Jay Cost Election Review, Part 4: The Midwest

[Nym90]

Today we continue our election review with the Midwest region.

From a national perspective, the elections of 1996 and 2008 turned out to have a lot in common. Bill Clinton won 54.7% of the two-party vote in 1996. This cycle, Barack Obama won 53.7% of the two-party vote. Accordingly, 1996 serves as a good baseline to analyze the 2008 electorate - and we have seen some interesting trends by using it. For starters, we found that, although nationwide the two Democrats pulled about the same vote, Obama did worse in the South Central division than Clinton, a bit better in the South Atlantic, and better still in the Pacific West. We also found that Obama typically did worse among rural and small town voters, but better among larger city voters. This is how we have been able to identify the uniqueness of Obama's majority voting coalition - by looking at another Democrat with roughly the same nationwide vote, we can use regional and sectional differences between the two to specify Obama's coalition.

When we turn to the Midwest, the differences appear to be less pronounced. In 1996 Bill Clinton won 100 electoral votes in the Midwest. This cycle, Obama won 97. They won the same states - except Obama swapped Missouri for Indiana. And, as far as share of the two-party vote goes, there were very few differences. Let's start with the East North Central.



We find very little change between 1996 and 2008. Overall, Obama improved upon Clinton by about 1.4%. Once again, he did a bit worse in the rural and small town areas - and a bit better in the larger urban areas. In fact, much of Obama's improvement upon Clinton is thanks to a 5-point increase in his share of the vote in the "mega city" category, i.e. metropolitan Chicago.

The story is basically the same when we examine the constituent states of this division: Wisconsin, Illinois, Indiana, Michigan, and Ohio. Obama improved a point or two upon Clinton, typically doing better in the urban centers and worse in the rural regions.

Even Indiana, which impressively flipped to Obama, was still pretty similar to 1996. For starters, Dole won the state, but Clinton was able to pull 46.9% of the two-party vote. Obama won a narrow, 50.5% victory there, thanks to improvement in the larger urban centers of the state: South Bend, Indianapolis, and the counties in the northwest that are part of greater Chicago. This improvement in Indiana was countered by a slight drop-off relative to Clinton in Ohio - where Obama was weaker than the 42nd President in the rural areas, small towns, and large towns.

The exit polls tell the same tale.



While George W. Bush won a 10-point victory among whites in 2004, Clinton and Obama split them with their Republican opponents.

If Obama did a few points better than Clinton in the East North central, he did a few points worse in the West North Central, as this graph makes clear.



Again, we see Obama improving in the larger urban areas while doing worse in the rural areas. Of course, in the East and West North Central divisions, Obama typically improved upon Kerry - which cannot be said about the East and West South Central divisions.

Once again, there are few differences when we toggle from state to state in the West North Central: North Dakota, South Dakota, Nebraska, Kansas, Missouri, Iowa, and Minnesota. Again and again, Obama ticks up a few points among voters in larger urban centers, and down a few points in rural and small town areas. [This is how he was able to flip the electoral vote allocated to Nebraska's second congressional district - he did about seven points better than Clinton in metro Omaha.] The only exceptions are Minnesota, where Obama consistently underperformed Clinton by 4-5 points across regions, and Missouri, where Obama only matched Clinton's strength in the larger urban areas of St. Louis and Kansas City.

The exit polls for the division generally confirm what the votes indicate.



Clinton was able to win the white vote in the West North Central, but Obama was not. This was due to the fact that he didn't improve upon Clinton in any state in this division, but did worse in Missouri and Minnesota. But notice there was improvement upon Kerry. George W. Bush seems to have been able to win lower income whites in 2004, but Obama seems to have flipped them back.

Looking at a straightforward map of countywide votes in the Midwest is not particularly illuminating. The counties won by both Obama and Clinton are largely identical. We can get a better sense of the changes in the Obama and Clinton coalitions by looking at the change of vote share in the counties from 1996 to 2008. This is what the map below shows. The darkest red counties shifted 10 points or more toward McCain from Dole (we are, again, talking about share of the two-party vote), while the bluest counties shifted ten points or more toward Obama. Every gradation of color is worth a 1% shift.

[Nym90]

Starting in the northwest corner of the map, we see that Obama generally underperformed relative to Clinton in western North and South Dakota, with the general exception of the Native American reservations in the center of the map (we saw a similar phenomenon in the West). Obama was able to move Grand Forks, Fargo, and to a lesser extent, Sioux Falls his direction. Similarly, we see the movement in eastern Nebraska resulting from Obama's targeting of Omaha.

