Which statement correctly describes the MSE from ANOVA and its use in pairwise t-tests?

MSE is the total sample variance ignoring group membership and is used only when groups have very different variances; otherwise you use the smaller of the group variances for pooling.

MSE is the mean squared error (within-group variance) pooled across groups and is used as the common variance estimate in pairwise t-tests because it reflects typical variation of observations around their group means.

MSE is the mean squared estimate of the grand mean's error and should be replaced by the median absolute deviation in all Bonferroni calculations because medians are robust to outliers.

MSE is the between-group mean square that measures differences among group means and is therefore used in pairwise tests to inflate the denominator and discourage rejections.