Ohio Ballot Order Bombshell

[minvis]

I just got the ballot order from Clermont County in Southwestern Ohio. * won this county by nearly 2-1. I wanted to see if the large third-party vote totals would appear in a * county. Although there were not a lot of third-party votes, I came across something amazing to me. Perhaps this was discussed in the Cuyahoga County postings, but I found out looking at Clermont County that although the number position of the candidate does rotate, the actual order of the candidates does not. For instance, the number position of the candidates in one precinct can be as follows:

Bush Kerry Nader Peroutka Badnarik
1 2 3 4 5

In the other precinct shared in the same location, it could be as follows:

Nader Peroutka Badnarik Bush Kerry
1 2 3 4 5

Notice that although Bush is in the first position in the first precinct and fourth in the second precinct, the order of the candidates themselves never changes.

Why does this matter? Well, if you are a programmer trying to come up with a way to change votes, this makes it very easy. If I know that my candidate always appears right before my opponent no matter what, I can write a program to change that by simply telling it for every n number of votes, change opponent's ballot number (x) to x-1.

Maybe I was naive but I assumed that rotating the ballot order also meant rotating the candidate's order as well. This makes computer fraud appear to be much easier to do than I first imagined.

[impeachthescoundrel]

1. Wonderful job

I am not a programmer and would never have known.

[FtWayneBlue]

2. True, but

A (Sandusky) county-wide report lists the same order as follows:
Badnarik, Bush, Kerry, Nader, Peroutka. Alphabetical, is all.
Or so it seems.........

[EuroObserver]

3. Always the same order, huh?

I suppose that's the strict meaning of the term "rotation": not the same as randomization.

As a (one-time) analyst/programmer myself, I can confirm that would makes it very easy to progarm a little fix...

[minvis]

6. Candidates Listing in the Precinct not on County-Wide Report

You are correct, the county-wide reports list the candidates alphabetically. What I am talking about is how they appear on the ballot when you vote in your precinct.

[L. Coyote]

71. Ballot orders are discussed in great detail online.

What you point out here is a side-affect of rotated ballot orders and how they are assigned. Because ballot order assignments follow the precinct number sequence, there are some built-in inequities that alter cross-vote probabilities for adjacent candidates in the order, ie. Bush and Kerry.

So votes are being switched without a conspiracy to modify tabulation. And they stay switched no matter how many recounts are done!

Ballot orders are discussed in detail in the 2004 Ohio Election - Presidential Ballot Orders and Cross-Voting page. It discusses all the ballot order combinations and the probabilities of vote-switching for the possible combinations.

Another Web page at the site 2004 Ohio Election - Analysis, Summary, Charts, and Spreadsheets and the article "How Kerry Votes were Switched to Bush Votes!" discusses specific examples in Cuyahoga County and some of the non-random aspects of cross-voting.
.

[Clark2008]

23. But that still has Bush just before Kerry

That's the point the OP is trying to make.

[snot]

4. kick!

[jdog]

5. That does make it sound awful easy. n/g

[emlev]

7. Link to more work on ballot order question

http://www.jqjacobs.net/bush/ohio_ballots.html

This is James Q. Jacobs study of Ohio ballot order and vote switching issues.

[understandinglife]

8. Very cool piece of information; Thank You!! (n/t)

[crispini]

9. Are these opscan precincts?

because if they are, it makes it even easier to screw things up, no actual deliberate fraud necessary. if these two precincts share a location and are not well organizied as far as a layout, it would be very easy for a voter to trot off with a ballot for precinct A and put it in precinct B's opscan ballot box. wups!

we had voters who actually tried to WALK OUT with ballots in my precint. Der! well, it's easy to get confused.

[minvis]

21. Opscan Ballot?

Yes, it was a county that used Opscan.

[AirAmFan]

10. You've learned something about ballot orders that's valid in Cuyahoga too

Once you know who was first on the ballot in a given precinct, then you know who was in positions 2, 3, 4, and 5. But I think you're taking your new knowledge too far. It's surely POSSIBLE that some programmer in some county still using punchcard ballots could have written a program to bias election results in the way you're suggesting. But most vote-tampering schemes that can evade detection here in the twenty-first century have to be much more complex than that. What you're suggesting doesn't provide the kind of "plausible deniability" to county election officials that makes election-tilting schemes viable today.

Ballot-tabulating programs for punchcards have been around for a well over a hundred years, and have gone through much scrutiny. Just Google "Hollerith", for Herman Hollerith, who invented a punch card system that was used during the Census of 1890. When punch card ballots first came into use, they were vulnerable to all kinds of abuse, both known and unknown to election officials. In the 1960s in some precincts, a voter could bring hs own pre-punched "flush" card (or punch one on the spot), to zero out all the votes on all the cards that preceded his in a counting run. Those kinds of problems have been fixed decades ago. The first place to look for abuse of punchcard voting now generally is in the punch card hardware, from "benign neglect" more than from deliberate positive measures to bias the vote. Just give minority precincts bent styluses that will not remove chads, especially if minority voters follow instructions and "push straight down". Or give them machines packed with ten or twenty years of chads, so that the ballot will not go all the way in.

The second place to look is misguided election-board policies such as having more than one ballot order per polling place. And a third place to look is hackable "central tabulators" than use insecure Windows Internet software to transmit precinct totals to the county and the state.

Electronic "direct entry" voting machines (Diebolds) are a different matter--they're still not well understood by many county election officials. But tabulating punchcards at the precinct level is something they know about, well enough to make the specific kind of tampering you're suggesting unlikely.

[k8conant]

12. We've definitedly found your second place in Cuyahoga:

"The second place to look is misguided election-board policies such as having more than one ballot order per polling place."

I don't think the different methods of voting are really such a "different matter"--only hand counting of paper ballots would not have to pass through the "tabulators" which may well be securely or insecurely hacked (I tend to think securely: in other words, in the original code and its associated system code, right from Diebold or Triad or Sequoia or ES&S).

[minvis]

26. Ballot Order

I agree that this simple code manipulation probably wasn't done here. I was more just amazed that the candidate order did not change no matter what position it was in on the ballot. I am looking at the precincts with a shared polling place in Clermont County. There are several and most have more than 2 precincts that shared polling locations. One township had 14 precincts, all at the same location. Many had 4 or more.

What is interesting so far is that many of the polling places that shared two precincts have this rotation:

Badnarik Bush Kerry Nader Peroutka
Bush Kerry Nader Peroutka Badnarik

Bush is always under Kerry in one precinct and Nader is always under Kerry in the other precinct. For instance, a potential Kerry vote in one precinct brought to the ballot box in the other precinct turns into a Bush vote and a potential Kerry vote in the other precinct becomes a null (Nader) vote in the other precinct. Obviously, it is bit more complicated when you have 3 or more precincts sharing the polling place. When I get done with all the shared precincts, I'll post them and let people see if they find anything more unusual.

[Bill Bored]

39. But wouldn't Bush votes become Badnarik votes or Kerry votes?

How could the systme be made to favor Bush more often than Kerry? (Gee I can't type/spell tonight!)

[minvis]

44. Yes,

but remember that the county I am looking at is a solidly Bush county. If I'm going to commit fraud for my candidate (Bush), I wouldn't have this run in any precincts I knew I was going to win anyway. I only run it in precincts that I know are huge Kerry precincts. By the way, Bush won all the precincts in Clermont county. There was one precinct where Kerry got within 4 votes, however.

[AirAmFan]

47. What do you mean, "I wouldn't have this run ..." Have WHAT run?

[AirAmFan]

46. Good question. See posts #32 and #36. My theory is that "caterpillar

crawl" is associated with long lines and overcrowding, which at least in Cuyahoga were concentrated in poorer areas that favored Kerry. Wherever the polls are overcrowded, the candidate most people favor is likely to lose votes to "caterpillar crawl". From what I've heard, places that favored Bush generally were not overcrowded and often had precinct staffers who accompanied each voter to a booth to make sure it was the right one and that the punchcard was inserted correctly.

Tests of my theory would entail looking for excessively high votes for minor candidates. In Cuyahoga, I don't see them in pro-Bush areas. It's possible that Bush lost more votes than Kerry to this phenomenon in some pro-Kerry counties--after all, Ohio has dozens of counties that used punchcards. But I've seen no newspaper stories, cable media broadcasts, or testimony about systematically overcrowded polls in Bush bastions. Have you?

[Bill Bored]

57. Yes I read that after I posted my question.

How about this:

You ensure that the shift is unidirectional. In other words, you allow some of the shifted Kerry votes to go to Bush, but none of the shifted Bush votes to go to Kerry. This way, without even knowing how anyone has actually voted, you know that Kerry's MARGIN will be reduced more than Bush's.

Let's say the ballot orders in 4 co-located precincts are:

4 Badnarik Bush Kerry Nader Peroutka
3 Bush Kerry Nader Peroutka Badnarik
2 Kerry Nader Peroutka Badnarik Bush
1 Nader Peroutka Badnarik Bush Kerry

If you randomly shift ballots upward from any row (precinct) to any other row, Bush will always come out ahead because Kerry will never get any of Bush's votes, but Bush will always get some of Kerry's. Of the shifted ballots, Bush will get 1/3 of Kerry's Precinct 1 votes, 1/2 of Kerry's Precinct 2 votes, and all of Kerry's Precinct 3 votes. Kerry will only get votes from the 3rd parry candidates, but never from Bush, unless you start shifting ballots in the opposite direction, from Precinct 4 downward.

Now, this also explains why some precincts, in this case, precincts 2, 3 and 4, could have more votes than registered voters, right? They are the ones that receive the shifted ballots, favoring Bush.

So, tell me, does this make sense? Could it have been done on the ground by steering voters from, in this case, lower numbered precincts to higher numbered ones but not the other way around? Could the lines for the lower numbered precincts have been engineered to be longer?

Tell me.

[AirAmFan]

60. The votes don't get reallocated at random: there's a pattern

that is knowable as soon as you know the entire configuration of ballot orders at a location.

Your hypothetical corresponds to five actual locations in Cuyahoga. One of them is listed at the bottom of page 4 in http://boe.cuyahogacounty.us/boe/PDF/votinglocations.pdf . ST._SAVA_CHURCH_HALL co-located precincts with IDs 0931, 0932, 0933, and 0941 . Look up the precinct results in the big "Canvass Report" at http://boe.cuyahogacounty.us/boe/results :

0931 BROADVIEW_HTS_3-A B 1 283 251 0 1
0932 BROADVIEW_HTS_3-B W 0 456 238 0 0
0933 BROADVIEW_HTS_3-C K 0 238 179 0 0
0941 BROADVIEW_HTS_4-A D 1 263 298 0 1

(The Canvass Report has everything but the initial of the first-line Presidential candidate ("W" for Bush)).

The tiny votes for minor candidates (Badnarik and Peroutka) indicate that "caterpillar crawl" could not have been very important at this site. Each received a total of 2.

But let's plunge ahead anyway. When four precincts share one location, there are 12 ways voters could use the wrong machine: With this configuration, a voter registered in 3A could use a machine set up for 3B, 3C, or 4A, and so on. If all these ways to go wrong are equally likely, then every 12 wrong votes intended for Kerry or for Bush are split equally among the other four candidates (not for all congugurations, but definitely for this one: Do the math). An estimate of the maximum total number of votes lost to caterpillar crawl by Kerry or by Bush at this location thus is 8.

