Prepared by: M.Riaz M.sc: Statistics I.U.B (2006-8): Lecturer NICAAS college R.Y.

K 1

1. Binomial distribution
Discrete variable
A variable which takes some specific values in a given range is called discrete variable.
Usually; discrete variables take values as whole integers
Such as 0, 1, 2, 3,…, n
Example
Number of children in a family
Define Bernoulli trial
Ans: A trial that gives only two possible outcomes is called Bernoulli trials.
Example:
There are many trials they have only two possible outcomes such as
i) Head and tail ii) Success and failure III) Alive and dead
iv) Right and wrong v) Good and defective
What are the Properties of Bernoulli experiments?
Ans: There are four properties of Bernoulli experiment
i) Successive trials are Independent
ii) The probability of success remains same for all trials
iii) The experiment is repeated a single time
iv) Each trial classified into two categories such as success(S) or failure(F)
Define binomial experiment.
Ans: An experiment consisting of “n” Bernoulli trials is known as binomial experiment.
What are the Properties of Binomial experiment?
Ans: There are four properties of Binomial experiment
i) Successive trials are Independent
ii) The probability of success remains same for all trials
iii) The experiment is repeated a fixed number of times say “n”
iv) Each trial classified into two categories such as success(S) or failure(F)
Explain briefly binomial random variable?
Ans: The random variable which denotes the number of successes of binomial
experiment is called binomial random variable. It is discrete variable and assume values
as X=0, 1, 2… n.
n
Why discrete probability function is b( x, n, p ) = c P x q n − x called the binomial
x

distribution?
n
Ans: b( x, n, p ) = c P x q n − x X=0, 1, 2… n. is called binomial distribution because its
x

successive terms are equal to binomial expansion of (q + p ) and they give the
n

probabilities of binomial random variable.

What is binomial probability function or mass function?
Ans: The formula used to find the probabilities of binomial random variable “X” is called
binomial probability function or binomial probability distribution. It is given by
n
b ( x , n, p ) = c P x q n − x X=0, 1, 2… n.
x

Where q = 1 − p P= probability of success of single trial.

n= number independent of trials
Define binomial probability distribution
Ans: Arrangement of all possible values of binomial random variable of “X” with their
probabilities is called binomial probability distribution and given as
X 0 1 ….. n
P(X) n n … n
c P 0 q n−0 c P 1 q n −1 c P n q n−n
0 1 n

Define binomial frequency distribution

Ans: If each probability of binomial distribution is multiplied by N then resulting
distribution is called binomial frequency distribution as given below
X 0 1 ….. n
NP(X) n n … n
N c P 0 q n −0 N c P 1q n−1 N c P n q n−n
0 1 n
Prepared by: M.Riaz M.sc: Statistics I.U.B (2006-8): Lecturer NICAAS college R.Y.K 2

Describe the properties of binomial distribution

Ans: The properties of binomial distribution are given below
i) Total probability of binomial probability distribution is one.
ii) Mean of binomial distribution is “np”
iii) Variance of binomial distribution is “npq”
iv) Shape of binomial distribution b( x, n, p ) depends upon “n and p”
1
When p = q = or p = q = 0.5 The distribution is always symmetrical
2
When p ≠ q The distribution is said to be skewed
1
When p > or p > 0.5 The distribution is said to be negatively skewed
2
1
When p < or p < 0.5 The distribution is said to be positively skewed
2
How can the mean and variance of binomial probability distribution be calculated
in terms of parameters?
Ans: Mean is calculated by multiplying by “n and p” Mean=np
Variance is calculated by multiplying by “n, p and q” Variance=npq
Why in binomial distribution variance is smaller than mean give reason?
Ans: We know that in binomial distribution mean is “np” and variance is “np(1-p). To
calculate variance mean is further multiplied by a number which is less than one hence
Mean> Variance
State the formula used to calculate binomial probabilities?
Ans: The formula used to find the probabilities of binomial random variable “X” is called
binomial probability function or binomial probability distribution. It is given by
n
b ( x , n, p ) = c P x q n − x X=0, 1, 2… n.
x

Where q = 1 − p P= probability of success of single trial.

n= number independent of trials
Which formulae’s should be used to find the Mean, Variance, Standard deviation
and Coefficient of variation of a binomial random variable?
Ans: To used the following formulae’s to calculate Mean, Variance, Standard deviation
and Coefficient of variation of a binomial random variable are given below
E ( X ) = Mean = np
Variance( X ) = np (1 − p ) or npq
S .D( X ) = np(1 − p) npq or
S .D( X )
Coefficien of var iation = C.V = ×100
E( X )
Derivation of binomial distribution
Let the binomial probability experiment having the following four properties
i) Each trial classified as success and failure
ii) Probability of success remains same = Constant
iii) Successive trials = independent
iv) Experiment repetition= Fixed number of times
Then
Total trials = n
Total success = X
P(Success)=P(S)=P
P(Failure)=P(f)=q
Total possible outcomes = 2 n
Each outcome has a sequence = {a1 , a 2 ,..., ai ,..., a n }
Where ai = Success and failure
First we consider no success i.e. P(X=0)
Then sequence of failure is = F .F .F ...F
P( X = 0) = P(F .F .F ...F ) = P( F ).P( F )...P( F ) = q.q.q...q = q n
And number of mutually sequence = n C 0 ; P 0 = 1
So, P ( X = 0)= n C 0 P 0 q n
Prepared by: M.Riaz M.sc: Statistics I.U.B (2006-8): Lecturer NICAAS college R.Y.K 3

