The phantom considers himself a man of science. Observe, measure, run the R correlation coefficient, analyze.
On the other hand, dealing with an election, the phantom lapsed.
the phantom knows that correlation is not causality, but when you look at the "Redskin Rule" which says if the Washington Redskins football team loses its home game immediately before election Tuesday, the incumbent President loses the election in every one of 18 elections since 1937, the phantom says, "There cannot be a causal connection, but there are few things in life with such a straight line correlation: It must mean something."
A statistician would say, presumably, not enough measurements. A study which is "insufficiently powered."
A rational person would say, "There cannot be a connection between the actions of two teams of football players and the decisions of 200 million voters. You cannot imagine a connection, therefore, there cannot be one."
On the other hand, one can imagine a connection: Since football is avidly followed by large percentage of the population, it is possible a significant number of "swing voters" may be subtly influenced by the perception of the effectiveness of the prototypical "establishment" institution, the home team of Washington, DC. So if "Washington" is seen to be dysfunctional, ineffective and the incumbent is running on the claim Washington is effective and a winner, seeing the government surrogate lose may cause vast numbers of people to perceive the federal government as being ineffectual. After all, citizens are polled all the time with nebulous questions like, "Is the country headed in the right direction?" which is supposed to provide a measurement of whether people perceive the "captain of the ship" to be directing the course of the ship of state effectively.
So, if the Redskins lose, it's well, they can't do anything right in Washington. Bunch of losers, vote them out.
If, on the other hand, the Redskins win, well, Washington is run by competent people, people who can order the SEALS to find and kill Osama Bin Laden.
All of this, of course, would operate subliminally. But if the subconscious of the American electorate were not thought to be important, there would be a lot of advertising shops out of business during election seasons.
The more plausible explanation, however, is this remarkable streak is simply a coincidence and if we had 100 games before 100 elections, we would see there is no correlation.
As Patty McKenzie, the intrepid organizer of the effort to win the election for Obama in the town of Hampton, New Hampshire, told the phantom: There are data which correlate with the likely outcome of the election, based on actual research and the correlations are easy to understand. If you walk up to a voter's door ten times before the election, and you confront him with a human being (as opposed to a TV spot) and you say, look, you say you are leaning toward Mr. Obama, are you voting for him or not? And if he says, yes, I guess I am. Then you pin him down to what time of the day on election day he intends to vote, then you get him thinking, picturing his day, he is more likely to follow through. That's a cause and effect correlation you can easily see.
And if you focus your efforts in areas which are closely contested, and ignore the states which are already safely in your column (NY, MD, RI) then you maximize the punch per energy spent, and you win the election.
Of course, if you look at the nitty gritty about what actually happens on that voter's front porch, you might not believe the visit was effective, but that's another story. At least the approach McKenzie suggests is understandable. The fate of an election based on the performance of a football team, no matter how the numbers cannot lie, they do.
You are learning grasshopper! There is hope for you.
ReplyDelete