Title: Leveraging Dynamic Structural Equation Modeling for Analyzing Intensive
Eye-Tracking Data: Illustration and Software Development

Author: Benedikt Langenberg

Abstract:
Consider an eye-tracking experiment that analyzes reading skills in children.
Vast amounts of data are generated with multiple measurements of the eyes’
position per second. Reading is a complex and dynamic task, where words can
affect the processing of the next words (both facilitate and hinder). Dynamic
structural equation modeling (DSEM) has increasingly gained attention in the
analysis of intensive longitudinal data (ILD), offering insights into diverse
psychological processes' dynamics. However, DSEM's application has been
largely confined to ecological momentary assessment, experience sampling, and
diary data. This presentation aims to demonstrate DSEM's application in
analyzing ILD from neuropsychological experiments, specifically in an
eye-tracking experiment encompassing 267 children. The outcome of interest was
the improvement in reading skills over the course of four years. Emphasizing
the necessity of accounting for dynamics, this presentation reveals
substantial bias in effect estimates when disregarded. Additionally, DSEM
provides novel insights into interindividual differences in the dynamic nature
of reading. The presentation culminates in introducing a user-friendly
small-scale R software package, serving as an interface to the well-known
Mplus software, facilitating DSEM estimation for experimental data.