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.