PONDICHERRY UNIVERSITY

(A Central University)

DIRECTORATE OF DISTANCE EDUCATION

MASTER OF COMMERCE

M.Com – Second Year

Course Code: 59 Paper Code: MCOM2004

STATISTICAL ANALYSIS

DDE – WHERE INNOVATION IS A WAY OF LIFE

 PONDICHERRY UNIVERSITY
(A Central University)

DIRECTORATE OF DISTANCE EDUCATION

MASTER OF COMMERCE

M.Com – Second Year

Course Code:59 Paper Code:MCOM2004

STATISTICAL ANALYSIS
 STATISTICAL ANALYSIS

Authors:
1. Unit I & II Dr.Kiruthika
Assistant Professor,
Department of Statistics,
Pondicherry University

2. Unit III, IV & V Dr. R.Vishnu Vardhan

Assistant Professor,
Department of Statistics,
Pondicherry University

1
 STATISTICAL ANALYSIS

TABLE OF CONTENTS

UNIT TITLE PAGE NO.

I 4
Univariate Analysis

II Probability Theory 32

III Correlation and Regression Analysis 58

IV Statistical Estimation and Testing 72

V Non Parametric Tests 97

2
M.Com – 2nd Year PAPER CODE: MCOM 2004

STATISTICAL ANALYSIS

UNIT I
Univariate Analysis: An overview of Central Tendency, Dispersion and Skewness. Sampling
and Data Collection: Sampling and Sampling (Probability and Non-Probability) methods;
Sampling and non-sampling errors; Law of Large Number and Central Limit Theorem;
Sampling distributions and their characteristics.

UNITII
Probability Theory: Probability – Classical, Relative, and Subjective Probability: Addition and
Multiplication Probability Models; Conditional Probability and Baye’s Theorem Probability
Distributions: Binomial, Poisson, and Normal Distributions their characteristics and
applications.

UNIT III
Correlation and Regression Analysis: Two Variables Case
UNIT IV
Statistical Estimation and Testing: Point and interval estimation of population mean, proportion
and variance; Statistical testing – hypotheses and errors; Sample size; Large and small sampling
tests –Z tests, T tests, and F tests.

UNIT V
Non Parametric Tests: Chi-square tests; Sign tests Wilcoxon Signed – Rank tests; Wald –
Wolfowitz tests; Kruskal – Wallis tests.
Note: Question Paper Shall covers 20% Theory and 80% Problems

REFERENCES:
Arora & Arora, Statistics for Management, Sultan Chand, New Delhi, 2009
Gupta S.P., Statistical Methods, Sultan Chand, New Delhi 2009
Levin, Richard I. and David S Rubin, Statistics for Management, Prentice Hall, Delhi 2009
Sonia Taylor,Business Statistics, Palgrave Macmillan, 2009.
Qaziahmed, Zubuirkhan, Shadabahmedkha, Numerical and Statistical Techniques Ane,
2010

3
 UNIT I

UNIVARIATE ANALYSIS

1.1 – Univariate Analysis

1.2 – Data Collection and Sampling Methods

1.3 – Sampling Distributions

1.1 Univariate Analysis

Statistics is collection, analysis and interpretation of numerical data. Statistics can be

broadly classified as descriptive statistics and inferential statistics. Descriptive statistics
includes collecting, summarizing, presenting and analyzing a data. The data can be presented
as diagrams such as bar chart, pie chart etc. or graphical representations such as histogram,
frequency curves etc. The data can also be presented as discrete or continuous tables. More
information about data can be obtained by finding the average (mean) and the variation in the
data specified by the standard deviation. Measures of central tendency and dispersion are
useful in describing the numerical data more precisely. The first section (Lesson 1.1) gives a
brief overview of various measures of central tendency, dispersion and skewness.
Measures of Central Tendency
Measures of central tendency give the average value of the data. The average values
are also known as measure of central tendency as the values tend to lie centrally within a data
set arranged according to magnitude. Measures of central tendency should be rigidly defined,
easy to calculate, based on all observations and suitable for further treatment. Measures of
central tendency are classified into mathematical averages like Arithmetic Mean, Geometric
Mean, Harmonic Mean and positional averages namely, Median and Mode.
Arithmetic mean is most commonly used measure of central tendency. It is rigidly
defined and based on all the observations. It is affected by the extreme values. In case of
extreme values, arithmetic mean does not represent the distribution properly.
Median divides the distribution into two equal parts. It is the positional average. It is
the central value such that the number of observations above and below it is equal. If the
number of observations is odd, then the median is the middle value in ordered (ascending order)
list of observations. If the number of observations is even then the median is the average of
the two middle values. Median is not affected by extreme values and it is located by inspection.
Median can also be located graphically by drawing ogives (cumulative frequency curves).

4
 Quartile divides the distribution into four equal parts. The first quartile (Q1) or lower
quartile has 25% of observations of the distribution below it and 75% of items are greater than
it. The second quartile (Q2) is the median and the third quartile (Q3) has 75% of items of the
distribution below it. Deciles divide the series into ten equal parts. There are nine deciles and
fifth decile is the median. Percentile divides the distribution into 100 equal parts. There are
99 percentiles (P1, P2, ..., P99).
Mode is a value which occur most number of times (highest frequency). The mode
may not be unique. Sometimes it may not exist. A distribution having one mode is known as
unimodal. The empirical relation between the arithmetic mean, median and mode is given by
Mean – Mode = 3 (Mean – Median)
Geometric mean is a special mean which gives more weights to small items. It is used
to find the rate of population growth and the rate of interest. It is also used in the construction
of index numbers.
Harmonic mean is also rigidly defined and not much affected by fluctuations of
sampling. It also gives more importance to small items and is useful only when small items
are to be given very high weightage.
The formulae for Arithmetic Mean, Geometric Mean and Harmonic Mean for
individual series (raw data), discrete frequency distribution and continuous frequency
distribution are given in Table 1. Table 2 gives the formulae for calculating Median and Mode
for continuous frequency distribution. The steps for calculating mode for discrete series are
given below.
1. Prepare a grouping table with 6 columns with ‘𝑥’ values in the margin.
2. Write the frequencies corresponding to ‘𝑥’ values in column 1.
3. In column 2, the frequencies are grouped in twos.
4. In column 3, the frequencies are grouped in twos excluding the first frequency.
5. In column 4, the frequencies are grouped in threes.
6. In column 5, the frequencies are grouped in threes excluding the first frequency.
7. In column 6, the frequencies are grouped in twos excluding the first two
frequencies.
8. Mark the highest frequency in each column.
9. Mode is the ‘𝑥’ value having highest frequency, obtained from the analysis table.

5
 Table 1: Formulae for Measures of Central Tendency

Measures of Individual Series Discrete Frequency Continuous Frequency

central tendency (Ungrouped data) Distribution Distribution
∑𝑛𝑖=1 𝑓𝑖 𝑚𝑖
∑𝑛𝑖=1 𝑥𝑖 ∑𝑛𝑖=1 𝑓𝑖 𝑥𝑖 𝑥̅ =
Arithmetic Mean 𝑥̅ = 𝑥̅ = 𝑁
𝑛 𝑁
∑𝑛𝑖=1 𝑓𝑖 𝑙𝑜𝑔𝑚𝑖
∑𝑛𝑖=1 𝑙𝑜𝑔𝑥𝑖 ∑𝑛𝑖=1 𝑓𝑖 𝑙𝑜𝑔𝑥𝑖 𝐺. 𝑀. = 𝑎𝑛𝑡𝑖𝑙𝑜𝑔 [ ]
Geometric Mean 𝐺. 𝑀. = 𝑎𝑛𝑡𝑖𝑙𝑜𝑔 [ ] 𝐺. 𝑀. = 𝑎𝑛𝑡𝑖𝑙𝑜𝑔 [ ] 𝑁
𝑛 𝑁
1
1 1 𝐻. 𝑀. =
𝐻. 𝑀. = 𝐻. 𝑀. = 1 𝑛 𝑓𝑖
Harmonic Mean 1 𝑛 1 1 𝑛 𝑓𝑖 ∑
∑ ∑ 𝑁 𝑖=1 𝑚𝑖
𝑛 𝑖=1 𝑥𝑖 𝑁 𝑖=1 𝑥𝑖

Table 2: Formulae for Median and Mode

Measures of
Continuous Frequency Distribution
central tendency
𝑁
– 𝑐.𝑓
2
𝑀𝑒𝑑𝑖𝑎𝑛 = 𝐿 + ( )x 𝑖
𝑓
Median where 𝐿 = Lower limit of median class; 𝑓 = frequency of median class
𝑐. 𝑓 = cumulative frequency of the class preceding median class
𝑖 = class interval of median class
𝑓1 − 𝑓0
𝑀𝑜𝑑𝑒 = 𝐿1 + ( ) x𝑖
2𝑓1 − 𝑓0 − 𝑓2
where 𝐿1 = Lower limit of modal class; 𝑓1 = frequency of modal class
Mode
𝑓0 = Frequency of the class preceding the modal class
𝑓2 = Frequency of the class succeeding the modal class
𝑖 = class interval

6
1. Individual Series: Let 𝑥1 , 𝑥2 , … , 𝑥𝑛 be set of 𝑛 observations.
Example 1: Calculate Arithmetic mean (A.M.), Median, Geometric Mean (G.M.) and
Harmonic Mean (H.M.) for the following data.
87, 63, 91, 72, 93, 79, 83, 94, 75
(i) A.M. = (87+63+91+72+93+79+83+94+75+88)/9 = 81.89
(ii) Arrange the values in ascending order : 63, 72, 75, 79, 83, 87, 91, 93, 94
Median = 83
(iii) G.M. = antilog[(log 87+ ... + log 75)/9] = 81.254
(iv) H.M. = 1/ [1/9 *(1/87 + ... + 1/75)] = 80.5903

Example 2: Find the median and mode for the data 3, 5, 2, 6, 5, 9, 8, 6, 3, 5.

Arrange the values in the ascending order: 2, 3, 3, 5, 5, 5, 6, 6, 8, 9

Median = (5 + 5) / 2 = 5

Mode = number occurring most frequently = 5

2. Discrete Frequency distribution: Let 𝑥1 , 𝑥2 , … , 𝑥𝑛 occur 𝑓1 , 𝑓2 , … , 𝑓𝑛 times

respectively and 𝑁 = ∑𝑛𝑖=1 𝑓𝑖 .
Example 3: Calculate mean weekly wages of the employees of a company from the
following frequency distribution.
Wages
(Rs.) 2550 2650 2750 2850 2950 3050 3150
(𝑥𝑖 )
Number of
employees 8 10 16 14 10 5 2
(𝑓𝑖 )
𝑓𝑖 𝑥𝑖 20400 26500 44000 39900 29500 15250 6300

∑𝑛
𝑖=1 𝑓𝑖 𝑥𝑖
𝑥̅ = = 181850/65 = Rs. 2797.69
𝑁
Example 4: Calculate Q1, Q2 and Q3 for the following series.
Size of
4 4.5 5 5.5 6 6.5 7 7.5 8
shoes
Frequency 10 18 22 25 40 15 10 8 7
Cumulative
10 28 50 75 115 130 140 148 155
frequency

Steps: 1. Arrange the data in ascending or descending order

7
 2. Find the cumulative frequencies.
3. Median = Size of (N+1)/2 th item
𝑁 = ∑𝑛𝑖=1 𝑓𝑖 = 155
Q1 = Size of (N+1)/4 th item = Size of 39th item = 5
Q2 = Size of (N+1)/2 th item = Size of 78th item = 6
Q3 = Size of 3*(N+1)/4 th item = Size of 117th item = 6.5
Example 5: Calculate mode for the following frequency distribution.
Value (𝑥𝑖 ) 100 110 120 130 140 150 160 170 180
Frequency (𝑓𝑖 ) 10 12 15 19 20 8 4 3 2

Grouping Table
Value (x) 1 2 3 4 5 6
100 10
22
110 12 37
27
120 15 46
34
130 19 54
39
140 20 47
28
150 8 32
12
160 4 15
7
170 3 9
5
180 2

Analysis Table
Column 110 120 130 140 150
No.
1 /
2 / /
3 / /
4 / / /
5 / / /
6 / / /
Total 1 3 5 4 1
Frequency

8
 The mode is 130 as this value occurs five times. But through inspection, the mode is
140.
Example 6: Calculate Geometric mean and Harmonic mean from the following data.
Size of items (𝑥𝑖 ) 6 7 8 9 10 11
Frequency (𝑓𝑖 ) 4 6 9 5 2 8
∑𝑛
𝑖=1 𝑓𝑖 𝑙𝑜𝑔𝑥𝑖
𝐺. 𝑀. = 𝑎𝑛𝑡𝑖𝑙𝑜𝑔 [ ] = antilog (31.41336/34) = antilog (0.923922) = 8.3930
𝑁

1
𝐻. 𝑀. = 1 𝑛 𝑓𝑖 = 34 / 4.131637 = 8.23
∑
𝑁 𝑖=1𝑥𝑖

3. Continuous Frequency Distribution: Let 𝑚1 , 𝑚2 , … , 𝑚𝑛 be mid-points of 𝑛 class

intervals with frequencies 𝑓1 , 𝑓2 , … , 𝑓𝑛 and 𝑁 = ∑𝑛𝑖=1 𝑓𝑖 .

Example 7: Calculate arithmetic mean, median, mode, geometric mean and harmonic
mean for the following frequency distribution.

Grade 30-40 40-50 50-60 60-70 70-80 80-90 90-100

No. of
1 3 11 21 43 32 9
Students

No. of
Grade Students c.f. mi fimi fi (logmi) fi / mi
(fi)
30-40 1 1 35 35 1.544068 0.02857143
40-50 3 4 45 135 4.959638 0.06666667
50-60 11 15 55 605 19.14399 0.2
60-70 21 36 65 1365 38.07118 0.32307692
70-80 43 79 75 3225 80.62763 0.57333333
80-90 32 111 85 2720 61.74141 0.37647059
90-100 9 120 95 855 17.79951 0.09473684
120 8940 223.8874 1.66285578
∑𝑛
𝑖=1 𝑓𝑖 𝑚𝑖
Arithmetic Mean : 𝑥̅ = = 8940/120 = 74.5
𝑁

𝑁
– 𝑐.𝑓 60−36
2
𝑴𝒆𝒅𝒊𝒂𝒏 = 𝐿 + ( ) x 𝑖 = 70 + ( ) x 10 = 70 + 5.581 = 75.581
𝑓 43

1 0 𝑓 −𝑓 43−21
𝑴𝒐𝒅𝒆 = 𝐿1 + (2𝑓 −𝑓 ) x 𝑖 = 70 + (86−21−32) x 10 = 70 + 6.67 = 76.67
−𝑓 1 0 2

∑𝑛
𝑖=1 𝑓𝑖 𝑙𝑜𝑔𝑚𝑖
𝑮. 𝑴. = 𝑎𝑛𝑡𝑖𝑙𝑜𝑔 [ ] = antilog(223.8874/120) = antilog(1.865728) = 73.405
𝑁

1
𝑯. 𝑴. = 1 𝑛 𝑓𝑖 = 120 / 1.66285578 = 72.165
∑
𝑁 𝑖=1𝑚𝑖

9
Example 8: Determine the quartiles and the median from the following table.
Income No. of
(Rs.) persons
Below 30 69
Below 40 236
Below 50 443
Below 60 508
Below 70 566
Below 80 593
Below 90 603

Income f c.f
Below 30 69 69
30-40 167 236
40-50 207 443
50-60 65 508
60-70 58 566
70-80 27 593
80-90 10 603

Median = size of N/2th item = 301.5th item

Median class is 40-50,
𝑁
– 𝑐.𝑓 301.5−236
2
𝑴𝒆𝒅𝒊𝒂𝒏 = 𝐿 + ( ) x 𝑖 = 40 + ( ) x 10 = 43.16
𝑓 207

Q1 = size of N/4th item = 150.75th item

First Quartile class is 30-40
𝑁
– 𝑐.𝑓 150.75−69
4
𝑸𝟏 = 𝐿 + ( ) x 𝑖 = 30 + ( ) x 10 = 34.895
𝑓 167

Q3 = size of 3N/4th item = 452.25th item

First Quartile class is 50-60
3𝑁
– 𝑐.𝑓 452.25−443
𝑸𝟑 = 𝐿 + ( 4 ) x 𝑖 = 50 + ( ) x 10 = 51.423
𝑓 65

Example 9: A time and motion study of a certain operation shows the following distribution
for 100 workers. Calculate mean, median and mode of the distribution.
Time
10-15 15-20 20-25 25-30 30-35 35-40 40-45
(minutes)
No. of
8 14 18 25 15 14 6
Workers

10
 Time fi c.f. mi di =( mi-27.5)/5 fidi
10 – 15 8 8 12.5 -3 -24
15 – 20 14 22 17.5 -2 -28
20 - 25 18 40 22.5 -1 -18
25 – 30 25 65 27.5 0 0
30 – 35 15 80 32.5 1 15
35 – 40 14 94 37.5 2 28
40 – 45 6 100 42.5 3 6
100 -9

∑ 𝑓𝑑
Mean = 𝐴 + x 𝑖 = 27.5 – (9x5)/100 = 27.05
𝑁
𝑁
– 𝑐.𝑓 50−40
𝑴𝒆𝒅𝒊𝒂𝒏 = 𝐿 + ( 2 ) x 𝑖 = 25 + ( ) x 5 = 27
𝑓 25

1 0 𝑓 −𝑓 25−18
𝑴𝒐𝒅𝒆 = 𝐿1 + (2𝑓 −𝑓 ) x 𝑖 = 25 + (50−18−15) x 5 = 27.06
−𝑓 1 0 2

Measures of Dispersion
The literal meaning of dispersion is “scatteredness”. Dispersion gives an idea about the
homogeneity or heterogeneity of the distribution. Measures of central tendency only tell us
about the concentration of the observations about the central part of the distribution. Consider
the series (1) 9, 9, 9, 9, 9 (2) 1, 2, 3, 4, 35 (3) 1, 4, 8, 12, 20. In all the cases it is seen that
there are 5 observations and the mean value is 9. Here the average value does not tell us about
the variation in the data. Measures of Dispersion gives complete information about the
variation in the data. According to Spiegel, “The degree to which numerical data tend to
spread about an average value is called the variation or dispersion of the data”. The average
specifies the representative value of a series while the dispersion tells how the values are
scattered from the central value. The main purpose of measuring dispersion is to test the
reliability of an average, to compare two or more series with respect to variability and to
facilitate as a basis for further statistical analysis. Measures of dispersion should be rigidly
defined, simple to understand and easy to compute, based on all observations and suitable for
further algebraic treatment. The dispersion may be measured either absolutely or relatively.
When the dispersion is measured and expressed in terms of the original data, it is called an
absolute dispersion. It cannot be used for comparison if expressed in different units. Relative
measure of dispersion is computed for comparing the variability. The coefficient of dispersion
of each group is calculated for comparing two series. The various measures of dispersion are
Range, Quartile deviation, Mean deviation and Standard deviation.

11
 Range is the simplest measure of dispersion. It depends only on the extreme values. It
is simple to compute but not reliable measure. It cannot be applied to open end cases. Range
is useful in studying the variation in the prices of shares, stocks and other commodities whose
price fluctuates from one period to another period. Range is used in construction of control
charts.

Range = Largest value – Smallest value = L - S

𝐿−𝑆
Coefficient of range =
𝐿+𝑆
Example 10: Find the range of the series 7, 10, 18, 12, 6, 15, 5, 3.
Range = 18 – 3 = 15
Quartile deviation is defined as half the distance between the third and first quartiles.
The quartile deviation gives an idea about the distribution of the middle half of the items around
the median. It is simple to understand and not affected by extreme values. Its value is affected
by sampling fluctuations.
𝑄3 −𝑄1
Quartile Deviation (Q.D.) =
2
𝑄3 −𝑄1
Coefficient of Q.D. =
𝑄3 +𝑄1

The range and quartile deviations are positional measures of dispersion.

Example 11: Calculate the semi-interquartile range and quartile coefficient from the following
data:

Age in years 20 30 40 50 60 70 80
Number of
3 61 132 153 140 51 3
members

Cumulative
Age Frequency
Frequency
20 3 3
30 61 64
40 132 196
50 153 349
60 140 489
70 51 540
80 3 543

Q1 = size of (N+1)/4 th item = size of 136th item = 40 years

Q3 = size of 3(N+1)/4 th item = size of 408th item = 60 years

12
 𝑄3 −𝑄1
Quartile Deviation (Q.D.) = = 10 years
2
𝑄3 −𝑄1
Coefficient of Q.D. = = 0.2
𝑄3 +𝑄1

Example 12: Compute Quartile deviation and coefficient of quartile deviation for the
following problem.

Class Cumulative
Frequency
Interval Frequency
250 – 260 8 8
260 – 270 10 18
270 – 280 16 34
280 – 290 14 48
290 – 300 10 58
300 – 310 5 63
310 – 320 2 65

𝑁
– 𝑐.𝑓
4
𝑄1 = 𝐿 + ( ) x 𝑖 = 268.25
𝑓

3𝑁
– 𝑐.𝑓
4
𝑄3 = 𝐿 + ( ) x 𝑖 = 290.75
𝑓

𝑄3 −𝑄1
Quartile Deviation (Q.D.) = = 11.25
2
𝑄3 −𝑄1
Coefficient of Q.D. = = 0.04025
𝑄3 +𝑄1

Mean deviation is the average of the deviations of a series from either the arithmetic
mean or median, ignoring the signs of deviation. Mean deviation is rigidly defined and based
on all values of the series. It is used in statistical analysis of economic business and social
phenomena. It is useful in forecasting business cycles. The main demerit of mean deviation
is that it ignores the algebraic signs of the deviation.
Standard deviation was introduced by Karl Pearson in 1893. Standard Deviation is
the most important and widely used measure of dispersion. It is also called Root Mean Square
Deviation or Mean Square Error. It is denoted by σ. It is defined as the positive square root of
the arithmetic mean of the squares of the deviations of the given value from the arithmetic
mean. It is rigidly defined; less affected by fluctuations of sampling and is the basis for
measuring the correlation coefficient, sampling and statistical inferences. The relative measure
of standard deviation is known as coefficient of variation. According to Karl Pearson,
“Coefficient of Variation is the percentage variation in mean, standard deviation being

13
considered as the total variation in the mean”. Coefficient of variation is used to compare the
variability in two or more series. The series with less coefficient of variation indicates that the
series is more consistent or more homogeneous. The series with high coefficient of variation
indicates that the series is less consistent or less stable. The variance is a measure of the
“average” of the squared deviations from the mean. It is denoted by σ2.
The formulae for Mean Deviation and Standard Deviation for individual series (raw
data), discrete frequency distribution and continuous frequency distribution are given in Table
3.
Table 3 : Formulae for Measures of Dispersion

Measure of Individual Series Discrete Frequency Continuous Frequency

Dispersion (Ungrouped data) Distribution Distribution
𝑀. 𝐷. (𝑎𝑏𝑜𝑢𝑡 𝑚𝑒𝑎𝑛)
∑|𝐷| 𝑀. 𝐷. (𝑎𝑏𝑜𝑢𝑡 𝑚𝑒𝑎𝑛) 𝑀. 𝐷. (𝑎𝑏𝑜𝑢𝑡 𝑚𝑒𝑎𝑛)
= ∑ 𝑓|𝐷| ∑ 𝑓|𝐷|
𝑛 = =
where D = 𝑋 − 𝑋̅ 𝑁 𝑁
Mean where D = 𝑋 − 𝑋̅ where D = mi − 𝑋̅
Deviation
𝑀. 𝐷. (𝑎𝑏𝑜𝑢𝑡 𝑚𝑒𝑑𝑖𝑎𝑛)
(M.D.)
∑|𝐷| 𝑀. 𝐷. (𝑎𝑏𝑜𝑢𝑡 𝑚𝑒𝑑𝑖𝑎𝑛) 𝑀. 𝐷. (𝑎𝑏𝑜𝑢𝑡 𝑚𝑒𝑑𝑖𝑎𝑛)
= ∑ 𝑓|𝐷| ∑ 𝑓|𝐷|
𝑛 = =
where D = 𝑋 − 𝑀𝑒𝑑𝑖𝑎𝑛 𝑁 𝑁
where D = 𝑋 − 𝑀𝑒𝑑𝑖𝑎𝑛 where D = mi - Median

Standard ∑𝑛𝑖=1(𝑥𝑖 − 𝑥̅ )2 ∑𝑛𝑖=1 𝑓𝑖 (𝑥𝑖 − 𝑥̅ )2 ∑𝑛𝑖=1 𝑓𝑖 (𝑚𝑖 − 𝑥̅ )2

Deviation 𝜎=√ 𝜎=√ 𝜎= √
𝑛 𝑁 𝑁
𝑛
where 𝑁 = ∑𝑖=1 𝑓𝑖
Coefficient
Coeff. Of M.D. = Mean Deviation / Mean or Median
of M.D.
Coefficient
of 𝜎
𝐶. 𝑉. = x 100
Variation 𝑥̅
(C.V.)

Example 13: Calculate the mean deviation from (1) arithmetic mean (2) mode (3) median
with respect to the marks obtained by 9 students: 7, 4, 10, 9, 15, 12, 7, 9, 7.

(1) Mean = 8.9

(2) Mode = 7
(3) Median 9

14
 Marks |𝑫| (Mean) |𝑫| (Median) |𝑫| (Mode)
7 1.9 2 0
4 4.9 5 3
10 1.1 1 3
9 0.1 0 2
15 6.1 6 8
12 3.1 3 5
7 1.9 2 0
9 0.1 0 2
7 1.9 2 0
21.1 21 23

∑|𝐷|
MD (about mean) = = 2.34
𝑛
∑|𝐷|
MD (about median) = = 2.33
𝑛
∑|𝐷|
MD (about mode) = = 2.56
𝑛

Example 14: Find the mean deviation about mean and standard deviation for the data given
below.
Height (inches) 59.5-62.5 62.5-65.5 65.5-68.5 68.5-71.5 71.5-74.5
No. of male
5 18 42 27 8
Students

Height fi mi fi mi |𝐷| = |𝑚𝑖 − 𝑋̅| 𝑓|𝐷| fi mi2

59.5-62.5 5 61 305 6.45 32.25 18605
62.5-65.5 18 64 1152 3.45 62.1 73728
65.5-68.5 42 67 2814 0.45 18.9 188538
68.5-71.5 27 70 1890 2.55 68.85 132300
71.5-74.5 8 73 584 5.55 44.4 42632
100 6745 226.5 455803

∑𝑛
𝑓𝑚
𝑋̅ = 𝑖=1𝑁 𝑖 𝑖 = 6745/100 = 67.45
∑ 𝑓|𝐷|
𝑀. 𝐷. (𝑎𝑏𝑜𝑢𝑡 𝑚𝑒𝑎𝑛) = = 226.5/100 = 2.26 inches
𝑁

Coeff. of M.D. = Mean Deviation / Mean = 2.26/67.45 = 0.0335

∑𝑛 2 ∑𝑛 2 ∑𝑛 2
𝑖=1 𝑓𝑖 (𝑚𝑖 −𝑥̅ ) 𝑖=1 𝑓𝑖 𝑚𝑖 𝑖=1 𝑓𝑖 𝑚𝑖
𝜎=√ =√ −( ) = 2.920188
𝑁 𝑁 𝑁
𝜎
𝐶. 𝑉. = 𝑥̅ x 100 = 4.33

15
Example 15: The price of shares of companies X and y are given below. State which
company’s share s more stable in value?

Company X 55 54 52 53 56 58 52 50 51 49
Company Y 108 107 105 105 106 107 104 103 104 101

∑𝑛
𝑖=1(𝑥𝑖 −𝑥̅ )
2
𝑋̅ = 53 𝜎𝑥 = √ =2.645
𝑛
𝜎
𝐶. 𝑉. (𝑋) = x 100 = 4.99
𝑥̅

∑𝑛 ̅)2
𝑖=1(𝑦𝑖 −𝑦
𝑦̅ = 105 𝜎𝑥 = √ =2
𝑛
𝜎
𝐶. 𝑉. (𝑌) = ̅ x 100 = 1.90
𝑦

Company Y shares are more stable in value s compared to Company X shares because
the coefficient of variation of Company Y is lower than Company X.
Example 16: Compute the absolute measure of variance and a relative measure of dispersion
from the following data.
Income
No. of families
(Rs.)
300-399 30
400-499 46
500-599 58
600-699 76
700-799 60
800-899 50
900-999 20

No. of 𝒎 − 𝟔𝟒𝟗. 𝟓
Income (Rs.) m 𝒅= fd fd2
families (f) 𝟏𝟎𝟎
299.5-399.5 30 349.5 -3 -90 270
399.5-499.5 46 449.5 -2 -92 184
499.5-599.5 58 549.5 -1 -58 58
599.5-699.5 76 649.5 0 0 0
699.5-799.5 60 749.5 1 60 60
799.5-899.5 50 849.5 2 100 200
899.5-999.5 20 949.5 3 60 180
N=340 ∑ 𝑓𝑑=-20 ∑ 𝑓𝑑2 =952

∑ 𝑓𝑑
𝑋̅ = 𝐴 + x 𝑖 = 643.6
𝑁

16
 ∑ 𝑓𝑑2 ∑ 𝑓𝑑 2
Variance: 𝜎 2 = { −( ) } x 𝑖 2 = 27965.39
𝑁 𝑁

σ = √27965.39 = 167.228
𝜎
Coefficient of Variation: 𝐶. 𝑉. = 𝑥̅ x 100 = 167.23/643.6 x 100 = 25.98%

Example 17: The index number of prices of cotton and coal shares is given below. Which of
the two shares is more variable in prices?
Index number Index number
Month of prices of of prices of
cotton shares coal shares
January 188 131
February 178 130
March 173 130
April 164 129
May 172 129
June 183 120
July 184 127
August 185 127
September 211 130
October 217 137
November 232 140
December 240 142

Cotton shares:
𝜎
𝑋̅ = 193.9 𝐶. 𝑉. = 𝑋̅ x 100 = 12.28%

σX = 23.81
Coal shares:
𝜎
𝑌̅ = 131 𝐶. 𝑉. = 𝑌̅ x 100 = 4.42%

σY = 23.81
The cotton shares are more variable in price than the coal shares.
Example 18: A manufacturer of television tubes has two types of tubes, A and B respectively.
The tubes have mean lifetime of 𝑋̅𝐴 = 1495 hours and 𝑋̅𝐵 = 1875 hours, and standard deviation
of σA = 280 hours and σB = 310 hours. Which tube has the greater relative dispersion?
C.V. (A) = σA / 𝑋̅𝐴 x 100 = 18.7%
C.V. (B) = σB / 𝑋̅𝐵 x 100 = 16.5%
Tube A has the greater relative dispersion.