The two dark blue counties in Kansas are Douglas and Riley counties, home to the University of Kansas and Kansas State University. Johnson County (suburban Kansas City) moved strongly toward Obama, although McCain still carried it. Emporia and Sedgwick (Wichita) counties also have substantial student populations, and moved toward Obama, while Bob Dole's home county of Russell, perhaps not unexpectedly, moved toward Obama. The other remarkable aspect of the Kansas map is that southeastern Kansas (the "little Balkans") moved heavily toward the Republicans - for an area of the country that was once a hub of socialist agitation, this is an interesting historical development.

The rightward drift of Missouri over the past decade is confirmed by this map. Kansas City in the west and St. Louis in the east moved toward Obama, and their metropolitan areas only moved marginally toward the Republicans. Boone County (University of Missouri) moved toward Obama. But the rest of the state is a sea of red.

The Upper Midwest has long been long viewed as something of a counterweight for Republicans - while the Mountain West appeared to shift toward Democrats, this region moved toward Republicans during the 90s and early 2000s. We still see some evidence of this in this year's results. While Iowa is fairly unremarkable, it did shift about a point toward McCain versus Dole. Given Obama's strong primary performance and McCain's opposition to ethanol, this is somewhat surprising. In Minnesota, the twin cities moved a bit toward Obama, and he performed well in a few spillover counties from Grand Forks and in a heavily Native American county in the Iron Range - but the rest of the state shifted toward McCain, including suburban Anoka and Dakota counties.

Wisconsin moved a point toward Obama vis-à-vis Clinton, but as we can see, much of this movement is confined to the southern portion of the state. Dane County (Madison) moved the most toward Obama, but it was certainly not the sole county in southern Wisconsin to shift.

In Michigan we see that McCain improved relative to Dole in the Upper Peninsula and the rural northern tier of the Lower Peninsula. In the more urban southern tier of the state, Obama improved markedly. In Illinois, Obama did well in the north, and surprisingly well in historically Republican northwest Illinois. Southern, agricultural Illinois swung heavily against the state's favorite son. In Indiana we see the greater Indianapolis area moved toward Obama, as did the areas around Elkhart and South Bend in the north. The southern portion of the state moved toward McCain. Finally in Ohio, we note McCain's strength in the Ohio River valley. Obama improved markedly in Hamilton and Franklin (Columbus) counties. But overall, the state moved about six points toward the Republicans from where Bill Clinton had it in his eight-point national victory.1

In conclusion, two points are worth noting. First, unlike the South, the Midwest did not shift much relative to 1996. Both elections are similar in that Clinton and Obama won about 54% of the nationwide two-party vote. Yet Clinton was significantly stronger in the South Central divisions than Obama, while Obama was slightly stronger in the South Atlantic. But in the Midwest, it was "back to the future:" regionwide the results in 2008 were noticeably similar to 1996. Obama did a bit better than Clinton in the East North Central, where he won a state (Indiana) Clinton couldn't, and a bit worse in the West North Central, where he couldn't win a state (Missouri) Clinton won.

Of course, it's not true that everything was static, which leads to the second point. The Midwest region saw the same trend we witnessed in the South and West: Obama doing worse than Clinton in the rural and small town areas, while doing better in the larger urban centers. Unlike the South Central and parts of the South Atlantic, Obama was generally able to do better than Kerry in Midwestern rural and small town areas.

What does this mean? Is this a sign of a political realignment? Not necessarily. As we noted last week, victorious presidential candidates form their own, unique voting coalitions that are never entirely replicated. So, this shift among urban and rural/small town voters might be due to the fact that Obama's political roots are more urban, while Clinton's are more rural. In many respects, a vote for president is quite personal - so voters might simply be responding to traits or qualities that they can relate to.

***
Endnote

[1]We also note that despite a surge in turnout nationwide, turnout in Ohio was little changed. In 2004, there were 5.72 million voters. In 2008 there were 5.77 million voters (according to the latest CNN numbers). Moreover, despite a massive turnout effort by both camps, turnout was down in most counties. The counties shaded below in red are counties where turnout fell.



We note first that urban counties such as Cuyahoga (Cleveland), Hamilton (Cincinnati), and Stark (Akron), and Mahoning (Youngstown) experienced low turnout. We also note that the counties along the Ohio River -- the 6th and 10th districts before the 1990 redistricting - had a drop in turnout. If you will recall from the primary campaign, these latter counties are part of Appalachia, and were heavily pro-Clinton. It is possible that some of these Clinton voters simply decided not to show up on election day. Although it is difficult to tell without precinct-level maps, we might also hypothesize that many of the white blue collar ethnics in east Cuyahoga county and in Hamilton, Stark, and Mahoning counties followed a similar path, offsetting a likely increase in African American turnout in these counties.