Monday, December 27, 2004

Default settings on Voting Machines

freepress.org

Default settings in Mahoning County
by Richard Hayes Phillips, Ph.D.

A disturbing story has been widely circulated that a vote for Bush was the default choice in the software of electronic voting machines in a number of states. By definition, “default” settings are built-in by the manufacturer to make sure their programs work properly, and can be changed by the user. Some examples of default settings on a home computer are screen savers, type face, and screen resolution.

According an article by Ann Harrison, posted at http://www.counterpunch.org/harrison12082004.html

in certain counties in Ohio, Florida, Pennsylvania, Texas, and New Mexico where touch screen voting machines were used, there have been complaints from voters who selected Kerry on the touch screen and saw their votes change to Bush on a summary screen. In addition, there was a specific problem with the Sequoia AVC Edge machine (not used in Ohio) where voters actually saw preselected default choices presented to them.

With touch screen machines, it is possible to set up a default choice for Bush that would not be seen by the voters. Their votes would be automatically cast for Bush unless they successfully overrode the default choice of the computer. Likewise, if they deliberately chose not to vote for president, their votes would be counted for Bush.

Mahoning County was by far the largest county in Ohio where touch screen voting machines were utilized. According to a report by the Youngstown Vindicator at http://www.vindy.com/basic/news/281829446390855.php

Mark Munroe, Chairman of the Mahoning County Board of Elections, said there were 20 to 30 machines that needed to be recalibrated during the voting process because some votes for a candidate were being counted for that candidate’s opponent. In addition, about a dozen machines needed to be reset because they “essentially froze.” Later on election night, problems arose in 16 precincts (11 in Youngstown, 2 in Boardman, and one each in Jackson Township, Craig Beach, and Washingtonville), causing election results to be delayed for three hours as Board of Elections employees checked the vote tallies of the touch screen machines.

Mahoning County utilized ES&S Ivotronic touch screen machines. The administrative password for these machines was reported on the ES&S website itself.

By default, the password is 1111. According to http://savethevote.com/issues/glitches.htm this password cannot be modified easily, which would mean that anyone who knows the password could change other default settings on the machines.

Lists of Election Day incident reports are posted at https://voteprotect.org/epc/index.php?display. Click on Research/Maps to find the state, and then the county, in which you are interested.

For Mahoning County there are 116 incident reports, 28 of them involving machine problems. Most of these are complaints consistent with the voting machine having a preselected default setting. There are several reports of machines miscasting votes, always away from Kerry, that do not specify a polling site. Most, however, do specify a polling place, and for these I have ascertained the precinct by calling the Board of Elections or referring to their website at http://www.electionohio.com/mahoning/findpollinglocation.asp.

In addition, we have the specific ID number for three machines (noted below):

TOUCH SCREEN COMPLAINTS, MAHONING COUNTY

Youngstown 1F, Mill Creek Community Center, 496 Glenwood Avenue, Youngstown (2 reports). Machine not working; “arrow moves to other candidates repeatedly.”

Youngstown 2A, Spanish Evangelical Church, 1408 Rigby Street, Youngstown (6 reports). Two of three machines not working properly, or vote for presidential candidate is “reversed.”

Youngstown 2B, Mary Haddow School, 2800 Oak Street Extension, Youngstown (1 report). Two machines down.

Youngstown 2C, Elizabeth Baptist Church, 7 South Garland Avenue, Youngstown (1 report). Machine # V0112442-C “not properly recording votes.” Brought to attention of presiding judge at 7:40 AM, recalibrated by Board of Elections employee at 9:35 AM, machine not shut down during that time.

Youngstown 2E, Price Memorial Zion Church, 920 Dryden Avenue, Youngstown (6 reports). Two voters “selected Kerry and Bush’s name came up.” They succeeded in voting for Kerry on their second attempts. One voter “had to scroll through and vote five times before his votes were finally recorded.” Two voters said the “presidential vote option never appeared while they were going through trying to vote,” it happened to a third voter at the same time, and on two different machines.

