Broward Co - Breakdown of Kerry Early Vote Discrepancy - With Graphs

[bj2110]

I'm not real sure what conclusions can be reached from this, but I definitely know what this suggests.

I'll start by saying that this data was extremely difficuly to come by (At least for a working stiff like me). It has taken probably a dozen calls to the Broward BOE, some hardcopy reports from the BOE, and a few e-mailed reports from the BOE. But, it's starting to come together now.

For further reference, please see one of my previous posts, linked here:
http://www.democraticunderground.com/discuss/duboard.php?az=show_mesg&forum=203&topic_id=184295&mesg_id=184295

As it stands now, I have compiled all early voting results, absentee voting results, and election day voting results by ballot sequence style. There were around 150 unique ballot style used for early voting and each of these corresponded to some number of distinct precincts. Likewise, there were about 100 unique absentee ballot "sequences" which corresponded to distinct early voting styles. Regardless, I now know enough correlating information to break out the Broward vote results into early voting, absentee voting, and election day voting by a sub-level with approximately 95 different variations. Each of these variations consists of some number of defined precincts.

First, please look at this graph, showing the Kerry % in each of these categories; early, absentee, and election day. There is a definite trend.



Obviously, the trend is that Kerry performed significantly better in the early voting stages. What's particularly intriguing is that it is so consistent, with only one ballot sequence showing election day Kerry %'s higher than the early voting %'s, and only a few showing higher absentee Kerry %'s than early voting. Also, it can be seen that as Kerry's % went down, the early voting variance goes up.

Another trend is that Kerry performed slightly worse in the absentee voting, and it is similarly consistent.

There are really only three ways to explain this discrepancy.

1) A significantly higher proportion of Kerry voters than Bush voters chose to make the trip prior to 11/2 and cast their ballot. And this was so consistent that only 1 out of 95 groups of precincts showed a higher election day % than early voting % for Kerry.

2) Something went wrong with the counting of the early votes (or something fraudulent occurred), which resulted in an abnormally high amount of Kerry votes being tallied.

3) Something went wrong with the counting of the election day votes and/or the absentee votes (or something fraudulent occurred), which resulted in an abnormally high amount of Bush votes being tallied.

To address possibility #1, I tested the tendency of sequences with Kerry totals against the proportion of early votes cast. See chart here:



Also, see the reverse tendency for the absentee votes cast:



As you can see, there is some amount of tendency for this to occur. However, the results are definitely scattered and do not correlate that well. Also, I have seen nor heard of any other reason why possibility #1 would be true.

Concerning possibilities #2 & #3, I think we can all agree on which option is more likely due to the relative amounts of motive and/or means for perpetrating such an act. Republicans may disagree, but that argument is obviously easily won.

That leaves us with two possible explanations: #1 above, which has only a shred of support from the scattergraph, or #3 above, which has a significantly larger amount of circumstantial evidence to back it up, even if there is no confession of guilt, or other similar smoking gun.

What do you think?

[DoYouEverWonder]

1. I've collected the tallies from about 1/3 of the counties in FL

for EV and ABS. I am inclined to believe that the EV are a good predictor for the real numbers for votes caste on election day. There is lots of evidence of fraud in the Absentee ballots. The EV ballots are much harder to tamper.


Kerry won the EV:

Bush 219,835
Kerry 267,726

but Bush won the ABS:

Bush 290,918
Kerry 234,311


These are the total votes from the 26 counties that I could find that posted this info.

If you have raw numbers for Broward I would love to add them in. Thanks.

[bj2110]

3. Here's The Broward Raw Totals

ABS BALLOTS 101493
BUSH 40913
KERRY 59552
OTHER 590


EARLY BALLOTS 176743
BUSH 46162
KERRY 128313
OTHER 1770


11/2 BALLOTS 431488
BUSH 157599
KERRY 266008
OTHER 8325

[DoYouEverWonder]

6. Thanks, here the new totals

for 27 counties

ABS:

Bush 331,831
Kerry 293,833


EV:

Bush 265,997
Kerry 396,039


It seems that prior to election day, even with the ABS vote going to Bush, Kerry still was easily beating Bush in FL.

[floridadem30]

48. I agree the early voting seems to be right but the absentee is not

[s-cubed]

2. I don't understand what your horizontal axis - distinct

ballot sequences - refers to and why it is significant. Could you explain? Thanks.