[Bill Bored]

72. AA Fan

My point is that perhaps they didn't go wrong equally. If you want to favor Bush, all you have to do is get voters to use the box or machine where Kerry's votes go to Bush and Bush's votes DON'T go to Kerry. There don't have to be 4 precincts together. Two would be enough. You don't have to know how the votes will be (or have been) cast because the odds are always in Bush's favor as long as you move ballots unidirectionally.

I haven't looked at the actual data; I'm just putting forward a hypothesis of how the whole thing can be non-random and favor Bush.

Each vote moved from Kerry to Bush is worth 2 votes of the margin between them.

Each vote moved from either Kerry or Bush to a 3rd party candidate is only worth 1 vote in the Bush-Kerry margin.

Are you saying there are no co-located precincts with this pattern:
High 3rd-party vote count
Lower than expected Kerry vote count
Higher than expected Bush vote count
???

I thought this is what had been found. And I assume, any vote for Nader will be an undervote since he wasn't on the ballot. We know there were 93,000 undervotes, right?

[AirAmFan]

76. Lower or higher than expected votes for major candidates are very

difficult to detect. We can detect unexpectedly high votes for minor candidates in ths case because their countywide total was way under one percent.

You're calling into question my simplifying statistical assumption that all wrong vote possibilities in a site were equally likely. Without that assumption or something equally strong, there's no way the problem of detecting miscounts and estimating votes lost to Kerry and to Bush countywide can be solved. So it's important to look for patterns in the data that would be inconsistent with any simplifying assumptions. I haven't found such patterns.

[L. Coyote]

78. Actually, this has been done and it is detectable.

At any given location with several precincts, the voting pattern can be analyzed. Adjacent precincts have similar voting patterns, unlike the urban/suburban schism. You simply need to value each precinct in a subset with probabilities for how cross-voting results, then sort and analyze.

The discussionin in the the 2004 Ohio Election - Presidential Ballot Orders and Cross-Voting page discusses this. Scroll down to the table with the heading: "2\3 Locations with Crawl 1 and Crawl 4 Ballots." Here the ballots have been separated into the two groups within the subset, precincts with 1.0 probability of Kerry cross-votes switching to Bush, and those with 0.0 probability at the same locations. The difference in non-vote percentage is 2.2% higher in the P = 0.0 group, the Kerry vote is 3% lower in the P = 1.0 group.

There is further discussion above and below the table, including how the vote-switching probability was skewed by non-random, higher numbers of precincts with collocation of Kerry and Bush.

[AirAmFan]

109. The whole thing can be RANDOM and still favor Bush too, if

the existence of chaotic conditions at polling places is biased toward relative prevalence in Kerry areas. I've developed a framework that gives Bush voters exactly the same chance to have their votes miscounted as Kerry voters. Incredibly excessive votes for minor-party candidates at a polling place bring them into play as sources of miscounted votes. The candidate in favor of whom most voters wanted to cast thier ballots will lost the most votes when things get chaotic.

My preliminary, very conservative estimate from this unbiased framework is that Kerry lost more than a thousand votes in Cuyahoga due to "Caterpillar Ballot crawl". See the new thread at http://www.democraticunderground.com/discuss/duboard.php?az=view_all&address=172x5139
. I'd appreciate your comments.

[AirAmFan]

66. And the underlying process in principle could shift as many votes

away from Bush as from Kerry, if all polling places were equally overcrowded and chaotic.

But we know that it was in overwhelmingly pro-Kerry areas where overcrowding was most prevalent. Not only does overcrowding of the polls discourage voters, both those who actually experience long lines and those who simply hear about them or see them on TV. Overcrowding combined with the "caterpillar ballot" scrambles some of the ballots that, despite high barriers, voters manage to cast.

Where there is order and no overcrowding, few votes are miscounted, regardless of the configuration of multiple ballot orders. But at an overcrowded, chaotic location, the more popular of the two major candidates loses the most votes. And that was Kerry pretty consistently, at least in Cuyahoga, according to all reports I've heard

[L. Coyote]

79. The correlation between Kerry support and non-votes is very troubling

and quite dramatic. It points to, at a minimum, whose votes are not being counted. But also, where cross-voting is most prevalent. So, even if we cannot "precisely" extricate one from the other, we know whose getting cheated out of votes.

This graph is a subset with about 24% of Cuyahoga votes, over 2% of Ohio votes.

This pattern is alarming, no matter which factor has the greater role.

In this subset of precincts, 110 of 352 precincts have a 1.0 probability that major candidate cross votes will switch. The random probability of 0.250 has been augmented to P = 0.31250 by the non-random ballot order distribution. Hence, the proportion of non-vote and Badnarik "evidence of cross-vote" increments are lower too! And this subset averages over 70% Kerry support "after vote switching."

[mgr]

135. Test it with Knapp's mapping

You have a hypothesis here that can be easily addressed using the information that Knapp did showing distribution of various voting irregularities in Cuyahoga, all you would have to do is plot it.

[AirAmFan]

40. In Cuyagoha, this would be a tough configuration for which to calculate

miscounts, because the Cuyahoga counting program provided no tallies of votes for Disqualified. The text file whose URL I provided in post #18 has a column of zeros for Disqualified, and a DUer in contact with the BOE Director in Cuyahoga verified that getting such counts would ential reprogramming and re-counting.

Do your data have precinct-level counts for "Disqualified"? If so, I've derived formulas that can estimate the total miscount for the county under reasonable statistical assumptions, and then allocate the miscount fairly to the candidates likely to have lost them.

[minvis]

42. No, but...

they do have precinct breakdowns for Undervotes, Overvotes and Write-Ins.

[AirAmFan]

108. Lack of tallies for "Disqualified" really hinders analysis of this configuration

In my Cuyahoga calculations, just posted to http://www.democraticunderground.com/discuss/duboard.php?az=view_all&address=172x5139 , I classify this "BW" configuration as part of "Case 25". See especially posts 10 and 11. I'd really appreciate your comments on this thread.

[k8conant]

11. I'm a programmer and I also looked at ballot switching problems in

Cuyahoga (where it happened extensively because voters used the wrong ballots). The only way we can tell about what minvis is pointing out here is to see the code that actually records the votes. Minvis makes a good point: that the order rotates (as in 12345 23451 34512 45123 51234) means that Bush (at 2) is always 1 less than Kerry (at 3). In the code there has to be a correlation set up between the number on the ballot (or the location on the screen) and the column where the candidate's total is being tabulated.

So...we still need to see that code.

[AirAmFan]

15. I'm not following your logic. I was simply pointing out that a certain

election board policy could misdirect votes intended for Kerry (or for Bush) even with unbiased counting programs. A counting program is looking for Kerry votes in a certain position on the ballot. But if the voter has punched out the chad for Kerry (or Bush) on a punch machine where he's in a different position, the counting machine does not detect this, and gives the vote intended and punched for Kerry to another candidate.

The fact that Ohio ballots were loaded up with about a hundred different choices for voters to make meant that each person took several minutes to vote. And the fact that minority poor and pro-Kerry precincts were "shorted" compared to Republican bastions in the suburbs made it much more likely that Kerry votes would be miscounted than Bush votes.

When is a voter likely to use the "wrong machine" and take a ballot stamped with one precinct's name to another precinct's punch, with a different ballot order? Why, when there are lines of unequal length for punch machines that have different ballot orders in the same polling place. Why stand in a two-hour line for your own precinct when there's a much shorter line for another precinct?

So the code for counting votes is irrelevant, once you know the process by which punchcards got to the counters. As I said in my initial post, a board policy that bleeds Democratic votes in overcrowded polling places on Election Day provides a "plausible deniablility" to the people in charge they wouldn't get from fairly crude manipulations of punchcard-counting code.

[k8conant]

85. Yes, and this has been discussed extensively before...

I'm surprised it was mentioned in Congress on the 6th. Maybe folks thought it wasn't fraudulent but it sure as heck was wrong (for the intent of the voter). This is why recounts won't come out right: the count was wrong in the first place.

[L. Coyote]

82. Yes, there are two issues being discussed here, is the code fixed and ...

the impact of cross-voting. The original post concerns "fixing" the counting devices. They are separate issues.

Because each precinct has a specific ballot order, this kind of "fix' would require different code for different precinc tabulators.

However, this can be accomplished irrespective of whether Kerry follows Bush in the order or not.

Confidence in elections cannot be restored unless and until there is no longer any "hidden code." All code must be in the public domain.

[RaulVB]

13. Kick (n/t)

[Melissa G]

14. kick

[VTGold]

16. When I saw the comma delimited fields in the "RobGeorgia" data I thought..

..of the same thing. I emailed BBV about it months back but don't know if it made any impact.

Its a "hide in plain site" way to manipulate the numbers - nothing changes - columns just get switched around.

[many a good man]

17. What's interesting

The fact names are always alphabetical doesn't mean much because the program will always "know" which one is ** any way.

What I would find interesting is IF Triad or ES&S used a different rotation formula in different counties. If the staggered start-alphabetical rotation meets state law, then they should use the same ballot rotation formula in ALL the counties they service. From a software development perspective, Why wouldn't you? How many "versions" of their software are certified, any how?

It may turn out to be just incompetence instead of fraud, but sounds like it might be yet another violation of state rules...

[AirAmFan]

18. The "ballot rotation" algorithm was very simple in Cuyahoga

Cleveland and suburbs): List the 1436 precincts in order by their ID numbers. Then divide them into blocks of five. The first in each block has Badnarik in position 1, the second has Bush, the third has Kerry, the fourth has "Disqualified" (Nader), and the fifth has Peroutka. You can check this by downloading the multi-megabyte precinct-level "Canvass Report" text file from http://boe.cuyahogacounty.us/boe/results , and then by downloading images of the individual precinct ballots. For example, an image of the ballot for Precinct 8 G in Cleveland is (or was) at http://boe.cuyahogacounty.us/boe/ballots/pdf/CLEVE08G.PDF .

This looks fair enough. But 1299 of the 1436 precincts were clustered into 447 multi-precinct polling places. And the algorithm for that clustering is much less transparent than the assignment of precinct ballot orders. See http://boe.cuyahogacounty.us/boe/PDF/votinglocations.pdf .

[AirAmFan]

19. "Cleveland 8 G" is number 322 in the list of precincts by numeric ID

number (2207), and so has Bush in position 1. So do all precincts whose rank number ends in a 2 or a 7. A last digit (in rank, not in absolute ID number) of 1 or 6 almays means Badnarik is at the top. 3 or 8 means Kerry, 4 or 9 means "Disqualified", and 5 or 0 means Peroutka on the coveted top line of the ballot.

[rdmccur]

22. Good observation

so for precinct n, n (mod5) = 1 implies Badnarik in position 1, n (mod5) = 2 implies Bush in position 1, n (mod5) = 3 implies Kerry in position 1, n (mod5) = 4 means disqualified in position 1, and n (mod5) = 0 implies Peroutka in position 1. Wasn't there a county where Bednarik got way too many votes to make sense and Kerry got very few? If so what was its rank number, n? N (mod5) means remainder upon division by 5 (easy to program).

[minvis]

27. Yes, Cuyahoga County

Several precincts in Cleveland had abnormally high third party votes. One had a high Badnarik count and another had a large Peroutka count. I can't remember the exact precincts, but there are previous postings on DU specifically on these Cleveland precincts.

