I am running a beta regression, my dependent variable is a proportion, beta-distributed such as y ~ B (µ,ɸ ), with µ its mean and ɸ, a constant precision parameter; the link function used, g(.), was logit. Pseudo-R2 was estimated 0.97 reflected that the goodness-of-fit measured from the beta regression was strong. ɸ was evaluated as 242.04. I know the larger ɸ the smaller the variance of y. But how to define "large ɸ"?