Thomas E Nichols wrote:
>> I had a similar problem with a PET series. However I can get results
>> (ie it doesn't crash or give nonsensical results) if I just omit the
>> scan data & condition number. I tried your method too which gave
>> almost identical results (only tried on one particular study).
>
> I am very intrigued/curious/worried... the results should be *exactly*
> identical.
Glad to hear that they should be. Looking at one set of volume stats the
differences only differ in 3 stats at 3dp - so in my study set level p =
0.039 in one and 0.040 in the other
voxel level p(fwe-corr) differs in two (out of six), 0.132 cf 0.133 and
0.162 cf 0.163.
This particular study I have a full symmetric set but was experimenting
taking some random scans out - our new study is more ambitious and
unfortunately has more holes, but hopefully few enough to make it a more
powerful study.
I can send you the SPM.mat files if you want to find where the 3dp
rounding error comes in (doesn't sound v. important).
Joel
|