Diebold strikes again in Ohio- Dr. Richard Phillips analysis shows huge manipulation

[philb]

UNOFFICIAL RESULTS IN SEVENTEEN OHIO COUNTIES CANNOT BE RIGHT (there are implications for other races)

Richard Hayes Phillips, Ph.D.
December 7, 2006

ABSTRACT: In the November 7, 2006 election in Ohio there were 350,669 more
ballots cast than the number of votes counted for United States Senator.
In 16 counties there were 268,987 uncounted votes, or 19.46% of ballots cast,
compared to 82,957, or 2.99%, in 71 other counties. Cuyahoga County alone
accounted for 148,928 uncounted votes, or 26.48% of ballots cast in the county,
and 42.47% of the statewide total of uncounted ballots. In Marion County there
were 1,275 more votes counted for United States Senator than the reported number
of ballots cast, which is an impossibility. If the rate of uncounted ballots
in the 17 suspect counties had been about 3%, as was the case elsewhere in the
state, there would have been about 42,000 uncounted ballots instead of 269,000.
This indicates that 227,000 votes may have been lost by the touch screen voting
machines, which were utilized in all 17 of the suspect counties.

NOTE: This analysis is based upon unofficial results. If the official results
should reduce these discrepancies, the question will remain as to how the
unofficial results could have been so erroneous in the first place.

* * * * *

In the 2006 general election, according to the official website of Ohio
Secretary of State J. Kenneth Blackwell, there were 4,177,498 ballots cast in
the State of Ohio. Of these, only 3,831,716, or 91.72%, contained votes for
Governor, and only 3,826,829, or 91.61%, contained votes for United States
Senate. These numbers create the appearance of undervote (or overvote) rates
of 8.28% and 8.39%, respectively, in the two most hotly contested statewide
races on the ballot. When the unofficial election results are examined
county by county, there is a strikingly abnormal distribution of undervotes
and overvotes. I have chosen the United States Senate race to examine in
detail because there were only two candidates on the ballot (and one write-in
candidate), which makes the mathematical analysis simpler than for the
Governor's race, in which there were four candidates on the ballot (and two
write-in candidates). A similar analysis could be done, and should be done,
or the Governor's race, and for all races in which the unofficial results
were very close.

The methodology was simple. The percentage of undervotes and overvotes for
each county was derived by fifth grade mathematics. The total number of
votes counted for the candidates combined was subtracted from the number of
ballots cast. The remainder is the number of uncounted ballots, or
undervotes plus overvotes. This number was divided by the number of ballots
cast to determine the percentage of ballots left uncounted in each county.
The complete data set for all 88 counties is appended to this paper.

There are 88 counties in Ohio. Of these 88 counties, according to unofficial
results posted by J. Kenneth Blackwell, 71 counties had rates of undervotes
and overvotes ranging from 0.88% (in Greene County) to 6.90% (in Holmes
County). In 62 of these 71 counties, the percentages were tightly clustered
between 2.00% and 4.50%. The rate in these 71 counties combined was 2.99%.

In 16 of the other 17 counties, including 4 of the 10 most populous counties
in the State of Ohio, the percentages of undervotes and overvotes were
clearly anomalous, ranging from 11.91% (in Montgomery County) to 26.48% (in
Cuyahoga County), with a combined rate of 19.46%, or six and one-half times
the rate in the rest of the state. Just four counties -- Cuyahoga, Lucas,
Montgomery and Stark -- accounted for 219,332 undervotes and overvotes, or
62.55% of the statewide total of 350,669. Cuyahoga County alone accounted
for 148,928 undervotes and overvotes, or 42.47% of the statewide total. It
is difficult to believe that more than one in four voters in Cuyahoga County
could not decide between Sherrod Brown and Mike DeWine.

UNOFFICIAL RESULTS: UNITED STATES SENATE

Ballots Cast Votes Counted Undervotes/Overvotes

16 counties 1,382,455 1,113,568 268,987 19.46%
71 counties 2,775,090 2,692,133 82,957 2.99%

In Marion County, Blackwell reported 19,853 total votes cast, and 21,128
votes counted for the United States Senate candidates -- an overcount of
1,275 votes. These are known as "phantom votes," because they are
apparitions, with no explainable origin. There can never be more votes
counted for an office than the number of persons voting in the election.


UNOFFICIAL RESULTS: UNDERVOTES AND OVERVOTES

County Ballots Votes Undervotes/ Voting
Cast Counted Overvotes Technology

Cuyahoga 562,498 413,570 148,928 26.48% touch screen
Morrow 15,679 12,242 3,437 21.92% touch screen
Belmont 29,045 23,192 5,853 20.15% touch screen
Coshocton 16,138 13,107 3,031 18.78% touch screen
Licking 70,705 57,704 13,001 18.39% touch screen
Jackson 12,025 9,974 2,051 17.06% touch screen
Lucas 164,003 139,003 25,000 15.24% touch screen
Tuscarawas 36,124 30,750 5,374 14.88% touch screen
Stark 139,646 119,011 20,635 14.78% touch screen
Perry 12,775 10,894 1,881 14.72% touch screen
Carroll 12,664 10,898 1,766 13.95% touch screen
Highland 14,351 12,358 1,993 13.89% touch screen
Wood 50,666 44,190 6,476 12.78% touch screen
Adams 9,592 8,378 1,214 12.66% touch screen
Hancock 28,692 25,114 3,578 12.47% touch screen
Montgomery 207,952 183,183 24,769 11.91% touch screen
Marion 19,853 21,128 -1,275 -6.42% touch screen


Note that there is no county falling between Holmes County (6.90%) and
Montgomery County (11.91%). The counties listed above are clearly anomalous.
The unofficial results cannot be right. And, of course, the unofficial
results in Marion County are impossible.

