1/14/2019

Learning objectives
8.1 Explain an interval estimator and the difference
between point estimation and interval estimation.
CHAPTER 8 8.2 Calculate a confidence interval estimate for the
population mean when the population standard
Interval estimation deviation is known.
8.3 Describe the factors that influence the width of a
confidence interval.
8.4 Discuss features of the t distribution.

continued
Copyright © 2016 McGraw-Hill Education (Australia) Pty Ltd Copyright © 2016 McGraw-Hill Education (Australia) Pty Ltd
Jaggia, Essentials of Business Statistics, 1e 8-1 Jaggia, Essentials of Business Statistics, 1e 8-2
Updated by: Azizur Rahman Updated by: Azizur Rahman

1 2

Learning objectives Fuel economy of ‘ultra-green’ cars

8.5 Calculate a confidence interval estimate for the • A car manufacturer advertises that its new ‘ultra-
population mean when the population standard green’ car obtains an average of 45 km/L and, based
deviation is not known. on its fuel emissions, has earned an A+ rating from
the Environment Protection Authority.
8.6 Calculate a confidence interval for the population
proportion. • Jared Beane, an analyst for a consumer advocacy
firm, selects a sample of 25 cars for testing
8.7 Determine the sample size necessary to estimate purposes.
the population mean and the population
proportion. • Each car is driven the same distance in identical
conditions in order to measure its fuel usage.

continued
Copyright © 2016 McGraw-Hill Education (Australia) Pty Ltd Copyright © 2016 McGraw-Hill Education (Australia) Pty Ltd
Jaggia, Essentials of Business Statistics, 1e 8-3 Jaggia, Essentials of Business Statistics, 1e 8-4
Updated by: Azizur Rahman Updated by: Azizur Rahman

3 4

Confidence interval for the population

Fuel economy of ‘ultra-green’ cars mean when σ is known
• The fuel usage (in km/L) for each ultra-green car is LO 8.1 Explain an interval estimator and the difference
given below. between point estimation and interval estimation.

• Confidence interval—provides a range of values

that, with a certain level of confidence, contains the
population parameter of interest.
– Also referred to as an interval estimate.
• Jared would like to use the data in this sample to:
– Estimate the mean km/L of all ultra-green cars with 90% • Construct a confidence interval as
confidence.
– Estimate the proportion of all ultra-green cars that obtain Point estimate ± margin of error
over 45 km/L, with 90% confidence.
– Determine the sample size needed to achieve a specified – Margin of error accounts for the variability of the estimator
level of precision in the mean and proportion estimates. and the desired confidence level of the interval.

Copyright © 2016 McGraw-Hill Education (Australia) Pty Ltd Copyright © 2016 McGraw-Hill Education (Australia) Pty Ltd
Jaggia, Essentials of Business Statistics, 1e 8-5 Jaggia, Essentials of Business Statistics, 1e 8-6
Updated by: Azizur Rahman Updated by: Azizur Rahman

5 6

1
 1/14/2019

Constructing a confidence interval for LO 8.2 Constructing a confidence

μ when σ is known interval for μ when σ is known
LO 8.2 Calculate a confidence interval estimate for the • Constructing the confidence interval
population mean when the population standard deviation
X −
is known. – Since Z =
 n
• Consider a standard normal random variable Z
P ( −1.96  Z  1.96) = 0.95  X − 
– We get P  −1.96   1.96  = 0.95,i.e.
as illustrated here.   n 
(
P  − 1.96 n  X   + 1.96 )
n = 0.95

which implies that if samples of size n are drawn repeatedly

from a given population, 95% of the computed sample
means, x 's , will fall within the interval   1.96  n
and the remaining 5% will fall outside the interval.
continued continued
Copyright © 2016 McGraw-Hill Education (Australia) Pty Ltd Copyright © 2016 McGraw-Hill Education (Australia) Pty Ltd
Jaggia, Essentials of Business Statistics, 1e 8-7 Jaggia, Essentials of Business Statistics, 1e 8-8
Updated by: Azizur Rahman Updated by: Azizur Rahman

