can not figure out simple question, please help

hi everyone!

I have the following question: while running univariate GLM, I have encountered the situation when mean differences between groups are the same, but in one test they reach statistical significance, whereas in other not. Probably the reason is that in one case std. deviations are much larger....can you please explain why this happens?

here's the data:

3 groups, N=51,47,48 Participants respectively
I'm trying to analyze the effects of groups on 2 variables: PS and satisfaction

when I run analysis groups-->PS I have the following means and St.Dev:
M1=3.7 Std.Dev=1.11
M2=3.81 Std.Dev=1.57
M3=3.43 Std.Dev=1.15

M1=6.38 Std.Dev=.68
M2=6.65 Std.Dev=.57
M3=5.68 Std.Dev=1.02

I'm interested in differences between M1 and M3 in the first analysis(the difference is 0.3) and M1 and M2 in the second case(the difference is again 0.3, but in the first case pairwise comparisons show no significance, whereas in the second case p=0.08

can you please explain why this happens?

thanks a lot!


New Member
Hey Antouanetta.

Can you show us the specific models (in terms of predictors and response variables along with distribution assumptions) to make this a bit clearer?