Posted by B. Douglas Ward on March 16, 2001 at 14:53:58:
In Reply to: 3dANOVA? posted by Christine Hooker on March 16, 2001 at 14:07:27:
Christine Hooker:
The answer depends on whether you are analyzing time series data for an
individual subject, or are you analyzing data across subjects?
If you are analyzing time series data for an individual subject, you should
use program 3dDeconvolve. You can input multiple stimulus functions, which
represent the timing of the different experimental conditions. The program
will then use multiple linear regression to estimate the response to each of
the separate stimuli. To calculate contrasts between conditions, you can use
the -glt option. See the documentation in file 3dDeconvolve.ps.
For analysis of data across subjects, you can use one of the 3dANOVA programs.
But this assumes that you have already performed the time series analysis on
the individual subjects, so that the time component has been removed. See the
documentation in file 3dANOVA.ps. Also, some nonparametric procedures are
available as possible alternatives to ANOVA; see the documentation in file
Nonparametric.ps.
Doug Ward