Re: statistical study after next Tues election

From: Kathy Dopp <kathy_at_directell_dot_com>
Date: Mon Nov 01 2004 - 12:27:01 CST

Douglas W. Jones said:
>
> On Oct 31, 2004, at 12:23 PM, Kathy Dopp wrote:
>
>> Douglas W. Jones said:
>
> DRE machines and lever voting machines do not permit overvotes to be
> recorded, so you get overvote counts of zero. The statistics I've seen
> show that DRE machines seem to be recording undervote rates of between 1
> and 2 percent,

Doug,

Aren't those two statements contradictory? How can DREs have zero
over-votes (that is what I thought) and also be recording overvote rates
of 1 to 2 %?

> This implies that around 1 percent of the voters who encounter DRE
> machines
> seem to have trouble expressing a preference that they would be able to
> express on mark-sense ballots, a source of major concern.

I still don't get what you're saying.

>
> If we could find differences in undervote rates between counties that
> correlated to the voting systems being used, after correction for
> demographics (for example, pairing counties by median income or some
> other census data), we'd begin to learn what it is that is causing this.
> My guess is that the central problem is one of bad GUI design, but this
> is only intuition.

Yes. I am all for examining undervote rates by county and considering
demographics as well as voting machine types. Let's do it.

>
>> The postal absentee plus paper-based early voting could be the
>> "control group" with which to compare the election results by county
>> to look for patterns.
>
> Yes, but you don't know that the early-voting rates correlate with the
> election day rates. I expect that undecided voters don't generally get
> sucked into early voting, while strongly committed voters don't mind
> voting early.

Yes, but I would expect to find that the pattern of "differences" between
early votes and election votes might be similar amoung various counties
and "not" expect to find differences (absentee vs. precinct) to vary by
voting machine type unless there was possible shanninigans going on. I'd
be looking to see if the amount or direction of differences were more
similar with op-scan or paperless e-voting counties (correction by
demographics could be looked at and would be more expected than
differences based on type of voting machine).

I have always envisioned this study as studying the patterns of
"Differences" between different types of election results or polls by
county. The pattern of differences would be expected to be the same or
vary by demographics, NOT by types of voting machines, unless there were
some shanninigans.

>
>> Are these three breakdowns of "in precinct, postal absentee, and early
>> voting" available on
>
> ... on what? Anyway, depending on the state rules and county
> flexibility, the early vote may not be separated from the absentee
> vote. In Iowa, for
> example, early voting is just on-the-spot absentee voting administered
> by
> county employees in what they call "satellite polling places". You sign
> the request for absentee ballot, they look you up in the voter rolls,
> and
> you vote, all in a space of minutes, and then you stuff your ballot in
> an absentee ballot privacy folder, stuff that in an envelope, and drop
> the
> whole thing, paperclipped to your absentee ballot request form, into the
> absentee ballot box. So, no distinction.

Is that true in all Iowa counties? Well that is perfect then because the
technology used is the same for both early and absentee, so I would lump
them together for comparing with precinct. Although, you are right that I
couldn't then directly compare Iowa's results with Florida's where early
voting in electronic voting counties is apparently done electronically,
not on paper ballots.

You're right, this study will not be easy, but there may be interesting
patterns that appear. I don't know yet if Florida election results are
available broken out by early voting vs. absentee vs. precinct, but in
Florida, I would lump the early voting with the precinct and compare
versus absentee paper ballot voting, looking at demographics in parts of
the study as you suggest.

>
> In some areas, early votes cast at satellite polling places are entered
> on DRE machines, while postal votes are on paper. These totals must be
> done separately, but whether they lump them for public release or keep
> them separate will depend on state rules.

If they are lumped together for public release, then that will make a
planned study comparing election results cast using different technologies
impossible. We need them separately.

>
>> I have chosen five swing states to study: Florida, New Jersey, Ohio,
>> Arkansas, and Iowa. Arkansas and Iowa are control states, having no
>> touchscreen machines.
>
> But we have a double handful of counties that use DRE voting machines
> here, mostly Microvote MV464 push-button machines. Not touch screen,
> but definitely not VVPT. On the other hand, we've got 99 counties, so
> these make up under 10% of the counties, and they aren't the big urban
> counties.

I know. That is why I have from the beginning considered Iowa as a
'control' state. IMO Florida is the primary state where we might find
differences between electronic and op-scan counties. ES&S voting
machines, from past articles I've read, are the most likely where any
anomolies may occur. Arkansas would be a control state of another type
because it has lots of electronic voting machines that are paperless but
not touchscreen ES&S, so anomolies on paperless e-voting machines of a
different type could show up there as well. Ohio and NJ each have just a
few touchscreen paperless e-voting machine counties using ES&S. If we can
obtain all the numbers, we can look at them in more than one way.

>
>>> Visit county web sites. You can grab sample ballots there, and these
>>> samples are important!
>>
>> Is that because residual errors may be due to ballot design? Darnit,
>> I need an assistant to do this study. It is a huge project just to
>> gather
>> all the data and info.
>
> Indeed. My preliminary analysis of the data from election 2000 in
> Florida shows lots of dependency on typography for optical mark-sense
> ballots.

What is typography? What do you mean?

>
>> Youch! What are you saying? The numbers posted on the county web
>> sites keep changing? Yikes.
>
> Depends on the state and the county. Some will put up unofficial
> early returns, first official canvass and second canvass as separate
> items. Some just keep one public portal that holds the current best
> available numbers and keeps changing until the lawyers are agreed.
>

Wow. Well wouldn't we want the final numbers for our study anyway? What
are some examples of why numbers would change?

Thank you.

Regards,

Kathy

==================================================================
= The content of this message, with the exception of any external
= quotations under fair use, are released to the Public Domain
==================================================================
Received on Tue Nov 30 23:17:02 2004

This archive was generated by hypermail 2.1.8 : Tue Nov 30 2004 - 23:17:43 CST