Component-based approaches to modeling networks of relationships between latent variables have experienced increasing popularity over recent years and have become reference methods in the analysis of multi-block data. Among others, these approaches include Partial Least Squares Path Modeling but also to its most recent variants and alternatives, Generalized Structured Component Analysis, Regularized Generalized Canonical Correlation Analysis and THEME-SEER.
When the networks are made of directed relationships, the path modeling has a predictive purpose while undirected paths imply the study of correlations/covariances between blocks.
In such methods the estimation of factor scores (defined as composite variables) starting from blocks of manifest variables is a key to explore relations between blocks of manifest data as well as to make predictions.
This specialized group is focused on methodological and applied contributions in component-based approaches to predictive and exploratory path modeling and their related methods.
All researchers and users interested in the methods and the applications mentioned above are highly welcomed to join us.