Fairfield County: Ballot mix-up causes Bush to lose votes

[jmknapp]

Found a case of ballot mix-up in a heavily Republican county in Ohio.

The precinct level results for Fairfield County moslty look clean, but there was clearly major confusion in this area in Violet Township. Several precincts showed high totals for Badnarik and Peroutka, as well as for “Disqualified,” that is, the Nader position.

One such example: Violet-R, Violet Twp Administrative Building. One other precinct, Violet-O, shared that polling place.

Out of 473 ballots cast for Violet-R, 43 votes went to Peroutka, and 29 went to Badnarik. These totals are of course highly unusual, 15% going to the third-party candidates. In the other precinct, Violet-O, 540 ballots were cast with 29 going to Disqualified and 10 to Peroutka. Also highly unusual totals, pointing to a ballot mixup.

The ballot orders were probably:

Vio-O: BKdpb
Vio-R: pbBKd

B Bush, K Kerry, d Disqualified, p Peroutka, b Badnarik

So ballots cast on O machines but run through R counting machines would map Bush->Peroutka and Kerry->Badnarik. Therefore, looks like 43 Bush ballots went from O to R, along with 29 Kerry ballots. By wiping them off the face of the two-party earth, Bush lost 43-29 = 14 votes margin.

Conversely, ballots cast on R machines but run through O counting machines would map Bush to Disqualified and Kerry to Peroutka. Looks like 29 Bush ballots went that way, along with 10 Kerry ballots. Bush loses 19 votes margin.

In all, about 111 ballots (about 11%) were evidently spoiled by mixing them up at this polling place. I say “about,” because there may have be onesie-twosie votes for Badnarik/Peroutka legitimately, but nowhere near these numbers, and certainly not for Disqualified.

[48pan]

1. Let me understand this

Is this saying that these ballots were already counted for someone other than Bush and he won't lose any of his current total? Too bad if true.

[MarkusQ]

2. Thanks Joe! For those of you tuning in late...

...it all relates to this thread started by Iceburg and this one started by AirAmFan, which are attempting to unravel / calculate the number of votes lost due to multi-precinct polling places. At first blush, it looks like a significant number of Kerry votes (on the order of 50,000) were lost this way. But the exact consequences are tricky to puzzle out. One data point which several people had pointed out would be useful was the corresponding effect from Bush-strong precincts. jmkapp has just provided such a case.

Stay tuned, or dust off your slide rule & pitch in...

--MarkusQ

[BlueDog2u]

3. Nice work

How many votes statewide did Kerry lose through this kind of mixup? MarkusQ says below about 50,000. That's a lot. How many did Bush lose? probably not as many, but this example shows how what DU and the Glibs are doing is authentically bipartisan in spirit and should concern all Americans, except of course those who endorse or practice deliberate fraud.

[jmknapp]

4. One Kerry voter's recollection

Turns out, I know someone who voted in Violet-R and I sent him an email telling him his (Kerry) vote might have been miscounted and asking if he remembered anything about the situation at the polling place. I just got this reply:

Well, as a matter of fact, Joe, I am in the Violet-R precinct.
When I walked in, there was no line, but all the punch card stations were occupied. The first one that became available was one for the other precinct and I was directed to it. I forget which one it was.
It may well have been O. (This was the first time I had voted at this location or in this precinct.) I remember it was the station in the NW corner of the room, and the R stations were along the west wall and the others were along the north wall. At this point I turned around and remarked to the poll worker that I was in Vi-R and asked if it would make any difference voting on a station marked for the other precinct. I was told it would not make a difference. So, that is where I cast my vote. The time was between 11AM and 12N and the location was the Violet Township office at Rustic and Stone Creek Drive. I do believe Bush and Kerry were the top two names on the ballot at the station where I voted.

Frankly, this is sort of scary. I think you're on to something.
In fact, it looks like you pretty much got it figured out.

[AirAmFan]

5. Would you please provide more information on the tallies

at this location?  THANK YOU for posting this.

With more information, I could apply the general formula for
counting and allocating "caterpillar crawl"
miscounts I've posted to reply #45 at
http://www.democraticunderground.com/discuss/duboard.php?az=view_all&address=203x147718
.

