vbpm
Variational Bayes estimation for a family of psychometric measurement models. Two models are provided. Variational Bayes factor analysis (vbfa) is a regularized partially confirmatory factor model spanning the confirmatory-exploratory continuum via spike-and-slab priors on the loadings (Chen, Guo, Zhang, and Pan, 2021 <doi:10.1037/met0000293>; Chen, 2023 <doi:10.3758/s13428-022-01884-7>; Jin and Chen, 2025 <doi:10.1080/10705511.2024.2432612>), with an optional dynamic (warm-started) regularization path, an orthogonal bifactor parameterization, and optional sparse residual (local dependence) estimation through a graphical spike-and-slab prior solved by QUIC (Jin, Chen, Yan, and Zhang, 2026 <doi:10.31234/osf.io/dehtv_v2>). Regularized MIMIC (vbmimic) extends this to multiple-indicators multiple-causes models, placing spike-and-slab priors on both the measurement and the structural part (Jin and Chen, 2025 <doi:10.1080/00273171.2025.2483253>). Companion tools compute SEM-like fit statistics, and sweep a factor-count window to report candidate fit, criterion, and between-candidate loading-correspondence measurements without selecting a count (Chen and Jin, 2026 <doi:10.48550/arXiv.2607.07159>). Data generators for either model family are also provided.
- A partially confirmatory approach to scale development with the Bayesian Lasso.
- Fully and partially exploratory factor analysis with bi-level Bayesian regularization
- Regularized Variational Approximation for Partially Confirmatory Factor Analysis
- Regularized Variational Bayesian Approximations for Variable Selection in Extended Multiple-Indicators Multiple-Causes Models
- Sparse Residual Estimation in Partially Confirmatory Factor Analysis
This page was built for software: vbpm
