About this Blog

RealClearPolitics HorseRaceBlog

By Jay Cost

HorseRaceBlog Home Page --> Swing State Review

Return To Pennsylvania

Michael Barone published an excellent essay on Pennsylvania earlier this week. In it, he analyzed McCain's prospects in the Keystone State, and from there pivoted to a broader discussion about the governing coalition Obama hopes to form.

I cannot recommend it more highly. Here, I'd like to supplement it, hoping to specify McCain's strategy in the Keystone State and its likelihood of success.

I suspect that McCain is aiming for a voting coalition roughly similar to the one Hillary Clinton fashioned in the April, 22 primary.

2008 D Primary Pennsylvania.jpg

As we can see, Obama performed well in the southeast, winning three of the five counties in Metropolitan Philadelphia. Meanwhile, Hillary dominated the rest of the state, winning the rural areas, metro Pittsburgh, and the smaller cities along the Northeast Extension of the Pennsylvania Turnpike - Allentown, Bethlehem, Wilkes-Barre, and Scranton.

Indeed, another look at the primary data indicates just how well Clinton performed in these areas.

Clinton Margin.jpg

Those are some very large margins of victory outside Philadelphia. Pittsburgh presents an interesting case. Thanks to strong African American support in the city and the neighborhoods in the east, Obama won 45% of the vote in Allegheny County. However, outside Allegheny County, he did quite poorly. He only cracked the 33% threshold in Butler County, which is an exurban county (Obama tended to do well in exurban counties nationwide in the primaries). In Fayette County, he barely managed to win one of five primary voters. As we can see from the map above, Obama's generally performed quite poorly throughout the western third of the state.

This data seems to indicate an opportunity for John McCain and Sarah Palin. If primary voting Democrats broadly preferred Clinton over Obama, it might be that Independents and persuadable Democrats can be brought to the GOP. But can they really? Can McCain make any progress in these areas? The fact that both sides have been stumping in the state recently suggests that the answer might be yes. In light of this, let's examine the following chart - which tracks the "Republican tilt" of these areas since 1964.

Republican Tilt.jpg

As we can see, the rural areas have consistently supported the Republican candidate at levels greater than his national average. Allentown and the cities along the Turnpike Extension have been pretty consistent over the last 40 years, displaying a Democratic tilt between 1 and 10%. Meanwhile, Pittsburgh is once again an interesting case. Contrary to conventional wisdom, Pittsburgh is not home to "Reagan Democrats." The steel industry took a huge hit in the early 1980s, and the town responded by favoring Mondale by nearly 13 points. George W. Bush, however, did very well here, performing better than any Republican since Richard Nixon in 1972. That had a lot to do with the region's cultural conservatism.

So, there is potential for McCain. Republicans tend to do all right in these ares, so McCain might have an opportunity here to do well. If he hopes to win the state, he would have ultimately have to pull more votes in these three places combined than Bush did in 2004.

However, this strategy faces a major obstacle: Philadelphia. Barone notes that the southeast, according to the most recent Survey USA poll, seems primed to favor Barack Obama by a margin greater than what John Kerry pulled. He offers a very cogent theory for why:

My hypothesis is that that is because places like the Philly suburbs are places where the recent decline in household wealth has been most conspicuous. Housing prices mean a lot more to you when your house started off at $400,000 and declined to $290,000 than they did when you started off (as may be typical of Scranton or a blue-collar town in metro Pittsburgh) at $140,000 and declined to $110,000. Newspaper coverage of our current economic distress focuses on the very poor (like a recent Washington Post story on North Carolina, which focused on an ex-convict in a cheap motel in Charlotte), but the people who are getting hurt most visibly in their lifelong project of accumulating wealth are the more affluent. They're the ones whose house values have most visibly and spectacularly declined, and whose 401(k) accounts and stock portfolios have tanked in the last few months as well. Folks in Scranton or in the cheap motel in Charlotte didn't expect to live comfortably ever after off their increased house values, 401(k)'s, and Merrill Lynch accounts; a $700 monthly check from Social Security is about what they have long expected and that's not in danger (yet). Folks in the Philly suburbs did expect to live comfortably off such assets.

It seems to me that this hypothesis might be generalizable to McCain's weaknesses in Colorado, Virginia, and North Carolina - where wealthier suburbanites have taken a huge hit in the last month.

Barone notes another problem for McCain in Philly - as the above chart makes clear, the trend-line has been bad for the GOP. George H.W. Bush did quite well in metro Philly - but since then it has been downhill for the Republicans. George W. Bush lost the state twice in part because he twice lost 4 of the 5 counties in metro Philly. Additionally, as the primary data indicates, Obama did quite well here. He carried Philadelphia County thanks to strong support from African Americans. He also won Chester and Delaware counties, and he split Bucks County with Hillary Clinton. Clearly, the problems Obama had in the northeast and the southwest were not present in the southeast.

There might be an additional complication for McCain - the fast-growing counties surrounding metro Philly. Since 2000, the Census Bureau estimates that Berks, Lancaster, and York counties have grown anywhere between 5% and 10%. Most recent estimates indicate that more than 1.2 million people live in these places. Bush carried Berks County by about 6 points in 2004, and he dominated Kerry in Lancaster and York - winning both by near 2:1 margins. So, for McCain, victories in these places are already built into a statewide Republican loss. He must hold the line here. This could be problematic if Barone's hypothesis about the political implications of the financial crash is accurate.

So, all in all, Pennsylvania is a very tough challenge for McCain-Palin. Barone is skeptical that the GOP can win the state, and I share that skepticism.

That being said, I will be interested in examining the data after the election. I suspect the emphasis on Pennsylvania recently is due to the fact Obama is weak somewhere - either in the northeast, the southwest, or both. If that turns out to be the case on Election Day, Pennsylvania might exhibit a sharper urban-rural divide than it has in any recent cycle. As Barone notes - that would make for a very interesting electoral coalition for Obama, one based upon the wealthy and the poor in big cities like Philadelphia.

-Jay Cost

Swing State Review: Virginia

Today we take our Swing State Review to Virginia. Historically, Virginia has not been much of a swing state, instead consistently favoring the Republican Party. The effect has been that the Democrats typically win Virginia when they have already secured 270 Electoral Votes.

The following chart makes this clear by examining the "Republican tilt" of Virginia against the three previous states in our review - Colorado, Pennsylvania, and Ohio.

