When are Post Hoc tests typically performed after an ANOVA?

When are Post Hoc tests typically performed after an ANOVA?

• A. After obtaining a significant ANOVA result.
• B. Before running ANOVA.
• C. Only when variances are unequal.
• D. Only with more than three groups.

Answer: (A)

An ANOVA tells you whether all group means are equal, but it doesn’t specify which pairs differ. Post hoc tests are used after a significant ANOVA result to compare specific group means while controlling for the increased chance of false positives that comes from making multiple comparisons. So the typical sequence is: run the ANOVA, if the F-test is significant, then perform post hoc tests to identify the particular group differences. If the ANOVA isn’t significant, there isn’t enough evidence to justify digging into which means differ. The other options don’t fit as the standard trigger. Post hoc tests aren’t done before the ANOVA, since there’s no evidence yet of any differences. They aren’t conditional only on variances being unequal—there are methods for both equal and unequal variances. And while you can have more than two groups, post hoc procedures are invoked after a significant result regardless of the exact number of groups (including three or more).

An ANOVA tells you whether all group means are equal, but it doesn’t specify which pairs differ. Post hoc tests are used after a significant ANOVA result to compare specific group means while controlling for the increased chance of false positives that comes from making multiple comparisons. So the typical sequence is: run the ANOVA, if the F-test is significant, then perform post hoc tests to identify the particular group differences. If the ANOVA isn’t significant, there isn’t enough evidence to justify digging into which means differ.

The other options don’t fit as the standard trigger. Post hoc tests aren’t done before the ANOVA, since there’s no evidence yet of any differences. They aren’t conditional only on variances being unequal—there are methods for both equal and unequal variances. And while you can have more than two groups, post hoc procedures are invoked after a significant result regardless of the exact number of groups (including three or more).