Central Limit theorem

Dr. Tanmoy Chakraborty

Assistant Professor,
Department of Mathematics,
SRM Institute of Science and Technology,
Kattankulathur - 603 203, Tamil Nadu, India.

(SRM IST, Ktr) Central Limit theorem 1 / 16

The Central Limit Theorem (CLT) is a fundamental concept in probability theory and statistics.
It states that, under certain conditions, the distribution of the sum (or average) of a large number
of independent, identically distributed random variables will be approximately normally distributed,
regardless of the original distribution of the variables themselves.

Here are the key points and implications of the Central Limit Theorem:

Independent and Identically Distributed (I.I.D.) Random Variables: The CLT applies to
a set of random variables that are independent and identically distributed. This means
that each random variable is drawn from the same distribution and is not influenced by
any other random variable in the set.
Sample Size: The CLT holds particularly well as the sample size increases. The larger the
sample size, the closer the distribution of the sample mean (or sum) will be to a normal
distribution.
Approximation: The CLT allows us to approximate the distribution of the sample mean
(or sum) even when the original distribution of the individual random variables is not
normal. This is extremely powerful because the normal distribution is well understood and
has many convenient properties for statistical inference.
Normal Distribution: According to the CLT, as the sample size tends to infinity, the
distribution of the sample mean (or sum) becomes exactly normal, regardless of the
original distribution of the individual random variables. For practical purposes, even
relatively small sample sizes often result in a good approximation to the normal
distribution.

(SRM IST, Ktr) Central Limit theorem 2 / 16

Central Limit theorem

Central Limit Theorem (Linderberg-Levy’s Form)

If X1 , X2 , · · · , Xn , · · · be a sequence of independent identically distributed RVs with
E (Xi ) = µ and Var (Xi ) = σ 2 , i = 1, 2, · · · and if Sn = X1 + X2 + · · · + Xn , then under certain
general conditions, Sn follows a normal distribution with mean nµ and variance nσ 2 as n
tends to infinity.

Property 1:
1 1 σ2
If X̄ = n
+ X2 + · · · + Xn ), then E (X̄ ) = µ and Var (X̄ ) = 2 (nσ 2 ) =
(X1
  n n
∴ X̄ follows N µ, √σn as n → ∞

(SRM IST, Ktr) Central Limit theorem 3 / 16

Worked Problem 1

The lifetime of a certain brand of an electric bulb may be considered a RV with mean 1200 h
and Standard deviation 250 h. Find the probability, using central limit theorem, that the average
lifetime of 60 bulbs exceeds 1250 h.

Solution:
Let Xi represent the lifetime of the bulb.

E (Xi ) = 1200 and Var (Xi ) = 2502

Let X̄ denote the mean lifetime of 60 bulbs.  
250
By Property 1 of Lindeberg-Levy form of CLT, X̄ follows N 1200, √
60

 
X̄ − 1200 1250 − 1200
P(X̄ > 1250) = P  250
> 250

√ √
60 60
√ !
60
=P z >
5
= P(z > 1.55),
where z is the standard normal variable
= 0.0606 (from the table of areas under normal curve)

(SRM IST, Ktr) Central Limit theorem 4 / 16

Worked Problem 2
A distribution with unknown µ has variance equal to 1.5. Use central limit theorem to find how
large a sample should be taken from the distribution in order that the probability will be at least
0.95 that the sample mean will be within 0.5 of the population mean.

Solution: Let n be the size of the sample, a typical number of which is Xi

Given:E (Xi ) = µ and Var (Xi ) = 1.5.
Let X̄ denote the sample mean. By Corollary under CLT,
√ !
1.5
X̄ follows N µ, √
n

We have to find n such that

P{µ − 0.5 < X̄ < µ + 0.5} ≥ 0.95
i.e., P{−0.5 < X̄ − µ < 0.5} ≥ 0.95
i.e., P{|X̄ − µ| < 0.5} ≥ 0.95
 

 |X̄ − µ| 
0.5 
i.e., P q < q ≥ 0.95

 1.5 1.5 

n n
√
i.e., P{|z| < 0.4082 n} ≥ 0.95
where z is the standard normal variable.
(SRM IST, Ktr) Central Limit theorem 5 / 16
Worked Problem 2 Cntd.,

The least value of n is obtained from

√
P{|z| < 0.4082 n} = 0.95

From the table of areas under normal curve

P{|z| < 1.96} = 0.95

√
Therefore, least n is given by 0.4082 n = 1.96,
i.e., least n = 24.
Therefore the size of the sample must be at
least 24.

