add r2 for lavaan

[mattansb]

m <- lavaan::sem("mpg ~ cyl
                   hp ~ am", 
                 data = mtcars)

lavaan::lavInspect(m, "rsquare")
#>   mpg    hp 
#> 0.686 0.033

performance::r2(m)
#> 'r2()' does not support models of class 'lavaan'.
#> [1] NA

^{Created on 2021-02-17 by the reprex package (v1.0.0)}

[DominiqueMakowski]

should we return like a list with the R2 for each response?

[mattansb]

🤷‍♂️ I guess it has so be somewhat consistent with other r2s...?

[DominiqueMakowski]

out of curiosity, is there a way to aggregate them to have a "unique" model-general R2 (the explained variance of the dataset made of all the responses)?

[mattansb]

Not that I know of...

[DominiqueMakowski]

https://link.springer.com/article/10.3758/s13428-020-01532-y

[jebyrnes]

In SEM we generally report individual response variable R2 by convention. FYI, this could easily be done for piecewiseSEM as well to cover all of your bases - there is a function within the package, or just lapply your way across each model piece.