Thursday, November 4, 2010

How it Might Have Been Different

Sorry for the lack of posting, but my political energy has all gone towards volunteering my time to my favorite candidates. One of those candidates was Scott McAdams, who ran in the race I am about to discuss. I have tried to keep my partisan views out of this piece as best as I can and to refer only to polls and election results in the most dispassionate way possible, but I feel it would be deceptive of me to not acknowledge my connection to the campaign.

Sometimes it's hard to articulate how approval voting or score voting would serve to create better electoral results. The typical election quickly boils down to just two good choices, and despite the rare holdout, most voters begrudgingly accept that they have a binary choice to make. But when there are only two options, every voting system produces the same result; the improvements only come by making more choices viable. Yet, for whatever reason, people have a hard time imagining how a better system encourages more choices, preferring instead to just complain about the lack of options or to simply demand more options, without supplying any mechanism that could encourage such a thing.

Approval voting and score voting are the mechanisms. And this election has actually provided an example to illustrate it: the three-way Alaska Senate race. Thanks to Public Policy Polling, we have some excellent information [PDF] pertaining to the candidate's favorability ratings. Favorability is a pretty good proxy for three-point score voting (where the scores are -1/0/+1). The first thing that jumps out when we look at these numbers, is that Miller and Murkowski had some of the worst favorability scores out of all Senate candidates across the entire country. Miller had 36% favorable versus an astounding 59% disfavorable, for a -23 net, and Murkowski had a 37/53, for a -16 net. But McAdams had one of the best favorability scores in the country, with a 50/30 for a +20 net.

This means that, under this simple score voting system, McAdams would likely have won this race, and it wouldn't have even been close. Under plurality though, he got only 24% of the vote, while Miller got 34% and Murkowski got up to 41% (41% is the total for all write-ins, but polling suggest that about 95% of those (so about 39% of the total) are for Murkowski). Looking at just these numbers it's clear that not only did a large number of Murkowski-and-McAdams approving voters chose to go with Murkowski, but that at least some Alaskans voted for Murkowski in spite of the fact that they did not approve of her.

Clearly, this is an example of tactical voting. No (deterministic) system is completely immune to tactical voting, and we can't be fully certain that voters would have voted precisely how the poll suggest they would have if they knew their favorability opinions would decide the outcome (indeed, some PPP favorability polls would have called narrow two-way races backwards). But, assuming the poll represents honest opinions, we can be certain that a majority of Alaskans (i.e., over 50%) would be disappointed with either a Miller win or a Murkowski win, while about half would have been satisfied with a McAdams win. And yet, the winner will be either Miller or Murkowski, and certainly not McAdams, depending on how many write-ins survive the auditing process.

Plurality voting breaks down when there are more than two good choices. That's why most elections see just two "viable" candidates. But simply adding more candidates (or even more parties) accomplishes nothing, because we don't have a system that can deal with more than two options. We have a two-party political system because we have a two-party voting system. But if we change to a better system, one where we can get better results by adding more candidates, only then can we find success by doing so.

No comments:

Post a Comment