5db83ef1f71e482 PDF

MANONMANIAM SUNDARANAR UNIVERSITY
DIRECTORATE OF DISTANCE & CONTINUING EDUCATION

TIRUNELVELI 627012, TAMIL NADU
B.Sc. MATHEMATICS - I YEAR
DJM1C - STATISTICS
(From the academic year 2016-17)
Most Student friendly University - Strive to Study and Learn to Excel
For more information visit: http://www.msuniv.ac.in

B.Sc. MATHEMATICS – I YEAR
DJM1C : STATISTICS
SYLLABUS
UNIT I : Correlation - Karl Pearson‟s coefficient of correlation, Lines of Regression -

Regression coefficient - Rank Correlation.
UNIT II : Probability - Definition - Applications of Addition and multiplication, theorems -

Conditional, Probability - Mathematical Expectations - Moment generating function - Special
Distributions, (Binomial Distributions, Poisson Distribution, Normal Distribution - Properties).
UNIT III : Association of Attributes - Coefficient of Association - Consistency - Time Series -

Definition - Components Of Time Series - Seasonal and cyclic variations.
UNIT IV : Sampling - Definition - Large samples. Small samples- Population with one samples
and population with two samples - Students - t - test - Applications - chi - square test and
goodness of fit - applications.
UNIT V : Index Numbers - Types of index numbers - Tests - Unit test commodity reversal test,
time reversal test, factor reversal test - Chain index numbers - cost of living index - Interpolation
- Finite differences operators - Newton‟s forward, backward interpolation formulae, Lagrange‟s
formula.
Books:
1. Statistics: S. Arumugam & others

2. Statistics: D.C.Sancheti & Kapoor
3. Statistics: Mangaladas & others
4. Statistics: T.Sankaranarayana & others
Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

1
UNIT I : CORRELATION AND REGRESSION
Correlation - Karl Pearson’s coefficient of correlation, Lines of Regression - Regression

coefficient - Rank Correlation.
CORRELATION AND REGRESSION

CORRELATION:
Definition:
Consider a set of bivariate data 𝑥𝑖 , 𝑦𝑖 ;i=1,2….n. If there is a change in one variable

corresponding to a change in the other variable we say that the variables are correlated.
If the two variable two variables deviate in the same direction the correlation is said to be
direct or positive.
Definition:
The covariance between x and y is defined by

𝑥 𝑖 −𝑥 𝑦 𝑖 −𝑦
Cov (x,y) = 𝑛
𝑐𝑜𝑣 (𝑥,𝑦)
Hence 𝛾𝑥𝑦 =
𝜍𝑥 𝜍𝑦
Example:
The heights and weights of five students are given below.
Height in c.m x
160 161 162 163 164
Weight in kgs y 50 53 54 56 57
𝐻𝑒𝑟𝑒 𝑥 = 162; 𝑦 = 54 ; 𝜍𝑥 = 2 𝑎𝑛𝑑 𝜍𝑦 = 6
Now 𝑥𝑖 − 𝑥 𝑦𝑖 − 𝑦 = −2 −4 + −1 −1 + 0 + 1 × 2 + 2 × 3 = 17
𝑥 𝑖 −𝑥 𝑦 𝑖 −𝑦 17 17× 12 17 ×3.46
∴ 𝛾𝑥𝑦 = = = = = 0.98
𝑛 𝜍𝑥 𝜍𝑦 5 2 6 60 60

2
Theorem:
𝑛 𝑥𝑖 𝑦𝑖 − 𝑥𝑖 𝑦𝑖
𝛾𝑥𝑦 = 1
𝑛 𝑥𝑖2 − 𝑥𝑖 2 2 𝑛 𝑦𝑖2 − 𝑦𝑖 2 1 2
Proof:
𝑥𝑖 − 𝑥 𝑦𝑖 − 𝑦
𝛾𝑥𝑦 = … … … … … … … … (1)
𝑛𝜍𝑥 𝜍𝑦
𝑥𝑖 − 𝑥 𝑦𝑖 − 𝑦 = 𝑥𝑖 𝑦𝑖 − 𝑥 𝑦𝑖 − 𝑦 𝑥𝑖 + 𝑛𝑥 𝑦
= 𝑥𝑖 𝑦𝑖 − 𝑥 𝑛𝑦 − 𝑦 𝑛𝑥 + 𝑛𝑥𝑦
= 𝑥𝑖 𝑦𝑖 − 𝑛𝑥𝑦
1
= 𝑥𝑖 𝑦𝑖 − (𝑛 ) 𝑥𝑖 𝑦𝑖
1
= 𝑛 [n 𝑥𝑖 𝑦𝑖 − 𝑥𝑖 𝑦𝑖 ] … … … … … … … . . (2)
Also,
1
𝜍𝑥2 = 𝑥𝑖 − 𝑥 2
𝑛
1
=𝑛 [ 𝑥𝑖2 − 2𝑥 𝑥𝑖 + 𝑛 𝑥 2 ]
1
=𝑛 [ 𝑥𝑖2 − 2𝑛 𝑥 2
+ 𝑛 𝑥 2]
1 1
=𝑛 𝑥𝑖2 − 𝑥𝑖 2
𝑛
1 2
= 𝑛 2 [𝑛 𝑥𝑖2 − ( 𝑥𝑖 )]
1 2 1 2
∴ 𝜍𝑥 = 𝑛 𝑥𝑖2 − 𝑥𝑖 … … … … … … … . (3)
𝑛

3
Similarly,
1 2 1 2
𝜍𝑦 = 𝑛 𝑦𝑖2 − 𝑦𝑖 … … … … … . (4)
𝑛
Substituting (2),( 3) and (4) in (1) we get the required result.
Theorem:
-1 ≤ 𝛾 ≤ 1
Proof:
𝛾𝑥𝑦 = 𝑛 𝜍𝑥 𝜍𝑦
1
𝑛
= 1 1
1 2 1 2
𝑥 𝑖 −𝑥 2 𝑦 𝑖 −𝑦 2
𝑛 𝑛
Let 𝑎𝑖 = 𝑥𝑖 − 𝑥 and 𝑏𝑖 = 𝑦𝑖 − 𝑦
2
2
𝑎𝑖 𝑏𝑖
∴ 𝛾𝑥𝑦 =
𝑎𝑖2 𝑏𝑖2
By Schwartz inequality we have,
2 2 2
𝑎𝑖 𝑏𝑖 ≤ 𝑎𝑖 𝑏𝑖
2
Hence 𝛾𝑥𝑦 ≤1
∴ │𝛾𝑥𝑦 │ ≤ 1
∴ −1 ≤ 𝛾 ≤ 1
Note: 1
If 𝛾 = 1 𝑡𝑕𝑒 𝑐𝑜𝑟𝑟𝑒𝑙𝑎𝑡𝑖𝑜𝑛 𝑖𝑠 𝑝𝑒𝑟𝑓𝑒𝑡 𝑎𝑛𝑑 𝑝𝑜𝑠𝑖𝑡𝑖𝑣𝑒.

4
Note: 2
If 𝛾 = −1 the correlation is perfect and negative.
Note: 3
If 𝛾 = 0 the variables are uncorrelated.
Note: 4
If the variables x and y are uncorrelated then cov (x,y) = 0
Problem: 1
Ten students obtained the following percentage of marks in the college internal
test (x) and in the final university examination (y). Find the correlation coefficient
between the marks of the two tests.
X 5 6 6 4 5 6 6 6 4 5
x 51 63 63 49 50 60 65 63 46 50
4 7 7 5 4 6 7 4 6 5
y 49 72 75 50 48 60 70 48 60 56
Solution:
Choosing the origin A = 63 for the variable x and B= 60 for y and taking 𝑢𝑖 =
𝑥𝑖 − 𝐴 and 𝑣𝑖 = 𝑦𝑖 − 𝐵.

5
We have the following table:
𝒙𝒊 𝒖𝒊 𝒚𝒊 𝒗𝒊 𝒖𝟐𝒊 𝒗𝟐𝒊 𝒖𝒊 𝒗𝒊
51 -12 49 -11 144 121 132
63 0 72 12 0 144 0
63 0 75 15 0 225 0
49 -14 50 -10 196 100 140
50 -13 48 -12 169 144 156
60 -3 60 0 9 0 0
65 2 70 10 4 100 20
63 0 48 -12 0 144 0
46 -17 60 0 289 0 0
50 -13 56 -4 169 16 52
Total -70 - -12 980 994 500
𝛾𝑥𝑦 = 𝛾𝑢𝑣
𝑛 𝑢 𝑖 𝑣𝑖 − 𝑢𝑖 𝑣𝑖
= 1 1
𝑛 𝑢 𝑖2 − 𝑢𝑖 2 2 𝑛 𝑣𝑖2 − 𝑣𝑖 2 2
10×500− −70 ×(−12)

= 1 1
10×980− −70 2 2 10×994− −12 2 2
4160
= 70×98.97
=0.6
Problem: 2
If x and y are two variable. Prove that the correlation coefficient between 𝑎𝑥 + 𝑏 and
𝑐𝑦 + 𝑑 is

6
𝑎𝑐
𝛾𝑎𝑥 +𝑏,𝑐𝑦 +𝑑 = 𝛾 𝑖𝑓 𝑎, 𝑐 ≠ 0.
│𝑎𝑐 │ 𝑥𝑦
Proof:
Let 𝑢 = 𝑎𝑥 + 𝑏and 𝑣 = 𝑐𝑦 + 𝑑
∴ 𝑢 = 𝑎𝑥 + 𝑏 𝑎𝑛𝑑 𝑣 = 𝑐𝑦 + 𝑑
1
𝜍𝑢2 = 𝑢−𝑢 2
𝑛
𝑎2
= 𝑥𝑖 − 𝑥 2
𝑛
=𝑎2 𝜍𝑥2
Similarly,
𝜍𝑣2 = 𝑐 2 𝜍𝑦2
Now,
𝑢 −𝑢 (𝑣−𝑣 )
𝛾𝑢𝑣 =
𝑛𝜍𝑢 𝜍𝑣
𝑎 𝑥−𝑥 𝑐(𝑦 −𝑦 )
= 𝑛 𝑎𝑐 𝜍𝑥 𝜍𝑦
𝑎𝑐
= 𝛾𝑥𝑦 .
𝑎𝑐
Problem: 3
A programmer while writing a program for correlation coefficient between

two variable x and y from 30 pairs of observations obtained the following results
𝑥 = 300; 𝑥 2 = 3718, 𝑦 = 210; 𝑦 2 = 2000; 𝑥𝑦 = 2100. At the time of
checking it was found that he had copied down two pairs 𝑥𝑖 , 𝑦𝑖 as (18, 20) and (12, 10)
instead of the correct values (10,15) and (20,15). Obtain the correct value of the
correlation coefficient.

7
Solution:
Corrected 𝑥 = 300 − 18 − 12 + 10 + 20
= 300
Corrected 𝑦 = 210 − 20 − 10 + 15 + 15
= 210
Corrected 𝑥 2 = 3718 − 182 − 122 + 102 + 202
= 3750
Corrected 𝑦 2 = 2000 − 202 − 102 + 152 + 152
= 1950
Corrected 𝑥𝑦 = 2100 − 18 × 20 − 12 × 100 + 10 × 15 + 20 × 15 = 2070
After correction the correlation coefficient is,
𝑛 𝑥𝑦 − 𝑥 𝑦
𝛾𝑥𝑦 =
𝑛 𝑥2 − 𝑥 2 1 2 𝑛 𝑦2 − 𝑦 2 1 2
30 × 2070 − 300 × 210

𝛾𝑥𝑦 = 1
1
30 × 3750 − 3002 2 [30 × 1950 − 2102 ]2
62100 −63000
= 1 1
112500 −9000 2 58500 −44100 2
−900
= 1
(22500 )2 (14400 )1/2
900
=-
150 × 120
1
=-
20
= - 0.05

8
Problem: 4
𝐼𝑓 𝑥, 𝑦 𝑎𝑛𝑑 𝑧 are uncorrelated variable each having same standard deviation

obtain the coefficient of correlation between 𝑥 + 𝑦 𝑎𝑛𝑑 𝑦 + 𝑧.
Solution:
𝐺𝑖𝑣𝑒𝑛 𝜍𝑥 = 𝜍𝑦 = 𝜍𝑧 = 𝜍 .
𝑋 𝑎𝑛𝑑 𝑦 are uncorrelated ⇒ 𝑥−𝑥 𝑦−𝑦 =0
𝑌 𝑎𝑛𝑑 𝑧 are uncorrelated ⇒ 𝑦−𝑦 𝑧−𝑧 =0
𝑍 𝑎𝑛𝑑 𝑥 are uncorrelated ⇒ 𝑧−𝑧 𝑥−𝑥 =0
Let 𝑢 = 𝑥 + 𝑦 𝑎𝑛𝑑 𝑣 = 𝑦 + 𝑧
𝑢 = 𝑥+𝑦 and 𝑣=𝑦+𝑧
Now ,
1
𝜍𝑢2 = (𝑢 − 𝑢)2
𝑛
1 2
=𝑛 𝑥 − 𝑥 + (𝑦 − 𝑦)
1
=𝑛 [ (𝑥 − 𝑥 )2 + (𝑦 − 𝑦)2 + 2 𝑥−𝑥 𝑦−𝑦 ]
= 𝜍𝑥2 + 𝜍𝑦2 [𝑠𝑖𝑛𝑐𝑒 𝑥 − 𝑥 𝑦 − 𝑦 = 0]
= 2𝜍 2
Similarly,
𝜍𝑣2 = 2𝜍 2
Now, 𝑢−𝑢 𝑣−𝑣 = 𝑥 − 𝑥 + 𝑦 − 𝑦 { 𝑦 − 𝑦 + 𝑧 − 𝑧 }]

2
= 𝑥−𝑥 𝑦−𝑦 + 𝑦−𝑦 + 𝑥−𝑥 𝑧−𝑧 +
𝑦−𝑦(𝑧−𝑧)
= 0+ n𝜍𝑦2 + 0 + 0
= n𝜍 2

9
𝑢 − 𝑢 (𝑣 − 𝑣)
𝛾𝑢𝑣 =
𝑛𝜍𝑢 𝜍𝑣
𝑛𝜍 2
= 𝑛 (2𝜍 2 )
= ½.
Problem: 5
Show that the variable 𝑢 = 𝑥𝑐𝑜𝑠𝛼 + 𝑦 𝑠𝑖𝑛 𝛼 𝑎𝑛𝑑 𝑣 = 𝑦𝑐𝑜𝑠𝛼 − 𝑥𝑠𝑖𝑛𝛼 𝑎𝑟𝑒 𝑢𝑛𝑐𝑜𝑟𝑟𝑒𝑙𝑎𝑡𝑒𝑑. If
1 2𝛾𝑥𝑦 𝜍 𝑥 𝜍 𝑦
𝛼 = 2 tan−1 𝜍𝑥 2 −𝜍𝑦 2
Solution:
𝑢𝑖= 𝑥𝑖 𝑐𝑜𝑠𝛼 + 𝑦𝑖 𝑠𝑖𝑛𝛼 𝑎𝑛𝑑 𝑣𝑖 = 𝑦𝑖 𝑐𝑜𝑠𝛼 − 𝑥𝑖 𝑠𝑖𝑛𝛼
𝑢 = 𝑥 𝑐𝑜𝑠𝛼 + 𝑦𝑠𝑖𝑛𝛼 𝑎𝑛𝑑 𝑣 = 𝑦𝑐𝑜𝑠𝛼 − 𝑥 𝑠𝑖𝑛𝛼
𝑢𝑖 – 𝑢 = (𝑥𝑖 − 𝑥 ) 𝑐𝑜𝑠𝛼 + 𝑦𝑖 − 𝑦 𝑠𝑖𝑛𝛼
The variable 𝑢𝑖 and 𝑣𝑖 are uncorrelated if 𝑢𝑖 − 𝑢 𝑣𝑖 − 𝑣 = 0
𝑥𝑖 − 𝑥 𝑐𝑜𝑠𝛼 + 𝑦𝑖 − 𝑦 𝑠𝑖𝑛𝛼 𝑦𝑖 − 𝑦 𝑐𝑜𝑠𝛼 − 𝑥𝑖 − 𝑥 𝑠𝑖𝑛𝛼 = 0
∴ (𝑥𝑖 − 𝑥) 𝑦𝑖 − 𝑦 cos2 𝛼 − 𝑥𝑖 − 𝑥 𝑦𝑖 − 𝑦 sin2 𝛼 − 𝑐𝑜𝑠𝛼𝑠𝑖𝑛𝛼[ 𝑥𝑖 − 𝑥 2

−
𝑦𝑖−𝑦2]=0
∴ 𝑛𝛾𝑥𝑦 𝜍𝑥 𝜍𝑦 (cos 2 𝛼 − sin2 𝛼) = 𝑛𝑐𝑜𝑠𝛼𝑠𝑖𝑛𝛼 𝜍𝑥2 − 𝜍𝑦2
1
∴ 𝛾𝑥𝑦 𝜍𝑥 𝜍𝑦 cos 2𝛼 = 𝑠𝑖𝑛2𝛼 𝜍𝑥2 − 𝜍𝑦2 .
2
2𝛾𝑥𝑦 𝜍𝑥 𝜍𝑦
∴ 𝑡𝑎𝑛2𝛼 =
𝜍𝑥2 − 𝜍𝑦2
1 2𝛾𝑥𝑦 𝜍𝑥 𝜍𝑦
∴ 𝛼= tan−1
2 𝜍𝑥2 − 𝜍𝑦2

10
Rank correlation:
Suppose that a group of n individuals are arranged in the order of merit or

efficiency with respect to some characteristics. Then the rank is a variable which takes
only the values 1,2,3…n. assuming that there is no tie.
1+2. …………….+𝑛 𝑛+1 1
Hence 𝑥= = and the variance is given by 𝜍𝑥2 = 𝑛2 − 1 .
𝑛 2 12
Theorem:
Rank correlation 𝜌 is given by
6 𝑥−𝑦 2
𝜌= 1- .
𝑛(𝑛 2 −1)
Proof:
Consider a collection of n individuals
let 𝑥𝑖 and 𝑦𝑖 be the ranks of the 𝑖 𝑡𝑕 individual in the two different rankings.
1 1
∴ 𝑥 = 2 𝑛 + 1 = 𝑦 and 𝜍𝑥2 = 12 𝑛2 − 1 = 𝜍𝑦2 .
2 2
Now, 𝑥−𝑦 = 𝑥−𝑥 − 𝑦−𝑦 (𝑠𝑖𝑛𝑐𝑒 𝑥 = 𝑦)
2 2
= 𝑥−𝑥 + 𝑦−𝑦 −2 𝑥 − 𝑥 (𝑦 − 𝑦)
= 𝑛𝜍𝑥2 + 𝑛𝜍𝑦2 − 2𝑛𝜌𝜍𝑥 𝜍𝑦
= 2n𝜍𝑥2 1 − 𝜌 (𝑠𝑖𝑛𝑐𝑒 𝜍𝑥2 = 𝜍𝑦2 )

1
= 6 𝑛 𝑛2 − 1 1 − 𝜌
6 𝑥−𝑦 2
1−𝜌= 𝑛(𝑛 2 −1)
6 (𝑥−𝑦)2
𝜌= 1- .
𝑛(𝑛 2 −1)
Problem: 6
Find the rank correlation coefficient between the height in cm and weight in
kg of 6 soldiers in Indian Army.

11
Height 165 167 166 170 169 172
Weight 61 60 63.5 63 61.5 64
Solution:
Height Rank Weight Rank x-y (𝒙 − 𝒚)𝟐

in in
height weight
x y
165 6 61 5 1 1
167 4 60 6 -2 4
166 5 63.5 2 3 9
170 2 63 3 -1 1
169 3 61.5 4 -1 1
172 1 64 1 0 0
Total - - - - -16
6 𝑥−𝑦 2 6×16
𝜌 =1− = 1- 6×35
𝑛(𝑛 2 −1)
= 1-0.457
= 0.543.
Problem: 7
From the following data of marks obtained by 10 students in physics and

chemistry. Calculate the rank correlation coefficient.
Physics 3 5 5 6 4 3 4 5 1 2
(P) 35 56 50 65 44 38 44 50 15 26
Chemistr 5 3 7 2 3 5 7 6 5 3
y (Q) 50 35 70 25 35 58 75 60 55 35
Solution:
We rank the marks of physics and chemistry and we have the following table.
P Rank in p Q Rank in Q x-y (𝒙 − 𝒚)𝟐

x y
35 8 50 6 2 4
56 2 35 8 -6 36

12
50 3.5 70 2 1.5 2.25
65 1 25 10 -9 81
44 5.5 35 8 -2.5 6.25
38 7 58 4 3 9
44 5.5 75 1 4.5 20.25
50 3.5 60 3 0.5 0.25
15 10 55 5 5 25
26 9 35 8 1 1
Total - - - - 185
We observe that in the values of x the marks 50 and 44 occurs twice. In the values
of 𝑦 the mark 35 occurs thrice.
2
Hence in the calculation of the rank correlation coefficient 𝑥−𝑦 is to be
corrected by adding the following correction factors
2(22 − 1) 2(22 − 1) 3(32 − 1)

+ + =3
12 12 12
2
After correction 𝑥−𝑦 = 188.
Now,
6 𝑥−𝑦 2
𝜌= 1- 𝑛(𝑛 2 −1)
6×188
= 1 - 10×99
1128
= 1- 990
= 1-1.139
= -0.139.
Problem: 8
Three judges assign the ranks to 8 entries in a beauty contest.
Judge Mr.x 1 2 3 4 5 6 7 8
Judge Mr.y 3 2 1 5 4 7 6 8
Judge Mr.y 1 2 3 4 5 7 8 6
Which pair of Judges has the nearest approach to common in beauty?

13
Solution:
Table for the rank correlation coefficients 𝜌𝑥𝑦 , 𝜌𝑦𝑧 , 𝜌𝑧𝑥
g y z
𝟐 𝟐 𝟐
x y Z x-y 𝒙−𝒚 y-z 𝒚−𝒛 z-x 𝒛−𝒙
- 2 0
1 3 1 -2 4 2 4 0 0
0 0 0
2 2 2 0 0 0 0 0 0
y - -
3 1 3 3 9 -2 4 -1 1
- 1 1
4 5 4 -2 4 1 1 1 1
3 - -
5 4 5 3 9 -1 1 -2 4
- 0 2
6 7 7 -1 1 0 0 1 1
- - 3
7 6 8 -1 1 -2 4 3 9
0 2 -
8 8 6 0 0 2 4 -2 4
- - -
Total - 28 - 18 - 20
6 𝑋−𝑦 2
𝜌𝑥𝑦 = 1- 𝑛 (𝑛 2 −1)
6× 28
= 1- 8× (82 −1)
168
= 1- 504
=1-0.333= 0.667.
6 ×18
𝜌𝑦𝑧 = 1- 8×63
108
= 1- 504
= 1-o.214
= 0.786.
6×20
𝜌𝑧𝑥 = 1- 8×63

14
120
= 1- 504
= 1-0.238
= 0.762.
Since 𝜌𝑦𝑧 is greater than 𝜌𝑥𝑦 and𝜌𝑥𝑧 the judges Mr. Y and Mr. Z have nearest
approach to common taste in beauty.
Problem: 9
The coefficient of rank correlations of marks obtained by 10 students in

mathematics and physics was found to be 0.8. It was later discovered that the differences
in rank in two subjects obtained by one of the students was wrongly taken as 5 instead of
8. Find the correct coefficient of rank correlation.
Solution:
6 (𝑥−𝑦 )2
𝜌𝑥𝑦 = 1- .
𝑛 𝑛 2 −1
Given 𝜌𝑥𝑦 = 0.8 and n=10
6 𝑥−𝑦 2
0.8 = 1- 10(10 2 −1)
6 𝑥−𝑦 2
= 1- 990
6 𝑥 −𝑦 2
= 1-0.8 = 0.2
990
6 (𝑥 − 𝑦)2 = 990 × 0.2
= 198
2
∴ 𝑥−𝑦 = 33
2
Corrected 𝑥−𝑦 = 33 − 52 + 82 = 72
6×72
Now, after correction 𝜌𝑥𝑦 = 1- 10(102 −1)
432
= 1- 990
= 1-0.486

15
= 0.564
The correct coefficient of rank correlation is 0.546.
Exercises:
1. Ten students got the following percentage of marks in two subjects.
Economics 7 6 3 9 2 7 8 9 6 3
78 65 36 98 25 75 82 90 62 39
Statistics 8 5 5 9 6 6 6 8 5 4
84 53 51 91 60 68 62 86 58 47
2. The following table shows how 10 students were ranked according to their
achievements in the laboratory and lecture portions of a biology course. Find the coefficient
of rank correlation.
Laboratory 8 3 9 2 7 1 4 6 1 5
8 3 9 2 7 10 4 6 1 5
Lecture 9 5 1 1 8 7 3 4 2 6
9 5 10 1 8 7 3 4 2 6
REGRESSION:
If there is a functional relationship between the two variable 𝑥𝑖 and 𝑦𝑖 the

points in the scatter diagram will cluster around some curve called a line of regression. If the
curve is a straight line it is called a line of regression between the two variables.
Definition:
It we fit a straight line by the principle of least squares to the points of the
scatter diagram in such a way that the sum of the squares of the distance parallel to the y-
axis from the points to the line is minimized we obtain a line of best fit for the data and its is
called the regression line of y and x.
Similarly we can define the regression line of 𝑥 𝑜𝑛 𝑦.
Theorem:
𝜍
The equation of the regression line of 𝑦 𝑜𝑛 𝑥 is given by 𝑦 − 𝑦 = 𝛾 𝜍𝑥 (𝑥 − 𝑥)
𝑦

16
Proof:
Let 𝑦 = 𝑎𝑥 + 𝑏 be the line of regression of on 𝑥.
According to the principle of least square the constants a and b are to be

determined in such a way that S= 𝑦𝑖 − 𝑎𝑥𝑖 + 𝑏 2 𝑖𝑠 𝑚𝑖𝑛𝑖𝑚𝑢𝑚
𝜕𝑠
= 0 ⇒ −2 [ 𝑦𝑖 − 𝑎𝑥𝑖 + 𝑏 𝑥𝑖 = 0
𝜕𝑎
⇒ 𝑥𝑖 𝑦𝑖 = 𝑎 𝑥𝑖2 + 𝑏 𝑥𝑖 ………(1)
𝜕𝑠
= 0 ⇒ −2 [ 𝑦𝑖 − 𝑎𝑥𝑖 + 𝑏 =0
𝜕𝑏
⇒ 𝑦𝑖 = 𝑎 𝑥𝑖 + 𝑛𝑏 … … … … … (2)
Equations( 1) and( 2) are called normal equations.
From (2) we obtain 𝑦 = 𝑎𝑥 + 𝑏………(3)
∴The line of regression passes through the point (𝑥, 𝑦).
Now, shifting the origin to this point (𝑥, 𝑦) by means of the transformation
𝑋𝑖 = 𝑥𝑖 – 𝑥 and 𝑌𝑖 = 𝑦𝑖 − 𝑦.
We obtain 𝑥𝑖 = 0 = 𝑦𝑖 and the equation of the line of regression becomes
𝑦 = 𝑎𝑥 ….………(4)
Corresponding to this 𝑙𝑖𝑛𝑒 𝑦 = 𝑎𝑥 the constant 𝑎 can be determined from the normal
equation .
a 𝑋𝑖2 = 𝑥𝑖 𝑦𝑖
𝑋 𝑖 𝑌𝑖
a= 𝑋𝑖2
𝑥 𝑖 −𝑥 (𝑦 𝑖 −𝑦 )
= 𝑥 𝑖 −𝑥 2
𝛾𝜍𝑥 𝜍𝑦
= 𝜍𝑥2
𝜍
= 𝛾 𝜍𝑥
𝑦
𝜍𝑦
The required regression line (4) becomes 𝑌 = 𝛾 𝜍 𝑋
𝑥

17
𝜍𝑦
∴ 𝑦 − 𝑦 = 𝛾 𝜍 (𝑥 − 𝑥).
𝑥
Definition:
The scope of the regression line of y on 𝑥 is called the regression coefficient of 𝑦 on

𝑥 and it is denoted by 𝑏𝑦𝑥 .
𝜍𝑦
Hence 𝑏𝑦𝑥 = 𝛾 𝜍
𝑥
The regression coefficient of 𝑥 on 𝑦 is given by

𝜍𝑥
𝑏𝑥𝑦 = 𝛾 .
𝜍𝑦
Theorem:
Arithmetic mean of the regression coefficients is greater than or equal to the

correlation coefficient.
Proof:
Let𝑏𝑥𝑦 and 𝑏𝑦𝑥 be the regression coefficients.
we have to prove ½ (𝑏𝑥𝑦 + 𝑏𝑦𝑥 ) ≥ 𝛾
Now, ½ (𝑏𝑥𝑦 + 𝑏𝑦𝑥 ) ) ≥ 𝛾
⇔ 𝑏𝑦𝑥 + 𝑏𝑥𝑦 ≥ 2𝛾
𝜍𝑦 𝜍𝑥
⇔𝛾 + 𝛾 ≥ 2𝛾
𝜍𝑥 𝜍𝑦
⇔ 𝜍𝑥2 + 𝜍𝑦2 − 2𝜍𝑥 𝜍𝑦 ≥ 0

2
𝜍𝑥 − 𝜍𝑦 ≥ 0.
This is always true.
Hence the theorem.
Theorem:
Regression coefficient are independent of the change of origin but dependent on

change of scale.

18
Proof:
𝑥 𝑖 −𝐴 𝑦 𝑖 −𝐵
Let 𝑢𝑖 = and 𝑣𝑖 =
𝑕 𝑘
Let 𝑥𝑖 = 𝐴 + 𝑕𝑢𝑖 and 𝑦𝑖 = 𝐵 + 𝑘𝑣𝑖
We know that, 𝜍𝑥 = 𝑕𝜍𝑢 , 𝜍𝑦 = 𝑘𝜍𝑣 𝑎𝑛𝑑 𝛾𝑥𝑦 = 𝛾𝑢𝑣

𝜍𝑦
Now 𝑏𝑦𝑥 = 𝛾𝑥𝑦 𝜍𝑥
𝑘𝜍𝑣
= 𝛾𝑢𝑣 𝑕𝜍𝑢
𝑘
= 𝑕 𝑏𝑢𝑣 … … … … … … . 1
𝑕
similarly 𝑏𝑥𝑦 = ( 𝑘 ) 𝑏𝑢𝑣 ……………(2)
From (1) and (2) ⇒ 𝑏𝑦𝑥 and 𝑏𝑥𝑦 depend upon the scales 𝑕 and 𝑘, but not on the
origins A and B .
Hence the theorem.
Theorem:
1−𝛾 2 𝜍𝑥 𝜍𝑦
The angle between two regression line is given by 𝜃 = tan−1 𝛾 𝜍𝑥2 +𝜍𝑦2
Proof:
The equations of lines of regression of y on x and x on y respectively are

𝜍𝑦
𝑦−𝑦 =𝛾𝜍 𝑥 − 𝑥 ………………………….(1)
𝑥
𝜍𝑥
𝑥−𝑥 = 𝛾 𝑦 − 𝑦 … … … … . (2)
𝜍𝑦
(2) can also be written as

1 𝜍𝑦
𝑦−𝑦= (𝑥 − 𝑥) … … … ….(3)
𝛾 𝜍𝑥
𝜍𝑦 𝜍𝑦
Slopes of the two lines (1) and (2) are 𝛾 and𝛾𝜍 .
𝑥 𝑥
Let 𝜃 be the acute angle between the two lines of regression.

19
𝜍𝑦 𝜍𝑦
𝛾 𝜍 − 𝛾𝜍
𝑥 𝑥
∴ 𝑡𝑎𝑛𝜃 = 𝜍𝑦 𝑦
1+ 𝛾𝜍
𝑥 𝛾𝜍𝑥
𝛾 2 −1 𝜍𝑥 𝜍𝑦
= 𝛾 𝜍𝑥2 +𝜍𝑦2
= 𝛾 𝜍𝑥2 +𝜍𝑦2
(since 𝛾 2 ≤ 1 𝑎𝑛𝑑 𝜃 𝑖𝑠 𝑎𝑐𝑢𝑡𝑒).
1 − 𝛾2 𝜍𝑥 𝜍𝑦
∴ 𝜃 = tan−1
𝛾 𝜍𝑥2 + 𝜍𝑦2
Problem: 10
The following data relate to the marks of 10 students in the internal test and the
university examination for the maximum of 50 in each.
Internal 2 2 3 3 3 3 3 3 4 4
marks 25 28 30 32 35 36 38 39 42 45
University 2 2 2 3 2 1 2 3 3 3
marks 20 26 29 30 25 18 26 35 35 46
i) Obtain the two regression equations and determine.
ii) The most likely internal mark for the university mark of 25.
iii) the most likely university mark for the internal mark of 30.
Solution:
(i) Let the marks of internal test and university examination be denoted by x and
y respectively.
1 1
We have 𝑥 = 𝑥𝑖 = 35 𝑎𝑛𝑑 𝑦 = 𝑦𝑖 = 29.
10 10
For the calculation of regression we have the following table.

