Regions of Significant Criterion Differences in Aptitude-Treatment-Interaction Research

The investigation of learning performance as an interactive function of individual differences of the learners and alternative modes of instruction is a timely topic in the area of instructional research. This research area is generally referred to as Aptitude-Treatment-Interaction (ATI) (Bracht, 1969; Cahen, 1969; Cronbach, 1967; Cronbach, and Snow, 1969; Kropp, Nelson, and King, 1967). The Johnson-Neyman technique (Johnson and Neyman, 1936) is a statistical tool that is particularly useful for studies designed to investigate aptitude-treatment-interactions (see Aiken, 1968; Carroll and Wilson, 1969;3 Johnson and Fay, 1950, for discussions of the application of the Johnson-Neyman method). As originally presented, the Johnson-Neyman technique defines regions in the area defined by two predictor variables where the expected values of a criterion variable for two groups are significantly different. (It is also applicable in the case of one predictor variable.) Generalizations of the procedure to more predictors, more groups, and more criteria have been provided by Abelson (1953), Johnson and Hoyt (1947), and Potthoff (1964).