NATIONAL OPEN UNIVERSITY OF NIGERIA

SCHOOL OF SCIENCE AND TECHNOLOGY

COURSE CODE: STT 211

COURSE TITLE: PROBABILITY DISTRIBUTION I

1
Course Code STT 211

Course Title Probability Distribution I

Writer Arowolo. OLATUNJI

School of Science and Technology
Lagos State Polytechnic. Ikorodu..

Course Editing Team

2
PROBABILITY DISTRIBUTION I

MODULE 1
UNIT 1
PREREQUISITES
The main prerequisites for understanding the content of this book is a knowledge of
elementary algebra, set theory, mathematical induction, differentiation and integration.
Set
A set is any well-defined list or collection of objectives. The objectives comprising the
set are called its elements or members.
A set will be denoted by capital letters or symbols such as X,Y,A,B,….. and its elements
will be denoted by lower case letters x, y, a,b,…..
Example 1.1 Toss a cubical die once. There are six possible numbers that can appear. We
can write
Ω = {1, 2, 3, 4,5, 6}
Where Ω is a set consisting of six elements 1, 2, 3, 4, 5, 6 called the element of the set Ω.
There are essentially two ways of specify a particular set. One way if possible is by
listing its elements as in example1.1 above. The other way is by stating properties which
characterize the elements in the set. The above set Ω can be written as:
Ω – {x: is an integer, 1 < x < 6}
If x is an element Ω, the notation x € Ω means that x belongs to Ω. The negation of this
assertion i.e. the statement that x does not belong to Ω will be denoted by
x€Ω
thus, for the above example 2 € Ω but 8 € Ω.
Definition 1.2
Subset: A set A is a subset of a set if each element in A also belongs to Ω.
In example 1.1 above, the set
A ={1,3,5}= (x: x €Ω and x is odd}
Is a subset of Ω, that is each element of A is in Ω.
Two sets A and B are called equal if and only if they contain exactly the same elements.
3
Throughout this book, whenever the word set is used, itwill be interpreted to mean a
subset of a given set denoted by Ω. The set which contains no elements is called the Null
set.
Definition
Let A be a set. The elements which are not included in A also constitute a subset. This is
known as the complement of A and is denoted by Ac.
In example 1,1 if A {1,3,5}
Ac = {2,4,6} = {x: x Ω, x even}.
Definition:
Two sets A, B define two related sets. One of these is the set of all elements which
belongs to both sets A and B. this is called the intersection of A and B and is denoted by
A∩B. the other is the set of all the elements which occur in either A or B or both. This is
called the union of A and B denoted by A B.

Example 1.2
Let Ω = {1,2,3,4,5,6}, A = {1,3,5} and B = {2,3,5}
Then
A∩B = {3,5}
A ∩B = {1,2,3,5}
A and B contain 3 element each while A ∩ B contains 2 elements and A ∩B contains 4
laments.
Note that the number of elements in A ∩ B is not the sum of the number of elements in A
and B.
Definition: Difference of Two sets
The different of A and B is the set of elements which belong to A but not to B and is
denoted by A/B
In example 1.2 above A/B = {x: x €A, x € B} Ω
A/B = {1}
B/A = {2}
The union of two sets A B can be divided into three disjoint sets

4
 A/B, A ∩B, B/A.
That is
A ∩ B = (A/B) ∩ (A ∩B) ∩ (B/A)
The number of element in a set will be denoted by nA. Thus,
n(A ∩ B) = n (A/B) + n(A∩B) ∩ (B/A)
Since
(A/B), (A ∩ B), (B/A) are disjoint sets
nA = n(A/B) + n(A∩B)
nB = n(B/A) + n(A∩B)
n(A/B_ = nA –n (A ∩B)
n (B/A) = nB-n(A∩B)
hence,
n(A ∩B) = nA-n (A ∩B) + n(A ∩B) + nB-n (A ∩B) = nA + nB – n(A ∩ B)
note: A/B = A ∩ Bc, B/A = B ∩ Ac
for any set A,
A ∩ Ac = Φ, A ∩ Ac = Ω.
For any two sets A and B, we have the following decomposition:
B = B ∩ Ω = B ∩ ( A ∩ Ac) = (B ∩ A) ∩ ( B ∩ Ac),
Since A ∩ B and Ac ∩ B are disjoints, we have
nB = n(A ∩ B) +n(Ac ∩ B)
de Morgan’s Law
(i) For any two sets A and B
(A ∩ B)c = Ac ∩ Bc

5
Proof:
Ac = ( A ∩ B)c ∩ (B/A),

Bc = ( A ∩ B)c ∩ (A/B).

Thus,

Ac ∩ Bc = (A ∩ B)c

Since B/A and A/B are disjoint, Alternatively,

(A B)c = { x: x A and X B}=Ac∩ Bc

In general, if A1, A2,…..An are any n sets

(A1 ∩ A2 ∩…∩An)c = A1c∩A2c…Anc.

Thus

Series

A sequence is a set of numbers occurring in order, and there is a simple rule by which the
terms are obtained. For example, 1, x, x2… is a sequence. If the terms of a sequence are
considered as a sum, for instance, 1 + x + x2 + …. The expression is called a series. A
series with a finite number of terms is called a finite series otherwise it is called a finite
series otherwise it is called an infinite series. The summation is shown by the symbol.

When the sum is taken from the first term (r = 1) to thee nth term (r = n).

Example

The series 1 + x + x2 + … + x10 can be written as

6
The above series is called a geometric series with common ratio x. the sum

Xr-1 = 1 ' () x =1

1–x

N, x =1

The commonratio is x. if – 1 < x < 1, xn -> 0 as n- > =, then the series converges to

1–x

The infinite series

Example

Find the sum of the series

Solution

(1 – p)r-1 = 1 + (1 –p) + 1 –p)2 +…….

The common ratio is 1 – p, from (1) let x = 1 –p, we have

(1 – p)r-1 = 1 = 1
1 –(1 –p) p
And
(1 – p)r-1 = 1 – (1 – p)n = 1 –(1 –p)n
1 – (1 – p) P

Exponential series
The function y = ex is called an exponential function. This function is one of the special
functions of analysis and can be defined as the sum of an infinite series. It is defined by
Ex =1 + x + x2 +…+ xn+ … = xr
2! n! r!
The notation r! is called factorial r and is defined as
R! = r(r – 1) (r-2) ….. 2.1
For example

7
 5! = 5 x 4 x 3 x 2 x 1
Where 0! = 1 and 1! = 1. R! defined above has no meaning unless r itself is a positive
integer.
The factorial of any negative integer is infinite.
Gamma Function
It is easily shown by direct integration that when m is an integer
M! =
Provided that m > - 1. It can be proved that (2) will also have a meaning for fraction m.
for instance, it is known that

Since m! = m(m-1)!. By setting m = 1 we have

2
The Binomial Theorem
If n is a positive integer,
(a+b)2 = an+ nC1a n-1 B + nC2an-2b2+….+bn
Where nCr = n! . Using the summation notation, we have
(n – r)!r!
(a +b)n using the summation notation, we have
Example
Find the sum of the series
Sn = Pn + nC1pn-1 (1 – P) + nC2pn-2(1-P)2 +…+ (1-p)nnCrP n-r (1-P)r
Substituting p and 1-p for a, b in the binomial theorem above, we have

Thus,
Generalization of the Bionomial theorem is the multinomial theorem. If n is a positive
integer
(a1+a2 + … ak)n = ∑…….∑ n! a1n1a2n2…aknk
N1 ! n2!....nk!
Sum is over all n1, n2….. nk where n1 + n2 + … + nk= n.

8
Product Notation
The product of the terms of a sequence x1, x2…., xn can be written as x1x2x3….xn. This
product is shown by the symbol.
Where the product is taken from x1, (the first term) to the nth term, xn. for example,
N1! N2!...nk! and a12…ak
Can be written as

Thus the multinomial theorem can be written as

Example
(a1 + a2 + a3)2 = ∑∑ 2!
3
Where n=2, k = 3. Possible values of n1 are
n1 =2, n2 =0, n3=0
n1 = 0, n2 = 2, n3 = 0
n1 = 0, n2 = 0, n3 =2
n1= 1, n2=1, n3 = 0
n1 =1, n2 = =, n3 =1
n1 = 0,n2 =1, n3 =
hence, we have
(aa + a2 + a3)2 = a12 + a22 + a23 + 2a1a2+ 2a1a3+2a2a3
Exponential Functions
These are function in which the variable occur in the index, for example ex32x are called
exponential functions, Let y = e h(x) then
For example, if y = e3x2 then
dy =6xe3x2
dx
Derivative of loge h(x)
If y = loge 2 x3 then
9
dy = 6x. 1 = 3
dx 2x3 x2

10
UNIT TWO
MATHEMATICS OF COUNTING
Probability had its origin in games of chance such as disc and card games. The number of
definitions you can find for it is limited only by the number of books you may wish to
consult. Probability can be defined as a measure put on occurrence of a random
phenomenon. probability theory is developed as study of the outcomes of trail of an
experiment.

Definition
An experiment is a phenomenon to be observed according to a clearly defined procedure.
Probabilities are numbers between 0 and 1, inclusive that reflect the chances of a
particular physical events occurring. If a die is tossed once, the possible outcomes are
1,2,3,4,5,6, let Ω = {1,2,3,4,5,6}. Then Ω is a set consisting of all possible outcome of
tossing a die once. This set is given a name, it is called a sample space.
Example 1.1
Write down the staple space for each of the following experiments
(i) Toss a coin 3 times and observe the total number of heads
(ii) A box contains 6 items of which 2 are defectives. One item is chosen one after the
other without replacement until the last defective items is chosen. We observe
the total number of items removed from the box.
Solution
(i) Possible outcomes are: number of heads is 0 when we have TTT, number of heads
is 1 when we have HTT or THT or TTH, number of heads is 2 when we have
HHT, THH, HTH and number of heads are 0,1,2,3. Hence the sample space Ω
= {0,1,2,3)
(ii) Total number of items removed is 2 if we have “the first is defective (D) and the
second is defective D (Note that the total number of items removed can not be
0 or 1) denoted by DD. The total number of items removed is 3 if we have

11
 GDD or DGD (where G denoted good item), and so on. Thus, the sample space
Ω = {2,3,4,5,6}
Definition 1.2 Event
Any subset A of a sample space Ω is called an event, where A < Ω.
Example 1.2
The following are examples of events
(i) An odd number occurs when a die is rolled once
Ω = {1,2,3,4,5,6}, A = {1,3,5}
(ii) Toss a coin 3 times, the total number of heads observed is even
Ω = {0,1,2,3} A = {0,2}

Definition 1.3 Trial of an Experiment

A trial is a single performance of an experiment
There are basically two methods for assigning probabilities to events. These are
(i) Relative frequency approach
(ii) Classical approach
(i) Relative Frequency Approach
If we are interested in the occurrence of an event A, we could perform a large
number of trials and define the relative frequency of A as
RF(a) = Number of trials in which A occurs
Total number of trials
The ratio RF(A) is called the empirical probability. If the number of trials is very large
RF(A) will (in most cases0 tend to a particular value called the probability that the event
A will occur, that is RF (A) is an estimate of P(A). the probability that A will occur.
Example 1.3
Suppose we are interested in the occurrence of a “6” when we roll a well balanced die
(fair die) the die is then rolled 100 times and “6” appeared 15 times. The relative
frequency of “6” is thus 15 which is the empirical probability of getting 6
100
(ii) Classical Approach:
12
In the above example, there are six ways the die can fall when it is rolled once.
So we can define a theoretical probability by
P(6) = number of ways of getting a “6” when rolled once = 1
Total number of possible outcomes when rolled once 6
The classical approach to probability gives
P(A) = nA
nΩ
This definition is valid only when outcomes are equally likely.

Example 1.4
A coin is rolled thrice. The possible outcomes assumed to be equally likely are
Ω= {HHH, HHT, HTH, HTT, TTT, TTH, THT, THH}
Let A be the event that 2 heads occur. Then
A = {HHT, HTH, THH}
n Ω = 8, nA = 3
hence
P(A) = 3
8
This classifiableapproach when applicable (the possible outcomes are equally likely) has
the advantage of being exact. Thus to compute a probability by using the above definition
one must be able to count.
(i) n Ω the total number of possible outcomes in the sample space and
(ii) nA, the number of ways in which event A can occur
equally like outcomes are also called equally probable outcomes. If an event cannot occur
its probability is 0, if it must occur its probability is .
the computation of nA and n Ω is easy if Ω has only a few possible outcome, as the
number of possible outcomes becomes large, this method of counting all possible
outcomes are cumbersome and time consuming.
Alternative methods of counting must therefore be developed. For example if one asks
for the probability of getting sum of numbers showing to be 30 when 7 dice are rolled
13
one must determine how many different ways are possible to get the sum to be 30. Such
possible ways include
(6,6,6,6,2,1,3) (2,5,2,6,6,3), (4,4,4,4,4,4,6)
In this chapter weintroduce nontechnical discussion of techniques of mathematics of
counting frequently needed in problems of finding nA and n Ω

Exercise 1.1
1. A die is rolled once. What are the probabilities of getting
(i) An even number
(ii) A prime number
(iii) An off prime number
(iv) An odd number
(v) An even number
Fundamental Principle of Counting
First Law of Counting
If an event A1 can occur in n1 ways and thereafter an event A2 can occur in n2 ways,
“both A1 and A2 can occur in this order in n1n2 ways.

Example 1.5
Roll a die first and then a con
A1 can occur in 6 ways (1,2,3,4,5,or 6) and A2 can occur in 2 ways (H or T).
Thus, by the above law, there are
6 x 2 =12
Possible ways for the outcomes
The outcomes are
(1,H) (1,T) (5, H) (5,T)
(2,H) (2,T) (6,H) (6, T)
(3, H) (3, T) (4, H) (4,T)
In general, if an event A1 can occur in n1 in different ways and if following this an event
A2 can occur in n2 different ways, and if following this second event, an event A3 can
14
occur in n3 different ways and so forth, then the events A1 and A2 and A3…. And Ak can
occur in this order in n1n2….nk ways.
Example 1.6
If a die is rolled 10 times. Let A1denote the outcome of the roll, i= 1,2 ,…. 10. A1 can
occur in 6 ways.
Thus thenumber of ways A1, A2,…. And A10 can occur is 610possible outcome.
That is
N Ω=610
Second Law of Counting
If an eventA1 can occur in n1 different ways and an event A2 can occur in n2 ways then
either A1 orA2 can occur in n1 +n2 different ways.

Example 1.7
Let us toss a die or a coin once. let a1 ne the event “the die shows an even number and a2
be the event “the coin lands heads”. a1 can occur in 3 ways (2, 4or6) and a2 can occur
only ocne.
\the number of ways in which an even number or a head be obtain is
3+ 1 =4
A1 can occur in 3 ways A2 can occur in 1 way
Therefore A1 or A2 can occur in 4 ways
In general, if events A1, A2….,Ak can occur in n1….nk different ways, then either or A1
+ n2 +… + nk different ways

Example 1.8
Four people enter a restaurant for lunch in which there are six chars. In how many ways
can they be seated.
Let A1, A2,A3,A4 denote the events “choice of chair by the four people”, the suffix
denoting order of seating. The first person to sit down has six choices. He can decide tosit
on any of the six vacant chairs. Therefore, there are 6 different ways A1 can occur, after

15
the first person has seated, the second person can sit on any of the remaining 5 chairs.
Thereafter the third person can sit on the remaining 4 chairs.
Thus, using the notation of law of counting
N1 =6, n2 = 5, n3 = 4, n4 = 3
6 x 5 x 4 x 3 =360 ways
They can be seated .
Example 1.9
How many 4 digits numbers can be formed from the digits 0,1,2,3,4,5 if the first digit
must not be 0 and repetition of digits are not allowed.
Let A1, A2, A3, A4, denote mthe events: select the first second, third and fourth digits
[Link] the first digits can not be 0, the first digit can e either 1,2,3,4, or 5
therefore A1 can occur in 5 ways, having chosen the first digit the second digit can be
selected from the remaining 5 digits. there the first and second digits have been chosen,
there remains 4 digits. The third digit can be chosen from the remaining 4 digits and the
fourth can be chosen from the remaining 3 digits. Therefore,A1, A2, A3 and A4can occur
in 5,5,4,3 ways respectively
Thus there
5 x5x4x3 = 300 numbers
Example 1.10
Ten candidates are eligible to fill 4 vacant position. How many ways are there of filling
them?
Let A1,A2, A3,A4 be the events denoting candidates that fill positions 1,2,3,4 respectively.
Candidate filling position 1 can be any of the tend candidates, therefore A1 can occur in
10 ways. Following this, the next position 2 can be filled by any of the remaining 9
candidates since position 1 had been filled by one candidate. Therefore A2 can occur in
ways. Similarly, A3 and A4 can occur in 8 and 7 ways respectively. Thus there are
10x9x8x7 = 5040 ways
Exercise 1.2
1. A student is to answer all the five questions in an examination. It is believed that
the sequence in which the questions are answered may have a considerable effect
16
 on the performance of the student. In how many different order can the question
be answered
2. If a woman has 10 blouses and 6 skirts, in how many ways can she choose a dress
assuming any combination of blouse and skirt matches
3. In a study of plants, five characteristics are to be examined. If there are six
recognizable differences in each of four characteristics and eight, recognizable
difference in the remaining characteristics. How many plants can be distinguished
by these five characteristics?
4. A bus starts with 6 people and stops at 10 different [Link] many different ways
can the 6 people depart if
(i) Any passenger can depart at any bud stop
(ii) No two passengers can leave at the same bus stop
5. Show that the number of ways of choosing r objects from n objects with
replacement is given by n”
Permutation and Combination
A permutation is an arrangement of objects in a definite order (called ordered sample). A
combination is a selection of objects without regard to order (unordered sample)
A group of objects, with regard to permutation and combination has three characteristics:
1. The way the objects in a group are arranged
2. The kind of objects in the group
3. The number of objects of each kind in the group

1.2 permutations:
Suppose that a set contain n objects. We are often interested in arranging the objects in a
definite order.
Two groups containing n object are said to form different permutations if they differ in
arrangements
Consider groups of letters a, b, c, d.
(a, b, c, d): (b, c, d, a): (c, a, d, b)

17
They are all different permutations because the arrangement of the letters is different in
each group.
Example 1.11
How many permutations of four letters can be formed from the letter a, b, c, d.
To answer this question we reason as follows:
Since are permuting 4 letters, four events are involved. Let A1 be the events denoting the
letter to occupy the ith position. A1 can occur in 4 ways, that is the first letter can be
either a, b, c, or d. after this event has taken place, A2 can occur in three ways (that is,
after having chosen the first letter, the letter occupying the second position can be chosen
from the remaining three letters. After the second event, there remains only two letters
from which one is to be chosen tooccupy the third position. So A3is 2 and it remains only
one letter.
A4 can occur only in one way. Thus, there are 4 x 3 x 2 x 1 = 24 permutation of the four
letters. Permutation of four objects from four objects is called permutation 4 and is
denoted by 4P4

Permutation of n Distinct Objects

Consider a set consisting of n district objects. permutation of this set consist of n events
A1, A2, ….An, where the object occupying the ith position is the outcome of A1, I = 1,
2,…..n A1 can occur in n1 ways, A2 can occur in n-1 ways and so on.
Thus there are n(n-1) (n-2)… x 3 x 2 x 1 permutation of the n objects. The number of
permutations of n distinct objects is n!.

Example 1.12
If n balls are distributed at random into r boxes, in how many ways can this be done if
(i) Each of the balls can go into any of the r boxes
(ii) No box has more than one ball.
Solution
(i) Let A1, A2….An where the object occupying the ith position is the outcome of A1 I
= 1,2,…..n. A1 can occur in n1 way, A2 can occur in n-1 ways and so on.
18
(ii) The first ball can go into any of the r boxes, the second any one of the remaining
(r-1) boxes etc, so in all there are r (r-1) (r-2)… (r-n+1) different ways (n < r).

19
Example 1.13
How many permutation of three letters can be formed from the letters a, b, c, d, e. abc,
bae, cba, cdb,….. are few of the required permutation. Since the permutation consist of 3
letters, 3 events are involved, A1, A2 A3. The first letter can be any of the 5 letters,
therefore A1 can occur in 5 ways. Similarly A3 can occur in 3 ways. Thus, there are
5 x 4 x 3 =60
Permutation of three letters from 5 letters. This is denoted by 5P3 (5 permutation 3).
Definition 1.4
The number of permutation of r (r < n) objects from n district objects is called nn
permutation r and is denoted by npr
It can easily be shown that
n
Pr =n(n-1)(n-2)…(n-r+1)
to see this we can argue as follows. Since we are permuting r objects, there are r events
A1, A2…..A3 involved.
The first position can be occupied by any of the n objects, following this the second
position can be occupied by any of the remaining n-1 objects and so on. Therefore A1,
A2,…Ar can occur in n, n-1, n-2, …n-(r-1) ways respectively. thus the number
ofpermutation is
n(n-1)(n-20…(n-r+1)
But
n!= n(n-1) (n-2)…(n-r+1) (n-r)(n-r-1)...1 (3.2.1)
(n-r)! = (n-r)(n-r-1)…2.1
n! = n(n-1)(n-2)…(n-r+1).
(n-r)!

20
Therefore
n
Pr=n(n-1) (n-2)…(n-r+1)
n
Pr= n!
(n-r)!
In example 1.13 above, n = 5, r = 3
5
P3= 5! = 5! = 5 x 4 x 3 = 60
(5-3) 2
Example 1.14
A group of students consist of 5 men and 3 women. The students are ranked according to
their performance in a quiz competition. Assuming no two students obtain the same score
(i) How many different ranking are possible?
(ii) If the men are ranked just among themselves and the women among themselves,
how many different rankings are possible?
Solution
(i) A possible ranking corresponds to a permutation of the students. The number of
possible permutation of the 8 students gives the number of different ranking
possible. Thus the answer is
8
P8 = 8! = 8! = 40320.
(8-8)!
(ii) There are 5P5 = 5! = 120 possible rankings of the men and 3P3 =3! = 6 possible
rankings of the women. It follows from the fundamental principle of counting
that there are
5
P5 x 3P3 = 5! X 3! = 72- possible rankings
Example 1.15
Four digits numbers are to be formed using any of the digits 1, 2, 3, 4, 5, 6, (No repetition
of digit is allowed).
(i) How many four digit number can be formed
(ii) How many 4 digit numbers greater than 3000 can be formed?
(iii) How many 4-digit even numbers can be formed?
21
 (iv) How many of these even 4-digit numbers are greater than 3000?
Solution
(i) This is permutation of 4 digits from 6 digits. Therefore the different permutation is
6
P4 = 6! =360.
2!
Alternatively, we can reason as follow:
The first digit to be selected can be any of the six given digits, so n1 = 6. The second
digits to be select can be any of the remaining 5 digits (since no reparation is allowed) so
n2 = 5. Similarly, n3 = 4 and n4 = 3. Thus, the answer is
N1 x n2 x n3 x n4 = 6 x 5 x 4 x 3 = 360
(ii) Since the number must be greater than 3000, the first digit must be chosen room 3,
4, 5 or 6 so n1 = 4. The second digit can be any of the remaining 5 digits, so n2
=5 similarly, n3 = 4, n4 = 3. Thus, there are
4 x 5 x 4 x 3= 240
4 digit numbers greater than 3000 that can be formed using the digits 1, 2, 3, 4,5, 6,
(iii) A number can be defined to be an even number of the last digit of the number
is even. Going by this definition, we see that the required number is even if the
last digit is 2,4, or 6. Since there is a restriction on the last digit. We have to
select the last digit first. The last digit can be selected from 2,4 or 6, so we
have 3 choices. Having chosen the last digit, we can now select the first,
second and third digit. After chosen the last digits there remains 5 digits, so n1
= 5 similarly, n2 = 4,n3 =3. Thus, the answer is 5x 4 x 3x 3 = 180
(iv) The last digit must be 2, 4 or 6 and the first digit must be 3, 4, 5or 6. If the last
digits is 2, then the number of ways of selecting the first digit is 4. However, if
the last digit is selected is 2 or 6 the number of ways of selecting the first digit
is 3 (that is 3, 5 or 6, 3, 4 or 5)
Case 1: n4 = 1 n1 = 4 n2 =4, n3=3
The number of 4 digits numbers that can be formed in this case is
4 x 4 x 3 x 1 = 48
Case II: n4 =2, n1 = 3n2 = 4, n3 = 3

22
The number of 4 digit numbers that can be formed in this case is
3 x 4 x 3 x 2 = 72
Hence, the total numbers of even 4 digit numbers greater than 3000 that can be formed is
48 +72= 120.

Example 1.16
The letters A,B,C, D and E are placed at random to form a five letters word (without
repetition). How many ways can a word be formed such that
(i) D directly follows A.
(ii) A and D follow each other
(iii) A, D and E follow each other
Solution
(i) For D to directly follow A, we must always have AD appearing in the word so
formed AD can be regarded as a letter so that we now have the letters B,C E
and AD. The number of ways of rearranging tehse new four letters is 4! Thus
there are 4! = 24 ways of forming a word such that D directly follows A
(ii) A and D follow each other; either we have AD or Da each having 4! Ways. Thus
there are 2x 4! Ways of forming a word such that A and follow each other
(iii) ADE can be regarded as a letter so that we now have the letters B.C ADE.
There are 3! Permutation of the new 3 letters. ADE can also be permuted in 3!
Ways, that is
ADE, AED, DAE, DEA, EAD, EDA.
Thus there are
3! X 3! = 36 ways
Permutation of Indistinguishable Objects
Consider n objects where n1 are of type 1, of type 2,…. Nk of type k. ,in shown many
ways can the n object be arranged.
For example in how many ways can the letters of the word book be arranged. here n= 4, k
= 3, n1= 1, n2= 2, n3 = 1. First give the two O’s suffixes bo1o2k. then treating the O’s as
different, the 4 letters may be arranged in 4! Ways. In every distinct arrangement, the
23
2O’s may be rearranged amongst themselves in 2! Ways without altering the permutation
for instance o1bo2k are the same when the suffixes are removed.
Therefore, the number of permutations of the letters of the word book is
4! = 6
2!
Book, ookb, oobk, obok, okob, koob.
In general, the number of ways in which n objects where n1 are of type 1, n2 of type 2…,
nk of type k can be arranged is given by
n!: n + n2 +…+ nk
n1!n2!...nk!

Example 1.17
How many distinct permutations are there of the letters of the word Television?
The ten letters to be permuted consist of 2e’s, 2i’s,IT, I, v, s, o, n. thus the number of
distinct permutation is
10! = 10!
2! 2! 1! 1! 1! 1! 1! 1! 2! 2!
1.3 Combinations
Definition 1.5
Two groups are said to form different “combination” if they differ in the number of
any kind of object in the groups. Consider a group of 4 letters a, b, c, d. the
combination abcd, bcda, cadb are identical combinations each of them contains the
same number of a, b, c, d: one b, one c and one d.
The combinations of the 4 letters taken 3 at a time are:
Abc, acd, abd, cbd
Therefore, there are 4 distinct combination for three letters from the four letters. Each
of tehse combinations has 3! = 6 permutation. For instance.
abc = abc, acb cab cba bac bca
acd = acd adc cad cda dac dca
abd = abd adb bad bda dab dba

24
 cdb = cbd cdb bcd bdc dbc dcb
The number of distinct permutation is 4P3=24
The number of distinct combinations is 4.
Therefore.
Number of combinations = number of permutations
3!
The number of distinct combination of 4 objects takn 3 at a time is denoted by 4C3.
Thus,
4
C3 =4P3/3!
In general,
n
Cr =nPr/r!= n!
(n-r)!r

Theorem 1.1
The number of distinct combinations of n objects taken r at a time (that is the number
of ways of choosing r objects out of n, disregarding order and without replacement) is
given by
Proof:
There are nPr = n!
(n-r)! permutations of n objects taken r at a time. If we disregard
order among the r objects, there are r! permutation that will give the same
combination. Therefore the number of combinations is the number of permutation
divided by r! thus.
n
Cr = nPr = n!
r! (n-r)!r!
Example 1.18
A club consist of 15 members. In how many ways can a committee of 3 to be chosen?
Solution
This can be done in 15C3 ways
15
C3 = 15! = 15 x 7 x 13 = 455 ways
25
 3! 12! 3

Example 1.19
A club consist of 10 men and 5 women, in how many ways can a committee of 6
consisting of 4 men and 2 women be chosen. The 4 men can be chosen from the 10
men in10C4 ways, the 2 women can be chosen from the 5 women in 5C2 ways. Hence
the committee can be chosen in (by the fundamental principle f counting)
10
C4 x 5C2= 10! X 15! = 2,100 ways
6!4! 2!3!
Exmaple1.20
Suppose we have a box containing n balls of which r are black and the remaining
white. A random sample of size k is selected without replacement. In how many ways
can the sample be selected such that it contains x black balls
The x black balls can be selected in rCxways and k-x white balls n be selected from n-
r white balls. Thus, there are
r
Cx Xn-rCk-r
ways of selecting a sample such that it contains x black balls and k-x white balls

Example 1.21
A committee of 4 men and 2 women is selected from 10 men and 5women. If two of
the men are feuding and will not serve on the committee together, in how many ways
can the committee be selected

Solution
The number of ways of selecting 4 men and 2 women is
10
C4 x 5C2 =2100
The number of ways of selecting the committee such that the two men are in the
committee is
8
C2 x 5C2 =280

26
Hence, the number of different committees that can be formed such that the two men
are not in the committee together is
2100 – 280 = 1820
Another method is to consider 3 cases
Case I: The two men say, A and B are not in the committee
8
C4 x 5C2= 700
Case II: A is the committee but not B:
8
C4 x 5C2 = 560
Case III: B is the committee but not A
8
C4 x 5C2 = 560
Thus, the total number of ways the committee can be selected is
700+ 560+ 560 = 1,820
Example 1.22
How man subsets can be formed, containing at least one member from a set of n
element

Solution
There are nCk subsets of size k that can be formed, the total number of subsets
containing is least one member is

Example 1.23
A student is to answer 5 out of 8 question in an examination. How many if he must
answer at least 2 of the first for 4 questions?

Solution
(i) By the combination law the answer of 8C5 = 56
(ii) Possible choices are (2,3) (3,2) (4,1) where (2,3) means answer 2 questions from
the first 4 questions and 3 from the remaining 4 questions. The number of ways
of doing this is 4C2 x 4C3.
Thus the answer is
27
 (4C2 x4C3) + ( 4C3 x 4C2) + (4C4 x 4C1)= 52

1.4 Partitioning
Dividing a population or a sample of n objects into k ordered parts of which the first
contain r object, the second r2 objects and so on is called ordered partition. If the
division is into k unordered parts, then the partition is said to be unordered. For
example suppose a class contains 15 students and we want to divide the class into 3
tutorial groups of 5 each. Three lecturers are available and each is to take each group.
In otherwise, we want to divide the 15 students into 3 ordered groups (A,B,C) this is
an ordered partition since there are
3! = 6 ways
The lecturers can be assigned to take any partition, for instance groups (A, B, C) can
be taken by
(L1, L2, L3) or (L2, L1, L3) or (L1, L3, L2)
Or (L2, L3, L1) or (L3, L2, L1) or (L3, L1, L2)
Where L1 means lecturer I ad (L1, L2, L3) means L1 takes groups A, L2 takes group B,
L3 takes group C and so on
15
There are C5 ways of selecting those students those students to be in group A,
following this, there are 10 students left and so there are 20C5 selecting those to be in
the second group, therefore by the fundamental principle of counting, there are
15
C5 x 10C5x 5C5= 15! X 10! X 5!= 15!
10! 5! 5! 5! 5! 5! 5! 5!
Ordered partitions. The number of ways in which n object can be divided into ordered
parts of which the fist contains r1 objects the second r2objects and so on is
n
Cr1. n-r1Cr2.n-r1-r2Cr3…n-r1-r2-…-rk-1Crk

= n! . (n-r2)! (n-r1-r2)! (n-r1-r2…rk-1! = n!

(n-r1)!r1! r2!(n-r1-r2)! (r3!)n r1-r2-r3)! Rk!(n-r1-r2…rk-1-rk)! r1!r2!r3!...rk!

Where n = r1 +r2+ … + rk.

28
Example 1.24
In how many ways can three committees of five, three and two persons be formed
from 10 persons.
We seek the number of ordered partitions of the 10persons
This is given by
10! =2520
5! 3! 2!

Example 1.25
In how many ways can 9 toys be divided among three children if each gets 3 toys
If the toys are numbered 1 through 9, the partition {(1, 2, 3), (4, 5, 6), (7, 8, 9}) means
child A gets toys 1, 2, 3 child B gets toys (4, 5,6) while child C gets toys 8, 8, 9. We
distinguish between {(1, 2, 3), (4, 5,6) (7, 8, 9)} and {(4, 5, 6), (1, 2, 3), (7,8, 9) so
these are ordered partitions. Thus there are
9! = 1680 ways
3! 3! 3!
When r1 = rj for I = j, we can distinguish between ordered and unordered partition for
example partition a set consisting of 8 objects numbered 1 to 8 into 3 parts of which
the first contains 2 objects, the second 4 objects and the third 2 objects. A partition is
{(1, 2), (3, 4, 5,6), (7, 8)}.
In an ordered partition we distinguish between the partition
{(1,2), (3,4, 5, 6), (7, 8)} and {(7, 8), (3, 4, 5, 6), (1, 2)}
But they are the same for unordered partition. Therefore, the number of unordered
partition is
8! . 1 = 210
2! 4! 2! 2!

