In other words, does $$ E\big[E[f(X)]\big] \ge E\big[f(E[X])\big] $$ holds for a convex function $f(x)$?
If so why? or why not?
In other words, does $$ E\big[E[f(X)]\big] \ge E\big[f(E[X])\big] $$ holds for a convex function $f(x)$?
If so why? or why not?
Copyright © 2021 JogjaFile Inc.
By Jensen, $E[f(X)] >= f(E[X])$. Therefore
$$E[E[f(X)]] = E[f(X)] ≥ f(E[X]) = E[f(E[X])$$
is true.
Any true inequality holds under expectation.
$A\geq B \Rightarrow E_X[A]=A\geq B=E_X[B]$