(SRM IST, Ktr) Central Limit theorem 6 / 16

 Worked Problem 3

If X1 , X2 , · · · , Xn are Poisson variates with parameter λ = 2, use the central limit theorem to
estimate P(120 ≤ Sn ≤ 160), where Sn = X1 + X2 + · · · + Xn and n = 75.

Solution: E (Xi ) = λ = 2andVar (Xi ) = λ = 1.5

√
By CLT, Sn follows N(nµ, σ n)
√
ie., Sn follows N(150, 150)
 
−30 Sn − 150 10
P{120n ≤ 160} = P √ ≤ √ ≤ √
150 150 150
= P{−2.45 ≤ z ≤ 0.85}
where z is the standard normal variable.
= 0.4927 + 0.2939 (from the normal tables)
= 0.7866

(SRM IST, Ktr) Central Limit theorem 7 / 16

Worked Problem 4

Suppose you roll a fair six-sided die 100 times and calculate the mean of the outcomes. What is
the approximate distribution of the sample mean according to the central limit theorem?

Solution:
Each roll a fair six-sided die follows a discrete distribution with

1+6
mean of µ = = 3.5
2
s r
(6 − 1 + 1)2 − 1 35
and a standard deviationσ = = ≈ 1.71
12 12
We are rolling the die 100 times, so according to the CLT, the sample mean will be approximately
normally distributed with mean µx = µ = 3.5 and standard deviation σx = √σn = √ 1.71
= 0.171
100
Therefore, the approximate distribution of the sample mean is

X̄ ∼ Normal(3.5, 0.1712 )

(SRM IST, Ktr) Central Limit theorem 8 / 16

Worked Problem 5

Suppose you have a population with mean (µ) of 50 and standard deviation (σ) of 10. You take
a random sample of size 100 from this population. What are the mean and S.D. of the sampling
distribution of the sample mean?

Solution:
According to CLT, the mean of the sampling distribution of the sample mean will be equal to the
population mean which is 50 in this case.
The S.D. of the sampling distribution of the sample mean is given by
σ
σ= √
n
where σ - population S.D and n - sample size
Substituting the given values,
10
σ= √ =1
100
So the mean of the sampling distribution of the sample mean is 50 and S.D. is 1.

(SRM IST, Ktr) Central Limit theorem 9 / 16

Worked Problem 6

Let X1 , X2 , · · · , X25 be i.i.d with the following PMF


 0.6 if k = 1
PX (k) = 0.4 if k = −1
0 if otherwise


And let Y = X1 + X2 + · · · + Xn , Using the CLT and continuity correction, estimates

P(4 ≤ Y ≤ 6)

Solution:
We have

EXi = (0.6)(1) + (0.4)(−1)

1
=
5
EXi2 = (0.6) + (0.4)
=1

(SRM IST, Ktr) Central Limit theorem 10 / 16

Therefore,

1 24
Var (Xi ) = 1 − =
25 25
√
2 6
thus σXi =
5
Therefore,

1
EY = 25 × =5
5
24
Var (Y ) = 25 × = 24
25
√
thus σY = 2 6

P(4 ≤ Y ≤ 6) = P(3.5 ≤ Y ≤ 6.5) (continuity correction)

 
3.5 − 5 6.5 − 5
=P √ ≤Y ≤ √
2 6 2 6
 
Y −5
= P −0.3062 ≤ √ ≤ 0.3062
2 6
≈ ϕ(0.3062) − ϕ(−0.3062) by the CLT
= 2ϕ(0.3062) − 1
≈ 0.2405

(SRM IST, Ktr) Central Limit theorem 11 / 16

Worked Problem 7

The guaranteed average life of a certain type of electric light bulbs is 1000
hours with a standard deviation of 125 hours. It is proposed to sample the
output so as to assure that 90% of the bulbs do not fall short of the
guaranteed average by more than 2.5 %. What should be the minimum
size of the sample? (The area under standard normal curve from z = 0 to
z = 1.28 is 0.4000)
Solution:
Given: µ = 1000, σ = 125. Let n be the size of the sample.
Now, ”Short of the guaranteed average by more than 2.5%” = (x̄ > 975).
9
By hypothesis, the probability of this event = 90% = 10
So,
9
P(x̄ > 975) = (1)
10
σ 125
The sampling distribution of x̄ is normal distribution with mean = µ = 1000 and S.d. = √
n
= √
n