Example 1.26

29
In how many ways can a family of 9 divide itself into 3 groups so that each group
contains 3 persons?
Solution
We are seeking for unordered partitions r1 = 3, r2=3 r3=3
The number of unordered partitions is
9! 1 = 280
3! 3!3! 3!
Since the three parts contain the same number of objects
The same relationship that exists between permutation and combination exist between
ordered and unordered partition of a set.

Example 1.27
In how many ways can a family of 10 be divided into three groups, one containing 4
and the others 3?
Solution
R1=4,r2 =, r3=3
The number of ordered partitions is
10! =4,200
4! 3! 3!
And the number of unordered partition is
10! 1 = 2,100
4! 3! 3! 2!
Since only two of the three parts contain the same number of objects

1.5 Selection of Non distinct Objects

The number of ways r districts balls can be distributed into n cells in r. the number of
ways if a specified cell contains exactly k balls (k = 0, 1,2…r) is rCk (n-1)r-k
That is the balls can be chosen in rCk ways and the remaining r-k balls can be paced
into the remaining n-1 cells in (n-1)r-k ways

30
Now, suppose the balls are non-district (indistinguis-shable) we can only talk about
number of balls in the ith cells
Let X1, x2…, xn denote the number of balls in the ith cell, then
x1 + x2 +…+xn = r
The number of distinct distribution in which no cell remains empty is
r-1
Cn-1
to see this let us assume that the r non-distinct objects are lined by and bars used to
divide them into groups. The r balls r-1 space of which n-1 are to be occupied by bars.
For example if r =9 and n = 5, we have
0∩0∩0∩0∩0∩0∩0∩0∩0
Thus 0/00/0/000/00 corresponds to
X1 =1,x2 = 2, x3 = 1, x4 = 3, x5 =2.
And there are 8C4 possible distribution of the bars. another possible distribution is
000/0/00/00
If x1> 0, that is cells can remain empty, the number of non-negative solutions of
X1 + X2 +…+ xn=r
Is the same as the number of positive solutions of
Y1 + y2 +…+ yn =r + n
Where y1 = x1 +1. Thus, there are
r + n-1
Cn-1 (1.2)
Distinct solution satisfying x1 + x2 +… + xn = r
The number of distinct distribution is the number of ways n – 1 spaces can be selected
out of the n + r-1 spaces

Application to Runs
Definition
A run is any ordered sequence of elements of two kinds
For example, by a run of wins we mean a consecutive sequence of wins. The
sequence WWWLWWLLWLWW gives 4 runs of wins. The first run is length 3, the
second run of length 2, the third of length 1 and the fourth of length 2.
31
 Suppose now that we have two letters (W and L) n non-distinct letters (L) and m non-
(n+m)
district letter (W). the total number of distinct orderings of W and L is Cn-r r runs
of W is equivalent to arranging the letters W into r cells none of which is empty. If
there are r runs on W,
The number of L run is necessarily r+ 1, r-1 or. Thus from (1.1) we have (m-1)Cr-
1distinct was of having r runs of W. hence there are.
(m-1)
Cr-1(n-1)Cr ways of having r runs of W and (r + 1) runs of 1…
A sequence representing r runs of W is

LL… WW…W LL… WW…W … WW… L..L

Y1 X1 Y2 X2 Xr Yr+1
Where y1 + y2 +… yr + 1= n, yi> x1 + x+2+ +… + xr = m, xi> 0.
Let y1 + 1, yi = yi, I = 2, …, r, y r+1 = yr+1 + 1
The number of non-negative solution to
Y1 y2 + … + y1+1 = n, (yi> 0)
Is the same as number of positive solution to
Y1+ y2 +…+ yr+1= n +2
Thus, the number of outcomes that result n r runs is n+1 Cr
Hence, n-1Crm-1Cr-1 is the total number of ways of having r runs of W
Exercise 1.3
1. How many 4 digits numbers can be formed from the digits 0,1, 2, 3, 4,5, 6, 7,8, 9
if the first digit must not be 0 and repetitions is allowed. how many of these
number are (i) even (ii) less than 5,000?
2. A code consist of five symbols. The first three symbols are letters and the last two
are digits. How many codes can be made if no letter nor digit is repeated in any
code word?
3. Suppose n objects are permuted at random among themselves. In how ways can
this be done such that k specified objects occupy k specified positions
4. How many distinctarrangements are there of the letters of word (1) University (ii)
biology (iii) Mississippi
32
 5. The number 1,2,3… n are arranged in random order. in how many ways can this
be done such that
(i) Q and 2 follow each other
(ii) 2, 3 and 4 follow each other in that order
6. Show that the number of distinct ordered sample of size r that can be drawn from a
population with n objects is
(i) Nr if sampling is with replacement and
(ii) n1 if sampling is without replacement
n –r!
7. A total of n balls are randomly placed into n cells. In how many ways can this be
done. In how many ways if each cell is occupied
8. A warehouse has 6 different containers to be distributed among 10 retailers. In
how many ways can this be done? How many way if no retailer receives more
than one container?
9. Prove the following identities
n
a. Cx = nCn-x (b) m+1Cx = nCx +nCx-1
10. (a) In how many ways can 4 boys and 2 girls be arranged to sit in a row?
(b) In how many ways if only the boys must sit together?
(c) In how many ways if the boy and no girl must sit together?
(d) In how many ways if no boy and no girl must sit together?
11. In howmany ways can a football team be selected from 15 players. In how many
ways if 6 particular players must be included in the team?
12. From a box containing 5 red, 4 white and 3 black marbles, three marbles are
drawn one after the other without replacement, how many ways can this be done if
(i) all are white (ii) 2 re white and 1 is red (iii) at least one is black
13. A pair of dice is rolled once. In how many ways can
(i) The sum of the two numbers appearing exceeds 8
(ii) The maximum of the two numbers is greater than 4
(iii) The minimum of the two numbers is greater than 4

33
14. A disciplinary committee of four is to be chosen from six men and five women.
One particular man and one particular woman refuses to serve if the other person
is on the committee. How many committees may be formed
15. Eleven people are to travel in two cars- salon and station wagon. The saloon has 4
sets and the station going 7 seats. In how many ways can party be split up?
16. In how many ways can a committee of 6 composing of 3 full professors, 2
associate professors and 1 senior lecturer be selected from 5 full professor, 10
associate professors and 20 seniors lecturers
17. A box contain 12 balls labeled 1, 2, 3… 12, suppose a random sample of size 4 is
selected. In how many ways can the sample be selected if balls labeled 2, 3, are
among the four selected.
18. Suppose a random sample of size r is drawn from a population of n objects. In
how many ways can this be done if k given object must be included in the sample
and (a) sampling is without replacement (b) Sampling is with replacement
19. I bought 2 tickets to a lottery for which n tickets were sold and 4 prizes to be given
in how many ways can the tickets be drawn such that 1 win at least a prize?
20. A committee of 8 is to be formed from 10 couples (10men and 10 women). In how
many ways can the committee be formed if no husband serves on it with his wife.
21. Lines are drawn to pass through six points. In how many ways can this be done if
each line passes a through only two points?
22. Interchanges may occur between any two of the n chromosomes of a cell
(a) In how many ways can exactly one interchange occur?
(b) In how many ways can exactly k interchanges occur?
(c) If n = 5, in how many ways can at most three inter changes occur?
23. Prove that (n1-n2)Ck = n1Crn2Ck-r where k < n1, n2
HINT: Select k objects from n1 + n2 obejcts of n1 of type 1 and n2 of typeII
24. A company is considering building additional warehouse at new locations. There
are ten satisfactory location and the company must decide how many and which
ones to select. How many choices are there?

34
25. Are there more samples obtainable in five draws from 10 objects with replacement
than 12 objects without replacement?
Each of fifty items is tested and found to be defective or non-defective. How many
possible outcomes are there?

MODULE TWO

UNIT ONE

ELEMENTARY PRINCIPLE OF THE THEORY OF PROBABILITY

In chapter 1, we considered some elementary methods of mathematics of counting

essential for determining probabilities of events. In this chapter, we continue our study by
discussing how to apply the knowledge gained in chapter 1 to determine probabilities of
events and general properties of probabilities. We present some theorems and definitions
that are basic to the understanding of commonly encountered problems.

Example 2.1

Suppose a fair die is rolled once. There are six possible outcomes. The sample space is

Ω = {1, 2, 3, 4, 5, 6}.

let A1, A2 and A3, represent the following events

A1 = “an even number occurs”

35
 A2 = “an odd number occurs

A3 = a prime number occurs”

The six possible outcomes in Ω are equally likely. The probability that A occurs is
denoted by P(A) and the probability that A does not occur by P(AC). if we let nA be the
number of outcomes that have attribute A, then

A1 = {2, 4, 6}
A2 = {1, 3, 5}
A3 = {2, 3, 5}

Thus,

nA1 3 3
P( A1 ) = = 3 / 6 = 1 / 2, p ( A2 ) = , P ( A3 ) =
nΩ 6 6

Example 2.2

Suppose that a box contains 10 items of which are defective. Two items are selected at
random without replacement. Find the probabilities that:

(i) both items are non-defective (ii) only one item is defective

(iii) both items are defective: (iv) at least one item is defective

Let A1, A2, A3, A4, denote the events

“both items are non-defective

“only one item is defective”

“both items are defective”

“at least one item is defective”

respectively. The number of ways of selecting 2 items from 10 is

10
C 2 = 45 ways.

36
So there are 45 elements in the sample space.

(i) The number of ways of selecting 2 items from the non-defective items is 6C2 = 15.
That is A1 can occur in 15 ways. Thus.

/ / 173
*+ 01- 34
,
/
-. + 02 54
P(A1) =
,

(ii) The number of ways of selecting 1 item from the 4 defective items and 1 item
from the 6 non-defective items is 4C1 × 6C1 = 24 so, A2 can occur in 2 ways.
Thus,

/ /
8+ 9 * + :5 ;
- -
-. + 54 34
,
P(A2) =

Similarly

/ /
8+ < :
,
-. + 54 34
,
(iii) P(A3) =

(iv) At least one defective means 1 or 2 defective items. So,

= / 273.
; :
34 34
P(A4) = P(1 defective item) + P(2 defective items) = P(A2) + P(A3) =

Example 2.3

Suppose a fair die is rolled twice. Find the probability that the sum of the numbers on the
two faces is (i) even, (ii) less than 5.

The sample space Ω, consists of 36 elements.

(i) Let A be the event “the sum of the two faces is even”. Possible outcome are:

(1,1) (1,3) (1,5) (2,2) (2,4) (2,6),…

(3,1) (5,1) (4,2) (6,2),…

A can occur in 18 ways. Thus,

37
 / 1⁄2
3;
><
P(A) =

(ii) Let B be the event “the sum is less than 5”. B occurs if the sum is 2, 3, or 4. The
sum is 2 if the outcome is (1,1) the sum is 3 if the outcome is (1,2) or (2,1) and the
sum is four if the outcome is (3,1), (1,3) or (2,2). Therefore B can occur in

1 + 2 + 3 = 6 ways

Thus

/ 176.
<
><
P(B) =

Example 2.4

3 balls are drawn at random with replacement from a box containing 8 red and 3 white
balls. Find the probability that (i) all 3 are red; (ii) 1 is red and 2 are white.

The sample space consists of 113 possible outcomes. Let A be “the evet all 3 are red” and
B the event “1 is red and 2 white”

/
01 ;A
02 33A
P(A) =

(The first red can be chosen in 8 ways the second in 8 ways and the third Red in 8 ways).

0B
02
P(B) = , nB = 8,3.3 + 3.8.3 + 3.3.8

(8 × 3 × 3) = number of ways of picking the first to be red, second white and third white
(RWW), 3 × 8 × 3 is for WRW and 3 × 3 × 8 is for WWR). Thus

.
>.;.>.>
33A
P(B) =

2.1 Properties of Probability

Definition 2.1: Mutually Exclusive events

Two events A1 and A2 are mutually exclusive if an only if

38
 A1 Aj = for every i ≠ j.

That is, two or more events are mutually exclusive if no two of them have points in
common. In example 2.1

A1 = {2, 4, 6}

A2 = {1, 3, 5}

A1 A2 = ,

Therefore A1 and A2 are mutually exclusive events.

Definition 2.1

If A1 and A2 are any two events, A1 A2 is the event that occurs if and only if both A1
and A2 occur. In general, if A1, A2, …, An occur.

In example 2.1. A1 A2 A3 is the event that an even prime occurs.

A1 A3 = {2}

Thus

The event A2 A3 is the event that an odd prime occurs A2 A3 = {3,5}.

Thus

P(A1 A2) =

Complementary Event

39
The event “A occurs” and “A does not occur” are mutually exclusive. The event “A does
nt occur” is called the complement of A and is denoted by AC.

Theorem 2.1

If AC is the complement of an event A, then P(Ac) = 1 – P(A). This theorem states that the
probability that an event will not occur is equal to 1 minus the probability that it will
occur.

In example 2.2 (iv); The event “no defective item” is the same as the event both items are
non-defective”.

From Theorem 2.1, we have

P(A4) = 1 – P(

A4c = A1 since the event “no defective item” is the same as the event” both items are non-
defective.

The following events are complementary events

(i) at most 2 and greater than 2

(ii) at least 4 and less than 4.

Definition 2.2

If A1 and A2 are any two events, A1 A2 is the event that occurs if t least one of A1 and
A2 occurs.

In general, if A1, A2,…, An are any events, A1 A2 is the event that occurs if at least one
of A1 i = 1, 2,…, n, occurs.

In example 2.1, the event A1 A3 is the event that an even or a prime number occurs.

A1 A3 = {2, 3, 4, 5, 6}.

40
Thus

P(A1 A2 ) = =

The event A1 A2 is the event that an event or an odd number occurs.

A1 A2 = {1, 2, 3, 4, 5, 6} = Ω,

thus

P(A1 A2) = P(Ω) = = 1.

Theorem 2.2

If A1 and A2 are any events, then

P(A1 A2) = P(A1) + P(A2) – P(A1 A2).

The proof of this theorem follows directly from the definition

dividing through by nΩ the result follows.

Corollary

(i) If A1 and A2 are mutually exclusive event then

P(A1 A2) = P(A1) + P(A2)

since A1 and A2 are mutually exclusive,

A1 A2 = ϕ

hence

41
 P(A1 A2 ) = 0

(ii) If A1 and A2 are any two events

P(A1 A2) ≤ P(A1) + P(A2)

In Example 2.1 above,

P(A1 A3) = P(A1) + P(A3) - P(A1 A3 ) = = .

(iii) If A and B are any two events defined on the sample space then

P(A) = P(A B) + P(A Bc)

P(B) = P(A B) + P(Ac B)

In general, if A1, A2,…An are any n mutually exclusive events, then

P(A1 A2 … An)= P(A1) +…+ P(An).

Theorem 2.2 can be generalized for n > 2 events. It can easily be shown that for any 3
events A1, A2, A3

P(A1 A2 A3) = P(A1) + P(A2) + P(A3) - P(A1 A2 )

- P(A1 A3) – P(A2 A3) +…+ P(A3).

To see this, let B = A1 A2, then

P(A1 A2 A3) = P(B A3) = P(B) + P(A3) – P(B A3 )

P(B) = P(A1) + P(A1) + P(A2) – P(A1 A2)

P(B A3) = P(A3 B) = P{(A3 (A1 A2)} = P{(A3 A1 ) (A3 A2)}

= P(A3 A1) + P(A3 A1) + P(A3 A2) – P(A1 A2 A3 )

42
(by theorem 2.2). thus

P(A1 A2 A3) = P(A1) + P(A2) – P(A1 A2) + P(A3)

- P(A1 A3) – P(A2 A3) + P(A1 A2 A3 )

Note:

A1 A2 A3 is the event that at least one of them will occur.

Definition 2.3

Two or more events are said to be exhaustive if their union equals the whole sample
space. In other words, A1, A2,…, An are exhaustive events if

P(A1 A2 … An) = 1

In other words it is certain that at least one of them will occur.

Note:

1. 0 ≤ P(A) ≤ 1

2. P(Ω) = 1.

3. For an sequence of mutually excusive events A1, A2,…

Examples:

2.5 A box contains 6 balls numbered 1 to 6. A ball was drawn from the box at random.
Find the probability that the number on the ball drawn was either 1, 2, or 6.

Solution

43
 Let A1, A2, A3 denote the events that the ball drawn was 1, 2 and 6 respectively.
A1 A2 A3 denote the event the number on the ball drawn was either 1, 2 or 6.
A1, A2 and A3 are mutually exclusive events. Thus

P(A1 A2 A3) = P(A1) + P(A2) + P(A3)

=1/6 + 1/6 + 1/6 = 12.

2.6 Four fair dice are tossed once, what is the probability that the sum of the numbers
on the four dice is 23?

Solution

The possible outcomes to get 23 are

(6,6,6,5), (5,6,6,6), (6,5,6,6), (6,6,5,6).

The sample space Ω consists of 64 elements.

Therefore, the probability that the sum of the numbers on the four dice is 23 is
4/64.

2.7 If P(A1) = 2/3, P(A1 A2) = ¼ and P(A1 A2) = 5/6, find P(A2).

Solution

From the addition law of probability

P(A1 A2) = P(A1) + P(A2) – P(A1 n A2)
We have
P(A2) = P(A1 A2) P(A1 n A2) – P(A1) = 5/6 + ¼ - 2/3 = 5/12.

2.8 Suppose a fair on is tossed three times, what is the probability that at least one
head occurs?
Solution
44
 Let A1 be the event that the first toss lands heads, A2 the event that the second toss
lands, and A3 the event that the third toss lands heads.
A1 A2 A3 is the event that at least one head occurs.

We are required to compute P(A1 A2 A3) The complement of A1 A2 A3 is

A1c n A2c n A3c. That is, the first does not land heads, the second does not and the
third does not.
A1c n A2c n A3c = {T, T, T}.
Thus
P(A1c n A2c n A3c) = 1/8.
Hence
P(A1 A2 A3) = 1 – 1/8 = 7/8.

2.9 If A1 A2, then P(A1) ≤ P(A2)

A2 = A1 (A2/A1).
Since A1 and A2/A1 are mutually exclusive, we have
P(A2) = P(A1) + P(A2/A1)≥ P(A1)
Since P(A2/A1) ≥ 0.
2.9 If A1, A2,…, An are n events, then

=
From de Morgan’s law, we have
=
hence

2.2 Conditional Probability

The conditional probability of an event A given that an event B has occurred is denoted
as P(A|B). the word “given” is represented by the upright stroke. We wish to determine

45
the probability that an event A will occur “conditional on” the knowledge that another
event B has occurred.
Suppose a fair die is rolled and it is known that an even number appeared uppermost. Let
A be the event that the number was greater than 3 and B the event that the number that
appeared was even. The problem is to find the conditional probability that the event A
occurred given that the event B has occurred, P(A|B). since we know that the number was
even, the number must be either 2,4 or 6. Therefore, the conditional sample space
contains 3 elements. The event A occurs if the number showing is 4 or 6, thus
P(A|B) = 2/3.
We can therefore define conditional probability of A given Bas the number of ways AnB
can occur divided by number of elements in the conditional sample space. That is,
,

Where ≠ 0, where ΩB is the condtional sample space given that B has occurred.
Divide the numerator and denominator by nΩ

In the above example,

n(A B) = 2
n(B) = 3
Definition 2.4
Let A and B be two events such that P(B) > 0. Then the conditional probability of A

given B, denoted by P(A|B) is defined to be P(A|B) = 2.2

Examples
2.11 Suppose a box contains 4 red balls and 3 black balls. Compute the probability that
(i) the second ball drawn is red if the first ball drawn was red; without
replacement,
(ii) the second ball drawn is red if the first ball drawn was black

46
 Solution

If the first ball drawn is red, there remains 6 balls, 3 red balls and 3 black balls.
The probability of the second ball being red is 3
6 = 12 . But if the first ball is black,

the box is left with 4 red and 2 black so the probability of the second ball being red
is then 4/6 = 2/3. Thus

P(2nd is re/first was red) = 3/6/ = ½

P(2nd is red/ first was black) = 4/6 = 2/3.

This shows that the probability of the event “the second ball drawn is red” depends
on the colour of the first ball drawn.

2.12 Suppose two fair dice are rolled. If the sum of the numbers appearing is 6, what is
the probability that one of the number is 2?

Solution

Let A be the event “one of the numbers is and B the sum is 6. there are five ways
for the event B to occur: (3,3), (2,4), (4,2), (5,1) and (1,5) and there are two ways
for the event AnB to occur: (2,4) and (4,2).

Thus,

P( A ∩ B) = 2 36 , P( B) = 5 36

Hence,

P( A ∩ B)
P( AΙB ) = = 2 5.
P( B )

47
2.13 There are two children in a family. If there is at least a girl in this family, what is
the conditional probability that both are girls.

Solution

The sample space is

Ω = {BB , GB , BG , GG}

Let A be the event “both children are girls and B ”a least a girl in the family.

B = {GB, BG, GG}, A = {GG}, A∩B = {GG}.

P ( AnB ) 1 4 1
P(A Ι B) = = = .
P( B ) 34 3

2.14 There are three children in a family. If there is at least one boy and at most two
boys in this family. What is the conditional probability that there are exactly two
boys in this family.

The sample space is

Ω = {BBB, BBG, BGG, BGB, GBB, GBG, GGB, GGG}

Let B be the event “at least one boy and at most 2 boys in the family” and let A be
the event “exactly two boys in the family”. Then

B = {BBG, BGG, BGB, GBB, GBG. GGB}

A∩B = {BBG, BGB, GBB}.

Therefore

P ( AnB ) 3 8 3 1
P(A Ι B) = = = = .
P( B ) 68 6 2

Exercises 2.1

1. A fair die is thrown twice.

48
 (i) If it is known that the sum of the numbers appearing was 8, what is the
probability that the difference between the two numbers was 2.

(ii) If it is known that the difference the two numbers was 3, what is the
probability that the sum of the two numbers was 7?

2. Two unbiased dice are thrown once. What is the probability that

(i) at least one 5 is thrown.

(ii) the sum is 10

(iii) the sum is 10 given that no 5 is thrown?

3. Suppose events A and B are sun that P(A) = 1/5, P(AnB) = 1/6.

Find (i) P(BIA); (ii) P(Ac ∪ Bc).

4. If A and B are two events defined on the same probability space, show that: (i)
P(A) = P(A∩B) + P(A ∩ Bc) = P(B) + P(A ∩ Bc) – P(Ac ∩ B).

5. Prove that P(Ac ∩ Bc) = 1 – P(A) – P(B) + P(A ∩ B).

6. Suppose a well balanced coin is tossed twice.

Find the conditional probability that

(i) both coins show a tail given that the first shows a head;
(ii) both are heads given that at least one of them is a head.
7. A red die and a green die are rolled once. Find the conditional probability that:

(i) the number on red die is odd, given that the sum of the two numbers
showing is 9;

(ii) the sum of the two numbers is 9 given that one of the numbers is odd and
the other even?

49
2.3 Bayes Theorem

suppose a box contains r red balls and b black balls. Two balls are drawn at random
1
without replacement. Assume that the probability of drawing any particular ball is r +b
.

Let A1 be the event “the first ball drawn is red and let A2 be the event” the second ball
drawn is red. Then

r r −1
P( A1 ) = , P( A2 Ι A1 ) =
r +b r + b −1

(
P A2 Ι A1 =
c
) r
r + b −1

The probability of the event A2 depends on A1 and A1c.

That is A2 is equivalent to A2 n A1 ∩ A1c

A2 = A1 ∩ A2 or A1c ∩ A2.

Therefore,

P(A2) = P(A2 ∩ A1) + P(A2 ∩ A1c) = P(A1) P(A1) P(A2 Ι A1) + P(A1c) P(A2 Ι A1c).

Theorem 2.3

If the probability of an event B, depends on k mutually exclusive and exhaustive events

A1, A2 +…+ Ak, then

P(B) = P(B ∩ A1) + P(B ∩ A2) +…+ P(B ∩ Ak)

k k
= ∑ P( B ∩ A ) = ∑ P( A ) P( BΙA ).
i =1
1
i =1
1 1

Example 2.15

Suppose a box contains 3 red balls, 2 black balls and 5 green balls. Two balls are drawn
at random without replacement. Find the probability that the second ball drawn is red. Let
A1 be the event “the first ball drawn is red A2 the event the first ball drawn is black and

50
A3, the event “the first ball drawn is green. Let B be the event “the second ball drawn is
red. The event B occurs if

(i) the first ball is red and the second is red or

(ii) the first ball is black and the second is red or

(iii) the first ball is green and the second is red.

Thus

B = B ∩ A1 or B ∩ A2 or B ∩ A3, P(B Ι A1) = 2/9, P(B Ι A2) = 3/9,

P(B Ι A3) = 3/9.

P(B) depends on A1, A2, A3 which are mutually and exhaustive events. Therefore,

P(B) = P(B ∩ A1) + P(B ∩ A2) + P(B ∩ A3)

= P(A1) P(B Ι A1) + P(A2) P(B Ι A2) + P(A3) P(B Ι A3) = 3/10.2/9 + 2/10.3/9 +
5/10.3/9

since P(A1) = 3/10, P(A2) = 2/10, P(A3) = 5/10.

Thus

P(B) = 1/90 (6 + 6 + 15) = 27/90 = 3/10.

Example 2.16

Suppose a factory has three machines M1, M2, M3 which produce 60%, 30% and 10% of
the total production respectively. Of their output, machine M1 produces 2% defective
items, machine M2 produce 3% defective items while machine M3 produces 4% defective
items. Find the probability that a part selected at random is defective.

51
 Solution

Let B be the event “a part selected at random is defective”. A defective item could have
been produced by either machine M1, M2 or M3. Thus

B = (B ∩ M1) U (B ∩ M2) U (B ∩ M3)

Since (B ∩ M1), (B ∩ M2), (B ∩ M3) are mutually exclusive events. The following
information is contained din the equation.

P(M1) = 60% = 0.6, P(M2) = 0.3, P(M3) = 0.1

P(B|M1) = 0.02, P(B|M2) = 0.03, P(B|M3) = 0.04

Hence

P(B) = P(M1) P(B|M1) + P(M2) P(B|M2) + P(B|M3) P(M3)

=(0.6 × 0.02) + (0.3 × 0.03) + (0.1 × 0.04) = 0.002 + 0.009 + 0.004 = 0.025.

suppose you are now asked, what is the probability that a given defective part was
produced by machine M1. that is, you are to find P(M1|B) =P (a part was produced by
machine M1 given that the part was defective).

NOTE:

1. P(M1 | B) ≠ P(B|M1), but

2. P(M1n B) = P(BnM1)
3
P(B) = ∑ P( M
i =1
1 n B)

hence,

P( M i n B) P( M i ) P( B | M i )
P(M1|B) = 3
= 3
2.4
∑ P( M
i =1
i n B) ∑ P( M
i =1
i ) P( B | M i )

52
Thus,

0.6 × 0.02 0.012

P(M1|B) = = = 0.48
0.025 0.025

0.3 × 0.03 0.009

P(M2|B) = = = 0.36
0.025 0.025

0.1× 0.04 0.004

P(M3|B) = = = 0.16
0.025 0.025

Equation (2.4) is an example of Bayes theorem which may be stated as follows.

Bayes Theorem

If A1, A2,…Ak are set of mutual exclusive and exhaustive events in a sample space Ω and
B is any other event in Ω such that P(B) > 0, then

P ( Ai ) P ( B | A i )
P(Ai|B) = k
; i = 1, 2,..., k 2.5
∑ P( A ) P ( B | A )
i =1
i i

Example 2.17

Suppose a college is composed of 70% male and 30% female students. It is known that
40% of the male students and 20% of the female students smoke cigarette. Find the
probability that a student observed smoking a cigarette is male?

Let M,F denote male and female respectively and S denotes smoker. The above problem
contains the following information.

P(M) = P(A student selected at random is male) = 0.7

P(F) = P(A student selected at random is female) = 0.3

P(S|M) = P(a student selected at random smokes given that the selected student is male) =
0.4
53
P(S|F) = P(a student selected at random smokes given that the selected student is female)
= 0.2

By Bayes theorem, we have

P(M|S) = P(a student observed smoking is male)

= P(A student selected at random is male given that the selected student is a
smoker)

P( M n S )
=
P( S )

Where P(S) = P(a student selected at random is a smoker)

= P(S ∩ M) + P(S ∩ F) = P(M)P(S|F)

= (0.7 × 0.4) + (0.3 × 0.2) = 0.28 + 0.06 = 0.34

Thus

P( M ) P ( S | M) 0.7 × 0.4 0.28 14

P(M|S) = = = = .
P(S ) 0.34 0.34 17

Example 2.18

A table has drawers. Drawer 1 contains two red and five black biros, drawer II contains
four red and three black biros and drawer III contains one red and six black biros. A
drawer is chosen at random and a biro is chosen from the drawer. Find the probability
that

(i) the biro chosen is black

(ii) the biro chosen is from drawer I if the chosen biro is black.

Solution

54
Let P(i) denote the probability that drawer I is selected (I – 1, 2, 3) and R and B
representing red and black biros respectively.

Then

P(1) = 1/3, P(2) = 1/3, P(3), = 1/3

P(B|1) = 5/7, P(B|2) = 3/7, p(B|3) = 6/7

From equation (2.3)

(i) P(B) = P(Bn1) + P(Bn2) + P(Bn3) = P(1)P(B|1) + P(2)P(B|2) + P(3)P(B|3)

= 1/3.5/7 + 1/3.3/7 +1/3.6/7 = 1/3 (5/7 + 3/7 + 6/7) = 2/3.

(ii) P(1|B) = P(the biro chosen came from drawer 1 given that the chosen biro is
black).

P(1 ∩ B) P( B ∩ 1) P(1) P( BΙ1) 13 . 57 5

= = = = 2 =
P( B) P( B ) P( B ) 3 14

Definition 2.5

Let A, B and C be three events such that P(C) > 0. then the conditional probability of A
∪ B given C is defined by

P(A ∪ B|C) = P(A|C) + P(B|C) – P(A ∩B|C).

Exercise 2.2

1. Suppose that a box contains 5 balls labeled 1 to 5. two balls are drawn at random
(one after the other without replacement).

What is the probability that

(i) the sum of the numbers on the two balls selected is even?

55
 (ii) The number on the first ball drawn is even if it is know that the sum of the
two numbers is even.
2. In a large population, it is observed that 30 percent of the people that are black
have caner and 25 percent of the people are not black have cancer. Assume that 10
percent of the population is black. What is the probability that a person selected at
random and found to have cancer is not black.

3. A vaccine produces immunity against smallpox in 98 percent of cases.

Suppose that, in a large population, 20 percent have been vaccinated. Find the
probability that a person who contracts smallpox has been vaccinated, assuming
that a vaccinated person without immunity has the same probability of contracting
smallpox as an unvaccinated person.

4. In a factory machines A,B,C produce respectively 20, 30 and 50 percent of the

total production. Of their output 3, 4, 5 percent respectively are defective items.
An item is drawn at random from the total production and is found defective.
What are the probability that it was produced by machine A, B,C?

5. In a faculty of a certain college, 60% of the students are female: 20% of the
females and 50% of the male are studying mathematics. If a student data card is
selected at random and the student is found to be studying mathematics, what is
the probability that the selected student is a male?

6. In JAMB examination each question has 5 possible answers, exactly one of which
is correct. If a student knows the answer he selects the correct answer. Otherwise
he selects one answer at random from the 5 possible answers. Suppose that the
student knows the answer to 70% of the questions.

(i) What is the probability that on a given question the student gets the correct
answer?

(ii) If the student gets the correct answer to a question, what is the probability
that he knows the answer?
56
 (iii) what is the expected score of the student in the examination?
7. A television set retailer finds out that 80% of this customers buy coloured T.V.,
and that 4 out of every 20 customers who buy coloured T.V set also buy antenna.
Calculate the probability that:

(i) a randomly selected customer buys antenna

(ii) a randomly selected customers who buys antenna has bought a

colored T.V

(iii) a randomly selected customer who has not bought an antenna has bought a
colored T.V set.

8. In a factory, machines A, B and C produce 20, 30 and 40 percent of the total

output, respectively. Of their output 5,4 and 3 percent respectively are defective
bolts. A bolt is chosen at random and found to be defective. What is the
probability that the bolt came from machine (i) A (ii) B (iii) C?

9. Four percent of an article manufactured by a company are defective. All the

articles produced are regularly inspected and those found defective are rejected. It
is found that one out of every eight of defective article produced is missed by the
inspector, while every good article passes inspection. What is the probability that a
customer buys a defective article produced by the company?

