Reply #115: Scenarios re: mispunching

[Iceburg]

Mistabulation errors can originate from:
a) mispunchs at the local precinct
b) misreads during the card reading process (machine reader problems, card stock problems, chad problems)
c) wrong/misplaced precinct header cards
d) ballot card mix-ups at the counting station (example dropping a deck of stacked ballot cards) before commencing the counting process). Trust me it happens!
e) miscounts during the central tabulation process due to flawed logic

In this manual recount we are only considering item a) above, the results of which will be compared to the post-precinct activities and its associated results.

Scenario 1 Mispunchs - Correct machine/votebook setup but ballot cards crossed precincts
====================================================================
Assumptions
1) Precinct identifier is stamped somewhere on the ballot

2) You know what machine had what ballot style designation. Each precinct will have multiple machines. Most of the machines were installed before poll opening. A few precincts will have had machines added after poll opening due to problems with the other machines. Be on alert for precincts with machines added after poll opening. In Cuyahoga, a significant percentage of the precincts where machines were added indicate gross mistabulation of ballots. It is likely in these cases the new machines were either configured incorrectly or placed in the wrong precinct with that vote location (essentially this is just a special case of machine configuration error). The BOE should have some audit records with respect to the machine#/ballot style relationship and when the machines were placed in the vote location.

3) Vote Books on machines were set-up correctly. Unfortunately there is know way of knowing that with precision without examining the machines/ballot frame/vote book which I assume have been reconfigured for other elections by now. A less precise method, would be to get the vote count/distribution by machine for each precinct. The BOE does not publish this in public places but they will have an audit record for each machine. This is very important information to obtain.

4) No ballots have been replaced others post-election(I call this the "Conservation of Ballots Law" -- no ballots created or destroyed)

Possible outcomes:

Used the right card on the right machine - detectable with pct stamp
Used the right card on the wrong machine - detectable with pct stamp
Used the wrong card on the right machine - detectable with pct stamp
Used the wrong card on the wrong machine - detectable with pct stamp

Scenario 2 - Incorrect machine/vote book setup in the precinct
====================================================================
Assumptions
1) The precinct has more than voting machine
2) You know what machine was alleged to have what ballot style designation. Same details as assumption #2 in Scenario 1.

It is highly unlikely that all machines in a given precinct will have been configured wrong -- if so it will be patently obvious (as in the case of Cleveland 4N and 4F).

So the goal here will be to identify the machine whose vote distribution data contradicts the data from the other machines' data in that specific precinct.

Unless we know the candidate vote counts by machine (the lowest level of granularity) it will be impossible to detect whether the ballots have been mis-punched.
-------------------------------------------------
To summarize, if you want to find out which ballots were mis-punched (as per scenario 1 and 2 above), you need to confirm that the precinct identifier is stamped on the ballot card plus, you need to know:
a) the exact ballot style deployed in the precinct
b) the ballot style to machine number relationship
c) the vote distribution by machine (precinct count alone is not sufficient)

It is possible to conduct the recount with no consideration to scenario #2 -- ie. machine configuration errors, but it will be at the risk of excluding one of the more likely causes of mistabulation.
I believe (from my analysis) that there was some random cross-precinct voting, but I strongly believe that the bulk of the mistabulations were due to precinct machine configuration problems, and problems during the central tabulation process.

Hope that helps.