LIP External Colloquium: Satoshi Usami, The University of Tokyo, Japan - "Longitudinal Data Analysis for Inferring Within-Person Relations: Recent Methodological Issues in Model Selection and Estimation"

LIP External Colloquium: Satoshi Usami, The University of Tokyo, Japan - will present:

"Longitudinal Data Analysis for Inferring Within-Person Relations: Recent Methodological Issues in Model Selection and Estimation"

Abstract:
In longitudinal research in psychology and related fields, the inference of within-person relations, that is, relations among changes occurring within individuals, is one of the central themes. In particular, the random intercept cross-lagged panel model (RI-CLPM), which is intended to infer reciprocal relations within individuals, has rapidly become widely used over the past decade. At the same time, debates concerning the appropriateness of using the RI-CLPM and the selection of related models are still ongoing.

In this presentation, focusing especially on longitudinal studies based on a small number of measurement occasions, I will explain the RI-CLPM and other candidate models and organize the key issues surrounding model selection. I will also briefly introduce recent methodological developments rooted in extensions of structural equation modeling (SEM), which is the methodological foundation of the RI-CLPM. For example, I will provide an overview of matrix decomposition-based estimation approaches and their applications for addressing the problems of improper solutions that frequently arise when the RI-CLPM, or models that extend it by incorporating measurement error, are estimated by maximum likelihood. I will also provide an overview of robust estimation and its significance when inferring within-person reciprocal relations as relations among latent variables, as in the RI-CLPM.

Please use this link for joining hybrid:

https://mpib-berlin.webex.com/mpib-berlin/j.php?MTID=med667fb14fa2a1f8e5f98f6189bb32f9

Meeting number: 2741 607 6544

Password: 8bKvH8EbCe6