Virginia Partisanship 1.jpg

Virginia's Republican tilt is apparent. In the last sixty years, only Truman and Johnson have carried the Old Dominion.

So why are the Democrats so bullish? Part of it is probably induced by their overall confidence. George W. Bush's job approval numbers are low, the public favors the generic Democrat over the generic Republican, and all of the "macro forces" of electoral politics seem to favor them. And remember, Republican tilt is a relative term. If the nation votes +10 Democrat, and Virginia votes +5 Democrat - it has a Republican tilt, but the Democrats still get its Electors!

Of course, Democratic confidence has been shaken with Obama's declining poll position. I think deriving one's confidence from summer polls is as silly as staking out the houses of potential vice-presidential candidates. For goodness sake, it's still August! That being said, the particulars of the Obama-McCain match up might disrupt the typical partisan electoral dynamics, producing a closer race than what some experts would otherwise predict.

So, the most fruitful question to ask is whether Obama can win the state in a close election.

Because this query is so similar to the one we asked about Colorado, it makes sense to approach it in a similar manner. Accordingly, let's divide Virginia into metropolitan regions - plus another "odds and ends" rural category. 1

Virginia Partisanship 2.jpg

There seems to be an angle for the Democrats here. From 1968 to 2000, metropolitan Washington favored the Republicans, but not in 2004. Some Democrats see an incipient realignment in the works here. Personally, I have such a hard time predicting one election, I don't know how anybody could predict the next ten! I'll just say that, realignment or not, there is certainly potential for Obama in metropolitan Washington, D.C.

That being said, Republicans have reasons to be cheery as well. All other areas represented in the graph - metro Virginia Beach, metro Richmond, and the rural areas - went for Bush in 2004. Additionally, the GOP has dominated the smaller metropolitan regions, as this graph and this graph make clear. The only exception to this trend is Charlottesville - which, like metro Washington, exhibited a Democratic tilt in 2004.

Comparing Virginia to Colorado, it is fair to say that, all else equal, Colorado is a more promising short-term opportunity for the Democratic Party. Assuming that metro Washington continues to favor the Democrats, the Republican Party is still left with a sizable base to fall back upon. You could have tripled Kerry's margin in metro Washington, and George W. Bush would still have won. That's how well he did in the rest of the state. Whether or not this persists in the medium- or long-term is outside the purview of this essay.2

In the short-term, that makes the state a relatively difficult capture for Obama (so long as we assume that his vice-presidential nominee does not give him a "favorite son" bump). This can be appreciated by examining how previous Democrats have done in statewide contests. The following picture reviews the most populous parts of the state.3

Virginia Democratic Margin of Victory 1.jpg

A close examination of the 2006 Senate contest indicates Obama's challenge in Virginia. Webb pulled an enormous margin out of metropolitan Washington, and yet he eked out the barest of victories - a margin of 9,300 votes out of more than 2 million cast. What tipped the race in Webb's favor? Virginia Beach. While John Kerry lost the area by about 6 points, Webb won it by 3 points.

This could be hard for Obama to mimic - for Webb's greatest strength in the region is one that favors McCain, not Obama: Virginia Beach has an extremely large concentration of military veterans. About 20% of the voting age public are veterans. Webb's military background probably made the difference in Virginia Beach, and thus the whole state. In 2004 John Kerry, himself a veteran, won 36% of the veteran vote in Virginia. Jim Webb won 42%. That difference was large enough to tip the election.

So, Webb's victory depended upon an angle Obama simply doesn't have. Does he have an alternative one? Maybe. One oppportunity manifested itself in the Democratic primary.

Virginia Obama Clinton.jpg

Notice that Obama did very well in metropolitan Washington, a sign that he might be able to improve on Kerry's margins there. He also did well in the south and east - in places like Richmond, Virginia Beach, and Danville. This was due in large measure to his overwhelming support from African Americans, as the following picture indicates.

Virginia African American.jpg

Note the close relationship between the two pictures. This is not a mirage. According to the exit polls, 42% of all Obama voters in Virginia were African American.

This might present a general election opportunity for Obama. Clearly, African Americans in Virginia are strongly behind him. If there is a surge in turnout among African Americans, such that they increase their share of the vote by 5%, and Obama does really well in metro Washington - he might be able to pull off the upset. Holding everything else constant from 2004, adjusting the racial demography of the electorate would cut Bush's margin of victory by about half, from 8 points to 4 points. Toss in a Webb-like performance in metropolitan Washington, and Obama carries the state.

Of course, it is one thing to write this - it is another thing entirely to do it. African Americans in Virginia vote in proportion to their overall population, which means that increasing their share of the vote 5% would be a monumental undertaking.

I'm not saying this is impossible. Frankly, I do not know. I just pulled that "5% increase" out of my you-know-what. If there was a sensible baseline I could use - believe me, I'd use it. But the fact is that there has simply never been a presidential candidate like Barack Obama, which in turn means that it is impossible to estimate what kind of effect his candidacy will have on the racial composition of the electorate.

There is another relevant subject to discuss, one we covered in our reviews of Ohio and Pennsylvania. Note Obama's poor margins in the southwestern part of the state. This area voted much like neighboring West Virginia and Kentucky, and nearby southern Ohio and western Pennsylvania. This is not an incredibly populous part of the state, but there are three important caveats to keep in mind. First, every previous non-voter who votes Democrat this cycle nets Obama one vote, but every typical Democrat who backs McCain nets the Republican two votes. Second, in the best-case scenario for the Obama campaign, the final result in Virginia will be close, something on the order of the 2006 Senate race. Every vote counts in that instance. Third, Obama's poor margin in the southwest indicates that he might have a more general problem among some downstate white voters. We might not be seeing it in the primary results in places like Roanoke, Lynchburg, or Danville because Obama did so well with African Americans. On the other hand, Obama did pretty well in Harrisonburg, which has a relatively small African American population - so it is hard to identify the extent of any potential problem with downstate whites. Ultimately, that's the limitation of countywide analysis - if there is intra-county variation (and there probably is), we can't see it.

I would note that McCain is still picking up a good portion of Clinton voters. My hunch is that these voters are not evenly distributed nationwide, that instead they are clustered in specific areas. I don't know how else to explain why, in a year like this, Arkansas, Kentucky, Tennessee, and West Virginia are out of Obama's grasp. Such voters might also be disproportionately concentrated in Virginia - which in turn could yield trouble for him on Election Day.