Note also that all 17 counties listed above utilized touch screen voting
machines, known in the trade as Direct Recording Electronic (DRE). 31 of 88
Ohio counties utilized optical scanners, and none of them had this problem.
Data on voting technology utilized in 2006 by each Ohio county is displayed
on a map provided by www.yourvotecountsohio.org

A word of caution is in order. In the 2005 general election in Ohio, several
counties reported incorrect figures for total ballots cast. The false
numbers were derived by counting absentee ballots twice, as the Diebold
tabulators were programmed to do. However, there is no evidence that this is
the cause of the abnormally high numbers of undervotes and overvotes reported
in 16 counties in the 2006 general election. The combined totals of absentee
and provisional ballots are reported by Blackwell for each county, and in no
case are these numbers anywhere near what would be required to account for
the discrepancies.

To the contrary, unofficial voter turnout in these 17 counties was 54.35%,
compared to 52.21% in the other 71 counties of Ohio. That is not much of a
difference. There were 1,402,408 ballots cast in these 17 counties. If the
rate of undervotes and undervotes had been about 3%, as was the case
elsewhere in the state, there would have been about 42,000. Instead there
were 267,712 (or 268,987 if one takes into account the 1,275 phantom votes
in Marion County).

This raises the disturbing possibility that some 227,000 votes were lost
by touch screen voting machines in Ohio. Surely this merits a full-scale
investigation. Whether or not this "affected the outcome," a phrase
generally intended to mean who won and who lost the election, is beside
the point. If 227,000 votes were not counted, the outcome was affected.


VOTER TURNOUT, UNITED STATES SENATE
UNOFFICIAL RESULTS: NOVEMBER 7, 2006

County Registered Ballots Percent Votes Undervotes/
Voters Cast Turnout Counted Overvotes

