My research interests lie at the intersection of psychometrics and quantitative methods in psychology. I am particularly interested in longitudinal data modeling, with a focus on traditional and intensive designs. My work encompasses various analytical strategies, including linear growth-curve modeling and advanced methods tailored to intensive longitudinal data, such as dynamic structural equation modeling (DSEM). I am especially interested in capturing, modeling, and interpreting temporal structures in psychological processes. In addition, as I continue to deepen my training, I am also interested in applying diverse statistical tools to address practical research problems in psychology.
More specifically, the research topics I am concerned with include, but are not limited to, the following:
Intensive Longitudinal data modeling (applied in psychology, education, medicine, and social sciences)
In recent years, the development of experience sampling methods (ESM) and wearable technologies has substantially broadened the scope of psychological research, enabling the systematic study of intensive longitudinal data (ILD). This shift enables researchers to capture dynamic processes as they unfold in daily life, offering new opportunities to understand temporal dependencies, contextual sensitivity, and individual differences in psychological functioning. Within this emerging paradigm, my research focuses on how advanced statistical modeling can be used to extract meaningful dynamic structures from ILD and apply them to explain psychological mechanisms and evaluate interventions.
A central component of my work involves modeling within-person dynamics. Building on DSEM, I aim to characterize temporal processes such as autoregressive effects, inertia, and recovery, while accounting for individual differences in these parameters. Rather than focusing solely on mean-level changes, I place particular emphasis on residual variance (amplitude) as a core indicator of psychological variability. This perspective highlights the importance of fluctuations themselves, suggesting that reductions in variability—such as emotional instability—may provide a more sensitive indicator of treatment effects than changes in average levels.
Another key interest lies in identifying dynamic heterogeneity. Many conventional models assume a homogeneous dynamic structure across individuals, potentially obscuring meaningful variation. To address this limitation, I explore Bayesian mixture modeling and latent class approaches to identify subgroups with distinct dynamic patterns. This line of work is further developed through a joint dynamic mixture modeling framework that integrates both mean and variability processes, enabling a more comprehensive representation of individual trajectories and improving predictions of distal outcomes.
I am committed to developing flexible and transparent modeling tools, particularly through Bayesian frameworks implemented in Stan and R. These efforts are motivated by the need to overcome the limitations of existing software and to promote reproducibility and extensibility in dynamic psychological modeling.
Psychometric models (applied in psychology and education)
My research interests focus on advancing psychometric models and their applications in psychology. I am interested in both the methodological development and applied use of IRT, including multidimensional, explanatory, and mixture IRT models, to better capture heterogeneity in latent traits and item functioning. In applied psychological contexts, I explore key issues such as classification decisions based on cut-off scores, differential item functioning (DIF), and the development of longitudinal reliability indices, aiming to improve the validity and interpretability of measurement across diverse populations and over time.
Beyond traditional response formats, I am also interested in modeling non-standard data structures, such as pairwise comparisons and rankings, using Thurstonian IRT frameworks. This line of work connects naturally with latent space approaches, including multidimensional scaling (MDS), where I investigate how response behavior and psychological representations can be jointly modeled within a unified latent structure. By integrating IRT with latent space models, my research seeks to bridge measurement theory with representation learning, providing a more comprehensive framework for understanding both individual differences and the structure of psychological constructs.
Quantitative methods in experimental psychology (applied in psychology, natural sciences, and medicine)
GLMM joint modeling to process experimental data
Using GLMM-based joint modeling frameworks to capture multiple processes simultaneously (e.g., accuracy and response time)
Causal Inference in Experimental Psychology
Bridging modern causal methodologies with experimental designs to better identify and interpret underlying psychological processes.
Semiparametric/nonparametric statistics
Semiparametric and nonparametric statistical modeling are another important component of my research agenda, particularly when working with complex psychological data that is nonlinear and heterogeneous.
In semiparametric modeling, I am primarily interested in combining interpretable parametric components with flexible nonparametric structures to better capture time-varying nonlinear processes. Specifically, I aim to retain theoretically meaningful parameters, such as treatment effects or group differences, while allowing temporal dynamics and covariate relationships to evolve in a data-driven manner. This approach is especially valuable for intensive longitudinal data (ILD), as psychological processes often exhibit nonlinear trajectories (e.g., adaptation, recovery, or threshold effects) that cannot be adequately represented by linear assumptions. Semiparametric models incorporating smooth functions (e.g., spline-based or Gaussian process components) can reduce model misspecification while preserving interpretability, providing a more realistic representation of psychological dynamics.
Concurrently, my interest in nonparametric modeling centers on analyzing structured behavioral data, particularly ranking and sequential data, which are prevalent in psychological and behavioral research. Unlike traditional scale-based measurements, these data types capture preferences, decision processes, and the temporal ordering of behaviors but often violate standard distributional assumptions. I plan to apply and extend nonparametric approaches, such as distance-based models, kernel methods, and latent-space representations, to investigate underlying structures in ranking and sequence data. This includes integrating methods from MDS, Thurstonian models, and unfolding models to better understand how individuals organize preferences and make decisions.