So, in conclusion, I would say that of the four states we have reviewed thus far, Pennsylvania is the most likely to go for Obama, Virginia the least, and Colorado and Ohio fall in between. It's certainly not impossible for Obama to flip Virginia, but it won't be easy.

Here are five things to watch on Election Day.

(1) Washington. Incipient realignment or not, Obama has a chance to improve on Kerry's numbers in metro Washington. Can he? If so, by how much? He'll need a big haul in northern Virginia to overcome McCain's advantages in other parts of the state.

(2) Virginia Beach. Home to a large number of veterans and African Americans - Virginia Beach plays to Obama and McCain's strengths. Keep an eye on Hampton, which has particularly large veteran and African American populations. In 2004, it went for Kerry by 15 points. If that number changes substantially in either direction, we'll get a sense of what kind of night we're in for.

(3) Richmond. Both Webb and Kaine did better than Kerry in metro Richmond - Kaine, who hails from Richmond, substantially so. If Obama can hold his own among suburban/exurban whites - enhanced African American turnout here might boost his margins and put the state in striking range.

(4) Downstate whites (outside Richmond and Virginia Beach). One reason Jimmy Carter kept Virginia so close in 1976 was his support outside metropolitan Washington. In an inverse of the 2004 results, Ford won metro DC while Carter was very strong in the rural areas and smaller towns like Roanoke, Blacksburg, and Bristol. These days, such voters generally vote Republican - but some still vote Democrat. Will they this cycle?

(5) African Americans. The Obama campaign has promised game-changing African American turnout. In the last two contests, African Americans constituted about 20% of Virginia's presidential vote, proportional to their share of the state's population. If, when the exit polls roll out on Election Night, we find something closer to 25% - expect the state to be very close.


[1] Unfortunately, our rural category for Virginia is even more diverse than the category we etched out for Colorado. These counties include places in the western, mountainous part of the state, as well as coastal counties that are not categorized as being part of Richmond or Virginia Beach. One implication of this is that there is a wide divergence in the racial composition of this "region." On average, African Americans make up about 19% of all people in non-metropolitan counties - but the standard deviation is an eye-popping 17%. Such an odds and ends category is as necessary as it is unfortunate. Breaking this category into subcategories would mean that the number of units to analyze in our review would simply be unmanageable.

[2] One's bet on the future depends upon two items. First, how one chooses to extend those Republican tilt graphs into the future. As I noted, my preference is to wait until I have actual data before I draw them forward. Second, changes in population. Americans' mating habits are easier to predict than their voting habits - so we can get some purchase on this. The following chart tracks the four most populous regions' share of the presidential vote over the last fifty-odd years.

Virginia Share of Statewide Presidential Vote 1.jpg

Metro Washington clearly has been on the rise, a growth that seems to have come largely at the "expense" of the rural areas. One of the advantages the Republicans have in Virginia that they do not possess in Colorado is that in Virginia the principal city, while recently leaning to the Democrats, makes up a smaller proportion of the statewide vote. Denver is leaning to the Democrats like metro Washington, but metro Denver matters more to the electoral politics of Colorado than Washington does to Virginia. That being said, Washington's trend line has a positive slope. So, in the future, we should expect Washington to become more and more important in the partisan results of the state. For those of you who are insanely interested in fine-grained data, as I am, you can check out the population trends in the rest of the state here and here.

[3] See how Kerry, Webb, and Kaine performed in the rest of the state here and here.

-Jay Cost

Swing State Review: Colorado

Today we continue our Swing State Review with an examination of Colorado, whose recent electoral behavior has been more dynamic than Pennsylvania or Ohio.

The following chart makes this clear by examining the "Republican tilt" of Colorado compared to Pennsylvania and Ohio. Republican tilt is measured by the margin of victory (or defeat) for the Republican presidential candidate in the state minus his national margin. So, if a state supports the Republican more than the whole country, it has a Republican tilt.

Colorado Partisanship.jpg

We noted in our Pennsylvania review that the Keystone State has exhibited a slight, pro-Democratic tilt over the years, which can be seen in this chart. For Ohio, we noted that it generally votes slightly more Republican than the nation. This too is clear.

Meanwhile, Colorado's behavior has been more variable. Of course, the variability is not entirely "random." Though the line bounces around quite a bit, it still has a downward slope, which indicates that it has moved from the right to the middle. When analysts suggest that Colorado is an emerging purple state, this is what they are on about.1

Thus, it's reasonable to assume that if the national vote is close this year, Colorado will be close, too. Let's take a closer look at the state to see if we can get an idea of what to expect.

Let's begin by investigating the partisanship of different parts of the state. Once again, we'll use Republican tilt as our metric, but instead of reviewing the state as a whole, let's break it into segments: the seven largest metropolitan areas, plus a "Small Town/Rural" segment that captures the rest of the state. 2

We'll examine the four most populous areas first.

Presidential Partisanship 2.jpg

Interestingly, we see Colorado becoming polarized over time. In 1952, these areas differed only slightly in their partisan preferences. They all sat on the Republican side of the aisle, and the differences among them were less than 20 points. Flash forward fifty-two years and we see significant polarization. Colorado Springs has become more Republican, Boulder strongly Democratic, Denver slightly Democratic,3 and the rural areas essentially unchanged.4

This indicates why Colorado is a swing state. Both parties have solid, sizable bases of support. Additionally, the most powerful part of the state, metropolitan Denver, sits in the middle. While it has a slight Democratic lean, both parties can play in metro Denver.

Let's take an even closer look by examining particular elections. We must be careful in how we do this. Because Colorado has only recently begun to behave as a swing state, we should favor recent statewide elections in our analysis. Building a set of expectations from older elections might bias our estimates toward the past alignment. So, let's stick to recent events.

Fortunately for us, Colorado offered two great data points in 2006. For Secretary of State, Republican Mike Coffman won a 1.6-point win over Democrat Ken Gordon. For State Treasurer, Democrat Cary Kennedy won a 2.6-point victory over Republican Mark Hillman.

Let's also include John Kerry's 4.7-point defeat as a baseline. The following chart reviews how the three Democratic candidates - Kerry in 2004, Gordon in 2006, and Kennedy in 2006 - performed statewide, as well as in the four most populous parts of the state.

Democratic Margin of Victory 1.jpg

What made the difference between victory for Kennedy and defeat for Kerry? Obviously, Kennedy improved relative to Kerry in all four quadrants, but there are two noteworthy features to her win.

First, she improved relative to Kerry in the rural areas. She didn't win them. In fact, she lost them by 6 points. But Kerry lost them by 15 points.

