back when tor was an enormous nonconformist, people would sometimes joke that he must be on steroids — especially when seen in the company of his less-than-beefy father. nevertheless, steroid use has always ranked near the bottom of my parental worry list.

simply put, few kids use steroids. according to 2008 monitoring the future data, about 1.5 percent of u.s. high school seniors reported steroid use in the past year. the number inched up from 1 percent in 1990 to a peak of about 2.5 percent in 2004, but has since fallen. that’s why i was a little surprised to hear that the state of texas had begun a program to routinely test every high school athlete for steroid use.

well, we now have results from the first wave of texas steroid tests. of the 10,117 high school athletes tested, steroid use was confirmed for a grand total of 4 (or .04 percent) students and suspected (based on a refusal to test or some other violation of testing rules) in 22 cases (or .22 percent). the remaining 10,091 athletes (or 99.74 percent) evidently tested clean.

proponents of the $6 million testing program point to the low violation rates as evidence of deterrence — see, the program is working! critics point to the inefficiency of spending so much money to catch so few users.

comparing the rates of self-reported use with the official test results, however, leads me to a different conclusion — the tests are ineffective as well as inefficient. that is, the kids are beating the tests. if texas mirrors the national rate of 1.5 percent and we assume that all 22 of the refusers were actually users, the UA tests are only catching about one in six of the estimated 150 past-year steroid users (10,117 *0.015=152).

i wouldn’t necessarily advise scrapping the program, but a small number of random tests would certainly bring a better bang for the buck. just as the internal revenue service strikes fear into all taxpayers with a handful of audits, a handful of steroid tests — administered randomly or to star performers such as state tournament participants — may be sufficient to deter use more generally. given these low base rates and the civil liberties issues involved, however, i’d be inclined to adopt a probable-cause or reasonable-suspicion standard for testing.

(a.p. link via auderey)