Youngstown 3C, Tabernacle Baptist Church, 707 Arlington Street, Youngstown (1 report). Machine # V0107749-C “malfunctioning, screen flashes and scrolls, shows different results or no results.”

Youngstown 5F, Martin Luther Lutheran Church, 420 Clearmount Drive, Youngstown (1 report). One touch screen machine broken, second one erased votes.

Youngstown 5G, Hillman Elementary School, 164 West Myrtle, Youngstown (1 report). When voter pressed Kerry, the screen “went blank” four times. Five attempts were required.

Boardman 20, St. John’s Greek Orthodox Church, 4955 Glenwood Avenue, Boardman (2 reports). Vote for Kerry goes to Bush. On third attempt, screen shows vote correctly. Poll worker dismisses complaint, saying machine is “temperamental.” Another voter needed two attempts to cast vote for Kerry.

Poland Village 1, North Elementary School, 361 Johnston Place, Youngstown (1 report). Pressed Kerry button, third party candidate appeared on screen. On the second attempt it worked.

Struthers IC, St. Nicholas Byzantine Church, 1898 Wilson Avenue, Youngstown (1 report). Machine # V0114343-C repeatedly marks Democratic vote as Republican. Many voters complained of this problem. Election worker had to restart and recalibrate the machine.

These are only some of the broken or improperly functioning machines. The Chairman of the Mahoning County Board of Elections reported many more than these. Most of the incident reports are precinct-specific, and they cry out for scrutiny of the 2004 and 2000 election results at the precinct level. However, the precinct boundaries in Youngstown have been completely redrawn since the 2000 election, making a meaningful comparison possible only at the ward level, not the precinct level.

In the table below, the 2004 results are listed on the top lines, and the 2000 results on the bottom lines. Percentages for candidates are derived from votes counted for president. Percentages for “undervotes,” ballots with no candidate selected or otherwise not counted, are derived from total ballots cast.

COMPARISON OF 2004 AND 2000 RESULTS IN YOUNGSTOWN
Cast Valid Republican Democratic Undervotes
WARD 1 2629 2596 209 8.05% 2339 90.10% 33 1.26%
2140 1977 131 6.62% 1796 90.84% 163 7.62%
WARD 2 3823 3781 295 7.80% 3434 90.82% 42 1.10%
3537 3315 205 6.18% 3018 91.04% 222 6.28%
WARD 3 3539 3488 450 12.90% 2981 85.46% 51 1.44%
3200 3038 386 12.70% 2545 83.77% 162 5.06%
WARD 4 5422 5372 1353 25.19% 3967 73.85% 50 0.92%
5023 4919 1222 24.84% 3419 69.51% 104 2.07%
WARD 5 5456 5371 1204 22.42% 4107 76.47% 85 1.56%
4946 4783 1064 22.25% 3542 74.05% 163 3.30%
WARD 6 3206 3178 420 13.22% 2723 85.68% 28 0.87%
2790 2619 432 16.49% 2105 80.37% 171 6.13%
WARD 7 4941 4890 1183 24.19% 3657 74.79% 51 1.03%
4587 4452 1107 24.80% 3117 70.01% 135 2.94%

As shown in the table above, the Republican percentage of the vote increased significantly in Wards 1 and 2, increased slightly in Wards 3, 4, and 5, and actually decreased in Wards 6 and 7. In fact, the actual number of Republican votes decreased in Ward 6. This trend can be studied more closely by comparing the changes for both parties in each ward, as shown in the table below:

GAINS OR LOSSES FOR BOTH PARTIES
Republican Democratic
WARD 1 78 12.60% 541 87.40%
WARD 2 90 17.79% 416 82.21%
WARD 3 64 12.80% 436 87.20%
WARD 4 131 19.29% 548 80.71%
WARD 5 140 19.86% 565 80.14%
WARD 6 -12 -1.98% 618 101.98%
WARD 7 76 12.34% 540 87.66%

