Baseball Musings
Baseball Musings
March 16, 2005
Population Testing

There are problems testing for steroids that I've discussed in other posts, mostly in letting a lot of true positives go by in order to reduce false positives. One of the questions I'm sure Congress will try to address with hearings is how widespread the use of steroids is in major league baseball. And yet, I'm sure we have the answer to that question in the data that's already been gathered.

We're looking at that data from the wrong perspective. We see numbers quoted indicating 5-7% of players tested positive in 2003. But we also know the test produces a lot of false negatives. The test for testosterone looks at the ratio of that hormone to epitestosterone. The normal ratio is 1:1, but 6:1 is considered positive. You can be doping, have a level of 4:1, and be considered clean. In fact, the way to beat the test is to keep your level below six. You still get the boost, but you don't get caught.

So in 2003 and 2004, I assume all players had their T/E ratio measured. I would expect those ratios to be normally distributed; therefore we can describe the distribution by calculating the mean and standard deviation.

Meanwhile, someone somewhere probably knows the mean and standard deviation for the general male population. Given that data, it's easy to calculate the probability of T/E ratios in baseball players being the same as the general population! Low probabilities would indicate widespread use of testosterone.

You see, one player having a T/E ratio of 3:1 is just random chance. One hundred players having a 3:1 ratio is a trend. It's the Bill James idea of evaluting defense by starting with the team and working back to the players. Study the population. That will tell you if there is widespread abuse. And it will tell you at what level the abuse is occuring. If the distribution shows abuse at 3:1 then set the positive level there.

Singling out individuals with tests full on uncertainty does not fix the problem. The tests are not conclusive enough. A study of the population is more likely to be conclusive in showing whether or not abuse is widespread, and at what levels the abuse is occuring. We would finally know the extent of doping.


Baseball Musings is holding a pledge drive during March. Click here for details.


Posted by David Pinto at 10:52 AM | Cheating | TrackBack (0)
Comments

Absolutely right. This whole thing is pretty idiotic, just an attempt to dupe the public. Why don't they just declare a "war on steroids," then we'd all know for sure that it was a big joke.

Posted by: Adam at March 16, 2005 12:14 PM

Yeah, it is a joke. And the 5-7% that were caught were only caught for the things that they were testing for. How about the dozens of things that WEREN'T tested? They need to test for everything, or at least as much stuff as they can realistically test.

Posted by: sabernar at March 16, 2005 12:41 PM

But David-- if they did that they'd only find out how significant the problem is; they still wouldn't know who to PUNISH. :-)

Posted by: john swinney at March 16, 2005 09:14 PM

Great points, David.

Posted by: Jason at March 17, 2005 08:17 AM
Post a comment









Remember personal info?