Nomination by Attrition
Facts in the Spotlight _ _ _
Number of federal primary runoff elections between 1994 and 2006: 104
Number of those elections that saw turnout declines: 101
Average turnout decline in those races: 35% _ _ _
Election Day in America is a day that arrives with some conflicting emotional baggage. On one hand, we can be filled with pride that we live in a country founded on the idea that the people should be able to decide their own fate, one in which every person has a say in the way their government is run. On the other hand, we are disappointed that relatively few of us take advantage of the voting franchise. We also tout majority rule as the basis for democracy, then use voting methods that either allow majorities to go unrepresented or make our turnout problem even worse.
Problems with Prefixes: If it is difficult to stir even half of the country's eligible voters to go to the polls, we find more trouble the more prefixes we add. Primary elections suffer from relatively low turnout. In the eight states that currently require primary winners to reach a certain threshold of support (usually 50%) to demonstrate true support in the party, and we have a primary runoff election, the numbers dwindle even further.
How much further? Between 1994 and 2006, 101 out of 104 runoffs for federal primary elections (for House and Senate) saw turnout declines from the initial primaries. The average drop-off during this period was 35%. 2006 saw the largest turnout drop for primary runoffs at 43%, meaning that in a watershed year like 2006, almost half of all primary voters did not take part in the final selection process to determine their parties' nominees. When a low-turnout primary winds up moving to a lower-turnout primary runoff, how can anyone be sure that the winner is the true consensus choice?
Consensus not achieved: Drops like this can have a major impact on the result. Take, for example, situations in which the second-place finisher in the first primary wins in the runoff. All well and good until one realizes that almost half of the primary voters from the first election opted out. In fact, in 29% of cases, the second-place finisher won in the runoff, even though far fewer primary voters showed up for the runoff.
So why have a threshold, and force a primary runoff at all? Think of an election, primary or general, in which several candidates are gunning for the same office - it's not unusual, as the current presidential season demonstrates. It's likely in such a case to have the first place finisher win with a vote percentage in the 20s. Indeed, FairVote found that in the five congressional elections from 1994 to 2002, there were 214 non-majority primary wins in U.S. House primaries and 29 in U.S. Senate primaries - with some winners waltzing to general election wins in lopsided districts after victories with well under 30%. For example, Mike Doyle of Pennsylvania won his 18th congressional district Democratic primary in 1994 with less than 20% of the vote while Massachusetts Democrat Michael Capuano won the 8th District after taking 23% in his primary in 1998.
By instituting a threshold, there can be more assurance that the winner is truly represents his or her party. Louisiana does away with primaries altogether with a blanket election, but still requires a runoff if no one reaches a certain threshold.
The instant remedy: The best way to be confident that the eventual nominee has the majority of his or her party behind them (and the way to eliminate the need for a costly, low-turnout runoff) is to use instant runoff voting (IRV) in primary elections. IRV allows voters to rank candidates on a ballot in order of preference. When one voter's first choice turns out to be a loser, their second choice is counted instead. In other words, with IRV voters can indicate who they would support if their first choice doesn't make the cut for a traditional runoff. Turnout never drops off because there is no second election from which to drop.
IRV is currently used in cities such as San Francisco, Burlington (VT), and Takoma Park (MD), and the city of Cary (NC) on October 9th became the first city in the American South to implement it - to rave reviews. The city avoided an expensive runoff, and an exit poll showed three-to-one support for IRV over traditional runoffs and 96% understanding of the system. Three North Carolina dailies then called for expanding IRV elections in their state. Hendersonville will soon follow with North Carolina's second IRV election.
Good news gets around: Local and national media are taking notice of Cary's big success. Check out this article in USA Today, which earlier in the year gave a ringing editorial endorsement to IRV. Also take a look at the newspaper endorsements from North Carolina and across the country.
Comment on Nomination by Attrition
Current Discussion
-
Posted by Jack, 10/30/2007 (2 years ago)
-
>And under Approval, strategic voters should ???bury??? >(i.e. not check off) all frontrunners except for >their preferred. Yes, but at least there is no insincere preference reversal as there is with strategic IRV voting. >To break a two-party duopoly, one must do three >things: (1) vote for third parties, (2) get others >to vote for third parties and (3) increase district >magnitude. #3 only applies to multi-winner elections yes? So it would not apply to election a mayor or the president or even federal Senate members. Even in multi-winner elections a suitable method must be used or it is of little help. #1 and #2 together is political suicide under plurality voting where the spoiler effect can result in the 'greater-of-two evils' winning from the point of view of the third party voters. Under IRV when the third party has a small level of support things work ok. But when the third party reaches the level of being a contender it can still act as a spoiler. Back on topic: One of IRV's main selling points is that it is packaged as a way to get rid of a real runoff election. This does appear to have a lot of appeal to people although I think that arguments that an IRV election chooses a 'majority supported' winner is fallacious. An obvious example of this is when many voters do not fully rank the candidates and too many ballots are exhausted.
Posted by AllAboutVoting, 10/29/2007 (2 years ago) -
And under Approval, strategic voters should "bury" (i.e. not check off) all frontrunners except for their preferred. To break a two-party duopoly, one must do three things: (1) vote for third parties, (2) get others to vote for third parties and (3) increase district magnitude.