Now we consider one success then sequence of one success and n-1 failures are in form
of
S .F .F ...F ; F .S .F ...F ;...; F .F .F ...S
P( X = 1) = P(S .F .F ...F ) + P(F .S .F ...F ) + ... + P(.F .F ....S )
P( X = 1) = Pq n −1 + Pq n−1 + ... + Pq n −1
And number of mutually sequence = n C1
So, P( X = 1)= n C1 Pq n −1
If we consider two success in “n” trials
P ( X = 2)= nC 2 P 2 q n−2
In general if we consider x success and n-x failures in “n” trials
P( X = x)= n C x P x q n− x Where X = 0,1,2,3,..., n
Properties of binomial distribution
i) Total probability of binomial distribution is one.
Proof: Let by definition
Total probability = P( X )
As X → b( x, n, p ) then its Probability function given as
P( X )= n C x P x q n − x Where X=0, 1, 2,..., n
P( X )= n C 0 P 0 q n −0 + n C1 P 1q n−1 + n C 2 P 2 q n −2 + ....+ nC n P n q n −n
P( X ) = (q + p ) = (1) n = 1
n

ii) Moment generating function

Proof: Let by definition
( )
m0 (t ) = E e tx = e tx P( X )
As X → b( x, n, p ) then its Probability function given as
P( X )= n C x P x q n − x Where X=0, 1, 2,..., n
m0 (t ) = e tx n C x P x q n − x = n
( )
C x Pe t q n − x
x

( ) 0
( ) 1
m0 (t )= n C 0 Pe t q n −0 + n C1 Pe t q n −1 + n C 2 Pe t q n − 2 + ...+ n C n Pe t q n − n ( ) 2
( ) n

( )n
m0 (t ) = q + Pe t Required result
Use it to find mean and variance
d
µ1′ = Mean = [m0 (t )]t =0
dt
µ1′ = Mean =
d
dt
n
[(
q + Pe t t = 0 )]
n −1 d
µ1′ = Mean = n q + Pe t ( dt
q + Pe t ) ( )
t =0

µ1′ = Mean = n(q + Pe [ )

t n −1
Pe t
]
t =0

[ [ ] ]
2
d
µ 2′ = 2
[m0 (t )]t =0 = d d m0 (t ) =
d
nPe t (q + Pe t )
n −1
t =0 t =0
dt dt dt t =0 dt
d d t
µ 2′ = nPe t
dt
(
q + Pe t ) n −1
+ np q + Pe t( ) n −1

dt
e
t =0

d
µ 2′ = nPe t (n − 1) q + Pe t ( )
n−2

dt
(
q + Pe t + nP q + Pe t ) ( )
n −1
et
t =0

[(
µ 2′ = nPe (n − 1)(q + Pe
t
)
t n−2 t
Pe + nP q + Pe ( )
t n −1
e t
)] t =0

µ ′ = [(n(n − 1) P e
2
2 2t
(q + Pe ) t n−2
+ nP q + Pe ( )
t n −1
e t
)] t =0

µ ′ = [(n(n − 1) P e
2
2 2(0 )
(q + Pe ) 0 n−2
+ nP (q + Pe )
0 n −1
e 0
)]
[(
µ 2′ = n(n − 1) P 2 (q + P ) n−2
+ nP(q + P )
n −1
)]
µ ′ = [(n(n − 1) P (1)
2
2 n−2
+ nP(1)
n −1
)]
µ 2′ = (n(n − 1) P + nP ) = n P − nP 2 + nP
2 2 2

Var ( x) = µ 2′ − (µ1′ ) = n 2 P 2 − nP 2 + nP − (nP) 2

2
Prepared by: M.Riaz M.sc: Statistics I.U.B (2006-8): Lecturer NICAAS college R.Y.K 4

Var ( x) = n 2 P 2 − nP 2 + nP − n 2 P 2 = −nP 2 + nP
Var ( x) = nP(1 − P) = nPq
Required result
iii) Characteristic function
Proof: Let by definition
Q x (t ) = M x (it )
As we know
( )
m x (t ) = q + Pe t
n

Q (t ) = (q + Pe )
x
it n
Required result
iv) Probability generating function
Proof:
Let by definition
G (θ ) = E (θ ) X
As X → b( x, n, p ) then its Probability function given as
P( X )= n C x P x q n − x Where X=0, 1, 2,..., n
C x (Pθ ) q n − x
x
G (θ ) = θ x P( x) = θ x nC x P x q n− x = n

G (θ )= n C 0 (Pθ ) q n− 0 + n C1 (Pθ ) q n −1 + n C 2 (Pθ ) q n − 2 + ...+ n C n (Pθ ) q n − n

0 1 2 n

G (θ ) = (q + Pθ ) Required result
n

Use it to find mean and variance

d
′ = Mean =
µ(1) [G (θ )]θ =1
dθ
µ1′ = Mean =
d
dθ
[
(q + Pθ )n θ =1 ]
n −1 d
′ = Mean = n ( q + Pθ )
µ(1) ( q + Pθ )
dθ θ =1
n −1
′ = Mean = n ( q + P )
µ(1) P
′ = Mean = np
µ(1)
Similarly
µ2′ = µ(2)
′ + µ(1)
′
µ3′ = µ(1)
′ + µ(2)
′ + µ(3)
′
µ4′ = µ(1)
′ + µ(2)
′ + µ(3)
′ + µ(4)
′

Example Find mean and variance of binomial distribution when “n=1”

