Analysis of Additional Votes Cast in 2004 over 2000 by Florida County

[bj2110]

Bear with me while I explain the procedure before I get to the eye-opening results. I'll apologize for the length up front. Sorry. Data is included at the end of the post in comma delimited form.

METHODS/ASSUMPTIONS:

1) Numbers are generated for new or available voters to Bush and Kerry in the 2004 election when compared to votes cast for Bush and Gore in 2000.


2) New or available voters are assumed to be what's left over after these assumptions:
a) All Bush voters in 2000 voted Bush in 2004
b) All Gore voters in 2000 voted Kerry in 2000
c) All other candidate voters in 2004 did not vote for Bush or Gore in 2000
d) a) & b) are likely false in and of themselves, however, these assumptions also hold true as long as an equal number of 2000 Bush voters switched to Kerry as Gore voters switched to Bush. As you will see, even this becomes a moot point based on the orders of magnitudes of these numbers.


3) The required margin of votes in the new/available voter data set is determined by comparing 2000 results with 2004 results, both total and by candidate.

For example, out of the 1000 people in a theoretical 2004 county, let's assume 550 voted Bush, 400 voted Kerry, and 50 voted for another candidate. In 2000, let's say the total voters were 900, and the count went 450 Bush, 380 Gore, 70 Other. In this scenario there are 100 extra voters + 20 fewer voters who picked neither Rep or Dem. So, we have 120 extra voters for Reps & Dems. In order for the 2004 results to be accurate, the new/available voters had to vote 100 Bush to 20 Kerry.


4) The vote percentages for the data set consisting of new/available voters is compared to the total data set for the county. In the previous example, the new/available voters voted 100/120 (83%) for Bush. But the county as a whole voted 55% Bush. I'm calling this a 28% variance between total votes and new/available votes. This would mean that over 12% of the total data set (120 votes out of 1000), there was a 28% difference.


5) These numbers are broken down by the average precinct size to illustrate how just a few switched votes at the precinct level can both make a huge difference, and can be rooted out in numbers.


RESULTS:

New/available voters represent anywhere from 12% - 31% of the total vote. This means that our sample size is significant.

60 of the 67 Florida counties exhibited a variance in favor of Bush. This may not be surprising by itself, since Bush supposedly increased his margin of victory in the state. But we have to consider that this variance is based on the final 2004 vote totals, meaning this increased margin is already taken into consideration.

5 counties (Holmes, Hardee, Union, Calhoun, and Baker) show more increased Bush votes than available voters. The only possible explanation of this is that not only did Bush receive every single one of the votes from people who didn't vote in 2000, he got every reduced third party vote, and then got more Gore voters to switch to him than Kerry got to switch from Bush. For example, Holmes County voted 77.3% Bush in 2004 over 8,298 votes. There were an 1,032 more votes available to Bush and Kerry than to Bush & Gore. Bush received 1,399 more votes in 2004 vs. 2000, or 136% of the new/available vote. This is a 58% variance on 12.4% of the data set. Un-freakin-believable.

39 counties have Bush-favored variances over 10%, 17 are over 20%, 7 are over 30%, and 4 are over 40%.

These variances are not isolated to only counties that Bush won. Bush-favored variances exist in 6 counties that Kerry took (Kerry only took 11). Kerry favored variances only exist in 7 counties, 6 of which have variances under 2.5%, the other is 6.5%.

Here's a big observation!: 23 of the 24 counties with the highest Bush variances used Optical Scan vote machines, including the top 19! Of the other 43 counties, 14 were touchscreen, 29 were optical scan. Considering there were only 15 counties total that used touchscreen, the top heavy overload of opscan counties is significant.

But, the most intriguing observation of all, IMO? Consider the average precinct size in this analysis, which was 971. The average county precinct size ranged from 1795 in Okaloosa to 302 in Liberty. The average new/available voters per precinct ranged from around 60 to 350. Based on the Bush margin, these new/available voters would be expected to vote at the same percentage as the total for the particular county, an average of about a 135-90 margin (county-county average). However, there was an average 29 vote variance in this tally, meaning if you expected a 135-90 margin based total percentage, we are actually getting an average of 164-61 in the new/available votes. The avg. precinct variances in favor of Bush ran from 1 vote up to 71 votes, with a mean of 29 and a median of 26. These are fairly consistent variances at a precinct level, don't you think?

