On the meaning of statistical interactions

The problems which can arise in predicting the presence or absence of statistical interactions when a theoretical model is not set into correspondence with the experimental design model are considered in the context of a specific study of developmental differences in distractibility. It is concluded that intuitive or incomplete theoretical analyses often lead to incorrect prediction concerning statistical interactions and to spurious or ambiguous conclusions based on the confirmation or disconfirmation of those predictions and that the analysis of the relationship between a theoretical model and the experimental design model will offer considerable protection against such undesirable outcomes.