Spokane County Election Integrity and Procedures
And why we should re-elect Vicky Dalton as Spokane County Auditor
Trump and his minions, including the local Spokane GOP, are still pushing his Big Lie of election fraud, an attempt to undermine confidence in electoral systems. In this setting it is particularly important that we as citizens understand the checks and balances, the meticulousness, with which elections in the State of Washington are conducted.
Election procedures are specified in Washington State Law and overseen by the Secretary of State in conjunction with the thirty-nine County Auditors. Our County Auditor, Vicky Dalton, has served us 23 years. With that longevity and attention to detail as the Auditor of the fourth most populous county in Washington State, she has become a leader among election officials in the State.
An earlier post, Ballot Processing, offers details, through the eyes of a Spokane County election observer, of the process of signature verification and separation of ballots from their envelopes. That separation makes the individual ballots anonymous. Ballots must still be identifiable by precinct or (rarely) by sub-precinct (they are so identified by a barcode) because the actual ballot layout (i.e. which contests are offered in that precinct) may vary depending on overlapping voting districts. Ballots are batched into precinct groups in preparation for feeding them into the tabulator (the ballot scanning, reading, and vote tabulating machine).
The tabulator is programmed to read the precinct number and then tally the votes cast in each race in each precinct. Before each election checks are run on the tabulator’s programmed accuracy prior to feeding it official ballots. The tabulator is not connected to the internet and lives in a separate locked room. In this process every ballot is scanned and stamped with a number (for later retrieval). The paper ballots are carefully preserved for possible later reference in the event of a recount or an audit. The ballots can be retrieved and matched with the scanned image if necessary.
By State law a hand recount is triggered under certain conditions you can read about here at the Secretary of State’s website. Between 2001 and 2021 there were 17 hand recounts (see the tally). The agreement between the machine count and the hand recount is striking—and serves as a check on the integrity of the machine counting system. (The Gregoire/Rossi race for governor in 2004 is a spectacular example. Within Spokane County nearly two hundred thousand votes were cast in this race. The result of the mandatory recount narrowed the margin between Gregoire and Rossi by only 20 votes, a change of one thousandth of one percent. Note: This accuracy was achieved when there were still multiple polling sites, many poll workers physically separated from the central office, and separate scanners. The whole state went to all-mail-in voting in 2011. In the current system the ballots are processed centrally.)
The recount history offers great assurance of the accuracy of the machine counting procedure, but in 2018, even before Trump was blathering his Big Lie, the Washington State legislature passed a law (2018 c 218 § 1) that added an additional check of the counting procedures: a Risk Limiting Audit (RLA). (For an explanation of the principles of RLAs click here.) An RLA is a random sample of election results based on statistical math that is used as an additional check that everything worked properly. Spokane County under Vicky Dalton had the honor of overseeing the first ever official RLA in the State during this year’s (2022’s) primary election. (Other counties are still doing test runs.) The specific race to be audited in a county is chosen by a procedure done at the office of the Secretary of State. (For a detailed explanation of that process see the quote at the bottom of this post.)
What follows is the account of this Spokane County RLA written by a Democratic Election Observer, Dan Simonson:
Spokane County’s (and Washington’s!) First Risk-Limiting Ballot Audit
On the morning of August 9th, I was asked by the Democratic party to attend a “Risk Limiting Audit” (RLA) at the Spokane County Election Center. As I found out, this was the very first [official] use of a Risk-limiting audit in Washington. This approach to auditing ballots was first developed over a 10-year period in Colorado, and the software program was developed from their extensive research and testing. A representative from the Secretary of State’s office was there to offer assistance and support, but made it clear that the County is responsible for the audit.
First task: generating a random number to seed the RLA program. I had previously been certified as an election observer, a requirement for this event as well. I arrived at 9 am, got my badge, but instead of going directly back to the ballot-processing area, we (3 - 4 Repubs, 2 Dems - myself and Carmela Conroy, a Democratic Party official) were escorted into a conference room and helped the election officers create a 20-digit random number to “seed” the audit program. This was done by rolling twenty 10-sided dice and recording the numbers (the dice were thrown one at a time into a lucite contraption that “rolls” them through a baffle).
Each observer and some of the staff got to toss their dice by dropping it into the clear lucite device, and it rolled out the bottom to be read and added to the number. Then, with that number in hand, the officers escorted us into our usual area. There were some tables set up and two computers were running. There was a screen facing us observers, so we could see what was being done on the computer.