2.4 Independence

The notion of independence is a basic tool of probability theory. Consider tossing

a die twice, and let A1 be the event that the first toss gives an even number and A2
the event that the second toss gives an even number. The event A1 ∩ A2 is the
event that both tosses give even numbers. The occurrence of A1 does affect the
probability of A2 occurring.

57
 Therefore

P(A2|A1) = P(A2) and P(A1 ∩ A2) = P(A1)P(A2|A1) = P(A1)P(A2).

Hence, we say that two events A1 and A2 are independent if

P(A1 ∩ A2) = P(A1)P(A2).

Example 1.19

Toss a fair die twice and let A be the event “the first toss shows 3” and B be the

event the sum of the two numbers showing is 7.

P(A) =

B ≡{(1,6), (6, 1), (2, 5), (3, 4), (4, 3), (5, 2)}

P(B) =

Therefore,

P(A ∩ B) =

Thus P(A ∩ B) = P(A)P(B) hence, A and B are independent events.

In general case of n events, we have the following definition.

Definition 2.6

The events A1, A2,…An are independent if

(i) P(Ai ∩ Aj) = P(Ai)P(Aj) for all I ≠ j

(ii) P(Ai∩ Aj ∩ Ak) = P(Ai)P(Aj)P(Ak) for all I, j, k such that i ≠ j ≠ k

(2.7)

58
(n-1) P(A1 ∩ A2 ∩ A3… An) = P(A1P(A2)P(A3)…P(An)

That is, the events A1, A2,…An are said to be mutually independent if

P(An1 An2…Ank) = P(An1)P(An2)…P(Ank)

For every subsequence of two or more events.

An1, An2…Ank is a subsequence of A1, A2…An if the subscript integers satisfy

1 ≤ n1< n2…nk ≤ n.

If n = 3, we have, A1, A2, A3 are independent if

(i) P(A1∩ A2) = P(A1)P(A2)

P(A1∩ A3) = P(A1)P(A3)

P(A2 ∩ A3) = P(A2)P(A3)

(ii) P(A1 ∩ A2 ∩ A3) = P(A1)P(A2)P(A3)

Condition (i) is called pairwise independent. We might think that pairwise independence
always implies independence. But this is not necessarily so, as illustrated by the
following example.

Example 2.20

Let a pair of fair dice be rolled once. Consider the events. A1 number appearing on the
first die is even, A2 = the number appearing on the second die is odd = {1,3, 5} and A3 =
the difference of the two numbers is even

A1 ≡ {(2, 1), (2, 2), (2, 3), (2, 4), (2, 5), (2, 6), (4, 1), (4, 2),…, (6, 1), (6, 2),…}

A2 ≡ {(1, 1), (1, 3), (1, 5), (2, 1), (2, 3), (2, 5), (3, 1),…}

A3 ≡ {(1, 1), (1, 3), (1, 5), (2, 2), (2, 4), (2, 6), (3, 1), (3, 3), (3, 5), (4, 2), (4,4),

(4, 6), (5, 1), (5, 3), (5, 5), (6, 2), (6, 4), (5, 6)}

A1∩ A2≡ {(2, 1), (2, 3), (2, 5), (4, 1), (4, 3), (4, 5), (6, 1), (6, 3), (6, 5)}

A2∩ A3≡ {(1, 1), (1, 3), (1, 5), (3, 1), (3, 3), (3, 5), (5, 1), (5, 3), (5, 5)

A1∩ A2 ∩ A3 = Φ.

Thus

59
 P(A1) = ½, P(A2) = ½, (P(A3) = ½

P(A1 ∩ A2) = , P(A1 ∩ A2) = , P(A2 ∩ A3) =

P(A1 ∩ A2) = P(A1)P(A2; P(A1 ∩ A3) =P(A1)P(A3); P(A2 ∩ A3) = P(A2)P(A3)

That is, A1, A2, A3 are pairwise independent but
P(A1 ∩ A2 ∩ A3) = 0 ≠ P(A1)P(A2)P(A3) =

Since condition (ii) is not satisfied, we conclude that A1, A2 and A3 are not independent.

Definition 2.7
If the events A1, A2,…An are independent then
P(A1 ∩ A2… An) = P(A1)P(A2)…P(An) (2.8)

Example 2.21
A man fires 10 shots independently at a target. What is the probability that he hits the
target (i) 10times; (ii) at least once
If he has probability 1/3 of hitting the target on any given shot.

(i) Let Ai be the event “he hits the target at the ith shot” (I = 1, 2, 3,…10)

A1 ∩ A2 ∩ …∩ A10 is the event he hits the target 10 times A1, A2,… A10 are
independent events.
Therefore, the probability of hitting the target 10 times is

P(A1∩ A2 ∩…, ∩ A10) = P(A1)P(A2)…p(A10) =

(ii) P(hitting the target at least once) = 1 – P(not hitting the target at all).
P(not hitting the target at all) = P(A1∩ A2 ∩ … ∩ A10)
Where, Ai ≡ not hitting the target at the ith shot. A1, A2, …, A10 are independent
events and P(Ai) = 1 - = .

Hence

P(Ā 1 ∩ Ā 2 ∩ … ∩ Ā 10) = P(Ā 1)P(Ā 2)…P(Ā 10) =

Examples
60
2.22 Suppose a box contains 5 red and 3 black balls. A ball is chosen at random from
the box and then a second ball is drawn at random from the remaining balls in the
box. Find the probability that
(i) both balls are black
(ii) both balls are red
(iii) the first ball is black and the second is red
(iv) the second is red
(v) the second is black.
Solution
Let A1 be the event ”the first ball is red
Ā 1 be the event “the first ball is black
Ā 2 be the event ”the second ball is red
Ā 2 be the event “the second ball is black
(i) P(both balls are black) = P(Ā1 ∩ Ā 2)
P(Ā1 ∩ Ā 2) = P(Ā1)P(Ā 2)D Ā1)
P(Ā1) = P(Ā2DĀ1) =

Thus
P(Ā1 ∩ Ā 2) =

(ii) P(both balls are red) = P(A1 ∩ A 2)

P(A1 ∩ A 2) = P(A1)P(A 2)DA1)
P(A1) = P(A2DA1) =

Thus
P(A1 ∩ A 2) =

(iii) P(A1 ∩ A 2) = P(first ball is black and the second ball is red) = P(Ā1)P(Ā 2) Ā1)
P(Ā1) = P(A2D Ā 1) =

Thus
P(Ā 1 ∩ A 2) =

61
(iv) P(A2) = P(A2 ∩ A1) + P(A2 ∩ Ā 1)
From (ii) and (iii) we have

(v) P(Ā 2) = P(A 1 ∩ Ā 2) + P(Ā 1 ∩ Ā 2) =

2.23 Two fair dice are rolled. Given that the dice show different numbers, what is the
probability that at least one die shows a 6?
Solution
Let A be the event: the dice show different numbers.
A ≡ {(1,2), (2,1), (1, 3), (3, 1), (1, 4), (4, 1), (1,5), (5, 1), (1, 6), (6, 1),
(2, 3),…(5, 6), (6, 5)
B ≡ At least one die show a 6.
≡ {(1,6), (6,1), (2, 6), (6, 2), (3, 6), (6, 3), (4,6), (6, 4), (5, 6), (6, 5)}
A ∩ B = {(1,6), (6,1), (2, 6), (6, 2), (3, 6), (6, 3), (4,6), (6, 4), (5, 6), (6, 5)}
P(BDA) =

2.24 Let A and B be any two events defined on the same sample space. Suppose P(A) =
0.3 and P(A B) = 0.6. Find P(B) such that
(i) A and B are independent
(ii) A and B are mutually exclusive.

Solution
(i) If A and B are independent, then P(A ∩B) = P(A)P(B)
Thus
P(A B) = P(A) + P(B) – P(A)P(B)
We have
0.6 = 0.3 + P(B) – 0.3 P(B) = 0.3 + 0.7 P(B)

62
 P(B) = = 0.43

(ii) If A and B are mutually exclusive, then

P(A ∩ B) = 0
Thus
P(A B) = P(A) + P(B)
0.6 = 0.3 + P(B)
P(B) = 0.3
2.25 Two women A, and B share an office with a single telephone. The probability that
any call will be for A is . Suppose that A is out of her office during the office

hours half of the time and B one third. Find the probability that forany call during
the working hours
(i) no one is in to answer the call
(ii) A call can be answered by the person being called
(iii) Two successive calls are for the same woman
(iv) A caller who wants A has to try more than two times to get her.

Solution
(i) P(A and B are not in the office) =

(ii) P(A call can be answered by the person being called)

= P(call for A and A is in the office) + P(call for B and B is in the office)
=

(iii) P(for AA) + P(for BB) =

(iv) P(X > 2) = 1 – P(X = 1) – P(X = 2).

Where X number of time for a caller who wants A has to try to get her. Thus,

P(X > 2) = 1 -

Exercise 2.3

63
1. (a) Show that if A and B are independent events, then
(i) A and Bc, (ii) Ac and Bc are also independent.
2. Let A and B denote two independent events such that A is a subset of B. prove that
either P(A) = 0 or P(B) = 1.
3. A man fires 10 shots independently at a target. The probability of hitting the target
at any shot is 1/3. Calculate the probability that
(i) noneof the shots hits the target (ii) At least one shot hits the target
(iii) The target is hit at least twice if it is know that it is hit at least once.
4. A box contains 6 red balls and 4 white balls. Three balls are drawn from the box
one after the other without replacement. Find the probability that
(i) the first two are white and the third red
(ii) the first two are white and the third white.
(iii) two are red and one is white
(iv) the second ball drawn is red
(v) the third ball drawn is white
5. A die is rolled 8 times. What is the probability that
(i) exactly 2 sixes appear.
(ii) at least 2 sixes appear.
(iii) at most 2 sixes appear.
6. Prove that if A1,…,An are independent events then
(i) P(A1 A2 … An) = 1 – [1 – P(A1)}{1 – P(A2)]…[1 – P(An)]
(ii) P(A1 A2 … An) ≤ 1- e –[P(A1) + P(A2) +…+ P(An)]
- (x + x +…+ x )
Hint: (1 – x1)(1 – x2)…(1 – xn) ≥ e 1 2 n xi ≤1
7. Show that P(A∩B∩C) = P(ADB∩C)P(BDC)P(C).
8. A die is tossed n times. What is the probability that a 6 appears at least two times
in the n tosses.
9. Suppose that A or B occurs is 0.7 while P(A) = 0.2, find P(B).
10. A boy decides to continue tossing a fair coin until he has thrown total of three
heads. Find Pn, the probability that exactly n tosses will needed.

64
11. Six blood samples are selected from 40 blood samples, of which four are
cancerous. What is the probability that exactly two of the blood samples selected
are cancerous?
12. Prove that if A1, A2,…An are any n events, then
P(A1 ∩ A2 … ∩ An) > 1 – {P(A2c) + P(A2c) + …+ P(Anc)}.
13. Prove that
(i) P(A ∩ Bc) = P(A) – P(A ∩ B)
(ii) P(A ∩ Bc) = 1 - P(A) – P(B) P(A ∩ B)
(iii) P(A) = P(A ∩ B) + P(A ∩ Bc)
(iv) P(A ∩ B) ≥ P(A) + P(B) – 1.
14. three coins have probability 0.5, 0.6 and 0.8 for heads respectively. One of then is
selected at random, that is, with equal chance for each, and tossed. If the outcome
is head, what is the probability that the coin with probability 0.8 for heads was
selected?
15. On a certain weekend there are 4 movies. Calculate the probability that at least one
of A and B will be selected by one or more of the 3 students.
16. A box contains n white balls numbered 1 to n, n black balls numbered 1 to n, and n
red balls numbered 1 to n. if two balls are drawn at random without replacement,
what is the probability that both balls will be of the same colour or bear the same
numbers.
17. On the first round, three fair coins are flipped at random. The coins resulting in
heads are flipped at random on the second round. If the second round results in
exactly one head, what is the conditional probability that the first round ended
inexactly two head?

65
 UNIT THREE
DISCRETE RANDOM VARIABLES
3.1 Introduction
This chapter introduces idea of a random variable and its probability density function.
A random variable is a variable whose actual numerical value is determined by
chance. There are two easily identifiable types of random variables, discrete and
continuous. A discrete variable is one that takes only a limited number of possible
values, otherwise the variable is called continuous. This chapter is devoted to discrete
random variables.
Section 3.2-3.4 are devoted to some special discrete random variables. Bernoulli,
Binomial Poisson, uniform,Geometric and negative bionmial.

Example 3.1
Consider variable X, the number of heads in three tosses of a coin. There are four
possible values (0, 1,2, 3,) of X.
The actual value assumed is due to chance therefore X is a random variable. The
sample space for this experiment is
Ω= {HHH, HHT, HTH, THH, HTT, THT, TTH, TTT}
X = 0 if the outcome is TTT
X = 1 if the outcomes is HTT, or THT, or TTH
X = 2 if the outcome is HHT or HTH or THH
X= 3 if the outcome is HHH
Let p be the probability of the con landing tail. Since landing tail and landing head are
exhaustive events the probability of the coin landing head is 1-p.
P(T ∩ T ∩ T) = P(T)P(T)P(T)
Since the outcome at each trials are independent.
P(TTT) = p.p.p=P3
66
 P(HTH) = P(T)P(H)P(H) = p (1-P) (1-P) = (1-P)2
P(HTH)= P(H)P(H)P(H)=( 1-P) (1-P) = P(1-P)2
P(HHT)= P(H)P(H)P(T) = (1-P)(1-P)P==(1-P)2
The probability of getting two heads = P (THH) + P(HTH) + P(HHT) = 3P (1-p)2
Similarly we have
P (o head) = P(TTT) = p3
P (1 head) = P(HTT) + P(TTH) + P(THT) = 3p2(1-p)
P (3 heads) = P(HHH) = (1-p)3
Thus we have the following table
Heads 0 1 2 3
Probability P3 3p2 (1- p) 3p (1 – p)2 (1 –p)3

P (0 head) = P (X is 0) = p3, p(X=1) 3p2 (1-p) and so on

The random variable X defined above is an example of what is called a discrete random
variable. A random variable is denoted by a capital letter such as X. Y, Z… and the
values that the random variable takes on is denoted by a lower case letter x, y, z…
The notation (P(X=x) means the probability that the random variable X takes on the value
x.

Definition 3.1
A random variable X on a sample space Ω is a function assigns to each element Ω one
and only one real number X ( ) = x, the space ofX is the set of real number
Φ = {x:x= x ( ), we Ω).
Definition 3.2
A random variable x is discrete if t can assume at most a finite or a countable infinite
number of possible values.
In the above example,
Ω=
Where w1 = HHH, w2 = HHT…, w8 = TTT

67
= {0, 1, 2, 3} and X (w1) = 3, X (w2) = 2.X(w3) = 2, X (w4) = 2
X (w5 = 1, X (w6) = 1, X (w7 = 1, X (w8) = 0
That is {w:X(w) = x1} id an event

Definition 3.3
The real valued function f defined on R by f(x) = p(X = x) is called the discrete
probability density function of X.
Let X be a discrete random variable and suppose that the values it can assume are x1,
x2…,xn
The probability can be written as
P(X = x1) = f(x1), =P(X = x2) = f(x2)…, P(X = xn) = f(xn)
Such that
F(x) is calledprobability density function of X
For an illustration, let us consider the following examples.
Example 3.2
Suppose a pair of fair dice is tossed onc, Let X,Y,Z represent the sum, maximum and
minimum respectively of the two numbers appearing find the probability density function
of
(i) X, (ii) Y, (iii) Z.
Solution
The sample space Ω = {( 1, 1), (1, 2) ,…, (6, 6)} consist of 36 elements
(i) P (X =2) = P {( 1, 1)} = 1
36
P ( X = 3) = P{( 1, 2), (2, 1) = 2
36
P (X = 4) = P{(1,3), (3, 1) , (2, 2)} = 3
36
.
.
P (X = 12) = P {(6, 6)} = 1
36
Thus,
F(2)= 1, f(3) = 2, f(4) = 3 ---
68
 36 36 36
In tabular form we have
X 2 3 4 5 6 7 8 9 10 11 12
F(x) 1 2 3 4 5 6 5 4 3 2 1
36 36 36 36 36 36 36 36 36 36 36

in function form

(ii) Y =Maximum of the two number

Possible values of Y are 1, 2, 3, 4, 5, 6
g (1) = P(Y=1)= P {(1,1)}= 1
36
g(2) = P(Y = 2) = P{( 1, 2), (2, 1) (2, 2)= 3
36
g(3) = P (Y = 3)= P (1, 3), (3, 1), (2, 3) (3, 2) (3, 3)} = 5
36
g(4) = P (Y=$) = P{( 1,4)(4,1) 2, 4) (4, 2) (3, 4) (4, 3) (4, 4) = 7
36
g(5) = P (Y = 5)
= P {(1,5), (5,1), (2, 5) (5,2) (3, 5) (5, 3) (4, 5)(5, 4) (5, 5) = 9
36
g(6) = P (Y = 6)
= P {( 1,6), (6,1) (2,6) (6,2) (3, 6) (6, 3)(4, 6) (6, 4) (5, 6) (6,5) (6,6)
= 11
36
Putting in form a table we have
Y 1 2 3 4 5 6
g(y) 1 3 5 7 9 11
36 36 36 36 36 36
This can be written in a functional form as

69
 (iii) Z = Minimum of the two numbers
The possible values of Z are 1, 2, 3, 4,5,6
P(Z-1) = P{(1,6), (6,1)(1,5)(5,1)(1,4)(4,1)
(1,3)(3,1)(1,2)(2,1)(1,1)! = 11
36
P(Z = 2) = P (2,6)(6,2)(2,5)(5,2)(2,4(4,2)
(2,3)(3,2) (2,2)= 9/36
P(Z= 4)P{(4,6) (6,4) (4, 5) (5,4) (4,4) 5.36
P(Z =5) = P(5,6) (6,4) (5,5) 3/36
P(Z= 6) = P(6,6) 1/36
Putting in form of a table we have

Z 1 2 3 4 5 6
h(z) 11 9 7 5 3 1
36 36 36 36 36 36

H(Z) = 13 – 2z z = 1, 2, 3, 4,5, 6
36
= 0 for other values of x.
The probability density function of a discrete random variable X has the following
properties
(i) O < f(x) < 1, x E R
(ii) {X:f (x) = 0} is a finite or countable infinite subset of R
(iii) ∑f(xi) = 1
The above densities can be represented in terms of a diagram as illustrated in figure (a),
(b) (c)

70
Example 3.3
Let X,Y,Z be the random viable introduced in example 3.1 above
The above three properties are satisfied, properties (i) & (ii) are immediate from
definition of probabilities. To check (iii) we have
∑f(x1) = 1/36+ 2/36+ 3/36 + 4/36 + 5/36+ 6/36 +5/36 + 4/36 + 3/36 + 2/36+ 1/36 =1

∑G (yi) = 1/36 + 3/36 + 5/36 + 7/36 + 9/36 + 11/36= 1

∑h(zi) = 11/36 + 9/36 + 7/36 + 5/36 + 5/36 + 3/36 + 1/36 = 1

Example 3.4

Suppose a box contains 1 balls of which 4 are red and 6 are black. A random sample of
size 3 is selected. Let X denote the number of red balls selected. Find the probability
density function of x if

(i) Sampling is without replacement

(ii) Sampling is with replacement

(iii) The possible values of X are 0, 1, 2, 3.

P(X = 0) = (P no red ball in the three balls selected) = 6C3 = 6 5 4 = 1

10c5 9 8 6

P(X = 1) = P(1 red and 2 black balls ) = 4C1 x 6C2 =1

10
C3 2

P(X = 2) = 4C2x 6C1 = 3

10
C3 10

P(X = 3) = 4C3/10C3 = 1

71
 30

Thus the p.d.f is

X 0 1 2 3
F(x) 1 1 3 1
6 2 10 30

As a check, we add up all the probabilities

1/6+ ½ + 3/10 + 1/30=1

(ii) Sampling with replacement

P(X = 0) =P (first ball is black, second black and the third black) = P(bbb).

The probability of black at any drawing is

6/10= 3/5

This value is constant since drawing is with replacement. Therefore

P(bbb) =

P(X = 1) = P(Rbb) + P(bRb) + P(bbR) =

Similarly ,

P(X = 2) =

P(X = 3) = P(RRR) =

Thus, the p.d.f of X is

X 0 1 2 3

f(x)

72
This can be written as

f(x) =
0 elsewhere

Example3.5
A box contains 6 balls labeled 1, 2, 3, 4, 5, 6 two balls are drawn at random one after
the other. Let X denote the larger of the two numbers on the balls selected, obtain the
probability density function of X if
(i) Sampling is without replacement (ii) Sampling is with replacement

Solution

(i) The possible values of X are 2, 3, 4, 5, 6. The larger of the two numbers can not be
1 since sampling is without replacement and we can not get same number twice.

P(X = 2) = P{(1, 2), (2, 1)} =

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

P(X =5) = P{(1, 5), (5, 1), (2, 5), (5, 2), (3, 5), (5, 3), (5, 4), (4, 5)} =

Similarly,

P(X = 6) =

Thus the probability density function of X is

X 2 3 4 5 6

73
 f(x) 1/15 2/15 1/5 4/15 1/3

In functional, form, this can be written as

f(x) = x = 2, 3, 4, 5

0 elsewhere

(ii) The possible values of X are 1, 2, 3, 4, 5, 6. The larger can be 1 in this case
since (1, 1) is a possible outcome. The sample space consists 36 possible
outcomes

P(X = 1) = P(1, 1) =

P(X = 6) = P{(1, 6), (6, 1), (2, 6), (6, 2).,,, (6, 6)} =

Hence the probability density function X is

X 1 2 3 4 5 6

f(x) 1/36 3/36 5/36 7/36 9/36 11/36

In functional form we have

f(x) = x = 1, 2, 3, 4, 5, 6

3.6 A couple decides that they will continue to have children until either they
have a boy and a girl in the family or they have four children. Assuming that boys
and girls are equally likely to be born. Let X denote the number of children in the
family. Find the probability density function of X.

The number of children in the family can either be 2, 3, or 4.

P(X = 2) = P{(B, G), (G, B)}

74
 That is, the couple will stop having more children if the first child is a boy and the
second of girl or the first is a girl and the second a boy.

Thus,

P(X = 2) = P(B, G) + P(G, B) = P(B)P(G) + P(G)P(B)

Similarly,

P(X = 3) = P{(BBG),(GGB)} = P(BBG) + P(GGB) =

Hence, the probability density function of X is

X 2 3 4

f(x) ½ ¼ ¼

Exercise 3.1
1. A fair coin is tossed until a head or five tails occur. Let X denote the number of
tosses of the coin. Compute the probability density function of X.
2. A box contains 2 red balls and 3 blue balls. Balls are successively drawn without
replacement until a blue ball is drawn. Let X denote the number of draws required.
Compute the p.d.f of X
3. The pdf of a random variable X is given by

X 1 3 4 6 8

f(x) K

Find (i) K (ii) P(X ≥ 3).

75
4. Toss a fair coin two times. Let X be the number of heads obtained. Find the pdf of
X.
5. A coin with probability p of a head is tossed until a head appears. Let X denote the
number of times the coin is tossed. Find the pdf of X.
6. A fair die is tossed twice. Let X denote the product of the two numbers appearing.
Find the pdf of X
7. A fair coin is tossed3 times. Let X represent the difference between the number of
heads and the number of tails obtained. Find the pdf of X.
8. The pdf of a random variable X is given by
f(x) = {k 2x, x = 1, 2, 3, …N, zero elsewhere}. Find the value ofK.

Definition 3.3
Probability Distribution function. Let X be a random variable with probability density
function f(x). The probability distribution function of X denoted f(x) is defined by
F(x) = p(X ≤ x) for x real.
=
Properties of the Probability distribution function
1. F is a non decreasing function, that is if a < b, then
F(a) < F(b).
2. lim F(b) = 1.
b-∞
lim F(b) = 0
b-∞
3. F is right continuous. That is F(b + ) = F(b).

Let A be any subset of R and let f(x) be the probability density function of X. ew can
compute P(X ε A) by noting that {w : X (w) ε A} is an event and that
{w : X(w) ε A} = U { w : X(w) = xi}.
XiεA

76
Thus
P(X ε A) =
If A is an interval with end points a and b, say A = [a,b].
Then
{(X ε A) = P(a ≤ X ≤b) =

Example 3.7
Consider the random variable X the sum of the two numbers appearing when a fair die is
tossed twice of Example 3.2. The probability distribution function for X is given by

X 2 3 4 5 6 7 8 9 10 11 12

F(x)

Suppose we wish to find the probability that X is between 4 and 9 (4 and 9 inclusive) we
write it as
P(4 ≤ X ≤ 9) = P(X = 4) + P(X = 5) + P(X = 6) + P(X = 7) + P(X = 8) + P(X = 9)
=

This can also be written as P(X≤ 9) – P(X ≤ 3) = = = F(9) – F(3)

3.2 SPECIAL DISCRETE RANDOM VARIABLES

3.2.1 In Section 3.1 we introduced discrete random variables and the probability density
function of some random variables were determined. Some variables are so common and
important that names are given to them.
In this section we shall consider in considerable detail a number of important discrete
random variables.

Bernoulli Random variables

77
Definition 3.4
Bernoulli Trail:
A random trail or experiment is which the outcome can be classified into one of two
mutually exclusive and exhaustive ways usually called success or failure is called a
Bernoulli trail. The random variable associated with Bernoulli trail is called a Bernoulli
random variable (X).let X=0 if outcome is a failure and X =1 if the outcome is a success.
That is any variable assuming only two values is called a Bernoulli random variable.
Suppose that we toss a coin once. Let the probability of it landing head be p. p= , if the

coin is fair and let denote the outcome of the toss. Then there are two possible values for
X, Heads or tails. These two values are mutually, exclusive and exhaustive and we may
associate the two possible outcomes of the toss with values 1, 0 of the random variables
X. That X = 1 when a head appears and X = 0 when a tail appears.
P(X = 1) = p, P(X = 0) = 1 –p.
The p.d.f. of X is

X 0 1

f(x) 1–p P

Or in functional form
F(x) = pX(1 –p)1-x, x = 0, 1
0 elsewhere 3.1
f(x) as defined above is called the Bernoulli probability density function and any variable
X having (3.1) has its probability density function is called a Bernoulli random variable
and is said to have the Bernoulli distribution.

3.2.2 The Binomial Random Variable

This is one of the most important random variables in statistics and the most important
discrete random variable. Consider n independent repetitions of Bernoulli trails. Let Xi, I
= 1, 2,…, n be Bernoulli random variables associated with the trails. The random

78
variables X1, X2,…Xn are independent Bernoulli random variables. Let us assume the
probability of success is p and failure 1-p and

P(x1 = 1) = p
Then, sum Sn = + x2 +---+ Xn is the number of successes in n Bemoulli trials. That is, Sn
is a counting- variable counting the number of successes in n repeated trials. This random
variable Sn is called the Binomial random variable. The possible values of Sn are 0, 1, 2,
3, 4,----, n.
P(Sn = 0) = p(no success)
= p(1sttrail is a failure) P(2nd trail is a failure)…P(nth trail is a failure).
= P(X1 = 0) P(X2 = 0) P(X3 = 0)…P(Xn = 0)
= (! – p) (1 – p)(1 – p)…(1 –p) = (1-p)n.
Sn is 1 if the sequence of outcome is
1 0 0 0 0 0 0 …0 or 0 1 0 …0 or 0.0 1 0 …0,…0 0 0 0 0 0 1
P(Sn = 1) = P(1, 0, 0, 0,…0) + P(), 1, 0, 0,…, 0) + …+ P(0, 0,…1)
= P(X1 = 1)P(X2 = 0)…P(Xn = 0) + P(X1 = 0)P(X2 = 1)…P(Xn = 0) +…+
+ P(X1 = 0)…P(Xn = 1)
= P(1 – p)…(1 – p) + (1- 1)P(1 –p)…(1 – p)…(1 – p)+ (1 – p)(1 – p)
= P(1-p)n-1 + P(1-p)n-1 +…+ P(1-p)n-1 = np(1-p)n-1

Similarly,
P(Sn = 2) = nC2 P2 (1-p)n-2
Where nC2 is the number of sequences in which exactly 2 have value 1 and the others 0
e.g. (1, 1, 0, 0, 0, 0,0,…, 0), (1, 0, 1, 0,…, 0)…
In general, it can easily be seen that
P(Sn = k) = nC2 pk (1-p)n-k, k = 0, 1,…, n
Where nC2 is the number of sequences in which exactly k have value 1 and others 0.
For example when n = 4, possible sequences of outcomes are given below.

Sequence Sn P(Sn)

79
 (0,0,0,0) 0 (1 – p)4

(1,0,0,0) 1 P(1 – p)3

(0,1,0,0) 1 P(1 – p)3

(0,0,1,0) 1 P(1 – p)3

(0,0,0,1) 1 P(1 – p)3

(1,1,0,0) 2 P2(1 – p)2

(1,0,1,0) 2 P2(1 – p)2

(1,0,0,1) 2 P2(1 – p)2

(0,1,1,0) 2 P2(1 – p)2

(0,0,1,1) 2 P2(1 – p)2

(0,1,0,1) 2 P2(1 – p)2

(1,1,1,0) 3 P3(1 – p)

(1,1,0,1) 3 P3(1 – p)

(1,0,1,1) 3 P3(1 – p)

(0,1,1,1) 3 P3(1 – p)

(1,1,1,1) 4 P4

Thus,
P(Sn = 0) = (1 – p)4
P(Sn = 1) = 4p(1 – p)3
P(Sn = 2) = 6p2(1 – p)2
P(Sn = 3) = 4p3(1- p)
P(Sn = 4) = p4

Theorem 3.1
Let Sn denote number of successes in n repeated Bernoulli trails, with probability of
success p. the probability density function of Sn is given by
80
 f(x) = P(Sn = x) = nCx px(1 – p)n-x x = 0, 1, …,n (3.2)
0
Definition 3.5
A discrete random variable X denoting total number of successes in n trails is said to
have the binomial distribution if

P(X = x) = nCxpx(1-p)n-x ; x = 0, 1, 2,…, n

0≤ p≤ 1
The conditions under which binomial distribution will arise are
(i) The number of trails is fixed
(ii) There are only two possible outcome ‘success’ or ‘failure’ at each trial.
(iii) The trails are independent
(iv) The probability p of success at each trail is constant
(v) The variable is the total number of successes in n trails.

Example 3.8
A soldier fires 10 independently at a target. Find the probability that he hits the target.
(i) once (ii) at least 9 times (iii) at most two times.
If he has probability 0.8 of hitting the target at any given time? Let X denote the number
of times he hits the target. Then X is a binomial variable with n =10 and p = 0.8
From equation (3.2), we have
P(X = x) = 10Cx (0.8)x (0.2)10-x
(i) P(X = 1) = 10Cx 0.8(0.2)9 = 8(0.2)9
(ii) P(He hits the target at least 9 times) = P(X ≥ 9)
= P(X = 9) + P(X = 10)
P(X = 9) = C9 (0.8)9 (0.2) = 10 × (0.8)9 × 0.2 = 2(0.8)9
10

P(X = 10) = 10C10(0.8)10 = (0.8)10

Hence,
P(X ≥ 9) = 2(0.8)9 + (0.8)10 = (0.8)9(2 +0.8) = (0.8)9(2.8) = 0.3758
(iii) P(at most twice) = P(X ≤ 2) = P(X = 1) + P(X = 2)
P(X = 0)= (0.2)10;
P(X = 1) = 8(0.2)9;
P(X = 2) = 45 (0.8)2 (0.2)8
Thus,
P(X ≤ 2) = (0.2)10 + 8(0.2)9 + 45(0.8)2 (0.2)8 = 0.00008.

