Friday, October 28, 2011

Significant or Not

Last night I was helping Jenna with her Statistic course by studying the topic of regression analysis.  The text pointed to the fact that correlation does NOT mean cause and effect.  Of course it is vary tempting to draw conclusions from statistical data.   Coincidentally, I also stayed up to 1am watching Game Six of the Rangers vs Cardinals on TV - a game that will be remembered and maybe even compared to the  1975 World Series Game Six.

I remember taking a Wall Street Journal article into Dr. Bolger's "Theory of Statistics" class back in the mid 70's.  The punch line of the article was that for equally matched teams (and all other variables treated equally), the probability of World Series teams going seven games is less than 31.25%. But surprisingly, from 1952 to 1976 there had been 14 out of 25 (56%) of the World Series games go the distance.  The article ended with the statement :  The first televised World Series was in 1947 (and coincidentally went 7 games).

By 2002 the probability had dropped to 48%  (still more that 31.25% but is it significant?).  I remember Dr. Bolger immediately taking the article into our classroom at Miami(of Ohio) and using it as re-inforcement to what we were learning at the time - "what is considered a significant difference from an actual sample to the theoretical distribution.

So was TV the cause?  Who cares we all want significant (extra games and innings) entertainment.

PS - This is the first World Series since 2002 that has gone the distance.  See:  Inside Science

No comments:

Post a Comment