Warnings from the past

Many people, on studying the matter, have come to the same conclusion.

Gove (1894):

Another serious objection is that a heavy additional burden is imposed on the voter by requiring him to indicate several names in the order of his preference instead of indicating but one choice, as in the other systems. This burden the voter would be very loth to assume, since it is with difficulty that voters are induced to discharge the duties already imposed upon them.

Lien (1925):

While the Hare system is meticulously more accurate as a plan of propor­tional representation than the Free List system so widely used on the continent of Europe, it is certain that for the election of members of our state legislatures or of our Congress, the Free List plan (with its close approximation to complete proportional representation) would be much more in accord with our voting habits and consequently much more acceptable and intelligible to the voters.

Weeks (1937):

All states except Alabama and Oklahoma did not require the voter to register more than a first choice for any office. It seems to have been quite common in all the states indicated above that a great many voters failed to avail themselves of the privilege of registering second or more choices, which resulted in the practical restoration of the plurality system in many primary races. This failure was due to several causes: ignorance of the voter; his desire not to have his vote counted for any but his first choice; or his refusal to accept what was thought to be a complicated system, which, it was felt, could be easily corrupted or readily subject to mistakes in the count, or which seemed to provide for an unfair method of evaluating choices.

Gosnell (1939):

Under the list plan, the voters would have to choose only one candidate, the best man from the group which they favored most. The results of the election would be known shortly after the single-preference votes had been counted. It is urged that the proposed system would probably lessen constitutional difficulties, simplify the task of the voter, save the expense and delay of the Hare count, and give substantially the same results.

Further reading here. Sources here.

New York City is the largest rollout of ranked ballots in human history

More on it soon. For now, the data.

Jurisdictions not included are: other Progressive Era adoptions (hard to imagine those populations being bigger), U.S. adoptions since late April 2021 (ditto), adoptions for military/overseas voters (ditto and see link), and state/provincial/municipal adoptions in other countries (ditto).

For various reasons, population does not reflect the pool of eligible voters.

Ballot exhaustion, STV edition

In a ranked-choice election, ballot exhaustion refers to the share of ballots that do not continue to the final round of counting. A ballot exhausts because the voter has not ranked a front-runner (should one exist). Analyses of single-seat elections sometimes show that, due to ballot exhaustion, the winner was not supported by a majority of voters. According to FairVote, this has occurred in 27.6 percent of modern “instant-runoff” races (which had three or more candidates).

What about ballot exhaustion in multi-seat elections (i.e., under STV)? This is a concern for the seat-maximizing party (taken here to include multi-party coalitions and party-like entities). If ballots do not flow among co-partisans — e.g., due to bullet voting for only the most popular candidate(s) — ballot exhaustion can be blamed for having changed a seat distribution. (I will not cover vote leakage here, which is a separate but related issue.)

The plot below gives rates of ballot exhaustion for three historic cases. Two of them, Cincinnati and Worcester (MA), were the subject of my 2018 article in Electoral Studies. Robert Winters provides the data for Cambridge (MA), which still uses STV. (See this blog post, with thanks to Mirya Holman, for a sense of more recent data.)

These figures are based on total ballots cast, not the valid-ballot totals, but a quick look at the data suggests they wouldn’t change too much (although they would be somewhat higher).

Also, these figures are based on exhausted ballots from the penultimate round of counting. Why? Say we have two parties squaring off. Competition is for the final seat, which either of these can win. Computing exhaustion from the final-round count would be akin to including votes for the ‘main loser’ in an IRV contest. Neither of the papers above does this.

I make no claims about what is a “normal” rate of ballot exhaustion under STV.

That said, it is possible to speculate about the variation. Worcester likely had the highest rates due to its weak Democratic Party (i.e., one that could not, for whatever reason, deter ‘excess’ candidate entry). Another factor in at least two cites was that large numbers of independents (and the very rare third party) often (but not always) ran hopeless campaigns, and their voters do not seem to have sent votes back to the “majors” (at least in large number). In Worcester’s first election, for example, there were 152 candidates to the nine-seat council (and 126 counting rounds, due to batch elimination). In Cincinnati, by contrast, the party organizations were both strong and adept at electioneering. Also, Cincinnati was a ‘movement model’ for STV adoptions elsewhere, so people took great pains to ‘get it right.’

I say more about all this in a very big project wrapping up. If ballot exhaustion matters, there are ways to deal with it. Here’s one. Here’s another — although it is not the main reason for that proposal.