A 6-hour workshop taught by Paul D. Umbach, Ph.D.
Overview of our multilevel modeling workshop
This workshop provides an introduction the basics of multilevel modeling, focusing on practical applications rather than statistical theory. Researchers from a range disciplines often collect data that have a hierarchical structure. Students are nested within schools (and/or classrooms), employees are nested within firms, and multiple test scores are nested within students. One consequence of this nested structure is that observations not statistically independent, thus violating a basic assumption of standard analytical techniques. Ignoring the nesting effect and proceeding with conventional, single-level methods of analysis (like linear regression) can yield misleading results. At the end of this workshop, you will know the basics of how to run and interpret multilevel models (sometimes called hierarchical linear models or general linear models), a statistical approach that accounts for the nesting effect and avoids these problems, as well as those associated with aggregation and disaggregation.
By the end of this workshop, participants should know the following
- The conceptual foundations of multilevel models.
- An appreciation of the advantages and disadvantages of multilevel modeling as compared with other approaches to nested data.
- Modeling slopes and intercepts as outcomes.
- How to build two-level models using HLM 7.
- Approach to building growth models in a multilevel context.
- How to interpret and explain the output from multilevel modeling software.
- Practical tools and strategies for developing and testing multilevel models.
- A clear understanding of the differences between fixed and random effects.
Who should attend?
The target audience for this workshop is a range of researchers, including institutional researchers, market researchers, policy analysts, student affairs professionals, assessment professionals, graduate students, and faculty, who use surveys in their work. It is important that participants have a working knowledge of ordinary least squares regression. If you need a refresher on ordinary least squares regression, we encourage you to enroll in our regression workshop prior to taking this workshop.
- Why we need multilevel modeling and why other approaches are inadequate
- Brief discussion of conceptual underpinnings of multilevel modeling
- Extending regression with random intercepts and slopes as outcomes
- Centering the variables we include in our models
- Building the two level model: Null model, random intercept model, and full model
- Using real data and HLM 7 to build models
- Interpreting two-level model output
- Using multilevel model to study growth
- Interpreting growth model output
- Introduction to extensions of the two-level model: multilevel models with categorical outcomes, three level models, and cross classified models.
We will be using HLM 7 for class demonstrations. You can download the HLM 7 student version at the SSI Scientific Software website. Please note that HLM only compatible with Windows. I will also provide output, data, and syntax for SPSS, Stata, and SAS.