Congratulations to the county and city clerks for November’s well-run election. In Colorado’s first and, so far, only county-coordinated Instant-Runoff Voting (IRV) election, the city and county clerks emphasized voter education and the voter experience to minimize invalid ballots. We in the League of Women Voters of Boulder County (LWVBC) were proud to partner with the clerks and their staff on voter education.
One surprise did arise, however. Almost all of the tabulation examples prior to the election consisted of only two scenarios:
• a candidate gets more than half the votes in the first round and is declared the winner;
• no candidate gets more than half the votes so the lowest candidate is eliminated and the votes on those ballots transfer to the next highest choice on each ballot. This process is repeated until one candidate has more votes than the other non-eliminated candidates.
Many voters were surprised when the actual tabulation was reported in neither of these ways. Instead of only eliminating the lowest candidate after Round 1, the two lowest candidates were eliminated simultaneously — called “batch elimination.”
The Boulder County Clerk followed CO Secretary of State Rule 26.5.4 which mandates batch elimination if the candidates with the lowest vote totals have no possible path to win. For instance, even if every one of Paul Tweedlie’s votes had transferred to Nicole Speer, Speer’s vote total would still be the lowest of all the non-eliminated candidates.
Eliminating Speer and Tweedlie simultaneously prevents the public from seeing how votes from Tweedlie were transferred, as distinct from votes for Speer, and thus obscures the opinions of the electorate. A major reason for changing voting methods is to help the public be more expressive. The LWVBC Voting Methods Team is advocating for Rule 26.5.4 to be changed to eliminate one candidate per round — called sequential elimination.
LWVBC member Neal McBurnett took the publicly available ballot data and ran the tabulation using sequential elimination. We present the official county’s batch elimination results and our preferred three-round sequential elimination results. Note that the numbers in Round 1 are the same for both versions.
You can see exactly how the votes for Tweedlie were transferred to the next highest candidate, if any, on each Tweedlie ballot. You can see that slightly more than half of Tweedlie’s votes transferred to Bob Yates. Similarly, you can see exactly how the new total of 6,487 votes for Speer were redistributed to the remaining candidates or to the exhausted pile if no candidate was ranked below Speer.
Another difference between the LWVBC table and the official results table is in the percentage calculations in Round 2 and Round 3. The total votes cast in this entire contest is 32,893 and doesn’t change. By showing only the percentage of continuing votes in the county’s report, the voters who only supported an eliminated candidate are not reflected in the final percentages. We believe all voters should be included.
For a yet more complete picture of the electorate, the number of votes each candidate received should be divided by the number of ballots. In Boulder, 1,315 voters did not fill in the mayoral contest. Some U.S. jurisdictions allow voters to mark “None of the Above.” In France, where voting is not compulsory, turning in a blank ballot is a way to record a protest vote. It sends a different message than not turning in a ballot at all. In the U.S., leaving a contest blank may be a protest vote, but is more likely an indicator of low voter engagement. We were pleased to see the Daily Camera report the council candidates’ support as a percentage of ballots in their Dec. 10 analysis, “How Boulder voted on this year’s ballot issues, city council and mayor candidates.”
You can see that when we consider all the City of Boulder ballots submitted to the clerk, Aaron Brockett’s win was just shy of majority support. “Majority” is a tricky term. IRV only promises a “majority of the continuing votes.” LWVBC and LWVCO have often pointed out that no voting method can guarantee a majority when there are three or more candidates, and we’ll continue educating everyone on the nuances of measuring a majority.
Congratulations again to the clerks for the successful administration of Boulder’s inaugural IRV contest.
Please reach out to us at vmteam@lwvbc.org to tell us what you think of our proposed changes to the IRV election report or to find out more about our Voting Methods Team.
Celeste Landry, Neal McBurnett, Mark Parsons, Jeanne Clelland and Martine Elianor for the League of Women Voters of Boulder County.