January 30, 2019

From: Rick Bourdon, Co-Chair, Open Democracy Action
Re: HB 728 (Ranked-Choice Voting)
Before: NH House Election Law Committee

Chairperson Cote and Members of the Committee,

I would like to speak to a serious structural bias in elections for members of New Hampshire's House of Representatives. I am referring specifically to the bias created by the majority-party-take-all nature of New Hampshire's multi-member districts, a bias that could be eliminated through the use of ranked-choice voting. Let's call it multi-member district bias (M-MD bias).

How M-MD bias works

In multi-member districts, particularly in large ones (4 to 11 seats), voters get to choose candidates for each seat. That sounds fair enough. In practice, however, voters tend to cast party-line votes, with the result that the party favored by a majority of voters in that district often wins all the seats instead of a proportionate share of them. This was particularly true in 2016, less so in 2018 as Republican majorities shrank. In 2016, while there were multi-member districts with Democratic voting majorities and all-Democrat house delegations, they were relatively few and with mostly small numbers of seats per district. The big districts were almost exclusively located in the southern part of the state where Republican majorities reigned. Hence a Republican advantage.

My analysis of 2016 data

In the spring of 2018, I analyzed 2016 election results. I calculated the efficiency gap (EG) for New Hampshire House races. The efficiency gap is a measure of how efficient one party is relative to another in its ability to convert votes into seats won. My goal was to determine whether NH districts were gerrymandered, i.e., whether the majority party in 2010 had engaged in partisan redistricting. The EG for 2016 House races was 9.8%, almost 2% above the threshold value used to identify gerrymandering. Upon further examination, however, it became clear that I hadn't  measured gerrymandering at all. Instead, I had measured M-MD bias.

The following table compares 2016 EGs and Republican seat advantages for small districts (1 to 3 members) and larger ones (4 to 11 members).

*Positive values indicate an advantage for Republicans.

**Calculated within each district by comparing the number of seats actually won to the number of seats that would have been won had there been proportional representation.

The EG for larger multi-seat districts was off the charts in favor of Republicans. In the small districts, there was no Republican advantage at all. The problem lay in the majority-party-take-all element inherent in large multi-member districts.

A quick scan of 2018 election results suggests that M-MD bias was much less of a factor in the more recent election than in 2016. A number of once solidly Republican M-MDs produced split delegations. There is no reason to believe, however, that those districts will remain split. M-MD bias might just favor a different party next time.

Ranked choice voting in multi-member districts

Use of ranked-choice voting would virtually eliminate M-MD bias. To illustrate that, I used 2016 data to simulate ranked-choice voting in several MM-Ds: Rockingham 5, Hillsborough 21, and Cheshire 1.  Results are shown in the table below.

With ranked choice voting, what had once been majority-party-take-all districts were now producing winners from both parties, not in perfect proportion to actual vote shares, but relatively so.

Efforts to end gerrymandering are necessary, but even if gerrymandering were eliminated, without changes in the way M-MDs are handled, the major source of bias in NH House races will remain. One solution is to use ranked-choice voting. Another is to break these districts up into many single-member districts. Read on, however, to see how neither alternative is, by itself, an optimal solution.

Beyond ranked choice: the case for more multi-member districts (and a new voting system)

There are two ways to think about electoral representation. The first is the most common. It examines representation at a macro-level, the level of a state or nation. We generally consider this kind of representation to be fair when the share of seats won by a political party parallels, at least roughly, that party's overall vote share. Fair representation, conceived in this way, is sabotaged by gerrymandering and fuels efforts to create nonpartisan redistricting commissions.

There is another, less common way to think of representation, however. It is experienced at a micro-level, the level of an individual district. Here, voters feel represented fairly if their votes result in a fair share of their own district's seats. Clearly, multi-member, majority-party-take-all districts fail in this respect. If you are a member of the minority party in one of these, you know that, unless political winds change dramatically, the likelihood of being represented by someone who shares your political philosophy is virtually zero. You have no one to think of as your representative. Perhaps surprisingly, micro-level representation is just as bad in single-member districts. They, too, are majority-party-take-all.

Fixing gerrymandering improves macro-level representation. It does nothing, however, to improve micro-level representation. So what would do both?

The most promising answer is what FairVote.org calls Fair Representation. It combines multi-member districts with ranked-choice voting. FairVote recommends districts with three to five seats each (any more than that makes ranking all candidates onerous). FairVote promotes Fair Representation primarily for Congressional races, but the advantages of the system would be the same for state-level contests. In the NH House, statewide results would approach actual proportional representation (macro-level fairness), and, within individual districts, each voter would be more likely to have at least one representative who shares that voter's political philosophy (micro-level fairness).

Fair Representation and gerrymandering

Unless I'm missing some subtle twist, it seems clear that gerrymandering multi-member districts that use ranked-choice voting would be extremely difficult. Successful gerrymandering requires both "packing"⎯cramming large numbers of a party's votes into a few districts⎯and "cracking"⎯creating many districts in which that party gets a minority, sometimes a slim minority, of votes. With Fair Representation, packing is easy enough, but cracking is virtually impossible. The critical majority-party-take-all element essential to cracking is gone.

So maybe the emphasis on nonpartisan redistricting commissions as the solution to gerrymandering is overrated. They're still a good idea. We do need to create districts that make sense and follow the rules. But changing our voting system is an even better idea.

Fair Representation, in my opinion, is the Holy Grail of voting systems. We're not there yet. But we can take a first step—institute ranked-choice voting.

I strongly encourage the Committee to vote Ought to Pass on HB 728.

Thank you for considering my testimony.