Multilevel Modeling for Longitudinal Data
Dr. Aidan Wright, University of Pittsburgh
July 24th-26th, 2019 - Days 1-2 will involve lectures and planned exercises, day 3 will allow for spill-over and individual consultations with your own data. Days 1-2 are anticipated to run from 9am-4pm with a 1hr break for lunch, and day 3 will depend on days 1-2.
Trainees (students and post-docs) - $500
Professionals (including faculty) - $750
NOTE: If also registering for R for Social Scientists, a 20% fee reduction will be applied to both registrations.
This 2.5-day workshop will offer a basic introduction to multilevel modeling (MLM; also known as mixed effects modeling and hierarchical linear modeling), with an emphasis on its application to modeling longitudinal data. MLM is a flexible analytic framework for analyzing data that has a nested or hierarchical structure. Studies that employ a longitudinal design (e.g., panel data, daily diaries, ecological momentary assessment) generate nested data, in the form of repeated observations nested within individuals, and therefore are usually best analyzed with MLM. The principles of MLM generalize across different forms of nesting (e.g., time-points within person; pupils within classrooms), however longitudinal data generally requires attention to specific issues (e.g., trends, cycles, temporal dependency) that make learning MLM as applied to this context advantageous. This workshop will focus on (1) understanding the fundamental issues associated with nested data and MLM, (2) designing and estimating MLMs in contemporary statistical software, and (3) interpreting the output of MLMs. Timing will not allow in depth discussion of longitudinal study design and methodology, although a brief overview of considerations associated with leveraging time in research will be offered. Thus, the focus is on the fundamentals of MLM, although the examples and exercises in the workshop will emphasize applications to longitudinal data. Those interested in learning fundamentals of MLM are welcome and should find this course helpful.
The primary software package for the course will be R. I recommend downloading RStudio (rstudio.com) and R prior to the start of the course. However, additional syntax will be provided for example models in Mplus, SAS, SPSS, and STATA for those who are not interested in using R.