20
𝟐 𝟐
𝒙𝒊 𝒙𝒊 -35 𝒙𝒊 − 𝟑𝟓 𝒚𝒊 𝒚𝒊 -29 𝒚𝒊 − 𝟐𝟗 (𝒙𝒊 -35)(𝒚𝒊 − 𝟐𝟗)
25 -10 100 2 -9 81 90
0
28 -7 49 2 -3 9 21
6
30 -5 25 2 0 0 0
9
32 -3 9 3 1 1 -3
0
35 0 0 2 -4 16 0
5
36 1 1 1 -11 121 -11
8
38 3 9 2 -3 9 -9
6
39 4 16 3 6 36 24
5
42 7 49 3 6 36 42
5
45 10 100 4 17 289 170
6
Tot 0 358 - 0 598 324
al
𝑥 𝑖 −𝑥 2 1
𝜍𝑥2 = = 𝑥𝑖 − 35 2
= 35.8
𝑛 10
𝑦 𝑖 −𝑦 2 1
𝜍𝑦2 = = 10 𝑦𝑖 − 29 2
= 59.8
𝑛
𝜍𝑥 = 5.98 𝑎𝑛𝑑 𝜍𝑦 = 7.73
𝑥𝑖 − 𝑥 (𝑦𝑖 − 𝑦)
∴𝛾=
𝑛𝜍𝑥 𝜍𝑦
324
= 10×5.98 × 7.73
324
=
462.254
= 0.7 (approximately)
𝜍𝑦
Now the regression of y on x is 𝑦 − 𝑦 = 𝛾 𝜍 (𝑥 − 𝑥)
𝑥
𝜍𝑦 𝑥 𝑖 −𝑥 (𝑦 𝑖 −𝑦 )
∴𝛾 =
𝜍𝑥 𝑛𝜍𝑥2

21
𝑥 𝑖 −𝑥 (𝑦 𝑖 −𝑦 )
= 𝑥 𝑖 −𝑥 2
324
= 358 = 0.905
𝜍 𝑥 𝑖 −𝑥 (𝑦 𝑖 −𝑦 ) 324
Similarly, 𝛾 𝜍𝑥 = = 598 = 0.542
𝑦 𝑦 𝑖 −𝑦 2
The regression line 𝑜𝑓 𝑦 𝑜𝑛 𝑥 is 𝑦 = 29 = 0.905 (𝑥 − 35)
(ie), 𝑦 = 0. 905 𝑥 − 2.675 ……..(1)
The regression line 𝑜𝑓 𝑥 𝑜𝑛 𝑦 𝑖𝑠 𝑥 − 35 = 0.542 (𝑦 − 29)
(ie), 𝑥 = 0.542𝑦 + 19.282…………..(2)
(1) and (2) are the required regression equations.
ii) the most likely internal mark for the university mark of 25 is got from the regression
equation 𝑜𝑓 𝑥 𝑜𝑛 𝑦 by putting 𝑦 = 25
(2)⇒ 𝑥 = 0.542 × 25 + 19.282 = 32.83
iii) The most likely university mark for the internal mark of 30 is got from the regression
equation 𝑜𝑓 𝑦 𝑜𝑛 𝑥 by putting 𝑥 = 30
(1)⇒ 𝑦 = 0.905 × 30 − 2.675 = 24.475
Problem: 11
The two variable x and y have the regression lines 3x+2y-26 = 0 and 6x+y-
31=0. Find
i)The mean values of x and y

ii)The correlation coefficient between x and y
iii)The variance of y if the variance of x is 25
Solution:
(i) Since the two lines of regression pass through (𝑥,𝑦)
we have 3𝑥+𝑦=26……(1)
6𝑥+𝑦=31……(2)
Solving( 1) and( 2) we get 𝑥 = 4 𝑎𝑛𝑑 𝑦 = 67

22
−3
(ii) As in the previous problem we can prove that 𝑦 = 𝑥 +
2
−1 31
13 𝑎𝑛𝑑 𝑥 = 𝑦+ represent the regression lines of y on x and x on y by
6 6
respectively.
−3
Hence we get the regression coefficients 𝑎𝑠 𝑏𝑦𝑥 = 2
𝑎𝑛𝑑 𝑏𝑥𝑦 = − 1/6
Now,
−3 −1
𝛾2 = × (6)=¼
2
1
𝛾= ±
2
1
Since both the regression coefficients are negative we take 𝛾 = − 2
iii) Given 𝜍𝑥 = 5
𝜍𝑦
We have 𝑏𝑦𝑥 = 𝛾 𝜍𝑥
−3 −1 𝜍𝑦
∴ =
2 2 5
𝜍𝑦 = 15 .
Problem: 12
If 𝜃 is the acute angle between the two regression lines.

Show that 𝜃 ≤ 1 − 𝛾2
Solution:
We know that if 𝜃 is the acute angle between the two regression on lines
we have,
tan 𝜃 = ………..(1)
𝛾 𝜍𝑥2 +𝜍𝑦2
We claim that 𝜍𝑥2 + 𝜍𝑦2 ≥ 2𝜍𝑥 𝜍𝑦 .

23
Suppose not, then 𝜍𝑥2 + 𝜍𝑦2 < 2𝜍𝑥 𝜍𝑦
(i.e.) 𝜍𝑥2 + 𝜍𝑦2 − 2𝜍𝑥 𝜍𝑦 < 0

2
𝜍𝑥 − 𝜍𝑦 < 0.This is impossible.
Hence 𝜍𝑥2 + 𝜍𝑦2 ≥ 2𝜍𝑥 𝜍𝑦

𝜍𝑥 𝜍𝑦 1
∴ 𝜍 2 +𝜍 2 ≤ 2.
𝑥 𝑦
1−𝛾 2 1
(1)⇒ tan 𝜃 ≤ 𝛾 2
1 − 𝛾2
∴ tan 𝜃 ≤
2𝛾
1−𝛾 2
Hence sin 𝜃 ≤ 1+𝛾 2
sin 𝜃 ≤ 1 − 𝛾 2 .
Exercise:
1.calculate the coefficient of correlation of correlation and obtain the lines of regression
for the following data.
X 1 2 3 4 5 6 7 8 9
Y 9 8 10 12 11 13 14 16 15
Correlation coefficient for a bivariate frequency distribution:
cos ⁡
(𝑥,𝑦)
The correlation coefficient between x and y is given by 𝛾𝑥𝑦 = 𝜍𝑥 𝜍𝑦
𝑛 𝑚 1 𝑛 𝑚
𝑖=1 𝑗 =1 𝑓 𝑖𝑗 𝑥 𝑖 𝑦 𝑗 −𝑁 𝑖=1 𝑔 𝑖 𝑥 𝑖 𝑗 =1 𝑓 𝑗 𝑦 𝑗
∴ 𝛾𝑥𝑦 = 2
𝑛 𝑔 𝑥2−1 𝑛 𝑔 𝑥 2× 𝑚 𝑓 𝑦 2−1 𝑚 𝑓 𝑦
𝑖=1 𝑖 𝑖 𝑁 𝑖=1 𝑖 𝑖 𝑗 =1 𝑗 𝑗 𝑁 𝑗 =1 𝑗 𝑗
Note: Since correlation coefficient is independent of origin and scale if x and y are
𝑥−𝐴 𝑦 −𝐵
transformed to u and v by the formula u = 𝑕
and v = 𝑘
then we 𝑕𝑎𝑣𝑒 𝛾𝑥𝑦 = 𝛾𝑢𝑣 .

24
Problem: 13
Find the correlation coefficient between x and y from the following table:
x 5 10 15 20
y
4 2 4 5 4
6 5 3 6 2
8 3 8 2 3
Solution:
𝑿 x1 x2 x3 x4 Total
𝒀 5 10 15 20
y1 4 2 4 5 4 F1=15
y2 6 5 3 6 2 F2=16
y3 8 3 8 2 3 F3=16
Total g1=10 g2=15 g3=13 g4=9 N=47
Correlation coefficient between x and y is given by

𝑖=1 𝑗 =1 𝑓 𝑖𝑗 𝑥 𝑖 𝑦 𝑗 −𝑁 𝑖=1 𝑔 𝑖 𝑥 𝑖 𝑗 =1 𝑓 𝑗 𝑦 𝑗
∴ 𝛾𝑥𝑦 = 2
𝑛 𝑔 𝑥2−1 𝑛 𝑔 𝑥 2× 𝑚 𝑓 𝑦 2−1 𝑚 𝑓 𝑦
𝑖=1 𝑖 𝑖 𝑁 𝑖=1 𝑖 𝑖 𝑗 =1 𝑗 𝑗 𝑁 𝑗 =1 𝑗 𝑗
Where i=1,2,3,4 and j=1,2,3.
𝑔𝑖 𝑥𝑖 = 50 + 150 + 195 + 180 = 575
𝑓𝑗 𝑦𝑗 = 60 + 96 + 128 = 284
𝑔𝑖 𝑥𝑖2 = 250 + 1500 + 2925 + 3600 = 8275
𝑓𝑗 𝑦𝑗2 = 240 + 576 + 1024 = 1840

25
𝑓𝑖𝑗 𝑥𝑖 𝑦𝑗 = 40 + 160 + 300 + 320 + 150 + 180 + 540 + 240 + 120 +
640+240+480=3410
1
3410 − (575×284)
47
𝛾𝑥𝑦 = 1 1
8275 − 575 2 × 1840 − 284 2
47 47
3410 × 47−(575×284)
=
8275 ×47−575 2 × 1840 ×47−284 2
160270 −163300
= 388925 −330625 × 86480 −80656
−3030 −3030
= = 241.5 ×76.3
58300 × 5824
−3030
= 18426 .5
= -0.16
Problem: 14
Find the correlation coefficient between the heights and weight of 100 students which
are distributed as follows.
Height in c.m Weight in Kgs Total

30-40 40-50 50-60 60-70 70-80
150-155 1 3 7 5 2 18
155-160 2 4 10 7 4 27
160-165 1 5 12 10 7 35
165-170 - 3 8 6 3 20
Total 4 15 37 28 16 100
Solution:
Let 𝑥𝑖 denote the mid value of the classes of weights and 𝑦𝑗 denote the mid value
of the classes of heights.

26
𝑥𝑖 − 55 𝑦𝑗 − 157.5
𝐿𝑒𝑡 𝑢𝑖 = 𝑎𝑛𝑑 𝑣𝑗 =
10 5
Then the 2 -way frequency table is given below.
𝑥𝑖 45
35 55 65 75 𝑓𝑗 𝑣𝑗 𝑓𝑗 𝑣𝑗 𝑓𝑗 𝑣𝑗2 𝑓𝑖𝑗 𝑢𝑖 𝑣𝑗
𝑦 𝑗
152.5 (2) (3) (0) (-5) (-4) (-4)

1 3 7 5 2 18 -1 -18 18
157.5 (0) (0) (0) (0) (0) (0)

2 -4 10 7 4 27 0 0 0
162.5 (- (-5) (0) (10) (14)
2) 35 1 35 35 (17)
5 12 10 7
1
167.5 (-6)- (0) (12) (12)
- 3 8 6 3 20 2 40 80 (18)
𝑔𝑖
4 15 37 28 16 10 - 57 133 (31)
0
𝑢𝑖
-2 -1 0 1 2 -
𝑔𝑖 𝑢𝑖
-8 -15 0 28 32 37
𝑔𝑖 𝑢𝑖2
16 15 0 28 64 12
3
𝑓𝑖𝑗 𝑢𝑖 𝑣𝑗
(0) (-8) (0) (17) (22) (3
1)

27
𝑖=1 𝑗 =1 𝑓𝑖𝑗 𝑢𝑖 𝑣𝑗 −𝑁 𝑖=1 𝑔𝑖 𝑢𝑖 𝑗 =1 𝑓𝑗 𝑣𝑗
𝛾𝑥𝑦 = 𝛾𝑢𝑣 =
𝑛 2 1 𝑛 2 𝑚 2 1 𝑚 2
𝑖=1 𝑔𝑖 𝑢𝑖 −𝑁 𝑖=1 𝑔𝑖 𝑢𝑖 × 𝑗 =1 𝑓𝑗 𝑣𝑗 −𝑁 𝑗 =1 𝑓𝑗 𝑣𝑗
1
31 − 37 × 57
100
=
1 1
123 − 372 × 133 − 572
100 100
3100 − 37 × 57
=
12300 − 372 × 13300 − 572
991
=
104.5 × 100.25
= 0.09.

28
UNIT II: PROBABILITY
Probability - Definition - Applications of Addition and multiplication, theorems - Conditional,

Probability - Mathematical Expectations - Moment generating function - Special Distributions,
(Binomial Distributions, Poisson Distribution, Normal Distribution - Properties).
PROBABILITY
Introduction:
In this chapter we develop the mathematical theory of probability and introduce the
concept of random variables which form the basis for various types of theoretical distributions.
Definition:
An experiment is defined as an action which we conceive and do or intend to do.
Each experiment ends with an outcome. For example, a research student in “statistics”
when undertaking a pre election sample survey.
An experiment is called a random experiment if, when repeated under the same
conditions, it is such that the outcome cannot be predicted with certainty but all possible
outcomes can be determined prior to the performance of the experiment.
Each performance of the random experiment is called trail. The collection of all possible
outcomes of a random experiment is called the sample space S. The elements of sample space
are called sample points.
Example:
When two cons are tossed at a time the outcome is an ordered pair (H,H) or (H, T) or
(T,H) or (T,T). Hence for the random experiment of tossing two coins, sample space S={(H,H),
(H,T), (T,H), (T,T)}.
Definition:
Any subset A of a sample space S is called an event.

29
The event S is called a sure event and the event 𝜑 is called an impossible event.
Definition:
Let S be a sample space associated with an experiment. Let A be event suppose an

experiment is repeated N times and suppose the event A happens 𝑓 times. Then 𝑓/𝑁 is called
the relative frequency of the event A. clearly 0 ≤ 𝑓/𝑁 ≤ 1.
The following desirable properties to be satisfied by a probability function P.
a) P(A) ≥0 for all events

b) P(A) ≤ 1for all events
c) P(S) = 1
d) If A and B are disjoint events
P(AUB) = P(A) +P(B)
Example:
When two cons are tossed at a time the outcome is an ordered pair (H,H) or
(H, T) or (T,H) or (T,T). Hence for the random experiment of tossing two coins,
sample space S={(H,H), (H,T), (T,H), (T,T)}.
Definition:
Any subset A of a sample space S is called an event.
The event S is called a sure event and the event 𝜑 is called an impossible event.
Definition:
Let S be a sample space associated with an experiment. Let A be event suppose an

experiment is repeated N times and suppose the event A happens 𝑓 times. Then 𝑓/𝑁 is called
the relative frequency of the event A. clearly 0 ≤ 𝑓/𝑁 ≤ 1.
The following desirable properties to be satisfied by a probability function P.

30
a) P(A) ≤ 1for all events
b) P(S) = 1
c) If A and B are disjoint events
P(AUB) = P(A) +P(B)
Proof:
Let 𝐴 ⊆ 𝐵.
Then B=A∪(𝐴 ∩B).A and 𝐴 ∩ 𝐵 disjoint events
of S.
Hence 𝑃(𝐵) = 𝑃(𝐴) + 𝑃(𝐴 ∩ 𝐵)
But 𝑃(𝐴 ∩ 𝐵) ≥ 𝑃(𝐴)
Hence 𝑃 𝐵 ≥ (𝑃𝐴)
Corollary:
For each A⊂S, 0≤P(A) ≤1.
Proof:
∅ ⊂A⊆S
Hence P(∅)≤P(A) ≤P(S)
Hence 0≤P(A) ≤1.
Theorem:
If A and B are any two events of a sample space S then
P(AUB)=P(A) + P(B) –P(A∩B).

31
Proof:
𝐴 ∩ 𝐵 = 𝐴 ∪ (𝐴 ∩ 𝐵) and A and 𝐴 ∩ 𝐵 are disjoint sets.
∴ P(A∪B) = P(A) + P (𝐴 ∩ 𝐵) ……………..(1)
Now, B = (A∩B) ∪(𝐴 ∩B) and A∩B and 𝐴 ∩B are disjoint sets.
∴ P(B) = P(A∩B) + P(𝐴 ∩B)
∴ P(𝐴 ∩B) = P(B) - P(A∩B)………………(2)
(1)⇒P(A∪B) = P(A) + P(B) –P(A∩B) [(2) in (1)]
Example:
Let S = {(H,H), (H,T) (T,H),(T,T)}.
Let us assign the probability of ¼ to each element of the sample space S.
Let A={(H,H),(H,T)}; B={(H,H), (T,H)}
A∪B = { (H,H) (H,T), (T,H)}
A∩B = {(H,H)}
1 1 3
P(A) = 2 , P(B) = 2, P(A∪B) = 4
1
And P(A∩ 𝐵) = 4
We have, P(A∪ 𝐵) = P(A) + P(B) – P(A∩B)

1 1 1
=2+2- 4
2+2−1 3
= =4
4
Hence it is verified.

32
Example:
Let S= {(i,j) /i,j ∈N, 1≤i≤ 6, 1 ≤j≤6} be the sample space of the random experiment of
throwing two dice. We assign the uniform probably of 1/36 to each of the 36 sample
points in the sample space S.
Let A denote the event of getting 1 in the second die.
Then A = {(1,1) (2,1), (3,1), (4,1), (5,1) (6,1) }

6 1
P(A) = =
36 6
Let B = {(1,2), (2,2) (3,2)}

3 1
Then P(B) = 36 = 12
A∩B = ∅, P(A∩ 𝐵) = 0
P(A∪ 𝐵) = 𝑃 𝐴 + 𝑃 𝐵 − 𝑃(𝐴 ∩ 𝐵)
6 3 9 1
= 36 + 36 = 36 = 4
Definition:
Let S denote a sample space associated with an experiment. Let

𝐴1 , 𝐴2 , 𝐴3 … . 𝐴𝑛,….. be a sequence of Subsets of S.
If 𝐴𝑖 ∩ 𝐴𝑗 = ∅ for all i, j with i ≠ j then the sequence of subsets is said to be mutually

disjoint.
∞
If 𝑛=1 𝐴𝑛 then the sequence of events is said to be exhaustive.

33
Example:
Let S= {(i,j) / i, j ∈ N, 1≤ 𝑖 ≤ 6, 1 ≤ 𝑗 ≤ 6}
Let A be an event of getting the sum i + j and odd number as B be an event of

getting the sum as an even number.
Clearly A∩B = ∅ and A∪B = S Hence A and B are mutually exclusive and exchaustive
events.
Theorem:
For any two events A and B, P (𝐴 ∩ 𝐵) = 𝑃 𝐵 − 𝑃 (𝐴 ∩ 𝐵)
Proof:
Let𝐴 ∩ 𝐵 and 𝐴 ∩ 𝐵 are disjoint events and
(𝐴 ∩ 𝐵) ∪ 𝐴 ∩ 𝐵 = 𝐵
P(B) = P [(A∩ 𝐵) ∪ (𝐴 ∩ 𝐵)]
P(B) = P(A∩ 𝐵) + 𝑃(𝐴 ∩ 𝐵)
∴ P (𝐴 ∩B) = P (B) – P (A∩ 𝐵)
Remark:
Similarly we shall get P (A∩ 𝐵) = P(A) – P (A∩ 𝐵)
Theorem:
If B⊂A, then (i) P(A∩ 𝐵 ) = 𝑃 𝐴 − 𝑃(𝐵)
(ii) P(B) ≤P(A)

34
Proof:
i).When B⊂A, B and (A∩ 𝐵 ) are mutually exclusive events and their union
is A.
∴ P(A) = P[B∪ (𝐴 ∩ 𝐵)]
P(A) = P(B) + P(A∩ 𝐵 )
∴ P(A∩ 𝐵 ) = P(A) – P(B).
ii) Using axiom (i)
P(A∩ 𝐵 )≥ 0 ⇒ 𝑃 𝐴 − 𝑃(𝐵) ≥ 0.
𝑃 𝐵 ≤𝑃 𝐴 .
Corollary:
If (A∩B) ⊂ A and (A∩B) ⊂(B) then𝑃(𝐴 ∩ 𝐵) ≤ 𝑃(𝐴) and 𝑃(𝐴 ∩ 𝐵) ≤ 𝑃(𝐵)
Law of addition of probabilities:
Statement:
If A and B are any two events (subsets of a sample space S) and are not disjoint
then, P(A∪B) = P(A) + P(B) – P(A∩B)
Proof:
We have, A∪ 𝐵 = 𝐴 ∪ 𝐴 ∩ 𝐵
Since A and (𝐴 ∩ 𝐵) 𝑎𝑟𝑒 𝑑𝑖𝑠𝑗𝑜𝑖𝑛𝑡.

35
P(A∪ 𝐵) = 𝑃 𝐴 + 𝑃(𝐴 ∩ 𝐵)
=P(A) + [P(𝐴 ∩B) + P(A∩ 𝐵) − 𝑃 (𝐴 ∩ 𝐵)]
= P(A) + P[(𝐴 ∩ 𝐵) ∪ (𝐴 ∩ 𝐵)] − 𝑃 (𝐴 ∩ 𝐵)
[∴ (𝐴 ∩ 𝐵) 𝑎𝑛𝑑 𝐴 ∩ 𝐵 𝑎𝑟𝑒 𝑑𝑖𝑠𝑗𝑜𝑖𝑛𝑡]
∴ 𝑃 𝐴 ∪ 𝐵 = P(A) + P(B) – P(A∩ 𝐵)
Theorem: (Extension of General law of addition of probabilities]
For n events 𝐴1 , 𝐴2 … … … , 𝐴𝑛 , wehave

∞ 𝑛
P( 𝑖=1 𝐴𝑖 ) = 𝑖=1 𝑝 𝐴𝑖 − 𝑃 𝐴𝑖 ∩ 𝐴𝑗 + ⋯ . .
𝑛−1
+ −1 𝑃(𝐴1 ∩ 𝐴2 ∩ … .∩ 𝐴𝑛 )for all 1 ≤ 𝑖 ≤ 𝑗 ≤ 𝑛.
Proof:
For two events 𝐴1 𝑎𝑛𝑑 𝐴2
We have,
𝑃 (𝐴1 ∪ 𝐴2 ) = 𝑃(𝐴1 ) + (𝑃(𝐴2 ) – 𝑃(𝐴1 ∩ 𝐴2 )
It is true for n=2
Suppose that, it is true for n=r

𝑟 𝑟
Then, P ( 𝑖=1 𝐴𝑖 ) = 𝑖=1 𝑃 𝐴𝑖 − 𝑃 𝐴𝑖 ∩ 𝐴𝑗 + ⋯ .
𝑟−1
+ −1 𝑃 𝐴1 ∩ 𝐴2 ∩ … .∩ 𝐴𝑟
for all 1 ≤ 𝑖 ≤ 𝑗 ≤ 𝑟.

36
Now ,
𝑟+1 𝑟
P( 𝑖=1 𝐴𝑖 ) =P 𝑖=1 𝐴𝑖 ∪ 𝐴𝑟+1
𝑟 𝑟
=P 𝑖=1 𝐴𝑖 + 𝑃 𝐴𝑟+1 − 𝑃 𝑖=1 𝐴𝑖 ∩ 𝐴𝑟+1
𝑟 𝑟
=P 𝑖=1 𝐴𝑖 + 𝑃 𝐴𝑟+1 − 𝑃 𝑖=1 𝐴𝑖 ∩ 𝐴𝑟 +1
(Distributive Law)
𝑟 𝑟−1
= 𝑖=1 𝑃 𝐴𝑖 − 𝑃 𝐴𝑖 ∩ 𝐴𝑗 +…+ −1 P(𝐴1 ∩ 𝐴2 ∩ … ∩ 𝐴𝑟 ) + 𝑃 𝐴𝑟+1 −
𝑟
𝑃 𝑖=1 𝐴𝑖 ∩ 𝐴𝑟+1
For all 1≤ 𝑖 ≤ 𝑗 ≤ 𝑟
𝑟+1 𝑟−1
= 𝑖=1 𝑃 𝐴𝑖 - 𝑃 𝐴𝑖 ∩ 𝐴𝑗 + ⋯ + −1 P
𝑟
(𝐴1 ∩ 𝐴2 ∩ … ∩ 𝐴𝑟 ) − 𝑖=1 𝑃 𝐴𝑖 ∩ 𝐴𝑟+1 − 𝑃 𝐴𝑖 ∩ 𝐴𝑗 ∩ 𝐴𝑟+1
+ −1 𝑟−1 P (𝐴1 ∩ 𝐴2 ∩ … ∩ 𝐴𝑟 ∩ 𝐴𝑟+1 )]
𝑟+1 𝑟
= 𝑖=1 𝑃 𝐴𝑖 − 𝑃 𝐴𝑖 ∩ 𝐴𝑗 + 𝑖=1 𝑃 𝐴𝑖 ∩ 𝐴𝑟+1 + −1 𝑟 P (𝐴1 ∩ 𝐴2 ∩
… ∩ 𝐴𝑟 ∩ 𝐴𝑟+1 )
𝑟+1
= 𝑖=1 𝑃 𝐴𝑖 − 𝑃 𝐴𝑖 ∩ 𝐴𝑗 + ⋯ + −1 𝑟 P (𝐴1 ∩ 𝐴2 ∩ … ∩ 𝐴𝑟 ∩ 𝐴𝑟+1 )
For all 1≤ 𝑖 ≤ 𝑗 ≤ 𝑟 + 1
Hence it is true for n=r and it is also true for n=r+1.
Hence by the principle of mathematical induction, it is true for all positive integral values of n.
Theorem: [Boole’s inequality]
For n events 𝐴1 , 𝐴2 , … 𝐴𝑛 we have

𝑛 𝑛
a) P ( 𝑖=1 𝐴𝑖 ) ≥ 𝑖=1 𝑃 𝐴𝑖 − 𝑛 − 1
𝑛 𝑛
b) P ( 𝑖=1 𝐴𝑖 )≤ 𝑖=1 𝑃(𝐴𝑖 )

37
Proof:
The result is now prove by mathematical introduction
P(𝐴1 ∪ 𝐴2 ) = P(𝐴1 ) + P(𝐴2 ) – P (𝐴1 ∩ 𝐴2 ) ≤1
P(𝐴1 ∪ 𝐴2 ) ≥P(𝐴1 ) + P((𝐴2 ) – 1
Hence it is true for n=2
Suppose that it is true for n=r such that

𝑟 𝑟
P( 𝑖=1 𝐴𝑖 )≥ 𝑖=1 𝑃 𝐴𝑖 − 𝑟 − 1
𝑟+1 𝑟
Then P ( 𝑖=1 𝐴𝑖 ) =P( 𝑖=1 𝐴𝑖 ∩ 𝐴𝑟+1 )
𝑟
≥ P( 𝑖=1 𝐴𝑖 )+P(𝐴𝑟+1 )-1
≥ 𝑃 𝐴𝑖 − 𝑟 − 1 + 𝑃 𝐴𝑟+1 − 1
𝑖=1
𝑟+1 𝑟+1
⇒P( 𝑖=1 𝐴𝑖) ≥ 𝑖=1 𝑃 𝐴𝑖 − 𝑟
∴ It is true for n=r+1 also
b) Let 𝐴1 𝐴2 ……..𝐴𝑛 be the events
We get
P(𝐴1 ∩ 𝐴2 ….∩ 𝐴n) ≥[P(𝐴1 ) + P(𝐴2 ) + ……..+P (𝐴𝑛 )] – (n-1)
= [1-P(𝐴1 ) + [1-P(𝐴2 )] + …….+ [1-P(𝐴𝑛 )] – (n-1)
= 1-P(𝐴1 ) – P(𝐴2 ) - ……-P (𝐴𝑛 )
⇒ P(𝐴1 ) + P(𝐴2 ) +…..P(𝐴𝑛 )≥1-P (𝐴1 ∩ 𝐴2 ∩…..∩ 𝐴𝑛 )
= 1-P(𝐴1 ∪ 𝐴2 ∪ … ∪ 𝐴𝑛 )
= P(𝐴1 ∪ 𝐴2 ∪ … ∪ 𝐴𝑛 )
⇒ P(𝐴1 ∪ 𝐴2 ∪ … ∪ 𝐴𝑛 )≤P(𝐴1 ) + P(𝐴2 ) +…..P(𝐴𝑛 )
Theorem:
For n events 𝐴1 , 𝐴2 , … 𝐴𝑛.

38
𝑛 𝑛
we have P[ 𝑖=1 𝐴𝑖 ] ≥ 𝑖=1 𝑃 𝐴𝑖 − 𝑃 𝐴𝑖 ∩ 𝐴𝑗 , 𝑓𝑜𝑟 𝑎𝑙𝑙 1 ≤ 𝑖 ≤ 𝑗 ≤ 𝑛
Proof:
We shall prove this theorem by the method of induction.
We have,
P(𝐴1 +𝐴2 +𝐴3 ) = P(𝐴1 ) +P(𝐴2 ) + P(𝐴3 )- [P(𝐴1 ∩ 𝐴2 ) + P(𝐴2 ∩ 𝐴3 ) + P(𝐴2 ∩ 𝐴1 )]

+ P(𝐴1 ∩ 𝐴2 ∩ 𝐴3 )
3 3
P( 𝑖=1) ≥ 𝑖=1 𝑃 𝐴𝑖 − 1≤𝑖≤𝑗 ≤3 𝑃 𝐴𝑖 ∩ 𝐴𝑗
Thus the result is true for n=3
Suppose that the result is true for n=r

𝑟
Then P( 𝑖=1 𝐴𝑖 ) ≥ 𝑖=𝑟 𝑃 𝐴𝑖 - 1≤𝑖≤𝑗 ≤𝑟 𝑃 𝐴𝑖 ∩ 𝐴𝑗
Now,
𝑟+1 𝑟
P( 𝑖=1 𝐴𝑖 ) =P[ 𝑖=1 𝐴𝑖 ∪ 𝐴𝑟+1 ]
𝑟 𝑟
= P( 𝑖=1 𝐴𝑖 + 𝑃 𝐴𝑟+1 − 𝑃 [( 𝑖=1 𝐴𝑖 ) ∩ 𝐴𝑟+1 )]
𝑟 𝑟
= P( 𝑖=1 𝐴𝑖 + 𝑃 𝐴𝑟+1 − 𝑃 [( 𝑖=1(𝐴𝑖 ∩ 𝐴𝑟+1 )]
𝑟 𝑟
≥ 𝑖=1 𝑃 𝐴𝑖 − 1≤𝑖≤𝑗 <𝑟 𝑃 𝐴𝑖 ∩ 𝐴𝑗 + 𝑃 𝐴𝑟+1 − 𝑃 𝑖=1 𝐴𝑖 ∩ 𝐴𝑟+1
From Boole‟s inequality we get

𝑟 𝑟
P[ 𝑖=1(𝐴𝑖 ∩ 𝐴𝑟+1 )] ≤ 𝑖=1 𝑃 𝐴𝑖 ∩ 𝐴𝑟+1
𝑟 𝑟
⇒ -P [ 𝑖=1(𝐴𝑖 ∩ 𝐴𝑟+1 )] ≥ − 𝑖=1 𝑃 𝐴𝑖 ∩ 𝐴𝑟+1
𝑟+1 𝑟+1 𝑟
∴ 𝑃( 𝑖=1 𝐴𝑖 )≥ 𝑖=1 𝑃 𝐴𝑖 − 1≤𝑖≤𝑗 ≤𝑟 𝑃 𝐴𝑖 ∩ 𝐴𝑗 − 𝑖=1 𝑃 𝐴𝑖 ∩ 𝐴𝑟+1
𝑟+1 𝑟+1
𝑃( 𝑖=1 𝐴𝑖 )≥ 𝑖=1 𝑃 𝐴𝑖 − 1≤𝑖≤𝑗 ≤𝑟+1 𝑃 𝐴𝑖 ∩ 𝐴𝑗
Hence, if the theorem is true for n=r, it is also true for n=r+1
Hence by mathematical induction,
the result is true for all positive integral values of n.

