In reading this story and exploring the corresponding map, I began to wonder just how those areas most dependent on government income voted over time. While the story focused in on one county in Minnesota, I wanted to broaden the scope and see if any interesting patterns emerged. What I decided to do was zoom in on those counties who had the highest degree of what we might call "dependence"--those who received more than 40% of their income from these sources--these are the counties colored dark red on the NYT map.
In total, there are 82 counties nationwide that fall into this category (I excluded the Alaska census area because voting data wasn't available for it). I produced a spreadsheet, seen below, that lists these counties--by state--along with their percentage of government "dependence" as well as two other variables that I thought might be revealing: their racial makeup shown as % white; and their poverty rate (both from Census data). Oftentimes in our politics there has been an assumption that poverty and government dependence is concentrated in minority communities. This data clearly refutes that.
Next, I wanted to see how each of these counties voted in recent elections. Rather than code each election, I picked a few that I thought might be of particular interest. I coded the two most recent presidential elections to see the most recent political behavior of these areas and to explore whether there was any noticeable short term change. I then decided to look at 1992. 1992 is of note in that it, like 2008, was an election contested during an economic downturn and might produce similar "pro-government" sentiment in those areas of greatest need. Similarly, I picked 1980. Also conducted during economic difficulties, 1980 is notable for the rise of Ronald Reagan and the ascendancy of anti-government rhetoric. Thus, as American politics began to transition away from the New Deal assumptions of the previous generation, we might wonder if areas strongly dependent on government resisted Reagan. Finally, I picked the 1964 Johnson landslide. The thinking here is that this election would probably represent the apex of pro-government voting.
In looking at the spreadsheet, very few clear cut trends emerge and there is certainly no correlation between government dependence and support for the Democratic party. Things are much more complicated (and hence interesting). Nonetheless, there are a number of things to take note of and explore. In 2008, of these 82 counties, only 30 voted for Barack Obama. This is the type of result that makes many Democrats' heads spin. How is it, they ask, that those voters most benefitting from programs championed by Democrats vote "against their interests"??? This phenomenon garnered a lot of attention a few years back with the publication of Thomas Frank's "What's The Matter With Kansas?" If we map this voting--coloring Obama counties Blue and McCain counties Red--we get the following:
The most interesting aspect of this map, I think, is the clustering of McCain counties in the Kentucky, Tennesse, Missouri region. These counties are, as the data suggests, overwhelmingly white and poor. They also tend to be mountainous, low population, counties in the Appalachian and Ozark ranges. Looking at the voting over time, these are also counties that have remained, for the most part, consistently Republican. Clearly, it seems as if there is more than just economics at play here. Indeed, these areas have long had a political culture that has confounded Democrats' ability to compete, going back generations. The one modern Democrat who performed well in this region, perhaps not surprisingly, was Bill Clinton. Not only did Clinton win 60 of the 82 counties nationwide, he did particurly well in this cluster of KY/TN/MO counties--so much so that he won all three states in both 1992 and 1996. No Democrat has won any of the three since.
How do we explain this change? While these counties and regions have changed little over the past decades, they were receptive to Clinton but not Obama (and Kerry). Going back further, these counties also supported Carter and Johnson. Is the shift a result of a broader movement against the Democrats, as hinted at in the original NYT story? Is it the result of the fact that the most recent Democratic nominees were northerners who were perceived as foreign to this region's culture and people? Clinton hailed from this area and both Carter and LBJ were southerners so there might be some credence to this hypothesis.
A few other observations. 1980 stands out as the year in which these counties' vote was most divided, with Reagan winning 42 and Carter 40 counties respectively. Thus, the notion that this was perhaps a "tipping point" election may have some confirmation.
Looking at the race variable, while I've so far discussed the overwhelmingly white KY/TN/MO counties, there are a number of overwhelmingly African American counties represented in the data. Specifically, Perry and Wilcox County Alabama, Marion County South Carolina, and Holmes, Jefferson, Humphreys, and Quitman Counties in Mississippi. Here, as expected, there was overwhelming support for all of the recent Democratic candidates (back during the lead up to the 2008 election I did a series of posts on the interesting political geography of these regions--see here and here). Also remember, when looking at the 1964 vote in these counties, that the Voting Rights Act had yet to be passed. Next, there are a few counties with large Hispanic populations--see those in Texas as well as Mora and Guadalupe in New Mexico. Finally, make note of some counties with a large Native American population--Apache in Arizona; Sioux in North Dakota; Buffalo and Shannon in South Dakota. Like with the heavily African American counties, those with large Hispanic or Native American populations have voted as we would expect.
Perhaps the most confounding set of counties--and those that I will need to research more deeply--are those found in Michigan. What is interesting about these counties is that they are overwhelmingly white, have high degrees of governement "dependence," but are not terribly poor. Their poverty rates all hover near the statewide average of 15%. So what explains this? Going back to the original NYT map, you can separate out the different components of government support. These counties show a heavy reliance on both Social Security and Medicare, suggesting a large elderly population. At the same time, though, these counties also have large reliance on unemployment insurance. In a lot of ways, these counties seem quite similar to Chisago County Minnesota which was the focus of the NYT story.
These are just a few observations based on a cursory examination of this data. There's a lot more that I hope to delve into, especially focusing on individual counties or regions, in the coming weeks. What does come through, though, is that voters and regions have voting histories and behaviors that don't fit into a simple narrative or explanation. This is especially the case when it comes to the correlation between reliance on government and party support. As the Times story makes clear, voters possess a series of often contradictory feelings and beliefs. For those on the left who aspire for an electorate that will vote strictly along economic lines, this data is bound to frustrate.