Michael wrote:
> Hello,
>
> I have a question about how to interpret the results from
> 3dANOVA2. I running a repeated measures anova (i.e., 3dANOVA
> -fa Drugs \
> -amean 1 D1 \
> -amean 2 D2 \
> -amean 3 D3 \
> -amean 4 D4 \
> -adiff 1 2 D1-D2 \
> -adiff 1 3 D1-D3 \
> -adiff 1 4 D1-D4 \
> -adiff 2 3 D2-D3 \
> -adiff 2 4 D2-D4 \
> -adiff 3 4 D3-D4 \
> -bucket ANOVA_results
>
> My question is about the -fa option, and how that relates to
> the results of the 6 -adiff options.
>
> 1) Does the -fa option for Drugs test the whether or not there
> is alot at least one difference among the 4 level means?
The -fa option is the F-test on the main effect of Drugs. i.e. is there significant variance across the 4 levels?
> 2) If so, do the -adiff options represent s form of a post-hoc
> test (similar to a Tukey's HSD test), to see which levels
> significantly differ based on the main effect for Factor A
These are a priori planned contrasts on the different levels of the factor Drug: they are not corrected for multiple planned comparisons. However, they could be used as posthoc tests for differences between levels, if the main effect of Drug is significant.
> 3) Would the activation maps for the main effect for Factor A
> correspond to the adiff activation maps? It appears that they
> don't when you set the set the p values (say p = 0.05) equal
> for the adiff T-statistic and the -fa F-statistic. I see
> activated areas in the contrast activation map that show up at
> p = 0.05, that don't show up in the -fa F-statistic activation
> map at p = 0.05.
They are two different tests: The main effect tests whether there are significant variance across all 4 levels of drug. A given difference tests only whether the two levels of drug differ. It is possible that two levels of drug differ significantly, but that the main effect over 4 levels is not significant, for example if 3 of the levels of drug are very similar.
>
> Your help would greatly be appreciated.
>
> Thanks