You are on page 1of 19

# Business

Statistics
Level 3

Model Answers
Series 3 2008 (Code 3009)
Business Statistics Level 3
Series 3 2008

How to use this booklet

Model Answers have been developed by Education Development International plc (EDI) to offer
additional information and guidance to Centres, teachers and candidates as they prepare for LCCI
International Qualifications. The contents of this booklet are divided into 3 elements:

(1) Questions – reproduced from the printed examination paper

(2) Model Answers – summary of the main points that the Chief Examiner expected to
see in the answers to each question in the examination paper, plus
a fully worked example or sample answer (where applicable)

(3) Helpful Hints – where appropriate, additional guidance relating to individual
questions or to examination technique

Teachers and candidates should find this booklet an invaluable teaching tool and an aid to success.

EDI provides Model Answers to help candidates gain a general understanding of the standard required.
The general standard of model answers is one that would achieve a Distinction grade. EDI accepts that
candidates may offer other answers that could be equally valid.

© Education Development International plc 2008

All rights reserved; no part of this publication may be reproduced, stored in a retrieval system or
transmitted in any form or by any means, electronic, mechanical, photocopying, recording or otherwise
without prior written permission of the Publisher. The book may not be lent, resold, hired out or
otherwise disposed of by way of trade in any form of binding or cover, other than that in which it is
published, without the prior consent of the Publisher.
QUESTION 1

(a) Describe the main characteristics of the normal distribution. (4 marks)

The weights of containers used by a company have a normal distribution with mean 50 grams and
standard deviation 5 grams.

(b) The containers are supplied in batches of 100. Calculate the probability that a randomly selected
batch will have a total weight of less than 5120 grams.
(8 marks)

The weight of the contents put in the containers is also normally distributed with mean 250 grams and
standard deviation 6 grams. The weight of the container and contents are independent of each other.

(c) Calculate the probability that the weight of a randomly selected filled container will be more than
318 grams.
(8 marks)

(Total 20 marks)

Page 1 of 17
MODEL ANSWER TO QUESTION 1

(a) Symmetrical, bell shaped curve with mean, median and mode equal.

(b) Mean weight of pack = n × x = 100 x 50 = 5000

Standard deviation for the pack = n × sd 2 = 100 × 52 = 50

x − x 5120 − 5000
z= = = 2.4
sd 50
Table probability for z = 2.4 = 0.992

(c) Weight of pack and container = x1+ 2 = x1 + x 2 = 50 + 250 = 300

Joint Standard deviation = sd12 + sd 22 = 52 + 62 = 61 = 7.81

x − x 318 − 300 18
z= = = = 2.30
sd 7.81 7.81
Table probability = 0.989

Required probability 1 – 0.989 = 0.011

Page 2 of 17
QUESTION 2

(a) Explain what is meant by the sampling distribution of the sample proportion. (4 marks)

The manufacturer of a long life battery is testing a new additive to the process. Before the new additive
was used a random sample of 12 items was selected and tested to destruction. Following the change in
additive a second random sample, of 10 items, was tested to destruction. The results were as follows:

Life in hours
Before new additive 25 28 36 29 27 26 31 40 25 36 28 22
After new additive 29 36 28 31 25 26 32 27 38 25

(b) Test whether the new additive has increased the life of the batteries. (12 marks)

(c) Explain what is meant by a type 2 error. State whether a type 2 error may have been committed in
your conclusions to part (b). Explain your answer.
(4 marks)

(Total 20 marks)

Page 3 of 17
MODEL ANSWER TO QUESTION 2

(a) If a large number of samples of a given size are taken the proportions for samples of size 30 or
more will be normally distributed with the mean of the proportion equal to population proportion and
the standard deviation of the sample proportion.
p (1 − p )
(standard error) = where n is the sample size.
n

(b) Null hypothesis: There has been no change in the life of the batteries.
Alternative hypothesis: The life of the batteries has been increased.