17
Skewness
The measures of central tendency and dispersion are not adequate to characterize a
distribution completely. They do not indicate whether the distribution is symmetric or not.
Skewness is a measure of degree of asymmetry of a distribution. Skewness gives an idea about
the shape of the curve. According to Croxton and Cowden, “When a series is not symmetrical
it is said to be asymmetrical or skewed”. The mean, median and mode of a skewed distribution
do not coincide. The relative measure of skewness is called coefficient of skewness. The
following are the coefficient of skewness:

𝑀𝑒𝑎𝑛−𝑀𝑜𝑑𝑒
1. Karl Pearson’s coefficient of skewness: 𝑆𝑘 = where σ is the standard
𝜎

deviation of the distribution.

𝑄3 +𝑄1 −2𝑄2
2. Bowley’s Coeficient of skewness: 𝑆𝑘 = 𝑄3 −𝑄1

In a symmetrical distribution, the quartiles are equidistant from the median but in a
skewed distribution, the quartiles are not equidistant from the median. Bowley’s coefficient of
skewness lies between ±1.

If the curve has longer tail towards the right then it is said to be positively skewed
(Mean>Median>Mode). If the curve has longer tail towards left then it is negatively skewed
(Mean < Median < Mode).

Example 19: Calculate Karl Pearson’s coefficient of skewness and Bowley’s coefficient of
skewness for the information given below.

Mean Standard Deviation Median First Quartile Third Quartile

26.01 17.5 23.67 11.09 38.54

Mode = 3 Median – 2 Mean = 18.99

𝑀𝑒𝑎𝑛−𝑀𝑜𝑑𝑒
𝑆𝑘 = = 0.4
𝜎

18
 𝑄3 +𝑄1 −2𝑄2
𝑆𝑘 = = 0.08
𝑄3 −𝑄1

Example 20: Compute Karl Pearsons’ coefficient of skewness for the following distribution.

Class Interval 35-40 40-45 45-50 50-55 55-60

Frequency 2 14 24 30 15

Class 𝒎 − 𝟒𝟕. 𝟓
f m 𝒅= fd fd2
Interval 𝟓
35-40 2 37.5 -2 -4 8
40-45 14 42.5 -1 -14 14
45-50 24 47.5 0 0 0
50-55 30 52.5 1 30 30
55-60 15 57.5 2 30 60
N=85 ∑ 𝑓𝑑= 42 ∑ 𝑓𝑑2 =112

∑ 𝑓𝑑
𝑋̅ = 𝐴 + x 𝑖 = 49.97
𝑁

∑ 𝑓𝑑2 ∑ 𝑓𝑑 2
𝜎 = √{ −( ) } x 𝑖 = 5.18
𝑁 𝑁

1 0 𝑓 −𝑓
𝑀𝑜𝑑𝑒 = 𝐿1 + (2𝑓 −𝑓 ) x 𝑖 = 51.43
−𝑓 1 0 2

𝑀𝑒𝑎𝑛−𝑀𝑜𝑑𝑒
𝑆𝑘 = = -0.282
𝜎

Kurtosis is used to study the peakedness of the distribution curve. A normal curve is
symmetrical and bell shaped which is known as mesokurtic curve. If the curve is peaked at the
top then it is known as leptokurtic curve. If the curve is more flat than normal curve then it is
known as platykurtic curve.
1.2 Data Collection and Sampling Methods
Statistical data may be classified as primary and secondary data. Primary data are
collected directly by the investigator to study a particular problem. There are many methods
for primary data collection namely, direct method, experiments and surveys. Surveys may be
conducted in different ways e.g. personal interview, telephone interview, self-administered
questionnaire. Questionnaire should be designed properly to get precise information from the
respondents.
Secondary data are the data which are already collected and is available for present study.
The various sources of secondary data are international publications, official publications of
government, publications of research institutions, financial institutions, journals and

19
newspapers. The investigators should check the suitability, adequacy and reliability of
secondary data before using it for analysis.
In a statistical enquiry, the population is set of all possible observations in the study. The
population may be finite or infinite. Population can also be classified as hypothetical or real.
It may be sometimes difficult to study the entire population, so a sample is drawn from the
population to arrive at valid conclusions. Sampling refers to drawing a sample from a
population. A sample is expected to represent the population. Sampling theory is a study of
relationship, existing between a population and samples drawn from the population. Sampling
theory is also useful in determining whether the observed difference between two samples is
due to chance variation or whether they are really significant. A representative sample is
obtained by random sampling. A sampling plan is a procedure for specifying how a sample
is to be drawn from the population. The samples selected must have same characteristics of
the population.
If we draw a sample from an urn, there is a choice of replacing or not replacing the sample
into the urn before a second drawing. Sampling where each member of the population are
chosen more than once is called sampling with replacement, while if each member cannot be
chosen more than once is called sampling without replacement. A finite population in which
sampling is with replacement is theoretically considered to be infinite. Since any number of
samples can be drawn without exhausting the population.
The principal steps in a sample survey are:
1. Objectives of the survey
2. Defining the population to be sampled
3. The frame and Sampling units
4. Data to be collected
5. The questionnaire or schedule
6. Method of collecting information
7. Non-respondents
8. Selection of proper sampling design
9. Organization of field work
10. Summary and analysis of the data
Sampling plan can be classified as probability (random) and non-probability sampling
methods. A random sample is one where each observation of the population has an equal
chance of being selected.

20
 The various random sampling methods are
1. Simple Random Sampling
2. Stratified Random Sampling
3. Systematic Sampling
Simple random sample is a sample selected such that each sample has equal probability
of being chosen. Simple random sampling forms the basis for the other random sampling
techniques. Let n denote the sample size and N represent the population size. The probability
of selecting any particular item from the population is 1/N. The samples may be selected with
replacement or without replacement. Simple Random Sampling With Replacement (SRSWR)
means that after an item is selected, the item is again returned to the frame, where it has the
same probability of being selected again. Simple Random Sampling Without Replacement
(SRSWOR) means that an item selected cannot be selected again. In otherwords, if an item is
selected with probability 1/N on the first draw then the probability of selecting another item
(not previously selected) on second draw is 1/N-1. This process continues until desired sample
is selected. Several methods are used for random selection of samples.
(i) Lottery Method: This is most popular method. In this method, all the items of the
population are numbered on separate slips of paper of same size and folded mixed up
in a container. A blindfold selection is made.
(ii) Table of random numbers: A table of random numbers consists of a series of digits
listed randomly generated sequence. There are several tables of random numbers such
as Tippet’s Random Number Tables, Fisher and Yates Table, Kendall and Babington
Smith’s Table. Since, the numeric system uses 10 digits (0, 1, 2, 3, 4, 5, 6, 7, 8, 9), the
chance that any particular digit is randomly generated is equal to the probability of
generating any other digit. To select a simple random sample, choose an arbitrary
starting point from the table of random numbers.
A simple random sample is more representative of the population and personal bias is
completely eliminated as compared to the judgement or purposive sampling. Some of the
limitations of simple random sampling are that it requires up-to-date frame and larger sample
size.
Stratified random sample is obtained by dividing the population into strata and then
drawing simple random sample from each stratum. A stratum is defined by some common
characteristic such as gender, age, education level, income, geographical area, economic status
and so on. The units within an each group (strata) are homogeneous. A simple random sample
is selected within each of the strata. Stratified random sampling is more efficient than simple

21
random sampling or systematic sampling as it ensures the representation of items across the
entire population.
In a systematic sample, the N items in the population is partitioned into n groups of k
items where k = n/N. To select a systematic sample, choose the first item at random from the
first k items in the frame and then select the remaining n-1 items by taking every kth item
thereafter from the entire frame.
In a non-probability sampling, an item or individual is selected without knowing their
probabilities of selection. The various non-probability sampling methods are
1. Judgement or Purposive Sampling
2. Quota Sampling
3. Convenience Sampling
In judgement sampling, the choice of the sample items depends on the judgement of
the investigator. Judgement sample is obtained according to the discretion of someone who is
familiar with relevant characteristic of the population. It is a simple method and is used to
obtain a more representative sample. The samples may not be representative due to individual
bias.
Quota sampling is similar to stratified sampling. It is stratified cum purposive sampling.
The data is collected by dividing the population into quota according to some characteristics.
In convenient sampling, convenient elementary units are chosen from a population. The
items selected are inexpensive or convenient to sample. It is suitable when the population is
not clearly defined and complete source list is not available. The samples are biased.
Sampling and Non-Sampling Errors
The errors involved in the collection, processing and analysis of a data may broadly be
classified as sampling errors and non-sampling errors. Sampling error comprises the difference
between the sample and the population which is due to improper selection of samples.
Sampling bias is a tendency to favour the selection of units that have particular characteristics.
Sampling errors are due to the fact that only a part of the population is used to estimate the
population parameters and draw inferences about the population. Increase in the sample size
usually results in the decrease in sampling error.
Sampling biases are due to the following reasons:
1. Sampling bias may be due to faulty selection of sample. This may be due to the use of
defective sampling technique for selection of a sample. The investigator may use
purposive or judgement sampling for obtain certain results. This bias can be overcome
by using simple random sampling.

22
 2. The investigators may sometimes substitute a convenient member of the population if
there is a difficulty in enumerating a particular sampling unit. This leads to bias.
3. Bias due to defective demarcation of sampling units is significant in area surveys such
as agricultural experiments in the field.
4. Sampling error may be due to improper choice of the statistic for estimating the
population parameters.
Non-sampling error is an error that results from the way in which the observations are
obtained. Non-sampling errors primarily arise at the stages of observation, ascertainment and
processing of the data. It is present in both complete enumeration survey as well as sample
surveys. Non-Sampling errors are more serious in complete enumeration as compared to a
sample survey.
Non-sampling errors arise due to the following factors:
1. Faulty planning or definitions: Non-sampling errors may be due to inadequate and
inconsistent data specification with respect to the objective of the survey. It may also
be due to errors in ill-designed questionnaire, errors in recording the measurements or
lack of trained investigators.
2. Response errors: These errors are introduced as a result of the response given by the
respondents. The respondent may misunderstand a particular question and give
improper information. The respondents may give wrong information due to self interest
or prestige. Sometimes investigator may affect the accuracy of the response by the way
the question is asked or recorded.
3. Non-response Biases: This bias occurs when full information is not obtained on all the
sampling units.
4. Errors in coverage: If the objectives of the survey are not precisely stated then this may
result in including the sampling units which are not to be included and excluding the
units which are to be included.
5. Compiling errors: Various operations of data processing such as editing and coding of
the responses, tabulation and summarizing the observations made in the survey are
source of error.
6. Publication errors: The errors committed during presentation and printing of the
tabulated results.

23
1.3 Sampling Distributions

In many applications, inferences about the population are made which are based on
statistics calculated from samples. In order to do this, we need to discuss about the sampling
distribution. A sampling distribution is a probability distribution consisting of all possible
values of a sample statistic.
Consider all possible samples of size N that can be drawn from a given population (either
with or without replacement). For each sample, a statistic such as mean or standard deviation
is computed, which will vary from sample to sample. In this way, a distribution of the statistic
is obtained which is known as sampling distribution. Sampling distribution for mean, variances
can be derived. The standard deviation of the sampling distribution of a statistic is known as
its standard error.
Sampling Distribution of Mean
The sampling distribution of the mean is the probability distribution of all possible values
of the sample mean. The sample mean is unbiased because the mean of all possible sample
means (of a given sample size n) is equal to the population mean. Let us consider a simple
example consisting of 4 typists. Each typist is asked to type the same page of the manuscript
and the number of typographical errors is recorded.
Typist Number of Errors
A 3
B 2
C 1
D 4

From the data, μ = 2.5 errors and σ = 1.12 errors. If we select samples of two typists with
replacement from this population, there are 16 possible samples. The average of all 16 sample
means is 2.5. Since the mean of the 16 sample means is equal to the population mean, the
sample mean is an unbiased estimator of population mean.

Sample Typists Sample Observations Sample Means

1 A, A 3, 3 3
2 A, B 3, 2 2.5
3 A, C 3, 1 2
4 A, D 3, 4 3.5

24
 5 B, A 2, 3 2.5
6 B, B 2, 2 2
7 B, C 2, 1 1.5
8 B, D 2, 4 3
9 C, A 1, 3 2
10 C ,B 1, 2 1.5
11 C, C 1, 1 1
12 C, D 1, 4 2.5
13 D, A 4, 3 3.5
14 D,B 4, 2 3
15 D, C 4, 1 3
16 D, D 4, 4 4
𝜇𝑋̅ = 2.5,
𝜎𝑋̅ =0.28

How the sample mean varies from sample to sample can be expressed as the standard
deviation of all possible sample means. This is known as the standard error of the mean. The
𝜎
standard error of mean is given by 𝜎𝑋̅ = .. The standard error is the spread of the values
√𝑛

from the mean of the means in a sampling distribution. In this example, although the sample
means vary from sample to sample, depending on which two typists are selected, the sample
means do not vary much as the individual values in the population.
As the sample size increases, the standard error of the mean goes to zero. The bigger the
sample the closer to the true mean we are likely to get and so have more confidence in that
estimate. So, we can say that if we are sampling from a normally distributed population with
a mean, , and standard deviation, , then the sampling distribution of the mean will also be
normally distributed for any size n with mean x- = and standard error of the mean, 𝜎𝑋̅ .
Table 3 gives standard errors of sampling distribution for various statistics under the
condition of random sampling from an infinite population or of sampling with replacement
from a finite population.

25
 Table 3 Standard Errors of Sampling Distribution

Statistic Standard Error

𝜎
𝜎𝑋̅ =
Mean √𝑛

𝑝𝑞
Proportion 𝜎𝑝 = √
𝑛
𝜎
𝜎𝑠 =
Standard deviation √2𝑛

𝜋 1.2533𝜎
Median 𝜎𝑚𝑒𝑑 = 𝜎√ =
2𝑛 𝑛

Sampling Distribution of Proportion

Consider a categorical variable which has two categories, such as the customer prefers a
particular brand of a product or not. Let π be the proportion of product in the entire population
and p be the proportion of product in the sample with the characteristic of interest.
The sample proportion is used to estimate the population proportion. The sample proportion
takes on values between 0 and 1.
p = X/n = Number of product having the characteristic of interest/ Sample size
𝜋(1−𝜋)
The standard error of proportion is given by 𝜎𝑝 = √ . The sampling distribution of
𝑛

the proportion follows Binomial distribution.

Central Limit Theorem
There are many instances where sampling is done from non-normally distributed
populations. The Central Limit Theorem deals with this situation. The Central Limit Theorem
provides the information required for constructing a sampling distribution.
The Central Limit Theorem states that as the sample size increases (gets large enough),
the sampling distribution of the mean is approximately normally distributed. For most
population distributions, regardless of shape, the sampling distribution of the mean is
approximately normally distributed if the samples of at least size 30 are selected. By applying
the theorem, the descriptive values for a sampling distribution (usually, the mean and the
standard error, which is computed from the sampling variance) is obtained and the probabilities
associated with any of the sample means in the sampling distribution are also obtained. The
most important reason why we need to construct sampling distributions is to obtain the

26
probabilities associated with varying amounts of sampling error. Once we have this
information, it can be used for statistical generalization and for testing the hypotheses. The
Central Limit Theorem allows us to directly compute the mean, variance, and standard error of
the sampling distribution without taking trouble of drawing all possible samples from the
population. The shape of the distribution of sample means is described by using the third
principle of the Central Limit Theorem. This principle states that the sampling distribution of
means will tend to look like a normal distribution, when the samples are selected from relatively
large numbers of observations. The theoretical proof of the central limit theorem requires
independent observations in the sample. This condition is met for infinite population and for
finite populations where samples are taken with replacement. In general, central limit theorem
is applied when the population size is large.

Questions
1. Distinguish between primary and secondary data.
2. Discuss the various methods of collecting primary data.
3. Explain the significance of sampling methods.
4. Describe briefly different types of sampling methods.
5. Explain sampling and non-sampling errors.
6. What are measures of central tendency? Also state the requisites of a good average.
7. State the various measures of central tendency and dispersion.
8. What do you understand by median and quartiles?
9. Briefly explain measures of dispersion. Why standard deviation is called as ideal
measure of dispersion.
10. What is the need for sampling as compared to complete enumeration?
11. Differentiate between simple random sampling with replacement and without
replacement.
12. Describe the following sampling methods with suitable examples.
a. Judgement sampling b. Cluster sampling c. Quota sampling

13. Describe different types of random sampling methods.

14. Explain non-probability sampling methods. State its disadvantages.
15. Discuss about sampling distributions and standard error.

27
 EXERCISES
16. The number of ATM transactions per day was recorded at 15 locations in a large city.
Find the mean and median number of transactions from the following data
35, 49, 225, 50, 30, 65, 40, 55, 52, 76, 48, 325, 47, 32, 60
17. (a) Find the mean, median and mode for the data: 8, 11, 4, 3, 2, 5, 10, 6, 4, 1, 10, 8,
12, 6, 5, 7
(b) Find the geometric mean for the data:
28.5, 73.6, 47.2, 31.5, 64.8
(c) Find the harmonic mean for the data:
3.2, 5.2, 4.8, 6.1, 4.2
18. In automobile mileage and fuel consumption testing, 10 automobiles were road tested for
300 miles in both city and country driving conditions. The following data were recorded
for miles-per-gallon performance.
City: 16.2 16.7 15.9 14.4 13.2 15.3 16.8 16.0 16.1 15.3
Country: 19.4 20.6 18.3 18.6 19.2 17.4 17.2 18.6 19.0 21.1
Comment about the difference in performance for city and country driving using mean,
median and mode.
19. In an examination of 675 candidates, the examiner supplied the following information:

Marks Number of
obtained Candidates

Less than 10% 7

Less than 20% 39

Less than 30% 95

Less than 40% 201

Less than 50% 381

Less than 60% 545

Less than 70% 631

Less than 80% 675

Calculate the arithmetic mean and mode of the percentage marks obtained.

28
20. The daily wages of the employees of a company is given in the following table. Calculate
i) the quartiles : Q1, Q2, Q3
ii) the deciles : D1, D2, D9
iii) the percentiles : P35,P60

Wages(Rs) Number of Employees

250.00-260.00 8
260.00-270.00 10
270.00-280.00 16
280.00-290.00 14
290.00-300.00 10
300.00-310.00 5
310.00-320.00 2

21. Calculate arithmetic mean from the following frequency distribution.

Output in units Number of workers
300-309 9
310-319 20
320-329 24
330-339 38
340-349 48
350-359 27
360-369 17
370-379 6

22. Compute the mode of the following distribution.

Size of item Frequency (f)
0-5 20
5 – 10 24
10 – 15 32
15 – 20 28
20 - 25 20
25 – 30 16
30 – 35 37
35 – 40 10
40 – 45 8

29
23. In a small branch of a bank in a rural area, the following is the average deposit balance
of current accounts during a month. Calculate the median, seventh decile and 85 th
percentile.
Deposit balance less No. of
than (Rs.) deposits
1000 500
900 498
800 480
700 475
600 440
500 374
400 300
300 125
200 25

24. Compute mean deviation about mean, standard deviation and coefficient of variation for
the following distribution.
Profits No. of
(Rs. in crores) companies
20-30 4
30-40 8
40-50 18
50-60 30
60-70 15
70-80 10
80-90 8
90-100 7

25. Calculate geometric for the following distribution.

Yield 7.5 – 10.5 10.5-13.5 13.5-16.5 16.5-19.5 19.5-22.5 22.5-25.5 25.5-28.5
No. of
5 9 19 23 7 4 1
farms

30
26. The following table shows the age distribution of persons in a particular region. Find
the median age.

Age Below Below Below Below Below Below Below 70 and

(years) 10 20 30 40 50 60 70 above
No. of
persons
2 5 9 12 14 15 15.5 15.6
(in
thousands)

27. Compute mean deviation about mean and mean coefficient of dispersion for the
following data.
Marks 10 15 20 30 40 50

Frequency 8 12 15 10 3 2

28. Obtain Karl Pearson’s coefficient of skewness and Bowley’s coefficient of skewness
for the following data.
Class
130-134 135-139 140-144 145-149 150-154 155-159 160-164
Interval
Frequency 3 12 21 28 19 12 5
29. In a moderately skewed frequency distribution, the mean is Rs.15 and the median is Rs.
14. If the coefficient of variation is 30% then find the Karl Pearsons’s coefficient of
skewness.
30. An analysis of the monthly wages paid to workers in the firms A and B belonging to the
same industry gives the following results. Which firm has greater variability in
individual wages?
Firm A Firm B
No. of workers 500 600
Average Monthly Wage (Rs.) 480 475
Variance of distribution of wages (Rs.) 400 625

31
 UNIT II

PROBABILITY THEORY

CONTENTS:

2.1 – Introduction to Probability theory

2.2 – Conditional Probability

2.3 – Probability Distributions

2.1 Introduction to Probability Theory

A probability is the numeric value representing the chance or possibility a particular
event will occur, such as a rainy day, a defective product or price of a stock etc. According to
Ya-Lin Chou, “probability is the science of decision making with calculated risks in the face
of uncertainty”. If the probability of success is based on prior knowledge of the process
involved then it is called a priori probability. In the empirical probability approach, the
probability is based on the observed data. Subjective probability differs from person to person.
The basic terminologies used in probability theory are given below.
1. Random Experiment: If in each trial of an experiment conducted under identical
conditions the outcome is not unique, but may be any one of the possible outcomes, then
such an experiment is called a random experiment.
e.g : tossing a coin, throwing a die, selecting a card from a pack of playing cards etc.
2. Outcome: The result of a random experiment is called an outcome.
3. Trial and Event: Any particular performance of a random experiment is called a trial
and outcomes or combination of outcomes is called events.
e.g: if a coin is tossed repeatedly, we may get head or tail.
Tossing of coin - Trial
Getting head or tail - Event.
Throwing a die – Trial, getting outcomes 1,2,3,4,5,6 – Event.
4. Exhaustive Events: The total number of possible outcomes of a random experiment
is known as exhaustive events.
e.g: In tossing of a coin, there are two exhaustive cases namely, Head and Tail
In throwing a die, there are 6 exhaustive cases namely 1, 2, 3, 4, 5, 6
5. Favourable Events: The number of cases favourable to an event in a trial is the
number of outcomes which entail the happening of the event.

32
 e.g. In drawing a card from a pack of cards, the number of cases favourable to
drawing an Ace is 4.
6. Mutually Exclusive Events: Events are said to be mutually exclusive if happening of
any one of the them precludes the happening of all the others (i.e.) if no two or more of
them can happen simultaneously in the same trial.
e.g. In tossing a coin, the events head and tail are mutually exclusive.

7. Equally Likely Events: Outcomes of trial are said to be equally likely if taking into
consideration all the relevant evidences, there is no reason to expect one is preference to
the others.
e.g: In a random toss of an unbiased coin, head and tail are equally likely.

8. Independent Events: Events are independent if the happening of an event is not

affected by the supplementary knowledge concerning the occurrence of any number of
the remaining events.
e.g : In tossing an unbiased coin, the event of getting a head in the first toss is
independent of getting a head in the second, third etc.
Mathematical or Classical or apriori Probability:
If a random experiment or a trial results in ‘n’ exhaustive, mutually exclusive and
equally likely outcomes, out of which ‘m’ are favourable to the occurrence of an event E, then
the probability ‘p’ of occurrence of E, denoted by P(E) is given by
𝑁𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝑓𝑎𝑣𝑜𝑢𝑟𝑎𝑏𝑙𝑒 𝑐𝑎𝑠𝑒𝑠 𝑚
𝑝 = 𝑃(𝐸) = =
𝑇𝑜𝑡𝑎𝑙 𝑛𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝑒𝑥ℎ𝑎𝑢𝑠𝑡𝑖𝑣𝑒 𝑐𝑎𝑠𝑒𝑠 𝑛
Remark:
1. Since m ≥ 0, n > 0 and m ≤ n, P(E) ≥ 0 and P(E) ≤ 1. => 0 ≤ p ≤ 1
2. The non-happening of the event E is called the complementary event of E and is denoted
by 𝐸̅ 𝑜𝑟 𝐸 𝑐 . The number of favourable cases of 𝐸̅ is (n-m). Then, the probability q
that E will not happen is given by ,
𝑛−𝑚 𝑚
𝑞 = 𝑃(𝐸̅ ) = =1− = 1−𝑝 => 𝑝 + 𝑞 = 1
𝑛 𝑛

Statistical or Empirical Probability:

If an experiment is performed repeatedly under essentially homogeneous and
identical conditions, then the limiting value of the ratio of the number of times the event occurs
to the number of trials, as the number of trial becomes indefinitely large, is called the
probability of happening of the event,

33
 𝑀
𝑃(𝐸) = lim
𝑁→∞ 𝑁

Sample space: The set of all possible outcomes of a given random experiment is called the
sample space associated with that experiment. Each possible outcome or element in a sample
space (S) is called a sample point or an elementary event.

Event: Every non-empty subset of S, which is a disjoint union of single element subsets of the
sample space S of a random experiment E is called an event.

Examples:
1. Tossing a single coin – S = {H, T} ; n(S) = 2 {n(S) is total number of sample points
in S.}
2. Tossing of two coins – S = {HH, HT, TH, TT} ; n(S) = 4.
3. Throwing a die - S = { 1, 2, 3, 4, 5, 6 } n(S) = 6.
 (1,1), (1,2) . . . (1,6) 
(2,1), (2,2),    (2,6)

 (3,1) (3,2)    (3,6) 
4. Throwing two dice –S =  
 (4,1)     (4,6)
 (5,1)     (5,6) 
 
 (6,1)     (6,6) 
E1 = Sum of points on two dice in 5
E1 = (1,4), (4,1), (2,3), (3,2) , n(E1) = 4
E2 = Sum of points on two dice is odd
(1,2), (1,4), (1,6), (2,1), (2,3), (2,5), (3,2), (3,4), (3,6),
 
(4,1), (4,3), (4,5), (5,2), (5,4), (5,6), (6,1), (6,3), (6,5) 
n(E2) = 18

Example 1: A ball is drawn at random from a box containing 6 red balls, 4 green balls and 5
blue balls. Find the probability that the ball drawn is
i) red ii) green iii) blue iv) red or green
i) P(red) = 6/15 = 2/5
ii) P(green) = 4/15
iii) P(blue) = 5/15= 1/3
iv) P(red or green) = 10/15 = 2/3

Example 2:a) Two cards are drawn at random from a pack of 52 cards. Find the probability
of drawing two aces.

34
 4𝐶2 1
𝑃(𝐴) = =
52𝐶2 221

b) Four cards are drawn from a pack of cards. Find the probability that
i) all are diamonds ii) two are spades and two are hearts

13𝐶4 13𝐶2 ×13𝐶2

𝑖)𝑃(𝐴) = 𝑖𝑖)𝑃(𝐴) =
52𝐶4 52𝐶4

Example 3: Out of 20 employees in a company, 5 are post graduates. Three employees are
selected at random. Find the probability that all the three are post graduates.
𝟓𝐶3
𝑃(𝐸) =
20𝐶3

Example 4: A coin is tossed two times. The sample space is S = {HH, HT, TH, TT}, where
H and T denote head and tail. What is the probability of getting at least one head?

Let A be an event of getting at least one head. A = {HH, HT, TH}

P(A) = P(HH) + P(HT) + P(TH)

= 1/4 + 1/4 + 1/4

P(A) = 3/4

Example 5: If two television tubes are picked in succession from a shipment of 240 television
tubes of which 15 tubes are defective. What is the probability that they will both be defective?
Prob. (both the selected tubes will be defective) = 15/240 x 14/239 = 7/1912
Example 6: From 25 tickets, marked with the first 25 numerals, one is drawn at random. Find
the probability that
i) it is a multiple of 5 or of 7,
ii) it is a multiple of 3 or of 7.

i) Numbers (out of the first 25 numerals) which are multiples of 5 are 5, 10, 15, 20 and
25, and the numbers which are multiples of 7 are 7, 14 and 21.
Pr ( a multiple of 5 or 7) = 8 / 25

ii) Numbers (among the first 25 numerals) which are multiples of 3 are 3, 6, 9, 12, 15,
18, 21, 24 and the numbers which are multiples of 7 are 7, 14, 21.
Pr ( a multiple of 3 or 7) = 10 / 25 = 2 / 5.

35
Example 7: Three groups of chicken contain respectively 3 girls and 1 boy, 2 girls and 2 boys,
1 girl and 3 boys. One child is selected at random from each group. Determine the probability
that the three selected children consist of 2 boys and 1 girl.

To select 2 boys and 1 girl:

Group No. I II III

(i) Girl Boy Boy

(ii) Boy Girl Boy

(iii) Boy Boy Girl

Prob = P(i) + P(ii) + P(iii)

3 2 3 1 2 3 1 2 1
= × × + × × + × ×
4 4 4 4 4 4 4 4 4
9 3 1 13
Prob = 32 + 32 + 32 = 32

Example 8: If two dice are thrown, what is the probability that the sum is

a) Greater than 8 b) neither 7 or 11

a) Let S denote the sum of the two dice.

𝑃(𝑆 > 8) = 𝑃(𝑆 = 9) + 𝑃(𝑆 = 10) + 𝑃(𝑆 = 11) + 𝑃(𝑆 = 12)

4 3 2 1 10 5
𝑃(𝑆 > 8) = + + + = =
36 36 36 36 36 18
b) Let A denote the event of getting the sum of 7 and B denote the event of getting
the sum of 11.
1 1
𝑃(𝐴) = , 𝑃(𝐵) =
6 18
𝑃(𝐴̅ ∩ 𝐵̅ ) = 1 − 𝑃(𝐴 ∪ 𝐵) = 1 − 𝑃(𝐴) − 𝑃(𝐵)
1 1 7
= 1− − =
6 18 9
Example 9: A factory finds that, average, 20% of the bolts produced by a given machine will
be defective for certain specified requirements. If 10 bolts are selected at random from the
day’s production of this machine. Find the probability

36
 a) that exactly 2 will be defective
b) that 2 or more will be defective
c) that more than will be defective
Let X be number of defective bolts.

a) 𝑃(𝑋 = 2) = 10𝐶2 (0.2)2 (0.8)8 = 45(0.04)(0.1678) = 0.3020

b) 𝑃(𝑋 ≥ 2) = 1 − 𝑃(𝑋 = 0) − 𝑃(𝑋 = 1)

= 1 − 10𝐶0 (0.2)0 (0.8)10 − 10𝐶1 (0.2)1 (0.8)0

= 1 − (0.8)10 − 10(0.2)(0.8)9

= 1 − 0.1074 − 0.2684 = 0.6242

c) 𝑃(𝑋 > 5) = 𝑃(𝑋 = 6) + 𝑃(𝑋 = 7) + 𝑃(𝑋 = 8) + 𝑃(𝑋 = 9) + 𝑃(𝑋 = 10)

= 10𝐶6 (0.2)6 (0.8)4 + 10𝐶7 (0.2)7 (0.8)3 + 10𝐶8 (0.2)8 (0.8)2 +

10𝐶9 (0.2)9 (0.8)1 + 10𝐶0 (0.2)10

= 0.00637

Results on probability:
1. 𝑃(∅) = 0.
2. 𝑃(𝐴̅) = 1 − 𝑃(𝐴).
3. 𝑃(𝐴̅ ∩ 𝐵) = 𝑃(𝐵) − 𝑃(𝐴 ∩ 𝐵).
4. 𝑃(𝐴 ∩ 𝐵̅ ) = 𝑃(𝐴) − 𝑃(𝐴 ∩ 𝐵).

