Election Review, Part 1

Election Review, Part 1

By Jay Cost and Sean Trende - January 7, 2009

Today's essay is the first in a series analyzing the results of the 2008 presidential election. Our goal here is not to describe every minute detail, but rather to highlight significant trends that emerged on Election Day.

To do that, we shall use three types of data: countywide vote results going back to 1988; exit poll data going back to 1996; and a range of socioeconomic information provided by the Census Bureau. Occasionally, we'll use this data to examine individual states, but the focus will be on divisions of the country, as defined by the Census Bureau:

Census Regions.jpg

As we can see, the Census Bureau identifies four broad regions, then nine divisions. We'll largely use the divisions.1

We'll spend two essays covering the South. The next installment will cover the South Atlantic.2 Today's focus will be on the East and West South Central divisions.

The area is historically the heartland of the Democratic party. Between 1880 and 1952 Republican candidates fared quite poorly: in seventy-two years, they carried Kentucky three times, Oklahoma and Tennessee twice, Texas once, and they were shut out in the remaining states. Eisenhower was the first to carry a majority of them in 1956. Nixon won every one in his 1972 landslide. Yet as recently as 1976, Jimmy Carter won every state in these divisions, while Bill Clinton won half of them in 1992 and 1996. Lately, they have moved squarely into the Republican column. George W. Bush won all of them in 2000 while losing the nationwide popular vote. Barack Obama is the first Democrat to win a Presidential election without carrying a single one.

Democrats like Clinton and to a lesser extent Carter were able to carry these states by forming biracial voting coalitions of lower-income whites and African Americans (African American turnout in the South was still relatively low in the 1970s). As we shall see, Obama was unable to do that. As a consequence, racially polarized voting - wherein whites voted for the Republican Party and African Americans voted for the Democratic Party - was at least as high in 2008 as it was in 1972, 2000, or 2004.

Let's first see how these divisions have voted across the last six electoral cycles.

Regionwide Share of Vote, East and West South Central.jpg

For the last thirty years, Republicans have consistently performed better here than they have nationwide - but when the GOP's fortunes decline nationally, candidates could expect a similar decline in the South Central. This trend was disrupted in the last election. Although John McCain did much worse than Bush nationwide, he only did slightly worse in the South Central.

To see why, let's examine countywide vote results. We'll divide the counties in both divisions into three groups: safe Democratic (i.e. won by Clinton in 1996 and John Kerry in 2004), safe Republican (i.e. won by Bob Dole in 1996 and Bush in 2004), and swing counties (i.e. won by Clinton in 1996 and Bush in 2004). [It is worth noting here that there were only a handful of swing counties that voted for Dole and Kerry, and none in the South Central divisions.]

Dem Vote by Partisanship, South Central Regions.jpg

As we can see, Obama did about as well as Clinton in the safe Democratic counties. What's more, he improved over Kerry in the safe Republican counties. But there is a surprise. In both divisions, Obama did slightly worse than Kerry in the swing counties. This is noteworthy, considering that nationwide he did significantly better.

Who lives in these counties? To answer this, we'll make use of the 2000 Census data on income, and 2007 Census estimates on population change and racial composition.

We'll use another tool based on the Census Bureau's statistical areas (CBSAs), which classify counties by metropolitan community. A brief explanatory note has been placed in this endnote. The bottom line is that we're going to classify counties by metropolitan status, then by population: from "Mega City" to "Rural."

Let's examine some demographic information for these swing counties.

Demographic Profile of Swing Counties, South Central Regions.jpg

When we compare the swing counties to the Democratic counties and the Republican counties, we find several important differences. First, the swing counties have more whites on average than the Democratic counties, and about as many as the Republican counties. They are also more rural than the strong partisan counties. As is the case for most rural areas nationwide, white income tends to be lower, and population growth has been stagnant if not shrinking.

We can appreciate this from another angle. Let's return to our CBSA classifications to see how Democrats have performed by metropolitan status going back to 1988. We'll look at rural counties, small town counties, and large town counties.

