On Tue, 22 Sep 2009 18:21:58 -0400, Bing Ye <[log in to unmask]> wrote: >Hi all: > >I am doing event-related analysis for my study. Subjects are doing a stroop test. there are two conditions reading the word and saying the color . I want to compare the incorrect responses when they say the color ink and correct responses when they read the word. but sometimes they only made a few mistakes when the say the color ink. My question is can i compare 3 incorrect responses to 64 corect responses. Would it yield the same or similar result when i compare 60 incorrect responses to 64 correct responses? It might have the same result, in the sense that the estimator is unbiased, but there's no way you can compare using only 3 incorrect responses---the sample size is far too small to get anything useful. There are various rules of thumb in fMRI (which is what I assume you're doing), one of which is how many trials of a given condition you need. > >Could anyone let me know? > >Thanks a bunch. > >bing >