Abstract: | Focuses on important issues which have been neglected in recent papers concerning the problem of using analysis of variance with an unbalanced design. The concept of population cell weights is introduced and used to define parameters for various models. Variance components associated with these parameters are also defined. The use of regression analysis is demonstrated, employing population weights to obtain the parameters of the different models. Finally, sample estimates of both parameters and components are introduced. Emphasis is given to the important issue of distinguishing between population parameters and sample statistics. Questions of the relations between population and sample are discussed. (20 ref) (PsycINFO Database Record (c) 2010 APA, all rights reserved) |