[EL] Ballot Exhaustion in Maine RCV Primary

Fri Jun 22 09:59:14 PDT 2018

Rob,

My focus on “exhaustion” is motivated by concern about the democratic deficiency of plurality winners that my OSU colleague Ned Foley has articulated<https://papers.ssrn.com/sol3/papers.cfm?abstract_id=2795124> quite cogently.* At a talk Ned has given a while ago, he laid out a nightmare 2020 election scenario in which you have Warren (D), Trump (R), and Kasich (I), and either Warren or Trump win a state’s electoral college votes with a mere plurality of the vote.

Ned argues that RCV/IRV is a solution, but I’m not convinced. If enough voters who choose Kasich have their ballot “exhausted” after the first round, you still have a plurality winner (as happened in San Francisco this time, and all of the other races we describe in our paper).

I also worry that RCV/IRV could actually make this scenario more likely by weakening the incentive (and increasing the difficulty) for strategic voting and encouraging more candidates to run. For example, we know that Bloomberg decided not to run in 2016 because he was worried he’d just split the Clinton vote. With RCV/IRV in place, this would have likely been less of a concern, but could still have produced the same outcome if enough first-choice Bloomberg voters didn’t rank a second choice and had their ballot exhausted. (Obviously, much depends on whether these hypothetical voters actually ended up voting in 2016 and who they supported.)

A couple of quick responses to your other points:

·         My guess is that turnout in Democratic primaries was higher across the board this year, so we need to be careful about how much of the increase in Maine we attribute to RCV.

·         You’re right that more people voted for mayor in San Francisco that voted for governor. On the other hand, once you exclude the “exhausted” votes, fewer ballots ended up making it to the final round that determined the winner of the mayoral election (229,408 continuing ballots in Round 9 vs. 244,137 in the gubernatorial primary).

·         You’re almost certainly right that “some voters are coming out primarily to back candidates who doesn't reach the final instant runoff,” but it depends on what we assume about their behavior under the counterfactual of no RCV. In the alternative 2016  scenario above, some of the hypothetical Bloomberg voters who hypothetically don’t rank a second choice may have actually stayed home in 2016 because he didn’t run. But some may have also begrudgingly voted for Trump or Clinton. Had Bloomberg run, his campaign rhetoric and advertisement might have convinced these voters that Clinton and Trump were both equally evil and they should rank Bloomberg and no one else. I don’t think we can (or should) assume that all those who had their ballots exhausted would not have voted in the absence of RCV.

·         As you note, the previous concern is particularly worrying if minority voters are most likely to have their ballots exhausted because they rank only their co-racial/co-ethnic candidates, and these candidates don’t make it to the final round. (I know Craig is doing some work on this question, so I look forward to seeing the results.) Again, how problematic this is depends on the counterfactual — what would Asian voters who supported Jane Kim and had their ballot exhausted have done if she had not run (anticipating strategic voting) or had not made it to a runoff in a two-stage election? Again, I don’t think we know for sure.

·         I agree that RCV could hypothetically change candidate strategies in a positive way (e.g., by reducing negative campaigning). Perhaps London Breed was ranked as the 2nd or 3rd choice of so many voters because she worked hard to persuade them, as you suggest. Alternatively, she could’ve made it to the top three simply because she had high name recognition because she was the president of the Board of Supervisors and acting mayor, and once voters had ranked the top one or two candidates they actually cared about, they filled in the rest based purely on name ID. (We know that name ID, and even ballot order, has a much bigger effect on nonpartisan local elections.)

*Back in the day, Fair Vote was also very concerned about plurality winners. Your website used to say<https://web.archive.org/web/20101030061205/http:/www.fairvote.org:80/Comparing-IRV-With-Plurality-Voting>: “Plurality voting, whereby the candidate with the greatest number of votes wins, is the norm in most American elections. As a result, time and again we witness some of our most powerful elected offices filled with candidates who were not supported by the majority of voters. … In fact, the prospect becomes very real that the winner of an election may even have been disliked by a majority of the population. This is the first and most basic problem with the plurality system.” Obviously, this problem is not solved by RCV with high levels of exhaustion.