CONCLUSIONS: (As much as someone like me can conclude)

The variance between the 2000 vote and the 2004 vote just doesn't seem possible. Theories that have been floating around lately, such as the vulnerability of the op-scan tabulators and the huge discrepancy between party registrations and actual vote results in many "dixiecrat" counties, are only further supported by this data.

Only 20 - 40 votes being switched from Kerry to Bush at a precinct level generates the sort of "landslide" victory that we saw in this election.

We have to investigate at a precinct level in order to narrow down the search to find deliberate intent to alter this election. There have to be precincts out there that have virtual mathematically impossible results when compared to the 2000 election.


NEXT STEPS:

Find and document probabilities that such a high portion of the total data set (20% - 30%) can be 20%, 30%, even 50% off from the total. Get these probabilities out. Think Olbermann.

Evaluate the assumptions that were made at the beginning to see if invalid assumptions could lead to any wrong conclusions.

Support Bev Harris and the people who are actually looking for proveable and not circumstantial (such as this) evidence that this election was fraudulent.

Thanks for listening!


Data Below, Comma Delimited

,,,,,,,,,,
County,# of Precincts,Total Votes,Total Vote Bush %,New/Available Votes,Expected Bush to Kerry Tally per Precinct for New/Available Voters,Change in Expected Bush-Kerry Tally per Precinct for New/Available Voters,Bush to Kerry Tally per Precinct for New/Available Voters,Available Voter Bush %,Variance Between Total Vote & New Available Vote,Data Sample Size % New/Available Votes to Total Votes Ratio
holmes,18,"8,298",77.25%,"1,032",44 to 13,33,78 to -20,135.56%,58.31%,12.4%
hardee,12,"7,246",69.65%,"1,091",63 to 28,44,107 to -16,117.51%,47.85%,15.1%
union,11,"4,675",72.64%,908,60 to 23,37,97 to -14,117.18%,44.54%,19.4%
calhoun,15,"5,961",63.41%,868,37 to 21,24,60 to -3,104.49%,41.08%,14.6%
putnam,55,"30,964",59.12%,"5,165",56 to 38,33,88 to 6,94.06%,34.94%,16.7%
gulf,16,"7,256",66.07%,"1,245",51 to 26,26,78 to 0,99.92%,33.85%,17.2%
baker,9,"9,955",77.73%,"1,916",165 to 47,71,236 to -24,111.06%,33.33%,19.2%
liberty,10,"3,021",63.79%,663,42 to 24,19,61 to 5,92.01%,28.22%,21.9%
hernando,56,"75,832",52.93%,"11,853",112 to 100,57,169 to 42,80.07%,27.14%,15.6%
washington,21,"10,365",71.09%,"2,488",84 to 34,29,113 to 5,95.42%,24.33%,24.0%
dixie,11,"6,440",68.84%,"1,869",117 to 53,41,158 to 12,92.88%,24.05%,29.0%
gilchrist,10,"7,012",70.37%,"1,740",122 to 52,41,163 to 11,93.91%,23.54%,24.8%
bradford,22,"10,851",69.61%,"2,308",73 to 32,24,97 to 8,92.68%,23.07%,21.3%
levy,20,"16,649",62.51%,"4,225",132 to 79,45,178 to 34,84.02%,21.51%,25.4%
highlands,27,"41,491",62.36%,"6,847",158 to 95,52,210 to 44,82.78%,20.42%,16.5%
okeechobee,18,"12,197",57.28%,"2,493",79 to 59,28,107 to 31,77.42%,20.13%,20.4%
jackson,28,"19,797",61.20%,"3,661",80 to 51,26,106 to 24,81.32%,20.12%,18.5%
lafayette,7,"3,325",73.98%,846,89 to 31,23,113 to 8,93.38%,19.40%,25.4%
columbia,25,"24,984",67.05%,"6,771",182 to 89,50,232 to 39,85.50%,18.44%,27.1%
sumter,42,"31,837",62.18%,"9,615",142 to 87,40,183 to 46,79.75%,17.57%,30.2%
walton,32,"23,974",73.22%,"5,942",136 to 50,32,168 to 18,90.39%,17.17%,24.8%
suwannee,19,"15,801",70.58%,"3,593",133 to 56,32,166 to 23,87.59%,17.00%,22.7%
brevard,221,"265,075",57.66%,"50,488",132 to 97,39,170 to 58,74.58%,16.92%,19.0%
