P.O.

Box 342-01000
Thika

Email: [email protected]

Web: www.mku.ac.ke

DEPARTMENT OF FINANCE AND ACCOUNTING

COURSE CODE:BMA3102

COURSE TITLE: BUSINESS STATISTICS II

Instructional Material for BBM- Distance Learning

1
 COURSE OUTLINE
PRE –REQUISITE: - BBM 112, BBM 212 and BBM 223.

PURPOSE: - To equip students with data collections skills, correlations, analysis hypothesis
formulation and testing methods

OBJECTIVES: - By the end of this module, the learner should be able to: -

(a) Compute the correlation coefficient, equation of the regression line and the coefficient of
determination.

(b) Test the hypothesis Ho : P = 0

(c) Draw the scatter plots and correlation.

(d) Test the difference between two variances or standard deviations.

(e) Test the difference between sample means, and between two proportions using the z –
test.

(f) State the null and alternative hypothesis

(g) State the five steps used in hypothesis testing.

(h) Test means when  is known using the z- test and chi-square.

(i) Test hypothesis, using the sign test, wilcoxon rank sum test and the Kruskal – Wallis test.

WEEK CONTENT

1 Measure of variation correlation, Pearson product moment coefficient r.

2 Regression analysis. Coefficient of determination. The rank correlation coefficient

3 Sampling theory, types of sampling mean and standard deviation in simple

sampling attributes.

4 Large sample testing the significance for a single proportion.

5 Definition used in hypothesis testing, steps in hypothesis testing – traditional

method. Z-test for a mean.

6 Z – Test for a proportion chi-square testing. CAT 1 sit in.

7 Test of proportions. Test the difference between sample means using z-test; test the
difference between two means for independent samples, using the t-test.

2
8 Testing hypothesis about proportion hypothesis testing of the difference between
proportions. Testing the hypothesis for equity of two variances.

9 Advantages and disadvantages of non parametric methods. Test of hypothesis

using the Wilcoxon rank sum test.

10 Test hypothesis using the Kruskal Wallis test

11 Revision: using revision questions

12 Tutorials and revision consultations

13 Examination and consultation

14 Examination and consultation

Course evaluation:

Cat 1 seat in 1 hour examination from week one to six taking 15% of overall marks.

Cat II distance cat to be handed in week 12 taking 15% of overall marks.

Final examination at the end of the semester 70% overall examination, any one of the
examinations not done will incomplete results

3
 TABLE OF CONTENTS
COURSE OUTLINE………………………………………………………………………………2
TABLE OF CONTENTS............................................................................................................... 4
CHAPTER 1: CORRELATION AND REGRESSION ........................................................... 6
1.1 WHAT IS CORRELATION? ............................................................................................... 6

1.2 PEARSON‟S PRODUCT MOMENT COEFFICIENT OF CORRELATION..................... 6

1.3 REGRESSION ANALYSIS ............................................................................................... 11

1.4 COEFFICIENT OF DETERMINATION ........................................................................... 14

1.5 THE RANK CORRELATION COEFFICIENT (R)........................................................... 16

1.6 TIED RANKINGS .............................................................................................................. 17

1.7 TIME SERIES ANALYSIS ................................................................................................ 18

CHAPTER TWO: SAMPLING THEORY .............................................................................. 23

2.1 SAMPLING: ....................................................................................................................... 23

2.2 TYPES OF SAMPLING ..................................................................................................... 24

2.3 MEAN AND STANDARD DEVIATION IN SIMPLE SAMPLING OF ATTRIBUTES 24

2.4 LARGE SAMPLES: TESTING THE SIGNIFICANCE FOR A SINGLE PROPORTION

................................................................................................................................................... 25

CHAPTER THREE: TEST OF HYPOTHESIS ...................................................................... 27

3.1 INTRODUCTION:- ............................................................................................................ 27

3.2 TRADITIONAL METHOD................................................................................................ 28

3.3 NULL AND ALTERNATIVE HYPOTHESIS .................................................................. 28

3.4 SIMPLE AND COMPOSITE HYPOTHESIS .................................................................... 28

3.5 TEST – STATISTIC ........................................................................................................... 28

4
 3.6 ACCEPTANCE AND REJECTION REGIONS ................................................................ 28

3.7 TYPE I AND TYPE II ERRORS ................................................................................. 29

3.8 ONE-TAILED AND TWO –TAILED TESTS ................................................................... 30

3.9 TESTS OF HYPOTHESIS CONCERNING LARGE SAMPLES ..................................... 31

3.10 TESTS OF HYPOTHESIS INVOLVING LARGE SAMPLES ARE BASED ON THE

FOLLOWING ASSUMPTIONS. ............................................................................................. 32

3.11 TESTING HYPOTHESIS ABOUT POPULATION MEAN ........................................... 32

CHAPTER 4: TEST OF PROPORTIONS............................................................................... 37

4.1 TESTING HYPOTHESIS ABOUT POPULATION PROPORTION................................ 37

4.2 HYPOTHESIS TESTING OF THE DIFFERENCE BETWEEN TWO MEANS ............. 41

4.3 HYPOTHESIS TESTING OF THE DIFFERENCE BETWEEN PROPORTIONS .......... 42

4.3 CHI- SQUARE TEST (Χ2) ................................................................................................. 46

4.4 CHI-SQUARE TEST .......................................................................................................... 47

4.5 TESTING THE HYPOTHESIS FOR EQUALITY OF TWO VARIANCES .................... 53

CHAPTER FIVE: NON PARAMETRIC STATISTICS ......................................................... 57

5.1. ADVANTAGES AND DISADVANTAGES .................................................................... 57

5.2 THERE ARE THREE DISADVANTAGES OF NONPARAMETRIC METHODS: 58

5.3 THE SIGN TEST .......................................................................................................... 58

5.4 THE WILCOXON RANK SUM TEST ....................................................................... 59

5.5 THE WILCOXON SIGNED-RANK TEST ................................................................. 62

5.6 THE KRUSKAL – WALLIS TEST ............................................................................. 62

SAMPLE PAPERS ..................................................................................................................... 67

5
 CHAPTER 1: CORRELATION AND REGRESSION

Purpose:- to equip the learner with methods of correlation and regression used to
determine the relationship between two or more numerical or quantitative variables.

Objectives: After completing this chapter you should be able to compute the following

 Pearson’s moment correlation coefficient

 The coefficient of determinations
 Equation of the regression line
 Draw scatter graph
 Draw the best of fit line
 Compute the Spearman’s rank correlation coefficient

A Statistician collects information for variables, which describe the situation. A variable is a
characteristics or attribute that can assume different values. That is Data are values,
measurements or observations that the variables can assume. Variables whose values are
determined by chance are called random variable.

A collection of data values forms a data set. Each value in the data set is called a data value or a
datum.

1.1 WHAT IS CORRELATION?

In statistics, correlation is a measure of the strength of a linear relationship between two sets of
numbers, i.e. if a change in one number is accompanied by a change in the other. The range of
the correlation coefficient is from -1 to +1. If there is a strong positive linear relationship
between the variables, the value of r will be close +1.

1.2 PEARSON’S PRODUCT MOMENT COEFFICIENT OF CORRELATION

r (PPMCC)

This provides a measure of the strength of association between two variables; one the dependent
variable, the other the independent variable, r, can range from +1 i.e. perfect positive correlation
where the variables change value in the same direction as each other, to -1 i.e. perfect negative
correlation where y decreases linearly as x increases.
There are several possible formulae but practical ones are: -

6
  x  x y  y 
 x  x  y  y 
2 2
r=

Where x = independent variable, = mean of the independent.

y = dependent variable, = mean of the dependent variable.

Or

x 2

= ∑ (x  x) ² = x  i

2 ( )
i
n

Syy = ∑ ( y  y) 2
= y
 2

( yi )

2
i
n

 x y   n
x y
Sxy = ∑ ( x  x) ( y  y) = i i

The correlation coefficient computed from the sample data measures the strength and direction
of a linear relationship between two variables. The symbol for the sample correlation coefficient
is r; the symbol for population correlation coefficient is ρ. The correlation coefficient explained
is called Pearson product moment correlation coefficient (PPMCC), Named after Statistician
Karl Pearson, pioneered the research in this area.
s xy
Gives r=
s xx s yy

Example 1.1: A football team has kept records of hours per month spent training and the goals
scored per month and wish to know if there is any correlation between training hours and goals
scored.

Hours per month x 12 21 27 16 19 22 33 10

Goals per month y 8 11 13 11 9 15 17 12
Table 1.1

Solution using table 1.1

Sum up the values and calculate the mean as column 1 and 2 table 2
160 96
x  20 y  12
8 8

7
Subtract the respective means from the independent and dependent variable to form column three
and four.

x y ( x  x) ( y  y) ( x  x)( y  y) ( x  x) 2 ( y  y) 2
12 8 -8 -4 32 64 16
21 11 1 -1 -1 1 1
27 13 7 2 14 49 4
16 11 -4 -2 8 16 4
19 9 -1 -3 3 1 9
22 15 2 3 6 4 9
33 17 13 5 65 169 25
10 12 -10 0 0 100 0

160 96 127 404 68

Table 1.2

Column five is the product three and four, while column six and seven are squares of the column
three and four respectively.

Sxy = 127 Sxx = 404 Syy = 68

127 127
r=   0  77
404  68 165  75

The correlation coefficient is +0.77 which indicates a reasonably strong positive linear
association between training hours and goals scored.

“Alternative”

Instead of finding the mean first you find the product of both independent and dependent
variable, square the variables respectively as table 3

x y xy x2 y2
12 8 96 144 64
21 11 231 441 121
27 13 351 729 169
16 11 176 256 121
19 9 171 361 81
22 15 330 484 225
33 17 561 1089 289
10 12 120 100 144
160 96 2036 3604 1214
Table1. 3

8
 160  160
Sxx = 3604   404
8

96  96
Syy = 1214   62
8

160  96
Sxy = 2036   116
8

116
r=  0  7329
404  62

Example1.2: The following marks have been obtained by a class of students in statistics (out of
100), table 4

x statistics I 45 55 56 58 60 65 68 70 75 80 85
y statistics II 56 50 48 60 62 64 65 70 74 82 90
Table1.4

Solution

Compute the coefficient of correlation for the above data.

Find the mean of both dependent and independent variable. Then subtract each variable
respectively.
   
