Florida voting machines rigged?

Kathy Dopp has posted an analysis of Florida's election results by county, as compared to expected results based on voter registration data. She notices that in counties which use optical scanner machines, the Republican turnout tended to be quite large, percentage-wise, compared to the Democratic turnout, whereas in counties which use touch-screen machines, there does not seem to be an overall pattern in terms of which voters turned out in greater numbers:

So, were the optical machines rigged?

First of all, I have compiled a map of which counties in Florida use optical vs. touch-screen tabulation machines, using the CNN Election 2004 map as a template:



Then, I created a map based on the turnout for each county as compared to the expected turnout from the voter registration rolls. The exact formula is:

(Bush votes / registered Republicans) - (Kerry votes / registered Democrats)

Say, for example, that there are 100 Republicans and 160 Democrats in a fictional Norfolk county. On Election Day, 80 votes are recorded for Bush and 100 for Kerry. That's a turnout rate of 80.0% for Republicans and 62.5% for Democrats. Subtract the latter from the former and you get a result of 17.5%.

If your result is positive, you color the county that shade of red (up to 100%), or blue if negative (down to -100%). In the example case, Norfolk county would be colored a light shade of red.

Here is that map, side-by-side with the voting machine map:



Notice that those areas which turned out unexpectedly well for Kerry form a ring around the southern coast of Florida. Most of those counties use touch-screen machines! Further, there's a cluster of five counties just inland from the coast which had high Bush vote counts relative to the Republican registration rolls, and used optical machines. Does this support the notion that the optical machines were rigged for Bush?

In other words, is there a pattern where counties which use optical vote-counting machines report irregularly high turnout for Bush, as compared to neighboring counties which use touch-screen machines?

Well, let's see. There's another cluster of touch-screen machines right in the center of Florida. Did they lean more towards Kerry than their neighboring counties? Looking at the map, there is no evidence that that is the case. Finally, Nassau county in the northeast corner of Florida uses a touch-screen machine, yet it reported a higher Republican-Democrat difference in turnout than the counties directly below it, which all use optical machines.

So so far, it seems that there just happen to be a bunch of counties on the southern tip which shifted unexpectedly to Kerry, and also happened to have touch-screen voting machines. Yes, that is a coincidence, but there's no larger pattern regarding the machines themselves, nor are there enough counties with touch-screen voting machines to detect a meaningful pattern even if there were one.

But what explains the extremely high voter turnout for Republicans in the northern counties, and in that county cluster inland from the southern coast? Actually, the turnout wasn't that high. If you base the map on total votes minus expected votes, rather than divided by, you get a wholly different picture:



So you see, there are only three counties which contributed substantially to Bush's victory in Florida, as compared to the expectations based on the voter rolls: Duval county in the northeast, which contains the densely-populated city of Jacksonville; and the Polk and Hillsborough counties in the center of Florida, one of which even used a touch-screen voting system! Two out of three using optical machines is not statistically meaningful.

In other words, all the northern counties colored deep red by the first map are very sparsely populated, and in fact many of them have such a low number of registered Republicans as compared to registered Democrats, that a shift in a couple hundred votes could cause them to be colored deep red even if the county itself had more people vote for Kerry than for Bush, since a couple hundred votes would constitute over 200% turnout for Bush as compared to the number of registered Republicans.

Also, many Floridians are old, and chose their party registration decades ago. There are certainly a large number of socially conservative people who registered as a Democrat years ago, but now find themselves voting Republican. Sort of the "Zel Miller" type of Democrat, or "Dixiecrat" as they're sometimes called (in fact, Dixie is the name of one of the Florida counties exhibiting that very result).

Notice also that, barring the five southern exceptions, every county which is colored deep red on the first map is located in the northern section of Florida. Also note that there is only one county in that northern section which actually uses a touch-screen machine, and it does not report any larger Republican-Democratic turnout ratio than its neighboring counties. Therefore, the entire anomaly is geographically-based. If you removed those northern counties, the anomaly would essentially disappear. And since there are basically no touch-screen machines in northern Florida, you have no way to actually control for geography.

To conclude, there is simply no pattern here that remains when you control for geographic location. Kathy does note that:


However, doing a simple geographic map of the counties, as I did, should be the first thing one should think of when you're considering the possibility of voter fraud, before you say:


Certainly Kathy made no explicitly false claims, and there's nothing wrong with asking questions, but I can only guess at what causes someone to be so curious as to be stumbling around in the dark looking for conspiracies like these. I attribute it to the irrational "Bush stole the election" mentality which has permeated the left since 2000.

It's sad, too, because when real fraud happens, how are well-meaning people supposed to know? We're so used to tuning out the conspiracy theories.