The mission of the Oregon Voter Rights Coalition is to restore public ownership and oversight of elections, and to ensure the fundamental right of every American citizen to vote and to have each vote counted as intended in a secure, transparent, impartial, and independently verified election process.
Oregon Voter Rights Coalitions Proposed Methodology for Election Verification
Submitted by Ginny on Mon, 04/02/2007 - 8:27am
Introduction
Verification System Standards
Sample of Ballots Verification
Statewide Races VerificationSample Size Table for (+/- ) 1% Error-Margin*
Election audit or verification methods are currently used or are being proposed in other states as well as at the federal level, and auditing a percentage of the precincts is an audit method commonly promoted. While that method may be good for checking the accuracy of a certain percentage of tally machines, it is not statistically valid to check the accuracy of the election results. Members of the Oregon Voter Rights Coalition with expertise in statistical methods have developed the following standards and methodology as a proposed quality control process to verify machine-tallied election results through a hand counted random sample of paper ballots. The goal of this verification process is to ensure that the certified election results accurately reflect the will of the voters. This methodology has also been vetted and approved by outside experts. Under this proposed methodology, the procedure would be used for statewide elections including President, U.S. Senator, Governor, Secretary of State, and statewide ballot measures. This process would be performed in each county by the County Clerk's office and would require minimal time, effort, and cost.
Note: A special thanks goes out to Oregon VRC members Jerry Adams, Ph.D., Jan White, and Nancy Matela for their expertise and diligent work on this project over the past 2 years, and to Betsy Salter for her dedicated work in support of this effort.
A system for independently verifying election results must meet these standards:
- The verification process must determine whether the tallies produced by electronic election equipment are accurate.
- Counties or precincts must provide easily-read paper ballots as the ballots of record.
- The verification process must involve local communities sufficiently to ensure transparency.
- Vote totals cannot be certified unless the hand-counted sample tally passes a statistical test based on the 99% Level of Significance (see Appendix A for details). If the verification fails the statistical test, a full hand count of all ballots is completed and the hand-tallied results are certified instead.
Statewide Races Verification
The sample size of hand-counted paper ballots needed for each state can be found in the table below, based on ballots cast in the state. The sample size will give no more than (+/-) 1% Margin of Error if drawn randomly from each county and precinct.
| Number of Votes Counted in the State | Sample Size for the State |
| 100,000 to 500,000 | 41,000 |
| 500,001 to 650,000 | 42,000 |
| 650,001 to 1,050,000 | 43,000 |
| 1,050,001 to 2,000,000 | 44,000 |
| 2,000,001 to 50,000,000 | 45,000 |
To determine the sample size for each county:
- Divide 45,000 (or the appropriate sample size) by the total ballots cast in the state. This provides the percentage of a county's ballots to be counted.
- Multiply the total ballots cast in the county by this percentage. (This provides the number of ballots for the sampling procedure.)
For example, if 3 million ballots are cast in a state, then the percentage of ballots to collect for each county in that state is:
45,000 / 3,000,000 or 0.015 (1.5%).
Every county in that state would randomly sample 1.5% of its ballots. For example: A precinct with 5,000 ballots cast would draw 1.5% of 5,000, or 75 ballots for recounting.
Collecting the Sample
The process of collecting the random sample of ballots must be transparent (publicly observed and recorded), secure, nonpartisan, swift, and inexpensive. One way to do this is to pull every "nth" ballot in each polling place as the votes come in. For example, if the sample percentage for the state is 1%, every 100th ballot is used in the sample. A nonpartisan group publicly pulls every 100th ballot, records the results for the chosen races, and puts each ballot back in the stack in the same place where it came from. The results from the random sample for each polling place are publicly posted, along side the machine tallied results.
For mail-in, absentee, and provisional ballots, a similar process of continuously collecting and posting a random sample needs to be completed, typically at the county level. Because the hand counting of ballots is performed during the election, the cost of such a process can be very small.
The polling place and the county level totals are then added together at the state level, and the proportion of votes is calculated for the top two candidates in each contest to be verified. The hand-counted sample proportion is compared to the machine-counted proportion and a statistical test is used to determine whether the machine-counted results are accurate.
VIDEO: “Protecting Your Vote”
