Hi Michael,
This is indeed a tricky issue. It partly comes down to a conceptual distinction - should we treat RTs as having been caused by brain activity (in which case we might not want to covary them out, as they are part of the observed behaviour) or as a confound of longer "time on task" (in which case we would) ?
Here is another paper by Grinband and colleagues that might be useful:
http://www.ncbi.nlm.nih.gov/pubmed/18775784
Best
Steve
On 25 Sep 2011, at 10:35, Jonathan Peelle wrote:
> Dear Michael,
>
>> my design is built like this-
>> participants see stimuli which can be from 6 relevant conditions
>> from time to time a catch question appears
>> so in total i have 7 conditions- 6 experimental conditions and 1 irrelevant "catch" condition
>>
>> for each of the 6 relevant stimuli types (but not for the 7th condition), i am measuring the response times.
>> i need to covary this out since it differs significantly from one condition to another.
>>
>> how can i achieve this?
>
> This is a tricky issue. One way to do this might be to define all of
> your 6 event types as a single condition, and then add parametric
> modulators that code for the 6 conditions (e.g. how each condition
> differ from the mean) and response time. This will help remove the
> variance associated with response time from the error (i.e.
> unexplained bits) of your data. To contrast conditions, you could
> contrast across the parametric modulators coding for each condition.
> Though parametric modulators are serially orthogonalized, so the order
> in which you enter things may make a difference (if they are not
> completely independent).
>
> Also, perhaps more importantly, if your conditions differ
> significantly in response time, there isn't really a good way to
> "covary out" response time. This is discussed by this very helpful
> Miller & Chapman paper:
>
> Miller GA, Chapman JP (2001) Misunderstanding analysis of covariance.
> Journal of Abnormal Psychology 110:40-48.
>
> So, depending on your data, it may be that there's no easy solution to
> disentangling the effect of condition from the effect of response
> time. If others have more helpful suggestions, I'll be interested to
> hear them though.
>
> Good luck!
>
> Best regards,
> Jonathan
>
>
> --
> Dr. Jonathan Peelle
> Department of Neurology
> University of Pennsylvania
> 3 West Gates
> 3400 Spruce Street
> Philadelphia, PA 19104
> USA
> http://jonathanpeelle.net/
____________________________________
Stephen M. Fleming PhD
Wellcome Trust Postdoctoral Fellow
http://web.me.com/stephen_fleming/web/Welcome.html
|