CHAPTER 9

Normal approximation to the binomial

A special case of the central limit theorem is the following statement.

Theorem 9.1 (Normal approximation to the binomial distribution)

If Sn is a binomial variable with parameters n and p, Binom (n, p), then
!
Sn − np
P a6 p 6b −−−→ P(a 6 Z 6 b),
np(1 − p) n→∞

as n → ∞, where Z ∼ N (0, 1).

This approximation is good if np(1 − p) > 10 and gets better the larger this quantity gets.
This means that if either 1 − p is small, then this is valid for large n. Recall that by
p or
Proposition 6.1 np is the same as ESn and np(1 − p) is the same as Var Sn . So the ratio
√
is equal to (Sn − ESn )/ Var Sn , and this ratio has mean 0 and variance 1, the same as a
standard N (0, 1).

Note that here p stays xed as n → ∞, unlike in the case of the Poisson approximation, as
we described in Proposition 6.3.

Sketch of the proof. This is usually not covered in this course, so we only explain
one (of many) ways to show why this holds. We would like to compare the distribution of
 p 
Sn with the distribution of the normal variable X ∼ N np, np (1 − p) . The random

variable X has the density

1 (x−np)2
p e− 2np(1−p) .
2πnp (1 − p)
The idea behind this proof is that we are interested in approximating the binomial dis-
tribution by the normal distribution in the region where the binomial distribution diers
signicantly from zero, that is, in the region around the mean np. We consider P (Sn = k),
and we assume that k does not deviate too much from
p np. We measure deviations by some
small number of standard deviations, which
√ is np (1 − p). Therefore we see that k − np
should be of order n. This is not much of a restriction since once k deviates from np by
many standard deviations, P (Sn = k) becomes very small and can be approximated by zero.
In what follows we assume that k and n − k of order n.
We use Stirling's formula is the following form

√
m! ∼ 2πme−m mm ,
121
122 9. NORMAL APPROXIMATION

where by ∼ we mean that the two quantities are asymptotically equal, that is,their ratio
tends to 1 as m → ∞. Then for large n, k and n−k

n!
P (Sn = k) = pk (1 − p)n−k
k! (n − k)!
√
2πne−n nn
∼√ p n−k
pk (1 − p)n−k
−k k
2πke k 2π (n − k)e −(n−k) (n − k)
 p k 1 − p n−k
  r
n  np k  n (1 − p) n−k r n
n
= n = .
k n−k 2πk (n − k) k n−k 2πk (n − k)
Now we can use identities

 
 np  k − np
ln = − ln 1 + ,
k np
   
n (1 − p) k − np
ln = − ln 1 − .
n−k n (1 − p)
y2 y3
Then we can use ln (1 + y) ∼ y − 2
+ 3
,y →0 to see that

!
 np k  n (1 − p) n−k  np  
n (1 − p)

ln = k ln + (n − k) ln
k n−k k n−k
 2  3 !
k − np 1 k − np 1 k − np
∼k − + −
np 2 np 3 np
 2  3 !
k − np 1 k − np 1 k − np
+ (n − k) + +
n (1 − p) 2 n (1 − p) 3 n (1 − p)
(k − np)2
∼− .
2np (1 − p)
Thus
 np k  n (1 − p) n−k (k−np)2
∼ e− 2np(1−p) .
k n−k
√
Now we use our assumption that k − np should be of order n to see that

√
k − np ≈ n,
√
n − k ≈ n (1 − p) − n,
k (n − k) ≈ n2 p (1 − p) ,
so
 9. NORMAL APPROXIMATION 123

r
n 1
∼p .
2πk (n − k) 2πnp (1 − p)

Example 9.1. Suppose a fair coin is tossed 100 times. What is the probability there will
be more than 60 heads?

Solution : np = 50 and
p
np(1 − p) = 5. We have

P(Sn > 60) = P((Sn − 50)/5 > 2) ≈ P(Z > 2) ≈ 0.0228.

Example 9.2. Suppose a die is rolled 180 times. What is the probability a 3 will be
showing more than 50 times?

Solution :
p
Here p = 16 , so np = 30 and np(1 − p) = 5. Then P(Sn > 50) ≈ P(Z > 4),
−42 /2
which is less than e .

Example 9.3. Suppose a drug is supposed to be 75% eective. It is tested on 100 people.
What is the probability more than 70 people will be helped?

