# Order statistic

• Dec 10th 2012, 10:29 PM
variousbubble
Order statistic
• Dec 20th 2012, 05:33 AM
estatistics
Re: Order statistic
I am a Statistician / psychologist who can help you with your statistics and statistical design in spss, amos, statistica, Econometrics, minitab, stata, r, gpower (in psychometrics, psychology, sociology, marketing, statistics, or psychometrics) as well with the whole project (master, bachelor, dissertation). Regression, ttests, factorial analysis, correlation, time series, cluster analysis, cross tabulation, Discriminant Function Analysis, Hierarchical Linear Modeling, Significance Testing, path analysis, anova, ancova, glm, non-parametric (e.g. bootstrap), reliability, standardization, validity, mediation and moderation analyses can be done. Normality check as well plagiarism check is offered. We can also evaluate questionnaires. Private lessons are offered also using skype. Payment by paypal. email:info@estatistics.eu, facebook: elias estatistics.
• Dec 28th 2012, 11:04 AM
abender
Re: Order statistic
Quote:

Originally Posted by variousbubble
Suppose that $\displaystyle X_1, \ldots, X_5$ are i.i.d. uniform random variables on the interval $\displaystyle (3,6)$. Let $\displaystyle \bar{X}$ be the sample mean, with mean $\displaystyle \mu_{\bar{X}}$ and standard deviation $\displaystyle \sigma_{\bar{X}}$. Find the probability that the first order statistic $\displaystyle Y_1$ and the fifth order statistic $\displaystyle Y_5$ differ from $\displaystyle \mu_{\bar{X}}$ by less than $\displaystyle \sigma_{\bar{X}}$.

$\displaystyle f_X(x)=\frac{1}{b-a}=\frac{1}{6-3} = \frac{1}{3} , \hspace{0.5cm} 3<x<6$

$\displaystyle F_X(x)=\int^{t=x}_{t=3}\frac{1}{3} dt = \frac{t}{3}\vert^{t=x}_{t=3} = \frac{x}{3}-1 , \hspace{0.5cm} 3<x<6$

$\displaystyle \mu = \frac{a+b}{2} = \frac{3+6}{2} = \frac{9}{2} = 4.5$

$\displaystyle \sigma^{\color{red}{2}\color{black}} = \frac{(b-a)^2}{12} = \frac{(6-3)^2}{12} = \frac{9}{12} = 0.75 \implies \sigma=\sqrt{0.75}$

$\displaystyle \mu_{\bar{X}} = \mu = 4.5$

$\displaystyle \sigma_{\bar{X}} = \frac{\sigma}{\sqrt{n}} = \frac{\sqrt{0.75}}{\sqrt{5}} = \sqrt{0.15} \approx 0.3873$

$\displaystyle P\left(\left[|Y_1-\mu_{\bar{X}}| < \sigma_{\bar{X}}\right] \cap \left[|Y_5-\mu_{\bar{X}}| < \sigma_{\bar{X}}\right] \right) \approx P\left(\left[4.1127 < Y_1 < 4.8873\right] \cap \left[4.1127 < Y_5 < 4.8873\right] \right)$

If both the sample min and sample max of the $\displaystyle X_i$s are bounded by 4.1127 and 4.8873, then all $\displaystyle X_i$s from the sample are bounded by 4.1127 and 4.8873.

Thus,

$\displaystyle P\left(\left[4.1127 < Y_1 < 4.8873\right] \cap \left[4.1127 < Y_5 < 4.8873\right] \right) = P\left(\left[4.1127 < X_1 < 4.8873\right] \cap \left[4.1127 < X_2 < 4.8873\right] \cap \cdots \cap \left[4.1127 < X_5 < 4.8873\right]\right)$

Since the $\displaystyle X_i$s are independent, we can rewrite the right-hand side from above as

$\displaystyle P\left(4.1127 < X_1 < 4.8873\right) \cdot P\left(4.1127 < X_2 < 4.8873\right) \cdot \cdots \cdot P\left(4.1127 < X_5 < 4.8873\right)$ ,

or, equivalently,

$\displaystyle \left[P\left(4.1127 < X_1 < 4.8873\right)\right]^5$,

which is,

$\displaystyle \tiny{\left[F_X(4.8873) - F_X(4.1127)\right]^5 = \left[\left(\frac{4.8873}{3}-1\right) - \left(\frac{4.1127}{3}-1\right)\right]^5 \approx 0.2582^5 \approx 0.00115}$.