ANCOVA

Question 1

Assumptions

Answer 1

Checking independence of the covarient
-look at test of between subjects effect and look at group sig value- we want this to be non significant- this means theres no sig difference between the DV across the IV e.g. – There are no significant difference between pre-treatment anxiety levels across the three conditions.

checking homogenity- look at test of between subjects effects- only interested in signifcance of the interaction with the group- again should be non significant

Question 2

Reporting the effect on ANCOVA

Answer 2

Look at test of between subject effects

report like this- A one-way ANCOVA showed that, after adjusting for pre-treatment anxiety, the main
effect of treatment group on post-treatment anxiety was statistically significant,
F (2, 26) = 5.53, p= .01, partial η2 = .299. This is a large effect size according to
Cohen (1988).

the 2 and 26 are df in group and error- 5.53 is the F value

Question 3

cohens 1988 effect sizes for cohens d

Answer 3

Effect size Partial η2
Small 0.02
Medium 0.13
Large 0.26

Question 4

post hoc tests

Answer 4

look at pairwise comparisons and estimates

In the estimates report the means as EMM=
and if it is signifcant on pairwise comaprisons= report each connection and if it is signifcant

e.g. Post-hoc Sidak comparisons of adjusted mean post treatment scores suggest that
participants in the new drug treatment group (EMM= 1.55) showed significantly
decreased anxiety scores in comparison to the control group (EMM= 3.51), p= .009.
There were no significant differences between post-treatment anxiety scores for the
control and exercise groups, or the exercise and new drug group, p≥ .284.