|
|
|
The first Cornell et al. critique of our analyses of the Florida data (expected vote based on registration vs. the actual vote http://UScountvotes.org) neglected to exclude the smaller counties in their analyses ( A Princeton critique also made the same mistake.) The explanations of Mebane et al. of the causes for this high correlation between voting machine type and election results may explain some of the phenomenon, but does not explain the phenomenon in all counties.
The reason that we statistically analyzed and compared only the medium-sized counties against each other is that everyone knows that county population size influences voting behavior, and that small counties in rural areas tend to vote Republican. As we stated well prior to the Mebane et al. critique, "the entire two groups of counties (E-touch and Op-scan users in Florida) cannot be validly compared, as county-size itself might account for the data". Here is a study of how population size affects cross-party voting.
Contrary to what our critics claimed, the mid-sized counties we included in our analyses are not "primarily in the Panhandle", as in the cases mentioned by Mebane et al. See our map of these counties.
For the 26 mid-sized counties with between 80,000 and 500,000 registered voters that we studied, the type of machine used was not significantly related to the number of registered voters in the county. Eight of these counties used E-touch machines, and 18 used Op-scan machines. There was no significant difference between these two groups of counties in either their numbers of registered voters or their proportion of registered Republicans to registered Democrats. Counties using E-touch machines showed significantly positive percent changes in vote for both Republican and Democrat candidates, with greater mean percent changes for the Democrat. However counties using Op-scan machines showed significant positive percent change only for the Republican candidate, the mean change for the Democrat being insignificantly greater than zero. The probability of this happening by chance would be less than one in one thousand. (p < .001)
The magnitude of the apparent effect of voting machine type on voter behaviour would seem to warrant further investigation based on our work.
We also noted that the FL 2000 results also have the same pattern as the FL 2004 results and we put the 2000 numbers out there so people could see them. Our group does not make the assumption that the FL 2000 election is necessarily a base model for measuring an honest, accurate election. We plan to go farther into historical analyses and also to compare Presidential race results with US Congressional and Senate race results and other measures over the next two years, to pinpoint and provide statistical evidence for those who would like to investigate further.
Here are two excellent writeups of our results by Elizabeth Liddle and Josh Mitteldorf and a terrific visual representation by Charlie Strauss. Here are responses by Elizabeth Liddle and Marc Sapir to the Critiques of Our work to develop methods to pinpoint areas with possible vote counting errors. and here is Josh Mitteldorf's and David Dodge's responses to the MITCalTech attempt to refute our work and another critique of another survey vote.CalTech.edu did on E-voting.
Here is the Mebane et al. response to this response
The Berkeley study, which drew the conclusion that votes had been miscounted in counties using touchscreen voting machines, also had the same flaw of being done at the county level. Here is another look at the Berkeley study.
After back and forth discussions of these analyses, statisticians agree that studies of voting data at the precinct level are needed before conclusions can be made to detect which voting precincts are likely to have miscounted votes.
So USCountVotes.org is forming a nonprofit to accomplish this Mission.
Kathy Dopp