Paper on causal mediation analysis with latent variables accepted for publication in Psychometrika

Psychometrika accepted our paper on latent variable modeling with causal mediator models when unobserved confounding is present in the data. We provide evidence that the new g-estimation based estimator is robust to confounding and highlight finite sample size properties. We posted a preprint on arxiv.org

Morelli, S., Faleh, R., & Brandt, H. (2025). RAPSEM: Identifying latent mediators without sequential ignorability via a rank-preserving structural equation model. arXiv preprint, arXiv:2509.23935. Accepted by Psychometrika. Github