39
Multiplication law of probability and conditional probability:
Theorem:
For two events A and B
P(A∩B) = P(A). P(B│A), P(A)>0
= P(B).P(A│B), P(B)>0
Where P(B│A) represents the 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.
Proof:
Suppose the sample space contains N occurrences of which 𝑛𝐴 occurrences belong to the
event A and 𝑛𝐵 occurrences belong to the event B.
Let 𝑛𝐴𝐵 be the number of occurrences favorable to the compound event A∩B then, the
unconditional probabilities are given by
𝑛𝐴 𝑛𝐵 𝑛 𝐴𝐵
P(A) = , P(B) = and P(𝐴 ∩ 𝐵) =
𝑁 𝑁 𝑁
Now, the conditional probability P(A│B) refers to the sample space of 𝑛𝐵 occurances,
out of which 𝑛𝐴𝐵 occurrences pertain to the occurrence of A, when B has already
happened.
𝑛 𝐴𝐵
P(A│B) = 𝑛𝐵
𝑛 𝐴𝐵
Similarly P(B│A) = 𝑛𝐴
n AB
Now, P(A ∩ 𝐵) = 𝑁
𝑛 𝐴𝐵 nA
= .
𝑛𝐴 𝑁
= P(B│A) P(A)
n AB 𝑛 nB
And P(A∩B) = = 𝑛𝐴𝐵 .
𝑁 𝐵 𝑁
= P(A│B) P(B)

40
𝑃(𝐴∩𝐵) 𝑃(𝐴∩𝐵)
∴ P(B│A) = and P(A│B) =
𝑃(𝐴) 𝑃(𝐵)
Thus the conditional probabilities P(B│A) and P(A│B) are defined iff P(A) ≠0
and P(B) ≠0 respectively.
Extension of multiplication law of probability :
Theorem:
For n events 𝐴1 , 𝐴2 , … 𝐴𝑛 . we have P(𝐴1 ∩ 𝐴2 ∩ … ∩ 𝐴𝑛 )= P(𝐴1 ) P(𝐴2 │𝐴1 )

P(A3│A1 ∩ A2)…P(An│A1 ∩A2 ∩……∩ An-1) …..
Where,
P(Ai│Aj ∩ Ak∩…….∩Ai) represents the conditional probability of the event Ai,

given that the events Aj, Ak,…Ai have already happened.
Proof:
We have for three events A1, A2, and A3
P(A1 ∩ A2 ∩ 𝐴3 ) = 𝑃(A1 ∩ 𝐴2 ∩ 𝐴3 )
= P(A1) P(A2∩A3│A1)
= P(A1) P(A2│A1)P(A3│A1 ∩ 𝐴2 )
It is true for n=2 and n=3
Suppose that it is true for n=k.
So that,
P(A1∩ 𝐴2 ∩ … .∩ 𝐴𝑘 ) = P(A1 ) P(A2│A1)P(A3│A1 ∩𝐴2 ) …
𝑃(𝐴𝑘 │𝐴1 ∩ 𝐴2 ∩. . .∩ 𝐴𝑘−1 )
Now,
P(𝐴1 ∩ 𝐴2 ∩ … ∩ 𝐴𝑘 ∩ 𝐴𝑘+1 )= P(A1∩ 𝐴2 ∩ … .∩ 𝐴𝑘 ) 𝑃(𝐴𝑘+1 │𝐴1 ∩ 𝐴2 ∩ … ∩ 𝐴𝑘 )
=P(A1) P(A2│A1)… 𝑃(𝐴𝑘 │𝐴1 ∩ 𝐴2 ∩. . .∩ 𝐴𝑘−1 × 𝑃(𝐴𝑘+1 │𝐴1 ∩ 𝐴2 ∩. . .∩ 𝐴𝑘 )
It is true for n=k+1 also

41
Since it is true for n=2 and n=3, by the principle of mathematical
induction, it is true for all integral values of n.
Theorem:
For any three events A,B,C;P(A∪ 𝐵│𝐶) = 𝑃(𝐴│𝐶) + 𝑃(𝐵│𝐶) − 𝑃(𝐴 ∩ 𝐵│𝐶)
Proof:
We have
P(A∪B) = P(A) + P(B) – P(A∩B)
⇒ P[(A∩C) ∪(B∩ 𝐶)] = 𝑃 𝐴 ∩ 𝐶 + 𝑃 𝐵 ∩ 𝐶 − 𝑃(𝐴 ∩ 𝐵 ∩ 𝐶)
Dividing both sides by P(C) we get

𝑃 𝐴∩𝐶 ∪(𝐵∩𝐶) 𝑃 𝐴∩𝐶 + 𝑃 𝐵∩𝐶 − 𝑃(𝐴∩𝐵∩𝐶)
= ,P(C) >0
𝑃(𝐶) 𝑃(𝐶)
𝑃(𝐴∩𝐶) 𝑃(𝐵∩𝐶) 𝑃(𝐴∩𝐵∩𝐶)

= + -
𝑃(𝐶) 𝑃(𝐶) 𝑃(𝐶)
𝑃[(𝐴∪𝐵)∩𝐶]
⇒ = P(A│C) + P(B│C) – P(A∩ 𝐵)│𝐶)
𝑃(𝐶)
⇒ P(A∪ 𝐵│𝐶) = 𝑃(𝐴│𝐶) + 𝑃(𝐵│𝐶) − 𝑃(𝐴 ∩ 𝐵│𝐶)
Theorem:
For any three events A, B and C P(A ∩ 𝐵 │𝐶) + P(A ∩ 𝐵│𝐶)=P(A│C)
Proof:
𝑃(𝐴∩𝐵 ∩𝐶) 𝑃(𝐴∩𝐵∩𝐶)
P(A∩ 𝐵 │𝐶) + 𝑃 𝐴 ∩ 𝐵│𝐶 = +
𝑃(𝐶) 𝑃(𝐶)
𝑃 𝐴∩𝐵 ∩𝐶 + 𝑃(𝐴∩𝐵∩𝐶)
= 𝑃(𝐶)
𝑃 (𝐴∩𝐶)
= = P(A│𝐶)
𝑃(𝐶)
Theorem:
For a fixed B with P(B)>0, P(A│𝐵) is a probability function

42
Proof:
𝑃(𝐴∩𝐵)
i. P(A│B) = ≥0
𝑃 𝐵
𝑃(𝑠∩𝐵) 𝑃(𝐵)
ii. A(S│B) = = 𝑃(𝐵) = 1
𝑃(𝐵)
iii. If {An} is any finite or infinite sequence of disjoint events, then
𝑃 𝑛 𝐴𝑛 ∩𝐵
P 𝑛 𝐴𝑛 │𝐵 = 𝑃(𝐵)
𝑃 𝑛 𝐴𝑛 𝐵
= 𝑃(𝐵)
𝑛 𝑃 (𝐴𝑛 𝐵) 𝑃 𝐴𝑛 𝐵
= = 𝑛
𝑃(𝐵) 𝑃 𝐵
= 𝑛 𝑃 𝐴𝑛 │𝐵
Hence the theorem
Theorem:
For any three events A,B and C defined on the sample space S such that B⊂C and
P(A>0),P(B│A) ≤P(C│A)
Proof:
𝑃(𝐶∩𝐴)
P(C│A) = 𝑃(𝐴)
𝑃[ 𝐵∩𝐶∩𝐴 ∪ 𝐵 ∩𝐶∩𝐴 ]
= 𝑃(𝐴)
𝑃(𝐵∩𝐶∩𝐴) 𝑃(𝐵 ∩𝐶∩𝐴)

= +
𝑃(𝐴) 𝑃(𝐴)
=P(B∩ 𝐶│𝐴)+ P( 𝐵 ∩ 𝐶│𝐴)
Now, B⊂C ⇒ B∩C=B
∴ P(C│A) = P (B│A) + P(𝐵 ∩ C│A)
⇒ P(C│A)≥P(B│A)
Theorem:
If A and B are independent events then A and 𝐵 are also independent events

43
Proof:
WE have P(A ∩ 𝐵 ) = 𝑃 𝐴 − (𝐴 ∩ 𝐵)
= P(A) – P(A) P(B) [ A and B are independent]
= P(A) [1-P(B)]
=P(A) P(B)
⇒ A and 𝐵 are independent events
Theorem:
If A and B are independent events then 𝐴 and 𝐵 are also independent events
Proof:
We are given P(A∩ 𝐵) = 𝑃 𝐴 𝑃(𝐵)
Now P(𝐴 ∩ 𝐵 ) = 𝑃(𝐴 ∪ 𝐵 )
=1-P(A∪ 𝐵)
= 1-[P(A) + P(B) – P(A∩ 𝐵)]
=1-[P(A) + P(B) – P(A) P(B)]
=1-P(A)-P(B) + P(A) P(B)
=[1-P(B)] –P(A) [1-P(B)]
= [1-P(B)] [1-P(A)]
= [1-P(A)] [1-P(B)]
= P(𝐴) P(𝐵)
∴ 𝐴 and 𝐵 are independent events
Theorem:
If A,B,C are mutually independent events when

A∪ 𝐵 𝑎𝑛𝑑 𝐶 𝑎𝑟𝑒 𝑎𝑙𝑠𝑜 𝑖𝑛𝑑𝑒𝑝𝑒𝑛𝑑𝑒𝑛𝑡.
Proof:
We are required to prove

44
P[(A∪ 𝐵) ∩ 𝐶] = 𝑃 𝐴 ∪ 𝐵 𝑃(𝐶)
L.H.S. = P[𝐴 ∩ 𝐶) ∪ (𝐵 ∩ 𝐶)] [ 𝐷𝑖𝑠𝑡𝑟𝑖𝑏𝑢𝑡𝑖𝑣𝑒 𝐿𝑎𝑤]
= P(A∩ 𝐶) + 𝑃 𝐵 ∩ 𝐶 − 𝑃(𝐴 ∩ 𝐵 ∩ 𝐶)
= P(A) P(C) + P(B) P(C) – P(A) P(B) P(C)
[A, B, and C mutually independent)
= P(C) [P(A) + P(B) – P(∩ 𝐵)]
= P(C) P(A∪ 𝐵) = 𝑅. 𝐻. 𝑆
Hence (A∪ 𝐵) 𝑎𝑛𝑑 𝐶 𝑎𝑟𝑒 𝑖𝑛𝑑𝑒𝑝𝑒𝑛𝑑𝑒𝑛𝑡.
Theorem:
Prove that if A,B and C are random events in a sample space and if A,B,C are pair wise
independent and A is independent of
(B∪ 𝐶), 𝑡𝑕𝑒𝑛 𝐴, 𝐵 𝑎𝑛𝑑 𝐶 𝑎𝑟𝑒 𝑚𝑢𝑡𝑢𝑎𝑙𝑙𝑦 𝑖𝑛𝑑𝑒𝑝𝑒𝑛𝑑𝑒𝑛𝑡
Proof:
We are given,
P(A∩ 𝐵) = 𝑃 𝐴 𝑃(𝐵)
𝑃 𝐵∩𝐶 = 𝑃 𝐵 𝑃 𝐶 (1)
𝑃 𝐴∩𝐶 = 𝑃 𝐴 𝑃 𝐶
P(A∩ (𝐵 ∪ 𝐶)] = 𝑃(𝐴) 𝑃(𝐵 ∪ 𝐶)
Now,
P[A∩ (𝐵 ∪ 𝐶)] = 𝑝[(𝐴 ∩ 𝐵) ∪ (𝐴 ∩ 𝐶)
= P(A∩ 𝐵) + 𝑃 𝐴 ∩ 𝐶 − 𝑃[ 𝐴 ∩ 𝐵 ∩ 𝐴 ∩ 𝐶 ]
= P(A) P(B) + P(A) P(C) – P(A∩ 𝐵𝐶)………………..(2)
And
P(A)P(B∪ 𝐶) = 𝑃 𝐴 [𝑃 𝐵 + 𝑃 𝐶 − 𝑃 𝐵 ∩ 𝐶 ]
= P(A) P(B) + P(A) P(C) – P(A) P(B∩ 𝐶)…………(3)

45
From (2) and (3) on using (1) we get
P(A∩ 𝐵 ∩ 𝐶) = 𝑃 𝐴 𝑃(𝐵 ∩ 𝐶)
=P(A) P(B) P(C)
Hence A, B, C are mutually independent
Theorem:
For any two events A and B, P(A∩ 𝐵) ≤ 𝑃 𝐴 ≤ 𝑃 𝐴 ∪ 𝐵 𝑃 𝐴 + 𝑃(𝐵)
Proof:
We have
A = (A∩ 𝐵 ) ∪ (𝐴 ∩ 𝐵)
We have P(A) = P{ (A∩ 𝐵 ) ∪ (𝐴 ∩ 𝐵)
= P(A∩ 𝐵 ) + 𝑃(𝐴 ∩ 𝐵)
But P 𝐴 ∩ 𝐵 ≥ 0
∴ 𝑃(𝐴) ≥ 𝑃(𝐴 ∩ 𝐵)
Similarly P(B) ≥ 𝑃 𝐴 ∩ 𝐵
⇒ P(B) – P(A∩ 𝐵) ≥= 𝑜
Now P(A∪ 𝐵) = 𝑃 𝐴 + [𝑃 𝐵 − 𝑃 𝐴 ∩ 𝐵 ]
P(A∪ 𝐵) ≥ 𝑃(𝐴)
⇒ 𝑃 𝐴 ≤ 𝑃(𝐴 ∪ 𝐵)
Also P (A∪ 𝐵) ≤ 𝑃 𝐴 + 𝑃 𝐵 From (2)
Hence from (1),(2) and (3) we get
P(A∩ 𝐵) ≤ 𝑃 𝐴 ≤ 𝑃 𝐴 ∪ 𝐵 ≤ 𝑃 𝐴 + 𝑃(𝐵)
Example:
Two dice, one green and the other red, are thrown. Let A be the event that the sum of the
points on the faces shown is odd and B the event of at least one ace (number „1‟)
a. Describe the

46
i) complete sample space.
ii) events A,B, 𝐵, A∩B, A∪B, and A∩ 𝐵 and find their probabilities assuming
all the 36 saple points have equal probabilities.
a. Find the probabilities of the events

i. (𝐴 ∪ 𝐵 )
ii. 𝐴∩𝐵
iii. 𝐴∩𝐵
iv. 𝐴∩𝐵
v. 𝐴∩𝐵
vi. 𝐴 ∪ 𝐵
vii. (𝐴 ∪ 𝐵 )
viii. 𝐴 ∩ 𝐴 ∪ 𝐵
ix. 𝐴 ∪ 𝐴 ∩ 𝐵
x. (A│B) and (B│A) and (A│𝐵) and (𝐵│𝐴)
Solution:
a. The sample space S, consists of the 36 elementary events

{(1,1); (1,2); (1,3); (1,4); (1,5); (1,6)
(2,1); (2,2); (2,3); (2,4); (2,5); (2,6)
(3,1); (3,2); (3,3); (3,4); (3,5); (3,6)
(4,1); (4,2); (4,3); (4,4); (4,5); (4,6)
(5,1); (5,2); (5,3); (3,4); (3,5); (3,6)
(4,1); (4,2); (4,3); (4,4); (4,5); (4,6)
(5,1); (5,2); (5,3); (5,4); (5,5); (5,6)
(6,1); (6,2); (6,3); (6,4); (6,5); (6,6)}
for example, the ordered pair (4,5) refers to the elementary event that the green
die shows 4 and the red die shows 5.
A= the event that the sum of the numbers shown by the two dice is odd.
= {(1,2); (2,1); (1,4); (2,3); (3,2); (4,1); (1,6); (2,5); (3,4); (4,3); (5,2); (6,1); (3,6);
(4,5); (5,4); (6,3); (5,6); (6,5)}
Therefore,
𝑛 (𝐴) 18
P(A) = 𝑛 (𝑆) = 36
B= the event that at least one face is 1

47
= {(1,1); (1,2); (1,3); (1,4); (1,5); (1,6); (2,1); (3,1) (4,1); (5,1); (6,1)}
therefore
𝑛 (𝐵)
P(B) = 𝑛 (𝑆)
11
= 36
𝐵 = the event that each of the face obtained is not an ace
= {(2,2); (2,3); (2,4); (2.5); (2,6); (3,2); (3,3); (3,4); (3,5); (3,6); (4,2); (4,3);
(4,4); (4,5); (4,6); (5,2); (5,3); (5,4); (5,5); (5,6); (6,2); (6,3); (6,4); (6,5); (6,6)}
therefore
𝑛 (𝐵 )
P(𝐵) = 𝑛 (𝑆)
25
=36
A∩B = the event that sum is odd and atleast one face is an ace.
= {(1,2); (2,1); (1,4); (4,1); (1,6); (6,1)}

𝑛(𝐴∩𝐵) 6 1
∴ 𝑃(𝐴 ∩ 𝐵) = = =
𝑛(𝑠) 36 6
A∪B = {(1,2); (2,1); (1,4); (2,3); (3,2); (4,1); (1,6), (2,5); (3,4); (4,3); (5,2); (6,1); (3,6);
(4,5); (5,4); (6,3); (5,6); (6,5); (1,1); (1,3); (1,5); (1,5) (3,1), (5,1)}
𝑛(𝐴∪𝐵) 23
∴ 𝑃(𝐴 ∪ 𝐵) = = 36
𝑛(𝑠)
A∩ 𝐵 = {(2,3); (2,5); (3,2); (3,4); (3,6); (4,1); (4,5); (5,2); (5,4) (5,6), (6,3) (6,5)}
𝑛(𝐴 ∩ 𝐵 )
𝑃 𝐴∩𝐵 =
𝑛(𝑆)
12
= 36
1
= 3
b. i. P (𝐴 ∪ 𝐵 ) = 𝑃 𝐴 ∩ 𝐵
= 1-P (𝐴 ∩ 𝐵)

48
1 5
= 1-6 = 6
ii) P(𝐴 ∩ 𝐵 ) = 𝑃(𝐴 ∪ 𝐵 )
= 1-P(A∪ 𝐵)
= 1- P(A) – P(B) + P(A∩ 𝐵)

18 11 6
=1- 36 - 36 + 36
13
= 36
iii. 𝑃 𝐴 ∩ 𝐵 = 𝑃 𝐴 − 𝑃(𝐴 ∩ 𝐵)
18 6
= 36 - 36
12
= 36
1
=3
iv) P(𝐴 ∩ 𝐵) = 𝑃 𝐵 − 𝑃(𝐴 ∩ 𝐵)

11 6
= 36 − 36
5
= 36
V) P(A∩ 𝐵 ) = 1 − 𝑃(𝐴 ∩ 𝐵)
1
= 1- 6
5
=6
vi) P(𝐴 ∪ 𝐵) = 𝑃 𝐴 + 𝑃 𝐵 − 𝑃(𝐴 ∩ 𝐵)

18 11 5
= (1-36 ) + - 36
36
2
= 3
vii) P( A ∪ B) = 1 − 𝑃(𝐴 ∪ 𝐵)
23 13
= 1-36 = 36

49
viii) P(𝐴 ∩ (𝐴 ∪ 𝐵)] = 𝑃[ 𝐴 ∩ 𝐴 ∪ (𝐴 ∩ 𝐵)
= P (𝐴 ∩ 𝐵) [ A∩ 𝐴 = ∅]
5
= 36
ix) P(A∪ (𝐴 ∩ 𝐵)] = 𝑃(𝐴) + 𝑃(𝐴 ∩ 𝐵) − 𝑃(𝐴 ∩ 𝐴 ∩ 𝐵)
= P(A) + P(𝐴 ∩ 𝐵)
18 5
= +
36 36
23
= 36
𝑃(𝐴∩𝐵)
x. P(A│B) = 𝑃(𝐵)
6/36
= 11/36
6
= 11
𝑃(𝐴∩𝐵)
P(B│A) = 𝑃(𝐴)
6/36
= 18/36
6
= 18
1
=3
𝑃(𝐴 ∩𝐵)
xi. P(𝐴│B) = 𝑃(𝐵)
13/36
= 25/36
13
=25
𝑃(𝐴 ∩𝐵 )
P (𝐵/𝐴) = 𝑃(𝐴 )

50
13/36 13
= 25/36 = 18
Example:
If two dice are thrown, what is the probability that the sum is a) greater than 8 and b) neither 7
nor 11?
Solution:
a.) If S denotes the sum on the two dice then we want P(S>8)
i. S=9, (ii) S=10, iii) S=11 iv) S=12
Hence by addition theorem of probability
P(S>8) = P(S=9) + P(S=10) + P(S=11) + P(S=12)
n(S) = 36
The number of favorable cases can be enumerated as follows
S=9: (3,6), (6,3), (4,5), (5,4)
i.e. 4 sample points

4
P(S = 9) = 36
S= 10; (4,6), (6,4), (5,5) i.e. 3 sample points

3
P(S=10) = 36
S=11: (5,6), (6,5) i.e. 2 sample points

2
P(S=11) = 36
S=12: (6,6) i.e. 1 sample point

1
P(S = 12) = 36
4 3 2 1 10 5
P(S>8) = 36 + 36 + 36 + 36 = 36 = 18
b. Let A denotes the event of getting the sum of 7 and B, the event of getting the sum of 1
with a pair of dice.

51
S=7; (1,6), (6,1), (2,5), (5,2), (3,4), (4,3)
ii. i.e. 6 distinct sample points

6 1
P(A) = P(S=7) = 36 = 6
S=11; (5,6), (6,5)

2 1
P(B) = P(S=11) = 36 = 18
Required probability = P(𝐴 ∩ 𝐵 )
= 1-P(A∪ 𝐵)
1
= 1- 𝑃 𝐴 + 𝑃 𝐵 (∴ 𝐴 𝑎𝑛𝑑 𝐵 𝑎𝑟𝑒 𝑑𝑖𝑠𝑗𝑜𝑖𝑛𝑡 𝑒𝑣𝑒𝑛𝑡𝑠) =1− −
6
1
18
7
= 9
Example:
An urn contains 4 tickets numbered 1,2,3,4 and another contains 6 tickets

numbered 2,4,,6,7,8,9. If one of the two urns is chosen at random and a ticket is drawn at
random from the chosen urn, find the probability that the ticket drawn bears of the
number
i. 2 or 4 (ii) 3 (iii) 1 or 9
Solution:
1. Required event can happen in the following mutually exclusive

ways,
I. First urn is chosen and then a ticket is drawn
II. Second urn is chosen and then a ticket is drawn
Since the probability of choosing any urn is ½ the required probability P is given by
P= P(I) + P(II)
1 2 1 2 5
=2 × + 2 × 6 = 12
4
1 1 1
ii) Required probability = 2 × 4 + 2 × 0

52
1
= 8
1 1 1 1 5
iii) Required probability = 2 × 4 + 2 × = 24
6
Example:
A card is drawn from a well – shuffled pack of playing cards. What is the
probability that it is either a spade or on ace.
Solution:
Let A and B denote the events of events drawing a spade card and an ace respectively. Then A
consists of 13 sample points and B consists of A sample points.
13 4
i.e. P (A) = 52 and P (B) = 52
The compound event A∩ 𝐵consists of only one sample point

1
i.e. P (A∩ 𝐵) = 52
The probability that the card drawn is either a spade or an ace is given by
P𝐴 ∪ 𝐵) = 𝑃 𝐴 + 𝑃 𝐵 − 𝑃(𝐴 ∩ 𝐵)
13 4 1
= 52 + - 52
52
4
= 13
Example:
A box contains 6 red, 4 white and 5 black balls. A person draws 4 balls from
that among the balls drawn there is at least one ball of each color
Solution:
The required event E that in a draw of 4 balls from the box at random there is at least
one ball of each color can materialize the following mutually disjoint ways.
i) 1 red, 1 white, 2black balls

ii) 2 red, 1 white, 1 black balls
iii) 1 red, 2 white, 1 black balls
Hence by the addition theorem of probability the required probability is given by,

53
P (E) = P(i) + P(ii) + P(iii)
6𝑐1 × 4𝑐1 × 5𝑐2 6𝑐2 × 4𝑐1 × 5𝑐1 6𝑐1 × 4𝑐2 × 5𝑐1
= + +
15𝑐4 15𝑐4 15𝑐4
1
= 15𝑐 [ 6×4×10+15×4×5+6×6×5]
4
4!
=15×14×13×12 [ 240+300+180]
24 × 720
= 15×14×13×12 = 0.5275
Example:
A problem is statistic is given to the three students A,B and C whose

chances of solving it are ½, -3/4, and ¼ respectively. What is the probability that the
problem will be solved?
Solution:
Let A ,B ,C denote the events that the problem is solved by the students A,B,C
respectively.
Then, P (A) = ½, P(B) = ¾ and P(C) = 1/4
𝑃 𝐴∪𝐵∪𝐶
= 𝑃 𝐴 + 𝑃 𝐵 + 𝑃 𝐶 − 𝑃 𝐴∩𝐵 − 𝑃 𝐴∩𝐶 − 𝑃 𝐵∩𝐶
+𝑃 𝐴∩𝐵∩𝐶
= 𝑃 𝐴 + 𝑃 𝐵 + 𝑃 𝐶 − 𝑃 𝐴 .𝑃 𝐵 − 𝑃 𝐴 .𝑃 𝐶 − 𝑃 𝐵 .𝑃 𝐶 +
𝑃 𝐴 . 𝑃 𝐵 . 𝑃(𝐶)
1 3 1 1 3 3 1 1 1 1 3 1
=2+ +4− .4 - 4 .4 − 2 . + 2.4 . 4
4 2 4
29
= 32
Example:
A bag contains 6 white 9 black balls. Four balls are drawn at a time. Find the
probability for the first draw to give 4 white and the second to give 4 black balls in each
of the following cases.
i. The balls are replaced before the second draw

54
ii. The balls are not replaced before the second draw
Solution:
1. The experiment of drawing 4 balls from a bag containing 6 white and 9 black
balls result in 15C4 ways and hence the sample space consist 15C4 sample points.
Let A be the event that the first drawing gives 4 white balls and B be the event that the
second drawing gives 4 black balls.
The event A consists of 6C4 sample points as there are 6 white balls and 4 are to be
chosen from them
6𝐶
P(A) = 15𝐶4
4
Now, if the drawn balls are not replaced our sample space is reduced to 11C4 points
only. The event B that the second draw results in 4 black balls.
𝐵 9𝐶4
P(𝐴 ) = 11𝐶4
Hence, P(A∩ 𝐵) = 𝑃 𝐴 × 𝑃(𝐵│𝐴)

6𝐶 9𝐶
=15𝐶4 × 11𝐶4
4 4
3
= 715
ii) consider the same experiment with replacement

6𝐶
P(A) = 15𝐶4
4
Whether A has occurred or not, the probability of drawing 4 black ball in the second
draw is 9𝐶4 │15𝐶4
P(A∩ 𝐵) = 𝑃 𝐴 × 𝑃(𝐵│𝐴)
= P(A).P(B), as B is independent of A
6𝐶 9𝐶 6
= 15𝐶4 × 15𝐶4 = 5926
4 4

55
Exercise:
1. A bag contains 6 balls of different colors and a ball is drawn from its. A speaks truth
thrice out of 4 times and B speaks truth 7 times out of times. If both A and B say that a
red ball was drawn, find the probability of their joint statement being true (Ans : 7/15)
2. A and B are two very weak students of statics and their chances of solving a problem
correctly are 1/8 and 1/12 respectively if the probability of their making a common
mistake is 1/1001 and they obtain the same answer, find the chance that their answer is
correct (Ans : 13/14)
3. A bag contains 10 balls, two of which are red three blue and 5 black. Three balls are
drawn at random from the bag, that is every ball has an equal chances of being included
is the three what is the probability that
i) The three balls are of different colors
ii) Two balls are of the same colors and
iii) The balls are all of the same color?

30 79 11
Ans: 120 ; 𝑖𝑖) 𝑖𝑖𝑖)
120 120
Mathematical expectation.
Definition:
Let x be a discrete random variable which can assume any of the values x1, x2,..xn with
corresponding probabilities Pi= P(x=xi) i=1,2,… Then the mathematical expectation of
x, denoted by E(x) is defined by
E(x) = 𝑖 𝑃𝑖 𝑋𝑖 provided the series is absolutely convergent.
Example:
1.Let X be the discrete random variable taking the values 1,2,…, 6 with corresponding
probabilities Pi = 1/6

56
6
Then E(x) = 𝑖=1 𝑃𝑖 𝑋𝑖
1 1 1
= 6 (1) + 6 2 + … … . + 6 6
1 7
=6 1+2+3+4+5+6 = 2
2. Let x be a random variable having the p.d.f
P(x) = { 𝑥
6
0
𝑖𝑓 𝑥 = 1,2,3
𝑜𝑡𝑕𝑒𝑟𝑤𝑖𝑠𝑒
Then E(x3+2x2) = E (x3) + 2 E (x2)

3 3 3 2
= 𝑖=1 𝑥 𝑃 𝑥𝑖 + 2 𝑖=1 𝑥 𝑃(𝑥𝑖 )
1 16 81 1 8 27
= (6 + 6 + ) + 2 (6 + + )
6 6 6
98 36
=6 + 2 × 6
170
= 6
85
= 3
Mathematical expectation of continuous random variable
Definition:
If x is a continuous random variable with p.d.f f (x) then

∞
mathematical expectation of x is defined to be E(X) = ∫−∞ 𝑥𝑓 𝑥 𝑑𝑥
Provided the integral is absolutely convergent.

∞
E (𝜓 𝑥 ) =∫−∞ 𝜓 𝑥 𝑓 𝑥 𝑑𝑥 𝑤𝑕𝑒𝑟𝑒 𝜓 𝑥 𝑖𝑠 𝑎 𝑓𝑢𝑛𝑐𝑡𝑖𝑜𝑛 𝑜𝑓 𝑟. 𝑣. 𝑋.
Example:
𝑥+2
Let x have a p.d.f f(x) = 𝑖𝑓 − 2 < 𝑥 < 4
18
0 otherwise
Solution:

57
4 𝑥+2
i. E(x) = ∫−2 𝑥( 18 ) 𝑑𝑥
1 𝑥3
= 18 [ 3 + 𝑥 2 ]4-2
1 64 −8
= 18 + 16 − + 4
3 3
1 108
= 18 [ ] =2
3
ii. E[(x+2)2] = E(x2+4x+4)
= E(x2) +4E(x) + 4
4 𝑥+2
= ∫−2 𝑥 2 ( 18 ) 𝑑𝑥 + 4𝐸 𝑥 + 4
1 𝑥4 2𝑥 3 4
= 18 [ 4 + ] + (4x2 + 4)
3 −2
1 128 16
= 18 [ 64 + − (4 − )]+12
3 3
1 320 4
= 18 [ + 3] + 12
3
= 18 + 12 = 30
Definition:
Let X be a r.v E(x) is called the mean value of x and is denoted by 𝜇.
Hence 𝑋 = 𝜇= E(X)
E(Xr), r≥1 is called the rth moment of X about the origin and is denoted by
𝜇𝑟1 .Hence 𝜇𝑟′ = 𝐸(𝑋 𝑟) and
𝑥 = 𝜇1′
E(X- 𝜇)2 is called the variance of X and is denoted by 𝜍2. The positive square root
𝜍 of the variance is called the standard deviation of X.
E(X- 𝜇)r is called the rth central moment of X and is denoted by 𝜇 r .
Hence 𝜇 r= E(X- 𝜇)r
Lemma:

58
2
𝜍 2 = E(X2) – [E(x)]2 = 𝜇2 − 𝜇1′
Proof:
𝜍 2 = E [(x − 𝜇)2 ]
= 𝐸(𝑥 2 − 2𝜇𝑥 + 𝜇 2 )
= 𝐸 𝑥 2 − 2𝜇𝐸 𝑥 + 𝜇 2
= 𝐸 𝑥 2 − 2[𝐸 𝑥 ]2 + [𝐸 𝑥 ]2
= 𝐸 𝑥 2 − [𝐸 𝑥 ]2
2
= 𝜇2′ − 𝜇1′
Lemma:
𝜇𝑟 = 𝜇𝑟′ − 𝑟𝑐1 𝜇𝜇𝑟+1 + 𝑟𝑐2 𝜇 2 𝜇𝑟−2 ……….
Proof:
𝜇𝑟 = 𝐸 (𝑋 − 𝜇)𝑟
=𝐸 𝑋 𝑟 − 𝑟𝑐1 𝜇𝑋 𝑟−1 + ⋯ . .
=𝜇𝑟′ − 𝑟𝑐1 𝜇𝜇𝑟−1 + 𝑟𝑐2 𝜇 2 𝜇𝑟−2 ……….
In particular, 𝜇1 = 𝐸 𝑋 − 𝜇 = 𝐸 𝑋 − 𝜇 = 𝜇 − 𝜇 = 0.
𝜇2 = 𝜇2′ − 2𝜇1′ 𝜇 + 𝜇 2 𝜇0
2
∴ 𝜇2 = 𝜇2′ − 𝜇1′ (Since 𝜇 = 𝜇1′ 𝑎𝑛𝑑𝜇0′ = 1)
𝜇3 = 𝜇3′ − 3𝜇2′ 𝜇1′ + 2 𝜇1′ 2
Problem:
A random variable X is defined as the sum of the numbers on the faces when
two dice are thrown. Find the expected value of X.
Solution:
The probability distribution of X is given by the following table.

