Mr.

Mohamed El-Sayed El-Dawoody Lecturer of Mathematical Statistics

LECTURE 3

Confidence Intervals and sample size

Introduction
 In lecture 2, we studied the point and interval estimates for the population mean when the
population standard deviation is known and when it is unknown.
 The point and interval estimates can be studied (extended) to other population parameters
that are important as the population mean.
 In this lecture, we will study and explain how to estimate the population proportion and
also, the population variance (standard deviation).
 We will estimate the confidence intervals for the population proportion using z distribution
and for the population variance using a new distribution called "chi-square distribution".
 Different problems and applications will be presented, in this lecture, to cover these two cases.
3. Confidence Intervals for the population proportion
 Many statistical studies involve finding a proportion of the population that has a certain
characteristic. It can be expressed as a fraction, decimal, or percentage.
 The population proportion is denoted by p and is given by the formula
X
p ,
N
where,
 N  The population size (always very large or infinite).
 X  The number of elements in the population with a certain characteristic.
 The best point estimate of the population proportion is the sample proportion taken from
it. It is denoted by p̂ and is given by the formula
X
pˆ  ,
n
where,
 n  The sample size (always finite and known).
 X  The number of elements in the sample with a certain characteristic.
 The complement of the sample proportion is denoted by qˆ and is given by the formula
n X
qˆ  .
n

LECTURE 3 PAGE 1 STAT 2040

Mr. Mohamed El-Sayed El-Dawoody Lecturer of Mathematical Statistics

 When p̂ and qˆ are given in decimals or fractions, then pˆ + qˆ = 1 . Also, when p̂ and qˆ are
given in percentages, then pˆ + qˆ = 100% . That is, qˆ can be written as qˆ = 1 - pˆ .
 The confidence intervals for the population proportion (p) are given by the formula
pˆ qˆ pˆ qˆ
pˆ  z  .  p  pˆ  z  . ,
2
n 2
n
where,
 n  The sample size.
 pˆ  The sample proportion.
 qˆ  The complement of the sample proportion
 z  2  The standard value of the random variable X.
Remark
 There are two basic conditions must be satisfied to use the above formula are:
n . pˆ  0 and n . qˆ  0.
 The margin of error when estimating the population proportion is given by
pˆ qˆ
E z. .
2
n
 The confidence intervals for the population proportion (p) can be written as
pˆ qˆ
C.I  pˆ  z  . .
2
n
Example 1

A random sample of 200 workers in a large factory found that 128 drive to work alone. Find p̂ and
qˆ , where p̂ is the proportion of workers who drive to work alone?
Solution

Since, n  200, X  128.

Then, the sample proportions p̂ and qˆ are given as
X 128
pˆ    0.64  64%,
n 200
n  X 200  128 72
qˆ     0.36  36%.
n 200 200
Hence, 64% of the people in the survey drive to work alone, and 36% drive with others.
Example 2

A survey conducted by Sallie Mae and Gallup of 1404 respondents found that 323 students paid
for their college education by student loans. Find the best point estimate and the 90% confidence
interval of the true proportion of students who paid for their college education by student loans?

LECTURE 3 PAGE 2 STAT 2040

Mr. Mohamed El-Sayed El-Dawoody Lecturer of Mathematical Statistics

Solution

 Since, we have
n  1404, X  323, z  2  1.65.
Then, the best point estimate of the population proportion is given as
X 323
p  pˆ    0.23,
n 1404
Also,
qˆ  1  pˆ  1  0.23  0.77.
 The 90% confidence interval of the population proportion is given as
pˆ qˆ pˆ qˆ
pˆ  z  .  p  pˆ  z  .
2
n 2
n
(0.23)(0.77) (0.23)(0.77)
0.23  (1.65)  p  0.23  (1.65)
1404 1404
0.23  (1.65)(0.01123119)  p  0.23  (1.65)(0.01123119)
0.23  0.019  p  0.23  0.019
21.1%  p  24.9%.
Hence, we can be 90% confident that the true proportion of students who pay for their college
education by student loans is between 21.1% and 24.9%.
Example 3

In 2008, 17% of American homes were protected by a home security system. A marketing firm
wanted to estimate the proportion of protected homes today. It chose a random sample of 200
homes and discovered that 26.5% had home security systems. Estimate the true proportion of
homes with security systems with 99% confidence?
Solution

Since, we have
n  200, pˆ  0.265, qˆ  0.735, z  2  2.58.
The 99% confidence interval of the population proportion is given as
pˆ qˆ pˆ qˆ
pˆ  z  .  p  pˆ  z  .
2
n 2
n
(0.265)(0.735) (0.265)(0.735)
0.265  (2.58)  p  0.265  (2.58)
200 200
0.265  (2.58)(0.03120697)  p  0.265  (2.58)(0.03120697)
0.265  0.081  p  0.265  0.081
18.4%  p  34.6%.
Hence, we can be 99% confident that the true proportion of homes with security systems is
between 18.4% and 34.6%.

