May you have a look @

http://www.r-inla.org/

and

http://people.bath.ac.uk/fl353/tutorials/oulu/tut_spde_lindgren.pdf

Dear Belay

Bayesian model is something different but any how thank you for your participation

The final chapter of this text shows you how - I use Bayesian estimation to do this as there tends to be lots of random effects and you are modelling variances with a likely skewed - non normal -distribution for the variance

https://www.researchgate.net/publication/260772180_Developing_multilevel_models_for_analysing_contextuality_heterogeneity_and_change_using_MLwiN_Volume_2

Book Developing multilevel models for analysing contextuality, he...

Thank you so much Dear Kelvyn