Figure 1.

Left: Coverage of centered percentile bootstrap and normal approximations for constructing confidence sets for $\tau \left(\widehat{c}\right)$. Right: Coverage of centered percentile bootstrap with smoothed target ${\tau }_{smoothed}\left(\widehat{c}\right)\triangleq P{(1+exp\left(aY\widehat{c}\left(X\right)\right))}^{-1}$ for varying values of a; a value of a = ∞ corresponds to $\tau \left(\widehat{c}\right)$. Results are based on 1000 Monte Carlo iterations, target coverage is .950. The performance of the ACI on this example can found in Section 5 under the example labeled “quad.”