oh - sorry - the problem was that I don't know how to calculate the intraclass correlation from a logistic model, given the info that SAS prints out.
But I figured out that if you take the intraclass correlation from the corresponding linear model and apply it to the variance from the logistic model, you get a value that is 2 times the correct standard error.
I don't yet know why it works, but it does work, so that's pretty cool