Random slopes

#1
Dear experts,

When should I include random slopes in my non-linear mixed effects model?

Is it solely when it results in a better Log Likelihood Ratio or is there a theoretical decision to be made, for example only include random slopes when each subject is in each leavel of treatment etc, i.e Barr (2013) Random effects structure for confirmatory hypothesis testing: Keep it maximal?. J Mem Lang

Thanks
 

hlsmith

Omega Contributor
#2
What do you mean by "non-linear effects model", logistic regression?

It is likely a combination of them, so given that they help better explain the phenomenon and they make theoretical sense.