Something happened in Broward county. Maybe it was a massive shift in voting patterns, or maybe it was fraud and/or flawed tabulation, but something happened.
I have experimented with various predictive equations in order to graph predicted results against actual results, aiming for the best fit. I factor in registration patterns in 2000 (including third party and "no affiliation" registrations), voting turnout and results in 2000, registration patterns in 2004, turnout in 2004, and decreased votes received by third party candidates in 2004.
My most recent approach predicts the actual results for most smaller counties fairly accurately. That is, the 2004 results in most small counties are consistent with results from 2000.
Larger counties seem to vary more because of large numbers of third party and "no affiliation" registrations. The equation I'm using for these counties is slightly different, and predictions are fairly accurate, but Broward, Palm Beach, and Pinellas still stick out as exceptions apart from the rest. Broward in particular is "way off the map."
First, the punch-line, then I’ll explain the calculation, which admittedly has many assumption built into it.
Here are the predicted and actual increases in vote totals from 2000 to 2004 for Broward county:
Predicted Increase from Gore Total to Kerry Total: 111,030
Predicted Increase from Bush2000 to Bush2004: 19,424
Net Gain: Kerry 91,606
Actual Increase from Gore Total to Kerry Total: 64,657
Actual Increase from Bush2000 to Bush2004: 65,797
Net Gain: Bush 1,140
Bush’s actual increase was 65,797. This would be as if the entire decrease in votes for third party candidates (1,221), and the entire projected turnout for newly registered third party registrants (66.57% * 76,211 = 50,734), went to Bush, and that’s only 51,955.
Here are the calculations for Broward county:
Start with voting and registration patterns in 2000
2000 Registration
Democrat: 456,789, 51.5%
Republican: 266,829, 30.1%
Other: 164,146, 18.5%
Total: 887,764
2000 Voting
Gore: 387,703, 67.4%
Bush: 177,902, 30.9%
Other: 9,538, 1.7%
Total: 575,143
Turnout: 64.79%
Consider the net gains from relative turnout and/or “cross over” votes (voters who voted differently than their registration). For Gore, for example, it would be calculated like this: (Votes for Gore) - (%Turnout * Registered Democrats)
Gore Net Gains: 91,770
Bush Net Gains: 5,035
Now consider each candidate’s net gains as a percentage of total net gains: for example, (Gore Net Gains) / (Gore Net Gains + Bush Net Gains)
% Net Gains Gore: 94.80%
% Net Gains Bush: 5.20%
Now look at the registration patterns for 2004
Democrat: 533,976 (up 77,187 from 2000), 50.5%
Republican: 283,736 (up 16,907 from 2000), 26.8%
Other: 240,357 (up 76,211 from 2000), 22.7%
Let’s say we know the %Turnout for 2004 and the number of votes received by third party candidates:
Turnout: 66.57%
Votes Other: 8,317 (down 1,221 from 2000)
Assuming that Bush and Kerry benefit equally from decreased votes for third party candidates, and assuming that the % Net Gains for each candidate are the same as what they were in 2000, we would predict the following:
Predicted Net Gains Kerry: 143,256
Predicted Net Gains Bush: 8,437
Predicted Total Kerry: 498,733, 70.8%
Predicted Total Bush: 197,326, 28.0%
Predicted Increase from Gore Total to Kerry Total: 111,030
Predicted Increase from Bush2000 to Bush2004: 19,424
Notice that Kerry’s predicted total is 70.8% of total votes, whereas Gore’s was 67.4%. This is primarily because turnout increased 1.7% in a county where 50.5% of registered voters are registered Democrat (down 1% from 2000), whereas only 26.8% are registered Republican (down 3.3% from 2000). The secondary reason Kerry’s percentage is higher than Gore’s is that votes for third party candidates decreased.
Now look at the actual results from 2004:
Net Gains Kerry: 96,882
Net Gains Bush: 54,811
Total Kerry: 452,360, 64.2%
Total Bush: 243,699, 34.6%
Increase from Gore Total to Kerry Total: 64,657
Increase from Bush2000 to Bush2004: 65,797
Notice that Kerry’s percentage of the total vote is only 64.2%, down 3.2% from Gore’s 67.4%.
Compare the Net Gains and % Net Gains from 2000 and 2004
Gore Net Gains: 91,770
Bush Net Gains: 5,035
% Net Gains Gore: 94.80%
% Net Gains Bush: 5.20%
Net Gains Kerry: 96,882
Net Gains Bush: 54,811
% Net Gains Kerry: 63.87%
% Net Gains Bush: 36.13%
Compare the predicted and actual increases in vote totals from 2000 to 2004
Predicted Increase from Gore Total to Kerry Total: 111,030
Predicted Increase from Bush2000 to Bush2004: 19,424
Net Gain: Kerry 91,606
Actual Increase from Gore Total to Kerry Total: 64,657
Actual Increase from Bush2000 to Bush2004: 65,797
Net Gain: Bush 1,140
Bush’s actual increase was 65,797. This would be as if the entire decrease in votes for third party candidates (1,221), and the entire projected turnout for newly registered third party registrants (66.57% * 76,211 = 50,734), went to Bush, and that’s only 51,955.
Palm Beach and Pinellas counties also depart significantly from predictions using the same method. Broward, Palm Beach, and Pinellas have very high numbers of registered voters (above 500,000) and high percentages of third party and "no affiliation" registrations (above 22.0%). However, Miami Dade, Hillsborough, and Orange have the same features and depart from predictions much less.
Broward: 1,058,000 registered, 22.7% in third parties
Palm Beach: 730,000 registered, 22.9% in third parties
Pinellas: 591,000 registered, 23.0% in third parties
Miami Dade:1,058,000 registered, 22.4% in third parties
Hillsborough: 621,000 registered, 23.3% in third parties
Orange: 532,000 registered, 24.7% in third parties
Interestingly, Hillsborough is the only county of these six in which Bush’s predicted increase in from 2000 is greater than Kerry’s predicted increase, and the actual results match the predicted results almost exactly.
Brevard and Hernando are medium-sized counties that seem to stick out most as exceptions. Both counties have relatively high % of third party registrations. Both counties have fewer registered Democrats than registered Republicans, but Hernando is close to even. Hernando departs a bit further than Brevard from predicted results. Given Hernando’s smaller size, and given that Hernando switched from Gore in 2000 to Bush in 2004, Hernando seems to be more exceptional.
But let's start with Broward. What happened in Broward county? Anyone?
