Estimation of Parameters

Confidence Intervals for Proportions

A university has a large dormitory system. A vote is needed to

change the system policy of ‘quite hours’. In a random sample of
34 residents, 23 say that they will vote in favor of change.
Construct a 95 percent confidence interval for the population
proportion favoring the change in policy.
The sample proportion of residents favoring the change of policy is
23
p= = 0.68.
34
Since both np = 23 > 5 X and n(1 − p) = 11 > 5 X, a 95%
confidence interval for the population proportion is
r r
p(1 − p) p(1 − p)
p − zα/2 < p < p + zα/2 .
n n
Confidence Intervals for Proportions

For 95% level of confidence, α = 0.05, we have α/2 = 0.025 and

z0.025 = 1.96. Hence
r r
0.68 · 0.32 0.68 · 0.32
0.68 − 1.96 < p < 0.68 + 1.96 .
34 34
Therefore, we get the following 95% confidence interval for the
population proportion:

0.5232 < p < 0.7368,

or, equivalently, the following 95% confidence interval for the

population percentage:

52.32% < p < 78.68%.

Confidence Intervals for Proportions

A medical researcher wants to estimate the percentage of

overweight people who have hypertension. He examined 125
individuals and found that 100 of them suffer from hypertension.
a) Find a point estimate for the true population percentage of
overweight people who have hypertension.
As we know, a sample percentage serves as a point estimate for a
population percentage. In our example,
100
p̂ = p = · 100% = 80%.
125
Confidence Intervals for Proportions
b) Construct a 90% confidence interval for the true population
percentage of overweight people who have hypertension.
In our problem, we have np = 100 > 5 and n(1 − p) = 25 > 5,
hence (1 − α) · 100% confidence interval has the form:
r r
p(1 − p) p(1 − p)
p − zα/2 < p < p + zα/2 .
n n
For a 90% confidence interval α = 0.1. From the table
z0.05 = 1.645. Therefore, a 90% confidence interval for p is
r r
0.8 · 0.2 0.8 · 0.2
0.8 − 1.645 < p < 0.8 + 1.645
125 125
or 0.8 − 0.059 < p < 0.8 + 0.059.
That is, 0.741 < p < 0.859 or 74.1% < p < 85.9% is a 90%
confidence interval for the true population percentage of
overweight people who have hypertension.
Confidence Intervals for the Variance
A social worker is interested in the average amount of money (in
TL) spent per day on medicine by elderly people. He drew a
random sample of 8 people and obtained the following data
16.0 55.0 28.0 21.9 10.5 12.5 26.3 25.6
I Find a point estimate for the population standard deviation.
We use a sample standard deviation as a point estimate for
a population standard deviation. Calculations:
P
x 195.8
x= = = 24.475.
n 8
We compute the sample variance as
P 2
2 x − n(x)2 6158.16 − 4792.24 1365.92
s = = = = 195.16.
n−1 7 7
Hence,
√ the sample standard deviation is
s = 195.16 = 13.96. and we get a point estimate for the
population standard deviation as: σ ≈ s = 13.96.
Confidence Intervals for the Variance
I Construct a 95% confidence interval for the population
standard deviation.
Using χ2 (n − 1) = χ2 (7) distribution, we can write a 95%
confidence interval for the population variance in the form
(n − 1)s 2 (n − 1)s 2
2
< σ2 < .
χ0.025 χ20.975
From the table χ20.025 = 16.01, χ20.975 = 1.69. Therefore,
7 · 195.16 7 · 195.16
< σ2 < or 85.33 < σ 2 < 808.36
16.01 1.69
is a 95% confidence interval for the population variance σ 2 .
Correspondingly,
9.24 < σ < 28.43
is a 95% confidence interval for the population standard
deviation σ.
Confidence Intervals for the Variance
A production manager studies the daily use of coal. He takes a
sample of 12 days and finds that the sample mean is 250 kg, with
a sample standard deviation of 35 kg. Assume that usage amounts
are normally distributed.
I Construct a 90% confidence interval for population variance.
Using χ2 (n − 1) = χ2 (11) distribution, we can write a 90%
confidence interval for the population variance in the form
(n − 1)s 2 2 (n − 1)s 2
< σ < .
χ20.05 χ20.95
From the table χ20.05 = 19.675, χ20.95 = 4.575. Therefore,
11 · 1225 11 · 1225
< σ2 <
19.675 4.575
or
684.88 < σ 2 < 2945.36
is a 90% confidence interval for the population variance σ 2 .
Confidence Intervals for the Variance

I Construct a 90% confidence interval for population standard

deviation.
We have found taht

684.88 < σ 2 < 2945.36

is a 90% confidence interval for the population variance σ 2 .

By taking the square roots, we obtain that

26.17 < σ < 54.27

is a 90% confidence interval for the population standard

deviation σ.