Hi,
Somebody has run 2 group level analyses that are identical except in one analysis the "group" column from the EV tab was set to all 1's while in the other values were used to indicate which of 3 groups the subject was in (the model is a 1-way ANOVA with 3 levels modeled using cell means, estimated with OLS).
The statistics from both analyses are identical, which doesn't seem possible if the variance estimates differed between the two. I was wondering if that was because we're using OLS. Is the group column info only used in Flame analyses?
Otherwise could it be because there are only 5-6 subjects per group and FSL ignores grouping when the groups are small?
Thanks,
Jeanette