Solution: Mean and variance of binomial distribution when “n” is one

X P(x) XP(X) X 2 P( X )
0 1 0 0
c P 0 q1 = q
0

1
1 c P1 q 0 = P P P
1

Total XP( X ) = p X 2 P( X ) = p

E( X ) = XP ( X ) = p
Var ( X ) = E ( X 2 ) − ( E ( X )) 2 = X 2 P( X ) − ( XP( X )) 2 = P − P 2 = P(1 − P ) = pq
Prepared by: M.Riaz M.sc: Statistics I.U.B (2006-8): Lecturer NICAAS college R.Y.K 5

Example: Find mean and variance of binomial distribution when “n=2”

Solution:
Mean and variance of binomial distribution when “n” is two

X P(x) XP(X) X 2 P( X )
0 2 0 0
c P0q2 = q2
0

2
1 c P 1q 1 = 2 pq 2 pq 2 pq
1

2 c p 2q0 = p2
2 2 p2 4 p2
Total XP( X ) = 2 pq + 2 p 2 X 2 P( X ) = 2 pq + 4 p 2

E( X ) = XP( X ) = 2 pq + 2 p 2 = 2 p(q + p) = 2 p((q + p) = 2 p(1) = 2 p

Var ( X ) = E ( X 2 ) − ( E ( X )) 2 = X 2 P( X ) − ( XP( X )) 2 = 2 pq + 4 p 2 − (2 p ) 2 = 2 pq + 4 p 2 − 4 p 2
Var ( X ) = 2 pq + 4 p 2 − 4 p 2 = 2 pq
S .D ( X ) = Var ( X ) = 2 pq
Example: Find mean and variance of binomial distribution when “n=3”
Solution: Mean and variance of binomial distribution when “n” is three
X n XP(X) X 2 P( X )
P(x)= c P x q n − x
x

0 3 0 0
c P0q3 = q3
0

3
1 c P 1 q 2 = 3 pq 2 3 pq 2 3 pq 2
1

3
c p 2 q1 = 3 p 2 q 6 p2q 12 p 2 q
2 2

3
3 c p3q 0 = p3 3 p3 9 p3
3

Total XP( X ) X 2 P( X )

E( X ) = XP ( X ) = 3 pq 2 + 6 p 2 q + p 3 = 3 p ( q 2 + 2 pq + p 2 ) = 3 p ( q + p ) 2 = 3 p (1) 2 = 3 p
2
Var( X ) = E( X 2 ) − (E( X ))2 = X 2 P( X ) − ( XP( X ))2
2
Var(X) = 3 pq + 12 p 2 q + 9 p 3 − ( 3 p) 2
= 3 pq 2
+ 12 p 2 q + 9 p 3 − 9 p 2
Var(X) = 3 p ( q 2 + 4 pq + 3 p 2 − 3 p ) = 3 p ( q 2 + 4 pq + 3 p ( p − 1 ))
Var(X) = 3 p ( q 2 + 4 pq − 3 p (1 − p )) = 3 p ( q 2 + 4 pq − 3 pq )
Var(X) = 3 pq ( q + 4 p − 3 p ) = 3 pq ( q + p ) = 3 pq (1) = 3 pq
S .D ( X ) = Var ( X ) = 3 pq
Prepared by: M.Riaz M.sc: Statistics I.U.B (2006-8): Lecturer NICAAS college R.Y.K 6

Example: Find mean and variance of binomial distribution when “n=4”

Solution:
Mean and variance of binomial distribution when “n” is four

X n XP(X) X 2 P( X )
P(x)= c P x q n − x
x

0 4 0 0
c P0q4 = q4
0

4
1 c P 1 q 3 = 4 pq 3 4 pq 3 4 pq 3
1

4
c p 2q2 = 6 p2q2 12 p 2 q 2 24 p 2 q 2
2 2

4
3 c p 3 q1 = 4 p 3 q 12 p 3 q 36 p 3 q
3

4
4 p4 16 p 4
c p 4q0 = p4
4 4

Total XP( X ) X 2 P( X )

E( X ) = XP( X ) = 4 pq 3 + 12 p 2 q 2 + 12 p 3 q + 12 p 4
E ( X ) = 4 p ( q 3 + 3 pq 2 + 3 p 2 q + 3 p 3 ) = 4 p ( q + p ) 3 = 4 p (1) 3 = 4 p
2
Var ( X ) = E ( X 2 ) − ( E ( X )) 2 = X 2 P( X ) − ( XP ( X )) 2
E(X 2) = X 2P(X )
E ( X 2 ) = 4 pq 3 + 24 p 2 q 2 + 36 p 3 q + 16 p 4
E ( X 2 ) = 4 p (q 3 + 6 pq 2 + 9 p 2 q + 4 p 3 )
E ( X 2 ) = 4 p (q 3 + 3 pq 2 + 3 pq 2 + p 3 + 9 p 2 q + 3 p 3 )
[
E ( X 2 ) = 4 p (q + p) 3 + 3 p 2 q + 6 p 2 q + 3 p 3 ) ]
[
E ( X 2 ) = 4 p (q + p) 3 + 3 p ( pq + 2 pq + p 2 ) ]
[
E ( X 2 ) = 4 p (q + p) 3 + 3 p( p + q) 2 ]
[
E ( X 2 ) = 4 p (1) 3 + 3 p(1) 2 ]
E ( X 2 ) = 4 p[1 + 3 p ] = 4 p + 12 p 2
X 2 P( X ) − ( XP ( X ) )
2
Var ( X ) = E ( X 2 ) − ( E ( X )) 2 =
Var( X ) = 4 p + 12 p 2 − (4 p) 2 = 4 p(1 + 3 p − 4 p) = 4 p(1 − p) = 4 pq
S .D ( X ) = Var ( X ) = 4 pq
Example: Find mean and variance of binomial distribution when for “n”
Solution:
Mean and variance of binomial distribution when for “n”
The binomial random variable “X” With parameters “n and p” has the probability
distribution
n
P( X = x) = b( x; n, p ) = c p x q n − x x = 1,2,3,..., n q + p =1
x
n n n n n −1
n n n −1 x −1 n − x
E( X ) = xP ( x) = x c p x qn−x = x c p pq = np c p x −1q n − x
x =0 x =0
x
x =1 x x −1 x =1
x −1

