|What is a Citizen Exit Pol ?
(also known as Parallel Elections)
Parallel Elections (PE)
A PE is a citizens' written exit poll. A PE is an attempt for citizens to regain control over elections by
conducting their own independent, unofficial election outside official
polling places, and invite voters to participate on their way out of
the polls. It is not unusual to have 50% participation, which is much
higher than exit polling, which polls about 5% of voters. *****
PEs participants sign an affidavit that they voted the same way in
the PE as they did in the official election, then fill out a secret
paper ballot, fold it and put it into a secured ballot box. The ballot
box is sealed and witnessed by the last voter and brought to a public
counting place and counted in public. These extra steps give PEs high
reliability and usefulness. The data is analyzed after the election
and we look for anomalies in the data, particularly in the differences
between the PE results and the official results.
Statistical analysis is important here, and we are now considering
how to show people to do this on their own.
The Birth of PEs
In March 2005 Former election official Ellen Brodsky conducted a parallel
election in Coconut Creek, Florida. Her purpose was to test the accuracy
of the voting machines used in a special election concerning a gambling
initiative. 67% of the voters in that precinct participated in the parallel
election. The PE gave a few voters an immediate opportunity to write
and sign affidavits stating that, for them, either the candidate or
the initiative had not appeared on the screen of the electronic voting
machine. The affidavits provided admissible court evidence, called "past
recollection recorded," to demonstrate that the electronic voting machines
did not consistently contain the required ballot information. To read
the entire description of this Brodsky's first parallel election, go
to Lynn Landes' website www.ecotalk.org.
The Genesis of the San Diego Special Election PE (July 26th, 2005)
In mid July 2005, election justice activist Jim Hamilton asked me (Judy
Alter) to help audit the mayoral run-off race in San Diego on July 26. When
Jim and I decided to conduct parallel elections, I consulted Ellen Brodsky. I
followed most of her procedures and modified some in cooperation with
the 23 volunteers who came from seven CA counties. We conducted five
parallel elections at polling sites that housed 11 precincts.
As part of our preparation, Jim contacted the San Diego Registrar of
Voters and requested the vote counting schedule for the absentee ballots
and the mandatory hand recount of 1% of the precincts, required by CA
election code. I thought we should witness these counting sessions as
part of the audit. I did not receive this information and we did not
The short lead time only gave us a week to publicize to the voting
public that citizens were going to conduct parallel elections. The more
information voters receive about parallel elections, the more they should
be willing to participate. We want as high a number of votes as possible
to compare to official voting results. The publicity, however, should
not include the locations of the parallel elections.
When planning to conduct parallel elections, gather a team: a person
experienced in rapid media response, a legal advisor familiar with election
law, a research statistician or two, and someone who can scout the best
poll sites. Brodsky recommended choosing polling sites for parallel
elections that have only one entrance and exit for voters. This single
entrance enables all voters to pass the parallel election table. Even
sidewalks on both sides of the table make it difficult for volunteers
to greet and hand information to all voters. In addition to a single
entrance, from our experience in San Diego, we recommend that volunteers
choose sites with previous records of high voter turnout.
Although a parallel election is non-partisan, we located the table
beyond 100 feet from the poll door and publicly measured the distance. To
set-up of a parallel election a card table is a minimum size but a longer
folding table is probably better. Hang a sign on the table identifying
it as a citizen audit parallel election.
The San Diego PE Gets up to Speed
At the beginning of the election, one volunteer should go into the
official poll to introduce him/herself to the poll workers and inform
them about the parallel election citizens are conducting outside their
site. Reassure them that the purpose of the parallel election is to
double-check the accuracy of the election machinery thereby enabling
voters to trust the official tallies.
Many of our volunteers at the five sites encountered hostility from
the poll workers. One told voters not to participate and another told
them to vote for different candidates than they had voted for in the
official election. Voters actually reported these incidences to us and
wrote them in affidavits. One inspector called the police. Even though
the police reassured the inspector that a parallel election was legal,
four police cars remained in the parking lot of that site for many hours
of the day. The voter participation rate there was only 25%, the lowest
of our five. After the police cars left, voters eagerly participated. At
other sites official election inspectors showed up off and on all day.
At only one site were the poll workers cooperative.
