= Solution
Use the usual <clinical trial> sampling assumption that all patients' responses are independent, with fixed positive arm sizes and positive known variances. Marginal <normal distributions> alone would not specify the variance of a treatment contrast without this assumption. Put $\overline Y_i=n_i^{-1}\sum_jY_{ij}$, $v_i=\sigma_i^2/n_i$, and $s_i=\sqrt{v_i+v_0}$. The <sample mean> contrast estimates $\delta_i$ and has <standard error> $s_i$. Thus \b[the one-sided Wald statistics] are
$$
\boxed{W_i=\frac{\overline Y_i-\overline Y_0}{\sqrt{\sigma_i^2/n_i+\sigma_0^2/n_0}},\qquad i=1,2.}
$$
Large positive values of these <Wald test> statistics provide evidence against the respective <null hypotheses>.
Back to article page