n −1 n −1 n −1 n −1
E ( X ) = np c p 0 q n −1 + c p 1 q n −2 + c p 2q n −3 + ... + c p n−1 q 0
0 1 2 n −1

n −1 n −1
E ( X ) = np q n −1 + c p 1 q n −2 + c p 2q n −3 + ... + p n −1
1 2

[
E ( X ) = np (q + p) n −1
= np(1) = np ] n −1

E ( X ) = E[ X ( X − 1) + X ] = E[ X ( X −)] + E ( X )
2
Prepared by: M.Riaz M.sc: Statistics I.U.B (2006-8): Lecturer NICAAS college R.Y.K 7

E ( X 2 ) = E[ X ( X − 1) + X ] = E [ X ( X −)] + np ………….. (A)

n n n
n n(n − 1) n − 2 x − 2 2 n − x
E [ X ( X −) ] = X ( X − 1) P ( X ) = X ( X − 1) c p x q n − x = X ( X − 1) c p pq
x =0 x =0
x
x=2 X ( X − 1) x − 2
n n−2
E [ X ( X −) ] = n(n − 1) p 2 c p x−2 q n− x
x−2
x=2
n− 2 n−2 n−2 n−2
E [ X ( X − 1)] = n(n − 1) p 2 c p 0 q n − 2 + c p 1 q n −3 + c p 2 q n −4 + ... + c p n − 2 q 0
0 1 2 n−2

[ ]
E [ X ( X − 1)] = n(n − 1) p 2 (q + p) n− 2 = n(n − 1) p 2 (1) n − 2 = n(n − 1) p 2
Put in A
E ( X 2 ) = n(n − 1) p 2 + np
2
Var ( X ) = E ( X 2 ) − ( E ( X )) 2 = X 2 P( X ) − ( XP ( X )) 2
Var ( X ) = n ( n − 1) p 2 + np − ( np ) 2 = n 2 p 2 − np 2 + np − n 2 p 2 = − np 2 + np = np − np 2
Var ( X ) = np − np 2 = np (1 − p ) = npq
S .D( X ) = Var ( X ) = npq
The recurrence formula for binomial distribution
Proof: Let by definition
As X → b( x, n, p ) then its Probability function given as
P( X = x)= n C x P x q n − x (i)
P ( X = x − 1)= n C x −1 P x −1 q n − x +1 (i)
Dividing (i) by (ii)
P( X = x) n
C x P x q n− x
=
P( X = x − 1) n C x −1 P x −1 q n − x +1
n!
P x q n− x
P( X = x) x!(n − x)!
=
P( X = x − 1) n!
P x −1 q n − x +1
( x − 1)!(n − x + 1)!
P( X = x) n!( x − 1)!(n − x + 1)! x n − x − x +1 −n + x −1
= P q P q
P( X = x − 1) n! x!(n − x)!
P( X = x) ( x − 1)!(n − x + 1)(n − x)! 1 −1
= Pq
P( X = x − 1) x( x − 1)!(n − x)!
P( X = x) (n − x + 1) P
=
P( X = x − 1) x q
(n − x + 1) P
P( X = x) = P( X = x − 1)
x q
Required result
Assignment
(n − x) P
P( X = x + 1) = P ( X = x) It is also recurrence formula do yourself
x +1 q
Question: Show that recurrance relation in binomial distribution
d
µ r +1 = Pq nrµ r −1 + µr
dP
Use it to find first four moments about mean and moments of skewness and kurtosis
Proof: Let we consider
µ r = E ( X − Mean )r
µr = ( X − Mean )r P(x)

As X → b( x, n, p ) then its Probability function given as

P ( X = x)= n C x P x q n − x And Mean = nP
µr = ( X − Mean)r n C x P x (1 − P) n− x
Differentiate with respect to P
Prepared by: M.Riaz M.sc: Statistics I.U.B (2006-8): Lecturer NICAAS college R.Y.K 8

d
dP
µr =
d
dP
[ ( X − Mean )r n C x P x (1 − P) n− x ]
d d x
n
C x P x (1 − P) n − x ( X − nP )r + ( X − nP )r n C x P x (1 − P) n− x P
d dP dP
µr =
dP
+ ( X − nP )r n C x P x d (1 − P) x
dP

d
n
C x P x (1 − P) n − x r ( X − nP ) ( X − nP ) + ( X − nP )r n C x P x (1 − P) n− x xP x −1 d P
r −1

d dP dP
µr =
dP d
+ ( X − nP )r n C x P x (n − x)(1 − P) n− x −1 (1 − P)
dP
C x P x (1 − P) n − x r ( X − nP ) ( X − nP )r n C x P x (1 − P) n− x xP x−1 (1)
n r −1
( − n) +
d
µr = d
dP + ( X − nP )r n C x P x (n − x)(1 − P) n − x −1 ( −1)
dP