59
𝑥𝑖 2 3 4 5 6 7 8 9 10 11 12
P(𝑥𝑖 ) 1 2 3 4 5 6 5 4 3 2 1
36 36 36 36 36 36 36 36 36 36 36
𝐸 𝑋 =
𝑥𝑖 𝑝(𝑥𝑖 )
=2(1/36)+3(2/36)+4(3/36)+5(4/36)+6(5/36)+7(6/36)+8(5/36)+9(4/36)+10(3/36)+11(2/3
6)+12(1/36)=252/36=7.
Problem:
A random variable X has the following probability functions.
𝑥𝑖 -2 -1 0 1 2 3
P(𝑥𝑖 ) 0.1 K 0.2 2k 0.3 k
Find (i) the value of k (ii) mean (iii)
variance (iv) 𝑝(𝑥 ≥ 2) (v) 𝑝(𝑥 < 2)
(vi) 𝑝(−1 < 𝑥 < 3).
Solution:
(i) 𝑝𝑖 = 1
⇒ 0.1+k+0.2+2k+0.3+k=1
⇒4k=0.4
⇒k=0.1.
𝑥𝑖 -2 -1 0 1 2 3
P(𝑥𝑖 ) 0.1 0.1 0.2 0.2 0.3 0.1
Hence the probability function is
(ii) Mean=𝐸 𝑋 = 𝑥𝑖 𝑝(𝑥𝑖 ).
=(-2)(0.1)+(-1)(0.1)+(0)(0.2)+(1)(0.2)+2(0.3)+3(0.1)
=0.8.
(iii) variance=𝐸 𝑋 2 − 𝐸(𝑋) 2

60
= 𝑥𝑖 2 𝑝(𝑥𝑖 )-0.82
=[4 0.1 + 1 0.1 + 1 0.2 + 4 0.3 + 9(0.1)
=2.8-0.64=2.16.
(iv) 𝑝 𝑥 ≥ 2 = 𝑝 𝑋 = 2 + 𝑝(= 3)=0.3+0.1=0.4.
(v) 𝑝 𝑥 < 2 = 1 − 𝑝 𝑥 ≥ 2 = 0.6.
(vi) 𝑝 −1 < 𝑥 < 3 = 𝑝 𝑋 = 0 + 𝑝 𝑋 = 1 + 𝑝(𝑋 = 2)
=0.2+0.2+0.3=0.7.
Problem:
𝑥+1
Let X have the p.d.f f(x)= , if -1<x<1
2
0, otherwise.
Find the mean and standard deviation of x.
Solution:
∞ 1 𝑥+1
𝜇 = 𝐸 𝑥 = ∫−∞ 𝑥𝑓 𝑥 𝑑𝑥 = ∫−1 𝑥 𝑑𝑥
2
1
𝑥3 𝑥2
=1/2 +
3 2 −1
𝟏
𝜇 =𝟑
𝜍 2 =E(X2)-[E(X)]2
1 𝑥2 1 2
= ∫−1 𝑥 + 1 dx-
2 3
1
𝑥4 𝑥3 1
=1/2 + −9
4 3 −1
2
𝜍2=
9

61
Moment Generating Function:
Definition:
The moment generating function (m.g.f) for any random variable X about the
origin is defined by 𝑀𝑋 𝑡 = 𝐸 𝑒 𝑡𝑥 =
𝑒 𝑡𝑥 𝑓 𝑥 𝑑𝑥 𝑖𝑓 𝑋 𝑖𝑠 𝑎 𝑐𝑜𝑛𝑡𝑖𝑛𝑢𝑜𝑠 𝑟. 𝑣 𝑤𝑖𝑡𝑕 𝑝. 𝑑. 𝑓 𝑓(𝑥)

𝑓 𝑥 =
𝑒 𝑡𝑥 𝑃( 𝑥), 𝑖𝑓 𝑋 𝑖𝑠 𝑎 𝑑𝑖𝑠𝑐𝑟𝑒𝑡𝑒 𝑟. 𝑣 𝑤𝑖𝑡𝑕 𝑝. 𝑑. 𝑓 𝑃(𝑥)
𝑥
Where the integration or summation is taken over the entire range of X and t is a real
parameter.
Definition:
More generally the m.g.f of a random X about a point a denoted by

𝑀𝑋=𝑎 𝑡 𝑖𝑠 𝑑𝑒𝑓𝑖𝑛𝑒𝑑 𝑏𝑦 𝑀𝑋=𝑎 𝑡 = 𝐸 𝑒 𝑡(𝑥−𝑎) .
Example :
6
, 𝑖𝑓 𝑥 = 1,2,3, …
Define P(x) = 𝜋 2𝑥 2
0, 𝑜𝑡𝑕𝑒𝑟𝑤𝑖𝑠𝑒
6 6 ∞ 1
Now, 𝑃 𝑥 = = 𝜋2 𝑛=1 𝑛 2
𝜋2𝑥2
6 𝜋2 1 1 𝜋2
=𝜋 2 × (Since 1 + 22 + 33 … += )=1
𝟔 6
∴ 𝑃 𝑥 𝑖𝑠 𝑎 𝑝𝑟𝑜𝑏𝑎𝑏𝑖𝑙𝑖𝑡𝑦 𝑑𝑖𝑠𝑡𝑟𝑖𝑏𝑢𝑡𝑖𝑜𝑛 𝑜𝑓 𝑡𝑕𝑒 𝑟. 𝑣 𝑋.
Now, the m.g.f of X, if it exists, is given by 𝑀𝑋 𝑡 = 𝐸 𝑒 𝑡𝑥 = 𝑒 𝑡𝑥 𝑃 𝑥

6
= 𝑒 𝑡𝑥
𝜋2𝑥2
6 ∞ 𝑒 𝑡𝑛
=𝜋 2 𝑛=1 𝑛 2

62
The above series, by ratio test ,diverges for t>0.
∴ 𝑀𝑜𝑚𝑒𝑛𝑡 𝑔𝑒𝑛𝑒𝑟𝑎𝑡𝑖𝑛𝑔 𝑓𝑢𝑛𝑐𝑡𝑖𝑜𝑛 𝑜𝑓 𝑋 𝑑𝑜𝑒𝑠 𝑛𝑜𝑡 𝑒𝑥𝑖𝑠𝑡.
Properties of moment generating function:
1. The rth derivative of m.g.f of r.v X at t=0 is 𝜇𝑟′ .
Proof:
Let 𝑀𝑋 𝑡 𝑏𝑒 𝑡𝑕𝑒 𝑚. 𝑔. 𝑓 𝑜𝑓 𝑡𝑕𝑒 𝑟. 𝑣 𝑋
𝜇 2′ 2 𝜇 𝑟′ 𝑟
Then 𝑀𝑋 𝑡 = 1 + 𝜇1′ + 𝑡 + ⋯+ 𝑡 +⋯
2! 𝑟!
𝑑𝑟 ′
𝜇 𝑟′ 𝑟! 𝑡𝜇 𝑟+1
∴ 𝑑𝑡 𝑟
(𝑀𝑋 𝑡 ) = 𝑟!
+(𝑟+1)!
+…
𝑑𝑟
At t=0, (𝑀𝑋 𝑡 )=𝜇𝑟′
𝑑𝑡 𝑟
Problem:
A random variable X has the probability function

1
P(x)=2𝑥 ; 𝑥 = 1,2,3 … 𝐹𝑖𝑛𝑑 𝑖𝑡𝑠 𝑖 𝑚. 𝑔. 𝑓
(ii) Mean and (iii) Variance.
Solution:
X is a discrete random variable
(i) (𝑀𝑋 𝑡 =E(𝑒 𝑡𝑥 ) = 𝑥 𝑃 𝑥
∞ 𝑡𝑥 1
= 𝑥=1 𝑒 (2 𝑥 )
𝑥
∞ 𝑒𝑡
= 𝑥=1 2
2
𝑒𝑡 𝑒𝑡
=2 + +…
2

63
2
𝑒𝑡 𝑒𝑡 𝑒𝑡
=2 1 + +
2 2
𝑒𝑡 1
=2 𝑒𝑡
1−
2
𝑒𝑡
𝑀𝑋 𝑡 = 2−𝑒 𝑡
𝑑 2−𝑒 𝑡 𝑒 𝑡 +𝑒 𝑡 𝑒 𝑡 2𝑒 𝑡
(ii) 𝑑𝑡 𝑀𝑋 𝑡 = =
(2−𝑒 𝑡 )2 (2−𝑒 𝑡 )2
𝑑
∴ 𝜇1′ = 𝑀𝑋 𝑡 =2
𝑑𝑡 𝑡=0
2
𝑑2 2−𝑒 𝑡 2𝑒 𝑡 +2𝑒 𝑡 2(2−𝑒 𝑡 )
(𝑀𝑋 𝑡 =
𝑑𝑡 2 (2−𝑒 𝑡 )4
8𝑒 𝑡 −2𝑒 2𝑡
= (2−𝑒 𝑡 )3
𝑑2
∴ 𝜇2′ = 𝑀𝑋 𝑡 =6
𝑑𝑡 2 𝑡=0
(iii) Variance 𝜇2 = 𝜇2′ − 𝜇1′ 2
=6-4 = 2
Problem :
Find the m.g.f of the r.v X having the p.d.f

−𝟏
𝒇 𝒙 = , −𝟏 < 𝑥 < 2
𝟑
𝟎, 𝒐𝒕𝒉𝒆𝒓𝒘𝒊𝒔𝒆
Solution:
X is a continuous random variable.

∞
𝑀𝑥 𝑡 = 𝐸(𝑒 𝑡𝑥 ) = 𝑒 𝑡𝑥 𝑓(𝑥)𝑑𝑥
−∞

64
𝟐
𝟏 𝟐 𝟏 𝒆𝒕𝒙
=𝟑 ∫−𝟏 𝒆𝒕𝒙 𝒅𝒙 =𝟑 when t≠ 0
𝒕 −𝟏
𝒆𝟐𝒕 −𝑒 −𝑡
= when t≠ 0
𝟑𝒕
𝟏 𝟐 𝟏 𝟐
when t= 0, 𝑀𝑥 𝑡 = 𝟑 ∫−𝟏 𝒅𝒙 =𝟑 𝒙 −𝟏 =1
𝒆𝟐𝒕 − 𝑒 −𝑡
𝑀𝑥 𝑡 = , when t ≠ 0
𝟑𝒕
𝟏 when t = 0
Some Special Distributions:
Introduction:
In this chapter we discuss some important distribution of random variable which are frequently
used is statistics. We make a detailed study of binomial distribution, Poisson distribution which
are of discrete type and normal distribution which is of continuous type.
Binomial Distribution:
Definition:
Let n be any positive integer and let 0 < 𝑝 < 1, 𝐿𝑒𝑡 𝑞 = 1 − 𝑝
𝑛𝑐𝑥 𝑝 𝑥 𝑞 𝑛 −𝑥 𝑖𝑓 𝑥 = 0,1,2, … . 𝑛
Define 𝑝(𝑥) =
0 𝑜𝑡𝑕𝑒𝑟𝑤𝑖𝑠𝑒
A discrete random variable with the above p.d.f. is said to have binomial distribution
and the p.d.f itself is called a binomial distribution.
Note: 1
The two independent constants n and p in the distributions are known as the parameters of the
distribution. If x is a binomial variate with parameters n and p we write as
𝑋~𝐵 𝑛, 𝑃 .

65
Note:2
In this experiment is repeated N times (say) then the frequency function of the binomial
distribution is given 𝑏𝑦 𝑓 𝑥 = 𝑁𝑃 𝑥 = 𝑁𝑛𝑐𝑥 𝑝 𝑥 𝑞 𝑛−𝑥 , 𝑥 = 0,1,2, … … . . 𝑛.
Theorem: 1
m.g.f of a binomial distribution about the origin is
(𝑞 + 𝑝𝑒 𝑡 )𝑛 .
Proof:
𝑛
𝑀𝑋 𝑡 = 𝐸 𝑒 𝑋𝑡 = 𝑒 𝑋𝑡 𝑝 𝑥 = 𝑒 𝑡𝑥 𝑛𝑐 𝑥 𝑝 𝑥 𝑞 𝑛−𝑥
𝑥=0
𝑛
= 𝑥=0 𝑛𝑐𝑥 𝑝𝑒 𝑡 𝑥 𝑛−𝑥
𝑞
= 𝑞 + 𝑝𝑒 𝑡 𝑛
.
Moments of binomial distribution:

𝜇 2′ 2
We know that for any random variable x the m.g.f is 𝑀𝑋 𝑡 = 1 + 𝜇1′ + 𝑡 +
2!
𝜇 𝑟′
…+ 𝑡𝑟 + ⋯ …
𝑟!
Theorem: 2 (Addition property of binomial distribution)
If 𝑋1 ~𝐵 𝑛1 , 𝑝 , 𝑋2 ~𝐵(𝑛2 , 𝑝) are independent random variable then 𝑋1 + 𝑋2 is 𝐵(𝑛1 +

𝑛2 , 𝑝).
Proof:
Given𝑋1 , and𝑋2 are independent random variable with parameters 𝑛1 , 𝑝 𝑎𝑛𝑑 𝑛2 , 𝑝

respectively.
Let us consider m.g.f of 𝑋1 and 𝑋2 about origin.
∴ 𝑀𝑋1 = 𝑞 + 𝑝𝑒 𝑡 𝑛1
𝑀𝑋2 𝑡 = 𝑞 + 𝑝𝑒 𝑡 𝑛2
Now, 𝑀𝑋1+𝑋2 𝑡 =𝑀𝑋1 𝑡 + 𝑀𝑋2 𝑡 ( since 𝑋1 , and 𝑋2 are independent)
= 𝑞 + 𝑝𝑒 𝑡 𝑛1
+ 𝑞 + 𝑝𝑒 𝑡 𝑛2
= 𝑞 + 𝑝𝑒 𝑡 𝑛 1 +𝑛 2

66
= m.g.f of the binomial 𝑋1+𝑋2 with parameters 𝑛1 + 𝑛2 and 𝑃.
Hence the uniqueness theorem 𝑋1+𝑋2 is a binomial variable with parameters
𝑛1 + 𝑛2 and 𝑃.
Theorem: 3
𝑛
Characteristic function of binomial distribution is 𝑞 + 𝑝𝑒 𝑖𝑡 .
Proof:
Let 𝑋 ~ 𝐵(𝑛, 𝑃). Hence 𝜑𝑥 𝑡 = 𝐸 𝑒 𝑖𝑡𝑥

𝑛 𝑖𝑡𝑥 𝑛 𝑖𝑡𝑥
= 𝑥=0 𝑒 𝑝 𝑥 = 𝑥=0 𝑒 𝑛𝑐𝑥 𝑝 𝑥 𝑞 𝑛−𝑥
𝑛 𝑥 𝑛−𝑥
= 𝑥=0 𝑛𝑐𝑥 𝑝𝑒 𝑖𝑡 𝑞
= (𝑞 + 𝑃𝑒 𝑖𝑡 )𝑛 .
Mode of Binomial Distribution:
Let 𝑋 ~ 𝐵 𝑛, 𝑃 . 𝑡𝑕𝑒𝑛 𝑃 𝑥 = 𝑛𝑐𝑥 𝑝 𝑥 𝑞 𝑛−𝑥 .Let x be the mode of the binomial

distribution.
Example:
The unbiased coins are tossed and number of heads noted. The experiment is
repeated 64 times and the following distribution is obtained.
No.of 0 1 2 3 4 5 Total
heads
Frequencies 3 6 24 26 4 1 64
Solution:
Here n=5 and N=64
Since the coins are unbiased p=1/2 =q. So that p/q =1
Now p(0)=qn=(1/2)5 =1/32
Hence f(0)=N qn =64×1/32 = 2

67
𝑛−𝑥 𝑝
Using the recurrence formula 𝑝(𝑥 + 1) = 𝑝(𝑥)
𝑥+1 𝑞
P(1)= 5(1/32).
Hence f(1)=10
𝑥 Probabilities Expected Observed

𝑝(𝑥) frequencies frequencies
𝑓(𝑥) = 𝑁𝑝(𝑥)
0 p(0)=1/32 2 3
1 P(1)=5/32 10 6
2 P(2)=10/32 20 24
3 P(3)=10/32 20 26
4 P(4)=5/32 10 4
5 P(5)=1/32 2 1
Total 64 64
Problem:
In a binomial distribution the mean is 4 and the variance is 8/3.Find the mode of the
distribution.
Solution:
Given mean =4 and Variance=8/3
∴ 𝑛𝑝 = 4 and npq=8/3
𝑛𝑝𝑞 8 2
=3×4 = 3
𝑛𝑝
q = 2/3,p=1/3
∴ 𝑛𝑝 = 4
n=12
Consider (n+1)p=13/3=4.3
Hence the mode is 4.

68
Problem:
A discrete random variable X has the mean 6 and variance 2. If it is assumed that
the distribution is binomial. Find the probability that 5≤ 𝑥 ≤ 7.
Solution:
Given np=6 and npq=2
Hence q=1/3 and p=2/3
Also n=9
Now, p(5≤ 𝑥 ≤ 7)=p(x=5)+p(x=6)+p(x=7)
2 5 1 4 2 6 1 3 2 7 1 2
=9C5 +9C6 + 9C7
3 3 3 3 3 3
126 26 27
= x25 + 84 × + 36 ×
39 39 39
25
= [126+168+144]
39
25
= x 438
39
32×438
= 19683
= 0.712
Problem :
An insurance agent accepts policies of 5 men all of identity age and in good health. The
probability that man of this age will be alive 30 years hence is 2/3. Find the probability that is 30
years (i) all five men (ii) at least one man (ii) almost three will be alive.
Solution:
Here n=5, p=2/3, q=1-p=1-2/3=1/3
It is a binomial distribution and x~B (5,2/3 ) Hence
P(X=x) = nCxPxqn-x
Probability of all 5 will be alive is

69
P(x=5) = 5C5 (2/3)5 = 32/243
ii) Probability of at least one being alive = 1- probability of no one being alive.
Probability of no one being alive
P(x=0) = 5C0 (1/3)5 = 1/243
Probability of at least one being alive
= 1-1/243
= 242/243
iii) Probability of almost 3 being alive = probability of one man being alive or
probability of 2 men being alive (or) probability of 3 men being a live.
= 1- [Prob. Of 4 men being a live (or) probability of 5 men being alive]
P(x≤ 3) = 1 − 𝑃(𝑋 > 3)
= 1- [P(x=4) + P(x=5)]
=1-[5C4 (2/3)4 (1/3) + 5C5 (2/3)5]
= 1-[5(16/243)+ 32/243]
= 1-112/243 = 131/243
Problem :
Six dice are thrown 729 times. How many times do you except at least 3
dice to show a five or six.
Solution:
Here n = 6, N= 729
P=prob. Of getting 5 or 6 with one dice = 2/6 = 1/3
Q = 1-1/3 = 2/3
The expected frequency of 0,1,2,..6. successor are the successive terms of
729 (1/3 + 2/3)6
Excepted number of times at least 3 dice showing five or six
= 729 (6C3 (1/3)3 + 6C4(1/3)4 (2/3)2 + 6C5 (1/3)5 2/3 + 6C6 (1/3)6]

70
= 729/36 × 233
= 169857/729
=233.
Problem :
If the m.g.f. of a r.v.x is of the form Mx(t) = (0.4et+0.6)8 find i) E(x)
ii) the m.g.f of the r.v Y= 3X+2
Solution:
We have the m.g.f of the binomial variable X ~ B(n,p) is (q+pet)n
Here p=0.4, q=0.6 and n=8
i) We have E(X) = 𝜇1 = 𝑛𝑝
E(X) = 8x0.4
= 3.2
ii) My (t) = M3X+2 (t)
= e2t Mx (3x)
= e 2t (0.6+0.4e 3t) 8
Poisson distribution:
We have the binomial distribution is determined by 2 parameters p and n. If the number

of trails is in definitely large and the probability p of success is in definitely small such that
np=𝜆,where 𝜆 is a constant then the limiting case of the binomial distribution when n→
∞ 𝑎𝑛𝑑 𝑝 → 0 becomes a distribution known as Poisson distribution
Definition:
A discrete random variable X is said to be follow a Poisson distribution if it assumes only

nonnegative integer values and its probability density function is given by
𝑒 −𝑥 𝜆𝑥
𝑝 𝑥 =𝑝 𝑋=𝑥 =𝑓 𝑥 = , 𝑖𝑓 𝑥 = 0,1,2 …
𝑥!
0, 𝑜𝑡𝑕𝑒𝑟𝑤𝑖𝑠𝑒

71
Where 𝜆 is the parameter of the distribution if X is a Poisson variate with
parameter 𝜆.we write X~𝑝(𝜆)
Example:
If fit a Poisson distribution to the following data
x 0 1 3 42
x 0 1 2 3 4 Total
f 5 2 3 1
f 123 9 14 3 1 200
Solution:
To fit a Poisson distribution we have to calculate all the excepted frequencies

𝑓𝑖 𝑥𝑖
Here the mean 𝜆 = 𝑓𝑖
59+28+9+4
= =0.5
200
𝜆 = 0.5
𝐻𝑒𝑛𝑐𝑒 𝑡𝑕𝑒 𝑝. 𝑑. 𝑓 𝑜𝑓 𝑝𝑜𝑖𝑠𝑠𝑜𝑛 𝑑𝑖𝑠𝑡𝑟𝑖𝑏𝑢𝑡𝑖𝑜𝑛 𝑖𝑠
𝑒 −0.5 0.5 𝑥
𝑝 𝑥 =
𝑥!
∴ 𝑝 0 = 𝑒 −0.5 = 0.6065
Hence f(0)=Np(0)
=200× (0.6065)
x
Probabilities using Expected Observed
p(x+1)=(λ/(x+1))p(x) frequencies frequencies
f(x)=Np(x)
0 P(0)=0.6065 121.3 123

72
=1213 1 P(1)=0.3033 60.66 59
2 P(2)=0.0758 15.16 14
Using recurrence formula 3 P(3)=0.0126 2.52 3

4 P(4)=0.0027 0.54 1
𝜆
VP(x+1)= 𝑝(𝑥)
𝑥+1
0.5
We have p(1) = 𝑝(0)
=0.3033
=60.66
Problem :
The probabilities of a Poisson variable taking the values 3 and 4 are equal calculate the
probabilities of variates taking the values 0 and 2
Solution
Take x be a Poisson variate with parameter λ
𝑒 −λ λ x
∴P(X=x) = 𝑥!
Given P(X=3)=P(X=4)
To find P(X=0) and P(X=1)
𝑒 −λ λ 3
P(X=3)=P(X=4) ⇒ 3!
𝑒 −λ λ 4
= ⇒ λ=4
4!
𝑒 −4 40
P(X=0)= =𝑒 −4 = 0.0183
0!
𝑒 −4 42
P(X=2)= =0.146.
2!

73
Problem :
Assuming that one in 80 births in a case twins. Calculate the probability of 2 or more
birth of twins on a day when 30 births occur using (i) binomial distribution (ii) Poisson
approximation.
Solution:
Assuming X to be a binomial variate with p=probability of twin births=1/80=0.125.
(q=0.9875) where n=30 we get
𝑥 30−𝑥
𝑝 𝑥 = 30𝑐𝑥 0.0125 0.9875
Probability of 2 or more births of twins on a day is
𝑝 𝑋 ≥ 2 = 1−𝑝 𝑋 <2
=1-[p(X=0)+p(X=1)]
=1-[(0.9875)30+30(0.0125)(0.9875)29
=1-0.6943(1.3625)
=0.054
Assuming X to be a Poisson variate with λ=np=0.375
𝑒 −𝜆 𝜆 𝑥 𝑒 −0.375 0.375 𝑥
We get 𝑝 𝑥 = =
𝑥! 𝑥!
𝑝 𝑋 ≥2 =1− 𝑝 𝑋 =0 +𝑝 𝑋 =1
=1-e-0.375+e-0.375(0.375)
=1-0.6873(1.375)
=0.0550.

74
NORMAL DISTRIBUTION:
Normal distribution is one of the most widely used distribution in application of

statistical methods.
∞ −𝑦 2
We have ∫−∞ 𝑒 2 𝑑𝑦 = 2𝜋
−𝑦 2
∞ 𝑒 2
Hence ∫−∞ 2𝜋 𝑑𝑦=1
− 𝑥 −𝜇 2
𝑥−𝜇 ∞ 𝑒 2𝜍 2
Put 𝑦 = then we have ∫−∞ 𝑑𝑥 = 1
𝜍 𝜍 2𝜋
Definition:
A continuous random variable X is said to follow a normal distribution if its

− 𝑥 −𝜇 2
𝑒 2𝜍 2
probability density function is given by f(x)= where −∞ < 𝑥 < ∞
𝜍 2𝜋
𝜇 𝑎𝑛𝑑 𝜍 are constants and 𝜍>0 and are called the parameters of the distribution and we
write 𝑋~𝑁(𝜇, 𝜍 2 ).
Fitting of normal distribution:
To fit a normal distribution to given date we first calculate the mean

𝜇 𝑎𝑛𝑑 𝑠. 𝑑 𝜍.Thus the normal curve fitted to the given data is given by
− 𝑥 −𝜇 2
𝑒 2𝜍 2
f(x)= where −∞ < 𝑥 < ∞
𝜍 2𝜋
Properties of normal distribution:
1. The normal probability curve is symmetrical about the ordinate at 𝑥 = 𝜇.The

ordinate decreases rapidly as x increases. The curve extends to infinity on either side of
the mean. The X-axis is an asymptote to the curve.
2. The mean, median and mode coincide the maximum ordinate at 𝑥 = 𝜇 is given by
1
.
𝜍 2𝜋
3. 𝜇 ± 𝜍 are the points of inflexion of the normal curve and hence the points of
inflexion are also equidistant from the median.

75
4.The area under normal curve is unity. The ordinate at 𝑥 = 𝜇 divides the area under
the normal curve into two equal parts symmetry also ensures that the first and third
quartiles of normal distribution are equidistant from the median of course on
either side.
5. 𝑝 𝜇 − 𝜍 < 𝑥 < 𝜇 + 𝜍 = 0.6826
𝑝 𝜇 − 2𝜍 < 𝑥 < 𝜇 + 2𝜍 = 0.9544
𝑝 𝜇 − 3𝜍 < 𝑥 < 𝜇 + 3𝜍 = 0.9973
6.Q.D:M.D:S.D=10:12:15.
Problem :
If X is normal distributed with zero mean and unit variance.
Find the expectation of X2.
Solution:
Given 𝑋~𝑁 0,1
1 −𝑥 2
Hence the normal is 𝑓 𝑥 = 𝑒 2
2𝜋
∞
Now 𝐸 𝑋 2 = ∫−∞ 𝑥 2 𝑓 𝑥 𝑑𝑥
1 ∞ −𝑥 2
= ∫ 𝑥2 𝑒
2𝜋 −∞
2 𝑑𝑥
1 −𝑥 2 ∞ −𝑥 2
= [−𝑥𝑒 2 ]∞
−∞ +∫−∞ 𝑒 2 𝑑𝑥
2𝜋
1
= 0 + 2𝜋
2𝜋
=1.
Problem :
If X is normally distributed with mean 8 and s.d 4. Find
(i)P( 5≤ 𝑋 ≤ 10) 𝑖𝑖 𝑃(10 ≤ 𝑋 ≤ 15)
(iii) P(𝑋 ≥ 5) 𝑖𝑣 𝑃(│𝑋 − 5│ ≤ 15.

76
Solution:
Given
𝑋~𝑁 8,4
𝑋−8
Hence the standard normal variate 𝑍 = 4
When X=5;z=-0.75
When X=10;Z=0.50
When X=15;Z=1.75
When X=20 ;Z=3
(i) P( 5≤ 𝑋 ≤ 10) = 𝑃(−0.75 ≤ 𝑍 ≤ 0.50)
=P(-0.75≤ 𝑍 ≤ 0) + 𝑃(0 ≤ 𝑍 ≤ 0.50)
=P(0≤ 𝑍 ≤ 0.75) + 𝑃(0 ≤ 𝑍 ≤ 0.50)
=0.2734+0.1915
=0.4649.
𝑖𝑖 𝑃 10 ≤ 𝑋 ≤ 15 =P(0.5≤ 𝑍 ≤ 1.75)
=P(0≤ 𝑍 ≤ 1.75) − 𝑃(0 ≤ 𝑍 ≤ 0.5)
= 0.4599 – 1.1915
= 0.2684
iii) P ( x≥5) = p(Z ≥1.75)
= 0.5 – p (0 ≤ 𝑧 ≤ 1.75)
= 0.5 – 0.4599
= 0.0401
iv) p(x ≤ 5) = p(z ≤-0.75)
= 0.5 – p(-0.75 ≤ 𝑧 ≤ 0)
= 0.5 – p( 0 ≤ 𝑧 ≤ 0.75)

77
= 0.5 – 0.2734
= 0.2266
Problem :
The marks of 1000 students in a university are found to be normally distributed with
mean to and s.d. 5. Estimate the number of students whose marks will be i) between 60
and 75 (ii) more than 75 (iii) less than 68.
Solution :
Let x denote the marks of students .
Hence X~N (70, 25)
The standard normal variate is

𝑋−𝜇 𝑋−70
𝑍= =
𝜍 5
i) To find p(60 < x < 75)
When x = 60; z= -2 and when x=75; z=1
P (60 < x < 75) = p (-2 < z < 1)
= p (-2 <z < 0) + p ( 0 < z < 1)
= p (0 < z < 2) + p (0 < z < 1)
= 0.4772 + 0.3413
= 0.8185
∴ The number of students whose marks is between 60 and 75 is 1000 x 0.8185 = 819.
(ii) To find p ( x> 75)
When x = 75; z =1
∴p (x > 75) = p(z > 1)
= 0.5 – p (0 <z < 1)
= 0.5 – 0.3413

78
= 0.1587
∴ The number of students whose marks is more than 75 is 159.
(iii) To find p (x < 68)

68−70
When x = 68, 𝑧 = = −0.4
5
∴p (x < 68) = p (z < -0.4)
= p(z > 0.4)
= 0.5 – p(0 < z < 0.4)
= 0.5 – 0.1554
= 0.3546.
Problem :
Assume the mean height of soldiers to be 68.22 inches with variance of 10.8 inches. How
many soldiers in a regiment of 2000 soldiers would you expect to be over six feet tall. Assume
heights to be normally distributed.
Solution :
Let the variable x denote the height in inches of the solders.
Mean 𝜇 = 68.22; 𝜍 2 = 10.8, 𝜍 = 3.286
Hence X ~N (68.22, 3.286)
∴p(x > 6 feet) = p (x > 72)

𝑥−𝜇
When x = 72; 𝑧 = 𝜍
72−68.22
= 3.286
3.78
= 3.286
= 1.15

79
∴p (x > 72) = p(z > 1.15)
= 0.5 – p (0 ≤ 𝑧 ≤ 1.15)
= 0.5 – 0.3749
= 0.1251
∴ The number of soldiers in the regimet of 2000 over 6 feet tall is 2000 x 0.1251= 250
Problem :
A set of examination marks is approximately distributed with mean 75 and S.D. of 5. If the
top 5% of students get grade A and the bottom 25% get grade B what mark is the lowest A and
what mark is the highest B?
Solution :
It x denote the marks in the examination.
Given x is normally distributed with mean 𝜇 = 75 𝑎𝑛𝑑 𝜍 = 5
ie) X ~N (75, 25)
Let 𝑥1 be the lowest marks for A and 𝑥2 be the highest marks for B. Given
p(x >𝑥1 ) = 0.05 and p(x <𝑥2 ) = 0.25
The standard normal variate
𝑥 1−𝜇
𝑧= 𝜍
𝑥 1 −75
= = 𝑧1
5
𝑥 2 −𝜇 𝑥 2 −75
Z= = = −𝑧2 …………. (1)
𝜍 5
P(0 < z <𝑧1 ) = 0.45
𝑧1 = 0.45
P(-𝑧2 < 𝑧 < 0) = 0.25
P( 0 < z <𝑧2 ) = 0.25
𝑧2 = 0.675
(1)⇒ 𝑥1 = 75 + 5𝑧1
𝑥1 = 83.225

80
𝑥1 ≈ 83
𝑥2 = 75 – 5𝑧2
𝑥2 = 71.625
𝑥2 ≈ 72
Hence the lowest mark for grade A is 83 and the highest mark for B is 72.
Problem :
In a normal distribution 31% of the items are under 45 and 8% are over 64.
Find the mean and standard deviation.
Solution :
Let x denote the normal variate with mean 𝜇 𝑎𝑛𝑑 𝑆. 𝐷. 𝜍
Given p(x < 45) = 0.31 and p (x > 64) = 0.08
45− 𝜇
When x = 45, 𝑧 = = −𝑧1
𝜍
45− 𝜇
When x = 64, 𝑧 = = 𝑧2
𝜍
P(0 < z <𝑧2 ) = 0.42 and p(−𝑧1 < 𝑧 < 0 ) = 0.19

From the area table we get
𝑧1 = 0.496 𝑎𝑛𝑑 𝑧2 = 1.405
(1)⇒ 45 –𝜇 = 0.496 𝜍
64 – 𝜇 = 1.405 𝜍
𝜍 = 9.99 ≈ 10
𝜇 = 49.96 ≈ 50
Problem :
Find the probability of getting between 3 heads to 6 heads in 10 tosses of a fair
coin using (i) binomial distribution (ii) the normal approximation to the binomial
distribution.
Solution :
(i). Take x as the binomial variate,

81
1 1
P = 2 ,𝑞 = , 𝑛 = 10
2
1
We have X ~ 𝐵 (10, 2)
Probability of getting at least 3 heads

= p (x ≥ 3)
= p(x=3) + p(x=4) + p(x=5) + p(x=6)

3 7 4 6 5 5 6 4
1 1 1 1 1 1 1 1
= 10𝑐3 + 10𝑐4 + 10𝑐5 + 10 𝑐6
2 2 2 2 2 2 2 2
1
= 120 + 210 + 252 + 210
210
792
= 210
= 0.7734
(ii). Taking the data as continuous it follows that 3 to 6 heads can be

considered as 2.5 to 6.5 heads.
Mean 𝜇 = 𝑛𝑝
1
= 10 2
= 5
𝜍= 𝑛𝑝𝑞 = 1.58
∴ 𝑋 ~ 𝑁 5, 1.58
𝑋− 𝜇
The standard normal variate is 𝑧 = 𝜍
2.5−5
For X = 2.5 ;𝑧 = 1.58
= -1.58
6.5−5
X = 6.5 ;𝑧 = 1.58
Z = 0.95
P( 2.5 ≤ 𝑥 ≤ 6.5) = 𝑝 (−1.58 < 𝑧 < 0.95)
= P ( -1.58 < z < 0) + p(0 < z < 0.95)

82
= p(0 < z < 1.58 ) + p ( 0 < z , 0.95)
= 0.4429 + 0.3289 = 0.7718.

83
UNIT III : ASSOCIATION OF ATTRIBUTES
Association of Attributes - Coefficient of Association - Consistency - Time Series – Definition -

Components Of Time Series - Seasonal and cyclic variations.
THEORY OF ATTRIBUTES
Attributes:
The qualitative characteristics of a population are called attributes and they cannot be
measured by numeric quantities. Hence the statistical treatment required for attributes is different
from that of quantitative characteristic.
Suppose the population is divided into two classes according to the presence or absence
of a single attribute. The positive class denotes the presence of the attributes and the negative
class denotes the absence of the attribute. Capital Roman letter such as A,B,C,D… are used to
denote positive Greek letters such as 𝛼, 𝛽, 𝛾, 𝛿 …… are used to denote negative classes.
For example If A represents the attribute richness then 𝛼 represents the attribute non-
richness (poor).
A class represented by n attributes is called a class of nth order.
For example,
A,B,C, 𝛼, 𝛽, 𝛾, 𝛿 are all of first order, AB, A𝛽, 𝛼𝐵, 𝛼𝛽 are of second order, and ABC,
A𝛽𝛾, 𝐴𝛽𝐶, 𝛼𝛽𝛾 are of the third order.
The number of individuals possessing the attributes in a class of nth order is called a class
frequency of order „n‟ and class frequencies are denoted by bracketing the attributes.
Thus (A) stands for the frequency of A the number of individuals possessing the attribute
A and (A𝛽) stands for the number of individuals possessing of the attributes A and not B.

