Statistical Methods Seminar Description
Overview: You probably work with nested data. Perhaps you study students within classrooms, professionals within corporations, patients within hospitals, or repeated observations from the same person. In each of these cases and many more, the data are hierarchically arranged and may require methods beyond multiple regression or analysis of variance. These methods fall under the heading of Multilevel Modeling (MLM), sometimes referred to as mixed modeling, hierarchical linear modeling, or random coefficients modeling.
Our goal for this course is to teach the fundamentals of multilevel modeling. We will begin by describing nested data structures and what this means for your data. We will cover key concepts in the analysis of both cross-sectional and longitudinal nested data. Finally we will close the week with a few advanced multilevel topics to give you ideas for future analyses.
Along the way we will do our very best to foster a relaxed atmosphere where you can learn, ask questions, and consult with us about your multielvel data and analysis questions.
- Nested and Multilevel Data
- Fixed and Random Effects
- Centering Predictors in MLM
- Interactions and Simple Slopes in MLM
- Growth Models in MLM
- Time-varying and Time-invariant covariates
- Power Analysis and Sample Size in MLM
Instructors: James P. Selig & Patrick Coulombe
James P. Selig received his Ph.D. in quantitative psychology from the University of Kansas. He is currently an Assistant Professor in the Educational Psychology program at the University of New Mexico where he teaches graduate courses in research methods and introductory and advanced statistics. His quantitative areas of interest include: longitudinal data analysis, multilevel modeling, structural equation modeling, interdependent data analysis, and mediation models. He is particularly interested in temporal design, or issues of time and timing in the design of longitudinal studies, and in the topic of time lag dependent effects. He currently serves on the editorial boards for Developmental Psychology, Parenting: Science and Practice, and The Journal of Humanistic Counseling.
Co-Instructor Patrick Coulombe – is a doctoral student in Quantitative Psychology at the University of New Mexico. He is interested in multilevel modeling and structural equation modeling, particularly as they apply to the analysis of longitudinal data. He is also interested in Bayesian statistics, Monte Carlo simulation, online research, and the issue of false positives in psychological research. He originally hails from sunny(ish) Canada.
This course will be helpful for researchers, with a basic background in statistics, wishing to learn how to analyze nested or hierarchically ordered data sets (e.g. students within schools, employees within organizations, and repeated measures within people).
Participants should have a good working knowledge of the principles and practice of multiple regression and elementary statistical inference. You do not need to know matrix algebra, calculus, or likelihood theory. Participants from a variety of fields— including sociology, psychology, education, human development, marketing, business, biology, medicine, political science, and communication— will benefit from the course.
The course will emphasize the use of SPSS for multilevel modeling. Some assistance will be available for questions related to other multilevel modeling packages. Nearly all the techniques taught in the course can be translated fairly easily to most other packages.
Below are links to course files for those who enrolled in the course. Please download these files onto your computer before the first day of the course. The files are password protected to respect the intellectual property rights of the instructors. By using your login information you agree not to share your login information or the content protected by it.
After May 1st: Click Here to Access The Multilevel Modeling Course Files