A way to increase statistical power. Variability due to factors that are not controlled or not known about is called noise. Noise can be reduced by (1) controlling conditions in a study to limit variability and (2) obtaining information about factors that may affect results and modelling these effects.
Used on pages 106, 110