I disagree with Hightower.

What you will find here is: a centrist's view of current events;
a collection of thoughts, arguments, and observations
that I have found appealing and/or amusing over the years;
and, if you choose, your civil contributions which will make it into a conversation.

He not busy bein' born, is busy dyin'. - Bob Dylan

Please refer to participants only by their designated identities.

suggestion for US citizens: When a form asks for your race, write in: -- American

Wednesday, January 30, 2013

Stasis fallacy 3

The stasis fallacy is the assumption that if the rules are changed, the actors will play the game in the same way they did before.

Nate Silver's Five Thirty Eight (Jan. 25, 2013) contains the very interesting article Electoral College Changes Would Pose Danger for Democrats by Micah Cohen in which there is discussed a variation of how the Electoral College votes.  Mr. Cohen ends his article with an example of the stasis fallacy by making the logically invalid assertion that with one alternative method Mr. Romney would have won the last election.

When voting in the Electoral College almost all states use the winner take all system: the candidate who gets the MOST popular (people) votes in the state gets ALL of that state’s electoral votes.  Exceptions are Nebraska and Maine which use the “congressional district” method which allows 1 electoral vote to the winner of each congressional district and two for the state at large.  (There was a variation in Virginia which would given the two statewide electors to the candidate who received the most district electors.  For our purposes, this is no different from the congressional district method.)

Mr. Cohen includes a complete compilation of how the votes that were cast in the 2012 election would have determined the electors if they had been selected by an alternative method:  a modified "congressional district" method.

"If the Virginia proposal had been adopted in 2012 and the other 47 states had adopted a system like Maine’s and Nebraska’s (where one electoral vote goes to the winner of each Congressional district and two votes go to the statewide winner), then 103 of Mr. Obama’s blue state electoral votes would have gone to Mr. Romney. Mr. Obama, by contrast, would have gained only 32 electoral votes in red states."
Mr. Cohen then moves from arithmetic to the area of logic as follows:
"And, by an Electoral College count of 277 to 261, Mr. Romney would be president."

This is not a valid conclusion, unless Mr. Cohen means that, after the election was conducted according to law, God came down and declared that the votes would be counted differently. However, the drift of the article is that a CHANGE to this new system would give the Republicans an advantage, which indicates that he means to be assuming that the new system would have been the law BEFORE the election.

Notice that I am not saying that Mr. Romney would not have won.  I am saying that the information at hand does not assure us that he would have won.

The flaw in the thinking in the article is the assumption that is implicit in counting the votes that were cast using method A and assuming that they would be cast in exactly the same way if the election were held by method B.  That in turn implies that the candidates would have campaigned (in election method B) in exactly the same way that they had campaigned in election method A.

This is the error.  Consider Texas.  In method A, Obama ignores Texas because it is hopeless.  In method B, Obama would likely campaign in Houston, Austin, Ft. Worth and perhaps Dallas.

For another example, consider Pennsylvania, 2012.  In method A, Obama, by the way he campaigned, piled up huge majorities in Philly and Pittsburg to overcome the Romney vote in the rural areas.  In this way he carried the whole state of PA.  In method B, those extra votes in the large cities will not be of nearly as much value (although they can still help win the 2 statewide electors).  Therefore, if we used method B, Obama would be moved to revise his campaign plan to be more consistent with the reality of that different method of counting.  

Once you recognize that the method of campaigning would change, then surely it becomes clear that the votes could change.

 Consider a simpler example from a different election. Al Gore obtained more popular votes for president than George Bush did in the presidential election of 2000. But Bush won the electoral vote and the presidency. We heard over and over again: “If the election of 2000 had been decided by popular vote, then Gore would have won.” The response was always some variation of: “But the election was not by popular vote”. For some purposes, that is a sufficient answer. But I think that this is worth a bit more reflection. Put aside your own political perspective and consider the logic of the statement: “If the election of 2000 had been decided by popular vote, then Gore would have won.” Again, suppose one is talking about God coming down after the election is over and decreeing that the voting system has been changed to popular vote. If that is what was meant by the statement, then the statement is obviously true, but also silly. For the statement to be meaningful it has to be about what would have happened if the rules had been changed (to popular vote election) before the election took place. To claim that the vote as actually cast would have been duplicated with those other rules, then one has to assume (among other things) that the candidates would have campaigned in exactly the same way. That is nonsense. In fact, if the rules had been changed, then the candidates would have campaigned very differently. With the existing system, it was totally irrelevant whether Gore won California (or Bush won Texas) by 1 million votes or 3 million votes. Either way he still got all of the electoral votes of that state and nothing else. There was no reason for either candidate to try to win the national popular vote. And they didn’t. If the rules had been changed to national popular vote, then Gore would have spent more time on the coasts and Bush on flyover country getting out their base voters. It is very possible - perhaps even likely - that Gore would have won.  But it is not certain. Also, even if one could know that Gore would have won a hypothetical national popular vote election it would have been because of how the votes had been cast in that completely different kind of election.  It would not have been because of how the votes were actually cast in 2000.

1 comment:

  1. Very interesting analysis. You bring up the flaw in the analysis statement that ends with, "all other things remaining equal". There has been much written regarding what is called the "butterfly effect" in the world of chaos theory, not that I am an expert or even knowledgeable in that area. But having read some on the subject I have a foggy idea of what it is. As I understand it a simple statement of the thesis is that outcome is highly sensitive to initial conditions. It seems to fit with Wayne's analysis that any changes in the rules at the beginning of the process can and most likely would have a ripple effect throughout the entire process making predictions of the eventual outcome much more complicated than just an arithmetic accounting for the initial change.

    Very good post Wayne. Thanks.