False Positives in NCAA/MLB Drug Testing
In order to maintain the integrity of NCAA athletics, they require that schools test a certain number of athletes each year. At Cornell, 10 football players and 8 athletes from other sports are tested. The consequences of failing a mandatory drug test are severe. Testing positive for any banned substance results in a full year suspension form participating in your sport. Although some players roll the dice and take banned substances to increase their performance and recovery, most athletes choose to abide by the rules. As an athlete myself, the possibility of a false-positive always worried me. Studying Bayes Theorem in lecture and the current homework assignment allowed me to break down the specific conditional probabilities behind these tests.
In the NCAA Drug-Testing Program, they acknowledge the possibility of a false-positive. They state, “No matter what screening methods may be used, including thin- layer chromatography and radioimmunoassay, thereis a finite probability of a false-positive result (i.e., the test is positive even though the student-athlete is actually ”clean”) (NCAA Drug Testing Program).” Thankfully, before any action is taken on a positive result, they screen these tests further by mass spectrometry to confirm the positive result and eliminate the finite probability of a false-positive.
In the MLB, they have another way of eliminating false-positives. They split the urine sample in two. If the first half of the sample tests positive, they test the second half of the sample (Gizmodo). Assuming that these two trials are independent events, the probability of two false positives is squared. If it was, say, 1/50, it now becomes 1/2500. However, it’s unlikely that the samples are independent because it’s usually another chemical in the urine that triggers the false-positive.
These probabilities are computed by these organizations in the same way that we analyze them in class. For example, in homework #6 with Dharmendra’s phone, we are calculating the probability of an event (her phone is off), given a certain observation. In drug testing, it’s the probability that the subject uses a drug given the observation of a positive test. Manipulating the Bayes Theorem into specific probabilities that we already know allows us to find this probability and further allows us to analyze the probability of false positives to give everyone a fair opportunity.
Resources:
- http://www.ncaa.org/sites/default/files/5.%20Drug%20Testing%20Program%20Book%202013-14.pdf
- http://gizmodo.com/5867293/the-myth-of-the-steroid-test-false-positive