Comparison of Democratic and Republican Populations and Vote Counts in Florida from 2000 to 2004 
Posted December 7, 2004 thepeoplesvoice.org 

By: Robert Millman

(Source data - Florida Board of Elections as of 12/3/04)

Methodology

The following chart tracks changes in registration and vote counts in Florida between the presidential elections of 2000 and 2004. Its purpose is to show in broad strokes the shifts in the registration and vote counts between the two major parties and to compare the registration shifts to the vote shifts--it does not measure vote counts or vote majorities. 

To show these changes, I calculated the ratio of Republicans to Democrats both in registration and voting, and then compared the ratio in 2000 to that in 2004. In calculating the registration and the vote columns, the raw numbers are reduced to an equivalent fraction of “X” over “100”. Only Republican and Democratic voter numbers were used—smaller parties and registered but undeclared voters were not included; neither were voters for third-party candidates.

Sample, using Broward County voter registration in 2000 (equation)

Broward County is rounded to 58
(58 Republicans for every 100 Democrats)

Example

All numbers entered in the Registration columns are a ratio of registered Republicans per 100 registered Democrats
All numbers entered in Vote columns are a ratio of Republican votes per 100 Democratic votes

Gain/Loss shows the difference between 2000 and 2004 in registration and vote ratios. It is a measure of increase or decrease county by county. When the 2000 number is subtracted from the 2004 number, a positive number shows an increase in the Republican registration or vote, a negative number shows an increase in the Democratic registration or vote.

Commentary

I first analyzed the Florida election using the vote counts posted at http://ustogether.org/Florida_Election.htm, and I was surprised by the pattern that emerged. Since then, the Florida Board of Elections has posted its official counts for the 2004 general election http://election.dos.state.fl.us/, and I recalculated the results, using the Florida BOE numbers. I was even more surprised by the pattern that emerged: Of the 15 Florida counties where Democratic registration increased, 13 counties lost Democratic vote share compared with 2000, one remained unchanged in proportion and in the one county where vote share improved for the Democratic candidate the vote gain was half of the registration gain.

I have heard the explanation that Republicans simply out-worked the Democrats in Florida, which is why Republicans did so well. But vote variations seem disproportionate to registration variations if you compare Group 1 and Group 2. Democrats’ vote share increased better than registration only in 6 counties where Democrats lost registration (Republicans gained); and out of the 15 counties where Democrats increased registration, they lost vote share in all but one. (And, in that one county, the Democrats picked up 22 points in registration, but picked up only 11 points in the vote.) 

This simple analysis shows that Democrats did better than registration shifts only in counties where they lost registration. I have to wonder why Democratic vote counts would run contrary to changes in registration in one group but not in another. Or, to put it more bluntly, does the Florida vote count pass the smell test? If someone can tell me how this could plausibly happen, and throughout a state where more than 7 million votes were cast, I would like to know. 

I am not a statistician and my method is perhaps unorthodox, but it is a simple way to compare changes in voting and registration. I arrived at these results using a calculator and paper and pen. I encourage anyone to check the figures.

© Copyright 12/5/04 by Robert Millman 508 Lindsay Ave., Glenville, NY 12302 Permission is granted for Print in print, email, blog, or web media if this credit is attached and the information remains unchanged. r.millman@att.net