In the social sciences, hierarchical, nested, or clustered data structures are common due to cluster or multi-stage sampling (e.g., students in classrooms), which can violate the independence assumption of basic statistical tests because of the design effect. Moreover, predictor variables can be located at different analytic levels (e.g., student and classroom characteristics) which must be considered when analyzing the data. Multilevel modeling is the most commonly used statistical method for dealing with hierarchical data structures though it should be noted that there are other approaches available, i.e., fixed-effect model, cluster-robust standard errors, and generalized estimating equations. The workshop introduces basic concepts of multilevel modeling with cross-sectional data using the open-source program R, e.g., hierarchical data structures, predictor centering, statistical control of predictors at Level 1, and different classes of multilevel models (i.e., random intercept model, and random intercept and slope model). The lecture will focus on R using the packages lavaan, lme4, and misty. Additionally, Mplus input and output files will be provided for all examples, but won’t be discussed due to time constraints. Note that basic knowledge of regression analysis and practical experience with the statistics program R is required for participation. It is recommended to have R (https://cran.r-project.org/) and RStudio (https://posit.co/download/rstudio-desktop/) installed on the computer to follow the practical examples.
Literature
Hox, J., Moerbeek, M., & van de Schoot, R. (2017). Multilevel analysis: Techniques and applications (3rd ed.). Taylor & Francis.
Luke, D. A. (2020). Multilevel modeling (2nd ed.). SAGE Publications Inc.
Snijders, T. A. B., & Bosker, T. J. (2012). Multilevel analysis: An introduction to basic and advanced multilevel modeling. SAGE Publications Inc
Webinar speaker
Dr. Takuya Yanagida is a senior scientist at the Faculty of Psychology at the University of Vienna, and an adjunct researcher in statistics at the National Centre for Learning Environment and Behavioural Research in Education at the University of Stavanger. He currently serves as associate editor for the Developmetrics section of the European Journal of Developmental Psychology. His primary research interests focus on quantitative methods in the social and behavioral sciences, as well as statistical programming in the open-source environment R. He held more than 50 workshops on various methodological topics, including multilevel structural equation modeling, longitudinal data analysis, mixture modeling, and missing data handling at numerous universities across Europe and Asia.