Second, Kennedy won metropolitan Denver by a comfortable margin - around 9 points. Kerry and Gordon both won metro Denver, but their margins (around 3 points each) were smaller. For Gordon, this was probably enough to keep him from winning.5

This gives us a pretty specific sense of what each candidate must do in November. However, the story is slightly more complicated than this. Until now we have operated under the assumption that Colorado's vote will be close if the nation's vote is close. While this is generally reasonable, there are two caveats to add.

First, while Denver has split down the middle in the last five presidential elections, 2004 saw it vote more Democratic than any time since 1952. Is this simply an "outlying" point around a central tendency of moderation, or is it a signal that the city is moving to the Democratic Party? It is too soon to say. If Denver continues to vote down the middle, we should expect Colorado to stay in the middle. If, on the other hand, it votes more Democratic, the whole state will shift to the left.6

Second, Colorado's population has been growing rapidly - by more than 30% in the last eighteen years. Much of this has been due to an increase in the Hispanic population. The following graph reviews changes in the Hispanic population in the most populous portions of the state.7

Hispanic Population As Percent of Whole.jpg

The electoral implications of this immigration have not been fully felt. In 2000, Hispanics accounted for about 17% of the state's total population, but 14% of the presidential electorate. In 2004, they were estimated to account for nearly 19% of the state's population, but just 9% of the presidential vote. This is about the share they made up in 1998 and 1996.

If Hispanics in Colorado begin to vote proportional to their numbers, Democrats should enjoy at least a short-term advantage. Despite George W. Bush's best attempts to attract Hispanics, John Kerry still won them in Colorado by better than 2-to-1.

Practically speaking, this means that our assumption about Colorado voting roughly as the country votes, while reasonable, is not a certainty. Simply put, Colorado is a state in flux. It's moved from reliably Republican to a swing state in recent years. It remains to be seen whether this is just a stop on the way to some final left-of-center destination.

We now have a good idea of the electoral landscape in Colorado. Let's tie all of these considerations into a "bottom line" conclusion. Here are four things to watch on Election Day.

(1) Metropolitan Denver. Denver is the critical battlefield because it's so large and sits in the middle. It will probably tip the state one way or the other on Election Day. Expect Obama to carry metro Denver. The question is: by what margin? Kerry won it by about 3 points. Obama will need to win it by close to 7-9.

(2) Hispanics. Will Hispanics comprise around 15% of the vote, or will they once again be counted lower than 10%? That could make a huge differences for Obama, assuming they go strongly for him. Will they? That's the other big question. From recent poll numbers, it looks as though Obama is on track to match Kerry's performance, but it is still early.

(3) Rural and small town areas. These parts of Colorado are often overlooked because no one place is particularly large. However, collectively they add up to the second largest category we have defined, so they should not be taken for granted. Can Obama improve over Kerry? Rural voters back east were not inclined to him during the primaries, but he did reasonably well with them in Oregon and New Mexico. If he can hold his own in rural Colorado, keeping his margin of defeat in the high single-digits, he'll be well positioned.

(4) Boulder and Colorado Springs. These towns are symbols of Colorado's political polarization over the last fifty years. Boulder has trended leftward, Colorado Springs rightward. This year, the parties have nominated candidates who supposedly possess cross-partisan appeal. Do they? I doubt it, but if they do, we might see Boulder and Colorado Springs break from their recent patterns. Electorally speaking, conflicting movement would cancel each other out. What would be consequential is if both places moved in the same direction. Combined, Boulder and Colorado Springs typically account for 20% of the state's vote.

We'll cover Virginia next.


[1] The current political alignment of the Centennial State, while new in many respects, nevertheless has roots in days long gone. The following picture makes that point.

Colorado Montage.jpg

These elections are separated by some 64 years, and yet the countywide voting patterns are extremely similar. This indicates that Colorado has exhibited both change and continuity over the years. The change has come from the fact that FDR did very well relative to prior or subsequent Democrats. A tight race in Colorado was not something any Democrat could typically count on then, or for many years to come. Today, they can. That's the change. The continuity is the similarity between the countywide vote patterns. Close races then and now could be expected to produce roughly similar countywide results.

[2] "Small town/rural" is an odds and ends category, and even though (as we'll see) the "region" as a whole votes Republican - there is significant variation from rural county to rural county. Southern Colorado has a sizeable Hispanic population and can generally be expected to support Democrats. Eastern Colorado, where the best farmland in the state is located, is solidly Republican.

[3] The Census Bureau identifies ten counties in metropolitan Denver, including recently created Broomfield County. In 2004, Bush won Arapahoe, Broomfield, Elbert, Jefferson, Park, and Douglas counties. Kerry won Adams, Clear Creek, Gilpin, and Denver counties. The only county to switch from 2000 to 2004 (excluding Broomfield, which did not exist at the time) was tiny Clear Creek County (population 8,956). It went for Bush in 2000 and Kerry in 2004. Overall, metropolitan Denver was essentially unchanged from 2000 to 2004 in terms of its absolute vote. It gave Bush 46.8% in 2000 and 47.8% in 2004. Kerry improved over Gore in part because Ralph Nader's role was greatly diminished.

[4] The less populous cities were more diverse to begin with, and they remain so today, as this chart makes clear. Note in particular the movement of Fort Collins, which has gone from being solidly Republican to a swing area. For the sake of brevity, we'll place the data on the less populous cities in the endnotes. However, this should not be taken to imply that they are not important. In fact, metro Fort Collins, metro Greeley, and metro Boulder are all about the same size, ranging from about 250,000 to 300,000 people. Pueblo and Grand Junction are less populous, with about 150,00 people each.

[5] An examination of the other parts of the state shows that Kennedy also improved on Gordon and Kerry in Greeley, Fort Collins, Pueblo, and Grand Junction. See here for the relevant chart. See here for the 2004 presidential map, here for the 2006 Secretary of State map, and here for the 2006 State Treasurer map.

[6] As everybody knows, Denver has grown by leaps and bounds in recent years. What is less remarked upon, however, is that the rest of the state has kept up.

This has had an important political consequence. Though Denver has grown, its voting power relative to the rest of the state has remained roughly constant. The following chart illustrates this point.

Share of Statewide Presidential Vote 1.jpg

When we combine this with the fact that metropolitan Denver sits roughly in the middle of the state and the country - we find another reason why Colorado is a swing state. Denver is relatively moderate, and it is not so large that it has "hegemonic" power over the rest of the state. [As this chart makes clear, the net change in voting power in the rest of the state has been roughly zero.]