Addition Theorem of Probability:

If A and B are any two events (subsets of sample space S) and are not disjoint, then
𝑃(𝐴 ∩ 𝐵) = 𝑃(𝐴) + 𝑃(𝐵) − 𝑃(𝐴 ∩ 𝐵).
Proof:
𝑛(𝐴 ∪ 𝐵)
𝑃(𝐴 ∪ 𝐵) = = 𝑛(𝐴) + 𝑛(𝐵) − 𝑛(𝐴 ∩ 𝐵)
𝑛(𝑠)
𝑛(𝐴) 𝑛(𝐵) 𝑛(𝐴∩𝐵)
= + − = 𝑃(𝐴) + 𝑃(𝐵) − 𝑃(𝐴 ∩ 𝐵)
𝑛(𝑠) 𝑛(𝑠) 𝑛(𝑠)

Corollary1: If the events A and B are mutually disjoint then

𝐴∩𝐵 =∅ => 𝑃(𝐴 ∩ 𝐵) = 𝑃(∅) = 0
Corollary2: For three non-mutually exclusive events to A, B and C
𝑃(𝐴 ∪ 𝐵 ∪ 𝐶) = 𝑃(𝐴) + 𝑃(𝐵) + 𝑃(𝐶) − 𝑃(𝐴 ∩ 𝐵) − 𝑃(𝐴 ∩ 𝐶) − 𝑃(𝐵 ∩ 𝐶) + 𝑃(𝐴 ∩ 𝐵 ∩ 𝐶)

37
Example 10: The chance of winning the race by person A is 1/5 and that of person B is 1/6.
What is the probability that the race will be won by A or B?
P(𝐴 ∪ 𝐵 ) = P(A) + P(B) = 11/30
Example 11: Find the probability of a 4 turning up at least once in two tosses of a die.
Let A1 be an event of ‘4’ turning up on the first toss.
Let A2 be an event of ‘4’ turning up on the second toss.
Since, A1 and A2 are not mutually exclusive,
𝑃(𝐴1 ∪ 𝐴2 ) = 𝑃(𝐴1 ) + 𝑃(𝐴2 ) − 𝑃(𝐴1 ∩ 𝐴2 )
1 1 1 1
= + − ( )( )
6 6 6 6
11
𝑃(𝐴1 ∪ 𝐴2 ) =
36
Example 12: A problem in Mathematics is given to three students X, Y and Z whose chances
of solving it are 3/4, 1/2 and 1/4 respectively. What is the probability that the problem will be
solved?
𝑃(𝑋 ∪ 𝑌 ∪ 𝑍) = 𝑃 (𝑋) + 𝑃(𝑌) + 𝑃(𝑍) − 𝑃(𝑋 ∩ 𝑌) − 𝑃(𝑋 ∩ 𝑍) − 𝑃(𝑌 ∩ 𝑍) + 𝑃(𝑋 ∩ 𝑌 ∩ 𝑍)
3 1 1 3 1 3 1 1 1 3 1 1 29
= + + − × − × − × + × × =
4 2 4 4 2 4 4 2 4 4 2 4 32

Example 13: In a city (based on a sample survey), the probabilities that a family owns a
television set, a washing machine or both television and washing machine are 0.86, 0.35 and
0.29 respectively. What is probability that a family owns either or both?
Let A be an event that a family owns a television set.
Let B be an event that a family owns a washing machine.
P(A) = 0.86
P(B) = 0.35
P(A ∩ B) = 0.29
𝑃(𝐴 ∪ 𝐵) = 𝑃(𝐴) + 𝑃(𝐵) − 𝑃(𝐴 ∩ 𝐵)
= 0.86 + 0.35 – 0.29
= 0.92
Example 14: A can hit a target 3 times in 5 shots, B 2 times in 5 shots, and C 3 times in 4 shots.
Find the probability of the target being hit at all when all of them try.
Let E1 be the event that A hits the target.
3 3
𝑃(𝐸1 ) = ̅̅̅1 ) = 1 − = 2/5
𝑎𝑛𝑑 𝑃(𝐸
5 5

38
 and E2 be the event that B hits the target,

2 2
𝑃(𝐸2 ) = ̅̅̅2 ) = 1 − = 3/5
𝑎𝑛𝑑 𝑃(𝐸
5 5
and E3 be the event that B hits the target,

3 3 1
𝑃(𝐸3 ) = ̅̅̅3 ) = 1 − =
𝑎𝑛𝑑 𝑃(𝐸
4 4 4
The required probability ‘p’ that the target is hit when they all try is given by

𝑝 = 𝑃[𝑎𝑡𝑙𝑒𝑎𝑠𝑡 𝑜𝑛𝑒 𝑜𝑓 𝑡ℎ𝑒 𝑡ℎ𝑟𝑒𝑒 ℎ𝑖𝑡𝑠 𝑡ℎ𝑒 𝑡𝑎𝑟𝑔𝑒𝑡]

= 1 − 𝑃[𝑛𝑜𝑛𝑒 ℎ𝑖𝑡𝑠 𝑡ℎ𝑒 𝑡𝑎𝑟𝑔𝑒𝑡]

̅̅̅1 ∩ ̅̅̅
= 1 − 𝑃(𝐸 𝐸2 ∩ ̅̅̅
𝐸3 )

̅̅̅1 )𝑃(𝐸
= 1 − 𝑃(𝐸 ̅̅̅2 )𝑃(𝐸
̅̅̅3 ),

by compound probability theorem, since E1, E2 and E3 are independent

 ̅̅̅
𝐸1 , ̅̅̅
𝐸2 𝑎𝑛𝑑 ̅̅̅
𝐸3 are also independent
2 3 1 47
𝑝=1− . . =
5 5 4 50
Example 15: The odds against A solving a certain problem are 4 to 3 and odds in favour of B
solving the same problem are 7 to 5. What is the probability that the problem is solved if they
both try independently?

Let E1 denote the event that ‘ A’ solves the problem.

Let E2 denote the event that ‘B’ solve the problem.

3 3 3 4
𝑃(𝐸1 ) = = 𝑃(𝐸1𝑐 ) = 1 − =
4+3 7 7 7
7 7 5
𝑃(𝐸2 ) = 𝑃(𝐸2𝑐 ) = 1 − =
12 12 12
4 5
P(atleast one of A and B solves the problem) = 1 − 𝑃(𝐸1𝑐 ∩ 𝐸2𝑐 ) = 1 − (7 × 12)

= 16/21

2.2 Conditional Probability

Let P(A) represent the likelihood that a random experiment will result in an outcome in
the set A relative to the sample space S of the random experiment. If we have prior information
that the outcome of the random experiment must be in a set B of S then this information must

39
be used to re-appraise the likelihood that the outcome will also be in B. This re-appraised
probability is denoted by
P(A|B) = conditional probability of the event A given that event B has already happened
When we know that a particular event B has occurred instead of S, we concentrate our attention
on B only and the conditional probability of A given B will be the ratio of that part of A which
is included in B (ie 𝐴 ∩ 𝐵) to the probability of B.

e.g : Let us consider a random experiment of drawing a card from a pack of cards.
A: drawing a king card ; n(A) = 4
4 1
𝑃(𝐴) = =
52 13

Now suppose that a card is drawn and we are informed that the drawn card is red. How does
this information affect the likelihood of the event A?
B: the card drawn is red ; n(B) = 26

Now, the P(A) must be computed relative to the new sample space B which consists of 26
sample points (red cards only).
Among the 26 red cards, 2(red) kings so 𝑛(𝐴 ∩ 𝐵) = 2.

Hence the required probability

𝑛(𝐴∩𝐵) 2 1
𝑃(𝐴|𝐵) = = =
𝑛(𝐵) 26 13

Multiplication Theorem of Probability:

𝑃(𝐴). 𝑃(𝐴|𝐵), 𝑃(𝐴) > 0

𝑃(𝐴 ∩ 𝐵) = {
𝑃(𝐵). 𝑃(𝐴|𝐵), 𝑃(𝐵) > 0

where P(B|A) represents conditional probability of occurrence of B when the event A has
already happened and P(A|B) is the conditional probability of happening of A, given that B has
already happened.

Independent Events: An event A is said to be independent of another event B, if the

conditional probability of A given P(A|B) is equal to the unconditional probability of B (ie) if
P(A|B) = P(A) and P(B|A)= P(B).

40
Multiplication theorem of probability for independent events:

If A and B are two events with positive probabilities {𝑃(𝐴) ≠ 0, 𝑃(𝐵) ≠ 0} then A
and B are independent if and only if 𝑃(𝐴 ∩ 𝐵) = 𝑃(𝐴). 𝑃(𝐵)

Pairwise independent events: The events 𝐴1 , 𝐴2 , … … . 𝐴𝑛 are said to be pairwise independent

if and only if

𝑃(𝐴𝑖 ∩ 𝐴𝑗 ) = 𝑃(𝐴𝑖 ). 𝑃(𝐴𝑗 ); 𝑖 ≠ 𝑗 = 1,2,3, … , 𝑛

Example 16: Given that P(A) = 1/3, P(B) = 1/4, P(A|B) = 1/6. Find P(B|A) and P(B|Ac).
P(B|A) = P(A|B) P(B) / P(A) = 1/8 , P(B|Ac) = 5/16
Example 17: Three marbles are drawn successively from a bag containing 5 blue marbles, 4
green marbles and 6 black marbles. Find the probability that they drawn in the order black,
green and blue if each ball is (a) replaced (b) not replaced
Let A be an event of drawing black marble on the first draw
Let B be an event of drawing green marble on the second draw
Let C be an event of drawing blue marble on the third draw
a) If each marble is replaced, then A, B and C are independent events
6 4 5 8
𝑃(𝐴𝐵𝐶) = 𝑃(𝐴)𝑃(𝐵)𝑃(𝐶) = × × =
15 15 15 225

b) If each marble is not replaced, then A, B and C are dependent events.

𝐵 𝐶
𝑃(𝐴𝐵𝐶) = 𝑃(𝐴)𝑃 ( ) 𝑃 ( )
𝐴 𝐴𝐵

6 4 5 4
= × 14 × 13 = 91
15

Example 18: A consumer research organization has studied the services under warranty
provided by the 50 new-car dealers in a certain city. The following table gives the results of
the findings of the study.

Good service under Poor service under

warranty warranty
Dealer in business(5 years or more) 16 4
Dealer in business (less than 5 years) 10 20

41
(i) What is the probability that a customer (who randomly selects one of the new-car
dealers), gets the dealer who provides good service under warranty?
(ii) What is the probability a customer (who randomly selects the dealer who has been in the
business for 5 years or more) gets the dealer who provides good service under warranty?
(iii) What is the probability that one of the dealers who has been in business less than 5 years
will provide good service under warranty?

Let A denote the selection of a dealer who provides good service under warranty.
Let D denote the selection of a dealer who has been in business for 5 years or more.
(i) P(A) = (16 + 10) / 50 = 0.52
(ii) P(A| D) = 16 / 20 = 0.80
10
50
(iii) P(A | D ) = P(A ∩ D ) / P(D ) =
c c c
30 = 0.20 / 0.60 = 1/3 = 0.33
50

Example 18: Find the probabilities of getting

(a) Three heads in three tosses of a coin
(b) Four sixes and then another number in five throws of a die.

(a) Pr.( Three heads in three tosses of a coin) = 1/2 x 1/2 x 1/2 = 1/8
(b) Pr. (Four sixes and then another number in five throws of a die)
= 1/6 x 1/6 x 1/6 x 1/6 x 5/6
= 5 / 7776
Example 19: A coin is tossed three times and the outcomes are HHH, HHT, HTH, HTT, THH,
THT, TTH, TTT. If A is the event that a head occurs on each of the first two tosses, B is the
event that a tail occurs on the third toss and C is the event that exactly two tails occur in three
tosses. Show that
(a) Events A and B are independent
(b) Events B and C are dependent
A = {HHH, HHT}
P(A) = 1/4
B = {HHT, HTT, THT, TTT}
P(B) = 1/2
C = {HTT, THT, TTH}
P(C) = 3/8

42
 A ∩ B = {HHT}
P(A ∩ B) = 1/8
B ∩ C = {HTT, THT}
P(B ∩ C) = 1/4
(a) Since P(A) . P(B) = 1/4 . 1/2 = 1/8 = P(A ∩ B) ,
Events A and B are independent.
(b) Since P(B) . P(C) = 1/2 . 3/8 = 3/16 ≠ P(B ∩ C) ,
Events A and B are dependent.
Bayes’ Theorem:

If 𝐸1 , 𝐸2 , … … . 𝐸𝑛 are mutually disjoint event with P (Ei)≠ 0 (i=1,2,….,n), then for

any arbitrary event A which is a subset of ⋃𝑛𝑖=1 𝐸𝑖 such that P(A) > 0, we have
𝑃(𝐸𝑖 )𝑃(𝐴|𝐸𝑖 ) 𝑃(𝐸𝑖 )𝑃(𝐴|𝐸𝑖 )
𝑃(𝐸𝑖 |𝐴) = 𝑛 = , 𝑖 = 1,2, … , 𝑛
∑𝑖=1 𝑃(𝐸𝑖 )𝑃(𝐴|𝐸𝑖 ) 𝑃(𝐴)

Remark:
1. The probabilities 𝑃(𝐸1 ). 𝑃(𝐸2 ) … … 𝑃(𝐸𝑛 ) are termed as ‘a prior’
probabilities.
2. 𝑃(𝐸𝑖 |𝐴) are called ‘likelihood’.
3. 𝑃(𝐸𝑖 |𝐴) are called ‘posterior’ probabilities.
4. If the events constitutes a disjoint partition of the sample space S and
P (Ei)≠ 0; i=1,2,…,n then for any event A in S, we have
𝑛

𝑃(𝐴) = ∑ 𝑃(𝐸𝑖 )𝑃(𝐴|𝐸𝑖 )

𝑖=1

Example 20: Basket I contains 1 white, 2 black and 3 red balls. Basket II contains 2 white, 1
black and 1 red ball. Basket III contains 4 white, 5 black and 3 red balls. One basket is chosen
at random and two balls are drawn. They happen to be white and red. What is the probability
that the balls are drawn from Basket I, II or III?
Let E1, E2, E3 be the event of choosing Basket I, II and III.
Let A be an event that the two balls drawn from the selected basket are white and red.
1
𝑃(𝐸1 ) = 𝑃(𝐸2 ) = 𝑃(𝐸3 ) =
3
1×3 1 2×1 1 4×3 2
𝑃(𝐴|𝐸1 ) = = , 𝑃(𝐴|𝐸2 ) = = , 𝑃(𝐴|𝐸3 ) = =
6𝐶2 5 4𝐶2 3 12𝐶2 11

43
 𝑃(𝐸1 )𝑃(𝐴|𝐸1 ) 33
𝑃(𝐸1 |𝐴) = =
𝐴 118
𝑃(𝐸1 )𝑃(𝐴|𝐸1 ) + 𝑃(𝐸2 )𝑃(𝐴|𝐸2 ) + 𝑃(𝐸3 )𝑃 (𝐸 )
3

𝑃(𝐸2 )𝑃(𝐴|𝐸2 ) 55
𝑃(𝐸2 |𝐴) = =
𝐴 118
𝑃(𝐸1 )𝑃(𝐴|𝐸1 ) + 𝑃(𝐸2 )𝑃(𝐴|𝐸2 ) + 𝑃(𝐸3 )𝑃 (𝐸 )
3

𝑃(𝐸3 )𝑃(𝐴|𝐸3 ) 30
𝑃(𝐸3 |𝐴) = =
𝑃(𝐸1 )𝑃(𝐴|𝐸1 ) + 𝑃(𝐸2 )𝑃(𝐴|𝐸2 ) + 𝑃(𝐸3 )𝑃(𝐴|𝐸3 ) 118

Example 21: An industrial unit has 3 machines- I, II and III which produce the same item.
Machines I and II, each produce 30% of the total output, Machine III produces 40% of the
remaining output. 2% of Machines I defective while Machine II and III, each produces 3%
defective items. All the items are put into one stockpile, and then one item is chosen at random.
Find the probability that

a) the selected item is defective.

b) What is the probability that the selected defective item was produced by Machine
I?

Let E1, E2, E3 be the events that the item chosen is produced by Machines I, II and III
respectively. Let A be an event that the item chosen is defective.
a) 𝑃(𝐴) = 𝑃(𝐸1 )𝑃(𝐴|𝐸1 ) + 𝑃(𝐸2 )𝑃(𝐴|𝐸2 ) + 𝑃(𝐸3 )𝑃(𝐴|𝐸3 )

= (0.3) × (0.02) + (0.03) × (0.03) + (0.4) × (0.03)

= 0.027
𝑃(𝐸1 )𝑃(𝐴|𝐸1 ) (0.02)×(0.3)
b) 𝑃(𝐸1 |𝐴) = =
𝑃(𝐴) 0.027

= 0.223

Example 22: The members of a consulting firm rent cars from three rental agencies: 60 percent
from agency 1, 30 percent from agency 2 and 10 percent from agency 3. If 9 percent of the
cars from agency one need a tune-up , 20 percent of the cars from agency 2 need a tune-up and
6 percent of the cars from agency 3 need a tune-up.
(a) What is the probability that a rental car delivered to the firm will need a tune-up?
(b) If a rental car delivered to the consulting firm needs a tune-up, what is the probability
that it came from agency 2?

44
Let E1, E2, E3 be the events that the car comes from rental agencies 1, 2 and 3 respectively. Let
A be an event that the car needs a tune-up.
a) 𝑃(𝐴) = 𝑃(𝐸1 )𝑃(𝐴|𝐸1 ) + 𝑃(𝐸2 )𝑃(𝐴|𝐸2 ) + 𝑃(𝐸3 )𝑃(𝐴|𝐸3 )

= (0.6) × (0.09) + (0.3) × (0.20) + (0.1) × (0.06)

= 0.12
𝑃(𝐸2 )𝑃(𝐴|𝐸2 ) (0.3)×(0.2)
b) 𝑃(𝐸2 |𝐴) = =
𝑃(𝐴) 0.12

= 0.5

Random Variable
A random variable is a function X(ω) with domain S and range (-∞, ∞) such that for
every real number a, the event [ω : X(ω) ≤ a] ∈ ℬ. In other words, we are considering a
function whose domain is the set of possible outcomes, and whose range is a subset of the set
of real.

Consider a random experiment (E) consisting of two tosses of a coin.

Outcomes: HH HT TH TT

Let X be a real number associated with the outcome of the experiment. Let X be number
of heads.

X : 2 1 1 0

Random variables are denoted by capital letters X, Y, Z,……

𝜔1 =𝐻
Examples: 1. If a coin is tossed 𝑆 = {𝜔1 , 𝜔2 }; 𝜔2=𝑇

1 𝑖𝑓 𝜔 = 𝐻
𝑋(𝜔) = {
0 𝑖𝑓 𝜔 = 𝑇

 (1,1), (1,2) . . . (1,6) 

(2,1), (2,2),    (2,6)

 (3,1) (3,2)    (3,6) 
2. If a pair of fair dice is tossed then 𝑆 =  
 (4,1)     (4,6)
 (5,1)     (5,6) 
 
 (6,1)     (6,6) 

and n(s) = 36. Let X be a r.v. with image set 𝑋(𝑆) = {1 , 2, 3, 4, 5, 6}

45
 1
𝑃(𝑋 = 1) = 𝑃{1,1} =
36
3
𝑃(𝑋 = 2) = 𝑃{(2,1), (2,2), (1,2)} =
36

Discrete Random Variable: A variable which can assume only a countable number of real
values and for which the value which the variable takes depends on chance (a real valued
function defined as a discrete sample space is called discrete random variable).

Probability Mass Function: If X is a discrete random variable, with distinct

values𝑥1 , 𝑥2 , … … . 𝑥𝑛 ,… then the function p(x) defined as

𝑃(𝑋 = 𝑥𝑖 ) = 𝑝𝑖 ; 𝑖𝑓 𝑥 = 𝑥𝑖
𝑝𝑋 (𝑥) = {
0 ; 𝑖𝑓 𝑥 ≠ 𝑥𝑖

Continuous Random Variable: A random variable X is said to be continuous if it can take

all possible values between certain limits. e.g. age, height, weight etc.

Probability density function:

𝑃(𝑥 ≤ 𝑋 ≤ 𝑥 + 𝛿𝑥)
𝑓𝑋 (𝑥) = lim
𝛿𝑥→0 𝛿𝑥

Distribution Function: Let X be a random variable. The function F defined for all real x by
𝐹(𝑥) = 𝑃(𝑋 ≤ 𝑥) = 𝑃{𝜔: 𝑋(𝜔) ≤ 𝑥} , −∞<𝑥 <∞
is called the distribution function of the random variable X.
Example 23: A random variable X has the following probability function:

X=x : 0 1 2 3 4 5 6 7

p(x) : 0 k 2k 2k 3k k2 2k2 7k2+k

i) Find k

ii) Evaluate P(X<6), P(X>6) and P(0<X<5),

iii) Determine the distribution function of X.

Since ∑7𝑥=0 𝑝(𝑥) = 1, we have

i) 𝑘 + 2𝑘 + 3𝑘 + 𝑘 2 + 2𝑘 2 + 7𝑘 2 + 𝑘 = 1

10𝑘 2 + 9𝑘 − 1 = 10

(10𝑘 − 1)(𝑘 + 1) = 0

46
  𝑘 = 1/10

ii) 𝑃(𝑋 < 6) = 𝑃(𝑋 = 0) + 𝑃(𝑋 = 1) + ⋯ + 𝑃(𝑋 = 5)

1 2 2 3 1 81
= + + + + =
10 10 10 10 100 100

P(𝑋 ≥ 6) = 1 − 𝑃(𝑋 < 6) = 19/100

4
𝑃(0 < 𝑥 < 5) = 𝑃(𝑋 = 1) + 𝑃(𝑋 = 2) + 𝑃(𝑋 = 3) + 𝑃(𝑋 = 4) = 8𝑘 =
5
iii) X=x: 0 1 2 3 4 5 6 7
𝐹𝑥 (𝑥) = 𝑃(𝑋 ≤ 𝑥): 0 1/10 3/10 5/10 4/5 81/100 83/100 1

2.3 Probability Distributions

Bernoulli Distribution

A random variable X is said to have a Bernoulli distribution with parameter p, if

its probability mass function is given by
𝑥 1−𝑥
𝑃(𝑋 = 𝑥) = { 𝑝 (1 − 𝑝) , 𝑓𝑜𝑟 𝑥 = 0,1
0 𝑜𝑡ℎ𝑒𝑟𝑤𝑖𝑠𝑒
A random experiment whose outcomes, are of two types, success S and failures F,
occurring with probabilities p and q respectively, is called a Bernoulli trial. If for this
experiment, a random variable X is defined such that if takes values 1 for S and 0 for F then X
follows Bernoulli distribution.
Let ‘p’ be the probability of success and ‘q’ be the probability of failure.
1
𝑥𝑃(𝑋 = 𝑥) = 0. 𝑃(𝑥 = 0) + 1. 𝑃(𝑥 = 1)
𝐸[𝑋] = ∑
= 0. 𝑞 + 1. 𝑝
𝑥=0

𝐸[𝑋] = 𝑝.
1 𝑥 2 𝑃(𝑋 = 𝑥) = 0. 𝑃(𝑥 = 0) + 1. 𝑃(𝑥 = 1)
𝐸[𝑋 2 ] = ∑ = 0. 𝑞 + 1. 𝑝
𝑥=0 =𝑝

𝑉[𝑋] = 𝐸[𝑋 2 ] − {𝐸[𝑋)}2 = 𝑝 − 𝑝2 = 𝑝(1 − 𝑝) = pq

The binomial distribution is called by this name because the values of the
probabilities are the success terms of binomial experiment (q+p)n

47
Binomial Distribution
Binomial distribution was introduced by James Bernoulli. Consider n independent
Bernoulli trial. Let occurrence of an event be success (S) Let occurrence of an event be failure
(F). Let p be the probability of success in any trial (and is constant for each trial). Let q= 1-p
be the probability of failure in any trial.
The probability of x successes and consequently (n-x) failures in n independent trails is a
specified order (say) SSFSFFS……FSF

P(SSFSFFS……FSF) = P(S)P(S)P(F)…….P(F)

= p.p.q…….q

= pxqn-x. (x factors and (n-x) factors)

The x successes in n trials occurs in 𝑛𝑐𝑥 ways and the probability for each of these ways is
same 𝑝 𝑥 𝑞 𝑛−𝑥 .

Hence the probability of x successes in n trials in any order is given by addition theorem
𝑛𝑐𝑥 𝑝 𝑥 𝑞 𝑛−𝑥 .

A random variable X is said to follow Binomial distribution if it assumes only non-

negative values and its probability mass function is given by

𝑛𝑐𝑥 𝑝 𝑥 𝑞 𝑛−𝑥 ; 𝑥 = 0,1,2, … , n; q = 1 − p

𝑃(𝑋 = 𝑥) = { 0 𝑜𝑡ℎ𝑒𝑟𝑤𝑖𝑠𝑒

The mean of Binomial distribution is 𝐸[𝑋] = 𝑛𝑝 and the variance is 𝑉[𝑋] = 𝑛𝑝𝑞

Example 24: If the probability is 0.70 that a student with very high grades will get into law
school, what is the probability that 3 of 5 students with very high grades will get law school?

X=3, n=5, p=0.70

𝑃[𝑋 = 3] = 5𝑐3 (0.70)3 (0.30)2 = 0.3087

Example 25: If the probability is 0.60 that a person shopping at a certain market will spend at
least Rs.100. Find the probability that among five persons shopping at this market 0, 1, 2, 3, 4
or 5 will spend at least Rs.100.

48
 x 𝑃(𝑋 = 𝑥) = 𝑛𝐶𝑥 𝑝 𝑥 𝑞 𝑛−𝑥
0 0.010
1 0.077
2 0.230
3 0.346
4 0.256
5 0.078

Example 26: Find the probability of getting exactly 2 heads in 6 tosses of a fair coin.
Probability of getting head ( p) = 1/2 , q = 1/2 , n = 6
𝑛
𝑃(𝑋 = 𝑥) = ( ) 𝑝 𝑥 𝑞 𝑛−𝑥 ; 𝑥 = 0, 1, … , 𝑛
𝑥
6 1 2 1 6−2 15
𝑃(𝑋 = 2) = ( ) (2) (2) =
2 64
Example 27: The mean and standard deviation of a Binomial distribution is 20 and 4
respectively. Find n, p and q.
Given np = 20 √𝑛𝑝𝑞 = 4
q = 16/20 = 0.8, p = 1 – q = 0.2, n = 20/0.2 = 100
Poisson Distribution
Poisson distribution was discovered by French mathematician and physicist Simeon
Denis Poisson. Poisson distribution is a limiting case of Binomial distribution under the
condition

1. n is large. (n→ ∞)
2. p (constant probability of success for each trail is very small)
3. np=λ is finite.

A random variable X is said to follow a Poisson distribution if it assumes only non-

negative values and its probability mass function is given by

𝑒 −𝜆 𝜆𝑥
𝑃(𝑋 = 𝑥) = { 𝑥! ; 𝑥 = 0,1,2 … . .
0 𝑜𝑡ℎ𝑒𝑟𝑤𝑖𝑠𝑒

E[X] = λ V[X] = λ

49
 Poisson distribution occurs when there are events which do not occur as outcomes of
a definite number of trials of an experiment but which occur at random points of time and
space. The following are some examples where Poisson distribution is used.

(i) Number of deaths due to a disease

(ii) Number of printing mistakes at each page of the book
(iii) Number of defective material in a packing of manufactured item.
(iv) Number of telephone calls received at a particular telephone exchange
(v) Number of cars passing a crossing per minute during busy hours of a day.

Example 28: A very large shipment of books contains 2% with defective bindings. Find the
probability that among 400 books taken at random from this shipment only five will have
defective bindings.

n=400, x=5, λ= np=8, p=0.02

𝑒 −8 85
𝑃(𝑋 = 𝑥) = { = 0.093
5!

Example 29: Suppose a telephone exchange receives telephone calls at the rate of 3 calls per
minutes on an average. Find the probability of receiving at most one call in one minute.
𝑒 −𝜆 𝜆𝑥
𝑃(𝑋 = 𝑥) = { ; 𝑥 = 0,1,2 … . .
𝑥!
0 𝑜𝑡ℎ𝑒𝑟𝑤𝑖𝑠𝑒

𝑒 −3 30 𝑒 −3 31
𝑃(𝑋 ≤ 1) = 0!
+ 1!
= 0.19915

Example 30: A manufacturer of cotter pins knows that 5% of his product is defective. If he
sells cotter pins in boxes of 100 and guarantees that not more than 10 pins will be defective,
what is the approximate probability that a box will fail to meet the guaranteed quality?

n=100.
Let p= probability of a defective pin = 5% = 0.05
λ = mean number of defective pins in a box of 100
= np =100*0.005 =5

Since ‘p’ is small, we may use Poisson Distribution.

Probability of x defective pins in a box of 100 is

50
 𝑒 −𝜆 𝜆𝑥 𝑒 −5 5𝑥
𝑃(𝑋 = 𝑥) = = ; 𝑥 = 0,1,2, . ..
𝑥! 𝑥!
Probability that a box will fail to meet the guaranteed quality is
10 10
𝑒 −5 5𝑥 −5
5𝑥
𝑃(𝑋 > 10) = 1 − 𝑃(𝑋 ≤ 10) = 1 − ∑ = 1 −𝑒 ∑
𝑥! 𝑥!
𝑥=0 𝑥=0

Example 31: If X and Y are independent Poisson variates such that

𝑃(𝑋 = 1) = 𝑃(𝑋 = 2)
𝑃(𝑌 = 2) = 𝑃(𝑌 = 3)
Find the variance of X-2Y
Let X~ P(𝜆) and Y~P(µ).
𝑒 −𝜆 𝜆𝑥
𝑃(𝑋 = 𝑥) = , 𝑥 = 0, 1, 2, 3, … ; 𝜆 > 0
𝑥!

