Example) Mathematical Statistics and Data
Analysis 3RD Edition Chapter 8, Q4
Suppose that X is discrete random variable
with P(X=0)=\frac{2}{3}\theta, P(X=1)=\frac{1}{3}\theta,
P(X=2)=\frac{2}{3}(1-\theta), P(X=3)=\frac{1}{3}(1-\theta). The following
10 independent observations were taken from such a distribution. (3, 0, 2, 1,
3, 2, 1, 0, 2, 1)
a) Find the method of moment estimate of
\theta.
b) Find the approximate standard error for
your estimate.
▷Think first!
- Based on the sample observations with
size 10, what is the sample mean?
- How can you find the expected value in
discrete case?
- From the results from two questions
above, what is your conclusion?
- We can find the sample variance by using
the second moment. So what is variance of the estimate?
▷Solution (a)
Sample mean
=\frac{3+0+2+1+3+2+1+0+2+1}{10}=\frac{15}{10}=\frac{3}{2}
E[X^2]=0\cdot P(X=0) + 1\cdot
P(X=1)+2\cdot P(X=2) + 3\cdot P(X=3)
= 0\cdot \frac{2}{3}\theta + 1\cdot \frac{1}{3}\theta+2\cdot
\frac{2}{3}(1-\theta)+3\cdot \frac{1}{3}(1-\theta)
= -2\theta + \frac{7}{3}
\therefore \frac{3}{2} = -2\theta +
\frac{7}{3} \rightarrow \theta = \frac{5}{12} \rightarrow
\hat{\theta}=\frac{5}{12}
▷Solution (b)
The variance of estimate,
Var(\hat{\theta}) = Var (\frac{7}{6}-\frac{1}{2}\bar{X}) =({\frac{1}{2}})^2
+ Var(\bar{X}) = \frac{1}{4}\cdot \frac{\sigma^2}{n} = hummm... what is the
variance of X?
Variance = \sigma ^2=E[X^2] - [E(X)]^2,
so we need the second moment!
E[X^2]=0^2\cdot P(X=0) + 1^2\cdot
P(X=1)+2^2\cdot P(X=2) + 3^2\cdot P(X=3)
=\theta (\frac{1}{3}-\frac{8}{3}-\frac{9}{3})+ \frac{8}{3}+\frac{9}{3}=
-\frac{16}{3}\theta +\frac{17}{3}
\sigma ^2=E[X^2] - [E(X)]^2 =
-\frac{16}{3}\theta +\frac{17}{3} -(-2\theta+\frac{7}{3})^2 = -2\theta
^2+4\theta+\frac{2}{9}
Let's go back the variance of estimate,
Var(\hat{\theta})=\frac{1}{4}\cdot \frac{\sigma^2}{n}=\frac{1}{40}(-4\theta
^2+4\theta +\frac{2}{9})
Now substituting \theta by its moment
estimate \hat{\theta}, the estimated variance is
Var(\hat{\theta})=\frac{1}{40}(-4\hat{\theta}^2+4\hat{}\theta
+\frac{2}{9}) = <\hat{\theta}=\frac{5}{12}
from (a)>
=\frac{1}{40}(-4\theta ^2+4\theta +\frac{2}{9})=\frac{1}{40}\cdot
\frac{172}{144}=0.029861
\therefore
SE(\hat{\theta})=\sqrt{Var(\hat{\theta})}=\sqrt{0.029861}=0.1728
No comments:
Post a Comment