Every Vote Counts?

A Snapshot of Election Day 06 Problems at the Polls in Montgomery County, Pennsylvania

March 1, 2007

Election Reform Network

921 Melrose Avenue

Melrose Park,PA 19027

215-782-8218

Introduction

This report is organized into two sections: the first is a description of findings related to the Election Reform Network's (ERN) comparison of the number of voter signatures in poll books for 25 election districts with the machine count of voters provided in the 2006 General Election Precinct Turnout Report, Montgomery County, Pennsylvania. The second section is an overview of problems and incidents in about 15 precincts in Montgomery County, where ERN poll watchers and other observers were placed. Together, we believe the two sections provide a useful "snapshot" of how elections operate in Montgomery County and point to some obvious directions for improvement.

This report, however, does not address directly what the ERN believes is the most fundamental question surrounding Montgomery County elections, the security and reliability of its current direct recording electronic (DRE) paperless vote machines, which are neither independently auditable nor recountable.

Section I. The Undercount Problem

Findings

Our findings in this section are based on what we believe to be a reasonably representative sample of 25 out of the 407 precincts in Montgomery County. The method employed was simply to compare the difference between the number of signatures in the poll books for the 25 precincts surveyed and the number of voters whose ballots were recorded in the machine count in those precincts, as provided by the 2006 General Election Precinct Turnout Report, Montgomery County, Pennsylvania. Findings indicate that out of the 25 precincts surveyed, 19 of them showed a disparity between the number of signatures and the machine count. In all but four cases, there were more signatures than votes recorded. (It is unclear how to interpret the cases where more votes are registered in the cartridge than the number of signatures, but considering their very low number, it is possible that the source of the problem is the failure of voters to sign in, rather than faulty machine counts.)

The average number of "undercounted"votes, defined as the difference between the number of electronically counted ballots and the number of signatures in the precinct poll books is slightly more than three per precinct. (Precincts with greater cartridge counts than signatures are not included in the calculation.)

What follows is the precinct-by-precinct breakdown of the comparative analysis.

Precinct Number of Signatures Cartridge Count Disparity

Bridgeport 1 289 289 0

Bridgeport 2 414 414 0

Bridgeport 3 334 334 0

Cheltenham 1-1 770 765 5

Cheltenham 1-2 600 602 2*

Cheltenham 7-3 649 649 0

Conshohocken 4 502 501 1

Conshohocken 5 352 352 0

E. Norriton 2-3 703 702 1

Lower Pottsgrove 3 829 827 2

Norristown 1-1 442 439 3

Norristown 1-2 530 527 3

Norristown 1-3 504 504 0

Norristown 2-1 469 464 5

Norristown 2-2 294 297 3*

Norristown 2-3 362 358 4

Norristown 3-1 728 731 3*

Norristown 3-2 421 420 1

Norristown 3-3 664 659 5

Norristown 4-1 551 537 14

Norristown 4-2 696 693 3

Norristown 4-3 751 741 10

Pottstown 5 789 782 7

Pottstown 6 906 905 1

West Conshohocken 493 494 1*

Conclusions

Montgomery County has long known that there is a danger of "undercounts" as a result of its Sequoia AVC Advantage machines. County officials have referred to the problem as the "fleeing voter syndrome," a turn-of-phrase for the problem of voters who make their selections on the screen and then do not have their ballot recorded because they fail to press the "cast vote" button before exiting the voting booth. While there is no decisive evidence that the problem we identify is not caused by machine error rather than poor user interface (see below), we believe that because of the relatively small numbers involved, it is more than likely that the "undercounts" are due to "fleeing voters" rather than machine errors or manipulation.

After 10 years with the Sequoia machines in Montgomery County, then, votes continue to be lost. Obviously, some voters still have not caught on. While we can pin blame on individual voters, it seems clear that the problem is a function of the fundamentally poor design of the Sequoia push-button electronic machines. That is, on top of the voter-verifiability, security, reliability, auditability and recountability deficits generic to all direct recording electronic machines, including the Sequoia, the awkward interface between the voter and the county's machines presents the dilemma of the inevitability of lost votes. The point is that even when the machine is recording and tabulating accurately, votes are lost when the button is not pressed.

What may be most disturbing about this problem is that despite knowledge of its existence, the Montgomery County Election Board has never done a quantitative study of the problem. Our findings from a small sample of total county precincts indicate that the problem is likely a significant one. An average of more than three lost votes per precinct could be easily higher during Presidential election years, when polling places are more crowded and voters are in a greater rush. But assuming an average of three votes lost per precinct, that amounts to a total of more than 1,200 votes lost countywide.

Considering that the margin of victory in a 2006 state legislative race in the county  was in the area of 100 votes, those seemingly inconsequential three votes begin to have serious implications. Those lost votes, of course, can be the difference in close elections, but they also represent the erosion of a fundamental American ideal: that every vote counts and must be counted.  The Election Reform Network believes that that the County could do far more to improve our elections and better strive to meet that most basic American ideal. 