7 8

LO 8.2 Constructing a confidence LO 8.2 Constructing a confidence

interval for μ when σ is known interval for μ when σ is known
• Constructing the confidence interval • Constructing the confidence interval
– Since we do not know , we cannot determine whether a – Level of significance (i.e. probability of error) = a.
particular x falls within the interval or not. – Confidence coefficient = (1 − a)
– However, we do know that x will fall within the interval – Confidence level = 100(1 − a)%
  1.96  n if and only if  falls within this interval
▪ A 100(1 − a)% confidence interval of the population
x  1.96  n mean  when the standard deviation  is known is
computed as
– This will happen 95% of the time, given the interval
x  za 2  n
construction. Thus, this is a 95% confidence interval for the
population mean. or, equivalently,
 x − za 2  n , x + za 2  n 


continued continued
Copyright © 2016 McGraw-Hill Education (Australia) Pty Ltd Copyright © 2016 McGraw-Hill Education (Australia) Pty Ltd
Jaggia, Essentials of Business Statistics, 1e 8-9 Jaggia, Essentials of Business Statistics, 1e 8-10
Updated by: Azizur Rahman Updated by: Azizur Rahman

9 10

LO 8.2 Constructing a confidence LO 8.2 Constructing a confidence

interval for μ when σ is known interval for μ when σ is known
• Constructing the confidence interval • Example: A sample of 25 generic-brand cereal
– za/2 is the z value associated boxes yields a mean weight of 450 grams of cereal
with the probability of a/2 per box.
in the upper tail.
– Construct a 95% confidence interval of the mean weight of
all cereal boxes. Assume that  = 0.30 g.
– A 95% confidence interval is computed as
 0.30
x  1.96 = 450  1.96 = 450  0.118
n 25
The mean weight of all cereal boxes falls between 449.88 g
– Confidence intervals and 450.118 g.
▪ 90%, a = 0.10, a/2 = 0.05, za/2 = z0.05 = 1.645
▪ 99%, a = 0.01, a/2 = 0.005, za/2 = z0.005 = 2.575
continued continued
Copyright © 2016 McGraw-Hill Education (Australia) Pty Ltd Copyright © 2016 McGraw-Hill Education (Australia) Pty Ltd
Jaggia, Essentials of Business Statistics, 1e 8-11 Jaggia, Essentials of Business Statistics, 1e 8-12
Updated by: Azizur Rahman Updated by: Azizur Rahman

11 12

2
 1/14/2019

LO 8.2 Constructing a confidence

Width of a confidence interval
interval for μ when σ is known
LO 8.3 Describe the factors that influence the width of a
• Interpreting a confidence interval confidence interval.
– Interpreting a confidence interval requires care. • The width of a confidence interval
Margin of error Confidence interval width
▪ Incorrect: The probability that  falls in the interval is 0.95.

▪ Correct: If numerous samples of size n are drawn from a given

population, then 95% of the intervals formed by the formula
za 2  n (
2 za 2  n )
x  za 2  n will contain . • The width of the confidence interval is influenced
• Since there are many possible samples, we will be right 95% of by the
the time, thus giving us 95% confidence.
– Standard deviation 
– Sample size n
– Confidence level 100(1 − a)%.

continued
Copyright © 2016 McGraw-Hill Education (Australia) Pty Ltd Copyright © 2016 McGraw-Hill Education (Australia) Pty Ltd
Jaggia, Essentials of Business Statistics, 1e 8-13 Jaggia, Essentials of Business Statistics, 1e 8-14
Updated by: Azizur Rahman Updated by: Azizur Rahman

13 14

LO 8.3 Width of a confidence interval LO 8.3 Width of a confidence interval

• Standard deviation • Sample size
– For a given confidence level 100(1 − a)% and sample – For a given confidence level 100(1 − a)% and population
size n, the greater the population standard deviation  , standard deviation  , the smaller the sample size n, the
the wider the confidence interval. wider the confidence interval.
▪ Example: Let the standard deviation of the population of ▪ Example: Instead of 25 observations, let the sample be based
generic-brand cereal boxes be 0.50 g instead of 0.30 g. on 16 cereal boxes. Compute a 95% confidence interval using
Compute a 95% confidence interval based on the same a sample mean of 450 g and a population standard deviation
sample information. of 0.30 g.
 0.50  0.30
x  1.96 = 450  1.96 = 450  0.196 x  1.96 = 450  1.96 = 450  0.147
n 25 n 16

