I've got a map of F-values that I'd like to transform to z-scores. It seems as if I can use 3dcalc and fift_t2z(F,df1,df2) to do this. However, the maps come out with no negative values.
Trying to figure it out, I ran "cdf -t2z fift XX 1 10" with several values of XX ranging from 0.01 to 5. Same deal. Outside of AFNI, I calculated the CDF values for several values of F and t (using their distributions), then calculated the inverse CDF for a normal distribution with mean 0 and sdev 1 to get the corresponding z-score. Using this method, for t-values, AFNI and my hand calculations agree. e.g., for t(10)=2: CDF=0.96, and z-score=1.79. Using this method for F-values, and we don't agree. For F(1,10)=2: CDF=0.81 and my z-score=0.89 but AFNI's is coming out 1.32. For F(1,10)=0.1: CDF=.24 and my z-score = -0.7 whereas AFNI's is coming out 0.31.
Am I using the cdf functions incorrectly or otherwise getting something wrong?
Tom