Example 3.9
A fair die is rolled four times. Find the probability of getting 2 sixes. Let us call a six a
success on a toss of a die and let X be number of sixes (successes) in 4 trails. X is
binomial random variable with n = 4 and p = 1/6. Thus,

81
 P(X = 2) = 4C2 ..
Example 3.10
Suppose that a certain type of electric bulb has a probability of 0.3 of functioning more
than 800hours. Out of 50 bulbs, what is probability that less than 3 will function more
than 800 hours. Let X be the number of bulbs functioning more than 800 hours.
Assuming that X has a Binomial distribution,

P(X = x) = 50 Cx(0.3)x(0.7)50-x
P(X < 3) = P(X = 0) + P(X = 1) + P(X = 2)
P(X = 0) = (0.7)50; P(X = 1) = 50.(0.3) (0.7)49
P(X = 2) = 50C2 (0.3)2(0.7)48.
Thus,
P(X < 3) = (0.7)50 + 15(0.7)49 + 110.25(0.7)48 = 0.0000046

Exercise 3.2
1. A fair coin is tossed 4 times. Compute the probability that (i) exactly two heads
occur (ii) at least 3 heads occur.
2. An investigation reveals that four out of every five patients are cured of malaria
when treated with a new drug. If a sample of ten patients is treated by the new
drug, compute the probability that (i) exactly six patients are cured, (ii) at most
four patients are cured
3. A man fires 12 shots independently at a target. The probability of hitting his target
is ¼.
(i) What is the probability of his hitting the target at least two times?
(ii) How many times must he fire so that the probability of his hitting the target
at least once is greater than 7/9?
4. Six children are born in a hospital in a given day. Calculate
(i) the probability that the number of boys is the same as the number of girls.
(ii) the probability that there are more girls tha boys
(iii) the most likely number of boys.
82
5. An experiment has 90 percent probability of success and 10 percent probability of
failure. The experiment is repeated four times. Find the probability of obtaining
(i) No success (ii) No failure (iii) two successes and two failures.
6. Let X be a Binomial random variable with paraments (n.p). show that
(i) P(X = k) first increases monotonically and then decreases monotonically.
(ii) The value of k that maximizes P(X = k) is the largest integer less than or
equal to (n + 1)P

(iii) P(X = k) =

7. A fair coin is tossed repeatedly until three are obtained. Find Pn, the probability
that exactly n tosses are needed.
n
8. Show that (i) Cxpx (1 – p)n-x = 1 (ii) n
Cxpx (1 – p)n-x = np
0<p<1.

3.3 Geometric Random Variable

Consider a Bernoulli trail with probability p of a success on one trail. The trail is
continued until a success occurs. Let X denote number of trails before the first success.
For example, a student decides to continue taking JAMB examination until he passes. X
in this case denote number of times he takes the examination before the first success.
The probability that the first x – 1 trails are failures and the xth trail is a success is given
by (1 – p)x=1 p.
To see this, the required probability is
P(F F F …F S) = P(F)P(F) …P(F)P(S) = (1 – p)(1 – p)(1 – p) … (1 – p)P = (1 –
p)x-1 p.
The probability that x trails are needed for the first success is the same as the probability
that the first x – 1 trails are failures and the xth is a success. Thus,
P(1 – p)x -1, x = 1,2…

P(X = x) = 0 otherwise
Where X is the number of trials before the first success and if we define Y as the number
of failures preceding the first success, we have

83
 P(1 – p)y y = 0,1,2…

P(Y= y) = 0 otherwise
Definition 3.6
A discrete random variable X is called a geometric random variable if its probability
density function is given by

(1 – p)x -1p x = ,1,2,3…

F(x) =p(X= x) = 0 otherwise

0 < p< 1 (3.3)

Where X is the number of independent Bernouilli trials taken for the first success to
occur (the successful trial is included in the count). To see that (f(x) is a probability
density function, all that needed to be checked is that
(1-p)x=1= 1 = P[1 + (1-p) + (1-p)2 +…]
From geometric series,
(1-p)x=1= .

Example 3.11
A fair is tossed until a head appears (a) What is the probability that three tosses are
needed? (b) What is the probability that at most three tosses are needed?
Solution
Let X denote the number of tosses until a success (a head) occurs. Since the coin is fair,
.

P(X = 3) = (1 – p)3-1

P(at most three tosses are needed) = P(X ≤ 3)

= P(X = 1) + P(X = 2) + P(X = 3)
2 2 3
1 1 1 1 1 1 1 1 7
= + . +  . = +  +  = .
2 2 2 2 2 2 2 2 8
84
3.12 A fair die is rolled until a six appears.
What is the probability that (i) at most 4 rolls are needed. (ii) at least 3 rolls are needed?
Solution
(i) P(at most 4 rolls are needed)= P(X ≤ 4)
= P(X = 1) + P(X = 2) + P(X = 3) + P(X = 4)
1
X is a geometric random variable with P =
6
x1
5 1
P(X = x) =  
6 6
2
1 5 1 5 1
P(X = 1) = , P(X = 2) = , , P( X = 3) =   .
6 6 6 6 6
3

P(X = 4) =   . .
5 1
6 6
Thus
2 3

P(X ≤ 4) = + . +   . +   .
1 5 1 5 1 5 1
6 6 6 6 6 6 6
x =1
1 4 5 1 1 − (5 / 6) 4
= ∑  = . = 1 − (5 / 6) 4
6 x =1  6  6 1− 5 / 6

(ii) P(at least 3 rolls are needed) = P(X ≥ 3) = P(X = 3) + P(X = 4) + …

x =1 x =1
1 1 − (5 / 6)  5 
∞ ∞ ∞
2 2
5 1 1 5 25
P(X ≥ 3) = ∑
x =3
f ( x) = ∑  
x =3  6 
=
6 6
∑  
x =3  6 
= . =  =
6 1− 5 / 6  6  36

or
11 25
P(X ≥ 3) = 1 – {P(X = 1) + P(X = 2)} = 1 - =
36 36

Example 3.13
The probability that a certain test yields a “positive” reaction is 0.6. what is the
probability that not more than 4 negative reactions occur before the first positive one?

Solution

85
Let X denote the number of negative reactions before the first positive one, then
P(X = x) = (0.4)x(0.6); x = 0,1, 2,…
Thus,
4 4 4
P(X ≤ 4) = ∑ P( X = x) = ∑ (0.4) X (0.6) = (0.6) ∑ (0.4) x = (1 − (0.4) 5 ) = 0.9898
0 x =0 x=0

The geometric random variable has an interesting property which is summarized in the
following theorem.
Theorem: 3.2
Suppose that X is a geometric random variable. Then for any given two positive integers
s and t,
P(X > s + t|X > s) = P(X > t).
Proof:
P ( X > s + t and X > s ) P ( X > s + t
P(X > s + t | X > s) = =
P( X > s) P( X > s)

P(X = x) = p(1- p)x-1, x = 1, 2, 3, …

∞
(1 − p) s +t
P(X > s + t) = P ∑
X = S + t +1
(1 − p ) X −1 =P
1 − (1 − p )
= (1 − p) s +t

∞
P(1 − p ) s
P(X > s + t) = P ∑ (1 − p)
S +1
X −1
=
1 − (1 − p )
= (1 − p ) s

Thus,
(1 − p ) s +t
P(X > s + t| X > s) = = (1 − p) t
(1 − p ) s

∞
p (1 − p) t
P(X > t) = ∑ p(1 − p) X −1=
t +1 p
= (1 − p) t .

Hence,
P(X > s + t |X > s) = P(X > t).
The above theorem states that if a success has not occurred during the first s repetitions of
Bernouli trails, then the probability that it will not occur during the next t repetitions is
the same as the probability that it will not occur during the first t repetitions of Bernoulli
trails. Therefore the distribution is said to have “no memory”.

86
Exercise 3.3
1. On a certain road the probability of an accident on any day is 0.05 (assuming not
more than one accident can occur on any day). Assuming independence of
accidents from day to day on this road, what is the probability that the first
accident of the year occurs in the month of March.
2. A fair die is rolled until a six appears. Calculate the probability that he has to
throw the die more than three times before he gets a six.
3. Let X be a geometric random variable with p = 0.2. Calculate the following
probabilities. (i) P(3 < x ≤ 6) (ii) P(2≤ X≤ 4) (iii) P(X ≤ 2).
∞
1
4. Prove that ∑ xp(1 − p)
x =1
x =1
=
p
.

5. Suppose a Bernoulli trail with probability p of success is continued until rth

success occurs. Let X be the number of independent trails needed in order to have
r successes. Show that
 k − 1 r
P( X = K ) =   p (1 − p ) k −r , k = r , r + 1,...
 r − 1
3.13 Pascal or Negative Binomial Random variable
A probability distribution closely related to the geometric distribution is the Pascal or
negative binomial distribution.
Suppose that independent repetition of Bernoulli trail needed to have “success” occurs
exactly r times.
X = x if and only if success occurs on the xth trail and success occurs exactly (r – 1) times
in the previous x -1 trails. The probability of this event is determined as follows:
The probability of r – 1 successes in x – 1 trails is given by the binomial formula
X=1
Cr – 1pr-1(1 – p)x=1 – (r – 1)= x -1 Cr– 1 pr – 1 (1 – p) x – r
Therefore, the probability of this event is
P.x-1Cr-1pr-1(11 – p)x-r+
hence,
f(x) = P(X= x) = x-1Cr-1 pr(1 – p)x-r, x = r, r + 1,… (3.4)
Similarly, if we let Y be the number of failures before the rth success we have
87
 P(Y = y) = p. (y+r-1Cypr-1(1 – p)y
FY(y) = y+r-1Cyr(1 – p)y, y = 0, 1, 2,…
Definition 3.7
A random variable having its probability density function given by (3.4) or (3.5) is said to
gave a Negative Binomial or Pascal distribution.
Example 3.14
Show that
Solution
By Binomial theorem,

Example 3.15
A fair die is rolled until two sixes occurs, find the probability that
(i) exactly 5 tosses are needed
(ii) at most 5 tosses are needed.
Solution
Let X be the number of tosses needed to get two sixes. X is a Pascal random variable
with pdf: f(x) = x-1C1p2(1-p)x-2, x = 2,3,..
Where p = ½ , r = 1.

(i)

(ii)

Thus,

88
3.3.2 The Hyper geometric Random Variable
Suppose that we have a box containing n items of which n1 are defective and n – n1 are
non-defective. Suppose that we choose at random k items from the box without
replacement. Let X denote the number of defective items in the k items selected. Then

(3.4)

The reader will notice that X = x if and only if we obtain x defective items from n1
defective items in the box (n1Cx ways of doing this).
Any random variable having its probability density function as given by (3.5) is called
hyper geometric random variable and is said to have hyper geometric distribution.

3.4 The Poisson Random Variable

Definition: Rare Event
An event is said to be rare if the probability p of observing the event is very small.
Consider n repeated Bernoulli trails, where n is very large and P very small, let X be be
the number of successes in n trials. Then
P(X = x) = nCxpx(1 – p)n-x.
Setting λ = n p, we have

Now as n ∞.
1.
2.
3.

Consequently,
n
Cxpx(1 – p)n-x =
(3.5) gives an approximation to the binomial distribution with λ = np, when n is large and
p is small, where e = 2.71828 is the base of natural logarithms.

Definition:
A random variable X is called a Poisson random variable if its probability density
function is given by

89
 (λ > 0) (3.8)
0 otherwise
X represents the total number of events which have occurred up to time t. When t= 1 then
f(x) corresponds to the probability density function of number of events in a unit interval.
For examples, the number of calls that come into a telephone exchange in a unit time
interval, the number of vehicles passing through a designated point in a unit time interval.
In order to motivate our discussion, let us consider the following examples.

Example 3.16
Suppose a rare disease occurs in 2 percent of a large population. A random sample of
10,000 people are chosen at random from this population and tested for the disease.
Calculate the probability that at least two people have the rate disease.

Solution
The probability p of having the disease is 0.02 and n = 10,000. Let X be the number of
people having the disease. X is a binomial random variable with parameters n= 10,000
and p = 0.02. we shall apply the result (3.5) since n is large and p very small.
Hence,
Np = 0.02 × 10,000 = 20,,
Thus,
P(X = x) =
P(X = ) = e-200
P(X = 1) = 200 e-200
Hence
P(X ≥ 2) = 1 – P(X < 2) = 1 – e-200 (1 + 200) = 1 – 201 e-200.

Example 3.17
On a given road, an average of five accidents occur every month. Calculate the
probability that over a year period there will be (i) no accident (ii) at most 2 accidents.

Solution
In this case, λ = 5, t = 12. From (3.8) we have

f(0) = e-λt = e-60

f(1) = = 60e-60
f(2) = = e-60
thus,
P(no accident) = f(0) = e-60
P(at most two accidents) = f(0) + f(1) + f(2) = e-60 + 60 e-60 + e-60

90
 = e-60 (1 + 60 + 1800) = 1861 e-60.
Tables for the Poisson distribution are available and brief abulation is given in the
Appendix.

Exercises 3.4
1. If 3% of the items manufactured in a factory are defective. Compute the
probability that in a simple of 100 items. (i) 2 items will be defective (ii) at least
2 items will be defective. Use Poisson approximation to the Binomial.
2. Use the Poisson approximation to calculate the probability that at least two sixes
are obtained when six dice are rolled once.
3. The telephone switchboard of a University has an average of two incoming calls
per minutes. Calculate the probability that, over a three-minute interval, there will
be (i) no incoming calls (ii) exactly one incoming call. (iii) at most two incoming
calls.
4. Let Pr be the negative binomial pdf with parameters r and P. prove that
where Pr(k) = P(X= k).

5. Prove that Hint: Use integration by parts.

6. A fair die is tossed 20 times. Let a success on the ith toss corresponds to at least a
five appears. What is the probability that
(i) exactly 10 failure prior to the first success?
(ii) exactly 10 failures prior to the fifth success?
(iii) exactly 8 failure and 3 successes for the first II throws.
7. if (i) show that (x + 1)f(x +1)= λ f(x).
n x
(ii) f(x) = Ccp (1 –p), show that (1 – p)(x + 1)f(x + 1) =p(n – x)f(x).
In each case, show that f(x) increases monotonically and then decreases
monotonically as x increases. Find x that maximizes f(x).
8. Suppose that the number of accidents occurring on a highway each day is a
Poisson random variable with parameter
(a) What is the probability that at least two accidents occur in a day?
(b) what is the probability that at least three accidents occur in a day given that
at least one accident occurs in a day?
9. Describe the Poisson distribution, stating clearly the meanings of the symbols
used. Show that the variance of the distribution is equal to its mean.
10. Show that f(x) = x-1Cn-1Pn(1-p)x-n, x, x = n, n+1,… is a pdf.
Note P-n= (1 – (1-p))-n = .
11. Show that f(x) = x = 1,2, …, 0< p<1is a pdf. Find the probability
generating function of X.
12. Solve the problem of exercise 3.2, No 7.

91
13. If n ≥ 8 and in s binomial distribution the probability of 7 successes in n trails is
equal to the probability of 8 successes in n trail, find the probability of success on
any given trail.
14. Let X be a random variable with moment generating function MX(t). If R(t) = In’’
MX(t) for all t. Find R”(0).
15. If a fair die is rolled repeatedly, find the probability that the first “six” will appear
on an odd-numbered roll.

p, x = -1
16. f(x) =
(1-p)2px, X = 0,1,2
(i) Show that f(x) is a pdf

UNIT FOUR
EXPECTATION OF DISCRETE RANDOM VARIABLE
In this chapter we introduce the concept of the mean value of a random variable. It is
closely related to the notion of weighted average. The moment and probability generating
functions of a random variable are also introduced.

4.1
Let X be a random variable having possible values x1, x2, …, xn, and let the experiment
on X be performed n times. For example let X be the outcome of rolling a die. There are
six possible values x1 = 1, x2 = 2, x3 = 3, x4 = 4. x5 = 5 and x6 = 6. Suppose the die is
rolled n times. The successive rolls constitute independent repetitions of the same
experiment. Let X1, X2, . …, Xn denote the outcomes of the experiment of rolling a die n
times (that is Xi denote the outcome of the ith toss). Then

Is the average of the numbers that appeared. Let fi denote the number of times xi occur.
Then we have

We know that
is
called the expectation of the random variable X. thus the expected value of a random
variable X is the long-run theoretical average value of X.

92
Definition 1
Mathematical Expectation. Let X be a random variable with probability density function
as follows:

Value of X, x x1 x2 x3 …. xk

Probability f(x1) f(x2) f(x3) f(xk)

The mathematical expectation of X, denoted by E(X), is defined to be

E(X) = x1f(x2) + x2f(x2) +…+ xkf(xk)
E(X) =
E(X) is weighted average of possible values of x, the weight attached to the value xi is its
probability f(xi). E(X) is also called the mean of X or the population mean.
We may express the result of (1) in words.
To compute the expected value or mean of a random variable, multiply each possible
value of the variable by its probability and add these products.

Examples
4.1 A fair coin is tossed three times. Let X denote the number of heads obtained. Find
the mathematical expectation of X.

Solution
The probability density function of X is as follows.

X 0 1 2 3

f(x)

Hence,
E(X) = (1 × 1/8) + (1 × 3/8) + (2 × 3/8) + (3 × 1/8)
= 0 + 3/8 + 6/8 +3/8 = 12/8 = 1.5

4.2 What ate the mathematical expectations of the variables X, Y and Z as defined in
example 3.1.1 of chapter 3.

Solution
The probability density function of X is

X 2 3 4 5 6 7 8 9 10 11 12

f(x) 1/36 2/36 3/36 4/36 5/36 6/36 5/36 4/36 3/36 2/36 1/36

E(X) = (2 × 1/36) + (3 × 2/36) + (4 × 3/36) + (5 × 4/36) + (6 × 5/36) + (7 × 6/36) +

93
 (8 × 5/36) + (9 × 4/36) + (10 × 3/36) + (11 × 2/36) + (12 × 1/36)
= 7.0

The p.d.f of Y is

Y 1 2 3 4 5 6

g(y) 1/36 3/36 5/36 7/36 9/36 11/36

E(Y) = (1 × 1/36) +(2 × 3/36) + (3 × 5/36) + (4 × 7/36) + (5 × 9/36) + (6 × 11/36)

= 4.47

The p.d.f of Z is

Z 1 2 3 4 5 6

h(Z) 11/36 9/36 7/36 5/36 3/36 1/36

E(Z) = (1 × 11/36) + (2 × 9/36) + (3 × 7/36) + (4 × 5/36) + (5 × 3/36) + (6 × 1/36)

= 2.53.

Exercises 4.1
1. Suppose a box contains 10 balls of which 4 are red and 6 are black. A random
sample of size 3 is selected. Let X denote the number of red balls selected. Find
(E(X) if
(i) Sampling is with replacement.
(ii) Sampling is without replacement.
2. A box contains 6 balls labeled 1, 2, 3, 4, 5, 6. Two balls are drawn at random one
after the other. Let X denote the larger of the two numbers on the balls selected.
Compute E(X) if (i) Sampling is with replacement, (ii) Sampling is without
replacement.
3. A box contains 3 balls and 2 white balls. A ball is without replacement one after
the other until a white ball is drawn. Find the Expected number of draws required.

Expectation of a Binomial Random Variable

The probability density function of a binomial variable X is defined by
f(x) = nCxpx(1 – p)n-x; x = 0, 1, 2, …,n.
the mean or expected value of X is
E(X) =
and since the term of x = 0 is zero,
E(X) =
Using the binomial theorem, we have

94
 since p + 1 – p = 1. Hence,
E(X) = np.

Example 4.3
A fair die is rolled 12 times, what is the expected number of six appear?

Solution
Let X be the number of sixes that appear. X is a binomial random variable with n = 12
and P = 1/6 (probability of a six). Hence
E(X) = np = 12x 1/6 =2.
That is, we expected to get 2 sixes when a die is rolled 12 times.

Definition 2
Let X be a discrete random variable having probability density function f(x).

then we say that X has finite expectation otherwise we say that X

does not have finite expectation.

Examples
4.4 The Random Variables X has Probability

X -2 -1 0 3 5

f(x) ¼ 1/8 1/8 ¼ ¼

Compute the expected value of the following random variables.

(i) X, (ii) 3X, (iii) X + 5, (iv) X2

Solution
(ii)
= -2/4 – 1/8 +3/4 +5/4 = 11/8 = 1
(iii) The possible values of 3x are -6, -3,0, 9, 15.
The probability density function of 3x is

3x -6 -3 0 9 15

f(x) ¼ 1/8 1/8 ¼ ¼

95
E(3X) = (-6 × ¼) + (-3 × 1/8) + (0 × 1/8) + (9 ×1/4) + (5 × ¼)
= -6/4 – 3/8 + 9/4 + 15/4 = 33/8

Note that
P(3X = -6) = P(X = -2_,
And so on and
E(3X) = 33/8 = 3 × 11/8 = 3E(X)
(iii)

x+5 3 4 5 8 10

f(x) ¼ 1/8 1/8 ¼ ¼

Note that P(X +5) = 3 =P(X = 3 – 5) = P(X = -2)

E(X + 5) = (3 × ¼) + (4 × 1/8) + (5 × 1/8) + (8 × ¼) + (10 ×1/4)
= ¾ + 4/8 + 5/8 + 8/4 + 10/4 + 51/8 = 6 = 5 + 1 = 5 + E(X).

(iv)

X2 4 1 0 9 25

f(x) ¼ 1/8 1/8 ¼ ¼

E(X2) = (4 × ¼) + (1 × 1/8) + (0 × 1/8) + (9 × ¼) + (25 × ¼)

= 1 + 1/8 + 9/4 + 25/4 = 77/8 = 9

Note that E(X2) ≠ [E(X)]2.

Definition 3
Let X be a random variable whose probability density function is given by

x x1 x2 …………. xk

f(x) f(x1) f(x2) f(xk)

Let ψ(X) be a function of X. then the mean or expected value of the new random ψ(X) is
given by
E[ψ(X)] = ψ(x1) f(x1) + ψ(x2) f(x2) +…+ψ(xk) f(xk)
That is
E[ψ(X)] = ψ(xi) f(xi)

Theorem 1
Let X be a random variable and let ψ(X) = a X + b where a, b are constants, then
96
 E(ψ(X)) = aE(X)+ b

Proof:
Suppose the probability density function of X is
{xi, f(Xi), I = 1, 2, …,k)

From the above definition we have

E(aX + b) = (ax1 + b) f(x1) + (ax2 + b) f(x2) + … + (axk + b) f(xk) + bf(x1) +…+ bf(xk)
= a[x1f(x1) + x2f(x2) +… xkf(xk)] + b[f(x1) + f(x2) +…+
f(xk)]=
since {E(X) =

Properties of Excitations
1. If c is a constant and P(X = c) = 1, then E(X) = c.
2. If a and b are constants and X has a finite expectation, then aX has finite
expectation and
(i) E(aX) = aE(X) + b.
(ii) E(aX + b) = aE(X) + b.
3. If X and Y are two random variables having finite expectations, then
(i) X + Y has random finite expectation and E(X + Y) = E(X) + E(Y)
(ii) E(X) ≥ E(Y) if P(X ≥ Y) = 1.
(iii) DEXD ≤ EDXD.
4. A bounded random variable has a finite expectation. That is if P(X ≤M) = 1, then
X has a finite expectation and DEXD ≤ M.

Definition 4
Variable. Let X be a random variable with mean E(X) = µ.
The variance of X, denoted by Var(X) is defined by
Ver(X) = E{X - µ)2} =
A general formula that is usually simple for computing the variable is given below
Var(X) = E{(X - µ)2} = E{X2 – 2 Xµ + µ 2}
= E(X2) – E(2µX) + E(µ 2) = E(X2) - 2µE(X) + µ 2 (property 2)
= E(X2) - 2µ 2 + µ 2 = E(X2) - µ 2.
Thus, the computing definition of variance is
Var(X) = E(X2) – [E(X)]2 (7)
The variance of X is interpreted as a numerical measure of spread or dispersion about its
mean.

Example 4.5
1. Find the variance of a random variable having the following probability density
function.

x 1 2 3 4 5 6

97
 f(x) 1/6 1/6 1/6 1/6 1/6 1/6

E(X) = (1 × 1/6) + (2 × 1/6) + (3 × 1/6) + (4 × 1/6) + (5 × 1/6) + (6 × 1/6).

= 1/6 + 2/6 +3/6 + 4/6 + 5/6 + 6/6 = 21/6

E(X2) = (12 × 1/6) + (22 × 1/6) + (32 × 1/6) + (42 × 1/6) + (52 × 1/6) + (62 × 1/6)
= 1/6 + 4/6 + 9/16 + 16/6 + 25/6 + 36/6 = 91/6.
Hence from (7) we have
Var(X) = 91/6 – (21/6)2 = 105/36 = 2 (11/12).

Example 4.6
Find the variance of a Bernoulli random variable with parameter p. that is X has
the following p.d.f.

X 0 1

f(x) 1–p P

E(X) – 0 × (1-p) + (1 × P) = p
E(X2) = 02 × (1-p) + (12 × p) = p.
Hence,
Var(X) = p - p2 = p(1 – p)

Properties of Variance
1. If c is a constant and P(X) = c) = 1, then Var (X) = 0.
2. If a and b are contents, then
(i) Var(aX) = a2 Var(X)
(ii) Var(aX + b) = a2 Var(X)
(iii) Var(X) ≥ 0
The proof of property (1) is very trivial and this is left as an exercise to the reader,
we shall now give a proof of (ii). From (7),
Var(aX) = E[aX2] – [E(aX)]2 = E a2 X2 – [aE(X)]2
= a2 E(X2) – a2[E(X)]2 = a2{(E(X2)]2} = a2 Var(X).

Definition 5
Standard deviation. Let X be a random variable with mean µ.
The standard deviation is the positive square root of the variance, and is given by

The variance of the binomial random variable.

98
To calculate the variable of X, we need E(X) and E(X2).
Var(X) = (E(X))2.

From (3) we have

E(X) =np.

n
E[X(X – 1)] = (xpx(1 – p)n-x = px(1 – p)n –x
(w2hen x = 1 or 1 the expression is zero)
= px(1 –p)n-x

= n(n -1) p2 px-2(1 – p)n-x

Let y = x – 2 = n(n – 1) p2 (ypy (1 – p)n-2-y = n(n – 1) p2 [ P + (1 – p)]n-1
Hence
E(X(X – 1) = n(n – 1) p2
E(X(X – 1)) = E(X2) – E(X) = n(n – 1)p2
E(X2) = n(n – 1) p2 + E(X) = n(n – 1) p2 + np.
Thus,
Var(X) = N(n – 1) p2 + np – n2p2 = n2p2 – np2 + np – n2p2 = np(1 – p).

The Variance of a Geometric Random Variable

The probability density function of X is given by
f(x) = p(1 – p)x-1, x = 1, 2, …
E(X) =
E(X) = P + 2p (1- p) + 3p(1 – p)2 + … = P{1 + 2(1 + p) + 3(1-p)2 + …}.
From the Binomial theorem
(1 – a)-2 = 1 + 2a + 3a2 + … DaD<D.
Putting a = (1 – p), we obtain1
E(X) = P. (1 – (1 – p))-2 =

E[X(X – 1)] =
= p[2.1(1-p) + 3.2 (1 – p)2 +…+ r(r -1)(1 – p)r-1+…].
Multiply both sides by 1-p,
(1 – p) E[X(X- 1)] = p(2.1(1 –p)2 + (3.2(1 – p)3 +…+ r(r-1)(1-p)r +…+]
E(X(X-1)] – (1-p)E[X(X-1)] = p[2(1-p) + 4(1-p)2 + 6(1 – p)3 +…+ 2r(1 – p)r +…]

= 2p(1 – p) [ + 2(1-p) + 3(1 – p)2 +…] = 2p(1- p).

Hence

99
 E[X(X)] – (1-p) E[X(X – 1)] =

E[X(X – 1)] {1 – (1-p)} =

which implies that

E[X(X – 1)] =

E(X2) = + E(X) =

Thus

Var(X) =

The Variance of a Poisson Random Variable

Let X be a poisson random variable with parament λ.

E(X) =

= λe-λ e-λ eλ =λ

E[X(X – 1)] =

= e- λ λ2

E(X2) – E(X) = λ2
E(X2) = λ2 + E(X) =λ2 + λ
Hence,
Var(X) = λ2 + λ – λ2 = λ.
This shows that the mean and variance of a passion random variable are both equal.
Exercise 4.2
1. Compute the variance and standard deviation of the random variable X defined in
exercises 4.1.1
2. A fair coin is tossed twice. Let X denote the number of heads that appear.
Compute
(i) the expected value of X
(ii) the variance of X
100
 (iii) the expected value of

3. Calculate the mean, variance and standard deviation of the random variable having
the following probability density function.

X -2 1 0 1 2

f(x) 3/10 1/5 1/10 1/5 1/5

4. Let X be any random variance, show that E(AX + B) = aE(X) + B

Where a, b are contants.
5. Let f(x) = pDXD, x = ± 1, ±2,…, 0 < p < 1

Show that f(x) is a pdf. Find E(X).

6. Let X be any random variable having finite expectation. Prove that
DEXD ≤ EDXD.
7. Let X be a random variable such that for some constant K, P(DXD ≤K). Prove that
X has finite expectation and DEXD ≤ k.
8. Show that if X and Y are any two dependent random variables with finite
variances, then Var(X + Y) ≠ Var(X) + Var(Y).

4.3 Probability Generating Function

In this section we shall introduce important mathematical concepts in probability theory.
The rth moment of a random variable X is defined by E(Xr), where E(Xr) and E(X - µ)r
are called the rth moment of X about 0 and µ respectively.
From (2)
E(Xr) =
Note that the second moment of X about µ is the variance of X. an indirect way of
calculating expectation is the use of probability generating function or moment
generating function. it is a mathematical device to simplify the calculations of moments
of X. its more general usefulness will not be apparent until we get to chapter 6.

101
Definition 6
The probability generating function of a non-negative, integer-valued random variable X
is defined by
E(Sx) = P(S) = (8)
From (6), we see that
P(S) = E(Sx)
By differentiating with respect to s we have
P’ (s) = E(XSX-1)
P’’(s) = E[X(X – 1) SX-2]
P(r) (s) = E(X(X – 1)(X – 2)…(X – r + 1)sX-1)
Putting s =1 in the above derivatives, we have
P’(1) = E(X)
P’’(1) = E[X(X – 1)] = E(X2) – E(X)
P(r) (1) = E[X(X – 1)(X – 2)…(X – + 1r)].
Thus the mean and variance of X can be obtained from P(s) by the following formulas.
E(X) = P’(1)
E(X2) – E(X) = P’’(!)
E(X2) = P’’(1) + E(X) = P’’(1) + P’(1)
And
Var(X) = E(X2) – [E(X)]2 = P’’(1) + P’(1) – [P’(1)]2
We now illustrate the use of these formulas with the following examples.

Examples 4.7
Let X be a poisson random variable wit parameter λ. Find the mean variance of X

f(x) =

The probability generating function is defined by

P(S) = E(SX)= (xpX(1 – p)n-x = n x
( (sp)x(1 – p)n-x.
By the Binomial theorem
= [sp + (1 –p)]n

102
Thus,
P(S) = [sp + (1 – p)]n (12)
Differentiating, we have
P’(s) =np [sp + (1 – p)]n-1
P’’(s) = n(n – 1)p2 [sp + 1 – p]n-2
Var(X) = p’’ (1) + - [p’(1) – [p’(1)]2 = n(n – 1) p2 + np – (np)2 = np(1 – p).
The table below summarizes some of the results of this chapter

Random Probability Mean Variance Probability Moment

variable density generating function generating
function function

Bernoulli Px(1-p)1-x P P(1-p) (sp+1 – p) pet + (1-p) -

∞< t <∞
n
Binomial Cxpx(1-p)n- Np Np(1-p) [sp +(1-p)]n [pet+(1-p)]n
x

x = 0, 1, 2,
….n

Geometric p(-p)x-1 P[1-s(1-p)]-1

x = 1,2,…

Poisson Λ Λ e λ(s-1) e λ(et – 1)

Uniform

X= a, a +
1,…b

The Moment Generating Function

Definition 7
Let X be a discrete random variable with pdf f(x). the moment generating function
denoted by Mx(t) is defined by
MX(t) = E(etX)
Where t is any real constant for which the expectation exists. (13)
103
 MX(t) =

RtX = 1 + tX + +…

E(etX) = E(1 + tX + + …+

Under fairly general conditions we shall assume that expectation of infinite sum equals
the sum of the expected value but this is true in general for finite sum

1 + tE(X) + + +…

That is

MX(t) = 1 + tE(X) +

Differentiating with respect to t, we have

M’X(t) = E(X2) +…+

M’’x(0) = E(X2+)
hence,
Var(X) = Mx’’(0) – [MX’(0)2]
And
Mx(t) = E(etX), Mx(r)(t) = E[Xr e tX] (15)
Putting t = 0, we have
M(r)(0) =E(Xr)
that is, E(Xr) is the rth derivative of Mx(t) evaluated at t=0. (16)

Note: It is assumed that ∑ etX f(x) converges for all values of t.

The domain of Mx(t) is all real number t such that ∑ etX f(x) converges. If Mx(t) is defined
for some t0> 0, then it is defined in the interval 0 ≤ t ≤ t0.

Examples 4.9
Let X be a binomial random variable with parameters n and p. then
MX(t) = E(etX) = n
CxPx(1 – p)n-x = n
Cx(Pet)x(1 – p)n-x
from the binomial theorem
(a + b)n = n
Cxaxbn-x = [pet + (1 – p)]n,
104
 on putting a = pet, b = 1- p.

Example 4.10
Let X be a Poisson random variable with parameter λ. Then

MX(t) = e-λeλet = eλ(et -1)

Differentiating, we have
M’X(0) = λete-λ(1-et)
M’x(0) = λete-λ(1-et)Dλ=0 = λ E(X)
Mx’’(t) = λet e-λ(1-et) + λ2 e2te-λ(1-et)
Mx’’(0) = λ + λ2 = E(X2)
Hence,
Var(x) = λ + λ2 – λ2 = λ.

