R help - Sep 2012 - variance of yBar.. for unbalanced a random effects model

Hi All,
I am analyzing a set of data collected by two-stage cluster sampling.  My
model is
y_ij = mu + T_i + e_ij
where T_i is the ith treatment and e_ij is random error for the ijth
individual.  I have MSE_within and MSE_between, which lead to MSE_T for the
model.

Suppose I have balanced data where i=1,...n treatments and all treatments
have m units,
YBARhat.. = estimated mean of y_ij
Var(YBARhat..) = MSE_T / n + MSE_within / nm.

Could you provide me the methods that is used to compute Var(YBARhat..)
when I have unbalanced data?  Any references are welcome.
Thank you very much
Han

	[[alternative HTML version deleted]]