84
Note:
1. Class frequencies of the type (A), (AB), (ABC) are known as positive
class frequencies.
2. Class frequencies of the type (𝛼), 𝛽 , 𝛼𝛽 , 𝛼𝛽𝛾 … . 𝑎𝑟𝑒 known as
negative class frequencies.
3. Class frequencies of the type (𝛼𝐵), 𝐴𝛽 , 𝐴𝛽𝛾 , 𝛼𝛽𝐶 …. are known as
contrary frequencies.
4. The classes of highest order are called the ultimate classes and their
frequencies are called the ultimate class frequencies.
Examples:
1. AB = (ABC) + (AB𝛾)
Consider, (AB𝛾) = 𝐴𝐵𝛾. 𝑁
=AB(1-C).N
=AB.N – ABC.N
=(AB) – (BC)
∴ 𝐴𝐵 = 𝐴𝐵𝐶 + (𝐴𝐵𝛾)
2. If there are two attributes A and B we have,
N = (A) + (𝛼) = (B) + (𝛽)
Hence N = (A) + (𝛼)
N = (AB) + (A𝛽) + (𝛼B) + (𝛼𝛽)
And N = (B) +(𝛽) = (AB) + (𝛼𝐵) + 𝐴𝛽 + (𝛼𝛽)
If there are three attributes A,B,C we have N = (A) + (𝛼)
We have N = (A) + (𝛼)
⇒ N = (AB) + (A𝛽) +( 𝛼𝛽) + (𝛼𝛽)
Thus,
N = (ABC) + (AB𝛾) + 𝐴𝛽𝐶) + 𝐴𝛽𝛾 + 𝛼𝐵𝐶 + 𝛼𝐵𝛾 + 𝛼𝛽𝐶 + (𝛼𝛽𝛾)
3. Consider two attributes A and B

Now, (𝛼𝛽) = 𝛼𝛽. 𝑁

85
= (1-A) (1-B).N
= (1-A-B+ AB). N
= N-A.N- B.N + AB.N
= N-(A) – (B) + (AB)
4. (AB) = AB.N
= (1-𝛾) 1 − 𝛽 . 𝑁
= (1-𝛼 − 𝛽 + 𝛼𝛽). 𝑁
= 𝑁 − 𝛼. 𝑁 − 𝛽. 𝑁 + 𝛼𝛽. 𝑁
= N- (𝛼) − 𝛽 + (𝛼𝛽)
5. (𝛼𝛽𝛾) = 𝛼𝛽𝛾. 𝑁 = 1 − 𝐴 1 − 𝐵 1 − 𝐶 . 𝑁 = N-A.N-B.N – C.N +

AB.N + AC.N + BC.N – ABC.N
= N-(A) – (B) –(C) + (AB) +(AC) +(BC) – (ABC)
6. N= (A) +(B) + (C) – (AB) – (BC) – (AC) + (ABC) + (𝛼𝛽𝛾)
Problem :
Given (A) = 30, (B) = 25, (𝛼) = 30 𝛼𝛽 = 20.
Find i) (N) (ii) (𝛽) (iii) (AB) (iv) (A𝛽) ( 𝑉) (𝛼𝐵)
Solution:
i) N = (A) + (𝛼) = 30 + 30 = 60
ii) (𝛽) = 𝑁 − 𝐵 = 60 − 25 = 35
iii) 𝐴𝐵 = 𝐴𝐵. 𝑁
= (1-𝛼) 1 − 𝛽 . 𝑁
= N-(𝛼) − 𝛽 + (𝛼𝛽)
= 60-30-35+20
= 15
iv. (A𝛽) = 𝐴𝛽. 𝑁 = 𝐴 1 − 𝐵 . 𝑁

= (A) – (AB)
= 30-15
= 15

86
v. (𝛼𝐵) = 𝛼𝐵. 𝑁 = 1 − 𝐴 𝐵. 𝑁
= (B)-(AB)
= 25-15
=10
Problem :
Given the following ultimate class frequencies of two attributes A and B. Find the
frequencies of positive and negative class frequencies and the total number of observations.
(AB) = 975, (𝛼𝐵) = 100, 𝐴𝛽 = 25, 𝛼𝛽 = 950.
Solution:
Positive class frequencies are (A) and (B)
(A) = (AB) + (𝐴𝛽) = 975 + 25 = 1000

(B) = (AB) +(𝛼𝐵) = 975 + 100 = 1075
Negative class frequencies are (𝛼) 𝑎𝑛𝑑 (𝛽)
(𝛼) = ( 𝛼𝐵) + 𝛼𝛽 = 100 + 950 = 1050
(𝛽) = 𝐴𝛽 + 𝛼𝛽 = 235 + 950 = 975
N = (A) + (𝛼) = (B) + (𝛽)
Taking,
N = (A) + (𝛼) = 1000 + 1050 = 2050
Problem :
Given the following positive class frequencies find the remaining class frequencies N =
20 (A) = 9; (B) = 12; (C) = 8; (AB) = 6; (BC= 4); (CA) = 4; (CA) = 4; (ABC) = 3
Solution:
There are three attributes A,B,C.

87
∴ The total number of class frequencies is 33=27.
We are given only 8 class frequencies and we have to find the remaining 19 class
frequencies. They are
Order 1:
(𝛼) = 𝑁 − 𝐴 = 20 − 9 = 11.
(𝛽) = N-(B) = 20-12=8
(𝛾) = 𝑁 − 𝐶 = 20 − 8 = 12
Order 2:
𝐴𝛽 = 𝐴 1 − 𝐵 . 𝑁
= (A) – (B)
= 9-6= 3
( 𝛼𝐵) = 1 − 𝐴 𝐵. 𝑁
= (B) – (AB)
= 12-6 = 6
(A𝛾) = 𝐴 1 − 𝐶 . 𝑁
= (A) – (AC)
= 9-4
=5
(𝛼𝐶) = 1 − 𝐴 𝐶. 𝑁
=(C) - (AC)
= 8-4=4
(B𝛾) = 𝐵 1 − 𝐶 𝑁
= (B) – (BC)
= 12-4 = 8

88
(𝛽𝐶) = (1 − 𝐵) 𝐶. 𝑁
= (C) - (BC)
= 8-4=4
(𝛼𝛽) = 1 − 𝐴 1 − 𝐵 . 𝑁 = 𝑁 − 𝐴 − 𝐵 + (𝐴𝐵)
= 20-9-12+ 6=5
(𝛽𝛾) = 1 − 𝐵 1 − 𝐶 . 𝑁
= N – (B) – (C) + (BC)
= 20-12-8+4
=4
(𝛼𝛾) = 1 − 𝐴 1 − 𝐶 . 𝑁
= N- (A) – (C) + (AC)
= 20-9-8+4
=7
Order 3:
(A𝛽𝛾) = 𝐴𝐵 1 − 𝐶 . 𝑁
= (AB) – (ABC)
= 6-3=3
(A𝛽𝐶) = 𝐴 1 − 𝐵 𝐶. 𝑁
= (AC) – (ABC)
= 4-3 = 1
(A𝛽𝛾) = 𝐴 1 − 𝐵 1 − 𝐶 . 𝑁
= (A) – (AC) – (AB) + (ABC)
= 9-4-6+3 = 2

89
(𝛼𝐵𝐶) = 1 − 𝐴 𝐵𝐶. 𝑁
= (BC) – (ABC)
= 4-3=1
(𝛼𝐵𝛾) = 1 − 𝐴 1 − 𝐶 . 𝐵. 𝑁
= (B) - (BC) – (AB) + (ABC)
= 12-4-6+3
=5
(𝛼𝛽𝐶) = 1 − 𝐴 1 − 𝐵 𝐶. 𝑁
=(C) - (AC) – (BC) + (ABC)
= 8-4-4+3=3
(𝛼𝛽𝛾) = 1 − 𝐴 1 − 𝐶 . 𝑁
= N-(A)-(B) – (C) + (AB) + (BC) + (CA) – (ABC)
= 20-9-12-8+6+4+4-3 = 2
Problem :
In a class text in which 135 candidates were examined for proficiency in English and
Maths. It was discovered that 75 students failed in English, 90 failed in Maths and 50 failed in
both. Find how many candidates i) have passed in Maths ii) have passed in English, failed in
Maths iii) have passed in both.
Solution:
Let A denote pass in English and B denote pass in Maths .
∴ (𝛼) denotes fail in English and (𝛽) denotes fail in Maths.
Given (𝛼)= 75; (𝛽) = 90; 𝛼𝛽 = 50; 𝑁 = 135

90
We have to find (i) (B) (ii) 𝐴𝛽 𝑖𝑖𝑖 (𝐴𝐵)
i) (B) = N-(𝛽)
= 135-90
= 45
ii) Consider, (𝛽) = 𝐴𝛽 + (𝛼𝛽)
⇒ 𝐴𝛽 = 𝛽 − 𝛼𝛽
= 90 − 50
= 40
iii) (AB) = (1-𝛼)( 1 − 𝛽). 𝑁
= N- (𝛼) − 𝛽 + (𝛼𝛽)
= 135-75-90 + 50
= 20
Problem :
Given N = 1200; (ABC) = 600; (𝛼𝛽𝛾) = 50; 𝛾 = 270;
𝐴𝛽 = 36; 𝛽𝛾 = 204; 𝐴 − 𝛾 = 192; 𝐵 − 𝛽 = 620.
Find the remaining ultimate class frequencies .
Solution:
Since there are 3 attributes there are 23=8.Ultimate class frequencies we are
given two.
Hence we have find the remaining six
They are (i) (AB𝛾) (ii)(A𝛽𝐶)
(𝑖𝑖𝑖) (𝛼𝐵𝐶) 𝑖𝑣 𝐴𝛽𝛾 𝑣 𝛼𝐵𝛾 𝑎𝑛𝑑 (𝑣𝑖)(𝛼𝛽𝐶)
To find the frequencies of positive classes: (A), (B), (C); (AB), (BC), (AC).

91
First order: (A) – (𝛼) = 192
𝐴 + 𝛼 = 1200(= 𝑁)
Adding,
2(A)=1200+192
2(A)=1392
(A)=696
(B) – (𝛽) = 620
(B) – (𝛽) = 620 (=N)
Hence (B) = 910
Now, (C) = N- (𝛾)
= 1200 – 270
= 930.
Second order:
(AB) = (A) – (A𝛽) = 696 − 36
= 660
(BC) = (B) – (B𝛾) = 910 – 204
= 706
We have, N= (A) + (B) + (C) – (AB) – (BC) – (AC) + (ABC) + (𝛼𝛽𝛾)
(AC) = (A) + (B) + (C) – (AB) – (BC) + (ABC) + (𝛼𝛽𝛾)
= 696 + 910 + 930 – 660 – 706 + 600 + 50= 620

92
Third order:
i. (AB𝛾) = 𝐴𝐵 1 − 𝐶 . 𝑁
= (AB) – (ABC)
= 660 – 600
= 60
ii. (A𝛽𝐶) = 𝐴𝐶 1 − 𝐵 . 𝑁
= (AC) – (ABC)
= 620 – 600
= 20
iii. (𝛼𝐵𝐶) = 1 − 𝐴 𝐵𝐶. 𝑁
= (BC) – (ABC)
= 706 – 600
= 106
iv. (A𝛽𝛾) = 𝐴 1 − 𝐵 1 − 𝐶 . 𝑁
= (A) – (AB) – (AC) + (ABC)
= 696 – 660 – 620 + 600
= 16
v. (𝛼𝐵𝛾) = (1 − 𝐴) 1 − 𝐶 𝐵. 𝑁
= (B) – (AB) – (BC) + (ABC)
= 910 – 660 – 706+ 600
= 144.
vi. (𝛼𝛽𝐶) = 1 − 𝐴 1 − 𝐵 𝐶. 𝑁
= (C) - (AC) – (BC) + (ABC)
= 930 – 620 – 706 + 600 = 204

93
Problem :
Given that (A) = (𝛼) = 𝐵 = 𝛽 = 𝑁/2
Show that i) (AB) ii) (𝛼𝛽) 𝑖𝑖 (𝐴𝛽) = (𝛼𝐵)
Solution:
i. (AB) = AB.N
= (1-𝛼) (1 − 𝛽). 𝑁
= N- (𝛼) − 𝛽 + (𝛼𝛽)
= N – N/2 – N/2 + (𝛼𝛽)
(AB) = (𝛼𝛽)
ii. (A𝛽) = 𝐴𝛽. 𝑁

= (1-𝛼) 1 − 𝐵 . 𝑁
= N – (𝛼) − 𝐵 + (𝐴𝐵)
= N – N/2 – N/2 + (𝛼𝐵)
(A𝛽) = (𝛼𝐵)
Problem :
Of 500 men in a locality exposed to cholera 172 in all were attacked, 178
were inoculated and of these 128 were attacked. Find the number of persons.
i) not inoculated not attacked
ii) inoculated not attacked
iii) not inoculated attacked

94
Solution:
Denote the attribute A as attacked and the attribute B as inoculated.
Hence 𝛼 denote “NOT ATTACKED”; 𝛽 DENOTES “NOT INOCULATED”.
Given, N= 500; (A) = 172; (B) = 178; (AB) = 128
To find (i) (𝛼𝛽) 𝑖𝑖 𝛼𝐵 𝑖𝑖𝑖 (𝐴𝛽)
i. (𝛼𝛽) = 𝛼𝛽. 𝑁
= (1-A) (1-B).N
= N-(A) –(B) + (AB)
= 500 -172-178 + 128
= 278
i. (𝛼𝐵) = 𝛼𝐵. 𝑁 = 1 − 𝐴 𝐵. 𝑁
= (B) – (AB)
= 178 – 128 = 50
iii). (A𝛽) = 𝐴𝛽. 𝑁 = 𝐴 1 − 𝐵 . 𝑁
= (A) – (AB)
= 172 – 128 = 44
Problem:
There were 200 students is a college whose results in the first semester, second semester and the
third semester are as follows: 80 passed in the first semester; 75 passes in the second semester.
96 passed in the third semester 25 passed in all the three semester 46 failed in all the three
semester 29 passed in the first two and failed in the third semester 42 failed in the first two

95
semester but passed in the third semester. Find how many students passed in atleast two
semesters
Solution:
Denoting “pass in first semester” as “A‟ Pass in second semester „B‟ and pass in the third
semester as „C‟ we get.
N = 200; (A) = 80, (B) = 75 ; (C) = 96
(ABC) = 25; (𝛼𝛽𝛾) = 46; 𝐴𝐵𝛾 = 29; 𝛼𝛽𝐶 = 42
We have to find (AB𝛾) + ( 𝛼𝐵𝐶) + 𝐴𝛽𝐶 + 𝐴𝐵𝐶
Consider, (C) = (AC) + (𝛼𝐶)
= (ABC) + (A𝛽𝐶) + 𝛼𝐵𝐶 + 𝛼𝛽𝐶
∴ 𝐴𝐵𝐶 + 𝛼𝐵𝐶 + 𝐴𝛽𝐶 = 𝐶 − 𝛼𝛽𝐶
= 96 – 42 = 54
∴ (ABC) + (𝛼𝐵𝐶) + 𝐴𝛽𝐶 + 𝐴𝐵𝛾 = 54 + 29 = 83
Thus the number of students who passed in atleast two semester is 83.
Problem :
Given (ABC) = 149; (AB𝛾) = 738; 𝐴𝛽𝐶 = 225 ; 𝐴𝛽𝛾 = 1196; 𝛼𝐵𝐶 =
204; 𝛼𝐵𝛾 = 1762; 𝛼𝛽𝐶 = 171; 𝛼𝛽𝛾 =
21842. 𝑓𝑖𝑛𝑑 𝐴 , 𝐵 , 𝐶 , 𝐴𝐵 , 𝐴𝐶 , 𝐵𝐶 𝑎𝑛𝑑 𝑁.
Solution:
N = (ABC) + (AB𝛾) + 𝐴𝛽𝐶 + 𝐴𝛽𝛾 + 𝛼𝐵𝐶 + 𝛼𝐵𝛾 + 𝛼𝛽𝐶 + (𝛼𝛽𝛾)
= 149 + 738 + 225 + 1196 + 204 + 1762 + 171 + 21842.
= 26287

96
(A) = (ABC) + (AB𝛾) + 𝐴𝛽𝐶 + 𝐴𝛽𝛾 = 149 + 738 + 225 + 1196
= 2308
(B) = (ABC)+ (AB𝛾) + 𝛼𝐵𝐶 + 𝛼𝐵𝛾 = 149 + 738 + 204 + 1762
= 2853
(C) = 749
(𝐴𝐵) = 𝐴𝐵𝐶 + 𝐴𝐵𝛾 = 149 + 738 = 887
(AC) = (ABC) + (𝐴𝛽𝐶) = 149 + 225 = 374
(BC) = (ABC) + (𝛼𝐵𝐶) = 353
Problem :
In a very hotly fought battle 70% of the solders at least lost an eye 75% at least lost an
ear 80% at least an arm and 85% at least lost a leg. How many at least must have lost all the
four?
Solution:
Denoting “loosing an eye” A, “loosing a ear by B” “loosing an arm by C” and

“loosing a leg by D”
We have
N= 100, (A) ≥ 70, 𝐵 ≥ 75, 𝐶 ≥ 80, 𝐷 ≥ 85.
To find the least value of ABCD
(ABCD) ≥ 𝐴 + 𝐵 + 𝐶 + 𝐷 − 3𝑁
≥ 70 + 75 + 80 + 85 − 300
= 10
(ABCD) ≥ 10
At least 10% of the soldiers lost all the four.

97
Problem :
A company producers tube lights and conducts a test on 5000 lights for production
defects of frames (F); chokes (C); starters (S) and tubes (T). The following are the records of
defects.
(F) = 130, (C)=120, (S) = 115, (T) = 86
(FC) = 100, (CS) = 130, (ST) = 75, (FT) = 60
(CT) = 54, (FS) = 37, (FCS) = 90 , (CST) = 85
(FST) = 112, (FCT) = 108, (FCST) = 5.
Find the percentage of the tube lights which pass all the four tests.
Solution:
Number of tube lights passing the four tests
= (1-F) (1-C) (1-S) (1-T) .N
= [1-(F+C+S+T) + (FC+CS+ST+FT+CT+FS) – (FCS+CST+STF+FCT) +

FCST].N
= N-[(F)+(C)+(S)+(T)]+ [(FC)+(CS)+(ST) + (FT)+(CT)+(FS)] –

[(FCS)+(FCT)+((FST)+(CST)] + (FCST)
= 5000- (130+20+115+86) + (100+130 +75+60+54+37)-(90+108+112+85)

+5
= 5000-451+456-395+5
= 5461-846=4615
Out of 5000 tube lights 4615 pass the four tests for defects.
Percentage of tube lights which pass the four tests

98
4615
= 5000 × 100 = 92.3%
Exercises:
1.Given the frequencies (A) = 1150, (𝛼) = 1120, 𝐴𝐵 = 1075 𝛼𝛽 =

985. 𝐹𝑖𝑛𝑑 𝑟𝑒𝑚𝑎𝑖𝑛𝑖𝑛𝑔 class frequencies and total number of observations.
2. Given the following ultimate class frequencies find the frequencies of the
positive and negative classes and the total number of observations.
(AB) = 733, (A𝛽) = 840, 𝛼𝐵 = 699; 𝛼𝛽 = 783.
3. A survey reveals that out of 1000 people in locality 800 like coffee, 700 like tea,
660 like both coffee and tea. Find how many people like neither coffee nor tea.
4. An examination result shows the following data. 56% at least failed in part I
Tamil, 76% at least failed in part II English 82% at least failed in major – chemistry
and 88% at least failed ancillary maths. How many at least failed in all the four?
5. In a university examination 95% of the candidates passed partI, 70% passed in

part II, 65% passed part III. Find how many at least should have passed the whole
examination.
Consistency of data:
Definition:
A set of class frequencies is said to the consistent if none of them is negative otherwise the given
set of class frequencies is said to be inconsistent.
We have the following set of criteria for testing the consistency in the case of single attributes
and three attributes.

99
Attributes Condition Equivalent positive Number
consistency class condition of
conditio
ns
A (A)≥0 (A)≥0
(𝛼)≥0 (A)≤N(Since (𝛼)=(1- 2
A)N≥0)
A,B (AB)≥0 (AB)≥0
(A𝛽)≥0 (AB)≤A
(𝛼B)≥0 (AB)≤B 22
(𝛼𝛽)≥0 (AB)≥(A)+(B)-N
A,B,C (ABC)≥0 i) (ABC)≥0
(AB𝛾)≥0 ii) (ABC)≤(AB)
(A𝛽C)≥0 iii) (ABC)≤(AC)
(𝛼BC)≥0 iv) (𝛼BC) ≤(BC)
(A𝛽𝛾)≥0 v)(ABC) ≥(AB)+(AC)-
(A) 23
(𝛼B𝛾)≥0 vi)(ABC)≥(AB)+(BC)-
(B)
(𝛼𝛽C)≥0 vii)(ABC)≥(AC)+(BC)-
(C)
(𝛼𝛽𝛾)≥0 viii)(ABC)≤(AB)+(BC)
+(AC)-(A)-(C)+(N)
Note:
In the case of 3 attributes conditions
(i) and (Viii)

⇒ (AB) +(BC) +(AC) ≥ 𝐴 + 𝐵 + 𝐶 − 𝑁 … . . . . 𝑖𝑥

100
Similarly,
(ii) and (vii)
⇒ (AC) +(BC) – (AB)≤ 𝐶 … … … … … … … … … … … . . 𝑥
(iii) and (vi)
⇒(AB)+(BC)-(AC) ≤(B) ………………………… ….(xi)
iv) and (v)
⇒ (AB) +(AC) – (BC) (A) ……………………(xii)
conditions (ix) to (xii) can be used to check the consistency of data when the class of
first and second order alone are known.
Problem :
Find whether the following data are consistent. N= 600; (A) = 300;(B) = 400;
(AB)=50.
Solution:
We calculate the ultimate class frequency (𝛼𝛽), 𝛼𝐵 𝑎𝑛𝑑 (𝐴𝛽)
(𝛼𝛽) = 𝛼𝛽. 𝑁 = 1 − 𝐴 (1 − 𝐵). 𝑁
= N-(A) –(B) +(AB)
= 600 – 400 +50
= -50
Since (𝛼𝛽) < 0, the data are inconsistent.

101
Problem :
Show that there is some error in the following data: 50% of people are
wealthy and healthy 35% are wealthy but not healthy 20% are healthy but not wealthy.
Solution:
Taking “wealth” as A and “health as “B” we get the following data
N=100, (AB) =50; (A𝛽) = 35, (𝛼B)=20
To check the consistency of data we find (𝛼𝛽)
(𝛼𝛽) = 𝛼𝛽. 𝑁 = 1 − 𝐴 1 − 𝐵 . 𝑁
= N-(A) – (B) + (AB)
But (A) = (AB) +(A𝛽)
= 50+35=85
(B) = (AB) + (𝛼𝐵)
= 50+20
= 70
(𝛼𝛽) = 100 − 85 − 70 + 50
= -5
(𝛼𝛽)<0
Hence there is error in the data.

102
Problem :
Of 2000 people consulted 1854 speak Tamil; 1507 speak Hindi; 572 Speak English;
676 speak Tamil and Hindi; 286 speak Hindi and English; 114 speak Tamil; Hindi and English.
Show that the information as it stands is incorrect.
Solution:
Let A,B,C denote the attribution of speaking Tamil, Hindi, English respectively.
Given, N= 2000, (A) = 1854, (B) = 1507 (C) = 572;
(AB)= 676; (AC)= 286, (BC) = 270, (ABC)= 114
Consider (𝛼𝛽𝛾) = 𝛼𝛽𝛾. 𝑁
=(1-A) (1-B) (1-C).N
=N – (A) – (B) –(C) + (AB) +(BC) +(AC) – (ABC)
=2000 – 1854 – 1507 – 572 +676 + 270 + 286 – 114
=- 815
∴ 𝛼𝛽𝛾 < 0.
Hence the data are inconsistent.
∴The information is incorrect.
Problem :
Find the limits of (BC) for the following available data.
N = 125, (A) = 48, (B) = 62, (C) = 45
(A𝛽) = 7 𝑎𝑛𝑑 𝐴𝛾 = 18

103
Solution:
To find (AB) and (AC)
(AB) = (A) – (A𝛽)
= 48-7 = 41
(AC) = (A) – (A𝛾)
= 48-18 = 30
Now, by condition of consistency (ix)
(AB) + (BC) +(AC) ≥ 𝐴 + 𝐵 + 𝐶 − 𝑁
41+(BC) + 30 ≥ 48 + 62 + 45 − 125
(𝐵𝐶) ≥ −41 … … … … … . . (𝑖)
Also using (xii)
(AB) +(AC)-(BC)≤ 𝐴
⇒(BC) ≥ 𝐴𝐵 + 𝐴𝐶 − 𝐴
= 41+30-48= 23
(BC) ≥ 23 … … … … … … … … … . 𝑖𝑖
Using (xi), (AB) + (BC) – (AC) ≤ 𝐵
⇒(BC) ≤ 𝐵 + 𝐴𝐶 − 𝐴𝐵
= 62+ 30-41
=51
∴(BC) ≤ 51 … … … … … … … … … . 𝑖𝑖𝑖

104
Using (x), (AC) + (BC) – (AB) ≤ 𝐶
⇒(BC) ≤ 𝐶 + 𝐴𝐵 − 𝐴𝐶
= 45+41-30
= 56
∴(BC)= 56 ……………………………..(iv)
From (i), (ii), (iii) and (iv) we get
23≤ 𝐵𝐶 ≤ 56
Problem :
Find the greatest and least value of (ABC) if (A)=50, (B)=60, (C)= 80, (AB) = 35, (AC)= 45 and
(BC)=42
Solution:
The problem involves 3 attributes and we are given positive class frequencies of first
order and second order only.
Using positive class conditions (ii), (iii), (iv) of consistency for 3 attributers
(ABC) ≤ 𝐴𝐵 ⇒ 𝐴𝐵𝐶 ≤ 35
(ABC) ≤ 𝐵𝐶 ⇒ 𝐴𝐵𝐶 ≤ 42
(ABC) ≤ 𝐴𝐶 ⇒ 𝐴𝐵𝐶 ≤ 45
⇒ 𝐴𝐵𝐶 ≤ 45 … … … … … … … (𝑖)
Using (v) (vi) and (vii)
(ABC) ≥ 𝐴𝐵 + 𝐴𝐶 − 𝐴
⇒ (𝐴𝐵𝐶) ≥ 35 + 45 − 50 = 30

105
(ABC) ≥ 𝐴𝐵 + 𝐵𝐶 − 𝐵
⇒ (ABC) ≥ 35 + 42 − 60 = 17
(ABC) ≥ 𝐴𝐶 + 𝐵𝐶 − 𝐶
⇒ (ABC) ≥ 45 + 42 − 80 = 7
Thus (ABC) ≥ 30
(ABC) ≥ 17
(ABC) ≥ 7
⇒ (ABC) ≥ 30 … … … … … … … 2
From (1) and (2) we get 30≤ 𝐴𝐵𝐶 ≤ 35
∴The least value of (ABC) is 30 and the greatest value of (ABC) is 35.
Problem :
(𝐴) (𝐵) (𝐶)

If = 𝑥; = 2𝑥, = 3x and
𝑁 𝑁 𝑁
(𝐴𝐵) (𝐴𝐶) 𝐵𝐶
= = = 𝑦 . prove that neither 𝑥 𝑛𝑜𝑟 𝑦 can exceed ¼ .
𝑁 𝑁 𝑁
Solution:
Clearly 𝑥 𝑎𝑛𝑑 𝑦 are positive integers. The condition of consistency
(AB) ≤ 𝐴
(𝐴𝐵) (𝐴)
⇒ ≤
𝑁 𝑁
y≤ 𝑥
Similarly,
(BC)≤ 𝐵 ⇒ 𝑦 ≤ 2𝑥

106
⇒ y≤ 𝑥 … … … … … … … … … … 1
Now, (AB)≥ 𝐴 + 𝐵 − 𝑁
𝐴𝐵 𝐴 𝐵
⇒ ≥ + − 1
𝑁 𝑁 𝑁
Thus, (AB) ≥ 𝐴 + 𝐵 − 𝑁
y≥ 3𝑥 − 1
Similarly
(BC) ≥ 𝐵 + 𝐶 − 𝑁
⇒y≥ 5𝑥 − 1
⇒y≥ 5𝑥 − 1 … … … … … … … . 2
(AC) ≥ 𝐴 + 𝐶 − 𝑁
By (1) and (2) 5x -1 ≤ 𝑦 ≤ 𝑥.
1
Taking 5x-1≤ 𝑥 𝑤𝑒𝑔𝑒𝑡 𝑥 ≤ 4
1
Taking y≤ 𝑥 𝑤𝑒𝑔𝑒𝑡 𝑦 ≤ 4
Neither 𝑥 𝑛𝑜𝑟 𝑦 can exceed ¼.
Exercises:
1. Examine the consistency of data when
i) (A)=800; (B)= 700, (AB)=660; (N)= 1000
ii) (A)=600; (B)= 500, (AB)= 50; N= 1000
iii) N=2100; (A)=1000, (B)=1300; (AB)=1100

107
iv) N=100; (A)=45; (B)=55, (C) = 50; (AB)=15 , (BC)= 25, (AC)= 20, (ABC)=12
v)N=1800; (A)=850; (B)=780; (C)=326; (AB)=250; (BC)=122; (AC)=144;(ABC)= 50
2. A market investigator returns the following data of 2000 people consulted 1754 liked
chocolates 1872 liked toffee and 572 liked biscuits, 678 liked chocolate and coffee, 236 liked
chocolates and biscuits, 270 liked chocolates and biscuits, 270 liked toffee and biscuits, 114
liked all the three .Show that the information it started must be incorrect.
3. If (A) = 50; (B)= 60; (C)=50; (A𝛽) = 5;
𝐴𝛾 = 20 𝑎𝑛𝑑 𝑁 = 100. Find the least and greatest value of (BC).
Independence and Association of Data:
Two attributes A and B are said to be independent if there is same proportion of A‟s
amongst B as amongst 𝛽‟s.
Thus A and B are independent iff
(𝐴𝐵) (𝐴𝛽 )
= ……………………………….(i)
𝐵 (𝛽 )
or
(𝐴𝐵) (𝛼𝐵 )
= ……………………………….(ii)
𝐴 (𝛼)
From (i) we get
(𝐴𝐵) (𝐴𝛽 ) (𝐴𝐵)+(𝐴𝛽 ) (𝐴)

= = =
𝐵 (𝛽 ) 𝐵 +(𝛽 ) 𝑁
𝐴 (𝐵)
∴(AB) = ………………….(1)
𝑁
𝐴 (𝛽 )
And (A𝛽) = …………………(2)
𝑁
Again from (1) we get

108
(𝐴𝐵) (𝐴𝛽 )
1- = 1-
(𝐵) (𝛽 )
𝐵 −(𝐴𝐵) 𝛽 − (𝐴𝛽 )
=
(𝐵) 𝛽
(𝛼𝐵 ) (𝛼𝛽 )
=
𝐵 (𝛽)
(𝛼𝐵 ) (𝛼𝛽 )
∴ =
𝐵 (𝛽)
𝛼𝛽 + (𝛼𝐵)
=
(𝛽)+ 𝐵
(𝛼 )
= 𝑁
𝛼 (𝛽 )
(𝛼𝛽) = …………………………(3)
𝑁
𝛼 (𝐵)
And (𝛼𝐵) = ………………………………(4)
𝑁
(1),(2),(3),(4) are all equivalent conditions for independent of the attribute A and B.
Association and Coefficient of Association:
𝐴 (𝐵)
If (AB) ≠ we say that A and B are associated. There are two possibilities.
𝑁
𝐴 (𝐵) 𝐴 (𝐵)
If (AB) > we say that A and B are positively associated and If (AB) < we
𝑁 𝑁
say that A and B are negatively associated.
𝐴 𝐵
Let us denote 𝛿 = 𝐴𝐵 − 𝑁
1
ie. 𝛿 = [ (AB) (𝛼𝛽) − 𝐴𝛽 (𝛼𝛽)]
𝑁
Note:
i. A and B are independent if 𝛿 = 0.

