I would like to run a simple t-test on whether means of two groups are significantly different. The data are double-clustered, so I cannot use clttest. What is the easiest way to do this?

The example is something like:

clttest age, by(treated) cluster(region year)