[7] See here for changes in the Hispanic composition of the rest of the state.

-Jay Cost

Swing State Review: Pennsylvania

Today we continue our Swing State Review, examining Pennsylvania. It is a common conception that the Keystone State has been "trending blue" in recent years. Commentators often note that Pennsylvania voted for Ronald Reagan twice and for George Bush in 1988, but has since voted Democratic.

While this is significant, it is not by itself a signal that Pennsylvania has been "trending blue." It is more appropriate to call Pennsylvania consistently and marginally blue. In the last fourteen presidential elections, Pennsylvania has voted more Democratic than the rest of the nation all fourteen times. On average, its Democratic tilt has been about 4 points. For instance, in 1952, Dwight Eisenhower beat Adlai Stevenson by 10.5 points. He won Pennsylvania by 5.8 points, for a Democratic tilt of 4.7 points. In 2004, George Bush beat John Kerry nationwide by 2.4 points. He lost Pennsylvania by 2.5 points, for a Democratic tilt of 4.9 points.

Interestingly, this statewide consistency masks significant intrastate change. It is frequently noted that metro Philadelphia has moved to the Democrats. This is quite true. However, this has been countered by movement to the GOP in central and western Pennsylvania. Taken together, these various shifts have altered the political contours of the state while keeping the aggregate vote results roughly consistent.

The following picture indicates this point. It illustrates the shift in the presidential vote from the 1968 presidential election to the 2004 election. These are two elections where the Democrats won modest victories: 3.6% in 1968 and 2.5% in 2004. Thus, they offer a good "apples to apples" comparison of how the state has shifted over time.

Pennsylvania 1.jpg

Clearly, the east has moved to the Democrats. All of the counties of metro Philadelphia have generally trended blue (Berks County being the sole exception). This is especially important as the surburban counties have grown in population. We also see some movement to the Democrats in the far northeastern counties, which have increasingly become part of greater New York.

Meanwhile, the central and western parts of the state have clearly been moving in the Republican Party's direction. All of the exterior counties of metropolitan Pittsburgh - Armstrong, Butler, Fayette, Washington, and Westmoreland - have shifted noticeably toward the GOP. The Republicans have also made gains in fast-growing York County. Lancaster County has not shifted, but it was very Republican to begin with. So, its growth has given the GOP more votes.

We can see the same patterns from two different angles. First, the 1976 election produced statewide results similar to the 2004 election. Jimmy Carter won a slight, 2.7-point victory. So, we can generate another "apples to apples" map comparing '76 to '04. By and large, we once again find the same pattern: movement to the Democrats in the east and to the Republicans in the center and west.

Second, the following picture examines the average partisan "swing" of each county for three groups of presidential elections: 1968-1980, 1984-1996, and 2000-2004.* This gives us a sense of each county's voting pattern relative to the state as a whole.

Pennsylvania 3.jpg

We are interested in how counties change color over time. In the east, red counties have turned purple or blue. This trend is countered in the west, where solid blue counties have turned purple or, in a few instances, red. The sparsely-populated central third of the state was Republican to begin with, and it has generally become more so.

Taken together, these pictures illustrate that while Philadelphia and much of the east have been moving toward the Democrats, the south and west have been moving to the Republicans.

As indicated above, the net effect of these shifts has so far been roughly zero. Pennsylvania still generally tilts Democrat by 4 or so points. Whether this pattern persists in the future depends on a number of factors outside our immediate concern. The important point for us now is that the Democrats still possess this advantage. So, Obama is favored because he is the Democrat in what many expect to be a close election.

Whereas in Ohio, Obama was on "offense" and McCain on "defense," matters are reversed in Pennsylvania. Accordingly, let us examine what McCain must do to flip Pennsylvania from blue to red. There are four areas he should focus on:

(1) Metropolitan Philadelphia. George W. Bush's numbers in metro Philly were not great in 2004. Statewide, Bush's vote share ticked up 2 points between 2000 and 2004, from 46.4 to 48.4. In the suburban counties surrounding Philadelphia, it barely inched up, from 46.7% to 47.0%. If McCain can improve on Bush's 2004 share in the suburban counties, he will be better positioned to win the state. He would not have to do much better - a shift of just 2 points in the suburban counties would close more than a third of Bush's statewide deficit.

(2) The Lehigh Valley. This is in Lehigh, Northampton, and Carbon counties. Formerly a center of American manufacturing, the economy has recently been invigorated by the tech and service industries. With this revitalization has come an increase in population. Bush was basically stagnant in this region between 2000 and 2004. He lost all three counties by slim margins in 2000. In 2004, he won Carbon County barely, and ticked up a bit in the more populous Lehigh and Northampton Counties, still losing both. This is an area McCain cannot afford to ignore: Republicans who have won close statewide races in recent years tend to win the Lehigh Valley.

(3) York and Lancaster Counties. York and Lancaster are in the southeast - in Pennsylvania Dutch Country. Situated just east of the Appalachian Mountains, these counties have some of the most productive farmland in the state, though this does not account for their political importance. They sit between Baltimore, Harrisburg, Philadelphia, and Washington - and their populations have been growing as people willing to commute look for new places to settle. Like most "exurbs" nationwide, Bush did exceedingly well here, winning more than 3 out of 5 voters in both counties in 2004. McCain will have to match that. A strong performance here is already built into a Republican loss. So, he cannot afford to underperform here.

(4) The West. Culturally conservative western Pennsylvania has been trending Republican in recent years. For a long while, their ties to American industry induced them to vote Democratic, even as the Party of Jackson moved to the left on cultural issues. With the decline of industry, the area's allegiance to the Democratic Party has weakened. Meanwhile, the region is still largely pro-life and pro-gun, which has given the GOP an opening. In fact, George W. Bush did better in the five outer counties of metropolitan Pittsburgh than any Republican since 1972.

To win Pennsylvania, McCain will have to find a way to account for Bush's weaknesses while mimicking his strengths. While it remains to be seen whether he can do this, we might get a sense of what to expect by examining the Democratic primary results. This could give us an idea of Obama's position in the state, and therefore what McCain needs to watch for.

Pennsylvania 4.jpg

This picture presents a mixed bag for the McCain campaign. First, it indicates that McCain can be reasonably optimistic about the western part of the state. Obama did poorly here, especially in the southwest. Given that these are areas where the political dynamic has increasingly favored the Republicans, anyway - Obama could have some rough sledding here.