𝑒 −µ µ𝑦
𝑃(𝑌 = 𝑦) = , 𝑦 = 0, 1, 2, 3, … ; µ > 0
𝑥!

𝜆2 𝑒 − 𝜆 µ2 𝑒 −µ µ3 𝑒 −µ
𝜆𝑒 −𝜆 = 𝑎𝑛𝑑 =
2! 2! 3!

𝜆𝑒 −𝜆 [𝜆 − 2] = 0 𝑎𝑛𝑑 µ2 𝑒 −µ [ µ − 3] = 0

𝜆 = 2 𝑎𝑛𝑑 µ = 3, since 𝜆 > 0, µ > 0.

Var(X)= 𝜆 = 2, 𝑎𝑛𝑑 𝑉𝑎𝑟(𝑌) = µ = 3

Covariance term vanishes since X and Y are independent.

Var(X - 2Y) = 2 + 4 *3 = 14

Example 32: The average number of trucks arriving on any one day at a truck depot in a
certain city is 12. What is the probability that on a given day fewer than 9 trucks will arrive at
the depot?
Let X be the number of trucks arriving on a given day.
𝑒 −12 𝜆𝑥
P(X < 9) = ∑8𝑥=0 = 0.1550
𝑥!

Normal Distribution

Normal distribution is a continuous distribution. It is most important probability

distribution in statistical analysis. A random variable X is said to have a normal distribution
with parameters µ (mean) and σ2 (variance) if the density function is given by
1 1 𝑥−𝜇 2
𝑓(𝑥; 𝜇, 𝜎 2 ) = exp [− { } ] ; −∞ < 𝑥 < ∞, −∞ < 𝜇 < ∞, 𝜎 2 > 0
√2𝜋 𝜎 2 𝜎

51
 Chief characteristics of Normal distribution:

(i) The normal curve is bell shaped and symmetrical.

(ii) Mean, median and mode of the distribution coincide.
(iii) β1 = 0 and β2 = 3.
(iv) X-axis is an asymptote to the curve.
(v) Area property :
a. P(µ - σ < X < µ + σ ) = 0.6826
68.26% of the values of a normal random variable are within plus or minus
one standard deviation of its mean.
b. P (µ - 2σ < X < µ + 2σ ) = 0.9544
95.44% of the values of a normal random variable are within plus or minus
two standard deviation of its mean.
c. P (µ - 3σ < X < µ + 3σ ) = 0.9973
99.73% of the values of a normal random variable are within plus or minus
three standard deviation of its mean.
(vi) The entire family of normal probability distribution is differentiated by its mean 𝜇 and its
standard deviation 𝜎.
(vii) The highest point on the normal curve is at the mean.

A random variable that has a normal distribution with mean ‘0’ and standard deviation ‘1’ is said
to have standard normal probability distribution. The letter ‘Z’ is used to denote a standard normal
random variable. The probability calculation with any normal probability distribution is made by
computing the area under the graph of the probability density function. The area under the normal curve
of standard normal distribution is computed and is available in tables that can be used in computing
probabilities. The formula used to convert any normal random variable X with mean 𝜇 and standard
𝑋−𝜇
deviation 𝜎 to the standard normal distribution is 𝑍 = .
𝜎

Example 33: X is normally distributed and the mean of X is 12 and S.D. is 4

(a) Find out the probability of the following:

i) X>20, ii) X<20, and iii) 0<X<12

(b) find 𝑥 ′ , when P(X > 𝑥 ′ ) = 0.24.

(c) Find 𝑥0′ 𝑎𝑛𝑑 𝑥1′ , 𝑤ℎ𝑒𝑛 𝑃(𝑥0′ < 𝑋 < 𝑥1′ ) = 0.50 𝑎𝑛𝑑 𝑃(𝑋 > 𝑥1′ ) = 0.25

(a) µ = 12, 𝜎 = 4, 𝑖. 𝑒. , 𝑋~𝑁(12,16).

52
 20−12
i) When X = 20, 𝑍 = =2
4

𝑃(𝑋 ≥ 20) = 𝑃(𝑍 ≥ 2) = 0.5 − 𝑃(0 ≤ 𝑍 ≤ 2) = 0.5 − 0.4772 = 0.0228

ii) 𝑃(𝑋 ≤ 20) = 1 − 𝑃(𝑋 ≥ 20) = 1 − 0.0228 = 0.9772

𝑋−12
iii) 𝑃(0 ≤ 𝑋 ≤ 12) = 𝑃(−3 ≤ 𝑍 ≤ 0) (𝑍 = 4
)

= 𝑃(0 ≤ 𝑍 ≤ 0) = 0.49865

𝑥 ′ −12
(b) When X > 𝑥 ′ , 𝑍 = 4
= 𝑧1 (𝑠𝑎𝑦)

then, we are given

𝑃(𝑋 > 𝑥 ′ ) = 0.24 => 𝑃(𝑍 > 𝑧1 ) = 0.24 𝑖. 𝑒. , 𝑃(0 < 𝑍 < 𝑧1 ) = 0.26

From tables

𝑧1 = 0.71

Hence

𝑥 ′ − 12
= 0.71 => 𝑥1′ = 12 + 4 × 0.71 = 14.84
4
(c) We are given

𝑃(𝑥0′ < 𝑋 < 𝑥1′ ) = 0.50 𝑎𝑛𝑑 𝑃(𝑋 > 𝑥1′ ) = 0.25

𝑥 ′ −12
When 𝑋 = 𝑥1′ , 𝑍= = 𝑧1
4

𝑥 ′ −12
And when 𝑋 = 𝑥0′ , 𝑍= 4
= −𝑧1

We have

𝑃(𝑍 > 𝑧1 ) = 0.25 => 𝑃(0 < 𝑍 < 𝑧1 ) = 0.25

𝑧1 = 0.67

𝑥 ′ −12
Hence = 0.67 => 𝑥1′ = 12 + 4 × 0.67 = 14.68
4

𝑥 ′ −12
4
= −0.67 => 𝑥1′ = 12 − 4 × 0.67 = 9.32

Example 34: X is a normal variate with mean 30 and S.D. 5. Find the probabilities that

𝑖) 26 ≤ 𝑋 ≤ 40, 𝑖𝑖)𝑋 ≥ 45, 𝑎𝑛𝑑 |𝑋 − 30| > 5

Here µ = 30, 𝜎 = 5.
𝑋−𝜇 26−30
i) When 𝑋 = 26, 𝑍= 𝜎
= 5 = −0.8
40−30
and when 𝑋 = 40, 𝑍 = 5 =2

53
 𝑃(26 ≤ 𝑋 ≤ 40) = 𝑃(−0.8 ≤ 𝑍 ≤ 2)
= 𝑃(−0.8 ≤ 𝑍 ≤ 0) + 𝑃(0 ≤ 𝑍 ≤ 2)
= 𝑃(−0.8 ≤ 𝑍 ≤ 0) + 0.4772
= 𝑃(0 ≤ 𝑍 ≤ 0.8) + 0.4772
= 0.2881 + 0.4772 = 0.7653
ii) 𝑃(𝑋 ≥ 45) =?
𝑋−𝜇 45 − 30
𝑋 = 45, 𝑍= = = 3
𝜎 5
𝑃(𝑋 ≥ 45) = 𝑃(𝑍 ≥ 3) = 0.5 − 𝑃(0 ≤ 𝑍 ≤ 3)

= 0.5 − 0.49865 = 0.00135

iii) P(|𝑋 − 30| ≤ 5) = 𝑃(25 ≤ 𝑋 ≤ 35) = 𝑃(−1 ≤ 𝑍 ≤ 1)

= 2𝑃(0 ≤ 𝑍 ≤ 1) = 2 × 0.49865 = 0.00135
P(|𝑋 − 30| > 5) = 1 − 𝑃(|𝑋 − 30| ≤ 5)
= 1 − 0.6826 = 0.3174

Normal distribution is most important of all the distributions in statistical theory. Most
of the discrete distributions namely, Binomial, Poisson distributions tend to normal distribution
as the sample size (n) increases. Almost all sampling distributions chi-square, Student’s-t and
Snedecor’s F tend to normal for large degrees of freedom.

Questions and Exercises

1. Define mutually exclusive events and exhaustive events with example.
2. Define random experiment and sample space with suitable examples.
3. An experiment has five possible outcomes I, II, III, IV, V, that are mutually exclusive.
Check whether the following assignments of probabilities are permissible. State reasons.
(a) P(I) = 0.21 P(II) = 0.26 P(III) = 0.58 P(IV) = 0.01 P(V) = 0.06
(b) P(I) = 0.18 P(II) = 0.20 P(III) = 0.22 P(IV) = 0.19 P(V) = 0.21
(c) P(I) = 0.10 P(II) = 0.10 P(III) = 0.30 P(IV) = 0.60 P(V) = -0.10
4. What is the probability of getting a 9 exactly once in 3 throws with a pair of dice?
5. From a bag containing 10 black and 5 white balls, a ball is drawn at random. What is the
probability that it is white?
6. A bag contains 10 white, 6 red, 4 black and 7 blue balls. 5 balls are drawn at random.
What is the probability that 2 of them are red and one black?
7. An urn contains 3 red and 4 black balls. Two balls are taken out at random. Find the
probabilities that the balls are (i) different colours, (ii) black colour, (iii) red colour.

54
8. If 3% of electric bulbs manufactured by a company are defective, find the probability
that in a sample of 100 bulbs (a) 0 (b) 3 (c) 5 bulbs are defective.
9. Two cards are randomly drawn from a deck of 52 playing cards. Find the probability
that both cards will be greater than 3 and less than 8.
10. State and prove the addition theorem of probability for two events when they are not
disjoint.
11. The odds in favour of certain event are 5:8, and odds against another event are 4:3. What
is the chance that at least one of them will happen?
12. The chance of getting an order by three salesmen A, B and C are 1/3, 3/4 and 1/2
respectively. What is the probability that any one of them will get the order?
13. Suppose that two people are randomly selected from a group of 4 women and 6 men
(a) What is the probability that both are women?
(b) What is the probability that one is a woman and other a man?
14. In a bolt factory, machines A, B and C manufacture respectively 25%, 35% and 40% of
the total. Of their output respectively 5, 4, 2 percents are defective. A bolt is drawn at
random from the product and is found to be defective. What are the probabilities that it
was manufactured by machines A, B and C?
15. Three newspapers A, B and C are published in a city and a recent survey of readers
indicate the following:
20% read A ; 16% read B ; 14% read C,
8% read both A and B ; 5% read both A and C ;
4% read B and C ; 2% read all three.
For one person chosen at random, compute the probability that
i) the person reads none of the papers,
ii) the person reads exactly one of the papers, and
iii) the person reads at least A and B if it is known that he reads at least one of the
papers.
16. Three urns are given each containing red and white chips as indicated.
Urn 1: 6 red and 4 white
Urn 2: 2 red and 6 white
Urn 3: 1 red and 8 white
i) An urn is chosen at random and a ball is drawn from the urn. The ball is red.
Find the probability that the urn chosen was urn 1.

55
 ii) An urn is chosen at random and two balls are drawn without replacement from
this urn. If both balls are red, find the probability that urn I was chosen. Under
these conditions, what is the probability that urn III was chosen.
17. State Bayes’ theorem. The members of a consulting firm rent cars from three rental
agencies: 60 percent from agency 1, 30 percent from agency 2, and 10 percent from
agency 3. If 9 percent of cars from agency 1 need a tune-up, 20 percent of the cars from
agency 2 need a tune-up and 6 percent of cars from agency 3 need a tune-up. What is
the probability that a rental car delivered to the firm will need a tune-up?
18. An electronic company manufactures MP3 players at three locations. The company at
location I manufactures 50% of the MP3 players and 1% are defective. The company at
location II manufactures 30% and 2% are defective. The company at location III
manufactures 20% and 3% are defective. If an MP3 player is selected at random, what
is the probability that it is defective?
19. From a lot of 10 items containing 3 defectives, a sample of 4 items is drawn at random.
Let the random variable X denote the number of defective items in the sample. Answer
the following, when the sample is drawn without replacement.
i) Find the probability distribution of X.
ii) Find P(X<1), P(X<1) and P(0<X<2)
20. In a survey of 500 persons were asked questions
(1) Do you own a cell phone?
(2) Do you own an ipod?
(3) Do you have internet connection?
The results of the survey were (no one answered no to all the three questions)
Cell phone 329
ipod 186
internet connection 295
cell phone and ipod 83
cell phone and internet connection 217
ipod and internet connection 63
Find the probability for the following events
(a) Answered yes to all three questions
(b) Had a cell phone but not an internet connection
(c) Had a cell phone but not an ipod or an internet connection
(d) Had an internet connection but not an ipod.

56
21. A random variable X has the following probability distribution:
X= x: 0 1 2 3 4 5 6 7 8
p(x): a 3a 5a 7a 9a 11a 13a 15a 17a
i) Determine the value of a.
ii) Find P(X<3), P(X>3), P(0<X<5).
22. Define Binomial distribution. The mean and variance of a binomial variate X with
parameters n and p are 16 and 8. Find (i) P (X = 0) (ii) P (X ≥ 2)
23. A machine manufacturing screws is known to produce 5% defectives. In a random
sample of 15 screws, what is the probability that there are
i) Exactly three defectives?
ii) not more than three defectives?
24. A multiple choice examination has 3 possible answers for each of 5 questions. What is
the probability that a student will get 4 or more correct answers just by guessing?
25. An automobile safety engineer claims that 1 in 10 accidents is due to driver fatigue. What
is the probability that at least 3 of 5 automobile accidents are due to driver fatigue?
26. The number of complaints that a establishment receives per day is a random variable
having Poisson distribution with λ = 3.3. Find the probability that it will receive only 2
complaints on any given day.
27. Suppose that 2 batteries are randomly chosen from a bin containing 12 batteries, of which
8 are good and 4 are defective. What is the expected number of defective batteries
chosen?
28. Define normal distribution and state its various properties.
29. The daily trading volumes (millions of shares) for stocks traded in a leading stock
exchange for 12 days are given below.
917, 983, 1046, 944, 723, 783, 813, 1057, 766, 836, 992, 973
Assuming the probability distribution of trading volume to be normal, compute the
probability that on a particular trading day, the trading volume will less than (i) 800
million shares and (ii) will be between 900 and 1020 million shares.
30. The time needed to complete a final examination in a particular college course is
normally distributed with mean of 80 minutes and a standard deviation of 10 minutes.
a. What is the probability of completing the exam in one hour or less?
b. What is the probability that a student will complete the exam in more than 60 minutes
but less than 75 minutes?

57
 UNIT - III
CORRELATION AND REGRESSION ANALYSIS
Contents:

3.1 – Regression Analysis

3.2 – Correlation Analysis

3.1 Regression Analysis

Managerial decisions often are based on the relationship between two or more variables. One
such technique is the correlation analysis and modeling these related variables is the Regression
Analysis. The main purpose of regression analysis is prediction. Usually, the regression model
is used to predict the values of a dependent variable based on the values of one or more
independent variables. Here, the discussion is limited to a two variable model with one
dependent (Y) and one independent (X) variable.
In general, the scatter diagram will be used to plot and explain the relationship between an X
variable on horizontal axis and a Y variable on the vertical axis. The nature of relationship
between two variables can take wither of the positive, negative and absence of relationships.
The simplest form of the relationship is the straight line or linear relationship. Apart from
straight line form, there are many other forms like exponential, power, cubic, parabolic and
many more to explain the nature of the data.

14

12

10

8
Y

0
0 2 4 6 8
X

Figure 3.1: Scatter Diagram showing linear relationship between X and Y variables

58
The straight line (linear) model can be represented as
Yi = β0+ β1 Xi + εi ; i = 1,2,…, n (3.1)
where
Yi is the response in the ith observation
Xi is the ith value of the independent variable
β0and β1 are the intercept and slope
εi is the random error and
n is the number of observations
The estimated regression equation can be obtained using least squares method.
Least Squares Method
Simple linear regression analysis is concerned with finding the straight line that fits the data
best. This can be addressed in two ways:
(a) Finding the straight line for which the difference between the actual values Yi and the
values that would be predicted from the line of regression.
(b) However, because these difference are positive and negative for some observations
instead, an approach called, the least squares method that minimizes the sum of the
squared differences is used.
2 2
̂𝑖 ) = ∑𝑛𝑖=1(𝑌𝑖 − (𝛽0 + 𝛽1 𝑋𝑖 ))
i.e., ∑𝑛𝑖=1(𝑌𝑖 − 𝑌
̂𝑖 = 𝛽0 + 𝛽1 𝑋𝑖
here, 𝑌
β0 and β1 are the unknown constants which are to be estimated. Using the least square method,
one can determine the values of β0 and β1 that minimizes the sum of squared differences. The
estimated regression equation and the expressions for determining β0 and β1 are given below:
𝑦̂𝑖 = 𝑏0 + 𝑏1 𝑥𝑖 ; i = 1,2,…,n (3.2)
where
𝑦𝑖 is the estimated value of the ith observation
b0 is the y intercept of the estimated regression line or model
b1 is the slope of the estimated regression line or model
𝑥𝑖 is the value of the ith observation of independent variable
In other words, the least square criterion can be given as min ∑𝑛𝑖=1(𝑦𝑖 − 𝑦̂)
𝑖
2

where
𝑦𝑖 is the observed value of the dependent variable for the ith observation
𝑦̂𝑖 is the estimated value of the dependent variable for the ith observation
The expressions for b0 and b1 can be derived using normal equations and are given below:

59
 𝑛 𝑛
(∑𝑖=1 𝑥𝑖 )(∑𝑖=1 𝑦𝑖 )
∑𝑛 ∑𝑛
𝑖=1 𝑥𝑖 𝑦𝑖 −
𝑖=1(𝑥𝑖 −𝑥̅ )(𝑦𝑖 −𝑦
̅) 𝑛
𝑏1 = 2 𝑜𝑟 𝑏1 = 2 (3.3)
∑𝑛
𝑖=1(𝑥𝑖 −𝑥̅ ) (∑𝑛
∑𝑛 2 𝑖=1 𝑥𝑖 )
𝑖=1 𝑥𝑖 − 𝑛

and 𝑏0 = 𝑦̅ − 𝑏1 𝑥̅ (3.4)
here
𝑥𝑖 is the value of the independent variable for ith observation
𝑦𝑖 is the value of the dependent variable for ith observation
∑𝑛
𝑖=1 𝑥𝑖
𝑥̅ is the mean value for the independent variable = 𝑛
∑𝑛
𝑖=1 𝑦𝑖
𝑦̅ is the mean value for the dependent variable = 𝑛

n is the total number of observations

In order to have a better understanding of the concept of regression, let us have an numerical
illustration using the following data on store size and its annual sales.
Problem:
It is to examine the relationship between the store size (i.e., square footage) and its annual sales,
a sample of 14 stores was selected.
Store 1 2 3 4 5 6 7 8 9 10 11 12 13 14
Square
footage
1.7 1.6 2.8 5.6 1.3 2.2 1.3 1.1 3.2 1.5 5.2 4.6 5.8 3.0
(in
1000’s)
Annual
sales
3.7 3.9 6.7 9.5 3.4 5.6 3.7 2.7 5.5 2.9 10.7 7.6 11.8 4.1
(in
lakhs)

For illustration and computational purpose, prepare the following table:

Table 3.1 Computation table

Square Annual
Store footage sales x2 y2 xy
(in 1000’s) (in lakhs)

1 1.7 3.7 2.89 13.69 6.29

2 1.6 3.9 2.56 15.21 6.24
3 2.8 6.7 7.84 44.89 18.76
4 5.6 9.5 31.36 90.25 53.2
5 1.3 3.4 1.69 11.56 4.42
6 2.2 5.6 4.84 31.36 12.32
7 1.3 3.7 1.69 13.69 4.81
8 1.1 2.7 1.21 7.29 2.97

60
 9 3.2 5.5 10.24 30.25 17.6
10 1.5 2.9 2.25 8.41 4.35
11 5.2 10.7 27.04 114.49 55.64
12 4.6 7.6 21.16 57.76 34.96
13 5.8 11.8 33.64 139.24 68.44
14 3 4.1 9 16.81 12.3
Total 40.9 81.8 157.41 594.9 302.3

Here, it clearly indicates that increasing store size has increased annual sales (figure 3.2). So,
store size (square foot in 1000’s) is independent variable, X and the annual sales purely depends
on the square foot of store size, Y.

14

12

10
Annual Sales

0
0 1 2 3 4 5 6 7
Square footage

Figure 3.2: Graphical view of linear relationship between

square footage and its annual sales

Using (3.3) and (3.4), the values of b0 and b1 can be computed as

(∑𝑛𝑖=1 𝑥𝑖 )(∑𝑛𝑖=1 𝑦𝑖 )
∑𝑛𝑖=1 𝑥𝑖 𝑦𝑖 −
𝑏1 = 𝑛
𝑛
(∑ 𝑥 )2
∑𝑛𝑖=1 𝑥𝑖2 − 𝑖=1 𝑖
𝑛
(40.9)(81.8)
302.3 −
𝑏1 = 14
(40.9)2
157.41 − 14

61
 302.3 − 238.9728
𝑏1 =
157.41 − 119.4864
63.32715
𝑏1 =
37.92358
⟹ 𝑏1 = 1.66986
𝑏0 = 𝑦̅ − 𝑏1 𝑥̅
𝑏0 = 5.84285 − (1.66986) ∗ (2.92143)
⟹ 𝑏0 = 0.96447
Therefore, b0 = 0.96447 and b1 =1.66986
So, the estimated regression equation is
Annual sales = 0.96447 + 1.66986 (square footage)
The slope of the estimated regression (b1 = 1.66986) is positive, implying that as square foot
of store size increases, annual sales increases. In fact, we can conclude that an increase in
square footage of store size of 2000 sq. ft is associated with an increase of 3.324 times in annual
sales. That is, if we wish to predict the annual sales for a store size of 2000 (=x) square feet,
we would have
Annual sales = 0.96447 + (1.66986) (2000)
Annual Sales = 0.96447 + 3339.72
Therefore, Annual Sales = 3340.6844
Hence, we would predict annual sales of $ 3340.6844 for this particular store size.
Once, the regression equation is estimated, the next step is to compute the coefficient of
determination, R2, is given by
𝑠𝑢𝑚 𝑜𝑓 𝑠𝑞𝑢𝑎𝑟𝑒𝑠 𝑑𝑢𝑒 𝑡𝑜 𝑟𝑒𝑔𝑟𝑒𝑠𝑠𝑖𝑜𝑛 𝑆𝑆𝑅
𝑅2 = =
𝑡𝑜𝑡𝑎𝑙 𝑠𝑢𝑚 𝑜𝑓 𝑠𝑞𝑢𝑎𝑟𝑒𝑠 𝑆𝑆𝑇
(∑𝑛𝑖=1 𝑦𝑖 )2
∑𝑛𝑖=1(𝑦̂𝑖 2
− 𝑦̅) 𝑏0 ∑𝑛𝑖=1 𝑦𝑖 + 𝑏1 ∑𝑛𝑖=1 𝑥𝑦 −
𝑅2 = = 𝑛
𝑛
∑𝑛𝑖=1(𝑦𝑖 − 𝑦̅)2 (∑ 𝑦 ) 2
∑𝑛𝑖=1 𝑦𝑖2 − 𝑖=1 𝑖
𝑛
For the considered problem, R2 can be computed as follows
(81.8)2
SSR = (0.964478) * (81.8)+ (1.66986) * (302.3) – = 105.74726
14
(81.8)2
SST = 594.9 - 14

SST = 594.9 – 477.94571

SST = 116.9542
𝑆𝑆𝑅 105.74726
∴ 𝑅2 = = = 0.9042
𝑆𝑆𝑇 116.9542

62
The ratio SSR/SST, which will take values between zero and one, is used to evaluate the
goodness of fit for the estimated regression equation.
R2 can be interpreted as the percentage of the total sum of squares that can be explained by
using the estimated regression equation. That is, for the considered problem, the R2 value is
0.9042, with this values, we can conclude that 90.42 % of the total sum of squares can be
explained by using the estimated regression equation, 𝑦̂ = 0.964478 + 1.66986 𝑥 to predict
annual sales. In other words, 90.42% of variability in annual sales can be explained by the size
of the store and annual sales.
Let us consider another example to explain the procedure of estimating regression equation in
a better manner. Data is collected from a sample of 10 restaurants located near college
campuses. The data is presented in the following table:
Student Quarterly
Restaurant Population sales
(in 1000’s) (in 1000’s)
1 2 58
2 6 105
3 8 88
4 8 118
5 12 117
6 16 137
7 20 157
8 20 169
9 22 149
10 26 202
Here from the given information, student population influences the quarterly sales of the
restaurants, hence, student population can be considered as independent variable (x) and
quarterly sales as dependent variable (y). Now, let us compute the following terms such as
mean of x and y, (𝑥𝑖 − 𝑥̅ ), (𝑦𝑖 − 𝑦̅), (𝑥𝑖 − 𝑥̅ )(𝑦𝑖 − 𝑦̅) and (𝑥𝑖 − 𝑥̅ )2.
Student Quarterly
Restaurant Population sales (𝑥𝑖 − 𝑥̅ ) (𝑦𝑖 − 𝑦̅) (𝑥𝑖 − 𝑥̅ )(𝑦𝑖 − 𝑦̅) (𝑥𝑖 − 𝑥̅ )2
(in 1000’s) (in 1000’s)
1 2 58 -12 -72 864 144
2 6 105 -8 -25 200 64
3 8 88 -6 -42 252 36
4 8 118 -6 -12 72 36
5 12 117 -2 -13 26 4
6 16 137 2 7 14 4
7 20 157 6 27 162 36
8 20 169 6 39 234 36
9 22 149 8 19 152 64
10 26 202 12 72 864 144
Total 140 1300 2840 568
Mean 14 130

63
 ∑𝑛𝑖=1(𝑥𝑖 − 𝑥̅ )(𝑦𝑖 − 𝑦̅) 2840
𝑏1 = = =5
∑𝑛𝑖=1(𝑥𝑖 − 𝑥̅ )2 568

𝑏0 = 𝑦̅ − 𝑏1 𝑥̅
𝑏0 = 130 − 5 ∗ 14
𝑏0 = 60
Thus, the estimated regression equation is 𝑦̂ = 60 + 5𝑥
In similar lines, the interpretation about the parameters of the estimated regression equation
can be given as follows:
The slope, b1 = 5 and is positive, which implies that as student population increases, sales
increases. In fact, we can conclude that an increase in the student population of 1000 is
associated with an increase of $5000 in expected sales. Suppose if we wish to predict quarterly
sales for a restaurant to be located near a campus with 16,000 students, we would compute the
quarterly sales, ŷ as, ŷ=60+5 (16) = 140. Hence, we would predict quarterly sales of
$140,000 for that particular restaurant. Further, the coefficient of determination can be
computed as R2 = SSR/SST = 14,200/15,730 = 0.9027
We can conclude that 90.27% of the total sum of squares can be explained by using the
estimated regression equation. In other words, 90.27% of the variability in quarterly sales can
be explained by the linear relationship between the size of the student population and quarterly
sales.

3.2 – Correlation Analysis

Correlation is another way of assessing the relationship between variables. To be more precise,
it measures the extent of correspondence between the ordering of two random variables. There
is a large amount of resemblance between regression and correlation but for their methods of
interpretation of the relationship. For example, a scatter diagram is of tremendous help when
trying to describe the type of relationship existing between two variables.

Measuring correlation

We make use of the linear product-moment correlation coefficient, also known as Pearson’s
correlation coefficient, to express the strength of the relationship. This coefficient is generally
used when variables are of quantitative nature, that is, ratio or interval scale variables.

64
Pearson’s correlation coefficient is denoted by r and is defined by
𝑛 ∑ 𝑥𝑦 − ∑ 𝑥 ∑ 𝑦
𝑟𝑥𝑦 =
√𝑛 ∑ 𝑥 2 − (∑ 𝑥)2 ∗ √𝑛 ∑ 𝑦 2 − (∑ 𝑦)2

The value of r always lies between –1 and 1 inclusive, that is, −1 ≤ r ≤1 . If Y increases when
X increases, we say that there is positive or direct correlation between them. However, if Y
decreases when X increases (or vice versa), then we say that they are negatively or inversely
correlated. The reader must have noticed that direct and inverse are terms that are used in the
context of variation or proportionality.

Interpretation of the correlation coefficient

The extreme values of r, that is, when r = ±1, indicate that there is perfect (positive or negative)
correlation between X and Y. However, if r is 0, we say that there is no or zero correlation.
When r = 0, we may not assert that there is no correlation at all between X and Y. Pearson’s
correlation coefficient is meant to measure linear relationship only. It should not be used in the
case of non-linear relationships since it will obviously lead to an erroneous interpretation.
The remaining values, falling in subintervals of [–1, 1], describe the relationship in terms of its
strength.

We observe that the strength of the relationship between X and Y is the same whether r = 0.85
or – 0.85. The only difference is that there is direct correlation in the first case and inverse
correlation in the second. We should bear in mind that r is the linear correlation coefficient and
that, as mentioned earlier, its value can be wrongly interpreted whenever the relationship
between X and Y is non-linear. That is the reason why we should have a look at a scatter diagram
of points (x, y) and verify whether the relationship is, for example, of quadratic, logarithmic,
exponential or trigonometric (briefly, non-linear) nature.
If r = 0, we should not jump to the conclusion that there is no correlation at all between X and
Y. Consider the case where there is perfect (but unsuspected) non-linear correlation between
the two variables.
With practice and experience, it is even possible to know approximately the value of r by
inspection of a scatter diagram. The location (amount of scattering) of the points with respect
to the least-squares regression line indicates the strength of the relationship between the
variables. The more scattered the points are, the weaker is the relationship and the closer is the

65
value of r to zero.
Example
Determine the degree of correlation between the prime lending rate and the inflation rate and
test whether it is significant at 0.05 level.
Inflation rate 3.3 5.8 6.2 6.5 7.6 9.1 11
Prime lending rate 5.2 6.8 8 6.9 9 7.9 10.8

Solution
We first summarise the data from the above table as follows:

Sample
number Prime lending
Inflation rate X2 Y2 XY
rate

1 3.3 5.2 10.89 27.04 17.16

2 5.8 6.8 33.64 46.24 39.44
3 6.2 8 38.44 64 49.6
4 6.5 6.9 42.25 47.61 44.85
5 7.6 9 57.76 81 68.4
6 9.1 7.9 82.81 62.41 71.89
7 11 10.8 121 116.64 118.8
Totals ∑ 𝑥 =49.5 ∑ 𝑦 = 54.6 2
∑ 𝑥 =386.79 ∑ 𝑦 2 =444.94 ∑ 𝑥𝑦 =410.14

Pearson’s correlation coefficient is calculated as

(7)(410.14) −(49.5)(54.6)

r= [( 7)(386.79) −(49.5)2 ][(7)(444.94) −(54.6)2 ] = 0.9082

Hence, there is a very strong direct correlation between inflation rate and prime lending rate.

