In rebus mediis est sapientia
Leonard Mlodinow is a physicist at the California Institute of Technology. His recent book is The Drunkard’s Walk. This is a book about statistical inference. We at Doombrothers have previously warned about statisticians (see Doombrothers November 22, 2008). However, we feel called upon to re-alert you about the dangers of this baleful pseudo-science.
Dr. Mlodinow, who clearly needs to buy a vowel, discusses the horrible statistical phenomenon of “regression to the mean.” What he speaks of with apparent unconcern and insouciance is the phenomenon that performance or gifts or acquisitions out of proportion to the ordinary are typically followed by gifts, or acquisitions or performances that are more in line with the ordinary. So, for example, landing on the front page of Sports Illustrated is associated with an extraordinary performance one week and will more typically be followed by a less than extraordinary performance the following weeks – thus the “jinx” of the magazine. Or, the extraordinary achievements of one man, say a Henry Ford, are followed by children with less than extraordinary gifts.
What Dr. Mlodinow DID NOT say to you is that this implies that we, at the very apex of achievement are gradually slipping downward in an inexorable slide toward mediocrity, the arithmetic mean. Statistically speaking, the more amazing are our lives or our feats the more rapidly we will fail and the further we will fall. As an illustrative aside, regression to the mean implies that one Doombrother is fated to be gap-toothed, barely literate and semi-articulate, wandering the hills and hollows of Appalachia and marrying my first cousin while the other is fated to a similar outcome in the swampy regions of the Middle Atlantic States.
Leonard Mlodinow is a physicist at the California Institute of Technology. His recent book is The Drunkard’s Walk. This is a book about statistical inference. We at Doombrothers have previously warned about statisticians (see Doombrothers November 22, 2008). However, we feel called upon to re-alert you about the dangers of this baleful pseudo-science.
Dr. Mlodinow, who clearly needs to buy a vowel, discusses the horrible statistical phenomenon of “regression to the mean.” What he speaks of with apparent unconcern and insouciance is the phenomenon that performance or gifts or acquisitions out of proportion to the ordinary are typically followed by gifts, or acquisitions or performances that are more in line with the ordinary. So, for example, landing on the front page of Sports Illustrated is associated with an extraordinary performance one week and will more typically be followed by a less than extraordinary performance the following weeks – thus the “jinx” of the magazine. Or, the extraordinary achievements of one man, say a Henry Ford, are followed by children with less than extraordinary gifts.
What Dr. Mlodinow DID NOT say to you is that this implies that we, at the very apex of achievement are gradually slipping downward in an inexorable slide toward mediocrity, the arithmetic mean. Statistically speaking, the more amazing are our lives or our feats the more rapidly we will fail and the further we will fall. As an illustrative aside, regression to the mean implies that one Doombrother is fated to be gap-toothed, barely literate and semi-articulate, wandering the hills and hollows of Appalachia and marrying my first cousin while the other is fated to a similar outcome in the swampy regions of the Middle Atlantic States.
We at Doombrothers have carefully studied this conundrum and we believe we have crafted an elegant, foolproof and realistic alternative to the fatal slippage and inherent disappointment of this doomful statistical nonsense. We have decided to EMBRACE THE MEAN. If we comport ourselves in an average fashion then we will not fall far away from the mean and thus will not be a disappointment to ourselves or to others!
Posts
