This may be another one of those cases where I should wait until there’s actually been an election. But there have been a number of instances lately where I commented on a particular district’s PVI and thought to myself, “Yeah, but that PVI is going to change a lot after this election.” Nobody is polling the presidential race at the level of individual districts (except, of course, the states that have only one House district), so short of breaking out the crystal ball, I can’t address House districts with much specificity.
However, what I can do is use current polling to predict likely percentage splits for each state in 2008, and plug those numbers into the PVI formula along with the 2004 percentages to calculate new PVIs for each state. (PVI, of course, is used in the context of House districts, but the formula is easy enough, and can be applied to pretty much any unit of analysis: states, counties, legislative districts, precincts, and so on. In fact, I’m surprised it isn’t, as a means of analyzing Senate races.)
For 2008 numbers, I’m just going to use today’s projection for each state from 538.com. (These numbers fluctuate a bit every day, so this post will already be out of date tomorrow.) Not to say that Nate’s prediction model is the be-all-and-end-all, just that it’s a good model for my purposes, since it basically pushes every leaner and accounts for third-party votes (so that each total actually adds up to 100). Over the flip, for each state, is the old PVI (reflecting the 2000 and 2004 elections), the new PVI (reflecting the 2004 and 2008 elections), and the difference.
Not surprisingly (since we’re measuring the same thing, although my numbers are blunted by being averaged out with 2004 results), the “difference” results look a lot like this map Chris Bowers put together at Open Left showing the biggest shifts in the new Obama electoral map. Big D+ shifts in the West and Great Plains, R+ shifts in Appalachia and the Northeast. (This doesn’t mean that, for instance, the Northeast is most rightward or going to give a smaller percentage to Obama than Kerry. It’s more like it’s standing still while the rest of the country moves left.)
| State | 00-04 PVI | 04-08 PVI | Difference |
|---|---|---|---|
| Alabama | R+9.6 | R+11.2 | R+1.6 |
| Alaska | R+13.6 | R+8.1 | D+5.5 |
| Arizona | R+3.7 | R+4.9 | R+1.2 |
| Arkansas | R+3.3 | R+6.3 | R+3.0 |
| California | D+5.9 | D+6.9 | D+1.0 |
| Colorado | R+2.8 | R+0.5 | D+2.3 |
| Connecticut | D+7.5 | D+6.8 | R+0.7 |
| Delaware | D+5.7 | D+4.2 | R+1.5 |
| Florida | R+0.8 | R+2.2 | R+1.4 |
| Georgia | R+6.6 | R+6.3 | D+0.3 |
| Hawaii | D+7.3 | D+6.6 | R+0.7 |
| Idaho | R+18.9 | R+13.4 | D+5.5 |
| Illinois | D+6.1 | D+5.9 | R+0.2 |
| Indiana | R+8.6 | R+5.8 | D+2.8 |
| Iowa | D+0.2 | D+1.5 | D+1.3 |
| Kansas | R+11.1 | R+10.2 | D+0.9 |
| Kentucky | R+8.3 | R+9.4 | R+1.1 |
| Louisiana | R+5.1 | R+7.3 | R+2.2 |
| Maine | D+4.0 | D+6.0 | D+2.0 |
| Maryland | D+7.8 | D+7.4 | R+0.4 |
| Massachusetts | D+13.6 | D+10.1 | R+3.5 |
| Michigan | D+2.6 | D+1.9 | R+0.7 |
| Minnesota | D+2.0 | D+3.8 | D+1.8 |
| Mississippi | R+8.7 | R+7.8 | D+0.9 |
| Missouri | R+2.2 | R+2.5 | R+0.3 |
| Montana | R+10.9 | R+5.6 | D+5.3 |
| Nebraska | R+15.1 | R+12.0 | D+3.1 |
| Nevada | R+1.1 | R+1.3 | R+0.2 |
| New Hampshire | D+0.5 | D+1.8 | D+1.3 |
| New Jersey | D+6.1 | D+3.7 | R+2.4 |
| New Mexico | D+0.4 | D+1.0 | D+0.6 |
| New York | D+11.3 | D+8.7 | R+2.6 |
| North Carolina | R+5.8 | R+4.2 | D+1.6 |
| North Dakota | R+13.3 | R+8.2 | D+5.1 |
| Ohio | R+0.9 | D+0.2 | D+1.1 |
| Oklahoma | R+12.8 | R+12.8 | R+0.0 |
| Oregon | D+1.6 | D+3.1 | D+1.5 |
| Pennsylvania | D+2.2 | D+2.1 | R+0.1 |
| Rhode Island | D+13.0 | D+10.6 | R+2.4 |
| South Carolina | R+7.8 | R+6.4 | D+1.4 |
| South Dakota | R+10.6 | R+7.0 | D+3.6 |
| Tennessee | R+4.1 | R+7.4 | R+3.3 |
