M E M O R A N D U M
DATE: November 21, 2006
TO: Interested Media
FROM: Jenny Nash
RE: AUDIT PLAN FOR SARASOTA COUNTY
As announced yesterday, the Florida Department of State will begin parallel
testing in Sarasota on Tuesday, November 28, 2006. This is the first testing
phase of an ongoing, comprehensive audit the Department is conducting of the
voting systems and election administrative procedures used in the U.S.
Congressional District 13 race in Sarasota County. The first parallel test will be
conducted on touchscreen voting machines which were prepared for Election
Day but not deployed.
The Department will continue to review all documents and data relevant to the
audit. On Friday, December 1st the second parallel test will take place. This
round of testing will be conducted on touchscreen machines utilized in Sarasota
County for precinct voting in the November General Election. Both parallel tests
will be videotaped. The Department has conferred with outside experts and
experts retained by both candidates in advance of the testing.
The public and representatives of the candidates are invited to observe. The
testing will take place from 6:30 a.m. to 7:30 p.m. at:
The Interim Government Operations Center
1001 Sarasota Center Boulevard
Sarasota, FL 34240
Jenny Nash
Deputy Chief of Staff for Communications
Florida Department of State
Sue M. Cobb, Secretary of State
850.245.6518
Make a Difference - Mentor!http://election.dos.state.fl.us/pdf/auditPlanSarasota.pdfParallel Test
DOE Audit for
Sarasota County FL
2006 General ElectionPurpose:
Sarasota County, Florida experienced an unexpected number of undervotes for the 13th
Congressional District race during the 2006 General Election. Although a number of factors may
have contributed to this undervote total, interested parties are concerned that the undervote for this
race suggests that the voting equipment may not have correctly captured the voters’ selection.
The Florida Division of Elections (DOE) has developed an extensive audit plan to ascertain if a
process, definition, machine, or tabulation anomaly contributed to this contest’s undervote total. As
part of DOE’s extensive audit, the Bureau of Voting Systems Certification (BVSC) will utilize a
test activity that has become known as a “parallel test”. Typically, a parallel test involves a random
selection of voting devices from the population of voting devices destined for deployment on
election day. This test sample would be segregated from the actual deployed devices, but otherwise
would undergo the same election day activities in “parallel” with the deployed voting devices,
except the voters would consist of a test team and the ballots cast would be defined by a
predetermined test script. The intent of this parallel activity is to ascertain the accuracy and
reliability of the deployed voting devices with consideration given to ballot style, layout, coding,
demographics, and operation.
Scope:
The application of the parallel test technique for this audit will deviate from the classical parallel
test in that the test script will be based on the audit data extracted from a sample of iVotronic
touchscreen devices. In addition, the test script will also take into consideration the voting
experience of several voters that were described in various news articles. The audit data for each
iVotronic touchscreen consists of two records: the event log and the ballot images. The event log
contains the timing element for when each ballot was cast. The ballot image file contains the voter
selections as they appeared on the review screen at the time the voter pressed the “VOTE” button.
However, the arrangement of the ballot images is random. Therefore, these ballot images cannot be
associated with the time that the ballot was cast.
BVCS requested each candidate to provide a list of two to four precincts that they believed warrant
close examination. From this list of precincts, BVSC staff identified four iVotronic touchscreens
(two from Jennings’ list and two from Buchanan’s list) that experienced the highest undervote
within their respective precinct. This selection should enhance the probability of revealing the
undervote anomaly should it exist. BVSC personnel then developed a test script from the audit data
extracted from each of these machines. The four iVotronic touchscreens and their precinct are:
iVotronic SN # Precinct Precinct selected by:
V0105192 105 Jennings’ organization
V0106437 118 Jennings’ organization
V0117973 76* Buchanan’s organization
V0106866 113* Buchanan’s organization
* Note: The Buchanan organization recommended a random selection. BVSC performed this
random selection utilizing MS Excel. The Jennings’ organization also identified precincts
118 and 31 in their initial selection and later added precincts 44 and 74.
-snip-
More...