Analysis of co-variance; ANCOVA (L11)

Question 1

When do we run an ANCOVA?

Answer 1

To test for differences between means when we know that an extraneous variable affects the outcome variable; used to control known confounding variables.

Question 2

What is a covariate?

Answer 2

The extraneous predictor being controlled by the ANCOVA.

Question 3

What are the advantages of an ANCOVA?

Answer 3

1) Reduces error variance - through explaining some of th unexplained variance (SSr)
2) Greater experimental control - through controlling known confounds (therefore gain greater insight into effect of predictor variable).

Question 4

How do you partition variance?

Answer 4

SSr is split into covariate and new SSR.

Question 5

Is ANCOVA an extension of the GLM?

Answer 5

Yes! Covariate is added to regression model first.

Question 6

What is the extra assumption of ANCOVA?

Answer 6

Homogeneity of regression slopes

- Relationship of covariate, IV and covariate*IV all have same direction of relationship