Massimo Stella PhD

The Exploratory Graph Model is now LIVE! TL;DR: • The paper introduces the Exploratory Graph Model (EGM), a new mathematical framework for analyzing psychological measurements using networks instead of traditional latent variable models • Key innovations: - Treats psychological dimensions as emerging from direct interactions between observed variables - Does not assume underlying latent variables - Provides a formal mathematical basis for community structures in networks - Allows for both exploratory and confirmatory network modeling • The researchers conducted simulation studies comparing EGM to traditional factor analysis (EFA): - Parameter estimates were most accurate when the analysis method matched the data generation method - Traditional fit measures could distinguish between EGM and factor models better than chance - Model detection accuracy improved when multiple fit measures agreed - Loading size and correlation strength affected which model fit better • Three empirical examples were analyzed (Misinformation Susceptibility Test, Beck Depression Inventory, and a Narcissism scale): - EGM fit better for some applications - Results demonstrate how EGM can be applied to real psychological data • Main implications: - Provides a new way to conceptualize psychological measurement without latent variables - Opens up new possibilities for network-based psychological research - Complements existing network approaches in psychology - May help improve measurement accuracy in psychological assessment • The paper represents a significant advance in psychometric modeling by formalizing network approaches as an alternative to traditional latent variable methods The paper can be found here: osf.io/preprints/psyarxiv/zk… (the main new functions) can be found in our EGAnet package here:EGM andEGM.Compare cran.r-project.org/web/packa…

Are you interested in NLP/LLMs? You might find our dynamic Exploratory Graph analysis pretty useful (video animation summarizing our paper below). In a new paper, I'm combining the powers of LLMs with Dynamic EGA! Stay tuned! youtu.be/kUsvFEaXfhs?feature…