I've put single question marks next to the data values I've
inferred from omissions in your post, and double question
marks next to the data values for which I see no information
so far.

Precinct Badnarik Bush Kerry Disqualified Peroutka  TOTAL
-------- -------- ---- ----- ------------ --------  -----
Vio-O       0?     ??   ??       29         10       540
Vio-R      29      ??   ??        0?        43       473

[jmknapp]

10. Tallies here

Precinct Badnarik Bush Kerry Disqualified Peroutka TOTAL
-------- -------- ---- ----- ------------ -------- -----
Vio-O 0 326 174 29 10 540
Vio-R 29 241 156 0 43 473

Of interest also is the Sheriff's race, where the opponent to Dave Phalen was Disqualified.

Precinct Phalen DQ
Vio-O 401 33
Vio-R 317 56

Since this is a two-person race, this perhaps gives a more positive measure of how many ballots were mixed up? Countywide, 686 votes went to the disqualified sheriff slot. Assuming that half the time the ballot order wasn't different when a mix-up occurred, the number of actual swaps would have been about double that, 1,372. That's about 2% of all ballots.

On the other hand, I like to know if it was possible to vote for Disqualified directly, or if the punch card template was designed to prevent that somehow. If it was possible to vote for Disqualified, maybe some voters punched it for a protest vote or something.

The problems are concentrated in Violet Twp. Out of 69 Disqualified votes for president, 56 were in Violet. Violet represents 28% of the votes in the county, but 81% of the disqualified presidential votes.

[AirAmFan]

6. Your data are BEAUTIFULLY consistent with an equally-likely

voting error hypothesis for the particular configuration of
ballot orders at the Vio-OR location.

Ask yourself, in how many ways could voters at this location
have gone wrong by using a "wrong" machine, and what
would all the mistaken outcomes have been?

Here are the precincts, the ballot orders, and the
wrong-machine KERRY-voter ballot position matches:

Vio-O Bush Kerry Disqual Peroutka Badnarik ---> Peroutka
Vio-R Peroutka Badnarik Bush Kerry Disqual ---> Badnarik

The wrong-machine BUSH-voter ballot position matches are

Vio-O Bush Kerry Disqual Peroutka Badnarik ---> Disqual
Vio-R Peroutka Badnarik Bush Kerry Disqual ---> Peroutka

So there are FOUR possible ways a voter could have gone wrong.
 Notice that the outcome for two of these wrong paths is
Peroutka, while only one wrong path leads to Badnarik and only
one leads to Disqual.

If all wrong paths were equally likely, we would expect to see
equality in the total number of wrong votes cast for
Disqualified and Badnarik, and double that common frequency
for Peroutka.  This expectation is borne out beautifully by
your data:

Disqualified and Badnarik each got 29, and Peroutka got just
shy of twice that number, at 53!

This case gives me epsilon more confidence that the theory
developed in my "Caterpillar" thread is
fundamentally sound.  Your data could have rejected the
equally-likely error hypothesis, but did not.  All the handful
cases I've examined in Cuyahoga so far have been consistent
with this hypothesis, so I'm going to try to develop
statistical evidence that's more than just anecdotal.

Within the next week or so, I plan to construct statistical
tests of hypotheses flowing from my theory and apply them
systematically to data for 1300 Cuyahoga precincts that voted
together in clusters of two or more precincts at the same
pollling places.

I'd appreciate any feedback about this work on the
"caterpillar" thread linked by Markus's second URL
in post #2 above.

[berniew1]

7. Did they use punch cards in Fairfield?

[AirAmFan]

9. Yes, if I'm reading this unlabeled map from the newspaper correctly

[L. Coyote]

8. Cross Precinct Voting -- What is the actual impact?

Definitely read the threads:

Chaos in Cuyahoga? 49,000 Votes Disappear into the Ether ...then found?

and

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

as an intro to this topic. These threads present statistics on non-vote correlations. In Cuyahoga county some interesting patterns have emerged. The pronounced Kerry voting precincts have higher non-vote percentages. For example, in an 90% Kerry support precinct, random cross voting can produce a 9 : 1 ratio of Kerry losses to Bush losses, given all 5 ballot orders in that precinct. In a precinct with only 3 ballot orders, this changes. There is always a candidate who does not lose. Which candidate cannot lose in which precinct makes a difference of up to 9 : 1!!