[rdmccur]

29. We need to get that info!

Some tinkering scheme may be herein revealed as reflected in your algorithm

[Zan_of_Texas]

84. Cuyahoga resources

MAPS, CHARTS
http://www.dailykos.com/story/2004/12/7/22452/6795

~~~~~~~~~~~~~~~~~~~
Ohio tally fit for Ukraine
Voter fraud in the Ukraine? Give me a break.

by Juan Gonzalez (Democracy Now co-host)
New York Daily News
November 30, 2004
http://www.nydailynews.com/news/politics/story/257365p-220441c.html

It has been a month now and we still don't have a clear count of the votes for our own presidential race from the state of Ohio. For those who may have forgotten, Ohio supposedly assured George W. Bush a second term in the White House - only the most important job on the planet.

snip

Back in 2000, the combined third-party votes in those two precincts - including the Nader vote - was 8. Cuyahoga, like most of Ohio's 88 counties, uses punch-card balloting. "That's terrible, I can't believe it," said City Councilman Kenneth Johnson, who has represented the 4th Ward since 1980. "It's obviously a malfunction with the machines."

But Peroutka and Badnarik polled unusually well in a few other black precincts. In the 8th Ward's G precinct at Cory United Methodist Church, for instance, Badnarik tallied 51 votes - nearly three times better than Bush's 19. And in I precinct at the same church, Peroutka was the choice on 27 ballots, three times more than Bush's 8. In 2000, independent candidates received 9 votes from both precincts. The same pattern showed up in 10 Cleveland precincts in which Badnarik and Peroutka received nearly 700 votes between them. In virtually all those precincts, Kerry's vote was lower than Al Gore's in 2000, even though there was a record turnout in the black community this time, and even though blacks voted overwhelmingly for Kerry.

snip

~~~~~~~~~

Odd vote results point to mix-ups at some precincts

Friday, December 10, 2004
Diane Solov and Diane Suchetka
Plain Dealer Reporters
http://www.cleveland.com/news/plaindealer/index.ssf?/base/cuyahoga/1102674912293811.xml

SNIP

Michael Vu, Cuyahoga County's election director, said the errors don't reveal a systemic problem that plagued the election. But he said the circumstances surrounding the erroneous votes will be investigated, and that they drive home the need for more poll worker training and more fail-safe measures to account for the human factor. "We'll have to make the necessary changes so they won't occur again," Vu said. "There are lessons learned in the largest election that Cuyahoga County has ever conducted, and those issues will be addressed for future elections."

It's difficult to tell exactly what went wrong in the anomaly precincts. Vu said poll workers are trained not to share books of voting pages, and the devices that hold them, with other precincts. But there are clear signs that, in some cases, poll workers erred in setting up the polling stations or misdirected voters. And voters, who often stood in long lines in cramped quarters, may have grabbed any open booth they could find, unaware there was a difference. In some cases, a combination of both factors conspired to produce bad votes.

The biggest problems occurred at Benedictine, where Election Day got off to a rough start on that rainy morning five weeks ago. Poll workers were locked out of the building until just before the polls opened at 6:30 a.m., forcing them to rush to set up the voting devices in the school cafeteria and complete other preparations. "The fact that we couldn't get in at 6 to get ourselves properly set up, anything could have happened," said poll worker Marjorie Baxter. Baxter recalled how hectic the day was and said she was so busy signing in voters that she didn't look up from her seat much. But she remembers poll workers trying their best to direct voters to the right place.

But Katie Daley, an observer for the Democratic Party who also spent the day at Benedictine, said voters waiting to cast their ballots formed a single line between the 4F and 4N precinct tables. They approached the booths, which were arranged in a semi-circle, as they became available, she said. "There was no distinction between precincts," Daley said. "Voters were being told to go to any machine that was open."

more

[AirAmFan]

32. The Benedictine High School polling place in Cleveland (listed on page 11

of http://boe.cuyahogacounty.us/boe/PDF/votinglocations.pdf ) is the precinct cluster where the "ballot rotation" went furthest afield. Get the individual precinct election results for the two precincts that voted together there and you'll see something very strange (download the text file I mentioned in post #18. Cleveland 4 F, rank 233, gave Peroutka 215 votes of his 1751 countywide total, and gave Badnarik no votes. Cleveland 4 N, rank 241, gave Badnarik 164 of his 1886 countywide total, and Badnarik 10. This is exactly the kind of pattern you'd expect to see if substantial numbers of people trying to vote for Kerry in both precincts used the other precinct's punch machine by mistake.

The ballot order for 4F was Kerry Disqualified Peroutka Badnarik Bush, while for 4N it was Badnarik Bush Kerry Disqualified Peroutka. Kerry was on line 1 for 4F, and Peroutka was on line 3. And Kerry was on line 3 for 4N, with Badnarik on line 1.

Each voter had to go to the right precinct to have her name found on the reigstration list and get a punchcard stamped on the back in blue with the name of the precinct. In the suburbs, it was very common for a precinct staffer to accompany each voter to a punch machine, to insert the punchcard for the voter, and then to draw the curtain closed. But at many inner city locations, such as Benedictine High on Martin Luther King Drive, voters were left to negotiate hours-long lines to punch out their chads on their own.

If 200 or so 4F voters tried to vote for Kerry on one of 4N's machines, their votes would have been miscounted for Peroutka. Not Badnarik, not Bush, and not Disqualified, but Peroutka. And if 160 or so 4N voters had punched Kerry's chad on one of 4F's machines, their votes would have been misdirected to Badnarik. Not Peroutka, not Disqualified, and not Bush. Since this is exactly the pattern shown in the results, we can be pretty confident that the "Ohio Caterpillar Ballot" robbed John Kerry of more than 300 votes at this one polling place. And there were 446 other clusters of precincts where something similar could have happened.

I call this the "Ohio Caterpillar Ballot" to draw attention to its similarity to the Palm Beach County Florida Butterfly Ballot that made Theresa LePore famous in 2000. Like a caterpillar, a candidate's name crawls away from the place where it would have to be in order for the intended candidate to be awarded the vote.

[rdmccur]

34. So was this problem an

accident or did it happen by design?
it is curious these two precincts vote at the same location and the order of the precincts is such that this kind of misvote can occur.

[AirAmFan]

36. I think this knd of bleeding of Kerry votes out to other candidates

is a consequence of the nutty policy of (1) having differently-set-up punch machines at the same location and of (2) overcrowding and long lines.

I'm not sure, but (1) may be a consequence of a decistion made a generation ago that has not been revisited in light of changes circumstances. When the 1976 constitutional provision cited in
post #28 was written, I'd imagine that precincts rarely were clustered in one polling place. Now the vast majority are. The decision not to revisit an antiquated procedure well may be partisan; it certainly benefits Republican incumbents.

(2) To the extent that inner-city locations were deliberately shorted on voting machines, compared to pro-Bush suburbs, "caterpillar crawl" was a deliberate predictable consequence of inevitably resulting overcrowding.on Election Day. There also could be more subtle intentional Republican election unfairness at work. Cuyahoga may have been deprived of its fair share of state funds to purchase voting machines for November just as it is deprived of its fair share of education funding, transit money, etc.

But caterpillar crawl is not the main consequence of inadequate election resources in pro-Kerry precincts. I'd put the maximum effect of Cuyahoga caterpillar crawl on Kerry's net vote at about 10,000. That has to be an order of magnitude lower than the effect of long lines at the polls on voter turnout. In her Ohio Challenge speech on the Senate floor, Barbara Boxer estimated the number of people who went to the polls in Columbus but left without voting at 5,000 to 10,000. The corresponding number for Cuyahoga has to be bigger. And the number of people who merely heard about the long lines, or saw them on TV, and stayed away from the polls entirely that day has to be on the order of 10 times the number who actually wasted their time in line.

[Occulus]

54. IOW, we got "punked". Again. n/t

[Zan_of_Texas]

87. It's insane or corrupt to place multiple precincts

in the same building with different ballot orders.

It's like putting banana peels on top of ice and expecting people not to fall -- no way this results in an accurate tally.

A poster from Wisconsin said they at least had different color ballots for different precincts in the same building. Apparently Ohio did not.

But, here's the other thing.

If you are running a tally for punch cards, the tabulating person places a beginning card and an end card to tell the ballot reader what is what.

This info is notes from an interview by Evan Davis and Terry Taylor with Triad heads Brett and Dwayne Rapp:

"We do not supply punch card ballots -- we use standard Documation ballot readers that have been available since the early 60s. We have an interface box that connects to the personal computer - header card tells the ballot reader which precinct is being tabulated. An end card at the end of the stack. Pink card is the CANCEL card. Green card tells it it's the end of stack. Ballot tabulator - reset, ready to count."

So, if you're counting punch cards with one tabulator in a building with two precincts, presumably all you have to do is shuffle some PRECINCT A cards into the stack to be read for PRECINCT B. Right?

Then later if it's obvious something went awry, you can fall back on the "voter mistake" excuse.

Or, perhaps there WAS enough voter mistake. But I doubt it.

The other thing that is really curious is the "inability" of officials to tell us what is in that "ZERO" disqualified slot. I'm betting lots of votes are in that black hole, or they wouldn't be so shy about telling us. Normal accounting procedures would show that vote, because every vote should be accounted for even if it goes in there.

[AirAmFan]

92. Thanks for this. I agree the unavailability of "Disqualified" tallies is an

IMPORTANT barrier to the most reliable kind of reconstruction of miscounted ballots due to "caterpillar crawl".

There's also an important point in the material you've posted about opportunities to reverse "caterpillar crawl" at the stage when decks of cards were being marshalled for precinct report runs.

Each card is stamped on the back in blue with the name of the precinct that issued it. If someone had fanned through handfuls of punchcards in each precinct's deck BEFORE running it through for a precinct report, cards with the "wrong" precinct stamp could have been pulled out and reallocated to the decks where they belonged.

Even if that opportunity to correct the problem had passed, there would have been YET ANOTHER CHANCE if ballots run through for each precinct's report had been stored together. In a thorough recount, the punchcards with the wrong stamp on the back could have been pulled out and new reports run or hand tallies made to completely reverse the problems in the initial precinct reports.

But I'm pretty sure that both these opportunities were lost. People associated with the recount reported that ballots had been re-sorted so that information of the kind I'm taliking about was lost forever. Maybe they were wrong and it's still possible for some professor's team of graduate student to do a FOIA request and find and tally all the wrong-stamp punchcards from each precinct run.

[AirAmFan]

110. All great insights. Over the past month, I've developed a framework in which

all these factors come into play. I've just posted my first, very conservative estimates of the effects of "Caterpillar Ballot crawl" on Kerry's Cuyahoga margin. I'm fairly certain that more than a thousand Cuyahoga voters went to the polls, waited in long lines, punched out the chad next to Kerry's name, but wound up adding to Bush's "margin ov victory"! See http://www.democraticunderground.com/discuss/duboard.php?az=view_all&address=172x5139 . I'd appreciate your comments.

