Wisconsin audits


Principles of Audits What Problems Have Occurred? Public comments on the Draft Wisconsin Audit Procedure Organizations Backing Audits Other States Audits Why Audit? Audits Home	Audits of Vote Counts Principles of the Methods On August 30, 2006, The Wisconsin State Elections Board will be considering the procedures for audits of the vote count for the November election. The audit shall be a counting of the votes, where the identification of each vote is determined by a visual inspection of the voter-verifiable paper record by a human being. This audit is an extremely important step forward for voter confidence, transparency, and quality control. But the procedure must be worthy of the name “audit.” If it does not meet basic principles of an audit, it will be subject to criticism and may actually increase cynicism about the accuracy of our vote counts. In order to be credible, the procedure for the audits must meet these criteria: We must be sure that the ballots being audited are the ones the voters submitted! Ballot bags should have either numbered seals, or a seal that is signed by the pollworkers. Timing Random selections shall occur after the initial unofficial results are announced, and as near as practical to the time of the audit. The audits shall take place as soon after polls close as possible without imposing unnecessary burdens on election officials. There shall be a state-wide schedule that is reasonably uniform. The audits are to be completed before the canvassing of the election for that county. (Perhaps the audit could be started coincidentally with the County Board of Canvassers at 9:00 a.m. on the Thursday following the election.) Random Selection Random selections shall be made by a method that is indisputably random, and that the average citizen can easily understand and observe to be random. The method shall be subject to approval by the appropriate governing body. The Brennan Center suggests that one way is to use a state lottery machine. At least one vote count for each election contest (state offices, congressional offices, ballot propositions, bond issues, etc.) shall be audited in each county. Each political party on the ballot shall have the right to pick some wards for audits. This shall be in addition to the randomly selected audits. Any selections that are automatically triggered must be in addition to the randomly selected audits. Discrepancies and Additional Audits In all instances where the hand count is performed, the hand count shall be the official count of the vote. Significant discrepancies shall be investigated by qualified review teams with little self-interest. This excludes vendors or the original election administrators. Significant discrepancies shall result in prompt audits of additional wards. The Brennan Center notes that Illinois law provides that where electronic and paper records in the Automatic Routine Audit do not match, the county notifies “the State Board of Elections, the State’s Attorney and other appropriate law enforcement agencies, the county leader of each political party and qualified civic organizations." Contests in which recounts are requested shall have 10% of the wards counted by hand as part of the recount. The wards shall be randomly selected. Professional Oversight When developing the policy, statistical analysis shall be done to determine the probability of discovering a problem, based upon the per cent sampled, the size of the population to be sampled, and an assumed problem rate. The procedures shall be reviewed by professional auditors and statisticians before adoption, and on an ongoing basis. This shall include evaluation of the method of random selection. Establishment of a Vote Count Audit Committee should be considered. Administration All functions, including random selections, shall be done at properly noticed public meetings. The local supervisor for the audit shall not be the person who administered that election. At least two people shall conduct the hand count, independent of the person who administered the election. A spreadsheet of ward-by-ward details shall be published promptly on the website of the State Elections Board. Minutes of each audit shall be kept and made available to the public. The State Elections Board shall publish a report describing and evaluating the audit and assessing its effectiveness, and, if applicable, identifying the discrepancies between the manual counts and electronic counts and any explanations for the discrepancies and recommending ways to reduce vote count errors in future elections. Items that cannot be accomplished in the initial 2006 procedure shall be included in the initial written proposal as items to be decided upon in 2007. These principles were derived using sources such as the Brennan Center report on the Machinery of Democracy, the Common Cause Report on Electronic Voting Machines, National Election Data Archive, and audit proposals for the states of Utah, Arizona, Connecticut, and Pennsylvania. Web page by Paul Malischke malischke@yahoo.com last updated June 17, 2007

Audits of Vote Counts

Principles of the Methods

On August 30, 2006, The Wisconsin State Elections Board will be considering the procedures for audits of the vote count for the November election. The audit shall be a counting of the votes, where the identification of each vote is determined by a visual inspection of the voter-verifiable paper record by a human being.

This audit is an extremely important step forward for voter confidence, transparency, and quality control. But the procedure must be worthy of the name “audit.” If it does not meet basic principles of an audit, it will be subject to criticism and may actually increase cynicism about the accuracy of our vote counts. In order to be credible, the procedure for the audits must meet these criteria:

We must be sure that the ballots being audited are the ones the voters submitted! Ballot bags should have either numbered seals, or a seal that is signed by the pollworkers.

Timing
Random selections shall occur after the initial unofficial results are announced, and as near as practical to the time of the audit.

The audits shall take place as soon after polls close as possible without imposing unnecessary burdens on election officials. There shall be a state-wide schedule that is reasonably uniform.

The audits are to be completed before the canvassing of the election for that county. (Perhaps the audit could be started coincidentally with the County Board of Canvassers at 9:00 a.m. on the Thursday following the election.)

Random Selection
Random selections shall be made by a method that is indisputably random, and that the average citizen can easily understand and observe to be random. The method shall be subject to approval by the appropriate governing body. The Brennan Center suggests that one way is to use a state lottery machine.

At least one vote count for each election contest (state offices, congressional offices, ballot propositions, bond issues, etc.) shall be audited in each county.

Each political party on the ballot shall have the right to pick some wards for audits. This shall be in addition to the randomly selected audits.

Any selections that are automatically triggered must be in addition to the randomly selected audits.

Discrepancies and Additional Audits
In all instances where the hand count is performed, the hand count shall be the official count of the vote.

Significant discrepancies shall be investigated by qualified review teams with little self-interest. This excludes vendors or the original election administrators.

Significant discrepancies shall result in prompt audits of additional wards.

The Brennan Center notes that Illinois law provides that where electronic and paper records in the Automatic Routine Audit do not match, the county notifies “the State Board of Elections, the State’s Attorney and other appropriate law enforcement agencies, the county leader of each political party and qualified civic organizations."

Contests in which recounts are requested shall have 10% of the wards counted by hand as part of the recount. The wards shall be randomly selected.

Professional Oversight

When developing the policy, statistical analysis shall be done to determine the probability of discovering a problem, based upon the per cent sampled, the size of the population to be sampled, and an assumed problem rate.

The procedures shall be reviewed by professional auditors and statisticians before adoption, and on an ongoing basis. This shall include evaluation of the method of random selection.

Establishment of a Vote Count Audit Committee should be considered.

Administration
All functions, including random selections, shall be done at properly noticed public meetings.

The local supervisor for the audit shall not be the person who administered that election. At least two people shall conduct the hand count, independent of the person who administered the election.

A spreadsheet of ward-by-ward details shall be published promptly on the website of the State Elections Board. Minutes of each audit shall be kept and made available to the public.

The State Elections Board shall publish a report describing and evaluating the audit and assessing its effectiveness, and, if applicable, identifying the discrepancies between the manual counts and electronic counts and any explanations for the discrepancies and recommending ways to reduce vote count errors in future elections.

Items that cannot be accomplished in the initial 2006 procedure shall be included in the initial written proposal as items to be decided upon in 2007.

These principles were derived using sources such as the Brennan Center report on the Machinery of Democracy, the Common Cause Report on Electronic Voting Machines, National Election Data Archive, and audit proposals for the states of Utah, Arizona, Connecticut, and Pennsylvania.

Web page by Paul Malischke malischke@yahoo.com last updated June 17, 2007