Degrees of freedom = n + m –2 = 12 + 10 –2 = 20
Critical t value = 1.72/2.53

x y x2 y2 (x − x ) (y − y )
2 2

25 29 625 841 19.51 0.49
28 36 784 1296 2.01 39.69
36 28 1296 784 43.34 2.89
29 31 841 961 0.17 1.69
27 25 729 625 5.84 22.09
26 26 676 676 11.67 13.69
31 32 961 1024 2.51 5.29
40 27 1600 729 112.01 7.29
25 38 625 1444 9.51 68.89
36 25 1296 625 43.34 22.09
28 784 2.01
22 484 55.01

353 297 10701 9005 316.917 184.1
Σx Σy Σx2 Σy2 (
Σ x−x ) Σ (y − y )
2 2

29.42 29.70

s=
(
Σ x−x +Σ y− y)
2
( )2

=
316.9 + 184.1
= 5.01
n+m−2 12 + 10 − 2

x −y 29.42 − 29.70 − 0.28
t= = = = 0.13
1 1 1 1 2.14
s + 5.01 +
n m 12 10

Conclusion: the calculated value of t is less than the critical value of t at both the 5% and 1%
significance level, there is insufficient evidence to reject the null hypothesis. The additive does not
increase the average life of the batteries.

(c) A type 2 error is when the null hypothesis is not rejected when it is in fact untrue.
A type 2 error might have been made as the null hypothesis was accepted.

Page 4 of 17
QUESTION 3

(a) Explain the factors which should be taken into consideration when constructing a general price
index such as the Retail Price Index or Consumer Price Index.
(10 marks)

(b) Explain two sampling methods you might use to choose the representative sample of items to
include in the consumer price index. For each sampling method give an advantage and a
disadvantage for using this method compared with other sampling methods.
(10 marks)

(Total 20 marks)

Page 5 of 17
MODEL ANSWER TO QUESTION 3

(a) Prices: It is unlikely that except for a minority of goods and services that a single price exists. Thus,
an average price needs to be constructed based upon a notional average price calculated by the
proportion of consumption made at the different prices.

Choice of Items: Not all the possible items available for consumption can be included in the index as
the processing would be too time complex and time consuming. A representative sample of items
has to taken which reflects the general pattern of consumption. This may be based upon a survey
of consumer expenditure in the preceding year.

Weights: The relative consumption of items needs to be reflected by the index. This can be dealt
with through the survey of consumer expenditure which will show the proportion of expenditure in
each consumption category.

Choice of Index: A wide variety of indices are available and one needs to be chosen. Generally, a
current weighted index based on present consumption patterns and prices is compared with a
historical pattern of consumption and prices. Similar to that below:
ΣWn Pn
x100
ΣWn Po

Choice of Base Year: The year in which the index is to begin has to be chosen. It is possible to
choose a year which after a period of significant price increases which reduces the apparent
increases in prices. The reverse is also true if a year is chosen which is just before a period of
significant price increases.

(b) Some form of random or non-random (quota) stratified sampling would seem the best approach.

Explain how a stratified random sample would be taken by allotting all the relevant items of the
sampling frame to different strata depending on characteristics eg food, fuel, etc. Selection of a
random sample in each strata. Advantages the sampling error can be calculated, the result can be
used in further statistical calculation and it reduces the sampling error. Disadvantages a sampling
frame is necessary, the relevant characteristics need to be identified the sample may be drawn from
a widely dispersed range of products.

Quota sampling involves taking a sample with a given number of people. Quota controls ie the
characteristics that the items should have are employed. These might relate to the “typical” family
and income. The advantages are that the sample is relatively cheap, there is no need for a
sampling frame and no need for call-backs. The disadvantages are there may be bias in the choice
of items, theoretically the standard error cannot be calculated so more complex statistical work is
not possible.

Page 6 of 17
QUESTION 4

A company records its electricity consumption against the temperature for 10 days.

Electricity Temperature 0C
Consumption
(Kilowatt
hours 000)
37 12
42 19
46 14
35 17
49 15
58 10
29 22
63 11
76 8
45 14

(a) Calculate and state the regression equation for electricity consumption based on temperature.

(10 marks)

The calculated value of the Coefficient of Determination is 68.69%.

(b) (i) Explain what the Coefficient of Determination measures and interpret the above figure.
(4 marks)

The critical value of the correlation coefficient at the 0.05 significance level for a sample of 10 items is
0.4733.

(ii) Use the result for the Coefficient of Determination to calculate the correlation coefficient and
test whether the correlation coefficient differs significantly from zero.
(6 marks)

(Total 20 marks)

Page 7 of 17
MODEL ANSWER TO QUESTION 4

(a)
o
Cx KwH 000 y x2 xy
12 37 144 444
19 42 361 798
14 46 196 644
17 35 289 595
15 49 225 735
10 58 100 580
22 29 484 638
11 63 121 693
8 76 64 608
14 45 196 630
142 480 2180 6365
2
Σx Σy Σx Σxy
n ∑ xy − (∑ x )(∑ y )
b=
n ∑ x 2 − (∑ x )2

10 × 6365 − 142 × 480
b=
10 × 2180 − 142 2

−4510
b= = -2.76
1636

∑y ∑x 480 142
a= −b a= − −2.76
n n 10 10

a = 48.0 − −39.15 = 87.15

y = 87.15 - 2.76x

(b) (i) The coefficient of determination measures the percentage of change in one variable
associated with the change in the other variable. In this instance it means that 68.89%
of the change in electricity consumption is associated with the change in temperature
(over the range of temperatures in question).