[emlev]

50. Benedictine site EIRS complaints

I don't see anything in here that looks particularly useful, but maybe you will:

Cleveland 04-F, 04-N
Benedictine High School (2):

032343
11/02/04, 7:07 AM PST
Long lines
BENEDICTINE HIGH SCHOOL, MLK DRIVE, CLEVELAND, CUYAHOGA County, Ohio
ONLY 15 OF 17 POLLING MACHINES ARE WORKING
11/02/04, 7:16 AM PST
Other ballot-related problem; Long lines; Other polling place problem
15600 Terrace Rd., East Cleveland, Cuyahoga County, Ohio
Long lines (excess of 30 min.). Voting location is too small, very hot inside. Poll workers not well trained, slow. Voters report that if they make a mistake or have a ballot problem, the poll workers are reluctant to provide a replacement ballot.

044052
11/02/04, 2:27 PM PST
Provisional ballot problem
Benedict King on MLK, Cuyahoga County, Ohio
No auto books or stickers for Prov. Ballots.

Source: http://epc.voteprotect.org

[AirAmFan]

52. Great source, thank you! First-hand observation of long lines is

quite useful in documenting the existence overcrowded conditions I theorize must be associated with "caterpillar ballot crawl".

[emlev]

59. Glad my many hours of work were useful.

It's tedious work correlating EIRS reports with polling place locations. I've done it for a few counties. If there are other polling places/precincts you want to see the EIRS complaints from, you're welcome to PM me. There aren't complaints from every precinct, of course. And most of the complaints don't specify precinct.

[EuroObserver]

56. Caterpillar Ballot

>>In the suburbs, it was very common for a precinct staffer to accompany each voter to a punch machine, to insert the punchcard for the voter, and then to draw the curtain closed. But at many inner city locations, such as Benedictine High on Martin Luther King Drive, voters were left to negotiate hours-long lines to punch out their chads on their own.<<

You're saying that people had to queue for hours to obtain a punch card, then queue again, punch card in hand, for a machine to punch it on (if they were not told to go back and join a different precinct's queue)?

Incredible.

[AirAmFan]

63. I'm not sure about this specific point. But I'm surmising it must have

taken much longer for people to actually vote on all the dozens of contests on the ballot than to get their name checked against a registration list and be given a ballot.Did you see the image of a Cleveland ballot at the URL I posted in #18? There are dozens and dozens of contests and referenda that a voter has to decide either to ignore or to take a long time to understand. If I'm remembering correctly, I think Chicago holds the record for number of separate voter choices on a punchcard ballot, at 88!

This is why voting machines in the US can't be replaced by reliable paper ballor systems. Democracy is being strangled here by a massive excess of pseudodemocracy. I believe it's time to cap the number of ballot initiatives, elections for judges, etc that can be on the ballot in a Federal election.

And there should always be a way for people to vote a "straight Party line ticket" to get out of the voting booth quickly. People should be able to tick one box, press one button, or pull one lever to vote for every Democratic, every Republican, every Green, etc at one fell swoop. Republican politics got a big boost in every state where "straight Party tickets" were eliminated, generally decades ago. They want the ballot to be long and confusing to intimidate first-time and poorly-educated voters.

[rdmccur]

73. Another reason I surmise

is that (in general) there are more registered Democrats than Republican.

[peacetalksforall]

20. An amazing thing about Ohio is what you would assume...

If any entity was plotting the theft of Ohio. they would do it exactly as it was done...with every possible scheme to provide variety to make it more difficult to find the theft, especially in a short time.

They arranged theft before the election, on the day of election, after the election, on a variety of machines from a variety of sources, lack of information - dis-information - contradictory information, late information, stupid information.

All levels of executives, officials, workers. With all varietys of trust.

It's a natural that they would scatter code, but they would have to keep a theme to track it.

The pity of Kerry's concession is that there was no arrangement to stress and enforce the sanctity of the data. So, we're left with stuff that thieves can twist and hardly anyway to get them under oath. Promise to never let the House report die. We must still keep working. Glad you're still working, minvis.

[DearAbby]

24. Kick

[minvis]

30. Still Working

Thanks. I can't stop working on this. I'm obsessed. I know my wife thinks I'm crazy continuing this, but there are just too many irregularities for me to chalk it up as "Oh well, we lost".

[rdmccur]

31. Same here.

[rdmccur]

74. I'm feel fairly sure that Rove

knew about something that would go *'s way. This implies that there was something contrived (and specifically in certain key states such as Ohio). I don't believe that the perpetrators relied on only disenfranchisement(e.g. long lines) and confusion at polling places. If they were going to this much trouble to perpetrate fraud they wanted a high confidence it would pay off with minimal risk of detection.

[BamaBecky]

25. wow! Can we steal one of those MACHINES & check the code? n/t

[VAMom]

64. No, don't steal it!

Just "borrow" it for a while.

[BamaBecky]

67. that's what I meant, but they are not going to allow us to BORROW it, Ha!

[Niche]

28. This is 'names of candidates' from Ohio Constitution...

§ 02a Names of candidates on ballots

The names of all candidates for an office at any election shall be arranged in a group under the title of that office.

The general assembly shall provide by law the means by which ballots shall give each candidate's name reasonably equal position by rotation or other comparable methods to the extent practical and appropriate to the voting procedure used.


At any election in which a candidate's party designation appears on the ballot, the name or designation of each candidate's party, if any, shall be printed under or after each candidate's name in less prominent type face than that in which the candidate's name is printed. An elector may vote for candidates (other than candidates for electors of President and Vice-President of the United States, and other than candidates for governor and lieutenant governor) only and in no other way than by indicating his vote for each candidate separately from the indication of his vote for any other candidate.

(Amended, effective June 8, 1976; SJR No.4.)

http://www.legislature.state.oh.us/constitution.cfm?Part=5&ExpandSections=Yes


POST NOTE: Is this the way the candidates were rotated in all states? and is this the way the rotation has happened in the past?

[AirAmFan]

33. Thank you for this! I was wondering about the statutory authority for

ballot rotation.

The biggest remaining gap in my knowledge about the strange ballot rotation results in Cuyahoga has to to with the decision to rotate PRECINCTS rather than POLLING PLACES. It sure would make sense not to have punch machines set up differently at the same polling place. It would be just as easy to make sure that all precincts assigned to the same location have the same ballot order. There's real potential for confusion and for deliberate mischief the way things were set up in November, at least in Cuyahoga.

[rdmccur]

35. I see you already said

essentially what I did above in a more general way.

[KerryOn]

37. But wait theres more...

If this is true with punch card ballots, then simply run them threw the wrong tabulator, and bingo, same WRONG results.

[truthpusher]

38. makes sense to me...what about other states? n/t

[minvis]

41. No ballot rotation here

I know in Chicago, where I am, there is no ballot rotation. When candidates run their commercials here, they always say punch such and such a number to vote for me. It doesn't matter what precinct you're in. Obviously, multiple precincts have to be located in one place due to available buildings in an area, but it's not a problem when the ballot order is the same.

[rdmccur]

43. Need to track down

the source of the ballot design rotation and precinct numbering scheme as who (pl?) as who assigned polling locations.

[minvis]

45. I would assume the county BOE

But, I'd like to know that as well. Was the Sec. of State responsible for the ballot design rotation?

[Niche]

48. SoS? Anyone else

I'm reading the Ohio laws and it seems that the SoS makes all these decisions... anyone else know details?

[rdmccur]

49. So it was coordinated needless to say.

[L. Coyote]

51. There have been several previous threads on this topic ...

"Uncounted Votes in Cuyahoga County" by R. H. Phillips. How many??

http://www.democraticunderground.com/discuss/duboard.php?az=show_topic&forum=203&topic_id=196518#198585

Another thread parallel with Chaos in Cuyahoga was:

Ohio's 'CATERPILLAR BALLOT': As bad as Florida's Butterfly?

http://www.democraticunderground.com/discuss/duboard.php?az=show_mesg&forum=203&topic_id=147718&mesg_id=147718

Chaos in Cuyahoga:

http://www.democraticunderground.com/discuss/duboard.php?az=show_mesg&forum=203&topic_id=166913&mesg_id=166913&page=

[AirAmFan]

53. Those threads jumped ahead to focus on arcane quantitative issues

and gave short shrift to important conceptual and electoral process issues. That's why I like the current thread and have learned much from it--people have documented aspects of the electoral process in Ohio that escaped notice in the other threads.

The issue I still wonder about most is, exactly how did it come about that Cuyahoga rotates ballot orders in PRECINCTS and not in POLLLING PLACES? Has this been going on just since HAVA passed in 2002? Since Blackwell became SOS? Since Taft became Governor? Since 1976 or before when ballot rotation got into the state Constitution?

[k8conant]

86. I wouldn't call the quantitative issues "arcane"...

vote counting is a quantitative issue, you know.

[L. Coyote]

88. Now you've made me read the dictionary. Websters says

arcane, 1. hidden or secret, 2. understood by only a few, esoteric.

Sadly, the later prevails! If such was not the case, many more people would be very upset.

[AirAmFan]

93. If I had it to do over again, I would have added another adjective to

the sentence you're parsing--"premature".

Quantitative work is much less confusing when there are only a handful of well-reasoned theories and hypotheses posing questions for the data to answer. I find most of what's in this thread much clearer than what's in the other threads, because we're struggling with conceptual and electoral process issues, not with exploratory data analysis lacking theories of what we'd expect to find in the data. I think there's a lot more real communication here than there.

[Joe Chi Minh]

128. Yes, it's interesting in that,

once gross and clearly partisan irregularities have been identified in quantitative terms, this kind of meta level of analysis arguably takes precedence in significance.

[LiberalHeart]

55. Be more concerned Nader was even on the ballot.....

Lucas County reprinted its ballots (except for 6,000 absentee ballots on which Nader's name was crossed off by election workers and a notice accompanied the ballots telling voters not to vote for him) -- this, after the court denied Nader's request to be on the ballot:

http://www.toledoblade.com/apps/pbcs.dll/article?AID=/20041002/NEWS09/410020352

[Bill Bored]

58. Dupe of post #57, just to make sure it's seen.

How about this:

You ensure that the shift is unidirectional. In other words, you allow some of the shifted Kerry votes to go to Bush, but none of the shifted Bush votes to go to Kerry. This way, without even knowing how anyone has actually voted, you know that Kerry's MARGIN will be reduced more than Bush's.

Let's say the ballot orders in 4 co-located precincts are:

4 Badnarik Bush Kerry Nader Peroutka
3 Bush Kerry Nader Peroutka Badnarik
2 Kerry Nader Peroutka Badnarik Bush
1 Nader Peroutka Badnarik Bush Kerry

If you randomly shift ballots upward from any row (precinct) to any other row, Bush will always come out ahead because Kerry will never get any of Bush's votes, but Bush will always get some of Kerry's. Of the shifted ballots, Bush will get 1/3 of Kerry's Precinct 1 votes, 1/2 of Kerry's Precinct 2 votes, and all of Kerry's Precinct 3 votes. Kerry will only get votes from the 3rd parry candidates, but never from Bush, unless you start shifting ballots in the opposite direction, from Precinct 4 downward.

Now, this also explains why some precincts, in this case, precincts 2, 3 and 4, could have more votes than registered voters, right? They are the ones that receive the shifted ballots, favoring Bush.

So, tell me, does this make sense? Could it have been done on the ground by steering voters from, in this case, lower numbered precincts to higher numbered ones but not the other way around? Could the lines for the lower numbered precincts have been engineered to be longer?

Tell me.