EXERCISES
Problems on Correlation
1. Five observations taken for two variables follow.
xi 4 6 11 3 16
yi 50 50 40 60 30
Compute and interpret the sample correlation coefficient. At α = 0.05 test for significant
correlation.
2. Nielsen Media Research provides two measures of the television viewing audience: a
television program rating, which is the percentage of households with televisions
watching a program, and a television program share, which is the percentage of
households watching a program among those with televisions in use. The following

66
 data show the Nielsen television ratings and share data for the Major League Baseball
World Series over a nine-year period (Associated Press, October 27, 2003).
Rating 19 17 17 14 16 12 15 12 13
Share 32 28 29 24 26 20 24 20 22
Compute the sample correlation coefficient. What does this value tell us about the relationship
between rating and share? At α = 0.05 level, does there appear to be a relationship between
rating and share?
3. A department of transportation’s study on driving speed and miles per gallon for
midsize automobiles resulted in the following data:
Speed (Miles per Hour) 30 50 40 55 30 25 60 25 50 55
Miles per Gallon 28 25 25 23 30 32 21 35 26 25
Compute and interpret the sample correlation coefficient. Test the correlation at α = 0.10.
4. The Dow Jones Industrial Average (DJIA) and the Standard & Poor’s 500 Index (S&P
500) are both used to measure the performance of the stock market. The DJIA is based
on the price of stocks for 30 large companies; the S&P 500 is based on the price of
stocks for 500 companies. If both the DJIA and S&P 500 measure the performance of
the stock market, how are they correlated? The following data show the daily percent
increase or daily percent decrease in the DJIA and S&P 500 for a sample of nine days
over a three-month period (The Wall Street Journal, January 15 to March 10, 2006).
DJIA 0.2 0.82 -0.99 0.04 -0.24 1.01 0.3 0.55 -0.25
S&P 500 0.24 0.19 -0.91 0.08 -0.33 0.87 0.36 0.83 -0.16
Compute the sample correlation coefficient for these data and test at α = 0.05.
5. The daily high and low temperatures for 14 cities around the world are shown (The
Weather Channel, April 22, 2009).
City High Low City High Low
Athens 68 50 London 67 45
Beijing 70 49 Moscow 44 29
Berlin 65 44 Paris 69 44
Cairo 96 64 Rio de Janeiro 76 69
Dublin 57 46 Rome 69 51
Geneva 70 45 Tokyo 70 58
Hong Kong 80 73 Toronto 44 39
What is the correlation between the high and low temperatures? Is the observed correlation
significant at α = 0.05? Discuss.
6. Education and crime rate ratings for selected US cities are given below. Education
rating is an index for public/teacher ratio, academic options in higher education the
higher the rating the better and other factors and crime is the crime rate per 100 people

67
 City Education Crime
(x) (y)
ordered data
New York 30 25
Detroit 31 16
Los Angeles 32 20
Boston 35 12
Chicago 35 10
Washington, DC 36 13
Compute and interpret the correlation coefficient and test at α = 0.05.
7. The data below summarized the relationship between number of employees (x) and
number of openings (y) at 11 Boston area hospitals.
∑ 𝑥 = 56,562; ∑ 𝑥 2 = 456,525,234; ∑ 𝑦 = 2611; ∑ 𝑦 2 = 818,149; ∑ 𝑥𝑦 =
18,267,023. Find the correlation coefficient, r and test at α = 0.05.
8. The number of hours of study and the exam scores of 10 students are collected. Find
the degree of correlation between the hours invested and scores obtained and test at α
= 0.05.
No. of Hrs of Study 4 6 8 4 2 1 5 7 4 6
Exam Score 85 80 92 70 65 60 89 82 81 95

Problems on Regression
9. Given are five observations for two variables, x and y.
xi 1 2 3 4 5
yi 3 7 5 11 14
Construct a simple linear regression model for the data and use the estimated regression
equation to predict the value of y when x = 4.
10. The following data were collected on the height (inches) and weight (pounds) of women
swimmers.
Height 68 64 62 65 66
Weight 132 108 102 115 128
a. Develop the estimated regression equation by computing the values of regression
coefficients.
b. If a swimmer’s height is 63 inches, what would her weight be?
11. Elliptical trainers are becoming one of the more popular exercise machines. Their
smooth and steady low-impact motion makes them a preferred choice for individuals
with knee and ankle problems. But selecting the right trainer can be a difficult process.
Price and quality are two important factors in any purchase decision. Are higher prices

68
 generally associated with higher quality elliptical trainers? Consumer Reports
conducted extensive tests to develop an overall rating based on ease of use, ergonomics,
construction, and exercise range. The following data show the price and rating for eight
elliptical trainers tested (Consumer Reports, February 2008).
Brand and Model Price ($) Rating
Precor 5.31 3700 87
Keys Fitness CG2 2500 84
Octane Fitness Q37e 2800 82
LifeFitness X1 Basic 1900 74
NordicTrack Audiostrider 990 1000 73
Schwinn 430 800 69
Vision Fitness X6100 1700 68
ProForm XP 520 Razor 600 55
Use the estimated regression equation to predict the rating for an elliptical trainer with a price
of $1500.
12. The cost of a previously owned car depends upon factors such as make and model,
model year, mileage, condition, and whether the car is purchased from a dealer or from
a private seller. To investigate the relationship between the car’s mileage and the sales
price, data were collected on the mileage and the sale price for 10 private sales of model
year 2000 Honda Accords (PriceHub website, October 2008).
Miles (1000s) 90 59 66 87 90 106 94 57 138 87
Price ($1000s) 7 7.5 6.6 7.2 7 5.4 6.4 7 5.1 7.2

a. Use the least squares method to develop the estimated regression equation.
b. Provide an interpretation for the slope of the estimated regression equation.
c. Predict the sales price for a 2000 Honda Accord with 100,000 miles.
13. A sales manager collected the following data on annual sales and years of experience.
Salesperson Years of Experience Annual Sales ($1000s)
1 1 80
2 3 97
3 4 92
4 4 102
5 6 103
6 8 111
7 10 119
8 10 123
9 11 117
10 13 136

69
 a. Develop an estimated regression equation that can be used to predict annual sales
given the years of experience.
b. Use the estimated regression equation to predict annual sales for a salesperson with
9 years of experience.
14. Bergans of Norway has been making outdoor gear since 1908. The following data show
the temperature rating (F°) and the price ($) for 11 models of sleeping bags produced
by Bergans (Backpacker 2006 Gear Guide).
Temperature
Model Price($)
Rating (F°)
Ranger 3-Seasons 12 319
Ranger Spring 24 289
Ranger Winter 3 389
Rondane 3-Seasons 13 239
Rondane Summer 38 149
Rondane Winter 4 289
Senja Ice 5 359
Senja Snow 15 259
Senja Zero 25 229
Super Light 45 129
Tight & Light & 25 199
a. Use the least squares method to develop the estimated regression equation.
b. Predict the price for a sleeping bag with a temperature rating (F°) of 20.
15. To avoid extra checked-bag fees, airline travelers often pack as much as they can into
their suitcase. Finding a rolling suitcase that is durable, has good capacity, and is easy
to pull can be difficult. The following table shows the results of tests conducted by
Consumer Reports for 10 rolling suitcases; higher scores indicate better overall test
results (Consumer Reports website, October 2008).
Brand Price ($) Score
Briggs & Riley 325 72
Hartman 350 74
Heys 67 54
Kenneth Cole Reaction 120 54
Liz Claiborne 85 64
Samsonite 180 57
Titan 360 66
TravelPro 156 67
Tumi 595 87
Victorinox 400 77
a. Use the least squares method to develop the estimated regression equation.

70
 b. Provide an interpretation for the slope of the estimated regression equation.
c. The Eagle Creek Hovercraft suitcase has a price of $225. Predict the score for this
suitcase using the estimated regression equation.
16. A personal watercraft (PWC) is a vessel propelled by water jets, designed to be operated
by a person sitting, standing, or kneeling on the vessel. In the early 1970s, Kawasaki
Motors Corp. U.S.A. introduced the JET SKI® watercraft, the first commercially
success-ful PWC. Today, jet ski is commonly used as a generic term for personal
watercraft. The following data show the weight (rounded to the nearest 10 lbs.) and the
price (rounded to the nearest $50) for 10 three-seater personal watercraft (Jetski News
website,2006).
Make and Model Weight (lbs.) Price ($)
Honda AquaTrax F-12 750 9500
Honda AquaTrax F-12X 790 10500
Honda AquaTrax F-12X GPScape 800 11200
Kawasaki STX-12F Jetski 740 8500
Yamaha FX Cruiser Waverunner 830 10000
Yamaha FX High Output Waverunner 770 10000
Yamaha FX Waverunner 830 9300
Yamaha VX110 Deluxe Waverunner 720 7700
Yamaha VX110 Sport Waverunner 720 7000
Yamaha XLT1200 Waverunner 780 8500
a. Use the least squares method to develop the estimated regression equation.
b. Predict the price for a three-seater PWC with a weight of 750 pounds.

71
 UNIT - IV

STATISTICAL ESTIMATION AND TESTING

CONTENTS:

4.1 – Point Estimate of the Population Mean and Variance

4.2 – Point Estimate for the population proportion and variance

4.3 – Statistical Hypotheses Testing

In statistical hypotheses testing, the main objective is to determine an appropriate value of a

parameter on the basis of a sample statistic. Usually, we can provide two estimates of the
populations namely, point and interval estimates. A Point estimate is a sample statistic that is
used to estimate an unknown population parameter. An Interval estimate is a range of values
used to estimate a population parameter. In other words, interval estimate indicates the error
by the extent of its range and by the probability of the true population parameter lying within
that range. An estimator is said to be a good estimator if it possess the qualities of unbiasedness,
consistency and efficiency.

4.1 Point Estimate of the Population Mean and Variance

It is already defined that point estimator helps in drawing inferences about a population
parameter by estimating the value of an unknown parameter using a single point. Sample mean
is the best estimator of the population mean since it is unbiased, consistent and most efficient
estimator. However, to claim that the sample mean is best estimator of population mean, either
the sample should be drawn from normal population or the sample should be sufficiently large
so that the sampling distribution approximates to normal distribution. Therefore, the point
estimate of the population mean is given by μ̂= ̅
X.

A point to be noted that the point estimator gets closer to the population parameter value as
sample size increases. However, the degree of error will not be reflected by point estimator
and thus we can make use of the concept of interval estimator. By doing this, the probability
value of the population parameter will lie between two values with middle value being
represented by point estimator. To obtain such interval, the estimate which can be used for the
population standard deviation, σ, is the sample standard deviation s, and is given by

√ ∑𝑛𝑖=1(𝑋𝑖 − 𝑋̅)2
𝜎̂ 𝑠 𝑛−1
𝜎̂𝑋̅ = = =
√𝑛 √𝑛 − 1 𝑛 −1

72
 ∑𝑛 ̅ 2
𝑖=1(𝑋𝑖 − 𝑋)
where s = √ is the unbiased estimator of the population standard deviation, σ
̂.
𝑛−1

Further, the interval estimates for the estimator of population mean can be observed under two
cases, namely with known and unknown σ. In practice, the data which are collected will have
sampling variability. This can be addressed in terms of sampling error. With point estimator,
we cannot explain the error involved in the samples collected. Our knowledge of sampling
error would indicate that the standard error provides a measure of this error in terms of a
probability value that the value of the population mean lie within a specified interval. This
interval is termed as interval estimate. Let us assume that the sampling distribution of the mean
follows normal distribution, then the 95% of the distribution lies within plus or minus Zα/2
𝜎
standard deviation of the mean, i.e., 𝑋̅ ± 𝑍𝛼⁄2 𝑆. 𝐸(𝑋̅) ⇒ 𝑋̅ ± 𝑍𝛼⁄2
√𝑛

𝜎
With 95% level, the confidence interval for the mean is 𝑋̅ ± (1.96) and the lower, upper
√𝑛
𝜎 𝜎
confidence intervals are given by(𝑋̅ − 𝑍𝛼⁄2 , 𝑋̅ + 𝑍𝛼⁄2 ). In general, it is known that the
√ 𝑛 √𝑛

point estimate will lie between these lower and upper confidence intervals.

Interval Estimation of the Population Mean (σ known and unknown)

Let x1, x2, …, xn be the 'n' random samples drawn from a normal population with mean µ and
variance σ2, i.e., X~N(µ, σ2), then the sampling distribution of the sample mean is
𝜎
𝑋̅ ~ 𝑁(𝜇, 𝜎 2 ). The confidence interval of the population mean is given by 𝑋̅ − 𝑍𝛼⁄2 ≤
√𝑛
𝜎
μ ≤ 𝑋̅ + 𝑍𝛼⁄2
√𝑛

But in most cases, the population standard deviation would be unknown value and we would
use the sample value to estimate the population value. Then the distribution of sample standard
deviation underpins students' t - distribution with (n-1) degrees of freedom. Therefore, the
100(1-α)% confidence interval for the population mean when σ is unknown can be given as
𝑠 𝑠
𝑋̅ − 𝑡𝑛−1 ≤ μ ≤ 𝑋̅ + 𝑡𝑛−1
√𝑛 √𝑛

The above expression is to be used when there are less than 30 samples. Suppose, the samples
drawn are greater than (or) equal to 30, then the expression can be rewritten as
𝑠 𝑠
𝑋̅ − 𝑍𝛼⁄2 ≤ μ ≤ 𝑋̅ + 𝑍𝛼⁄2
√𝑛 √𝑛

73
4.2 Point Estimate for the population proportion and variance

In some situations, the point estimates can be obtained using proportions calculated using a
sample instead of using sample mean, we can use the sample proportions to give point estimates
of the population proportions and the standard error of the proportion are as follows:

Estimate of the population proportion, 𝑃̂ = 𝑝

𝑃̂ (1−𝑃̂ )
Estimate of the standard error, 𝜎̂𝑝 = √ 𝑛

If more than one sample is taken from a population then the population proportion is obtained
𝑛1 𝑝1 +𝑛2 𝑝2
by the pooled sampled proportion, i.e., 𝑃̂ = 𝑛1 +𝑛2

Interval estimate of a population proportion

If the population is normally distributed or the sample size is large, then the confidence interval
for a proportion is given by

𝑝(1 − 𝑝) 𝑝(1 − 𝑝)
𝑝−𝑍∗√ ≤𝑃 ≤ 𝑝+𝑍∗√
𝑛 𝑛

Problems on Means and Proportions

1. A random samples of five values was taken from a population, 8.1,6.5, 4.9, 7.3 and 5.9.
Estimate the populations mean, standard deviation and standard error of the estimate
for the population mean.
Solution: Given a random sample of five values and let denote it by X, now compute
mean and standard deviation for the given data i.e.,
Sample number Xi (Xi - X̅ ) (Xi - ̅
X)2
1 8.1 1.56 2.4336
2 6.5 -0.04 0.0016
3 4.9 -1.64 2.6896
4 7.3 0.76 0.5776
5 5.9 -0.64 0.4096
Total 32.7 6.112

∑𝑛𝑖=1 𝑋𝑖
𝑋̅ = ; 𝑛 = 5 𝑎𝑛𝑑 𝑖 = 1,2, … , 5
𝑛
8.1 + 6.5 + 4.9 + 7.3 + 5.9
𝑋̅ =
5

74
 32.7
𝑋̅ = = 6.54
5

∑5𝑖=1(𝑋𝑖 − 𝑋̅)2 6.112

𝑠= √ =√ = √1.528 = 1.2361
𝑛−1 4
̂
𝜎 𝑠 1.2361 1.2361
Standard error of mean, 𝜎̂𝑋 = = = = = 0.5528
√𝑛 √𝑛 √5 2.2361

Therefore, the values of the unbiased estimates of mean, standard deviation and standard error
of mean are 6.54, 1.2361 and 0.5528 respectively.
2. (case of σ known) 8 samples measuring the length of cloth are sampled from a
population where the length is normally distributed with population standard deviation
0.2. Calculate 95% confidence interval for the population mean based on the 8 samples,
4.9, 4.7, 5.1, 5.4, 4.7, 5.2, 4.8 and 5.1.
Solution: Given that 8 samples are drawn from normally distributed population and relates to
the length of cloth.
Population standard deviation = 0.2
In order to obtain the 95% confidence interval for the mean, first we need to compute sample
mean. Since the population standard deviation is known, the expression for obtaining
𝜎
confidence interval is given by 𝑋̅ ± 𝑍𝛼⁄2
√𝑛

Here, 𝑋̅ = sample mean which is to computed

𝑍𝛼⁄2 = the table value (𝑍0.05⁄ = 1.96)
2

σ = population standard deviation = 0.2

n = sample size = 8
∑𝑛𝑖=1 𝑋𝑖
𝑋̅ = ; 𝑛 = 8 𝑎𝑛𝑑 𝑖 = 1,2, … , 8
𝑛
4.9 + 4.7 + 5.1 + 5.4 + 4.7 + 5.2 + 4.8 + 5.1
𝑋̅ =
8
39.9
𝑋̅ = = 4.9875
8
0.2
The confidence interval for the mean is 4.9875 ±(1.96)
√8

0.2
⇒ 4.9875 ± (1.96)
2.8284
⇒ 4.9875 ± (1.96)(0.0707)
⇒ 4.9875 ± 0.1386
Therefore, 95% confidence interval for the mean is (4.8489, 5.1231)

75
 3. (case of unknown σ) The masses in gram of 13 ball bearings taken at random from a
batch are 21.4, 23.1, 25.9, 24.7, 23.4, 24.5, 25.0, 22.5, 26.9, 26.4, 25.8, 23.2 and 21.9.
Calculate 95% confidence interval for the mean mass of the population assuming that
the samples were drawn from the normal population.
Solution: Given that 13 samples were drawn at random from a normally distributed population.
Now, we need to calculate the 95% confidence interval for the mean. In this problem,
information about population standard deviation is not known, so, we need to compute the
sample standard deviation from the data.
The expressions for obtaining sample mean and standard deviation are

∑𝑛𝑖=1 𝑋𝑖 ∑𝑛 (𝑋𝑖 − 𝑋̅)2

𝑋̅ = 𝑎𝑛𝑑 𝑠 = √ 𝑖=1
𝑛 𝑛−1
𝑠 𝑠
and the 95% confidence interval for mean when σ is unknown is (𝑋̅ − 𝑡𝛼⁄2 , 𝑋̅ + 𝑡𝛼⁄2 )
√𝑛 √𝑛

Sample number Xi (Xi - ̅

X)2
1 21.4 7.8832
2 23.1 1.2269
3 25.9 2.8639
4 24.7 0.2424
5 23.4 0.6524
6 24.5 0.0854
7 25.0 0.6277
8 22.5 2.9162
9 26.9 7.2484
10 26.4 4.8062
11 25.8 2.5354
12 23.2 1.0154
13 21.9 5.3255
Total 314.7 37.429

314.7
𝑋̅ = = 24.2077
13

∑13 (𝑋𝑖 − 𝑋̅)2 37.429

𝑠 = √ 𝑖=1 =√ = √3.1191 = 1.7661
13 − 1 12

t0.025 at 12 degrees of freedom = 2.1788

therefore, 95% confidence interval for mean is
1.7661
⇒ 24.2077 ± (2.1788)
√13

76
 ⇒ 4.9875 ± (1.96)(0.4898)
⇒ 4.9875 ± 1.0672
The lower and upper 95% confidence interval for the mean of the given samples is (23.1405,
25.2749)
4. A random samples of 20 children in a large school were asked a question and 12
answered correctly. Estimate the proportion of children in the school who answered
correctly and the standard error of this estimate.
Solution: Given that a random sample of 20 children in a large school were selected and asked
a question, i.e., n =20 of which 12 children answered correctly i.e., X = 12 (say).
𝑋 12
Now, the sample proportion, p = 𝑛 = = 0.6
20

𝑝̂(1−𝑝̂)
The standard error of this sample proportion is 𝜎̂𝑝 = √ 𝑛

0.6 ∗ (1 − 0.6)
⇒ 𝜎̂𝑝 = √
20

0.6 ∗ 0.4
⇒ 𝜎̂𝑝 = √
20

0.24
⇒ 𝜎̂𝑝 = √ = 0.1095
20

Therefore, the sample proportion and the standard error values are 0.6 and 0.1095

5. 80 samples were drawn at random from a normally distributed population with a sample
proportion of 0.26. Using this information, construct the 95% confidence interval for
the sample proportion.
Solution: Basing on the given information, it is understood that n is large i.e., n =80 and
p=0.26.
Therefore, 95% confidence interval for the sample proportion can be obtained as
𝑝(1−𝑝)
𝑝 ± 𝑍𝛼⁄2 ∗ √ 𝑛

0.26 ∗ (1 − 0.26)
⇒ 0.26 ± (1.96) ∗ √
80

0.26 ∗ (0.74)
⇒ 0.26 ± (1.96) ∗ √
80

77
 ⇒ 0.26 ± (1.96) ∗ (0.0490)
⇒ 0.26 ± (0.09611)
Thus, the lower and upper 95% confidence intervals are (0.16389, 0.35611).

Z- test for two Proportions

Let p1 and p2 are the two proportions from two independent normally distributed populations
with sample sizes n1 and n2. Here the statement to be test is that the sample proportions of two
populations are one and the same. Thus, the null and alternative hypothesis can be defined as
H0: the sample proportions, p1 and p2 are equal, i.e., p1 = p2
H1: the sample proportions, p1 and p2 are not equal, i.e., p1 ≠ p2
The test statistic for testing the proportion based hypothesis is
𝑝1 − 𝑝2
𝑍= ~ 𝑁(0,1)
1 1
√𝑃𝑄 ( + )
𝑛 𝑛 1 2
𝑛1 𝑝1 +𝑛2 𝑝2
where 𝑃̂ = and 𝑄̂ = 1 − 𝑃̂
𝑛1 +𝑛2

If Zcal ≤ Zcri at α % level of significance, there is no evidence to reject the null hypothesis,
otherwise.
Example:
A Business Week/Harris survey asked senior executives at large corporations their opinions
about the economic outlook for the future. One question was, “Do you think that there will be
an increase in the number of full-time employees at your company over the next 12 months?”
In the current survey, 220 of 400 executives answered yes, while in a previous year survey,
192 of 400 executives had answered yes.
Solution: From the given information, the proportions for two populations can be calculated
as follows. Sample proportion for current survey is 220/400 = 0.55 and the sample proportion
for the previous survey is 0.48.
𝑛1 𝑝1 +𝑛2 𝑝2 400∗0.55+400∗0.48
and 𝑃̂ = = = 0.515 and 𝑄̂ = 1 − 𝑃̂ = 1 − 0.515 = 0.485 with
𝑛1 +𝑛2 400+400

n1 = 400 and n2 = 400 are

Now on substituting the above computed values, the Z test statistic is given by

𝑝1 − 𝑝2 0.55 − 0.48
𝑍= = = 1.99
1 1
√𝑃𝑄 ( + ) √0.515 ∗ 0.485 ( 1 + 1 )
𝑛1 𝑛2 400 400

78
Z0.05 = 1.96 and from the results it can be concluded that there is an evidence to reject the null
hypothesis. Thus, there will be an increase in the number of full-time employees in the
company over the next 12 months.

4.3 STATISTICAL HYPOTHESES TESTING

Hypothesis: Statement about the population (or) parameters of the population. In other words,
it is a statement of the perceived value of a variable or relationship between two or more
variables that can be measured.
In dealing with a hypothesis test, we have to formulate our initial research into two statements
such as Null and Alternative hypothesis.
Null Hypothesis (H0): It is the hypothesis of no difference and is formulated in anticipation
of being rejected as false.
Alternative Hypothesis (H1): It is the positive proposition which states that a difference exists.
Errors: In statistical hypotheses testing, usually we come across four possible states of
decision making, of which two states provides correct decision making and the rest two states
have wrong decision making process. The following matrix or table will given an idea about
these four possible states.
H0 is true H0 is false
Accept H0 Correct Type II error
Reject H0 Type I error correct

These two type I and II errors are to be considered to minimize the error in decision making
process.
Type I error:
It is the probability that rejecting H0 when it is true and is denoted by α and is given as
α = P(reject H0 | H0 is true)
The maximum probability of occurring type I error is the level of significance.
Type II error:
It is the probability that acepting H0 when H1 is true and is denoted by β and is given as
β = P(accept H0 | H1 is true)

in testing a particular hypothesis, sample size plays a prominent role. Basing on the number
of samples involved in the study, the selection about an appropriate statistical test will be done.
In general, the branch of statistical tests is divided into small and large samples. It is the rule

79
of thumb that if number of samples are less than 30, then we need to make use of exact sampling
tests like, t, F and chi-square. If the number of samples are greater than 30, then the large
sample tests namely Z test should be used. However, if the number of samples are at larger
number then the t and F or exact sampling distributions approximates to normal distribution.

Small sample tests

Statistical tests based on t- distribution

The major applications branched from this t-distribution are
 One sample t-test
 Two samples or Independent samples t-test and
 Paired samples t-test
One Sample t-test

The following is the procedure of one sample t-test

1. Let x1, x2, …, xn be the random samples drawn from a normal population with a
specified or known population mean, µ0 (say)
2. Define the null and alternative hypothesis as:
H0: there is no difference between the estimated sample mean and population mean.
H1: there is a difference between the estimated sample mean and population mean
3. The test statistic to test the above defined hypothesis is

x̅-μ
t= |s | ~ t n-1 degrees of freedom
⁄ n
√
4. Now, compare t with tn-1 degrees of freedom at α% level of significance.
5. If t ≤ tn-1 then there is no evidence to reject the H0, otherwise.
Example: The mean outer diameter of a bearing of 25 randomly selected units is observed to
10.6 cm with standard deviation of 1.15cm. Is it significantly different from mean of 11 cm.

Solution: In the given problem, information about the population mean and sample statistics
such as sample mean and sample standard deviation are given.

i.e., population mean, µ = 11 cm

sample mean, x̅ = 10.6 cm

sample standard deviation, s = 1.15 cm

80
Hypothesis: H0: x̅ = μ against H1: x̅ ≠ μ

The test statistic for testing the above defined hypothesis is

x̅-μ
t= |s | ~ t n-1 degrees of freedom
⁄ n
√

10.6-11 0.4
t= | |⇒𝑡= ⇒ 𝑡 = 1.74
1.15⁄ 0.23
√25
tn-1 = t25-1 = t24 at 0.05 level of significance is 2.06.
therefore, t < tn-1 i.e., 1.74 < 2.06, hence there is no evidence to reject the null hypothesis and
it can be concluded the sample mean is not significantly different from the given population
mean.
Example: A random sample of 10 boys had the following I.Q.’s: 70, 120, 110, 101, 88, 83,
95, 98, 107, 100. Do these data support the assumption of a population mean I.Q. of 100? Find
a reasonable range in which most of the mean I.Q. values of samples of 10 boys lie.
Solution: Given that, a random sample of 10 boys are selected and their I.Q.'s are measured.
Now the hypothesis to be tested is that the mean I.Q. values of samples support the assumption
of population mean I.Q. of 100 or not.
First define, null and alternative hypothesis as
H0: mean I.Q. values of samples do not support the assumption of population mean I.Q. of 100
H1: mean I.Q. values of samples support the assumption of population mean I.Q. of 100

Now, to claim any one of the hypothesis, we need to compute sample mean and sample standard
deviation.
Sample mean,

∑𝑛𝑖=1 𝑋𝑖
𝑋̅ = ; 𝑛 = 10 𝑎𝑛𝑑 𝑖 = 1,2, … , 10
𝑛
972
𝑋̅ = = 97.2
10
Sample standard deviation

∑𝑛𝑖=1(𝑋𝑖 − 𝑋̅)2
𝑠= √ = 14.27352
𝑛−1
Therefore, test statistic value is

81
 97.2-100 2.8
t= | |⇒𝑡= ⇒ 𝑡 = 0.6203
14.27352⁄ 4.514
√10
tn-1 = t10-1 = t9 at 0.05 level of significance is 1.8331.
Therefore, t < tn-1 i.e., 0.6203 < 1.8331, hence there is no evidence to reject the null hypothesis.
Thus, mean I.Q. values of 10 samples support the assumption of population mean I.Q. of 100.

t-test for two independent samples

This test is applicable only when there are two independent samples and the objective of
comparison between these two groups. Further, it is assumed that the samples are drawn at
random from a normally distributed populations.
 Let there be two groups, G1 and G2 with x̅1 and x̅2 , s1 and s2 as their respective sample
means and sample standard deviations
 The hypothesis, in general can be defined as follows:
H0: There is no mean difference between two groups
H1: There is a mean difference between two groups
 To test the defined hypothesis, the test statistic based on t-distribution is given by

x̅1 -x̅2
t= || || ~ t n1+n2-2 degrees of freedom
1 1
s√n + n
1 2

1 𝑛1
where, 𝑠 = √n (∑𝑖=1(𝑥1𝑖 − 𝑥̅1 )2 + ∑𝑛𝑖=1
2
(𝑥2𝑖 − 𝑥̅2 )2 ) and
1 +n2 -2

∑𝑛𝑖=1
1
𝑥1𝑖 ∑𝑛𝑖=1
2
𝑥2𝑖
𝑥̅1 = 𝑎𝑛𝑑 𝑥̅2 =
𝑛1 𝑛2
 If t ≤ t n1+n2-2 degrees of freedom at α% level of significance, then there is no evidence
to reject the null hypothesis, otherwise.
Example: Two types of teaching aids A and B, say, were introduced and as a response,
students' scores for each teaching aid is collected independently. Analysis showed the
following results n1 = 6, n2 =5, x̅1 =7.5 and x̅2 =7.2, s1 = 0.024 and s2 = 0.032. It is to be
tested that which teaching aid is better for students?

Solution: It is stated that two teaching aids A and B are adopted independently on two groups
of students and the responses are collected as scores.