Naturally, having the machine combination that converts Kerry votes to Bush votes the highest Kerry percentage precincts yields more cross votes than in a 50/50 precinct.

Of 1436 precincts, 154 have only one ballot type (= no cross precinct voting possible). All others can have cross precinct voting. Here are the tallies derived from Joe Knapp's spreadsheet:

# precincts - # Ballots - Count

10	5	1
9	4	1
7	5	1
7	3	1
6	5	2
6	4	3
6	3	1
5	5	4
5	4	22
5	3	6
4	4	28
4	3	26
4	2	6
3	3	106
3	2	43
3	1	2
2	2	176
2	1	18
1	1	134

Total Locations = 581

129 locations have fewer ballot orders than precincts at the location. This provides more opportunity to ensure that a certain candidate will or will not receive cross precinct votes.

[MarkusQ]

11. We should also look at...

1. Other counties. We should, if possible, try to get state-wide data in a unified format.

2. Relationship between ballot order, precinct packing, and expected vs. "actual" ratios between the candidates.


We've got a good mechanism here, that could have been exploited in at least two different ways:

First, by crafting the ballot order & precinct clumping, it would (in an egregious example) be possible to give most of Kerry's votes to Bush without producing more than about 20% windfall to Badnarik. This would have been detected, but more subtle cases might not.

Second, by simply crowding Kerry-leaning precincts, with no tweaking of ballot order, it would be simple to reduce Kerry strongholds toward 50%/50% (not counting 3rd party votes).

Were these things done? Was something else done? I haven't had coffee yet, but I'll be back...

--MarkusQ

[Emit]

12. This might have nothing to do with your research, but

I had seen this article some a couple of days ago about Summit County eliminating 149 of its precincts. The potential relationship that I see to what you all are working on here is that more voters in less precincts in a highly democratic area, plus the potential of ballot order switching, could cause quite a number of votes being switched. Anyway, just food for thought.
http://www.ohio.com/mld/ohio/news/politics/10393379.htm
<<snip>>
Posted on Sat, Dec. 11, 2004


R E L A T E D L I N K S
• Summit County precinct changes
• Election Day criticisms may yield changes

Cuts in precincts add to voter lines

Akron hit hardest by Summit elections board's reductions in polling sites

By David Knox, Carl Chancellor and Kymberli Hagelberg

Beacon Journal staff writers


Since the 2000 presidential election, the Summit County Board of Elections has eliminated 149 of its precincts -- more than any other Ohio county.

Why did Summit get rid of so many precincts -- nearly a quarter of the 624 it had in 2000?

Although the two Democrats on the board remember it differently, the answer given by Republicans, when the decision was made three years ago, was simple: to save money by reducing the need for increasingly hard-to-find poll workers.

The Republicans say they were betting that precincts could be consolidated without inconveniencing voters, thanks to more efficient electronic-voting machines that were expected to replace punch-card machines before this year's presidential election.
<<snip>>
But both Jones and Pry said they understood that the county had no choice but to reduce precincts. ``We were under a mandate from the secretary of state to consolidate,'' Pry said.

But that's not true, said Carlo LoParo, spokesman for Ohio Secretary of State J. Kenneth Blackwell. ``There's no requirement for reductions.''

<<snip>>

[AirAmFan]

14. Thank you--this is an important insight into why so many precincts

are clustered together in polling places.Before I read your post, I suspected that changes in the physical landscape of the county were forcing precincts to be clustered together in a declining number of large spaces available for use by the BOE on one or two days a year. Aging plumbing, heating, ventilation, roofs, and windows eventually cause old schools and churches to be boarded up or torn down. As residential real estate values rise, more and more public and commercial spaces are converted to condos and can't be used anymore as polling places.

I had forgotten about the need for staff and volunteers. In a way, problems stemming from clustered precincts are OUR FAULT: I saw a news report about the increasing average age of election day staff and volunteers at polling places. If we want to see why not every voter had someone knowledgable to insert her ballot into the holder properly, on the correct machine, all we have to do is look in the mirror!