Example 4.11
Let X be a geometric random variable with parameter p. them
Mx(t) =

The fourth equality follows from sun to infinity of geometric series Further
applications of Mx(t) will be considered later in Chapter 6.

Exercise 4.3
1. Find the probability generating function of the geometric distribution with
parameter P. hence determine the mean and variance of the distribution.
2. Let X be uniformly distributed on (0, 1, 2,…n). Find the mean and variance of X.

Hint:

3. Let X be a random variable with finite variance. Prove that for any real number a,
Var(X) = E[X – a)2] – [E(X) + a]2.
4. A fair die is tossed 50 times. Let X be the number oftimes six appears. Evaluate

105
 (i) E(X) (ii) E(X2) (iii) VAR(X)
5. Find the expected value and variance of the random variable X os Example 3.1.1.
6. Show that if c is a constant, Var(x + c) = Var(X).
7. Let X be uniformly distributed on (0, 1, 2,..N). Find P(S), mean and variance of X.
8. Let X be defined by

x 1 2 3
1 1 1
2 4 4
f(x)

Find P(S) and hence the mean and variance of X.

9. Prove that if X has variance σ2, and mean µ, then

every real number a.

10. Derive the MGF for the following: (a) Bernoulli, (b) Negative Binomial, (c)
Binomial and use it to find the mean and variance.

11. A random variable X has the pdf f(x) defined by

c(x + 2), x = 1, 2, 3, 4
f(x) =
0 elsewhere
Find (i) c, (ii) moment generating function of X, (iii) use result of (ii) tofind the
mean and variance of X.

12. Prove that for any random variable, X, E(X2) > [E(X)]2.

13. Show that [E(X – a)2] is minimized when a = E(X).

14. A fair die is rolled until all the 6 sides appeared at least once. Find the expected
number of rolls needed.

p, x = -1
15. Let f(x) =
Px(1-p)2, x = 0,1, 2,…
(a) Show that f(x) is a pdf. (b) Find Mx(t), the moment generating function of
(b) Hence, the mean and variance of X.

106
 MODULE THREE (UNIT 1)

CONTINUES RANDOM

5.1 INTRODUCTION

In the chapter so far, we consider discrete random variables and their probability
density functions e.g. Binomial, Poisson, Geometric. These are essentially counting
variables. However, there exist random variables whose set of possible uncountable,
such random variables are called continuous random variables.

In this chapter, we shall study in considerable detail, a number of important continuous

random variables and their characteristics. We shall consider an idealized range space
for a random variable X, in which all possible real number (in some specific intervals or
set of intervals) many be considered as possible outcomes. We shall essentially replace
the summation in the previous chapters in the discrete probability function case by
integration for the continuous probability density function case.

Example 5.1

Let a point be chosen at random from the interval (0, 2) and let X represent the point
chosen. Then 0 ≤ X≤ 2. the range space of X consists of unaccountable real numbers. Any
real number in the interval (0, 2) is a sample point. Let A be the event that the point
chosen is 0.3. From the basic definition of probability we have

107
 number of po int A 1
P ( A) = = = 0.
number of po int s in Ω ∞

Hence,

In general, if X is a continuous random variable, probability that X takes a specific value

is 0. random variables denoting measurements of physical quantities such as time,
weight, temperature, length are continuous random variables. The proability that X lies
between two defined values is given by

P(a ≤ X ≤ b).

P(X ε A) = ∫A
f ( x)dx

For any set A of real numbers

P(X) is denoted by F(X).

Definition

The distribution function F of a random variable X is the function

P(X) = P(X ≤ x), - ∞ < x < ∞ ,

If we know the probability distribution function of X, then we can calculate the

probability that X fails in a specific interval. For example

P(a < X ≤ b) = P(X ≤ b) – P(X ≤ a) = F(b) – F(a).

Properties of Distribution Function

108
A probability distribution function is any function F satisfying the following conditions:

(i) 0 ≤ F(x) ≤ 1 for all x

(ii) F(x) is a non-decreasing function of x.

(iii) F(-∞) = 0 and F(+∞) = 1.

(iv) F(x +) = F(x) for all values of x where F(x+) = Lim F(x + δ).
δ−0

Graphs of a typical distribution function

Example 2

Show that

e x x<0
F ( x) = 
1 x >0

is a probability distribution function

Solution

109
0 ≤ ex ≤ 1 for x ≤ 0 and ex is a non decreasing continuous function, hence F(x) is a
probability distribution function.

Probability density function [Link]

For discrete random variable

x
F(x) = P(X ≤ x) = ∑ f ( y)
−∞

Where f(y) = P(X = y).

Similarly, for continuous random variables, there exists a function f(x) such that

x
F(x) = Px(X ≤ x) = ∫−∞
f ( y )dy

Such a function f is called probability density function of X. to see this consider the
following.

The probability that the random variable X takes on a value between x and x + Δx is
P(x < X < x + Δx) = P(X ≤ x + Δx) – P(X < x)

= F(x + Δx) – F(x).

this is the probability associated with an interval of length Δx

F ( x + ∆x ) − F ( x )
∆x

is the probability per unit length in the interval. If we now let Δx tends to zero, we have

F ( x + x ) − F ( x)
Lim = F ' ( x ) = f ( x)
∆x − 0 ∆x

x
Hence, F(x) = ∫−∞
f ( y )dy where F(-∞) = 0

110
F(x) represents the probability per unit distance along the x acis and

d
F(x) = F(x)
dx

This shows that the p.d.f of X is obtained by differentiating the distribution function of X.

Properties of probability density function

A probability density function is any function satisfying the following conditions.

(i) f(x) ≥ 0

∞
(ii) ∫−∞
f ( x) dx = 1

a ∞
(iii) P(X < a) = ∫−∞
f ( x) dx = 1 - ∫a
f ( x) dx = 1 – P(X > a)

(iv) if A is any interval, P(X ε A) = ∫−∞

f ( x) dx .

Figure (2) represents a graph of the function f(x). the area enclosed by the curve f(x)
between a and b is

b
∫a
f ( x) dx = P(a ≤ X ≤ b) = F(b) – F(a).

Graphical Representation of a Continuous Random Variable.

111
Theorem

(i) The function F is non decreasing

(ii) Lim F(x) = 0 and Lim F(x) = 1.

x−∞ x −∞

Proof:

(i) Suppose x1 ≤ x2 and let A = { X ≤ x1}, B = {X ≤ x2), then

A ⊂ B since x1 ≤ x2

Thus

P(A) ≤ P(B)).

x
(ii) Lim F(X) = Lim
x−∞ x −∞ ∫
−∞
f ( y )dy = 0

x
Lim F(x) = Lim
x−∞ x −∞ ∫
−∞
f ( y )dy =1

Examples

e − λx , x ≥0
Exponential Distribution. Let f(x) = 
0, x<0

The graph of f(x) is shown in Figure 2 below

f(x)

112
Figure 3. Density function of exponential distribution.

The random variable X having its probability density function given by (1) is said to have
exponential distribution with parameter λ.

5.4 Normal Distribution

1 − 12 x 2
Let f(x) = e , − ∞ < x < ∞.
2π

The graph of f(x) is shown in Figure 4 below.

Figure 4. Standard Normal Distribution

113
5.5 Let the random variable X have the probability density function f(x) – 1/2e-1/2x 0
≤ x ≤ ∞. Find (i) P(2 ≤ X ≤ 7) (ii) P(X ≥ 10).

Solution

7
−1 / 2 x − 2x 7
P(2 ≤ x ≤ 7) = ∫ 1 / 2e dx = − e Ι = e −1 − e − 7 / 2 = 0.338
2 2

And

∞ ∞ −5
P(X ≥10) = ∫ 1 / 2 e −1 / 2 x dx = − e −1 / 2 x Ι =e .
10 10

Suppose that X is a continuous random variable whose probability density function is

given by

A e-2x 0< x < ∞

f(x) =

0 otherwise

(i) what is the value of a? and (ii) find P(X > 1).

Solution

(i) From property (ii) of a probability density function, we have

∞
∫−∞
f ( x) dx = 1

implying that

∞ ∞
∫ ae −2 x dx = 1 = a ∫ e − 2 x dx = 1 / 2
0 0

==>a=2
114
hence

2e -2x 0 < x < ∞

f(x) =

0 otherwise

∞
−2 x
(ii) P(X > 1) = ∫ 2e
1
dx = e-2 = 0.14

5.2 Expectation and variance of continuous random variables.

Let X be a continuous random variable with probability density function f(x). Suppose we divide the
range of X into small intervals, each of length Δx (by the definition of f(x)).

The expected value of X is therefore approximately

∑ xf ( x) ∆x
The limiting value of this as Δx 0

∫ xf ( x) dx.
This leads us to the following definition.

Definition 3

If X is a continuous random variable with probability density function f(x), then

∞
E(x) = ∫−∞
xf ( x) dx

In general,

∞
E(Xr) = ∫ −∞
x r f ( x) dx

115
 ∞
E(Φ(X)) = ∫−∞
Φ ( x) f ( x) dx.

∫ (x − µ )
2 2 2
The variance σ = E[(X – μ) ] = f(x) dx = E(x2) – μ2
−∞

Where μ = E(x).

Examples

5.6 Suppose X is a continuous random variable with p.d.f f(x) = λe-λx, 0 < x < ∞.

Find (i) the mean and variance of X and (ii) the probability distribution function F(x), and the probability
that X lies between 2 and 5.

Solution

∞ ∞
E(X) = ∫ x.λe −λx dx = ∫ xλ e −λx dx
−∞ 0

By integrating by parts

∞ d − λx ∞ ∞ 1
= ∫ −x (e ) = − xe −λx Ι + ∫ e −λx dx = .
−∞ dx 0 0 λ

Similarly,

∞ ∞ d − λx ∞ ∞
E(x2) = ∫ x 2 .λe − λx dx = ∫ − x 2 (e ) = − x 2 e − λx Ι + 2 ∫ xe −λx dx
0 0 dx 0

0 2 2
= 2∫ x o − 2 x dx = xλl − 2 x dx = 2 E ( x) =
o λ∫ λ2

116
 2 1 1
hence, var(x) = − =
λ 2
λ 2
λ2

x
xf ( y ) dy = x − λy [ =1− e1 x
f ( y )dy = ∫ λl − λydy = −e
F(x) = ∫− ∫ 0
=
(u)
5
p (2 < x < 5) = ∫ λe 2 x dx = F (5) − F (2)1 − e −5 − 1 + e − µ = e − µ − e − µ
2

5.7 Let X be a continuous random variable with probability density function

1
6

0<x<6
F(x) =

elesewhere

find (1) F(x) (ii), E(x), E(x2) and var (x).

solution

1x
(1) F(x) = ∫0 6
= p( x≤ x)
0 1 1 1 2 6 1 1
E(x) = ∫0 x 6dx = 6 − 2 x [ 0 = 6 − 2 − 36 = 3

6 1 1 1 1 1
E ( x 2 ) = ∫ x 2 − dx = − x 3 [ 60 = − − 6 3 = 12
0 6 6 3 6 3

117
hence var(X) = 12-9 =3

5.3 symmetric probability Density Function

definition 4

a probability density function f(x) is called a symmetric probability function if

f(-x) =f(x) for all x.

1 x2
1
for example. If f(x) = 2 xe isasynmericfunction.
2

similarly,

1
f(x) = −0< x < a
2x

is a symmetric probability density function

Definition.

A probability density function f(x) is said to be symmetric about a if

F(a + x)=f(a – x) for all x.

If x = 0. then f(x) is symmetrical about 0. for example, if

1
−1
F(x) = 2π e ( x − π ) 2 =< x <=
2
118
Then

1 1 ( µ + x- µ )2 1 − 1
F(( µ − x) = e = e
2x 2 2π 2

1 1 ( µ + x- µ )2 1 − 1
F( µ − x) = e = e
2x 2 2π 2

hence, f( µ +x) = f( µ -x). thus f(x) is symmetrical about µ .

theorem 1

if f(x) is a symmetric probability density function then

f(x) = 1-f(-x)

where f(x) is the distribution function of x.

proof
x −x
f(-x)= ∫
−
f ( y )dy = ∫
−
f (− y )dy

since f(y) =f(-y). Let z = -y, then

119
 x = = x
f(-x) = ∫
−
f ( z )(−1)dz = ∫ f ( z )dz =
x ∫
−
f ( z )dz − ∫ f ( z )dz = 1 − f ( x)
−

and hence

f(x) = 1-f(x)

terminology

when we speak of the probability distribution of a random variable X we mean its

probability density function and when we speak of the distribution function or
cumulative distribution function we always mean f(x).

5.2 some important continuous random variables.

5.2.1 The Normal Distribution

one of the most important continuous random variables is the Normal random
variable, X. The range of X is the real line.

Definition

The Normal Distribution. The continuous random variable X has the normal
distribution with mean µ and varianceo3 if µ has the probability density function.

− 1 (x − µ )
3
1
F(x) = e =< x <=
2π 2 02

The graph of f(x) is given in Figure (4)

It can be verified that f(x) satisfies

120
 f ( x ) dx = 1
1. ∫

2. ∫ xf ( x) dx =µ

( x − µ ) 2 f ( x)dx = 0 2
3. ∫
Properties of the Normal Distribution.

f ( x ) dx = 1
1. ∫ this can be verified.

x−µ
Thus let z =
σ

∞ 1 −
1
(x − µ )2 dx = ∞ 1 −
1

∫
−∞
σ 2π
e 2
σ2 ∫
−∞
2x
e 2
dz = 1

1 1
∞ − y2 ∞ ∞ − (z2 + y2 )
1 1
I = ∫ dy = ∫ ∫
2 2 2
e e dz dy.
2π −∞ 2π −∞ −∞

Let us introduce polar coordinates to evaluate this double integral:

Z = rcos θ, y = rsin θ
r2
1 2π ∞ 1 2π − ∞ 1 2π
I = 2
∫ ∫ re −r 2
dr dθ = ∫ −e 2
Ι dθ = ∫ dθ = 1.
2 0 0 2π 0 0 2π 0

where dzdy = rdrdθ, Hence

(x − µ)2
1
∞ 1 −
∫
−∞
σ 2π
e 2
σ2
dx =1.

121
2. E(X) = μ, Var(X) = σ2.

The p.d.f is completely specified when μ and σ are known.

When μ = 0 and σ2 = 1. X is said to have a standardized normal distribution. This

is, the pdf of X is

1 − 12 x 2
f(x) = e ,− ∞< x<∞
2π

The importance of the standardized normal distribution is the fact that it is

tabulated.

3. f(x) is symmetrical about the point μ, hence F(-x) = 1 – F(x).

Most available tables are tabulated only for positive values of x since F(-x) can be
found from f(x) and f(x) are given at the back of this book. The graph of f(x) is
bell-shaped curve centred at its mean.

X −µ
4. IF X has a normal distribution with mean μ and variance σ2, then Y = has a
ο
standardized normal distribution. The proof of this is left as an exercise. Y is said
to have a standard Normal distribution.

5. The moment generating function of X is defined by

(x − µ)2
1
∞ 1 −
Mx(t) = E(etx) = ∫ −∞
e tx σ
2π
e 2
σ2
dx.

X −µ 1
Let y = , then dy = dx.
σ σ

x = μ + y σ.

Thus,
1
∞ 1 − y2
Mx(t) = ∫ e t ( µ + yσ ) e 2
.σ dx dy.
−∞ σ
2π
122
 e tµ
1 1 1
∞ − ( y − tσ ) 2 tµ + σ 2 t 2 1 ∞ − ( y − tσ ) 2
=
2x
∫ −∞
e 2
dy = e 2

2π
∫−∞
e 2
dy .

Let y – tσ = y, then
1 1
1 ∞ − ( y − tσ ) 2 1 ∞ − y2

2π
∫
−∞
e 2
dy =
2π
∫ −∞
e 2
dy = 1.

Hence,

1
tµ + σ 2 t 2
2
Mx(t) = e

Examples

5.8 If X has a standard normal distribution i.e. mean of X, μ=0, and variance = 1. Find

(i) P(X ≤ 0.5) (ii) P(X ≥ 0.5)

(iii) P(X ≤ -0.3) (iv) P(0.3 ≤ X ≤ 0.5)

(v) P(-0.3 ≤ X ≤ 0.5)

Solution
∞
(i) P(X ≤ 0.5) = ∫
−∞
f ( x) dx = F (0.5).

From table (3) F(0.5) is found by locating the first two digits (0.5) in the column
headed x, the desired probability of 0.6915 is found in the row labeled 0.5 and
the labeled 0.00. That is
1
0.5 1 − x2
P(X ≤ 0.5) = ∫−∞
2π
e 2
dx = 0.6915.

and

(ii) P(X ≥0.5) = 1 – P(X < 0.5) = 1 – 0.6915 = 0.3085.

(iii) P(X ≤ -0.3) = F(-0.3) = 1 – F(0.3)

From the table, F(0.3) = 0.6179.

123
(iv) P(0.3 ≤ X ≤ 0.5) = F(0.5) – F(0.3) = 0.6915 – 0.6179 = 0.0736

(v) P(-0.3 ≤ X ≤ 0.5) = F(0.5) – F(-0.3) = 0.6915 – 0.3821 = 0.3074.

Example 5.9

If x is a normal variable = 5 and variance 02 = 16, find (i) P(X) < 6)

(ii) P(3 < x < 7) (iii) P{X-5\>2}.

Solution

From property (4)

L M / N HO L M / 0.5987
IJ K <J4 3
5 5 5
(i) P (X <(6) = PH

U U M / W'0.5 U O U 0.5X
>J 4 IJ4 VJ4
5 5 5
(ii) P(3 < X < 7) = PH

= Ф(0.5) – Ф(-0.5) = 0.6915 – 0.3085 = 0.3830

(iii) P{|X-5|>2} = P(X – 5 > 2 or X – 5 < -2) = P(X > 7 or X < 3}

= P(X > 7) + P(X<3) = PHO Y M = HO U M

VJ4 >J4
5 5

= P(Y > 0.5) + P(Y < -0.5) = 0.3085 + 0.3085 = 0.6170.

Notation:

When we write X is N (o2), we mean that X has a Normal probability distribution with
mean and variance o2

5.3.2 The Normal Approximation to the Binomial

Let X be a binomial random variable with parameters n and p, then

E(X) = np, var (X) = np (1-p)

An approximation to the distribution of X is given by De Moivre Laplace limit theorem.

124
Theorem (De Moivre laplace Limit Thoerem)

If X denote a binomial random variable with mean up and variance np(1-p), then for any
a < b,

^ ' )_
lim N \] L L ab / ФWaX ' ФW] X.
0Z[ `)_W1 ' _X

The accuracy of this approximation is quite good for values of n greater than 10 P(1-p).

This theorem says that if X is B(n,p) then large n, X is N(np, np (1-p).

Example 5.10

Let X be a binomial random variable with parameters n = 40, p = 0.20. find the
probability that X = 5.

Since X is discrete and normal distribution is for continuous random variable, we write

U Y M
5.4J5 IJ5 4.4J5
[Link] [Link] [Link]
P (X = 5) = {(4.5 < 5.5 = PH

= p (0.26 < y < 0.79) = Ф (0.79) – Ф(0.26) = 0.18226.

Notation: Whenwe write X is N0 (µ o), we mean that X has a normal probability

distribution with mean µ variance 02

5.2.4 The Exponential Distribution

A continuous random variable, X is said to have an exponential distribution with

parameter > 0 if its pdf is given by,

f W( X / g hi lf ( m 0 n
Jjk

0 (U 0
The pdf has the graph shown in Figure 3. It can easily be shown that

125
 [
o hi Jjk / 1.
e

The exponential distribution plays an important role in point processes and in describing
random phenomena. A number of real life situations can be modeled using an
exponential distribution.

For example, time to failure of an equipment.

Properties of the Exponential Distribution

(1) E(X) and standard deviation of X are the same. The expected value of X is given by

E(X) = pe (hi Jjk q(

[

On integrating by parts we have

[ [
∞
[
1
o (hi Jjk q( / o '(q ri Jjk s / '(i Jjk | = o i Jjk q( /
e e
0 e h

(ii) The variance of X is given by

Var(X) = E(X2) – (E(x))2

∞
E(X2) = pe ( : hi Jjk q( / pe '( : qri Jjk s / '( : i Jjk | = 2 pe (i Jjk q( /
[ [ [ :
0 j,

Hence,

' / .
: 3 3
j, j, j,
Var(X) =

Thus, the mean and standard deviation of an exponential distribution are the same.

(ii) Lack of Memory of the Exponential Distribution

A random variable X is said to be without memory =, or memorable, if

P(X> s + t\X> t) = P (X > s) for all s, t > 0

P(X > s + t) X > t = P (X > s + t and X > t = P(X> s + t

P(X> t) P(X > t)

126
By the definition of memoryless, we have,

P (x > s) = 1 – F(x + t)

1 – F (t)

That is,

1 – F(s) = F(s+t)

1-F(t)

(1-F(s)) (1-F(t) = 1 –F (s + t)

F(X) = P(X< x) = dy = 1-ee

1 – F (x) = ex

P (X > s) = 1-F (s) = ex

P(X > s + t ) = 1 – F(s + t) = e

(p(x>t) = 1 – F (t) = e

P(X > s) = e

e-µ = e

thus, an exponential random variable is memoryless. The only continuous random

variable X assuming non negative values satisfying

P(X < s + t/X > t ) = p (X > s) for all s, t > 0

(iii) The moment generating functions of X is given by (definition 7 of chapter 4).

5,2,5 The Gamma Distribution

Definition 6

The gamma function, denoted by I’, is defined as follows:

I (x) =

By integrating by parts, we have

∞
F(X) = pe 'v kJ3 qWi Jw Xqv / 'v kJ3 i Jw x = pe W( ' 1Xv kJ: i Jw qvn
[ [
0
127
 = (x – 1) pe v kJ: i Jw qv / W( ' 1XyW( ' 1X
[

Γ(x) = (x – 1)Γ(x-1) = (x-1). (x-2) Γ(X-2) = (X-1)!Γ(1)\

(1) = pe i Jw qv / 1.
[

Hence

Γ(x) = (x-1)!

It is also easy to verity that

ΓH M / pe v J, i Jw qv / √{
3 [ -

:
(i)

pe / 1.
[ w |}- ~ }€w
 W|X
(ii)

Definition 7

Let X be a continuous random variable assuming only non-negative values. Then X is said
to have a gamma distribution if its pdf is given by

i Jjk ( Y 0
jWjkX|}-
f W( X / \ W0X n
0 i‚ƒi„…i†i
λ and n are called the paramenters of the distribution. f(x) is denoted by Γ(n, λ).

Properties of the Gamma distribution

E(X) = , ˆ]†W^ X /
0 0
j j,
1.

To see this, we have

E(X) = pe (. q( / pe q( / p ^ 0 i Jjk q(.

[ jWjkX|}-~ }‰Š [ j| I | ~ }‰Š j| [
W0X W0X W0X e

Let y = λx, dy = λdx

[ [
v 0 1 1 [
o ( 0 i ' h( q( / o H M i Jw qv / 0‹3 o v 0 i Jw qv.
e e h h h e

Hence,
128
 . / .
j| W0‹3X 0
j|Œ- j
E(X) =
W0X

Similarly,

E(X2) = pe ( : . q( / p ( 0‹3 i Jjk q(.

[ jWjkX|}- ~ }‰Š j| [
W0X W0X e

[
yW) = 2X
o ( 0‹3 i Jjk q( / .
e h0‹:

Hence,

. / .
j| W0‹:X 0W0‹3X
W0X j|Œ, j,
E(X2) =

Thus

' / .
0W0 ‹3X 0, 0
j, j, j,
Var(X) =

The moment generating function of x is given by

Mx(t) = pe i k q( / p ( 0J3 i JkWjJX q(

[ j| I |}- ~ }‰Š j,[
W0X W0X e

. p Ž(Wh ' X0J3 i JkWjJX q(.

j| 3 [
W0X WjJX|}- e
=

Let y = x(λ – t), dy = (λ – t) dx. Therefore

0
p v 0J3 i Jw qv / . y W) X / x ‘
j| [ j| j
W0XWjJX| e W0XWjJX | jJ
Mx(t) = (8)

2. If X is a random variable having Γ(n, λ) distribution and Y is a poisson random

variable with parameter λx, then

P(X ≤ x) = P(Y ≥ n)

That is

pe qv / ∑[ .
[ j| w|}- ~ }‰ WjkX“ ~ }‰Š
W0X 0 ”!

3. If n = 1, f(x) becomes
129
 f(x) = λe-λx.

Hence, the exponential distribution is a special case of the Gamma distribution.

5.2.6 The Chi-square Distribution

3
:
A special case of the gamma distribution in which λ = and n replaced by n/2, where n
is a positive integer is called the chi-square distribution. The pdf is given by
|

( H M i J,k
| 3 , |
J3
,
1
/ ( , J 3 i J, , ( Y 0
| Š
:
f W( X /
yH M 2, y H M
0 |
0
: :

A continuous random variable having pdf f(x) given by (7) is said to have a chi-square
distribution with n degrees of freedom (denoted by χn). Because of its importance, the
chi-square distribution is tabulated for various value of the parameter n) see table 4). If
X has a chi-square distribution with n degrees of freedom, we have

(i) E(X) = n;

(ii) Var(X) = 2n;

3
:
(iii) Mx(t) = (1 – 2t)-n/2; t > .

The graph of f(x) is shown in figure below.

130
5.2.7 The uniform distribution

A continuous random variable X assuming all values in the interval (a, b) where both a
and b are finite is said to be uniformly distributed over the interval (a, b) if its pdf is
given by

] L ( L an
3
f W( X / –—J˜
0 i‚ƒi„…i†i
The graph of f(x) is shown in figure below:

Properties of the Uniform Distribution

131
 , ™k WX /
˜‹ — ~ š› J ~ œ›
: W— J ˜X
1. E(X)

i ˜ / 1 = ] = =
W˜X,
:!

i ˜ = 1 = a = =
W—X,
:!

/ xWa ' ] X = = =‘

~ š› J ~ œ› 3 r— , J ˜ , s W— AJ ˜A X ,
W— J ˜X —J ˜ :! >!

™I WX / x = =‘
3 r— , J ˜ , s W— A J ˜A X ,
—J ˜ :! >!

/ /
W— , J ˜ , X W—J˜XW— ‹ ˜X —‹ ˜
:W—J˜X :W— J ˜X :
Mx’(0) =

Similarly,
W— A J ˜ A X
>W—J˜X
Mx’(0) =

Hence,

' / – / Ž2Wa : = ]a = ] : X ' 3W] = a X

W— A J ˜ A X ˜‹ — —, ‹ ˜— ‹ ˜ , X ˜‹ — 3
>W—J˜X : > : <
Var(X) =

] L ( L an
kJ ˜
F(x) = P(X ≤ x) = \— J ˜
1 (Ya

5.2.8 The Beta Distribution

A continuous random variable X is said to have a beta distribution if its pdf is given by α,
β > -1

( Ÿ W1 ' (X  , 0 U ( U 1n
Ÿ ‹   ‹ 3X!
f(X) = \ Ÿ!  !
0 i‚ƒi„…i†i
f(x) becomes the pdf of a uniform distribution over (0, 1) when α = β = 0.

Properties of the Beta distribution

132
 ( Ÿ W1 ' (X  q( / pe ( ¡ ‹ Ÿ W1 – (X  q(.
3 WŸ ‹   ‹ 3X! WŸ ‹   ‹ 3X!
E(xr) = pe ( ¡
Ÿ!  !  ! Ÿ!
1.

From definition (Beta function)

/ pe ( Ÿ W1 ' (X  q(
Ÿ!  ! 3
WŸ ‹   ‹3X!

and
3 W† = ¢ X! £!
o ( ¡ ‹ Ÿ W1 – ¢X  q( /
e W† = ¢ = £ = 1X!

We have

9 / .
WŸ ‹   ‹ 3X! W¡ ‹ ŸX!  ! WŸ‹  ‹3X!W¡‹ ŸX!
 ! Ÿ! W¡ ‹   ‹3X! Ÿ!W¡‹ ‹3X!
E(xr) =

The moment generating function of X does not have a simple form.

5.2.9 The Weibull Distribution

A random variable is said to have a Cauchy distribution with parameter θ, -∞ < θ < ∞, if
its pdf is given by

f(x) = . , '∞ U ( U ∞
3 3
¤ Ž3 ‹ WkJ ¥X, 

For an illustration of a Cauchy distribution random variable. See example 6.12

Exercises

1. If X is N(0,1), determine the following probabilities

(i) P(X > 0.9)

(ii) P(X< -0.7)
(iii) P(0.7 < X < 0.9)
(iv) Find a such that P(X < -a) 0.10
(v) P(|X| < 1.5)
2. A random value X has the pdf.

133
 §( 0 U ( U 1n
f W( X / ¦
0 i‚ƒi„…i†i

Determine the value of the constant c. find F(x) and the mean and variance of X.

3. A random variable X has the pdf.

( 0 L ( L 1
f W(X / ¨2 ' ( 1 ( L 2n
0 i‚ƒi„…i†i

(i) Show that f(x) is actually a pdf

(ii) Find the means and variance of X.
4. Let X bed a continuous random variable having probability density function
3
:
f(x) = e|x|, -∞< x <∞
Find (i) P(-2<X<2) (ii) EW(X) (iii) Var (X)

5. Let X be a Normal random variable with mean 8 and variance 2. Find P(-2<X<6).
Determine the values of a and b such that P(X< a) = 0.25 and P(X>b) = 0.25

6. Suppose that X is N(5, 9). Find a constant c such that P(X > c) = 2P(X< c).

7. Let a point be chosen uniformly in the interval (a, b). ). Let let X denote the
distance of the point chosen from a. find the probability distribution function F(x)
of X.

Show thatpJ[ i J,k q( / √2{

[ - ,
8. (a)

(b) Show that f(x) = 1-|α|, -1 < x < 1 is a pdf

Show that pj i Jw qv / ∑”J3 .

[ w “}- jŠ ~ }‰
9.
W”X k©e k!

10. The lifetime in hours of an electric light bulb is a random variable with probability
density function

f(x) = ª exp H' (M , ( Y 0

3
:4e

Determine the value of θ

11. A probability density function f of random variable X is defined by

134
 0, (U0
¯(, 0U(U1
f W( X / n
® :, 1U(U§
3

­ 0, ( Y§

Where c is a constant

(i) Calculated the value of c and hence the mean and variance of X.
(ii) Sketch the graph of f.
12. A random variable X has pdf.

F(x) = k exp {-(x-2)2/9} < x <

(i) Find the value of k,

(ii) What is the probability distribution of X?
13. Suppose (x) = f(x) , where F(x) = f(x) x > 0

1 – F(x)

Show that F(x) = 1 – exp – {pe hW( Xq(°

j

14. The pdf of a random variable X is defined by

f(x) = ½ e|x|, -∞< x <∞

Find P(1<) /x/ < 2

15. (a) If X is N (o2) find k (as a function of u and 02 such that P(X > k) = 2P (X <k)

16. Let X be a random variable with pdf

(a) f(x) = λe-λ(x-a) x ≥ a;

(b) f(x) = ½ -λ(x-a) -∞< x <∞

Find the M.G.F of X and hence the mean and variance of x in each case

(a) F(x) = e x > a;

(b) F(x) = 1/2 e < x <
Find the M.G.F of X and hence the mean and variance of X in each case

17. A random variable X has mean 5 and variance 3. If the pdf of X is given by f(x) =
3
—J ˜
, if a < x < b and zero otherwise. Find a and b

135
18. If f(x) = in 1/x, 0 < x < 1 and 0, otherwise show that f(x) is a pdf

¢ ik ' ∞ U ( U £ n
19. Let X be a random variable with zero expectation and probability density function
given by f(x) = ¦
0 i‚ƒi„…i†i
Find α.

20. Let f(x) = ½ (k) = ½ (k-1)/ (1+/x/)k, k ≥ 1-∞ < x< ∞show that f is pdf

136
 UNIT 2
JOINTLY DISTRIBUTED RANDOM VARIABLES
6.1 Bivariate Distribution
In our study so far, we have considered only univariate distribution, Univariate
distribution depends on only one random variable but we are frequently interested in
specifying two or more random variables for a member of a population and interested in
probability statements. For example, the age A and the systolic blood pressure, P may be
of interest and we would consider (A,P) as a single experimental outcome. We might
study the height H and the weight W of a chosen person given rise to the outcome (h, w).
A probability distribution which depends on two random variables. Table (1) gives the
joint variables. Table (1) gives the joint probability density function of two discrete
random variables X, number of white marbles, and Y, number of blue marbles in a
sample of 3 chosen at random from a box containing 4 red, 3 white and 5 blue marbles.

Table 1
X/Y 0 1 2 3

0 1/55 3/22 2/11 1/22 21/55

1 9/110 3/11 3/22 0 2755

2 3/55 3/44 0 0 27/220

3 1/220 0 0 0 1/220

35/220 21/44 7/22 1/22 1

We denote the probability that X takes the value of x and Y the value y by f(x, y) where
f(x, y) = P(X = x, Y = y).