82
On conducting such experiment, it is given that the sample means and sample standard
deviations obtained from two teaching aids are n1 = 6, n2 =5, x̅1 =7.5 and x̅2 =7.2, s1 = 0.024
and s2 = 0.032

Here, the hypothesis to be defined is

H0: There is no mean difference between two teaching aids

H1: There is a mean difference between two teaching aids

Then, the test statistic value can be obtained by substituting the given values in the following
formula

x̅1 -x̅2
t= || || ~ t n1+n2-2 degrees of freedom
1 1
s√n + n
1 2

7.5 − 7.2 0.3

t= || || ⇒t = | | ⇒t = 29.4025
1 1 0.0102
0.01685√6 +
5

t9 at 0.05 level of significance is 1.8331

Therefore, t > t n1+n2-2 i.e., 29.4025 > 1.8331, hence there is an evidence to reject the null
hypothesis. This means that there is difference between two teaching aids and the teaching aid
A is comparatively better than teaching aid B.

Note: If data pertaining to random samples of two populations are given, then the sample means
and sample standard deviations are to be computed and then substituted in the test statistic for
obtaining outcome.

t-test for paired samples

This test is applied to paired data of observations from one sample only when each individual
gives a pair of observations. For example, the comparison of prices of certain commodities
estimated before and after raising of shares. The point to be noted is that the number of random
samples which are drawn should be same during before and after experimentation. For this
test, the hypothesis is defined as follows:

H0: There is no true discrepancy between before and after values

H1: There exists true discrepancy between before and after values

83
The test statistic to test the defined hypothesis is given as

𝑑̅
t= |s | ~ t n-1 degrees of freedom
⁄ n
√
∑𝑛
𝑖=1 𝑑𝑖
where 𝑑̅ = ; 𝑑𝑖 = 𝑥𝑖 − 𝑦𝑖 ; here xi and yi are used to denote before and after values and
𝑛

the sample standard deviation is

∑ 𝑛
(𝑑 𝑖 − 𝑑̅ )2
𝑠= √ 𝑖=1
𝑛−1

Now, compare t with tn-1 degrees of freedom at α% level of significance. If t ≤ tn-1 then there
is no evidence to reject the H0, otherwise.

Example: The marks or opinions in form of scores are collected on 12 trainees before
commencement of a training program and at the end of the program. Is there any change in
the scores indicating the effect of the training program?

Trainee No. 1 2 3 4 5 6 7 8 9 10 11 12
Scores before training 6 8 8 6 5 9 6 7 6 6 4 8
Scores after training 8 8 10 7 6 10 9 8 5 7 4 6

Solution: It is stated that the effectiveness of the training program is observed by collecting
data in terms of scores before and after the program from 12 trainees. Now, basing on the
given information, the hypothesis can be defined as:

H0: There is no effectiveness of the training program

H1: Training program is effective

To test the defined hypothesis the test statistic is given by

𝑑̅
t= |s | ~ t n-1 degrees of freedom
⁄ n
√
∑𝑛
𝑖=1 𝑑𝑖
where 𝑑̅ = ; 𝑑𝑖 = 𝑥𝑖 − 𝑦𝑖 ; here xi and yi are used to denote before and after values and
𝑛

the sample standard deviation is

84
 2 𝑛
√∑𝑛𝑖=1(𝑑𝑖 − 𝑑̅ ) (∑𝑛𝑖=1 𝑑𝑖 )2
𝑠= = ∑ 𝑑𝑖2 −
𝑛−1 𝑛
𝑖=1

∑𝑛
𝑖=1 𝑑𝑖 9
Here, 𝑑̅ = = 12 = 0.75
𝑛

Scores before Scores after

Trainee No. di di2
training training
1 6 8 -2 4
2 8 8 0 0
3 8 10 -2 4
4 6 7 -1 1
5 5 6 -1 1
6 9 10 -1 1
7 6 9 -3 9
8 7 8 -1 1
9 6 5 1 1
10 6 7 -1 1
11 4 4 0 0
12 8 6 2 4
27

2
√ ∑𝑛𝑖=1(𝑑𝑖 − 𝑑̅ ) 20.25
⇒𝑠= = = 1.84
𝑛−1 11

𝑑̅ 0.75
t= |s | ⇒ t= | | ⇒ 𝑡 = 1.92
⁄ n 1.84⁄
√ √12

tn-1 = t12-1 = t11 at 0.05 level of significance is 2.20.

Therefore, t < tn-1 i.e., 1.92 < 2.20, hence there is no evidence to reject the null hypothesis.
This helps to claim that there is no effectiveness of training program.

F - test for equality of two sample variances

 Let us suppose two random samples are drawn from two independent normal
populations, xi and yj, say
 Here the main objective is to test whether the sample variances obtained from these
two independent samples are equal or not.

85
  To meet this objective, the null and alternative hypothesis can be defined as
H0: The population variances are equal, i.e., 𝜎𝑥2 = 𝜎𝑦2
H1: The population variances are not equal, 𝜎𝑥2 ≠ 𝜎𝑦2
 The test statistic which is used to test the hypothesis is as follows
𝑠𝑥2
𝐹= 2 ~𝐹(𝑛1 −1,𝑛2 −1) degrees of freedom
𝑠𝑦

1 1 2
where 𝑠𝑥2 = ∑𝑛𝑖=1
1
(𝑥𝑖 − 𝑥̅ )2 ; 𝑠𝑦2 = ∑𝑛𝑗=1
2
(𝑦𝑗 − 𝑦̅) and
𝑛1 −1 𝑛2 −1

∑𝑛𝑖=1
1
𝑥𝑖 ∑𝑛𝑗=1
2
𝑦𝑗
𝑥̅ = ; 𝑦̅ =
𝑛1 𝑛2
Here, 𝑠𝑦2 𝑎𝑛𝑑 𝑠𝑦2 are the sample variances and unbiased estimates of 𝜎𝑥2 𝑎𝑛𝑑 𝜎𝑦2 respectively
 If Fcal ≤ 𝐹(𝑛1 −1,𝑛2 −1)degrees of freedom at α% level of significance, then there is no
evidence to reject the null hypothesis, otherwise.
Example: The sample variances in 25 males and 25 females are 5.0 and 9.0. can we conclude
that the variance in prices is same in both males and females?
Solution: Given n1 = 25, n2 = 25, 𝑠12 = 5.0 and 𝑠22 = 9.0
𝑠𝑥2 9.0
Therefore, 𝐹 = 2 = = 1.8
𝑠𝑦 5.0

F(24, 24) at 5% level of significance is 1.98

Here Fcal < F(24, 24), 1.8 < 1.98, hence, we accept and conclude that there is no variability in both
the groups with respect to prices.
Example: The College Board provided comparisons of Scholastic Aptitude Test (SAT) scores
based on the highest level of education attained by the test taker’s parents. A research
hypothesis was that students whose parents had attained a higher level of education would on
average score higher on the SAT. During 2003, the overall mean SAT verbal score was 507
(The World Almanac, 2004 ). SAT verbal scores for independent samples of students follow.
The first sample shows the SAT verbal test scores for students whose parents are college
graduates with a bachelor’s degree. The second sample shows the SAT verbal test scores for
students whose parents are high school graduates but do not have a college degree.
College
485 534 650 554 550 572 497 592 487 533 526 410 515 578 448 469
Grads
High
School 442 580 479 486 528 524 492 478 425 485 390 535
Grads

86
Formulate the hypotheses that can be used to determine whether the sample data support the
hypothesis that students show a higher variability in verbal score on the SAT if their parents
have not attained a higher level of education. At α = 0.05, what is your conclusion?
Solution:
Basing on the given information, we need to test whether the sample data support the
hypothesis that students show a higher variability in verbal score on the SAT if their parents
have not attained a higher level of education
Let us denote the SAT verbal test scores for students whose parents are college graduates with
a bachelor’s degree as Group A and the other as group B
The null and alternative hypothesis for this problem is
H0: The sample variability between groups A and B are equal
H1: The sample variability between groups A and B are not equal
Sample College Grads High School Grads
number 𝑛1 𝑛2
2
xi ∑(𝑥𝑖 − 𝑥̅ )2 yj ∑(𝑦𝑗 − 𝑦̅)
𝑖=1 𝑗=1
1 485 1600 442 2025
2 534 81 580 8649
3 650 15625 479 64
4 554 841 486 1
5 550 625 528 1681
6 572 2209 524 1369
7 497 784 492 25
8 592 4489 478 81
9 487 1444 425 3844
10 533 64 485 4
11 526 1 390 9409
12 410 13225 535 2304
13 515 100
14 578 2809
15 448 5929
16 469 3136
Totals 525 3310.125 487 2454.667

𝑛1
1 3310.125
𝑠𝐴2 = ∑(𝑥𝑖 − 𝑥̅ )2 = = 220.675
𝑛1 − 1 15
𝑖=1
𝑛2
1 2 2454.667
𝑠𝐵2 = ∑(𝑦𝑗 − 𝑦̅) = = 223.1515
𝑛2 − 1 11
𝑖=1

87
 2
𝑠𝐴 220.675
Therefore, 𝐹 = 2 = = 0.9889
𝑠𝐵 223.1515

F(15,11) at 0.05 level of significance is 2.72. hence, from the computed results, it is evident that
there is no variability between the groups A and B. In other words, the variability in the SAT
verbal test scores for students whose parents are college graduates with a bachelor’s degree
and with SAT verbal test scores for students whose parents attained higher degree is equal.
EXERCISES

Problems on tests for single Proportion

1. A builder claims that heat pumps are installed in 70% of all homes being constructed
today in the city of Richmond, Virginia. Would you agree with this claim if a random
survey of new homes in this city shows that 8 out of 15 had heat pumps installed? Use
a 0.10 level of significance.
2. A commonly prescribed drug for relieving nervous tension is believed to be only 60%
effective. Experimental results with a new drug administered to a random sample of
100 adults who were suffering from nervous tension show that 70 received relief. Is this
sufficient evidence to conclude that the new drug is superior to the one commonly
prescribed? Use a 0.05 level of significance.
3. Suppose that, in the past, 40% of all adults favored capital punishment. Do we have
reason to believe that the proportion of adults favoring capital punishment today has
increased if, in a random sample of 15 adults, 8 favor capital punishment? Use a 0.05
level of significance.
4. It is believed that at least 60% of the residents in a certain area favor an annexation suit
by a neighboring city. What conclusion would you draw if only 110 in a sample of 200
voters favor the suit? Use a 0.05 level of significance.
5. A new radar device is being considered for a certain defense missile system. The system
is checked by experimenting with actual aircraft in which a kill or a no kill is simulated.
If in 300 trials, 250 kills occur, accept or reject, at the 0.04 level of significance, the
claim that the probability of a kill with the new system does not exceed the 0.8
probability of the existing device.
6. In a controlled laboratory experiment, scientists at the University of Minnesota
discovered that 25% of a certain strain of rats subjected to a 20% coffee bean diet and
then force-fed powerful cancer-causing chemical later developed cancerous tumors.
Would we have reason to believe that the proportion of rats developing tumors when

88
 subjected to this diet has increased if the experiment were repeated and 16 of 48 rats
developed tumors? Use a 0.05 level of significance.
Problems on tests for Proportions between two populations

7. Consider the following results for independent samples taken from two populations.
Sample 1 Sample 2
𝑛1 = 400 𝑛2 = 300
𝑝1 = 0.48 𝑝2 = 0.36
a. What is the point estimate of the difference between the two population
proportions?
b. Develop a 90% confidence interval for the difference between the two population
proportions.
c. Develop a 95% confidence interval for the difference between the two population
proportions.
8. Consider the following hypothesis test.
𝐻0 : 𝑝1 − 𝑝2 ≤ 0
𝐻0 : 𝑝1 − 𝑝2 > 0
The following results are for independent samples taken from the two populations
Sample 1 Sample 2
𝑛1 = 200 𝑛2 = 300
𝑝1 = 0.22 𝑝2 = 0.16
With α = .05, what is your hypothesis testing conclusion?
9. The Professional Golf Association (PGA) measured the putting accuracy of
professional golfers playing on the PGA Tour and the best amateur golfers playing in
the World Amateur Championship (Golf Magazine, January 2007). A sample of 1075
6-foot putts by professional golfers found 688 made puts. A sample of 1200 6-foot putts
by amateur golfers found 696 made putts.
1. Estimate the proportion of made 6-foot putts by professional golfers. Estimate the
proportion of made 6-foot putts by amateur golfers. Which group had a better
putting accuracy?
2. What is the point estimate of the difference between the proportions of the two
populations? What does this estimate tell you about the percentage of putts made
by the two groups of golfers?
3. What is the 95% confidence interval for the difference between the two population
proportions? Interpret his confidence interval in terms of the percentage of putts
made by the two groups of golfers.

89
10. Chicago O’Hare and Atlanta Hartsfield-Jackson are the two busiest airports in the
United States. The congestion often leads to delayed flight arrivals as well as delayed
flight departures. The Bureau of Transportation tracks the on-time and delayed
performance at major airports (Travel & Leisure, November 2006). A flight is
considered delayed if it is more than 15 minutes behind schedule. The following sample
data show the delayed departures at Chicago O’Hare and Atlanta Hartsfield-Jackson
airports.
Chicago O’Hare Atlanta Hartsfield-Jackson
Flights 900 1200
Delayed Flights 252 312
a. State the hypotheses that can be used to determine whether the population
proportions of delayed departures differ at these two airports.
b. What is the point estimate of the proportion of flights that have delayed departures
at Chicago O’Hare?
c. What is the point estimate of the proportion of flights that have delayed departures
at Atlanta Hartsfield-Jackson?
d. What is your conclusion?
11. In a test of the quality of two television commercials, each commercial was shown in a
separate test area six times over a one-week period. The following week a telephone
survey was conducted to identify individuals who had seen the commercials. Those
individuals were asked to state the primary message in the commercials. The following
results were recorded.
Commercial A Commercial B
Number Who saw Commerical 150 200
Number Who Recalled Message 63 60
a. Use α = .05 and test the hypothesis that there is no difference in the recall
proportions for the two commercials.
b. Compute a 95% confidence interval for the difference between the recall
proportions for the two populations.
12. During the 2003 Super Bowl, Miller Lite Beer’s commercial referred to as “The Miller
Lite Girls” ranked among the top three most effective advertisements aired during the
Super Bowl (USA Today,December 29, 2003). The survey of advertising effectiveness,
conducted by USA Today’s Ad Track poll, reported separate samples by respondent
age group to learn about how the Super Bowl advertisement appealed to different age
groups. The following sample data apply to the “The Miller Lite Girls” commercial.

90
 Age Group Sample size Liked the Ad a lot
Under 30 100 49
30 to 49 150 54
a. Formulate a hypothesis test that can be used to determine whether the population
proportions for the two age groups differ.
b. What is the point estimate of the difference between the two population
proportions?
c. Conduct the hypothesis test. At α = .05, what is your conclusion?
d. Discuss the appeal of the advertisements to the younger and the older age groups.
Would the Miller Lite organization find the results of the USA TodayAd Track poll
encouraging? Explain.
13. A 2003 New York Times/CBSNews poll sampled 523 adults who were planning a
vacation during the next six months and found that 141 were expecting to travel by
airplane (New York Times News Service, March 2, 2003). A similar survey question
in a May 1993 New York Times/CBSNews poll found that of 477 adults who were
planning a vacation in the next six months, 81 were expecting to travel by airplane.
a. State the hypotheses that can be used to determine whether a significant change
occurred in the population proportion planning to travel by airplane over the 10-
year period.
b. What is the sample proportion expecting to travel by airplane in 2003? In 1993?
c. Use α = .01 and test for a significant difference. What is your conclusion?
d. Discuss reasons that might provide an explanation for this conclusion.
Problems on tests for means when variances are known

14. The following results come from two independent random samples taken of two
populations. If 𝑛1 = 50, 𝑥̅1 = 13.6, 𝜎1 = 2.2 and 𝑛2 = 35, 𝑥̅2 = 11.6, 𝜎2 = 3.0.
a. What is the point estimate of the difference between the two population means?
b. Test whether the means are equal or not with α = 0.05.
a. When conducting a hypothesis test with the values given for the standard deviation,
sample size, and α, how large must the increase from 2007 to 2008 be for it to be
statistically significant?
b. Use the result of part (d) to state whether the increase for J.C. Penney from 2007 to
2008 is statistically significant.
15. A manufacturer claims that the average tensile strength of thread A exceeds the average
tensile strength of thread B by at least 12 kilograms. To test his claim, 50 pieces of each

91
 type of thread are tested under similar conditions. Type A thread had an average tensile
strength of 86.7 kilograms with known standard deviation of 𝜎𝐴 = 6.28 kilograms,
while type B thread had an average tensile strength of 77.8 kilograms with known
standard deviation of 𝜎𝐵 = 5.61 kilograms. Test the manufacturer’s claim at α = 0.05.
Problems on tests for means when variances are unknown

16. The following results are for independent random samples taken from two populations.
If 𝑛1 = 20, 𝑥̅1 = 22.5, 𝑠1 = 2.5 and 𝑛2 = 30, 𝑥̅2 = 20.1, 𝑠2 = 4.8.
a. What is the point estimate of the difference between the two population means?
b. What is the degrees of freedom for the t distribution?
c. At 95% confidence, what is the margin of error?
d. What is the 95% confidence interval for the difference between the two population
means?
17. Consider the following hypothesis test.
𝐻0 : 𝜇1 − 𝜇2 = 0
𝐻0 : 𝜇1 − 𝜇2 ≠ 0
The following results are from independent samples taken from two populations.
𝑛1 = 35, 𝑥̅1 = 13.6, 𝑠1 = 5.2 and 𝑛2 = 40, 𝑥̅2 = 10.1, 𝑠2 = 8.5.
a. What is the value of test statistic?
b. With α = 0.05, what is your hypothesis testing conclusion?
18. Consider the following data for two independent random samples taken from two
normal populations.
Sample 1 10 7 13 7 9 8
Sample 2 8 7 8 4 6 9
a. Compute the two sample means.
b. Compute the two sample standard deviations.
c. What is the point estimate of the difference between the two population means?
d. What is the 90% confidence interval estimate of the difference between the two
population means?
19. FedEx and United Parcel Service (UPS) are the world’s two leading cargo carriers by
volume and revenue (The Wall Street Journal, January 27, 2004). According to the
Airports Council International, the Memphis International Airport (FedEx) and the
Louisville International Airport (UPS) are 2 of the 10 largest cargo airports in the world.
The following random samples show the tons of cargo per day handled by these
airports. Data are in thousands of tons.

92
 Memphis
9.1 15.1 8.8 10 7.5 10.5
8.3 9.1 6 5.8 12.1 9.3
Louisville
4.7 5 4.2 3.3 5.5
2.2 4.1 2.6 3.4 7
a. Compute the sample mean and sample standard deviation for each airport.
b. What is the point estimate of the difference between the two population means?
Interpret this value in terms of the higher-volume airport and a comparison of the
volume difference between the two airports.
c. Develop a 95% confidence interval of the difference between the daily population
means for the two airports.
20. Are nursing salaries in Tampa, Florida, lower than those in Dallas, Texas? Salary data
show staff nurses in Tampa earn less than staff nurses in Dallas (The Tampa Tribune,
January 15, 2007). Suppose that in a follow-up study of 40 staff nurses in Tampa and
50 staff nurses in Dallas you obtain the following results.
Tampa Dallas
𝑛1 = 40 𝑛2 = 50
𝑥̅1 = $56,100 𝑥̅2 = $59,400
𝑠1 = $6000 𝑠2 = $7000
a. Formulate hypothesis so that, if the null hypothesis is rejected, we can conclude that
salaries for staff nurses in Tampa are significantly lower than for those in Dallas.
Use α = 0.05.
b. What is the value of the test statistic?
c. What is the p-value?
d. What is your conclusion?
21. Injuries to Major League Baseball players have been increasing in recent years. For the
period 1992 to 2001, league expansion caused Major League Baseball rosters to
increase 15%. However, the number of players being put on the disabled list due to
injury increased 32% over the same period (USA Today, July 8, 2002). A research
question addressed whether Major League Baseball players being put on the disabled
list are on the list longer in 2001 than players put on the disabled list a decade earlier.
a. Using the population mean number of days a player is on the disabled list, formulate
null and alternative hypotheses that can be used to test the research question.

93
 b. Assume that the following data apply:
2001 Season 1992 Season
𝑛1 = 45 𝑛2 = 38
𝑥̅1 = 60 𝑑𝑎𝑦𝑠 𝑥̅2 = 51 𝑑𝑎𝑦𝑠
𝑠1 = 18 𝑑𝑎𝑦𝑠 𝑠2 = 15 𝑑𝑎𝑦𝑠
What is the point estimate of the difference between population mean number of days
on the disabled list for 2001 compared to 1992? What is the percentage increase in the
number of days on the disabled list?
c. Use α = 0.01. What is your conclusion about the number of days on the disabled
list? What is the p-value?
d. Do these data suggest that Major League Baseball should be concerned about the
situation?
22. Periodically, Merrill Lynch customers are asked to evaluate Merrill Lynch financial
consultants and services. Higher ratings on the client satisfaction survey indicate better
service, with 7 the maximum service rating. Independent samples of service ratings for
two financial consultants are summarized here. Consultant A has 10 years of
experience, whereas consultant B has 1 year of experience. Use α = 0.05 and test to see
whether the consultant with more experience has the higher population mean service
rating.
Consultant A Consultant B
𝑛1 = 16 𝑛2 = 10
𝑥̅1 = 6.82 𝑥̅2 = 6.25
𝑠1 = 0.64 𝑠2 = 0.75
a. State the null and alternative hypotheses.
b. Compute the value of the test statistic.
c. What is your conclusion?
Problems on F-test for variances
23. A sample of 16 items from population 1 has a sample variance 𝑠12 = 5.8 and a sample
of 21 items from population 2 has a sample variance 𝑠22 = 2.4 Test the following
hypotheses at the .05 level of significance.
𝐻0 : 𝜎12 = 𝜎22
𝐻𝑎 : 𝜎12 ≠ 𝜎22
24. Consider the following hypothesis test.
𝐻0 : 𝜎12 = 𝜎22
𝐻𝑎 : 𝜎12 ≠ 𝜎22
What is your conclusion if 𝑛1 = 21, 𝑠12 = 8.2, 𝑛1 = 26 and 𝑠22 = 4.0? Use α = 0.05

94
25. The variance in a production process is an important measure of the quality of the
process. A large variance often signals an opportunity for improvement in the process
by finding ways to reduce the process variance. Conduct a statistical test to determine
whether there is a significant difference between the variances in the bag weights for
two machines. Use a .05 level of significance. What is your conclusion? Which
machine, if either, provides the greater opportunity for quality improvements?
Machine 1 2.95 3.45 3.5 3.75 3.48 3.26 3.33 3.2
3.16 3.2 3.22 3.38 3.9 3.36 3.25 3.28
3.2 3.22 2.98 3.45 3.7 3.34 3.18 3.35
3.12
Machine 2 3.22 3.3 3.34 3.28 3.29 3.25 3.3 3.27
3.38 3.34 3.35 3.19 3.35 3.05 3.36 3.28
3.3 3.28 3.3 3.2 3.16 3.33
26. On the basis of data provided by a Romac salary survey, the variance in annual salaries
for seniors in public accounting firms is approximately 2.1 and the variance in annual
salaries for managers in public accounting firms is approximately 11.1. The salary data
were provided in thousands of dollars. Assuming that the salary data were based on
samples of 25 seniors and 26 managers, test the hypothesis that the population variances
in the salaries are equal. At a .05 level of significance, what is your conclusion?
Problems on t-test for single mean
27. A machinist is making engine parts with axle diameters of 0.700 inch. A random sample
of 10 parts shows a mean diameter of 0.742 inch with a standard deviation of 0.040
inch. Compute the statistic you would use to test whether the work is meeting the
specifications. Also state how you would proceed further.
28. The mean weekly sales of soap bars in departmental stores were 146.3 bars per store.
After an advertising campaign the mean weekly sales in 22 stores for a typical week
increased to 153.7 and showed a standard deviation of 17.2. Was the advertising
campaign successful?
29. The heights of 10 males of a given locality are found to be 70, 67, 62, 68, 61, 68, 70,
64, 64, 66 inches. Is it reasonable to believe that the average height is greater than 64
inches? Test at 5% level of significance assuming that for 9 degrees of freedom P (t >
1.83 ) = 0.05.

95
30. A random sample of 16 values from a normal population showed a mean of 41.5 inches
and the sum of squares of deviations from this mean equal to 135 square inches. Show
that the assumption of a mean of 43.5 inches for the population is not reasonable.
31. A manufacturer of gun powder has developed a new powder which is designed to
produce a muzzle velocity equal to 3000 ft/sec. Seven shells are loaded with the charge
and the muzzle velocities measured. The resulting velocities are as follows: 3005; 2935;
2965; 2995; 3905; 2935 and 2905. Do these data present sufficient evidence to indicate
that the average velocity differs from 3000 ft/sec.
32. The average length of time for students to register for summer classes at a certain
college has been 50 minutes with a standard deviation of 10 minutes. A new registration
procedure using modern computing machines is being tried. If a random sample of 12
students had an average registration time of 42 minutes with s.d. of 11.9 minutes under
the new system, test the hypothesis that the population mean has not changed, using
0.05 level of significance.
33. The average breaking strength of steel rods is specified to be 18.5 thousand pounds. To
test this a sample of 14 rods was tested. The mean and standard deviations obtained
were 17.85 and 1.955 thousand pounds respectively. is the result of the experiment
significant?

96
 UNIT – V

NON PARAMETRIC TESTS

CONTENTS:

5.1 Introduction to Non Parametric Tests

5.2 Sign Test

5.3 Wilcoxon Signed Rank test

5.4 Kruskal-Wallis H test

5.5 Wald-Wolfowitz Run test

5.6 Chi-Square test

5.1 Introduction to Non Parametric Tests

In statistics, most of the methods require an assumption that the data follows a normal
distribution. Sometimes, this may not be applicable when data does not follow normal. Further,
the normality assumption can be obtained by adapting a suitable transformation method. In
some situations, the researchers will be in need of those methods which are free from
distributional assumptions that the data follows a particular distribution. Those methods are
usually named as Non Parametric methods. Another way of usage of these methods arise when
your data is of nominal or ordinal type. So, in general, for a statistical method to be classified
as non-parametric, it must satisfy at least one of the following conditions: (Conover, 1998)

1. The method can be used with nominal/ ordinal data.

2. The method can be used with interval/ ratio data when no assumption can be made
about population probability distribution.
Here, a point to be noted is that if the data with level of measurement interval/ ratio has the
necessary probability distribution then parametric methods result in providing more powerful
statistical procedures. It is also to be noted that in many cases, both non-parametric and
parametric methods can be applied, in such cases non-parametric method is as good as (or)
almost as powerful as parametric method. Finally, it can be understood that whenever there are
less restrictions on data measurement and fewer assumptions about population distributions,
non-parametric methods are regarded as more generally applicable than parametric methods.

97
In this chapter, a detailed discussion is made on each of the non-parametric method like Sign,
Wilcoxon Signed Rank test, Wald-Wolfowitz runs test, Kruskal-Wallis test and Chi-Square
tests with numerical illustrations.

5.2 Sign Test

Sign test is a non-parametric analogue to one sample Z test under small and large sample sizes.
The applicability of Sign test will come into role when we need to test the data values against
a perceived hypothesis statement with the following inclusions:

 When n ≤ 20, the small sample case of Sign test is to be used by replacing one sample
t-test with underlying distribution as Binomial.
 When n > 230, the large sample case of Sign test is to be used by replacing one sample
Z test with underlying approximated distribution as normal.
Now, let us see and understand how the small and large sample versions of the Sign test can be
used to test the hypothesis and helps in drawing proper conclusions about the defined
hypothesis.

Assumptions

1. The samples are drawn at random

2. Continuous distribution and
3. Measurements can be expressed using plus or minus signs
The sign test purely depends on the usage of the measure, median to note the number of counts
that fall above and below the median. Generally, for those counts which lie above the median
a plus sign is given and a negative sign is assigned for the values which are below the median.
Let ‘p’ be the proportion or probability that 50% of all values lie above and 50% of all values
lie below the median value. It is expected that the number of counts of distribution is
approximated symmetrically around median. Here, the data so collected is represented in terms
of plus or minus sign; hence this non-parametric test is called the SIGN TEST. The number of
plus signs is the test statistic.

In small sample case under the assumption that null hypotheses is true, the underlying sampling
distribution is Binomial distribution with p = 0.5. The hypothesis under one tail and two tail
test can be defined as : (usually α = 0.05 or 0.01)

98
 Table 5.1

One tail test

Two tail test
Upper one tail Lower one tail
H0 : Sample median ≤ H0 : Sample median ≥ H0 : Sample median =
Population median Population median Population median
H1 : Sample median > H1 : Sample median < H1 : Sample median ≠
Population median Population median Population median

Procedure: Consider an experiment where the pairs of observations on two things are being
compared and different pairs are observed under different conditions. These number of paired
observations (n) are the number of trials with probability of success equal to 0.5. In this case,
we can handle the situation using a Binomial distribution.

X ~ Bin(n, p) with probability mean function

P(X =r) = nCrprqn-r ;

Mean µ = np and variance σ2 = npq.

In general, there will be five steps of conducting any test of hypothesis in both parametric/ non-
parametric analysis.

1. Define null hypothesis

2. Define alternative hypothesis
3. Set the level of significance
4. Test statistic and
5. Inference
Let us consider an example to illustrate the procedural format of conducting sign test.

Example: A company has conducted a study to know the preferences of individuals in terms of
taste for two brands of a product. The following table contains the list of preferences indicated
by 10 individuals

Table 5.2

Individuals 1 2 3 4 5 6 7 8 9 10
Brand A vs Brand B + + + - + + - + - +

99
At α = 0.05, test for a significant difference in the preferences for the two brands. Here, +
indicates preference for Brand A over Brand B.

From the data listed in Table 5.2, seven plus signs are observed indicating that the individuals’
preference is towards brand A rather than brand B.

Now, on using the binomial probabilities, it is easy to determine the p-value for the test. The
binomial probabilities can be obtained using the table ____ given in appendix with p = 0.5.

For the mentioned problem, hypothesis can be stated as

H0 : Preference for Brand A = Preference for Brand B

H1 : Preference for Brand A ≠ Preference for Brand B

Upon using the binomial probabilities, with n = 10; p = 0.5 and number of successes as 7, p-
value is observed to be 0.000, indicating that the chance of preferring brand A of a product is
very high over brand B.

So, with respect to taste of a product, it is evident to claim that individual’s preference differs
significantly for the two brands of a product.

In large sample case, under the null hypothesis H0 : p =0.5 and a sample size of n > 20, the
sampling distribution for the number of plus signs can be approximated by a normal
distribution.

The mean and standard deviation under normal approximation are µ = 0.5n; 𝜎 = √0.25𝑛.

