Dear Takeuchi,
On Fri, 11 Apr 2014 00:10:29 +0900, Hikaru Takeuchi <[log in to unmask]> wrote:
>Dear Prof. Christian Gaser,
>
>Thank you very much for the information.
>I was using the newest version (v66). I tried several things but the updating SPM8 to the newest versions, using SPM5 or SPM12b didn't work. Actually it didn't work even in the case of single regression analyses.
The TFCE toolbox should run best with SPM8 (have not yet prepared for SPM12, probably that runs).
To give you feedback about your design I need the columns of your design matrix and your given contrast.
Best,
Christian
>But for whatever the reasons, the older version (v56) did work to correlation analyses (at least began running).
>For the newest version (which didn't work to correlation analyses), I was using Matlab 2009a and SPM8 and windows 7, but the use of SPM12b or SPM5 did not help, either. Also the newest version of TFCE doesn't seem to be recognized by SPM5. Is this o.k.?
>Among the versions that work to correlation analyses, I will try the newest version, if it compromise the validity of the results, please let me know.
>Thank you very much.
>Sincerely,
>
>Takeuchi
>-----Original Message-----
>From: Christian Gaser [mailto:[log in to unmask]]
>Sent: Thursday, April 10, 2014 4:32 PM
>To: [log in to unmask]; SUBSCRIBE SPM Hikaru Takeuchi
>Cc: Christian Gaser
>Subject: Re: SPM's TFCE's application to multiple regressions and ANCOVAs
>
>Dear Takeuchi,
>
>first, please check that you have updated to the newest version.
>
>In order to apply permutation tests so-called exchangeability blocks in the design are permuted. However, the permutation can only be done for two blocks (or one covariate) in one contrast. Multiple regression should be possible, if your given contrast is defined for only one parameter. Of course you can additionally define nuisance parameters.
>For the design with the main effects of covariates of ANCOVAs I do need more information about the designs and the contrast you have defined. Again, only two parameters can be compared in one contrast.
>
>Best,
>
>Christian
>
>On Wed, 9 Apr 2014 14:00:05 +0100, Hikaru Takeuchi <[log in to unmask]> wrote:
>
>>Dear Prof. Christian Gaser and TFCE users,
>>
>>I would like to apply SPM’s TFCE toolbox which was developed by Prof. Christian Gaser (http://dbm.neuro.uni-jena.de/tfce/) to multiple regression analyses or main effects (such as 1/2 1/2) of covariates of ANCOVAs when there are several covariates in ACOVAs.
>>But when I tried it, the following error message is returned and the analysis cannot work.
>>
>>“Running 'Estimate TFCE'
>>Whitening of the data is not yet supported Failed 'Estimate TFCE'
>>エラー ==> cg_tfce_estimate at 146
>>Maximal two exchangeability blocks allowed.
>>In file "C:\Programs\MATLAB\R2009a\work\spm8\toolbox\TFCE\cg_tfce_estimate.m" (???), function "cg_tfce_estimate" at line 146.
>>The following modules did not run:
>>Failed: Estimate TFCE”
>>
>>
>>Does this mean the multiple regression analyses or ANCOVA’s covariates’ analyses cannot be run by TFCE of this toolbox (unlike FSL’s counterpart)?
>>I am sorry if it is an amateurish question. I could not find the manual or the detailed explanation of this toolbox.
>>Thanks in advance for help.
>>Sincerely,
>>
>>Takeuchi
|