lmer model: non-normal residual

Question

I sometimes run into problems that the residuals of my lmer models are not normally distributed. I was wondering what to do in such circumstances?

Is gamm (generalized additive mixed model) the non-parametric equivalent to lmer? Or is glmer (generalized linear mixed-effects model) the non-parametric equivalent to lmer?

My variables are always continuous so glmer with binomial or poison families do not fit. Whenever I fit the models with Gaussian, the results tell me to go back to lmer; I also tried gamma family and it also doesn't work.

I am aware of the following questions and answers but they don't solve my problem:

• non normal distribution in lmer

• Problem with non-normal residuals (lmer function)

• R lmer Model Diagnosis qqnorm

• Heavy-tailed errors in mixed-effects model

I am also aware that you can log transform the data, but it doesn't always work on my data.

Could anyone tell me which types of models I should try? Thanks!