polk,161,"210,777",58.61%,"44,020",160 to 113,46,206 to 67,75.50%,16.90%,20.9%
broward,"1,032","704,376",34.60%,"130,454",44 to 83,20,64 to 63,50.44%,15.84%,18.5%
bay,59,"74,998",71.19%,"16,959",205 to 83,45,250 to 37,86.98%,15.79%,22.6%
franklin,10,"5,930",58.55%,"1,372",80 to 57,21,102 to 35,74.20%,15.65%,23.1%
pasco,172,"190,866",54.07%,"49,783",156 to 133,45,201 to 88,69.53%,15.47%,26.1%
citrus,41,"69,462",56.86%,"13,480",187 to 142,50,237 to 91,72.19%,15.33%,19.4%
taylor,14,"8,580",63.71%,"1,810",82 to 47,18,101 to 29,77.90%,14.19%,21.1%
volusia,179,"228,358",48.85%,"47,202",129 to 135,34,163 to 101,61.83%,12.99%,20.7%
pinellas,380,"455,203",49.57%,"65,539",85 to 87,22,107 to 65,62.26%,12.69%,14.4%
santa rosa,43,"67,213",77.35%,"17,549",316 to 92,50,365 to 43,89.54%,12.20%,26.1%
palm beach,695,"542,835",39.03%,"116,909",66 to 103,19,85 to 83,50.42%,11.38%,21.5%
wakulla,12,"11,763",57.61%,"3,323",160 to 117,29,189 to 88,68.16%,10.55%,28.2%
charlotte,83,"79,730",55.69%,"13,558",91 to 72,17,108 to 55,66.20%,10.51%,17.0%
okaloosa,52,"89,707",77.65%,"19,971",298 to 86,40,338 to 46,87.93%,10.29%,22.3%
martin,50,"72,430",57.09%,"10,957",125 to 94,22,148 to 72,67.35%,10.26%,15.1%
glades,13,"4,188",58.33%,878,39 to 28,7,46 to 21,68.56%,10.23%,21.0%
desoto,15,"9,495",58.05%,"1,846",71 to 52,12,84 to 39,68.04%,9.99%,19.4%
nassau,24,"32,664",72.65%,"8,921",270 to 102,35,305 to 66,82.14%,9.49%,27.3%
osceola,93,"82,178",52.46%,"27,332",154 to 140,28,182 to 112,61.82%,9.36%,33.3%
manatee,136,"143,539",56.62%,"35,371",147 to 113,24,171 to 89,65.93%,9.31%,24.6%
marion,143,"139,644",58.20%,"38,716",158 to 113,25,183 to 88,67.48%,9.29%,27.7%
flagler,34,"38,475",51.03%,"11,696",176 to 168,31,206 to 138,60.01%,8.99%,30.4%
escambia,91,"142,990",65.30%,"27,650",198 to 105,25,224 to 80,73.60%,8.30%,19.3%
st. lucie,75,"99,913",47.56%,"23,015",146 to 161,25,171 to 136,55.68%,8.12%,23.0%
lake,103,"123,938",60.02%,"36,017",210 to 140,27,237 to 113,67.67%,7.65%,29.1%
seminole,126,"186,115",58.10%,"50,213",232 to 167,26,258 to 141,64.62%,6.53%,27.0%
clay,64,"81,230",76.18%,"24,418",291 to 91,24,315 to 67,82.50%,6.32%,30.1%
indian river,57,"61,321",60.12%,"12,398",131 to 87,14,144 to 73,66.41%,6.29%,20.2%
hillsborough,360,"461,520",53.00%,"107,702",159 to 141,19,177 to 122,59.28%,6.28%,23.3%
st. johns,68,"86,262",68.60%,"26,519",268 to 122,21,289 to 101,74.03%,5.43%,30.7%
lee,225,"241,433",59.82%,"59,078",157 to 106,13,170 to 92,64.78%,4.97%,24.5%
sarasota,167,"195,530",53.51%,"37,059",119 to 103,10,129 to 93,58.10%,4.59%,19.0%
alachua,69,"111,022",42.89%,"28,474",177 to 236,19,196 to 217,47.38%,4.49%,25.6%
orange,263,"387,752",49.62%,"110,870",209 to 212,11,220 to 202,52.20%,2.58%,28.6%
monroe,30,"39,525",49.24%,"6,569",108 to 111,6,113 to 106,51.80%,2.56%,16.6%
madison,11,"8,307",50.51%,"2,192",101 to 99,5,105 to 94,52.83%,2.32%,26.4%
miami-dade,872,"768,553",46.66%,"146,371",78 to 90,1,79 to 89,47.20%,0.53%,19.0%
hamilton,10,"5,079",54.97%,"1,184",65 to 53,0,65 to 54,54.56%,-0.41%,23.3%
hendry,26,"9,774",58.89%,"1,729",39 to 27,0,39 to 28,58.36%,-0.53%,17.7%
duval,285,"379,614",57.76%,"117,410",238 to 174,-2,236 to 176,57.20%,-0.56%,30.9%
jefferson,15,"7,477",44.11%,"1,913",56 to 71,-2,55 to 73,42.86%,-1.24%,25.6%
leon,116,"136,314",37.85%,"34,935",114 to 187,-6,108 to 193,35.87%,-1.98%,25.6%
collier,96,"128,352",65.04%,"36,827",250 to 134,-10,240 to 144,62.55%,-2.49%,28.7%
gadsden,27,"20,984",29.80%,"6,380",70 to 166,-15,55 to 181,23.29%,-6.51%,30.4%

