> > Is this result reliable? No. chisq.test is a non parametric goodness-of-fit test for *counts*. In my opinion you need a one sample t.test.