− rn n C x P x (1 − P) n − x ( X − nP ) + ( X − nP ) n C x P x (1 − P) n − x xP x −1
r −1 r

d
µr = d
− ( X − nP ) n C x P x ( n − x)(1 − P) n − x −1
r
dP
dP

− rn ( X − nP ) n C x P x (1 − P) n − x + ( X − nP ) n C x P x (1 − P) n − x xP x −1
r −1 r

d
µr = d
− ( X − nP ) n C x P x (n − x)(1 − P) n − x −1
r
dP
dP

d
dP
µ r = −rn ( X − nP )r −1 P( x) + ( X − nP )r nC x P x (q ) n − x P x xP −1 − (n − x)q −1 [ ]
d x (n − x )
µ r = −rnE ( X − nP )r −1 + ( X − nP )r n C x P x (q) n− x P x −
dP P q
d xq − P(n − x)
µ r = −rnE ( X − nP )r −1 + ( X − nP )r n C x P x (q) n− x P x
dP Pq
d x(1 − P) − nP + xP
µ r = −rnE ( X − nP )r −1 + ( X − nP )r n C x P x (q) n− x P x
dP Pq
d x − xP − nP + xP
= − rnE ( X − nP ) ( X − nP )r n C x P x (q) n− x P x
r −1
µr +
dP Pq
d 1
µr = − rnµ r −1 + ( X − nP )r n C x P x (q) n− x P x (x − nP )
dP Pq
d 1
µr = − rnµ r −1 + ( X − nP )r +1 n C x P x (q) n− x P x
dP Pq
d 1
µr = − rnµ r −1 + ( X − nP )r +1 P( x)
dP Pq
d 1
µr = − rnµ r −1 + µ r +1
dP Pq
d 1
µr + rnµ r −1 = µ r +1
dP Pq
d
µ r +1 = Pq rnµ r −1 + µr
dP
Hence proved
Now first four moments about mean
Put r = 0
d
µ 0+1 = Pq 0(nµ 0−1 ) + µ0 Where µ 0 = 1
dP
d
µ 0+1 = Pq 0 + 1 = Pq(0) = 0
dP
Prepared by: M.Riaz M.sc: Statistics I.U.B (2006-8): Lecturer NICAAS college R.Y.K 9

µ1 = 0
Put r = 1
d
µ1+1 = Pq 1(nµ1−1 ) + µ1 Where µ 0 = 1 and µ1 = 0
dP
d
µ 2 = Pq nµ 0 + 1 = Pq(n + 0 ) = nPq = Var ( x)
dP
Put r = 2
d
µ 2+1 = Pq 2(nµ 2−1 ) + µ2 Where µ 2 = nP(1 − P) and µ1 = 0
dP
d
µ 3 = Pq 2(nµ1 ) + nP (1 − P )
dP
d d d
µ 3 = Pq 0 + n P (1 − P ) = nPq P (1 − P ) + (1 − P ) P
dP dP dP
µ 3 = nPq(P(−1) + (1 − P)1) = nPq(− P + (1 − P) )
µ 3 = nPq(1 − 2 P ) = nPq(1 − P − P ) = nPq(q − P )
Put r = 3
d
µ 3+1 = Pq 3(nµ 3−1 ) + µ 3 Where µ 2 = nPq and µ 3 = nP − 3nP 2 + 2nP 3
dP
d
µ 4 = Pq 3(nµ 2 ) +
dP
(
nP − 3nP 2 + 2nP 3 )
d d 2 d 3
µ 4 = Pq 3n(nPq) + n P − 3n P + 2n P
dP dP dP
(
µ 4 = Pq 3n(nPq) + n(1) − 3n(2 P) + 2n(3P 2 ) )
µ4 = Pq[3n 2
Pq + n − 6nP + 6nP ] 2

µ 4 = 3n P q + nPq[1 − 6 P + 6 P 2 ]
2 2 2

µ 4 = 3n 2 P 2 q 2 + nPq[1 − 6 P(1 − P)]

µ 4 = 3n 2 P 2 q 2 + nPq[1 − 6 Pq ]
µ 32 (nPq(q − P ))2 (nPq )2 (q − P )2 (q − P )2
Moment of skewness β1 = = = =
µ 23 (nPq )3 (nPq )3 nPq
Moment of kurtosis
µ 3n 2 P 2 q 2 + nPq[1 − 6 Pq ] 3n 2 P 2 q 2 nPq[1 − 6 Pq ] [1 − 6 Pq ]
β 2 = 42 = = + = 3+
µ2 (nPq ) 2
(nPq ) 2
(nPq ) 2
nPq
Q: Discuss the statement that in a binomial distribution mean=5 and Standard
deviation=5.
Ans: Given that µ =5 and σ = 5 σ 2 = 25
Mean < Variance
It is wrong because in binomial distribution Mean > Variance
1
Example.8.3: Let “X” have a binomial distribution with n = 4 and P = . Find
3
3
P ( X = 1), P ( X = ), P ( X = 6) , P ( X = −1), P ( X = 0.5) and P ( X ≤ 2)
2
Solution: Let “X” be binomial random variable and n=4 then “X=0, 1, 2, 3, 4”
1
X → b(x,4, ) And P ( X = x)= n C x P x q n − x
3
1 2
i) P ( X = 1)= 4 C1 ( )1 ( ) 3 = 0.40
3 3
3 1 2
ii) P ( X = )= 4 C x ( ) x( ) 4− x = 0 Because it is based on discrete variable which is
2 3 3
assume whole integers?
1 2
iii) P ( X = 6)= 4 C x ( ) x( ) 4− x = 0 Because it’s out of range
3 3
Prepared by: M.Riaz M.sc: Statistics I.U.B (2006-8): Lecturer NICAAS college R.Y.K 10

