2021 Methods Workshop
Presented by the Learning Performance Research Center
Free to all attendees through sponsorship by the Berry Family Distinguished Professorship!
May 12, 2021 – 10:00 a.m. to 5:00 p.m.
Virtual gathering on Zoom
About the Workshop
Longitudinal modeling and missing data analysis:
An introduction within the structural equation modeling framework
This one-day workshop commences with a brief introduction to the popular structural equation modeling methodology. Special emphasis is placed then on its key notions of implied matrix, identification, fit function, model fitting, and parameter estimation. Next, the widely used intercept and slope model is introduced as are its non-linear extensions. The Health and Retirement Study (HRS) dataset is employed to illustrate these models. Issues pertaining to missing data are briefly discussed then (to be revisited later), as are issues related to how to optimally analyze such data from nationally representative studies. In the second part of the workshop, inferential problems resulting from missing data are covered along with missing data patterns and mechanisms. The methodology of full information maximum likelihood (FIML), with its robust version and auxiliary variables, is then introduced, and the HRS examples revisited in light of the applicability of the FIML method on that data set. Throughout the workshop, the popular latent variable modeling software Mplus is used. All data sets utilized will be made available to the attendees, as will be an elaborated lecture notes volume covering all topics of the workshop and containing all its slides.
Presenter: Dr. Tenko Raykov
Tenko Raykov is Professor of Quantitative Methods in the Graduate Program in Measurement and Quantitative Methods at Michigan State University. His research focuses on a variety of areas with many empirical applications, including latent variable and structural equation modeling, behavioral and social science measurement, scale construction and development, multilevel modeling, longitudinal data modeling, analysis of incomplete data sets (missing data analysis), latent class analysis (finite mixture modeling), item response theory and modeling, as well as survival and duration analysis. At Michigan State University since 2005, he teaches courses in structural equation and latent variable modeling, psychometric theory, scale construction and development, multilevel modeling, univariate and multivariate statistics, and item response theory and modeling, as well as units on missing data analysis, latent class analysis, and statistical software (Stata and R).
Registration has closed! We hope to see you next year!
Free to all attendees for 2021!
2019 – Matthew McCrudden, Pennsylvania State University – Mixed Methods
2018 – Greg Hancock, University of Maryland – Latent Growth Curve Modeling
2017 – Roy Levy, Arizona State University – Foundations of Bayesian Statistical Modeling
2016 – Todd D. Little, Texas Tech University – Measurement, design, and analysis issues in longitudinal modeling with a particular focus on the longitudinal CFA model as the basis for both panel and latent growth curve modeling.
2015 – Greg Hancock, University of Maryland – Second Course in Structural Equation Modeling
2014 – Bethany Bray, Pennsylvania State University – An Introduction to Latent Class and Latent Profile Analysis
2013 – Greg Hancock, University of Maryland – A First Course in Structural Equation Modeling