Expected Value of 4th Order Statistic

521 Views Asked by At

I am trying to solve for the expected value of fourth order statistic, $$E[X_{(4)}]. $$

Suppose I have the following random variables:

$$ X_1, X_2, ... X_5 $$

With pdf: $$ f(x) = e^{-x} $$ $$ x>0 $$

And their corresponding order statistics:

$$ X_{(1)} < X_{(2)} ... <X_{(5)}$$

First, I have solved for the CDF: $$ F(x) = \int_0^x{f(t)}dt = (1-e^{-x})$$

Using the following equation for the PDF of the kth order statistic, I have found the PDF of the 4th order statistic.

$$f_{x_{(k)}} = n f(x)*\begin{pmatrix} n-1\\k-1\end{pmatrix} *F(x)^{k-1}*(1-F(x))^{n-k}$$

$$f_{x_{(4)}} = 5*e^{-x}*\begin{pmatrix} 5-1\\4-1\end{pmatrix} *(1-e^{-x})^{4-1}*(1-F(x))^{5-4}$$

$$f_{x_{(4)}} = 20e^{-2x}(1-e^{-x})^3 $$

Now to solve for the expected value, $E[X_{(4)}]. $

$$ E[X_{(4)}] = \int_0^{\infty}x*f_{x_{(4)}}dx$$

$$ E[X_{(4)}] = \int_0^{\infty}x*20e^{-2x}(1-e^{-x})^3 dx$$

$$ E[X_{(4)}] = ? $$

First, I am not sure that I can possibly evaluate this integral analytically, and I would be fine to evaluate numerically. However, when I try to evaluate numerically using a tool like Wolfram, it says that the integral does not converge, and it will not solve.

Can somebody please verify that I have taken the correct approach and applied all formulas correctly.

Thank you very much in advance!