Note that the Republican share of the increased vote was 17.79% in Ward 2, which is 11.61% more than what Bush received in the 2000 election. This is also the ward where half the complaints (14 of 28) about touch screen voting machines were filed. In Ward 1, the Republican share of the increased vote was 12.60%, which is 5.98% more than what Bush received in the 2000 election. There were complaints filed here as well, in Precinct 1F, where the touch screen machine recorded “other candidates repeatedly.” In Ward 3, where a complaint was filed in Precinct 3C, Bush’s share of the increased vote was 12.80%, which is 0.10% more than he received in the 2000 election. This might seem insignificant were it not for the complaint that the touch screen “flashes and scrolls, shows different results or no results,” and for voting patterns in the other wards of Youngstown.

In Ward 4, Bush’s share of the increased vote was 19.29%, which is 5.55% less than he received in the 2000 election. In Ward 5, Bush’s share of the increased vote was 19.86%, which is 2.39% less than he received in the 2000 election. In Ward 6, Bush actually had a net loss of 12 votes, or -1.98%, which is 18.47% less than he received in the 2000 election. In Ward 7, Bush’s share of the increased vote was 12.34%, which is 12.46% less than he received in the 2000 election.

It is entirely possible that the increases in Bush’s percentage of the vote in Wards 1, 2, and 3 of Youngstown were due to default settings for Bush on the touch screen voting machines. Both the statistical evidence and the anecdotal evidence are entirely consistent with this conclusion. Even in the wards where Bush lost ground to his Democratic opponent, default settings for Bush may have served to cut his losses. The lack of posted complaints specific to these wards should not be construed to mean that all the touch screen machines in these wards were operating properly.

Touch screen machines are touted as a technological advance because they prevent all “overvotes,” ballots with two or more candidates selected for the same office, and because they prevent “undervotes,” ballots with no candidate selected, unless by deliberate choice of the voter.

“Undervotes” have long been a real problem, undermining the ability of election results to reflect the will of the electorate. Undervotes are especially common with punch cards. If the machines are not emptied frequently they become filled with chads, which can prevent the voter from punching fully through the card, or prevent the voter from fully inserting the punch card into the machine, in which case the punches will not be aligned properly. The machine will then fail to record a vote.

It is an affront to voting rights for election officials to refuse to examine by hand the punch cards that the machines failed to identify as valid votes. There are always people who go to the polls and choose not to vote for a presidential candidate, but not many.

Undervotes are less common with optical scanners, which are machines that scan paper ballots for marks made by voters in spaces next to candidates’ names. There are always some ballots with marks that are not detected by the scanner, and these must be examined by hand if the will of the voter is to be determined. Again, it is an affront to voting rights for election officials to fail to do so. Machines should be utilized to minimize the number of ballots that must be read manually, but should not be assumed to eliminate this task altogether.

In Mahoning County, optical scanners were used for all voting in the 2000 election, and for absentee ballots in the 2004 election. Again, the 2004 data are listed on the top lines, and the 2000 data on the bottom lines.

COMPARISON OF UNDERVOTES, 2004 AND 2000
Cast Counted Uncounted
YOUNGSTOWN 29,016 28,676 340 1.17%
26,223 25,103 1120 4.27%
SUBURBS 86,919 86,233 686 0.79%
78,347 77,128 1219 1.56%
ABSENTEE 18,354 17,966 388 2.11%
12,319 12,011 308 2.50%
COUNTY WIDE 134,289 132,875 1414 1.05%
116,889 114,242 2647 2.26%

The percentage of uncounted votes in 2004 was twice as high for absentee ballots with optical scanners as for Election Day ballots with touch screens. The problem with touch screens lies not in the number of undervotes, but in the reason why the number is so low.

One way to examine the Mahoning County election results is to look for the precincts with the most undervotes.