Adams 16,966 9,592 56.54% 8,378 1,214 12.66%
Allen 68,892 35,138 51.00% 33,758 1,380 3.93%
Ashland 35,499 19,458 54.81% 18,636 822 4.22%
Ashtabula 62,265 34,415 55.27% 33,614 801 2.33%
Athens 42,573 17,525 41.16% 17,078 447 2.55%
Auglaize 32,759 16,877 51.52% 16,427 450 2.67%
Belmont 43,396 29,045 66.93% 23,192 5,853 20.15%
Brown 27,436 14,290 52.08% 13,831 459 3.21%
Butler 236,998 114,908 48.48% 110,908 4,000 3.48%
Carroll 20,169 12,664 62.79% 10,898 1,766 13.95%
Champaign 25,162 14,553 57.84% 14,168 385 2.65%
Clark 90,057 49,972 55.49% 48,934 1,038 2.08%
Clermont 127,759 65,531 51.29% 63,308 2,223 3.39%
Clinton 25,826 12,749 49.37% 12,394 355 2.78%
Columbiana 79,855 37,143 46.51% 36,104 1,039 2.80%
Coshocton 20,768 16,138 77.71% 13,107 3,031 18.78%
Crawford 28,923 17,072 59.03% 16,430 642 3.76%
Cuyahoga 1,054,670 562,498 53.33% 413,570 148,928 26.48%
Darke 34,357 20,435 59.48% 19,726 709 3.47%
Defiance 25,769 13,168 51.10% 12,981 187 1.42%
Delaware 108,733 64,927 59.71% 63,248 1,679 2.59%
Erie 51,017 30,645 60.07% 29,795 850 2.77%
Fairfield 96,566 54,246 56.18% 52,441 1,805 3.33%
Fayette 15,476 8,510 54.99% 8,340 170 2.00%
Franklin 766,490 342,958 44.74% 333,466 9,492 2.77%
Fulton 28,633 16,269 56.82% 15,728 541 3.33%
Gallia 23,320 9,976 42.78% 9,862 114 1.14%
Geauga 63,299 40,006 63.20% 38,765 1,241 3.10%
Greene 104,200 57,168 54.86% 56,663 505 0.88%
Guernsey 25,429 12,936 50.87% 12,365 571 4.42%
Hamilton 566,661 282,190 49.80% 273,129 9,061 3.21%
Hancock 49,855 28,692 57.55% 25,114 3,578 12.47%
Hardin 17,604 9,805 55.70% 9,388 417 4.25%
Harrison 10,936 6,234 57.00% 5,967 267 4.28%
Henry 19,618 11,697 59.62% 11,201 496 4.24%
Highland 25,679 14,351 55.89% 12,358 1,993 3.89%
Hocking 18,035 9,800 54.34% 9,465 335 3.42%
Holmes 18,204 8,564 47.04% 7,973 591 6.90%
Huron 35,645 18,635 52.28% 18,020 615 3.30%
Jackson 21,748 12,025 55.29% 9,974 2,051 17.06%
Jefferson 48,693 26,344 54.10% 25,364 980 3.72%
Knox 36,348 20,784 57.18% 20,180 604 2.91%
Lake 154,048 86,684 56.27% 84,134 2,550 2.94%
Lawrence 41,470 19,209 46.32% 18,731 478 2.49%
Licking 102,300 70,705 69.12% 57,704 13,001 18.39%
Logan 29,730 16,488 55.46% 15,955 533 3.23%
Lorain 190,767 98,664 51.72% 96,324 2,340 2.37%
Lucas 296,541 164,003 55.31% 139,003 25,000 15.24%
Madison 23,597 13,583 57.56% 13,106 477 3.51%
Mahoning 175,923 96,027 54.58% 92,497 3,530 3.68%
Marion 43,814 19,853 45.31% 21,128 -1,275 -6.42%
Medina 121,000 65,600 54.21% 64,384 1,216 1.85%
Meigs 15,690 7,912 50.43% 7,661 251 3.17%
Mercer 30,985 15,510 50.06% 14,966 544 3.51%
Miami 68,309 36,686 53.71% 36,089 597 1.63%
Monroe 10,054 6,266 62.32% 5,925 341 5.44%
Montgomery 375,459 207,952 55.39% 183,183 24,769 11.91%
Morgan 9,605 5,480 57.05% 5,370 110 2.01%
Morrow 23,962 15,679 65.43% 12,242 3,437 21.92%
Muskingum 51,907 29,004 55.88% 27,739 1,265 4.36%
Noble 8,650 5,453 63.04% 5,124 329 6.03%
Ottawa 28,773 17,734 61.63% 17,238 496 2.80%
Paulding 13,619 7,749 56.90% 7,409 340 4.39%
Perry 20,888 12,775 61.16% 10,894 1,881 14.72%
Pickaway 30,917 17,383 56.22% 16,860 523 3.01%
Pike 18,320 9,632 52.58% 9,508 124 1.29%
Portage 103,718 55,283 53.30% 53,557 1,726 3.12%
Preble 27,699 16,007 57.79% 15,552 455 2.84%
Putnam 24,259 14,545 59.96% 14,031 514 3.53%
Richland 90,641 46,124 50.89% 44,900 1,224 2.65%
Ross 42,351 23,008 54.33% 22,262 746 3.24%
Sandusky 40,477 23,129 57.14% 22,265 864 3.74%
Scioto 47,285 26,072 55.14% 25,466 606 2.32%
Seneca 34,982 20,399 58.31% 19,818 581 2.85%
Shelby 29,476 17,523 59.45% 16,926 597 3.46%
Stark 270,515 139,646 51.62% 119,011 20,635 14.78%
Summit 373,289 199,256 53.38% 193,188 6,068 3.05%
Trumbull 141,393 81,011 57.29% 78,552 2,459 3.04%
Tuscarawas 56,295 36,124 64.17% 30,750 5,374 14.88%
Union 30,004 17,076 56.91% 16,554 522 3.06%
Van Wert 20,347 10,642 52.30% 10,299 343 3.22%
Vinton 8,666 4,598 53.06% 4,412 186 4.05%
Warren 127,843 66,684 52.16% 64,859 1,825 2.74%
Washington 39,547 23,134 58.50% 22,369 765 3.31%
Wayne 70,067 39,130 55.85% 37,753 1,377 3.52%
Williams 25,329 13,174 52.01% 12,687 487 3.70%
Wood 93,272 50,666 54.32% 44,190 6,476 12.78%
Wyandot 15,428 8,303 53.82% 8,006 297 3.58%

TOTAL 7,851,499 4,177,498 53.21% 3,826,829 350,669 8.39%

--------------------------------------------------------------------

OFFICIAL RESULTS IN TEN OHIO COUNTIES CANNOT WITHSTAND SCRUTINY

Richard Hayes Phillips, Ph.D.
December 21, 2006

A fortnight ago I posted a study showing that the unofficial election results
in 17 of 88 Ohio counties cannot be right. Marion County reported more votes
counted than ballots cast, which is impossible. The other 16 counties had a
combined rate of undervotes equaling 19.46% -- six and one-half times the rate
in the rest of the state (2.99%). In Cuyahoga County, 26.48% of ballots cast,
more than one in four, reportedly contained no vote for United States Senate,
the most hotly contested race on the ballot.

All 17 suspect counties utilized Diebold touch screen voting machines, known
in the trade as Direct Recording Electronic (DRE). 31 of 88 Ohio counties
utilized optical scanners, numerous counties utilized touch screens provided
by other vendors, and none of them had this problem.

If the rate of uncounted ballots in the 17 suspect counties had been about 3%,
as was the case elsewhere in the state, there would have been about 42,000
undervotes instead of 269,000. This raised the disturbing possibility that
227,000 votes might have been lost by the touch screen voting machines.

This analysis was based upon unofficial results. I stated at the time that
even if the official results were to reduce these discrepancies, the question
would remain as to how the unofficial results could have been so erroneous in
the first place. Now that official results have been posted on the website of
Ohio Secretary of State J. Kenneth Blackwell, I am able to compare the official
and unofficial results for the 17 suspect counties and analyze the changes.


