Jasjeet Sekhon sent me the following note regarding analyses of votes for Bush and Kerry in counties in Florida:

Hi Sam and Andrew,
I just saw your blog entries on the e-voting controversy.
A week ago I posted a short research note about the optical voting
machine vs. DRE issue in Florida. It is entitled “The 2004 Florida
Optical Voting Machine Controversy: A Causal Analysis Using Matching”.
In this note, I try to obtain balance on all of the baseline variables
I can find, and the results give NO support to the conjecture that
optical voting machines resulted in fewer Kerry votes than the DREs
would have. Of course, one really needs precinct-level data to make
inferences to many of the counties in the state.

Also of interest to you may be a research note by Jonathan Wand.
He has also obtained a ZERO effect for optical
machines by using Walter’s and mine robust estimator. In that
analysis, ALL of the counties are used. Wand introduces a key new
variable: he uses campaign finance contributions as a covariate. But
has he notes the linearity assumption is dubious with this dataset.

Cheers,
Jas.

My comments on these:

First, it’s funny that they are testing whether Kerry was disadvantaged by optical-scan machines, since I thought people were worried about electronic voting machines (as I discussed in my post last week). But the statistical analysis shouldn’t depend too much on what you’re looking for in any case.

The analyses seem reasonable. Jas’s paper focuses on the use of matching to identify comparable counties for comparison. This is sort of like what we do visually when looking at Bruce Shaw’s scatterplot but is more general in that you can match on many variables as well. In this particular example, I don’t know if matching is so necessary, but it’s a good general tool and described clearly in his paper.

Wand’s paper is cool too, although I don’t like that he presented counties in alphabetical order in his Figure 1. Any meaningful order (e.g., population, or support for Bush in 2000, or whatever) would be preferable. Also, Wand uses an overdispesed model for robust multinomial data. This is fine, but realistically you can treat counts of this magnitude as continuous, so this might be methodological overkill. Can’t hurt though, I guess.