= Chi-squared stochastic ordering
Chi-squared distributions are stochastically increasing in their <statistical degrees of freedom>. For any real $0<r<s$, take <independent> $Q_r\sim\chi_r^2$ and $W\sim\chi_{s-r}^2$. Gamma additivity gives
$$
Q_r+W\sim\chi_s^2,\qquad W>0\ \text{almost surely}.
$$
For <integer> $r,s$, this is the familiar construction by adding extra <independent> squared <standard normal distribution>[standard normal] variables. For noninteger <statistical degrees of freedom>[degrees of freedom], multiplying the gamma <moment-generating functions> proves the same statement.
Under this <coupling> the larger-degree variable is larger on every <sample path>. Hence $F_s(x)\leq F_r(x)$ for every $x$, and in fact the inequality is strict for $x>0$: average $F_r(x-W)$ and use $W>0$ and the strictly positive chi-squared <probability density function>[density]. Therefore every quantile at a fixed <probability> in $(0,1)$ strictly increases with the <statistical degrees of freedom>[degrees of freedom]. Comparing chi-squared critical values must account for this change; the distribution is not fixed while only its label changes.
Back to article page