1 2
iv) P ( X = −1)= 4 C x ( ) x( ) 4− x = 0 In binomial random variable we assume positive
3 3
values it is also impossible
1 2
v) P ( X = 0.5)= 4 C x ( ) x( ) 4− x = 0 Because it is based on discrete variable which is
3 3
assume whole integers?
vi) P ( X ≤ 2) = P ( X = 2) + P ( X = 1) + P ( X = 0) = 0.89
Example:8.6: Six dice are thrown 729 times. How many times do you expect at least
three dice to show a five or a six?
Solution Given that N = 729 and n = 6
Feverable cases
P (event ) =
Total possible outcomes
When a single die throw possible outcomes
S = {1,2,3,4,5,6}
Feverable outcomes "5 or 6" occur = {5,6}
2 1
P (5 or 6) = =
6 3
233
P ( X ≥ 3) = P ( X = 3) + P ( X = 4) + P ( X = 5) + P ( X = 6) =
729
The required expected numbers are
233
NP ( X ≥ 3) = 729 = 233
729
Example: 8.7: A certain event is believed to follow the binomial distribution. In 1024
samples of 5, the result was observed one 405 times and twice 270 times. Find “p and q”.
Solution: Given that according to condition
NP( X = 1) = 1024( 5 C1 )( p)1 (q) 5−1 = 10254(5 pq 4 )
10254(5 pq 4 ) = 405 (i)
NP( X = 2) = 1024( 5 C 2 )( p) 2 (q) 5−2 = 10254(10 p 2 q 3 )
10254(10 p 2 q 3 ) = 270 (ii)
Now dividing eq (i) dividing by (ii)
10254(5 pq 4 ) 405
=
10254(10 p q ) 270
2 3

(q ) = 45
(2 p ) 30
30q = 90 p
q = 3p
1− p = 3p
1 = 3p + p = 4p
1 3
p = = 0.25 q = 1 − p = 0.75 =
4 4
Q.8.3 (a): A die is rolled five times and a 5 or 6 is considered a success. Find the
probability of i) no success ii) at least 2 success iii) At least one but not more than 3
success.
b) Using the binomial distribution, find the probability of
i) 3 success in 8 trials when “p==0.4” ii) 2 failures in 6 trials when “p=0.6”
iii) 2 or fewer success in 9 trials when “p=0.4”
1 2
Solution: a) given that n = 5 p= q=
3 3
i) p ( x = 0) = ? ii) p ( x ≥ 2) = ? iii) p (1 ≤ x ≤ 3) = P ( x = 1) + P9 x = 2) + P ( x = 3)
b)
i) ) Given that P ( x = 3) = ? n = 8 p = 0 .4 q = 0 .6
i) ) Given that P ( x = 6 − 2 = 4 Success) = ? n = 6 p = 0.6 q = 0 .4
i) ) Given that P ( x ≤ 2) = ? n = 9 p = 0 .4 q = 0 .6
Prepared by: M.Riaz M.sc: Statistics I.U.B (2006-8): Lecturer NICAAS college R.Y.K 11

Q.8.5 (a) if the probability of getting caught copying someone else’s exam is 0.2, find the
probability of not getting caught in 3 attempts. Assume independence
Solution: P=0.2 q=0.8 n=3 P(X=0) =0.512
Q.8.6 (b): If on the average rain falls on twelve days in every thirty, find the probability
that i) The first three days of a given week will be fine and the remaining wet
ii) Rain will fall on just three days of a given week.
12
Solution: P (rain) = = 0 .4 q = 0 .6
30
i) P( First three days fine and remaining wet ) = q 3 P 4 = (0.4) 3 (0.6) 4 = 0.0055
i) P( First three days fine and remaining wet ) = 7 C 3 q 3 P 4 = 35(0.4) 3 (0.6) 4 = 0.2903
Q.8.7: An insurance salesman sells policies to 5 men all of identical age and in good
health. According to the actuarial tables, the probability that a men of this particular age
will be alive 30 years hence 2/3. Find the probability that in 30 years i) All men ii) At
least 3 men iii) Only two men iv) At most one man will be alive
2 1
Solution: n = 5 P= = 0.67 q= = 0.33
3 3
Q.8.10(b): An irregular six-faced die is thrown and the expectation that in 10 throws it
will give five even numbers is twice the expectation that it will give four even numbers.
How many times 10000 sets of 10 throws would you expect it to give no even number?
Solution: Given that n = 10 N = 10000
Expectation of 5 even numbers is = N (10 C 5 )P 5 q 5
Expectation of 4 even numbers is = N(10 C 4 )P 4 q 6
Now according to conditions
NP ( X = 5) = 2 NP ( X = 4)
N (10 C 5 )P 5 q 5 = 2 N (10 C 4 )P 4 q 6
252 P 5 q 5 = 2(210) P 4 q 6
P4q6 q
0.6 = =
P5q5 P
q
0 .6 =
P
0.6 P = q
0 .6 P = 1 − P
0 .6 P + P = 1
1 .6 P = 1
1
P= = 0.625 q = 1 − P = 0.375
1 .6
Expectation of no even numbers is = NP ( X = 0) = 10000(10 C 0 )(0.625) 0 (0.375)10 = 0.55 = 1
Q.8.17: Let “X” be a random variable having a binomial distribution with
Parameters n = 25 and p = 0.2 . Evaluate P ( X < µ − 2σ )
Solution: As we know that
mean = nP = 25(0.20) = 5 And Var ( X ) = nPq = 25(0.20)(0.8) = 4 S .D ( X ) = 2
P ( X < µ − 2σ ) = P ( X < 5 − 2(2)) = P ( X < 5 − 4) = P ( X < 1)
x 10 − x
1 1
b) Given binomial probability function P ( X = x) = ( C x ) 10
Find the median
2 2
and mode of the distribution.
1 1
Solution: Given that n = 10 P= q=
2 2
1
As we know that when p = q = then binomial distribution will be symmetrical
2
1
So we get Mean = Median = Mode = np = 10 × = 5
2
Prepared by: M.Riaz M.sc: Statistics I.U.B (2006-8): Lecturer NICAAS college R.Y.K 12

