B1491
Title: Generalized additive latent and mixed models
Authors: Oystein Sorensen - University of Oslo (Norway) [presenting]
Abstract: Generalized additive latent and mixed models (GALAMMs) are presented for analysis of clustered data with responses and latent variables depending smoothly on observed variables. A scalable maximum likelihood estimation algorithm is proposed, utilizing the Laplace approximation, sparse matrix computation, and automatic differentiation. Mixed response types, heteroscedasticity, and crossed random effects are naturally incorporated into the framework. The methods are implemented in the R package galamm, which is briefly demonstrated. The models developed were motivated by applications in cognitive neuroscience. By combining semiparametric estimation with latent variable modelling, GALAMMs allow a more realistic representation of how brain and cognition vary across the lifespan while simultaneously estimating latent traits from measured items. Simulation experiments suggest that model estimates are accurate even with moderate sample sizes.
