Week 7: Three Factor ANOVA

Question 1

What does a 3 factor ANOVA produce?

Answer 1

Main effect of IVa, IVb and IVc
Interaction between A and B (first order)
Interaction between A and C (first order)
Interaction between B and C(first order)
Interaction between A, B and C (second order interaction)

Question 2

How do you set up a 3 factor ANOVA in jamovi?

Answer 2

1x row for each participant

1x column for each IV and the DV

Question 3

How do you do a 3 factor ANOVA?

Answer 3

Linear models - General linear model
DV - DV
IVs in fixed factors box
Ask for the effect of partial eta squared and assumption of homogeniety tests

Question 4

How do you report the effects of a 3 factor ANOVA? (non-significant 3-way interaction)

Answer 4

Start with the most complicated
If there is a significant 2-way interaction - need to do tests of simple main effects with bonferonni adjustments to tell you where the differences are (may need to follow with post hoc if more than 2 levels)

Discuss main effects - report if significant or not

Question 5

What should you do if a main effect is significant but there this variable is not included in the significant 2 way interaction

Answer 5

You can explore this variable further using tukey

Question 6

What does a signifcant interaction mean?

Answer 6

That the effect of one IV is inconsistent across levels of the other IV

Question 7

How do you report the effects of a 3 factor ANOVA? (significant 3-way interaction)

Answer 7

If the 3-way interaction is significant you need to: briefly report main effects and 2 way interactions

Report the significant 3 way interaction..

Then have two options:
1. Tests of simple interactions: filter to one level of the variable you are interested in, then click linear and general linear model and so it separately for each level
If you repeat twice, the alpha level should be divided by 2 for the new alpha (bonferonni)
This can be followed with simple effects test (bonferonni adjustment again) - look at cohens d for magnitiude and cause

2. asking for tests of simple main effects
   - need bonferonni adjustments again
   - need to calculate cohens d for each