They started up the specialized RLA program on the computer, and to conduct the audit, the first thing they had to do was initialize the program's random number generator by entering the truly random 20 digit number we just created to act as a seed. As you may be aware, computers generate random numbers all the time for various programmatic requirements - but since they are generated by algorithm rather than by a physical process, those numbers are not truly random - they are called “pseudorandom”. Initializing the program with the seed made sure we had absolute rather than “pseudo” randomness.
A single county-wide race was audited
We were auditing one specific race: State Representative position 2 for Leg District 3 - Candidates were Poulson, Spurlock, and Ormsby. And even more specifically, we were focused only on any irregularities in votes between the 2nd and third place candidates. Why? As I was told, since the top-two vote getters are chosen, the differences between one and two aren’t decisive - but between 2 and 3 are.
Using statistical calculations based upon the size of the races and the amount of difference between candidates, the software told them how many ballots they would need to count. And the sample was not only of ballots in that legislative district, but across all races - which meant that some of the chosen ballots didn’t even have the studied race on them! The reason for choosing ballots randomly across the whole county was that we could also check to make sure that ineligible voters (for example, from a different legislative district) were not able to vote in the examined race.
As I recall, they only needed 67 ballots to do the survey, and the computer spit out a list of all of the particulars (Carton#, batch#, serial number in the batch, etc.). The boxes were brought into the room, and under our close supervision, the staff found the ballots, took them out of the boxes (leaving a paper trail on each) and replaced the others as before.
The stack of 67 chosen ballots were then brought over to the computer, and two staff members processed them - one reading the ballot, the other tapping the appropriate button on the screen (see photo). Their choices were: named candidate, write-in, blank vote, or “not on ballot”. In an additional nod to security, the two staff members entering and observing were replaced halfway through the ballots - I thought that was very clever!
Finally, all the ballots were entered, the RLA program analyzed the responses, and lo and behold, we had 100% agreement! Election security and transparency in Spokane County has been affirmed.
I have to wonder if November election challenger, former LD4 State legislator Bob McCaslin, is aware of any of these details. He is, however, (along with Rob Chase, McCaslin’s compatriot representing LD4) schooled by the Mike Lindell funded and produced “cyber symposium” espousing Lindell’s conspiracy theory of nationally widespread election fraud. McCaslin refused an interview with Al Merkel (in which Ms. Dalton displayed her expertise) after he saw the questions to be asked. One might presume McCaslin deemed it poor strategy to display his ignorance of actual election procedures.
Keep to the high ground,
A detailed explanation of how a race is randomly chosen for a Risk Limiting Audit by the Secretary of State’s Office, graciously provided to me via email by Hannah Taylor, Results Audit Specialist with the SOS’s office:
The target contest for the risk-limiting audit was selected by lot draw. The Office of the Secretary of State created a list of eligible contests for Spokane County. For a primary election, these contests needed to have three or more candidates. If a contest had only two candidates, both would proceed to the General Election, so it makes the most sense to look at contests where three or more candidates are on the ballot to make sure that the correct two finishers proceed to the General Election in November. We also include local measures when they involve a large enough percentage of a county’s voters. Based on these criteria we created the following list:
Legislative District 3 State Representative Pos. 2
Legislative District 4 State Representative Pos. 1
Commissioner District 2
Commissioner District 4
Commissioner District 5
Spokane Valley Fire Department Proposition No. 1
Each of these contests was printed on a slip of paper and put into a capsule, then put into a bingo cage and mixed up. Then we pulled out a capsule and read the contest name—that’s it! This is actually the very same process our office uses to determine the ballot order for candidates.
The contest selected could have been a countywide contest. Whether it is or not, we always sample ballots from the entire universe of accepted and tabulated ballots. I often get the question, “Why don’t you just look at the ballots that have the contest on them?” The trouble with that is that we’d be relying on the voting system to tell us which ballots to look at, and if a problem existed we likely wouldn’t find it.
Spokane County has the distinction of conducting the first official risk-limiting audit in the state of Washington. The state law passed in 2018 requires all counties to conduct post-election audits, and RLAs are one option available. The RCW for audit of election results is 29A.60.185.
Please feel free to contact me if you have any additional questions! I love discussing audits.
Results Audit Specialist
Office of the Secretary of State
(360) 725-5795 (office) | (360) 764-3834 (cell)