109
ii. A and B are positively associated if 𝛿 > 0 and negatively associated if 𝛿 < 0.
Coefficient of association:
There are several measures indicating the intensitivity of association between two
attribution
A and B.
The most commonly used measures are the Yule‟s coeficiency of association Q and
coefficient of colligation Y which are defined as follows.
𝐴𝐵 𝛼𝛽 − 𝐴𝛽 𝛼𝐵
𝑄=
𝐴𝐵 𝛼𝛽 + 𝐴𝛽 𝛼𝐵
𝑁𝛿
𝑄=
𝐴𝐵 𝛼𝛽 + 𝐴𝛽 𝛼𝐵
𝐴𝛽 𝛼𝐵
1−
𝐴𝐵 𝛼𝛽
𝑌=
𝐴𝛽 𝛼𝐵
1+
𝐴𝐵 𝛼𝛽
Problem :
Check whether the attributes A and B are independent given that (i) = 30 (B)=
60, (AB)= 12, N= 150
(ii)(AB) = 256, (𝛼𝐵) = 768, 𝐴𝛽 = 48 , 𝛼𝛽 = 144.
Solution:
Given class frequencies are of first order condition for independence is
𝐴 (𝐵)
(AB) =
𝑁
Consider,

110
𝐴 𝐵 30×60
= == = 12 = 𝐴𝐵
𝑁 150
𝐴 𝐵
∴ (AB)=
𝑁
Hence A and B are independent.
ii) (A) = (AB) + (A𝛽) = 256 + 48 = 304
(B) = (AB) + (𝛼𝐵 ) = 256 + 768 = 1024
(𝛼) = 𝛼𝐵 + 𝛼𝛽 = 768 + 144 = 912
(𝛽) = 𝐴𝛽 + 𝛼𝛽 = 48 + 144 = 192
N = (A) + (𝛼) = 304 + 912 = 1216
𝐴 𝐵 304 × 1024
Now = = = 256 = 𝐴𝐵
𝑁 1216
𝐴 𝐵
∴ (AB) =
𝑁
Hence A and B are independent.
Problem :
In a class test in which 135 candidates were examined for proficiency in

physics and chemistry, it was discovered that 75 students failed in physics, 90 failed in
chemistry and 50 failed in both. Find the magnitude of association and state if there is
any association between failing in physics and chemistry.
Solution:
Denoting “fail in Physics” as A and “fail in Chemistry” as B we get
(A) = 75, (B) = 90, (AB) = 50, N= 135
The magnitude of association is measured by

111
𝐴𝐵 𝛼𝛽 − 𝐴𝛽 𝛽 (𝛼𝐵 )
𝑄=
𝐴𝐵 𝛼𝛽 + 𝐴𝛽 (𝛼𝐵 )
(𝛼) = 𝑁 − 𝐴 = 135 − 75 = 60
(𝛽) = 𝑁 − 𝐵 = 135 − 90 = 45
(𝛼𝐵) = 𝐵 − 𝐴𝐵 = 90 − 50 = 40
(A𝛽) = 𝐴 − 𝐴𝐵 = 75 − 50 = 25
(𝛼𝛽) = 𝛼 − 𝛼𝐵 = 60 − 40 = 20
50×20−25×40
𝑄=
50×20+20×40
𝑄=0
∴ A and B are independent hence failure in physics and chemistry are completely
independent of each other.
Problem :
Show whether A and B are independent or positively associated or negatively

associated in the following cases.
i) N = 930, (A) = 300, (B) = 400, (AB) = 230
ii) (AB) = 327, (A𝛽) = 545, 𝛼𝐵 = 741, 𝛼𝛽 = 235
iii) (A) = 470, (AB) = 300, (𝛼) = 530, 𝛼𝐵 = 150
iv. (AB) = 66, (A𝛽) = 88, 𝛼𝐵 = 102; 𝛼𝛽 = 136
Solution:
𝐴 𝐵 300 × 400
i) = = 129.03
𝑁 930
𝐴 (𝐵)
Now, 𝛿 = 𝐴𝐵 − 𝑁

112
= 230 – 129.03
= 100.97
Here 𝛿 > 0
Hence A and B are positively associated.
𝐴𝐵 𝛼𝛽 − 𝐴𝛽 (𝛼𝐵 )
ii) 𝑄 =
327×235−545×741
=
327×235+545 × 741
76845 −4038845
=
76845 +403845
−32700
=
480690
= -0.6803
𝑄 < 0.
Hence A and B are negatively associated.
iii) N= (A) + (𝛼)
= 470 + 530
= 1000
(A) = (AB) + (𝛼𝐵)
= 300 + 150
= 450
𝐴 (𝐵) 470 × 450

Now, = = 2115
𝑁 1000

113
𝐴 (𝐵)
∴ 𝛿 = 𝐴𝐵 −
𝑁
= 300- 2155
= - 1825
∴ 𝛿 < 0.
𝐻𝑒𝑛𝑐𝑒 𝐴 𝑎𝑛𝑑 𝐵 𝑎𝑟𝑒 𝑛𝑒𝑔𝑎𝑡𝑖𝑣𝑒𝑙𝑦 𝑎𝑠𝑠𝑜𝑐𝑖𝑎𝑡𝑒𝑑.
iv. 𝑄 =
66×136−88×102
= = 0.∴ A and B are independent.
66×136+88×102
Problem :
Calculate the co-efficient of associate between intelligence of father and son from
the following data.
Intelligent father with intelligent sons 200.Intelligent fathers with dull sons 50.
Dull fathers with intelligence sons 110. Dull fathers with dull sons 600. Comment on
the result.
Solution:
Denoting the “intelligence of fathers” as A and intelligence of sons” by B
we have
(AB) = 200, (A𝛽) = 50 , 𝛼𝐵 = 110, 𝛼𝛽 = 600
𝐴𝐵 𝛼𝛽 − 𝐴𝛽 (𝛼𝐵)
𝑄=
𝐴𝐵 𝛼𝛽 + 𝐴𝛽 (𝛼𝐵)

114
200× 600−50×110
=
200 ×600+50×110
= 0.91235
Since 𝑄 is positive it means that intelligent fathers are likely to have intelligent sons.
Problem :
Investigate from the following data between inoculations against small pox
prevention from attack.
Attacked Not attacked Total

Inoculated 25 220 245
Not inoculated 90 160 250
Total 115 380 495
Solution:
Denoting A as “inoculated” and B as “attacked” we have (AB)= 25, (A𝛽) =

220, 𝛼𝐵 = 90 𝑎𝑛𝑑
(𝛼𝛽) = 160.
𝑄=
25 𝑋 160−220 𝑋 90
=
25 𝑋 160+220 𝑋 90
400−19800
=
400+19800
−15800
=
23800
= -0.6638.

115
Attributes A and B have negative association.
i.e. “Inoculation” and “attack from small pox” are negatively associated.
Thus inoculation against small pox can be taken as the preventive measure.
Problem :
From the following data compare the association between marks in physics and
chemistry in MKU and MSU
University MSU MKU

Total number of 200 1600
candidate
Pass in physics 80 320
Pass in chemistry 40 90
Pass in physics 20 30
and chemistry
Solution:
Denoting “pass in physics” as A and “pass in chemistry” as B.
We have,
MKU MSU
N=1600 N=200
(A) = 320 (A)=80
(A)= 90 (A)= 40
(AB) = 30 (AB) = 20
From the above data we get the rest of the class frequencies for MKU and MSU.

116
MKU MSU
(A𝛽) = 𝐴 − (𝐴𝐵) (A𝛽) = 𝐴 − (𝐴𝐵)
= 320 – 30 = 80 – 20
= 290 = 60
(𝛼𝐵) = 𝐵 − (𝐴𝐵) (𝛼𝐵) = 𝐵 − (𝐴𝐵)
= 90-30 = 40-20
= 60 = 20
(𝛼𝛽) = 𝑁 − 𝐴 − (𝛼𝛽) = 𝑁 − 𝐴 − 𝐵 +
𝐵 + (𝐴𝐵) (𝐴𝐵)
= 1600- 320 – 90 +30 = 200 – 80 – 40 + 20
= 1220 = 100
We now find the coefficient of association between A and B for MKU and MSU
Passed Failed Total

90 65 155
Married
260 110 370
Unmarried
Total 350 175 525
3. From the figures given in the following table compare the association between literacy
and un employment in rural and urban areas- and given reasons for the difference if any

117
Urban Rural
Total adult males 25 lakhs 200 lakhs
Literate males 10 lakhs 40 lakhs
Unemployed 5 lakhs 12 lakhs
males
Literate and 3 lakhs 4 lakhs
unemployed males
Time series:
Definition:
Time series is a series of values of a variable over a period of time arranged

chronologically
Components of a time series:
The various forces affecting the values of a phenomenon in a time series may be broadly
classified into the following three categories generally known as the components of a time series.
1. Longtime trend (or) secular trend

2. Short term fluctuations (or) periodic movements
3. Irregular fluctuations
1. Long time trend:
The general tendency of a time series is to increase or decrease or stagnate over a period
of several years. Such a long run tendency of a time series to increase or decrease over a period
of time is known as secular trend or simply trend. Though the term “long” is a relative term it
depends upon the nature of the series under consideration.

118
The long term trend does not mean that the series should continuously move in one
direction only. It is possible that different tendencies of increases and decrease persist together.
A graphical representation indicating a long term increase or decrease or stability is given is the
following figures.
2. Short term fluctuations:
In most of the time series a number of forces repeat themselves periodically over a period
of time preventing the values of the series to move in a particular direction. The variations
caused by such forces are called short term fluctuations. This short term fluctuations may
broadly be classified into (a) seasonal variation (B) cyclical variation
a) Seasonal variation:
Generally seasonal variations are considered as short term fluctuations

that occur within a year. These fluctuations may be regular as well as irregular
with in a period of one year
b) Cyclical variation
If the period of oscillation for the periodic movements is a time series is

greater than one year then it is called cyclical variation. Generally oscillatory

119
movement is nay business activity is due to the out time of the business cycles
normally having four phases namely prosperity recession, depression and
recovery. The period between two successive peaks or though is known as the
period of the cycle. In cyclical variation generally the period of a cycle is three to
eleven years.
3. Irregular fluctuations
The fluctuation which are purely random and due to unforeseen and unpredictable forces
are called Irregular fluctuations
Measurement of trends
A graphical representation of a time series exhibits the general upwards and

downward tendencies
The following are the four study of measurement of the trend in a time series
i) Graphic method
ii) Method of curve fitting by the principles of least squares.
iii) Method of semi averages
iv)Method of moving averages.
i) Graphic Method
This is the simplest method of determining the trend. In this method all values of the time
series are plotted on a graph paper and a smooth curve is drawn by free hand to pass through as
many points as possible. The smoothing of the curve eliminates the other components such as
seasonal, cyclic and random variations.
ii) The method of curve fitting:
This is the best method of fitting a trend and it is commonly used in practice.

120
iii) Method of semi averages
In this method the whole time series data is classified into two equal parts with respect to
time. Having divided the given series into two equal parts we calculate the arithmetic mean for
each part. These means are called semi-averages. Then these average are plotted against the mid
values of the respective period covered by each part. The line joining these points give the
straight line trend for the time series.
iv) Method of moving averages
This method for measuring the trend consists of obtaining a series of moving average of
successive m terms of the time series. This averaging process smoothens the fluctuations and the
UPS and down in the given data. It has been observed and proved mathematically that if a trend
is liner the period of the moving average is taken to be the period of oscillation.
Measurement for seasonal variation
There is a simple method for measuring the seasonal variation which involves
simple averages.
Simple average method
Step 1:
All the data are arranged by years and months.
Step 2:
Compute the simple average 𝑥i for ith months
Step 3:
Obtain the overall average x of these average 𝑥𝑖 and
𝑥1+𝑥 2+⋯…….+𝑥 12
𝑥= 12

121
Step4:
Seasonal indices for different months are calculated by expressing monthly

average as the percentage of the overall average x
𝑥𝑖
Thus seasonal index for ith month = × 100 Take X = x – 1987 and Y = y-42
𝑥
Then the line of best fit become
Y=ax + b
The normal equations are 𝑥𝑦 = 𝑎 𝑥2 + 𝑏 𝑥
𝑦=𝑎 𝑥 + 𝑛𝑏, 𝑤𝑕𝑒𝑟𝑒 𝑛 = 11
From the table,
-19 = 110 a
−19
⇒a = = - 0.17
110
17 = 11b
17
⇒b11 = 1.55
∴ The line of best fit is Y= - 0.17 x + 1.55
ie. 𝑌 – 42 = − 0.17 (𝑥 − 1987) + 1.55
𝑦 = −0.17𝑥 + 1987 × 0.17 + 1.55 + 42
𝑦 = −0.17𝑥 + 381.34 𝑖𝑠 𝑡𝑕𝑒 𝑠𝑡𝑟𝑎𝑖𝑔𝑕𝑡 𝑙𝑖𝑛𝑒 𝑡𝑟𝑒𝑛𝑑
Problem:
Use the method least squares and fit a straight line trend to the following data given from 82 to
92. Hence estimate the trend values for 1993.

122
Year 82 83 84 85 86 87 88 89 90 91 92
Production 45 46 44 47 42 41 39 42 45 40 48
in quintals
Solution:
Let the line of best fit be
Y=ax + b Take X = x – 1987 and Y = y-42
Then the line of best fit become
Y=ax + b
The normal equations are 𝑥𝑦 = 𝑎 𝑥2 + 𝑏 𝑥
𝑦=𝑎 𝑥 + 𝑛𝑏, 𝑤𝑕𝑒𝑟𝑒 𝑛 = 11
−19
From the table, -19 = 110 a ⇒a = = - 0.17
110
X= Y= y-42
X x-1987 Y XY 𝑋2
-5 3 -15 25
1982 45
-4 46 4 -16 16
1983
-3 44 2 -6 9
1984
-2 47 5 -10 4
1985
-1 42 0 0 1
1986

123
0 41 -1 0 0
1987
1 39 -3 -3 1
1988
2 42 0 0 4
1989
3 45 3 9 9
1990
4 40 -2 -8 18
1991
5 48 6 30 25
1992
0 - 17 -19
1993 110
17
17 = 11𝑏 ⇒b11 = 1.55
∴ The line of best fit is Y= - 0.17 x + 1.55
ie. 𝑌 – 42 = − 0.17 (𝑥 − 1987) + 1.55
𝑦 = −0.17𝑥 + 1987 × 0.17 + 1.55 + 42
𝑦 = −0.17𝑥 + 381.34 𝑖𝑠 𝑡𝑕𝑒 𝑠𝑡𝑟𝑎𝑖𝑔𝑕𝑡 𝑙𝑖𝑛𝑒 𝑡𝑟𝑒𝑛𝑑
From the line trend
When x =1982, y=44.4
X = 1983, y=44.23, x=1984, y=44.06
X= 1985, y=43.89, x= 1986, y=43.72
X=1987, y= 43.55, x=1988, y=43.38 X=1989, y=43.21, x=1990, y= 43.04

124
X = 1991, y = 42.87, x= 1992, y=42.7
Thus the trend values are 44.4, 44.23, 44.06, 43.89, 43.72, 43.58, 43.38, 43.21, 43.04,
43.04, 42.87, 42.7
Problem:
Calculate the seasonal variation indices from the following data

125
Seasonal
Month Monthly sales in lakhs Total 𝑥𝑖 indices

𝑥𝑖
𝑥
x 100
I II III IV
1991 1992 1993 1994
January 10 11 11.5 13.5 46 11.5 11.5

12
x100=95.8
February 8.5 8.5 9 10 36 9 9

12
x100 = 75
March 10.5 12 11 12.5 46 11.5 11.5

12
x100 = 95.8
April 12 14 16 18 60 15 15
12
x100 = 125
May 10 9 12 15 46 11.5 11.5

x100 = 95.8
12
June 10.5 10.5 11 14 46 11.5 11.5

x100 = 95.8
12
July 12 14 13 17 56 14 14
x100 = 116.7
12
August 9 8 11 16 44 11 11
x100 = 91.7
12
September 11 11 12.5 13.5 48 12 12

12
x100 = 100
October 10 9.5 11.5 13 44 11 11

12
x100 = 91.7
November 11 12.5 10.5 14 48 12 12

x100 = 100
12
December 12 13 15 16 56 14 14
12
x100 = 116.7
Total 144
12

126
I year II III IV V2 VI
production in 4 yearly 4 yearly period trend
quintals moving total moving moving total values (V)/
average 2
1982 45 - - - -
1983 46 - - - -
1984 44 182 45.50 90.25 -
1985 47 179 44.75 88.25 45.13
1986 42 174 43.50 85.75 44.13

1987 41 169 42.25 83.25 42.88
1988 39 164 41.00 82.75 41.63
1989 42 167 41.75 83.25 41.38

1990 45 166 41.50 85.85 41.63
1991 40 175 43.75 - 42.93
1992 48 - - -
Problem:
Compute the trend values by the method of A yearly moving average for the data
given in problem 1.
Problem:
Determine the suitable period of moving average for the data given in problem 1

127
We observe that the data has peaks at the following years 1983, 1985, 1985, 1990 and
1992.
Thus the data shows 3 cycles with varying periods 2,5,2 respectively.
Hence the suitable period of moving average is taken to be the A.M.
periods.
2+5+2
Hence = 3 is the period of moving average.
5
Problem:
Compute the seasonal indices for the following data by simple average method

128
Princes in Season
different season 1990 1991 1992 1993 1994
Summer 68 70 68 65 60
Monson 60 58 63 56 55
Autumn 61 56 68 56 55
winter 63 60 67 55 58
Solution:
Year Summer Monsoon Autumn Winter Total

1990
68 60 61 63
1991
70 58 56 60
1992
68 63 68 67
1993
65 56 56 55
1994
60 55 55 58
total
331 292 296 303
average 244.4
66.2 58.4 59.2 60.6
Seasonal index 66.2 58.4 59.2 60.6 𝑥 = 61.1
x100 x100 = x100 = x100
61.1 61.1 61.1 61.1
= 108.3 95.6 69.9 = 99.2

129
Exercises:
1. Room the data given below calculate the seasonal indicates assuming that trend is
absent
Year I quarter II quarter III quarter IV quarter

40 35 38 40
1990
42 37 39 38
1991
41 35 38 40
1992
45 36 36 41
1993
44 38 38 42
1994
2. Compute the seasonal index for the following data assuming that there is no need to
adjust the data for the trend
Quarter 1989 1990 1991 1992 1993 1994

I 3.5 3.5 3.5 4.0 4.1 4.2
II 3.9 4.1 3.9 4.6 4.4 4.6
III 3.4 3.7 3.7 3.8 4.2 4.3
IV 3.6 4.8 4.0 4.5 4.5 4.7

130
UNIT IV: SAMPLING
Sampling - Definition - Large samples. Small samples - Population with one samples and
population with two samples - Students – t – test - Applications - chi - square test and goodness
of fit - applications.
TESTS OF SIGNIFICANCE (Large sample)
INTRODUCTION:
Any statistical investigation usually deals with the study of some characteristics of a
collection of objects
SAMPLING:
Definition:
A finite subset of population is called a sample and the number of objects in a sample is
called the sample size.
Some of the important types of sampling are (i) purposive sampling (ii) Random
sampling (iii) Simple sampling (iv) stratified sampling.
(i)Purposive sampling:
If the sample elements are selected with a definite purpose in mind then the sample
selected is called purposive sample.
(ii) Random sampling:
A random sample is one in which each element of the population has an equal
chance of inclusion in the sample.
(iii) Simple sampling:
Simple sampling is a special type of random sampling in which each element of the
population has an equal and independent chance of being included in the sample.

131
(iv) Stratified sampling:
The sample which is the aggregate of the sampled individuals of each stratum is called
stratified sample and the technique of selecting such sample is called stratified sampling.
TESTS OF SIGNIFICANCE FOR LARGE SAMPLES
I. Tests for proportion or percentage
(A) Single proportion (B) Difference of proportions.
II. Tests for means.
(A) (i) Test for single mean if standard deviation of the population σ is
known. (ie) 𝐻0∶ 𝜇 = 𝜇0 , 𝜍 is known.
(B) (ii) Tests for single mean if σ is not known 𝐻0 : 𝜇 = 𝜇0 ,σ is
unknown.
(C) (i) Test for equality of means of 2 normal populations with
Known standard deviations (ie) 𝐻0 : 𝜇1 = 𝜇2 ; 𝜍1 , 𝜍2 is known.
(ii) Test for equality of means of 2 normal populations with same standard
deviation though unknown 𝐻0 : 𝜇1 = 𝜇2, 𝜍1 = 𝜍 = 𝜍2 .
III. Test for standard deviations.
(A) : Test for single standard deviation 𝐻0 : 𝜍 = 𝜍0
(B) : Test for equality for 2 standard deviation (ie)𝐻0 : 𝜍1 = 𝜍2
1. Test of significance for proportions and percentages.

I(A) Single proportions.
If X is the number success in independent trials with constant probability
of success P for each trial we have E(X)=nP
and V(X)=variance(X)=nPQ where Q=1-P. It has been proved that for large n,
the binomial distribution tends to a normal distribution. Hence for large n,
𝑋 ~𝑁(𝑛𝑝, 𝑛𝑃𝑄)

132
𝑋−𝐸 𝑋 𝑋 − 𝑛𝑃
∴𝑍= = ~𝑁 0,1
𝑆. 𝐸 𝑜𝑓 𝑋 𝑛𝑃𝑄
I (B)Difference of proportions.
Suppose we want to compare 2 distinct populations with regard to possession of an

attributes. Let a sample of size 𝑛1 be chosen from the first population and sample of size 𝑛2 be
chosen from the second population.
Let 𝑋1 be number of persons possessing the attribute A in the first sample and 𝑋2 be the number
of persons possessing the same attribute in the second sample
𝑋1 𝑋2
𝑝1 = ; 𝑝2 =
𝑛1 𝑛2
As before 𝐸(𝑝1 ) = 𝑃1 and 𝐸 𝑝2 = 𝑃2 where 𝑃1 𝑎𝑛𝑑 𝑃2 are the proportions in
𝑃1 𝑄1 𝑃2 𝑄2
the populations.𝑉 𝑝1 = 𝑎𝑛𝑑 𝑉 𝑝2 = .
𝑛1 𝑛2
𝑝 1 −𝑝 2
∴𝑍= 1 1
~ 𝑁(0,1)
𝑃𝑄 +
𝑛1 𝑛2
An unbiased estimate of population proportion P based on both the samples is

(𝑛 1 𝑝 1 +𝑛 2 𝑝 2 )
given by 𝑃 = .Suppose the population proportion𝑠 𝑃1 𝑎𝑛𝑑𝑃2 are given to be
𝑛 1 +𝑛 2
different (ie)𝑃1 ≠ 𝑃2.
𝑝1 − 𝑝2 − (𝑃1 − 𝑃2 )
𝑍= ~𝑁(0,1)
𝑃1 𝑄1 𝑃2 𝑄2
𝑛1 + 𝑛2
Problem:
A coin is tossed 144 times and a person gets 80 heads. Can we say that the coin is unbiased one?
Solution:
Set the null hypothesis 𝐻0 : the coin is unbiased. Given n=144.
Probability of getting a head in a toss P=1/2.HenceQ=1/2.Let X=number of

successes=number of getting heads=80.
1
80 − 144 2 80 − 72 8
𝑍= = = = 1.33 < 1.96.
1 1 36 6
144 2 2
Since │Z│<1.96. We accept the hypothesis at 5% level of significance.

133
Hence the coin is unbiased.
Problem:
A die is thrown 10000 times and a throw of 1 or 2 was obtained 4200 times. On the assumption
of random throwing do the data indicate an unbiased die?
Solution:
P=Probability of getting 1 or 2=1/3 .Hence Q=2/3
Given n=10000,X=4200.The null hypothesis 𝐻0 :the die is unbiased.
1
𝑋 − 𝑛𝑃 4200 − 10000 3 4200 − 3333.3 866.7
∴𝑍= = = = = 18.4
𝑛𝑃𝑄 2 47.14 47.14
10000(9)
Since │z│>3, 𝐻0 is rejected and hence the die is biased one.
Problem:
A manufacturer claimed that at least 95% of the equipment which he supplied to a factory
conformed to specification. An examination of a sample of 200 pieces of
equipment revealed that 18 were faulty. Test his claim at a significant level of (i)5%
(ii)1%.
Solution:
Out of a sample of 200 equipments 18 were faulty.
X=200-80=182
182
𝑝= = 0.91.
200
Set the null hypothesis 𝐻0 :P=0.95,Q=0.05.𝐻1 : 𝑃 < 0.95.
𝑝−𝑃 . 91 − .95 . 04
𝑍= = =− = −2.6
𝑃𝑄 . 95 × .05 . 0154
𝑛 200
(i) Since the alternative hypothesis is left tailed and the significant
value of Z at 5% level of significant for left tail is -1.645.
Z=-2.6<-1.645.

134
Hence we accept the null hypothesis at 5% level of significance.
(ii) The critical value of Z at 1% value of significance for left tailed test is -
2.33 and Z=-2.6<-2.33.Hence 𝐻0 is accepted at 1% level.
Problem:
A sample of 1000 products from a factory are examined and found to be 2.5% defective. Another
sample of 1500 similar products from another factory are found to have only 2% defective. Can
we conclude that the products of the first factory are inferior to those of the second?
Solution:
Given 𝑛1 = 1000, 𝑛2 = 1500. Proportion of defectives in the first factory

𝑝1 = 25 1000 = .025 𝑝2 = 30 1500 = .020
Proportion of defective in the second factory 𝑝2 = 30 1500 = .020
𝑛1 𝑝1 + 𝑛2 𝑝2 25 + 30
∴𝑃= = = .022
𝑛1 + 𝑛2 1000 + 1500
Hence Q=1-.022=.978.
Null hypothesis 𝐻0 : 𝑃1 = 𝑃2
𝑝 1 −𝑝 2
Test hypothesis 𝑍 = 1 1
~𝑁(0,1)
𝑃𝑄 +
𝑛1 𝑛2
. 025 − .020 . 005

𝑍= = = .83 < 1.9
1 1 . 022(.978)
. 022(.978) 1000 +
1500 600
The difference of proportion is not significant on 5% level. Hence this

hypothesis is accepted and the two factories are producing similar products. Hence
one is not inferior to the other.
Problem:
A machine puts out 16 imperfect articles in a sample of 500 articles. After the machine
overhauled it. Puts out 3 defective articles in sample of 100.Has the machine improved?
Solution:
Given 𝑛1 = 500, 𝑛2 = 100. 𝑝1 =Proportion defectatives in the first

sample=16/500=.032
3
𝑝2 = = .03
100

135
Set the null hypothesis 𝐻0 : 𝑃1 = 𝑃2
Alternative hypothesis 𝐻1 : 𝑃1 > 𝑃2
𝑛1 𝑝1 + 𝑛2 𝑝2 16 + 3 19
∴𝑃= = = = .032 ; 𝑄 = 1 − .032 = .968
𝑛1 + 𝑛2 500 + 100 600
𝑝1 − 𝑝2 . 032 − .030 . 002 . 002

𝑍= = = =
1 1 1 1 . 019
𝑃𝑄 𝑛 + 𝑛 . 032 × .968 + . 032 × 6 500
1 2 500 100
= .105.
Sine Z<1.645 it is not significant at 5% level of significance. Hence we can

accept the null hypothesis and conclude that the machine has not been improved.
II (B) Test of significance for difference of sample means.
Consider two different normal populations with 𝜇1 and 𝜇2 and s.d 𝜍1 and 𝜍2 respectively. Let a
sample of size 𝑛1 be drawn from the first population and an independent sample of size 𝑛2 be
drawn from the second population. Let 𝑥1 be the mean of the first sample from the first
population and 𝑥2 be the mean of second sample from the second population. If the sample sizes
𝜍12
are large we know 𝑥1 is a normal variate with mean 𝜇1 and variance and 𝑥2 is an independent
𝑛1
𝜍22
normal variate and normal variate with mean 𝜇2 and variance .
𝑛2
𝑥 1 −𝑥 2
The test statistic becomes 𝑍 = which can be tested at any level of
𝜍2 2
1 +𝜍 2
𝑛1 𝑛2
significance.
Problem:
The number of accidents per day were studied for 144 days in Madras city and for 100 days in
Delhi city. The mean numbers of accidents and the s.ds were respectively 4.5 and 1.2 for Madras
city and 5.4 and 1.5 for Delhi city. Is Madras city more prone to accidents than Delhi city?
Solution:
Given 𝑛1 = 144 ; 𝑥1 = 4.5; 𝑥2 = 5.4.

𝑛2 = 100; 𝜍1 = 1.2; 𝜍2 = 1.5.
Set the null hypothesis 𝐻0 : 𝜇1 = 𝜇2

136
𝑥1 − 𝑥2 4.5 − 5.4
𝑍= = = −4.99.
𝜍12 𝜍22 1.22 1.52
+ 𝑛 144 + 100
𝑛1 2
∴ │𝑍│ = 4.99 > 3 we reject the hypothesis that the two cities have the same
accident rates. However since Delhi city has higher rate of accident than Madras city.
Therefore Delhi more prone to accidents.
Problem:
The mean yields of rice from two places in a district were 210 kgs and 220 kgs per acre from 100
acres and 150 acres respectively. Can it be regarded that the sample were drawn from the same
district which has the s.d of 11kgs per acre?
Solution:
𝑛1 = 100; 𝑥1 = 210; 𝜍 = 11
𝑛2 = 150; 𝑥2 = 220
𝑥1 − 𝑥2
∴𝑍=
𝜍 1 𝑛1 + 1 𝑛2
210 − 220 −10

𝑍= = = −7.04
11 1
100 +
1
150 11 250 15000
𝑍 = 7.04 > 3.The value is highly significant and hence we reject the null hypothesis.
Hence the samples are certainly not from the same district with the s.d 11.
Test of significance for equality of standard deviations of a normal population .

If we want to test whether the two independent samples with known standard deviations
𝑠1 𝑎𝑛𝑑 𝑠2 have come from the same population with standard deviation σ. Under the hypothesis
𝑠1 −𝑠2
𝐻0 : 𝜍1 = 𝜍2 the test statistics is 𝑍 = .
𝜍 1 1
2𝑛 1 + 2𝑛 2

137
Problem:
The s.d of weight of all students in a first grade college was found to be 4 kgs. Two samples are
drawn. The s.ds of the weight of 100 undergraduate students is 3.5kgs and 50 post graduate
students are 3 kgs. Test the significance of the difference of standard deviations of the samples at
5% level.
Solution:
Given 𝑛1 = 100; 𝑠1 = 3.5; 𝜍 = 4; 𝑛2 = 50; 𝑠2 = 3.
Set the null hypothesis 𝐻0 : 𝜍1 = 𝜍2. Then 𝐻1 : 𝜍1 ≠ 𝜍2
𝑠1 − 𝑠2 3.5 − 3
∴𝑍= = = 1.02
𝜍 1 + 1 2𝑛 4 1 1
2𝑛1 2 200 + 100
│𝑍│ =1.02<1.96. It is not significant at 5% level of significance.
Problem:
The mean production of wheat of a sample of 100 plots is 200kgs per acre with s.d of 10 kgs.
Another sample of 150 plots gives the mean production of wheat as 220kgs. With s.d of 12kgs.
Assuming the s.d of the 11kgs for the universe find at 1% level of significance ,whether two
results are consistent.
Solution:
Given σ=11 and
size mean S.D

𝑛1 = 100 𝑥1 = 200 𝑠1 = 10
Sample 1
𝑛2 = 150 𝑥2 = 220 𝑠2 = 12
Sample 2
Set the null the hypothesis 𝐻0 : 𝜇1 = 𝜇2 and 𝜍1 = 𝜍2 .For 𝐻0 : 𝜇1 = 𝜇2

138
𝑥1 − 𝑥2 200 − 220 −20 −155
𝑍= = = = = −14.1
11
𝜍 1
𝑛1 +
1
𝑛2 11 1
100 +
1
150 11 10 600
∴ │𝑍│ =14.1>3.Hence the two means differ significantly at 5% level even a 1% level.
For 𝐻0 : 𝜍1 = 𝜍2 .
𝑠1 −𝑠2 10−12
𝑍= = = −1.99
𝜍 1 1 1 1
2𝑛 1 + 2𝑛 2 11 200 + 300
∴ │𝑍│ = 1.99 > 1.96 and │𝑍│ = 1.99 < 2.58.
Hence the difference of s.d is significant at 5% level and not significant and 1% level.
∴At 1% level the difference between s.d is not significant but between means it is
significant. Hence we can conclude that at 1% level the two results are not consistent.
TEST OF SIGNIFICANCE (SMALL SAMPLES)
TEST OF SIGNIFICANCE BASED ON t-DISTRIBUTION (t-test)
Consider a normal population with mean µ and s.d σ . Let 𝑥1 , 𝑥2 , … 𝑥𝑛 . be a random

𝑥 −𝜇
sample of size n with mean 𝑥 and standard deviation s. We know that 𝑍 = 𝜍 is the
𝑛
standard normal variate N(0,1).
Hence the test statistics is in small sample becomes

𝑥 −𝜇 𝑥 −𝜇 𝑥 −𝜇
𝑍= = 𝑠 . Now let us define 𝑡 = 𝑠 .
𝑠 𝑛 𝑛−1 / 𝑛 𝑛 −1 𝑛−1
This follows students‟s t-distribution with n-1 degrees of freedom
1.Test for the difference between the mean of a sample and that of a population
Under the null hypothesis 𝐻0 : 𝜇 = 𝑥 .
The test statistic

139
𝑥 −𝜇
𝑡=𝑠 ~𝑡𝑛−1. Which can be tested at any level of significance with n-1
𝑛−1
degrees of freedom.
II. Test for the difference between the means of two samples
II.A. If 𝑥1 and 𝑥2 are the means of two independent samples of sizes 𝑛1 and 𝑛2 from a normal
𝑥 1 −𝑥 2
population with mean µ and standard deviation σ. It found that ~𝑁 0,1 .
𝜍 1 1
𝑛1 + 𝑛2
𝑥1 − 𝑥2
𝑡=
𝑛1 𝑠12 + 𝑛2 𝑠22 1 1
𝑛1 + 𝑛2 − 2 𝑛1 + 𝑛2
Which follows a t-distribution with d.f 𝑣 = 𝑛1 + 𝑛2 − 2 .
II.B. suppose the sample sizes are equal (ie) 𝑛1 = 𝑛2 = 𝑛.Then we have 𝑛 pairs of values.
Further we assume that the 𝑛 pair are independent .Then the test statistic 𝑡 in (1) becomes
𝑥 1 −𝑥 2
𝑡= .
𝑛 𝑠2 2
1 +𝑠 2 2
2𝑛 −2 𝑛
𝑥 1 −𝑥 2
∴𝑡= is a students 𝑡 variate with
𝑠12 +𝑠22
𝑛−1
𝑣 = 𝑛 + 𝑛 − 2 = 2𝑛 − 2.
II. (C) suppose the sample size are equal and if then n pairs of values in this case are not
independent.
𝑥 −𝜇
The test statistic 𝑡 = 𝑠 to test whether the means of differences is
𝑛−1
significantly different from zero. In this case the d.f is n-1.
Confidence limits (Fiducial limits). If𝜍 is not known and 𝑛 is small then
𝑠𝑡 .05 𝑠𝑡 .05
1. 95% confidence limits for µ is 𝑥 − ,𝑥 +
𝑛−1 𝑛−1
𝑠𝑡 .01 𝑠𝑡 .01
2. 99% confidence limits for µ is 𝑥 − ,𝑥 +
𝑛−1 𝑛−1

140
Problem:
A random sample of 10 boys has the following I.Q (intelligent quotients). 70, 120, 110, 101, 88,
95, 98, 107, 100. Do these data support the assumption of a population mean of a population
mean I.Q of 100?
Solution:
Given n=10; µ=100 . Set 𝐻0 : µ=100

𝑥 −𝜇
Under 𝐻0 test statistics 𝑠 ~𝑡 𝑛 −1 where𝑥 and 𝑠 can be calculated from the sample
𝑛−1
data as 𝑥 = 972 10 = 97.2 and
2
2
𝑥𝑖 − 𝑥 1833.60
𝑠 = = = 183.36.
𝑛 10
Hence 𝑠 =13.54.
97.2 − 100 −2.8 × 3

∴𝑡= = = −.6204
13.54 13.54
9
∴ │𝑡│ = .62(nearly).
The table value for 9 d.f at 5% level of significance is 𝑡.05 =2.26
∴ │𝑡│ = .62 < 𝑡.05. Hence the difference is not significant at 5% level. Hence 𝐻0 may
be accepted at 5% level hence the data support the assumption of population mean 100.
Problem:
It was found that a machine has produced pipes having a thickness .05 mm. to determine whether
the machine is in proper working order a sample of 10 pipe is chosen for which the mean
thickness is .53mm and s.d is 0.3mm .test the hypothesis that the machine is in proper working
order using a level of significance of (1) .05 (2) .01
Solution :
Given µ= .50,𝑥=.53;𝑠 = .03; 𝑛 = 10.
Set the null hypothesis 𝐻0 :µ=50

141
𝑥 −𝜇 .53−.50
Under the null hypothesis the test statistic is 𝑡 = 𝑠 = × 9
.03
𝑛−1
. 03 × 3
= = 3.
. 03
(i)The table value for 𝑣 = 9d.f at 5% level of significance is 𝑡.05 =2.26
(ie)│t│=3>𝑡.05.
∴The difference is significant at 5% level of significance.
∴The null hypothesis is rejected at 5%level of significance .
(ii) The table value for 𝑣 = 9 d.f at 1% level of significance is𝑡.01 = 3.25.
Hence │t│=3< 𝑡.01.
∴The difference is not significant at 1% level of significant .