COMPARISON OF UNOFFICIAL AND OFFICIAL RESULTS, UNITED STATES SENATE

-- Unofficial Results -- -- Official Results --
Ballots Votes Ballots Votes
County Cast Counted Undervotes Cast Counted Undervotes

Cuyahoga 562,498 413,570 148,928 26.48% 468,056 452,832 15,224 3.25%
Morrow 15,679 12,242 3,437 21.92% 12,952 12,481 471 3.64%
Belmont 29,045 23,192 5,853 20.15% 24,484 23,556 928 3.79%
Coshocton 16,138 13,107 3,031 18.78% 13,865 13,366 499 3.60%
Licking 70,705 57,704 13,001 18.39% 60,726 58,923 1,803 2.97%
Jackson 12,025 9,974 2,051 17.06% 10,669 10,288 381 3.57%
Lucas 164,003 139,003 25,000 15.24% 146,652 142,304 4,348 2.96%
Tuscarawas 36,124 30,750 5,374 14.88% 31,913 31,385 528 1.65%
Stark 139,646 119,011 20,635 14.78% 143,753 139,264 4,489 3.12%
Perry 12,775 10,894 1,881 14.72% 13,368 11,189 2,179 16.30%
Carroll 12,664 10,898 1,766 13.95% 11,566 11,053 513 4.44%
Highland 14,351 12,358 1,993 13.89% 15,064 12,981 2,083 13.83%
Wood 50,666 44,190 6,476 12.78% 47,089 45,515 1,574 3.34%
Adams 9,592 8,378 1,214 12.66% 9,972 8,570 1,402 14.06%
Hancock 28,692 25,114 3,578 12.47% 26,147 25,622 525 2.01%
Montgomery 207,952 183,183 24,769 11.91% 219,153 188,836 30,317 13.83%
Marion 19,853 21,128 -1,275 -6.42% 22,224 21,604 620 2.79%

SUBTOTAL 1,402,408 1,134,696 267,712 19.09% 1,277,653 1,209,769 67,884 5.31%


As shown in the table above, in four counties (Adams, Highland, Montgomery and
Perry) the egregious errors in the election results have not been corrected. There
are still far more ballots cast than votes counted. In fact, the discrepancies are
even greater than in the unofficial results. In Montgomery County alone, there
were 30,317 voters, or 13.83%, who did not vote for United States Senator. Either
that, or the official results are not true and correct – which is, of course, the
case. And this is not the only problem with the official results.

In 11 of the counties where, according to unofficial results, ballots cast had far
exceeded the number of votes counted, the reported number of ballots cast has been
revised downward. In Marion County, which originally reported 1,275 more votes
counted than ballots cast, which is impossible, it is now acknowledged that the
number of ballots cast had been underreported in the unofficial results. But in
Stark County, where 20,635 undervotes, or 14.78% of ballots cast, were originally
reported, the Board of Elections now reports, in the official results, 4,107 more
ballots cast than in the unofficial results:


DIFFERENCE BETWEEN UNOFFICIAL AND OFFICIAL RESULTS

Ballots Votes Absentee/
County Cast Counted Undervotes Provisional

Cuyahoga -94,442 +39,262 -133,704 30,791
Morrow -2,727 +239 -2,966 271
Belmont -4,561 +364 -4,925 529
Coshocton -2,273 +259 -2,532 157
Licking -9,979 +1,219 -11,198 1,464
Jackson -1,356 +314 -1,670 464
Lucas -17,351 +3,301 -20,652 3,694
Tuscarawas -4,211 +635 -4,846 497
Stark +4,107 +20,253 -16,146 4,488
Perry +593 +295 +298 369
Carroll -1,098 +155 -1,253 244
Highland +713 +623 +90 646
Wood -3,577 +1,325 -4,902 1,664
Adams +380 +192 +188 265
Hancock -2,545 +508 -3,053 851
Montgomery +11,201 +5,653 +5,548 10,272
Marion +2,371 +476 +1,895 698

SUBTOTAL -124,755 +75,073 -199,828


In Stark County, according to Blackwell’s website, there had been, at the time the
unofficial results were reported, 4,488 absentee and provisional ballots yet to be
examined – from which, no doubt, the 4,107 additional ballots were drawn. Thus
there is no indication from the Board of Elections that the number of ballots cast
was overreported in the unofficial results. Rather, the number of votes counted
was underreported. These numbers deserve closer scrutiny:


COMPARISON OF RESULTS, STARK COUNTY

Sherrod Mike Richard
Brown DeWine Duncan

Unofficial 68,266 50,741 4
Official 79,900 59,353 11
Difference 11,634 8,612 7


If there were only 4,107 additional ballots in Stark County, how did Sherrod Brown
gain 11,634 votes, and how did Mike DeWine gain 8,612 votes? Where did these votes
come from? If there were only 4,488 unexamined absentee and provisional ballots,
where did the 20,253 newly counted votes come from? These numbers are impossible.

In fact, there are three other counties (Coshocton, Cuyahoga and Tuscarawas) where
the number of newly counted votes exceeds the number of absentee and provisional
ballots that remained to be examined on Election Night. In Cuyahoga County, there
were 39,262 newly counted votes, drawn from only 30,791 unexamined absentee and
provisional ballots. These numbers also deserve closer scrutiny:


COMPARISON OF RESULTS, CUYAHOGA COUNTY

Sherrod Mike Richard
Brown DeWine Duncan

Unofficial 291,469 122,101 0
Official 319,568 133,235 29
Difference 28,099 11,134 29


If there were only 30,791 absentee and provisional ballots in Cuyahoga County that
emained to be examined on Election Night, how did Sherrod Brown gain 28,099 votes
and Mike DeWine gain 11,134 votes? These numbers are impossible.

