TABLE 2.

The elements of $y_{\text{test}}$ cross-tabulated by their administered treatment $A_i$ and our model’s estimate of the better treatment $\widehat{d}$(x _i).

	$\widehat{\mathit{d}}$(x i) = 0	$\widehat{\mathit{d}}$(x i) = 1
A i = 0	P	Q
A i = 1	R	S