In last week's Iowa Democratic Party caucuses, a number of local gatherings were, after several hours of peer pressure and guilt-tripping, tied.  Party rules, in place in advance of the caucuses, provide for a tie break by way of a coin flip.  In a number of these tied votes, Team Hillary won the toss and elected to receive.  (There's no kicking to the clock in caucuses.  Stay with me, though, I'm about to make a serious point.)

Apparently, more than a few coin flips went Senator Sanders's way, but that hasn't stopped some observers from alleging corruption by way of bad statistical inference.  Here's Counter Punch's John V. Walsh, with an instructively complete demonstration of the error.
What are the odds of one of two candidates winning all six coin tosses if the outcomes are random, that is, if the tosses are fair, unbiased and with honest coins?

The calculation is so simple that a schoolboy or schoolgirl can do it. The formula is simply 1/2 raised to the power of 6 – that is, 1/2 taken six times and multiplied.

The probability of winning all six tosses by chance alone is 1/64. That is 0.016 or 1.6 in 100 or 1.6%. Not even 2%! In many areas of science including many areas of biology, one must demonstrate that the result of one’s experiments is unlikely to happen by chance alone. If the probability of getting the results by chance alone is less than less than 5%, the result reported is considered to be “significant,’ that is, not likely to be a chance finding. Such a result is publishable in highly respected journals.

Since the probability of the outcome in Iowa was 1.6%, it is quite unlikely, highly improbable that the coin tosses resulted from chance and were honest. And if the results did not occur by chance alone, then the coin tosses were manipulated, fixed! Why has no one in the mainstream media looked into this?
Perhaps there are enough mathematically numerate people in the mainstream media to understand that, in Probability Jeopardy, 1/64 is the answer to "What is the probability of a fair coin coming up heads six times in six tosses, Alex?"  The judges will also accept "What is the probability of a fair coin coming up tails six times in six tosses?"

A better way of framing the problem is "On any given Sunday, what are the odds of six visiting teams winning the coin toss in the early games?"  Different captain, different referee, different choice of heads or tails by each captain, and a much harder problem to frame.  But that's how the local tie breaks go down.  Team Clinton or Team Sanders designates a captain to call the toss, the party organization provides a coin flipper, and the winner receives.  Simplest explanation is that Team Clinton had some lucky toss-callers that evening.

No comments: