For example, say you suspect that some employees are on drugs. You decide to test all of them with a drug test that's 99% accurate. One employee tests positive, what's the chance that they were actually on drugs?
Most people would say 99%, but that's not correct. The answer is that you can't know without more information. Counter-intuitively, you first need to know the percentage of employees on drugs. For example, if you have 100 employees, 1 is on drugs, and 1 tests positive, then there is only a 50/50 chance that you caught the right person. If none are on drugs then you certainly caught an innocent person. On the other hand, if all of the employees are on drugs then the employee is certainly guilty, and only catching one is a statistically anomalous result.