 |
 |
Dear Andrew, There isn't an issue per se with different group sizes, but that usually requires a correction of the statistical method as the variances are likely to be different between groups (larger in the smaller group, just because of the N). Usually, the basic two sample t test used in SPM assumes unequal variance to start with, which although it entails a bit less power, as it reduces the degrees of freedom, is more robust. So, in short, I wouldn't worry too much about it, although the more similar they are, the better as the assumption of sphericity is "less" broken. Best, Joao On 19 January 2013 02:51, Andrew Martin <[log in to unmask]> wrote: > Dear all > Is there any support for both target and control size being the same size > (approximately) in fMRI studies. If one group was 15 and the other 30 would > this pose any statistical problems (power to detect an effect in both groups > etc)? > Thanks > Andrew