QAM – II

Problem Set – 1
Sampling Distributions, Confidence Interval Estimation, Sample Size Decisions
(Course Instructor: Gaurav Garg)

1) If X1, X2, …, Xn is a random sample from a N(μ, σ) population, write the distribution of following
statistics:
a.

b.

c.

d.
Ans:
a. Chi Square with 1 d.f.
b. F(7, n-7)
c. Chi Square with 1 d.f.
d. F(1,1)

2) In the last election a state representative received 52% of the votes cast. One year after the election
the representative organized a survey that asked a random sample of 300 people whether they
would vote for him in the next election. If we assume that his popularity has not changed, what is the
probability that more than half of the sample would vote for him?
Solution: Given that, 𝝅 = 0.52, n = 300
For large n,
Required Probability is
Prob ( p > 0.5) = Prob
= Prob ( Z > - 0.69338)
= Prob (Z < 0.69 ) approximately
= 0.75490

3) A company services home air conditioners. It has been found that times for service calls follow a
normal distribution with mean 60 minutes and standard deviation 10 minutes. A random sample of 4
service calls was taken. What is the probability that the sample mean service time is more than 65
minutes? What is the probability that more than two calls in the sample take more than 65 minutes?
Solution: Time for service call X ~ N(60, 10), n = 4.
Sample mean
= 0.15866.

Y =Number of service calls taking more than 65 minutes in the sample of 4.

Y~ Bin (4, p), where
Required Probability 0.09146.

4) Suppose 40% Lucknowites have enjoyed Tunde’s Kebab at least once. If a random sample of size 300
is drawn from the population, what is the probability that 44% or fewer in the sample have enjoyed
Tunde’s Kebab at least once?
Ans: 0.9207

5) A carload of steel rods has arrived at Cybermatic Construction Company. The car contains 50,000
rods. Claude Ong, manager of Quality Assurance, directs his crew measure the lengths of 100

Page 1 of 6
 randomly selected rods. If the population of rods has a mean length of 120 inches and a standard
deviation of 0.05 inch, what is the probability that Claude's sample has a mean between 119.985 and
120.0125 inches?
Ans: 0.9925

6) Albert Abbasi, VP of Operations at Ingleside International Bank, is evaluating the service level
provided to walk-in customers. Accordingly, he plans a sample of waiting times for walk-in
customers. If the population of waiting times has a mean of 15 minutes and a standard deviation of 4
minutes, what is the probability that Albert's sample of 64 will have a mean less than 14 minutes?
Ans: 0.0228

7) Pinky Bauer, Chief Financial Officer of Harrison Haulers, Inc., suspects irregularities in the pay roll
system. If 10% of the 5,000 payroll vouchers issued since January 1, 2000, have irregularities, what is
the probability that Pinky's random sample of 200 vouchers will have a sample proportion of
between .06 and .14?
Ans: 0.9412

8) Catherine Chao, Director of Marketing Research, needs a sample of Kansas City households to
participate in the testing of a new toothpaste package. If 40% of the households in Kansas City prefer
the new package, what is the probability that Catherine's random sample of 300 households will
have a sample proportion between 0.35 and 0.45?
Ans: 0.9232

9) Some quality control experiments require destructive sampling. The test to determine whether the
item is defective destroys the item. The cost of destructive sampling often dictates small samples. For
example, suppose a manufacturer of printers for personal computers wishes to estimate the mean
number of characters printed before the print-head fails. Suppose the printer manufacturer tests n =
15 randomly selected print-heads and records the number of characters printed until failure for
each. These 15 measurements (in millions of characters) are listed below followed by EXCEL
summary statistics:
Mean 1.2387
Standard Error 0.0499
Median 1.25
Mode #N/A
Standard Deviation 0.1932
Sample Variance 0.0373
Kurtosis 0.0636
Skewness -0.4913
Range 0.7
Minimum 0.85
Maximum 1.55
Sum 18.58
Count 15

Making the appropriate assumptions, form a 99% confidence interval for the mean number of
characters printed before the print-head fails. If population standard deviation is known as 0.2, form
a 99% confidence interval for the mean number of characters printed before the print-head fails.
Compare the width of the two confidence intervals.
Solution:
n = 15, = 1.2387, s1 = 0.1932,
Page 2 of 6
 Assume that sample has been drawn from normal population. 99% CI Estimate of unknown
population mean μ is given by

or (1.2387 – 0.1485, 1.2387 + 0.1485)

or (1.0902, 1.3872).

When σ =0.2, 99% CI Estimate of unknown population mean μ is given by

or (1.2387 – 0.1332, 1.2387 + 0.1332)

or (1.1055, 1.3719).

For fixed confidence coefficient, the width of the confidence interval is higher in case when
we do not know σ and use sample standard deviation s1 in its place. In such a case we use t
distribution in place of N(0,1) distribution to obtain critical values. The width is higher
because t distribution has larger spread or variability.

10) One of the few negative effects of quitting smoking is weight gain. Suppose that the weight gain in the
12 months following a cessation in smoking is normally distributed with a standard deviation of 6
pounds. To estimate the mean weight gain a random sample of 13 quitters was taken and their gain in
weights (in pounds) are listed here:
16 23 8 2 14 22 18 11 10 19 5 8 15
Determine the 90% confidence interval estimate of the mean 12-months weight gain for all quitters.
Solution:
Sample Size = 13,
sample mean = 13.1538
Population Standard Deviation = 6
90% Confidence Interval for population mean is given by
Z(0.05) = 1.645
Thus, required Confidence Interval is
Or (10.4164, 15.8912).