[MarkusQ]

17. I would agree, if it turns out that the effect was equally distributed...

...but if there's significant bias to the clustering, then I suspect "lack of volunteers" is not a reasonable explanation.

--MarkusQ

[MarkusQ]

13. The egregious example fleshed out...

Someone asked, so I'll put it where everone can see it.

Suppose there is an election with four candidates: Able, Baker, Charlie, and Disqualified, that the real race is between Baker and Charlie, that you work for Baker, and that you have absolute control over 1) where the precincts vote, 2) which of the four posible ballot orders they use, and 3) how many machines they use.

Further, and unrealistically, assume that absolutely no one will ever question your actions as long as you follow the rules, and thus you can be as blatant as you wish in rigging the election. How bad can you make things for Charlie?

Here's one stab:

• Take an area in which 90% of the people want to vote for Charlie.
  
• Break them into five precincts (1-5), with 80%, 5%, 5%, 5%, and 5% of the voters respectively.
  
• Give precinct 1 ballot order ABCD and the others BCDA.
  
• Put them all in the same room.
  
• Give precint 1 a single voting machine; give the others four each.
  
• Neglect to tell the poll workers how important it is to keep the precicnts seperate.
  

What does this do?

Well, 16/17ths of the voters will be using machines set up for BCDA, and 80% of them will be using ballots for ABCD. Although 90% of them think they are voting for Charlie, over 90% * 80% * 16/17 > 67% of them will actually vote for Baker. The actual vote for Baker will be more than that, when you take into account the other permutations, but I have to go change a diaper, so I'll leave the details as an exercise for the reader...

--MarkusQ

P.S. (On edit) 5/6ths was wrong, corrected to 16/17ths

[L. Coyote]

16. Aaahh yes! The many ways to legally fix an election no recount can

possibly detect or alter! Let's list a few:

Voter suppression.
Scrubbing the voter rolls.
Overloading the "enemy" precincts with too few machines.
Placing machines with high spoilage rates in "enemy" territory.
Buy new E-Vote machines for the rich.
Spend a billion to create a negative image of enemy.
Create a situation of utter fear of foreign terrorists.

I could rant on!. Back to the topic...

Add to that, arranging the multiple machine precincts so that cross-voting benefits your candidate.

The situation is more complex, but that's the idea being investigated. The machines have to have equal numbers of ballot orders. However, I think that is on a county level to stay within the letter of the law, that is to keep the election fixing perfectly legal! So there certainly is some leeway to manipulate.

[anamandujano]

15. reminds me of-- "my mother in law lost money with Enron"

Can't look at any more numbers. I suspect we'll be hearing news of the few glitches that harmed Bush trumpeted throughout the lands. Then they can say "no harm, no foul", just like Bush did with Enron.

And Martha Stewart will have a few more dems to keep her company in prison.

[MarkusQ]

18. I think you may be missing the point of this thread...

...even if there is systematic bias, we might still see the problem on both sides of the track. But if (for example) one candidate lost 1,000 votes and the other lost 100,000 votes, it is hardly "no harm, no foul". In order to get at the truth, we have to look at all the cases, not just the ones that might "help" one candidate or the other.

This is a non-partisan effort to get at the truth, not an attempt to spin a few isolated facts to favour one side or the other.

--MarkusQ

[anamandujano]

19. talking about the spin factor and sound bites

You know how the propaganda arm of the repukes works. Do you really think they will explain the significance of all the hard work you number crunchers are doing? Surely you've been watching the telling of this election story. As always, I hope that my dark thoughts don't come to pass.

[MarkusQ]

20. Numbers trump. Always.

There is a widespread misconception in the world that spin can trump anything. But explicit, objective, quantified facts win in the long run. Why? Because spin is limited to a narrow range of phenomena where subjective "facts" can (temporarily at least) out weigh real facts. That only works for a little while, and only for very limited goals. Spin can't put people on the moon with spin, it can't even feed them or keep them warm. But if you do the math, these and many more things are possible.

--MarkusQ

[anamandujano]

21. If it's not on the TV news

it just ain't so.

And how I wish that weren't so.

[MarkusQ]

22. No, the truth doesn't watch TV. That's just people. (n/t)