Obama ran better in the Lehigh Valley than he did in the west, but that is not saying much. Indeed, he failed to break 40% in any of the counties in the Lehigh Valley. Plus, in lightly populated Carbon County, he barely won one out of five voters. There might be an opportunity for McCain here.

On the other hand, Obama ran very well in the Philadelphia suburbs. He carried Chester and Delaware counties, and kept it close in Montgomery. So, he might be well-positioned to prevent McCain from improving on Bush here.

Furthermore, Obama won Lancaster County. This is similar to his win in Delaware County, Ohio. Both Lancaster and Delaware are fast-growing exurban counties. Is it possible that Obama might be able to blunt the GOP's nationwide advantage in the exurbs? Maybe, though I'd note that Obama was roundly defeated in the exurban counties of Pittsburgh - Butler and Westmoreland.

The primary map points to another area we should watch as the election returns come in - the old coal country of Scranton and Wilkes-Barre. Lackawanna County, home of Scranton, is typically a safe Democratic county in presidential elections. However, it is not a lock. Michael Dukakis eked out at 3-point victory in the county in 1988. Luzerne County, home of Wilkes-Barre, is more in the middle, though it slightly favors Democrats. Obama did quite poorly here, winning only about 25% of the vote in either county. Let's add this part of the state as a fifth place for McCain to focus on. If voters in these areas are disinclined to Obama, McCain might pick up some support here that a "generic Republican" would not.

Overall, it is far too early to make any predictions about Pennsylvania. Nevertheless, given the last half-century of elections, it is not unreasonable to expect the state to vote more Democratic than the country at large. If this holds true this cycle, Obama should have the advantage in the Keystone State. That being said, McCain certainly has a shot here, and if this turns out to be an election where the old rules of electoral politics are suspended, we might see a surprise on Election Day.


[*] "Swing" is defined here as it was in our review of Ohio. It measures a county's partisan behavior relative to the state. If, for instance, the state voted +2 Republican, and Philadelphia County voted +20 Democrat, we would say that Philadelphia County had a Democratic swing of 22 points. Each county gets a score like this for every presidential election, and those scores are averaged to produce the picture above.

-Jay Cost

Swing State Review: Ohio, Part II

In the previous installment, we analyzed partisan voting patterns in Ohio. Today's essay will apply this analysis to the upcoming election.

Let's encapsulate the last essay's observations in a simple visual presentation. The following picture measures the partisan "swing" of each Ohio county. 1

Ohio Countywide Partisanship.jpg

The lightest red, lightest blue, and purple counties generally vote as the state does. They are bellwethers. A few observations about them are noteworthy.

First, Ohio's mid-sized cities are in the political center. The purple counties in the northwest are near Toledo (in Lucas County). Dayton is in Montgomery County, another purple county. Ditto for Springfield, in Clark County, and Canton, in Stark County.

Second, there are lots of swing counties in the south. We noted last week that Democrats tend to win when they form an inverted "C" on the map - winning the counties along Lake Erie, the Ohio-Pennsylvania border, and the Ohio River. However, such a pattern is not a guarantee. While the eastern border is generally reliable territory for Democrats, the southern border is not.

Let's apply these insights to the pragmatic task of winning a close election in the Buckeye State.

In a close election, a generic Democrat can generally count on many counties along Lake Erie and the eastern border, but he or she must supplement this base. There are several ways to do that:

(1) Win the counties surrounding Cleveland (Medina, Lake, and Ashtabula Counties).

(2) Win the south/southeastern counties in Ohio's sixth and eighteenth congressional districts.

(3) Win Columbus (Franklin County).

(4) Win mid-sized cities like Canton, Dayton, and metro Toledo.

(5) Minimize losses in the Cincinnati and Columbus suburbs/exurbs.

Of course, complete success would mean a blowout, not a close election. To win a close one, the Democrat need not do all this. Instead, he or she can win by mixing-and-matching several of these tasks. Indeed, if you look at the close elections we reviewed in the last essay, you'll see that Democratic victories all look a little different.

Meanwhile, the generic Republican can count on the western and central parts as his base. He or she must then stop the generic Democrat from doing several of these tasks. Much like the Democrat, there are many possible combinations for a GOP victory. 2

What might a McCain or Obama victory look like? By and large, we'll have to wait and see. Nevertheless, we can make a few reasonable points now, thanks to the primary results. However, we have to be careful with the data. After all, there was no McCain-Obama ballot question. So, the results cannot tell us how well one will do against the other. Nevertheless, they might tell us about a candidate's strength in one part of the state relative to his strength in another. 3

Let's begin with Obama-Clinton.

Clinton v. Obama.jpg

These results indicate potential problems and opportunities for Obama. We'll begin with the potential problems, of which there are two.

Start with the counties on the eastern border. Counties like Trumbull, Mahoning, Belmont and Monroe are solidly Democratic. So, we should not expect them to vote McCain. They'll go for Obama. The question is, by how much? They swung heavily for Bill Clinton, but less so for Gore and Kerry. Clearly, Obama was not the first choice of Democrats in these counties. Is this an indication that the swing voters here - the friends, neighbors, and relatives of the primary voters who voted so overwhelmingly for Clinton back in March - will be inclined to McCain? Possibly, but not necessarily.

Next, notice the Appalachian counties in the south and southeast. Unlike Trumbull and Mahoning, many of these are swing counties. Clinton defeated Obama by such large margins here that I would give McCain the advantage. Indeed, I think Obama would, too. After all, he chose to advertise in North Dakota and Montana, but not Kentucky and West Virginia, which these counties are a lot like.

This is the bad news for Obama. However, it is far from devastating. As mentioned above, there are many ways a Democrat can win Ohio. The primary map indicates the possibility of other such routes for Obama.

First, the Appalachian counties are not very populous, and many are losing people. Obama can suffer losses here, so long as they are not extreme. For instance, suppose that turnout in 2008 is identical to 2004, that Obama performs as well in those dark green counties as Kerry, and that in the rest of the state he performs as well as Bill Clinton did in 1996. Under those conditions, Obama would win. It would be closer than Clinton's victory - more like 2% than 6% - but a win is a win.