12
1 3 t
Q.8.23 (b): If the m.g.f of “X” is M 0 (t ) = + e , find E(X), Var(X) and
4 4
P ( X ≥ 10)
Solution: As we know that m.g.f of binomial distribution
[
M 0 (t ) = q + pe t ]
n
(A)
12
1 3 t
M 0 (t ) = + e (B)
4 4
Comparing eq (A) and (B) then we get
n = 10 P = 3 q= 1
4 4
1
Question: Show that if two symmetrical binomial distributions ( P = q = ) of degree
2
“n” (and of the same number of observations) are so superposed, the rth term of the
one coincide with rth term of the other, the distribution formed by adding
superposed term is symmetrical binomial of degree (n+1).
Solution:
For symmetrical distribution
1
P=q=
2
n
1 1
Binomial probability frequency distribution of degree “n” is N +
2 2
Let
(
Tr = rth term = N n C r -1 ( p ) (q )
r −1 n − r +1
)
r −1 n − r +1

Tr = rth term = N ( n
C r -1 ) 12 1
2
r −1+ n − r +1 n
1 1
Tr = rth term = N C r -1 ( n
) 2
= N C r -1 ( n
) 2
Now
Tr +1 = (r + 1)th term = N ( n C r )( p ) (q )
r n−r

r n− r
1 1
Tr +1 = (r + 1)th term = N C r ( n
) 2 2
r + n−r n

Tr +1 = (r + 1)th term = N n C r ( ) 12 = N n Cr ( ) 12
Then according to condition
n n
1 1
(
Tr + Tr +1 = N C r -1n
) 2
+ N Cr ( n
) 2
n
1
Tr + Tr +1 = N
2
[( n
C r -1 + ) ( n
Cr )]
n
1 n! n!
Tr + Tr +1 =N +
2 (r − 1)!(n − r + 1)! r!(n − r )!
n
1 n! n!
Tr + Tr +1 =N +
2 (r − 1)!(n − r + 1)(n − r )! r (r − 1)!(n − r )!
n
1 n! 1 1
Tr + Tr +1 = N +
2 (r − 1)!(n − r )! (n − r + 1) r
n
1 n! r + n − r +1
Tr + Tr +1 =N
2 (r − 1)!(n − r )! (n − r + 1)r
n
1 n! n +1
Tr + Tr +1 =N
2 (r − 1)!(n − r )! (n − r + 1)r
n
1 (n + 1)n!
Tr + Tr +1 =N
2 r (r − 1)!(n − r + 1)(n − r )!
Prepared by: M.Riaz M.sc: Statistics I.U.B (2006-8): Lecturer NICAAS college R.Y.K 13

n
1 (n + 1)!
Tr + Tr +1 = N
2 r!(n − r + 1)!
n
1 n +1
Tr + Tr +1 = N c
2 r

n
1 n +1 2
Tr + Tr +1 =N c
2 r 2
n
1 n +1 1
Tr + Tr +1 = 2 N c
2 r 2
n +1
n +1 1
Tr + Tr +1 = 2 N c
r 2
rr n +1− r
n +1 1 1
Tr + Tr +1 = 2 N c
r 2 2
n +1
1 1
Tr + Tr +1 = 2N +
2 2
n +1
1 1
Tr + Tr +1 =(r+1)th term of binomial distribution 2 N + symmetrical frequency
2 2
binomial probability distribution of degree (n+1).
Question: Find cummulent generating function of binomial distribution.
Proof:
Let by definition
K (t ) = log M x (t )
As we know that
M x (t ) = (q + Pe t ) n
K (t ) = log(q + Pe t ) n
K (t ) = n log(q + Pe t )
K (t ) = n log(1 − p + Pe t )
We know that
t 0 t1 t 2
et = + + + ....
0! 1! 2!
t 0 t1 t 2
K (t ) = n log 1 − p + p( + + + ....)
0! 1! 2!
t1 t 2
K (t ) = n log 1 − p + P(1 + + + ....)
1! 2!
t1 t 2
K (t ) = n log 1 − p + p + P( + + ....)
1! 2!
t1 t 2
K (t ) = n log 1 + P( + + ....)
1! 2!
Since
t2 t3 t4
log(1 + t ) = t −+ − + ....
2 3 4
t1 t 2 t 3 t 4 p 2 t1 t 2 t 3 p3 t1 t 2 t 3 p 4 t1 t 2 t 3
K (t ) = n P( + + + + ..) − ( + + + ..) 2 + ( + + + ...) 3 − ( + + + ...) 4
1! 2! 3! 4! 2 1! 2! 3! 3 1! 2! 3! 4 1! 2! 3!
t1 t 2 t 3 t 4 p 2 t 2 t 4 2t 3 2t 4 p 3 t 3 3t 4 p4 t 4
K (t ) = n P( + + + + ..) − ( + + + ..) + ( + + ...) − ( + ...)
1! 2! 3! 4! 2 1! 4! 2! 3! 3 1! 2! 4 1!
t1 t2 t3 t4
K (t ) = n P + ( p − p 2 ) + ( p − 3 p 2 + 2 p 3 ) + ( p − 7 p 2 + 12 p 3 − 6 p 4 ) (A)
1! 2! 3! 4!
As we know that general expression of cummulent
t1 t2 t3 t4
K (t ) = K 1 + K 2 + K 3 + K 4 + .... (B)
1! 2! 3! 4!
Prepared by: M.Riaz M.sc: Statistics I.U.B (2006-8): Lecturer NICAAS college R.Y.K 14

