Monday, September 24, 2012

HOUSE EFFECT


Nate Silver, the statistics guru who now writes for the NY TIMES, is well aware of the fact that a sizable share of his readership consists of freaked out junkies like myself who crave not merely daily but hourly analyses of the upcoming presidential election.  Since he rigorously confines himself to an analysis of polling data, eschewing speculation or political commentary, he is on some days hard put to come up with something new to satisfy our cravings.  Today, for example, having nothing much in  the way of new data to crunch, he devoted an entire column to analyzing the extremely outré circumstances under which the entire election could turn on the results in the Congressional District that includes Omaha, Nebraska.  Nebraska is one of only two states [the other is Maine] that award electoral votes by Congressional District, and in 2008, in that overwhelmingly Republican state, Obama actually snatched a single electoral vote from the sea of red by carrying that district by a single percentage point.  The odds of the election turning on that electoral vote, by the way, are apparently a thousand to one, but the Nebraska Republicans, stung by the insult to their purity, have rearranged the district so that it is unlikely to occur again.

Another wonkish topic Silver spent some time discussing is the tendency of some polling firms systematically to tilt more toward one side or the other than the other firms that are doing polling.  Rasmussen, for example, can be relied upon to tilt Republican.  This tendency is called by statistics nerds "the house effect" [for reasons a quick Google search failed to disclose.]  It is not that any firms actually cook their data.  Rather, the technique they use for forming their sample or making their contacts has built in biases one way or the other.

For example, pollsters regularly distinguish between the universe of Registered Voters [RVs] and the universe of Likely Voters [LVs].  A poll of RVs will tilt more to the Democrats than a poll of LVs, because the sub-populations strongest in their support for the Democrats [Hispanics, young people] are also less likely to turn out and vote than the sub-populations favoring the Republicans [rich people, old people, white people].  Silver has no patience with those who average polls of RVs with polls of LVs.  There is also a good deal of disagreement among pollsters as to how one identifies LVs.

There are other sources of bias, or a House Effect.  Rasmussen leans Republican in part because it calls people only on landlines.  But young people, who tend to favor Obama, are more likely to be reachable only on cell phones.

Since I obsess a good deal about these matters, I got to thinking last night at one a.m. about other examples of House Effects.  One that occurred to me concerns my life-long tendency to daydream.  As readers of my Memoir will know, I spend a goodly part of my waking hours in my head, daydreaming.  Sometimes I daydream about having magical powers, with which I correct much that is wrong in the world.  I also do my share of daydreaming about sexual conquests, of course.  And a not inconsiderable portion of my daydreaming is actually a form of work, in which I give extended lectures to imaginary audiences as a way of thinking through  a theoretical problem.  On occasion, I write imaginary reviews of books I have published.

I have noticed that my daydreams exhibit a significant House Effect.  They tend to biased in my direction.  There are limits, of course.  I do not write reviews of my books that begin, "Not since Immanuel Kant burst upon the scene has an author so stunned the philosophical world ...", but the reviews do tend to be somewhat more enthusiastic about the arguments of which I am singularly proud, and unusually forgiving of such shortcomings as I am willing to acknowledge.

You might think that such self-regarding daydreaming would, from the standpoint of a utilitarian calculation, be a net minus, since the momentary pleasure from the imagined review would be more than compensated for by the eventual pain of realizing that it was only a daydream.   But like Walter Mitty, I am undeterred by such hard-eyed calculations.  I have long since become inured to the disappointments of reality, while remaining enraptured by the enticements of fantasy.

There is one odd counter-example, however,  As I am now only fifteenth months from my eightieth birthday, I have begun musing on the possibility of throwing myself a big party on December 27, 2013.  I thought maybe I would hold it in Paris [which would have the effect of keeping the attendance within manageable limits].  I would invite everyone:  family, friends, former students, colleagues, everyone who reads this blog.  We would take over an entire restaurant in Paris and spend the evening celebrating -- me.  But somehow, the prospect does not enchant.

I think the problem is like that attendant upon the fantasy of being present at your own funeral.  You can count on the speeches at the funeral being encomia -- de mortuis nil nisi bonum and all that -- but the problem is that you would be dead.   An eightieth birthday bash carries with it the suggestion that you are pretty well finished and are just putting a cap on it.

I guess I will wait until I am ninety.

3 comments:

  1. As I can handle only a small amount per week of campaign speculation, I almost checked out of this post early on. So glad I went on. Love the imaginary lectures and book reviews. in which I too indulge. As for birthday celebrations, it isn't a question of 80 OR 90. You could do BOTH!

    ReplyDelete
  2. When my grandfather died, the minister who did the eulogy before he was put in the ground or "laid to rest" used my name instead of his. His middle name was Harley and mine is Harris but otherwise our first and last names are the same. So at his funeral the minister went on and on "as we lay to rest John Harris Doe" and continued the using the name over and over. There I was watching my own name being used to bury someone else. Sometimes you don't have to daydream about such things, they really happen. Annd I can only comment "the reports of my death have been greatly exaggerated". Not an original comment but in my case it really happened.

    ReplyDelete