Second, many counties where Obama performed well have plenty of (as Paul Begala famously put it) "egg-heads and African Americans." Ohio University is in Athens County. Ohio State is in Franklin County. Greene County, in the southeast, has several colleges and universities, including Wright State and the famously liberal (and currently closed) Antioch College. Plus, there are sizeable African American populations in Cuyahoga, Franklin, Hamilton, Lucas, and Montgomery counties. If the Obama campaign amplifies turnout, this is where we will see its effects.

Third, Obama did relatively well in Butler County, north of Cincinnati, and he won Delaware County, north of Columbus. Both are large exurban counties, which have been a critical part of the GOP's success. The counties around Cincinnati and Columbus gave Bush a net of 175,000 votes in 2004 (and remember that Bush won the whole state by about 120,000 votes).

Obama claims he can win Republicans: can he win exurban Republicans? According to this map, possibly. He would not need to win these counties outright. It would be enough for him to need to minimize the size of McCain's victories. The gas price issue may be the crucial factor. When gas was $2 or less, exurbans swapped a longer commute for affordable homes and quiet, open spaces. Now that it is $4 or more, they are getting squeezed.

What about McCain? Unfortunately for our purposes, McCain was the all-but presumptive nominee by the time of the Ohio primary. This probably skewed the results in his favor, limiting our ability to use the data. Nevertheless, we might be able to make a few, extremely narrow conclusions.

McCain v. Huckabee.jpg

Note that McCain seemed to be strong in typical Democratic areas - the counties along Lake Erie, the PA-OH border, and the Ohio River. He also performed very well in metropolitan Columbus, winning Franklin County and its surrounding areas by over-sized margins. A similar result occurred in Cincinnati - he won Hamilton and Clermont counties by large margins. His showing in Butler County was strong, too.

This might be a sign that McCain is well-suited to counter Obama along the eastern border, as well as the suburbs and exurbs. However, that is far from a slam-dunk. After all, McCain was running against Huckabee, a candidate whose appeal never broadened beyond evangelicals. So, all we can say for sure is that McCain should be better at this than Huckabee, which is not saying much.

So, to conclude:

What To Watch in the Buckeye State

(1) If the national vote is close, expect Ohio to be close. It's a bellwether.

(2) Watch the mid-sized cities. They tend to vote with the winner. If Canton, Dayton, Springfield, and metro Toldeo go for the same candidate - he'll have the edge.

(3) Watch Franklin County (Columbus). If recent history is any guide, it will go for Obama. The question is by how much.

(4) Watch the exurbs. Obama promises to appeal to Republicans. These Republicans here are probably his best bet. McCain should still win the exurban counties of Cincinnati and Columbus, but Obama will be in good shape if he can turn them pink.

(5) Watch the eastern border. There are lots of "working class whites" here, the ones Obama had trouble with in the primaries. But it's not the strong Democrats he needs to worry about. It's the swing voters. If they vote McCain, these counties will be a lighter shade of blue than what Obama needs.

(6) Watch the south. It voted heavily for Clinton in the primary, and there are good reasons to expect it to support McCain. The bigger the margins, the better for the GOP.

(7) Watch Hamilton County (Cincinnati). Obama promises game-changing GOTV efforts. If he delivers, the first sign of success should be here. Traditionally, Hamilton County votes Republican, but just barely (and by steadily decreasing margins over the years). If Obama amplifies African American turnout enough to flip it, that's a sign that his plan's on track.

We'll review Pennsylvania next.


[1] This chart was created by examining all close elections (where the two major parties are separated by 10 points or less) over the last ten years to see how each county voted relative to the state. If, for instance, the state voted +2 Republican, and Cuyahoga County voted +20 Democrat, we would say that Cuyahoga County had a Democratic swing of 22 points. Each county gets a score like this for every close election, and those scores are averaged to produce the picture above.

[2] So, we see an intersection between the structures of politics and the personalities in politics. Each victorious candidate generates a winning coalition from the party base, plus his or her own unique personality, background, and message.

[3] For instance, suppose that Obama defeated Clinton by 6 points in one county, but lost by 40 points in another county. It is reasonable to infer from this that Obama is better positioned in the first county than the second. We have to be careful with this kind of inferential work because the data is not fine-grained enough to make some types of judgments. For instance, if Obama lost the second county by 6 points, rather than 40, we would be unwise to make any judgment between the +6 and -6 counties. The bottom line is that modest judgments can be made if we are careful with the data we have.

-Jay Cost

Swing State Review: Ohio, Part 1

Today I initiate a new series of essays that will discuss the electoral landscape of the swing states.* Each essay will consider a different state in some detail, outlining how partisanship in the state typically manifests itself, and what this manifestation might mean for November.

We'll begin with Ohio, which typically has been a crucial component of Republican voting coalitions ever since the Civil War. That being said, Ohio has not been a state that the Republican Party could necessarily count on. Since 1960, it has voted with the winning party in every cycle. That's 11 in a row. Plus, Ohio's popular vote has tracked closely with the national popular vote. Over those 11 presidential elections, Ohio's vote has deviated from the national vote by a scant 2%.

Ohio behaves like this because of its diversity. It has large populations firmly rooted in one party or the other. A great example of this can actually be seen in the famous election of 1896. In Ohio, Democrat William Jennings Bryan won 47% of the vote against William McKinley, of Canton. At first blush, that seems strange. Why didn't McKinley dominate his home state? The answer had to do with the state's diversity. McKinley was the candidate of prosperous American industry, Bryan of struggling farmers. Because Ohio was a state with sizeable farming and industrial elements, it split its vote between the two.

While the nature of the Republican and Democratic bases have changed - the fact remains that both parties have large enough bases that competitive elections are inevitable. The relatively narrow slice of voters in the center of the state is thus endowed with the power to swing many elections one way or the other.

Because the zeitgeist currently belongs to the Democrats, it is fitting to begin our investigation with the last presidential election Democrats won - the anticlimactic contest of 1996. Bill Clinton won Ohio by 6.4 points. How did he manage it?

1996 Election for President.jpg

Clinton's victory is visually striking. It is a kind of inverted "C." He dominated the counties along Lake Erie and the Ohio-Pennsylvania border. Prior to the election of 1932, these places were solidly Republican. With FDR in 1932, they switched sides and generally have not switched back.

Furthermore, note the cluster of counties that Clinton carried in the southern part of the state. These are the poorest counties in all of Ohio. There is little farming or industry here. They are actually part of Appalachia, and many of them are classified as economically distressed. Typically in these counties, you'll find the biggest employers to be various governmental agencies (local school districts, the state, etc) and Wal-Mart.

