Ellen,

Below are more specific instructions. I hope this is clearer.


Q1:
I was just unclear about what you meant with "load your threshold subbrick
in the lower subbrick selector within the afni define function window, and
your amplitude in the
first subbrick selector"

Right now, I have been loading my Anatomy dataset with the MannWhitney test
as the functional overlay but I was not sure which subbrick of the mann
whitney test lay where. is there a specification for which subbrick I am
looking at, at a given time?

A1:
Load your mann_whitney test as your functional dataset in 'switch function'. According to the mann whitney.pdf help file, the output of this test should include the estimate of the treatment effect as the 'intensity' and the rank sum statistic as the 'threshold'. These should correspond to the 0 and 1 subbricks of your mann_whitney test, respectively.

Q2:
Also, "Slide the threshold slider until your p value displays .001", does
this mean the number next to the slider is the p-value and the number
directly below this (at the base of the threshold slider) is the actual
threshold or is it the other way around? becasuse the number at the base
does not reach below .3174, and I need a p-value of .05. Sorry, these are
silly questions, I'm just confused.

A2:
Go to 'define datamode'. Select subbrick 0 for the 'func' and subbrick 1 for 'thresh'. I think it is the default to load this way. Now look at your slider bar. The value on the left of the bar is your mann_whitney test statistic value). Below that, at the bottom, is the p value for the statistic. If you are limited to .3174, use the p value selector to be more sensitive (it is the drop down bar right under the p value itself). Set it to 1 or 2, or whatever allows you to get a p value of .001. Now look at your mann whitney statistic at the left of your slider. Enter that as your -1thresh in 3dmerge when clustering.


Q3:
Finally, in the alpha sim test results, the alpha value is the actual
result of the test, so to find out the cluster volume for the p-value .05,
would I need to look at the alpha value of .0500 and the corresponding
cluster?

A3:
No, use 3dmerge to cluster and threshold your results by specifying the minimum cluster size and mann_whitney statistic (-1thresh) associated with your p value from alpha sim (.001). Any clusters that survive have a corrected p value of .05.


Christine Smith