Hence the null hypothesis is accepted at 1% level of significance.
Problem:
A group of 10 rats fed on a diet A and another group of 8 rats fed on a different diet B recorded
the following increase in weight in gms.
Diet 5 6 8 1 12 4 3 9 6 10
A
Diet 2 3 6 8 1 10 2 8 - -
B
Test whether diet A is superior to diet B .
Solution :
Given 𝑛1 = 10; 𝑛2 = 8.
5+6+⋯+10 64
Mean of the first sample 𝑥1 = = 10 = 6.4.
10
2+3+⋯+8 40
Mean of the second sample 𝑥2 = = = 5.0.
8 8
Standard deviation 𝑠1 and 𝑠2 of the first and second sample can be found as

142
𝑥2 𝑥 2
𝑠12 = − =10.24 and 𝑠22 = 10.25
𝑛 𝑛
𝑥1 − 𝑥2 6.4 − 5
𝑡= =
10 × 10.24 + 8 × 10.25 1 1
𝑛1 𝑠12 + 𝑛2 𝑠22 1 1
10 + 8 − 2 10 + 8
𝑛1 + 𝑛2 − 2 𝑛1 + 𝑛2
1.4
= = .875.
11.525 . 1 + .125
Table value for t at 5% level of significance for 𝑛1 + 𝑛2 − 2 = 16 𝑑. 𝑓 𝑖𝑠

𝑡.05 =2.12.
Since t=.875 <𝑡.05 the difference is not significant at 5% level of significance .
Hence the null hypothesis may be accepted.
Problem:
The table gives the biological values of protein from 6 cows milk and 6 buffalo‟s
milk . Examine whether the differences are significant .
Cow‟s milk Buffalo‟s milk
1.8 2.0
2.0 1.8
1.9 1.8
1.6 2.0
1.8 2.1
1.5 1.9
Solution:
Mean value of protein of cow‟s milk =1.6
Mean value of protein of buffalo‟s milk =1.9
Variance of protein of cow‟s milk =.03
Variance of protein in buffalo‟s milk=0.1

143
We notice that the two sets of observations are independent .
Given 𝑛1 = 𝑛2 = 6; 𝑥1 = 1.8; 𝑥2 = 1.9; 𝑠12 = .03;
𝑠22 = .01.
𝑥 1 −𝑥 2
Set null hypothesis 𝐻0 : 𝑥1 = 𝑥2 . Under this null hypothesis the test statistic is 𝑡 =
𝑠2 2
1 +𝑠 2
𝑛 −1
and the d.f 𝑣 = 2𝑛 − 2 = 10.
−.1 −.1
= = = −1.11.
. 03 + .01 . 04
5 5
The table value for 𝑣 = 10d.f at 5% level of significance
is 2.23.
│t│=1.11<2.23.Hence the difference is not significant .
Hence the hypothesis is accepted .
Problem:
Ten soldiers participated in a shooting competition in the first week. After intensive training they
participated in the competition in the second week. Their scores before and after coaching were
given as follows.
Soldiers 1 2 3 4 5 6 7 8 9 10
Score 67 24 57 55 63 54 56 68 33 43
before(x)
Score 70 38 58 58 56 67 68 75 42 38
after(y)
Do the data indicate that the soldier have been identified by the training ?
Solution:
Here we are connected with the same set of the soldiers in the 2 competitions and their scores
which are related to each other because of the intensive training .we compute the difference in
their scores 𝑧 = 𝑦 − 𝑥and calculate the mean 𝑧 and the s.d 𝑧 as follow

144
2
𝑧=𝑦−𝑥 𝑧−𝑧 𝑧−𝑧
𝑥 𝑦
67 70 3 -2 4
24 38 14 9 81
57 58 1 -4 16
55 58 3 -2 4
63 56 -7 -12 144
54 67 13 8 64
56 68 12 7 49
68 75 7 2 4
33 42 9 4 16
43 38 -5 -10 100
- 50 482
- -
50 𝑧−𝑧 2 482
𝑧=10 = 5; 𝑠 2 = = = 48.2
10 10
Set the null hypothesis 𝐻0 : 𝑧=0.

𝑧 −0 5 15
Under the null hypothesis the test statistic is 𝑡 = 𝑠 = × 9 =6.94 = 2.16
48.2
𝑛−1
The table value for 𝑣 = 9d.f at 55 level of significance is 𝑡.05 = 2.26.
∴ │𝑡│ = 2.16 < 𝑡.05.
The difference is not significant on 5% level of significance .
Hence the null hypothesis is accepted .We can conclude that there is no significant
improvement in the training .

145
TEST BASED ON 𝝌𝟐 - DISTRIBUTION
INTRODUCTION:
The 𝜒 2 distribution has number of application in statistics. It has three important

applications based on 𝜒 2 distribution.
I. 𝜒 2 - test for population variance.
II. 𝜒 2 -test to test the goodness of fit.
III. 𝜒 2 -test to test the independence of attributes.
𝝌𝟐 -TEST.
𝑰. 𝝌𝟐 -test for population variance
Let 𝑥1 , 𝑥2 , … 𝑥𝑛 be a random sample from a normal population with variance 𝜍 2 .Set

𝑛 𝑥 𝑖 −𝑥 2 𝑛𝑠 2
the null hypothesis 𝐻0 : 𝜍 2 = 𝜍02 . Then the test statistic is 𝜒 2 = 𝑖=1 =
𝜍0 𝜍02
𝑛𝑠 2
where 𝑠 2 is the variance of the sample. Then 𝜒 2 = 𝜍 2 defined above follows a𝜒 2
0
distribution with 𝑛 − 1 degrees of freedom.
Problem:
A random sample of size 25from a population gives the sample standard deviation 8.5.Test the
hypothesis that the population s.d is 10.
Solution:
Given σ=10,n=25,s=8.5 𝐻0 : 𝜍 = 10
𝑛𝑠 2 25×8.52
𝜒2 = = = 18.06.
𝜍02 100
The table value of𝜒 2 for 24 d.f =36.415 at 5% level of significance.
It is not a significant. Hence the null hypothesis is accepted.

146
Problem:
Test the hypothesis that σ=8 given that s=10 for a random sample of size 51.
Solution:
Given 𝑛1 =51,σ=8,s=10.
Let 𝐻0 : 𝜍 = 8.
𝑛𝑠 2 51×10 2
𝜒2 = = = 79.7.
𝜍02 82
Since𝑍 = 2𝜒 2 - 2𝑛 − 1= 2 × 79.7- 2 × 51 − 1
=2.58
│𝑧│ = 2.58 > 1.96.
Hence the difference is significant at 5% level of significance and hence the

hypothesis is rejected at 5% level of significance.
𝟐
-TEST TO TEST THE GOODNESS OF FIT
𝑰𝑰. 𝝌
The 𝜒 2 -distribution can be used to test the goodness of fit. This test can also be applied
to test for compatibility of observed frequencies and theoretical frequencies. Let
𝑜1 , 𝑜2 , … 𝑜𝑛 be the observed frequencies and 𝑒1 , 𝑒2 , … 𝑒𝑛 be the corresponding expected
𝑛 𝑛
frequencies such that 𝑖=1 𝑜𝑖 =𝑁= 𝑖=1 𝑒𝑖 where N is the number of members in the
population.
𝑛 𝑜 𝑖 −𝑒 𝑖 2
Define 𝜒 2 = 𝑖=1 .It is a 𝜒 2 variable with n-1 degrees of freedom.
𝑒𝑖

147
Problem:
The theory predicts that the proportion of an object available in four groups A,B,C,D should be
9:3:3:1. In an experiment among 1600 items of this object the members in the four groups were
882,313,287 and188.use 𝜒 2 -test to verify whether the experimental result support the theory.
Solution:
The observed frequencies 𝑜𝑖 are 882,313,287,118.
𝑜𝑖 =882+313+287+118=1600
The expected frequencies are in the ratio 9:3:3:1.
∴ The expected frequencies 𝑒𝑖 are 900,300,300,100.
𝑒𝑖 =1600= 𝑜𝑖 .
𝑜 𝑖 −𝑒 𝑖 2
∴ 𝜒2 = .
𝑒𝑖
882−900 2 313−300 2 287−300 2 118−100 2

= + + + = 4.7266
900 300 300 100
Degrees of Table value of 𝜒 2 for 3 d.f at 5% level of significance is 7.851.
Calculated 𝜒 2 =4.7266<7.852=table value of 𝜒 2 .It is not significant. Hence the null

hypothesis may be accepted at 5% level of significance and hence we may conclude that
experiment results support the theory.
Problem:
Fit a poisson distribution for the following data and test the goodness of fit.
x 0 1 2 3 4 5 6 Total
f 273 70 30 7 7 2 1 390

148
Solution:
𝑓𝑥 70+60+21+28+10+6 195
𝑥= =273+70+30+7+7+2+1 = 390
𝑓
𝜆 = 1 2.
The theoretical frequencies of the poisson distribution are given by
𝑥
𝑁𝑒 −𝜆 𝜆𝑥 390 1
𝑓 𝑥 = = ; 𝑥 = 0,1,2, … 6.
𝑥! 𝑒𝑥! 2
freedom =4-1 =3 .
390 1 0
The expected frequencies are by𝑓 0 = 𝑒 2
= 236.4;
390 1 1
𝑓 1 = = 118.2 ………………….,
𝑒1! 2
390 1 6
𝑓 6 = 𝑒6! 2
= 0.005.
Thus the observed and expected frequencies can be shown below
𝑜𝑖 273 70 30 7 7 2 1 390
𝑒𝑖 236.4 118.2 29.5 4.9 .6 .1 0 389.7
Since the sum of the expected frequencies is 389.7.It can be adjusted in the last two
frequencies by adding .3.
𝑜𝑖
273 70 30 17 390
𝑒𝑖 236.4 118.2
29.5 5.9 390
Set up the null hypothesis 𝐻0 :Poisson distribution can be fitted well.

149
𝑜 𝑖 −𝑒 𝑖 2
The test statistics is 𝜒 2 = .
𝑒𝑖
2 2 2 2
273 − 236.4 70 − 118.2 30 − 29.5 17 − 5.9
= + + + = 46.3.
236.4 118.2 29.5 5.9
Degrees of freedom=7-1-1-3=2.
The table value of 2 d.f. at 5% level is 5.99.
2
Since 𝜒 2 =46.3>5.99=The table value of 𝜒.05 it is much significant at 5% level of significance.
Hence the hypothesis is rejected at 5% level and hence the poisson distribution is not a good fit
to the data.

150
UNIT V: INDEX NUMBERS
Index Numbers - Types of index numbers - Tests - Unit test commodity reversal test, time
reversal test, factor reversal test - Chain index numbers - cost of living index – Interpolation -
Finite differences operators - Newton’s forward, backward interpolation formulae, Lagrange’s
formula.
INDEX NUMBERS
Index Numbers :
An index number is widely used statistical device for comparing the level of a certain
phenomenon with the level of the same phenomenon at some standard period.
In the computation of an index number, if the base year used for comparison is kept constant
throughout, then it is called fixed base method. If on the other hand, for every year the previous
year is used as a base for comparison, then the method is called chain base method.
Index numbers can be broadly classified into two types.
I. Unweighted or simple index number

II. Weighted index number.
Two standard methods of computation are
A) Aggregate method
B) Average of price relatives method.
I. A). Aggregate method:
In this method total of current prices for various commodities is divided by the total of the
base year. In symbols if 𝑝0 denotes the price of the base year and 𝑝1 the price of the current
year.
𝑃1
𝑝01 = 𝑥 100 𝑤𝑕𝑒𝑟𝑒 𝑃1 is total of the current year
𝑃0
𝑝0 𝑖𝑠 𝑡𝑕𝑒 𝑡𝑜𝑡𝑎𝑙 𝑜𝑓 𝑡𝑕𝑒 𝑏𝑎𝑠𝑒 𝑦𝑒𝑎𝑟.

151
Example :
From the following data construct the simple aggregative index number for 1992.
Price in Price in
Commodities 1991 1992
Rs Rs
Rice 7 8
Wheat 3.5 3.75
Oil 40 45
Gas 78 85
Flour 4.5 5.25
Solution :
Construction of price index taking 1991 as base year.
Price in 1991 Price in 1992

Commodities
Rs Rs
Rice 7 8
Wheat 3.5 3.75
Oil 40 45
Gas 78 85
Flour 4.5 5.25
Total 133.00 147.00
𝑃1
∴aggregate index number 𝑃01 = × 100
𝑃0
147
= × 100
133
= 110.5

152
I. B) Average of Price Relatives method (Simple index numbers)
Price relatives denoting the price of a commodity of a base year as 𝑃0 and the price of the
𝑃
current year as 𝑃1 the ratio of the prices 𝑃1 is called the price relatives.
0
𝑃
Index number for the current year is = 𝑃01 = 𝑃1 × 100
0
𝑃1
×100
𝑃0
i) The Arithmetic mean index number 𝑃01 = 𝑛
ii) The Geometric mean index number
1
𝑃1 𝑛
𝑃01 = 𝜋 × 100, 𝑤𝑕𝑒𝑟𝑒 𝜋 𝑑𝑒𝑛𝑜𝑡𝑒𝑠 𝑡𝑕𝑒 𝑝𝑟𝑜𝑑𝑢𝑐𝑡
𝑃0
𝑃
log 𝑃1 × 100
0
𝐻𝑒𝑛𝑐𝑒 log 𝑃01 =
𝑛
Problem:
For the above example, we find the index number of the price relatives taking 1991 as the base
year using i) Arithmetic mean ii) Geometric mean.
Solution:
Price in Price in 𝑷𝟏 𝑷𝟏
𝐥𝐨𝐠
Commodities 1991 1992 𝑷𝟎 𝑷𝟎
𝑷𝟎 𝑷𝟏 × 𝟏𝟎𝟎 × 𝟏𝟎𝟎
Rice
7 8 114.3 2.0580
Wheat 3.5 3.75 107.41 2.0298
Oil
40 45 112.5 2.0512

153
Gas
78 85 109.0 2.0374
Flour
4.5 5.25 116.7 2.0671
Total
559.6 10.2435
i) Using arithmetic mean the index number
559.6
𝑃01 = = 111.92
5
ii) Using geometric mean the index number
10.2435
log 𝑃01 = = 2.048%
5
∴ 𝑃01 = 𝑎𝑛𝑡𝑖𝑙𝑜𝑔 2.0487 = 111.87
Problem:
From the following data of the whole sale price of rice for the 5 years construct the index
numbers taking (i) 1987 as the base (ii) 1990 as the base.
Years 1987 1988 1989 1990 1991 1992

Price
of rice 5.00 6.00 6.50 7.00 7.50 8.00
per kg
Solution :
i) Construction of index numbers taking 1987 as base.
Index
Price of
Years numbers (base
rice per kg
1987)

154
1987 5.00 100
6
1988 6.00 × 100 = 120
5
6.5
1989 6.50 × 100 = 130
5
7
1990 7.00 × 100 = 140
5
7.5
1991 7.50 × 100 = 150
5
8
1992 8.00 × 100 = 160
5
From the index number table we observe that from 1987 to 1988 these is a increase of
20% in the price of rice per kg; for 1987 to 1989 there is a increase of 30% in the price of
rice per kg etc.
ii) Construction of index numbers taking 1990 as base

Price of Index
Years
rice per kg number (Base 1990)
5
1987 5 × 100 = 71.4
7
6
1988 6 × 100 = 85.7
7
6.5
1989 6.50 × 100 = 92.9
7
1990 7 100
7.5
1991 7.5 × 100 = 107.7
7
8
1992 8 × 100 = 114.3
7

155
Problem:
Construct the whole sale price index number for 1991 and 1992 from the data given below using
1990 as the base year.
Whole sale prices in

Rupees per quintal
Commodity
1990 1991 1992
Rice
700 750 825
Wheat
540 575 600
Ragi
300 325 310
Cholam
250 280 295
Flour
320 330 335
Ravai
325 350 360

156
Solution: Taking 1990 as base year
Commodity 1990 1991 1992 Relatives for 91 Relatives for 92
750 825
× 100 × 100
Rice 700 750 825 700 700
= 107.1 = 117.9
575 600
× 100 × 100
Wheat 540 575 600 540 540
= 106.5 = 111.1
325 310
× 100 × 100
Ragi 300 325 310 540 300
= 108.3 = 103.3
280 295
× 100 × 100
Cholam 250 280 295 250 250
= 112 = 118
330 325
× 100 × 100
Flour 320 330 335 320 320
= 103.1 = 101.6
350 360
× 100 × 100
Ravai 325 350 360 325 325
= 107.7 = 110.8
Total 644.7 662.7
Index Number (using AM) 107.5 110.5

157
Problem:
From the following average prices of the three groups of commodities given in rupees per unit
find (i) fixed base index number (ii) chain base index numbers with 1988 as the base year and
Commodity 1988 1989 1990 1991 1992
A 2 3 4 5 6
B 8 10 12 15 18
C 4 5 8 10 12
Solution:
i) Fixed base index number
Commodity 1988 1989 1991 1992

1990
3 5
× 100 × 100 6
A 100 2 2 × 100 = 300
2
= 150 = 250
12
10 8 15
× 100 × 100 18
B 100 8 8
× 100 × 100 = 225
8
= 125 = 188
= 150

158
8
5 4 10
× 100 × 100 12
C 100 4 4
× 100 × 100 = 300
4
= 125 = 200
= 200
Total 300 400 688 825

550
Index number (AM) 100 229.3 275

133.3 183.3
ii) Chain base index numbers:
Commodity 1988 1989 1990 1991 1992
2 3 5 6
2 2 4 4 5
× 100
A 3
× 100 × 100 × 100 × 100
= 133.3
= 100 = 150 = 125 = 120
8 10 12 15 18
8 8 10 12 15
B
× 100 × 100 × 100 × 100 × 100
= 100 = 125 = 120 = 125 = 120
4 5 10 12
4 4 8 8 10
× 100
C 5
× 100 × 100 × 100 × 100
= 160
= 100 = 125 = 125 = 120
Total

159
300 400 413.3 375 360
Index
number
100 133.3 137.8 125 120
(AM)
II. Weighted Index numbers :
Standard methods of computing weighted index number are
II-A weighted aggregative method
II – B weighted average of price relatives method.
II – A weighted aggregative method:
Though there are many formulae to calculate index number in this method we give below
some standard formulae which are very often used.
a) Laspeyre’s index number:
According to Laspeyre‟s method the prices of the commodities in the base year as well as the
current year are known and they are weighted by the quantities used in the base year.
𝑃1 𝑞 0
𝐿𝐼01 = × 100
𝑝0 𝑞0
b) Paasche’s index number:
According to paasche‟s method current year quantities are taken as weights and
hence paasche‟s index number is defined.
𝑃1 𝑞 1
𝐿𝐼01 = × 100
𝑝0 𝑞1

160
C) Marshall – Edgeworth’s Index number:
According to this method the weight is the sum of the quantities of the base
period and current period.
𝑃1 𝑞 0 + 𝑃1 𝑞 1
𝑀𝐼01 = × 100
𝑝0 𝑞0 + 𝑝0 𝑞1
d) Bowley’s Index number:
The arithemetic mean of Laspeyre‟s and paasche‟s index number is

defined to be Bowley‟s index number.
1 𝑃1 𝑞 0 𝑃1 𝑞 1
𝐵𝐼01 = 2 + × 100
𝑝0 𝑞0 𝑝0 𝑞1
e) Fisher’s index number:
𝑃1 𝑞 0 𝑃1 𝑞 1
𝐼01 = × x 100
𝑝0 𝑞0 𝑝0 𝑞1
𝐼01 = 𝐿𝐼01 × 𝑃𝐼01
f) Kelley’s Index number:
According to Kelley, weight may be taken as the quantities of the period which is
not necessarily the
base year or current year. The average quantity of two or more years may be taken as
the weight.
𝑃1 𝑞
𝐾𝐼01 = x 100. Where q is the average quantity of two or more years.
𝑃0 𝑞
Example :
Calculate i) Laspeyre‟s (ii) Paasche‟s iii) Fisher‟s index number for the following data given
below. Hence or otherwise find Edgeworth and Bowley‟s index number.

161
Base year Current
1990 year 1992
Commodities
Quantity Price Quantity
Price
A
2 10 3 12
B
5 16 6.5 11
C
3.5 18 4 16
D
7 21 9 25
E
3 11 3.5 20
Solution:
1 1
Commodities 990 992 𝑷𝟎 𝒒𝟎 𝑷𝟎 𝒒𝟏 𝑷𝟏 𝒒𝟎 𝑷𝟏 𝒒𝟏
𝑷𝟎 𝒒𝟎 𝑷𝟏 𝒒𝟏
A 2 10 3 12 20 24 30 36
B 5 16 6.5 11 80 55 104 71.5
C 3.5 18 4 16 63 56 72 64
D 7 21 9 25 147 175 189 225
E 3 11 3.5 20 33 60 38.5 70
Total 343 370 433.5 466.5
𝑃1 𝑞 0
i) Las Peyre‟s Index number = × 100
𝑝0 𝑞0
433.5
= × 100 = 126.4
343

162
𝑃1 𝑞 1 466.5
ii) Paasche‟s index number 𝑃𝐼01 = × 100 = × 100 = 126.1
𝑝0 𝑞1 370
𝑃1 𝑞 0 𝑃1 𝑞 1
iii) Fisher‟s Ideal index number = × × 100
𝑝0 𝑞0 𝑝0 𝑞1
433.5 × 466.5
= × 100
343 × 370
= 126.2
𝐿𝐼01 + 𝑃𝐼01 126.4+126.1
iv)Bowly‟s Index numbers = =
2 2
= 126.25
𝑃1 𝑞 0 + 𝑃1 𝑞 1
v) Edge – Worth‟s Index number = × 100
𝑝0 𝑞0 + 𝑝0 𝑞1
433.5 + 466.5
= × 100
343 + 370
900
= × 100
713
= 126.2
II – B. Weighted Average of Price Relative Method :
In this method the index number is computed by taking the weighted Arithmetic mean of
price relatives. Thus if P is the price relative and V is the value weights 𝑃0 𝑞0 then the index
𝑃𝑉
number 𝑃01 = 𝑉
Example :
Index number using weighted arithmetic mean of price relatives.
Qua
Price Price
ntity P
in in 𝑉
Commodity in 𝑃1 PV
1990 1992 𝑃0 𝑞0
1990 𝑃0
𝑃0 𝑃1
𝑞0 × 100

163
Rs. Rs. 15
Coconut oil 750 108 8100
50 54 lit.
Groundnut Rs. Rs. 25
1125 106.7 120037.5
oil 45 48 lit.
Rs. Rs. 30
Gingili oil 1290 104.7 135063
43 45 lit.
350
Rice Rs. 7 Rs. 9 2450 128.6 315070
kg
Total 5615 - 651170.5
𝑃𝑉
Weighted index number = 𝑉
𝟔𝟓𝟏𝟏𝟕𝟎.𝟓
= ≅ 𝟏𝟏𝟔
𝟓𝟔𝟏𝟓
Ideal Index number :
An index number is said to be ideal index number if it is subjected to the following three
test.
i) The time reversal test

ii) The factor reversal test
iii)The commodity reversal test
i) The time reversal test :
Let 𝐼(01) denote the index number of the current year 𝑦1 relative to the base year 𝑦0
without considering percentage, and 𝐼(01) denotes the index number of the base year 𝑦0 relative
to the current year 𝑦1 without considering the percentage. If 𝐼(01) × 𝐼 10 = 1, then we say that
the index number satisfies the time reversal test.
ii) The factor reversal test:
In this test the prices and quantities are interchanged, without considering the percentage,
𝑃1 𝑞 1
satisfying the following relation 𝐼(𝑝𝑞 ) × 𝐼(𝑞𝑝 ) = , where 𝐼(𝑝𝑞 ) is the price index of the
𝑝0 𝑞0

164
current year relative to the base year and 𝐼(𝑞𝑝 ) is the quantity index of the current year relative to
the base year.
iii) The Commodity reversal test:
The index number should be independent of the order in which different commodities are
considered. This test is satisfied by almost all index numbers.
Problem:
Construct Fisher‟s index number and show that it statistics both the factor reversal test
and time reversal test.
Commodity aA bB cC dD
Base year price in 55 66 44 33
Rupees
Base year quantity
in Quintals 50 40 120 30
Current year price 7 8 5 4
in Rupees
Current year 35
quantity in Quintals 60 50 110
Solution:
Base Current
Commodity 𝑷𝟎 𝒒𝟎 𝑷𝟎 𝒒𝟏 𝑷𝟏 𝒒𝟎 𝑷𝟏 𝒒𝟏
year year
𝑷𝟎 𝒒𝟎 𝑷𝟏 𝒒𝟏
A 5 50 7 60 250 300 350 420
B 6 40 8 50 240 300 320 400
C 4 120 5 110 480 440 600 550

165
D 3 30 4 35 90 105 120 140
Total 1060 1145 1390 1510
𝑃1 𝑞 0 𝑃1 𝑞 1
Fisher‟s Index number is 𝐼01 = × × 100
𝑝0 𝑞0 𝑝0 𝑞1
1390 1510
= × × 100
1060 1145
Time reversal test
𝑃1 𝑞 0 𝑃1 𝑞 1
Now, 𝐼01 = ×
𝑝0 𝑞0 𝑝0 𝑞1
1390 1060
= ×
1510 1390
1390 1510 1060

Now, ,𝐼 01 × 𝐼 10 = × 1145 × =1
1060 1390
Factor reversal test
𝑃1 𝑞 0 𝑃1 𝑞 1 1390 1510
𝐼 𝑃𝑞 = × = × 1145
𝑝0 𝑞0 𝑝0 𝑞1 1060
Interchanging the factors
𝑝1 𝑞0 𝑃1 𝑞 1
𝐼 𝑃𝑞 = ×
𝑝0 𝑞0 𝑃0 𝑞 1
1145 1510
= ×
1060 1390
1390 1510 1145 × 1510

Now, 𝐼 𝑝𝑞 × 𝐼 𝑞𝑝 = × 1145 ×
1060 1060 × 1390
1510 𝑃1 𝑞 1
=1060 = 𝑝0 𝑞0

166
Hence the factor reversal test is also satisfied.
Problem:
Find the missing price in the following data if the ratio between Laspeyre‟s and Paasche‟s index
numbers is 25:24.
Current
Base Year
year
Commodities
Price Quantity Price Quantity

A 1 15 2 15
B 2 15 - 30
Solution :
Let the missing price be x
Base Current
Commodities 𝑷𝟎 𝒒𝟎 𝑷𝟎 𝒒𝟏 𝑷𝟏 𝒒𝟎 𝑷𝟏 𝒒𝟏
year year
𝑷𝟎 𝒒𝟎 𝑷𝟏 𝒒𝟏
A 1 15 2 15 15 15 30 30
B 2 15 X 30 30 60 15x 30x
Total 45 75 30+15x 30+30x
Laspeyre‟s index number
𝑃1 𝑞0
𝐿𝐼 01 = × 100
𝑝0 𝑞0

167
30+15𝑥
= × 100
45
Paache‟s index number,
𝑃1 𝑞1 30 + 30𝑥
𝑃𝐼 01 = × 100 = × 100
𝑝0 𝑞1 75
Given 𝐿𝐼 01 ∶ 𝑃𝐼 01 = 25 ∶ 24
30 + 15𝑥 30 + 30𝑥
∴ × 100 ∶ × 100
45 75
= 25:24
30+15𝑥 30+30𝑥
∴ 24 = 25
45 75
72 30 + 15𝑥 = 45 30 + 30𝑥
8 30 + 15𝑥 = 5 30 + 30𝑥
8 × 15 2 + 𝑥 = 5 × 30 1 + 𝑥
4 2+𝑥 =5 1+𝑥
∴ 8 + 4𝑥 = 5 + 5𝑥
5𝑥 − 4𝑥 = 8 − 5
𝑥=3
Hence the missing price is Rs. 3
Interpolation:
Definition:
Interpolation is the process of finding the most appropriate estimate for missing data. It is the art
of reading between the lines of a table.
It is also possible that we may require information for future in which case the process of
estimating the most appropriate value is known as extrapolation. There are two methods in
interpolation.

168
i) Graphic method
ii) Algebraic method
i) Graphic method is a simple method in which we just plot the available data on a graph
sheet and read off the value for the missing period from the graph itself.
ii) Algebraic method:
There are several methods used for interpolation of which we deal with the
following.
i). Finite differences
ii) Gregory – Newton‟s formula
iii) Lagrange‟s formula
Finite Differences:
Definition:
We define an operator ∆ which is known as the first order difference on 𝑈𝑥 𝑎𝑠
∆𝑈𝑥 = 𝑈𝑥+𝑕 − 𝑈𝑥 𝑤𝑕𝑒𝑟𝑒 𝑥 = 𝑎, 𝑎 + 𝑕, 𝑎 + 2𝑕, … … .. in particular.
(i) =∆𝑈𝑎 = 𝑈𝑎+𝑕 − 𝑈𝑎 𝑖𝑖 ∆𝑈𝑥 = 0 if 𝑈𝑥 is constant.
Example :
The difference table for the following data is given below.
x 0 1 2 3 4
𝑼𝒙 8 11 9 15 6

169
Solution :
X 𝑼𝒙 ∆𝑼𝒙 ∆ 𝟐 𝑼𝒙 ∆ 𝟑 𝑼𝒙 ∆ 𝟒 𝑼𝒙
8
3
0
11 -5
1
-2 13
2
9 8 -36
3
6 -23
4 15 -15
-9
6
The differences ∆𝑈𝑥 , ∆2 𝑈𝑥 𝑒𝑡𝑐.are called forward differences. In contrast to the forward
differences we have another kind of differences known as backward differences.
Theorem : [Fundamental Theorem for finite differences]
If 𝑈𝑥 is a polynomial of degree n hence
∆𝑟 𝑈𝑥 = constant if r=n
0 if r=n
i.e.) the 𝑛𝑡𝑕 order difference of a polynomial of degree n is constant and differences of
order higher than n are zero.

