Re: Please Evaluate This Ballot Selection Procedure

From: Charlie Strauss <cems_at_earthlink_dot_net>
Date: Wed Jan 17 2007 - 08:54:07 CST

This recount is similar to one that has been widely promoted. In that
scheme 10% of the ballots are chosen at random and recounted, the
results are compared statistically to the whole. The procedure below
is an elaboration on what is meant by random but does not change the
essential statitical approach.

This approach has advantages and disadvantages. The disadvantage of
this approach is that it breaks the hard connection between the
electonic count and the paper count. That is, the electronic count
is seldom done in a way that individual ballots can be paired with
their paper counter parts, or done in a way that those electronic
ballots are visibly tracable to the electronic vote total. Indeed
it's this latter correspondence that is the whole black box issue
that one seeks to establish. If one were counting entire batch
totals for the machine where one has the machines total then one can
compare this directly to the handcount batch total and the answers
must agree exactly if one is to say no errors were made.

Thus the statisitcal approach is not useful for auditing individual
machines. Instead the sampled approach only works when it's numbers
are large enough to have the statistical power to estimate the final
totals across all machines with satisfactory precision.

The advantage of this process is that it does sample all machines, it
has a predictable workload and so can be planned and budgeted for
every precinct, and that workload is small enough that it may be
possible to audit every race.

There's some techincal issues in its implementation. If a discrepancy
is noted it does not tell you where the problem was. Who performs
it? the same precinct workers that ran the regular election or a new
set off line? Without an immediate check on the accuracy of the
sample in a precinct how can we tell if it was done right?

On Jan 16, 2007, at 10:10 PM, Kathy Dopp wrote:

> I would really appreciate your comments on this selection procedure
> for "randomly selecting" ballots that was proposed by Paul Velleman.
>
> What do you think of it? Thank you.
>
> ---------- Forwarded message ----------
> From: Paul Velleman <pfv2@cornell.edu>
> Date: Jan 16, 2007 6:31 PM
> Subject: Re: [ei] Re: Hand Counting vs Auditing
>
>
> OK. This is probably unwise, since I am sure there are better ways,
> and I haven't studied them. So this comes with a caveat: I haven't
> studied all of the issues involved. So this is just to show a
> reasonable method can be devised easily and not a finalized or
> studied proposal.
>
> There is nothing wrong with a systematic or partly systematic sample
> for this purpose. We want to get a representative sample and give
> every flawed ballot an equal chance to be audited. We also want to
> make it difficult to subvert the audit.
>
> I would do the following:
>
> 1) Generate a fresh set of random numbers. For example, use
> www.random.org, which is public, cannot be influenced, and generates
> truly random numbers. Generate random integers between 0 and 9.
>
> 2) Stack up the ballots. I don't care how, as long as it doesn't
> involve actually reading them. If they come out of the box in order,
> that's fine.
>
> 3) Optionally, shuffle the ballots. This isn't important, but could
> help make citizens more confident.
>
> 4) Take the first random integer. Count that many down from the top
> of the stack.
>
> 5) Then take every nth ballot. For example, for a 10% audit, examine
> every 10th ballot.
>
> 6) Whenever any of the observing parties requests it, take the next
> random digit and count past that many ballots, then resume systematic
> sampling.
>
> This is not a simple random sample, but it should be sufficient for
> auditing. In fact, if the ballots have some natural order (e.g. the
> order in which they were cast) it would have the advantage of being
> somewhat stratified--for example, it would be guaranteed to sample
> roughly the same proportion of early votes as of late ones.
>
> -- Paul
> _______________________________________________
> OVC-discuss mailing list
> OVC-discuss@listman.sonic.net
> http://lists.sonic.net/mailman/listinfo/ovc-discuss

_______________________________________________
OVC-discuss mailing list
OVC-discuss@listman.sonic.net
http://lists.sonic.net/mailman/listinfo/ovc-discuss
==================================================================
= The content of this message, with the exception of any external
= quotations under fair use, are released to the Public Domain
==================================================================
Received on Tue Jan 1 14:12:45 2008

This archive was generated by hypermail 2.1.8 : Tue Jan 01 2008 - 14:12:51 CST