MAHONING COUNTY PRECINCTS WITH 2.5% OR MORE UNDERVOTES
Precinct Reg Cast Turnout Bush Kerry Undervotes
YOUNGSTOWN 5G 687 316 46.00% 19 247 44 13.92%
YOUNGSTOWN 3G 684 306 44.74% 26 263 13 4.25%
YOUNGSTOWN 3D 686 225 32.80% 16 192 8 3.56%
STRUTHERS 1B 460 237 51.52% 50 178 7 2.95%
CAMPBELL 4B 542 288 52.77% 53 225 8 2.78%
LOWELLVILLE 1 517 329 63.64% 88 230 9 2.74%
YOUNGSTOWN 1F 595 265 44.54% 16 235 7 2.64%
YOUNGSTOWN 3C 813 266 32.72% 34 218 7 2.63%

There are 312 precincts in Mahoning County. In only 8 precincts was the percentage of undervotes 2.5% or more, and 5 of these precincts were in Youngstown.

Youngstown 3C, Tabernacle Baptist Church, 707 Arlington Street, Youngstown, was the one where the touch screen “flashes and scrolls, shows different results or no results.” As a result, 7 undervotes were cast.

But without question, the most interesting of these precincts is Youngstown 5G, Hillman Elementary School, 164 West Myrtle, Youngstown. There were 9 complaints about long lines, slow pace, and “planned confusion,” despite having 6 voting machines. More importantly, this is the precinct where a voter pressed Kerry and the screen “went blank” four times. Five attempts were required to cast a vote. Apparently other voters had this problem also, as there were 44 undervotes, 13.92% of the ballots cast, three times the rate of any other precinct in Mahoning County. It is obvious that nearly all of these were Kerry voters, as he won the precinct with 247 votes to 19 for Bush. It is also obvious that the default setting on the defective machine was for no vote at all. The complaint was not that Bush’s name appeared on the screen when voters pressed Kerry, but that the screen “went blank.” This precinct serves to illustrate how many votes could default to an opposing candidate with one machine programmed in this manner.

Another way to examine the Mahoning County election results is to look for precincts with zero undervotes. In the 2000 election, when optical scanners were used, there were 11 precincts out of 416, or 2.6%, where every ballot cast was counted for president. In the 2004 election there were 21 precincts out of 312, or 6.7%, where every ballot cast was counted for president.

Without question, the most interesting of these precincts is Youngstown 2E, Price Memorial Zion Church, 920 Dryden Avenue, Youngstown. There were many reports of voter suppression and intimidation here.

Since the 2000 election, the 19 precincts of Ward 2 have been consolidated into 12. Price Memorial Zion Church was the polling place for more than one precinct, and voters were not told which side of the room to go to. One voter had to go home, check her registration card, and return again to vote. An 88-year-old woman who had voted at the same polling place for 40 years learned only on Election Day that her precinct had been changed and that her new polling place was farther away.

Another voter who had recently changed her address to this precinct was told by the Republican challenger that her name did not appear on the rolls. When she stated her intention to vote at the precinct for her old address, the Republican challenger took her name and called ahead, so she felt she would be challenged again if she went there. She was so upset by the challenge that she was afraid to vote even by provisional ballot, fearing that it would not count. There were complaints at 7:00 AM, 9:57 AM, 10:26 AM, and 11:45 AM that the Republican challenger was intimidating voters, standing too close, leaning and looking over their shoulders.

But the most damning evidence about Youngstown Precinct 2E is the fact that no undervotes were reported. This is the precinct where two voters selected Kerry and Bush’s name came up, where one voter had to scroll through five times before his votes were finally recorded, and where three voters said that the presidential option never appeared at all while they were trying to vote, and that this happened on two different machines. How could a presidential option fail to appear on the screen, even on one occasion, let alone three times with three different voters, and no undervotes be cast? By setting up a touch screen machine with an unseen default candidate, the one whose name sometimes appears against the will of the voters.

If touch screen machines are to be used, against the better judgment of this writer, it is essential to allow inadvertent undervotes by having a default setting for no vote at all, and that the machines function properly so that the number of undervotes is small. To steer undervotes into the column of a candidate chosen by a computer hacker is nothing short of fraud. The number of votes stolen in this manner may seem small in any given precinct, but it can happen almost everywhere at once, automatically, undetected, without a paper trail.