m on November 20th, 2008

(This is how I spent GIS Day)

I was surprised in my previous post by how young and black Louisiana was (in 2000), yet how not for Obama it went. Only 10 of 64 parishes were carried by the Democrat, though they included three of the four most populous parishes. I wondered if maybe there was a rupture between the demographic data I had (from the 2000 Census) and the election data from this year.

That rupture, of course, could be Katrina. The hurricane obviously changed the demographics of the state somewhat, but how much? I’m certain that the campaigns canvassed a good demo picture of the state, but I don’t have (I don’t think) access to that data. Instead, I wonder if we can detect a Katrina effect knowing simply that there was a huge change in the population voting for president between 2004 and 2008. And once we detect it, can we measure it. And once we do that, can we speculate as to a possible result in 2008 that did not involve Katrina.

So let’s start with the demographic data. In 2000, Louisiana was about 1/3 African-American (145k against a total population of 445k). Over half of that population was in four parishes (in descending order): Orleans (where New Orleans is), East Baton Rouge (Baton Rouge), Caddo (Shreveport), and Jefferson (New Orleans suburbs). To give a sense of the drop off: in 2000, there were over 10,000 African-Americans in Jefferson Parish. The next largest population was in Ouachita Parish, which had not even 5000. Jefferson is also notable in that it was only 23% black. African-Americans made up larger portions of the populations of the other three parishes, topping out with 67% of Orleans Parish. Here, then, is a simple chart.

African-Americans vs. others in 2000, by parish

African-Americans vs. others in 2000, by parish

What’s clearest here is that, surprise, Louisiana has some pretty sparsely populated parishes, a few of which, actually, end up having black majorities. The real numbers are so small, though, that they just look like little chips on the map. But since it’s the people that matter, not the space, I made cartograms of the state for the past three presidential elections (Data: 2000, 2004, 2008). To make the cartograms, I used Tom Gross’s Cartogram Geoprocessing Tool, which itself is based on cart, by Michael T. Gastner and M. E. J. Newman. As with the last post, the colors change at 5% and 10% margins of victory. The Republican (Bush and McCain) is red, and the Democrat (Gore, Kerry, Obama) is blue.

Presidential election cartograms sized by 2000 population

Presidential election cartograms sized by 2000 population

Now things get a little clearer (and far more cetacean). Orleans is reliably, strongly Democratic, as are the teeny parishes on the western edge of Mississippi. Yet Obama was able to flip Caddo Parish back to blue and carried Baton Rouge’s parish, which neither Kerry nor Gore could. We can speculate as to why other parishes went more for McCain than they did for Bush (like 42% black St. Landry Parish, which Gore managed to carry but went soft to the GOP in 2004 and 2008), but my primary interest here is in speculating about Katrina, which means focusing on Orleans Parish.

Orleans Parish in 2000 was about 2/3 African-American. Gore carried it by 54%, Kerry by 55%, and Obama by 59%. This is predictable based on the higher African-American support for Obama in comparison to Gore and Kerry. It wasn’t that much bigger (it would be hard for it to be so), but it was bigger. But if we look at the raw vote totals, perhaps we can get a sense of a parish that has lost its population. Here we have the total number of votes cast, the percentage that went to the Democrat, and the change from the previous election:

2000: 181,221 (76%)

2004: 195,269 (78%) (up 8%)

2008: 146,287 (79%) (down 25%)

Interesting. Obama got a larger share of the vote (and won by a larger percentage), but got fewer total votes. Nearly 50k fewer people voted in Orleans Parish in 2008 than in 2004. How did the vote count change statewide?

2000: 1,768,656

2004: 1,928,049 (up 9%)

2008: 1,958,059 (up 2%)

So even though turnout was actually up in Louisiana in 2008 generally, Orleans Parish suffered an intense drop. In fact, in 2008, Orleans dropped to third among parishes by total number of votes cast. In the previous two elections, it led the way. Only six parishes had fewer votes in 2008 than in 2004: Orleans, Jefferson, St. Bernard, Plaquemines, Red River, and Cameron. Plaquemines and St. Bernard are, of course, the two parishes just southeast of Orleans Parish. Interestingly, St. Bernard had an even steeper drop in votes cast than Orleans Parish. Not even half as many votes were cast in that parish in 2008 as had been in 2004. Cameron and Red River Parishes had relatively small drops of votes cast. Cameron’s 17% drop was big proportional to it, but both Parishes put together had just over 800 fewer votes total. Considering that of the Katrina parishes (the other four), over 77k fewer votes were cast, those western parishes are a drop in the bucket.

So despite the general tick up of 2% in the Louisiana voting, over 77k fewer votes were cast in Orleans, St. Bernard, Jefferson, and Plaquemines parishes. In fact, not including those four parishes, the statewide increase was actually not 2% but over 6%! And if you consider further that only two parishes had fewer votes in 2004 than 2000 (Tensas and Madison), and that both decreased by under 3% each, it’s not that far of a stretch to assume that without Katrina, there would’ve been increases in those four parishes. So it’s obvious that something huge, and Katrina is a great culprit, suppressed the hell out of the vote in southeastern Louisiana.

So now to finish: absent Katrina, could Obama have won?

Statewide, 1,958,059 votes were cast this year. Obama came 198,049 short of a majority. So even if the 50k Orleans Parish lost would’ve all voted for Obama (which is close to assumable, considering he won 80% of the votes in the Parish as it was), he still would have only started to chip away at that imposing 200k vote difference. Additionally, the other three depressed parishes went strongly for McCain, so the big strides made by the extra votes in Orleans would’ve been nibbled away by the extra McCain support in the other parishes. It’s possible that Jefferson would have gone a little less strongly for McCain (Obama carried a smaller portion of the parish in comparison to Kerry, but not much smaller), but there’s still no way. Furthermore, in St. Bernard, McCain got about 10,000 fewer votes than Bush did, while Obama only got about 6,500 fewer votes than Kerry.

But there’s one more thing to take into account, which is visible in the cartograms: East Baton Rouge Parish’s going for Obama could, in fact, be the result of Katrina relocation. Over 14,000 more people voted in that parish this year than in 2004. And in that parish, McCain got about 5,000 fewer votes than Bush did, but Obama got over 17,000 more than Kerry did. Furthermore, though many parishes posted gains of 10k or more voters between 2000 and 2004, only East Baton Rouge did so between 2004 and 2008.

So, sadly, it looks like, tragedy that Katrina was, it did not prevent Obama from carrying Louisiana. Orleans Parish, large as it is, simply was not large enough to make the difference. Furthermore, ultimately, the racial makeup (assuming a level of correlation between racial identity and support for Obama) was not big enough to push the vote in Obama’s direction. There is no doubt that Orleans Parish, which was both the blackest and most vote gutted parish in the state underwent an immense demographic upheaval. But compared to the rest of the state, it wasn’t big enough. Furthermore, while it is true that many parishes in Louisiana have large African-American populations, many of them–especially the parishes adjoining Mississippi–are so small in terms of population that they could not overwhelm the huge McCain strongholds like St. Tammany Parish (which McCain carried by almost 60,000 votes).

That said, the new taste for Democrats in Baton Rouge and Shreveport bodes well for the future.

Tags: , , , , ,