(ii) The correlation coefficient = 0.6869 = −0.83

Null hypothesis: the correlation coefficient does not differ from zero.
Alternative hypothesis: the correlation coefficient does differ from zero.

The calculated value of the correlation coefficient is greater that the critical value of the
correlation coefficient therefore reject the null hypothesis. The correlation coefficient
does differ from zero.

Page 8 of 17
QUESTION 5

(a) Explain what is meant by and give a business example of:
(i) mutually exclusive events
(ii) conditional probability.
(4 marks)

A company decides to advertise its service on 2 or 3 television channels which have national coverage.
Viewer statistics show that 48% of the population view channel A, 30% view channel B and 25% view
channel C. 12% view channels A and B, 4% view channels A and C and 2% view channels B and C.
1% view all three channels.

(b) If an advertisement was placed on all three channels, using a Venn diagram or otherwise, find the
percentage of the population that would be able to:
(i) have the opportunity to see the advertisement
(ii) see the advert on only one channel.
(7 marks)

(c) If the company only had sufficient funds available to place the advert on 2 channels, assuming the
intention is to maximise the audience, which 2 should it choose and what percentage of the
population would be able to see the advertisement.
(3 marks)

All three channels show a free phone number where viewers can contact the firm for more information.
The number called identifies the channel that the viewer was watching and records identify whether the
caller bought the service. This analysis shows channel A calls represented 40% of respondents and
20% of these bought the service. For channel B the percentages were 35% and 30% respectively and for
channel C 25% and 25% respectively.

(d) (i) Of those that called what percentage bought the service? (3 marks)

(ii) If a person bought the service what is the probability it was a person calling the channel A
number?
(3 marks)

(Total 20 marks)

Page 9 of 17
MODEL ANSWER TO QUESTION 5

(a) (i) Mutually exclusive events are events where the occurrence of one event prevents the
occurrence of a second event. For example, a business project cannot be both
successful and unsuccessful.

(ii) Conditional probability is where the outcome of one event depends upon the outcome of
another event. For example, the probability that a person will be on sick leave may
depend upon their age.

(b)

A
33

11 3
1
B C
17 1 20

(i) P ( A ∪ B ∪ C ) = 86%

(ii) P (one only) 33+17+20 = 70%

(c) P( A ∪ B ) = 66%
P( A ∪ C ) = 69%
P( B ∪ C ) = 53%

Therefore, A and C should be chosen

(d) (i) 0.4 x 0.2 + 0.35 x 0.3 + 0.25 x 0.25
0.08 + 0.105 + 0.0625 = 0.2475 or 24.75%

(ii) Probability bought on Channel A = 0.08 = 0.32
Probability bought 0.2475

Page 10 of 17
QUESTION 6

(a) Explain when you would use the Chi-squared test of significance. (4 marks)

A random sample of consumers was carried out to investigate if there was any relationship between the
type of television bought and the social class of the purchaser. The results were as follows:

Type of Television
Social Class CRT LCD Plasma
A 60 45 25
B 65 70 45
C 155 120 130
D&E 120 85 80

(b) Test whether there is any association between the type of television bought and the social class of
the purchaser.
(12 marks)

(c) By combining the data for the number of people who bought plasma televisions estimate the 99%
confidence interval for the population proportion for plasma televisions bought.
(4 marks)

(Total 20 marks)

Page 11 of 17
MODEL ANSWER TO QUESTION 6

(a) The chi square test tests the degree of association between two variables or randomness goodness
of fit to a known distribution. It is based on the differences between observed data and data that
could be expected to occur.

(b) Null hypothesis: there is no association between the type of television and social class.
Alternative hypothesis: there is association between the type of television and social class.