Vlad

[The Ohio State University]
Vladimir Kogan, Associate Professor
Department of Political Science
2004 Derby Hall | 154 N. Oval Mall, Columbus, OH 43210-1373
510/415-4074 Mobile
614/292-9498 Office
614/292-1146 Fax
http://u.osu.edu/kogan.18/
kogan.18 at osu.edu<mailto:kogan.18 at osu.edu>

From: Rob Richie [mailto:rr at fairvote.org]
Sent: Friday, June 22, 2018 11:18 AM
To: Kogan, Vladimir
Cc: Election Law Listserv
Subject: Re: [EL] Ballot Exhaustion in Maine RCV Primary

Thanks for inviting me to respond, Vlad. Just enjoyed being in a forum with your "ballot exhaustion" coauthor Craig Burnett, by the way, and I welcome folks on the list joining me at 9 am Monday morning at Third Way in DC <https://www.thirdway.org/events/ranked-choice-voting-lessons-from-the-states> for a conversation about RCV in the states with Third Way's Lanae Erickson Hatalsky and FairVote New Mexico's Maria Perez, who will talk both about Santa Fe's first use of RCV in March and the unanimous vote of the council in the state's 2nd largest city Las Cruces to use RCV next year.

As you might suspect, I don't see Vlad's numbers as a problem - indeed, quite the opposite. A few reasons why:

* Turnout: More Maine Democrats came out to vote for governor than any previous gubernatorial primary. That follows a string of mayoral elections with ranked choice voting where turnout was a good bit higher<http://www.fairvote.org/the_facts_of_ranked_choice_voting_voters_like_it_high_turnouts_are_trending> than projected: San Francisco (2nd most mayoral votes ever, this month and a lot more votes cast  for mayor with RCV than top of the ballot for governor and US Senator without RCV), Santa Fe, St. Paul and Minneapolis. We can safely put the "RCV will turn voters off" claim to rest. But that higher turnout almost certainly does mean some voters are coming out primarily to back candidates who doesn't reach the final instant runoff. (2 additional notable facts from San Francisco: voters were six times more likely to invalidate their vote for governor with an overvote than invalidate their vote for mayor with an overvote, while 95% of backers of the Asian American candidate Jane Kim who did particular well with API voters ranked at least one other candidate 2nd.)

* Contrast with runoffs and plurality:  Of the Maine Democrats who had their ballot count in the first round (final numbers aren't in, but will be in 126,000 range), 93% had their ballot count for one of the two finalists after five candidates were defeated in the RCV tally. That "exhaustion" compares quite well to the average of federal primary runoffs drawing 62% of first round voters <http://www.fairvote.org/federal_primary_runoff_elections_2016> back to the polls. We did an overall analysis comparing "exhausted votes" with RCV vs. "exhausted voters" in runoffs <http://www.fairvote.org/ranked_choice_voting_outperforms_runoffs_in_upholding_majority_rule> last winter that will look at better for RCV when we update it this summer. And of course the fact that of four out of every five backers of defeated candidates were able to express a preference that counted in the final round is a lot better than a single-choice plurality system that would have allowed the winner to take the nomination with 33% of the first round vote.

* Deeper significance of RCV beyond the "binary choice" analysis: It's misleading with RCV to determine an RCV winner's mandate only by their final round "instant runoff" percentage. What that percentage misses is that the winner has often worked hard to be ranked by backers of the candidate who finishes second, In San Francisco, for example, voters this year were still limited to ranking three candidate in the 8-candidate election for mayor. (That limit will be lifted next year, btw.) The winner London Breed was ranked 2nd or 3rd by nearly half of the backers of her finalist opponent Mark Leno and was ranked in the top three by 63% of all voters. That's not just of theoretical value. Leno easily could have finished third behind Jane KIm and missed the final instant runoff, so Breed wanted to connect with his voters. Indeed, she was ranked 2nd or third by more than a fifth of the backers of all other 7 candidates, including the most liberal and conservative candidates.

* Getting the runoff candidates right: Furthermore, RCV is more reliable than runoffs for ensuring that the final two candidates are representative of voter opinions The Top 2 primary in California this year ultimately didn't shut out Democrats from winnable congressional districts this November, but Democratic forces had to literally spend millions of dollars to ensure that didn't happen. Without that kind of intervention, that "shut out" in fact did happen to Republicans in a special state senate election in Georgia last year in a GOP-leaning district and to Democrats in a statewide race for state treasurer in 2016. I know some mathematicians can have their concerns about RCV in this area too, but RCV certainly is more reliable than runoffs for ending up with the two strongest candidates in the final round.

Thanks again for the invite, Vlad, and happy to take this offline to not bore the list.
Rob Richie, FairVote

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://webshare.law.ucla.edu/Listservs/law-election/attachments/20180622/bac8f32d/attachment.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: image001.png
Type: image/png
Size: 3605 bytes
Desc: image001.png
URL: <http://webshare.law.ucla.edu/Listservs/law-election/attachments/20180622/bac8f32d/attachment.png>