[badc0der]

1. flawed assumptions

The assumptions that all Gore voters voted for Kerry and that gore voters (many of whom likely felt disenfranchised after 2000) showed up this year in the same numbers as Bush voters is attractive, but IMO is too much of a leap. If those two things had happened I do not believe that any amount of fraud on the GOP's part would have put FL in their column.

I think better metrics are:

provisional votes cast/votes cast
Should be (relatively) constant

correlation between absentee and in person votes
Should be very high and favor republicans by a couple of points

correlation between votes for minor races and votes for president
It is difficult to change totals without causing this to diverge significantly from what is predicted by exit polls.

[jsamuel]

2. Every single poll (exit or otherwise) showed Gore voters went 7% for Bush

BUT 11% BUSH VOTERS WENT KERRY!!!

Where did BUSH's EXTRA 11 MILLION VOTES COME FROM?

[badc0der]

3. something I saw

Had the breakdown of the "who did you vote for in 2000?" exit poll question as something like:
37/40/3/17 gore/bush/nader/none
but I can't seem to find it atm.

Meaning that a lot of gore voters stayed home and the new voter essentially decided it.

I don't know what the growth patterns in FL are exactly but Cubans and retirees (two segments I believe are growing in Florida) typically swing republican so on its face a Florida victory seems plausible. Of course this is based on 2000 results and exit polls both of which are questionable.

[ewulf]

5. Retirees swing Republican?

For real? I guess they just LOVE the idea of privatizing social security, and that Medicare bill passed last year was just Great as far as their concerned?

Hm...