717 2 6
x  65.18  65 y  65.5 4 65
11 11 11

Table 1.5, Shows the working of the following working in respective columns

 ( x  x)( y  y)  1401.89 ,  x  x   y  y 
2 2
 1413.63 and  1646.47

1401.89
r=  0  91889  0  92
1413.64 1646.47

9
 x y ( x  x) ( y  y) ( x  x)( y  y) ( x  x) 2 ( y  y) 2
45 56 -20 -9 192.64 407.31 91.12
55 50 -10 -15 158.28 103.67 241.86
56 48 -9 -17 161.10 84.31 307.84
58 60 -7 -5 39.83 51.58 30.75
60 62 -5 -3 18.37 26.85 12.57
65 64 - -1 0.28 0.03 2.39
68 65 2 - -1.54 7.94 0.05
70 70 4 4 21.46 23021 19.84
75 74 9 8 83.01 96.40 71.48
80 82 14 16 243.82 219.58 270.75
85 90 19 24 484.64 392.76 598.02
717 721 1401.89 1413.64 1646.47
Table 1.5

Or “alternative “

Instead of finding the mean first you find the product of both independent and dependent
variable, square the variables respectively as table 6

x y xy x2 y2
45 56 2520 2025 3136
55 50 2750 3025 2500
56 48 2688 3136 2304
58 60 3480 3364 3600
60 62 3720 3600 3844
65 64 4160 4225 4096
68 65 4420 4624 4225
70 70 4900 4900 4900
75 74 5550 5625 5476
80 82 6560 6400 6724
85 90 7650 7225 8100
717 721 48398 48149 48905
Table 1.6

10
  
717 2
s xx  48149   1413  6 3
11

7212  
s yy  48905   1646  7 2
11

717 721    
s xy    
48398 1401 9 0 9
11

1401.89
0  91889 0  92
r= 1413.64 1646.47

The performance of both statistics papers has a strong positive linear relationship which means
the papers are related.

The significance of the correlation coefficient

As stated before the range of the correlation is between -1 and +1. When the value of r is near +1
or -1, there is a strong linear relationship. When the value is near zero, the linear relationship is
weak and nonexistent. Since r is computed from the samples, there are two possibilities when r is
not: either the value of is high enough to conclude that there is a significant linear relationship
between the variables, or the value of r is due to chance.

To make this decision, you use hypothesis testing procedure. The traditional method is:-

 State the hypothesis

 Find the critical value
 Compute the test value
 Make the decision
 Summarize the result

1.3 REGRESSION ANALYSIS

Regression analysis is a statistical technique which can be used for short to medium term
forecasting which seeks to establish the line of “best fit” to be observed data. The data could be
shown as a scatter diagram and the line of best fit. The purpose of the scatter plot, as indicated
above is to determine the nature of the relationship. The possibilities include a positive linear
relationship, a negative linear relationship, a curvilinear relationship, or discernible relationship.
If the value of the correlation coefficient is significant, the next step is to determine the equation
of the regression line which is the data‟s line of best fit. The purpose of the regression line is to
enable the researcher to see the trend and make predictions on the basis of the data.
11
Example 1.3: the following data was collect for the sales of wedges for a period of 7 years

Year 1 2 3 4 5 6 7

Sales of wedges in 14 17 15 23 18 22 27
thousands

Table1.7

As there seven years of readings in table 7, the data are set out as follows

Year Sales of xy x2 y2
x wedges
y
1 14 14 1 196
2 17 34 4 289
3 15 45 9 225
4 23 92 16 529
5 18 90 25 324
6 22 132 36 484
7 27 189 49 729
28 136 596 140 2776
Table 1.8

Draw a scatter diagram?

The general form the equation of any straight line on a graph is y = α + βx, where α and β are
constant and where α represents the fixed element and β represents the slope of the line.

To find the values of α and β, it is necessary to solve two simultaneous equations known as the
normal equations which are

 xy    x    x 2

Where n is the number of pairs of figures?

136  7  28  596  28  140 

596  28  140  544  28  112 

52
28   52    1  86
28

12
 136 28  1.86
  12
7

Regression line y = 12 + 1.86x

Sales (in 000‟s of units) = 12 + 1.86 (no of years).

To use this expression for forecasting, we merely need to insert the number of the year required.
For example 8th year sales = 12 + 1.86(8) = 26.88

The slope of the line, β, is called the regression coefficient.

To transpose the normal equation so as to be able to find α and β

  n xy  x y

 y    x   
n n n x 2  ( x) 2
s xy
α = y  x  =
s xx

7 596 28  136

  1  86
7 140  (28) 2

140 1 86 28

   12  6
7 7

y  12  6  1  86 x

Or

There are several methods for finding the equation of the regression line. Two formulae are
given here. These formulae use the same values that are used in computing the value of the
correlation coefficient. The development of these formulae is beyond the scope of this module.

Formulas for the regression line y = α + βx

y x (xxy)

2

α= ( n( x )  ( x)
2 2

13
 n( xy)  ( x y)
β = n( x 2 )  ( x) 2
 
Where α is the y intercept and β is the slope of the line.

Example 1.4: The values of the regression line for the data are given by

N =6, ∑x = 153 .8 ∑y = 18.7, ∑ xy = 682.77 and ∑ x2 = 5859.26.

Find the equation of the regression line.

Solution

Substituting in the formulae above you get

(18.7)(5859.26)  (153.8)(682.77)
α = (6)(5859.26)  (153.8) 2 = 0.396

(6)(682.77  (153.8)(18.7)
β = (6)(5859.26)  (153.8) 2 = 0.106

hence the equation of the regression line is y = 0.396 + 0.106x

1.4 COEFFICIENT OF DETERMINATION

This measure denoted by r2 (because it is the square of the correlation coefficient r) calculates
what proportion of the variation in the actual values of y May be predicted by change in value
of x .

 (Y Y )
2
exp lained var iation

2 E
r is the ratio =
total var iation  ( y  y)
2

Where YE = estimate of y given by the regression equation for each value of x .

Table 8, example 3 is modified using regression formula obtained yE = 12 + 1.86x to table 9

136
Y   19.43
7
E E E Y Y

Year Sales of Y Y Y (Y  Y ) 2 Y  Y    2

x wedges y
1 14 13.86 -5.57 31.02 -5.43 29.48
2 17 15.72 -3.71 13.76 -2.43 5.90
3 15 17.58 -1.85 3.42 -4.43 19.62

14
 4 23 19.44 0.01 0 3.57 12.74
5 18 21.30 1.87 3.49 -1.43 2.04
6 22 23.16 3.73 13.91 2.57 6.60
7 27 25.00 5.59 31.24 7.57 57.30
136 96.84 133.68
Table 1.9

96 84
r2 =  0  7244  72%
133  68

Alternative formula for r2 is given

(nxY x  Y )
2
2
r =
(n  x 2  ( x) 2 )(n  y 2  ( y) 2 )

(7 596 28 136) 2

=  0  7222
(7(140)  28 2 )(7(2776)  136 2

Pearson Product Moment correlation coefficient =

= 0.8498

Example 1.5

Fit a straight line to the following points

x 0 1 2 3 4
y 1 1.8 3.3 4.5 6.3
Table1. 10

Solution

The regression line is obtained by using the table 11 below

∑
x 0 1 2 3 4 10
y 1 1.8 3.3 4.5 6.3 16.9
xy 0 1.8 6.6 13.5 25.2 47.1
x2 0 1 4 9 16 30
y2 1 3.24 10.89 20.25 39.69
Table 1.11

The gradient (slope) is (the values are obtain from table 14

15
 5 47.1  10  16.9 66.5
   1.33
5  30  (10) 2 50
The y –intercept (fixed value) is

30 1.33  10
   3.34
5 5

y  3  34  1  33x

1.5 THE RANK CORRELATION COEFFICIENT (R)

This provides a measure of the association between two sets of ranked or ordered data R can also
vary from +1, perfect positive rank, Correlation to -1, perfect negative rank correlation. This
coefficient is also known as the spearman rank correlation coefficient.
The formula is as follows

R = 1

6 d 2

n(n 2  1)

Where d  difference between the pairs of ranked values.

n  Number of pairs of rankings

Example1. 6: A group of 8 mathematics students are tested in statistics and calculus. Their
rankings in the tests were: -

Student A B C D E F G H

Statistics 2 7 6 1 4 3 5 8

Calculus 3 6 4 2 5 1 8 7

Table 1.12

Find the Spearman‟s correlation between Statistics and calculus.

16
Solution

The table 13 shows the solution of the above problem by finding the difference and squares in
column four and five respectively

student statistics calculus d d2

A 2 3 -1 1
B 7 6 1 1
C 6 4 2 4
D 1 2 -1 1
E 4 5 -1 1
F 3 1 2 4
G 5 8 -3 9
H 8 7 1 1
22
Table1. 13

6 22
R = 1  0  738
8(64  1)

As the rank correlation coefficient is +0.74, we are able to say that there is a reasonable
agreement between the student‟s performances in the two types of tests.
1.6 TIED RANKINGS
3 4 1
The table above has the students when the student E and F tied same 3rd position so 3
2 2

Example1.7: A group of 8 mathematics students are tested in statistics and calculus. Their
rankings in the tests were: -

student calculus statistics d2

A 3 2 1
B 6 7 1
C 4 6 4
D 2 1 2
E 5 3 2
F 1 3 6
G 8 5 9
H 7 8 1
25
Table1.14

17
A slight adjustment to the formula is necessary if some students obtain the same ranking.
t3  t
The adjustment is , where t is number of tied rankings.
12

And adjusted formula is

t3  t
1  6( d 2  )
R= 12
n(n 2  1)

For example assume that students E and F achieved equal marks in statistics and were given joint
third place.

1 2 3 2
1  6(25  )
R= 2 12  0  69
8(64  1)

As will be seen, the R-value has moved also from +0.74 to 0.69

1.7 TIME SERIES ANALYSIS

Time series analysis uses some form of mathematical or statistical analysis on past data arranged
in a time series, e.g. sales by month for the last ten years. Time series analyses have the
advantage of relative simplicity but certain factors need to be considered.

(a) Are the past data representative, for example, do they contain the results of a
recession/boom, a major shift of taste etc.
(b) Time series methods are more appropriate where short term forecasts are required. Over
the longer period external pressures, internal policy changes make historical data less
appropriate.
(c) Time series methods are best suited to relatively stable situations. Where substantial
fluctuations are common and or conditions are expected to change, then the time series
methods may give relatively poor results.

1.7.1 Time series analysis – moving average

If the forecast for the next month‟s sales say December was actual sales for November, then the
forecasts obtained would fluctuate up and down with every random fluctuation. If he forecast for

18
next month‟s sales was the average of sales for several preceding months then, hopefully, the
random fluctuations would cancel each other out, i.e. would be smoothed away.

Example1. 8

Month Actual 3 monthly 6 12 monthly

sales moving monthly moving
average moving averages.
average
January 450
February 440
March 460 450
April 410 437
May 380 417
June 400 397
July 370 383 423
August 360 377 410
September 410 380 397
October 450 407 388
November 470 443 395
December 490 470 410
January 460 425 424
Table 19

Any month‟s forecast is the average of the preceding, n, month‟s actual sales:

For example, 3 monthly moving average forecasts were prepared as follows:

april‟s forecast = jan sales + feb sales + march sales

3
450  440  460
=  450
3

440 460 410 

May forecast =  437
3

Similar logic applies for 6 and 12 monthly moving averages.

