Hi Ivy,
I would tend to agree with Doug. Trials in which the subject does not
respond may be fundamentally different from trials in which the subject does
respond, so modeling them with separate regressors makes sense to me.
Hope this helps,
Daniel
-----Original Message-----
From: SPM (Statistical Parametric Mapping) [mailto:[log in to unmask]] On
Behalf Of Douglas Burman
Sent: Wednesday, May 17, 2006 12:08 AM
To: [log in to unmask]
Subject: Re: [SPM] parametric modulation and response times
Ivy,
My suggestion would be to model the group of trials without a reaction time
as a separate condition
of no interest. When we have missing response time data in our study, often
it is because subjects
are too slow to respond before the end of the trial -- but sometimes the
subjects just simply
aren't paying attention (perhaps from dozing off). In any case, if you
don't know what the
subjects are doing on a trial, it is really risky to make assumptions about
their behavior in a model.
Doug Burman
==============Original message text===============
On Tue, 16 May 2006 12:38:28 pm CDT Ivy Estabrooke wrote:
Hello:
I am trying to use response times as a parametric modulator in analyzing my
event-related study. The study has two phases, first an implicit encoding
task which requires an overt resposne (living vs. non-living), and the
second is a recognition task (old vs. new) However, for some scans (about
15%) the subject did not respond and therefore I have no response time.
Has anyone encountered this problem? I am debating using the mean response
time for the missing values or the maximum response time for the subject?
Another thought is to use the mean maximum response time of the group?
Any thoughts?
Any help would be greatly appreciated!
Ivy
===========End of original message text===========
|