Democratic Share of Two Party Vote by CBSA Region, 1988-2008.jpg

We find Obama doing worse than any of the previous four Democratic candidates in the rural and small town counties, which is consistent with our findings on the swing counties. We also find Obama doing better in the large towns than Kerry. Later on, we'll discuss his performance among African Americans. For now, it's sufficient to say that enhanced turnout and support among African Americans probably aided him in the larger communities.4

All in all, we have a good intuition of what has happened in the South Central Divisions. Clinton's victories were likely formed by a biracial voting coalition that included lower income whites in rural and small town counties. In 2000 and 2004 Bush flipped these divisions by winning many of these voters. In 2008, while the rest of the country moved toward Obama and the Democrats, they did not. In some places, they seem to have inched a bit closer to the GOP. 5

The countywide vote data we have examined so far cannot confirm these intuitions - no matter how many different ways we slice and dice the numbers. However, this is where we can bring exit poll data to bear. We'll develop a kind of "divisional exit poll" for the East and West South Central divisions by averaging each state's results according to population.

That produces the following results:

Select Exit Poll Results, South Central Regions.jpg

It is important to note that these averages occasionally mask some important differences. For instance, consider the East South Central division. Whites in Kentucky and Tennessee were more supportive of Clinton, Kerry, and Obama than whites in Mississippi and Alabama. Additionally, in the West South Central division, Obama underperformed Kerry among whites in Arkansas and Louisiana. In Oklahoma and Texas, he ran about the same.6

Still, across both divisions, we see essentially the same trend. McCain was typically able to match Bush's share of the white vote - and he was apparently able to tick up a few points among some subgroups: white Democrats, white Independents, and poorer whites in particular. This seems due to his slightly larger share of the vote in the "swing counties," which, as we saw above, are more rural and (among whites) slightly poorer than strongly partisan counties. Also noteworthy is that much of the movement among whites from Clinton to McCain came from white women; McCain and Bush closed a ten-point gender gap that Clinton had opened up in the divisions.

Our final question is a fairly obvious one. If McCain did at least as well among Bush with white voters in these divisions, how did Obama improve upon Kerry overall?

Obama did not improve upon Kerry in Oklahoma, Arkansas, Louisiana or Tennessee. He did improve in the other states. That appears to have been due in part to increased support from African Americans. For instance, in Mississippi, Bush won 10% of the African American vote in 2004. McCain won just 2%. In Texas, where Obama saw his largest gains, this pattern was supplemented by improvement among Hispanics: Kerry won 50% of Hispanics; Obama won 63%.7

In conclusion, we would note the following. While McCain's victories here were not surprising, they do carry some significance. After all, for the last twenty years (counting George H.W. Bush as a Texan, given that he began his political career there), the Chief Executive of the nation has hailed from the West South Central division - and the overlap between Clinton and Bush's voting coalitions here seems to have been pretty significant. Clinton was able to forge a coalition in this part of the country that united African Americans with lower income, rural/small town whites. Bush brought many of the latter into a coalition built upon the Republican base. So, while they don't get much press, these voters have, in a strange way, been ruling the roost for twenty years.

But this era is now over. Even as much of the rest of America swung toward Obama and the Democrats - these voters stuck with the Republican Party. In fact, it appears the GOP actually won a few more of them. For almost twenty years, you could say that the "first choice" of these swing voters would become President, but that's no longer the case. That seems pretty significant.



[1]The reason for this is that we want to develop a relatively broad explanation of the election. If we looked at the results state-by-state, we would surely get plenty of detail, but we would probably lose track of the big picture. On the other hand, examining a broad region all at once might cause us to overlook some important intra-regional differences. So, the divisions serve as a nice middle ground.

[2]The South Atlantic division saw as much electioneering as any other. Three states (Virginia, North Carolina, and Florida) flipped from the Republican column to the Democratic column, and one more (Georgia) nearly did. And, as we'll see, voter behavior in the South Atlantic was often quite different than it was in the rest of the South. Accordingly, we'll hold off on the South Atlantic until the next installment in this series - so that we can dedicate our full attention to what happened in that part of the country.

