Wednesday, August 24, 2011

Magical Thinking about Top Two Elections

The Hankster recently posted an article that mentioned Washington state’s new top two general election, which is preceded by an open primary. Upon clicking the link I found my friend Richard Winger having another one of his automatic fits triggered by the words “top two.” Top two, if not the root of all evil, at least is the bane of third parties, and the friend of incumbents.

But upon close examination, anyone can see that Richard is espousing unscientific mysticism, or magical thinking.

There is no causal relationship of any sort between a top two process for taking the final vote on candidates for office, and favoritism for incumbents or bias against minor parties.

There is only one causal relationship attributable to the final top two vote: namely, the winner receives a majority of the votes.

Richard’s argument that top two is biased against third parties and favors incumbents is pure voodoo. The causes of incumbents repeatedly winning re-election are well-known in political science. Incumbents have forged strong connections with the needed elites in their district. Incumbents generally have more campaign money, more name recognition, more activists in their campaigns, and easier access to the local media. Third parties regularly loose elections because they lack all the advantages of incumbency.

Gerrymandered districts also contribute to incumbent advantage. Party primaries are an additional factor that favors incumbents, because incumbent enthusiasts are more likely to turn out for primaries than are uncommitted middle Americans. Rigged districts and holding primaries also favor extremists, and can foster polarization.

Here is the key point missed by Richard: These factors have their determinative effects in the politics that take place long before the top two elections occurs.

Every sausage maker knows that a good sausage is not determined by the gut that holds the sausage together, but by the ingredients poured into the gut. But Richard is the only cook around who ignores the ingredients, and blames the foul taste of bad sausage on the gut, which everyone else regards as neutral in taste.

All the events that occur before the final top two vote are what account for the persistence of immoderate politicians, polarized legislatures, the failures of third parties, and the repeated re-election of incumbents. This is the case for Louisiana, California, Washington, Wisconsin, and every other state with these dysfunctional patterns.

Richard recognizes the dysfunctionality, but blames the victim – the final top two vote – rather than dealing with the intransigent causes. These causes are the natural consequences of a political system dominated by our a-constitutional two-party system. Putting the gut on the sausage is the last step. If the sausage is bad, it is all the steps taken prior to the last step that are to blame. The last step is neutral.

William J. Kelleher, Ph.D.
Face Book:
Twitter: wjkno1
Internet Voting Explained on


richardwinger said...

In 2008, Washington was the only state that used top-two for congressional elections. No incumbent from either house of Congress was defeated in Washington state. But across the nation, 28 members of Congress were defeated for re-election.

Again, in 2010, Washington was the only state that used top-two for congressional elections. No incumbent from either house from Washington state was defeated. But across the nation, 62 congressional incumbents were defeated.

In Louisiana, which switched to partisan congressional elections in 2008, two incumbents were defeated from that state that year, more than had been defeated in the previous 30 years under top-two elections. And again in 2010, an incumbent in Louisiana lost.

DLW said...

Are you going to believe the evidence or are you going to believe a guy with a PhD?

And why no mention of IRV as a better alternative to FPTP than a "top two primary"? Or even what if better rules were used in the first stage of the "top two primary" or if it was guaranteed to be a three-way, using IRV, in the second stage?

A little imagination can go a long ways...

wjk said...

Causal Relationships

To me it seems important to try and reach some sort of consensus as to the validity of the IMPLIED causal relationships suggested by Richard's statements. I have no doubt that he is right about the facts, but I dispute the causal implications. One reason why I think this issue should be clarified is that magical thinking spreads rapidly and widely. Then it becomes an obstacle to progress, even though it is unscientific. Here on The Hankster there have been comments to the effect that top two is a devilish device of the two-party system, intended to permanently side-line third parties and Independents.

When the issue was being debated here in CA, third party leaders cried that their existence was under threat. After the Proposition was passed by the voters, Ralph Nader condemned Prop 14 as sounding the Death Knell for third parties and Independents, citing Richard as his authority, at

Suppose we read Richard's statements as a claim of causation:

Factual premise: "In 2008, Washington was the only state that used top-two for congressional elections."

