[Math] Sampling Distribution of sample mean for Poisson Distribution
statistics
I am particularly struggling with part b, I don't know where to begin. For part a, I think the answer is that the sampling distribution is a Poisson(n$\lambda$).
Best Answer
The sampling distribution of a Poisson(λ) distributed random variable is given by:
As you have correctly suggested the sum is Poisson(nλ) and therefore, substituting nλ for λ and $\Sigma x_i$ for $x_i$ the sampling distribution for $\Sigma X_i$ is given by:
Let us first center everything, using $\bar X_k=X_k-\lambda$ and $\bar M=M-\lambda$. Then
$$
\mathrm{Cov}(M,S^2)=\mathbb E(\bar MS^2)=\mathbb E(\bar X_1S^2)=\frac1{n-1}\mathbb E(\bar X_1U),
$$
where
$$
U=\sum\limits_{k=1}^n(\bar X_k-\bar M)^2.
$$
Note that $U$ is a linear combination of products $\bar X_k^2$ and $\bar X_k\bar X_i$ for $i\ne k$. Amongst these products, many will not contribute to the expectation of $\bar X_1U$ since $\mathbb E(\bar X_1\bar X_k\bar X_i)=0$ for every $k\ne i$ and $\mathbb E(\bar X_1\bar X_k^2)=0$ for every $k\ne1$.
Hence, one needs only the coefficient of $\bar X_1^2$ in $U$, which is $c_n=\left(\frac{n-1}n\right)^2+(n-1)\frac1{n^2}=\frac{n-1}n$. This yields $\mathbb E(\bar X_1U)=c_n\mathbb E(\bar X_1^3)$ and $\mathrm{Cov}(M,S^2)=\frac1{n-1}c_n\mathbb E(\bar X_1^3)=\frac1n\mathbb E(\bar X_1^3)$.
Finally, the third central moment of the Poisson distribution with parameter $\lambda$ is $\mathbb E(\bar X_1^3)=\lambda$ hence
$$
\mathrm{Cov}(M,S^2)=\frac\lambda{n}.
$$
For any distribution with unknown mean $\mu$ and variance $\sigma^2$,
$\bar{X}=\frac1n \sum_1^n X_i$ is an unbiased estimate of $\mu$
$ \frac1{n-1}\sum_1^n (X_i-\bar{X})^2$ is an unbiased estimate of $\sigma^2$
and this is also true for a sample Poisson distribution. So you can build your sample mean, sample standard deviation and estimate of the standard error of the mean this way if you wish.
But a property of Poisson distributions is that the population mean and variance are in fact equal, so if your aim is estimating that parameter, then producing these two numbers might not be so helpful. Since the sample mean is a sufficient statistic for the Poisson distribution, you might prefer to use $\bar{X}$ as an estimator of the variance, $\sqrt{\bar{X}}$ as an estimator of the standard deviation, and $\sqrt{\frac{\bar{X}}{n}}$ for the standard error of the mean.
Best Answer
The sampling distribution of a Poisson(λ) distributed random variable is given by:
$$ P(X_i = x_i) = f(x_i) = \frac{e^{-\lambda}\lambda^{x_i}} {x_i!} $$ where $x_i \in \{0, 1, 2 ...\}$
As you have correctly suggested the sum is Poisson(nλ) and therefore, substituting nλ for λ and $\Sigma x_i$ for $x_i$ the sampling distribution for $\Sigma X_i$ is given by:
$$ P(\Sigma X_i = \Sigma x_i) = g(\Sigma x_i) = \frac{e^{-n\lambda}(n\lambda)^{\Sigma x_i}} {(\Sigma x_i)!} $$ where $\Sigma x_i \in \{0, 1, 2 ...\}$
the sampling distribution for the sample mean, $\bar{X}$, is derived using the relation $\Sigma X_i =n\bar{X} $ as follows:
$$ h(\bar{x}) = P(\bar{X} = \bar{x}) = P(\Sigma X_i = n\bar{x}) = g(n\bar{x}) $$
$$ = \frac{e^{-n\lambda}(n\lambda)^{n\bar{x}}} {(n\bar{x})!} $$ where $\bar{x} \in \{0, \frac{1}{n}, \frac{2}{n} ...\}$
For $ n \ne 1$ This is not a poisson distribution as the pdf is not of the form $ \frac{e^{-\lambda}\lambda^{\bar{x}}} {\bar{x}!}$.
The sum distribution is stretched by factor 1/n as can be seen in these plots
(Note that the probabilities are only defined for the plotted circles, trendlines are just for visualising).
N = 2, λ = 3
N = 10, λ = 2