On CRAN and GitHub
Network Psychometrics in R
EGAnet represents a set of measured variables as a network and estimates how many dimensions organize it. The package is built around reproducible code and visualizations suitable for publication.
On CRAN and GitHub
The function map
From raw items to a finished network model, every stage of the EGA workflow is its own set of composable functions.
itemStability() · invariance() · net.loads() · UVA() · CFA()
simEGM() — simulate from the Exploratory Graph Model
Version 2.5.0
node.shape (and its shape shortcut) now actually changes the node shape, instead of being validated but silently ignored in favor of a hardcoded solid-circle-with-hollow-border look. The same solid-inside/hollow-border style is preserved for any shape with a matching base R "hollow" pch counterpart (e.g., node.shape = 15 draws solid squares with hollow-square borders), and the default (node.shape = 19) renders identically to before. The five "fillable" pch codes (21:25) use their own built-in fill + border instead. The legend keys also now match each community's actual node shape (previously always a circle, regardless of node.shape)
label was colliding with ggnet2's own, differently-behaved native label argument, since {EGAnet}'s network plot.* methods now forward the full ggnet2 argument set. This caused custom node text passed via label to be drawn as a second, overlapping text layer on top of the original node names (rather than replacing them). ggnet2's own label mechanism is now always disabled in favor of {EGAnet}'s own label layer, and passing label directly now raises an informative warning pointing to node.label (the correct argument for custom node text) and label.size = 0 (to remove labels) instead of silently doing the wrong thing. label is intentionally *not* turned into an automatic shortcut for node.label, since composite plots (bootEGA, hierEGA, invariance, dynEGA.Group/.Individual, compare.EGA.plots) also forward arguments to ggpubr::ggarrange, which has its own unrelated labels (plural) argument for panel captions -- label vs. labels would be an easy typo to make and miss