For example, the probability that the sample contains 1 white and 2 blue marbles is
denoted by f(1, 2). From the table,
137
 f(1, 2) = 3/22
This is calculated thus:

P(X = 1, Y = 2) =

The reader should verify the entries in table (1).

Note:
(i) ∑∑ f(x, y) = 1.
(ii) 0 ≤ f(x, y) ≤ 1 for all (x, y).
(iii) the row totals form what is called the marginal distribution of X denoted by fx(x)
and the column totals is called the marginal distribution of Y denoted by fy(y).
From table (1) the probability of obtaining a particular value of X irrespective of the
value of Y is given below.

Table 2
x 0 1 2 3

fx(x) 21/55 27/55 27/220 1/220

Similarly for Y, we have

y 0 1 2 3

fY(x) 35/220 21/44 7/22 1/22

If we know the joint probability density function of two random variables X and Y, then
we can compute the marginal probability density function fx(x) by summing over y and
the marginal probability density function fy(y) by summing over x i.e.
fx(x) =

fy(y) = (1)
Note: It may happen that X is discrete while Y is continuous, but in most applications it
is either both are discrete or both are continuous.

138
Definition 1
Two discrete random variables X and Y are said to be independent if their joint density
function is give by
f(x, y) = fx(x) fY(y)
i.e.
P(X = x, Y = y) = P(X = x) P(Y = y) for all pairs of x, y.

In the above example,

fx(1) =

fy(2) =

f(1, 2) =

hence X and Y are dependent (not independent) random variables.

6.2 Expectation of Sum and Product of Jointly Distributed Discrete Random

Variables
Let f(x, y) be the joint probability density function of X and Y and fx(x), fy(y), the
marginal probability densities of X and Y respectively. Then
E(X) = ∑ xfx(x)
and
E(Y) = ∑y yfy(y).
For example, consider the variables X and Y having the joint pdf given by the table (1).
The marginal pdf is given by table (2). Then

and
E(X + Y) = E(X) +E(Y) = 2.99.
Expectation of Product
The expectation of product of two discrete random variables is defined as
139
 E(XY) =

For the above example we have

( (0 ×0) × +( (0 ×1) × ( (0 ×2) × +( (0 ×3) ×

+ + +

+ + ×

= +

Variance of Sum of Two Random Variables

From the definition variance E(X2) – [E(X)]2 replacing X by X + Y we have Var(X + Y)
= E(X + Y)]2
= E(X2 + Y2 + 2XY) – [E(X)]2 – [E(Y)]2 – 2E(X)E(Y)
= E(X2) + E(Y2) + 2E(XY) – [E(X)]2- [E(Y)]2 – 2E(X)E(Y)
= E(X2) – [E(X)]2 + E(Y2) – [E(Y)]2 + 2[E(XY) – E(X)E(Y)]
= Var(X) + Var(Y) + 2[E(XY) – E(X)E(Y)]
= Var(X) + Var(Y) + 2Cov(X, Y)]
Where Cov(X<Y)= E(XY) – E(X)E(Y)

Definition 2: Covariance
The covariance of two random variables X and YU denoted by Cov(X< Y) is denoted by
Cov(X, y) = E{(X –E(X))((Y – E(Y)}
= E[XY – XE(Y) – YE(X) + E(X)E(Y) = E(XY) – E(X) E(Y)

Example 1
Given the following joint probability density function of X and Y. Find
(i) The marginal probability density function of X and Y
(ii) The expectation of X and Y
(iii) The covariance of X and Y

140
(iv) The variance of X + Y.
X/Y 1 2 3

3 .1 .2 0

4 0.2 0 .1

6 .15 .15 .1

(i) Summing along the rows we have

x 3 4 6

fX(x) .3 .3 .4

Summing along the column, we have

y 1 2 3

fY(y) 0.45 0.35 0.2

(ii) E(X) = (3 × 0.3) + (4 × 0.3) + (6 × 0.4) = 4.5

Similarly E(Y) = (1 × 0.45) + (2 × 0.35) + (3 × 0.2) = 1.75
(iii) Cov(X, y) = E(XY) – E(X)E(Y)
E(XY) = ∑∑xyf(x, y)
= ((3 × 1) × 0.1) + ((3 × 2) × 0.2) + ((4 × 1) × 0.2)
+ ((4 × 3) × 0.1) + ((6 × 1) × 0.15) + (12 × 1.5) + ((6 × 3) × 0.1) = 8.0
Hence,
Cov(X, Y) = 8.0 – (4.5)(1.75) = 0.125
(iv) E(X2) = 21.9
Var(X) = 21.9 – 20.25 = 1.65
E(Y2) = 3.65
Hence,
141
 Var(X + Y) = Var(X) + Var(Y) + 2Cov(X, Y) = 1.65 + 0.25 = 2.49.

6.3 Independent Random Variables

X and Y are said to be independent if
P(X = x, Y = y) = P(X = x) P(Y = y)
For all pairs (x. y). That is,
f(x, y) = fx(x)fy(y) for all pairs (x, y).
thus for any subset A and Bof R
P(X ε A, Y ε B) = P(X ε A) P(Y ε B).

Example 2
Two random variables X and Y have the joint probability distribution given by
X/Y 1 2

1 0.2 0.4

2 0.3 0.1

Show that X and Y are not independent.

Solution
The marginal pdf of X is
x 1 2

fX(x) 0.6 0.4

And that of Y is
Y 1 2

fy(y) 0.5 0.5

142
 P(X = 1, Y = 1) = 0.2
P(X = 1) = 0.6, P(Y = 1) = 0.5
P(X = 1) P(Y =1) = 0.30 ≠ P(X = 1, Y =1) = 0.2
Hence X and Y are not independent.
Consider two independent random variables X and Y. then
E(XY) = E(X)E(Y).
To see this, note that
E(XY) =

Thus if two random variables X and Y are independent, then

E(XY) = E(X)E(Y) (3)
That is Cov(X, Y) = 0.
Note that independent and zero covariance are not the same. That is if E(XY) = E(X)E(Y)
it does not imply that X and Y are independent.

Definition 3: Correlation Coefficient

Let X and Y be two random variable having finite non-zero variance. One measure of
degree of dependence between X and Y is the correlation coefficient denoted by p(X, Y)
and defined by

ρ = ρ (X, Y) =

it can be shown that -1 ≤ p ≤ 1.

If X and Y are independent random variables then
(i) ρ=0
(ii) Var(X + Y) = Var(X) + Var(Y)
(iii) Var(aX + bY) = a2Var(X) + b2Var(Y) where a and b are constants.
(iv) Var(X – Y) = Var(X) + Var(Y).

6.4 Conditional Probability Density Function

In chapter Two, we defined the conditional probability of an event A given event B by
143
 P(ADB) =

Similarly, this idea can be extended to random variables. Thus

P(X =x DY =y) =

Definition 4
If X and Y are two jointly distributed discrete random variables, trhe conditional
probability density function of X given Y = y is denoted and defined by
fXDy(xDy) = P(X = xDY = y) =

It follows that

fXDy(xDy) =

And
F(x, y) = fXDy(xDy)fy(y).
Example 3
Let the joint probability density function of X and Y be given by the table below.
X/Y 1 2 3

3 .1 .2 0

4 .2 0 .1

6 .15 .15 .1

Compute the conditional probability density function of X given that Y = 2.

Solution
From the joint p.d.f. table we have using the formula
fXDy(xDy) =

fXDy=2(3D2) =

fXDy=2(4D2) =

144
 fXDy=2(6D2) =

Hence, the conditional pdf of X given Y = 2 is

X 3 4 6

fXDy=2(xD2) 0

It is left as an exercise to show that the conditional pdf of X given that Y= 1,3 are

1. fXD1(xD1)

x 3 4 6

2 4 3
9
fXD1(xD1)
9 9

2. fXD3(xD3)

X 3 4 6

fXD3(xD3) 0

Definition 5: Conditional Expectation

The conditional expectation of X given Y = y is defined by
E(XDy) = ∑xfxDy(xDy)
Where fXDy(xDy) is the conditional probability density function of X given Y = y.

Definition 6: Conditional Variance

The conditional variance of X given y is defined by
Var(XDy) = E(X2Dy) – [E(XDy)]2

Example 4
Calculate the conditional expectation and conditional variance of X given that Y = 2
where the joint pdf of X and Y is given in Example 3. From example 3, we have
145
 E(XDY = 2) =

E(X2DY = 2) =

Hence condition variance of X given that Y = 2 is

Var(XDY = 2) = E(X2DY = 2) – [E(XDY = 2)]2 =

Exercise 6.1
1. Two discrete variables X and Y have the joint probability distribution given
below.

X/Y -1 5 7

0 0.1 0.2 K

1 0.05 0 0.15

2 0.3 0.1 0

(a) Find the value of k, (b) Find the marginal distribution of X and Y and

(c) Are X and Y independent?

2. Given the following joint probability density function of random variables X and
Y.

X/Y 1 2 3

1 1/8 1/8 ¼

2 3/8 0 0

3 0 1/8 0

Find

146
 1. The marginal probability density function of X and Y and hence E(X) and
E(Y).

2. Are X, Y independent?
3. Find the conditional pdf of Y given that X = x for which E(YDx) is defined.

3. The joint probability distribution of two random variables X and Y is given by

X/Y -1 0 3

2 .1 0 .1

3 .2 .1 0

4 0 .3 .2

(a) Determine whether or not X and Y are independent.

(b) Calculate the correlation coefficient, P(X, Y).
(c) Find the variance of X + Y.

4. Suppose a box has 3 balls labeled 1, 2, 3. Two balls are drawn at random one after
the other without replacement. Let X and Y denote the number on the first and
second balls drawn respectively

(a) Find the joint probability distribution of X and Y

(b) Determine whether or not X and Y are independent.

5. Suppose the situation is as an exercise 4, except now that the two balls are selected
with replacement.

(a) Find the joint probability distribution of X and Y

(b) Determine whether or not X and Y are independent.

(c) Find the Var(X + Y).

147
6. Let X and Y be two random variables having the joint probability density function
given by the following table.

X/Y -3 -1 0 2 5

-1 1/20 1/10 0 0 3/20

3 1/40 1/40 1/20 1/20 0

4 3/40 1/40 1/40 1/20 1/20

5 1/20 1/40 1/10 1/20 1/10

Compute the following probabilities

(i) X is even (ii) Y is odd

(ii) X + Y is even (iv) XY is odd

(v) X > 0 and Y > 0 (vi) X is even given that Y is even

(vii) X > 0 given that Y is ≤ 0.

6.5 Multinomial Distribution

Multinomial distribution is a generalization of the binomial distribution in which each of

n independent identical experiments can result in any of k possible outcomes with
probabilities
p1, p2,…pk, ∑pk = 1.
For k= 2 we have the binomial distribution.
Consider an experiment, such as rolling a die, that can result in only a finite number of k
distinct outcomes at let Xi be the number of the n experiments that result in outcome
number i.
Then
P(X1 = x1, X2 = x2 = … Xk = xk) = x1, x2, … xk)

148
Where c x1, x2, … xk) = number of possible sequences of outcomes for the n experiments
to yield
X1 = x1, X2 = x2 = … Xn = xn.
From our knowledge of mathematics of counting,
c x1, x2, … xk) =

can be regarded as the number of ways n objects can be partitioned into k classes such
that class contains xi objects (I = 1, 2, …k). Thus,
P(X1 = x1, X2 = x2,… Xk = xk) =

Example 5
Suppose that a fair die is rolled 12 times. Find the probability that 1 and 6 appear 2 times
each. 3 thrice each, 3 times, and 4 and 5 once each.

Solution
Let Xi be the number of times ith outcome appears.

n=12, x1 = 2, x2 = 3, x3 = 3, x4 = 1, x5 = 1, x6 = 2, pi =

Hence,

P(X1 = 2, X2 = 3, x3 = 3, x4 = 1,x5 = 1, x6 = 2)

Example 6

In a certain large population, 70% are right-handed, 20% left handed and 10% are
ambidextrous. If 10 persons are chosen at random from the population, what is the
probability that (i) all are right-handed? (ii) 7 are right-handed, 2 are left handed and 1 is
ambidextrous.

Solution
149
 P1 = 0.7,P2 = 0.2,P3 = 0.1

Let X1, X2, X3 denote number that are right-handed and ambidextrous respectively, then

P(X1 = x1, X2 = x2 = X3 = x3) =

(i) P(X1 = 10, X2 = 0 = X3 = 0) = (0.7)10(0.2)0(0.1)0 = (0.7)10 = 0.028

(ii) P(X1 = 7, X2 = 2 = X3 = 1) = (0.7)7(0.2)2(0.1)1 = 0.0119.

6.6 Continuous Random Variables

In this section we will consider a pair of continuous random variables X and Y and some
of their properties. The results in the earlier section, which have been proved for discrete
random variables are also applicable to continuous random variables. The joint
probability density function of X and Y is denoted by f(x, y) such that

and

f(x,y) ≥ for all (x, y) ε R

R

(that is, summation in discrete case is replaced by integrals) and the marginal probability
density functions are given by

fX(x) =

fY(y) =

The joint probability distribution function F is defined by

F(x, y) = P(X ≤ x, Y ≤ y)

Where X, Y are defined on the same probability space.

F(x, y) =

150
If B is in the range space of (X, Y)

P(X, ε < = B) = P(B) =

and

The marginal probability distribution function are defined yb

Fx(x) = P(X ≤ x) and Fy(y) = P(Y ≤ y).

From the joint probability distribution function we can get the marginal distribution
functions by the following relationship

FX(x) = F(x, ∞) =

FY(y) = F(∞, y) =

Let X and Y be any continuous random variables. Then X and Y are independent if and
only if

f(x, y) = fX(x)fY(y), -∞< x <∞, -∞< y <∞

The conditional probability density function of X is defined by

Similarly, the conditional probability function of Y given X is

The conditional expectation is defined by

E(YDX=x) =

151
Example 7

Let X and Y have the joint pdf

2 0 <x < y < 1

f(x, y) =
0 elsewhere
Find
(i) The marginal probability density functions.
(ii) Determine whether or not X and Y are independent
(iii) Find the conditional expectation of X given Y = y.

Solution
The marginal probability density function is given by
(i) fX(x) =
fX(x) =
and
fY(y) =
=
Hence

2(1 – x) 0<x<1
fx(x) =
0 elsewhere
and
2y 0<y <1
fy(y) =
0 elsewhere

(ii) fx(x)fY(y) = 2(1 - x) 2y = 4y(1-x) ≠ f(x, y)

So X and Y are not independent.

(iii) E(XDy) =

=
Example 8
Let the random variables X and Y have the joint pdf.

x+y 0 < x < 1, 0 < y < 1

152
 f(x, y) =
0 elsewhere
Find (i) the marginal probability density functions, (ii) are X, Y independent? And (iii)
find the Cov(x, Y).

Solution

(i) fX(x) =
=

fY(y) =
=
Hence,
+x 0<x<1
fX(x) =
0 elsewhere

+y 0 <y< 1
fY(y) =
0 elsewhere

(ii) f(x, y) ≠ fX(x)fY(x)). Thus, X and Y are not independent.

(iii) E(X) =
E(Y) =
E(Y) =

Hence
E(XY) = 1/3
153
Thus,
Cov(X, Y) = 1/3 – 7/12.7/12. = -1/144.
Example 9
Let the random variables X and Y have the joint pdf
0 <y< x < 1
f(x, y) =
0 elsewhere
Find the marginal p.d.f of X and of Y and the conditional pdf of Y given X = x.

Solution

Therefore
1, 0<x<1
fX(x) =
0 elsewhere

FY(y) =
Therefore

-log y 0 <y< 1
fy(y) =
0 elsewhere

0 <y< x < 1
fYDx(yDx) = =
0
Hence X and Y/x are uniformly distributed over (0, 1), (0, x) respectively.

Example 10
Let the random variables X and Y have the joint pdf

x2+ 0 ≤x ≤ 1, 0 ≤ y ≤ 2
f(x, y) =
0 elsewhere
Find the P(X + Y ≤ 2).

154
 Solution
Let A ≡{x, y: x + y ≤ 2}.
P(A) =

=
Exercise 6.2
1. Suppose that X and Y have the joint p.d.f
K e-λ(x+y)x≥ 0, y ≥ 0
f(x, y) =
0 elsewhere
Find (i) K and (ii) the marginal pdf of X and of Y.
2. Suppose that X and Y have the joint pdf
C(x-y) 0 <x <2
f(x, y) = -x< y < x.
0 elsewhere

(a) Find (i) c and (ii) the marginal pdf of Xand of Y.

(b) Compute the following (i) P(x< 1) and (ii) P(0 < Y < 2).
3. Suppose that X and Y have the joint pdf

0 ≤x ≤2
f(x, y) =
0 elsewhere
(i) Find K using the fact that ∫ ∫ f(x, y) dxdy = 1.
(ii) Are X and Y independent random variables.
4. Two random variables X and Y have the joint pdf given by
c(x – 2xy + y)0 ≤x ≤ 1, 0 ≤ y ≤ 1
f(x, y) =
0 elsewhere
(a) Show that X and Y are uniformly distributing over [0, 1].
(b) Find the joint probability distribution function of X and Y. Hence, find the
marginal distribution functions of X and Y.
5. Suppose that X and Y have the joint pdf
λ2e-λy 0 ≤x ≤y
f(x, y) =
0 elsewhere
(i) Find the marginal probability density functions of X and of Y.
6. Suppose that X and Y have the joint pdf

f(x, y) = -∞< x <∞

0 -∞< x <∞
(i) Determine the value of c.
155
 (ii) Are X and Y independent random variables?
7. Let X and Y have the probability density function defined by
2 if (x,y) is inside the triangle having variances at
f(x, y) = (0, 0), (0,1), (1, 0)
0 elsewhere
Find the marginal probability density function of X and Y
8. Let X1/X1 + X2) (ii) E(X1IX1 + X2 = m).
9. A fair die is successively rolled. Let X and Y denote, respectively, the number of
rolls necessary to obtain a 6 and a 5. Find (a) E(X) (b) E(XIY = 1), and (c)
E(XIY = 5).
10. Let the random variables X and Y have the joint pdf

F(x, y) = 0 < x <∞; 0< y <∞. Find E(XIY = y).

11. If it is assumed that, in single – car accidents in a certain area, the probabilities of
minor severe, and fatal injuries to the driver are 0.5, 0.4 and 0.1 respectively, find
the probability that in 10 accidents there are 4 minor, 4 severe and 2 fatal injuries
to the 10 drivers.

156
Example 2
Let X have the poisson pdf with parameter λ. Find the pdf of Y = X2
; ^ / 0, 1, 2, …
~ }‰ j±
k!
f(x) =
Solution

P(Y = y) = P(X2 = y) = P(X = + `v)

Possible values of Y are: 0, 1, 4, 9, 16…

=f(`v), x = `v
(single-valued inverse function)

~ Jj j`
=¨ , v / 0, 1, 4, 9, 16 …n
√w
0 i‚ƒi„…i†i
Since X is a positive random variable), y = 0, 1, 4, 9, 16,…
Example 3
Suppose X is a random variable having probability density function f(x) given by

X -2 -1 0 1 2 3
f(x) .15 .25 0.15 .2 .15 .1

Find the p.d.f. of Y = X2

Solution
Possible values of Y are 0, 1, 4, 9.
This is not a one-to-one function
P(Y = 0) = P(X = 0) = 0.15
P(Y = 1) = P(X = -1 or 1) = P(X = -1) + P(X = 1) = 0.25 + 0.2 = 0.45
P(Y = 4) = P(X = -2 or 2) = P(X = -2) + P(X = 2) = 0.15 + 0.15 = 0.3
P(Y = 9) = P(X = 3) = 0.1
Hence, the pdf of Y is given by
y 0 1 4 9
h(y) 0.15 0.45 0.3 0.1

Example 4

3 k
f W( X / H M , ( / 1,2
Let X has the pdf given by

:
0 elsewhere
Find the pdf of Y defined by
1 if X is even
Y=
1 if X is odd
Solution

157
 P(Y = 1) = P(X = 2, 4, 6, 8,…

3 : 3 5 3 <
= P(X =2) + P(X = 4) + P(X = 6) +…
/ H M = H M = H M = …/
3
: : : >

P(Y = -1) = .
:
Similarly,
>
Hence, the pdf of Y is given by

1 2
Y 1 1

3 3
g(y)

Continuous Case
Let X be a continuous random variable having probability density function f(x). we wish
to find the probability density of a variable Y = g(X). We shall discuss methods for
finding the pdf of Y.
Let FY(y) be the probability distribution function of Y. Then, expressing FY(y) in terms of
X, we obtain
FY(y) = P(Y ≤ y) = P(g(X)) ≤ y)

= P (X ≤ g-1(y)) if Y is a strictly increasing function of X

P (X ≥ g-1(y)) if Y is a strictly increasing function of X
Where X = g-1(y) is the inverse of the transformation Y = g(X) and g(X) is a one-to-one
function.
Hence,
FY(y) = FX(g-1(y)).
Thus, the pdf of Y is given by
FY(y) = FY(y) = FX(g -1(y)).

Let X be uniformly distributed over (0, 1). Find the pdf of Y = log~ W1 ' (X, ( Y 0.
J3
Example 5
j

FY(y) = P(Y ≤ y) = P log ~ W1 ' (X L vX / NW' logW1 ' ( X L hvX

3
Solution
j
= P(log (1 – X) ≥ -λy) = P(1 – X ≥ e-λy) = P(X ≤ 1 - e-λy)
The pdf of X is given by
1 0<x<1
f(x) =
0 elsewhere
and
158
 FX(u) = pe fW( Xq( / pe 1 q( / µ
´ ´

Hence
FY(y) = (1 - e-λy)

Thus, the pdf of Y is given by

λe-λy y≥0
f(x) =
0 elsewhere
Example 6
Let X has the pdf, f(x) = e-x, x > 0. Find the pdf of Y = 2X + 1.

P(Y ≤ y) = P(2X + 1 ≤ y) = PH^ L M

wJ3
Solution
:
P(X ≤ x) = pe i Jk q( / 1 ' i Jk
k

FXH M / P H^ L M / 1 ' i JWwJ3X/: .

wJ3 wJ3
: :

FY(y) = F’y(y) = i J,WwJ3X .

On differentiating, we obtain
3 -

:
Hence,
i J,WwJ3X , v m 1
3 -

:
fY(y) =
0 elsewhere
Example 7

λ( jJ3 0≤x≤1
let X has the density function given by

fx(x) =
0 elsewhere
Show that Y = - loge X is an exponential random variable.

Solution
µ
P(X ≤ u) = pe h( jJ3 q( / ( j | / µ j
P(Y ≤ y) = P(-loge X ≤ y) = P(X ≥ e-y) = 1 – P(X ≤ e-y)
´
0
thus,
P(X < e-y) = e-λy
P(Y ≤ y) = 1 – e-λy; fY(y) = λe-λy, y ≥ 0
Hence,
FY(y) = P(Y ≤ y) = 1- e-λy,
λe-λy , y≥0
Fy(y) = F’Y(y) =
0 elsewhere
This shows that Y is an exponential random variable.

Example 8
159
Let X be a random variable having pdf
2x 0≤x≤1
fx(x) =
0 elsewhere
Find the pdf of Y = 4x2.
Solution
w
FY(y) = P(Y ≤ y) = P(4X ≤ y) = PH^ : ≤ M
2
5
w J w ‹√w ‹√w J√w
= PH^ :
≤ M / PH √ ≤ ^ ≤ M= F XH M - FXH M
5 : : : :
´ ´ µ
FX(u) = pe fW(Xq( / pe 2( q( / ( : |0 / µ :

Thus,
:
w J√w
FY(y) = FXH√ M ; FXH M /0
: :

since X cannot be less than zero. Hence,

:
w w
FY(y) = H√ M / .
: 5
Thus, the pdf of Y is
3
0<Y<4
5
F’Y(y) = fY(y) =
0 elsewhere
that is, Y is uniformly distributed over (0, 4).

Example 9
Let X be a continuous random variable having probability density function f(x). derive an
expression for the probability density function of the random variable Y = X2.
Solution
FY(y) = P(Y ≤ y) = P(X ≤ y) = W'`v ≤ ^ ≤ = `vX = FX(+`v) - FXW'`v).
2

Differentiation gives
3 J3
F’X(`v) = fXW`vX, F’XW'`v) = fXW'`vX.
:√w :√w
Hence
3
(fXr`vs + fXr'`vs; y > 0
:√w
F’Y(y) = fY(y) =
0 elsewhere

Example 10
Lex X be a random variable having the pdf.
3
(x + 1); -1 ≤ x < 1
:
f(x) =
160
 0 elsewhere
2
Find the pdf of Y = X .
Solution
From example 10above, we have
3
(fXW`vX + fXW'`v)
:√w
fY (y) =
fX(`v) = W`v = 1X
3
:

W1X / \ : v 0 U v U 1 n
3 -
hence,

fI Wv X / ¦ r`v = 1s = W'`v = 1X¸ /

3 3 3 , 3

i‚ƒi„…i†i
:√w : : :√w
0

Note: The transformation Y = X2 mapped x:-1 ≤ x < 1 onto the set y: 0 < y < 1.

Example 11
Let X be a continuous random variable having pdf f(x). Derive the formula for the pdf of
Y = |X|
FY(y) = P(Y ≤ y) = P(|X| ≤ y) = P(-y < X < y) = FX(y) – FX(-y)
By differentiation we see that

Wfk WvX = Wfk W'v X v Y 0n

FY(y) = g
0 v ≤0

Theorem 1
Let Y = g(X) be a differentiable strictly increasing or strictly decreasing function on an
interval D and let D * denote the range of g. let X e a continuous random variable with
pdf f(x) such that f(x) = 0 for x ϵ D. Then the pdf of Y is given by
fY(y) = fX(g-1(y)) ¹ ¹ ; v » ¼*
ۼ}- WwX
€w

fY(y) = f(x) ¹ ¹
-1

€k
where x = g (y) is the inverse function of g. equivalently

€w
where x is expressed in terms of y, y ϵ D* and x = g -1(y).

Proof:
Case 1: suppose that g(X) is strictly increasing
FY(y) = P(Y ≤ y) = P(g(X) ≤ y) = P(X ≤ g -1(y)) = FX(g-1(y))
€
Thus by differentiation, we have
€w
F’Y(y) = fX(g-1(y)) (g-1(y)).
Since g(X) is strictly increasing so also is g-1(y).

Ž½J3 Wv XX / ¹ Ž½J3 WvX¹.

€ €
Hence
€w €w
¾½J3 WvXs Y 0. Thus
€
€w
Since

161
 Ž½J3 WvX¹ / fI W( X ¹ ¹ ; ( / ½J3 WvX.
€ €k
FY(y) = fX(g -1(y))¹
€w €w

Case II: When g(X) is a strictly decreasing function of X

FY(y) = P(Y ≤ y) = P(g(X) ≤ y) = P(X ≥ g-1(y))
= 1 – P(X ≤ g-1(y)) = 1 – FX(g-1)y)).
Thus,
F’Y(y) = fY(y) = -fX(g-1)y)).

Therefore, g-1(y) is negative and g-1(y) = ¹ g ' 1WyX¹.

Since g(X) is strictly decreasing so also is g-1(y).
€ € €
€w €w €w

fy(y) = fX(g-1(y)) ¹ g ' 1WyX¹ = fX(x) ¹ ¹; x = g -1(y)

€ €k
Thus,
€w €w

Example 12
Let X be a random variable having the pdf
1 0<x<1
f(x) =
0 elsewhere
Find the pdf of Y = -2 loge x.

Solution

fY(y) = fX(x) ¹ ¹; x = e-y/2

€k
Y is a strictly decreasing differentiable function of X. hence by Theorem 1
€w
Y = -2loge x,

/ ' e-y/2,¹ ¹ / e-y/2

€k 3 €k 3
then

€w : €w :

3
Hence
:
fY(y) = 1, e-y/2, y ≥ 0.

3 -y/2
Thus
:
e y>0
fY(y) =
0 elsewhere
Example 13
Let X be a random variable with pdf
2x 0≤x≤1
fX(x) =
0 elsewhere
2
Find the pdf of (i) Y = 2X – 5 (ii) Y = X .

Solution
(i) Y is a strictly increasing function of X. thus,
162
 f¿ WvX / fI W( X ,( /
€k w‹4
€w :
/
€k 3
€w :

f¿ WvX / 2(, / ( /
3 w‹4
Thus,
: :

' 5 ≤ v ≤ '3
w‹4
Hence
:
fY(y) =
0 elsewhere

q( 1 -
(ii) Y is strictly increasing on (0, 1) and 0 ≤ Y ≤ 1.
^ / = `v ])q / v J,
qv 2

f¿ WvX / fI W( X
€k
thus,
€w

f¿ WvX / 2( ¹ v ¹ / 2. v . v J, , ( / `v / 1.
3 - - 3 -
J J
, ,
: :

Hence,
1 0≤y≤1
fY(y) =
0 elsewhere

Let X be a random variable having the N(µ, σ2). Find the pdf of (i) O /
IJ À
Example 14:Chi-square Distribution
Á
(ii) Z = Y2.

O/ / Â
IJ À €k
Solution
Á €w
(i) , X = µ + Yσ,

fY(y) = fX(x) ¹ ¹; x µ + yσ
€k
Y is a strictly increasing function of X, hence
€w

fI W( X / i J, .
3 - Wk J ÀX,

Á √:¤ Á,

fI Wμ = vÂX / i J,w
- ,
Substituting for x, we see that
3
Á √:¤

f¿ WvX / i J,w . Â / i J,w , '∞ U v U ∞.

- , - ,
3 3
Thus

Á √:¤ Á√:¤
That is Y is a standard normal random variable (N(0, 1)).

(ii) Z = Y2

163
 = f¿ WvX ¹ ¹
0 €w €w 0
fZ(z) = fY(y) ¹ ¹
€Ä '∞ €Ä '∞

v / Å,, / Å ,
- €w 3 -
ۀ :

fÄ WÅX / i J,Ä . Æ , = i J,Ä . Æ ,

3 - 3 - 3 - 3 -
Hence

√:¤ : √:¤ :

- - -

0U Å U ∞n
- - } É
H M, È , ~ ,
i / Ç
} -
Ä , J Ä ,
,
√:¤ √¤
i‚ƒi„…i†i
=
0

Definition 3

A continuous random variable X that has the pdf

( , J 3 i J, n , 0 U ( U ∞
-
H M,
fW( X / Ç
| Š
,
|
H M
,

0 elsewhere

is said to have a chi square distribution with n degrees of freedom and is written ^ : W0X .

Let X be a random variable having the N(µ, σ2) distribution.

IJµ :
Then the pdf of O / H M lƒ^ : W3X.
Á

Proof: See Example 14.

The pdf of Y = g(X) when g is not one to one.

Theorem 2

Let X be a continuous random variable having pdf f(x) such that f(x) > 0, x » D and f(x) = 0, x ∉
D. Suppose D can be partitioned into two sets D1 and D2 such that the transformation Y = g(x)
which is a monotone differentiable function in D and is a strictly increasing function on D1 and
strictly decreasing differentiable function of X on D2, then the pdf of Y is given by

f¿ WvX / fI W( X ¹€w¹ = fI W( X n€w¹

€k €k
Ë0Ì- Ë0Ì,

Where x = g-1(y), D = D1∪ D2

164
Consider the following examples.

Example 15

Let X be a continuous random variable having pdf f(x) given by

i J,k ' ∞ U ( U ∞
- ,
3
fW( X /
√:¤

Find the pdf of (i) Y = |^| (ii) Y = X2 .

165
(i) The graph of Y is given below.

Figure 1

Y is strictly increasing on (0, ∞) = D1 and strictly decreasing on (-∞, 0) = D2

/ 1.
€k
€w
On D1, y = x,

fI WvX ¹ ¹ / i J,w
- ,
€k 3
€w √:¤

/ 1.
€k
€w
On D2, y = x,

fI W'vX ¹ ¹ / i J,w .
€k 3 - ,

€w √:¤

Thus,

f¿ WvX / fI WvX ¹ ¹ = fI W'vX ¹ ¹ / i J,w = i J ,w ,

€k €k 3 - ,
3 - ,

€w Ì €w Ì √:¤ √:¤
- ,

Hence

(ii) The function Y = X2 is a strictly decreasing differentiable function of X on (-∞, 0) and is

= (-∞, 0) and D2 = (0, ∞) so that D = D1∪ D2.

a strictly increasing function of X on (0, ∞). Thus D = (-∞, ∞) can be partitioned in to D1

Thus, by Theorem 2, we have

f¿ WvX / fI W( X ¹ ¹ = fI W( X ¹ ¹
€k €k
€w Ë0Ì €w Ë0Ì
- ,

166
 1
i J ,k
- ,
f W(X /
√2{

( / E`v

on D1, x = '`v])q ¹€w¹ / v J,

€k 3 -

fI r'`vs ¹€w¹ / i J,w . : v J,

€k 3 -
3 -

√ :¤

On D2, x = + `v])q

fI r`vs ¹ ¹ / i J,w . v J,
€k 3 -
3 -

€w √:¤ :

Hence
- -

vm0
} 
w,~ ,

f¿ WvX / i J,w . v , = i J,w . v , / Î n

- - - -
3 3 3 3 √:¤
√:¤ : √:¤ :

0 i‚ƒi„ i†i
Example 16

Let X be a uniform random variable over the interval H , M. Find the probability density
J¤ ¤
: :
function of Y = tan X.