Finally, the vote totals for Richard Duncan, a write-in candidate, cannot be right.
By comparing the unofficial and official results, one sees that his vote totals
actually decreased, sometimes substantially, in four counties (Coshocton, Highland,
Lucas, and Marion). This is especially ridiculous in light of the fact that, in
many counties, write-in votes are not counted on Election Night, and thus are not
included in the unofficial results. For example, Richard Duncan received 29 write-
in votes in Cuyahoga County, none of which appeared in the unofficial results.


COMPARISON OF RESULTS FOR RICHARD DUNCAN

County Unofficial Official Difference

Coshocton 6 2 -4
Highland 31 10 -21
Lucas 129 15 -114
Marion 19 0 -19


How did Richard Duncan lose 4 of his 6 votes in Coshocton County, 10 of his 31
votes in Highland County, 114 of his 129 votes in Lucas County, and all of his
19 votes in Marion County? These numbers are impossible.

There were 17 Ohio counties for which the unofficial results could not have been
correct. Of these, only 7 counties (Belmont, Carroll, Hancock, Jackson, Licking,
Morrow and Wood) have posted official results that withstand scrutiny. The
obviously incorrect results in the other 10 counties may be attributable to the
same problem that existed in the 2005 election – the programming of Diebold
tabulators to report two sets of numbers for ballots cast. One, “times counted,”
may or may not be correct; the other, “cards cast,” equals “times counted” plus the
number of absentee ballots, which thus are counted twice. The result is, at best,
an erroneously high number of undervotes and, at worst, a window of opportunity for
altering the vote count. Secretary of State J. Kenneth Blackwell had a choice of
which, if either, of these numbers to post as the number of ballots cast. Ohio had
the good sense to get rid of Blackwell. Now Ohio needs to get rid of Diebold.

------------------------------------------------------------------------------

TWO MORE COUNTIES WHERE OFFICIAL RESULTS CANNOT WITHSTAND SCRUTINY

Richard Hayes Phillips, Ph.D.
December 22, 2006

Dale Tavris has performed a similar analysis of undervotes in the 2006 Senate
race. Using official results only, he identified not four, but six counties
with inexplicably high percentages of undervotes:

"Furthermore, there were six counties that were definite and extreme outliers
(all Diebold) compared to the other counties. Those six counties (Mercer,
Darke, Highland, Montgomery, Adams, Perry) had undervote rates ranging from
11.2% to 16.3%, with an average of 13.8%, while the other 82 Ohio counties
had undervote rates ranging from 0.62% to 6.76%, with an average of 3.37%.
The undervotes in the six outlier counties amounted to almost a quarter
(24.9%) of the undervotes in the whole state, whereas the total votes in
those six counties amounted to only 7.1% of the total votes in the state."

Two of these counties -- Darke and Mercer -- did not appear as outliers in
the unofficial results. At that time, the undervote rates for these two
counties were reported as 3.47% and 3.51%, respectively. The differences
between the unofficial and official results are nothing short of ridiculous:


COMPARISON OF RESULTS, DARKE COUNTY

Ballots Votes
Cast Counted Undervotes

Unofficial 20,435 19,726 709 3.47%
Official 23,350 20,187 3,163 13.55%
Difference 2,915 461 2,454 84.19%

COMPARISON OF RESULTS, MERCER COUNTY

Ballots Votes
Cast Counted Undervotes

Unofficial 15,510 14,966 544 3.51%
Official 17,483 15,532 1,951 11.16%
Difference 1,973 566 1,407 71.31%


There you have it. The percentage of provisional and late-arriving absentee
ballots containing no vote for Senator was 84.19% in Darke County, and 71.31%
in Mercer County. Either that, or the official results are not true and
correct -- which is, of course, the case.

When the unofficial results were posted on Blackwell’s website, it was
reported that there were 300 unexamined absentee and provisional ballots in
Darke County, and 534 in Mercer County. Somehow, another 461 votes were
counted in Darke County, and another 566 in Mercer County. These numbers
cannot be right. And there is surely no legitimate way to account for an
additional 2,915 ballots cast in Darke County, and 1,973 in Mercer County.
Again, the problem lies in the Diebold tabulators, which are programmed to
produce two sets of numbers for ballots cast. Very likely, Darke and Mercer
counties reported the number for “times counted” in their unofficial results,
and the number for “cards cast” in their official results. There is no
legitimate reason for any accounting system to keep two sets of books.
It is an open invitation to error and fraud. Diebold tabulators need to be
decertified in Ohio and in every other state of the union.

[peacetalksforall]

1. Recommended. Save this one for ammunition and activism.

Kick it up.

Everyone should learn from this and know where to find it when they need it.

[Jackpine Radical]

2. Can we get it into the Research forum?

[philb]

4. When there are lots of machines with undervotes over 20% as here & Sarasota, its clear

that the results were manipulated. And these aren't the only races manipulated in this manner.

Its hard to believe that the authorities suggest that lots of machines with undervotes over 20% should be disregarded
and the results taken seriously.
Its too bad the public at large doesn't seem to understand or care enough to allow such manipulation to go unpunished, and without consequences.

