Multilevel Modeling in the Social Sciences (Chapel Hill, NC)


  • Michelle Dion, McMaster University

Across the social sciences, we encounter multilevel, hierarchical, or clustered data, such as students nested in classrooms or schools or voters nested in electoral districts or states. Often, we have theories about how characteristics at different levels of analysis are related to lower level outcomes. For instance, student characteristics help explain student learning outcomes, but so do the characteristics of their teachers and classroom peers. We may even hypothesize that students with certain characteristics will learn differently if they share certain characteristics with their teachers or peers. To incorporate characteristics from each level into our analysis, we use multilevel models, which are also known as hierarchical linear models or mixed models in some social science disciplines. If our outcomes of interest have non-normal distributions, we may use multilevel/hierarchical/mixed generalized linear models. Repeated observations over time, sometimes referred to as pooled time series, panel data, or time series cross-section data (TSCS), can also be analyzed as a case of multilevel modelling. These models have applications in education, psychology, political science, sociology, public policy, and public health.

In this workshop, we will examine and apply all these types of multilevel models in social science applications. The workshop will both present the statistical models and demonstrate how to apply the models using the R programming language in R Studio, which is free software that includes all the core functionalities of statistical software (e.g. command text editor, command console, results pane, and graphics viewer). Though the primary tool used for teaching and demonstrations during class will be R, comparable commands or syntax for STATA will be provided when available. Workshop participants will be provided sample datasets and command syntax and encouraged to follow along on their laptops or lab machines during workshop meetings. We will also set aside time to discuss how to apply these methods to workshop participants’ projects and to demonstrate how to visualize or graph model results.

The workshop will assume understanding of linear regression and previous experience with applied statistical analysis using R, STATA, SAS, or SPSS. Participants are not expected to be familiar with R, though some experience using statistical software syntax will be helpful. Participants are not assumed to be familiar with advanced mathematics, such as matrix algebra. Registered participants who have no experience using R will be given recommendations for interactive, self-guided learning tools to familiarize themselves with R and R Studio before the workshop. Participants will be provided with sample data and R command files for all models covered in the workshop.

The workshop will cover multilevel models at the level of and as covered in:

  • Stephen W. Raudenbush and Anthony S. Bryk. Hierarchical Linear Models: Applications and Data Analysis Methods, 2nd edition. Thousand Oaks, CA: Sage. [Chapters 1-10]
  • Andrew Gelman and Jennifer Hill. Data Analysis Using Regression and Multilevel/Hierarchical Models. New York: Cambridge UP, 2007. [Chapters 11-15]
  • W. Holmes Finch, Jocelyn E. Bolin, Ken Kelley, Multilevel Modeling Using R, 1st edition, Boca Raton, FL: CRC Press, 2014. [Chapters 1-8]

Fee: Members = $1700; Non-members = $3200

Tags: multilevel models, hierachical linear models, MLM, HLM, multilevel modeling

Course Sections

Section 1

Location: University of North Carolina -- Chapel Hill, NC

Date(s): June 17 - June 21

Time: 9:00 AM - 5:00 PM


  • Michelle Dion, McMaster University