The materials for a parallel election are readily available and inexpensive:
a box, a roster book, such as a composition book, 5-6 clipboards, markers
and pens, a tablet and envelope for written voting irregularities,
and the printed documents. The essential documents include the voter
instruction sheet (see attached document) and the ballot. It must not
look in any way like the official ballot but needs to contain the same
information. Because it is difficult to estimate the number of ballots
to print ahead of time, print more than you think you will need. Also
have available an information sheet about voting justice and election
problems for further voter education.
The set-up and procedure of a parallel election utilize the same safeguards
as an official one. We sealed the parallel election ballot box inside
and out and asked the first voter to witness that the box was empty
before voting. We sealed the lid to the ballot box with masking tape
that first voter signed as did a volunteer.
After voters had voted in the official election, us we asked participating
voters to sign and print their names in a roster book agreeing that
they were "voting voluntarily in this parallel election and pledging
to vote the same way as in the official election." The signatures provide
evidence for legal procedures if they are required to challenge the
election. Though Ellen Brodsky asked voters to sign their ballots, the
San Diego group wanted to maintain the use of a secret ballot, so, instead
we used a roster book for voters' signatures.
Voters then picked up a clipboard with a ballot, stepped away from
the table to fill out their ballot privately, folded and deposited it
into the box. The process took voters less than a minute. We invited
voters who turned in their absentee ballot on election day to participate
also and simply to mark an A for absentee in the roster book and on
their PE ballot. We identified the ballots as absentee because officials
do not count them on the precinct scanners and we wanted to get an accurate
total of the election day voters.
Voters in the parallel election in San Diego, made many positive comments
as they participated. The people who turned us down said they did not
have time or just said, "no thank you." The volunteers conducting the
parallel election felt exhilarated in their civic engagement with the
public. I felt very moved by the trust the voters showed us, fellow
citizens, especially in their willingness to sign their names before
The election started and ended at the same time as the regular election:
7:00 a.m. to 8:00 p.m. Like official elections, we asked the last voter
to witness the sealing of the slot in the ballot box and to sign the
tape as did one of the volunteers. While one volunteer packed up the
supplies, another witnessed the closing of the official poll and waited
to copy the official totals that the poll workers were required to post
on the precinct door. This important chore required patience by all
involved. In two of our 5 sites, the poll workers refused to allow volunteers
to witness the closing processes of the official poll. This was another
of several violations of election regulations we encountered.
Whereas only 25% of voters participated at the site where the police
cars remained most of the day, the response rate at the other four sites
was 38%, 57.5%, 58.5%, and 66%. The overall response rate for our five
sites was 49.8%: 1516 out of 3045. These voters wanted to double-check
the accuracy of Diebold Accu-Vote scanners used to tally the ballots
at the precincts.
To reassure voters that we volunteers would not tamper with the parallel
election paper ballots, the volunteers signed the sealed box and signed
again on the last page of the roster book. We wrote on that page a pledge
that we would not tamper with the ballots. Then we put the unused ballots,
if any, into the envelope in which they came and marked the number of
empty ballots next to the total number we started with. We put the official
posted totals into the parallel election roster book to keep the book
and official tally together, and we taped the ballot envelope and the
roster book to ballot box. We then put the box into the trunk of one
of our cars and when possible photographed this act.
The San Diego PE Results
Earlier in the day, Channel 10 News broadcasters had picked up the
police call on their scanner and had gone to the poll site where we
were conducting a parallel election. They provided wonderful unsolicited
news coverage to our action by broadcasting an interview with the Registrar
of Voters, Mikel Haas. Haas announced on T.V. that what we were doing
was legal. The news about our parallel elections with that interview
led the 9:00 news.
When the PE volunteers arrived at Denny's where we arranged to count
the ballots we found that the management did not want us to videotape
in the restaurant to protect the privacy of their customers. But we
found the director of Channel 10 news waiting for us. He invited us
to count our PE ballots in their conference room where we counted with
two cameras rolling, theirs and one brought by one of our volunteers.
The parallel election led the 11:00 pm news including a short interview
with me and shots of us counting the ballots.