Solution : Here Sn is the number of successes, n = 100, and p = 0.75. We have

p
P(Sn > 70) = P((Sn − 75)/ 300/16 > −1.154)
≈ P(Z > −1.154) ≈ 0.87.

(The last gure came from a table.)

When b − a is small, there is a correction that makes things more accurate, namely replace
a by a − 21 and b by b + 12 . This correction never hurts and is sometime necessary. For
example, in tossing a coin 100 times, there is positive probability that there are exactly 50
heads, while without the correction, the answer given by the normal approximation would
be 0.

Example 9.4. We toss a coin 100 times. What is the probability of getting 49, 50, or 51
heads?

Solution : We write P(49 6 Sn 6 51) = P(48.5 6 Sn 6 51.5) and then continue as above.
124 9. NORMAL APPROXIMATION

In this case we again have

p = 0.5,
µ = np = 50,
σ 2 = np(1 − p) = 25,
p
σ = np(1 − p) = 5.
The normal approximation can be done in three dierent ways:

P(49 6 Sn 6 51) ≈ P(49 6 50 + 5Z 6 51) = Φ(0.2) − Φ(−0.2) = 2Φ(0.2) − 1 ≈ 0.15852

or

P(48 < Sn < 52) ≈ P(48 < 50 + 5Z < 52) = Φ(0.4) − Φ(−0.4) = 2Φ(0.4) − 1 ≈ 0.31084
or

P(48.5 < Sn < 51.5) ≈ P(48.5 < 50+5Z < 51.5) = Φ(0.3)−Φ(−0.3) = 2Φ(0.3)−1 ≈ 0.23582
Here all three answers are approximate, and the third one, 0.23582, is the most accurate
among these three. We also can compute the precise answer using the binomial formula:

51  100
X
100

1 37339688790147532337148742857
P(49 6 Sn 6 51) = k
=
k=49
2 158456325028528675187087900672
≈ 0.2356465655973331958...

In addition we can obtain the following normal approximations

P(Sn = 49) ≈ P(48.5 6 50 + 5Z 6 49.5) = Φ(−0.1) − Φ(−0.3) = Φ(0.3) − Φ(0.1) ≈ 0.07808

P(Sn = 50) ≈ P(49.5 6 50 + 5Z 6 50.5) = Φ(0.1) − Φ(−0.1) = 2Φ(0.1) − 1 ≈ 0.07966
P(Sn = 51) ≈ P(50.5 6 50 + 5Z 6 51.5) = Φ(0.3) − Φ(0.1) ≈ 0.07808
Finally, notice that
0.07808 + 0.07966 + 0.07808 = 0.23582
which is the approximate value for P(49 6 Sn 6 51) ≈ P(48.5 < 50 + 5Z < 51.5).

Continuity correction
If a continuous distribution such as the normal distribution is used to approximate a
discrete one such as the binomial distribution, a continuity correction should be used.

For example, if X is a binomial random variable that represents the number of successes in
n independent trials with the probability of success in any trial p, and Y is a normal random
variable with the same mean and the same variance as X . Then for any integer k we have
that P (X 6 k) is well approximated by P (Y 6 k) if np (1 − p) is not too small. It is better
approximated by P (Y 6 k + 1/2) as explained at the end of this section. The role of 1/2
is clear if we start by looking at the normal distribution rst, and seeing how we use it to
approximate the binomial distribution.
 9. NORMAL APPROXIMATION 125

The fact that this approximation is better based on a couple of considerations. One is that
a discrete random variable can only take on only discrete values such as integers, while a
continuous random variable used to approximate it can take on any values within an interval
around these specied values. Hence, when using the normal distribution to approximate the
binomial, more accurate approximations are likely to be obtained if a continuity correction
is used.

The second reason is that a continuous distribution such as the normal, the probability of
taking on a particular value of a random variable is zero. On the other hand, when the
normal approximation is used to approximate a discrete distribution, a continuity correction
can be employed so that we can approximate the probability of a specic value of the discrete
distribution.

For example, if we want to approximate P (3 6 X 6 5) = P (X = 3 or X = 4 or X = 5) by

a normal distribution, it would be a bad approximation to use P (Y = 3 or Y = 4 or Y = 5)
as the probability of Y taking on 3, 4 and 5 is 0. We can use continuity correction to see
that

P (3 6 X 6 5) = P (2.5 6 X 6 5.5)
and then use the normal approximation by P (2.5 6 Y 6 5.5).