▪ This confidence interval width has increased from 0.24 to ▪ This confidence interval width has increased from 0.24 to
2(0.196) = 0.39. 2(0.147) = 0.29.

continued continued
Copyright © 2016 McGraw-Hill Education (Australia) Pty Ltd Copyright © 2016 McGraw-Hill Education (Australia) Pty Ltd
Jaggia, Essentials of Business Statistics, 1e 8-15 Jaggia, Essentials of Business Statistics, 1e 8-16
Updated by: Azizur Rahman Updated by: Azizur Rahman

15 16

LO 8.3 Width of a confidence interval LO 8.3 Width of a confidence interval

• Example: Confidence interval for IQ scores
• Confidence level
– IQ tests are designed to yield scores that are approximately
– For a given sample size n and population standard
normally distributed. A reporter is interested in estimating
deviation  , the greater the confidence level 100(1 − a)%,
the average IQ of employees of a large high-tech company
the wider the confidence interval.
in Sydney. She gathers the IQ scores of 22 employees and
▪ Example: Instead of a 95% confidence interval, compute a records the sample mean IQ as 106. She assumes that the
99% confidence interval based on the same sample population standard deviation is 15.
information.
 0.30 – Compute 90% and 99% confidence intervals for the
x  2.576 = 450  2.576 = 450  0.155
n 25 average IQ at the company.
▪ This confidence interval width has increased from 0.24 to
2(0.155) = 0.31.

continued continued
Copyright © 2016 McGraw-Hill Education (Australia) Pty Ltd Copyright © 2016 McGraw-Hill Education (Australia) Pty Ltd
Jaggia, Essentials of Business Statistics, 1e 8-17 Jaggia, Essentials of Business Statistics, 1e 8-18
Updated by: Azizur Rahman Updated by: Azizur Rahman

17 18

3
 1/14/2019

Confidence interval for the population

LO 8.3 Width of a confidence interval
mean when  is unknown
• Solution: LO 8.4 Discuss features of the t distribution.

– For a 90% confidence interval z a / 2 = z0.05 = 1.645 • The t distribution

– For a 99% confidence interval z a / 2 = z0.005 = 2.576 – If repeated samples of size n are taken from a normal
population with a finite variance, then the
15
– The 90% confidence interval is 106  1.645 = 106  5.26 statistic T follows the t distribution, X −
T =
22 with (n − 1) degrees of freedom df. S n
15 ▪ Degrees of freedom determine
– The 99% confidence interval is 106  2.576 = 106  8.24
the broadness of the tails of the
22
distribution—the fewer the degrees
– Note that the 99% interval is wider than the 90% interval. of freedom, the broader the tails.

continued
Copyright © 2016 McGraw-Hill Education (Australia) Pty Ltd Copyright © 2016 McGraw-Hill Education (Australia) Pty Ltd
Jaggia, Essentials of Business Statistics, 1e 8-19 Jaggia, Essentials of Business Statistics, 1e 8-20
Updated by: Azizur Rahman Updated by: Azizur Rahman

19 20

LO 8.4 Confidence interval for the LO 8.4 Confidence interval for the
population mean when  is unknown population mean when  is unknown
• Summary of the tdf distribution • The tdf distribution with various degrees of freedom
– Bell-shaped and symmetric around 0 with asymptotic
tails that get closer and closer to the horizontal axis, but
never touch it.

– Has slightly broader tails than the z distribution.

– Consists of a family of distributions where the actual shape

of each one depends on the df. As df increases, the tdf
distribution becomes similar to the z distribution; it is
identical to the z distribution when df approaches infinity.

continued continued
Copyright © 2016 McGraw-Hill Education (Australia) Pty Ltd Copyright © 2016 McGraw-Hill Education (Australia) Pty Ltd
Jaggia, Essentials of Business Statistics, 1e 8-21 Jaggia, Essentials of Business Statistics, 1e 8-22
Updated by: Azizur Rahman Updated by: Azizur Rahman