We counted the ballots using the method that election officials use
to conduct a manual recount: one person reads out the candidate's name,
one witnesses this, and a third person records the vote. If possible,
two people can record to double-check each other. After this first step,
we then sorted the ballots into piles for each candidate to double-check
the first, oral tally. Another person counted the names in the roster
book and compared the total number of voters to the total number of
ballots. At the end of this careful and laborious counting process we
then compared our results to the official results and three statisticians
Though a mayoral election is non-partisan, the leading candidate, Donna
Frye, is a Democrat. She won 43% of the total votes in the official
tallies but in our small sample, she garnered 50.1%. Two known Republican
candidates came in second and third. In our parallel election results
these candidates earned 2% less than in the official results. We analyzed
the votes at our five sites by political party and found that at three
sites more Republican voters participated than Democrats. In the two
sites where Frye won, a similar percentage of voters voted in both elections. In
two sites one of the Republican candidates received the same percentage
of PE votes as in the official election and in two others sites we found
the same pattern for the other Republican candidate. These observations
at first led us the think that we had a representative sample of the
The statisticians who analyzed our results agreed that the differences
between the PE tallies and the official ones could not have happened
by chance. Said another way, the probability of these results occurring
is one in a million. They suggested that we probably did not have a
random sample but we would not be able to determine why the differences
were so great unless we recounted our precincts. Other explanations
included that our PE voters were biased, that they did not vote honestly
with us, or that the scanners counted wrong.
We requested a partial recount of our 11 precincts, 12 more precincts
that we chose as allowed by law in the process of a recount, and all
of the ballots hand-counted in the mandatory 1% of precincts that we
had wanted to witness when it occurred the first time. The RoV had told
us that he did not select these precincts randomly as required by law. We
found the recount results inconclusive. The people recounting found
4 variations in our 11 precincts, and 10 variations in the surprise
12 precincts. In the 1% manual recount ballots, that constituted 48.8%
of the 11000 ballots they found no variations.
The officials told us they had not sealed the ballots from July 26
until August 23rd because they knew we were requesting a recount. In
fact during the one and a half days of the recount we only saw 4 out
of 31 packages of ballots that were sealed in their plastic bag. Thus
we do not know whether or not anyone tampered with the ballots.
Since our PE included only .6% of the voters in the SD run-off election
and our partial recount covered only 4.4% of the ballots we conclude
that our numbers were too small to demonstrate whether or not the machine
tallies were accurate. And the partial recount revealed serious problems
with the way officials conducted it. I believe that we must conduct
many more parallel elections in larger elections to understand their
Observations and Recommendations
The following recommendations come from our experience in San Diego:
- Use teams of three rather that two. Have one volunteer oversee voters
signing the roster book. Trade roles of greeting voters and replenishing
the clipboards with ballots.
- If possible send in relief volunteers for the last 3-4 hours of
the election. Make sure they are trained in the protocol: voters must
sign the roster book and then vote.
- Become familiar with official election procedures and voters rights. Be
able to contact your legal advisor throughout the day.
- Organize parallel elections teams with at least one person familiar
with the neighborhood services near the poll, especially food and
- Strategize with your statistician and media person planning to analyze
and respond rapidly to any and all news coverage.
- Associate conducting parallel elections with the possibility of
- Plan and arrange for collecting donations to fund both the materials
for the parallel election and recounts if required.
Parallel elections serve as a form of exit poll with fewer limitations. In
exit polls, interviewers ask voters to participate by using numerical
intervals such as every 3rd or 6th voter. The interviewer does not remain
at the poll all day while PE volunteers do. If the interviewer needs
to choose between two voters who emerge from the poll at the same time,
that choice introduces bias into the sample. Thus, what interviewers
intend to be a random sample may not always be random. In contrast to
an exit poll a parallel election has voters vote in secret; they do
not say out loud for whom they vote. Participation is not limited to
a random sample of voters but invites all voters to participate. Both
surveys, however, are limited because voters participate, OR NOT, voluntarily
so neither may produce a random sample.
In California, we are applying what we learned in San Diego about how
to conduct parallel elections to our goal to conduct them in as many
CA counties as possible for the Nov. 8 special election. Since there
is evidence that in the 2003 Recall election Arnold Schwarzenegger only
won on Diebold touch-screen machines and scanners, we are very concerned
that the upcoming election tallies may not reflect the true choices
of the electorate.