Posted by Jack, 10/29/2007 (2 years ago) -
>[Jack]Even without a computer, it???s intuitive that >the rational choice under approval is to vote for >one, not for ???one or more.??? I disagree. See below. >[Jack]That is, without information about how the rest >of the electorate plans to vote I sort of agree with this part. In a zero information case an approval voters best option is to only support the candidates that they most like. There is no less-of-two-evils type candidate that they should be motivated to also vote for. As you note this is also true for plurality in a zero information situation. However, most elections that I have participated in are far, far from zero information situations. The media does a good job of polling and reporting on these polls. In that context there a likely to be a number of 'real contenders' identified before the voter actually votes. Then a voter's best strategy under plurality is to vote for whichever 'real contender' they can best stomach. In other words to "vote for the lesser of two evils". And a voter's best strategy under approval is to vote for whichever 'real contender' they can best stomach and also to vote for everyone that they like better. Note that strategic voting under IRV also differs based on information about who the front runners are. In strategic IRV voting voters should bury all front runners except for one.
Posted by AllAboutVoting, 10/29/2007 (2 years ago) -
I'm skeptical of any computer model that claims to know voter preferences and model the voter's strategic calculations. We're talking a lot of data about a lot of issue dimensions. Even without a computer, it's intuitive that the rational choice under approval is to vote for one, not for "one or more." That is, without information about how the rest of the electorate plans to vote, it makes sense to treat an approval ballot like a regular, old plurality ballot.
Posted by Jack, 10/29/2007 (2 years ago) -
The claim that IRV is the "best way" is simply false. IRV is one of the worst of 60+ voting methods that have been analyzed through Bayesian regret calculations by Princeton math Ph.D., Warren D. Smith. A far better and simpler method, that can be done on standard voting machines (unlike IRV), and plausibly will break the two-party system (unlike IRV), and reduces spoiled ballots (opposite effect of IRV, which massively increases them) is Approval Voting. We just change the "vote for one" rule to "vote for one, or more", and presto - we have a better and simpler method than the archaic plurality and Instant Runoff Voting methods. As an added benefit, Approval Voting is much less harmed by strategic voting than IRV is (although FairVote has used some craftily misleading arguments to say otherwise). Quality democracy requires a good voting method that is also simple to use and count. In that sense, Approval Voting is the hands down winner over IRV, so says an overwhelming consensus among credentialed election method experts. http://rangevoting.org/Consensus.html Clay Shentrup San Francisco, CA clay@electopia.org 415.240.1973
Posted by Clay Shentrup, 10/29/2007 (2 years ago)
I really don't like in-line reply. It takes discussion off-topic. That said, we are discussing two points. One is the relative susceptibility of Approval and IRV to strategic voting. The other is creating space for third-party elected officials. With respect to the first point... Hypothesis: the incentive to bullet vote under Approval is more intuitive to more voters than the incentive to insincerely reverse preferences under IRV. A potentially useful comparison is the bloc vote. Though the strategic context is different, the ballot design and expectation of voters is similar, except that overvotes cannot happen under Approval while they can under bloc vote. Under bloc vote, there are myriad cases of undervoting (or bullet voting) where voters withhold votes from favored frontrunners in order to boost other favored candidates. Boston is one example. Others are documented on fairvote.org, if memory serves me. The same logic is at play under Approval. Voters can intuitively reason that voting for a favored frontrunner increases the chances of "bumping" another favored candidate out of the running. I am not sure how intuitive the oft-cited IRV anomaly is to voters. I am inclined to say it is less intuitive than the rationality of bullet voting under Approval. If it were more intuitive, Clay Shentrup probably would not need to paste the same stock response into every IRV-related blog entry he finds. Nevertheless, numerous cases of public elections under IRV exist, and numerous more elections will be held, providing ample opportunity for skeptics to conduct natural experiments. Simply pretest a sample of voters to determine their preferences and, after the election, posttest to see if they strategically reversed those preferences when marking the ballot. With respect to the second point, how to create space for third parties in public office... Casual observation and, increasingly, the literature suggest that changes in electoral rules follow changes in party systems. So, to an extent, this internecine feuding among partisans of various single-winner systems is a waste of everyone's time - insofar as the objective of reform is to create space for third parties. Which leads to some points about your third point, which is not "back on topic." In fact, it is a new topic. Too much of the internecine feud - and that's all it is - is grounded in hazy speculation on voters' "mental preferences" (to borrow a term) and hyper-abstract mathematical models that make questionable assumptions about the fullness of a voter's information as (s)he enters the voting booth. Not enough of the feud is grounded in empirical - or even plausible - claims. And when the arguments get hard to make, parties bring in new points like your third one here about whether runoffs and/or IRV "manufacture" majorities. To do any good for anyone other than members of the proverbial choir, the discussion needs to be structured and grounded in empirical reality. In-line replies and the introduction of non-germane claims make arguments (1) verbose and (2) hard to follow. Together with the knee-jerk posting of stock text, the constant introduction of new, non-germane claims says something about the strategy of IRV's most vocal opponents. It is to advance other systems by bashing instant runoff voting. Hypothesis: very few are paying attention to a discussion that's mired in noise pollution.