NB

A moving average can be used as a forecast as shown above but when graphing moving averages
it is important to realize that being averages , they must be plotted at the midpoint of the period
to which they relate.

19
1.7.2 CHARACTERISTICS OF MOVING AVERAGES

(a) The different moving averages produce different forecasts.

(b) The greater the number of periods in the moving average, the greater the smoothing
effect.
(c) If the underlying trend of the past data is thought to be fairly constant with substantial
randomness, then the greater number of periods should be chosen.
(d) Alternatively, if there is thought to be some change in the underlying state of the data,
more responsiveness is needed, therefore fewer periods should be included in the moving
average.

1.7.3 LIMITATIONS OF MOVING AVERAGES

(a) Equal weighing is given to each of the values used in the moving average calculations,
whereas it is reasonable to suppose that the most recent data is more relevant to current
conditions.
(b) A n period moving average requires the storage n  1 values to which is added the latest
observation.
(c) The moving average calculation takes no account of data outside the period of the
average, so full use is not made of all the data available.
(d) The use of the unadjusted moving average as a forecast can cause misleading results
when there is an underlying seasonal variation.

Exercise 1

i. In a competition to win a new car, entrants were asked to rank 8 features of the car in
order of importance. The features were labeled A, B, C, D, E, F, G and H. The entries
for a randomly selected insurance salesman and a randomly selected mother with
young children are given below.

Rank 1 2 3 4 5 6 7 8

Insurance A E B H D G F C
salesman

Mother D H A B E F G C

Table 1.15

(a) Calculate spearman’s rank correlation coefficient for these data.

20
 (b) Stating your hypothesis clearly tests, at the 5 % level of significance, whether or not
there is evidence of a correlation.

(c) Explain what could be said about the criteria these two entrants used when making
their choices.

The values of spearman’s rank correlation coefficient between judges’ orders were
0.6190 and 0.8571 respectively.

(d) State, giving a reason, which of these two types of persons you believe the competition
was aimed at.

ii. Given the data, table 16

x 1 2 3 4 5 6 7 8 9

y 9 8 10 12 11 13 14 16 15

Table 1.16

(a) Calculate the coefficient of correlation?

(b) Obtain the line of regression.

(c) Estimate the value of y which should correspond to x = 6.2

iii. Calculate the coefficient of correlation between the values of x and y and obtain the
lines of regression for the following data.

x 78 89 97 69 59 79 61 61

y 125 137 156 112 107 136 123 108

Table 1.17

iv. Two judges in a beauty contest rank the ten competitors in the following order.

X 6 4 3 1 2 7 9 8 10 5

Y 4 1 6 7 5 8 10 9 3 2

Table 1.18

Do the two judges appear to agree in their standard?

21
 Suggested References

i. DAVID S. Moore and George P. McCabe (1993), Introduction to the Practice of

Statistics (second Edition) W.H. FREEMAN AND COMPANY NEW YORK
p94 -117 and p161 -183

ii. Ajit C. Tamhane and Dorothy D. Dunlop (2000) Statistics and Data Analysis (from
Elementary to Intermediate) PRENTICE HALL, Upper Saddle River , p347 - 384

iii. Allan G. Bluman, (2009), Elementary Statistics A Step by Step Approach, (seventh
Edition) MAGRAW –HILL INTERNATIONAL EDITION, P533 – 573

iv. Murray R. Spiegel and Larry J. Stephens (2009) Statistics (fourth Edition) SCHAUM‟S
SERIES P316 -352,

22
 CHAPTER TWO: SAMPLING THEORY

Purpose: to equip the learner with sampling techniques and identifying the confidence
interval for the mean, proportion and standard deviation.

Objective: By the end of his chapter, the learner should be able to:-

a) Find the confidence interval for the mean when σ is known.

b) Determine the maximum sample size for finding a confidence interval for the mean.

c) Find the confidence interval for the mean when σ is known.

d) Find the confidence interval for a proportion.

e) Determine the minimum sample size for finding a confidence interval for a proportion.

f) Find a confidence interval for variance and a standard deviation

2.1 SAMPLING:
Population or universe or census in statistics I used to refer to any collection of individuals or
their attributes or of results of operations which can be numerically specified. A population
containing a finite number of individuals or members is called a finite population. A population
with infinite number of members is known as infinite population.

A part or small section selected from the population is called a sample and the process of such
selection is called sampling. The aim of the theory of sampling is to get as much information as
possible, ideally the whole of the information about the population from which the sample has
been drawn. Given the form of the parent population we would like to estimate the parameters of
the population or specify the limits within which the population parameters are expected to lie
with a specified degree of confidence.

The fundamental assumption underlying most of the theory of sampling is random sampling
which consists in selecting the individuals from the population in such a way that each individual
of the population has the same chance of being selected.

23
2.2 TYPES OF SAMPLING
(i) Purposive sampling in which sampling attributes may be regarded as the drawing of
samples from a population whose members possess the attribute A or not –A. for
example in sampling from a population of men, persons who are smokers and non-
smokers. The choosing or drawing of an individual in sampling may be called an
„event‟ or „trial‟ and the possession of the specified attribute A by the individual
selected a “success”.
(ii) Random sampling -
(iii)Simple sampling we mean random sampling in which each event has the same
probability ρ of success and the probability of an event is independent of the success
or failure of events in the preceding trials. For example counting the number of
success in the throwing of a dice or tossing of a coin is a case of simple sampling.
(iv)Stratified sampling.

2.3 Mean and standard deviation in simple sampling of attributes

A simple sample of , n, members

This is clearly identical with that a series of, n, independent trials and with constant probability ρ
of success. The probabilities of 0, 1, 2, ------ n success are the terms in the binomial expansion of
(q  p) n where (q  1  p)

The distribution so obtained is called sampling distribution of the number of successes in the
sample. The expected value or the mean value, of the number of successes is therefore np and
the standard deviation (also called the standard error) of the number of successes is
By the proportion of successes in a sample, we mean the number of successes divided by the
number of members in the sample. The mean and the standard error of the proportion of
successes can be obtained by dividing the corresponding results for the number of success by n .

Mean of the proportion of successes = p

pq
Standard error of the proportion of success =
n

24
2.4 Large samples: Testing the significance for a single proportion
Suppose large number n of independent x trials. The probability of success in each success in
each trial is p

X ~ B(n, p)
Assuming the hypothesis to be correct

The mean of the sampling distribution is np

The variance of the sampling distribution is npq , q  1  p

For large, n, then, X ~ N (np, npq)

x np
Z is distributed as a standard normal variation.
npq

NB:

(i) We use the proportion of successes p

(ii) For large samples, n   , and small p , the binomial distribution may be approximated
by normal distribution.
(iii)By the testing of a statistical hypothesis is meant to a procedure for deciding whether to
accept or reject the hypothesis. The procedure usually consists in assuming or
accepting the hypothesis as correct and then calculating the probability of getting the
observed or more extreme sample.
(iv)The probability level below which we reject the hypothesis is called the level of
significance usually, two levels are used i.e. 5% and 1%

Example 2.1: In a locality of 18000 families a sample of 840 families was selected, of these 840
families, 206 families were found to have a monthly income of Ksh7000 or less. It is desired to
estimate how many out of the 18000 families have a monthly income of Ksh3000 or less. Within
what limits would you place your estimate.
p = proportion of families having monthly income of Ksh3000
206 103
p   0  245 q  0  755
840 420

Assume that the condition of this problem will give a simple sample.
103
Mean = np = 18000  420  4414.29

25
Standard Error (S.E) of the proportion of families having monthly income of Ksh3000

0.245 0755
 0.015  1.5%
840

Hence taking 0.245 to be the estimate of the families having a monthly income of Ksh3000 or
less the limits are 24.5%  (3  1.5)% = 20% and 29% between 3600 and 5220 families.

Exercise 2
1. A sample of 10 measurements of the diameter of a sphere gave a mean x  438
centimeters (cm) and a standard deviation S = 0.06 cm. find the (a) 95% and (b) 99%
confidence limits for the actual diameter.

2. The intelligence quotients (IQs) of 16 students from one area of a city showed a mean
of 107 and a standard deviation of 10, while IQ of 14 students from another area of the
city showed a mean of 112 and a standard deviation of 8. Is there a significant
difference between IQs of the two groups at significance level of (a) 0.01 and (b) 0.05?

3. The standard deviation of the life times of a sample of 200electric light bulbs in 100
hours. Find the (a) 95% and (b) 99% confidence limits for the standard deviation of all
such electric light bulbs.

4. A survey of 1721 people found that 15.9% of individuals purchase religious book at a
Christian bookstore. Find the 95% confidence of the true proportion of people who
purchase their religious books at a Christian book store.

Suggested References

1. DAVID S. Moore and George P. McCabe (1993), Introduction to the Practice of

Statistics (second Edition) W.H. FREEMAN AND COMPANY NEW YORK p432 -440

2. Ajit C. Tamhane and Dorothy D. Dunlop (2000) Statistics and Data Analysis (from
Elementary to Intermediate) PRENTICE HALL, Upper Saddle River , p197 -208

3. Allan G. Bluman, (2009), Elementary Statistics A Step by Step Approach, (seventh

Edition) MAGRAW –HILL INTERNATIONAL EDITION, p355

4. Murray R. Spiegel and Larry J. Stephens (2009) Statistics (fourth Edition) SCHAUM‟S
SERIES P203 -207,

26
 CHAPTER THREE: TEST OF HYPOTHESIS

Purpose: - The learner should be able to state and test hypothesis to make decision
based on the sample data.
Objective:- After completing this chapter, the learner should be able to:-

a) Define the terms used in hypothesis testing

b) State the null and alternative hypotheses

c) Find critical values for z-test

d) State five steps used in hypothesis testing

e) Test means when σ is known, using the z –test

f) Test proportions, using the z –test

g) Test variances or standard deviations, using the chi- square test

h) Test hypothesis, using confidence intervals

i) Explain the relationship between type I and type II errors and the powers of a test.

3.1 Introduction:-
In hypotheses testing, the researcher must define the population under study, state the particular
hypotheses that will be investigated, give the significance level, select a sample from the
population, collect the data, perform the calculations required for the statistical test, and reach a
conclusion.

Hypotheses concerning parameter such as means and proportions can be investigated. The three
methods used to test hypotheses are:-

 Traditional method

 The p –value method

 The confidence interval method

27
A hypothesis is an assumption, belief, or opinion which may or may not be true. E.g. it may be
believed that a given drug cures 90% of the patients taking it or the average height of soldiers in
the army is 168cms. The testing of a statistical hypothesis is the process by which this belief or
opinion is tested by statistical means. We accept the hypothesis as being true, when it is
supported by the sample data. We reject the hypothesis when the sample data fail to support it.