[AirAmFan]

61. See post #60 above

[lonestarnot]

62. EEWW y Someone needs to keep on this for sure!

[GetTheRightVote]

65. That would make coding easier to fix, good eye.

:mad:

[nashville_brook]

68. wow, it's like a simple card trick!

[Snivi Yllom]

69. or, it's just in shifting alphabetical order

conspiracy or just the ABCs?

[L. Coyote]

81. Just ABCs, and nothing is wrong as long as other factors are random, but

the other factors are not random. Given 5 candidates, the probability that any one cross-vote will go to one particular other candidate is 0.25. Closer analysis reveals that this probability is not the case.

If there is conspiracy, it is not in the sequence or rotation, it has to be in other aspects, like deciding to have multiple ballot orders at locations, deciding how many precincts will vote at a location in which areas, challenging one precinct but not the other at a location, suffling precincts, combining precincts at locations, the worst performing machines at the highest Kerry voting precincts, etc.

The very fact that Bush and Kerry are adjacent in the alphabetical sequence is itself a factor. Look at the high percentage of locations with only 2 ballot orders. Combine this with the fact that the rotation is assigned by precinct number. Which precincts are combined at a location? Ones adjacent to each other in precinct number!

We have seen this in hindsight. Those organizing the election know this in foresight!

(Spelling edit.)

[L. Coyote]

70. Yes, this is a problem, however the real problems are not the order ...

but having more than one ballot order at a polling location. Only then can cross-votes occur. Too late to change that!

Minvis' post 26. "Ballot Order" states: "What is interesting so far is that many of the polling places that shared two precincts have this rotation"

That's referring to collocation of Kerry and Bush in the same position. When collocation occurs, Kerry votes can be switched to Bush votes, and vice-versa. What needs to be analyzed is the number of times this happens. Is it random?

In Cuyahoga County it is not. This is discussed in detail in the Web site mentioned above, in post 7. It discusses all the ballot order combinations and the probabilities of vote-switching for the possible combinations. Another Web page at the site discusses specific examples in Cuyahoga County and some of the non-random aspects of cross-voting. That study points out "How Kerry Votes were Switched to Bush Votes" and the study continues.

Because of the method of assigning ballot orders to precincts, by numerical order, and because Kerry follows Bush in the order, the probabilities of Bush/Kerry collocation are augmented because adjacent numbers in the precinct sequence are likeliest to be at the same locations. This changes the probability of Kerry votes being switched to Bush votes in Cuyahoga, as that study points out.

Knowing this, the next, broader analysis ought to be whether the number of locations with multiple ballot orders is randomly distributed. If there are many more in Kerry strongholds, then more Kerry cross-votes get switched to Bush than vice-versa. In Cuyahoga County, a Kerry stronghold, only 2.94% of the precincts have only one ballot order when at multiple precinct locations, and only 9.46% of locations have only one precinct. At all remaining precincts, 87.6% of them, cross-voting is possible. Is this the same in Bush strongholds? What about in other Kerry counties?

AirAmFan's post 46 makes an important point. Cross-voting is not random. Overcrowding and long lines are factors, especially if imbalanced. People tend to go to open voting machines in the wrong precinct if their own precinct has none. Also, precinct populations alter the probabilities of cross-voting. This too is discussed in the linked articles above.

An important point not yet made in this thread is that not all cross-voting leaves a 3rd party or non-vote trail. When the major candidates are collocated, there is no trail for cross-voting in one direction. So, when this cross-voting pattern happens in a 75% Kerry precinct, when Kerry votes switch to Bush and Bush votes switch to Badnarik, the number of Badnarik votes will indicate, if cross-voting is random, that 3 times as many Kerry votes getting switched to Bush votes. In a 90% Kerry precinct, this can be 9 to 1.

In the final analysis, as has been pointed out, other factors play a bigger role. Punch cards result in more lost votes than random cross-voting. Suppression loses more vote than punch cards.

This does not mean there was even-handedness on the part of election officials. They were in a position to know all this in advance. It was their job to understand these factors and respond to them. Was that response partisan? Did they act to steal votes or to ensure the right to vote. The differnce may be the difference between a jail cell or the governor's chair!

The Cuyahoga analysis is being accomplished. What is needed is a broader analysis to understand the statewide pattern, a comparison with Cuyahoga on really broad questions. What is the percentage of locations with multiple ballot orders? Are these random patterns, or do they correlate with candidate support levels?

For a good breakdown of the Cuyahoga data in summary format, download the spreadsheet Summary: Cuyahoga County Ballot Order and Precinct Location Analysis from the site linked above. You can see all the Badnarik vote %, etc.
.

[rdmccur]

75. Do you know whether

the information and analyses discussed at these sites has been sent to Conyers, Arnebeck, etc.?

[L. Coyote]

77. Some of the people working with Arneeck know. Dem Comm. ?? n/t

[EuroObserver]

80. kick n/t

[minvis]

95. Cross-Vote Liklihood in Clermont County

Your percentages of one ballot order in combined precincts and polling locations with only one precinct pretty much hold up in Clermont County. 3 of the 60 multiple precinct locations have one ballot order (5%) and 12 of the 72 locations have only one precinct (16.67%). That leaves 57 of the 72 locations as possible cross-voting eligible (79.17%).

I haven't gone through all the multiple precincts yet, but it seems at first glance that more of the 'crawls' are in the 1/4 category Kerry switches to Bush or Kerry switches to Nader than in the 2/3 category. More than would normally be randomly that way. I'll post the actual breakdowns when I'm finished, probably sometime tomorrow.

Also, I'm expecting ballot order information from Butler County in the next couple days.

[minvis]

102. Sorry, my numbers above were wrong

Clermont only had 51 multiple precinct locations and only 64 total polling locations. So the numbers work out this way.

3 out of 51 multiple precinct locations have the same ballot order (5.88%). 13 out of 64 polling locations have only one precinct (20.31%). By the way, that's twice the percentage that Cuyahoga County had. That leaves 48 out of 64 precinct locations that could have cross voting (75%).

[AirAmFan]

107. Re: "important point not yet made in this thread" about Badnarik miscounts

in cases where we can't see expected Kerry votes disappearing into the "black hole" of "Disqualified":

LC: I know you have high hopes for these cases, but are they being realized? I suspect not. I've just posted details of my conservative calculation of at least a thousand net Cuyahoga Kerry votes lost to "Caterpillar Ballot crawl". I'd be grateful for your comments. Here's what I say about "Calculation Class C" configuarations of ballot orders, which contributed very little to my 1000 estimate:

"In any case, a formula for cases 25 through 28 of the table ("Calculation Class C" is on shakier ground and less reliable than the formula used for cases 4 through 24 ("Calculation Class B"). In a way, we're like the motorist who lost car keys in a dark parking lot, and looked for them under lampposts because the light was better, not in the dark places where the keys were more likely to be. But let's press on."

See http://www.democraticunderground.com/discuss/duboard.php?az=view_all&address=172x5139

[Bill Bored]

83. This is all very interesting but

Badnarik and Peroutka got a total of about 27,000 votes statewide.

L. Coyote has found a correlation between Kerry's percentage and the percentage of undervotes, but statewide, there were only 93,000 undervotes altogether -- comparable to 2000.

Some of the 3rd party votes had to be legit too right? In 2000 the Libertarian, Browne, got 13,475 -- comparable to Badnarik's vote in 2004.

Nader on the other hand got 118,000 votes in 2000, and this year, when he wasn't really on the ballot in OH, there were 93,000 undervotes. But there were 94,000 undervotes in 2000, when Nader was on the ballot. If votes were switched to the crossed-out Nader, you'd expect to see more undervotes in 2004, unless there were a lot more DREs in 2004, which aren't supposed to allow as many undervotes. (I think there are.)

In any case, unless the 2004 undervotes went 100% to Kerry, the only way that ballot rotation could affect the 2004 outcome would be to find more votes switched from Kerry to Bush than the other way around. This is why I proposed a method by which this could have been done using ballot rotation and the redirection of voters by operatives on the ground, or by consistently unequal line lengths, in a specific direction (non-randomly) to switch votes from Kerry to Bush, but not the other way around. I don't know if this happened, but it seems that without this, or some actual hacking of the tabulators, there wasn't enough to change the outcome, was there?

A proper hand recount of all the paper ballots in the state would show what happened, so why don't we just get this done? Is it the money or are there other obstacles at this point?

[L. Coyote]

89. More analysis is needed to define the number of cross-votes, esp...

in other counties. In Cuyahoga we are close to knowing. And, yes, you are correct, it is not enough to change the outcome. Kerry probably knew this much on Nov. 3 when he woke up.

But remember, only some of the cross-votes show up in the 3rd party columns, and the stronger the Kerry support, the smaller the proportion of thse will be to the actual cross-vote total. Some just switch a Kerry vote to a Bush vote, or vice-versa.

Cross-voting is just one part of the puzzle. And it needs to be remedied once and forever. More importantly, suppression must end, and equipment that actually counts all the votes must be provided to minority areas too.

Just because one factor does not have the potential for changing the outcome is insufficient reason to stop analyzing the election. Everyone should keep working, studying, learning new skills, etc.... Many surprises lie ahead, hidden from obvious view.

[Bill Bored]

104. I'm with you!

After all, a few votes swapped per precinct on average is all it would take to get 100,000 vote margin.

[rdmccur]

90. In the recent Madsen update,

Madsen notes in the beginning that
Datamax and ChoicePoint used their government contracted databases to
red line black and democratic districts in OH and FL to suppress
minority votes in a variety of ways. Manipulation based on the candidate order of punch card ballots in various precincts may be one way this was done. What connections, if any, have these companies in Ohio?

[rdmccur]

91. I googled Choicepoint + Ohio

and came up with (amongst others) this: http://www.white-knights911.com/Dove/edove1040.htm

Anyone else been to this site?

[Bill Bored]

94. Uh, wasn't ChoicePoint the co. that compiled the FL "felons" list? n/t

[rdmccur]

97. Yes

[rdmccur]

96. Posted on Kos

http://www.dailykos.com/story/2005/1/9/194437/8201

[thanatonautos]

99. What a nightmarish story. I hope this guy makes out a proper affidavit.

This should get into the Federal Lawsuit if it's true.

[rdmccur]

98. kick!

[EuroObserver]

100. kick n/t

[L. Coyote]

101. kick, n/t

[minvis]

103. Union Township E Precinct

This precinct is shared with four other precincts - Union Township D, Union Township H, Union Township L1L and Union Township M1M. Here is the breakdown in ballot order for those precincts:

Union Township D - Kerry, Nader, Peroutka, Badnarik, Bush
Union Township E - Nader, Peroutka, Badnarik, Bush, Kerry
Union Township H - Bush, Kerry, Nader, Peroutka, Badnarik
Union Township L1L - Badnarik, Bush, Kerry, Nader, Peroutka
Union Township M1M - Bush, Kerry, Nader, Peroutka, Badnarik

Notice that a potential Kerry vote in Precinct E could become a Bush vote in Precinct D. I'm not saying that this happened here, but just that it was possible.

Also, on a sidenote, Jamboi in the Kos posting is correct that Nader was still on the ballot in Clermont County. I have the copies of the actual ballots for each of the precincts and Nader is listed on all of them.

