Post-hoc

Question 1

When do we need to do a post hoc test and why?

Answer 1

• We do it when ANOVA was found to be significant (must be 3 or more means)
• Post-hoc tells you where the differences in the test are. ANOVA just tells you a difference is somewhere.

Question 2

Type 1 Error Rate vs Familywise Error Rate

Answer 2

• Making comparision with more than two comparisons within a single set of data
  – Based on type 1 error (specified by alpha)
  – 5% with a single comparison
  – Multiple tests on the same data result in finding a difference when one isn’t there
• Familywise error rate: More comparisons, alpha>0.05
  – Probability of making at least one type 1 error in a set or “family” of statistical comparisions.

Question 3

Familywise error increases based on the number of ____

Answer 3

comparisons

Ex:
1 Comparison: 0.05
3 comparisons: 0.143

Question 4

Fischer’s LSD

Answer 4

• Most liberal post-hoc
• No control of familywise error rate (Type 1 error)
• Most powerful (easy to find differences)

Question 5

Bonferroni

Answer 5

• Most Conservative
• Controls for Type 1 Error
• Easiest to run for a t-test