[OnTheOtherHand]

9. no analogy to Sarasota

These reports don't support undervote rates over 20%; they support the inference that the unofficial turnout figures were wrong. We've seen those before -- IIRC, turnout calculations that double-count voters. In Cuyahoga, where the numbers are largest, the unofficial turnout figure was way out of trend with 2002 turnout, as well as inconsistent with all the race results. A local blogger contemporaneously reported here on his investigation into the unofficial figure.

If someone has some evidence supporting the inferences that the BoE was lying or mistaken, that Cuyahoga had eye-popping midterm turnout, and that close to 100,000 votes or more were deleted but somehow showed up in the unofficial turnout figure anyway -- and only in the unofficial turnout figure -- that would be good to know. (For instance, precincts where 60 or 80 or 100 more people signed in than were recorded in the tallies.) Regardless, the fact pattern isn't analogous to Sarasota.

[Ellipsis]

3. Is that distant thunder I hear?

K&R

For those who don't know Dr Phillips, or the accomplishments of one beloved geomorphologist, some background.

http://www.scoop.co.nz/stories/WO0412/S00167.htm

[TankLV]

5. Could you make it a little longer - I'm all thru reading it for the comming year....

[emlev]

6. Here's a link to it on Richard Hayes Phillilps's website

http://web.northnet.org/minstrel/ohio.2006.htm

His other published studies on Ohio can be found there as well.

[Ellipsis]

7. Thanks for this

I have his declaration in pdf, which is what I really wanted to link to.

[emlev]

10. You're welcome.

[philb]

11. Similar Free Press supportive article

Free Press Article
This is consistent with what we found based on polling in Ohio - see:
http://www.freepress.org/departments/display/19/2006/2250

[Stevepol]

8. K&R!!

[Ms. Toad]

12. Cuyahoga County is not a good county to use for this analysis...

There were anomalies relevant to Cuyahoga County that (at a minimum) need to be addressed as to their impact on the statistical analysis:

1. In multiple precincts in Cuyahoga County, voters bypassed the sign in books and went directly to the voting lines. They were handed a voter card, and were allowed to vote without creating any record that they showed up at the polls. It is not clear how the number of voters in the statistical analysis was determined. This quirk should result in more votes than voters if the number of voters used for this analysis is taken from the poll books. If the number of voters is taken from the voting machine tally of number of ballots cast, it should be a wash. This was estimated to involve about 12,000 voters. http://www.columbusdispatch.com/news-story.php?story=231323 . (Note: this was documented in Cuyahoga County – I do not know whether it happened in other places, as well.)

2. Cuyahoga County voters were also uniquely positioned with respect to the constitutional amendment to permit gambling. The measure would have permitted slot machines at 7 horse tracks throughout the state, and the creation of two free-standing gaming parlors in Cuyahoga County. The constitutional prohibition on gambling (outside of race tracks) was to remain intact in the remaining 87 counties in the state.

The NIMBY factor or, in the alternative, the potential for additional jobs in a very depressed local economy and revenue for the county, not to mention easy access to gambling sites likely motivated some voters to head for the polls who did not care about any other issue or candidate. That the measure earned 54% approval in Cuyahoga County (http://www.newsnet5.com/politics/10268783/detail.html) compared to 43% statewide (http://www.sos.state.oh.us/SOS/ElectionsVoter/results2006.aspx?Section=1856) is some indication that voters in Cuyahoga County viewed this matter differently than the rest of the state - it is not unlikely that there were voters who turned out just to vote for this issue and skipped the rest of the ballot.

As a general matter, the undervote across the state may have been impacted by three hot issues which were attractive to individuals who are otherwise indifferent to politics: the smoking amendment (permitting smoking in certain areas - supported by bars, bowling alleys, presumably their patrons, and the tobacco industry), the smoking law (banning nearly all public smoking - opposed by the same folks) and the gambling issue. People (generally – not specific to Cuyahoga County) may have gone to the polls and voted for or against these three issues without bothering to vote for anyone or anything else.

Note, I'm not saying everything is hunky dory in Ohio – certainly the first problem above is a clear indication things are still really bad. Nonetheless, statistics don't exist in a vacuum and the local factors which likely impacted the statistics being reported need to be discussed in the analysis in order to make it more than a rallying cry for the already convinced. I am aware of the Cuyahoga County quirks because I am in the Cuyahoga County news market. There may well be other quirks in other localities (particularly like the bypassing the pollbook quirk).

[Bill Bored]

13. Ah Ms. Toad! Hopping out of the woodwork again eh?

I'm curious: Is there some reason why you only seem to post on this forum when someone raises questions about these Ohio elections?

What's your prescription for election integrity if everything isn't hunky dory in Ohio? Just curious.

[OnTheOtherHand]

14. Bill, with respect

Describing another poster as "hopping out of the woodwork," and completely ignoring the substance of her post, is the sort of reception I expect to get -- but I don't see why anyone else should have to put up with it.

[Ms. Toad]

16. I live in Ohio,

one county away from Cuyahoga County. These are local issues for me. Because of this I have information that doesn't necessarily make national news - and isn't apparent from the information on the Secretary of State's website. (Even being local, I missed the article about the 12,000 folks who voted without signing in and had a devil of a time tracking it down when someone told me about it a couple of days later.)

