Thursday, 20 December 2007
Maybe Steroids Don’t Work?
Well the Mitchell report is making lots of press, and enough names have been named to generate some interesting questions. An Article in the Milwaukee Journal Sentinel compares the results of 46 baseball players named in the report in the first two seasons they have allegedly used steroids, with some surprising results.
Of the nineteen pitchers named in the report, fourteen had better than career average performances in the first year after the suspected first-use date. Now that is pretty significant if we assume that the chances of a better than average year is 50-50, like a coin flip. If you flip a coin nineteen times, the probability you get 14 or more heads is about 3% or one in thirty. For the hitters, a similar effect occurred; nineteen improved out of twenty-seven players. As a coin flip simulation, that is even less likely. The probability is about 2 ½ % or about one in forty chances.
Now those results would make a great case for steroid use, but when we look at the second year, we find that only nine of the nineteen pitchers were above their career average, and only seventeen of the twenty-seven hitters were above average. The results for hitters is still positive, with about a 12% probability of chance, which is definitely not a strong statistical significance; but for the pitchers, less than half improved on their career average.
One world-class statistician, Professor Paul Velleman of Cornell, suggests that the result could just be the well-known placebo effect. Hitters and pitchers performed better in their first year because they assumed that steroids would make them perform better.
I wrote recently about Pete Rose, who was banned for life for gambling. I don’t remember any suggestion that he ever was suspected of trying to cheat as part of that. He just loved to gamble. Now we have a bunch of players who actually cheated in many people’s eyes and some of them will be going to the hall of fame. So what is the message we send to young athletes with this result? If we are confused, I bet they are too.