From: Jerry Lobdill <lobdillj_at_charter_dot_net>

Date: Thu Aug 10 2006 - 16:46:33 CDT

Date: Thu Aug 10 2006 - 16:46:33 CDT

I seem to have a typo in my original message. I left out the phrase

"the number of vote counts to audit and" in the next to last

paragraph. So that paragraph should read:

The fact is that you have consistently claimed in emails to me that

the only inputs your technique requires are (1) the number of vote

counts in the county, (2) the number of hypothetical corrupted vote

counts, (3) the desired probability of detecting corruption, and that

with these inputs you calculate the number of vote counts to audit

and what you have labeled in the above referenced document as "Exact

Probabilities of Finding One or More Corrupt Counts". Exact

probabilities? How exact? What if the actual number of corrupted

precincts (which we do not know in any situation) is different from

what you used to calculate your "exact probability"? In what sense is it exact?

Jerry

On Wed. 9 Aug 2006 at 21:26:09 Kathy Dopp wrote:

"Jerry Lobdill provided some sample data from Tarrant County which has

large variation in precinct sizes that produces a drop of 2.5% in the

probability of detecting outcome altering miscount (from 95% to 92.5%)

in a tiny range of small candidate margins if miscounts are thusly

targeted to the fewest largest precincts. So Frank and I added an

adjustment that uses the largest precinct sizes and the total number

of ballots cast in any county to ensure that precinct count audits

probabilities still never drop even a little below the desired

probability."

***************

Let's DO have some honesty here, Kathy!

When, on 7/29/06, I sent you the only data from Tarrant County that I

have ever sent you, the data I sent was a frequency distribution plot

of vote count sizes from Tarrant County and the table from which it

was plotted. ( See

http://www.myimagehosting.com/pic.php?u=778Ke5d2&i=41167) I did this

in the context of an extended email exchange with you in which you

had been increasingly impatient and inappropriately defensive--not to

mention nasty. It was because of a total break down of communication

with you that I posted my message about NEDA going down a primrose

path to this list.

I see that you have been busy trying to slither around and alter your

position on how to do audits so that you can claim that you were

always considering the possibility that vote count size variation is

important. Your paper,

"How Can Independent Paper Audits Detect and Correct Vote Miscounts",

which carries your date of 7/29/2006 available at

http://www.myimagehosting.com/pic.php?u=778Ke5d2&i=41168 makes no

mention whatever of the distribution of vote count sizes. Nor do any

of your papers carrying a previous date make such a claim. This has

been my theme song all along.

Please explain how you and Frank Stenger have used the data I sent

you (Download it at the URL given above.) to come up with your new

paper. Explain how you used that data to conclude that the

distribution caused a "2.5% drop in the probability of detecting

outcome altering miscount (from 95% to 92.5%) in a tiny range of

small candidate margins if miscounts are thusly targeted to the

fewest largest precincts." Your new paper does not say how you

arrived at that result. Nor does it mention anything about the effect

of vote count size distribution. Nor does it describe the

"adjustment that uses the largest precinct sizes and the total number

of ballots cast in any county" that you say that you and Frank

added. This is the very first time that I have heard you say that

this might be important. I suppose that is progress.

The fact is that you have consistently claimed in emails to me that

the only inputs your technique requires are (1) the number of vote

counts in the county, (2) the number of hypothetical corrupted vote

counts, (3) the desired probability of detecting corruption, and that

with these inputs you calculate what you have labeled in the above

referenced document as "Exact Probabilities of Finding One or More

Corrupt Counts". Exact probabilities? How exact? What if the actual

number of corrupted precincts (which we do not know in any situation)

is different from what you used to calculate your "exact

probability"? In what sense is it exact?

As soon as Arthur Keller reports back on the evaluation of my paper

and whatever you submitted I will offer mine as a download to this

group. I hope you will do the same.

Jerry Lobdill

_______________________________________________

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 Thu Aug 31 23:17:06 2006

*
This archive was generated by hypermail 2.1.8
: Thu Aug 31 2006 - 23:17:10 CDT
*