Two-way mixed ANOVA FACTORIAL ANOVAs

Question 1

two-way mixed ANOVA

Answer 1

• participants take part in all levels of one IV, but only one level of other IV

Question 2

mixed ANOVA SPSS data input

Answer 2

• every p gets own row
• between-subjects factor you need a column for grouping variable (levels in same column)
• for within-subjects factor you need a column for each IV level

Question 3

main effects - between-subjects IV

Answer 3

• check Levene’s significance for all levels of IV (if p>.05 we can assume homogeneity) (no fix for factorial designs)
• SPSS output from ‘tests of between subjects effects’
• if significant and more than 2 levels, post hoc / if not, no need for post hoc

Question 4

main effects - within-subjects IV

Answer 4

• no need to check Mauchly’s if only 2 levels, check if more
• SPSS output from ‘tests of within subjects effects’
• if significant only post hoc if more than 2 levels

Question 5

main effects notes

Answer 5

• read primary IV output first, then secondary IV after

Question 6

interaction in mixed ANOVA

Answer 6

• does primary IV depend on secondary IV
• remind yourself which IV is within vs between
• for INTERACTION check ‘WITHIN subjects effects’
• for residual variance d.f. use error for main IV as no interaction error in ‘within-subjects effects’ is provided

Question 7

interaction error in mixed ANOVA

Answer 7

because there is a between-subjects element we cannot remove the individual differences from the interaction error term
… therefore we use ‘within’ subjects error and interaction

Question 8

partial eta squared

Answer 8

for interaction is also from within subjects effects
- between subjects from between subjects effects
- within subjects from within subjects effects

Question 9

profile plots and simple effects for interaction in mixed ANOVA

Answer 9

• if interaction is significant, check profile plot for directionality and conduct t-tests
• simple effects are used to compare the n levels of main iv separately for each level of secondary IV
• if main/primary IV is between, use independent t-tests BUT if main IV is within use paired
• remember to report Cohen’s d for each t-test even if not significant