21 22

LO 8.4 Confidence interval for the Constructing a confidence interval for

population mean when  is unknown  when  is unknown
LO 8.5 Calculate a confidence interval estimate for the
• Example: Compute ta,df for a = 0.025 using 2, 5 and population mean when the population standard deviation
50 degrees of freedom. is not known.
– Solution: Turning to the Student’s t distribution table in • A 100(1 − a)% confidence interval of the population
Appendix A of the text, we find mean  when the population standard deviation  is
▪ For df = 2, t0.025,2 = 4.303 not known is computed as x  ta 2,df s n
▪ For df = 5, t0.025,5 = 2.571
or, equivalently,  x − ta s n , x + ta s n
▪ For df = 50, t0.025,50 = 2.009.  2,df 2,df 

– Note that the tdf values change with the degrees of freedom.
Further, as df increases, the tdf distribution begins to where s is the sample standard deviation.
resemble the z distribution.

continued
Copyright © 2016 McGraw-Hill Education (Australia) Pty Ltd Copyright © 2016 McGraw-Hill Education (Australia) Pty Ltd
Jaggia, Essentials of Business Statistics, 1e 8-23 Jaggia, Essentials of Business Statistics, 1e 8-24
Updated by: Azizur Rahman Updated by: Azizur Rahman

23 24

4
 1/14/2019

LO 8.5 Constructing a confidence LO 8.5 Constructing a confidence

interval for  when  is unknown interval for  when  is unknown
• Example: Recall that Jared Beane wants to estimate • Using Excel to construct confidence intervals
the mean km/L of all ultra-green cars. Use the
– Use functions (AVERAGE, STDEV.S) to compute sample
sample data for 25 cars to construct a 90% average and sample standard deviation.
confidence interval of the population mean. Assume
that km/L follows a normal distribution. – Excel’s [Link] function finds a particular tdf value for a
given cumulative probability. For a 95% confidence
– Solution: Since the population standard deviation is not
interval with a = 0.05 and df = n – 1 = 100 – 1 = 99, we
known, the sample standard deviation has to be computed
need to find t0.025,99.
from the sample. As a result, the 90% confidence interval is
s 4.84 – Input ‘=[Link](0.975, 99)’ in an empty cell, then press
x  ta / 2, df = 43.44  1.711 = 43.44  1.66
n 25 <Enter>.
– Thus, Jared concludes with 90% confidence that the
average km/L of all ultra-green cars is between 41.78 and
45.10 km/L.
continued
Copyright © 2016 McGraw-Hill Education (Australia) Pty Ltd Copyright © 2016 McGraw-Hill Education (Australia) Pty Ltd
Jaggia, Essentials of Business Statistics, 1e 8-25 Jaggia, Essentials of Business Statistics, 1e 8-26
Updated by: Azizur Rahman Updated by: Azizur Rahman

25 26

Confidence interval for the population LO 8.6 Confidence interval for the
proportion population proportion
LO 8.6 Calculate a confidence interval for the population
proportion.
• Thus, a 100(1 − a)% confidence interval for the
• Let the parameter p represent the proportion of population proportion is
successes in the population, where success is
defined by a particular outcome. p (1 − p ) or
 p (1 − p ) p (1 − p ) 
p  za 2  p − za 2 , p + za 2 
 
n n n
• P is the point estimator of the population proportion
p.
where p is used to estimate the population
• By the central limit theorem, P can be approximated parameter p.
by a normal distribution for large samples; that is,
where np > 5 and n(1 − p) > 5.

continued continued
Copyright © 2016 McGraw-Hill Education (Australia) Pty Ltd Copyright © 2016 McGraw-Hill Education (Australia) Pty Ltd
Jaggia, Essentials of Business Statistics, 1e 8-27 Jaggia, Essentials of Business Statistics, 1e 8-28
Updated by: Azizur Rahman Updated by: Azizur Rahman

27 28

LO 8.6 Confidence interval for the

Selecting the required sample size
population proportion
LO 8.7 Determine the sample size necessary to estimate
• Example: Recall that Jared Beane wants to estimate the population mean and the population proportion.
the proportion of all ultra-green cars that obtain over
45 km/L. Use the sample information to construct a • Precision in interval estimates is implied by a low
90% confidence interval of the population proportion margin of error.

