When constructing a CI for an ANOVA question, we take the estimate of the variance to be SSr / (n - k). This gives a different estimation than the normal formula of estimating a population variance using a sample (pg 23 of the Tables). Is the reason for this because we are assuming the same variances for the different populations, but assuming (potentially) different means? I guess the standard sample variance estimator is a version of the SSr / (n - k) formula, with the assumption that the various treatments also have the same mean. Is that correct? Thanks
