mb -
Please provide any appropriate constructive criticism and/or comments on the following.
I finally managed to get all the Florida Presidential and Senatorial vote totals (pre-recounts, but not important)for Bush, Gore, McCollum and Nelson in a spreadsheet. All third parties were excluded.
All counties were separately ranked 1 to 67 by the Bush to McCollum and Gore to Nelson vote ratios.
The counties were then effectively grouped into quartiles depending on these ratios. This was done separately for Bush/McCollum and Gore/Nelson.
For Bush, the approximate quartile thresholds for Bush/McCollum were :
Quartile I -II boundary = 1.065
Quartile II-III boundary = 1.104
Quartile III-IV boundary = 1.200
For Gore/Nelson :
Quartile I -II boundary = 0.815
Quartile II-III boundary = 0.909
Quartile III-IV boundary = 0.979
The Bush/McCollum Quartile II-III boundary for example means that 1.104 is the ratio of Bush to McCollum voters in the county that has half of the counties both above and below it when ranked on that ratio.
The lower ratios for Gore/Nelson are presumably primarily due to Nader voters also voting for Nelson, but this was of no interest.
In an attempt to judge either missing or excess Bush or Gore votes county by county, using the same party Senatorial vote as a reference, the following arbitrary procedure was used.
For counties in either of the extreme quartiles (I and IV) add or subtract just enough Bush or Gore votes to bring those counties to the edge of the adjacent quartile.
For example, all of the Bush counties in the first quartile would have enough Bush votes added to bring those counties up to a Bush/McCollum ratio of 1.065. Those counties in the fourth quartile would have enough Bush votes subtracted to bring those counties down to a Bush/McCollum ratio of 1.2.
Separately, Gore votes were also adjusted.
When finished, approximately 25% of the counties will have a number of added Bush votes, and a different 25% will have a number of subtracted Bush votes. The same will be true for Gore, but the list of counties will be independent of the Bush counties.
The added and subtracted Bush and Gore votes are then added up for a state total. However, due to several factors, some of these numbers are more useful than others, and it is the individual county figures that are most likely to have significance.
Overall State Results -
Bush missing votes - 50K
Bush excess votes - 10K
Gore missing votes - 2.8K
Gore excess votes - 84K
County components of above -
Bush missing -
Brevard - 4.5K
Broward - 9.1K
Charlotte - 3.8K
Collier - 4.0K
Duval - 3.4K
Lee - 8.5K
Palm Beach - 11.6K
Sarasota - 2.8K
Bush Excess -
Highlands - 4.2K
Holmes - 1.7K
Gore Missing -
All counties less than 400 votes missing
Gore Excess -
Broward - 17.8K
Collier - 2.4K
Dade - 32.6K
Highlands - 3.4K
Lee - 6.0K
Martin - 6.4K
Orange - 2.2K
Palm Beach - 4.7K
Saratota - 2.9K
Discussion -
The calculation of the excess votes, i.e. votes for President in excess of votes for Senator, is pretty useless because we know that a large number of voters will simply vote for president, and not for Senator. This is easily seen in the large excess of Gore votes in Broward and Dade counties which make up most of the state total.
It is the missing votes that CAN have more significance. If we look at Palm Beach county, Gore is shown with an excess of 4.7K, but this is probably just a case of the single vote Democratic Presidential voters, as in Broward and Dade, overpowering possible missing Gore votes. So whatever missing Gore votes there are here cannot be seen without trying to estimate how many single Gore voters there were in the county.
The missing Bush votes in Palm Beach are more interesting at 11.6K. The ratio of Bush to McCollum votes in this county was 0.989, indicating that Bush was actually less popular than the losing Republican incumbent Senator who was, I believe, heavily involved in the impeachment proceedings, among those who voted for McCollum.
Additionally, we have to look how the Bush/McCollum ratio of 0.989 for Palm Beach county fits in the distribution of all 67 counties for this measure. It in fact ranks third, meaning that fully 64 of the 67 counties have a higher ratio of Bush to McCollum votes. Also from the quartile boundaries above, half the counties have Bush vote totals more than 10% above the McCollum totals, and 75% of the counties have Bush vote totals above more than 6.5% above the McCollum totals. Yet, in Palm Beach the Bush vote total is 1.1% BELOW the McCollum vote total.
Remember that these missing Bush votes are derived ONLY from ballots that have successful votes for Senator McCollum. These are NOT likely to be confused voters. I would be interested in any reasonable explanation for this result that doesn't include ballots that have tampered with by the addition of an extra Presidential punch, for Gore for example.
Thanks, Don |