– Solution: Note that p = 6 / 25 = 0.24 . In addition, the normality • A larger n reduces the margin of error for interval
assumption is met since np  5 and n(1 − p )  5 . Thus, estimates.
0.24(1 − 0.24) • How large should the sample size be for a given
0.24  1.645 = 0.24  0.141
25 margin of error?
– With 90% confidence, Jared reports that the percentage of
cars that obtain over 45 km/L is between 9.9% and 38.1%.

continued
Copyright © 2016 McGraw-Hill Education (Australia) Pty Ltd Copyright © 2016 McGraw-Hill Education (Australia) Pty Ltd
Jaggia, Essentials of Business Statistics, 1e 8-29 Jaggia, Essentials of Business Statistics, 1e 8-30
Updated by: Azizur Rahman Updated by: Azizur Rahman

29 30

5
 1/14/2019

LO 8.7 Selecting the required sample LO 8.7 Selecting the required sample
size size
• Selecting n to estimate  • Selecting n to estimate 
– Consider a confidence interval for  with a – For a desired margin of error E, the minimum sample
known  . Let E denote the desired margin of error. size n required to estimate a 100(1 − a)% confidence
interval of the population mean  is
– Since E = zα 2 σ n
2
 zα 2 σˆ 
we may rearrange to get  zα 2 σ 
2
n =  
n =    E 
 E 
where ̂ is a reasonable estimate of  in the planning
If  is unknown, estimate it with ̂ .
stage.

continued continued
Copyright © 2016 McGraw-Hill Education (Australia) Pty Ltd Copyright © 2016 McGraw-Hill Education (Australia) Pty Ltd
Jaggia, Essentials of Business Statistics, 1e 8-31 Jaggia, Essentials of Business Statistics, 1e 8-32
Updated by: Azizur Rahman Updated by: Azizur Rahman

31 32

LO 8.7 Selecting the required sample LO 8.7 Selecting the required sample
size size
• Example: Recall that Jared Beane wants to • Selecting n to estimate p
construct a 90% confidence interval of the mean – Consider a confidence interval for p. Let E denote the
km/L of all ultra-green cars. desired margin of error.
– Since
– Suppose Jared would like to constrain the margin of error to
p (1 − p ) where p is the
within 1 km/L. Further, the lowest figure in the population is E = zα 2
35 km/L, whereas the highest is 53 km/L. n sample proportion,
– How large a sample does Jared need to compute the 90%
confidence interval of the population mean?  zα 2 
2

we may rearrange to get n =   p (1 − p )

 za ˆ 
2

   1.645  4.50  = 54.80

2
 E 
– The sample size is n =  2  =  
 E   1 
  – Since p comes from a sample, we must use a reasonable
which is rounded up to 55. estimate of p, that is, p̂.

continued continued
Copyright © 2016 McGraw-Hill Education (Australia) Pty Ltd Copyright © 2016 McGraw-Hill Education (Australia) Pty Ltd
Jaggia, Essentials of Business Statistics, 1e 8-33 Jaggia, Essentials of Business Statistics, 1e 8-34
Updated by: Azizur Rahman Updated by: Azizur Rahman

33 34

LO 8.7 Selecting the required sample LO 8.7 Selecting the required sample
size size
• Selecting n to estimate p • Example: Recall that Jared Beane wants to
– For a desired margin of error E, the minimum sample construct a 90% confidence interval of the proportion
size n required to estimate a 100(1 − a)% confidence of all ultra-green cars that obtain over 45 km/L.
interval of the population proportion p is – Jared does not want the margin of error to be more than
2
 zα 2  0.10.
n =   pˆ (1 − pˆ ) – How large a sample does Jared need for his analysis of the
 E 
population proportion?
2
 zα 2 
2
 1.645 
 pˆ (1 − pˆ ) =  0.10  0.50(1 − 0.50) = 67.65  68
where p̂ is a reasonable estimate of p in the planning n = 
stage.  E   

continued
Copyright © 2016 McGraw-Hill Education (Australia) Pty Ltd Copyright © 2016 McGraw-Hill Education (Australia) Pty Ltd
Jaggia, Essentials of Business Statistics, 1e 8-35 Jaggia, Essentials of Business Statistics, 1e 8-36
Updated by: Azizur Rahman Updated by: Azizur Rahman

35 36