[bj2110]

4. Nothing significant to the chart. It is not a quantitative axis....

What's important is the relationship between the types of voting methods, i.e. early, absentee, & election day.

Ballot sequences in general were described to me as the style of ballot as printed for absentee ballots, i.e. which local candidates, etc... were included.

There also styles for early voting. Why these were different, I'm not sure. It had something to with how many pages the ballot used.

Hope that helps. :)

[bj2110]

5. kick for all the damn graph work. nt

[bj2110]

7. Can anybody add anything to this? What are other reasons why...

... the early vote is skewed to Kerry? Or vice versa, why is the election day vote skewed to Bush? What gives?

Thoughts?

[rdmccur]

28. Another DUer Ignatzmouse

found similar trends between the early and election day vote patterns
in North Carolina. Certainly election day vote or vote tabulation fraud
accounts for these trends (additionally the exit poll results, as well as all the other discrepancies discovered on election day reinforce this hypothesis). Notice I say hypothesis not conclusion. But taking a cue from Dr. Freeman, this just spotlights the need for a full investigation of the election (not just some msm 'bury your head in the sand' mantra).

[Pobeka]

8. What is the x-axis? Is it chronological? Different for each method?

It's hard to get excited about this until I know what the x-axis is.

[bj2110]

9. The x-axis of the first graph is nothing. Simply a list of the 95...

... different groups of precints in order of decreasing Kerry %. The point is that in any particular group of precincts, Kerry's early vote % is significantly higher than both his election day % and his absentee % for that same group of precincts. And this variance seems unusually consistent over the county.

[FogerRox]

16. I worked in Broward County--- there was a concentrated effoert to get the

vote out ----early. The idea is that if there was problem--in early voting you would have days --instead of hours to redress the problem.
this was a COunty wide effort in the K/E Offices.

[L. Coyote]

23. Well, it obviously worked! Good job. n/t

[icehenge]

10. Taking time to digest....

the charts but I will post a response in 24hrs.
Interesting work.

[JBear]

11. Interesting Conclusion - what about other states??

As I recall, the early analysis of North Carolina was equally interesting.

In the counties where all tallying was done by hand, the early votes, absentee and day of election votes were clearly from the same population of voters (no significant difference in the groups).

There was a slight shift in the early vote population towards Kerry in the counties that had DRE systems and a slight shift towards Bush in the absentee (not sure of the percentage, but it was a slight shift compared to optical).

In the optical scan counties, there were major shifts in the early vote for Kerry and for Bush in the day of election voting and absentee.

This was put out fairly early after the election on blackboxvoting.org as I recall.

What does it prove? Who knows....

:bounce:

[rdmccur]

41. Yes, this was Ignatzmouse's work

Very good analysis!

[zapped 1]

12. 58000 absentees never mailed out..

...so yes, there was something wrong with that.
Major media covered this very strongly right up until the election.
In case you don't already know, the BOE said they sent them to the post office (if I am not mistaken,via Courier.hmm..) and the Post Office claims that they mailed them out. FBI "investigated' for an hour or two. The way they "remedied'' the problem was to resend them via overnight mail to the on either thursday or Friday before the election. Many complained that they recieved them AFTER the election and many, like my father, never recieved the ballots at all.
Keep in mind most of these people are retired and go back and forth and vote democratic. As usual, the so-called "glitch works out in *'s favour. I have yet to hear an official explanation on exactly who was responsible for this. Was the Post Office lying?

[bj2110]

13. I'm not sure that helps explain the low absentee %'s....

you would think that the 58,000 absentee ballots that weren't mailed were distributed consistently between Bush & Kerry voters. Of course, maybe they weren't, and if they weren't, and there were some choices made as to which absentee ballots weren't mailed, then that would help explain it.

Still doesn't address why Kerry's early vote % is so high compared to election day %.

Somebody has to have a valid reason for this. If there isn't one, then the early vote discrepancy is a glaring neon sign, imo. One that has also been shown at the county by county level in other states, such as North Carolina, see: Ignatzmoose. (sp?)