3.2 Traditional method

Every hypothesis testing situation begins with a statement of the hypothesis. A statistical
hypothesis is a conjecture about a population parameter. This conjecture may or may not be true.
There are two types of statistical hypothesis and alternative hypothesis.

3.3 Null and alternative hypothesis

A null hypothesis, generally denoted by the symbol Ho, is a statistical hypothesis that states tht
there is no difference between a parameter and specific value, or that these arte no difference two
parameters. I.e. Hypothesis to be tested for possible rejection or nullification under the
assumption that it is true, the hypothesis is assigned a numerical value Ho: μ = 150cms.

An alternative hypothesis is any other hypothesis which we are willing to accept when the null
hypothesis Ho is rejected, denoted by H1 or HA
H o :   150cms , then H 1 :   150cms or H 1 :  > 150cms or H 1 :  < 150cms

3.4 Simple and composite hypothesis

A statistical hypothesis is said to be simple hypothesis if it completely specifies the value(s) of
the parameter(s). for example, if we hypothesize that   150cms or p  0.4 , then we have
stated simple hypothesis. If a hypothesis, e.g. if we hypothesize that μ > 150cms, the hypothesis
is composite as it does not assign a specified value to the parameter μ

3.5 Test – Statistic

A statistics which provides a basis for testing a null hypothesis is called a test-statistic. Every
test-statistic has a probability (sampling) distribution which gives the probability of obtaining a
specified value of the test – statistic when the null hypothesis is true. The most commonly used
2
test – statistic are z, t, x orF

3.6 Acceptance and Rejection Regions

All possible values which a test – statistic may assume can be divided into two mutually
exclusive groups: one group consisting of values which appear to be consistent with null
hypothesis, and the other having values which lead to the rejection of the null hypothesis. The
first group is called acceptance region and the second rejection region is also called critical

28
region. The value(s) that separates the critical region from the acceptance region is called the
critical value(s).

3.7 Type I and Type II errors

When testing hypothesis, two types of errors are likely to be made. On the basis of sample
information, we may reject a null hypothesis Ho, when it is, in fact true or we may accept a null
hypothesis Ho, when it is actually false.

Decision Accept Ho Reject Ho (or accept H1)

Ho is true Correct decision (no error) Wrong decision (type I error)
Ho is false Wrong decision (type II error) Correct decision ( no error)
Table 3.1

A legal analogy

In court trial, the supposition of law is that the accused (the defendant) is innocent. (Ho).

After having heard the evidence presented during the trial, the judge arrives at a decision.
Suppose that the accused is, in fact, innocent (Ho is true but the finding of the judge is guilty. The
judge has rejected a true null hypothesis and in so doing has made a type I error. If, on the hand,
the accused is, in fact, guilty (i.e. Ho is false) and finding of the judge is innocent, the judge has
accepted a false null hypothesis and by accepting a false hypothesis, he has committed a type II
error.

These situations may be summarized as under: -

(i) The hypothesis is true but our test rejects it.

(ii) The hypothesis is false but our test accepts it
(iii)The hypothesis is true and our tests accepts it
(iv)The hypothesis is false and our test rejects it.

The probability of making a type I error is conventionally denoted by α and that of committing α
type II error is indicated by β.

α= P (type I error) = p (reject Ho/Ho is true)

β= p (type II error) = p (accept t Ho/Ho is false)

29
 H0:μ=μ1

H0:μ≠μo

Figure 3.1
The probabilities of α and β are the shaded and dotted areas respectively of the distributions
under the null hypothesis and under the alternative hypothesis.
THE SIGNIFICANCE LEVEL
The significance level of a test is the probability used as a standard for rejecting a null hypothesis
Ho is assumed to be true. The value α is known as the size of the critical region, which is most
frequently

  0.05 &   0.01

A test of significance is a rule or procedure by which sample results are used to decide whether
to accept or reject a null hypothesis. A value of the statistic is said to be statistically significant
when the probability of its occurrence under Ho is equal to or less than the significance level α,
that is the value falls in the rejection region Ho in this case is rejected.

3.8 One-tailed and two –tailed tests

Three kinds of problems in test of hypothesis, they include
(i) Two tailed test,
(ii) Right-tailed test and
(iii)Left-tailed test.
Two tailed test is that where the hypothesis about the population mean is rejected for value of x
falling into either tail of the sampling distribution. When the hypothesis about population mean
is rejected only for value of x falling into one of the tails of the sampling distribution, then it is
known as one tailed test. If it is the right tail then it is called right-tailed test or one –sided
alternative to the right and if it is on the left tail then it is one sided alternative to the left and is
called tailed test.

30
 Figure3.2
The following table gives critical values of z for both one- tailed and two- tailed tests at various
levels of significance
Level of 0.10 0.05 0.01 0.005 0.0002
significance
Critical value of -1.28 - -2.33 -2.58 -2.88
z for one tailed 1.650
tests 1.28 2.33 2.58 2.88
1.650
Critical value of -1.645 -1.96 -2.58 2.81 -3.08
z for two tailed And And And And And
tests 1.645 1.96 2.58 2.81 3.08
Table 3.2

3.9 Tests of hypothesis concerning large samples

Though it is difficult to draw a clear-cut line of demarcation between large and small samples, it
is generally agreed that if the size of sample exceeds 30 it should be regarded as a sample. The
tests of significance used for large samples are different from the ones used for small samples for
the reason that the assumptions we make in case of large samples do not hold for small samples.

31
3.10 Tests of hypothesis involving large samples are based on the following assumptions.
(i) The sampling distribution of statistic is approximately normal
(ii) Values given by the samples are sufficiently close to the population value and can be
used in its place for the standard error of the estimate.

3.11 Testing hypothesis about population mean

For testing hypothesis about population mean, the following procedure is adopted: -
(a) State the null hypothesis that there is no contradiction between the sample mean and
population mean.
(b) Find the standard error of the mean by using the formula: -
x
x 
n
(c) Compute limits within which the sample mean will fall at 95% or 99% confidence levels
if the population mean is true.
(d) Find out whether the sample mean does lie within those limits or not. If, the sample mean
lies within those limits then null hypothesis is accepted, otherwise, it is rejected.

Example 3.1: The mean lifetime of a sample of 100 light tubes produced by a company is found
to be 1570 hours with standard deviation of 80 hours. Test the hypothesis that the mean lifetime
of the tubes produced by the company is 1600 hrs.
Solution:
H0:μ=1600
H0:μ≠1600
At 95% level of confidence

Population mean = sample mean  1.96 x

x  80  8
x 
n 100

Population must lie between the following ranges:

1570  1.96(8)
1570  15.7

1554.3 to 1585.7

As the population mean 1600 lies outside these limits so the null hypothesis is rejected.

32
Alterative method
The null hypothesis is that there is no
difference between the sample mean and
hypothetical population mean

Figure 3.3

H0:μ=μo

H0:μ≠μo


x  s
z , where x [since  is unknown for large sample]
x n

1570  1600
z  3.75
80
100
The critical value z  1.96 for a two tailed test at 5% level of significance
Since the computed value of z  3.75 falls in the rejection region

Example 3.2: A child welfare officer asserts that the mean sleep of young babies is 14 hours a
day. A random sample of 64 babies shows that the mean sleep was only 13 hours 30 minutes
with standard deviation of 3 hours. At 5% level of significance, test the assertion that mean sleep
of babies is less than 14 hours a day.
Solution:
H0:μ=14 hours

H0:μ< 14 hours

This is one tailed test as we are interested in the left hand tail of the distribution.
 13 3  14  3
1 2
x  2 8 16
z =    
x 3 3 3 3 9
64 8

33
z Should be within 1.645 at 5% level of significance H1 is accepted

Figure 3.4

I.e. population mean is less than 14 hours a day.

Exercise 3
1. (a) Write short notes on the following: -
(i) Null and alternative hypothesis
(ii) One tailed and two tailed tests
(iii) Type I and type II errors

(iv) Acceptance and rejection regions.

(b) A sample of 100 motorcar tyres has a mean of 20,000 km and a standard deviation
of 800 km. A second sample of 150 tyres has a mean life of 22,000 km and a standard
deviation of 900 km. is it true to say that the two samples were drawn from the same
population?
2. A researcher reports that the average salary of assistant professors is more than
₤42,000. A sample of 30 assistant professors has a mean salary of ₤43,260 at α = 0.05,
test the claim that assistant professors earn more than ₤42,000 per year. The standard
deviation of the population is ₤ 5,230.
3. (a) What is meant by a p- value?
(b) State whether the null hypothesis should be rejected on the basis of the given P-
value
(i) P-value = 0.258, α = 0.05, one tailed test
(ii) P-value = 0.0684, α = 0.10, two tailed test
(iii)P-value = 0.0153, α = 0.01, one tailed test
(iv) P-value = 0.0232, α = 0.05, two tailed test
(v) P-value = 0.002, α = 0.01, one tailed test

34
Suggested References

1. DAVID S. Moore and George P. McCabe (1993), Introduction to the Practice of

Statistics (second Edition) W.H. FREEMAN AND COMPANY NEW YORK
p 455 -458

2. Ajit C. Tamhane and Dorothy D. Dunlop (2000) Statistics and Data Analysis (from
Elementary to Intermediate) PRENTICE HALL, Upper Saddle River , p237 – 256

3. Allan G. Bluman, (2009), Elementary Statistics A Step by Step Approach, (seventh

Edition) MAGRAW –HILL INTERNATIONAL EDITION, p399 -452

4. Murray R. Spiegel and Larry J. Stephens (2009) Statistics (fourth Edition) SCHAUM‟S
SERIES P245- 277,

35
 CAT 1
1. A group of 25 students took examinations in both pure mathematics and statistics. Their
marks out of 150 in mathematics, x, and in statistics, y, were recorded and are
summarised below.
∑ x = 1978, ∑ x2 = 175840 ∑ y = 2123 ∑ y2 = 202257 ∑ xy = 181572

i. Calculate Sxx, Syy and Sxy

[6marks]
ii. Find the product moment correlation coefficient between the marks in pure
Mathematics and Statistics,
[2marks]
iii. Starting your hypotheses clearly tests, at the 5% level of significance, whether or
not there is evidence of a correlation.
[4marks]
iv. State an assumption needed for the test in part (iii) to be valid.
[1mark]
2. Explain the following terms
i. The difference between one tailed and two tailed test
[2marks]
ii. Null and Alternative Hypothesis
[2marks]

3. The breaking strengths of cables produced by manufacturer have a mean of 1800 Kg and
a standard deviation of 100 Kg. by a new technique in the manufacturing process, it is
claimed that breaking strength can be increased. To test this claim, a sample of 50 cables
is tested and it is found that the mean breaking strength is 1850Kg. can we support the
claim at the 0.01 significance level?
[6 marks]
4. Given that H0:  ≤ 15
H1:  > 15

A sample of 40 provides a sample mean of 16.5 and a sample standard deviation of 7

i. At   0.02 , what is the critical value for z and what is the rejection rule?
ii. Compute the value of the test statistics z
[7marks]

36
 CHAPTER 4: TEST OF PROPORTIONS

Purpose: - The learner should be able to test differences between two means,
proportions and variances using z-Test
OBJECTIVES:
After the end of this chapter, you should be able to:-

 Test the difference between sample means using the z –test

 Test the difference between two means for independent samples, using the t –test

 Test the difference between two means for dependent samples

 Test the differences between two proportions.