[rdmccur]

105. Kick!

[AirAmFan]

113. This is what I called "Case 9" for Cuyahoga. Calculation of miscounts

should be straightforward. See http://www.democraticunderground.com/discuss/duboard.php?az=view_all&address=172x5139 and expecially posts #10 and #8.

[rdmccur]

106. Kick!

[nolabels]

111. For Pete's sake already, this is not a Democracy, what are you thinking?

Don't you get it yet? You have no choice, you let Bushco and the corporate whores have it or your choice goes to them anyway.

It's not that hard to figure out, so just get over it already :shrug:

[goclark]

112. Kick for the Republic!

[rdmccur]

114. Kick!

[minvis]

115. Another Discovery

As I've been putting the Clermont info. into Excel, I came across an interesting thing about the ballot order and its precincts. When I put the precincts in numerical order, according to their BOE's own system, I noticed that the first precinct (Loveland City A) had the ballot order Badnarik, Bush, Kerry, Nader, Peroutka and the second precinct (Loveland City B) had the order Bush, Kerry, Nader, Peroutka, Badnarik and so on for the rest of the precincts. In other words, from Precinct 1 to Precinct 191, the ballot orders went as follows: 1,2,3,4,5; 2,3,4,5,1; 3,4,5,1,2; 4,5,1,2,3; 5,1,2,3,4 etc.

Now, of course, the precincts that shared polling locations did not follow the numerical precinct order. They jumped around a bit, but that got me to thinking. If I know what the ballot order is going to be in each precinct and I know that many precincts need to share polling locations, I can possibly group the precincts that benefit my candidate the most, while making the third-party votes not appear too large. The best configuration would be one where the ballot crawls either 1 position or 4 positions. This makes a Kerry vote made in the wrong precinct either a Bush vote or a 'Disqualified' vote.

Guess how many precincts had the crawl position of 1 or 4 in Clermont County? 133 out of 175.

[rdmccur]

116. I think you better send

this to Conyers! Maybe to Arnebeck as well.

[AirAmFan]

117. So Clermont's ballot rotation was just a tad more sophisticated than

Cuyahoga's (see post #18 above). Cuyahoga's ballots were assigned BWKDP BWKDP BWKDP ...

Neither of these schemes really is random. It would be more accurate to call them "serpentine". But then again it's comforting to know the algorithm for precinct ballot orders apparently did not use the expected proportion of Kerry votes.

I doubt there is much design to the clustering of precincts in Cuyahoga--the "caterpillar ballot crawl" payoff just is not that big, even with massive long lines and confusion as in the African-American areas of Cleveland on Election Day. In the absence of long lines and confusion systematically concentrated in pro-Kerry areas, "caterpillar crawl" would have little effect on election results. See my post in the Ohio state forum.

Has anyone heard reports of extremely long lines on election day in pro-Kerry sections of Clermont?

[minvis]

119. Pro-Kerry precincts

Bush won all 191 precincts in Clermont County. The only precinct that was even close was Felicity Village. Bush won 124 to 120. Felicity Village shared a polling location with Franklin Township A, C & D. The polling place was New Felicity High School Gym. Here's the ballot order for each of the precincts.

Felicity Village: Nader, Peroutka, Badnarik, Bush, Kerry
Franklin Twp. A: Peroutka, Badnarik, Bush, Kerry, Nader
Franklin Twp. C: Bush, Kerry, Nader, Peroutka, Badnarik
Franklin Twp. D: Kerry, Nader, Peroutka, Badnarik, Bush

[AirAmFan]

120. This is what I call Case 7b, configuration WKDP in post #10 at

http://www.democraticunderground.com/discuss/duboard.php?az=show_topic&forum=172&topic_id=5139 . You should be able to calculate and re-allocate miscounts at that voting location very easily, using a variant of the the formula in post #8 in that thread. (post #11 if you have tallies for miscounts). But, since Bush won that location by a relatively small margin, there should be a relatively small change in his net margin due to miscounts there. When the major vote splits fairly evenly among major party candidates, "caterpillar ballot crawl" is unlikely to affect election results significantly.

Your statement that Bush took all 191 precincts and therefore all Clermont polling places means to me that you are using a model likely to show relatively few miscounted votes, and likely to show losses for BUSH, not losses for Kerry. The candidate most people intended to vote for is the one who loses the most votes when things go awry and votes get miscounted. And places with the long lines and chaotic conditions necessary for "caterpillar crawl" tend to be in poorer communities starved for election-day resources. In well-off pro-Bush areas, I understand it is common for a precinct staffer to accompany each voter to a booth, insert the ballot into the machine for her, and draw the curtain.

What do the distributions of polling-place proportions look like for Badnarik, Peroutka, and "Disqualified"? Without dozens of outliers in such distributions, "caterpillar crawl" analysis is unlikely to find many miscounted votes. I haven't examined all the thousands of theoretically possible configurations of ballot orders at polling places, just those 101 different configuations that actually occurred in Cuyahoga's clusters of sizes 1 to 10. But all of the patterns I've looked at made it fruitless to look for shifts of votes between major party candidates unless there were also proportionate shifts of votes to Badnarik, Peroutka, or Disqualified at suspect locations.

Now, there are many other avenues by which Kerry may have lost votes in Clermont. You may want to apply other models to your data, in addition to the "caterpillar crawl" model . And you may want to broaden your analysis to other counties where Kerry took a significant number of voting locations.

[L. Coyote]

122. I have to take exception to a conclusion you make.

You write: "When the major vote splits fairly evenly among major party candidates, "caterpillar ballot crawl" is unlikely to affect election results significantly."

Actually, there are some unstated assumptions here. Most importantly, you are assuming a random pattern in cross-voting, that is, equal amounts of crossing in all possible directions and combinations.

Take the example of a location with two ballot orders and 2 precincts as follows:

Precinct A: bBKdp
Precinct B: BKdpb

Precinct A Kerry voters crossing to vote in precinct B punch out or mark the second position, a Bush vote in their own precinct. Bush voters crossing the same way do not vote for Kerry. This is a one-way inequality. Kerry does not gain by a Bush A to B cross-vote, but Bush does gain from a Kerry A to B cross-vote.

First, the non-randomness of the ballot order combinations must be assessed. Even then, there is no way of detecting how many Kerry voters voted for Bush in a A to B cross-vote. Anomalous 3rd party votes for Badnarik only indicates how many Bush voters crossed.

If a conspiracy is indicated by the non-random distribution of combinations, the possibility that the conspiracy extended to the ground, to making best use of the non-random distribution should not be ignored.

This all needs much more study before dismissing the possibility of impact just because one candidate or the other has a majority of votes at the location.

[L. Coyote]

121. The pattern of clustering of precincts is important and impacts switching

and switching probabilities. It is important to remember the details, and not give a blanket "doesn't matter" to the issue because the effect is not grand.

At the time the precinct clustering decisions and reassignments were made, there still existed the possibility of a 300-some-vote margin like Florida 2000. Remember, every switched major candidate vote counts twice (minus 1 for Kerry and plus 1 for Bush, or vice-versa).

It is very significant, statistically, that so many more crawl 1 and crawl 4 collocations occured. This is not random.

And it is important to distinguish the two precincts at such locations. Cross-voting from one switches a Kerry vote to Bush, from the other does not. This knowledge can be manipulated on the ground to ensure only one-way cross-voting by unscrupulous individuals.

We need to know if this pattern, more crawl 1 and crawl 4 collocations in Clermont and Cuyahoga, repeats in other counties. If so, it is likely other than coincidence.

[AirAmFan]

123. I think I finally may understand where we diverge in looking at ballot

rotation. Your statement, "it is important to distinguish the precincts at such locations", indicates you may be making different statistical assumptions about the process leading voters to "wrong" machines.

So that I could derive a simple formula for estimating the total number of miscounts at a location (posts #8 and #11 in my thread), I assumed

(1) All possible wrong votes for Kerry supporters are equally likely, and

(2) All possible wrong votes for Bush supporters are equally likely.

These assumptions allow me to simply add up the vote-gain coefficients gBK+gPK+gDK and gBW+gPW+gDW, without putting individual coefficients on the gXYs. It sounds as though you want to put coefficients on them. But where do you get the coefficients, so that you can churn out numerical estimates?

[L. Coyote]

124. This is certainly a difficult problem because there are no overt clues in

the vote record, no non-votes, no 3rd party votes. Some deductions might be possible from vote results, given adjacent precincts tend to vote the same. But this will not be valid with small sample sizes. An overarching trend would be convincing of something being wrong.

Even with the assumption that crosss-voting one-way can be used to quantify cross-voting other ways, you still need to use a probability index to determine the true multiplier. If in 2 ballot/2 precinct locations the number of crawl 1 and crawl 4 ballot collocations is greater that the 2 and 3s, then the probabilities of major candidate switches increases, and the evidentiary trail of non- and 3rd-party-votes is proportionally diminished. This is the case in Cuyahoga.

Another factor is the actual demographic inequality of the collocated precincts. The number of potential cross-voters in each precinct is a factor in equations assuming equal cross-voting. How many are in precincts with a 1.0 probability of a Kerry to Bush switch, how many are in the precincts with a Kerry to disqualified switch probability. Only after this is known can an appropriate multiplier be arrived at for the location (or subset) in question.

In locations with 2 ballot orders and 3 precincts (2\3 in notation form), the probability matrix is more complex and a different formula prevails. At some of these locations, from one precinct you have 1.0 probability of a Kerry to Bush switch, at others you have a 0.50 probability because of KKB collocation instead of KBB. With KKB Kerry voters can cross-vote for their own candidate, with KBB they cannot.

And it just gets more complicated with 4\4, 2\4, 3\4, etc.

[AirAmFan]

125. Remember Occam's Razor. Some estimates are better than none,

especially when time is short.

It sounds as if you think I used the razor to give the data too close a shave. But sometimes you need to make sweeping assumptions to get the data to tell you anything in time to make a difference in the real world.

Models always can be extended. But Occam's Razor always makes you justify a more complex model with demostrated shortcomings of a simpler one.

[L. Coyote]

127. Actually, I'm showing how to do calculations in place of assumptions.

The analysis in the 2004 Ohio Election - Presidential Ballot Orders and Cross-Voting page and associated Ballot Order discusses all the ballot order combinations, the probabilities of vote-switching for the possible combinations, and the factors which modify the random probabilities.

The known non-random aspects of cross-voting can replace the assumptions. The subdivisions the county according to the numbers of precincts and ballot orders at locations is actually a subdivision to probability subsets. The subsequent sort bases the probabilities of the specific precincts takes this a step further. One precinct at a 2\2 location will have a P = 1.0 for a Kerry vote switching to Bush for example, the other at the same location will have P = 0.0, for example.

For example, one way to approach the baseline (true) Badnarik and Peroutka vote is to find out what that vote percentage is in all precincts that have 0.0 probability of cross-votes going to those candidates.

The individual precinct cross-vote probabilities are an essential distinction in this analysis.

The data clearly shows that cross-voting is not a random factor in Cuyahoga. An estimate based on that assumption will be wrong.

[AirAmFan]

129. Assumptions underlie all calculations. Sometimes they're explicit,

sometimes they're hidden or obscured.

I'm still not quite sure what you're doing with the data and how that differs from what I've done. Once I'm sure what you're doing, I should be able to infer the statistical assumptions underlying your calculations.