[FogerRox]

20. Post Office--no--IIRC the BOE found like 10k on like the friday before nov

Oct 30th IIRC

[mgr]

14. Excellent, this is exciting data, there seems to be something here

Have you statistically compared the three, absentee, early, and election day results with total results at the precinct level. I would think that any precinct that has significant variations between the three I would analyze more closely. I assume you realized that one could count the early voting, and tamper with the absentee, or election day ballots once one counted those results.

I would follow this by mapping the precincts, and compare these to the database of voting complaints, then indicate the ones with these odd deviations, next map the distribution of synagogues (use yellow pages in Netscape), or any other indicator of ethnic distribution (census data is good for everything but Jews). If these start lining up, start looking at the polling books, as you have now hit the hammer with the nail.

I've been looking to do the same with Palm Beach but have been waiting for the BOE to get things together, and not be so wary of what I would be up to.

Mike

[bj2110]

27. That's exactly the path I'm on. However,....

... this data at the precinct level does not exist. It can only be compared with a group of precincts that shared both early voting and absentee ballot styles. There are approx 780 precincts in Broward, but only 150 early voting ballot styles. And for some reason, there were only around 100 different absentee ballot types. Early votes are reported by ballot style, as that is how they are tracked at the early voting polling places, which house any number of different election day precincts. Absentee ballots are reported by what the BOE explains to me as "sequences", meaning the different printing set-ups used to produce the absentee ballots (I think).

That said, there are only about 100 distinct groups of precincts that can be compared at all levels, early, absentee, & election day.

Good thoughts on ethnic distributions, census data, etc... Thanks.

[Land Shark]

15. similar trends seen in Washington state

where early voting takes the form of liberal "absentee" rules, and absentees were much better for Democrats than Sequoia touch screens on election day. The advantage here is that with respect to the governor's race, there was a hand recount of the paper absentees. In the county I studied, Democrat Gregoire won the governor's race approximately 98000 to 96000 on paper, but in the remaining portion of the vote, which is purely touch screen, Republican Rossi won 50,400 to 42,145, a difference of over 8000 votes but it is less than one third of the total voting.
You can download the full study here, which has much more evidence and argument
www.votersunite.org/info/SnohomishElectionFraudInvestigation.pdf

[lostnfound]

18. I love that study. But the effect is just the opposite I think.

In Broward, bj's study shows absentees favored Bush compared to election-day voting.

[lostnfound]

17. What I think is you've accomplished a Herculean job..

Hi BJ2110,

I know what a challenge it is to get consistent groups of precincts. I wonder if you can share your spreadsheet with me? I'd like to play with the numbers, try some additional plotting techniques, and get a sense of your approach (remember I was working on matching Pinellas precincts before?).

For example, I'm curious what a sort by variance would yield. The last data point on your graph makes me wonder what kind of precinct yields a 30% difference in K% for absentee vs early. I guess the answer is likely to be 'snowbirds' in their beachfront vacation condos vs Florida crackers (not a derogatory term IMHO). There had to have been a HUGE GOP effort to get snowbirds to vote in Florida if they had any ability to do so. I recall some anecdotal stories to that effect. Wonder how many of those votes were illegal.

I tend to think that absentee ballots were sorted and intentionally delayed for some precincts. It seems a simple matter to do so. If I'm not mistaken, at the other end of the process, LePore was found to be labelling them 'D' or 'R' on the outside of the envelope when they were received. Hmm. I'll have to search for that story.

[FogerRox]

21. I was in Miramar--I saw almost no Bush teams in Broward County

the Bush GOTV effort was not visible to me

[demo dutch]

30. They pretty much rode Broward off since it's the largest Dem area

[bj2110]

32. But it's an ideal place to hide votes if you had the opportunity. nt.

[bj2110]

37. PM me with your e-mail & I'll send you the spreadsheet. nt

[johan helge]

19. Thanks for the hard work.

Your first graph, I assume, is not unexpected: Kerry's % in absentee votes is low, because of military votes? Kerry's % in early votes is high, because of angry Democrats who can't wait to vote against Bush?

Perhaps this explanation can be checked using sources like:

a) % of the absentee, early, election day voters, respectively, who are registered Democrats
b) exit poll data
c) data from earlier elections

[bj2110]

26. You seem to have rattled off what the GOP's tag line response is...

... and although it may be accurate, I haven't seen anything to confirm or deny it. Which is why I included the second and third graphs to show precinct groups and the respective relationship between % Kerry and % Early / Absentee.

