STAT2006 Assignment 3

Due Date: 15:00, 19th March, 2014

1. (a) Let Y be an exponential random variable with mean 1 and X , 1 + 2 Y, 2 > 0. Find the pdf
of X and remember to state the support of X. X is said to follow a shifted exponential
distribution with location parameter 1 and scale parameter 2 .
(b) Let X1 , X2 , ..., Xn be a random sample which Xi are identically distributed as X. Find the
method-of-moments estimator for 1 and 2 .
(c) When 2 is fixed, show that the likelihood function is strictly increasing in 1 when 1 x(1)
and is equal to zero when 1 > x(1) , where x(1) , min{x1 , x2 , ..., xn } is the sample minimum.
Hence find the maximum likelihood estimator of 1 and 2
2. Let X1 , X2 , ..., Xn be a random sample from N (, 2 ), then the pivotal quantity
(n 1)S 2
2 (n 1), and we can make use of its quantiles a, b to construct a 100(1 )%
2
confidence interval for . The quantiles a, b need to satisfy the constraint
(n 1)S 2
 
G(b) G(a) = Pr a b =1
2
where G is the CDF of 2 (n 1). Obviously there are many possible choices for a and b.

(a) Construct the 100(1 )% confidence interval for in terms of the quantiles a, b defined
above. Let k be the length of the confidence interval. Express k in terms of n, s2 , a and b.
(b) Show that the k is minimized when a, b also satisfy
n a n b
a 2 e 2 b 2 e 2 = 0
Combining with the constraint above, we can numerically solve for the optimal pair of
quantiles a, b to minimize the length of the confidence interval.

3. A test was conducted to determine whether a wedge on the end of a plug fitting designed to hold a
seal onto the plug was doing its job. The data taken were in the form of measurements of the force
required to remove a seal from the plug with the wedge in place (say, X) an the force required
without the plug (say, Y ). Assume that the distributions of X and Y are N (X , 2 ) and
N (Y , 2 ), respectively. Thirteen independent observations of X are
3.01 2.44 3.13 2.92 3.02 2.45 1.81 2.38 2.91 2.56 2.16 2.72 2.77
Thirteen independent observations of Y are
1.53 1.55 1.56 1.63 1.27 1.40 1.92 1.40 1.54 1.09 1.47 1.49 1.38

(a) Construct box-and-whisker diagrams of these data on the same figure. Is there any obvious
differences between these observations? (You may enter the data into a statistical software,
e.g. R and use the built-in functions, e.g. boxplot() inside.)
(b) Further confirm your findings by constructing a two-sided 95% confidence interval for
X Y . Is wedge necessary?

4. A manufacturer sells a light bulb that has a mean life of 1450 hours with a standard deviation of 25
hours. A new manufacturing process is being tested, and there is interest in knowing the mean life
of the new bulbs. How large a sample is requried so that [x 5, x + 5] is an approximate 95%
confidence interval for ? You may assume that the change in the standard deviation is minimal.

1
5. It is reported that in a telephone poll of 2000 adult, 925 of them are nonsmokers, y1 = 600 of
nonsmokers and y2 = 500 of smokers said yes to a particular question. Let p1 , p2 equal the
proportions of nonsmokers and smokers would say yes to this question respectively

(a) Find a two-sided 95% confidence interval for p1 p2 .

(b) Find a two-sided 95% confidence interval for p, the proportion of adult who would say yes to
this question.
Y , S 2 , and S 2 be the respective sample means and unbiased estimates of the variances
6. Let X, X Y
2
obtained from independent samples of sizes n and m from the normal distribution N (X , X ) and
2
X
2 2 2
N (Y , Y ), where X , Y , X , and Y are unknown. If 2 = d, a known constant,
Y

Y 2)(X
(a) Argue that (X 2
Y)
is N (0, 1).
dY /n+ /m
2
(n1)SX (m1)SY2
(b) Argue that 2
dY
+ 2
Y
is 2 (n + m 2).
(c) Argue that the two random variables in (a) and (b) are independent.
(d) With these results, construct a random variable (not depending upon Y2 ) that has a t
distribution and that can be used to construct a confidence interval for X Y .

7. Let X1 , X2 , . . . , Xn be a random sample of size n from an exponential distribution with unknown

mean of = .

(a) Show that the distribution of the random variable W = (2/) ni=1 Xi is 2 (2n).
P

(b) Use W to construct a 100(1 )% confidence interval for .

(c) If n = 7 and x = 93.6, give the endpoints for a 90% confidence iinterval for the mean .

8. Using a random sample of size n from the normal distribution N (, 2 ), find the 100(1
)%

confidence interval for of minimum length based on the statistic T = (X )/(S/ n), where X
and S 2 are the unbiased estimator of and 2 , respectively.