expected value of cross-terms in sample variance derivation

174 Views Asked by At

I have been trying to work through the proof of the population variance estimator using the sample variance. In the 4th line of the proof (link below), $$\sum_{j\neq i} \mathsf E[Y_i\cdot Y_j]$$ ...is simplified in the next line to $(n-1)μ^2$

How does that step come about?
I tried expanding the summation over yj, but to no avail.

https://en.wikipedia.org/wiki/Variance#Sample_variance

1

There are 1 best solutions below

2
On BEST ANSWER

Note: This is for sampling with replacement.

When $i\neq j$ then the random variables are independent and identically distributed, so: $$\mathsf E(Y_i\cdot Y_j) = \mathsf E(Y_i)\mathsf E(Y_j) = \mu^2$$.   Where $\mu$ is the population mean.

So for a particular value of $i$, there are $n-1$ values of $j$ such that $j\neq i$   Hence:

$$\sum\limits_{i=1}^n\left(\ldots -\tfrac 2n \sum\limits_{j\neq i}\mathsf E(Y_i\cdot Y_j)\ldots\right) \\ = \sum\limits_{i=1}^n\left(\ldots -\tfrac 2n (n-1)\mu^2\ldots\right)$$