ggplot2 SEM models with tidygraph and ggraph
@drsimonj here to share a ggplot2-based function for plotting path analysis/structural equation models (SEM) fitted with Yves Rosseel’s lavaan package.
SEM and its related methods (path analysis, confirmatory factor analysis, etc.) can be visualized as Directed Acyclic Graphs with nodes representing variables (observed or latent), and edges representing the specified relationships between them. For this reason, we will use Thomas Lin Pedersen’s tidygraph and ggraph packages. These packages work together to work with relational structures in a tidy format and plot them using ggplot2.
Below is a function
ggsem(), which takes a fitted lavaan object and returns a ggplot2 object representing the nodes, edges, and parameter values. It handles regression paths, correlations, latent factors, and factor loadings.
library(tidyverse) library(tidygraph) library(ggraph)
Continue reading →