Below is a table on how to use the continuity correction for normal approximation to a
binomial.

Binomial Normal

If P (X = n) use P (n − 0.5 < X < n + 0.5)

If P (X > n) use P (X > n + 0.5)

If P (X 6 n) use P (X < n + 0.5)

If P (X < n) use P (X < n − 0.5)

If P (X > n) use P (X > n − 0.5)

126 9. NORMAL APPROXIMATION

9.1. Exercises
Exercise 9.1. Suppose that we roll 2 dice 180 times. Let E be the event that we roll two
ves no more than once.

(a) Find the exact probability of E.

(b) Approximate P(E) using the normal distribution.
(c) Approximate P(E) using the Poisson distribution.

Exercise 9.2. About 10% of the population is left-handed. Use the normal distribution
to approximate the probability that in a class of 150 students,

(a) at least 25 of them are left-handed.

(b) between 15 and 20 are left-handed.

Exercise 9.3. A teacher purchases a box with 50 markers of colors selected at random.
The probability that marker is black is 0.6, independent of all other markers. Knowing
that the probability of there being more than N black markers is greater than 0.2 and the
probability of there being more than N + 1 black markers is less than 0.2, use the normal
approximation to calculate N.
 9.2. SELECTED SOLUTIONS 127

9.2. Selected solutions

Solution to Exercise 9.1(A): The probability of rolling two ves in a particular roll is 1
36
,
so the probability that we roll two ves no more than once in 180 rolls is
   
180    179
 180   180  1

35 35
p=  +  ≈ .0386.
  36   36 36
0 1

Solution to Exercise 9.1(B): we are interested in the number of successes to be 0 or

1, that is,P (0 6 S180 6 1). Since the binomial is integer-valued, we apply the continuity
correction and calculate P (−0.5 6 S180 6 1.5) instead. We nd that the expected value is
p
µ = 180 · p = 5 and the standard deviation is σ = 180p(1 − p) ≈ 2.205. Now, as always,
we convert this question to a question about the standard normal random variable Z ,
 
−0.5 − 5 1.5 − 5
P (−0.5 6 S180 6 1.5) = P 6Z6 = P (−2.49 < Z < −1.59)
2.205 2.205
= (1 − Φ (1.59)) − (1 − Φ (2.49))
= (1 − 0.9441) − (1 − 0.9936) = 0.0495.

Solution to Exercise 9.1(C): We use λ = np = 5 (note that we found this already in

(b)!). Now we see that

50 51
P(E) ≈ e−5 + e−5 ≈ 0.0404.
0! 1!
Solution to Exercise 9.2: Let X denote the number of left-handed students in the class.
We use Theorem 9.1 with X ∼ Binom(150, 0.1) below. Note that np = 15.
 
X−15
(a) P(X > 25) = P √
13.5
> √10
13.5
≈ 1−Φ(2.72) ≈ 0.00364. Note that in this approximation
we implicitly use that the tail of this probability distribution is small, and so instead of a
two-sided interval we just used a one-sided interval.

We can see that the result is really close to the two-sided estimates as follows.
 
135 X − 15 10
P(150 > X > 25) = P √ > √ >√
13.5 13.5 13.5
≈ Φ(36.74) − Φ(2.72) ≈ 0.00364.

Finally, with the continuity correction the solution is

P(150 > X > 25) = P(150.5 > X > 24.5)

 
135.5 X − 15 9.5
=P √ > √ >√ ≈ Φ(36.87) − Φ(2.59) ≈ 0.00480.
13.5 13.5 13.5

(2) Similarly to the rst part

128 9. NORMAL APPROXIMATION

P(15 6 X 6 20) = P(14.5 < X < 20.5)

   
5.5 −0.5
=Φ √ −Φ √ ≈ Φ(1.5) − 1 + Φ(0.14) ≈ 0.4889.
13.5 13.5
Solution to Exercise 9.3: Let X denote the number of black markers. Since X ∼
Binom(50, 0.6) we have
   
N − 30 N − 29
P(X > N ) ≈ 1 − Φ √ > 0.2 and P(X > N + 1) ≈ 1 − Φ √ < 0.2.
2 3 2 3
From this we deduce that N 6 32.909 and N > 31.944 so that N = 32.