Ged,
Treatment would be the nested factor within the two groups. The
second-level analysis tests for differences between groups in pre/post
post-activation as a random effects analysis, thereby implying that each
group is representative of a larger population.
I don't have a reference handy, but the principles can be derived from any
standard textbook on statistics.
Doug
At 09:22 PM 5/24/2006 +0100, Ged Ridgway wrote:
>Douglas Burman wrote:
>>Ivy,
>>One way to do this would be to enter both 'before' and 'after treatment'
>>data into a design matrix at the first level of analysis, then create an
>>'after-before' contrast to evaluate the difference for each individual.
>>At the 2nd level (random effects analysis), use a 2-sample t-test to
>>determine whether the 'after-before' difference for the group of AD
>>subjects is different than that observed for the control subjects.
>
>Hi Doug,
>
>Is this treating subject as a random effect and group as a fixed effect,
>or the other way around, or both random?
>
>As you can tell I don't know much about this, but it seems that group
>should be a fixed effect since Ivy is asking specifically about AD vs
>control and not about a sample from a population of various possible diseases.
>
>I suppose "group" is complicated as it's not really a factor in the sense
>that you have subject A at both before and after, but they will only be in
>one group not both. I believe this kind of design is described as having
>subject as a "nested" factor within group (my interpretation of McCulloch
>and Searle's GLMM ch1), but I probably don't know what I'm talking about...
>
>It seems like this kind of design/factor will occur commonly in important
>questions like disease vs control, but most of the random effects stuff I
>can recall seeing is focussed on fMRI with subject as a random effect
>after getting summary stats from the first level; I don't think I've seen
>a clear discussion of how to analyse this more complicated mixed effects
>model. Can you/anyone recommend any refs?
>
>Many thanks,
>
>Ged.
Dept. of Communication Sciences & Disorders
Northwestern University
2240 Campus Drive
Frances Searle Building, Room 2-356
Evanston, IL 60208
phone 847-467-1549
fax 847-491-4975
email: [log in to unmask]
|