Speaker

Dr. Denny Borsboom
Faculty of Social and Behavioural Sciences
University of Amsterdam

Denny Borsboom is Professor of Psychological Methods and Director of the Social and Behavioural Data Science Centre at the University of Amsterdam. His work has focused on conceptual analyses of psychometric models, the measurement problem in psychology, and on substantive psychological research in a number of domains, including intelligence, personality, and psychopathology. He is the founder of the Psychosystems Project, which is dedicated to the development of network approaches to psychometric problems and the construction of statistical models; in addition, he directs the Theory Methods Lab, which focuses on the development of methodological tools designed to facilitate the development of formalized theories in psychology. He publishes extensively on topics such as validity, network psychometrics, psychometric theory and has received wide recognition for his work, including President of the Psychometric Society in 2025.

Title

Theory Construction Methodology

Abstract

Scientific theory is the most successful instrument for understanding, predicting, and controlling the world that humans have so far produced. Despite this fact, we do not explicitly teach, practice, or systematize theory construction. Instead, common methodological approaches treat the development of theory as being outside of the bounds of scientific method. In this talk, I will argue that this viewpoint creates an incomplete picture of both science and methodology: many aspects of theory formation are in fact amenable to methodologically systematic approaches and can be assisted by modern modeling techniques. I will outline one of these approaches: Theory Construction Methodology (TCM). TCM is a practical sequence of steps that can be followed to construct explanatory theories. First, the theorist identifies a domain of empirical phenomena that the theory should accommodate. Second, through analogical abduction and related approaches, the theorist constructs a proto-theory: a loose set of suggestive principles that are suspected to hold explanatory force with respect to the empirical phenomena identified in the previous step. Third, the proto-theory is used to construct a formal model. The formal model can be either a set of model equations that encode explanatory principles, a graphicalrepresentation of causal relationships, or a (possibly agent-based) simulation model. Fourth, the theorist develops the explanatory force of the model. This is done by representing the empirical phenomena under investigation in terms of the formal model chosen and researching whether the model indeed reproduces these phenomena accurately. Fifth, the theorist studies the overall adequacy of the theory. In the past years, we have developed various methodological techniques that can be used to assist the researcher in each of these steps, and we are currently working on AI-based methods to support the TCM methodology. I will discuss the status of this research program and outline future directions.