I am having an issue running glmer, namely:

Error in pwrssUpdate(pp, resp, tol = tolPwrss, GQmat = GHrule(0L), compDev = compDev, : pwrssUpdate did not converge in (maxit) iterations

From what I’ve read online, it seems this error can indicate issues in the dataframe/model specification rather than being an issue of maxit being too low. Would you mind having a look at the data attached, the glmer command below and seeing if any clear error jumps out? I would be extremely grateful as I’ve tried many things already (see bullet points below!).

I am predicting a very skewed score of childhood mental health, the CBCL, from sex and birthweight. Children are nested within families which are nested within testing site.

m2 <- glmer(cbcl ~ sex  +  bw_kgs  + sex:bw_kgs +  (1|site_id/family_id),
            data = d,
            family = Gamma(link=identity),
             control = glmerControl(optimizer = "bobyqa" ,
                                      optCtrl = list(maxfun= 100000)),
            nAGQ = 0,
            verbose = 1)

Interestingly, the model previous to this (m1) included just sex and the random effects, and ran fine.

Things I’ve tried:

• switching optimiser from bobyqa to nloptwrap
• Standardizing birthweight (bw_kgs)
• Removing NAs rows from the data (NAs in any predictor or the outcome)
• Scaling the outcome without centering it (as this would involve 0s, which glmer can’t seem to handle)
• Winsorizing high values of the outcome, cbcl

Some of these experiments resulted in a different but related error:

Error in (function (fr, X, reTrms, family, nAGQ = 1L, verbose = 0L, maxit = 100L, : (maxstephalfit) PIRLS step-halvings failed to reduce deviance in pwrssUpdate

Hope someone can help!
Thanks.
data4github.xlsx