Solution

1 '{ {
fk W^ X / ¨{ 2 U ( U 2n

0 elsewhere

Y = tan X is a strictly increasing function of X in the interval H , M Ïi)§i,

–¤ ¤
: :

f¿ WvX / fW( X ¹ ¹ ; ( / ])J3 v

€k
€w

.
3 3
¤ 3 ‹ w,

Thus, the pdf of Y is given by

167
 3
f¿ WvX /
¤W3 ‹ w , X
-∞< y < ∞

This pdf is called the Cauch pdf.

Exercise 7.1

1. Let X be a uniform random variable over the interval (-1, 1).

(i) Let Y = 2X – 1. Find the pdf of Y.

Find the pdf of (a) Æ / |^|WaXÐ / sin H M ^

¤
:
(ii)

2. Let X has the pdf given by

, ( / 1, 2, … ÒWiÓi)Xn
:Š
fW( X / \:ь-J :

0 elsewhere

Find the pdf of Y defined below.

2 if X is even

Y=

5 is X is odd

Hence, find the mean and variance of Y.

3. Let X has pdf

fW( X / g2(i
Jk ,,
0 U ( U ∞n

0 elsewhere

Find the pdf of (i) Y = X2 (ii) Y = |^|.

4. Suppose that X is a continuous random variable defined over (0, 1) such that P(X ≤ 0.45)
= 0. Let Y = 1 – X. Find the value of K such that P(Y ≤ k) = 0.40.

Suppose that X has the pdf f(x) = e-x, x > 0. Find the pdf of the following random
(b) Å / .
3
5.

WI‹3X,
variables, (a) Y = X3

6. Suppose that X is a geometric random variable with parameter p. Find the pdf of

168
 (a) Y = X2 (b) Z = X + r, where r is a constant.

7. Find the pdf of Y = ex if X has

(i) Normal distribution with mean µ and variance σ2

(ii) Exponential distribution with mean λ.

8. Let X be uniformly distributed over (0, 1).

Find the pdf of Y = Xn.

9. Let X be a continuous random variable having probability distribution function F. show

that Y = F(X) is a uniform random variable over (0, 1).

10. Let X be uniformly distributed over (0, 1). Find the pdf of

(a) Y = eX (b) Z = X2 + 1
3
(c) Ð/ I‹ 3
(d) Y = X1/β, where β ≠ 0.

11. Let X be exponentially distributed with parameter λ. Find the pdf of Y = loge X.

12. Let X be have gamma density with parameters (α, λ). Find the pdf of

(i) O / √^ (ii) Z = cX, where C > 0.

An angle θ is chosen at random from H , M. Let X = sin θ. Find the pdf of X.

–¤ ¤
: :
13.

14. Let X be exponentially distributed with parameter λ.

Let Y = n if n ≤ X < n + 1 where n is a nonnegative integer. Find the pdf of Y.

15. Let X and Y be independent random variable each having uniform pdf on (0, 1,…N).

Find the pdf of (i) min (x, Y) (ii) Max (X, Y).

7.2 Functions of Two Random Variables

The method discussed in section 7.1 of finding the pdf of a function of one random variable of
the continuous type will now be extended to functions of two random variables

Let X and Y be continuous random variables and Z = X + Y.

The problem of finding the pdf of Z is somewhat more involved. The following steps are useful.

1. Introduce a second random variable W = K(X, Y).

169
2. Obtain the joint pdf of Z and W, say d(z, w).

3. Obtain the desired pdf of Z by simply integrating d(z, w) with respect to w. that is

∫ d(z, w) dw.

The following will be assumed in obtaining the jointing pdf of Z and w.

(i) The functions h(X, Y) and k(X, Y) satisfy the following conditions.

(a) The equations

Z = h(x, y) and w = k(x, y)

may be uniquely solved for x and y in termsof z and w, ssay x = g1(z, w),

y = g2(x, w)
Ôk Ôk Ôw Ôw
, , ,
ÔÄ ÔÕ ÔÄ ÔÕ
(b) The partial derivatives exist and are continuous

From the result of section 7.1 it can be proved using results of advanced calculus that under the
assumptions (i) and (ii), the joint pdf of Z and W is given by d(z, w) = f(x, y) |Ö| where x = g1(z,
w), y = g2(z, w) and
Ôk Ôk

Ö / ×Ôw
ÔÄ ÔÕ
Ôw
×.
ÔÄ ÔÕ

The determinant J, is called the Jacobian of the transformation (x, y)-
(z, w). some of the
important random variables we shall be interested in are

X +Y, XY, X/Y, min. (X, Y), max. (X, Y), etc.

Distribution of Sum

Let Z = X + Y and W = X, the Jacobian,

0 1
Ö / ¹ ¹ / '1
1 '1
Thus d(z, w) = f(x, y) |Ö| = f(w, (z – w)),

FZ(z) = pJ∞W„, Å ' „ Xq„ / pJ∞ f W(, Å ' (Xq(

∞ ∞

If X and Y are nonnegative independent random variables, then

FZ(z) = pe fI W( Xf¿ WÅ ' ( Xq(

Ä

170
Example 17

1 0<x<1 0<y<1

Let X and Y have the joint pdf f(x, y) = 0 elsewhere

Find the pdf of Z = X + Y.

Solution

Let W = X; Z = X + Y, then x = w; y = z – w and

0 1
Ö / ¹ ¹ / '1
1 '1
Hence, d(z, w) = 1.1 = 1.

Let A be the set A = {(x, y), 0 ≤ x ≤ 1, 0 ≤ 1}. A is shown in figure below.

f(x)

x
Figure 2

Under the transformation, Z = X + Y, W = X, the boundaries of A are x = 0, x = 1, y = 0 and y =

1. The set B in the z, w –plane which is the mapping of A under the one-to-one transformation is
determined as follows:

x = 0 implies that w = 0; y = 1 implies that w = 1

y = 0 implies that z = w; y = 1 implies that z = w + 1.

171
Thus, the boundaries of B are the lines w = 0, w = 1, z = w, z = w + 1. This is show in the figure
below.
z
z =w+1

w=1
B

z =w
w

Figure 3

Hence

d(z, w) = 1

thus, (z, w) ∈ B

qWÅX / pe 1 q„ / Å
Ä
0<z<1

pÄJ3 1 q„ / 1 ' Å 1<z<2

Z 0<z≤1

d(z)= 2-z 1<z<2

0 elsewhere

Example 18

Suppose a point (x, y)is selected at random in a circle centre (0, 0) and radius 1 and that this
point is uniformly distributed over the circle. Find the pdf of

Ø / `W^ : = O : X, the distance from the origin.

The joint pdf of X and Y is

3
¤
if (x, y) lies inside or on the circle

d(z)=

0 elsewhere

172
Let W = tan-1(Y/X) then tan W = Y/X, x = r cos w, y = r sin w.

y
r

w 1
x
B

2 x
Figure 4

The Jacobian is

Ú( Ú(
Ö / ٠چ ڄ Ù ¹cos „ '† sin „
¹ / †W§Ûƒ : „ = ƒl) : „ X / †.
Úv Úv sin „ † cos „
Ú( ڄ
hence

(r, w) » B
3
¤
r

d(r, w) = f(x, y) |Ö|= 0 elsewhere

Thus,

2† 0L†L1n
d(r) = ∫ d(r, w) dw = r pe q„ / ¦
:¤ 3
¤ 0 i‚ƒi„…i†i
Example 19

3x 0<y<x

If f(x, y) = 0 < x < 1. Find the pdf of Z = X – Y.

0 elsewhere

Let W = X, A = {(x, y), 0< y < x, 0 < x < 1} and B ={(z, w) : z=x – y, w = x}

173
 y Z

A B

x W
1 1

Figure 5

The Jacobian is

1
Ö / ¹ ¹
0
'1 1
hence

3w 0<z<w

d(z, w) = 0< w<1

0 elsewhere

qWÅX / / pÄ 3 „q„
3

W1 ' Å : X
>
:
0<z<1

d(z) = 0 elsewhere

Distribution of Product

Suppose X and Y are continuous random variables having the joint pdf (x, y). show that the pdf
of Z = XY is given by

p ¹k¹ f H(, kM q(
3 Ä

Solution
Ä
Let W = X, thus, x = w, y = Õ. The Jacobian

174
 1 0
Ö / Ü JÄ
3
3Ü / Õ
Õ, Õ

Hence,

qWÅ, „ X / f H„, M |Ö|

Ä
Õ

Thus,

qWÅX / p ¹ ¹ f H„, M q„
3 Ä
Õ Õ
(2)

Example 20

Let X and Y be independent random variables with the following pdfs.

2( 0L(L1n 4v > /15 1≤v ≤2n

fW( X / ¦ , ½WvX / g
0 i‚ƒi„…i†i 0 i‚ƒi„…i†i
Find the pdf of Z = XY.

Solution

The joint pdf of X and Y is f(x. y) = f(x) g(y).

Thus, from 2.

qWÅX / p ¹ ¹ ½W„ X½ H M q„, where W = X

3 Ä
Õ Õ

Ä >
qWÅX / p ¹Õ¹ . 2„ 34 HÕM q„ / Å > p ÕA q„.
3 5 ; 3
34

Let A = {(x. y): 0< x < 1, 1 ≤ y ≤ 2}. The transformation, Z = XY, W = X gives the set B
determined as follows:

When x = 0, w = 0; when x = 1, w = 1,

When y = 1, Z = w, when y = 2, z = 2w.

175
The lines w = 0; w = 1; x = w, z = 2w are the boundaries of B as shown in the figure below.

B
1

Figure 6

Thus

pÄ/: ÕA q„ / Å, 0UÅU1
;Ä A Ä 3 5
qWÅX /
34 4

/ pÄ/: ÕA q„ / W4 ' Å : X, 1UÅU2

;Ä A 3 3 5Ä
34 34

0UÅU1
5Ä

qWÅX / Ç 5Ä
4 n
W4 ' Å :X
1UÅU2
34

Distribution of Quotients

Suppose X and Y are continuous random variables having the joint pdf f(x, y). show that the pdf
of Z = Y/X is given by

fÄ WÅX / o |(|fW(, (ÅXq( ' ∞ U Å U ∞

∞

J∞

Let W = X, then x = w, y = zw, the Jacobian is

1 0
Ö / ¹ ¹ / „.
Å „
Hence

qWÅ, „X / fW„, ńX|Ö| / fW„, ńX|„|

Thus

fÄ WÅX / p|„| fW„, ń Xq„. (3)

176
Example 21

Let X and Y be independent random variables having the respective gamma densities Γ(α1, λ)
and Γ(α2, λ). Find the probability density function of Z = Y/X.

jÝ- k Ý- }- ~ }‰Š
fI W(X /
WŸ- X
x>0

f¿ WvX /
jÝ, k Ý, }- ~ }‰Š
WŸ, X
y>0

The joint pdf of X and Y is given by

fW(, vX / fI W( Xf¿ WvX / .

jÝ-Œ Ý, k Ý- }- ~ }‰WŠŒX
WŸ- XWŸ, X

From (3) we have

fÈ WÅX / p „fW(, ń Xq„ / p „ i q„

jÝ- ŒÝ, Õ Ý- }- WÄÕXÝ, }- jWՋÕÄX
WŸ- X WŸ, X

p ( i q„
jÝ- ŒÝ, È Ý,}- [ Ÿ-Œ Ý J3 JjkW3 ‹ ÄX
,
WŸ- X WŸ, X e
=

But

pe „ Ÿ- ‹Ÿ, J3 i JjÕW3‹ ÄX q„ /
[ WŸ- ‹ Ÿ, X
WjW3‹ÄXXÝ-Œ Ý,

Hence

,
WŸ- ‹ Ÿ, X È Ý, }-
WŸ- X WŸ, X W3‹ÄXÝ- Œ Ý,
fZ(z) = z>0

Example 22

Let X and Y be independent exponential random variables with parameter λ. Find the
I
I‹ ¿
joint pdf of Z = X + Y and W = .
Solution
The joint pdf of X and Y is given by
fX,Y (x, y) = λ2 e-λ(x + y) x > 0, y > 0
The transformation gives
x = wz, y = z(1 – w)

|Ö| = z
The Jacobian of the transformation is

Åh: i JjÄ , Å Y 0, 0 U „ U 1n
fÈ,Õ WÅ, „X / g
The joint pdf of Z and W is given by

0 , i‚ƒi„…i†i
177
Example 23
3
x ≥ 1, y ≥ 1, find the joint pdf of Z = XY
k, w ,
Let X and Y have the joint pdf fX,Y(x,y) =
and Ð /
I
¿
.
Solution
v (
From the transformation z = wy, w = x/y, we have
Ö / Ü 3 JkÜ /
J:k
w
w w,
|Ö| / 2„.

fÈ,Õ WÅ, „X / fÈ,Õ W(, v X|Ö|J3 / .

3 3
Thus
È , :Õ
Å m 1, 0 U „ U Å
3
:ÕÄ ,
=
= 0, otherwise.

Let X1 and X2be random variable with joint pdf fk-, Š, , (x1, x2). Let Y1 = X1 + X2,
Example 24

Y2 = X1 - [Link] the joint pdf of Y1 and Y2 in terms of fk-, Š , (x1, x2).

,

f¿-, Þ, , (y1, y2) = fk-, Š , (x1, x2)|Ö|

Solution
,
(3 / Wv3 = v: X, (: / Wv3 ' v: X
3 3
: :
Ôk- Ôk- 3 3

Ö/ ×Ôw- × / ×:3 3× / ' :.

Ôw Ôw, : 3

'
, Ôk,
Ôw- Ôw, : :

f¿-, Þ, ,(y1, y2) / f( , 2 M

v1 = v2 v1 ' v2
H
3
Hence

: 1, (2 2
Example 25
Let X and Y be independent exponential random variables with parameters λ1 and λ2
I
respectively.
I‹ ¿
(i) find the joint pdf of U = X + Y and V =
(ii) show that U and V are independent if λ1 = λ2.
Solution

fX,Y(x, y) = λ1λ2i JWj- k‹ j, wX

The joint pdf of X and Y is given by

x = uv and y = u – uv.

fu,v(u,v) = fX,Y(x,y) |Ö|.

The joint pdf of u and v is given by

fu,v(u,v) = λ1 λ2i JWj- ´ß‹ j, ´W3JßXX |Ö| / µh3 h: i J´ßWj- J j, XJ ´j,

Substituting for x and y in terms of u and v

(ii) Putting λ1 = λ2 = λ. We obtain

fu,v(u,v) = uλ2e-λu, u ≥ 0, 0 < v < 1

178
 f´ Wµ X / pe µh: i Jj´ qÓ / µh: i j´ , µ m 0
3

fß WÓ X / pe µh: i Jj´ qµ / 1 0 U Ó U 1.
[

Example 26

f W(, v X / gi ^ Y 0, v Y 0n
JWk‹wX
The joint pdf of X and Y is given by

0 ۏ…i†„lƒi
find (i) P(X > 2, Y > 3), (ii) P(Y < X) and (iii) the pdf of Z = X/Y.

Solution

P(X > 2, Y > 3) = à1 f W(, v Xq( qv / p> p: f W(, vXq( qv

[ [
Let A = {x, y)! x > 2, y > 3), we have

¾p: i JWk‹wX q(áqv / p> i Jw ¾p: i Jk q(áqv / i J: p> i Jw qv / i J4

[ [ [ [ [
=p>

P(Y < X) = à1 f W(, vXq(qv / pe xpw i JWk‹wX q(‘ qv / pe i J:w qv / .

[ [ [ 3
(ii) Let A {(x, y); y < x}

(: = v : ≤ †: n
3
Example27
Let f(x, y) = \ ¤¡ ,
0 i‚ƒi„…i†i
find (i)fX(x) and (ii) fZ(z) where Z = √^ : = O : .

à : = ( : / :
à : ' ( : n
Solution
fk W( X / pJ[ f W(, vXqv / p , qv / â
[ 3 w
¤¡ ¤¡ ,
√† : ' ( : ¤k ,
(i)

fZ(z) = P{X2 + Y2 ≤ z2) = àk, ‹w, ãÄ , f W(, v Xq(qv / àk,‹w, ãÄ , q(qv

3
¤¡
(ii)
/ / Wƒl)§i àk, ‹w, ãÄ , q(qv / area of a circle radius z.)
¤Ä , Ä,
¤¡ , ¡,
Fz = P(z ≤ z) = N H ≤ (M / à± f W(, v Xq( qv
I
¿ ãÈ
(iii)
Þ

= pe pe i q(qv / 1 '
[ wÄ JWk‹wX 3
ċ3

fÄ WÅX / , 0 U Å U ∞.
3
Differentiating we obtain
Wċ3X,
Example 28

FZ(z) = P(X + Y ≤ z) = àk‹wäÄ f W(, vXq( qv / pJ[ pJ[ f W(, vXq( qv.
[ ÄJw
Distribution of sum. Let X and Y have the joint pdf f(x, y) and let Z = X + Y then

Fz(z) = pJ[ pJ[ f WÓ ' v, vXqÓ qv / pJ[ ¾pJ[ f WÓ ' v, vXqváqÓ.

[ ÄJw Ä Ä
Changing variable, x = v – y

FZ(z) = pJ[ fWÅ ' v, vXqv.

[
Thus

Example 29: Student’s t-Distribution

degrees of freedom. Find the pdf of Æ /

Iç
Let X be a standard normal distribution and Y a chi-square random variable with v

√¿
where X and Y are independent.
179
 and W = Y then y = w and x = Å .
I √ß √Õ
Solution
Let Æ /
√¿ √ß

1 0
Then Jacobian of the transformation is

Ö / å' Å3 √ Õ √ Õå / √w
.
:
A √Õ
}
æ , √ß

dZ,w(z,w) = fx(x)fY(y) ¹ ¹/ i J,k v ,J3 i J,

- , è 
Thus
√Õ 3 3 √é
ç
æ √:¤ H M:è/, æ
,
substituting for x and y, we have
è}-
-É, ê -É, ê ê
i . |„ i . / èi , .
3 3 è ê é Õ ,
J J3 J √ J J
, è , , è ,
√:¤ H M: ,
ç √æ √:¤H , M:,
ç
dZ,w(z,w) =
,

„ √Ð 1 √Ð
è}-
The marginal pdf of z is given by
-É, ê ê [ É,
qÄ WÅX / o o „
ê
i . q„ / i q„
,
J J
è}- J ë3‹ ì
, è
√ˆ √ˆ
, è , ,

√2{y H : M 2 √2{y H : M 2
æ | æ è
, , e

H1 = M; / H1 = M,„ /
Õ Ä, €´ 3 Ä, :´
: ß €Õ : ß í,
3‹
Let u =
ç

2 2µ qµ
è}-
[
q È WÅ X / o î ï i J´ 3
,

√2{Óy H : M 2 1= H1 = M
æ è
J3 Ä, Ä,
, e
æ : æ
1
è}-
J3
oµ i J´ qµ.
,
è}-

√2{y H : M 2, H1 = M
æ è Ä, ,
æ

pe µ, i J´ qµ / y H M.
[ è ߋ3
Note
J3
:
Hence
èŒ-
qÈ W Å X /
HM
. , '∞ U Å U ∞.
, 3
ç èŒ-
√:¤ßH , M É, ,
ë3‹ ì
ç
The above distribution, dZ(z) is called a t-distribution.

Note
(i) That a t-distribution is completely determined by the parameter v, the number of
degrees of freedom of the random variable that has the chi-square distribution.
(ii) The t-distribution is a bell-shaped distribution, symmetric about zero.
(iii) The t-distribution is flatter than the Normal distribution. As the sample size
becomes large, t-distribution approaches the standard normal distribution (n > 30
is sufficient).
(iv) The values of the t-distribution for various values of degrees of freedom have been
tabulated.
180
Example 30
Exercise 7.2

1. Let X1 and X2 be independent normal random variables N(µ 1, σ21), N(µ 2, σ22)
respectively. Find the pdf of Y = X1 + X2.
I
2. Let X and Y be independent random variables having the respective pdfs.
f(x) = e-x, x ≥ 0, g(y) = 2e-y y ≥ 0. Find the pdf ofÆ /
¿
3. Let X and Y have the joint pdf f(x, y) = λ2 e-λ(x+y), x > 0, y > 0. Find the pdf of
Z = X + Y.
4. Let X and Y be independent random variables each having the normal probability
density function N(0, σ2). Find the probability density function of
¿ ¿, ¿ |¿|
I I, |I| |I|
(i) (ii) (iii) (iv)
5. Suppose X and Y are independent normal random variables with mean 0 and
variance σ2. Find the pdf of (i) X + Y (ii) X2 + Y2 .

the joint pdf of Z = X + Y and Ð /

If X and Y are independent exponential random variables with parameter λ. Find
I
6.
I‹ ¿
. Hence find the pdf of Z and of W.

Æ / |O ' ^ |.
7. Let X and Y have uniform distribution over the interval (a, b). Find the pdf of

8. Let X be uniformly distributed over (0, 1) and let Y be uniformly distributed over
(0, x). Find
(i) The conditional pdf of Y given X = x.
(ii) The joint pdf of X and Y
(iii) The marginal pdf of Y.
9. Let f(x, y) = c(y – x)2, 0 ≤ x < y ≤ 1
= 0, elsewhere
Find the pdf of Z = X + Y.

7.3 Sum of Independent Identically Distributed RandomVariables

The Moment Generating Function Technique
Let X1, X2,….Xn be independent identically distributed random variables. In this section
we will find the pdf of sum of the X’s. The moment generating function technique will be
used to establish the pdf of sum Bernoulli, Poisson, exponential, Normal, Gamma random
181
variables. All the probability distributions above except Bernoulli and exponential has the
following remarkable and very useful property called Reproductive properties.

Definition: Reproductive Property

A probability distribution F is said to have reproductive property if when two or more
independent random variables having the probability distribution F are added, the
resulting random variable has the distribution F. we shall establish this result using
moment generating function defined in Chapter 4, section 4.3.

Moment Generating Function

The moment generating function MX(t) of a random variable X is defined by
MX(t) = E(etX)
The domain of Mx(t) is all numbers t such that etX has finite expectation.

Theorem 3
Suppose that a random variable X has mgf Mx(t). Let Y = aX + b. Then, the mgf of the
random variable Y is given by MY(t) = ebt Mx(at).
Proof:
MY(t) = E(ety) = E(e(aX + b)t) = E(eatX + bt) = E(eatX ebt) = ebt E(eatX) = ebt MX(at).

Theorem 4
Let X and Y be two random variables with mgf’s Mx(t) and MY(t), respectively. If Mx(t)
= MY(t) for all values of t, then X and Y have the probability distribution. That is, the
mgf uniquely determines the probability distribution of the random variable. The proof
of this theorem is outside the scope of this work, the reader is referred to volume II of this
book.

This theorem will now be used to show that a linear function of a random variable also
has a normal distribution.

Example 31
Suppose that X has distribution N(µ, σ2).
Let Y = aX + b. show that Y also has a Normal distribution with mean aµ + b and
variance a2 σ2. The mgf of Y is
MY(t) = E(ety) = ebt(MX(at))
from theorem I.

1 1 [
MX(t) = E(etx)
-/,WŠ}ðX, -/,WŠ}ðX,
i q( / o i i q(
[
J k J
o i / , ñ,
 2{  2{
k
√ √
ñ
J[
Let ƒ / , then MX(t) becomes
J[
kJ À
Á
1 [
i À [ J3/:ó, ‹Áò
™I WX / o i WÁò‹ ÀX
i J3/:ò
Âqƒ / o i qƒ
,

Â√2{ J[ √2{ J[

182
 õ÷

pJ[ i J3/:ó ‹Áò qƒ.

ôõð ‹ ö÷ [ ,
÷
√÷ø
Making the change of variable u = s – σt, we see that
ñ, › ,
Mx(t) = i À‹ ,Á p i J´ qµ / μ = i
- , ,
 3 [ , ´‹
,
√ :¤ J[

MX(at) = i ˜À‹ Á ˜  /:
, , ,
Substituting at for t, we have

 : ]:  :
Hence,
ñ, œ,›,
W X
™¿  / i –i — ˜À
= ù/i W—‹˜ÀX‹
/ i ú‹ Á  ˜ /:
, , ,

2
,

Where γ = b + µa = E(Y), Vara = E(Y), Var(Y) = a2σ2.

This is the mgf of a normal variable with mean γ = aµ + b and variance a2σ2. Thus, by
theorem 2, Y is a normal random variable with mean aµ + b and variance a2σ2.

Let X be N(µ, σ2). Show that O /

IJ À
Example 32
Á
is N(0, 1).

O/ = ' .
IJ À I À
Solution
Á Á Á
Let a = and b = ' , then Y = aX + b.
3 À
Á Á
From Example 312, Y is a normal random variable with mean aµ + b and variance a2σ2.

aµ + b = ' / 0
À À
Substituting for a and b, we see that
Á Á
(i)
aσ = ,/1
2 2 Á,
Á
(ii)
Thus Y is standard normal (N(0, 1)) random variable.

Theorem 5
Let X and Y be independent random variables and Z = X + Y.
Let MX(t), MY(t) and MZ(t) be the mgfs of the random variables X, Y and Z respectively.
Then
MZ(t) = MX(t) MY(t) (4)

Proof:
By definition,
MZ(t) = E(eZt) = E(e(X + Y)t) = E(eXt+eYt) = E(e+Xt) E(eYt) = MX(t) MY(t).

In general, if X1, X2,…, Xn are independent random variables.

™È WX / ™I- WX™I, WX … ™I| WX

The mfg of Z = X1 + X2 + …+ Xn is

183
If X1, X2,..…, Xn are independent identically distributed with mgf MX(t), then
MZ(t) = [MX(t)]n.

Example 33: Sum of Binomial Random Variables

Let X and Y be independent binomial random variables with parameters (n1, p), (n2, p)

i I H 3 M _ I W1 ' _X0- Jk / Ž_i  = W1 ' _X0, .

)
respectively. Find the pdf of Z = X + Y.
MX(t) = E(etX) = ∑k©e
0-
(

MZ(t) = MX(t) MY(t) =Ž_i  = W1 ' _X0- ‹0,

The mgf of Z is given by (from Theorem 5).

but this is the mgf of a binomially distributed random variable with parameters (n1 + n2,
P).
0- ‹0,
ûÄ _ È W1 ' _X0- ‹0, JÄ Å / 0, 1, 2, … )3 = ): n
fÈ WÅX / g
thus, by theorem 2, Z is a binomial random variable and its pdf is given y

0 i‚ƒi„…i†i
In general,if X1, X2,…XK are independent binomial random variables with parameters
(n1, p), (n2, p),…(nk, p) respectively, then the moment generating function of Z = X1,

MZ(t) = Ž_i  = W1 ' _X0- ‹0,‹‹ 0“ .

X2,..…, Xk is given by

Thus Z is a binomial variable with parameters (n, p) where n = n1 + n2 … nk.

Example 34: Sum of normal random variables

Suppose that X and Y are independent random variables with distribution N(µ 1, σ21) and
N(µ 2, σ22) respectively. Show that Z = X + Y is a normal random variable with mean µ 1 +

MX(t) = i À-  ‹ 3/:Á -
µ 2 and variance σ21 + σ22
, ,

MY(t) = i À,  ‹ 3/:Á ,
, ,

MZ(t) = MX(t)MX(t) = i À-  ‹ 3/:Á - i À, ‹,Á ,

- ,
Hence
, , ,

= i WÀ- ‹ À, X‹ 3/: rÁ -‹ Á ,s / i À‹3/: Á

, , , , ,

where µ = µ 1 + µ 2, σ2 = σ21 + σ22. Since i À‹3/: Á is the mgf of N(µ, σ2), hence by
, ,

Theorem 4, z is a normal random variablewith mean µ = µ 1 + µ 2 and variance

σ2 = σ21 + σ22.
In general, if X1, X2,..…, Xn are independent N(µ 1, σ21) then
Z = X1 + X2 +.…+ Xk
is N(µ, σ2) where µ = µ 1 + µ 2 +…+ µ n, σ2 = σ21 + σ22 +…+ σ2k

Example 35: Sum of Poisson Random Variables.

Suppose that X and Y are independent poisson random variable with mean λ1 and λ2
respectively. Show that Z = X + Y is a Poisson random with mean λ1 + λ2
™I WX / ü Wi I X / ∑[
k©e i / i Jj- r3J~ s .
I ~ }‰- jŠ - ›
k!

MY(t) = i Jj- r3J~ s .

›
Similarly,

184
 MZ(t) = i Jj- r3J~ s i Jj, r3J~ s / i JW3J~ XWj- ‹ j, X / i JW3J~ Xj , h / h3 = h:
Hence,
› › › ›

The mgf of Poisson distribution with parameter λ is i JjW3J~ X , thus Z is a poisson random
›

variable with mean (parameter) λ = λ1 + λ2. In general, if X1, X2,…,Xn are independent
P(λ1), then
Z = X1 + X2 +… + Xn
is a Poisson random variable with parameter λ = λ1 + λ2 +…+ λn.

Example 36
Let X and Y be independent gamma random variable with parameters (n1, λ) and (n2, λ)
respectively. Show that Z = X + Y is a gamma random variable withparameter (n1 + n2,
λ).

[
h (3 i
0- 0J3 Jjk [
W(X0- J3 i JkWjJX
Solution
™I WX / ü Wi I X / o i k q( / o h0- q(
J[ yW)3 X e y W)3 X
h0- [
h0- [
/ o ( 0J3 i JkWjJX q( / o µ0- J3 i J´ qµ
yW)3 X e Wh ' X0-J3 yW)3 X e
But pe µ0- J3 i J´ qµ / y W)X.
[

0-
MX(t) = x ‘ .
Hence,
j
jJ

0,
MY(t) = x ‘ .
j
jJ
Similarly

0- 0, 0- ‹0, 0
MZ(t) = x ‘ . x ‘ / x ‘ / x ‘ . ) / )3 = ): .
Thus,
j j j j
jJ jJ jJ jJ
This show that Z is a gamma random variable with parameters (n1 + n2, λ).
In general, if X1, X2 ,… + Xk are independent gamma random variables with parameters
(n1, λ), I = 1,2,…k. then Z = X1 + X2 +… + Xk has a gamma distribution with parameters
(n, λ) where n = n1 + n2 +…+ nk.

Example 37
Suppose that Xi has χ2 (n1) (chi-square distribution with ni degrees of freedom) I = 1,
2,…k where the Xi’s are independent random variables.

Let Z = X1 + X2 +… + Xk. Show that Z has distribution χ2n where n = n1 + n2 +…+ nk.
Recall (5.9) that chi-square distribution is a special case of the gamma distribution in
0
which λ = 1/2 and n replaced by n/2. Since the mgf of gamma (n, λ) is given by
x ‘ ;
j
jJ
substituting for λ = 1/2 and replacing n by n/2 we have
0/: 0/:
MX(t) = x ‘ / x ‘
3/: 3
3/:J 3J:
.
The mgf of Xi is therefore given by
185
 0/:
MX(t) / x ‘
3
3J:
I = 1, 2,…

|-Œ|, ŒŒ|“
Hence, the mgf of Z is
|

MZ(t) = ™I- WX™I, WX … ™I“ WX / x ‘ /x ‘ .

3 , 3 ,
3J: 3J:
where n = n1 + n2 +…+ nk which is the mgf of χ n distribution.
2

Example 38

σ2. Show that (i) E(^ý ) = µ, where ^ý = (X1 + X2 +…+ Xn) (ii) Var(^ý ) = and
3 Á,
Let X1, X2,… Xn be independent identically normal variables with mean µ and variance

0 0
O/
∑|
- WI- J ÀX
,

Á,
2
(iii) has a χ (n).

E(^ý ) = ü H ∑03 ^þ M / ∑03 üW^þ X / . )μ / μ

3 3 3
0 0 0
Var(^ý ) = VarH ∑03 ^þ M /
(i)
∑ ˆ]†W^þ X / ) : /
3 3 3 Á,
0 0, 0, 0

O/ / = = =
∑|
(ii)
- WI- JÀX WI- JÀX, WI, JÀX, WI| JÀX,
,

Á, Á, Á, Á,
I- JÀ : I, JÀ : I| JÀ :
(iii)

/ H M = H M == H M /Y12 + Y22 +…+ Yn2.

Á Á Á
From Example 14, we know that Y12 has χ2(1) and from Example 30, since Y12 are
independent χ2(1) we have that Y is χ2(n).

Example 39
Let X1, X2,…Xn be independent identical normal random variables with mean µ and
variance σ2
Show that
: / - ,-
∑| WI JIýX,
Á
has a χ2(n-1).

