Interval Estimation

Confidence interval for the population variance and

standard deviation

In this part of the course, we assume that all samples are drawn
from normally distributed populations with unknown parameters.
As we know, in this case, a (1 − α)100% confidence interval has
the form:
(n − 1)s 2 2 (n − 1)s 2
< σ < .
χ2α/2 χ21−α/2

Taking the square roots, we obtain a (1 − α)100% confidence

interval for the population standard deviation:
s s
(n − 1)s 2 (n − 1)s 2
< σ < .
χ2α/2 χ21−α/2
Confidence interval for the population variance and
standard deviation
Problem. Twelve bank tellers were randomly sampled and it was
determined that they made an average of 3.6 errors per day with a
sample standard deviation of 0.42 errors. Assume that the number of
errors bank tellers make per day is normally distributed.
Construct a 90% confidence interval for the population standard
deviation of the number of errors per day.
I In our problem n = 12, so we use χ2 (11) distribution. For a 90%
confidence interval α = 0.1, so α/2 = 0.05 and the needed
2 2
confidence interval can be written as χ11s
2 < σ 2 < χ11s
2 .
0.05 0.95
Confidence interval for the population variance and
standard deviation
Problem. Twelve bank tellers were randomly sampled and it was
determined that they made an average of 3.6 errors per day with a
sample standard deviation of 0.42 errors. Assume that the number of
errors bank tellers make per day is normally distributed.
Construct a 90% confidence interval for the population standard
deviation of the number of errors per day.
I In our problem n = 12, so we use χ2 (11) distribution. For a 90%
confidence interval α = 0.1, so α/2 = 0.05 and the needed
2 2
confidence interval can be written as χ11s
2 < σ 2 < χ11s
2 .
0.05 0.95
I From the table we find χ20.05 = 19.68, χ20.95 = 4.57. So a 90%
confidence interval for the population variance is
11 · (0.42)2 11 · (0.42)2
< σ2 < or 0.098 < σ 2 < 0.425.
19.68 4.57
Confidence interval for the population variance and
standard deviation
Problem. Twelve bank tellers were randomly sampled and it was
determined that they made an average of 3.6 errors per day with a
sample standard deviation of 0.42 errors. Assume that the number of
errors bank tellers make per day is normally distributed.
Construct a 90% confidence interval for the population standard
deviation of the number of errors per day.
I In our problem n = 12, so we use χ2 (11) distribution. For a 90%
confidence interval α = 0.1, so α/2 = 0.05 and the needed
2 2
confidence interval can be written as χ11s
2 < σ 2 < χ11s
2 .
0.05 0.95
I From the table we find χ20.05 = 19.68, χ20.95 = 4.57. So a 90%
confidence interval for the population variance is
11 · (0.42)2 11 · (0.42)2
< σ2 < or 0.098 < σ 2 < 0.425.
19.68 4.57
I Taking the square roots, we get that (0.31, 0.65) is a 90%
confidence interval for the population standard deviation.
Confidence interval for the population variance and
standard deviation
Travel agencies studied frequent flyer miles. One part of the study
concerned ”blackout days” on which frequent flyer miles cannot be
used. Data for five airlines appear below:
14 0 34 32 7
Assume that the numbers of blackout days are normally distributed.
I Calculate the mean number of blackout days for the sample.
The sample mean is
P
x 87
x= = = 17.4.
n 5
Confidence interval for the population variance and
standard deviation
Travel agencies studied frequent flyer miles. One part of the study
concerned ”blackout days” on which frequent flyer miles cannot be
used. Data for five airlines appear below:
14 0 34 32 7
Assume that the numbers of blackout days are normally distributed.
I Calculate the mean number of blackout days for the sample.
The sample mean is
P
x 87
x= = = 17.4.
n 5
I Find the sample variance and standard deviation.
The sample variance is
P 2
2 x − n(x)2 2425 − 5 · 302.76 911.2
s = = = = 227.8.
n−1 4 4
√ √
The sample standard deviation is s = s 2 = 227.8 = 15.09.
Problem (continued)
I Construct a 90% confidence interval for the true mean
number of blackout days.
Problem (continued)
I Construct a 90% confidence interval for the true mean
number of blackout days.
I We have: n = 5, x = 17.4, s = 15.09. Since the sample size is
small, to constract a confidence interval, we use t distribution.
A (1 − α)100% confidence interval for a population mean is
given by the condition
 
s s
P x − tα/2 √ < µ < x + tα/2 √ = 1 − α.
n n
Problem (continued)
I Construct a 90% confidence interval for the true mean
number of blackout days.
I We have: n = 5, x = 17.4, s = 15.09. Since the sample size is
small, to constract a confidence interval, we use t distribution.
A (1 − α)100% confidence interval for a population mean is
given by the condition
 
s s
P x − tα/2 √ < µ < x + tα/2 √ = 1 − α.
n n
I For a 90% confidence interval α = 0.1, so α/2 = 0.05. From
the table t0.05 = 2.132. So, we get a 90% confidence interval
for the population mean:
 
15.09 15.09
17.4 − 2.132 · √ , 7.4 + 2.132 · √ = (3.04, 31.74).
5 5
Problem (continued)
I Construct a 90% confidence interval for the population standard
deviation of the number of blackout days.
Problem (continued)
I Construct a 90% confidence interval for the population standard
deviation of the number of blackout days.
I We have: n = 5, s = 15.09. When we sample from a normal
population, a (1 − α)100% confidence interval has the form
(n − 1)s 2 2 (n − 1)s 2
< σ < .
χ2α/2 χ21−α/2
As n = 5, we use the χ2 (4) distribution. For a 90% confidence
interval α = 0.1 ⇒ α/2 = 0.05 and, from the table,
χ20.05 = 9.488, χ20.95 = 0.711. So, a 90% confidence interval for the
population variance is:
4 · (17.4)2 4 · (17.4)2
 
, = (96.04; 1281.6).
9.488 0.711
Problem (continued)
I Construct a 90% confidence interval for the population standard
deviation of the number of blackout days.
I We have: n = 5, s = 15.09. When we sample from a normal
population, a (1 − α)100% confidence interval has the form
(n − 1)s 2 2 (n − 1)s 2
< σ < .
χ2α/2 χ21−α/2
As n = 5, we use the χ2 (4) distribution. For a 90% confidence
interval α = 0.1 ⇒ α/2 = 0.05 and, from the table,
χ20.05 = 9.488, χ20.95 = 0.711. So, a 90% confidence interval for the
population variance is:
4 · (17.4)2 4 · (17.4)2
 
, = (96.04; 1281.6).
9.488 0.711
I To find a 90% confidence interval for σ, we take the square roots:
√ √
that is, ( 96.04, 1281.6) = (9.8, 35.8) is a 90% confidence
interval for σ.