ABSTRACT
Starting from the notion that organismic or subject variables (like age and sex of Ss) may not be interpreted conclusively in factorial designs, it is proposed to replace the traditional two-variable paradigm (involving dependent and independent variables only) by a three-variable paradigm where independent variables are subdivided into randomized treatment variables and nonrandomizable subject variables. In terms of the three-variable paradigm, ANOVA models are specified in such a way as to allow for conclusive evaluation and interpretation of factorial designs, including subject variables. The numerical procedure is exemplified by an example from Ivo Kohler's perceptual research.
