When I saw this op-ed by Phil Kiesling the other day (recommending nonpartisan primary elections as a way of reducing polarization in Congress), I had several thoughts.
John Sides offers a more thorough, research-based discussion of the effects of open primaries, which I’ll discuss below. But first my immediate reactions to Kiesling’s op-ed:
1. The proposal seemed reasonable to me. Electorates in partisan primaries are much more partisan (of course) than general election voters.
2. There will certainly be side effects. (I was going to say “unanticipated consequences” but then I realized that’s not the right thing to call them since I’m anticipating them right now.) For example, you could have a district that’s 60% in favor or party A, but then in the primary there could be 5 candidates from party A and only 1 each of parties B and C (because the candidates from those parties did a better job of coordination and picked just one candidate). The resulting general election–a matchup between the two leading vote-getters in the primary–could then be B vs. C, which doesn’t seem quite right.
3. On second thought, though, I don’t see the above problem as so serious. Every now and then a congessmember gets elected isn’t a good match for the views of his or her district. If it’s a big problem, the congressmember won’t be reelected. If the larger effects are beneficial, it’s hardly a tragedy for the wrong winner to be sitting in Congress for two years.
4. The larger problem I’d worry about is that the aforementioned need for coordination would lead to an unofficial prmary-before-the-primary where each party tried to winnow down its field to one candidate. Related to this is the nature of multiple-candidate voting, which is inherently unstable, both because differences between candidates of the same party can be minor and difficult to perceive, and because of the potential for strategic (or, as the British would say, tactical) voting, not wanting to waste one’s vote on a candidate who doesn’t have a chance to be in the top two.
5. That said, the above problems are already occurring in closed primaries. So, to the extent that polarization is a concern, open primaries might well be a big deal.
6. My final thought about Kiesling’s proposal was that political scientists would be inclined to bash it. I’ve noticed for many years that political scientists tend to be dismissive of campaign reform proposals, partly from research experience–it’s rare to find that one factor can make a big difference–and partly because of a general support of the political system as it is.
In his discussion, John didn’t disappoint me regarding point 6. This is not to say that the study he cites has any mistakes. I just think that John has put a political scientist’s spin by emphasizing the null results. John summarizes the estimated effects as:
(1) not big enough to do anything near what the proponents of primary reform claim, and
(2) not big enough to affect the day-to-day dynamics of policymaking in state and national legislatures.
I’ll buy the first point–certainly, advocates exaggerate the importance of their pet ideas. But I don’t know about #2. Remember that zero Republicans voted for the recent health care bill. And we all know about the California state legislature. I’d think that even a few crossover legislators of each party and a few centrist candidates running and winning election, might make a big difference in how the legislature operates.
Changing priorities?
One more thing. It’s my impression that the people who propose political reforms of this sort are generally on the center-left of the political spectrum. (For example, Phil Kiesling, who wrote above-linked op-ed, is identified as a Democrat.) With all the struggle with the health care bill, lots of Democrats were upset about polarization. Now that the bill has passed, a lot of people on the center-left might start to have warmer feelings about political polarization and party discipline. I say this not to call into question any of the research that John was cited, but just to reflect upon the timing of some of these proposals. My impression is that voters on the whole remain unhappy about polarization, even if they are unlikely to go so far as to express such feelings at the ballot box.
Point 4 is exactly why primaries exist, isn't it? So that the main election can have just one candidate from each party, so that you don't have (say) Hillary Clinton and Barack Obama splitting the Democrat vote and letting John McCain win even though most people want a Democrat.
You can reduce the pain involved in this kind of election by not using first-past-the-post; for example, if voters can specify an order of preference and there's some algorithm so that when a candidate is eliminated, all voters who preferred that candidate get their votes shifted to their second choice, then you can avoid the vote-splitting problem. There's a theorem that no voting system can be without one of a short list of important flaws, but one can do much better than first-past-the-post (and some places, including I think Australia, do).
Looking from abroad, this is just bizarre.