(SRM IST, Ktr) Central Limit theorem 12 / 16

 √
x̄ − 1000 n(x̄ − 1000)
So, z = 125
= is a
√
n
125
standard normal variate.
√ √
n(975 − 1000) n
When x̄ = 975, z = =−
 125
√  5
So, P(x̄ > 975) = P z > − 5n
 √   √ 
From (1), P z > − 5n = 10 9
= 0.9 or 1 − P z < − 5n = 0.9
 √ 
or P z < − 5n = 1 − 0.9 = 0.1
Bysymmetry
√ 
of normal curve,
 √ 
P z > 5n = 0.1 or P 0 < z < 5n = 0.5 − 0.1 = 0.4
√
n
So the area under standard normal curve enclosed between z = 0 and z = 5
is 0.4.

(SRM IST, Ktr) Central Limit theorem 13 / 16

Worked Problem 8
Assume that 3000 male students of a university are normally distributed with mean 68.0 inches
and standard deviation 3.0 inches. 80 samples consisting of 25 students each are obtained. In
how many samples would you expect to find the mean between 66.8 and 68.3 inches.
Given area under standard normal curve enclosed between z = −2 and z = 0 is 0.4772 and
between z = 0 and z = 5 is 0.1915

Solution:
The sample mean X̄ is normally distributed with mean = population mean = 68 and
σ 3
S.d. = √ = √ = 0.6
n 25
X̄ − 68
So, z = is standard normal variate.
0.6
66.8 − 68
When X̄ = 66.8, z = = −2
0.6
68.3 − 68
When X̄ = 68.3, z = = 0.5. Therefore, Probability of X̄ lying between 66.8 and 68.3
0.6

= P(66.8 < X̄ < 68.3)

= P(−2 < z < 0.5)
= P(−2 < z < 0) + P(0 < z < 0.5)
= 0.4772 + 0.1915 = 0.6687
So, the expected number of required type of samples
= 80 × 0.6687 = 53.496 ≈ 53

(SRM IST, Ktr) Central Limit theorem 14 / 16

Worked Problem 9

Distribution of marks scored in an examination is normal. Samples of four students are drawn
and it is seen that the probability of the sample mean to be less than 61 is 0.44, to be more
than 80R is 0.04. Find the mean and S.D of the distribution.
Given 0z ϕ(t)dt = 0.06, 0.10, 0.46 according as z = 0.15, 0.25 and 1.75

Solution:
Let mean = µ and S.D. = σ of the population.
Since the population is normal so the sampling distribution of x̄ (sample mean) is normal with
mean = population mean = µ and S.D. = √σn = √σ = σ2
4
By Problem

P(x̄ < 61) = 0.44 (2)

P(x̄ > 80) = 0.04 (3)

x̄ − µ 2(x̄−µ)
Now, z = σ = σ
has standard normal distribution.
2
2(61 − µ)
When x̄ = 61, z =
σ
2(80 − µ)
When x̄ = 80, z =
σ

(SRM IST, Ktr) Central Limit theorem 15 / 16

∴ From (2), we have ,
 
2(61 − µ)
P z< = 0.44
σ
 
2(61 − µ)
P < z < 0 = 0.50 − 0.44 = 0.6
σ
 
2(61 − µ)
P 0<z<− = 0.50 − 0.44 = 0.6
σ
 
2(µ − 61)
P 0<z<− = 0.50 − 0.44 = 0.6 (4)
σ

We are given a data

P(0 < z < 0.15) = 0.6 (5)
From (4) and (5) we have

2(µ − 61)
= .15
σ
or 2(µ − 61) = 0.15σ (6)

Similarly from (3) and from other hypothesis we have

2(80 − µ) = 1.75σ (7)

Solving (6) and (7) we get µ = 62.5 and σ = 20

(SRM IST, Ktr) Central Limit theorem 16 / 16