170
Proof:
Let 𝑈𝑥 = 𝑎0 𝑥 𝑛 + 𝑎1 𝑥 𝑛 −1 + ⋯ … . . + 𝑎𝑛−1 𝑥 + 𝑎𝑛. Where

𝑎0 , 𝑎1 … … . , 𝑎𝑛 𝑎𝑟𝑒 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡𝑠 𝑎𝑛𝑑 𝑎𝑛 ≠ 0.
∆𝑈𝑥 = 𝑈𝑥+𝑕 − 𝑈𝑥
= 𝑎0 (𝑥 + 𝑕)𝑛−1 + ⋯ … . . + 𝑎𝑛−1 𝑥 + 𝑕 + 𝑎𝑛 − 𝑎0 𝑥 𝑛 + 𝑎1 𝑥 𝑛−1 +

⋯ . . +𝑎𝑛−1 𝑥 + 𝑎𝑛
= 𝑎0 𝑥 𝑛 + 𝑛𝑐1 𝑥 𝑛−1 𝑕 + ⋯ … . +𝑕𝑛 + 𝑎1 (𝑥 𝑛−1 + 𝑛 − 1𝑐1 𝑥 𝑛−2 𝑕 + ⋯ . +𝑕𝑛−1 + ⋯
+ 𝑎𝑛
− 𝑎0 𝑥 𝑛 + 𝑎1 𝑥 𝑛−1 + ⋯ . + 𝑎𝑛
= 𝑎0 𝑛𝑕𝑥 𝑛 −1 + 𝑏2 𝑥 𝑛 −2 + ⋯ … . +𝑏𝑛−1 𝑥 + 𝑏𝑛 ,where 𝑏2 , 𝑏3 … . . 𝑏𝑛 are constants

independent of x and 𝑎0 𝑛𝑕 ≠ 0.
∴ ∆𝑈𝑥 is a polynomial of degree n-1 continuing this process we get ∆2 𝑈𝑥 is a

polynomial of degree n-2.
∆3 𝑈𝑥 is a polynomial of degree n-3 etc.
∴ ∆𝑛 𝑈𝑥 = 𝑎0 𝑛 𝑛 − 1 𝑛 − 2 … . .2.1 𝑕𝑛 𝑥 0
= 𝑎0 𝑛! 𝑕𝑛
= 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡. 𝑎𝑛𝑑 ∆𝑟 𝑈𝑥 = 0 𝑓𝑜𝑟 𝑟 > 𝑛.
Problem:
Find first and second differences for
𝑥
𝑖)𝑈𝑥 = 𝑎𝑏 𝑐𝑥 𝑖𝑖 𝑈𝑥 = taking interval of differencing as h.
𝑥 2 +7𝑥+12

171
Solution :
i) ∆𝑈𝑥 = 𝑈𝑥+𝑕 − 𝑈𝑥
= 𝑎𝑏 𝑐 𝑥+𝑕
− 𝑎𝑏 𝑐𝑥
= 𝑎𝑏 𝑐𝑥 𝑎𝑏 𝑐𝑕 − 𝑎𝑏 𝑐𝑥
= 𝑎𝑏 𝑐𝑥 𝑏 𝑐𝑕 − 1
∆2 𝑈𝑥 = 𝑏 𝑐𝑕 − 1 ∆ 𝑎𝑏 𝑐𝑥
= 𝑏 𝑐𝑕 − 1 2
𝑎𝑏 𝑐𝑥
𝑥
𝑖𝑖) 𝑈𝑥 =
𝑥2 + 7𝑥 + 12
𝑥 3
= − (by partial fraction)
𝑥+4 𝑥+3
4 3 4 3
∆𝑈𝑥 = − − −
𝑥+1 +4 𝑥+1 +3 𝑥+4 𝑥+3
4 3 4 3
= − 𝑥+4 − 𝑥+4 + 𝑥+3
𝑥+5
4 7 3
= − 𝑥+4 + 𝑥+3
𝑥+5
4 11𝑥 10 3
𝑆𝑖𝑚𝑖𝑙𝑎𝑟𝑙𝑦, ∆2 𝑈𝑥 = − + −
𝑥+6 𝑥+5 𝑥+4 𝑥+3
Problem:
If 𝑈0 = 1, 𝑈 = 5, 𝑈2 = 8, 𝑈3 = 3, 𝑈4 = 7, 𝑈5 = 0 𝑓𝑖𝑛𝑑 ∆5 𝑈0 .

172
Solution :
X 𝑼𝒙 ∆𝑼𝒙 ∆𝟐 𝑼𝒙 ∆ 𝟑 𝑼𝒙 ∆ 𝟒 𝑼𝒙 ∆ 𝟓 𝑼𝒙
0 1
4
-
1 5 -1
7
3
1
2 8 -8 24
7
-5 -61
3 3 9 -37
-20
4
4 7 -11
-7
5 0
Hence , ∆5 𝑈0 = −61
Problem:
Estimate the missing term in the following table.
X
0 1 2 3 4
x
𝑈𝑥 1 3 9 -
81
Explain why the resulting value from 33

173
Solution :
Let the missing term in 𝑈𝑥 𝑏𝑒 𝑎.
X 𝑼𝒙 ∆𝑼𝒙 ∆𝟐 𝑼𝒙 ∆ 𝟑 𝑼𝒙 ∆𝟒 𝑼 𝒙
0 1
2
1 3 4
6 a-19
2 9 a-15 124 – 4a
a-9 105-3a
3 Q 90-2a
81-a
4 81
Since 4 values of 𝑈𝑥 are given it is a polynomial of degree 3.
Hence by fundamental theorem of finite differences ∆4 𝑈𝑥 = 0 𝑓𝑜𝑟 𝑎𝑙𝑙 𝑥.
In particular ∆4 𝑈0 = 0
Hence 124 – 4a =0
a = 31
Problem:
Give an estimate of the population in 1971 from the following table.
Year 1941 1951 1961 1971 1981 1991

Population in lakhs 363 391 421 ? 467 501

174
Solution :
Let the population in 1971 be „a‟.
𝑈0 = 363, 𝑈1 = 391, 𝑈2 = 421, 𝑈3 = 𝑎, 𝑈4 = 467, 𝑈5 = 501
Since five values are given
∆5 𝑈𝑥 = 0, for all 𝑥
In particular, ∆5 𝑈0 = 0
𝐸 − 1 5 𝑈0 = 0
(𝐸 5 − 5𝐸 4 + 10𝐸 3 − 10𝐸 2 + 5𝐸 − 1)𝑈0 = 0
∴ 𝑈5 − 5𝑈4 + 10𝑈3 − 10𝑈2 + 5𝑈1 − 𝑈0 = 0
∴ 501 − 5 × 497 + 10𝑎 − 10 × 421 + 5 × 391 − 363 = 0
∴ 501 − 2335 + 10𝑎 − 4210 + 1955 − 363 = 0
10𝑎 − 4452 = 0
𝑎 = 445.2 𝑙𝑎𝑘𝑕𝑠
Hence the estimated population in 1971 is 445.2 lakhs.
Problem:
Find the missing figures in the following table
x 0 5 10 15 20 25
𝑈𝑥 7 11 ? 18 ? 32

175
Solution :
Here ,two values are missing. Let the missing values be a and b.
𝑈0 = 7, 𝑈1 = 11, 𝑈2 = 𝑎, 𝑈3 = 18, 𝑈4 = 𝑏, 𝑈5 = 32.
Since four values are given, we have ∆4 𝑈𝑥 = 0 𝑓𝑜𝑟 𝑎𝑙𝑙 𝑥.
In particular, ∆4 𝑈0 = 0 𝑎𝑛𝑑 ∆4 𝑈1 = 0
𝐸 − 1 4 𝑈0 = 0
∴ 𝐸 4 − 4𝐸 3 + 6𝐸 2 − 4𝐸 + 1 𝑈0 = 0
𝑈4 − 4𝑈3 + 6𝑈2 − 4𝑈1 + 𝑈0 = 0
𝑏 − 72 + 6𝑎 − 44 + 7 = 0
𝑖𝑒) 6𝑎 + 𝑏 = 10𝑞 … … … . . (1)
Taking ∆4 𝑈1 = 0
𝐸 4 − 4𝐸 3 + 6𝐸 2 − 4𝐸 + 1 𝑈1 = 0
𝑈5 − 4𝑈4 + 6𝑈3 − 4𝑈2 + 𝑈1 = 0
32 − 4𝑏 + 108 − 4𝑎 + 11 = 0
4𝑎 + 4𝑏 = 151 … … … . . 2
Solving (1) and (2) we get,
A = 14.25, b=23.5
Hence the missing values are 14.25 and 23.5.
Problem:
Given that 𝑈0 𝑡𝑜 𝑈8 = 80; 𝑈1 + 𝑈7 = 10;
𝑈2 + 𝑈6 = 5; 𝑈3 + 𝑈5 = 10 𝑓𝑖𝑛𝑑 𝑈4.

176
Solution :
Since 4 values are given
∴ ∆𝑛 𝑈𝑥 = 0 𝑓𝑜𝑟 𝑎𝑙𝑙 𝑛 ≥ 4 𝑓𝑜𝑟 𝑎𝑙𝑙 𝑥 .
𝐼𝑛 𝑝𝑎𝑟𝑡𝑖𝑐𝑢𝑙𝑎𝑟 ∆8 𝑈0 = 0
𝐻𝑒𝑛𝑐𝑒 𝐸 − 1 8 𝑈0 = 0
𝑈8 − 8𝑈7 + 28𝑈6 − 56𝑈5 + 70𝑈4 − 56𝑈3 + 28𝑈2 − 8𝑈1 + 𝑈0 = 0
𝑈0 + 𝑈8 − 8 𝑈1 + 𝑈7 + 28 𝑈2 + 𝑈6 − 56 𝑈3 + 𝑈5 + 70𝑈4 = 0
∴ 80 − 80 + 140 − 560 + 70𝑈4 = 0
∴ 70𝑈4 = 420
∴ 𝑈4 = 6.
Problem:
Given that 𝑈1 + 𝑈2 + 𝑈3 = 25, 𝑈4 = 29, 𝑈5 + 𝑈6 = 113. Find the

polynomial 𝑈𝑥 and hence find 𝑈10.
Solution :
Since three values are given 𝑈𝑥 is a polynomial of degree 2.
Let 𝑈𝑥 = 𝑎𝑥 2 + 𝑏𝑥 + 𝑐
𝑈1 = 𝑎 + 𝑏 + 𝑐, 𝑈2 = 4𝑎 + 2𝑏 + 𝑐, 𝑈3 = 9𝑎 + 3𝑏 + 𝑐
Given, 𝑈1 + 𝑈2 + 𝑈3 = 25
∴ 14𝑎 + 6𝑏 + 3𝑐 = 25 … … . . 1
𝑁𝑜𝑤, 𝑈4 = 24
⇒ 16𝑎 + 4𝑏 + 𝑐 = 24 … … … … … 2
𝑈5 + 𝑈6 = 113
61𝑎 + 11𝑏 + 2𝑐 = 113

177
Solving (1), (2) and (3) we get
A = 2, b=-1, c=1
∴ 𝑈𝑥 = 2𝑥 2 − 𝑥 + 1
𝑁𝑜𝑤, 𝑈10 = 100𝑎 + 10𝑏 + 𝑐

= 200 − 10 + 1
= 191
Problem:
If 𝑈1 = 12 − 𝑥 4 + 𝑥 ;
𝑈2 = 5 − 𝑥 4 − 𝑥 ,
𝑈3 = 𝑥 + 18 𝑥 + 6 𝑎𝑛𝑑 𝑈4 = 9.
Obtain a value of x assuming second differences constant.
Solution :
∆3 𝑈1 = 0
𝐻𝑒𝑛𝑐𝑒 𝐸 − 1 3 𝑈1 = 0
∴ 𝐸 3 − 3𝐸 2 + 3𝐸 − 1 𝑈1 = 0
𝑈4 − 3𝑈3 + 3𝑈2 − 𝑈1 = 0
∴ 94 − 3 𝑥 + 18 𝑥 + 6 + 3 5 − 𝑥 4 − 𝑥 − ? 12 − 𝑥 4 + 𝑥 = 0
𝑖𝑒) 𝑥 2 − 107𝑥 − 218 = 0
𝑥 − 109 𝑥 + 2 = 0
𝑥 = 109 𝑜𝑟 − 2
Newton’s Formula:
Consider the function y=f(x). Let f(𝑥0 ) = 𝑦0 , 𝑓 𝑥1 = 𝑦1 … … … 𝑓 𝑥𝑛 = 𝑦𝑛 .

178
Newton-Gregory formula for forward interpolation:
∆𝑈𝑎 ∆2 𝑈𝑎
𝑈𝑥 = 𝑈𝑎 + 𝑥 − 𝑎 + 𝑥−𝑎 𝑥−𝑎−𝑕
1! 𝑕 2! 𝑕2
∆𝑛 𝑈𝑎
+ ⋯…. 𝑥 − 𝑎 𝑥 − 𝑎 −𝑕 …… 𝑥 −𝑎 − 𝑛 − 1𝑕
𝑛! 𝑕𝑛
Newton’s formula for backward interpolation
∇𝑈𝑎 +𝑛𝑕 ∇2 𝑈𝑎 +𝑛 𝑕
𝑈𝑥 = 𝑈𝑎+𝑛𝑕 + 𝑥 − 𝑎 + 𝑛𝑕 +
1!𝑕 2!𝑕 2
∇𝑛 𝑈𝑎+𝑛𝑕
𝑥 − 𝑎 + 𝑛𝑕 (𝑥 − 𝑎 + 𝑛 − 1 𝑕 + ⋯ … . + 𝑥 − 𝑎 + 𝑛𝑕 …
𝑛! 𝑕𝑛
… (𝑥 − 𝑎 + 𝑕)
Problem:
𝑈75 = 246; 𝑈80 = 202; 𝑈85 = 118 𝑎𝑛𝑑 𝑈 90 = 40 𝑓𝑖𝑛𝑑 𝑈79.
Solution :
Here, a= 75, h=5
To find 𝑈𝑎+𝑟𝑕 = 𝑈79
∴ 𝑎 + 𝑟𝑕 = 79
75 + 5𝑟 = 79
4
𝑟= =0.8
5

179
By Newton-Gregory formula for equal intervals,
𝑟 𝑟 𝑟 −1
𝑈𝑎+𝑟𝑕 = 𝑈𝑎 + 1! ∆𝑈𝑎 + ∆2 𝑈𝑎 + ⋯ …
2!
X 𝑼𝒙 ∆𝑼𝒙 ∆ 𝟐 𝑼𝒙 ∆ 𝟑 𝑼𝒙
75 246
-44
80 202 -40
-84 46
85 118 6
-78
90 40
0.8 −44 0.8 0.8 − 1 0.8 0.8 − 1 0.8 − 2

∴ 𝑈79 = 246 + + −40 + 46
1 1.2 1.2.3
= 246 − 35.2 + 3.2 + 1.472
= 215.472
Problem:
By using Gregory – Newton‟s formula find 𝑈𝑥 for the following data. Hence estimate
(i) 𝑈1.5 𝑖𝑖 𝑈9
𝑈0 𝑈1 𝑈2 𝑈3 𝑈4
1 11 21 28 29

180
Solution :
X 𝑈𝑥 ∆𝑼𝒙 ∆ 𝟐 𝑼𝒙 ∆ 𝟑 𝑼𝒙 ∆ 𝟒 𝑼𝒙
10
0 1
0
1 11
10 -3
2 21
-3 0
3 28
7 -3
4 29
1 -6
1
Here the third order difference are constant and hence required function is a
polynomial of degree 3.
∇𝑈𝑎 ∆2 𝑈0 𝑕2
𝑈𝑥 = 𝑈𝑎 + 𝑥 − 𝑎 + 𝑥−𝑎 𝑥−𝑎−𝑕 + ⋯ ….
1! 2!
Here, a=0 and h=1
10 0 −3
∴ 𝑈𝑥 = 1 + 𝑥 − 0 × + 𝑥−1 × +𝑥 𝑥−1 𝑥−2 ×
1! 2! 3!
𝑥 𝑥−1 𝑥−2
= 1 + 10𝑥 −
2
1
= 2 + 20𝑥 − 𝑥 3 + 3𝑥 2 − 2𝑥
2
1
𝑈𝑥 = −𝑥 3 + 3𝑥 2 + 18𝑥 + 2
2
1
i) ∴ 𝑈1.5 = [−(1.5)3 + 3(1.5)2 + 18 1.5 + 2]
2
1
= −3.375 + 6.75 + 27 + 2
2
= 16.188

181
1
ii) 𝑈9 = −9)3 + 3 × 92 + 18 × 9 + 2
2
1
= −729 + 243 + 162 + 2 = −161
2
Problem:
Population was recorded as follows in a villages.
Year 1941 1951 1961 1971 1981 1991
Population 2500 2800 3200 3700 4350 5225
Estimate the population for the year 1945.
Solution :
Year
Population 𝑈𝑥 ∆𝑼𝒙 ∆ 𝟐 𝑼𝒙 ∆ 𝟑 𝑼𝒙 ∆ 𝟒 𝑼𝒙 ∆ 𝟓 𝑼𝒙
x
300
1941 2500 100
1951 2800 400 0
1961 3200 100 50
1971 3700 500 50 -25
1981 4350 150 25

650 75
1991 5225
225
875

182
We have to find 𝑈1945
Here a = 1941 and h=10
∴ 𝑈𝑎+𝑟𝑕 = 𝑈1945
𝐻𝑒𝑛𝑐𝑒 1941 + 10 = 1945
𝑟 = 0.4
Applying Newton – Gregory formula

300 100 0
𝑈1945 = 2500 + 0.4 × + 4 0.4 − 1 × + 0.4 0.4 − 1 0.4 − 2 × +
1 2! 3!
50 25
0.4 0.4 − 1 0.4 − 3 × + 0.4 0.4 − 1 0.4 − 2 0.4 − 3 0.4 − 4 ×
4! 5!
= 2500 + 120 − 12 − 2.08 + 0.75
= 2606.67 ≅ 2607
4. From the following data estimate the number of persons whose daily wage is
between Rs. 40-50.
Daily wage in Rs. No. of persons
0-20 120
20-40 145
40-60 200
60-80 250
80-100 150

183
Solution :
The less than cumulative frequency table of the above data is given by
Wage less than x No. of persons (c.f)
20 120
40 265
60 465
80 715
100 865
The difference table
X 𝑈𝑥 ∆𝑼𝒙 ∆ 𝟐 𝑼𝒙 ∆ 𝟑 𝑼𝒙 ∆ 𝟒 𝑼𝒙
20 120
145 55
40 265
200 -5
60 465
250 50 -145
80 715
150 -150
100 865 -100
Number of persons whose earnings is between Rs. 40-50 is 𝑈50 − 𝑈40
We have 𝑈40 = 265
Find 𝑈50
𝑈50 = 𝑈𝑎+𝑟𝑕
𝑎 = 20, 𝑕 = 20
50 = 20 + 20𝑟
𝑟 = 1.5

184
By Newton-Gergory formula,
145 55 −5
𝑈50 = 120 + 1.5 × + 1.5 1.5 − 1 × + 1.5 1.5 − 1 1.5 − 2 × +
1 2! 3!
−145
1.5 1.5 − 1 1.5 − 2 1.5 − 3 4!
= 120 + 217.5 + 20.625 + 0.3125 − 3.3984

= 355
Number of persons whose earnings is between Rs. 40-50 is
𝑈50 − 𝑈40 = 355 − 265

= 90
Problem:
The following data gives the melting point of an alloy of lead and zinc. 𝜃is the
temperature in degrees centigrade and x is the temperature of lead.
X 40 50 60 70 80 90
𝜃 184 204 226 250 276 304
Find 𝜃 𝑤𝑕𝑒𝑛 𝑖 𝑥 = 42 𝑖𝑖 𝑥 = 38
Solution :
To find 𝑈42 , a = 40, h=10
Hence 𝑈𝑎+𝑟𝑕 = 𝑈42

𝑎 + 𝑟𝑕 = 42
40 + 10𝑟 = 42
𝑟 = 0.2
20 2
∴ 𝑈42 = 184 + 0.2 × + 0.2 0.2 − 1 ×
1! 2!
= 184 + 4 − 0.16 = 187.84

185
X 𝜃 ∆𝜃 ∆2 𝜃 ∆3 𝜃
40 184
20 2
50 204 0
22 2
60 226 0
24 2
70 250 0
26 2
80 276
28
90 304
(ii) To find 𝑈38
𝑈40 + 10 𝑟 = 𝑈38
𝐻𝑒𝑛𝑐𝑒, 40 + 10𝑟 = 38
𝑟 = −0.2
20 2
∴ 𝑈38 = = 184 + −0.2 x + −0.2 −0.2 − 1 x
1! 2!
= 184 − 4 + 0.24
= 180.24
Problem:
The following table gives the census population of a town for the years 1931 – 1971. Estimate
the population (i) for the year 1965, (ii) for the year 1933 by using an appropriate interpolation
formula.
Y
ear 1931 1941 1951 1961 1971
Population
in lakhs 36 66 81 93 101

186
Solution :
i) To find the population in the year 1965
Population
𝛁𝑼𝒙 𝛁 𝟐 𝑼𝒙 𝛁 𝟑 𝑼𝒙 𝛁 𝟒 𝑼𝒙
Year 𝑈𝑥
1931 36
30
1941 66 -15
15 12
1951 81 -3 -13
12 -1
1961 93 -4
8
1971 101
To find 𝑈1965
𝐻𝑒𝑟𝑒 𝑎 + 𝑛𝑕 = 1971, 𝑕 = 10
𝑈𝑎+𝑛𝑕+𝑟𝑕 = 𝑈1965
𝑎 + 𝑛𝑕 + 𝑟𝑕 = 1965
∴ 1971 + 𝑟𝑕 = 1965
𝑟 = −0.6

187
Applying Newton‟s backward difference formula for interpolation we get,
−0.6 8 −0.6 −0.6+1 −0.6 −0.6+1 −0.6+2
𝑈1965 = 101 + + −4 + −1 +
1! 2! 3!
−0.6 −0.6+1 −0.6+2 −0.6+3
−13
4!
= 101 − 4.8 + 0.48 + 0.056 + 0.4368

= 97. 1728.
ii) To find the population in 1993
Population
Year ∆𝑼𝒙 ∆ 𝟐 𝑼𝒙 ∆ 𝟑 𝑼𝒙 ∆ 𝟒 𝑼𝒙
𝑈𝑥
1931 36
30 -15
1941 66
15 12
1951 81
12 -3 -13
1961 93
8 -1
1971 101
-4
To find 𝑈1933
𝑎 = 1931 𝑎𝑛𝑑 𝑕 = 10
𝑈𝑎+𝑟𝑕 = 𝑈1931
𝐻𝑒𝑛𝑐𝑒 𝑎 + 𝑟𝑕 = 1933
∴ 1931 + 10𝑟 = 1933
𝑟 = 0.2
Applying Newton –Gregory‟s formula we get,
0.2 0.2 − 1 0.2 0.2 − 1 0.2 − 2

𝑈1933 = 36 + 0.2 × 30 + −15 + × 15
2! 3!
0.2 0.2 − 1 0.2 − 2 0.2 − 3
+ × −13
4!

188
= 36 + 6 + 1.2 + 0.576 + 0.4368
= 44.2128
Lagrange’s formula
The Lagrange‟s formula becomes
𝑥 − 𝑎2 𝑥 − 𝑎3 … … 𝑥 − 𝑎𝑛 𝑥 − 𝑎1 𝑥 − 𝑎3 … … 𝑥 − 𝑎𝑛
𝑈𝑥 = 𝑈𝑎 1 + 𝑈
𝑎1 − 𝑎2 𝑎1 − 𝑎3 … … 𝑎1 − 𝑎𝑛 𝑎2 − 𝑎1 𝑎2 − 𝑎3 … … 𝑎2 − 𝑎𝑛 𝑎 2
𝑥 − 𝑎1 𝑥 − 𝑎2 … … 𝑥 − 𝑎𝑛−1
+ ⋯+ 𝑈
𝑎𝑛 − 𝑎1 𝑎𝑛 − 𝑎2 … … 𝑎𝑛 − 𝑎𝑛−1 𝑎 𝑛
Problem:
Find 𝑈5 given that 𝑈1 = 4; 𝑈2 = 7; 𝑈4 = 13; 𝑎𝑛𝑑 𝑈7 = 30.
Solution :
Take 𝑎1 = 1, 𝑎2 = 2; 𝑎3 = 4; 𝑎4 = 7 𝑎𝑛𝑑 𝑥 = 5.
𝑆𝑢𝑏𝑠𝑡𝑖𝑡𝑢𝑡𝑖𝑛𝑔 𝑖𝑛 𝐿𝑎𝑔𝑟𝑎𝑛𝑔𝑒 ′ 𝑠𝑓𝑜𝑟𝑚𝑢𝑙𝑎 𝑤𝑒 𝑔𝑒𝑡,
𝑈5
5−2 5−4 5−7 5−1 5−4 5−7
= × 4+ ×7
1−2 1−4 1−7 2−1 2−4 2−7
5−1 5−2 5−7 5−1 5−2 5−4
+ × 13 + × 30
4−1 4−2 4−7 7−1 7−2 7−4
3 × 1 × −2 4 × 1 × −2 4 × 3 × −2
= × 4+ ×7+
−1 −3 −6 1 −2 −5 3 × 2 × −3
4 × 3×1
× 13 × 30
6×5×3
4 28 52
= − + +4
3 5 3
= 17.06

189
Problem:
Find the form of the function 𝑈𝑥 for the following data. Find 𝑈3
X 0 1 2 5
𝑈𝑥 2 3 12 147
Solution :
Here 𝑎1 = 0; 𝑎2 = 1; 𝑎3 = 2; 𝑎4 = 5
∴ 𝑈𝑎1 = 2; 𝑈𝑎2 = 1; 𝑈𝑎3 = 12; 𝑈𝑎4 = 147
Applying Lagrange‟s formula, we get
𝑥−1 𝑥−2 𝑥−5 𝑥−0 𝑥−2 𝑥−5 𝑥−0 𝑥−1 𝑥−5

𝑈𝑥 = ×2+ ×1+ × 12 +
0−1 0−2 0−5 1−0 1−2 1−5 2−0 2−1 (2−5)
𝑥−0 𝑥−1 𝑥−2

× 147
5−0 5−1 5−2
− 𝑥 3 − 8𝑥 2 + 17𝑥 − 10 3 𝑥 3 − 7𝑥 2 + 10𝑥
= + − 2 𝑥 3 − 6𝑥 2 + 5𝑥
5 4
𝑥 3 − 3𝑥 2 + 2𝑥
+ × 147
60
1 3
= 𝑥 −12 + 45 − 120 + 147 + 𝑥 2 96 − 315 + 720 − 441
60
+ 𝑥 −204 + 450 − 600 + 294 + 120
1
= 60𝑥 3 + 60𝑥 2 − 60𝑥 + 120
60
∴ 𝑈𝑥 = 𝑥 3 + 𝑥 2 − 𝑥 + 2
∴ 𝑈3 = 33 + 32 − 3 + 2
= 35

190
Problem:
Determine by Lagrange‟s formula the percentage number of criminals under 35

years.
% number of
Age
criminals
Under 25 years 52.0
Under 30 years 67.3
Under 40 years 84.1
Under 50 years 94.4
Solution :
To find 𝑈35 . 𝑎1 = 25; 𝑎2 = 30; 𝑎3 = 40; 𝑎4 = 50

𝑈𝑎1 = 52; 𝑈𝑎2 = 67.3; 𝑈𝑎3 = 84.1, 𝑈𝑎4 = 94.4
Applying Lagrange‟s formula we get
𝑈35 =
35−30 35−40 35−50 35−25 35−40 35−50
× 52.0 + × 67.3 +
25−30 25−40 25−40 30−25 30−40 30−50
35−25 35−30 35−50 35−25 35−30 35−40
× 84.1 + × 94.4
40−25 40−30 40−50 50−25 50−30 50−40
−1 × 52.0 3 × 67.3 1 × 84.1 1 × 94.4

= + + −
5 4 2 20
= −10.40 + 50.38 + 42.05 − 4.72
= 77.31
Hence the estimated % number of criminals under 35 years is 77.31.
Prepared by
Ms. C. KANI
Assistant Professor of Mathematics, St. Jude‟s College,
Thoothoor - 629 176, Kanyakumari District.

191

5db83ef1f71e482 PDF

Uploaded by

Copyright:

Available Formats

5db83ef1f71e482 PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

5db83ef1f71e482 PDF

Uploaded by

Copyright:

Available Formats

MANONMANIAM SUNDARANAR UNIVERSITY

DIRECTORATE OF DISTANCE & CONTINUING EDUCATION

B.Sc. MATHEMATICS - I YEAR

Most Student friendly University - Strive to Study and Learn to Excel

For more information visit: http://www.msuniv.ac.in

UNIT I : Correlation - Karl Pearson‟s coefficient of correlation, Lines of Regression -

UNIT II : Probability - Definition - Applications of Addition and multiplication, theorems -

UNIT III : Association of Attributes - Coefficient of Association - Consistency - Time Series -

1. Statistics: S. Arumugam & others

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

Correlation - Karl Pearson’s coefficient of correlation, Lines of Regression - Regression

CORRELATION AND REGRESSION

Consider a set of bivariate data 𝑥𝑖 , 𝑦𝑖 ;i=1,2….n. If there is a change in one variable

The covariance between x and y is defined by

The heights and weights of five students are given below.

𝐻𝑒𝑟𝑒 𝑥 = 162; 𝑦 = 54 ; 𝜍𝑥 = 2 𝑎𝑛𝑑 𝜍𝑦 = 6

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

Substituting (2),( 3) and (4) in (1) we get the required result.

By Schwartz inequality we have,

If 𝛾 = 1 𝑡𝑕𝑒 𝑐𝑜𝑟𝑟𝑒𝑙𝑎𝑡𝑖𝑜𝑛 𝑖𝑠 𝑝𝑒𝑟𝑓𝑒𝑡 𝑎𝑛𝑑 𝑝𝑜𝑠𝑖𝑡𝑖𝑣𝑒.

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

If 𝛾 = −1 the correlation is perfect and negative.

If 𝛾 = 0 the variables are uncorrelated.

If the variables x and y are uncorrelated then cov (x,y) = 0

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

10×500− −70 ×(−12)

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

A programmer while writing a program for correlation coefficient between

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

Corrected 𝑥 2 = 3718 − 182 − 122 + 102 + 202

Corrected 𝑦 2 = 2000 − 202 − 102 + 152 + 152

Corrected 𝑥𝑦 = 2100 − 18 × 20 − 12 × 100 + 10 × 15 + 20 × 15 = 2070

After correction the correlation coefficient is,

30 × 2070 − 300 × 210

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

𝐼𝑓 𝑥, 𝑦 𝑎𝑛𝑑 𝑧 are uncorrelated variable each having same standard deviation

𝑋 𝑎𝑛𝑑 𝑦 are uncorrelated ⇒ 𝑥−𝑥 𝑦−𝑦 =0

𝑌 𝑎𝑛𝑑 𝑧 are uncorrelated ⇒ 𝑦−𝑦 𝑧−𝑧 =0

𝑍 𝑎𝑛𝑑 𝑥 are uncorrelated ⇒ 𝑧−𝑧 𝑥−𝑥 =0

𝑢 = 𝑥+𝑦 and 𝑣=𝑦+𝑧

= 𝜍𝑥2 + 𝜍𝑦2 [𝑠𝑖𝑛𝑐𝑒 𝑥 − 𝑥 𝑦 − 𝑦 = 0]

Now, 𝑢−𝑢 𝑣−𝑣 = 𝑥 − 𝑥 + 𝑦 − 𝑦 { 𝑦 − 𝑦 + 𝑧 − 𝑧 }]

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

𝑢𝑖= 𝑥𝑖 𝑐𝑜𝑠𝛼 + 𝑦𝑖 𝑠𝑖𝑛𝛼 𝑎𝑛𝑑 𝑣𝑖 = 𝑦𝑖 𝑐𝑜𝑠𝛼 − 𝑥𝑖 𝑠𝑖𝑛𝛼

𝑢 = 𝑥 𝑐𝑜𝑠𝛼 + 𝑦𝑠𝑖𝑛𝛼 𝑎𝑛𝑑 𝑣 = 𝑦𝑐𝑜𝑠𝛼 − 𝑥 𝑠𝑖𝑛𝛼

𝑢𝑖 – 𝑢 = (𝑥𝑖 − 𝑥 ) 𝑐𝑜𝑠𝛼 + 𝑦𝑖 − 𝑦 𝑠𝑖𝑛𝛼

The variable 𝑢𝑖 and 𝑣𝑖 are uncorrelated if 𝑢𝑖 − 𝑢 𝑣𝑖 − 𝑣 = 0

𝑥𝑖 − 𝑥 𝑐𝑜𝑠𝛼 + 𝑦𝑖 − 𝑦 𝑠𝑖𝑛𝛼 𝑦𝑖 − 𝑦 𝑐𝑜𝑠𝛼 − 𝑥𝑖 − 𝑥 𝑠𝑖𝑛𝛼 = 0

∴ (𝑥𝑖 − 𝑥) 𝑦𝑖 − 𝑦 cos2 𝛼 − 𝑥𝑖 − 𝑥 𝑦𝑖 − 𝑦 sin2 𝛼 − 𝑐𝑜𝑠𝛼𝑠𝑖𝑛𝛼[ 𝑥𝑖 − 𝑥 2

∴ 𝑛𝛾𝑥𝑦 𝜍𝑥 𝜍𝑦 (cos 2 𝛼 − sin2 𝛼) = 𝑛𝑐𝑜𝑠𝛼𝑠𝑖𝑛𝛼 𝜍𝑥2 − 𝜍𝑦2

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

Suppose that a group of n individuals are arranged in the order of merit or

Rank correlation 𝜌 is given by

Consider a collection of n individuals

= 𝑛𝜍𝑥2 + 𝑛𝜍𝑦2 − 2𝑛𝜌𝜍𝑥 𝜍𝑦

= 2n𝜍𝑥2 1 − 𝜌 (𝑠𝑖𝑛𝑐𝑒 𝜍𝑥2 = 𝜍𝑦2 )

Manonmaniam Sundaranar University, Directorate of Distance & Continuing Education, Tirunelveli.

Height Rank Weight Rank x-y (𝒙 − 𝒚)𝟐