Dear Donald,
> Once you decide how your going to model it, I usually go with the rule that
> you need at least 5-10 observations per parameter.
It is not perfectly clear what is meant by "observations". Does it
refer to number of scans OR number of trials (OR number of subjects)?
Should we perhaps try to incorporate all of the above into a rule of
thumb using a simple equation?
thanks for your help,
Nu
On Fri, Sep 23, 2011 at 4:28 PM, MCLAREN, Donald
<[log in to unmask]> wrote:
> I wouldn't think of it as the maximum number of covariates, but the maximum
> number of parameters. The reason for this is that if you have one covariate
> (say age) and you want to use it in the model, you could: (a) model it as a
> single regressor across all groups, or (b) split it by groups into N
> regressors/parameters to allow the effect to vary by group.
>
> Once you decide how your going to model it, I usually go with the rule that
> you need at least 5-10 observations per parameter.
>
> Best Regards, Donald McLaren
> =================
> D.G. McLaren, Ph.D.
> Postdoctoral Research Fellow, GRECC, Bedford VA
> Research Fellow, Department of Neurology, Massachusetts General Hospital and
> Harvard Medical School
> Office: (773) 406-2464
> =====================
> This e-mail contains CONFIDENTIAL INFORMATION which may contain PROTECTED
> HEALTHCARE INFORMATION and may also be LEGALLY PRIVILEGED and which is
> intended only for the use of the individual or entity named above. If the
> reader of the e-mail is not the intended recipient or the employee or agent
> responsible for delivering it to the intended recipient, you are hereby
> notified that you are in possession of confidential and privileged
> information. Any unauthorized use, disclosure, copying or the taking of any
> action in reliance on the contents of this information is strictly
> prohibited and may be unlawful. If you have received this e-mail
> unintentionally, please immediately notify the sender via telephone at (773)
> 406-2464 or email.
>
>
>
> On Fri, Sep 23, 2011 at 10:01 AM, Cyril Pernet <[log in to unmask]>
> wrote:
>>
>> I Gordon
>>
>> don't know for others but for me as long as your matrix is not rank
>> deficient and p<n I don't see any problem ...
>> of course the more data points the more likely the effect is likely to be
>> 'true' ... now if your covariates are the same across subjects and only want
>> to regress this out (without comparisons) then it will fits across your 39
>> subjects ..
>>
>> Cyril
>>
>>
>> In a full factorial comparison of the functional imaging results of 3
>> groups (N=12,13,14) I would like to include a number of covariates of
>> interest. It has been suggested that in normal epidemiological studies a
>> rule of thumb suggests 1 covariate for each 20 samples. Even by removing
>> covariates that are strongly correlated with each other I have 3 remaining.
>> Is there a feeling in the SPM community that this 1/20 ratio is applicable
>> here also.
>>
>>
>>
>> Thanks,
>>
>>
>>
>> Gordon
>>
>>
>>
>> Gordon D. Waiter PhD CSci MIPEM CPhys MInstP
>> Aberdeen Biomedical Imaging Centre
>> Division of Applied Medicine
>> University of Aberdeen
>> Research MRI Centre
>> Lilian Sutton Building
>> Foresterhill
>> Aberdeen
>> AB25 2ZD
>>
>> Tel: +44 (0)1224 559725
>> Fax: +44 (0)1224 559718
>>
>> [log in to unmask]
>> www.abdn.ac.uk/ims/imaging
>>
>>
>>
>>
>>
>>
>> The University of Aberdeen is a charity registered in Scotland, No
>> SC013683.
>>
>>
>>
>> The University of Edinburgh is a charitable body, registered in
>> Scotland, with registration number SC005336.
>>
>
|