Response to your possible explanations:

a) That data cannot be extracted. Registration data is based on full, complete precincts. It is impossible (and against the law, according the Broward BOE, see my reference thread) to try and distinguish the identity of voters, which is effectively what we would be doing by identifying demographics / registration info against voting method. This isn't at all feasible at the precinct level. I wish it was.

b) What sort of exit poll data is available for Broward Co, or Florida, for that matter? If there is exit poll data on early voting, this would be very helpful. But I doubt there is.

c) Unfortunately, this is the first presidential election to use the early voting processin Florida. There aren't any previous elections to compare to. The absentee votes could be compared, though. I will definitely do this.

Also, another post here suggested comparing other races on the ticket. This is great idea. I will do this also.

[johan helge]

46. As far as I remember, three months after the election

a report from the firms behind the exit poll will be published. Perhaps that report will be useful.

[skids]

22. What would make this more compelling...

Is a comparison with other races on the ticket.

If any "problems" with the machines were only "problems" for high ticket races, then Houston, we have a "problem."

[bj2110]

24. Great thought. I can get that up pretty quickly. Check back. nt

[L. Coyote]

25. The Senate race was much closer. Did the same thing happen?

And the Senate race is equally likely to be subject to the same impacts, whatever they are. I would look further down ticket but compare the Senate race too.

[DoYouEverWonder]

33. Here's a link to the Broward BOE results

http://www.browardsoe.org/elections/elections.php3?LID=1

They actually have a nice interface with the data in a couple of different formats. Have fun.

[skids]

38. Already have that link.

(And in fact have processed it into a CSV file, if anyone needs.)

The data that the original poster has presented is much more detailed than what they have there. What the original poster has is a "Group" version of the "Precinct report" (or maybe the ward report). This specifies the absentee and early vote in separate columns for each precinct or small group of precincts, not just as totals.

Plus he's already got the spreadsheet organized to do the charts.

I posted a similar analysis, though with (voluntarily chosen) larger precinct group sizes, some days earlier for King County, WA, which explains how there are a lot of interesting things to be seen with access to this level of "horizontal market" voter data.

http://abrij.org/~bri/my2c/wabs.html

[bj2110]

39. Here's the graph for the Castor/Martinez race....

Looks about the same. Still a pretty high ticket race, though. I can't imagine that whatever was done to the presidential race (if anything) wasn't also done to the high-profile senate race. I'll work on lower tickets next....

[berniew1]

45. There was vote machine fraud in the Senate race, "default to Martinez"

http://www.flcv.com/fraudpat.html
http://www.flcv.com/EIRSFla.html

[NationalEnquirer]

29. Here in Dade county..

From what I saw at the polling places, early voting went heavily for Kerry. Democrats were so enthused!

[demo dutch]

31. Wish someone would look at the numbers in Dade, because

Kerry did not win big, and I find that suspicious. There were a lot young Cubans voting for Kerry, and for example at U of M people stood in line till late at night!

[rdmccur]

43. Second that one!

[berniew1]

35. Vote machine fraud also documented in Dade and Palm Beach

http://www.flcv.com/fraudpat.html

[berniew1]

34. Widespread vote machine fraud has been documented in Broward County

as well as in other Florida touchscreen counties. In many precincts machines were set to "default to Bush" after an initial Kerry vote, and would only be noticed by those who checked the review screen. Of those on such machines who did not notice the late default to Bush, the Kerry votes would go to Bush. Of those who did notice the switch, about 60% were unable to correct the selection to Kerry. So between 40 and 100 % of Kerry voters on such machine lost their vote.
This pattern also happened in other Florida touchscreen counties, and also in 8 or more other states in similar fashion. These switches were noted by looking at the election day incident reports by voters to the election hotline manned by several non-partisan groups interested in fair elections at www.voteprotect.org

There was also vote switching by some machines in the Castor/Martinez U.S. Senate race.
Documentation of precincts with known vote machine fraud in Broward County( and other counties) can be found at http://www.flcv.com/fraudpat.html
http://www.flcv.com/EIRSFla.html

There was also widespread dirty tricks in the minority precincts of Broward County and many other Florida counties, as well as similar dirty tricks in other states.
http://www.flcv.com/dirtytrf.html
and widespread suppression of minority voting
http://www.flcv.com/EIRSFLa2.html

[berniew1]

36. Anyone with direct experience of any of this please contact me.

[seaclyr]

40. Nice work! n/t

[rdmccur]

42. I would urge you to contact Ignatzmouse

about this. I think he could contribute greatly here.
He has another thread running on page 1.

