Jay Cost's primary vote calculator to predict the outcome of the Democratic race. Most who have played with it have come up with some kind of scenario where Hillary leads in the popular vote.

Now, I don't mean to pat myself on the back, but a few days before Jay's calculator came out, I had my own estimate coming to this conclusion. But this calculator provides some more concrete ways of estimating the popular vote. Let's look at this in more detail (especially given all the calls for Hillary to drop out).

Before we do a state-by-state assessment, people who followed me from know about my obsession with political geography. In case you didn't know, I've hand-programmed maps for every congressional election going back to 1972, with about half the states going back to their origins. I love maps and their use at displaying political data. This Hillary-Obama race gives a perfect opportunity to analyze along these lines.


So, let's look at this map:

Take a close look at this map. It is the Hillary/Obama results on a county-by-county basis for the states that have voted in primaries east of the Mississippi and that border states that have not yet held primaries (plus AL, MS, and GA, for educational purposes). The bluest counties are counties that she carried with at least 60% of the vote; the greenest counties are counties that he carried with at least 60%.

Looking on a county-by-county basis gives us a better idea what may happen in upcoming elections than the state-by-state vote. States are large, diverse places, while counties tend to be more compact and uniform, and can hence give us a better idea what is going on on the micro level.

One of the most common arguments against Hillary's ability to win the popular vote, and hence have an argument to present to Superdelegates, has been the fact that she hasn't won over 60% of the popular vote in more than a handful of states. To win the popular vote, she will have to win over 60% of the popular vote in at least a couple of states, maybe more. Contrary to many people, I think this is possible, given this map.

Take a look at some things. First, take a look at Illinois. One common story is that he will have the advantage in Indiana because it borders Illinois. There is probably some truth to this, due to spillover media. But look closer. In truth, Hillary carried large portions of Southern Illinois. Obama's huge advantage was built almost entirely in the Northern section of the state, near Wisconsin (which he also won big). As he got further South, he wasn't able to carry that portion of the state, even with his "home state advantage." In Southern Indiana -- where (at least) two of the state's nine Congressional districts are located -- districts with heavy Democratic registration among white, rural voters -- Hillary will have a large base of operations. But more on that later.

Also, look at MS, AL, and GA. These states are often thought to be purely Obama country. But this map shows that Hillary did quite well in the Northern portions of those states. The same is true of Western Virginia. Again, more later. This is just to set the stage.

South Carolina is a little more difficult, since it is a three-way race. I added a red tint for John Edwards' performance in the counties. We can still see the same basic trend -- as we get farther from the coast, the counties tend to get less green (except for the awesomely-named Horry County in the Northern-East section of the state -- Fertile . . . I mean, Myrtle Beach is there and the county as a whole is whiter than the surrounding areas).

Now part of what is going on here is the so-called black belt which runs from Texas across to Virginia. Originally named for the black, fertile soil, it has since become an appropriate descriptor for the racial composition of the counties as well. Looking at this map:

We can see the nice correlation between race and Obama's performance. Obama did not fare well in Northeast MS, Northern AL, and Northern GA (as well as Western VA) at least in part because there is such slight AA presence there.

Indeed, consider this blowup map of MS.

Note how Obama performed best in the areas in the West and middle of the state -- in the Northeast and Southeast, his performance was not as strong. These also happen to be the whitest portions of the state.

But that can't be the entire story. After all, WI is almost uniformly white, and Obama performed well there. He did well in portions of Ohio.

But there is another aspect to this as well. Consider Virginia. Here is a blow-up of how the state voted.

Now, let's try something different. Let's overlay a topographic map of the state with the voting map.

And now the trend becomes clearer. As we get into mountainous country, into Appalachia, Obama's percentages, even in good states, drops precipitously.

In Mississippi, we see a similar result:

Here, the Delta region is plain as punch; when the hills start, we start to also see Obama's vote share drop (this is mitigated in the south-central portion of the state due to the aforementioned black belt running through). Thus, it isn't just the whiteness of districts, it is white, poor, Appalachian counties that constitute Hillary's wheelhouse. Now again, this isn't anything new. Its just that the maps drive home just how stark this divide is.

So what does this mean for us. Well, consider THIS map:

The black line traces the exact contours of Appalachia, according to the United States Government. You can see the nearly perfect correlation here with Hillary's vote performance. The correlation is scary in some states, particularly Ohio, Georgia (where the Atlanta suburbs provide the only mitigation) and Virginia; in most states it gives a pretty good description of where Obama starts to have a chance Note: TN is all blue, save for the Southwest portion and the greater Nashville area in the center. This is because, although Appalachia stops east of Nashville, you are still very much in hill country until you get to the western third of the state, where Obama at least occasionally has a good showing. BTW, did you know that Northwest TN was Davy Crockett's old district? True story.

