Dear Heather,
It is generally difficult to do a power analysis for
Neuroimaging because the power depends on a specified
alternate hypothesis and will change from voxel to voxel.
In our unit we use between 12 and 16 subjects for random-effect
analyses.
Our quantitative analyses of sensitivity and specificity have
been pursued in a Bayesian context and are summarised in:
Friston KJ, Glaser DE, Henson RN, Kiebel S, Phillips C, Ashburner J.
Classical and Bayesian inference in neuroimaging: applications.
Neuroimage. 2002 Jun;16(2):484-512.
Friston KJ, Penny W, Phillips C, Kiebel S, Hinton G, Ashburner J.
Classical and Bayesian inference in neuroimaging: theory.
Neuroimage. 2002 Jun;16(2):465-83.
I hope this helps - Karl
At 06:38 27/06/2005 -0700, you wrote:
>Dr. Friston:
>I recently read your Neuroimage 1999 paper entitled
>"How many subjects constitute a study?" In the paper
>you outline the number of subjects needed for a
>fixed-effects analysis and conjunction analysis. Also,
>it is suggested that a random-effects analysis needs
>more subjects than a fixed-effects. Although, the
>power analysis for the random-effects is not included
>in this paper. Have you published any papers outlining
>how many subjects for a random-effects given desired
>power levels etc?
>
>I am a graduate student at George Mason University,
>VA, and am applying for a research grant. I'll be
>conducting a random-effects analysis on four groups
>(those with mild cognitive impairment, early
>Alzheimer's Disease, Alzheimer's disease, and healthy
>volunteers). I have no preliminary data and would like
>to know the lowest subject number possible to have
>sufficient power.
>
>Thank you in advance for taking the time to answer my
>question. I truly appreciate it.
>
>Take care and good luck in your research endeavors,
>Heather Strasser
>
>
>
>____________________________________________________
>Yahoo! Sports
>Rekindle the Rivalries. Sign up for Fantasy Football
>http://football.fantasysports.yahoo.com
>
>
|