0
$\begingroup$

I am trying to finish a problem, my method requires to prove variance of mean equals mean of covariance, but I have trouble proving it. Is it correct? Or more condition needed?

Now I use $$Var(X)=E(X^2)-E(X)^2$$ $$Cov(X_i,X_j)=E(X_iX_j)-E(X_i)E(X_j)$$ but fail to show whether$$Var(E(X_i))=E(Cov(X_i,X_j))$$is true.

id.est.$$\frac{1}{n}\sum_{i=1}^{n}{(E(X_i)-E(X))^2}=\frac{1}{n^2}\sum_{i=1}^{n}\sum_{j=1}^{n}{Cov(X_i,X_j)}$$

It seems that this problem assumes$$E(X_i)=E(X)$$and$$Var(X_i)=Var(X)$$

$X_i$ is a finite set consisting of $n$ real numbers for each $i \in \left\{1,...,n\right\}$.

$X$ means the set of all $n^2$ real numbers from $X_1,...,X_n$.

$\endgroup$
  • 1
    $\begingroup$ Are you trying to ask if $$ \sum_i \left(E(X_i) - \sum_j E(X_j)\right)^2 =\sum_{i,j} Cov(X_i,X_j)$$ or some variant of this? $\endgroup$ – Calvin Khor Jul 12 '18 at 19:45
  • $\begingroup$ @CalvinKhor no... $\endgroup$ – user571299 Jul 12 '18 at 19:52
  • $\begingroup$ Well as written, the answer below is right. Once you take an expectation, there is no more randomness. The expectation of a random variable is not random. Since variance is defined with expectations, the same is also true for variance. Maybe you want to consider conditional expectations? $\endgroup$ – Calvin Khor Jul 12 '18 at 19:53
  • $\begingroup$ @CalvinKhor but $Cov(X_i,X_j)$ is not a constant. $\endgroup$ – user571299 Jul 12 '18 at 20:00
  • $\begingroup$ It is a constant. For each index $i$ and $j$. That is why I introduced a discrete sum which can be interpreted as making the index uniformly random $\endgroup$ – Calvin Khor Jul 12 '18 at 20:02
1
$\begingroup$

Counterexample. Suppose $n=2$, $X_1 = (1,0)$ and $X_2 = (1,0)$, so $X = (1,0,1,0)$. Then $X_1X_2 = (1,0) = X_1$ by multiplying the entries termwise, and under the interpretation that $\mathbb E((x_1,\dots,x_n)) = \frac{1}{n}\sum_{i=1}^n x_i$,

$$ \mathbb E (X_1) = \mathbb E(X_2) = \mathbb E(X) =1/2,$$ $$ \operatorname{Var}X_1 = \operatorname{Var}X_2 = 1/4, $$ $$\operatorname{Cov}(X_1,X_2) = \operatorname{Cov}(X_2,X_1) = E(X_1 X_2)-E(X_1)E(X_2) = 1/4, $$

So $\text{LHS} = 0$ but $\text{RHS} =1/4 $.

$\endgroup$

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.