I also have two degrees in math, including one in applied math. Math, particularly statistics, gets a bad rap, because statistics are viewed as easily manipulable to prove anything. When statistics are trotted out that do not properly take into account (often entirely innocently) the real non-number crunching assumptions that had to be made in order to do the the number crunching, it adds to the perception that statistics are nothing but a d*** lie. They aren't - they are very valuable tools when used properly and carefully. When I have knowledge (generally because I know something based on personal connections or because it is a local issue) I post it with the hope that the information will be integrated into the statistical analysis so that if it still reaches the same conclusion after the integration, the analysis can better withstand careful scrutiny.

My prescription for Ohio - I don't have a complete prescription, but # 1 on my list is to get rid of Blackwell (yay - one down)

Second is to address all of the human foibles and/or easily addressable security issues that keep cropping up. (Just a couple of examples - I am sure that a review of election process in each local jurisdiction would generate more.)

With respect to election day - good training, and perhaps some practice runs before election day, are crucial. If anyone had described the setting to me in Cuyahoga County ahead of time, I would have predicted what apparently happened with at least 12,000 voters. You place a voting line next to voting terminals without anyone checking to see whether the folks in the voting line had actually signed in to vote and you will have lots of folks voting without signing in. This could have been prevented, or at least minimized, by physical arrangement (can't get to the voting booth line without going through the poll book line), a hard stop at the poll book (one voter at a time let past as a booth opens up), giving the voter the voting card at the poll book - or at least giving the voter some kind of token required to be traded for the voting card.

Another example is examining the security measures (regardless of what equipment is used). In one nearby county, they used to leave bags of voted ballots in an unguarded hallway in an unlocked building before counting them. That is far easier security issue to address and fix than convincing the powers to be that they spent hundreds of thousands of dollars on faulty equipment. If the equipment is electronic, make sure those having access to it are never alone with it (no sleepovers, equipment always kept in locked rooms - perhaps with two keys needed to access the locked room), limit electronic access to the software by need - a poll worker does not need to have - and should not have - the same access as the IT staff. There is enough awareness of these kinds of issues - including among current poll workers who have some access to the elections officials - that a security audit should be a relatively easy thing to convince the election officials to have implemented. An audit should be relatively inexpensive, but if money is an issue, concerned local folks could even raise money to pay for it.

A final example is to be more realistic in the distribution of voting equipment - and to prepare for emergency redistribution on election day. They have voting records for years of voting. It should not be that hard to track voting trends in particular precincts, or at particular address blocks, and to divide the voting equipment so that everyone has equal access. Relatively speaking, none of the voting equipment is expensive enough that a few extra could not be made reserved (even by rental, if necessary) to redistribute on election day.

Third (or maybe even second) I would address the voter suppression (making registration and voting more difficult for folks likely to lean in a particular direction). The Voter ID law should go, as it disproportionately disenfranchises the poor, elderly, and minority (as it was designed to do). Purging voter rosters based on factors more likely to be applicable to low income/minority voters needs to stop (deletion or eligible for challenge based on returned registration postcards, criminal record, etc.).

All of the above are things (except perhaps getting rid of the voter ID law - for which there is inexplicable support even here on DU) are relatively easy to do, relatively non-controversial, and can make a big difference in election integrity. If I had energy to devote to this cause, that is where I would put my energy. (Generally speaking, my current energy for activism is devoted to GLBT issues - with occasional blips near election time if I have enough extra time and energy to campaign for a particular candidate or issue).

Dealing with equipment specific issues is relatively low on my list. I have yet to see anything that convinces me that the current equipment is more of a threat than the relatively mundane issues I have noted above. Most any equipment is manipulable - to a greater or lesser extent. With every kind of voting equipment ever used, people have figured out a way to stuff the ballot box. The term vote shaving came from shaving the gears on the voting machines to make an incremental difference in the vote tallied for the candidate associated with a particular gear. These machines were formerly used in Ohio, and are still being supported in New York by folks who are opposed to DREs with paper trails - and even by some who are opposed to paperless voting. These machines are the original paperless voting machines - on most models there is absolutely no record of any individual vote. The lack of local security and the disenfranchisement of elderly/poor/minority voters are much more crucial in my mind - and we should easily be able to garner bipartisan support for at least the first two general areas of concern noted above.

[OnTheOtherHand]

15. I think you're already too subtle for the OP

As far as I could tell from the rather lengthy OP, it isn't really an undervote analysis at all; it focuses on discrepancies between the unofficial and official returns. Cuyahoga stands out because its unofficial turnout figure was much higher than its official figure, and didn't jibe with the votes cast.

Your points are good ones nonetheless.

[Ms. Toad]

17. Perhaps I misread,

but I thought that the analysis was based solely on unofficial results, specifically the difference between the senate/governor's races and the total ballots cast as all three were reported unofficially on the SOS website - with a precaution that in past elections the official results were significantly different than the unofficial ones and the analysis should be reviewed once the official results are available.

The gist of what I took from the OP was that there were several counties in which there were over votes (more votes than ballots cast) - which should be impossible. Depending on where the ballots cast numbers came from, a situation like the one in Cuyahoga County, in which voters were allowed to cast ballots without signing in, could explain it. (Ballots cast could have come from a number of sources - the sign in roster and the total ballots cast recorded electronically are two sources - I couldn't tell the original source for the number - other than that it was unofficially reported on the secretary of state's site.)