Additionally, observe the three counties in the middle of the state that Clinton won. These are Montgomery County (Dayton), Clark County (Springfield), and fast-growing Franklin County (Columbus). Each of these counties have sizeable African American populations, which makes them more electorally balanced than their neighbors. Also, note the suburban counties that surround Franklin, Hamilton (Cincinnati), and Cuyahoga (Cleveland). By and large, Clinton did not win them, but he kept Dole's margins relatively small. Neither Al Gore nor John Kerry could do that to George W. Bush, which is a big reason both of them lost.

Bob Dole, for his part, generally won central and western Ohio. These are areas with vast expanses of farmland. However, you're much more likely to find people in these parts making a living through industrial work. As we'll see in a little bit - the central and western parts of Ohio constitute the Republican base in the state.

The pattern of Clinton's victory is neither unique nor accidental. Instead, this inverted "C" forms the geographical core of the Democratic coalition. When Democrats win, their victories tend to take this shape. Consider, for instance, the 2006 election for Attorney General, which the Democratic candidate won by about the same amount as Clinton did.

2006 Election for Attorney General.jpg

There are some modest differences in how the vote is dispersed, but by and large we see the same essential pattern. That is noteworthy. Even though the election for Attorney General is not nearly as salient as the election for president, the same basic dispersion of the vote remains.

Sooner or later, a political party in power will mess things up really well, giving the opposition an opportunity for a big win. Republican governor Bob "No Contest" Taft had succeeded in doing that by 2006 - which meant that the Democrats had an opportunity to win some offices by large margins. Their victory in the Attorney General race was relatively narrow, but Sherrod Brown's victory over Mike DeWine in the Senate contest was broader, and thus illustrative of what the state looks like when Democrats win a bigger majority.

2006 Election for Senate.jpg

Brown's 12-point victory did not occur because the state voted homogeneously. That inverted "C" is still the essential pattern. Instead, the differences between Brown's win and the other Democratic victories are two-fold. First, the inverted "C" is more pronounced in Brown's victory. The counties in the east and south that vote Republican in other elections flipped to the Democrats; the counties that vote Democratic did so more intensely. Second, the counties outside the "C," while still red, are generally a much lighter shade. DeWine still carried the western and central counties, but his margins were much smaller. Importantly for the Democrats, DeWine was relatively weak in the suburban counties that surround Columbus and Cincinnati, and he was defeated outright in the suburban counties around Cleveland.

These days, gleeful Democrats and gloomy Republicans forget that partisan triumphs are part of a larger partisan cycle. The competitive character of elections promotes the cycle: defeat inspires innovation, reinvigoration, and eventually restoration. If we take a broad enough time frame, we can appreciate this. In fact, the Democratic triumph of 2006 ultimately depended upon the Democratic defeat of 2002, which serves as a good transition to examine how Ohioans behave when they are feeling more like elephants and less like donkeys. The following picture examines Taft's landslide reelection for governor.

2002 Election for Governor.jpg

This is an inverse of the 2006 Senate race. What we see is that the red counties became more red, and the counties of the inverted "C" became light-blue or even pink.

What about a more modest Republican victory? George W. Bush's 2004 win in Ohio is instructive. It is illustrated below.

2004 Election for President.jpg

As mentioned above, Bush failed to win Franklin County. He also lost it in 2000. This is noteworthy. Prior to Bush, the last victorious Republican to win Ohio without winning Franklin County was Benjamin Harrison, who bested Grover Cleveland in the state (but nevertheless lost reelection) in 1892. A lot has changed since then - but it is interesting that, despite improving statewide relative to Dole statewide, he was unable to improve on Dole's performance here. Moving forward, weakness in Franklin County might put more pressure on Republican candidates to win other parts of the state by larger margins. Of course, this might be mitigated by continued growth in "exurban" counties, where Republicans had been doing quite well (prior to 2006, of course). Of the 10 fastest-growing counties in Ohio, Bush won all 10, with an average of 65% of the vote.

Beyond Franklin, Bush generally was successful in Ohio. Several factors contributed to his victory. First, he consolidated the Republican hold over the central and western portions of the state - most counties there gave him margins of 20 points and higher.

Second, he kept Kerry's margins pretty minimal in the northeast. Bush was unable to make much of a dent in Cuyahoga County (Cleveland) or Mahoning County (Youngstown), but everywhere else in the northeast moved noticeably in his direction.

Third, he generally improved upon Dole's performance in the suburban and exurban counties that surround Cincinnati, Columbus, and Cleveland. We indicated this earlier by noting that Bush carried the fastest growing counties in the state. What we see in a comparison of the 1996 and 2004 maps is that the counties surrounding Ohio's three largest cities all became less blue and more red.

Fourth, relative to the results in 1996, Bush actually flipped most counties in the southeast. Remember that these counties are the economically distressed parts of the state. If Clinton's appeal to these voters was based on economics, Bush's was based on culture.

The countywide results we see in Bush's 2004 victory are not unique. Typically, that is exactly what we see when Republicans win by a modest margin. You can confirm that by checking out Taft's first election in 1998, Bush's first election in 2000, and the ever-so-narrow victory of Republican Mary Taylor for Auditor in 2006. Between these three and Bush's reelection, you'll note some variation between the counties. But by and large, Republican victories look generally similar.

All in all, we have a good sense of what Democratic and Republican victories look like. The following picture specifies this idea further by examining how the map changes as the statewide votes change.


This demonstrates the "flow" of vote patterns in Ohio. As one party or another begins to run away with the election, the picture changes in a systematic fashion. These results date back to 1996, but a trip further back in time shows the same pattern.

Old and New.jpg

These days, everybody talks about this place or that place "emerging blue." A few years back, the talk was about "emerging red." That's all well and good, and some places are indeed emerging - but this picture tells us we need to be careful with all that talk. The fact remains that vote choices depend largely upon partisanship - which, even though it can and does change at the margins, remains a fundamentally stable political characteristic, typically passed on from parent to child. This is how we can explain elections separated by 40 years or more in basically the same terms. There is a great deal of stability to American electoral behavior. We need to remember that.

This concludes the general overview of voting in Ohio. In the next installment, we'll use this analysis as a foundation to understand exactly what McCain and Obama need to do to win the Buckeye State.

[*]I'd like to thank Sean Oxendine for the helpful advice he gave in the start-up phase of this project, as well as for letting me use his maps prior to my figuring out how to make my own.

-Jay Cost