E.4 Summary

This introduction to multiple comparisons illustrates just some of the many methods available for conducting multiple comparisons for hypothesis testing. Many of the same methods can be used when addressing the multiplicity problem caused by simultaneous analysis of multiple response variables, such as that which occurs with multiple primary endpoints or study objectives. Adjustments based on methods discussed here can also be implemented for simultaneous confidence intervals.

The Tukey-Kramer and Dunnett's methods presented in this section using one-way ANOVA (Chapter 6) are widely used when response data have a normal distribution. These methods can also be used with more complex models, such as two-way ANOVA (Chapter 7), repeated measures ANOVA (Chapter 8), and analysis of covariance (Chapter 11), although these methods can be somewhat more conservative with these models.