Advances in Moderation Analysis: An Introduction to Mixture Regression
Moderation analyses in research typically are conducted using either interaction terms or multiple group analyses. In this way, scholars can examine whether the relationship between variables differs across the levels of another variable.
However, it may be that the relationships among variables differ for certain “unobserved” or “latent” subgroups within a population (i.e., groups that are not directly observed). Mixture regression — a type of finite mixture model/latent class analysis — is a method to examine whether the relationships between independent and dependent variables differ for latent classes. That is, mixture regression identifies classes for whom the relationships between the independent and dependent variables differ. This method to examine moderation is highly flexible, identifying a multiple of groups for whom variables are differentially related and allows for examining the characteristics of those groups.
By attending this webinar — presented by W. Justin Dyer, Ph.D. — you’ll receive an accessible introduction to mixture regression and be able to understand:
- basic concepts of moderation;
- how mixture regression can address moderation questions that previous standard methods cannot; and
- how to conduct a mixture regression.
You’ll learn how to identify subgroups of families and individuals who may not follow typical patterns, which can otherwise be difficult. You’ll learn about the theory and techniques related to mixture regression that will enable to you to employ this important new method.
Approved for 1 CFLE contact hour of continuing education credit.
Intended Audience: Researchers, new professionals, students
About the Presenter
W. Justin Dyer, Ph.D., is an associate professor at Brigham Young University, where he has taught courses on family and structural equation modeling. His research area includes fatherhood with a particular emphasis on fathers of children with disabilities and incarcerated fathers. His work also has focused on statistical innovations that enable researchers to better address family research questions, an area in which he has published several articles. He has taught numerous methodology workshops, including a yearly workshop on longitudinal structural equation modeling.
On-Demand Webinar Recording and Classroom Use
Even if you can't watch this webinar live, your registration will still grant you access to watch the recording at your convenience. The fee for this webinar is $25 for NCFR student members, $45 for NCFR members, $85 for nonmembers.
License for classroom use by one professor is available for $105 for NCFR members, $185 for nonmembers.
License for departmental use (multiple professors) is available for $155 NCFR member / $305 nonmember.
Departmental license for CFLE-approved programs is $125.