Saturday, December 08, 2012

Early voting key to Obama's reelection?

Jack Cashill, who is always putting forward interesting and controversial (in an accurate sense of the word) ideas, is one of my favorite writers, not least of all because he plugged me into the local political scene. His recent contention that early voting was a key to Obama's reelection victory is hardly an exception to that rule. He writes:
In the week before the election, I was invited to speak at my alma mater, Purdue University. As I drove around West Lafayette, what caught my eye was the absence of Obama signs. I did not see one, not even in the faculty neighborhoods. As Obama's unpopularity grew during the last four years, his team at some point decided to concede Indiana. From their perspective, Indiana lacked one tactical asset that Ohio had -- early voting. 
Early voting is a boon for the vote harvesters. The vote harvester's mission is to gather unthinking collectives of potential voters -- nursing home residents, college students, skid-row dwellers, recent immigrants -- and get them to vote. Harvesting does not necessarily mean fraud, but it clearly encourages the same. In James O'Keefe's Project Veritas videos, we saw how easy it was for even a congressman's son -- in this case, Pat Moran, son of Jim -- to cross the line from harvesting to cheating.
After reading this, I decided to try and quantify the alleged phenomenon, but immediately became skeptical (and bemused) upon finding out that Indiana did have early in-person voting. Ohio had five weeks of it to Indiana's four. Seems like a pretty marginal difference. Not potent enough to explain why Obama's share of the vote dropped over 6 points in Indiana yet less than one point in Ohio, anyway.

Still, I ran correlations at the state level on the change in Obama's share of the vote from 2008 to 2012 and the number of days before the general election that in-person early voting began. Team Obama didn't enjoy stronger performance in 2012 relative to 2008 in early voting states. To the contrary, the president performed relatively more poorly the earlier voting began, though the relationship is statistically insignificant (r = .16, p = .25).


Parenthetically, why no link to the phrase "change in Obama's share of the vote from 2008 to 2012"? So far as I can tell, it doesn't exist. Until now, that is! The following table shows the improvement (decline) in Obama's share of each state's popular vote in 2012 from 2008:

StateImp/(Dec)
1. Alaska2.92
2. New Jersey1.07
3. Mississippi0.79
4. Louisiana0.65
5. Maryland0.05
6. Rhode Island(0.16)
7. Alabama(0.38)
8. New York(0.48)
9. Arizona(0.67)
10. California(0.77)
11. South Carolina(0.81)
12. Ohio(0.83)
13. Vermont(0.89)
14. Florida(1.02)
15. Oklahoma(1.12)
16. Massachusetts(1.15)
17. Hawaii(1.30)
18. North Carolina(1.35)
19. Minnesota(1.41)
20. Maine(1.44)
21. Virginia(1.47)
22. Washington(1.49)
23. Georgia(1.51)
24. District of Columbia(1.55)
25. Iowa(1.94)
26. Arkansas(1.98)
27. New Hampshire(2.15)
28. Colorado (2.17)
29. Texas(2.30)
30. Oregon(2.51)
31. Pennsylvania(2.52)
32. Connecticut(2.53)
33. Tennessee(2.76)
34. Nevada(2.79)
35. Michigan(3.22)
36. Delaware(3.33)
37. Kentucky (3.37)
38. Wisconsin(3.44)
39. Idaho(3.47)
40. Nebraska(3.51)
41. Kansas(3.66)
42. New Mexico(3.92)
43. Illinois(4.32)
44. Wyoming(4.72)
45. South Dakota(4.88)
46. Missouri(4.91)
47. Montana (5.55)
48. North Dakota(5.92)
49. Indiana(6.02)
50. West Virginia(7.04)
51. Utah(9.66)

Fittingly (and tangentially), Utah was the reddest state and Hawaii the bluest--excluding DC, that is, which, when one really thinks about it, is even more fittingly the Democrats' best than Hawaii is.

While Jack may not be on empirically solid footing in this instance, the GOP should still heed his general counsel of "tightening the electoral process", as returns out of Pennsylvania highlighted:
It's one thing for a Democratic presidential candidate to dominate a Democratic city like Philadelphia, but check out this head-spinning figure: In 59 voting divisions in the city, Mitt Romney received not one vote. Zero. Zilch.

...

Was there not one contrarian voter in those 59 divisions, where unofficial vote tallies have President Obama outscoring Romney by a combined 19,605 to 0?

The unanimous support for Obama in these Philadelphia neighborhoods - clustered in almost exclusively black sections of West and North Philadelphia - fertilizes fears of fraud, despite little hard evidence.

6 comments:

Anonymous said...

Early and often

Jokah Macpherson said...

Thanks for compiling Obama's change in vote share by state.

The popularity of early voting is a phenomenon that baffles me. I knew a lot of people that waited in line for hours to vote early specifically so that they could "avoid the crowds"; meanwhile, I was in and out in 10 minutes on election day. There's something to be said for the robustness of large numbers of small polling locations.

Ed Tom Kowalsky said...

I can see those Philly hoodlums voting for Obama over Romney 20,000 to squat. The flash mobsters know who's greasing their chitlins.

Anonymous said...

..

Here's an HBD Dictionary:

http://www.humanbiologicaldiversity.com/#Dictionary

..

Audacious Epigone said...

Yep, it's been on my sidebar for awhile now.

DPG said...

I hadn't seen a chart with Obama's state by state change in vote share.

What sticks out to me is that only 5 states swung towards Obama, and the 2nd largest was New Jersey. (Alaska, the largest, can be explained by the absence of Palin on the ticket, rather than any positive feelings towards Obama.)

This interests me because it seems like strong evidence that Sandy had an outcome on the election. I live in New Jersey and I can't think of any other reason why Obama would have outperformed. Wall Street New Jerseyans can be compared with Connecticut, which was -2.53 points lower for Obama. South Jersey can be compared to Pennsylvania and Delaware, which were -2.52 and -3.33 lower for Obama, respectively. I wonder if there was a noticeable effect nationally on moderates' opinion of Obama due to all the coverage of him being "presidential" following the storm. Any way to test this?

Perhaps we could compare presidential favorability ratings and their local/regional/national breakdowns following other disasters. For instance, how much did Bush's ratings bounce in NY relative to the rest of the nation following 9/11?

We might use Nate Silver's model with data before the storm to estimate Obama's non-Sandy share of the vote in New Jersey.

Then, using a model of ratings bumps in local vs. national opinion, we could use the bump in New Jersey vote share to extrapolate how much of an effect it had on the national election.