Yes, this does make sense, thank you very much for explaining it so
thoroughly!
I am replying to the mailing list, including the attachment [which was
originally so big it was rejected], so that it will be archived for others!
With kind regards,
Johannes Keyser
Institute of Cognitive Science
University of Osnabrück
On 08/14/2012 06:56 AM, MCLAREN, Donald wrote:
> This is the expected behavior. Its likely that one condition has more
> variance than another, thus there is a difference.
>
> If you look at the average of all 4 conditions with a one-sample
> t-test compared to the average of all 4 conditions, you will see a
> similar value in both methods. If you look at the one-sample t-test of
> each condition, you will see that the ResMS is larger for one test
> than the others. This is one of the effects of pooled variance models
> - it combines the variance across all conditions, rather than only in
> the subset of observations of the contrast. A similar is effect is
> seen with a two-sample t-test with the contrast [1 0] compared to the
> one-sample t-test. Contrasts are the same, but the variances are
> difference.
>
> Does that make sense?
>
> Best Regards, Donald McLaren
> =================
> D.G. McLaren, Ph.D.
> Research Fellow, Department of Neurology, Massachusetts General Hospital and
> Harvard Medical School
> Postdoctoral Research Fellow, GRECC, Bedford VA
> Website: http://www.martinos.org/~mclaren
> 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, Aug 10, 2012 at 3:50 PM, Johannes Keyser<[log in to unmask]> wrote:
>> Yes indeed, they are identical - I created a difference image using imcalc,
>> 'i1-i2'.
>> Visual inspection shows the expected irregular, tiny deviations from zero,
>> max = 1.1921e-07, min = -1.1921e-07, sum = 8.2748e-06.
>>
>> Would you like anything in addition?
>> Thank you very much for taking an interest in this!
>>
>> Kind regards,
>> Johannes Keyser
>>
>>
>> On 08/10/2012 09:24 PM, MCLAREN, Donald wrote:
>>>
>>> Could you check the con_ images as well. They should be identical.
>>> I'll take a closer look over the weekend.
>>>
>>> Best Regards, Donald McLaren
>>> =================
>>> D.G. McLaren, Ph.D.
>>> Research Fellow, Department of Neurology, Massachusetts General Hospital
>>> and
>>> Harvard Medical School
>>> Postdoctoral Research Fellow, GRECC, Bedford VA
>>> Website: http://www.martinos.org/~mclaren
>>> 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, Aug 10, 2012 at 3:16 PM, Johannes Keyser<[log in to unmask]> wrote:
>>>>
>>>> Hello, and thanks again for the quick and highly useful reply!
>>>>
>>>> Your explanation does make a lot of sense to me, however there is a
>>>> splinter of doubt left:
>>>> Using error term 1, I end up with a much more conservative T-map (almost
>>>> factor 3!) than when using the analogous one-sample T-test.
>>>> The attachment shows this case, along with the GLM-flex code that
>>>> generated it (you might need to zoom in).
>>>>
>>>> Shouldn't the GLM-flex's contrast, with the correct error term, closely
>>>> correspond to T-test?
>>>> If that's a wrong assumption, what is my conceptual error?
>>>>
>>>> Thank you so much for your help, I greatly appreciate it!
>>>> Johannes Keyser
>>>> Institute of Cognitive Science
>>>> University of Osnabrück
>>>>
>>>
>>
>
|