The purpose of the primaries is for a party to select their candidate in the election. Opening that up to the outside seems to defeat the point: they're no longer the party's candidate. Plus. it's a great way of screwing the opposition: vote for people like Orly Taitz as their candidate. or even vote for a Republican as the Democrat candidate.
You're attacking the wrong problem. The issue is that your system only realistically allows two parties at a time.
Let each party select their candidates however they want and instead make it feasible to get a wider spectrum of parties electable for congress. For example, make state congress elections strictly proportional, with each running party getting congressmen according to the number of votes cast.
@Anne
The editorial is advocating a 'non partisan blanket primary' or jungle primary, which can result in two candidates from the same political party on the final ballot. It eliminates a pair of ‘first past the post’ elections (one in the primary, then one in the general election) with a pair of elections in which the second would function as a run-off system for the two leading candidates in the first election. As an example, in the 2008 presidential election, Obama / Clinton might have been able to get enough support than any of the Republican candidates.
A jungle primary would result in more centrist elections for heavily partisan districts, as voters on the right / left would elect candidates from the opposite political party closer to their position. Someone who is more into political data could analyze Washington State, which recently switched primary systems.
I agree with the above posters. Changing the primary system is an ugly kludge fix to a more systematic problem. The underlying problem is that picking the candidate with the most votes only gives reasonable results when there are two candidates.
There are a number of good alternatives. IRV (mentioned by Anne) is used in Australia and mitigates spoiler effects. Might I dare to dream that we might one day use a Condorcet method!
The other systematic problem with our voting system is that neither house has representatives that are,.. well,… representative of the ideological diversity of the population. It would be great to see an election method used that creates proportional representation.
@Cody
That just moves the spoiler effect problem to the first election. Suppose in an evenly decided district, 5 democrats run and two republicans. The 5 democrats split 50% of the vote with 10% a piece, and the two republicans get 25% each. The general election then has only republicans. IRV is like a simulated series f run-off elections, where the weakest candidate is removed at each step, and voters specify their preferences ahead of time.
My personal preference would be to just dispose of the party system all together. I understand it's human nature to classify things by categories and it helps inform me about someone's views, in a general sense, by looking at their party label – but I still think it's utterly worthless. I think the labels promote absurd partisanship since nobody likes to look like they are going against their in-group. It also centralizes power in a way that prevents independent, objective thought about issues since you always have to worry about your party leadership harassing you.
Obviously what I'm suggesting is never going to happen and might even have unintended effects since not all voters (or many) are particularly well informed and they use party as a tool to make up for that.
"There's a theorem…"
Indeed, and Kenneth Arrow won a Nobel prize for it!
http://en.wikipedia.org/wiki/Arrow%27s_impossibil…
But there are much better systems than those currently in use here.
"My personal preference would be to just dispose of the party system all together."
Of course, the "party system" isn't mentioned in the Constitution. It arose because there were incentives in the way things were set up that encouraged parties, and in the U.S. basically just two parties. The only way to change that bias is to change the incentives, which would probably require constitutional amendments, so it will never happen. Probably (subjective probability of course!)
@Ian
I agree that changing the voting system is the best fix, but IRV replicates the results of Plurality in a nonpartisan election, like a blanket primary. Condorcet methods do better, but they still have problems and I don't know how well they would work since they still suffer from favorite betrayal and spoiler style problems.
The only real fix (other than Proportional Representation) would be Score voting, which is a cardinal rating of each candidate, and thus removes things like spoilers and vote splitting. As a nice side effect, centrist candidates can often garner some points from both sides of the spectrum, and if they're pretty good, they will probably win.
I live in Washington State where we currently use the top two primary, and it is wildly popular among the electorate for reasons that always escape me. At the end of the day if you live in Eastern Washington, the primary gives you the chance to vote between two Republicans in the general election — each trying to out do the other to prove their Republican bona fides. If you live in Western Washington it will be between two Democrats (this isn't always the case, but it happens more often than not).
At the end of the day, the primaries are about who can mobilize the base, and the dominate parties do a much better job of doing it than the weaker party. I suppose if you had perfectly balanced political districts with equal numbers of Democrats and Republican this would be a great system. But as long as we segregate ourselves politically, this system just ensures major party domination.