 Test the difference between two variances or standard deviations

4.1 Testing hypothesis about population proportion

For testing hypothesis about population proportion, the following procedure is adopted: -
(a) State the null hypothesis that there is no contradiction proportion. (The required measure
as found from the sample.)
(b) Select the confidence level on the basis of one tail or two tests.
(c) Compute the standard error of proportion of using the following formula : -
pq
Standard error of proportion  p 
n
Where p = sample proportion, q  1  p , n = sample size.
(d) Compute the values of : p  appropriate number  p
(e) Check where the sample proportion lies. If the sample means lies between the above
limits then null hypothesis is accepted and vice versa.
Example 4.1 A sample of 400 electors selected at random gives 51% majority to political party
XY. Could such a sample have been drawn from population with a 50 – 50 division of political
opinion
Solution:
p  0.5 p̂  0.51
n  400
q  0.5 q̂  0.49

H0: Po=0.5

H0:Po≠ 0.5

37
Po is the population proportion.

Figure 4.1

pq  (0.50)(0.49)
 p    0.025
n 400

At 5 % significance sample proportion should fall within the following limits.

p  1.96 p  0.50  1.96(0.025)  0.50  0.049

Or 0.451 to 0.0549

A sample proportion lies between these limits, the null hypothesis is accepted.

Example 4.2 A sales clerk in a departmental store claims that 60% of the shoppers entering the
store leave without make a purchase. A random sample of 50 shoppers showed that 35 of them
left without buying anything. Are the sample results consistent with the claim of the sales clerk?
Use a level of significance of 0.05

Solution
H0: Po=0.6

H0:Po≠ 0.6

Test: two – tails

35
p  0.7
50

Using the z statistic we have

38
 p p o 0.7 0.60
z   1.44
po qo (0.6)(0.4)
n 50

This is one tailed test so the critical value of z is 1.645 at 5% level of confidence. Since the
computed value of z =1.44 which is less than the critical value of z = 1.645, therefore the null
hypothesis cannot be rejected. Hence based on this sample data, we cannot reject the claim of the
sales clerk.
Example 4.3: A firm purchases a very large quantity of metal off cuts and wishes to know the
average weight an off cut. A random sample of 625 off cuts is weighed and it is found that the
mean sample weight is 150 grams with a sample standard deviation of 30 grams. What is the
estimate of the population mean and what is the standard error of the mean. What would be the
standard error if the sample size was 1225?
Solution:
The sample mean is 150g so that the estimate of the population mean is 150g
x  150  ̂  150 g Since n >30

Where ̂ means best “estimate of” where n=625

s 30
Standard error of the mean =   1.2 gms
n 625

When n  1225
30
sx   0.857 gms.
1225

It will be seen that increasing the sample size reduces the standard error. This accord with
common sense, we would expect a larger sample to be better than a smaller one.
x  150 g And s x  1.2 g

 At 95% confidence level

  x  1.96s x

 150  1.96(1.2)
 150  2.35
This means that we are 95% confident that the populations mean lies within the confidence zone;
somewhere between 147.65gm and 152.35g
At 99% confidence level the limits are   150  2.58(1.2 gms )

39
Range from 146.9 to 153.1gms.
Example 4.4 A random sample of 400 rail passengers is taken and 55% are in favour of proposed
new timetables. With 95% confidence what proportion of all rail passengers are In favour of the
timetables?
Solution:
n  400 p  0.55 q  1  0.55  0.45

As np  220 i.e. well over s , the normal approximation can be used

pq  0.55 0.45
s ps    0.025
n 400

 We are 95% confident that the population proportion is ps  1.96s ps

 0.55  1.96(0.025)
 0.55  0.049

= 0.501 to 0.599

Example 4.5 It is required to test hypothesis that 50% of households have freezers. A random
sample of 400 households found that 54% of the sample had freezers. The significance level is
5%

Solution:
H0: ∏=50% of all house holds.
H0: ∏≠ 50% of all house holds.

pq  0.5 0.5
s ps    0.025
n 400

0.54 0.50
z  1.6
0.025

At the 5% level of significance for a two- tailed the appropriate value 1.96
 As the calculated z score is 1.6 we can say that difference is not significant and that Ho should
not be rejected.

40
4.2 Hypothesis Testing of the Difference between Two Means
Where two random samples are taken, frequently it is required to know if there is a significant
difference between the two means. This hypothesis test follows the general pattern except that,
the standard error calculations differs.

The distribution of sample mean differences is normally distributed and remains normally
distributed whatever the distribution of the populations from which the samples are drawn.
Where n >30 i.e. large samples, the normal area tables are used, when n < 30, the t distribution
applies.

s A 2 sB 2
Standard error of the difference of means = s X A  X B  
nA nB

s A = standard deviation of sample A, size n A

s B = standard deviation of sample B, size n B

X A X B
z
S( X A XB )

Example 4.6 Machine A and machine B produce identical components and it is required to test if
the mean diameter of the components is the same. A random sample of 144 from machine A had
a mean of 36.40mm and a standard deviation of 3.6mm, whilst a random sample of 225 from
machine B had a mean of 36.90mm and a standard deviation of 2.9mm. Are the means
significantly different at the 5% level?

Solutions:

H0: mean of A= mean of B

H0: mean of A≠ mean of B

I.e. as we are not concerned with the direction of the variation this is a two tail test.

3.6 2 2.9 2
s X A XB    0.357
144 225

36.40 36.90
z  1.401 (Ho would be accepted)
0.3569

41
The score for a two failed test at 5% level is  1.96  as the calculated z score of 1.4 is within
this value there is nothing to suggest that there is any difference between.

4.3 Hypothesis Testing of The Difference Between Proportions

In a similar manner it may be required to test the differences between the proportions of a given
attribute found in two random samples.

Sample 1 Sample
2
Sample size n1 n2
Sample proportion of p1 p2
success
Population proportion  1 2
of success
Table 4.1

The null hypothesis will be ∏1 =∏2 i.e. that the two samples are from the same population. This
being so the best estimate of the standard error of the difference of p1 and p2 is given by polling
the samples and finding the pooled sample proportion thus.

p1 n1 p 2 n 2
P
n1  n 2
And the Standard error is

pq pq
s P!  P2  
n1 n2

( p1 p 2 ) ( 1 2 )

z
And s P  P2

But where the null hypothesis is ∏1 =∏2 the second part of he numerator disappears.

Example 4.7 A market research agency take a sample of 1000 people and finds that 200 know of
brand X after an advertising campaign a further sample of 1091 people is taken and it is found
that 240 know of brand X.

Solution:

It is required to know if there has been an increase in the number of people having an awareness
of brand X at the 5% level.

42
 Ho :  2  1
H1 :  2   1 (One tail test)
200 240
P1   0.2 P2   0.22
1000 1091

200 240
P  0.21 q  1  p  1  0.21  0.79
Pooled sample proportion 1000  1091 and

0.21 0.79 0.21 0.79

 s ( P1  P2 )    0.0178
1000 1091

0.20 0.22
z  1.12
0.0178

The critical value for a one-tailed test at the 5% level is -11.64 so that as the calculated value is
within this value we conclude there is insufficient evidence to reject the null hypothesis.

Example 4.8 in random sample of 100 persons taken from village A, 60 are found to be
consuming tea. In another sample of 200 persons taken from village B, 100 persons are found to
be consuming tea. Do the data reveal significant difference between the two villages so far as
habit of taking tea is concerned?

Solution:

Let us take the hypothesis that there is no significant difference between the two as far as the
habit of taking tea is concerned i.e. ∏1 =∏2

We are given

x1 60
P1    0.6, n1  100
n1 100
x 100
P1  2   0.5, n2  200
n 2 200

The appropriate statistics to be used here is given by

p1 n1 p 2 n 2 (0.6)(100) (0.5)(200) 60  100

P    0.53
n1  n 2 100  200 300

43
q  1  0.53  0.47

pq pq  (0.53)(0.47) (0.53)(0.47)
 P!  P2      0.0608
n1 n2 100 200

( p1 p 2 ) 0.6 0.5 0.1

z    1.64
 P  P2 0.0608 0.0608

Accept the null hypothesis since computed value z is less the 5%

Example 4.9 Suppose that you are as a purchase manager for a company. The following
information has been supplied to you by two manufactures of electric bulbs: -

Company A Company
B
Mean life (in hours) 1300 1248
Standard deviation 82 93
(in hours)
Sample size 100 100
Table 4.3

Which brand of bulbs are you going to purchase if you desire to take a risk of 5%?

Solution: let us take the hypothesis that there is no significant difference in the quality of the
brands of bulbs.
i.e.

H o : 1   2
. H 1 : 1   2

x1 x 2 1300  1248 52 52

z     4.19
S1 2
S 2
822
93 2
67.24  86.49 12.399
 2 
n1 n2 100 100

Since our computed value z = 4.19 is greater than critical value of z= 1.96

(5% level) we reject the null hypothesis, hence the quality of two brands of bulbs differ
significantly.

44
Example 4.10: A sample of 100 motor car tyres has a mean of 20,000 miles and a standard
deviation of 800 miles. A second sample of 150 miles has a mean life of 22,000 miles and
standard deviation of 900 miles.

Is it true to say that the two samples were drawn from the same population?

Sample A Sample B
x 20,000 22,000
S2 8002 9002
n 100 150
Table 4.4

Solution:
The samples were drawn from the same sample

H o : 1   2  0
H 1 : 1   2  0 Samples are not drawn from the sample population.

Standard error of difference between means:

2 2
S1 S2 800 2 900 2
      11,800
x1  x 2
n1 n2 100 150

= 108.6

x1 x 2 20000 22000

z 
x1  x 2 108.6
 18.4

z  values  1.96
Under Ho, and at 0.05 level of significance with a two-tailed test, critical

Conclusion: Ho is rejected i.e. samples

could not have been drawn. From the

45
 Same population, since the calculated
z value falls in the rejection region

Figure 4.2

4.3 Chi- Square Test (χ2)

The chi-square is denoted by the Greek letter χ2. It is frequently used as a test statistic in testing a
hypothesis concerning the difference concerning the difference of a sample and a corresponding
set of expected or theoretical frequencies called the number of degrees of freedom (d.f), where
the term degree of freedom represents the number of independent random variables that express
the chi-square.

PROPERTIES

(1) For every increase in the number of degrees of freedom, there is a new χ2 – distribution.

