Model-Building Strategies in Response Surface Analysis: A General Framework for Investigating Congruence Processes
Résumé
Research into congruence processes investigates how the combinatory effects of two variables predict variations in an outcome variable. Response surface analysis (RSA) offers an optimal approach to identify various types of congruence processes, by estimating polynomial models of the congruence and interpreting the resulting response surface. To date, the capabilities of RSA are limited to the analysis of few congruence processes (10 families), and the method lacks a general framework for generating and testing hypotheses about congruence processes. The present research addresses these limitations gap. Conceptually, we develop a systematic model-building framework based on components of the cubic function, rationales for their combination and an expanded typology of 37 polynomial families encompassing over 2,200 hypotheses (94% new). Procedurally, we propose a three-step identification strategy that guides researchers in selecting and validating best-fitting hypotheses using multiple criteria. We illustrate the functioning of this framework by means of simulation data and the R package RSAtools.
| Origine | Fichiers produits par l'(les) auteur(s) |
|---|---|
| Licence |