Hi all, I am designing a 3-way ANOVA, 4*2*3 between groups design, testing for main effects and the triple interaction. I am trying to calculate sample size using g*power and am wondering why I get different results when I run it as a 4*2*3 design vs a 4*3*2 design. To illustrate, here is what I ran first:

4*2*3 design

F tests

ANOVA: Fixed effects, special, main effects and interactions

A priori

Effect size = .25

Alpha = .05

Power = .8

Numerator df = 3 [(J-1) * (K-1) = (4-1) * (2-1) = 3]

Number of groups = 8 [4 levels A * 2 levels B = 8]

Output: Total sample size = 179 (per each AxB cell)

To look at for each level of C, multiply by levels of C: 3 * 179 = 537

So total sample size is 537.

However, if I run it as a 4*3*2 design (meaning, just swapping which variables I consider to be B & C)...

F tests

ANOVA: Fixed effects, special, main effects and interactions

A priori

Effect size = .25

Alpha = .05

Power = .8

Numerator df = 6 [(J-1) * (K-1) = (4-1) * (3-1) = 6]

Number of groups = 12 [4 levels A * 3 levels B = 12]

Output: Total sample size = 225 (per each AxB cell)

To look at for each level of C, multiply by levels of C: 2 * 225 = 450

So total sample size is 450.

Am I calculating something incorrectly? Appreciate any guidance!