Nov 08 Excerpt: Audit Organization

Lack of Appropriate Notification to Selected Towns and to the Public

The Secretary of the State’s office notifies the towns by means of an e-mail and voice mail that they have been selected to participate in a post-election audit. However, e-mail addresses change, messages can be lost in transit or incorrectly classified as spam, phone numbers change, and part time registrars often go days without listening to voice mail or checking emails.

Despite a requirement that municipalities provide the Secretary of the State with three business days notice of the schedule and location of an audit  and despite repeated calls to registrars, we were unable to determine dates, times, and locations of all audits with the required advanced notice.  Although we note continuing improvement in this area, we also note the following problems:

• Several towns indicated they had first learned of their selection from another town or read of their selection in the newspaper. Several learned of their selection from our call.
• One municipality called us mid morning to notify us of an audit beginning at 4:00pm that same day. They explained that they could not provide longer notice as they had determined the audit date and time that same morning.
• One municipality stated several times that they were not selected for audit as the Secretary of the State had not notified them of the audit. We also passed this information on to the Secretary of the State’s Office. Unknown to us until after the fact, that municipality performed the audit one day after the close of the audit period.

Incorrectly Completed Forms

An analysis of the eighty-four (84) district reports submitted by the municipalities to the Secretary of the State, we note that fifteen (15) reporting forms were not accurately completed.  Although this represents an improvement from past audits, it is difficult to create comprehensive statistics or to rely on the audit as a vehicle for assessing the scanners’ ability to count every ballot.

• Seven (7) districts submitted forms that omitted either the number of machine-counted ballots to be included in the audit or the number of hand-counted ballots that were actually counted .
• Two (2) districts audited only four races of the five required by the Secretary of the State.
• Four (4) districts submitted report forms on which Column D or Column F was not completed as required[1].

C. Guidance on Counting Procedures Inadequate, Inconsistently Followed

In our August report we warned of ongoing problems with counting due to a lack of detailed guidance on counting procedures. Unfortunately, our prediction was largely confirmed by our observers. We base this conclusion on the discrepancies in the official audit reports submitted to the Secretary of State’s office, and in statistics that can be derived from those reports (Covered in the “Audit Statistics” section of this report). Specifically, observers note confusion over definitions and the lack of “blind counting” as major concerns. In addition, observers’ overall impressions do not paint a picture of confidence in the counting methods used to create the audit reports:

• In nineteen (19) audits, observers had concerns that the auditing was not well organized.
• In thirteen (13) audits, observers had concerns with the integrity of the counting and totaling process.
• In thirteen (13) audits, observers had concerns that the manual count was inaccurate.
• In eight (8) audits, observers had concerns that the results on the reporting forms were inaccurate.

In more than half of the observations, observers noted that audit procedures were not followed:

• When using the hash mark counting method, twenty-nine (29) observed that a second official did not verify that votes were read accurately by the first official, nor that hash marks were recorded accurately.
• When using the sort and stack counting method, at least thirteen (13) observers reported instances where a second official did not verify that ballots were piled accurately.
• When using the sort and stack counting method, at least seven (7) observers reported instances where a second official did not verify that piles were counted accurately.

Confusion in Definitions

There continues to be confusion in the definitions of “ballots with questionable votes” (marks that the machine may have misread) and those that should be considered “undisputed ballots”. In this round of audits, there was also confusion over how to classify ballots with two votes for cross-endorsed candidates and how to count those votes.

On the official reporting form, some towns fail to classify any ballots as having any questionable votes.   Other towns mistakenly find the difference between the machine count and the hand count in a particular race and record it as the number of ballots with questionable votes. These conclusions are supported by the detail in the ‘Audit Statistics’ section of this report.

Reportedly, several towns classified ballots with two votes for cross-endorsed candidates as ballots with questionable votes and, further, failed to provide instructions on how to count such votes. Some counting teams counted two votes for the same candidate as two votes; other teams considered them overvotes and did not count them at all.  Many officials commented that two votes for cross-endorsed candidates were the cause of discrepancies between hand counts and tabulator totals in the audit reports.

Lack of “Blind Counting” Procedures

Blind counting is a method of counting without pre-conceived knowledge of the expected outcome.  When counting teams know the tabulator totals or know the differences between their counts and the machine totals, there is a natural human tendency to seek to make the hand count match the tape by taking shortcuts or finding a ready explanation.  This lowers the credibility and confidence in the audit results.

• In thirteen (13) observations, counters knew the tabulator totals while they were counting.

• In twenty-one (21) observations, counters knew the difference between their initial hand counts and the tabulator totals while they were recounting.