From Latent Variables to Composites: Rethinking Construct Modeling in SEM

Povzetek

Formative constructs are widely used across business and social science research, yet their conceptualization, modeling, and validation remain challenging. This seminar revisits the foundations of formative construct modeling, clarifying the distinction between formative constructs and the formative measurement of latent variables. We will discuss how recent methodological developments have changed the way these constructs can be incorporated into Covariance-based Structural Equation Models (CB-SEM). Particular attention will be given to recent contributions by Jörg Henseler and co-authors, including the Henseler-Ogasawara (H-O) specification, which provides a flexible framework for modeling composites within SEM. Building on these developments, the seminar will compare covariance- and component-based approaches to SEM and discuss practical implications for model specification, estimation, and validation. The aim is to provide participants with an updated perspective on formative constructs and practical guidance for making theoretically and methodologically sound modeling choices in applied research.

O predavateljici

Dr. Laura Trinchera je profesorica statistike in podatkovne znanosti na poslovni šoli NEOMA Business School v Franciji, kjer vodi področje odličnosti Umetna inteligenca, podatkovna znanost in poslovanje (AI, Data Science & Business). Diplomirala je iz poslovnih ved in ekonomije (2004) ter doktorirala iz statistike (2008) na Univerzi Federico II v Neaplju v Italiji. Njeno raziskovalno delo se osredotoča na presečišče podatkovne znanosti in psihometrije, pri čemer je posebej strokovno usposobljena na področjih modeliranja strukturnih enačb, metod delnih najmanjših kvadratov (PLS), algoritmov za klasifikacijo ter statističnega učenja. Svoje raziskave je objavila v številnih mednarodno priznanih znanstvenih revijah, med drugim v Structural Equation Modeling: A Multidisciplinary Journal, International Journal of Production Economics, Journal of Organizational Behavior, Food Quality and Preference, Computational Statistics & Data Analysis, International Journal of Information Management in Management Decision. Je tudi sourednica knjige Handbook of Partial Least Squares: Concepts, Methods and Applications. Kot gostujoča raziskovalka je delovala na več uglednih mednarodnih ustanovah, med drugim na University of California, Santa Barbara, University of Michigan, Ann Arbor, University of Hamburg, Karlovi univerzi v Pragi in HEC Paris. Poleg tega je bila zunanja predavateljica na ESSEC Business School, Sciences Po Paris in Sorbonne University Abu Dhabi.

Gostitelj na Ekonomski fakulteti je red. prof. dr. Matej Černe.

Veselimo se vaše udeležbe na seminarju.