If $K_1, \dots, K_n$ are i.i.d. Poisson distributions with parameter $\beta$ I have worked out that the maximum likelihood estimate is $$\hat\beta (k_1, \dots, k_n) = \frac{1}{n} \sum_{i=1}^n k_i$$ for data $k_1, \dots, k_n$. Therefore we can define the corresponding estimator $$T = \frac{1}{n} \sum_{i=1}^n K_i .$$ My question is how would you work out the variance of this estimator?

In particular, as each $K_i$ follows a Poisson distribution with parameter $\beta$ I know, from the properties of the Poisson, that the distribution $\sum_{i=1}^n K_i$ will follow a Poisson distribution with parameter $n \beta$, but what is the distribution of $T$?

  • 1
    $\begingroup$ You don't need the distribution of $T$ to work out its variance, just the basic properties of variances. $\endgroup$
    – Glen_b
    Jun 15 '13 at 13:57

$T$ is distributed... as a Poisson variable scaled by $n$. Hence the variance of $T$ is $1/n^2 \times n\beta$.


Remember that $$ \mathbb{Var}\left(\sum_{i=1}^n a_i X_i\right) = \sum_{i=1}^n a_i^2\,\mathbb{Var}(X_i) + 2 \sum_{1\leq i<j\leq n} a_i\,a_j\,\mathbb{Cov}(X_i X_j) \, , $$ always. But, if the $X_i$'s are independent, what is the value of $\mathbb{Cov}(X_i X_j)$? That's all you need to answer the question.


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.