West Virginia is tailor-made for Hillary Clinton

by: Clem Guttata

Mon Apr 28, 2008 at 13:50:42 PM EDT

A diary last week about Obama's ability to excite swing voters in other parts of the country (NY Times Thursday: Obliterating Hillary's Electability Meme) including a fascinating graphic from the New York Times:

Below the break, I apply this decision tree to 2000/2005 census data and 2004 voting data to project Clinton's margin of victory in the state.

The decision tree

Let's work our way down N.Y. Times decision tree. The first stop is "is a county more than 20 percent black?" This is an easy one in W.Va. The answer for all 55 counties is no. McDowell County has a 10.6% black population (2006 census projection). The state average is 3.3%. This is a lilly white state. In fact, in 40 counties there are less than 1000 African-American residents.

Educational attainment

The next two questions are about the percentage of high school graduates in a county. Counties below 78% have heavily trended toward Clinton, counties above 87% to Obama. According to the 2000 census, in 41 of 55 W.Va. counties the percentage of high school graduates is less than 78%. (The state average is 75.2%). No county has a high school graduation rate above 83.8.

According to the decision tree, Clinton won 89% (704 / (704+89)) of counties with 78% or less of the population having a high school education. For our analysis, Clinton wins 89% of these W.Va. counties:

CD1: Barbour, Doddridge, Gilmer, Grant, Preston, Ritchie, Taylor, Tucker, Tyler, Wetzel

CD2: Berkeley, Braxton, Calhoun, Clay, Hampshire, Hardy, Jackson, Lewis, Mason, Morgan, Pendleton, Randolph, Roane, Upshur, Wirt

CD3: Boone, Fayette, Greenbrier, Lincoln, Logan, McDowell, Mercer, Mingo, Monroe, Nicholas, Pocahontas, Raleigh, Summers, Wayne, Webster, Wyoming

Although these are 41 of state's 55 counties, they only represent 49% of the votes cast for Kerry and Nadar in the 2006 Presidential election. (Because independents can vote in this year's Democratic party primary, I'm using the Kerry+Nadar vote to weight counties.)

Region

The next question is "where is the county?" in the "northest or south" or "west or midwest"? Now, you might think this is a really straight-forward question. In most states it would be.

Here in West Virginia, that's not so.

We've been called the Northern-most state in the South, the southern-most state in the North, and some places we're the eastern-most part of the mid-west.

I'm going to punt on the question of what parts of West Virginia are in the South. Of the remaining 14 counties I'll say that the eastern-most county of West Virginia, Jefferson (home of Harper's Ferry) is definitely in the Northeast or South. That's a 70% pickup for Clinton of another 2.8% of the electorate. (Jefferson is in CD2.)

Household income

The next question is: "in 2000, were many households poor?" Here, my census data doesn't quite match up with the N.Y. Times criteria. They ask if "at least 47% earned less than $30,000". The figure I have readily available is, is 2005 median income above or below $30,000? (They should match closely.)

Of the 13 remaining counties, 11 had median incomes in 2005 below $30,000. According the decision tree, Clinton has won 68% of counties like these.

CD1: Brooke, Hancock, Harrison, Marion, Marshall, Mineral, Monongalia, Pleasants, Wood

CD2: Putnam

CD3: Cabell

These 11 counties are another 32.5% of the 2004 vote.

Population Density

We're down to two counties -- and a less friendly factor for Clinton. Population density favors Obama. Ohio and Kanawha counties are both far in excess of the minimum 61.5 people per sq. mile threshold. Clinton only wins 29% of the counties like these (Obama wins 71% of them).

Together, Ohio and Kanawha comprise 16% of the electorate. (Ohio is in CD1, Kanawha in CD2.)

Popular Vote and Delegate Tallies

Calculating a popular vote margin based on the likelihood of winning individual counties is, admittedly, a blunt-force approach. Nonetheless, it does provide a baseline for the huge advantage demographics provide Sen. Hillary Clinton's in West Virginia.

Here are the weighted total votes by Congressional District:

CD1: 109,442 total, 74,756 for Clinton
CD2: 113,674 total, 71,305 for Clinton
CD3: 107,488 total, 91,853 for Clinton
Sum: 330,604 total, 237,914 for Clinton

The CD percentages are 68%, 63%, and 85% with 72% statewide. This yields a +2, +2, and +4 delegate advantage from the CDs and a +1 and +3 advantage in the state-wide totals -- a +12 total delegate advantage.

In summary

Demographics are not destiny. Clinton has lost ground to Obama in many of these demographics between Ohio and Pa. On the other hand, there may be other demographic advantages, including gender and age, that this analysis does not capture.

Also, this anlayis points out challenges and opportunties for both campaigns.

If Obama can keep Hillary's margin of victory in CD-3 below 75%, that's a four delegate swing. If Obama can get Hillary's margin of victory in CD-1 or CD-2 below 58.3% (quite possible in CD-2), that's another 4 delegate swing per CD.

For Hillary, the key is to run up at least one CD for to keep the +4 swing intact and to insure at least a 64.28 state-wide total to avoid a 2 delegate swing.

To date, Obama is investing more resources in West Virginia than Clinton -- he has opened far more field offices and already made a media buys. We'll know in a couple short weeks if Clinton is been able to hold on to her commanding advantage in West Virginia or not.