Introduction to Statistics and

Econometrics
Case Problem
SPECIALTY TOYS
Specialty Toys
Specialty faces the decision of how many
Weather Teddy units to order for the coming
holiday season. Members of the management
team suggested order quantities of
15000, 18000, 24000 or 28000 units. The wide
range of order quantities suggested indicate
considerable disagreement concerning the
market potential. The product management
team asks you for an analysis of the stock-out
probabilities for various order quantities,
an estimate of the profit potential, and to help
make an order quantity recommendation.
Specialty expects to sell Weather Teddy for
$24 based on a cost of $16 per unit. If
inventory remains after the holiday
season, Specialty will sell all surplus inventory
for $5 per unit.
After reviewing the sales history of similar
products, Specialtys senior sales forecaster
predicted an expected demand of 20,000
units with a 0.95 probability that demand
would be between 10,000 units and 30,000
units.
Questions
Approximate the demand distribution using
Normal distribution and sketch the distribution.
Compute the probability of a stock-out for the
order quantities suggested by members of the
management team.
Compute the projected profit for the order
quantities suggested by the management team
under three scenarios: worst case in which sales
is 10,000 units, most likely case in which sales is
20,000 units and best case in which sales is
30,000 units.
Questions
One of Specialtys managers felt that the profit potential
was so great that the order quantity should have a 70%
chance of meeting demand and only a 30% chance of any
stock-outs. What quantity would be ordered under this
policy, and what is the projected profit under the three
sales scenarios?
Provide your own recommendation for an order
quantity and note the associated profit projections.
Normal Distribution
20,000
.025
10,000 30,000
.025 .95
At 30,000,
30000 20000
1.96
5102
Mean and Standard deviation are
20000, 5102
x
x
z

=

= = =
=
= =
Stock out situation
The management team suggested order
quantities of 15000, 18000, 24000 or 28000
units.
If order quantity is 15000,
15000 20000
0.98
5102
[ 15000] [ 0.98]
0.3365 0.5 0.8865
x
z
P X P Z

= = =
> = >
= + =
Stock out situation
For order quantity being 18000, the
probability of stock out is 0.6517.
At 24,000, the probability of stock out is
0.2177.
At 28,000, the probability of stock out is
0.0582.
Profit Projection
For order quantity being 15,000, profit
projection is
Sales
Unit Sales Total Cost at $24 at $5 Profit
10,000 240,000 240,000 25,000 25,000
20,000 240,000 360,000 0 120,000
30,000 240,000 360,000 0 120,000

At 18,000,
Sales
Unit Sales Total Cost at $24 at $5 Profit
10,000 288,000 240,000 40,000 -8,000
20,000 288,000 432,000 0 144,000
30,000 288,000 432,000 0 144,000

At 24,000,
Sales
Unit Sales Total Cost at $24 at $5 Profit
10,000 384,000 240,000 70,000 -74,000
20,000 384,000 480,000 20,000 116,000
30,000 384,000 576,000 0 192,000

At 28,000,
Sales
Unit Sales Total Cost at $24 at $5 Profit
10,000 448,000 240,000 90,000 -118,000
20,000 448,000 480,000 40,000 72,000
30,000 448,000 672,000 0 224,000

Profit potential being high
Order quantity should have a 70% chance of meeting demand
and only a 30% chance of any stock-outs.
For Q=22,653, the projected profits under
the three scenarios are
20000
0.52 22,653
5102
Q
z Q

= = =
Sales
Unit Sales Total Cost at $24 at $5 Profit
10,000 362,488 240,000 63,265 -59,183
20,000 362,488 480,000 13,265 130,817
30,000 362,488 543,672 0 181,224

Assignment Problem 3
Are male college students more easily bored than
their female counterparts? This question was
examined in the article Boredom in Young Adults
Gender and Cultural Comparisons (Journal of
Cross-Cultural Psychology, 1991). The authors
administered a scale called the Boredom
Proneness Scale to 97 male and 148 female U.S.
college students. Does the accompanying data
support the research hypothesis that the mean
Boredom Proneness Rating is higher for men than
for women? Test the appropriate hypothesis using
a 0.05 significance level.
Gender Sample
Size
Sample
Mean
Sample
Standard
deviation
Male
Female
97
148
10.40
9.26
4.83
4.68
Solution
0 1 2
1 1 2
1 2
1 2
1 2
1 2
2 2
1 2
1 2
0.05
:
:
97 148
10.40 9.26
4.83 4.68
1.14
1.83
0.3885
1.645
H
H
n n
X X
s s
X X
Z
s s
n n
Z


=
>
= =
= =
= =

= = =
+
=
As , we reject the null hypothesis at
5% level of significance.
Hence it can be concluded that the mean
Boredom Proneness Rating is higher for
men than for women at 5% level of
significance.
Z Z