Anyway, this map doesn't tell the full extent of the story -- in many of the TN, OH, and VA counties, Hillary was well over 60% of the vote, even reaching 90% in one Virginia county. In other words, even though she hasn't reached 60% of the vote in many states, she's done it in several counties. The question is: Are there states with enough Appalachian counties left to push her across the 60% threshold? I think the map above tells us "perhaps so."

Given this background, let's return to Jay's calculator.


Jay's default is the 63%-of-the-Kerry vote turnout that has been the norm for closed primaries to date (and let us be clear -- this just Jay's default, not his actual judgment). I think this is unlikely to be the case. I think it will be much higher. PA is going to be the center of the political universe for the next few months. It is the only game in town. The GOTV operations will be in full swing. And people have been registering and switching parties at a furious pace. I am going to put it turnout at 80% -- closer to an open primary.

As for the predicted results, I will go ahead and use the estimate from this excellent blog at MyDD. I think it is an excellent analysis. I happen to think that it is a little bit generous to Obama in its classifications. Consider that in the RCP average, he has been flatlined at around 36% for almost a month now. Indeed, he has never been above 43% of the vote here in any poll. Hillary's numbers have bounced around significantly.

I have seen numbers like this before, and it was the Bush-Kerry race, where Kerry was usually static around 46, 47 percent of the vote, while Bush bounced around. It is common in races where people aren't sold on the "incumbent," but aren't sure about the challenger. In the end, the incumbent usually gets his or her votes. This is why last-minute voters break so heavily for Hillary. Throw in a modest Bradley effect, and she could win here by over 20 points.

Regardless, let's assume 80% turnout, with a 16-point Hillary win, and let's move on.


People are assuming that Obama will do well in Indiana, in part because of the close proximity to the Chicago media market. We can test this somewhat. Take a look at Western MO.

Looking at this, I am fairly confident that Obama did not enjoy any significant spillover in "home state" support. I can tell much more easily where St. Louis and the University of Missouri are located than I can Illinois. There is some "greening" of the map as we move further North, but we would expect this, given results in IA and WI. This is consistent with what we saw in Illinois as well, except that the state as a whole was biased toward Obama (as one would expect from his home state), so downstate was still somewhat green. NOTE: You could do this for WI as well, but it will be hopelessly biased, since the counties bordering IL comprise the great Milwaukee and Madison areas. But even here, his performance in counties bordering IL runs about 10% below what they did in Illinois counties just across the border.

So I don't accept much of a spillover. Moreover, I suspect that the eighth and ninth districts in the South will go heavily for Clinton, though the University of Indiana may give him a bit of a boost in the Ninth. Moreover, looking at the Ohio maps, she ran about ten points ahead of Obama in the counties bordering IN -- which bodes well for his performance in the Sixth and Third. The Fourth and Fifth are harder to predict, though given their small-town and rural feel, I would guess there would be a good turnout.

Many have speculated that Obama will do well in the First and the Seventh. I suspect he will do well in the latter (Indianapolis). Marion county is about 1/4 black, and much of the inner suburban area is in this district.

But what about the First, with Gary? In truth, Lake County is about 26% black. But the rest of the county is ethnic whites -- not exactly Obama's wheelhouse. Moreover, there is a long history of racial tension here, with the current congressman, Peter Visclosky, defeating an appointed AA representative back in the 80s, much to the chagrin of local AAs. In short, I'm not sure how many votes Obama really gets here.

Which to my mind, shapes up to a Clinton win. I will call it 8%, but that is just a guess, and I am guessing that it is on the pro-Obama side.

North Carolina

North Carolina is an exceedingly difficult state to predict. As Jay Cost has noted, it is demographically varied, and has several population centers. I agree with him. So I will guesstimate. Looking at my map above, it is more Appalachian than South Carolina by far -- the region only grazes the northeastern tip of that state (where Obama did his worst by far). It is more so than Virginia or Georgia as well. Adjusting for the greater App. presence (three Congressional districts -- 1/4 of the state's total -- fall largely within the region) let's bring Obama's margin down from the 63% or so he got in GA and VA to around 59%.

Now you have to remove NoVa and Atlanta. Charlotte, Raleigh-Durham, and the Piedmont Triangle (Winston Salem, etc), have about 1M people. NoVa has about 2.5M, 1M alone of whom are located in Fairfax County. There is nothing on this scale in North Carolina. The three large NC areas are more akin to areas like Hampton and Richmond in terms of their impact on the state, which still went heavily for Obama, but weren't responsible for his margin -- that came from NoVa. Add in that the state is much whiter than GA (about ten points), and I think a 12-point Obama win is reasonable.

