1
$\begingroup$

How do I use the moment generating function of the chi-square distribution to show that if $x_1\sim\chi^2(n_1), x_2\sim\chi^2(n_2),\dots ,x_k\sim \chi^2(n_k)$ then $y=x_1+x_2+\dots+x_k\sim\chi^2(n_1+n_2+\dots+n_k)$ so indicate that you can add random variables having chi-square distributions by adding their degrees of freedom?

$\endgroup$
3
$\begingroup$

HINTS: Start with the moment generating function of $X \sim \chi^2(n)$: $$ \mathcal{M}_X\left(t\right) = \left(1-2 t\right)^{-n/2} $$ Now, notice that for the sum of $m$ independent random variables $Z = X_1 + X_2 + \cdots+X_n$, the moment generating function is $$ \mathcal{M}_Z(t) = \mathbb{E}\left(\mathrm{e}^{Z t}\right) = \mathbb{E}\left(\mathrm{e}^{\left( X_1 + X_2 + \cdots+X_n\right) t}\right) = \mathbb{E}\left(\mathrm{e}^{X_1 t} \cdot \mathrm{e}^{X_2 t} \cdots\mathrm{e}^{X_n t}\right) $$ Using independence: $$ \mathcal{M}_Z(t) = \prod_{k=1}^n \mathbb{E}\left(\mathrm{e}^{X_k t} \right) = \prod_{k=1}^n \mathcal{M}_{X_k}(t) $$ Now piece these two facts together, and notice that the functional form of $\mathcal{M}_Z(t)$ is that of a $\chi^2$ distribution, and read off its degrees of freedom parameter.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.