Hello Maren,
If you want your model to account for specific values of some measure you just need to add those values as an additional column in the design.
Many Regards
Matthew
> Hi Matthew,
>
> I need to come back to my initial question on ANOVAS using randomise. Where in the design matrix can I account for the actual values of the cognitive measure that is my factor with the 4 levels? My question was whether I can get an FA map that shows areas of FA change that depends on how my subjects perform on a cognitive measure (memory task with 1 factor and 4 levels). So I have response times from the memory task and need to bring them into my design matrix. The ANOVA description on the website only shows how to model mean effects of subjects, categorial effects, and the F tests, but not where to model the actual response times.
>
> Thanks, Maren
>
>
> -----Ursprüngliche Nachricht-----
> Von: "Maren Strenziok" <[log in to unmask]>
> Gesendet: Jun 10, 2011 10:14:28 PM
> An: [log in to unmask]
> Betreff: Re: [FSL] ANOVA with non-timeseries data
>
>> Thanks! Maren
>>
>>
>> -----Ursprüngliche Nachricht-----
>> Von: "Matthew Webster" <[log in to unmask]>
>> Gesendet: Jun 10, 2011 3:35:39 PM
>> An: [log in to unmask]
>> Betreff: Re: [FSL] ANOVA with non-timeseries data
>>
>>> Hello,
>>> Randomise outputs p-value files, these values will be "compatible" between versions provided that the p-values were not created via pre-4.1.8 usage of the -e option.
>>>
>>> Many Regards
>>>
>>> Matthew
>>>
>>>> Hi Matthew,
>>>>
>>>> thanks for your reply. I overlooked the exchangeability-blocks file issue on the website. I have an older Randomise version than implemented in FSL 4.1.8. I am concerned to upgrade to the newer FSL version at this point becasue I am in the middle of an analysis where we will add more subjects soon. I wonder whether the files created with the new version will be compatible with those that I would create with a newer version. Can you let me know how to go with that in general? I just started using FSL using only one version so far, but I have heard from colleagues that for other programs, like Freesurfer, it is highly recommended to refrain from using files that were created in different versions for the same analysis.
>>>>
>>>> Thanks, Maren
>>>>
>>>>
>>>> -----Ursprüngliche Nachricht-----
>>>> Von: "Matthew Webster" <[log in to unmask]>
>>>> Gesendet: May 26, 2011 11:56:54 AM
>>>> An: [log in to unmask]
>>>> Betreff: Re: [FSL] ANOVA with non-timeseries data
>>>>
>>>>> Hi Maren,
>>>>> The ANOVA: 1-factor 4-levels section of the FEAT documentation should help with this:
>>>>>
>>>>> http://www.fmrib.ox.ac.uk/fsl/feat5/detail.html#higher
>>>>>
>>>>> Note that you will also need to create an exchangeability-blocks file to respect the repeated-measures in your input data ( and make sure you are using the FSL 4.1.8 version of randomise, as previous versions had an issue with the exchangeability-block -e option ).
>>>>>
>>>>> Many Regards
>>>>>
>>>>> Matthew
>>>>>> Hi,
>>>>>>
>>>>>> I am trying to figure out whether I can run a 1-way repeated measures ANOVA with non-timeseries data (FA maps from DTI scans) using Randomise. I have data from a cognitive test with one factor and 4 levels. So far, I correlated difference measures between levels of that factor using the one-sample t-test with covariates, but I would like to know whether there are white matter regions that show altered FA values across the levels of the factor. Can anyone tell me whether it is possible to set up a design matrix for this or am I wasting my time?
>>>>>>
>>>>>> Thanks, Maren
>>>>>>
>>>> ___________________________________________________________
>>>> Schon gehört? WEB.DE hat einen genialen Phishing-Filter in die
>>>> Toolbar eingebaut! http://produkte.web.de/go/toolbar
>>>>
>> ___________________________________________________________
>> Schon gehört? WEB.DE hat einen genialen Phishing-Filter in die
>> Toolbar eingebaut! http://produkte.web.de/go/toolbar
> ___________________________________________________________
> Schon gehört? WEB.DE hat einen genialen Phishing-Filter in die
> Toolbar eingebaut! http://produkte.web.de/go/toolbar
>
|