Now we come to WVA and KY. Here is where Hillary gets her big wins. WV is entirely Appalachian. It is one of the whitest states in the country. It is also one of the poorest. And the most Democratic in terms of registration. Hillary routinely won upwards of 70% of the vote in neighboring counties -- in fact, I believe she has carried every county which borders the state except for six counties in VA (one of which is Loudon, a DC suburb).

Charleston is something of a population center, but it is more like Knoxville in TN, where Obama pulled even, but no more. Add in gritty blue collar towns like Wheeling, Morgantown, and Huntington, and it is not impossible to imagine a 70%+ Hillary result. But let's call it more in the mid 60s -- say 65%, for a 30% win.


KY is similar. Remember, Hillary did well in Southern Illinois, and walloped Obama in Southeast Missouri. She received around 80% of the vote in several counties along the TN-KY border. And Southern Ohio was Hillary country as well.

He will likely perform strongly in Louisville and the Cincinnati suburbs, and may have a good show in Lexington. But the 1st (Jackson Purchase), 2nd (Pennyrile) and Fifth (Appalachia) will be big Obama Clinton counties. The "Old Seventh," which is basically an extension of West Virginia will run up huge Hillary margins. And notice that in KY, rural counties make up entire Congressional districts.

Basically, I think we're looking at a strong Hillary showing on the order of 15-20%. Maybe higher. We'll call it 20% and move on.


I will go ahead and use Jay's defaults of 10 point Obama wins in MT and SD, and a 5-point win in OR. To tell the truth, I have no idea what will happen in these states. Yes, Obama crushed in caucuses in neighboring states, but they were caucuses. These are primaries. I have no frame of reference here. MT in particular is difficult to predict, since there is a heavy blue collar Dem base in the unionized, mountainous West. OR is whiter and poorer than CA or WA, and Hillary did well in the non-binding WA primary (at least compared to the caucus result). Regardless, these are small states. Leave them as is.

Puerto Rico

That leaves Puerto Rico as the wild card. No one knows what will happen here. The Democratic and Republican parties don't exist in the state. Hillary has the support of most of the politicians here; Obama had the Governor, but he was just indicted. Moreover, it is dicey using results from Mexican-Americans in TX and AZ here; Puerto Ricans are different culturally, linguistically, and even ethnically.

Then again, PR has a strong connection with NY, and she performed well in the PR community in her primary, as well as in Hudson county in New Jersey. I haven't seen any indication that Obama has performed well with Puerto Ricans anywhere. And there's the FALN pardons. So Cost's 25% victory for Hillary sounds about right, maybe even conservative.

That leaves the question of turnout. This is the million dollar question. We have no way of knowing how many people will turn out. The island has 4M people. Cost put his default at 1M. I think it will almost certainly be higher. In 2004, turnout for the delegate to the Congress -- a non-voting member -- resulted in 2 million votes cast. FOR A NON-VOTING MEMBER. Imagine the excitement -- in a state where American party labels mean nothing -- to be able to pick a President. I think the 2004 delegate race is a floor. We'll set it there, understanding that it might go higher.


All told, this gives Hillary around a 100K margin of victory, using Obama's best count system (use caucus estimates, don't us FL or MI). In truth, I think the best system credits FL -- both were on the ballot, neither campaigned, and even though the delegates don't count, the votes were still cast. Under this count, she wins by almost a half million votes -- exactly Gore's popular vote win over Bush.

And therein lies the rub. Are the Democrats, who still feel victimized by 2000, going to go with the person who very narrowly won the bizarre system of delegate allocations? Who won because of Texas' primacaucus, and the refusal to seat FL and MI?

But then again, are they going to not nominate the first African-American who won the most delegates? Or will they gamble on the notion that balck's will still turn out Democrat, or will at worst stay home, while the women and blue collar Dems who supported Hillary might really vote for McCain?

It's a mess for Democrats under that scenario. I don't know what the Superdelegates will do. Heck, we don't even know who all the Superdelegates are at this point! But she has a reasonable pathway to get to this point. And once she gets there, it is anybody's ballgame.

(BTW, also note that calling for her to drop out before PA, PR, KY and WV vote would be akin to asking Obama to drop out because he was trailing with GA, AL, and MS yet to vote. It's cutting her off before literally her best states vote).

UPDATE: I should also add that this isn't a prediction, it is a scenario. Predictions are foolish at this point. But I think it is a plausbile scenario, and it is why she continues to fight.

Oxendine writes The Race 4 2008 Blog.

Page Printed from: at April 07, 2008 - 11:55:25 PM PDT


Post a Comment

<< Home