X1 - µ = Xi - ^ý + ^ý - µ
We shall proceed as follows:

(X1 - µ)2 = (X1 - ^ý )2 + (^ý - µ)2 + 2(Xi - ^ý )(^ý - µ)

ý : ∑0  ∑0 W ý XW ý X
∑0þ©3WX þ ' μX: / ∑0þ©3 WXþ ' ^ X = þ©3WX ' μX = 2 þ©3 X þ ' ^ ^ ' μ
:
Summing, we have

∑0þ©3WX þ ' ^ý XW^ý ' μX / W^ý ' μX ∑0þ©3WX þ ' ^ý X\

since ∑W^þ ' ^ý X / 0, „i ƒii …]
WX ' ^ý X: = ∑0þ©3WX
 ' μX:
∑0þ©3WX þ ' μX: / ∑0þ©3 þ

dividing through by σ2, we have

/ =
∑|
-W J ÀX
, ∑WI- JIý X ∑|
-W J ÀX
 ,

Á, Á, Á,

186
 W J ÀX,
From Example 37, we have ∑0þ©3
Á,
is χ2(n) and (it is left as an exercise to show that
∑W J ÀX,
Á,
is χ2(1).
∑W J ÀX, ∑W J ÀX,
Á, Á,
Let Z = and Y =
∑W J 
 X,
Á,
2
S =
 and S are stochastically independent, we have
Since X 2

E(etZ) = E(et(Y + S2) = E(etY)E(ets2)

0/:
but MZ(t) = E(etZ) / x ‘
3
3J:
, since Z is χ(n),
/:
MY(t) = E(etY) / x ‘
3
3J:
(Y is Z2(1))

|}-
hence,

E(e ) / x ‘
3 ,
3J:
tS 2

and this is the mgf of a chi-square distribution with n – 1 degrees of freedom. This shows
that S2 has a chi-square distribution with n – 1 degrees of freedom.

Exercise 7.3

which is N(µ, σ2). Show that ^ý / - , is NHμ, M.

I ‹ I ‹‹ I| Á,
1. Let X1, X2,…Xn be independent identically distributed random variables, each of

0 0
2. Let X1, X2,…Xnbe independent identically distributed exponential random
variables with parameters λ. Show that Z = X1 + X2 +…+ Xn has a gamma
distribution with parameters n and λ.
3. Let X1, X2,…Xn be independent N(µ 1, σ21). Show that
Y = a1 X1 + a2X2 +…+ anXn is a normal random variable with mean µ = a1µ 1 +
a2µ 2 +…+ anµ n and variance σ2 = a21σ21 + a2σ22 + …+ a2n σn2.
∑WI JIýX,
0J3
4. Let S2 = where X1, X2,… Xn are independent identically distributed
W0J3Xó ,
Á,
normal random variable with mean µ and variance σ2. Show that has a chi-
square distribution with n – 1 degrees of freedom.
5. If X has χ2(n), show that E(X) = n and Var(X) = 2n.
6. If X –χ2k1 and Y - χ2k1 where X and Y are independent random varaible. Prove that
X + Y - χ2k1+k2.
If X – N(0, 1) prove that (i) Y = X2 has χ2(1)
IýJ À :
If X – N(µ, σ2) prove that H M ' χ: W1X
7. (a)

Á/√0
(b)
∑WI JIýX,
0J3
8. Let S2 = where Xi are independent identically distributed random
:Á 8
0J3
(iii) Show that ^ý and S are stochastically independent.
variables N(µ, σ2), prove that (i) E(S2) = σ2 (ii) Var(S2) = .
2

187
 function of (i) Y = -2 In X and (ii) Z = '2 ∑0þ©3 ln ^þ , where the Xi’s are
9. Suppose X is uniformly distributed over (0, 1). Find the moment generating

independent uniformly distributed over (0, 1) random variables.

10. Let X1, X2,.. be independent identically distributed geometric random variable
with parameter p. Find the moment generating function of Y = X1 + X2 +…+ Xn.

χ2n. Find the pdf of Æ / Þ.

I
11. Let X and Y be independent random variables with the respective pdf N(0, 1) and


|

k2. Find the pdf if Æ / .

I/”-
12. Let X and Y be independent chi-square random varaibles with parameters k1 and
¿/”,
13. Let X1, X2 be independent random variables each having an exponential pdf with
parameter λ. Y1 = X1 +, …+ Xi; 1 ≤ i ≤ n. Find the joint pdf of Y1, Y2,…Yn.

conditional pdf of X1 given t has the binomial pdf a H , M, where t = X1 + X2

Let X1, X2 be independent Poisson variables with mean λ. Show that the
3
14.
:

7.4 Order Statistics

7.4.1 Independent Identically Distributed Random Variables
Let X1, X2,…, Xn be independent continuous random variables, each having distribution
function F(x) and probability density function f(x). define
X(1) = smallest of X1, X2,… Xn
X(2) = second smallest of X1, X2,… Xn
.
.
.
X(n) = largest of X1, X2,… Xn
That is, X(1), X(2),… X(n) are random variable obtained by arranging X1, X2,… Xn in an
increasing order.
X(1) = min (X1, X2,… Xn)
X(n) = max (X1, X2,… Xn).
The random variables X(i) is called the ith order statistics and the ordered values x(1) ≤ x(2)
≤ … ≤ x(n) are known as the order statistics corresponding to the random variables X1,
X2,…Xn.
For Example, let x1, x2, x3 be a random sample of size 3 from a random variables X
having probability density function f(x) -∞ < x < ∞.
The possible values of X(1), X(2) and X(3) are as follows:
x(1) = x1 , x(2) = x2 , x(3) = x3 if x1< x2< x3
x(1) = x2 , x(2) = x1 , x(3) = x3 if x2< x1< x3
x(1) = x3 , x(2) = x2 , x(3) = x1 if x3< x2< x1
.
.
x(1) = x3 , x(2) = x1 , x(3) = x2 if x3< x1< x2

188
There are 3! = 6 permutations of (x1, x2, x3) and hence 6 possible ordered statistics. We
shall now find the joint probability density function of X(1), X(2), X(3). Since for any

» » » » »
N ë(: ' U ^W3X U (: = , (3 = U ^W:X U (3 = , (W>X U (> = ì / » f W(: X
permutation, say (x2< x1< x3)

2 2 2 2 2
» f W(3 X » f W(> X / »> f W(3 XfW(: Xf W(> X,
as illustrated in the Figure (1) below

(+) (+) (+)

x2 x1 x3
Figure 1

Since there are 3! Possible permutations, the region D is partitioned into six dejoint sets
D1, D2=,…D6. We have X(1) X(2) X(3).

» » » » »
N ë(3 ' U ^W3X U (3 = U (W:X U (: ' , (> ' U ^W>X U (: ' ì
Thus, for x1< x2< x3

2 2 2 2 2
/ 3! »> f W(3 XWf W(: XXfW(> X
Dividing by ϵ3 and letting ε -> 0, we have
f(x(1), x(2), x(3)) = n! f(x1 f(x2) f(x3); x1< x2< x3
in general, we can extend this idea to a sample of size n and we have for x1 x2 … xn
fx(1), x(2), … x(n) (x1, x2,… xn) = n! f(x1)f(x2)…f(xn);
x1< x2<… < xn.
the marginal density function of X(i) the ith order statistic can be obtained by integrating

fk- W( X / p … p fkW-X ,kW,X…kW|X W(3 (: , … (0 X q(: q(> … q(0

the probability density function (5)

fkW-X W(þ X / o … o fkW-X…,kW|X W(3 , (: , … (0 X q(3 q(: … q(þJ3 q(þ‹3 … q(0 .

Example 33

1 0 U ( U 1n
f W( X / ¦
Let x1, x2, x3 be a random sample of a random varaible X having the p.d.f.

0 i‚ƒi„…i†i
Find
(i) the joint p.d.f. of X1, X2 and X3
(ii) the joint p.d.f of X(1), X(2) and X(3) the ordered statistics
(iii) the marginal p.d.f of X(1).

fk-,k, ,kA W(3 , (: , (> X / fk- W(3 Xfk, W(: XfkA W(> X ' 1; 0 U (3 , (: , (> < 1
Solution

fkW-X,kW,X,kWAX r(W3X , (W:X , (W>X s / 3! fk- W(3 Xfk, W(: XfkA W(> X; 0 U (3 , (: , (> < 1

189
 3 3
fk- W(3 X / o o 3! q(W>X
q(W:X
k- k,
3
1 1 3!
/ 3! o W1 ' (: Xq(: / 3!  ' (3 ' (3 :  / Ž1 ' 2(3 ' (3 : 
k- 2 2 2
/ 3W1 ' (3 X:

3W1 ' (X: 0 U ( U 1n

fkW-X W(X / g .
Thus,

0 i‚ƒi„…i†i
Another method of finding the joint marginal p.d.f of X(1), X(2),… X(n) is by direct
reasoning as follows:
(i) For X(i) to equal x, I – 1 of (x1, x2….xn) should be less than x and n-1 should be
greater than x and 1 of them equals x.

i – 1 observations n-I observations

^' x ^=

: :
P(I – 1 of the Xi’s are less than x) = [F(x)]i-1
P(n – 1 of the Xi’s are greater than x) = [1 – F(x)]n-i
Thus for any given set of partition of (X1, X2…Xn) into 3 as stated above, the required

into the 3 group above is WþJ3X!3!W0JþX! . ½lÓi) av

0!
probability is from Chapter 1 section 14 number of ways of partitioning a set of n object

fk- W( X / Ž W(Xþ©3 Ž1 '  W(X0©3 fW(X. (6)

0!
WþJ3X!3!W0JþX!
(ii) The joint p.d.f. of X(i) and X(j)I < j
(i – 1) values (j – i – 1)values (n – j) values

xi xj
Figure 3

The probability that exactly i of the Xi’s lie in (-∞, x) and (n-1) lie in (x, ∞) is
n
Ci [F(x)]i[1 – F(xi)]n-1
since F(x) = P(X ≤ x) = P(any of the x1, …, Xn falls in (-∞, x).

kW-X W(X / Nr^WþX ≤ (s / ∑0©3 nCj [F(x)]j [1 – F(x)]n-j

the event {X(i) ≤ x} occurs if and only i or more of the Xi’s lie in (-∞, x). Thus
–∞ < x < ∞ (8)

kW-X W(X / ∑0©3 nCj [F(x)]j [1 – F(x)]n-j

If i = 1, we have

By Binomial theorem ∑0©3 Cj [F(x)] [1 – F(x)] =1

(9)
n j n-j

kW-X W(X / 1 ' nCj [F(x)]0 [1 – F(x)]n-0 = 1-[1 – F(x)]n

Thus
–∞ < x < ∞ (10)
190
 kW-X W(X / [F(x)]n
Similarly, when we put i = n in (8)
–∞ < x < ∞ (11)

fkW-X W(X / k- W(X / ) [1 – F(x)]n-1 f(x)

By differentiating, we have

kW|X W(X / kW|X W(X = n[F(x)]n-1f(x)

–∞ < x < ∞
–∞ < x < ∞

2. The Joint Distribution Function of X(1), X(n).

To find the joint distribution function of Xi and Xn, we proceed as follows.

{(x1, xn) : X(1) ≤ x1, x(n) ≤ x(n)} = {(x1, xn) : X(n) ≤ x(n)}\{x, y)}:
X(1)> x1, x(n) ≤ xn}.
See figure 5 below.
X(n)

X(n)

Figure 5
We know that P(X(n) ≤ xn) = [F(xn)]n, therefore
P(X(1)> x1, X(n)) = P(x1< X1 ≤ xn, x1< x2 ≤ xn… x1< Xn ≤ xn)
= (F(xn) – F(x1))n
Thus
fkW-XkW|X W(3 , (0 X / P(X(1) ≤ x1, X(n) ≤ xn) = [F(xn)]n – {F(xn) - F(x1)}n (12)

Ú:
from which the density function can be derived by differentiation:
fkW-X,kW,X W(3 , (0 X /  W( , ( X
Ú(3 Ú(0 kW-XkW|X 3 0
= n(n-1) f(x1)f(xn) [F(xn) – F(x1)]n-2, x1 ≤ xn (13)
Distribution Function of the Range of a Random Sample
The random variable R, defined by R = X(n) – X(1) is called the range of the observed

fkW-X,kW|X (x1, xn) dx1dx2

random variables.
P(R ≤ r) = P(X(n) – X(1) ≤ r) = ∫àk
| Jk- ã¡
= pJ[ pk )W) ' 1X [F(xn) – f(x1)]n-2 f(x1) f(xn) dxndx1.
[ k- ‹¡
-

k- ‹¡ Wk- ‹¡XJWk- X
1
Making the change of variable y = F(Xn) – F(x1), dy = f(xn) dxn

o Ž W(0 X '  W(3 X0J: f W(0 Xq(0 / o v 0J: qv / Ž W(3 = †X

k- )'1
'  W(3 X0J3

191
 FR(r) = P(R ≤ r) = npJ[Ž W(3 = †X '  W(3 X0J3 f W(3 Xq(3
Thus
[

FR(r) = n(n - 1) pJ[  W(3 XfW(3 = †XŽ W(3 = †X' W(3 X0J: q(3
[
(14)

Example 40
Let X1, X2,… Xn be a random sample from uniform distribution on (0, 1). Find
(i) The distribution function of R = X(n) – X(1)
(ii) The probability density function of R.

P(R ≤ r) = pe Ž W(3 = †X '  W(3 X0J3 f W(3 Xq(3

k
Solution

Where F(x) = pe 1 q( / (;
k
(i)

F(x1 + r) = 1 if x1 + r > 1
= x1 + r if x1 + r < 1.

P(R ≤ r) = npe † 0J3 q(3 = ) p3J¡W1 ' (3 X0J3 q(3

3J¡ 3
Thus

W XW X 0J: 0 ≤ † ≤ 1n
fR(r) = F’R(r) = g) ) ' 1 1 ' † †
FR(r) = n(1 – r)rn-1 + rn

0 ۏ…i†„lƒi
(ii) .
Another method of finding the probability density function of R is by using the formula

f W†X / pJ[ fkW-X ,kW|X W(3 , † = (3 Xq(3

[
(1) for sum of random variables.

[ f W( Xf W† = ( XŽ W† = ( X '  W( X0J: q( †Y0n

/ g)W) ' 1X pJ[
0 †U0
Example 41
Let X1, X2, … Xn be independent random variables having the uniform distribution on (a,
b). Find the pdf of (i) X(1) = min (X1, X2,… Xn) (ii) X(n) = max (X1, X2,… Xn).

Solution

fk W( X / , ] U ( U a.
3
The pdf of X is
—J˜

fkW-X W(X / Ž W(XþJ3 Ž1 '  W( X0J3 f W(XfW(X

0!
The pdf of X(i) is given by
WþJ3X!W0J3X!

F(X) = p˜ q( /
k 3 kJ˜
where
—J˜ —J˜
( ' ] 0J3 1 )Wa ' (X0J3
fkW-X W(X / ) x1 ' ‘ / ,] U ( U a
a'] a'] Wa ' ]X0
,] U ( U a n
0W—JkX|}-
kJ˜ 0J3
fkW-X W(X / ) x ‘ / \
3
W—J˜X|
i‚ƒi„…i†i
—J˜ —J˜
(ii)
0

192
Exercise 7.4
1. Let X1, X2,…Xn be independent and each is uniformly distributed over (a, b). find
the joint p.d.f of X(1), X(2),… X(n).
2. Let X1, X2,… Xn be independent random variables each having an exponential
density with parameter λ. Find the pdf of (i) X(1) = min (X1, X2,… Xn)
(ii) Xn = max (X1, X2,… Xn).
3. Let X1, X2,… Xn (n odd) be independent random varaibles each uniformly
distributed over (0, 1). Find the p.d.f of the median of the sample.
Note:When n is odd, the median is ^H|Œ-M.
,
4. Show that if n people are distributed at random along road Y miles long, then the

W0J3X” 0
x1 ' ‘ , U
probability that no 2 people are less than a distance k miles apart is
¿
w 0J3
5. Let X1, X2,… X2n-1 be a random sample from uniform distribution on (0, 1). Show
that the median of the sample has a beta distribution with parameter (n + 1, n + 1).

193
 MODULE FOUR
UNIT 1

LIMIT THEOREMS

The Law of Large Numbers and the Central Limit Theorem

The most important theoretical results in probability are limit theorems. In this chapter
we prove often useful tools in probability, the Chebychev’s inequality. This inequality is
then used to deduce the law of large numbers for independent and identically distributed
random variables. In section 8.5 the central limit theorem which is a fundamental result in
the theory and applications f probability theory is given.

8.1 Chebychev’s Inequality

Chebychev;s inequality gives an upper bound in terms of variance of a random variable X
for probability that X deviates from its mean by more than k units.

Theorem 8.1:Chebychev’s Inequality

Let X be a random variable (discrete or continuous) with mean µ and variance σ2. Then
for any positive number k we have
PDX-μD Y k} L
Á,
”,

PDX-μD m k σ}} L
Or equivalently
3
.
”,

Proof:
Let X be a non-negative random variable such that E(X) = µ <∞.
Define another random variable Y as follows:
0 if X < k
Y =
k if X > k
This new variable Y is a discrete variable having two values 0,k. The probability density
function of Y can be written thus:

y 0 k

P(Y = y) P(X < k) P(X > k)

Hence,
E(Y) =OP(X < k) + kP(X > k)
That is
E(Y) = kP(X ≥ k).
194
Since the variable X ≥ Y for all possible values, we have
E(X) ≥ E(Y) = kP(X ≥ k),

WIX
Thus,
P(X ≥ k) ≤
”
. (2)

Equation (2) is called the Markov inequality and can be generalized thus: For any j ≥ 0, k
>0
é 
P{DWD ≥ k} ≤ 
”
(3)
The proof of Chebychev’s inequality is an immediate consequences of (3) by putting j =

/
WIJ KX, Á,
2 and W = X - µ, we obtain

”, ”,
P{DX - µD ≥ k} ≤
2 2
Since (X - µ) ≥ k <==>DX-µD> k, we have
Á,
”,
P{DX-µD> k } = P {(X - µ)2 ≥ k2} ≤ .

Á,
Thus

”,
P{DX-µD≤ k } ≤ .

Let X be a random variable having Poisson distribution with mean λ and variance σ2. Use
Example 8.1

Chebychev’s inequality to show that (i) P{PDX-λD ≥ 1} ≤ λ (ii) P H^ Y M ≤ .

>j 5
: j

Using Chebychev’s inequality with µ = λ, and σ2 = λ we have

Solution .

j
”
(i) P{DX-µD ≥ k } ≤ , .
Putting k = 1 we obtain
P{DX-µD ≥ k } ≤ λ.
i 2 / ///Y Â2 /
σ2 4 42
k λ h
On substituting for σ2 we have
(ii)

 :
λ2 = 4k2, k2 = H M ,  / .

: :

j 5
Thus
: j
P{DX-µD ≥ } ≤ .
Since DX-λD ≥ if and only if X-λ <
j Jj
: :
P{DX-λD ≥ } = P{X-λ <- n M = N H^ ' h Y M
or X-λ> λ/2, we see that
j j j
: : :
= P(H^ U M = N H^ Y M ≤
j >j 5
: : j
Since P H^ U M m, „i …]Ói
j
:
PH^ Y M ≤ .
>j 5
: j
195
8.2 The Law of Large Numbers of Bernoulli Trails
Let X1, X2,…, Xn be n independent and identically distributed Bernoulli random variables
and let X = X1 + X2 +…+ Xn be Binomial random variables (number of successes), with
parameters n and p. The mean and variable of X are np and npq respectively.
The mean µ grows as n increases but the standard deviation grows only as √), Using

P{DX - npD> ε} ≤ , ≤ ,.
0
Chebychev’s inequality
0

Theorem 8.2
Let X be the number of successes in n independent Bernoulli trails with probability p of

N ¦ ' N Y ¸ ≤ , ≤ ,
success. For any ε > 0.
I  3
0 0 5 0

lim0Z[ N ¦ ' N Y ¸ / 0
I
and
0
Proof:

ü H M / N, ˆ]† H M / , . )_ /
3
Applying Chebychev’s inequality
I I 
0 0 0 0
±
昡H M
|
/ 'Y 0 ]ƒ )'Y ∞

», 0»,
_ / W1 ' _ X_ ≤ fۆ ]‚‚ _, 0 ≤ _ U 1
3
5

N ¦¹ ' _¹ m »¸ 'Y 0 ]ƒ )'Y ∞

I
Hence
0
I
0
This means that for large number we can be almost certain that will be very close to
probability p of success. This shows that the relative frequency of success in independent
Bernoulli trails converges (in a probabilistic sense) to the theoretical probability p of
success at each trail. This important result can be stated more precisely as the law of large
numbers.

The Law of Large Numbers

8.3 The Weak Law of Large Numbers:
Let X1, X2,…, Xn be independent and identically distribution random variables with all
E(Xi2) < ∞ and let Sn – X1 + X2 +…+ Xh.
We know that

ü H |M / ü W
0 X / . ) ü W^3 X / ü W^3 X

ó 3 3
E(X1) = E(X2) = … E(Xn), E(Sn) = nE(X1) and Var(Sn) = nVarX1
0 0 0
VarH M / ˆ]†W
0 X / / ˆ]†W^3 X
ó| 3 0昡I- 3
0 0, 0, 0
ó
0
This shows that expectation of is equal to the expectation of Xi and the standard
ó|
0
deviation of is
196
 ¦
昡I ó€I-
¸/
0 √0
Which tends to zero as n tends to infinity. Thus, the distribution of Sn/n becomes more
and more concentrated near E(X1).
We state the more general result in the following theorem.

Theorem 8.3: A (WLLN) Weak Law of Large Numbers for Independent Random
Variables. Let X1, X2…Xn be independent and identically distributed random variables

lim0Z[ N H¹ | ' !¹ Y "M / 0

with finite mean µ and σ2. Let Sn = X1 – X2 + … + Xn. then for δ > 0.
ó
0
(4)
Proof:
ó
Applying Chebychev’s inequality to |, we have
0
! / ü H M . ˆ]† H |M / ˆ]†W
0 X
ó| ó 3
0 0 0,
N ¦¹
0 – üW^X¹ Y "¸ ≤
3 :Á , ±
” Á ,0

For at least one k satisfying n ≤ k ≤ m}.

Corollary 1:
Let f(x) be a continuous real function on [0, 1]. Then as n -> ∞ uniformly with respect to

ü ¦f H |M¸ Z f W_X.
p ε [0, 1]
ó
0
Corollary II: Weierstress Approximation Theorem.
Let f(X) be a continuous real function on [0, 1] defined on the real interval 0 ≤ p ≤1 and

Pn(p) = ∑0”©e f H MnCkpk (1 – p)n – k.

”
let
0
Then Pn(p) -> ∞ f(p) as n -> ∞ and the convergence is uniform in p. Pn(x) is called the
Bernstein Polynomials.

Theorem 8.5
A WLLN for independent but not necessarily identical random variables.

E((X2) < ∞ and let Sn = X1 + X2 +… + Xn and µ = EH |M, then

ó
Let X1, X2,…Xn be independent random variables with all
0
N ¦¹ ' !0 ¹ m »¸ ≤ , , ˆ]†W^3 X = ˆ]†W^: X =  = ˆ]†W^0 X°.  Y 0.
ó| 3
0 » 0
When VarXi ≤ M for all I, that is when the variances are all bounded bya constant which

N ¦¹ | ' !0 ¹ m »¸ ≤ , 'Y 0 ]ƒ )'Y ∞.

ó
does not depend on n, then we have
#
0 » 0

8.5 The Central Limit Theorem

The central limit theorem is a fundamental result in the theory and application of
ó|
0
probability. The law of large numbers asserts that in a series of n independent trails
197
with constant probability p of success in one trail tends (in a certain probabilistic sense)
ó|
0
to p as n increases. But this assertion does not tell us anything about the distribution of

as n becomes large. The answer to this question is given by the so called central limit
theorem. The theorem asserts that under quite general conditions the sum of independent
variables has the Normal distribution in the limit.

Theorem 8.6: Central limit theorem

Let X1, X2, …, Xn be a sequence of independent random variables with E(Xi) = µ and
Var(Xi) = σi2 I = 1, 2,…Let Sn = X1 + X2 +…+ Xn. Then under general conditions,

0 ' üW
0 X
0 ' ∑0þ©3 !þ

Æ0 / /
`ˆ]†
0 `∑0þ©3 Âþ :
Has approximately the standard Normal distribution.

The Identically Distribution Case

Theorem 8.7
Let X1, X2, …, Xn be independent identically distributed random variables. Then

$|
JK
Æ0 / Æ0 /
ó| J 0K |
√0Á, `Á, /0
(i) (ii)

tend to the standard normal distribution as n -> ∞.

What is remarkable about this theorem is that all we need to know about X is its mean
and variance and there is no need to specify the nature of its distribution. Also X can be
discrete, continuous or both. The outline of the proof is given below,
198
Proof:
Let MX,(t) = E(etX), Sn(t) = [E(etX)]n
/ ' √). .
ò0 J 0K ò0 K
√0Á, √0Á Á
Let Zn =

The m.g.f of Zn is
0
™È| WX / i W`0K/ÁX x™I H M‘

Á √0

Taking log we have

) ™È| WX / '√)  ' ) ) ™I H M.

K 3
Á Á √0

Using Maclaurin series

=
WeX ,
:!
M(t) = 1 + M’(0)t + M’’

From above we have

) ™È| WX / '√)  = ) ) x1 = = = %‘

K K rK, J Á , s ,
Á Á √0 :0Á,
(7)

where A is the remainder term from Maclaurin series. Similarly, when the Maclaurin
series

= =
I, IA
: >
In (1 + x) = x -

is used to expand In H1 = =  = %M . ƒil)½ ( / = = %,

K WK, ‹ Á , X K rK, ‹ Á , s ,
√0Á 0 Á, √0Á :0Á ,

we find that

In H1 = =  = %M
K WK, ‹ Á , X
√0Á 0 Á,
:
=  = %- x =  = %‘ = 
K WK, ‹ Á , X 3 K WK, ‹ Á, X
√0Á 0 Á, : √0Á 0 Á,
=

It is left as an exercise to show that for large n.

¹ = = %¹ U 1.
K WK, ‹ Á, X
√0Á :0Á,

On substituting in (7) we obtain

199
 !
) ™È| WX / '√)
Â
! W! : ' ! : X 1 ! W! : =  : X :
:
=) ¨ = = %' = = %
= … &
 √) 2) : 2 Â√) 2) :

It can easily be shown that as n -> ∞

) ™È| WX / .
,
:

Hence,
lim ™È| WX / i  ,
, /:

0Z[

the moment generating function of a random variable having the standard normal
distribution.

Let Xn- n > 1, and X be random variables such that lim0Z[ ™I' WX / ™I WX, '∞ U  U
Theorem

∞. Then Lim ^)(/ ^(] ]‚‚ _Ûl)ƒ „…i†i ^( is continuous.

Example 8.2
A boy throws a fair die 100 times. What is the probability that his mean score will exceed
3.
Let X represent the outcome of each toss.
Possible values of X are 1, 2, 3, 4, 5, 6.
E(X) = 3.5. Var(X) = 2.92.
That is µ = 3.5. σ2 = 2.92
Let X be the mean score. By the central limit theorem X is normally distributed with
mean 3.5 and Varσ2/n.

/ .
Á, :.d:
0 3ee

Hence,

Æ/
IýJ >.4
√e.e:d

has a standard normal distribution.

NW^ý Y 3X / N HÆ Y M / NWÆ Y '2.94X / 0.9984.

Je.4
e.3Ve

200
Example 8.3
Marks in an I.Q examination are normally distributed with mean 55 and standard
deviation 10. What is the probability that
(i) the mean mark of a group of 10 students will be above 50
(ii) the mean mark of a group of 20 students will be between 40 and 50.
(iii) the sum of the marks of the 10 students will be less than 500.

Solution
Let X1, X2,… X10 be the mark scored by each student respectively.

ÂI- / ÂI, / ÂI| / 10.

E(X1) = E(X2) =…= E(X10) = 55

(i) ^ý = mean, be the central limit theory (assuming n = 10 is large enough for this
Iý J 44
Á/√0
distribution) has a standard Normal distribution.

/
IýJ 44 Iý J 44
3e/√3e √3e

NW^ý Y 50X / N H Y M / 0.9306.

IJ ý 44 4e J 44
√3e √3e

(ii) The mean mark of a group of 20 students between will be between 40 and 50. In

NW40 U ^ý U 50X / N W^ý U 50X ' NW^ý L 40X

this case, n = 20.

NW^ý L 40X / N HÅ L M / NWÆ L '3 √5X / 0.00

5e J 44
√4

NW^ý U 50X / N HÅ U M / NWÆ U ' √5X / 0.01

4e J 44
√4

P(40 <^ý < 50) = 0.01 – 0.00 = 0.01

Hence

(iii) S10 = X1 + X2 +…+ X10

ó-. J 0K
√0Á,
has a standard Normal

/ / / '1.53
ó-. J 44e 4ee J 44e J4e
√3eee 3e√3e 3e√3e

Hence,
P(Sn< 500) = 0.057.
201
8.6 Normal Approximately to the Binomial Distribution
Let X1, X2, …, Xn be n independent identically distributed Bernoulli trails where
1 with probability P
Xi =
0 with probability 1 - P
E(Xi) = p and Var(Xi) = p(1 – p). From theorem 2, the distribution
ó| J Wó| X
√0Á,
--> standard Normal.

where Sn = X1 + X2 + … + Xn
E(Sn) = np
Var(Sn) = np(1 – p), σ2 = p(1 – p).
Note that Sn has binomial distribution with mean np and variance np(1-P).

P(Sn ≤ y) = ∑”©e
w n
Ck Pk (1 – p)n-k
For large n

P(Sn ≤ y) = N gÆ0 ≤
w J 0
(
`0W3 J X

)/ H M
w J 0
`0W3 J X

where Zn is standard Normal random variable. Thus, from above we have

ó| WvX / NW
0 ≤ v X / ∑”©e Ckpk (1 – p)n-k = ) H M

w w J 0
0W3 J X
n

Example 8.3
It is claimed that 60% of the voters I a given ward are going to vote for party A. assuming
that all voters will vote, and that there are 100 voters. What is the probability that Party A
receives at least 50 votes.

NW
0 Y 50X / ∑3ee

Let p = 0.6. prob. Of voting for A. n = 100. Assuming independent voting
”©4e 100Ck p (1 –p)
k 100-k
=∑3ee
”©4e 100Ck (0.6) (0.4)
k 100-k

Using normal approximation (6) above, we have

P(Sn > 50) = N HÆ m M / N HÆ m M / 0.9793

4e – 3ee 9e.< 4e J <e
`3eeWe.<XWe.5X √:5

8.7 The Normal Approximation to the Poisson Distribution

202
If a random variable X has a poisson distribution whose mean is λ, then for large λ the
standardized random variable.
IJ j
Æ/
√j

has a standard Normal Distribution. you will recall that E(X) = λ , Var(X) = λ,
Continuity correction may be introduced.
It has been shown that the approximation is good for λ > 5.

Example 8.4
A system suffers random breakdown at a constant rate of 10 per month. Find the
probability that there will be at least 8 breakdowns in any month.
Let X be the number of time the system breaks down in any month. Then X has a poisson
distribution with λ = 10. Thus

NW^ Y 8X / ∑[
~ -. W3eX“
”©; ”!

Using Normal approximation, we have

I J 3e
√3e
has N(0.1)

NW^ m 8X / N HÆ Y M / N HÆ Y M / 0.7357
; J 3e J:
√3e √3e

Exercise 8
1. It is claimed that 30% of the voters in a given local government area are going to
vote for party A in a local government election. Assuming that all voters will vote,
there are 4000 voters and the claim is based on a proper (unbiased) sampling
method. What is the probability that Party A will receive more than 1,500 votes.
2. (a) Suppose that a system consists of components each of which has a
probability of 0.05 of failing during a specific time. The system functions
properly when at least 150 components function. Assuming these
components function independently of one another, what is the probability
that the system functions properly during a specific time.
(b) Suppose that the above system is made of n components each having
probability of 0.90 of not failing during a given time. The system will

203
 function if at least 80 percent of the components function properly.
Determine n so that the probability that the system functions properly
during a specific time is 0.96.
3. Let X be a non-negative interger-valued random variable whose probability
generating function PX(s) = E(sX) is finite all s.
Use Chebychev’s inequality to verify the following inequalities:

N W^ ≤  X ≤ . 0 ≤ ƒ ≤ 1,  any positive inequalities:

*± WòX
ó“
(a)

N W^ Y  X ≤ .ƒ Y 1
*± WòX
ó“
(b)

4. Prove the Chebychev’s inequality.

Let X be a random variable (discrete or continuous) with E(X) = µ and Var(X) =
σ2. Then for any positive constant number k we have
.
3
”,
P{|X - µ| ≥ k σ} ≤

5. Prove that (i) 1 – x ≤ e-x for real x, (ii) log x ≤ x -1 for x > 0, and

(iii) ∏,
þ©3 ^þ ≤ i lf ^þ ≤ 1, Ò / 1,2, …
Ñ
J ∑- I

204