I am using lme4 package to build a multi-level model. It can be said that my data has three levels: company level, industry level, year level.
However, adding all three levels in a model returns a singular fit. That is why I am thinking of leaving only company and industry levels and using robust standard errors to omit heteroskedasticity issue. I found some tips on obtaining robust st. errors (robust standard errors for mixed-effects models in lme4 package of R).
Nevertheless, I want also to receive p-value based on it. Can lmerTest package implement robust standard errors? Also, should I use robust standard errors in the first place? Any chance dummy variables for years would be enough?