MULTIPLE TESTING: EVALUATING NBA PLAYERS

When multiple hypothesis tests are carried out at the same time, the guarantees provided by the hypothesis-testing framework no longer hold. In this video we explain why this is the case, illustrating the phenomenon with two real-world examples where we evaluate the performance of NBA players. We show that Bonferroni's correction allows us to still control false positives in multiple-testing scenarios, but also produces an increase in false negatives.