Overcoming the problems of variable interaction: Notes regarding empirical and conceptual issues in innovation theory building

Innovation research has been characterized by findings which are either considered unstable or inconsistent. This article addresses some of the methodological issues that arise when one attempts to test interactive models of innovation, describes how one particular algorithm — AID-3 — can be used to refine the use of general linear models, and proposes a research strategy based on a building mode of innovation theory. As such it functions as a “how to” for those interested in designing innovation studies and also as a theoretical piece aimed at re-organizing the goals and outputs of innovation research.