Summary

Linear model analysis usually sets one factor level as the “Intercept,” estimates its mean, and then gives the differences between this intercept and the means of the other factor levels. The differences in means are accompanied by standard errors—SEDs. In simple situations, the SED is approximately 1.4 times the SEM.

$(SED=SEM\times \sqrt2)$.

We use standard errors to calculate confidence intervals. As long as the sample sizes are not very small, an approximate 95% CI is ±2 standard errors.