Section II: Election Day Operations

Findings:

The ERN monitored some 15 polling places last November 7 and observed numerous problems, which fall into the following categories:

Late openings

ERN observers identified one polling place, Norristown 2-3, which opened 30 minutes late. Observers witnessed about five people who went inside and turned around and left the poll without voting before 7:30 a.m. Others who may have come to vote but never entered the building and were unseen by the observer inside would also have left without voting. The ERN was unable to make a determination as to whether any discouraged voters were able to return to vote later in the day. In addition, there were rumors of other polling places opening late, however, they lacked specificity and could not be confirmed.

Voters wrongly directed to other polling places

Some 20 potential voters who professed to be long-time residents at their current address who have voted regularly at their polling place, St. Patrick's Church, were directed to alternate locations to vote, according to an ERN poll watcher. In the case of two family members who lived next door to one another, one was allowed to vote and one was sent elsewhere. The same thing happened to a husband and wife who live at the same address. The election judge claimed that he was told by Election Board personnel to send people to alternate locations. The judge thought that the county should have sent out a letter notifying people of a redistricting (there was none).

At another precinct, a similar story unfolded. One man who claimed he had voted there in 2004 was not allowed to vote this time. However, his mother, who lived at the same address, was able to vote.

Rampant Confusion related to Provisional Ballots

Another election judge told voters that if their name did not appear on the rolls, they were not entitled to vote by provisional ballot (the opposite is true). He said that he was told this by Election Board personnel at a training. When he was finally prevailed upon by the ERN poll watcher to call the Election Board for advice, he was advised to offer provisional ballots when people were not found on the rolls. However, the judge remained reluctant to offer provisional ballots or call the Election Board on individual cases. The ERN poll watcher suggested to the judge that he encourage people who were at the wrong polls to go to their correct polling place. The judge seemed confused and disinterested.

Pollworkers Leave Polling Place

Two pollworkers departed their polling place, leaving the judge of elections without any back-up for about 90 minutes.

Voters Wrongly Asked for Identification

Many voters in another precinct were asked for identification before giving their names and being looked up in the poll books, in clear violation of the law.

Machine Breakdowns and Mechanical Errors by Pollworkers

Pottstown 3-1: Machine down for 20 minutes. Judge called Election Board and machine restarted.

Norristown 4-3: Two machines did not start up, causing a long line during the morning rush. Warehouse was called and judge was told to reinsert the cartridges; the machines started.

Lower Salford 2: Machine down for 2 hours

Upper Salford: Machine down after 7 votes; machine replaced at about 2 pm.

Whitpain10: Machines had not been turned on, while people attempted to vote. Six votes lost.

Conclusions

Taken individually, most of our findings concerning election day operations on the ground are neither extraordinary nor catastrophic. Problems such as confusion over provisional ballots and registration foul-ups may not be uncommon in other jurisdictions. Taken together, however, confusion over provisional ballots, registration problems, machine breakdowns, inadequate training, late openings, and understaffed polling places paint a disturbing and dysfunctional picture.

The first prerequisite for any solution is the understanding of a problem by asking the right questions. The ERN undertook this citizen study because Montgomery County election officials were not asking any questions. When we asked for reports on recent prior elections, we received no response. We again asked for a report on the last November election and even identified the key issues we thought should be addressed. The Department of Voter Services responded by providing a summary listing of countywide machine problems, citing eight election districts out of the four hundred and seven total, with one-three word descriptions and no indication of impact. A computer print-out of provisional ballots was also included. In contrast, the ERN identified machine problems in five additional election districts out of the approximately fifteen that we monitored. Again, knowing the size and scope of a problem is vital for finding solutions.

Considering the relatively few polling places where ERN observers were posted, the problems we found were numerous and troubling. Sadly, we find little reason to believe that what we discovered on election day does not reflect similar problems across the county. Some problems, however, are clearly more serious than others. A late opening by 30 minutes and an apparent massive mistaken rerouting of some 20 people at one polling place - whatever the real source of the problem - demands investigation and solution. In addition, there are clearly some major problem areas needing analysis and remedies that cut across jurisdictional lines. Provisional ballot rules and training to implement them need a complete reassessment and probably revamping. Registration foul-ups may also fit that category. The SURE registration system probably needs some major holes to be plugged.

Unfortunately, the ERN finds that when it comes to election issues, it is common for one level of government to slough off a problem and drop it into the lap of another. Of course, the only hope of real solutions is based on different levels of government collaborating for change rather than passing the buck. If election administration is to truly enter the 21^st century, it is not going to be via inappropriate unchecked, insecure technologies that undermine democratic accountability and voter confidence. It will be based on a proactive approach to solving difficult problems through creativity and collaboration between different levels of government and between government and the citizens it serves. The ERN hopes that Montgomery County will start leading the way.

Lastly, ERN observers discovered first-hand that voter disenfranchisement leads to a profound undermining of what we have in common as citizens. There were too many instances on November 7 of people being angry and dispirited about their inability to vote. They genuinely felt left out and shunted aside. The good news is that people care and want to participate. The challenge is to put systems in place that can activate that desire to be part of the exercise of democracy. If we can do that, as a county and as a country, we will all be stronger for it

* The cartridge count was greater than the number of signatures in these four precincts.