Intersectional approaches to cognitive aging: a practical guide to modeling heterogeneous trajectories with GLMM-trees.

Understanding cognitive aging through an intersectional lens requires analytical methods that can flexibly identify subgroup-specific trajectories across multiple social identities. However, traditional longitudinal models often require researchers to specify how these identities interact in advance, limiting their ability to uncover unexpected patterns of heterogeneity. This article presents a step-by-step guide to applying Generalized Linear Mixed Model Trees (GLMM-trees)-a flexible, recursive partitioning method that integrates mixed-effects modeling with decision-tree algorithms-to uncover distinct cognitive aging patterns shaped by intersecting sociodemographic factors. We demonstrate the utility of this method using data from the U.S. Health and Retirement Study (HRS), with cognitive outcomes measured across five waves. The protocol outlines materials, analytic steps, visualization tools, and strategies to avoid overfitting. We illustrate how GLMM-trees can detect previously unobserved subgroups defined by combinations of education, race, gender, and income that differentially influence both baseline cognitive performance and change over time. By enabling data-driven detection of heterogeneity, GLMM-trees offer a powerful tool for researchers seeking to apply intersectional frameworks to aging research and other domains involving complex longitudinal data.