# Importance of absolute values of the covariance matrix in the nonlinear mixed models

I am fitting a nonlinear mixed model (three-parameter logistic function) without any hierarchical structure. I have adopted an unstructured variance-covariance structure for the random effects. Is it possible to know (out of the variance-covariance components, or absolute values of the variance-covariance components) which parameter is more relevant than the others to explain the variability of my observations? Are the units of the parameters in the three-parameter logistic function playing an important role? (I came across with this reference, but could not find something similar for nonlinear mixed models).

Any hint is more than welcome.

• Generally, high variances in the variance-covariance matrix of the random effects show that the corresponding random effect is important. – papgeo Nov 28 '19 at 3:23