Hello Stuart,
>It was argued, however, that there was no need to to do that because the
>data was heading for a second level analysis and only the number of
>subjects will ultimately contribute to the df and not the number of
>scans.
>
Yes, this is true. the second level inference (for one sample t test) will
have a DF of n-1 (n being the number of subjects)
Very best wishes
Paul
-----------------------------------------------------------------
Paul Fletcher,
Box 189,
Research Department of Psychiatry,
University of Cambridge,
Addenbrooke's Hospital,
Hills Road,
Cambridge,
UK
CB2 2QQ
Tel 01223 336 988
Fax 01223 336 581
|