Implied minor premise: [Top two causes incumbent re-elections]

Conclusion: [Therefore] "No incumbent from either house of Congress was defeated in Washington state."

Richard then assumes that the absence of top two is some sort of necessary condition for the defeat of incumbents: "But across the nation, [where top two could not cause incumbents to win] 28 members of Congress were defeated for re-election."

Richard's statement about Washington in 2010 also implies that top two caused the incumbent re-elections.

In reference to Louisiana, he again assumes that the absence of top two was some sort of necessary condition for the defeat of incumbents.

But incumbents were winning disproportionately long before top two appeared on the scene; so, the causes for this must be other than top two.

Indeed, as I stated in my original post, factors other than top two cause incumbents to win. And, factors other than top two cause third parties and Independents to fail.

Think about what happened in medicine when doctors acted on bad science, and about how medicine improved when doctors started acting on good science. One of the primary aims of science is to get causal relationships straight. But Richard's statements don't do that.

In my opinion, the results of an election are caused by factors in the political culture, but not by the top two final vote. That vote is completely neutral as a causal factor. To put the blame for third party and Independent failures on the wrong object can only have undesirable results. For example, it can result in a continued misunderstanding of the causes of those failures, a continuation of failed practices with wasted energy, or despair about participation in politics.

However, if the correct causal relations are identified, then plans for action can be formulated based on the scientific knowledge.


wjk said...

OK folks, put on your thinking caps and gulp down some strong coffee. This is from the Election Law group at Linkedin.

Prof G__ writes:
There are many reasons for the incumbency advantage, as Kelleher points out. But to claim that there is "no causal relationship of any sort between a top two process ... and favoritism for incumbents and bias against third parties" is simply wrong.

The top-two, or any candidate selection mechanism, has all sorts of strategic effects on the behavior of elites, candidates, parties, and voters. The best treatment of the strategic rationale is Gary Cox's book "Making Votes Count." The rules of the game, including the top two, are NEVER "neutral."


I agree with Professor G__ that the rules of a game are never neutral, and that those rules have some influence on the strategic thinking of the players of the game. But these truisms don’t support the claims of Winger and Nader that top two necessarily favors the two-party system and disadvantages third parties and Independents.

I suggest these points:

1. The one certain causal consequence of a top two final vote is that the winner will have a majority of the votes. No other causal consequence is as sure.

2. What happens below, or prior to, the top two is of much greater causal consequence than the top two final vote itself.

3. The top two final vote does not, of itself, cause specific forms of political behavior in the processes leading up to that final vote. E.g., striving to form majority coalitions may not be necessary to get into the top two.

4. That there will be a top two final vote does not necessarily favor the formation of a two-party system in the primary stages. Strong third parties can win office, so can a charismatic Independent individual.

5. In a state like California, numerous single member districts allow for innumerable unforeseeable opportunities for candidates outside the control of the elites in the two major parties. In the recent special election for a seat in Congress, the favorite of the Democratic party elites, Bowen, lost.

Practical Consequences of this Discussion

6. Preaching on faith that two-party domination is an inevitable consequence of an electoral system with a top two final vote, as Winger and Nader do, can discourage political competition, and can become a self-fulfilling prophesy.

DLW said...

If you're playing Russian roulette, wouldn't you care about your odds of getting the bullet? Richard is saying that empirically, it seems "top two primary" guns are very rarely ever loaded. This is not magically refuted by the fact that the other guns got only one bullet in them.

I'm reminded here of the dispute in "In the Name of the Rose". What is clear is that incumbency advantage is the equivalent of "whether Jesus owned his cloak", while the real issue are the differences in the ownership of property between the Benedictines and the Franciscans.

The real issue here is what sorts of election rules give third party dissenters from the two major parties a constructive role in the political system. And that seems like the sort of thing that autonomous independents ought to care a good deal about...