Using lavaan, my CFA model returns loadings in the ‘Std.all’ column. How can I graph these values? For example:
library(semPlot)
model_object <- cfa('factor =~ item1 + item2 + item3', data = data.frame(item1 = rnorm(10), item2 = rnorm(10), item3 = rnorm(10)))
semPaths(model_object, whatLabels = 'std')
Hey everyone, I recently encountered a similar situation when trying to get a visual overview of the standardized loadings, and I ended up experimenting a bit to customize the plot to my liking. I found that semPlot’s semPaths function can be really handy, but sometimes the default settings don’t display everything as clearly as I’d like. One trick that worked for me was to play around with the argument that controls what labels get printed. For example, instead of just using whatLabels = ‘std’, I sometimes combine it with other settings to ensure clarity around the numbers I care about.
I also tried exporting the fully standardized summary directly from lavaan with summary(model_object, standardized = TRUE) and then parsing the ‘Std.all’ column. With that information in hand, I used base R plotting functions or even ggplot2 to create a customized graph, which gave me a bit more control over the appearance.
I’m curious – have any of you tried mixing semPaths with manual adjustments to the plot? Or perhaps even switching to a different visualization method after extracting the loadings? It would be great to hear what other creative solutions have worked for you! 
I have tackled this problem by combining built-in semPlot functionalities with some custom extraction. I would extract the standardized estimates directly using lavaan’s summary call and then use a plotting package like ggplot2 for more tailored visuals. While semPaths with whatLabels = ‘std’ offers a quick solution, creating a custom plot often gives better control over label placement and graph aesthetics, especially when the default visualization does not meet all requirements. My own workflow enhanced the clarity of factor loadings in a way that semPaths alone did not.
hey, i exracted std loadings using inspect(model_object, ‘std’) then plotted with ggplot2. this let me tweak colors & labels better than semPaths. customizing plot manually solved my issues.
Hey everyone, I tried a slightly different approach for visualizing these standardized loadings by stepping away from semPaths a bit. I extracted the loadings using lavaan’s summary function as well, but then I ended up playing around with the qgraph package to see how a network graph might offer a new perspective. I found that by feeding the standardized loadings into qgraph, I could naturally display how strong and in what direction each item was related to its factor, which was pretty neat. The customization potential with node colors, edge widths, and labels felt super flexible―almost like you can build a story out of your latent variables instead of just a static diagram. Has anyone else tried using qgraph or even interactive components like in plotly to explore these kinds of relationships? I’m curious to know if visualizing the model as a network really helps you interpret the structure any better, or if you prefer sticking with the more traditional semPaths plots. I’d love to hear your experiences and any tips you might have! 