Comparing (A) and (B), then we get

K 1 = np = µ1′ = Mean
K 2 = n( p − p 2 ) = np(1 − p) = npq = µ 2 = Var ( x)
K 3 = n( p − 3 p 2 + 2 p 3 ) = np (1 − p − 2 p (1 − p )) = np (q − 2 pq ) = npq (1 − 2 p ) = npq (q − p ) = µ 3
K 4 = n( p − 7 p 2 + 12 p 3 − 6 p 4 ) = np(1 − 7 p + 12 p 2 − 6 p 3 ) = np(1 − p − 6 p + 12 p 2 − 6 p 3 )
K 4 = np(q − 6 p(1 − 2 p + p 2 ) = np(q − 6 p(1 − p) 2 ) = np(q − 6 pq 2 ) = npq(1 − 6 pq)
µ 4 = k 4 + 3k 22 = npq(1 − 6 pq) + 3(npq) 2

Theorem: If n → ∞ then binomial distribution approaches to normal distribution.

Proof: We know in binomial distribution
X → b ( x , n, p )
Mean = np = µ ′
Var ( x) = np(1 − p) = npq = µ 2 = σ 2
S .D( x) = npq = σ
Now we consider standardized normal variate
X − E( X ) X − µ
Z= =
S .D( X ) σ
X − np
Z=
npq
Let by definition of moment generating function
M z (t ) = E (etZ )
X − np
t( )
npq
M z (t ) = E (e )
X np
t −t ( )
npq npq
M z (t ) = E (e )
tX np
− t
npq npq
M z (t ) = E (e e )
np tX
− t
npq npq
M z (t ) = e E (e ) (A)
In binomial distribution “m.g.f” is
M x (t ) = (q + Pe t ) n
t
Replacing “t” by npq
tx t
npq npq
E (e ) = (q + Pe )n Put in (A)
np t
− t
M z (t ) = e npq (q + Pe npq ) n
Taking log on both sides
np t t
− t
npq npq n − npt npq
ln M z (t ) = ln e (q + Pe ) = ln e + n ln(q + Pe )
npq
t
− npt npq
ln M z (t ) = + n ln(q + Pe )
npq
t 0 t1 t 2
et = + + + ....
0! 1! 2!
t t t
( )0 ( )1 ( )2
− npt npq npq npq
ln M z (t ) = + n ln(1 − p + p ( + + + ....))
npq 0! 1! 2!
− npt t t2
ln M z (t ) = + n ln(1 − p + p(1 + + + ....))
npq npq 2npq
− npt t2 t
ln M z (t ) = + n ln(1 − p + p + p( + + ....))
npq npq 2npq
Prepared by: M.Riaz M.sc: Statistics I.U.B (2006-8): Lecturer NICAAS college R.Y.K 15

− npt t t2
ln M z (t ) = + n ln(1 + p( + + ....))
npq npq 2npq
Since
t2 t3 t4
log(1 + t ) = t − + − + ....
2 3 4
− npt t2
t p2 t t2 2 p3
ln M z (t ) = + n p( + + ....) − ( + + ....) + (...) 3 − ...
npq npq 2npq 2 npq 2npq 3
− npt npt npt 2 np 2 t 2 1
ln M z (t ) = + + − + .....involving ( ) and higher term
npq npq 2npq 2npq n
t 2 pt 2 1
ln M z (t ) = − + .....involving ( ) and higher term
2q 2 q n
t2 1 p 1
ln M z (t ) = ( − ) + .....involving ( ) and higher term
2 q q n
t2 1− p 1
ln M z (t ) = ( ) + .....involving ( ) and higher term
2 q n
t2 q 1
ln M z (t ) = ( ) + .....involving ( ) and higher term
2 q n
t2 1
ln M z (t ) = + .....involving ( ) and higher term
2 n
1
Applying limit n → ∞ then all the term equal to zero involving ( ) and higher term
n
2
t 1
ln M z (t ) = + n → ∞.....involving ( ) and higher term
2 n
2
t
ln M z (t ) = + 0 + 0 + 0 + ...
2
t2
ln M z (t ) =
2
Taking antilog
t2
M z (t ) = e2

It is m.g.f of standard normal distribution. Hence its proved when n → ∞ then binomial
distribution approaches to normal distribution.
b( X , n, p ) → N (np, npq ) When n → ∞

Prepared by: Muhammad.Riaz

M.sc: Statistics (Silver medalist)
I.U.B (2006-08)
Lecturer: NICAAS College R.Y.K
03337368518