Did Obama Do Better in States with Bad Economies?
A few weeks ago, I noted a post by Jim Gimpel, who found that the economic climate in the battleground states was worse than in other states.
With some help from Jim in sourcing data, I’ve delved a little deeper. Consider what follows suggestive, rather than conclusive. For each state, I subtracted Kerry’s vote share from Obama’s, to get a sense of how much better or worse Obama did. I plot that against several economic indicators:
- The percent unemployed, as of October 2008.
- The raw change in that percent, from December 2007 until October 2008.
- The percent change in that percent, over the same time period.
- The number of home foreclosures per 10,000 households, as of October 2008.
- The percent change in that number, from October 2007 to October 2008.
Here’s the first plot. The rest are below the fold.
The upshot here: the relationships are modest at best, and even the stronger relationships may derive from Hawaii’s outlier status. Simple regression models with one economic indicator and a dummy variable for Hawaii reveal no statistically significant relationships.
Further evidence of a uniform swing?
Comments
John,
If your DV is Obama-Kerry, shouldn’t there be some form of control for what the state looked like in 2004? You do the 07-08 measure, but I wonder what a longer differential would do. (With foreclosures there’s not much point to doing that, as those are so huge in 2008 compared to anything since the Depression that differences are essentially the same thing)
I’m inclined to believe the uniform swing theory because priming on the economy comes from national media sources, but I’d like to put the differential story to bed.
Posted by: Matt Jarvis | January 30, 2009 01:41 PM
The errors on that fit are enormous. I wager that the errors on the slope of that line include both positive and negative correlations. This plot shows that there is no correlation.
Posted by: Chris | January 30, 2009 04:13 PM
You know, John, if you scrapped all your graphs but the first one you could probably get The Times or Post to publish an op-ed where you explain the deliciously counterintuitive finding that the bad economy actually helped McCain, and Obama won despite the economic crisis.
Posted by: Alex F | January 30, 2009 07:29 PM
Matt: A longer differential is worth testing. I used shorter ones b/c of the evidence that it’s only the most recent changes in economic indicators that influence election outcomes.
Chris: I’m not sure what plot you’re referring to. Your conclusion echoes the point of the post.
Alex F: An excellent idea. Look for it next week, under the title “Voodoo Obamanomics” or some such.
Posted by: John Sides | January 30, 2009 09:26 PM
I agree with Matt, with a modification.
(1) Calculate the recent change, as you’ve done.
(2) Calculate the same change for Dec 03-Oct 04.
(3) On X, use (1) minus (2).
If Y is the difference in Obama’s margin relative to Kerry’s margin, then X has to be the difference between the situation Obama faced and the situation Kerry faced.
Posted by: Adam | February 2, 2009 11:43 AM