1
$\begingroup$

We say that $(x_1,x_2,\dots)$is an infinitely exchangeable sequence of random variables iff for any permutation $\pi$, $p(x_1,\dots,x_n)=p(x_{\pi(1)},\dots,x_{\pi(n)})$.

Let $(x_1,x_2,\dots)$ be an infinite sequence of iid r.v, and $x_0$ an independent r.v. from the sequence.

Then why is $(x_1+x_0,x_2+x_0,\dots)$ infinitely exchangeable but not iid?

|cite|improve this question
$\endgroup$
1
$\begingroup$

Exchangeability is a generalization of IID, so IID implies Exchangeability but not the other way around. As a trivial example take $(X_1,X_2)$ where $X_1$ is bernoulli (-1,1) and $X_2=-X_1$, which is exchangeable but not iid.

Assuming each $x_i$ is independent in your example and non-constant, then it's not IID because $\mbox{Cov}(x_1+x_0,x_2+x_0)=\mbox{Var}(x_0,x_0)\neq 0$ (recall IID would imply 0 covariance).

|cite|improve this answer
$\endgroup$
  • 3
    $\begingroup$ Perhaps you should change the first sentence to "All iid sequences are exchangeable, but the reverse is not true." $\endgroup$ – jaradniemi Apr 13 '16 at 18:50
  • 3
    $\begingroup$ Note that the example given here, while it is exchangeable, is not infinitely exchangeable! $\endgroup$ – kjetil b halvorsen Apr 13 '16 at 19:45

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.