11) Obtain 95% CI estimate no. of smokers in the IIML campus. Total population is 1500. A sample of 100
has 30 smokers.
Ans: (316, 585)

12) 100 fish are caught from a lake. They were marked and dropped back to the lake. After some time,
when those fish were well mixed with other fish, another sample of 100 fish was taken. This sample
has 15 marked fish. Obtain 95% CI estimate no. of fish in a lake.
Ans: (455, 1250)

13) Brian Vanecek, VP of Operations at Portland Trust Bank, is evaluating the service level provided to
walk-in customers. Accordingly, his staff recorded the waiting times for 64 randomly selected
walk-in customers and determined that their mean waiting time was 15 minutes. Assuming the
population standard deviation to be 4 minutes, construct 90% confidence interval for the population
mean of waiting times.
Ans: (9.46, 34.09)

Page 3 of 6
14) Elwin Osbourne, CIO at GFS, Inc., is studying employee use of GFS e-mail for non-business
communications. A random sample of 200 e-mail messages was selected. Thirty of the messages
were not business related. Construct 90% confidence interval for the population proportion.
Ans: (14.1775, 15.8225)

15) Brian Vanecek, VP of Operations at Portland Trust Bank, is evaluating the service level provided to
walk-in customers. Brian would like to minimize the variance of waiting time for these customers,
since this would mean each customer received the same level of service. Accordingly, his staff
recorded the waiting times for 15 randomly selected walk-in customers, and determined that their
mean waiting time was 15 minutes and that the standard deviation was 4 minutes. Assuming normal
distribution, construct 95% confidence interval for the population variance of waiting times.
Ans: (8.58, 39.79)

16) A telephone company plans to ascertain the conditions of telephone poles in the region it services
and the cost of their repair. There are altogether 10000 poles, a list of which is maintained by the
company. From this list in a pilot survey, a simple random sample of 100 poles was selected without
replacement. Crews were sent out to examine the conditions of the poles selected, and to calculate
the cost of needed repairs. The results of the inspection were as follows: sample average repair cost
= Rs. 830; sample standard deviation of repair cost = Rs. 110. How many poles must be sampled if
the estimate of the total cost of repairing all telephone poles is to be within Rs. 100000 of the true
total cost with probability 90%?
Solution:
Population Size = Total number of poles = N = 10,000
Error of Estimation = |Estimate of Population Total – True Population Total|
= |N x Population Mean – N x Sample Mean|
= N x |Population Mean – Sample Mean|
From the question,
N x |Population Mean – Sample Mean| = 100,000
Or 10,000 x |Population Mean – Sample Mean| = 100,000
Or |Population Mean – Sample Mean| = 10
2
 z 
Required Sample Size n is given by n    /2  ,
where  e 
e = |Population Mean – Sample Mean| = 10
σ = population standard deviation. Since σ is unknown, we use its estimate given
by pilot survey which is 110.
zα/2 = Two sided critical value of N(0,1) distribution for 0.10 as significance level =
1.645
Using the above formula, we get n = 327.429025 or 328 poles approximately.

17) Municipal board of a city wants to get the 15,000 faulty street lights of the city repaired. In order to
get the work done by a private contractor, it wishes to estimate the total expenditure. A pilot sample
of 100 lights was collected which yielded an average repair cost of Rs. 1250 with a standard
deviation of Rs. 250. What should be the appropriate sample size to obtain a reasonably good
estimate of total expenditure within Rs. 250,000 of the true total cost with probability 90%?
Solution:
Population Size = Total number of poles = N = 15,000

Error of Estimation
=|Estimate of Population Total – True Population Total|

Page 4 of 6
 = |N x Population Mean – N x Sample Mean|
= N x |Population Mean – Sample Mean|

250,000 = N x |Population Mean – Sample Mean|

250,000 = 15,000 x |Population Mean – Sample Mean|
Or
|Population Mean – Sample Mean| = 16.6667

Required Sample Size n is given by

where
e = |Population Mean – Sample Mean| = 16.6667
σ = population standard deviation. Since σ is unknown, we use its estimate given by
pilot survey which is 250.

zα/2 = Two sided critical value of N(0,1) distribution for 0.10 as significance level =
1.645
Using the above formula, we get n = 609 poles approximately.

18) A market research group wishes to estimate the average number of customers visiting a
supermarket per day in the month of December 2011. The group wants that its estimated value to be
within 5% of the actual value with a probability of 0.95. The earlier studies indicated that the
coefficient of variation of the number of customers visiting the supermarket in the given month was
24%. The cost of collection of data is Rs. 125/- per day. Without ignoring finite population correction,
calculate the cost involved in estimating the required value. If the research group decides to settle for
an estimate with a probability of 0.90, how much reduction in cost can be achieved?
Solution:
e = 0.05μ, N = 31, α = 0.05, σ/ μ = 0.24

(1 - α)100% Confidence Interval Estimate for unknown population mean μ is given by

(T-h, T+h), where T = sample mean and . When population size is small, we use fpc.
In such a case

or
or

or

Cost = 24 × Rs. 125 = Rs. 3000/-

When α = 0.10,

Cost = 21 × Rs. 125 = Rs. 2625/-

Reduction in Cost = 3000 – 2625 = 375/-

Page 5 of 6