[3]The Census Bureau groups certain counties into metropolitan or micropolitan areas according to their interrelationship. For instance, the Pittsburgh Metropolitan Area consists of the counties that the Census Bureau deems are inter-related with the city and each other: Allegheny, Armstrong, Beaver, Butler, Fayette, Washington, and Westmoreland counties. If you live in one of these counties, you are said to live in metropolitan Pittsburgh. This is a Census Bureau Statistical Area, or CBSA.

We're going to make use of this metric in the following way. We'll divide the counties into their CBSAs, and then we'll rank the CBSAs according to population:

Labels for CBSAs.jpg

[4] If we examine Democratic performance in the larger metropolitan areas, from "Small City" to "Mega City," we find the following:

Democratic Share of Two Party Vote by CBSA Region 2, 1988-2008.jpg

In the East South Central, Obama still underperformed Clinton in the cities, though he did improve upon Kerry, Gore, and Dukakis. In the West South Central, we find the same result for the small and large cities: Obama improved upon Kerry, Gore, and Dukakis, but still trailed Clinton. However, in the so-called "Mega Cities" - i.e. metro Dallas and Houston - Obama did better than Clinton. In fact, Obama won Dallas and Harris Counties - a feat Clinton could not pull off in 1996.

As we'll see, these successes was likely due to the large percentage of Hispanics and/or African Americans in these areas. When it comes to Dallas and Houston, Obama's gains might also have been due to a shift among upper-income white voters in Texas. In 2004, Bush won them by 57 points. In 2008, McCain won them by 49 points. White median family income is higher in metro Houston and metro Dallas than in any other type of area in the state, so it is reasonable to infer that this movement among high-income whites helped him here.

Interestingly, Obama did not significantly improve statewide among whites in Texas. The reason? He actually declined relative to Kerry among poorer whites, thus balancing his improvement among wealthier whites.

[5] The swing counties' movement from Democratic to Republican has produced some prominent changes to the electoral map:

South Central, 1996-2008.jpg

We see that Clinton won almost all of the same counties that Obama did - and then some (Obama added some important counties such as Harris and Dallas in Texas, and Jefferson (Birmingham) in Alabama). Both Clinton and Obama performed well in the heavily Hispanic counties in the Rio Grande Valley, in the "black belt" (named originally for the rich soil) across eastern Mississippi and western Alabama, and the Mississippi delta region.

But the similarities end there. Clinton carried a good portion of the "Little Dixie" area of Southeastern Oklahoma; Obama did not. Clinton performed well in his home state of Arkansas; Obama did not. Clinton did well in the rural Fourth and Sixth Congressional Districts in Tennessee and the Fifth District in Alabama; Obama did not. And perhaps most strikingly, Clinton carried the "Old Seventh" Congressional district (dismembered in 1992) in Eastern Kentucky; Obama did not. This is quite a change: most of these counties have voted Democratic since they were organized by the United Mine Workers in the 1930s.

Of course, Kerry also underperformed Clinton in these sections of these states. What is significant is that while Obama won back much of what Kerry had lost from Clinton in most other areas of the nation, the Republican trend persisted here.

[6] As mentioned in the previous footnote, he improved among wealthier whites in Texas. This does not show up in the divisional exit poll because McCain did better than Bush among wealthier whites in Louisiana and Arkansas, and about the same in Oklahoma.

[7] In Louisiana and Arkansas, African American voters appear to have been more supportive of Obama than Kerry; however, this movement toward Obama seems to have been overwhelmed by McCain's improvement among whites.

Generally speaking, Obama's strong performance among African Americans and Hispanics aided him greatly in this division. Another look at the 2008 electoral map should drive the point home. This time, we'll supplement it with maps of the African American and Hispanic populations.

Race, Ethnicity, and the 2008 Vote.jpg

There is a very tight correlation here, which is consistent with everything else we have found.

Sean Trende can be contacted at

Jay Cost and Sean Trende

Author Archive

Follow Real Clear Politics

Latest On Twitter