Degrees of freedom = (R-1)(C-1) = (4-1)(3-1) = 6
Critical chi-squared = 12.59/16.81

Observed data Expected frequencies

Totals

60 45 25 130 52 41.6 36.4
65 70 45 180 72 57.6 50.4
155 120 130 405 162 129.6 113.4
120 85 80 285 114 91.2 79.8
Totals 400 320 280 1000

Contributions to chi-squared

1.231 0.278 3.570
0.681 2.669 0.579
0.302 0.711 2.430
0.316 0.421 0.001 chi-squared = 13.189

Conclusion: the calculated value of chi-squared is greater than the critical value of chi-
squared at 0.05 significance level reject the null hypothesis there is evidence to show
association between type of television and social class. However, the calculated value of chi-
squared is less than the critical value of chi-squared at 0.01 significance level there is
insufficient evidence to show an association.

(c) Proportion of Plasma Televisions: p = 280/1000 = 0.28

p (1 − p ) 0.28(1 − 0.28)
ci = p ± 2.58 = 0.28 ± 2.58
n 1000

= 0.28 ± 0.037 = 0.253 to 0.317

Page 12 of 17
QUESTION 7

(a) Explain the difference between a 1 tailed and a 2 tailed test of significance. (4 marks)

A company has two forms of packaging, A and B for its product. It conducted two random samples of the
customers who had returned the product as damaged identified by type of packaging.

Packaging A Packaging B
Items returned 27 22
Sample size 1000 750

(b) Test whether the proportion returned differs due to the type of packaging used.
(12 marks)

(c) If a 1 tailed test had been carried out, would the conclusion you have reached in part (b) remain the
same? Explain your answer.
(4 marks)

(Total 20 marks)

Page 13 of 17
MODEL ANSWER TO QUESTION 7

(a) A one tail test of significance states the alternative hypothesis in terms of the direction of
change either greater than or less than. A two tail test states the alternative hypothesis
without specifying the direction of change “differs from”.

(b) Null hypothesis: there is no difference in the proportion of items returned due to the
type of packaging.
Alternative hypothesis: there is a difference in the proportion of items returned due
to the type of packaging.

Critical z value 1.96/2.58

p1 = 27/1000 = 0.027, p2 = 22/750 = 0.02933

27 + 22 49
Pooled value of p = = = 0.028
1000 + 750 1750

0.027 − 0.02933 − 0.00223
z= = = -0.28
⎛ 1 1 ⎞ 0.02722(0.02233)
0.028(1 − 0.028)⎜ + ⎟
⎝ 1000 750 ⎠

Conclusion the calculated value of z is less than the critical value of z. There is insufficient
evidence to reject the null hypothesis. The proportion returned does not differ due to the
packaging.

(c) A one tail test would test at z = 1.65, and the solution would fall within the acceptance area.
The conclusion would be the same. The proportion returned does not differ due to the
packaging.

Page 14 of 17
QUESTION 8

(a) Explain the circumstance in which a t test is used in preference to a normal distribution based z test,
when testing for the mean of one sample.
(4 marks)

The mean weight of a random sample of 12 bags of sugar is 0.988 kg with standard deviation 0.015 kg.

(b) Test, at the 0.01 significance level, whether the mean weight of the sample is significantly less than
the specified mean weight of the bags of sugar given as 1.0 kg.
(8 marks)

(c) If the company introduces a new production technique which reduces the standard deviation to 0.09
kg, and assuming the sample size remains 12, what weight should the company set as the target
weight so that 99% of the time the sample mean weight of the bags exceeds 1 kg?
(4 marks)

(d) Explain the impact of a larger sample size upon the value of the standard error.
(4 marks)

(Total 20 marks)

Page 15 of 17
MODEL ANSWER TO QUESTION 8

(a) A ‘t’ test is used when the sample size is small less than 30 and the population standard
deviation unknown compared with a large sample with known population standard deviation.

Null hypothesis: the mean weight of bags does not differ from 1.0 kg.
Alternative hypothesis: the mean weight of bags is less than 1.0 kg.

Degrees of freedom = n-1 = 11
Critical t value = 2.72

x−μ 0.988 − 1.000 − 0.012
t= = = = -2.77
σ n 0.015 12 0.00433

Conclusions: the calculated value of t is greater than the critical value of t. Reject the null
hypothesis, there is evidence to suggest the mean weight of bags is less than
1.0 kg.

(c) Target weight = mean + t × se = 1.0 + 2.72 × .0026 = 1.0 + 0.007 = 1.007kg

σ
se = or p(1 − p)
(d) As the standard error, n n , the larger the sample size the smaller

will be the standard error. For example, the sample size increases by a factor of 4 the SE will half.

Page 16 of 7
Page 17 of 17