Understanding Hypothesis Testing.

Simple statistical procedures, like testing for differences in the means or proportions of groups, become complicated when many groups are involved. This setting arises naturally and frequently in science: a study may include multiple treatment arms, important sub-groups, interim analyses, or multiple outcomes. We must correct for the number of statistical tests-called multiple comparisons-or face increased risk of invalid results. To truly grasp the problem, we must delve into the specifics of hypothesis testing.