(2) This possesses additive property so that when χ2 and χ22 are independent and have a chi-
square distribution with n1 and n2 degree of freedom, χ21+ χ22 will also be distributed as a chi-
square distribution n1+n2 degrees of freedom.

(3) Where the d.f is 30 and less, the distribution of χ2 is skewed. But, for degrees of freedom
greater than 30 in a distribution, the values of χ2 are normally distributed.

(4) The χ2 function has only one parameter, the number of degrees of freedom. The χ2
distribution is positively skewed on the right, especially when the number of degrees of
freedom is small.

(5) χ2 distribution is a continuous probability distribution which has the value zero at its lower
limit and extends to infinity in the positive direction. Negative value of χ2 is not possible
because the differences between the observed and expected frequencies are always squared.

46
4.4 Chi-square test
The χ2 test is one of the simplest and most widely used non-parametric tests in statistical
work. It makes no assumptions about the population being sampled. The quantity χ2
describes the magnitude of discrepancy between theory and observation, i.e. with the help of
χ2 test we can know whether a given discrepancy between theory and observation can be
attributed to chance or whether it results from the inadequacy of the theory to fit the observed
facts. If χ2 is zero, it means that the observed and expected frequencies completely coincide.
The greater the value of χ2 the greater would be the discrepancy between observed and
expected frequencies.

The formula for computing square is: -

 2

 (O E ) 2

E , where O = Observed frequency

E = Expected or theoretical frequency

The calculated value of χ2 is compared with the table value of χ2 for given degrees of
freedom at specified level of significance. If the calculated value of χ2 is greater than the
table value, the difference between theory and observation is considered to be significant.

On the other hand, if the calculated value of χ2 is less than the table value, the difference
between theory and observation is not considered significant, i.e. it could have arisen due to
fluctuations of sampling.

Conditions for application of χ2 test

Basic conditions must be met in order for chi-square analysis to be applied.

(1) The experimental data (sample observation) must be independent of each other.

(2) The sample data must be drawn at random from the target population

(3) The data should be expressed in original units for convenience of comparison, and not in
percentage or ratio form.

(4) The sample should contain at least 50 observations.

(5) There should not be less than five observations in any one cell.

47
Example 4.12: A random sample of 400 householders is classified by two characteristics:
whether they own a colour television and by what type of householder (i.e. owner –occupier,
private tenant, council tenant). The results of this investigation are given below:

OWNER COUNCIL PRIVATE TOTAL

OCCUPIER TENANT TENANT
Colour TV 150 60 20 230
No colour TV 45 68 57 170
195 128 77 400
Table 4.6

It is required to test at the 5% level, the following hypothesis.

Ho: The two classifications are independent. (I.e. no relation between classes of householder and
colour

TV ownership)

H1: the classifications are NOT independent

The expected frequency in each cell in the table is found by apportioning the total of the type of
house holder in the ratio of colour TV: no occurs TV

230: 170 = 23: 17

The 195 owner occupiers are split in the 23: 17 proportion

195 23 195 17
:  112 : 83
40 40
128 23 128  17
:  74 : 54
40 40
77 23 77  17
:  44 : 33
40 40

48
EXPECTED FREQUENCIES

OWNER COUNCIL PRIVATE TOTAL

OCCUPIER TENANT TENANT
Colour 112 74 44 230
TV
No colour 83 54 33 170
TV
195 128 77 400
Table 4.7

The χ2 calculations can now be made

Observed Expected (O-E) (O-E)2 (O  E) 2

frequencies (O) frequencies E
(E)
150 112 38 1444 12.89
45 83 -38 1444 17.40
60 74 -14 196 2.65
68 54 14 196 3.63
20 44 -24 576 13.09
57 33 24 576 17.45
χ2 =67.11
Table 4.8

To find the appropriate χ2 value from tables, this is done by establishing ѵ, the degree of

freedom. This is found by multiplying the number of columns less one.

Ѵ = (rows-1) (columns-1)

= (2-1) (3-1) = 2 degrees of degrees of freedom.

The value of the cut-off point of χ2 for 2 degrees of freedom is 5.999. As calculated value 67.11
is greater than the table value we reject the null hypothesis and accept there is a connection
between the type of house holder and colour TV ownership.

49
Example 4.13

Observed frequencies 364 376 218 89 33 13 21

Expected frequencies 339 397 234 92 27 6 10
(based on Poisson
distribution)
Table 4.10

The last three classes should be combined together after grouping the position would be as
follows

Observed Expected (O-E) (O-E)2 (O  E) 2

frequencies frequencies E
(O) (E)
364 339 25 625 1.844
376 397 -21 441 1.111
218 234 -16 256 1.094
89 92 -3 9 0.0978
33 27 6 36 1.091
16 7 9 81 5.0625
10.3002
Table 4.11

Ѵ = (2-1) (6-2) = 4

Χ2 value is 11.070 calculated values 10.300, we accept the null hypothesis

Example 23: Test the hypothesis that the number of parts demanded does not depend on the day
of t he week. In a sample study the following information was obtained:

Day Mon Tue Wed Thur Fri Sat Total

No of parts 1124 1125 1110 1120 1126 1115 6270
demanded
Table 4.12

Test the hypothesis that the number of parts demanded does not depend on the day of the week.
(The table value of χ2 for S d.f at 5% level of significance is 11.071

50
Solution:

Ho: The number of parts demanded does not depend on the day of the week

H1: The number depends on the day of the week.

The numbers of spare parts demanded in a week are 6720 and if all days are same we should
6720
 1120
expect 6 i.e. 1120 spare parts on each day of the week.

Days Observed Expected (O-E)2 (O  E) 2

frequencies frequencies E
(O) (E)
Monday 1124 1120 16 0.014
Tuesday 1125 1120 25 0.022
Wednesday 1110 1120 100 0.089
Thursday 1120 1120 0 0
Friday 1126 1120 36 0.032
Saturday 1115 1120 25 0.022
Table 4.13

The table value of χ2 for 5 d.f at 5% level of significance is 11.07. The computed value of χ2 is
much less than the table value. The hypothesis is accepted and we conclude that the demand for
spare parts is independent of the day of the week.

Example4.14: A survey of 320 families with S children each revealed the following distribution.

No of boys 5 4 3 2 1 0
No of girls 0 1 2 3 4 5
No of families 14 56 110 88 40 12
Table 4.14

Is this result consistent with the hypothesis that make and female births are equally probable?

Solution:
1
p
2

On the assumption that male and female births are equally probable, the probability of a male is

The expected number of families can be calculated by the use of binomial distribution.

51
  2 ~ B(5, 1 2 )
f ( x)  5 C x p x q 5 x x  0,1,2,3,4,5
1 , for [ p  q  1 ]
 5 Cx ( ) 5 2
2

To get the expected frequencies, multiply f(x) by the total number N=320

x f(x) Expected No of (O-E)2 (O  E) 2

frequency families
E
N f(x)
0 5C0 ( 1 2) 5  132 10 14 16 1.6

1 5 C1 ( 1 2) 5  5 32 50 56 36 0.72

2 5C2 ( 1 2) 5  10 32 100 110 100 1.0

3 5C3 ( 1 2) 5  10 32 100 88 144 1.44

4 5C 4 ( 1 2) 5  5 32 50 40 100 2.0

5 5C5 ( 1 2) 5  132 10 12 4 0.4

7.16
Table 4.15

2 
 (O  E)
2

 7.16
E

The table value of χ2 for ѵ = 6-1 = 5 at 5% level of significance is 11.07. The computed value of

χ2=7.16 is less than the table value. Therefore, the hypothesis is accepted. Thus, it can be
conclude that male and female births are equally probable.
4.5 Testing The Hypothesis For Equality Of Two Variances
The test for equality of two population variances is based on the variances in two independent by
selected random samples drawn from two normal populations.

Under the null hypothesis σ12 = σ22

2
S1
1
2 2
S1
F 2 , reduces to F 
S2 S22
2 2

Which follows F-distribution with ѵ1 and ѵ2 degrees of freedom, It is convenient to place larger

sample variance in the numerator for computational purpose.

If the computed value of F exceeds the table value of F, we reject the null hypothesis. I.e. The
alternate hypothesis is accepted.

Example 4.15: Two sources of raw materials are under consideration by a company. Both
sources seem to have similar characteristics, but the company is not sure about their respective
uniformity. A sample of 10 lots from source A yields a variance of 225 and a sample of 11 lots
from sources B yields a variance of 200. Is it likely that the variance of source A is significantly
greater than the variance of sources B at α = 0.01?

Solution:
H0 : 1   22
2
Null hypothesis the variance of source A and that of source B are same

H1 : 1   2
2 2

S 12
F 2 , where is S 1  225
2
, S 2  200
2

S2

225
F  1.125
200

53
The table value of F for ѵ1=9 and ѵ2=10 at 1% level of significance is 4.94. Since the computed

value of F is smaller than the table value of F. the null hypothesis is accepted. Hence the
population variances of the two populations are the same.

Example4.16: In a manufacturing process a sample of 10 items is found to have a standard

deviation of 5 mm. the same items are produced by a different process and a sample of 12 items
is found to have a standard deviation of 4.2 mm.

Is the new process more consistent than the old at the 5% level?

Solution: The sample details are

Sample 1 Sample 2
S1  5 S 2  4.2
S 1  25 S 2  17.64
2 2

n1  10 n2  12

Table 4.16

Degrees of freedom = 10-1 = 9 and 12-1 = 11

Null hypothesis H 0 :  1   2 , there is no difference between the population variability‟s.
2 2

Alternative hypothesis H 1 :  2   1 , that the new process is more consistent than the old.
2 2

2
S 25
Fscore  1 2   1.42
S2 17.64

The F score where ѵ1=9 and ѵ2=11, at the 5% level is 2.90.

Therefore, the null hypothesis cannot be rejected. Based on the sample results there is no
significant difference between the two processes.

54
Exercise 4

1. (a) A survey found that the average hotel room rate in Mombasa is KSh7, 500 and the
average room rate in Nakuru is KSh6850. Assume that the data were obtained from
two samples of 50 hotels each and that the standard deviations of the populations are
KSh475 and KSh410 respectively. At α = 0.05, can it be concluded that there is a
significant difference in the rates?

(b) Find the 95% confidence interval for the difference between the means for the data
in example above.

2. Explain the difference between testing a single mean and testing the difference
between two means.

A random sample of size 36 was taken from a population distributed N (μ, 3.92). The value of
Sample mean x was 15.6
the

(a) Find a 90% confidence interval for μ. it is believed that the value of μ is 17.0

(b) Use your confidence interval to comment on this belief.

3. A social historian believes that changes in attitudes to left-handedness around 1980

have led to a change in the number of left handed people less than 40 years. A random
sample of 200 people was asked their age and whether they were left or right handed.
The results are summarized in the table, 36 below.