To have a better understanding about the practical applicability of sign test under large sample
case, let us have an example of a national survey which is conducted to see whether the annual
income will fulfill the dreams to come true or not. For which, the data was collected from the
adults of median annual income. A sample of 225 individuals in a country are randomly
selected, of which 122 individuals reported that the amount of income is needed to make their
dreams come true is less than $ 152,000 and the rest 103 individuals reported that the amount
needed is more than $ 152,000.

Now, the problem here is to test whether the median amount of annual income needed to make
dreams come true in that country is $ 152,000 or not.

100
On using the sign test, we have that n = 225. Now, we compute the mean and standard deviation
for the given problem.

μ = 0.5𝑛 = 0.5(225) = 112.5

𝜎 = √0.25𝑛 = √0.25(225) = √56.25 = 7.5

With 0.05 level of significance and based on the response from the individuals, (i.e., x= 122),
we can calculate the Z value as

𝑥 − 𝜇 122 − 112.5 9.5

𝑍= = = = 1.2667
𝜎 7.5 7.5

The standard normal probability table shows that the area in the tail towards left of Z = 1.2667
is 0.8962. Under two tail test, p-value = 2(0.8962) = 1.7924. This depicts that at α = 0.05, we
accept H0 indicating that individual’s perception about median amount of annual income
$152,000/- to make dreams come true is not true.

Let us consider another example with a different set of information about a study. Taken a
situation of competition in the personal computer market intensity, a sample of 500 purchases
showed 202 Brand A computers, 158 Brand B computers and 140 other computers. It is
necessary to test the hypothesis that brand A and brand B have the same share of the personal
computer market. (α = 0.05)

Using sign test, we have n = 500-140 = 360 individuals were able to purchase Brand A and
Brand B personal computers. From this, we can compute mean and standard deviation as

μ = 0.5𝑛 = 0.5(360) = 180

𝜎 = √0.25𝑛 = √0.25(360) = 9.486

Here n = 360, it can be assumed that the sampling distribution is normally distributed
(approximated).

Using 0.05% level of significance, the test statistic can be computed as

𝑥−𝜇
𝑍=
𝜎

101
Here x = 202; µ = 180; σ = 9.486

202 − 180 22
⇒𝑍= = = 2.32
9.486 9.486

From the standard normal probability table, the area in the tail to the left of Z = 2.32 is 0.9898.
under two tail test, the p-value is 2(0.9898) = 1.96

With the p-value, we do not reject the hypothesis

i.e., H0 : Personal computer market for Brand A = Personal computer market for Brand B

H1 : Personal computer market for Brand A ≠ Personal computer market for Brand B

Therefore, it can be concluded that both Brand A and Brand B have the same share of the
personal computer market.

5.3 Wilcoxon Signed Rank test

When an experiment is conducted to generate a matched pairs of observations with a normality

assumption, then, in such situations, the standard test which is derived from exact sampling
distribution, namely the t-test for paired sample is generally used. However, if there is a non-
normality and particularly for samples the t-test may not be valid. Hence, a non-parametric
analogue to paired samples t-test is the Wilcoxon- Signed rank sum test, which is used to test
the null hypothesis that the median of a distribution is equal to some value.

The method considers ‘n’ matched pairs as one sample. The following are the assumptions of
the test:

 Each paired sample is randomly distributed.

 The differences obtained through matched pairs should be symmetrically distributed.
In the case of large samples of paired data (n ≥ 10), it is suggested to have a normal
approximation. If n ≤ 10, we shall test the hypothesis using the critical values of the statistic.

A test was conducted of two overnight delivery services. Two samples of identical deliveries
were setup so that both delivery services were notified of the need for a delivery at the same
time. The hours required to make each delivery is reported in table 5.3. The problem which is
to be addressed is that is there any difference in the delivery times for the two services?

102
 Table 5.3
Delivery 1 2 3 4 5 6 7 8 9 10 11
Service 1 24.5 26 28 21 18 36 25 21 24 26 31
Service 2 28 25.5 32 20 19.5 28 29 22 23.5 29.5 30

In order to verify the given statement of the problem that is there any similarity in the delivery
times of two services, first we need to define the hypothesis in the following way.

H0 : The delivery times of both services are identical

against H1 : The delivery times of both services are not identical

If the null hypothesis cannot be rejected, indicates that the delivery time for the two services
does not differ. However, if H0 is rejected, we will have an evidence to claim that the delivery
time for the two services differ in time.

The following way outlines the procedures of conducting the test.

1. Compute the difference for each matched sample.

2. Consider the absolute difference of the differences between the matched pairs.
3. Rank these absolute differences from lowest to highest. Suppose, if there is any zero
difference is present, then ranking is done by discarding that zero difference pair.
4. Once, the ranking is done for the absolute differences, assigning of signs to those ranks
on par with the original difference in the data.
5. Calculate the sum of the signed ranks, denoted as T.
6. The sampling distribution of T for identical populations can be approximated to normal
distribution when n ≥ 10.
i.e., Mean = µT = 0
𝑛(𝑛+1)(2𝑛+1)
Standard Deviation = 𝜎𝑇 = √
6

7. The test statistic, to test the defined hypothesis is

𝑇 − 𝜇𝑇
𝑍=
𝜎𝑇
8. The Z value can be used to compute the p-value. This can be done using the standard
normal probability table.
9. If p-value ≤ 0.05 (= α), then it is evident to reject the null hypothesis, otherwise.

103
Here, for the considered problem, the following table 5.4, gives us the absolute values, ranks
and signed ranks.

Table: 5.4

Deliver Service Service Differenc Absolute Ran Signed

y 1 2 e Difference k Rank
1 24.5 28.0 -3.5 3.5 7.5 -7.5
2 26.0 25.5 0.5 0.5 1.5 +1.5
3 28.0 32.0 -4.0 4.0 9.5 -9.5
4 21.0 20.0 1.0 1.0 4.0 +4.0
5 18.0 19.5 -1.5 1.5 6.0 -6.0
6 36.0 28.0 8.0 8.0 11.0 +11.0
7 25.0 29.0 -4.0 4.0 9.5 -9.5
8 21.0 22.0 -1.0 1.0 4.0 -4.0
9 24.0 23.5 0.5 0.5 1.5 +1.5
10 26.0 29.5 -3.5 3.5 7.5 -7.5
11 31.0 30.0 1.0 1.0 4.0 +4.0
T = -22

The tied observations are assigned the average ranks. The last column of the above table 5.4
takes the sign as that of the original difference. It is noted that the sum of signed ranks is -22,
which is significantly from zero (since the sum of signed rank values would be approximately
zero, if the populations are identical). Here, the numbers of matched pairs is 11, hence the
sampling distribution of T can be approximated by a normal distribution. The mean and
standard deviation for the given problem are as follows:

µT = 0 (assuming identical nature of two services)

𝑛(𝑛 + 1)(2𝑛 + 1) 11(12)(23) 3036

𝜎𝑇 = √ =√ =√ = √506 = 22.494
6 6 6

Suppose, if there are any matched pairs with zero difference, then 𝜎𝑇 should be calculated by
discarding such observations.

Now, the test statistic value is computed as

𝑇 − 𝜇 𝑇 −22 − 0
𝑍= = ⟹ 𝑍 = −0.9780
𝜎𝑇 22.494

104
The two tailed p-value can be obtained using the standard normal probability table, i.e.,
2(0.1660) = 0.332 > 0.05. This indicates that there is no evidence to reject the H0, and it can be
concluded that the delivery times of services one and two are identical.

Let us consider another example for a better understanding of the concept.

In a study of bilingual coding, 6 bilingual (French and English) college students an article
written in French, and each answers a series of 25 multiple-choice questions covering the
content of the article. The questions are written in French in one examination and the other
examination is in English. The score (total correct) is:

Table 5.5
Examination in French 11 12 16 22 25 25
Examination in English 10 13 17 19 21 24
Is this evidence at the 5 percent significance level that there is difficulty in transferring
information from one language to another?
H0 : There is no difficulty in transferring information from one language to another.

H1 : There is a difficulty in transferring information from one language to another.

In order to test the given hypothesis, the test statistic is to be computed. This requires
computation of T, µT and σT. The table 5.6 consists of absolute values, ranks and signed ranks

Table: 5.6
Examinatio Examinatio
Studen Differenc Absolute Ran Signed
n n
t e Difference k Rank
in French in English
1 11 10 1 1 2.5 +2.5
2 12 13 -1 1 2.5 -2.5
3 16 17 -1 1 2.5 -2.5
4 22 19 3 3 5 +5
5 25 21 4 4 6 +6
6 25 24 1 1 2.5 +2.5
T = 11

The tied observations are assigned the average ranks. Absolute difference 1 is repeated four
times thus the ranks 1,2,3 and 4 that are to be assigned to these numbers are replaced by the
average of the four ranks i.e., 2.4. Here n = 6, the hypothesis is tested using the critical values

105
of the statistic. The overall rank sum is T = 11. The mean and standard deviation for the given
problem are as follows:

µT = 0 (assuming identical nature of two languages)

𝑛(𝑛 + 1)(2𝑛 + 1) 6(7)(13) 546

𝜎𝑇 = √ =√ =√ = √91 = 9.5394
6 6 6

Now, the test statistic value is computed as

𝑇 − 𝜇 𝑇 11 − 0
𝑍= = ⟹ 𝑍 = 1.1531
𝜎𝑇 9.5394

The Wilcoxon table value at n=6 and α = 0.05 is 0 which is less than the Z value, i.e., 1.1531
> 0. This indicates that there is evidence to reject the H0, and it can be concluded that the there
is a difficulty in transforming information from one language to another.

5.4 Kruskal-Wallis H test

In usual hypothesis testing problems, we come across the situations where there will be a need
to compare two groups. But, if there are more than two groups which are to be compared then
the conventional independent samples t-test (when data is assumed to be normal and
interval/ratio scale) or the non-parametric approach of Mann-Whitney U test fails to provide
such comparisons. Hence, Kruskal and Wallis framed a statistical methodology that
accommodates the problem of ‘k’ populations at a time. So, the test was named after them as
KRUSKAL-WALLIS test. Whenever, the data from k ≥ 3 populations or ordinal (or) whenever
the normality assumption is violated then the Kruskal-Wallis test is an alternate statistical
procedure for testing whether the populations are identical (or) the populations are with same
median.

Assumptions

 The samples in each population (group) are drawn at random.

 The populations (groups) are independent of each other
 The populations have similar distribution shape and variability

106
This test is based on the sum of the ranks for each sample and using the rank position of the
population, it assesses whether the population median values come from population with the
same global median. That is, in hypothetical terms

H0 : All the ‘k’ populations are identical (have same median)

H1 : All the ‘k’ populations are not identical

The following are the steps which are to be used in computing the Kruskal-Wallis test value
under the setup of ‘k’ populations. Let ‘k’ be the number of populations and nk be the number
of samples in each of the kth population.

1. First rank all the nk samples in descending order. (i.e., the lowest value will be assigned
rank 1)
2. If there are any ties in ranking the data samples, assign the ranks by taking the average
of these ranks.
3. Now, compute the totals for each of the k populations, R1,R2,…Rk (say)
4. The Kruskal-Wallis test statistic is computed using
𝑘
12 𝑅𝑖2
𝐻= ∑ − 3(𝑛 + 1)
𝑛(𝑛 + 1) 𝑛𝑖
𝑖=1

Here k = number of populations

ni = size of the ith population
Ri = rank sum of the ith population
𝑛 = ∑𝑘𝑖=1 𝑛𝑖 = Total number of observations
5. To determine the critical value, we can now use the χ2 distribution table (Since the
Kruskal-Walls test approximates to χ2 distribution with k-1 degrees of freedom.)
2
i.e., 𝐻 ~𝜒𝑘−1 d.f
If computed H > critical value at α = 0.5, we will have an evidence to reject the null hypothesis,
otherwise accept.

In order to understand the methodology of this test, let us consider an example of best selling
candies which are with high calories. Consider data on calorie content which is taken from the
samples of M & M’s, Kit Kat and Milky Way II. Here, the problem is to test for significant
differences in the calorie content of these candies (α = 0.5)

107
 M & M’s 230 210 240 250 230

Kit Kat 225 205 245 235 220

Milky Way II 200 208 202 190 180

From the given information, we have k = 3; n1 = n2 = n3 = 5 and n = 5+5+5 = 15. Now, we

need to compute 𝑅𝑖2 , for this, first rank all the observations in descending order.

M & M’s Rank Kit Kat Rank Milky Way II Rank

230 10.5 225 9 200 3

210 7 205 5 208 6
240 13 245 14 202 4
250 15 235 12 190 2
230 10.5 220 8 180 1
R1 = 56 R2 = 48 R3 = 16

Here, the lowest value is 180, observed in Milky Way II is assigned rank 1, the next is 190 and
is given rank 2 and this process is continued until all the data observations are given ranks. If
we observe, the observation 230 appears twice with ranks 10 and 11, so, the average this turns
to be 10.5. from the rank sums, it can be understood that the calorie content is more in M &
M’s candy and least in Milky Way II. However, there is a need to have evidence about the
significant difference in calorie content of these candies, for which, the hypothesis can be
defined as

H0 : The calorie content is same in all the three candies

H1 : The calorie content is not same in all the three candies

The test statistic, H is given by

12 (56)2 (48)2 (16)2

𝐻= [ + + ] − 3(15 + 1)
15(15 + 1) 5 5 5

12 3136+2304+256
= 15(16) [ ] − 3(16)
5

12 5696
= 240 [ ] − 48
5

= (0.05)(1139.2) − 48

= 56.96 − 48

108
𝐻 = 8.96

2 2
For k = 3, the 𝜒𝑘−1 = 𝜒3−1 = 𝜒22 at 0.05 level is 7.378

∴ 𝐻 > 𝜒22 ⟹ 8.96 > 7.37

This creates an evidence to reject the null hypothesis and we can claim that the calorie content
is not same in all the three candies. However, it can be concluded from the rank sums that, the
samples of M & M’s candy contain high calorie content and next to it is samples of Kit Kat
and the least calorie is observed in Milky Way II.

Now consider another example where three admission test preparation programs are being
evaluated. The scores obtained by a sample of 20 people who used the programs provided the
following data. α = 0.05.

Program
A B C
540 450 600
400 540 630
490 400 580
530 410 490
490 480 590
610 370 620
550 570

The null and alternative hypothesis for the above problem can be defined as

H0: There is no difference among the three test preparation programs.

H1: There is a significant difference among the three test preparation programs.

From the given information, we have k = 3; n1 = 6, n2 = n3 = 7 and n = 6+7+7 = 20. Now, we

need to compute 𝑅𝑖2 , for this, first rank all the observations in descending order.

A Rank B Rank C Rank

540 11.5 450 5 600 17
400 2.5 540 11.5 630 20
490 8 400 2.5 580 15
530 10 410 4 490 8
490 8 480 6 590 16
610 18 370 1 620 19
550 13 570 14
R1= 58 R2= 43 R3= 109

109
The test statistic, H is given by

12 (58)2 (43)2 (109)2

𝐻= [ + + ] − 3(20 + 1)
20(20 + 1) 6 7 7

12 3364 1849 11881

= 20(21) [ + + ] − 3(21)
6 7 7

= (0.02857)(2522.0952) − 63

= 72.0563 − 63

𝐻 = 9.0563

2 2
For k = 3, the 𝜒𝑘−1 = 𝜒3−1 = 𝜒22 at 0.05 level is 7.378

∴ 𝐻 > 𝜒22 ⟹ 9.0563 > 7.378

This creates evidence to reject the null hypothesis and we can claim that the test preparation
programs are different from each other. However, it can be concluded from the rank sums that,
the scores obtained from test C are the highest among the three tests, scores from test A are
next highest and scores obtained in test B are the least.

5.5 Wald-Wolfowitz Run test

This test is used to examine whether the random samples have come from populations with
same distribution. Let ‘r’ and ‘s’ be the number of samples drawn randomly from two
independent populations. Let x1, x2,…xr be the ordered sample from first population, P1 (say)
and y1, y2, …,ys be the ordered sample from second population, P2 (say). Here, the hypothesis
is to test if the samples have been drawn from same population i.e., P1 = P2.

Let us now define the hypothesis as

H0 : Two samples come from populations having same distribution

H1 : Two samples come from populations having different distribution

Let ‘U’ denote the number of runs. A run is defined as a sequence of letters of one kind
surrounded by a sequence of letters of other kind. To obtain r, let us combine two samples by
arranging the observations in the order of magnitude, as

110
 x1x2y1y2x3y3…xkyk+1…

2𝑛1 𝑛2
𝐸(𝑈) = 𝜇 = +1
𝑛1 + 𝑛2

2𝑛1 𝑛2 (2𝑛1 𝑛2 − 𝑛1 − 𝑛2 )
𝑉(𝑈) = 𝜎 2 =
(𝑛1 + 𝑛2 )2 (𝑛1 + 𝑛2 − 1)

and the test statistic for testing the hypothesis is given by

𝑈−𝐸(𝑈)
𝑍= ~𝑁(0, 1) asymptotically
√𝑉(𝑈)

Here U is the observed number of runs. n1 and n2 are the number of observations in first and
second population.

To understand the procedure of conducting Wald-Wolfowitz run test, let us consider an

example of two methods of teaching. A group of students was randomly divided into two
groups. One group was taught to read using a uniform method of teaching under teacher’s
direction. The second group was taught to read using an individual method of teaching with
programmed workbook. The following are the test scores which projects the reading ability of
two groups of students.

First
227 176 252 149 16 55 234 194 247 92 184 147 88 161 171
Group
Second
202 14 165 171 292 271 151 235 147 99 63 284 53 228 271
Group

Test the equality of the distribution of scores of two groups.

As per the procedure, first we need to define the hypothesis then an arrangement of given scores
of two groups is to be made as per the order of magnitude.

H0 : The distribution of scores from first and second groups are same.

H1 : The distribution of scores from first and second groups are not same.

111
Arrangement of sequence and defining runs:

Data
14 16 53 55 63 88 92 99 147 147 149 151 161 165 171
Group 2 1 2 1 2 1 1 2 1 2 1 2 1 2 1
Data 171 176 184 194 202 227 228 234 235 247 252 271 271 284 292
Group 2 1 1 1 2 1 2 1 2 1 1 2 2 2 2

Total number of runs U = 23 and n1 = n2 = 15.

2(15)(15)
𝐸(𝑈) = 𝜇 = +1
15 + 15

2 × 225
= +1
30

450
= +1
30

= 15 + 1

𝐸(𝑈) = 𝜇 = 16

2(15 × 15)(2(15 × 15) − 15 − 15)

𝑉(𝑈) = 𝜎 2 =
(15 + 15)2 (15 + 15 − 1)

2(225)(2(225) − 30)
=
(30)2 (29)

450(450 − 30)
=
(900)(29)

450 × 420 189,000

= =
900 × 29 26,100

⇒ 𝑉𝑎𝑟(𝑈) = 7.2413

𝑈 − 𝐸(𝑈) 23 − 16
∴𝑍= ⇒𝑍=
√𝑉(𝑈) √7.2413

7
=
2.6909

112
 ⇒ 𝑍𝑐𝑎𝑙 = 2.6013

At α = 0.05, the Zcri value is 1.96

∴ 𝑍𝑐𝑎𝑙 > 𝑍𝑐𝑟𝑖 ⇒ 2.6013 > 1.96

This claims that the distribution of test scores is not the same in first and second groups.

Let us consider another example of winning or losing of a local softball team. The following
are the outcomes of the last 24 games played by a local softball team. The letter W signifies a
win and L a loss. W, L, L, L, W, L, L, W, L, L, W, L, L, W, L, W, L, L, L, L, W, L, W, L. Is
this dataset consistent with randomness?

In order to test the given statement about the randomness, first we need to define the
hypothetical statements which are given below

H0 : Wins and losses are random.

H1 : Wins and losses are not random.

From the given information, we need to identify the pattern of runs which is denoted with a
flower bracket as below

𝑊
⏟ 𝐿𝐿𝐿
⏟𝑊 ⏟ 𝐿𝐿
⏟𝑊 ⏟ 𝐿𝐿
⏟𝑊 ⏟ 𝐿𝐿
⏟𝑊 ⏟⏟
𝐿𝑊⏟ 𝐿𝐿𝐿𝐿
⏟ 𝑊 ⏟⏟
𝐿𝑊⏟⏟
𝐿

Total number of runs = U =16, n1 = 8 and n2 = 16

2(8)(16)
𝐸(𝑈) = 𝜇 = +1
8 + 16

2 × 128
= +1
24

256
= +1
24

= 10.6667 + 1

𝐸(𝑈) = 𝜇 = 11.6667

113
 2(8 × 16)(2(8 × 16) − 8 − 16)
𝑉(𝑈) = 𝜎 2 =
(8 + 16)2 (8 + 16 − 1)

2(128)(2(128) − 24)
=
(24)2 (23)

256(256 − 24)
=
(576)(23)

256 × 232 59,392

= =
576 × 23 13,248

⇒ 𝑉𝑎𝑟(𝑈) = 4.4830

𝑈 − 𝐸(𝑈) 16 − 11.6667
∴𝑍= ⇒𝑍=
√𝑉(𝑈) √4.4830

4.3333
=
2.1173

⇒ 𝑍𝑐𝑎𝑙 = 2.0466

At α = 0.05, the Zcri value is 1.96

∴ 𝑍𝑐𝑎𝑙 > 𝑍𝑐𝑟𝑖 ⇒ 2.0466 > 1.96

This provides evidence to claim that the wins and losses of the last 24 games are not at random.

5.6 Chi-Square test

The Chi-Square test is a distribution free statistic, usually use for testing the goodness of fit of
data whether it is consistent with respect to the hypothesized distribution and another use is to
test for independence between two attributes or categorical variables. Many statistical tests are
approximated to Chi-Square distribution. It describes the magnitude of difference between
observed and expected frequencies.

Chi-Square test for Goodness of fit

This test is applied when you have a categorical variable from a single population. It is used to
compare observed data with expected data under a specific hypothesis. Means, the interest lies
in knowing the “goodness of fit” between observed and expected data. For this test, the null

114
hypothesis can be stated as that “there is no significant difference between the expected and
observed result”. In simple, the Chi-square test for goodness of fit is used to how much the
observed data is fit for a hypothesized distribution.

Let oi, denotes the number of counts in ith category of a categorical variable and ei represent
the expected count of the ith category of the same categorical variable. Here i= 1, 2,…k; k being
the number of categories. The expected value of ith category can be computed as

𝑒𝑖 = 𝑛 ∗ 𝑝𝑖 ; pi is the proportion or percentage of ith category and n is total count.

For this particular test, the hypothesis takes the following form

H0 : The data are consistent with a specified distribution.

H1 : The data are not consistent with a specified distribution.

Then the Karl Pearson χ2 test is given by

𝑘
2
(𝑜𝑖 − 𝑒𝑖 )2 2
𝜒 = ∑[ ] ~𝜒𝑘−1 𝑑. 𝑓.
𝑒𝑖
𝑖=1

2 2
If 𝜒𝑐𝑎𝑙 ≤ 𝜒𝑘−1 , there will not be any evidence to reject the null hypothesis, otherwise.

The implementation of the procedure of goodness of fit can be illustrated using the following
example of Acme toy company prints baseball cards.

The company claims that 30% of the cards are rookies, 60% of veterans and 10% are All-stars.
A random sample of 100 cards is taken, of which, 50 rookies, 45 veterans and 5 All-stars are
observed. Is this consistent with Acme’s claim? Test this at α = 5% level of significance.

For the given problem, first we need to define the null hypothesis as

H0 : The proportions of rookies, veterans and All-stars is 30%, 60% and 10% respectively.

H1 : At least one of the proportion in null hypothesis is false.

The expected frequencies can be computed as 𝑒𝑖 = 𝑛 ∗ 𝑝𝑖

i.e., 𝑒1 = 100 ∗ 0.30 = 30; 𝑒2 = 100 ∗ 0.60 = 60; 𝑒3 = 100 ∗ 0.10 = 10

115
 2
(𝑜1 − 𝑒1 )2 (𝑜2 − 𝑒2 )2 (𝑜3 − 𝑒3 )2
𝜒 = + +
𝑒1 𝑒2 𝑒3

(50 − 30)2 (45 − 60)2 (5 − 10)2

= + +
30 60 10

(20)2 (−15)2 (−5)2

= + +
30 60 10

400 225 25
= + +
30 60 10

= 13.33 + 3.75 + 2.5

𝜒 2 = 19.58

2 2
The 𝜒𝑘−1 = 𝜒3−1 = 𝜒22 at 0.05 is 5.99

2
∴ 𝜒𝑐𝑎𝑙 = 19.58 > 𝜒22 = 5.99

This claims to reject the null hypothesis and with evidence we can conclude that the proportions
claimed by Acme’s are not correct.

Another example is as follows

Over a long period of time, a research team monitored the number of car accidents which
occurred in a particular county. The following table summarizes the data relating to the day of
the week on which the accident occurred.

Day M Tu W Th F S Su
No. of Accidents 60 54 48 53 53 75 77

Investigate the hypothesis that these data are a random sample from uniform distribution.
The null and alternative hypothesis can be defined as
H0 : The data are a random sample from uniform distribution.

H1 : The data are not a random sample from uniform distribution.

The expected frequencies are based on uniform distribution and each ei can be computed as

116
 1
(60 + 54 + 48 + 53 + 53 + 75 + 77) = 60
7

The Chi square statistic is computed using the following table

(𝑜𝑖 − 𝑒𝑖 )2
Day 𝑜𝑖 𝑒𝑖 𝑜𝑖 − 𝑒𝑖 (𝑜𝑖 − 𝑒𝑖 )2
𝑒𝑖
M 60 60 0 0 0.000
Tu 54 60 -6 36 0.600
W 48 60 -12 144 2.400
Th 53 60 -7 49 0.817
F 53 60 -7 49 0.817
S 75 60 15 225 3.750
Su 77 60 17 289 4.817
2
𝜒 =13.200
2 2
The 𝜒𝑘−1 = 𝜒7−1 = 𝜒62 at 0.05 is 12.592

2
∴ 𝜒𝑐𝑎𝑙 = 13.2 > 𝜒62 = 12.592

This claims to reject the null hypothesis and with evidence we can conclude that the dta does
not follow uniform distribution. i.e., The accidents occurring on different days of the week are
not uniform.

Chi-square test for independence of attributes

This test is used to determine whether there is a significant association between two categorical
variables. The null and alternative hypothesis can be stated as

H0 : The two categorical variables are not associated.

H1 : The two categorical variables are associated.

Let us consider two attributes (categorical variables) A and B in such a way that A and B are
divided into r and c subcategories i.e., A1, A2, …Ar and B1, B2, …Bc respectively. Let (Ai) and
(Bj) denote the number of frequency (count) possessing attribute A and B respectively. In other
words, Oij represents the observed frequency of ith row and jth column of the categorical
variables A and B. In order to test the null hypothesis, first the expected frequencies are to be
computed in the following way:

117
 𝑅𝑖 ×𝐶𝑗 𝑅𝑜𝑤 𝑇𝑜𝑡𝑎𝑙∗𝐶𝑜𝑙𝑢𝑚𝑛 𝑇𝑜𝑡𝑎𝑙
𝐸𝑖𝑗 = (or) 𝐸𝑖𝑗 =
𝑁 𝑇𝑜𝑡𝑎𝑙 𝑠𝑎𝑚𝑝𝑙𝑒 𝑠𝑖𝑧𝑒

Here Ri is the sum or total of the ith subcategory of attribute A

Cj is the sum or total of jth subcategory of attribute B

N is the total number; i = 1, 2, …r; j = 1, 2, …c

The number of degrees of freedom for the test is (r-1)(c-1) with ‘r’ number of rows and ‘c’
number of columns.

Now, we need to compute the χ2 test statistic as

𝑟 𝑐 2
2
(𝑂𝑖𝑗 − 𝐸𝑖𝑗 ) 2
𝜒 = ∑∑[ ] ~𝜒(𝑟−1)(𝑐−1) 𝑑. 𝑓.
𝐸𝑖𝑗
𝑖=1 𝑗=1

2 2
If 𝜒𝑐𝑎𝑙 > 𝜒(𝑟−1)(𝑐−1) , then we do not accept the null hypothesis.

Let us consider an example of a public opinion poll surveyed a simple random sample of 1000
voters. Respondents were classified by gender (male or female) and by voting preference
(republic, democrat or independent). The following table shows the distribution of the voters.

Voting Preferences
Row Total
Republican Democrat Independent
Male 200 150 50 400
Gender
Female 250 300 50 600
Column Total 450 450 100 1000

Do the men’s voting preferences significantly differ from the women’s preferences? Use a 0.05
level of significance.

In the given information on voting preferences, the statement to be tested is defined in terms
of hypothesis as

H0 : Gender and voting preferences are independent.

H1 : Gender and voting preferences are not independent.

The expected frequencies are calculated as follows

118
 𝑅1 × 𝐶1 400 × 450 180000
𝐸11 = = = = 180
𝑁 1000 1000

𝑅1 × 𝐶2 400 × 450 180000

𝐸12 = = = = 180
𝑁 1000 1000

𝑅1 × 𝐶3 400 × 100 40000

𝐸13 = = = = 40
𝑁 1000 1000

𝑅2 × 𝐶1 600 × 450 270000

𝐸21 = = = = 270
𝑁 1000 1000

𝑅2 × 𝐶2 600 × 450 270000

𝐸22 = = = = 270
𝑁 1000 1000

𝑅2 × 𝐶3 600 × 100 60000

𝐸23 = = = = 60
𝑁 1000 1000

(𝑂11 − 𝐸11 )2 (𝑂12 − 𝐸12 )2 (𝑂13 − 𝐸13 )2 (𝑂21 − 𝐸21 )2 (𝑂22 − 𝐸22 )2
𝜒2 = + + + +
𝐸11 𝐸12 𝐸13 𝐸21 𝐸22
(𝑂23 − 𝐸23 )2
+
𝐸23

(200 − 180)2 (150 − 180)2 (50 − 40)2 (250 − 270)2 (300 − 270)2
= + + + +
180 180 40 270 270
(50 − 60)2
+
60

(20)2 (−30)2 (10)2 (−20)2 (30)2 (−10)2

= + + + + +
180 180 40 270 270 60

400 900 100 400 900 100

= + + + + +
180 180 40 270 270 60

𝜒 2 = 2.22 + 5.00 + 2.50 + 1.48 + 3.33 + 1.67

⟹ 𝜒 2 = 16.2

Degrees of freedom: (r-1)(c-1) = (2-1)(3-1) = 2

𝜒22 at 0.05 level of significance = 5.99

2
∴ 𝜒𝑐𝑎𝑙 > 𝜒22 ⇒ 16.2 > 5.99

119
With this result, there is evidence to reject the null hypothesis. Thus, we can conclude that the
voting and gender preference are not independent. In other words, we can say that there is a
relationship between gender and voting preference.