[Bill Bored]

118. I have to read this again, it seems so cool, but

I assume you know that in Clermont, Butler and Warren and 9 other counties, Kerry got fewer votes than Ellen Connally, a Black Dem candidate for Ohio Chief Justice, right? And in these 3 counties alone, Kerry lost to Bush by 130,000 votes.

Have you found anything with ballot order in Clermont that would account for this, and if so, how would it affect the Chief Justice race there (Moyer vs. Connally)?

[minvis]

130. Connally Votes

Yes, I am aware of the excess Connally votes compared to Kerry votes in these 3 S.W. Ohio counties. I have looked at the numbers for the Connally/Moyer race and have just begun to look at the ballot order for that race and nothing is immediately apparent. I'll certainly post anything interesting that I find in the future.

[skids]

144. I'm looking at the Clermont returns...

...and none of the telltale signs of crawl in the presidential election are there (3rd party votes, undervotes.)

Not that it isn't worth a look, but if it occured, the BOE would have had to have fudged the return figures.

Butler is a bit more "promising" with one precinct with a lot of 3rd party votes (TRENTON CITY 5EA) and a few with high undervote
(MIDDLETOWN CITY 3AC, MIDDLETOWN CITY 3BI among the highest) which
could be crawl going into the Nader column.

As far as figuring out if swapping machines had anything to do with
the Conally thing, do note that it isn't just the rotation that matters, but the order of the races on the tickets. If a difference in districting inserts a race higher up in the ticket, pushing the Connally/Moyers race downwards, then all votes to races including and below the inserted race are corrupted in very complicated ways.

There's at least one polling location in Cuyahoga that served precincts with different race orders, so I guess it should be watched out for in S.W. Ohio as well. But I doubt it is crawl -- could be a similar systematic error though. I'll eyeball lower ticket results for Clermont and see if anything stands out.

[Bill Bored]

131. They have the crawl position to switch the votes

but how can we tell if it happened?

Are there any locations with a single ballot order that we can use as a control to get Kerry/Bush and Kerry/Connally ratios without the possibility of switching?

Do you have a link to this data or to the other 2 counties?

Thanks!

[Bill Bored]

132. BTW I chked the 2000 results for these three counties.

Gore and Kerry both got about 30% in Clermont.
Warren voted about 28% Dem and Butler about 34% Dem in both elections.

Either this crawl problem has happened before, or we're barking up the wrong tree.

I'd like to see a hand recount of these three counties, or at least one of them.

[L. Coyote]

133. I would look at precinct level data first

Does anyone have precinct level data for these counties for 2004 and 2000?

Conclusions based on county data may be premature.

[minvis]

134. Precinct level data

Butler and Warren County's websites have the precinct totals for 2000 and 2004. Clermont only has the 2004 precinct level data. I think I may contact Clermont again and have them send me the precinct level data back to 1996.

I'm still putting together the data in Excel for Clermont and Butler counties. When I finish I'll let everyone know and you can PM me if you'd like a copy.

[BrightKnight]

126. deleted - n/t

[minvis]

136. Any idea on number of voting machines per precinct

We all know about the shorting of machines in minority precincts in Columbus and Cleveland. Does anyone have information on machine allocation by precinct in the other counties? This came to my mind as I was going through some of the numbers of total votes in these S.W. Ohio counties. A couple of precincts in Butler County had over 1300 votes in one precinct.

If we use the formula that the average voter took between 3-5 minutes to vote, that would work out to anywhere between 12 and 20 voters per machine, per hour. The polling place was open 13 hours. That's between 156 and 260 voters per machine. In order for over 1300 votes to be cast in these precincts, there would have to be at least 6, but more likely 8-10 machines. Of course, this is even assuming that the voting was constant throughout the day which I'm sure it was not in these counties. There was probably the big rushes in the morning and the evening.

I bring this up in this discussion because these precincts in Butler County did share the polling place with others. If I'm the Sec. of State and I already know the ballot order in these counties, perhaps I could group some precincts with the same ballot order in the same location and therefore not have to worry about people going to a voting booth in the wrong precinct. This way I can still put 4-5 machines per precinct but still send the voter to any available voting booth and not have to worry about the votes switching. Obviously, I would only do this in precincts where I am the strongest and it doesn't look too suspicious.

Am I crazy and just grasping at straws here?

[Bill Bored]

138. See this from the other thread, in case it sheds some light.

These three counties strongly favored Bush. Perhaps there are precincts within them that strongly favored Kerry or Gore. Not sure we'd know, unless they were different in 2000, but the locals would know. Are there any heavily minority precincts for example? See my post below:

<http://www.democraticunderground.com/discuss/duboard.php?az=view_all&address=201x7613#7773>

I agree we have to look at the precinct level data. Any ones that strongly favor Kerry can be suppressed just by randomly shuffling some ballots to the wrong machines. Bush comes out ahead even if you don't look at the ballots, as long as you know they favored Kerry. This only works in pro-Kerry precincts with the right crawl of course. Anywhere else, you have to target the switching non-randomly. But why not do it the easier way?

[minvis]

139. Waiting on Butler County info

I should be getting the ballot order info from Butler County sometime next week. Butler has at least 20 or so solid Kerry precincts, mostly from the area around Miami of Ohio University. It was my plan to concentrate on these precincts first for any anamolies. I'd also be interested in seeing if their ballot sequence stays the same as it did in Clermont, i.e. Kerry always is listed after Bush. I'm assuming it will.

[Bill Bored]

140. Awesome minvis!

This could really explain a lot. While these 3 counties are heavily Repub, if there are areas of Dem support, they can be hacked, and all the more likely if there are Repukes running things at the County level.

What do you think of the random switching effect favoring Bush in strong Kerry areas? Has this been brought up before??? I'm so overloaded with all this stuff, I'm not sure who said what anymore! I'm glad there are some like you who are specializing!

Just rereading the thread, I see that AA Fan has suggested this, but I'm suggesting that you get the biggest bang for your buck in locations where you have just 2 ballot orders and heavy Kerry support. This way, any random shifting of ballots will favor Bush and there won't be any change in thrid party votes to tip us off. Right?

[minvis]

146. Looking forward to info on Butler County

I think we'll get a good sample from Butler County. What is also interesting in almost all of Kerry's precincts there is that the turnout was quite a bit lower than the county average. Most of them had turnouts in the 40 and 50 percent range. The county average was over 70%. This tells me that it was much like pro-Kerry precincts in Franklin and Cuyahoga counties that had long lines and would be even more susceptible to voting at wrong machines for their precinct.

[minvis]

137. I have the Excel file ready for Clermont County

In case anyone is interested, I have completed compiling the Excel file for Clermont County with the ballot order groupings, crawl numbers, votes cast, percentages, etc. Please PM me if you're interested in a copy of the file.

Just a couple added comments I have on the data. The one rather large Badnarik vote total of 8 votes in Stonelick Township B does in fact line up in the ballot order (#4) with Kerry's ballot order in Stonelick Township E. Stonelick Township precincts A,B & E shared the same polling location. Secondly, as I mentioned in an earlier post, a vast majority of the 'caterpillar crawls' in the shared precincts were either 1 or 4 which meant a Kerry vote could become a Bush or Disqualified vote. A crawl of 2 or 3 for either Bush or Kerry would have resulted in large third party or disqualified votes something Bush wouldn't want in one of his best counties. Finally, nearly half of the shared precincts were 2/2 (2 ballot orders in two precincts).

[skids]

141. Where is the ballot order information...

...on the county website?

[minvis]

145. No

I had to send away for them. They actually photocopied the actual ballot for each precinct and sent it to me.

[skids]

142. In case folks missed it... cuyahoga down-ticket rotations.

Since this thread floated up, FYI the full cuyahoga ballot rotations including lower races are linked here:

http://www.democraticunderground.com/discuss/duboard.php?az=show_mesg&forum=203&topic_id=285460&mesg_id=285460

[Bill Bored]

143. Another suggestion

3% of the vote in each county was manually recounted. The precincts were selected non-randomly in violation of Ohio law.
See: <www.votecobb.org>

Since these turned to be close, if not perfectly correct, what were the ballot orders in THESE precincts? Has anyone looked at this yet?

[minvis]

147. Recount precincts

I found out the precincts recounted in Clermont county were the 15 precincts with the lowest vote counts. They were as follows with the following ballot order:

Moscow Village - P,b,B,K,N
Monroe Twp. H - K,N,P,b,B
Pierce Twp. H - P,b,B,K,N
Newtonsville Village - N,P,b,B,K
Milford C - K,N,P,b,B
Batavia Vlg. B - P,b,B,K,N
Union Twp. V - b,B,K,N,P
Union Twp. G1G - B,K,N,P,b
Bethel Village B - N,P,b,B,K
Washington Twp. B - B,K,N,P,b
Williamsburg Village A - P,b,B,K,N
Goshen Twp. A - N,P,b,B,K
New Richmond Vlg. B - B,K,N,P,b
Union Twp. F - P,b,B,K,N
Monroe Twp. D - N,P,b,B,K

This breaks down to 1 precinct with Badnarik at the top, 3 with Bush at the top, 2 with Kerry at the top, 4 with Nader at the top and 5 with Peroutka at the top - relatively random.

What is interesting is that none of them have large third party votes, undervotes or overvotes except for Goshen Twp. A with 4 overvotes, but no Badnarik, Peroutka or undervotes. In other words, these precincts were pretty straight-forward. If I were a Democratic member and had the choice of choosing some of these precincts, I definitely would have looked at Miami Twp. L and Y with 6 and 7 Badnarik votes respectively and Stonelick Twp. B with 8 Peroutka votes. These three had the highest third-party votes in Clermont County.

[Bill Bored]

148. Were any of these collocated? n/t

[minvis]

149. Just one

Moscow Village and Washington Twp. B. The crawl was 2 for that collocation. There were no Badnarik or Peroutka votes in either and only 1 undervote in Washington Twp. B.

[minvis]

150. Kerry Precincts in Butler County

I thought while I waited for the ballot order information, which I confirmed today will be faxed to me tomorrow, I would look at the differences between not only the Kerry and Connally votes but also another race as well. I decided on a distinctly local race rather than a statewide race since that would give me an indication of the general voting patterns of the county. I chose the Sheriff's race since that was one of the only countywide races that was contested in Butler County.

We all know that Connally received more votes countywide here than Kerry (about 5300 more), but the Democratic candidate for Sheriff, Dale Richter, also got more votes than Kerry (just under 4000 more). As we know, almost without exception, down ballot candidates from the same party tend to get less votes than the top of the ballot candidates. Next, I tried to look at where they picked up these extra votes. That's when I found something that astounded me.

When I looked at precincts that Kerry won, all but 3 of the 34 precincts, both Connally and Richter received less votes than Kerry. So, all the gain was in the precincts that Bush won.

Any ideas on why this would be the case?

[Bill Bored]

151. One answer would be that in the Bush precincts,

votes were switched from Kerry to Bush while the other races were left alone. But the quesiton is how was this done? Was the ballot order exploited, or something else? If it was ballot order, it coundn't have been done with random shuffling of ballots in a Bush-heavy precinct because that would favor Kerry; it would have to be unidirectional.

Out of curiousity, in the Kerry precincts, was his margin of victory, expressed as a percentage, higher or lower than Connally's?