LECTURE 3 PAGE 3 STAT 2040

Mr. Mohamed El-Sayed El-Dawoody Lecturer of Mathematical Statistics

Sample Size
 There is an important question in estimation of the population proportion. How large should
the sample be in order to make an accurate estimate?
 To estimate the sample size that helps the researchers to make an accurate estimate of the
population proportion, we will use the margin of error formula as follows:
pˆ qˆ z  /2 pˆ qˆ
2
 z  /2 
E z.  E n  z  . pˆ qˆ  n   n  pˆ qˆ   .
2 n 2
E  E 
 That is, the minimum sample size needed for an interval estimate of the population
proportion is given by the formula
2
z 
n  pˆ qˆ   /2  .
 E 
Example 4

A researcher wishes to estimate, with 95% confidence, the proportion of people who own a home
computer. A previous study shows that 40% of those interviewed had a computer at home. The
researcher wishes to be accurate within 2% of the true proportion. Find the minimum sample size
necessary?
Solution

Since, we have
z  2  1.96, E  0.02, pˆ  0.40, qˆ  0.60.
Then, the minimum sample size needed to estimate the proportion of people who own a home
computer is given by
2 2
z   1.96 
n  pˆ qˆ   /2   (0.40)(0.60)    2304.96  2305.
 E   0.02 
Hence, to be 95% confident that the estimate is accurate within 2% of the true proportion, the
researcher needs a sample of at least 2305 people.
4. Confidence Intervals for the population variance and standard deviation
 In statistics, the variance and standard deviation of a variable are as important as the mean.
Both of them measure the variation in the data set of the variable under study.
 In this section, we will explain how to find the confidence intervals for the population
variance (σ2) and standard deviation (σ) of a variable.
 To calculate confidence intervals for σ2 and σ, a new statistical distribution will be studied. It is
called the "chi-square distribution".

LECTURE 3 PAGE 4 STAT 2040

Mr. Mohamed El-Sayed El-Dawoody Lecturer of Mathematical Statistics

 The chi-square distribution has the following properties:

 The chi-square variable is denoted by (pronounced "ki").
 The chi-square variable cannot be negative (That is, ).
 The chi-square distribution curve is skewed to the right (Positively skewed).
 The total area under the chi-square distribution curve is equal to 1 (or, 100%).
 The chi-square distribution has a family of curves based on the "degrees of freedom".
 The chi-square distribution becomes somewhat symmetric at about 100 degrees of freedom.
Remark
 Two different values are used to estimate the confidence intervals for the population variance
and standard deviation. This is because the chi-square distribution is not symmetric.
 One value is located on the right side of the distribution, and the other is located on the left
side of the distribution. See the following figure:
 These two values are denoted by and
and are given by the formulas:

where, (1   )100%  Confidence Level and

.
Example 5

Find the values of and for a 90% confidence interval when the sample size is 25?
Solution

Since, we have
(1   )100%  90%  1    0.90    1  0.90  0.10   / 2  0.05.
Then, we get
  right   /2, d.f
2 2

 0.05,
2
24  36.415.

 left  1 /2, d.f

2 2

 0.95,
2
24  13.848.

Remark
The values of  right
2
and  left
2
are found from the chi-square distribution table as follows:

LECTURE 3 PAGE 5 STAT 2040

Mr. Mohamed El-Sayed El-Dawoody Lecturer of Mathematical Statistics

Theorem
 The confidence intervals for the population variance (σ2) are given by the formula
(n  1) s 2 (n  1) s 2
2  .
 right
2
 left
2

 The confidence intervals for the population standard deviation (σ) are given by the formula
(n  1) s 2 (n  1) s 2
  .
 right
2
 left
2

where, n  The sample size & s 2  The sample variance (known or calculated).
Example 6

Find the 95% confidence interval of the variance and standard deviation for the nicotine content of
cigarettes that are manufactured if a random sample of 20 cigarettes has a standard deviation of
1.6 milligrams? Assume the variable is normally distributed.
Solution

Since,
(1   )100%  95%  1    0.95    1  0.95  0.05   / 2  0.025.
Then, we have
 right   /2, d.f  0.025,19  32.852.
2 2 2

 left  1 /2, d.f  0.975,19  8.907.

2 2 2

Now, we find that:

 The 95% confidence interval for the population variance is given by
(n  1) s 2 (n  1) s 2
2 
 right
2
 left
2

(20  1) (1.6)2 (20  1) (1.6)2

2 
32.852 8.907
1.48    5.46.
2

LECTURE 3 PAGE 6 STAT 2040

Mr. Mohamed El-Sayed El-Dawoody Lecturer of Mathematical Statistics

Hence, we can be 95% confident that the true variance for the nicotine content of all cigarettes
manufactured is between 1.48 and 5.46 milligrams.
 The 95% confidence interval for the population standard deviation is given by
1.48    5.46
1.22    2.34.
Hence, we can be 95% confident that the true standard deviation for the nicotine content of all
cigarettes manufactured is between 1.22 and 2.34 milligrams.
Example 7

Find the 90% confidence interval of the standard deviation for the number of named storms per
year in the Atlantic basin. A random sample of 10 years has been used and we obtain the results:
10 5 12 11 13 15 19 18 14 16
Assume the variable is approximately normal.
Solution

Since,
 x 
x i 10  5  .......  16 133
   13.3.
n 10 10
 s 
2  x i2  n x 2 1921  10(13.3)2 152.1
   16.9.
n 1 10  1 9
And where,
(1   )100%  90%  1    0.90    1  0.90  0.10   / 2  0.05.
Then, we have
 right   /2, d.f  0.05, 9  16.919.
2 2 2

 left  1 /2, d.f  0.95, 9  3.325.

2 2 2

Now, the 90% confidence interval for the population standard deviation is given by
(n  1) s 2 (n  1) s 2
2 
 right
2
 left
2

(10  1) (16.9) (10  1) (16.9)

2 
16.919 3.325
8.99    45.74.
2

8.99    45.74
2.99    6.76.
Hence, we can be 90% confident that the standard deviation for the number of named storms is
between 3.0 and 6.8 storms.
‫تمت بـحمـد اللـه‬

LECTURE 3 PAGE 7 STAT 2040