[Eloriel]

9. But here's some data re the FL Latino vote --

Bush wins by 5% in Florida yet his Latino vote in Florida was down 15%!!!
http://www.democraticunderground.com/discuss/duboard.php?az=view_all&address=203x42375
Link: http://www.pww.org/article/articleview/6081/1/240

See these threads for as complete a record of discussions here as I've been capable of pulling together myself:

VOTE FRAUD Links - a DU Compendium
http://www.democraticunderground.com/discuss/duboard.php?az=show_topic&forum=201&topic_id=1984#

VOTE FRAUD Links Compendium - Thread #2
http://www.democraticunderground.com/discuss/duboard.php?az=view_all&address=201x3223

ALSO SEE: VOTE FRAUD? What can we do? http://www.democraticunderground.com/discuss/duboard.php?az=show_topic&forum=104&topic_id=2701028#

[ryesteve]

10. "who did you vote for in 2000?" exit poll

Actually, it was 37% Gore, 43% Bush... which makes me very suspicious that the exit poll data we're seeing now is oversampling Republicans. Given that Gore won the popular vote, but there's a 6% spread in this exit poll answer, I find it very hard to believe that so many people who were motivated enough to make it to the polls to vote against Bush in 2000 didn't feel the need to do the same in 2004.

[MrUnderhill]

11. Nah... you would see the same thing today... it's the "bandwagon" effect

Ask people who they are rooting for to win the Super Bowl and you'll get one answer. Ask them the day AFTER the game who they rooted for and you'll see a measurable shift toward the victor.

It's the same thing in politics. If you ask people TODAY if they voted and who they voted for, you're likely to see a 6-8% victory for Bush (when he supposedly won by 3%). That's just the way polling is.

[ewulf]

4. Good question!

Maybe new voters? Bj2110's analysis does have some merit when one factors in exit poll data.

[cdp]

8. CNN and USATODAY

both had about 10% of 2000 Gore voters going for Bush
and 10% of 2000 Bush voters going for Kerry

[bj2110]

7. That analysis has been done for NC

and the same results were seen, although they weren't as dramatic as what was seen here. I've had a difficult time getting a hold of the absentee/early vote vs. election day votes in Florida simply because the state doesn't provide a breakdown of this by ocunty, you have to get these #'s from each individual county.

But to your point, there are just as many flawed assumptions in considering the early & absentee votes as valid data sets. Military tend to vote Republican, more Dems voted early, etc... But the underlying theme is that over large amounts of votes, i.e. every single county in a state, you'd expect such a large % of the data (12% - 30%) to correlate with the rest. It just doesn't. And given the exit poll data below concerning vote switchers and the 17% "new" voters who decided this election, it seems totally implausible to think that Bush carried upwards of 85%, 95%< even over 100%! of this vote in many counties.

Statewide, the new/available votes went 61%/39% in favor of Bush, but until you look at these numbers by county (and then by precinct, but precincts changed, so this is difficult), you don't see how glaringly ludicrous this increase actually is.

[goldengreek]

6. This is great work!

We just need to keep playing around like this. Once Bev Harris gets finished we'll get to the nuts and bolts of this thing.

These variances really are ridiculous, and since just a few votes here and there tucked away in each county could have carried this election, as you say, then we may be well on our way to nirvana.

[bj2110]

12. And there's a few DU'ers who are saying we should move on...

I just don't understand that at all.

[bj2110]

13. Deep Dive on Hardee County - "Cards Cast" vs. "Counted"

Hardee was #2 on my list of variances between new/available vote % and total vote %, so I decided to do a little further research. I've got a request in with Holmes county for their precinct results.

Hardee has a precinct-precinct GEMS pdf report on its website. The first page lists turnout statistics. There is a column here which is called "Cards Cast", and this # is used to generate % turnout along with registration #'s.

In the Presidential results section (and all other races), there is a column called "Times Counted."

You would think "Cards Cast" and "Times Counted" would be the same, right?

Well, interestingly enough, "Cards Cast" is exactly (I mean exactly) double "Times Counted".

"Times Counted" is equivalent to the recorded vote counts for the Bush & Kerry (minus other candidates & undervotes), but "Cards Cast" is used to generate turnout #'s. Turnout, by the way, was 118% turnout for the county, because of the apparently doubled # of cards cast.

Can anyone think of something which justifies this discrepancy. What is the specific difference between "Cards Cast" & "Times Counted"?

I'll post more here later.