Could anyone help me on how to approach to define an unbiased estimator of the difference μ1 μ2, if both symbols represent means from two independent random samples (X1, ... Xn and Y1, ... Yn) with variances σ12 and σ22.