>
Assignment Problem 2
A manufacturer has just marketed a new appliance
with a one year warranty. The product development
plan anticipates that the cost of meeting the warranty
terms will be Rs 5000 per appliance, on the
average, with a standard deviation of Rs 4000. As the
warranty period expires for the first appliances
sold, the manufacturer will note the actual costs of
warranty servicing (X). Two testing approaches have
been suggested for deciding whether or not the
average warranty costs are exceeding the Rs 5000
target. (1) Wait for warranty data on the first 100
appliances sold and conclude that the average cost
will exceed the target if > Rs 5800. (2) Wait for
warranty data on the first 400 appliances sold and
conclude that the average cost will exceed the target
if > Rs 5400.
A timely conclusion about actual warranty
costs is desirable because the manufacturer
wishes to reduce the warranty terms if
these are proving too generous to be
profitable. On the other hand, an
unnecessary reduction of warranty terms
would adversely affect future sales of the
industrial appliance. Assume that the
warranty data on the initial sales will
constitute random sample observations.
Questions
Define the parameter and the alternatives H
0
and H
1
which
are of interest in this situation.
Sketch the power curves for the two test approaches on the
same graph, assuming that the anticipated standard deviation
of warranty costs is accurate. How do the risks of the two
approaches compare if the mean warranty cost is on target?
If it is Rs1000 higher than the target?
The test approach with n=400 was eventually chosen and the
sample results were sample mean =Rs 5340 and s= Rs 3840.
Using the specified decision rule for this approach, what
conclusion should be drawn about average warranty costs?
Estimate the alpha risk of your test when =Rs 5000. Estimate
the beta risk of your test when the mean warranty cost is
Rs1000 higher than the target.
Hypotheses
Type I error
P[Rejecting H
0
| H
0
is true]
0
1
: 5000
: 5000
H
H

=
>
Type I error
Rule 1
Reject if . Here .

Rule 2
Reject if . Here .
0
H 5800 X >
100 n =
5800| 5000 0.0228 P X ( = > = =

0
H
5400 X >
400 n =
5400| 5000 0.0228 P X ( = > = =

Power of a test
The power of a statistical hypothesis test
is the probability of rejecting the null
hypothesis when the null hypothesis is
false.
Power = (1 - )
Table showing Power
Rule 1 Rule 2
5100 0.0401 0.0668
5400 0.1587 0.5
5700 0.4013 0.9332
6000 0.6915 0.9999
6300 0.9878 1
Power Curve
0
0.2
0.4
0.6
0.8
1
1.2
5000 5100 5400 5700 6000 6300
Rule 1
Rule 2
As , Rule 2 will not reject the null
hypothesis. Therefore, we conclude that
the warranty cost does not exceed Rs
5000.
400 Rs 5340 Rs 3840 n X s = = =
5340 X =
Risks
If is known,
If is unknown,
Beta risk when the mean warranty cost is
Rs 1000 higher than the target.
If is known, is negligible.
If is unknown, is negligible.

0.0228 =

0.0188 =

Assignment Problem 1
An aluminum company is experimenting with a new
design for electrolytic cells in smelter pot rooms. A
major design objective is to maximize a cells
expected service life. Thirty cells of the new design
were started and operated under similar
conditions, and failed at the following ages (in days):
Two items of concern to management are: (1) the
mean service life of this design, and (2) the
comparative performance of this design with the
standard industry design, which is known to have a
mean service life of 1,300 days. Management does not
want to conclude that the new design is superior to
the standard one unless the evidence is fairly strong.
Assuming the distribution of service life of the cell is normal; calculate an
appropriate 95% confidence interval for the mean service life of cells of
the new design. Justify your choice of one or two-sided confidence interval.
Should management conclude that the new design is superior to the
standard one with respect to mean service life? Comment.
It has been suggested that the logarithms of service life are more normally
distributed than the original observations. Take logarithms (to base 10) of
the service-life data and calculate the same type of confidence interval as
in (a) for the mean log-service-life of cells of the new design. Cells of the
standard design are known to have mean log-service-life of 3.095 (to base
10). Can management claim with confidence that the mean log-service-life
is greater for cells of the new design than for cells of the standard design?
Graph histograms of the original data and the log-data. Does the
distribution of log-service-lives appear to be more normal than the
distribution of service-lives, as suggested? Comment
A management objective is to obtain a large total service life for the cells.
Is the mean service life or the mean log-service-life the more relevant
measure here? Explain.
Confidence interval
The confidence interval is
This does not support that the new design is
superior to the standard one.
1
( ) ,
n
s
X t
n

| |

|
\ .
( )
1255.361,
0.05
1379.133 399.0149 (29) 1.699 X s t = = =
Logarithms of service times
The confidence interval is
This does not support that the new
design is superior to the standard one.
0.05
3.120485 0.134747 (29) 1.699
Y
Y s t = = =
( )
3.078,
0
1
2
3
4
5
6
7
8
9
10
500 800 1100 1400 1700 2000 2300 More
F
r
e
q
u
e
n
c
y
Bin
Histogram
0
1
2
3
4
5
6
7
8
9
10
2.79 2.89 2.99 3.09 3.19 3.29 3.39 More
F
r
e
q
u
e
n
c
y
Bin
Histogram