[bj2110]

44. I've sent him a message. nt

[Pacifist Patriot]

47. What do I think? I think my head hurts.

It's at times like these that I bemoan my lack of mathematical ability and interest. My brain leans heavily towards liberal arts. On the other hand, I am extremely grateful to you and all of the other number crunchers who are still beating the bushes to expose the problems. Keep up the good work.

[floridadem30]

49. A republican filed a lawsuit in this county and then dropped it.

Here is the link www.floridabaptistwitness.com3487.article It is a very interesting article that explains anomolies that took place and how votes shot up on election day.

[bj2110]

50. Anybody know why it was dropped? nt

[findTruth]

51. Request electronic copy of database in NM, Ohio and Florida!

Republic party in Washington State asked the State Patrol for an electronic copy of that database, the Washington State Identification System. The list contains more than 1.2 million records.

We should do the same thing in Ohio, NM, Florida and other states!!

[ignatzmouse]

52. E-Voting Coincides With Low Turnout

It's very good work. Thanks for jumping on it.

The first thing that detractors generally mention is that the Democrats had a large "Get Out The Vote" campaign and that will skew the numbers toward the early voting. I would like to point out, however, that if the early voting numbers were that weighted, it should correlate to a very high voter turnout. In Broward County, the voter turnout was 66.70%, the second lowest turnout of any Florida county won by Kerry. I have Broward as using ES&S's iVotronic touchscreen for election day. Looking at other Kerry touchscreen counties, I also find Palm Beach at 62% and Miami-Dade at 67.70%, all on the low-end. Excepting the infamous Volusia county, the Kerry optical scan counties appear within a more normal range of turnout. (Note the figures quoted are from information I downloaded early on and should be re-verified.)

KERRY OPTICAL SCAN COUNTIES TURNOUT
Volusia 67.50%
Orange 72.60%
St. Lucie 72.80%
Monroe 77.10%
Alachua 78.20%
Gadsden 78.50%
Leon 79.60%
Jefferson 80.70%

KERRY E-VOTING COUNTIES TURNOUT
Palm Beach 62.00%
Broward 66.70%
Miami-Dade 67.70%

Bush's winning e-voting counties, however, offer a markedly different turnout.

BUSH E-VOTING COUNTIES TURNOUT
Martin 73.50%
Hillsborough 74.30%
Indian River 75.10%
Collier 75.50%
Lake 76.70%
Pinellas 77.40%
Sumter 79.00%
Nassau 79.20%
Lee 79.50%
Sarasota 81.40%

Clearly, something odd is going on. Either we have Bush bumped numbers especially in Hillsborough and Pinellas or there is some kind of suppression in the three Kerry e-voting counties. The absentee ballots are definitely one place to look. If the GOTV was so high, why are the absentees so low? If the GOTV was high enough to give the early voting such a boost, why are the overall turnout numbers so low? Secondly, why are Kerry's turnout numbers the lowest in e-voting counties as opposed to paper trail op-scan counties? Does the paper trail also provide an explanation for the difference between early voting and election day results in e-voting counties? Overall, the Republican strategy was clearly more sophisticated than just swapping votes. The same goal can be accomplished by keeping turnout low in heavily minority and Democratic areas -- thus target counties/precincts may record a nice win on paper for the opponent (Kerry) and yet effect the overall race much less. On the face, it's somewhat invisible. Hope that is of some help and can be used to further your interesting study.

[bj2110]

53. Very good insight. I did a quick take on % Turnout vs. Early Vote Variance

I didn't find much correlation in the sub-level Broward #'s. Very scattered, here's the graph:



% Variance represents the % difference between early voting and election day voting.

There's not much correlation at this level within Broward between the two.

As you said, another good way to illustrate and verify the hypothesis is to look at other counties in total, specifically, the difference between op scan and touchscreen methods.

Thanks.