Avtar Dhanota, one of Hanlan's Masters scullers is an inveterate writer of letters to the editor. Today on June 11, The Globe & Mail published another of his missives. This time Avtar was commenting on a recent article about statistical results reached by using false correlations.
"It's very easy to mislead with statistical tests if we don't use our common sense...If we start with a useless hypothesis, we arrive at useless results. [For example],
A researcher amputates one leg of a cockroach, puts the cockroach on the table and shouts at it to run. Of course, it will try to run to hide. Then the researcher amputates all the remaining legs except one, shouts at the poor cockroach and it moves a little.
When the researcher amputates the last leg, then shouts at it to move, it can't. The researcher hypothesizes that when we amputate all the legs of cockroaches, they lose their hearing. Perfect correlation, but it does not imply causation."