Equivalence Tests: Demonstrating that groups don't meaningfully differ

This seminar will introduce equivalence tests, which are statistical methods designed to use data to explicitly prove that the outcomes for two groups do not differ meaningfully. This contrasts the goal of most classical statistical methods (e.g., t-tests, chi-square tests, ANOVA, etc.), which aim to use data to prove that the groups in question actually differ from one another on the outcome measure. The seminar will explain what kinds of scientific questions equivalence tests can answer, how these tests work, and why they provide more credible evidence of equivalence than that offered by simply finding a non-significant effect with a classical method. It will offer practical advice on how and when to use equivalence tests in research studies. This workshop will focus on core concepts and simple applications of equivalence tests (e.g., comparing two groups continuous outcome measures).