Im building a linear mixed effects model of a variable, lets say the headway time maintained by drivers in car following situations. I want to use a training dataset to come up with a linear regression model and validate this model by running it through several sets of validation data and compare the results. This is fine as long as I consider only the fixed effects, however I cant seem to find a way to use the mixed effects in my model. Ill try to explain this more clearly below.

Lets say I have 80 sets of data in my training set. I can estimate the fixed effects and random effects for the data. I would of course come up with 1 set of fixed effects and 80 sets of mixed effects. So validating the model considering only the fixed effects is fine. How do I use the random effects in my model?

I use the linreg function in matlab to estimate fixed effects and nlmefit to estimate the mixed effects. Any views are greatly appreciated!!