Life Imitating Art: New Jersey Edition

My last post centered around a set of example elections that serve to highlight the short-comings of instant runoff voting (IRV). Going into this Tuesday's election, we have two highly-publicized real-world elections that also perfectly highlight how IRV fails: the New Jersey governor race and special election for the US House seat for New York's 23rd district. Here, I will cover the New Jersey race

In New Jersey, Democrat John Corzine and Republican Chris Christie are very nearly tied, while Independent Chris Daggett has faded into the background after reaching a peak of about 14% in earlier polls. On the 5th of October, a piece was printed in the Times of Trenton claiming that IRV is the panacea for this problem, an article which was recently quoted on and supported by Independent Political Report, a pro-third-party website. But IRV would not be a good choice for the independent candidate in this election, and would elect the same winner as plurality would. Let's walk through the logic.

First, the data. Political Wire points us to the latest FDU Public Mind Poll. One of the questions they asked was whether the voter would vote for Corzine (D) or Christie (R); the answers fell for Corzine, 44% to 43%; but while 4% were undecided, 6% named Daggett, even unprompted. They also asked the question naming all three possibilities; at that point, Daggett's support climbs to 14%, but then Christie takes the lead, 41% to 39%.

What does this tell us? Daggett's support is significant, and comes about 2:1 from otherwise-Democratic voters. But even if we were to double his vote share to 28%, while pulling in at the same 2:1 proportion, the election would go 29% for Corzine with only 28% for Daggett (with 36% going to Christie, giving him a clear win.) With double support, Daggett still comes in third! And since IRV decides who to eliminate based on first-place support, there is no way that Daggett can avoid being the first of these three to be eliminated if this election were held under IRV. We do see, however, that once he is eliminated, his supporters second-choice votes could change the outcome from Christie to Corzine. In other words, IRV isn't pro-third-party; in this election, it's pro-Democrat.

Digging deeper into the FDU poll, we find a different set of questions, which might shed some more light on what's happening here. Instead of posing the election as an either/or choice, they ask whether the voter views the candidate favorably or unfavorably, and whether that view is somewhat strong or very strong. The results are illuminating. I'll leave the names out (as well as the percentages for "no opinion" and "haven't heard of the candidate"); with just this information, which candidate do you think voters would most want to win?

Secret Name	Very Favorable	Somewhat Favorable	Somewhat Unfavorable	Very Unfavorable
Mystery Candidate A	15%	24%	17%	37%
Mystery Candidate B	17%	24%	20%	24%
Mystery Candidate C	6%	22%	15%	8%

Looking at just these numbers, who hard to tell which candidate "deserves" to win. B has the most "very favorable" ratings, and A has the most "very unfavorable"; but B has more "somewhat unfavorable". Meanwhile C has the fewest "somewhat unfavorable" and by a large margin the fewest "very unfavorable", but also the fewest "very favorable". Really, the choice is between two options that much of the electorate will absolutely despise and a few will cheer, and one choice that won't ruffle any feathers but will only have a few ecstatic celebrations. Is it more important to make a few people very happy at other's expense, or is it better to shy away from the extremes to have an overall slightly-satisfied balance? To give in to extremes, or to reach an equitable compromise?

I think the answer is to compromise. I think the preferences of the voters of New Jersey would be most closely matched by electing Candidate C; and that candidate (if you haven't already figured it out), is Daggett. But under plurality, as well as under IRV, voters are pushed away from the real compromise.

There are several ways to take the FPU data and interpret it as a score voting election. I chose to take "very unfavorable" = 0, "somewhat unfavorable" = 1, "no opinion" = 2, "somewhat favorable" = 3, "very favorable" = 4 (and "haven't heard of" = X; see rangevoting.org's front page.) These values give final scores of Corzine = 1.6, Christie = 1.9, and Daggett = 2.0. Now, you may or may not agree that that's a fair rubric for scoring (if you have suggestions, please share in the comments), but to me it doesn't seem like a score that could even be coming from the same electorate that intends to vote 44% for Corzine, 43% for Christie, and 6% for Daggett. It seems to me like a large number of Daggett supporters are going to begrudgingly vote for Corzine, even though they don't like him. Those people will be wide-open to the false promises of IRV, when what they really need is score voting.