Let us have one more illustration to have a clear picture about the computational aspects and
the practical application of this test. A market research firm has distributed samples of a new
shampoo to a variety of individuals. The following data summarizes the comments of these
individuals about the shampoo as well as provide the age group into which they fall.

Age Group (in years)

Rating
15-20 21-30 Over 30
Excellent 18 20 41
Good 25 27 43
Fair 17 15 26
Poor 3 2 8

Do these data prove that different age groups have different opinions about the shampoo at 5
percent level of significance?
From the above explained situation on comments given by individuals about the shampoo, the
null and alternative hypothesis can be defined as

H0 : Age group and Ratings (opinions) are independent.

H1 : Age group and Ratings (opinions) are not independent.

Before going to compute the expected frequencies, we need to calculate the row and column
totals which are given in the following table.

Age Group (in years) Row

Rating
15-20 21-30 Over 30 Total
Excellent 18 20 41 79
Good 25 27 43 95
Fair 17 15 26 58
Poor 3 2 8 13
Column Total 63 64 118 245

The expected frequencies can now be calculated as

𝑅1 × 𝐶1 79 × 63 4977
𝐸11 = = = = 20.3142
𝑁 245 245

120
 𝑅1 × 𝐶2 79 × 64 5056
𝐸12 = = = = 20.6367
𝑁 245 245

𝑅1 × 𝐶3 79 × 118 9322
𝐸13 = = = = 38.0489
𝑁 245 245

𝑅2 × 𝐶1 95 × 63 5985
𝐸21 = = = = 24.4286
𝑁 245 245

𝑅2 × 𝐶2 95 × 64 6080
𝐸22 = = = = 24.8163
𝑁 245 245

𝑅2 × 𝐶3 95 × 118 11210
𝐸23 = = = = 45.7551
𝑁 245 245

𝑅3 × 𝐶1 58 × 63 3654
𝐸31 = = = = 14.9143
𝑁 245 245

𝑅3 × 𝐶2 58 × 64 3712
𝐸32 = = = = 15.1510
𝑁 245 245

𝑅3 × 𝐶3 58 × 118 6844
𝐸33 = = = = 27.9347
𝑁 245 245

𝑅4 × 𝐶1 13 × 63 819
𝐸41 = = = = 3.3428
𝑁 245 245

𝑅4 × 𝐶2 13 × 64 832
𝐸42 = = = = 3.3959
𝑁 245 245

𝑅4 × 𝐶3 13 × 118 1534
𝐸43 = = = = 6.2612
𝑁 245 245

(𝑂11 − 𝐸11 )2 (𝑂12 − 𝐸12 )2 (𝑂13 − 𝐸13 )2 (𝑂21 − 𝐸21 )2 (𝑂22 − 𝐸22 )2
𝜒2 = + + + +
𝐸11 𝐸12 𝐸13 𝐸21 𝐸22
(𝑂23 − 𝐸23 )2 (𝑂31 − 𝐸31 )2 (𝑂32 − 𝐸32 )2 (𝑂33 − 𝐸33 )2
+ + + +
𝐸23 𝐸31 𝐸32 𝐸33
(𝑂41 − 𝐸41 )2 (𝑂42 − 𝐸42 )2 (𝑂43 − 𝐸43 )2
+ + +
𝐸41 𝐸42 𝐸43

121
 (18 − 20.3142)2 (20 − 20.6367)2 (41 − 38.0489)2 (25 − 24.4286)2
= + + +
20.3142 20.6367 38.0489 24.4286
(27 − 24.8163)2 (43 − 45.7551)2 (17 − 14.9143)2
+ + +
24.8163 45.7551 14.9143
(15 − 15.1510)2 (26 − 27.9347)2 (3 − 3.3428)2 (2 − 3.3959)2
+ + + +
15.1510 27.9347 3.3428 3.3959
(8 − 6.2612)2
+
6.2612

(−2.3142)2 (−0.6367)2 (2.9511)2 (0.5714)2 (2.1837)2 (−2.7551)2

= + + + + +
20.3142 20.6367 38.0489 24.4286 24.8163 45.7551
(2.0857)2 (−0.1510)2 (−1.9347)2 (−0.3428)2 (−1.3959)2
+ + + + +
14.9143 15.1510 27.9347 3.3428 3.3959
(1.7388)2
+
6.2612

4.5548 0.4054 8.7090 0.3265 4.7685 7.5906 4.3501

= + + + + + +
20.3142 20.6367 38.0489 24.4286 24.8163 45.7551 14.9143
0.0228 3.7431 0.1175 1.9485 3.0234
+ + + + +
15.1510 27.9347 3.3428 3.3959 6.2612

𝜒 2 = 0.2242 + 0.0196 + 0.2288 + 0.0133 + 0.1921 + 0.1659 + 0.2916 + 0.0015

+ 0.1339 + 0.0351 + 0.5737 + 0.4828

⟹ 𝜒 2 = 2.3625

Degrees of freedom: (r-1)(c-1) = (4-1)(3-1) = 6

𝜒62 at 0.05 level of significance = 12.59

2
∴ 𝜒𝑐𝑎𝑙 < 𝜒22 ⇒ 2.3625 < 12.59

With this result, there is evidence to reject the null hypothesis. Thus, we can conclude that the
voting and gender preference are not independent. In other words, we can say that there is a
relationship between gender and voting preference.

122
 EXERCISES

Problems on Sign Test

17. The following table lists the preferences indicated by 10 individuals in taste tests
involving two brands of a product.
Individual Brand A versus Individual Brand A versus
Brand B Brand B
1 + 6 +
2 + 7 -
3 + 8 +
4 - 9 -
5 + 10 +
With α = 0.05, test for a significant difference in the preferences for the two brands. A plus
indicates a preference for brand A over brand B.

18. The following hypothesis test is to be conducted.

H0 : Median ≤ 150
H1 : Median > 150
A sample of size 30 yields 22 cases in which a value greater than 150 is obtained, three cases
in which a value of exactly 150 is obtained, and five cases in which a value less than 150 is
obtained. Use α = 0.01 and conduct the hypothesis test.

19. Are stock splits beneficial to stockholders? SNL Securities studied stock splits in the
banking industry over an 18-month period and found that stock splits tended to increase
the value of an individual’s stock holding. Assume that of a sample of 20 recent stock
splits, 14 led to an increase in value, four led to a decrease in value, and two resulted in
no change. Suppose a sign test is to be used to determine whether stock splits continue
to be beneficial for holders of bank stocks.
a. What are the null and alternative hypothesis?
b. With α = 0.05, what is your conclusion?

20. A poll asked 1253 adults a series of questions about the state of the economy and their
children’s future. One question was, “Do you expect your children to have a better life
than you have had, a worse life, or a life about as good as yours?” The responses were
34% better , 29% worse, 33% about the same, and 4% not sure. Use the sign test and a

123
 0.05 level of significance to determine whether more adults feel their children will have
a better future than feel their children will have a worse future. What is your conclusion?

21. Nielsen Media Research identified American Idol and Dancing with the Stars as the
two top-rated prime-time television shows (USA Today, April 14, 2008). In a local
television preference survey, 750 individuals were asked to indicate their favorite
prime-time television show: Three hundred thirty selected American Idol, 270 selected
Dancing with the Stars, and 150 selected another television show. Use a .05 level of
significance to test the hypothesis that there is no difference in the preference for the
American Idol and Dancing with the Stars television shows. What is your conclusion?

22. Competition in the personal computer market is intense. A sample of 450 purchases
showed 202 Brand A computers, 175 Brand B computers, and 73 other computers. Use
a .05 level of significance to test the null hypothesis that Brand Aand Brand B have the
same share of the personal computer market. What is your conclusion?

23. The median annual income of subscribers to Shutterbug magazine is $75,000 (Home
Theater website, August 18, 2008). A sample of 300 subscribers to Popular
Photography & Imaging magazine found 165 subscribers with an annual income over
$75,000 and 135 with an annual income under $75,000. Can you conclude that the
median annual income of Popular Photography & Imaging subscribers differs from the
median annual income of Shutterbug subscribers? Use α = 0.05.

24. The median number of part-time employees at fast-food restaurants in a particular city
was known to be 18 last year. City officials think the use of part-time employees may
be increasing. A sample of nine fast-food restaurants showed that seven restaurants
were employing more than 18 part-time employees, one restaurant was employing
exactly 18 part-time employees, and one restaurant was employing fewer than 18 part-
time employees. Can it be concluded that the median number of part-time employees
has increased? Test using α = 0.05.

25. Net assets for the 50 largest stock mutual funds show a median of $15 billion (The Wall
Street Journal, March 2, 2009). A sample of 10 of the 50 largest bond mutual funds
follows.

124
 Bond Fund Net Assets Bond Fund Net Assets
Fidelity Intl Bond 6.1 T Rowe Price New Income 6.9
Franklin CA TF 11.7 Vanguard GNMA 15
American Funds 22.4 Oppenheimer Intl Bond 6.6
Vanguard Short Term 9.6 Dodge & Cox Income 14.5
PIMCO: Real Return 4.9 iShares: TIPS Bond 9.6
Using the median, can it be concluded that bond mutual funds are smaller and have fewer net
assets than stock mutual funds? Use α = 0.05.
a. What are the hypotheses for this test?
b. What is the p-value? What is your conclusion?

26. The median annual income for families living in the United States is $56,200 (The New
York Times Almanac, 2008). Annual incomes in thousands of dollars for a sample of
50 families living in Chicago, Illinois, are shown. Use the sample data to see if it can
be concluded that the families living in Chicago have a median annual income greater
than $56,200. Use α = 0.05. What is your conclusion?
66.3 60.2 49.9 75.4 73.7
65.7 61.1 123.8 57.3 48.5
74 146.3 92.2 43.7 86.9
59.7 64.2 56.2 48.9 109.6
39.8 60.9 79.7 42.3 52.6
60.9 43.5 61.7 54.7 95.2
70.4 43.8 57.8 83.5 56.5
51.3 42.9 87.5 43.6 67.2
48.7 79.1 61.9 53.4 56.2
57 49.6 109.5 42.1 74.6

27. A Pew Research Center survey asked adults if their ideal place to live would have a
faster pace of life or a slower pace of life (USA Today, February 13, 2009). A
preliminary sample of 16 respondents showed 4 preferred a faster pace of life, 11
preferred a slower place of life, and 1 said it did not matter.
a. Are these data sufficient to conclude there is a difference between the preferences
for a faster pace of life or a slower pace of life? Use α = 0.05. What is your
conclusion?
b. Considering the entire sample of 16 respondents, what is the percentage who would
like a faster pace of life? What is the percentage who would like a slower pace of
life? What recommendation do you have for the study?

125
 28. An engineering firm is involved in selecting a computer system, and the choice has
been narrowed to two manufacturers. The firm submits eight problems to the two
computer manufacturers and has each manufacturer measure the number of seconds
required to solve the design problem with the manufacturer’s software. The times for
the eight design problems are given below.
Design problem 1 2 3 4 5 6 7 8
Time with computer A 15 32 17 26 42 29 12 38
Time with computer B 22 29 1 23 46 25 19 47
Determine the p-value of the sign test when testing the hypothesis that there is no difference in
the distribution of the time it takes the two types of software to solve problems.

29. To test the hypothesis that the median weight of 16-year-old females from Los Angeles
is at least 110 pounds, a random sample of 200 such females was chosen. If 120 females
weighed less than 110 pounds, does this discredit the hypothesis? Use the 5 percent
level of significance. What is the p-value?

30. In 2004, the national median salary of all U.S. financial accountants was $124,400. A
recent random sample of 14 financial accountants showed 2007 incomes of (in units of
$1,000)
125.5, 130.3, 133.0, 102.6, 198.0, 232.5, 106.8,
114.5, 122.0, 100.0, 118.8, 108.6, 312.7, 125.5
Use these data to test the hypothesis that the median salary of financial accountants in 2007
was not greater than in 2004. What is the p-value?

31. An experiment was initiated to study the effect of a newly developed gasoline detergent
on automobile mileage. The following data, representing mileage per gallon before and
after the detergent was added for each of eight cars, resulted.
Mileage Mileage
Car
without Additive with Additive
1 24.2 23.5
2 30.4 29.6
3 32.7 32.3
4 19.8 17.6
5 25.0 25.3
6 24.9 25.4
7 22.2 20.6
8 21.5 20.7

126
Find the p-value of the test of the hypothesis that mileage is not affected by the additive when
a. the sign test is used; and (b) the signed rank test is used.

32. An engineer claims that painting the exterior of a particular aircraft affects its cruising
speed. To check this, the next 10 aircraft off the assembly line were flown to determine
cruising speed prior to painting, and were then painted and reflown. The following data
resulted.
Cruising Speed (knots)
Aircraft
Not Painted Painted
1 426.1 416.7
2 418.4 403.2
3 424.4 420.1
4 438.5 431
5 440.6 432.6
6 421.8 404.2
7 412.2 398.3
8 409.8 405.4
9 427.5 422.8
10 441.2 444.8
Do the data uphold the engineer’s claim?

33. Ten pairs of duplicate spectrochemical determinations for nickel are presented below.
The readings in column 2 were taken with one type of measuring instrument and those
in column 3 were taken with another type.
Sample Duplicates Sample Duplicates
1 1.94 2 6 1.96 1.98
2 1.99 2.09 7 1.95 2.03
3 1.98 1.95 8 1.96 2.03
4 2.07 2.03 9 1.92 2.01
5 2.03 2.08 10 2 2.12
Test the hypothesis, at the 5 percent level of significance, that the two measuring instruments
give equivalent results.

Problems on Wilcoxon Signed-rank Test

34. Two fuel additives are tested to determine their effect on miles per gallon for passenger
cars. Test results for 12 cars follow; each car was tested with both fuel additives. Use α
= 0.05 and the Wilcoxon signed-rank test to see whether there is a significant difference
between the median miles per gallon for the additives.

127
 Additive Additive
Car 1 2 Car 1 2
1 20.12 18.05 7 16.16 17.2
2 23.56 21.77 8 18.55 14.98
3 22.03 22.57 9 21.87 20.03
4 19.15 17.06 10 24.23 21.15
5 21.23 21.22 11 23.21 22.78
6 24.77 23.8 12 25.02 23.7

35. A sample of 10 men was used in a study to test the effects of a relaxant on the time
required to fall asleep. Data for 10 subjects showing the number of minutes required to
fall asleep with and without the relaxant follow. Use a 0.05 level of significance to
determine whether the relaxant reduces the median time required to fall asleep. What is
your conclusion?
Relaxant Relaxant
Subject No Yes Subject No Yes
1 15 10 6 7 5
2 12 10 7 8 10
3 22 12 8 10 7
4 8 11 9 14 11
5 10 9 10 9 6

36. Percents of on-time arrivals for flights in 2006 and 2007 were collected for 11 randomly
selected airports. Data for these airports follow (Research and Innovative Technology
Administration website, August 29, 2008). Use α = 0.05 to test the hypothesis that there
is no difference between the median percent of on-time arrivals for the two years. What
is your conclusion?
Percent on Time
Airport 2006 2007
Boston Logan 71.78 69.69
Chicago O’Hare 68.23 65.88
Chicago Midway 77.98 78.4
Denver 78.71 75.78
Fort Lauderdale 77.59 73.45
Houston 77.67 78.68
Los Angeles 76.67 76.38
Miami 76.29 70.98
New York (JFK) 69.39 62.84
Orlando 79.91 76.49
Washington (Dulles) 75.55 72.42

128
 37. The PGA Players Championship was held at the Sedgefield Country Club in
Greensboro, North Carolina, August 11–17, 2008. Shown here are first-round and
second-round scores for a random sample of 11 golfers. Use α = 0.05 to determine
whether the first- and second-round median scores for golfers in the Players
Championship differed significantly.
Golfer 1st Round 2nd Round
Marvin Laird 63 74
Jimmy Walker 70 73
Kevin Chappell 72 70
Kevin Duke 65 71
Andrew Buckle 70 74
Paul Claxton 69 73
Larry Mize 72 71
Chris Riley 68 70
Bubba Watson 70 68
Carlos Franco 71 71
Richard Johnson 72 69

38. Fifteen cities, of roughly equal size, are chosen for a traffic safety study. Eight of them
are randomly chosen, and in these cities a series of newspaper articles dealing with
traffic safety is run over a 1 month period. The number of traffic accidents reported in
the month following this campaign is as follows:
Treatment group 19 31 39 45 47 66 74 81
Control group 28 36 44 49 52 52 60 72
Determine the exact p-value when testing the hypothesis that the articles have not had any
effect.
Problems on Kruskal-Wallis Test
39. A sample of 15 consumers provided the following product ratings for three different
products. Five consumers were randomly assigned to test and rate each product. Use
the Kruskal-Wallis test and α = 0.05 to determine whether there is a significant
difference among the ratings for the products.
Product
A B C
50 80 60
62 95 45
75 98 30
48 87 58
65 90 57

129
40. Forty-minute workouts of one of the following activities three days a week will lead to
a loss of weight. The following sample data show the number of calories burned during
40- minute workouts for three different activities. Do these data indicate differences in
the amount of calories burned for the three activities? Use a .05 level of significance.
What is your conclusion?
Swimming Tennis Cycling
408 415 385
380 485 250
425 450 295
400 420 402
427 530 268

41. Condé Nast Traveler magazine conducts an annual survey of its readers in order to rate
the top 80 cruise ships in the world (Condé Nast Traveler, February 2008). With 100
the highest possible rating, the overall ratings for a sample of ships from the Holland
America, Princess, and Royal Caribbean cruise lines are shown. Use the Kruskal-Wallis
test with α = 0.05 to determine whether the overall ratings among the three cruise lines
differ significantly. What is your conclusion?
Holland America Princess Royal Caribbean
Ship Rating Ship Rating Ship Rating
Amsterdam 84.5 Coral 85.1 Adventure 84.8
Maasdam 81.4 Dawn 79 Jewel 81.8
Ooterdam 84 Island 83.9 Mariner 84
Volendam 78.5 Princess 81.1 Navigator 85.9
Westerdam 80.9 Star 83.7 Serenade 87.4

42. A large corporation sends many of its first-level managers to an off-site supervisory
skills training course. Four different management development centers offer this
course. The director of human resources would like to know whether there is a
difference among the quality of training provided at the four centers. An independent
random sample of five employees was chosen from each training center. The employees
were then ranked 1 to 20 in terms of supervisory skills. A rank of 1 was assigned to the
employee with the best supervisory skills. The ranks are shown. Use α = 0.05 and test
whether there is a significant difference among the quality of training provided by the
four programs.

130
 Course
A B C D
3 2 19 20
14 7 16 4
10 1 9 15
12 5 18 6
13 11 17 8

Problems on Wald-Wolfowitz Run test

43. A production run of 50 items resulted in 11 defectives, with the defectives occurring on
the following items (where the items are numbered by their order of production): 8, 12,
13, 14, 31, 32, 37, 38, 40, 41, 42. Can we conclude that the successive items did not
constitute a random sample?

44. The win-lose record of a certain basketball team for its last 50 consecutive games was
as follows-
WWWWWWLWWWWWWLWLWWWLLWWWWLWWWLLWWW
WWWLLWWLLLWWLWWW
Apply run test to test that sequence of wins and losses is random.

45. A random sample of 8 household is selected from a village A whose daily expense on
milk is 11, 15, 17, 19, 25, 27, 31, 33. Another sample of 9 households is selected from
village B whose expense on milk is 12, 16, 20, 22, 28, 30, 36, 38, 42. Test whether the
households of the two villages are same on spending daily milk expenses.

46. Two makes of cars were sampled randomly, to determine the mileage (in thousands)
until the brakes required relining. Calculate the probability value for the null hypothesis
that make A and make B are similar.
Make A Make B
30 22
41 26
48 32
49 39
61

131
 47. A random sample of 8 men’s heights and an independent random sample of 8 women’s
heights were observed as follows
Men’s Women’s
Height (M) Height (M)
65 62
67 63
69 64
70 65
71 66
73 68
76 69
77 71
Use run test to check whether the distribution of heights is same in men and women.

48. 12 people are polled to find out if they use a certain product and the outcomes are
recorded according to their sex by the symbols M and F. Use run test to check for the
randomness if the observed sequence is M M F F F M F F M M M M.

49. A machine is adjusted to dispense acrylic paint thinner into a container. Would you say
that the amount of paint thinner being dispensed by this machine varies randomly if the
contents of the next 15 containers are measured and found to be 3.6, 3.9, 4.1, 3.6, 3.8,
3.7, 3.4, 4.0, 3.8, 4.1, 3.9, 4.0, 3.8, 4.2 and 4.1 liters? Use a 0.01 level of significance.

50. A series of 20 coin tosses might produce the following sequence of heads (H) and tails
(T). H H T T H T H H H H T H H T T T T T H H. Test for the randomness of occurrence
of heads and tails.

51. A college is interested in whether there is any predictable change in the academic
performance of international students from the first to the second semester of their first
year at a university. A random sample of 20 students is selected and their first and
second semester grade point averages are
First 1.53 2.00 1.93 3.90 2.14 1.52 0.91 1.95 3.00 1.67
Sem 2.78 1.21 1.66 1.75 2.96 1.50 2.25 2.66 1.87 3.50
Second 3.67 2.74 3.50 3.27 1.97 1.54 3.42 1.04 2.45 2.09
Sem 2.00 3.00 1.78 2.31 2.25 2.20 0.91 1.52 1.61 2.56
Problems on Chi-Square for Goodness of fit

132
 52. Test the following hypotheses by using the χ2 goodness of fit test.
H0: pA = 0.40, pB = 0.40, and pC = 0.20
Ha: The population proportions are not
pA = 0.40, pB = 0.40, and pC = 0.20
A sample of size 200 yielded 60 in category A, 120 in category B, and 20 in category C. Use α
= 0.01 and test to see whether the proportions are as stated in H0.
a. Use the p-value approach.
b. Repeat the test using the critical value approach.

53. Suppose we have a multinomial population with four categories: A, B, C, and D. The
null hypothesis is that the proportion of items is the same in every category. The null
hypothesis is
H0: pA = pB = pC = pD = 0.25
A sample of size 300 yielded the following results.
A: 85 B: 95 C: 50 D: 70
Use α = 0.05 to determine whether H0 should be rejected. What is the p-value?

54. During the first 13 weeks of the television season, the Saturday evening 8:00 p.m. to
9:00 p.m. audience proportions were recorded as ABC 29%, CBS 28%, NBC 25%, and
independents 18%. A sample of 300 homes two weeks after a Saturday night schedule
revision yielded the following viewing audience data: ABC 95 homes, CBS 70 homes,
NBC 89 homes, and independents 46 homes. Test with α = 0.05 to determine whether
the viewing audience proportions changed.

55. M&M/MARS, makers of M&M® chocolate candies, conducted a national poll in

which more than 10 million people indicated their preference for a new color. The tally
of this poll resulted in the replacement of tan-colored M&Ms with a new blue color. In
the brochure “Colors,” made available by M&M/MARS Consumer Affairs, the
distribution of colors for the plain candies is as follows:
Brown Yellow Red Orange Green Blue
30% 20% 20% 10% 10% 10%
In a follow-up study, samples of 1-pound bags were used to determine whether the reported
percentages were indeed valid. The following results were obtained for one sample of 506 plain
candies.

133
 Brown Yellow Red Orange Green Blue
177 135 79 41 36 38
Use α = 0.05 to determine whether these data support the percentages reported by the company.

56. Where do women most often buy casual clothing? Data from the U.S. Shopper Database
provided the following percentages for women shopping at each of the various outlets
(The Wall Street Journal, January 28, 2004).
Outlet Percentage Outlet Percentage
Wal-Mart 24 Kohl’s 8
Traditional department Stores 11 Mail order 12
JC Penney 8 Other 37
The other category included outlets such as Target, Kmart, and Sears as well as numerous
smaller specialty outlets. No individual outlet in this group accounted for more than 5% of the
women shoppers. A recent survey using a sample of 140 women shoppers in Atlanta, Georgia,
found 42 Wal-Mart, 20 traditional department store, 8 JC Penney, 10 Kohl’s, 21 mail order,
and 39 other outlet shoppers. Does this sample suggest that women shoppers in Atlanta differ
from the shopping preferences expressed in the U.S. Shopper Database? What is the p-value?
Use α = 0.05. What is your conclusion?

57. The American Bankers Association collects data on the use of credit cards, debit cards,
personal checks, and cash when consumers pay for in-store purchases (The Wall Street
Journal, December 16, 2003). In 1999, the following usages were reported.
In-Store Purchase Percentage
Credit card 22
Debit card 21
Personal check 18
Cash 39
A sample taken in 2003 found that for 220 in-stores purchases, 46 used a credit card, 67 used
a debit card, 33 used a personal check, and 74 used cash.
a. At α = 0.01, can we conclude that a change occurred in how customers paid for in-
store purchases over the four-year period from 1999 to 2003? What is the p-value?
b. Compute the percentage of use for each method of payment using the 2003 sample
data. What appears to have been the major change or changes over the four-year
period?
c. In 2003, what percentage of payments was made using plastic (credit card or debit
card)?

134
58. The Wall Street Journal’s Shareholder Scoreboard tracks the performance of 1000
major U.S. companies (The Wall Street Journal, March 10, 2003). The performance of
each company is rated based on the annual total return, including stock price changes
and the reinvestment of dividends. Ratings are assigned by dividing all 1000 companies
into five groups from A (top 20%), B (next 20%), to E (bottom 20%). Shown here are
the one-year ratings for a sample of 60 of the largest companies. Do the largest
companies differ in performance from the performance of the 1000 companies in the
Shareholder Scoreboard? Use α = 0.05.
A B C D E
5 8 15 20 12

59. How well do airline companies serve their customers? A study showed the following
customer ratings: 3% excellent, 28% good, 45% fair, and 24% poor (BusinessWeek,
September 11, 2000). In a follow-up study of service by telephone companies, assume
that a sample of 400 adults found the following customer ratings: 24 excellent, 124
good, 172 fair, and 80 poor. Is the distribution of the customer ratings for telephone
companies different from the distribution of customer ratings for airline companies?
Test with α = 0.01. What is your conclusion?

Problems on Chi-Square for Independence of Attributes

60. The following 2 X 3 contingency table contains observed frequencies for a sample of
200. Test for independence of the row and column variables using the χ2 test with α =
0.05.
Row Variable A B C
P 20 44 50
Q 30 26 30

61. Visa Card USA studied how frequently consumers of various age groups use plastic
cards (debit and credit cards) when making purchases (Associated Press, January 16,
2006). Sample data for 300 customers shows the use of plastic cards by four age groups.
Age Group
Payment 18–24 25–34 35–44 45 and over
Plastic 21 27 27 36
Cash or check 21 36 42 90

135
 a. Test for the independence between method of payment and age group. What is the
p-value? Using α = 0.05, what is your conclusion?
b. If method of payment and age group are not independent, what observation can you
make about how different age groups use plastic to make purchases?
c. What implications does this study have for companies such as Visa, MasterCard,
and Discover?

62. With double-digit annual percentage increases in the cost of health insurance, more and
more workers are likely to lack health insurance coverage (USA Today, January 23,
2004). The following sample data provide a comparison of workers with and without
health insurance coverage for small, medium, and large companies. For the purposes of
this study, small companies are companies that have fewer than 100 employees.
Medium companies have 100 to 999 employees, and large companies have 1000 or
more employees. Sample data are reported for 50 employees of small companies, 75
employees of medium companies, and 100 employees of large companies.
Health Insurance
Size of Company Yes No Total
Small 36 14 50
Medium 65 10 75
Large 88 12 100
a. Conduct a test of independence to determine whether employee health insurance
coverage is independent of the size of the company. Use α = 0.05. What is the p-
value, and what is your conclusion?
b. The USA Today article indicated employees of small companies are more likely to
lack health insurance coverage. Use percentages based on the preceding data to
support this conclusion.

63. The National Sleep Foundation used a survey to determine whether hours of sleeping
per night are independent of age (Newsweek, January 19, 2004). The following show
the hours of sleep on weeknights for a sample of individuals age 49 and younger and
for a sample of individuals age 50 and older.

Hours of sleep
Age Fewer than 6 6 to 6.9 7 to 7.9 8 or more Total
49 or younger 38 60 77 65 240
50 or older 36 57 75 92 260

136
 a. Conduct a test of independence to determine whether the hours of sleep on
weeknights are independent of age. Use α = 0.05. What is the p-value, and
what is your conclusion?
b. What is your estimate of the percentage of people who sleep fewer than 6
hours, 6 to 6.9 hours, 7 to 7.9 hours, and 8 or more hours on weeknights?

64. Samples taken in three cities, Anchorage, Atlanta, and Minneapolis, were used to learn
about the percentage of married couples with both the husband and the wife in the
workforce (USA Today, January 15, 2006). Analyze the following data to see whether
both the husband and wife being in the workforce is independent of location. Use a .05
level of significance. What is your conclusion? What is the overall estimate of the
percentage of married couples with both the husband and the wife in the workforce?
Location
In Workforce Anchorage Atlanta Minneapolis
Both 57 70 63
Only one 33 50 90

65. On a syndicated television show the two hosts often create the impression that they
strongly disagree about which movies are best. Each movie review is categorized as
Pro (“thumbs up”), Con (“thumbs down”), or Mixed. The results of 160 movie ratings
by the two hosts are shown here.
Host B
Host A Con Mixed Pro
Con 24 8 13
Mixed 8 13 11
Pro 10 9 64
Use the chi-square test of independence with a .01 level of significance to analyze the data.
What is your conclusion?

137
References

1. Levin D M, Krehbiel T C, Berenson M L and Viswanathan P K (2011), Business

Statistics – A First Course, 5/e, Pearson.
2. Sheldon M Ross (2006), Introductory Statistics, 2/e, Academic Press.
3. Gupta S C and Kapoor V K (2002), Fundamentals of Mathematical Statistics, 11/e, Sultan
Chand and Sons.
4. Gupta S C and Kapoor V K (2007), Fundamentals of Applied Statistics, 4/e, Sultan
Chand and Sons.
5. Spiegel M R and Stephens L J (2010), Schaum’s Outlines – Statistics, 4/e, Tata McGraw
Hill Education Private Limited.
6. Pillai R S N and Bagavathi V (2005), Statistics – Theory and Practice, S.Chand and
Company Limited.
7. Pillai R S N and Bagavathi V (2003), Practical Statistics, 2/e, S.Chand and Company
Limited.
8. Irwin Miller and Marylees Miller (2002), Mathematical Statistics, 6/e, Prentice-Hall of
India Pvt. Ltd.
9. Anderson D R, Sweney D J and Williams T A(2005), Statistics for Business and
Economics, 8/e, Thomson.

*******

138