There were also under votes - and the OP particularly focused on the governor and senate races - in which there were far fewer total votes in the governor's race and Senate race than the number of ballots cast, with Cuyahoga County standing out like a big sore thumb claiming half of the under votes for the state in one/both races (don't recall which, at the moment). The gambling issue, and the unique impact it would have had on Cuyahoga County, could account for a bunch of people in Cuyahoga County going to the polls who didn't bother to vote for the governor or senate candidate. At a minimum, it should be explicitly taken into account (the first place to start would be by including the number of votes on Issue 3 in the analysis).

[OnTheOtherHand]

18. I shd have been clearer -- the OP has parts...

Part is based just on unofficial results; part is based on changes from unofficial to official results. Actually, it is almost funny: first RHP concluded (rightly, at least in part) that the unofficial results "cannot be right"; then, when the official results didn't comport well with the unofficial results, he declared that the official results "cannot withstand scrutiny." I don't know whether the official results can withstand scrutiny or not, but the fact that they contradict unofficial returns isn't much basis for judgment. (EDIT TO ADD: What he actually says about the official returns in Cuya isn't that the turnout figure doesn't match the unofficial count, but that the changes in candidate vote counts are inconsistent with the previously reported numbers of uncounted ballots. That's a fair point, although by itself I hardly take it as evidence of fraud. RHP also notes in his initial article, "If the official results should reduce these discrepancies, the question will remain as to how the unofficial results could have been so erroneous in the first place" -- and reiterates that in his second article. So I had that in mind when I wrote what I wrote.)

But the fact that part of the OP addresses official returns doesn't supersede your analysis. My main point is that there is good reason to doubt that the unofficial turnout figure for Cuyahoga corresponded with actual (ETA: turnout or) votes cast for anything. A similar thing happened in 2004, to much kerfuffle. Lots of problems panned out in Cuya in 2004, but that one didn't. (ETA: I really should not try to write about this stuff before my first cup of coffee!)

So, when I said that the OP "isn't really an undervote analysis at all," what I meant was (inter alia) that the official returns from Cuyahoga are pretty comparable to the unofficial returns except that the turnout figure has been fixed, and the results at precinct level don't show large undervotes. Actually, the SOVC reports "blank" votes, "over vote(s)," and "under votes" for each precinct -- their blank vote is what most people would call an undervote.

I guess here is where I should interject a semantic quibble with you and also with myself. I think what you mean by "overvotes" is actually "phantom votes" -- more votes counted than voters. Generally "overvote" means an attempt to vote for too many candidates. That's why RHP calculates "undervotes and overvotes" as part of one calculation. Political scientists have taken to calling these "residual votes." I just said "undervotes" because generally most residual votes are undervotes, especially in DRE jurisdictions. Diebold apparently reserves the term "undervote" for the circumstance where a voter is allowed to vote for multiple candidates but votes for fewer than the maximum (or maybe they actually define it in a way that makes it impossible on touch screens, I don't remember) -- so, as I said, their blank vote is what most of us would call an undervote.

Now, part of the OP does try to address more solid undervote figures. Perhaps most strikingly, Montgomery County reports 219,153 "cards cast," but only 189,201 votes counted for governor. Weird. On the other hand (tm), Cuyahoga in their 2800-page SOVC reports (on p. 37) 676,560 "cards cast" -- way above even their unofficial turnout figure. Yet on p. 74 they have, for governor, 468,056 "times counted," 454,469 "total votes," 12,510 "blank voted," and 1,077 "over voted." (It's easy to spot-check at the precinct level that the cards cast numbers are often far above the total vote numbers within precincts.) Evidently the "cards cast" figure soars way above the vote count. I have no way of knowing whether votes were somehow deleted in Cuyahoga County, but as a data analyst, my initial thought is that the "cards cast" figure exaggerates turnout -- not that all the other figures understate it. (ETA: RHP seems to end up near this point at the end of his second article, although I don't think the "cards cast" anomalies owe solely to double-counting of absentee ballots.)

Oh, did you go here yet? http://www.callahansclevelanddiary.com/?p=110 (I think I linked to this in another post -- maybe not even.) I'm reluctant to quote an e-mail quoted on a blog without having obtained permission and verification from the original author, but the gist is that even the BoE couldn't figure out exactly what the "cards cast" figure meant. Diebold machinations? Well, I would be more alarmed if I thought there were a chance in hell that 676,560 people had voted in Cuyahoga in 2006. As far as I know, that turnout level is totally out of trend not only with past Cuya numbers, but turnout figures from other jurisdictions around the country where the numbers are considered relatively sane and reliable.

Your point about Issue 3 may still hold, but it would be (possibly) explaining why (e.g.) the Cuyahoga undervote or blank vote for governor was as high as 2-3%, which is not shockingly high to begin with.

I've spent about an hour trying to put together some of these basic facts. As you know, it takes so much less time to allege evidence of manipulation than it does to investigate it. The Sarasota numbers are alarming: they provide strong evidence that Jennings should have won by thousands of votes. So far, the Cuyahoga numbers are merely weird. Since the Cuyahoga SOVC does provide undervote (blank vote) numbers, we could analyze those much as people have analyzed the Sarasota undervotes, and they might tell an interesting story.

[truedelphi]

19. Thank yo for these clarifications. n/t