Exercise C.3.7

Let SS be a sample space, and let XX and XX' be random variables such that X(s)X(s)X(s) \ge X'(s) for all sSs \in S. Prove that for any real constant tt,

Pr{Xt}Pr{Xt} \Pr\{X \ge t\} \ge \Pr\{X' \ge t\}

Both of them, expanded, are:

Pr{Xt}=sS:X(s)tPr{s}Pr{Xt}=sS:X(s)tPr{s} \Pr\{X \ge t\} = \sum_{s \in S:X(s) \ge t}\Pr\{s\} \\ \Pr\{X' \ge t\} = \sum_{s \in S:X'(s) \ge t}\Pr\{s\}

Each term of the second sum is present in the first sum, because X(s)X(s)X(s) \ge X'(s). This makes them at least equal. There might be additional terms in the first sum (when X(s)t>X(s)X(s) \ge t > X'(s). Thus it can also be greater.