TY - UNPB
T1 - A Hitchhiker’s Guide to Bayesian Hierarchical Drift-Diffusion Modeling with dockerHDDM
AU - Hu, Chuan-Peng
AU - Geng, Haiyang
AU - Zhang, Lei
AU - Fengler, Alexander
AU - Frank, Michael
AU - Zhang, Ru-Yuan
PY - 2022/11/1
Y1 - 2022/11/1
N2 - Drift diffusion models (DDM) are widely used to investigate decision-making processes in psychology, behavioral economics, neuroscience, and psychiatry. As one of the most cited software packages, HDDM (Hierarchical Bayesian estimation of DDMs), a python library, has been useful in helping researchers with minimal coding experience fit DDMs and other sequential sampling models to their experimental data. Despite the popularity of HDDM, its compatibility issues during installation and the lack of advanced Bayesian modeling functionalities, unfortunately, hamper its further applications in research practices. To circumvent these challenges, we integrated Bayesian modeling Python package ArviZ into HDDM and encapsulated them into a virtualized ready-to-use package in Docker, called dockerHDDM. Augmented by ArviZ, dockerHDDM provides richer data analysis functions and data visualization tools. This tutorial provides a hands-on guide on how to use dockerHDDM to efficiently conduct Bayesian hierarchical analysis of DDMs and is expected to facilitate the implementation, analysis, and reproducibility of DDMs. The workflow showcased here can be further generalized into broader applications of Bayesian data analysis.
AB - Drift diffusion models (DDM) are widely used to investigate decision-making processes in psychology, behavioral economics, neuroscience, and psychiatry. As one of the most cited software packages, HDDM (Hierarchical Bayesian estimation of DDMs), a python library, has been useful in helping researchers with minimal coding experience fit DDMs and other sequential sampling models to their experimental data. Despite the popularity of HDDM, its compatibility issues during installation and the lack of advanced Bayesian modeling functionalities, unfortunately, hamper its further applications in research practices. To circumvent these challenges, we integrated Bayesian modeling Python package ArviZ into HDDM and encapsulated them into a virtualized ready-to-use package in Docker, called dockerHDDM. Augmented by ArviZ, dockerHDDM provides richer data analysis functions and data visualization tools. This tutorial provides a hands-on guide on how to use dockerHDDM to efficiently conduct Bayesian hierarchical analysis of DDMs and is expected to facilitate the implementation, analysis, and reproducibility of DDMs. The workflow showcased here can be further generalized into broader applications of Bayesian data analysis.
U2 - 10.31234/osf.io/6uzga
DO - 10.31234/osf.io/6uzga
M3 - Preprint
BT - A Hitchhiker’s Guide to Bayesian Hierarchical Drift-Diffusion Modeling with dockerHDDM
PB - PsyArXiv
ER -