Left Right
handed handed

Age ≤ 40 17 93

Age > 40 5 85

Table 4.17

Stating your hypothesis clearly test the social historian’s belief. Use a 5% level of
significance.

55
Suggested References

1. DAVID S. Moore and George P. McCabe (1993), Introduction to the Practice of

Statistics (second Edition) W.H. FREEMAN AND COMPANY NEW YORK, p530

2. Ajit C. Tamhane and Dorothy D. Dunlop (2000) Statistics and Data Analysis (from
Elementary to Intermediate) PRENTICE HALL, Upper Saddle River , p 591 – 593

3. Allan G. Bluman, (2009), Elementary Statistics A Step by Step Approach, (seventh

Edition) MAGRAW –HILL INTERNATIONAL EDITION, p399 -452, 589

4. Murray R. Spiegel and Larry J. Stephens (2009) Statistics (fourth Edition) SCHAUM‟S
SERIES

56
 CHAPTER FIVE: NON PARAMETRIC STATISTICS

Purpose: - to Equip the student with information how to use nonparametric tests

OBJECTIVES: By the end of the chapter you should be able to:-

a) State the advantage and advantages of nonparametric methods

b) Test hypothesis, using sign test

c) Test hypothesis, using the Wilcoxon rank sum test.

d) Test hypothesis using the Kruskal Wallis test

e) Compute the Spearman rank correlation Coefficient μ, σ and p.

Introduction

Statistical tests such as z, t and F tests are called parametric tests. Parametric tests are statistical
tests for population parameters such as means, variances and proportions that involve
assumptions about populations from which the sample were selected. However, Statisticians
have developed a branch of statistics known as nonparametric statistics or distribution statistics
free to use when the population from which the samples are drawn is not normally distributed.
Nonparametric statics can also be used to test hypothesis that do not involve specific population
parameters, such as

5.1. Advantages and Disadvantages

There are five advantages that nonparametric methods have over parametric methods:

1. They can be used to test population parameters when the variable is not normally
distributed.

2. They can be used when the data nominal or ordinal.

57
 3. They can be used to test hypotheses that do not involve population parameters.

4. In some cases, the computations are easier than those for the parametric counterparts.

5. They are easy to understand.

5.2 There are three disadvantages of nonparametric methods:

1. They are less sensitive than their parametric counterparts when the assumptions of the
parametric methods met. Therefore, larger differences are needed before the null
hypotheses can be rejected.

2. They tend to use less information than parametric tests

3. They are less efficient than their parametric counterparts when the assumptions of the
parametric methods are met.

5.3 The Sign Test

The simplest nonparametric test, the sign test for single samples, is used to test the value of a
median for a specific sample. When using the sign test, the researcher hypothesizes the specific
value for the median of a population, and then he or she selects a sample of data and compares
each value with the conjectured median. If it is below the conjecture median, it is assigned minus
sign. If the data value is above the conjectured median it is assigned a plus sign. And if it is
exactly the same as the conjectured median, it is assigned zero. Then the number of plus and
minus signs are compared. If the null hypothesis is true, the number of plus signs should be
approximately equal to the number of minus signs. If the null hypothesis is not true, there will
be a disproportionate number of plus or minus signs.

The test value is the smaller number of plus or minus signs

Example5.1. A convenience store owner hypothesis that the median number of scones she sells
per day is 40. A random sample of 20days yields the following data for the number of scones
sold each day.

18 43 40 16 22

30 29 32 37 36

39 34 39 45 28

36 40 34 39 52

58
 At α =0.05, test the owner‟s hypothesis.

Solution

Step 1: State the hypotheses and identify the claim

H0: median = 40(claim) and H1: median ≠ 40

Step 2: Find the critical value. Compare each value of the data with the median. If the value is
greater than the median, replace the value with a plus sign and vice versa. The completed table
follows.

- + 0 - -

- - - - -

- - - + -

- 0 - - +

Table 5.2

Using n =18 (omit the zeros) and α =0.05 for two -tailed test, the critical value is 4.

Step 3: Compute the test value. Count the number of plus and minus signs obtained in step, and
uses the smaller value as he test value. Since there are 3 plus signs and 15 minus signs, 3 is the
test value.

Step 4: Make the decision. Compare the test value 3 with critical value 4. The null hypotheses is
rejected since 3 < 4

Step 5: Summarize the results. There is enough evidence to reject the claim that the median
number of scones sold per day is 40

5.4 The Wilcoxon Rank Sum Test

 Wilcoxon signed Rank test the population of differences assumed to have a symmetric
distribution. In the wilcoxon tests, the values of the data for both samples are combined
and then ranked. If the null hypothesis is true – meaning that there is no difference in the
population distributions – then the values in each sample should be ranked approximately
the same. Therefore, when the ranks are summed for each sample, the sums should be
approximately equal, and the null hypothesis will not be rejected. If there is a large
difference in the sums of the ranks, then the distributions are not identical and the null
hypothesis will be rejected.

59
  The first test to be considered is the wilcoxon rank sum tests for independent samples.
For this test, both sample sizes must be greater than or equal to 10. The formulas needed
for the test are given next.

 Formula for the Wilcoxon Rank Sum Test when samples are independent.

R  R
z
R
Where

n1 (n1 n 2  1)
R 
2

n1n 2 (n1 n 2  1)

 R 
12

R = Sum of ranks for smaller sample size (n1)

n1 = smaller of sample sizes

n2 = larger of sample sizes

n1 ≥ 10 and n2 ≥ 10

Note that if both samples are the same size, either size can be used as n1

Example 5.2: Two independent samples of army and navy recruits are selected and the time in
minutes it takes each recruit to complete an obstacle course is recorded as shown in the
table. At α = 0.05, is there any difference in the times it takes the recruits to complete the
course?

Army 15 18 16 17 13 22 24 17 19 21 26 28

Navy 14 9 16 19 10 12 11 8 15 18 15

Table 5.3

Solution:

Step 1: State the hypothesis and identify the claim

Ho: there is no difference in the times it takes the recruits to complete the obstacle course.

H1: there is a difference in the times it takes the recruits to complete the obstacle course.

60
Step 2: Find the critical value. Since α = 0.05

And this test is a two-tailed test; use the z-values of +1.96 and -1.96 from normal tables.

Step 3: Compute the test value.

(a) Combine the data from the two samples, arrange the combined data in order and rank
each value. Be sure to indicate the group.

Time 8 9 10 11 12 13 14 15 15 16 16 17

Group N N N N N A N A N A N A

Rank 1 2 3 4 5 6 7 8.5 8.5 10.5 10.5 12.5

Time 17 18 18 19 19 21 22 24 25 26 28

Group A N A A N A A A N A A

Rank 12.5 14.5 14.5 16.5 16.5 18 19 20 21 22 23

Table 5.4

(b) Sum the ranks of the group with the smaller sample size. The sample size for navy is
smaller

R = 1+2+3+4+5+7+8.5+10.5+14.5+16.5+21=93

(c) Substitute in the formulas to find the test value

n1 (n1 n2  1) 11(11  12  1)

R    132
2 2

n1n 2 (n1 n 2  1)  (11)(12)(11 12  1) 

 R    264  16.2
12 12

R  R 93  132
z 
R 16.2
 2.41

Step 4: Make the decision. The decision is to reject the null hypothesis, since -2.41 < -1.96

61
Step 5: Summarize the results. There is enough evidence to support the claim that there is a
difference in times it takes the recruits to complete the course.

5.5 THE WILCOXON SIGNED-RANK TEST

When the samples are dependent, as they would be in a before-and-after test using the same
subjects, the wilcoxon signed-rank test can be used in place of the t-test for dependent sample.

Example 29: in a large supermarket, the owner wishes to see whether the number of shop lifting
incidents per day will change if the number of uniformed security officers is doubled. A sample
of 7 days before security is increased and 7 days after the increase shows the number of
shoplifting incidents.

Number of shoplifting
incidents

Day Before After

Monday 7 5

Tuesday 2 3

Wednesday 3 4

Thursday 6 3

Friday 5 1

Saturday 8 6

Sunday 12 4

Table 5.5

Is there enough evidence to support the claim, at α = 0.05 that there is a difference in the number
of shoplifting incidents before and after the increase in security?

Solution:

5.6 THE KRUSKAL – WALLIS TEST

The U test is non parametric test for deciding whether or not two samples come from the same
population. A generalization of this for k samples is to be provided by the Kruskal-Wallis H test.
Suppose that we have k samples of sizes N1, N2… Nk with the total size of all samples taken
together being by N = N1+N2+………. +Nk. suppose further that the data taken together are

62
ranked and that the sums of the ranks for the samples are R1, R2… Rk respectively. If we define
statistic
2


k
Rj

12
H  3( N  1)
N ( N  1) j 1 N j

2 2 2
12 R R R
 ( 1  2  ................  k )  3( N  1)
N ( N  1) n1 n2 nk

Where

R1 = sum of ranks in sample 1.

n1 = size of sample 1 .etc.

Then it can be shown that the sampling distribution of H is very nearly a chi-square distribution
with k-1 degrees of freedom, provided N1, N2, ….Nk are all at least 5.

The H test provides a non parametric method in the analysis of variance for one-way
classification, or one- factor experiments, and generalizations can be made.

Example5.3: a company wishes to purchase one of five different machines: A, B, C, D or E. in

an experiment designed to determine whether these is a performance difference between the
machines, five experienced operators each work on the machines for equal times. Table below
shows the no of units produced by each machine. Test the hypothesis that there is no difference
between the machines at the (a) 0.05 and (b) 0.01 significance levels.

A 68 72 77 42 53

B 72 53 63 53 48

C 60 82 64 75 72

D 48 61 57 64 50

E 64 65 70 68 53

Table 5.6

Solution:

Since there are five samples (A, B, C, D, and E)

k=5

63
And since each sample consists of five values, we have N1 =N2 =N3 = N4 = N5 and N = 5 x 5 =
25.

By arranging all the values in increasing order of magnitude and assigning appropriate ranks to
the ties, we replace above table with the table below.

Sum of ranks

A 17.5 21 24 1 6.5 70

B 21 6.5 12 6.5 2.5 48.5

C 10 25 14 23 21 93

D 2.5 11 9 14 4 40.5

E 14 16 19 17.5 6.5 73

Table 5.7

R1 = 70, R2 = 48.5, R3 = 93, R4 = 40.5, R5 = 75

Thus

12 70 2 48.5 2 93 2 40.5 2 73 2
H [     ]  3(25  1)
25(25  1) 5 5 5 5 5
 6.44

For k-1 = 4 degrees of freedom at the 0.05 significance level.

0.95
2
 9.49
We have since 6.44< 9.49

We cannot reject the hypothesis of no difference between machines at the 0.05 level and
certainly cannot reject it at the 0.01 level. Therefore, we can either accept the null hypothesis or
reserve the judgment, that there is no difference between the machines at both levels.

