I seek to simultaneously analyse 3 variables describing the content of a corpus of texts. These 3 variables have negative binomial distributions as they result from a terms count process.
The variables are correlated with each other (correlation coefficients of 0.7 and 0.4) so I need to analyse them in a model that take into consideration this correlation. The 'gsem' model allows for considering the correlation of residuals for Gaussian distributions but not for non-Gaussian distributions. Am I missing an option of the gsem model? Or is there another model that allow me to take into the account the correlation between non-Gaussian dependent variables?
Thank you
The variables are correlated with each other (correlation coefficients of 0.7 and 0.4) so I need to analyse them in a model that take into consideration this correlation. The 'gsem' model allows for considering the correlation of residuals for Gaussian distributions but not for non-Gaussian distributions. Am I missing an option of the gsem model? Or is there another model that allow me to take into the account the correlation between non-Gaussian dependent variables?
Thank you
Comment