I'm looking to identify outliers in a logistic regression model, e.g.

mod1 <- glmer(proportion ~ a*b*c + (1 | ID), family = binomial, weights = 
n, data = my_dat)  

The model itself converges. However, when I try to determine influence using influence.ME (in order to calculate Cook's and DFBETAS), the operation fails to converge:

mod1.est <- influence(mod1, obs = T)

And produces a warning message:

In checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv,  :
Model failed to converge with max|grad| = 0.00125887 (tol = 0.001, component 

I've tried the following fixes, but the warning message remains the same:

  1. I converted my ID variable from an integer to a factor (it becomes a factor with three levels, 67 observations at each level):

    my_dat$ID <- as.factor(my_dat$ID)  
  2. I inputted ID as the grouping variable in the influence function:

    mod1.est2 <- influence(mod1, group = "ID")  

Any advice about what might be going on and/or how to identify outliers using a different method would be much appreciated. Thank you!


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.