ANOVA follow up tests & effect size

Question 1

What do linear contrasts allow you to do that pairwise tests don’t?

Answer 1

Compare groups of means

Question 2

How do simple effects treat sources of variance?

Answer 2

They combine main effect and interaction variance, then repartition it to explain the focal variable.

Question 3

What are three conditions which would require bonferroni corrections?

Answer 3

1. Predictions are post hoc
2. Performing five or more comparisons
3. Desire to be conservative (e.g. large data set or novel exploration)

Question 4

What are three main issues with over reliance on p values?

Answer 4

1. Arbitrary criterion
2. No practical significance
3. Increasing sample size will eventually lead to significance

Question 5

What two factors affect the difference between biased and unbiased effect size estimates?

Answer 5

1. Sample size

2. Error variance