Path analysis is a form of a structural equation model in which all the variables in the model are observed and all the paths of the model are estimated simultaneously. Simply put, path analysis is a series of linear regressions. Hence, the same assumptions that we have for linear regression hold for path analysis. Path models are often accompanied with a path diagram, so the reader and researcher alike can visualize the complex system of relations.
Mediation is something we can test for in a path model. Mediation occurs when a variable X affects a variable Y partly or completely through an intermediate variable M. Although both path analysis and mediation analysis are commonly used in the social sciences, they have become popular tools in other fields.
In this workshop, we will focus on the terminology and notation commonly used in path analysis and structural equation models; understand how to determine if a model is under-identified, just-identified, or over-identified; interpret direct, indirect, and total effects; discuss when different estimators may be implemented in path analyses; and cover the commonly used fit measures to determine if there are issues with the model.
The workshop will include a hands-on component using the R statistical software package. Participants should install the most recent versions of both R and RStudio on their computers prior to the workshop. To download and install R, visit https://cran.r-project.org/ and to download and install RStudio, visit https://www.rstudio.com.The lavaan package will be used to estimate our path models.
The workshop is appropriate for faculty, research staff, graduate students, and advanced undergraduate students. No previous experience with path modeling is required, but knowledge of linear regression is essential.
Instructors: The Cornell Statistical Consulting Unit (CSCU)
Date: October 13, 2022
Time: 9:00am - 12:00pm Arizona Time
Location: Zoom, link will be sent out 2 days before workshop
Workshop Fee: $50