Do the mean and variance of a random variable define it completely? In particular,

if $X$ is a random variable, such that $E(X) = \operatorname{Var}(X) = u$,

does that imply that $X \sim \operatorname{Poisson}(u)$ ?

If so, then a proof that a RV is Poisson would require only to show the above relationship which is very useful. Otherwise, what measures can uniquely define a random variable, or a Poisson Random Variable?


No, for example you can take a standard normal distribution, multiply it by $\sqrt u$ and add $u$ to get a different distribution with mean and variance $u$. (You can do this trick with any non-constant starting distribution with finite mean and variance: take a linear transformation to get any mean and variance you want).

What is true is that if all the moments exist and match what they would be for a Poisson distribution (i.e. $\mathbb E(X^k)$ has the correct value for every $k$) then the distribution is Poisson.

[edit] The previous paragraph wouldn't necessarily be true if you were trying to determine whether $X$ was equal to some arbitrary distribution, but it is true for comparing with a Poisson distribution (and many other common distributions). Here is a thread on mathoverflow discussing when it is true.


A general random variable is completely defined by its distribution; if the support of a r.v. is an infinite set, finite number of point parameters like mean and variance can't define the type of a random variable.


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.