--> Why don't statistical experts check the validity of commonly used imaging software (SPM, FSL, AFNI)?

they do -- actually doing what people recenty(ish) started doing, ie using low cluster determining threshold is known to be wrong since 1994 as described in the seminal Karl Friston's paper (as pointed out by Guillaume). Being quite cheeky here, I answer back 'why don't neuroscientists who employ neuroimaging as the research tool go to courses to make sure they understand the tool they use. That would prevent having to do another article like the cluster failure one'. No point pointing the finger, we simply need to increase education and standards.

--> But on the other hand, can the neuroscientific significant be determined by purely statistics?

nope and you are right -- context/priors are important, effect size matter as much as above/below threshold and in that respect Bayesian stats are nicer. In all cases, if you expect a region a show an effect, even if not significant in your given experiment, you should plot parameter estimates in there, see look at how 'high' it is with regards to the threshold and discuss this -- see https://www.amstat.org/newsroom/pressreleases/P-ValueStatement.pdf


Cyril



從我的 iPhone 傳送

> Angstadt, Mike <[log in to unmask]> 於 2016年7月22日 下午9:49 寫道：
>
> The corrected version has a note in the help text mentioning that the issue has been corrected. If you grep the usage text for Eklund, it should show up something like this:
>
> $ 3dClustSim | grep Eklund
> ** It was pointed out to me (by Anders Eklund and Tom Nichols) that smoothing
> $
>
> Whereas the old version:
> $ 3dClustSim | grep Eklund
> $
>
> If I had to guess, I would suspect that your version may not be new, based on the 2011_12_12_1014 string there. I have a version like that here:
> $ afni -version
> Precompiled binary =Debian-x86_64-1-gnu: Mar  6 2016 (Version AFNI_2011_12_21_1014)
> $
>
> That is definitely not the corrected version.
>
> -Mike
>
>> -----Original Message-----
>> From: SPM (Statistical Parametric Mapping) [mailto:[log in to unmask]]
>> On Behalf Of Yang Hu
>> Sent: Friday, July 22, 2016 5:43 AM
>> To: [log in to unmask]
>> Subject: [SPM] How to make sure that I used the corrected version of
>> 3dClustSim
>>
>> Dear Experts,
>>    After I read the Cluster Failure paper and John's blog, I checked my
>> previous analyses with 3dClustSim. For example, my command is like this:
>>
>>    3dClustSim -mask mask.hdr -fwhmxyz 11.9 11.9 10.9 -niter 10000 -prefix
>> SPMmask -NN 2 -nodec -2side
>>
>>    When running the command, it showes sth like a version information in the
>> terminal:
>>    "++3dClustSim: AFNI version=FNI_2011_12_21_1014 (Mar 6 2016) [64 bit]
>> ..."
>>
>>    I just want to make sure that this is not the corrected version of
>> 3dClustSim, right? If so, what should the command in the updated
>> 3dClustSim be like?
>>
>>    Thanks for your help!
>>
>> Best,
>> Yang
> **********************************************************
> Electronic Mail is not secure, may not be read every day, and should not be used for urgent or sensitive issues