| Texas | R+10.6 | R+8.1 | D+2.5 |
| Utah | R+21.8 | R+17.9 | D+3.9 |
| Vermont | D+8.0 | D+11.1 | D+3.1 |
| Virginia | R+3.6 | R+2.1 | D+1.5 |
| Washington | D+3.7 | D+4.6 | D+0.9 |
| West Virginia | R+4.3 | R+5.4 | R+1.1 |
| Wisconsin | D+0.6 | D+2.3 | D+1.7 |
| Wyoming | R+19.5 | R+14.0 | D+5.5 |
Now you might be sitting there thinking “Yeah, but I really want to know how Congressional District X is going to change!” Well, here’s a very rough method you might use: take the difference from the state where the district is, and apply it to the district. I’ll use my home district (WA-07) as an example. It’s currently D+30.3. Add D+0.9, and the adjusted PVI is D+31.2.
(Again, this is a rough method… different parts of different states are reddening or bluening at different rates. For instance, Seattle might not be bluening as fast as the sagebrush parts of the state; it’s kind of maxed out on liberalism, while eastern Washington has lots in common with Idaho, Montana, etc., where Obama is making up the most ground. On the other hand, because 2000 and the huge Nader effect that occurred is dropping out of the equation, maybe the PVI will shoot up even more in Seattle. Hard to tell, so just exercise your judgment.)
One last question, as a bonus. Does this rough method change the most and least liberal districts in the nation? Well, NY-16 is still safe in its position at #1, although it drops from D+43.4 to a nice moderate D+40.8. However, the most conservative district changes, as Utah is rapidly going one direction and Alabama is going the other. UT-03 falls from R+26.2 to R+22.3, while AL-06 goes up from R+25.2 to a batshit insane R+26.8.
I’m sure districts with large black populations will see a big swing towards the dems due to Obama. This could be enough to win districts like NC-08 and OH-01.
I thought Obama would have a homestate boost in these states. But apparently 2008 will make them more republican.
There are three states where Democrats pick up 5 or more PVI points: Alaska, Montana, and Wyoming. The one state with the largest Republican gain (+3.5) is Massachusetts.
Crisitunity, you rock.
While an interesting measure, PVI certainly has its problems. Anyone who lives in the south knows that Democrats are a lot stronger at the local, state, and congressional levels than at the senate and presidential levels down here. So Democrats winning districts with solid Republican PVIs are not really upsets at all.
Another problem is using the national averages. I’ve often noticed that overall voter turnout tends to be lower in “uncompetitive” states than in “competitive” states (not the least because the former get very little if any campaigning or advertising – I didn’t see a single Gore or Kerry ad here in NC in the last two elections). I suspect that the lower turnouts tend to depress the “minority” party more than the “majority” party. It’s not much fun to vote when you know your candidate has no chance to win.
Thus the strong numbers for Obama in certain “blood red” states isn’t as surprising as the PVIs might suggest – the mere fact that he’s actually paying attention there will automatically boost Democratic turnout. And if there are a lot more states with high turnouts this year as a result, it will change the national averages, too, even if there had not been any real change in the partisanship of the average voter (of course, this is an example – I think there has been a real change – it just points out that this statistic may be too simplistic to merit all the weight people put on it – probably loses validity as you move to lower levels in terms of races).
It seems that the states shifting most rapidly to the left (Alaska, North Dakota etc.) are also the states that Obama is surprising people with how close he is running to McCain.