Author: Seung Hyun Min
ggsem is a web-based application and an R package, enabling users to draw path diagrams for structural equation modeling (SEM) and small-to-medium sized networks interactively using the ggplot2 engine.
As the core of the ggsem package, its web app (built with Shiny) provides an interface that allows extensive customization, and creates CSV outputs, which can then be used to recreate the figures either using the web app or in a typical ggplot2 workflow. This will be useful to generate reproducible figures.
Unique features of this R package include options for users to draw gradient lines, gradient arrows, self-loop arrows, as well as interactively change the layout of nodes and edges. Also, ggsem provides layouts of the nodes and edges from the igraph package for networks, understands the string input form in laavan’s syntax, and supports layouts from the semPlot package for SEM diagrams.
Coding is optional as users can create high-quality figures of network and SEM diagrams from start to finish within the ggsem application.
Nevertheless, the outputs from ggsem can be recreated as a raw ggplot output in script-based workflow, allowing users to further modify the graphical output using other packages.
The SEM diagram was created using data and a model without coding (see Chapter 6).
The two networks were separately generated from the ggsem app, annotated with ggplot2 functions, and combined with patchwork in script-based workflow (see Chapter 4).
The stable version of ggsem can be installed using
install.packages() from the Comprehensive R Archive Network
(CRAN).
install.packages("ggsem")The examples on this website use the development version of the package, which can be directly downloaded using the code below:
install.packages("devtools")
devtools::install_github('smin95/ggsem')The ggsem app can be initiated online using the link: https://smin95.shinyapps.io/ggsem/.
It can also be started locally in RStudio:
ggsem::ggsem()To access an updated tutorial (sample codes and figures) of the package, please visit https://smin95.github.io/ggsem/.
The creation of the package was in part inspired by this blog post: https://drsimonj.svbtle.com/ggsem-plot-sem-models-with-ggplot2.
If you find any issues with ggsem (both application and package), please contact me via email (seung.min@mail.mcgill.ca).