[minvis]

152. Definitely Higher

Many of the precincts won by Kerry were won by huge percentages, on the order many of the Cleveland precincts. Some of the them were as high as 85% for Kerry. For that same precinct, Connally barely broke 60%.

[Bill Bored]

154. I think that sounds reasonable.

So we are left with a possible scenario in which Bush may have gained votes in Repub counties while Kerry lost them in Dem counties, but not in all precincts. Or something like that?

[floridadem30]

153. kick

[Carolab]

155. Why just punch cards?

Wasn't switching candidate order also being done on optiscan ballots? And were there differently ordered touchscreens too?

[skids]

156. There weren't many opscan in OH.

13 counties, most very small. Small counties will generally have less polling places serving multiple precincts, not a whole lot less though.

Hancock and Clermont were two. They look clean as far as 3rd party votes go.

And 7 e-voting, which had the ballot programmed in it so there wasn't precinct cross-voting there.

Theoretically, optical scan ballots should be printed with the candidates names on them and keyed to prevent cross-voting. Not that a motivated vote stealer couldn't figure out how to screw that up, but...

[Carolab]

157. The reason I asked is that in Minnesota we use optiscan ballots.

Almost exclusively and the SoS is pushing for the AutoMark...

There were some very strange results in the 6th Congressional district here, which happened to be a very nasty race between Patty Wetterling and Mark Kennedy for senate. There were 8,000-9,000 votes for 3rd party and write-in candidates in that district versus totals in just the hundreds for those same candidates in the other 7 congressional districts in the state. These numbers were live on the SoS website at 1:20 a.m. on 11/3/04, then changed a few hours later. From a post above, which talked about throwing Kerry votes to third-party candidates, as a result of switching candidate order on ballots from precinct to precinct, it made we wonder whether there was a different order on the ballots in some or all of the precincts in the 6th congressional district that could have accounted for this. The total number of votes was over 50,000. Later on, those votes were moved mostly to Kerry or to Bush. But without examining the actual ballots, etc., who knows whether the difference was properly allocated? Kerry was supposed to win by a lot more than he supposedly did in Minnesota.

[skids]

158. I guess you would have to research the machines there.

...and find out how the machines are programmed to discriminate between ballots. If you, for example, had a set of precincts that were centrally tabulated, and they have to manually reconfig the machines for each batch of ballots, then ballot order could be a factor if the staff fed the ballots through with the wrong ballot key programmed into the scanner.

That is, assuming ballot order is rotated in Minnesota. If it isn't, then the error was probably due to something else.

But if you want Minnesota analysed, my best advice to you is to sign up on the Minnesota mailing list at uscountvotes.org, and when they start taking precinct level data (my best guess is a week or two and that will start), help them by obtaining and submitting it. Also look around for any groups looking to do an informal hand recount, as in many states, citizens can request that after the election for informational purposes, as long as they raise the money to pay the fees.

Aside from paper ballots, opscan is one of the best technologies to use, as long as it is the "precinct count" variety where the opscan machines are present at each precinct, not the "central count" where they don't get counted until they are delivered to a central location. The reason the difference is important is because in central count, voter's don't get to see whether the machine detects an overvote and correct their ballot.

[minvis]

159. Across the board

The rotation of ballot order was done no matter what type of voting you were using in Ohio, (punch card, opscan, electronic). It was mandated by Ohio law that the order of the candidates had to switch from precinct to precinct. The initial idea was so that candidates at the top of the ballot didn't have an unfair advantage.

[minvis]

160. Just got the Butler County info.

It confirmed by suspicion that the ballot order was not random at all. Butler County did the same thing as Clermont. The first precinct (Fairfield City 1AA) had the ballot order Badnarik, Bush, Kerry, Nader, Peroutka . The next precinct had Bush listed first, then the next precinct Kerry first. This continues all the way to the last precinct. I still have to go through all the collocated precincts and look at the numbers. I'll definitely post what I find.

Also, an added bonus that I wasn't expecting. They also sent me how many voting machines were in each precinct. Knowing what happened in Franklin County, I will definitely look at the Pro-Kerry precincts and see how many machines were allocated there.

[Bill Bored]

161. How come they're so cooperative?

(Not that there's anythng wrong with that!)

[minvis]

162. I don't know

I've been doing all of this through e-mail with the BOEs. I have been extra nice in asking for things. Everytime I ask for something, I always thank them so much for their time. I also think that I'm usually dealing with the assistants, underlings, etc. and not the head of the BOE. They tend to be just office workers, not political hacks.

It was pretty funny when I got the info from Butler County, I e-mailed back with a few questions because the ballot rotation information was not clear like it was with Clermont. Clermont gave me photocopies of the actual ballots for the precincts, so I could actually see the order. Butler had it in a spreadsheet kind of format with the shorthand of 1-5, 2-5, etc (1-5 meaning Badnarik was first, 2-5 meaning Bush was first). When she wrote back she was amazed that I figured out what that meant. She said most people don't understand it the first time.

[minvis]

163. Voting Machine Info from Butler

After I looked at all the information that Butler County faxed to me, I noticed that not only did they send me how many machines were in each precinct, but also broke it down by how many were originally assigned to each precinct as of Sept. 22, 2004 and how many more were assigned and where after the voter registration numbers were in in mid-October.

There were 1924 machines originally allocated for all the precincts. An additional 385 machines were then added to certain precincts after the voter registration numbers for each precinct came in making a total of 2309 machines. The total number of registered voters for the county was 238,022. That works out to an average of 103.1 registered voters per machine. Contrast that to the known problems in Franklin County, their average was over twice this. I believe it was close to 230 registered voters per machine.

Much like Franklin County, however, the allocation of machines for Kerry and Bush precincts was not even. In fact, the largest precinct was Liberty Twp. 4DI with 1834 registered voters. They got 11 additional machines on top of the 9 originally allocated for a total of 20 machines. If you look at one of Kerry's precincts with the highest number of registered voters, Oxford 4HB, with 1287. They got 2 additional machines on top of the 8 originally allocated. That makes 10 machines for 1287 potential voters, an average of 128.7 voters per machine. Liberty Twp. 4DI's average was 91.7. As an aside, guess what the turnout in each of these precincts was. Liberty Twp. 4DI - 86.15%, Oxford 4HB - 51.59%.

[minvis]

164. Collocated Precincts

I first looked at Trenton City 5EA precinct which had the 30 Badnarik votes. It was collocated with Trenton City 4EJ precinct. The ballot order for these two precincts were as follows:

5EA: Kerry, Nader (DQ'd),Peroutka, Badnarik, Bush
4EJ: Badnarik, Bush, Kerry, Nader (DQ'd), Peroutka

I then contacted the Butler BOE to ask if Nader was, in fact, taken off the ballot because in Clermont County, right next to Butler County, he was not removed from the ballot. The Butler BOE did confirm that he was. With that information, now look at what the rotation is for these two precincts.

5EA: Kerry, Peroutka, Badnarik, Bush
4EJ: Badnarik, Bush, Kerry, Peroutka

So, close to 30 Kerry votes in 4EJ became 30 Badnarik votes in 5EA.

Next, I looked at where the highest percentage of undervotes appeared and, lo and behold, they were in Kerry precincts or very competitive precincts that Bush "officially" won. Here is the breakdown for the precincts with over 3% undervotes. The county average, by the way, was 1.29%.

Precincts Undervote Pct. Percent Vote
Middletown City 3AC 7.27% 54-38 Bush
Middletown City 3BI 6.63% 83-10 Kerry
Middletown City 3BC 5.12% 80-15 Kerry
Hamilton City 2EA 4.58% 53-42 Kerry
St. Clair Twp. 4KE 4.17% 61-35 Bush
Oxford Twp. 4HO 3.95% 59-36 Bush
Middletown City 3AI 3.75% 54-42 Kerry
Wayne Twp. 4MC 3.65% 65-31 Bush
Middletown Ctiy 3BF 3.57% 81-15 Kerry
St. Clair Twp. 4KG 3.50% 70-26 Bush
Ross Twp. 4JE 3.41% 70-27 Bush
Wayne Twp. 4MF 3.23% 68-29 Bush
Hamilton City 2BB 3.18% 85-12 Kerry
Middletown City 3AJ 3.10% 59-38 Bush
Middletown City 3DC 3.06% 52-45 Bush
St. Clair Twp. 4KA 3.04% 55-42 Bush

What was really interesting, however, was that in Middletown 3AC, with the highest undervote percentage of 7.27%, there was even more undervotes, 37, in the Presidential race than in the Senatorial race, 20. There was no other precinct in the county that had more undervotes for President than another race. Add on top of that the ballot order for 3AC and its collocated precinct 3AB is as follows.

3AB: Peroutka, Badnarik, Bush, Kerry, Nader (DQ'd)
3AC: Badnarik, Bush, Kerry, Nader (DQ'd), Peroutka

A Kerry vote in 3AB becomes a DQ or undervote in 3AC.

[Carolab]

165. Have you shared this information?

Especially in light of the new Conyers' FBI request?

[minvis]

166. I will

after I finish getting everything in my Excel spreadsheet. I've done the same thing with Clermont County as well, so I was going to send it to them when I had completed both. I'm also waiting for uscountvotes.org to be ready to accept my information as well.

[findTruth]

167. Include info about votes for C. Ellen Connally

Do you have all the votes counting at precinct level including
those for C. Ellen Connally? She got total of 19,129 more votes
than Kerry in Auglaize, Brown, Butler, Clermont, Darke, Highland,
Mercer, Miami, Putnam, Shelby and Warren counties. I'd like to
figure out how the votes might be "migrated".

In Ohio:

Total votes cast -- 5,722,391
Total votes for president -- 5,627,903
Total undervotes -- 94,488

Based on exit poll:
Kerry's votes would be around -- 2,981,365
Bush's votes would be around -- 2,729,659

Final count:
Kerry's votes are -- 2,741,165
Bush's votes are -- 2,859,764
Third parties votes -- 26,974

If Kerry's votes were reduced by 240,200 and Bush' votes were
increased by 130,105 based on the exit poll, I'd like to see
how the rest of 110,095 votes were migrated to third parties or
non-president candidates, which caused the 94,488 undervotes and
Connally's total of 19,129 more votes than Kerry in the above
counties.

I think examing only the votes for president might not give us
the whole picture...

[minvis]

168. Connally votes

To be honest, I'm not that convinced of the larger Connally vote being proof of fraud. Don't get me wrong. I believe dirty tricks and vote theft was done in Ohio and other states as well, I just don't think a down ticket candidate getting more votes than the Presidential candidate is automatically a sign of fraud. What many of you may not know and I didn't know it either until I saw copies of the actual ballots is that the Connally/Moyer race officially was non-partisan. There is no notation on the ballot of who the Democratic or Republican candidate is in that race.

I did, however, look at not only Connally's precinct vote totals, but also every other Democratic candidate running in Butler County as well. I compared their precinct vote totals to Kerry's precinct vote totals. What I did find that was interesting is that in the precincts that Kerry won, every down ticket Democratic candidate received less actual votes than Kerry. That is just what one would expect. However, in the Bush precincts, many of the down ticket Democrats got more actual votes than Kerry. I'd definitely say that the number of votes siphoned off from Kerry was sizable, but not the size of the Connally differential, perhaps 1,000 or so.

I've got the information on an Excel spreadsheet, so feel free and PM me and I'll be happy to send you the info.