Under what conditions can I get the explained variance of every variable in a regression model?
In particular, suppose I have several observed variables, and a target variable that will be modeled by regression. My understanding is that it is not possible to get the % variance explained by each variable if there are any correlations between the observed variables. However, is it possible to define some latent variables and constrain the model so that the latent variables are orthogonal by definition? How do I do this in lavaan?