Exercise 5

1. Referring to the table below, test the hypothesis Ho that there is no difference between
machines I and II against the alternative hypothesis H1 that there is a difference at the
0.05 significance level.

Day 1 2 3 4 5 6 7 8 9 10 11 12

64
 Machine I 47 56 54 49 36 48 51 38 61 49 56 52

Machine 71 63 45 64 50 55 42 46 53 57 75 60
II

Table 5.8

2. Work out problem 1. Using a normal approximation to the binomial distribution.

3. A sample of 40 grades from a statewide examination is shown in table below. Test the
hypothesis at the 0.05 significance level that the median grade for all participants is (a)
66 and (b) 75. Work the problem first by hand, supplying all the details for the sign
test. Follow this with the MINITAB solution to the problem.

71 67 55 64 82 66 74 58 79 61

78 46 84 93 72 54 78 86 48 52

67 95 70 43 70 73 57 64 60 83

73 40 78 70 64 86 76 62 95 66

Table 5.9

4. Referring to the table below, determine whether there is a difference the 0.05
significance level between cables made of alloy I and alloy II. Work the problem first
by hand, supplying all the details for the Mann-Witney U test. Follow this with the
MINITAB solution to the problem.

Alloy I Alloy II

18.3 16.4 22.7 17.8 12.6 14.1 20.5 10.7 15.9

18.9 25.3 16.1 24.2

19.6 12.9 15.2 11.8 14.7

5.A company wishes to purchase one of five different machines A, B, C, D, OR E. in an

experiment designed to determine whether there is a performance difference between
the machines, five experienced operators each work on the machines for equal times.
Table below shows the number of units produced by each machine. Test the hypothesis
that there is no difference between the machines at the (a) 0.05 and (b) 0.01
significance levels. Work the problem first by hand, supplying all the details for the
Kruskal-Wallis H test. Follow this with the MINITAB solution to the problem.

65
 A 68 72 77 42 53

B 72 53 63 53 48

C 60 82 64 75 72

D 48 61 57 64 50

E 64 65 70 68 53

Table 5.11

Suggested References

1. DAVID S. Moore and George P. McCabe (1993), Introduction to the Practice of

Statistics (second Edition) W.H. FREEMAN AND COMPANY NEW YORK. p562-570
2. Ajit C. Tamhane and Dorothy D. Dunlop (2000) Statistics and Data Analysis (from
Elementary to Intermediate) PRENTICE HALL, Upper Saddle River p 580 -594

3. Allan G. Bluman, (2009), Elementary Statistics A Step by Step Approach, (seventh

Edition) MAGRAW –HILL INTERNATIONAL EDITION, p669 -692
4. Murray R. Spiegel and Larry J. Stephens (2009) Statistics (fourth Edition) SCHAUM‟S
SERIES

66
SAMPLE PAPERS

BMA3102: STATISTICS II ASSIGNMENT

Answer all questions

1. A random sample X1, X2,… X10 is taken from a normal population with mean 100 and
standard deviation 14.
a) Write down the distribution of X , the mean of this sample. (2marks)
b) Find P(| X -100|) > 5 (3marks)

2. A random sample of the invoices, for books purchased by customers of a large

bookshop, was classified by book cover (hardback, paperback) and type of book (novel,
textbook, general interest). As part of the analysis of these invoices, an approximate
 2 statistic was calculated and found to be 11.09.
Assuming that there was no need to amalgamate any of the classification, carry out an
appropriate test to determine whether or not there was any association between book
cover and type of book. State your hypotheses clearly and use a 5% level of
significance. (6marks)
3. For one of the activities at a gymnastics completion, 8 gymnasts were awarded marks
out of 10 for each of artistic performance and technical ability. The results were as
follows:-
Gymnast A B C D E F G H

Technical ability 8.5 8.6 9.5 7.5 6.8 9.1 9.4 9.2

Artistic 6.2 7.5 8.2 6.7 6.0 7.2 8.0 9.1

performance

The value of the product moment correlation coefficient for these data is 0.774.
a) Stating your hypothesis clearly and using a 1% level of significance, interpret
this value.
b) Calculate the value of the rank correlation coefficient for these data
(6marks)

67
 c) Stating your hypotheses clearly and using a 1% level of significance, interpret
this coefficient. (3marks)
d) Explain why the rank correlation coefficient might be better one to use with
these data.
4. The results of an experiment to investigate the amount of chemical compound, y
grams, that dissolved in 100grams of water as x0C are recorded below
x 5 10 15 20 25 30 35 40 45 50 55 60

y 9 11 13 17 21 24 27 29 31 35 38 42

∑x2 = 16250, ∑y2 =8641, ∑xy =11295)

a) Calculate Sxx , Syy and Sxy (3marks)
b) Find the equation of the regression line of y on x in the form y =a + bx
(3marks)
c) Test, at the 5% level of significance, whether or not there is evidence that the
gradient of the regression line is positive. (8marks)
d) Explain how residuals can be used to decide on the suitability of a regression
model. (3marks)

5. Past records from a large supermarket show that 20% of people who buy chocolate
bars buy the family size bars. On one particular day a random sample of 30 people was
taken from those that had bought chocolate bars and 2 of them were found to have
bought a family size bar.
a) Test at the 5% significance level, whether or not the proportion p, of people who
bought a family size bar of chocolate that day had decreased. State your
hypotheses clearly.
(6 marks)
The manager of the supermarket thinks that the probability of a person buying a
gigantic chocolate bar is only 0.02. To test whether this hypothesis is true the manager
decides to take a random sample of 200 people who bought bars.
b) Find the critical region that would enable the manager to test whether or not
there is evidence that the probability is different from 0.02. The probability of
each tail should be as close to 2.5% as possible. (6 marks)

c) Write down the significance level of this test. (1mark)

68
 UNIVERSITY EXAMINATIONS 2010

SCHOOL OF APPLIED SOCIAL SCIENCES DEPARTMENT

OF INFORMATION TECHNOLOGY BACHELOR OF

BUSINESS INFORMATION TECHNOLOGY UNIVERSITY

EXAMINATION AUGUST 2010

BMA3102: STATISTICS II

TIME 2HOURS
AUGUST SERIES

INSTRUCTIONS: ANSWER QUESTIONS ONE (COMPULSORY) AND ANY OTHER

TWO QUESTIONS

QUESTION ONE (COMPULSORY) 30 MARKS

a) Define a statistic
(2 marks)
A random sample x1, x2 …... xn is taken from a population with unknown mean μ
b) For each of the following state with reason whether or not it is a statistic
x1 x 4
i. (2 marks)
2

ii.
 x2
 2 (2marks)
N
c) Two judges in a beauty contest rank the ten competitors in the following order.
x 6 4 3 9 2 7 1 5 10 8

y 4 1 6 7 5 8 2 3 9 10

Do the two judges appear to agree in their standards? (6marks)

d) A random sample of size 36 was taken from a population distributed N (μ, 3.92). The
value of the sample x¯ was 15.6.
i. Find a 90% confidence interval for μ (5 marks)

69
 It is believed that value of μ is 17.0
ii. Use your confidence interval to comment on this belief. (2marks)
e) A random sample x1, x2… x10 is taken from a normal population with 100 and standard
deviation 14.
f) A tennis coach believes that taller players are generally capable of hitting faster serves.
To investigate this hypothesis he collects data on the 20 adult male players he coaches.
The height, h, in metres and the speed of each player‟s fastest serve, v, in Km per hour
were recorded and summarized as follows:-
∑h = 36.22, ∑v = 2275, ∑h2 = 65.7396, ∑v2 = 259853 and ∑hv = 4128.03
i. Calculate the Pearson Moment Correlation Coefficient for these data. (4marks)

ii. Comment on the coach‟s hypothesis (2marks)

Question 2 (20marks)
a) Penshop have stores selling stationary in each of 6 towns. The population, P, in tens of
thousands and monthly turnover, T, in thousands of Pounds for each of the shops are as
recorded below
Town Agnet Bonns Claster Doggis Edgeton Figland

P(0000‟s) 3.2 7.6 5.2 9.0 8.1 4.8

T(000‟s) 11.1 12.4 13.3 19.3 17.9 11.8

i. Represent these data on a scatter diagram with T on the vertical axis (4 marks)

ii. Which town‟s shop might appear to be under achieving given the populations of
the towns?
(2marks)
iii. Suggest two other factors that might affect each shop‟s turnover. (2 marks)
2 2
You may assume that ∑P = 264.69, ∑T = 1286 and ∑PT = 574.25.
b) Find the equation of the regression line of T on P
(8marks)
c) Estimate the monthly turnover that might be expected if a shop were opened in Gratton, a
town with a population of 68000 (2marks)

d) Why might the management of Penshop be reluctant to use the regression line to estimate
the monthly turnover they could expect if a shop were opened in Haggin, a town with a
population of 172,000?

Question 3 (20marks)
a) Write shot notes on the following :-
i. Null and alternative hypothesis (2marks)
ii. One tailed and two tailed tests (2marks)
iii. Type I and Type II errors (2marks)
iv. Acceptance and rejection regions (2marks)

70
 b) A sample of 100 motorcar Tyres has a mean of 20000Km and a standard deviation of
800Km. a second sample of 150 Tyres, has a mean life of 22000Km and a standard
deviation of 900 Km.
i. Is it true to say that the two samples were drawn from the same population?
(10marks)
ii. Which is your rejection or acceptance as you conclude about the two populations
(2marks)
Question 4 (20 marks)
a) Explain briefly what you understand by the terms
i. Population
ii. Statistics (2marks)
b) A rugby player scores an average of 0.4 tries per march in which he plays
i. Find the probability that he scores 2 or more tries in a march (5marks)
The team‟s coach moves the player to a different position in the team believing he will then
score more frequently. In the next five marches he scores 6 tries.
ii. Stating your hypothesis clearly, test at the 5% level of significance whether or not
there is evidence of an increase in the number of tries the player scores per match
as a result of playing in a different position (5marks)
c) A child welfare officer asserts that the mean sleep of young babies is 14 hours a day. A
random sample of 64 babies shows that their mean sleep was only 13hours 30 minutes,
with a standard deviation of 3 hours. At 5% level of significance, test the assertion that
mean sleep of babies is less than 14 hours a day. (8marks)

Question 5 (20 Marks)

a) What is time series analysis? (3marks)
b) What is the moving averages system of forecasting (2marks)
c) The table below shows the number of bags of sugar per week
Week 1 2 3 4 5 6 7 8

No of Bags 340 330 323 343 350 345 365 342

Calculate thee weeks moving averages and six weeks moving averages (5marks)

d) A new diet program claims that participants wi8ll lose on average at least eight pounds
during the week of the program. A random sample of 40 people participating in the
program showed a sample mean weight loss of seven pounds. The sample standard
deviation was 3.2 pounds.
i. What is the rejection rule with   0.05 [3]
ii. What is your conclusion about the claim made by the diet program? [4]
iii. What is the p-value? [3]

71