02-11-2014, 10:44 PM
OK, then let me ask you about this. From 1998 to 2011 the NFC won the coin toss in the Super Bowl every time. If I am not mistaken, the probability of that happening is 1/2^14 or 0.006%. So this would fall far below any alpha level used in behavioral research. But we wouldn't say that the 14-year run is statistically significant, even though it is extracted from a larger population of coin tosses over 48 Super Bowls, would we? Or if we would claim statistical significance, what's the point (other than claiming that the NFC must have cheated during that time period)? That 14-year run has no bearing on future coin tosses.
If there's one thing people have learned in Berkeley, it's that you can't put "Cal football" and "Rose Bowl" in the same sentence. The words just don't fit. You'd be better off linking "covered wagons" with "Mars travel." (Bruce Jenkins, Cal '70, SF Chronicle, Nov. 11, 2017)