Differentiability of statistical functionals and consistency of the jackknife (Q2366727)

An unknown parameter of an unknown distribution \(F\) can be considered as a function \(T(F)\) on a space of d.f.s. containing \(F\). The estimate of the parameter is obtained by evaluating \(T\) at the empirical d.f. based on i.i.d. samples from \(F\). Various types of differentiability of statistical functionals are studied (Fréchet, Gâteaux), and some interesting examples of statistical functionals are given. The jackknife estimator of the variance is defined using empirical distributions and its consistency is proved. The consistency of the jackknife estimator for \(g\circ T\) is also proved, where \(g\) is a real- valued function on \(\mathbb{R}^ d\) and \(T\) is a \(d\)-vector, whose components are functionals on a certain convex set of \(p\)-dimensional distributions. The results are not limited to the use of the sup-norm metric \(\rho_ \infty\).