PPL 212 Lecture Notes - Lecture 13: Statistical Hypothesis Testing
Document Summary
Ppl 212 class notes: probabilities and natural frequencies 2/22/17. Tests are fallible, so we can"t always trust these results. But even highly accurate tests for rare events can be unreliable ex: A drug test always correctly identi es a drug user as a drug user: whenever s takes drugs and take the test, there is a positive result 100/100 test. The test also gives a false positive 1% of the time: whenever s doesn"t take drugs and takes the test, the test will be positive 1/100 times. Was it more likely caused by sam taking drugs or a false positive: most likely, yeah sam uses drugs. You can"t know this without knowing the prevalence of drug users. You need all 3 of these things: the prevalence of the phenomenon, the sensitivity (accuracy of test, the false positive rate, using natural frequencies. Since 1% of the population are drug users: 100/10000 are drug users, 9900/10000 or (99%) are not drug users.
