determining vote count audit sample size to detect outcome-altering miscount

From: Kathy Dopp <kathy_dot_dopp_at_gmail_dot_com>
Date: Sat Jul 22 2006 - 19:31:45 CDT

Determining Vote Count Audit Sample Size to Detect Any
Outcome-Altering Vote Miscount

Everyone (NEDA and OVC),

However, we have a great solution.

Frank Stenger wrote a simple 5-line maple (or mathematica) function to
calculate the exact sample sizes needed to detect outcome-altering
vote miscounts. He uses the Gamma natural log function to calculate
the sample sizes using the following inputs:

N = # total vote counts

C = #minimum corrupt counts to alter any election ( C = (minimum
candidate margin)/(maximum margin swing per machine))

P = desired probability for detecting at least one vote miscount

maple and mathematica algorithms are time-tested for calculating the
sample sizes to any desired accuracy, in fact, much more accuracy than
we need. His solution is correct, without question.

The program can be refined (round up to the nearest integer sample
size; and permit the input of candidate margin and max margin-swing
per machine) and in other ways.

I am proposing that the Open Voting Consortium and the National
Election Data Archive write a joint paper on determining vote count
audit percentages. There has been much discussion on the OVC list (by
Laird, Ed, and Jerry) on how to generate charts to make this new
method for calculating audit percentages easily understood by the

We may be able to apply the maple program on NEDA's server and on
OVC's server to display the exact vote count audit sample sizes needed
for any election and county. Or we could create numerous charts or
tables and post those charts or tables to use to look up values.

The program's source code would be made publicly available.

Does everyone agree on doing such a joint "vote count audit" project
between the OVC and NEDA?

Kathy Dopp
National Election Data Archive
OVC-discuss mailing list
= The content of this message, with the exception of any external 
= quotations under fair use, are released to the Public Domain    
Received on Mon Jul 31 23:17:07 2006

This archive was generated by hypermail 2.1.8 : Mon Jul 31 2006 - 23:17:09 CDT