The command for an HLM model in Stata is:

Code:
mixed depvar indepvars . . . || levelvar: revars, . . ., covar(___)
I would expect that any variable in revars would also have to be in indepvars. To estimate the random effect of a variable, that variable itself would have to be on the right hand side of the model. In HLM7, which I used some time ago, you couldn't specify a random effect without a basic Level-1 effect.

However, Stata doesn't throw an error when I put a variable in revars only. In fact I can put zero variables in indepvars and one variable in revars, and the model runs. I get an estimate of the random effect of that var. What is Stata doing in this case? Is it just estimating whether the intercept randomly varies? And is there a set of equations that can show what is happening at Level 1 and Level 2 here? Thanks and happy new year!