Finding the Variance of 100 Trials given a probability density function

394 Views Asked by At

Given a probability density function:

$$\boxed{\begin{array}{c|r:r:r:r:r} x & -10& -5& 0& 5& 10\\ \hline P(x)& 0.1 & 0.2 & 0.3 & 0.2 & 0.2 \end{array}}$$

1) Find the Standard Deviation:

Sorry the makeshift table I made up there might be confusing. But this is how I found the standard deviation:

$\newcommand{\Var}{\mathsf{Var}}\newcommand{\E}{\mathsf {E}}\Var(X) = \E(X^2) - (\E(X))^2$

$(\E(X))^2 = ((-10\cdot 0.1) + (-5\cdot 0.2) + (0\cdot 0.3) + (5\cdot 0.2) + (10\cdot 0.2))^2 = 1$

$\E(X^2) = ((-10^2\cdot 0.1) + (-5^2\cdot 0.2) + (0^2\cdot 0.3) + (5^2\cdot 0.2) + (10^2\cdot 0.2)) = 40$

$\Var(X) = \E(X^2) - (\E(X))^2 = 40 -1^2 = 39$

$\newcommand{\SD}{\mathsf{SD}}\SD(X) = \sqrt{\Var(X)} = \sqrt{39}$


2) If we have 100 independent replications of this experiment with measurements $X_i$ and the average of these 100 random variables is $\bar X$, then calculate $\Var(\bar X)$.

This is where it gets tricky for me. My professor said it was along lines of

$\Var(\bar X) = \Var(\tfrac 1 n \cdot \sum X)$

$\Var(\tfrac 1 n \sum X) = \tfrac 1{n^2}\Var(X_1 + X_2 + \cdots + X_n)$

So do I just multiply the variance I got for 1) by $\tfrac 1{100^2}$?

In other words does $\Var(X_1 + X_2 + \cdots + X_n)$ equal the $\Var(X)$ I found in 1)?


3) Use Chebyshev’s inequality to bound the probability. $\mathsf P\{\operatorname{abs}(\bar X − 1) > 2\}$

I'm completely lost at this one.

1

There are 1 best solutions below

0
On BEST ANSWER

1) Looks technically correct, though I have not checked the calculations.$\newcommand{\Var}{\mathsf{Var}}\newcommand{\E}{\mathsf {E}}$

2) Independence means all covariances are zero, so $\Var(\sum X_k)$ equals $\sum\Var(X_k)$ , and identicallity means that equals: $n\Var(X)$ .

3) Chebyshev’s inequality is: $~\mathsf P\Big\{\big(X-\E(X)\big)^2\geq k^2\Big\}~{\large\leq}~\dfrac{\Var(X)}{k^2}$

Alternatively stated (for $k\geqslant 0$) as: $~\mathsf P\Big\{\big\lvert X-\E(X)\big\rvert \geq k\Big\}~{\large\leq}~\dfrac{\Var(X)}{k^2}$

So...