Statistics for Managers

Using Microsoft® Excel

5th Edition

Chapter 8
Confidence Interval Estimation

Chap 8-1
Chapter Outline

1. Confidence Intervals for the Population

Mean, μ
 when Population Standard Deviation σ is
Known
 when Population Standard Deviation σ is
Unknown
2. Confidence Intervals for the Population
Proportion, π
3. Determining the Required Sample Size

Chap 8-2
 Point Estimates
 A point estimate is a single number. For the population
mean the point estimate is the sample mean. For the
population proportion the point estimate is the sample
proportion.
 A confidence interval provides a range of values which
have a predetermined likelihood of containing the true
population parameter (i.e. population mean or proportion).

Lower Confidence Upper Confidence

Limit Point Estimate Limit

Width of
confidence interval
Chap 8-3
 Confidence Interval Estimates

 The general formula for constructing a confidence interval is:

Point Estimate ± (Critical Value) (Standard error of the point estimate)

Chap 8-4
 Confidence Interval Estimates

 A confidence interval gives a range estimate of values:

 Based on all the observations from a single sample
 Takes into consideration variation in sample statistics from sample
to sample (i.e. sampling variation)
 Stated in terms of level of confidence
 Ex. 95% level of confidence, 99% level of confidence
 Can never be 100% confident

Chap 8-5
 Level of Confidence

 Set by the researcher and determines the likelihood that the interval
will contain the unknown population parameter
 It is a percentage (less than 100%), usually set to be high values
such as 90%, 95% and 99%.
 It will implicitly determine the magnitude of the critical value.
 Higher confidence levels are associated with larger confidence
intervals.

Chap 8-6
 Confidence Level
 Suppose confidence level = 95%
 Also written = 0.95
 A specific interval either will contain or will not contain the true
parameter
 Hence, the level of confidence provides a reassurance in the
procedure that we follow for constructing the confidence interval:
 In the long run, 95% of all the confidence intervals that can be
constructed will contain the unknown true parameter

Chap 8-7
 Confidence Interval for μ
(σ Known)
Assumptions
 Population standard deviation σ is known
 Population is normally distributed
 If population is not normal, use large sample (Central Limit Theorem)
Confidence interval estimate:
σ
X ± Z critical
√n
(where Zcritical is the standardized normal distribution critical value
for a probability of half of the one minus the level of confidence
value in each tail)

Chap 8-8
 Finding the Critical Value, Z
Consider a 95% confidence interval:
1    .95

α α
 .025  .025
2 2

Z units: Z= -1.96 0 Z= 1.96

Lower Upper
X units: Confidence Point Estimate Confidence
Limit Limit

Zcritical=norm.s.inv(0.025) Zcritical=norm.s.inv(0.975) Chap 8-9

 Finding the Critical Value, Z
Commonly used confidence levels are 90%, 95%,
and 99%
Confidence Confidence
Level Coefficient Z value

80% .80 1.28

90% .90 1.645
95% .95 1.96
98% .98 2.33
99% .99 2.58
99.8% .998 3.08
99.9% .999 3.27
Chap 8-10
 Confidence Intervals and
Level of Confidence (1-)
Sampling Distribution
of the Sample Mean
/2 1  /2
𝐗
x
Intervals μx  μ
extend from x1
x2
𝜎
𝑋 − 𝑍 𝑐𝑟𝑖𝑡𝑖𝑐𝑎𝑙 (1-)*100% of
√𝑛 intervals constructed
to contain μ;
𝜎 ()*100% do not.
𝑋 + 𝑍 𝑐𝑟𝑖𝑡𝑖𝑐𝑎𝑙
√𝑛 Confidence Intervals
Chap 8-11
 Confidence Interval for μ
(σ Known) Example
 You collect a sample of 11 observations from a large
normal population. The sample mean is calculated to
be 2.20 in your sample. We know from past testing
that the population standard deviation is 0.35.
 Construct a confidence interval for the true
population mean at the 95% level of confidence.

Chap 8-12
 Confidence Interval for μ
(σ Known) Example
σ σ
¿X
X ±Z Z critical
n √n
 2.20  1.96 (.35/ 11 )
 2.20  .2068
(1.9932 , 2.4068)
 We are 95% confident that the true population mean is somewhere
between 1.99 and 2.41.
 Alternatively, please note that although the true mean may or may
not be in this interval, 95% of intervals formed in this manner will
contain the true mean.
Chap 8-13
 Confidence Interval for μ
(σ Unknown)

 If the population standard deviation, σ, is unknown, we can

substitute the sample standard deviation, s, into our formula
instead.
 However, this introduces extra layer of uncertainty into our
estimation, since sample standard deviation also varies from
sample to sample (like the sample mean).
 So we use the t-distribution instead of the standard normal
distribution to take this additional uncertainty into account.
Chap 8-14
 Confidence Interval for μ
(σ Unknown)
Assumptions
 Population standard deviation, σ, is unknown.
 Population is normally distributed.
 If population is not normal, use large sample.
Use t-distribution instead of the standard normal distribution.
Confidence Interval Estimate:
s
¿ X ± t critical , n −1
√n
(where t is the critical value of the t-distribution with n-1
degrees of freedom and an area of α/2 in each tail)

Chap 8-15
 Student’s t Distribution
 The critical t-value depends on two factors:
a) The level of confidence, 1-
b) Degrees of freedom: Denoted by d.f. = n -1
 Degrees of freedom measures the number of observations that are
free to vary after sample mean has been calculated.

 The critical t-value will be larger when

a) The level of confidence set by the researcher is larger.
b) When the degrees of freedom (i.e. the sample size) is smaller.

Chap 8-16
 Student’s t-distribution
Note: t Z as n increases

Standard
Normal
(t with df = ∞)

t (df = 13)
t-distributions are bell-
shaped and symmetric, but
have ‘fatter’ tails than the t (df = 5)
normal

0 t
Chap 8-17
 Critical values for the t-distribution
• Suppose we have a sample of 30 observations: Hence, n = 30.
• The degrees of freedom df = n - 1 = 29
• If the level of confidence is 95%, then 1-  = 0.95
• df
Which .25 .10
implies that 05
the center of the distribution is 95%.
• As before, we have 5% left for tails, which implies that each
tail has half of the 5% under it: /2 = 0.05/2 = 0.025
1 1.000 3.078 6.314
• The critical t-value is calculated using the
2 0.817
t.inv.2t() command1.886 2.920
in Excel.
/2 = .025
• It requires two inputs: First input is the area
/2 = .025
3 0.765
under 2.353
the tails, which is known as the level
of significance () and is equal to one minus 1- = 0.95
the level of confidence. In this example it is
5%, i.e. = 0.05. Second input is the degrees
of freedom, which is 30-1=29.
-2.045 0 2.045 t
• Hence tcritical = t.inv.2t(0.05,29) =
tcritical tcritical Chap 8-18
 Confidence Interval for μ
(σ Unknown) Example
Suppose that you collected a sample of 36 observations where the
sample mean and the sample standard deviation are calculated to be 50
and 8 respectively. Construct a 95% confidence interval for μ.
 d.f. = n – 1 = 36-1 = 35, so = t.inv.2t(0.05,35)=2.03
 The confidence interval is constructed as follows:

=[]

We can be 95% confident that the true population mean lies

somewhere between 47.29 and 52.71.
Chap 8-19
 Confidence Intervals for the
Population Proportion, π

 An interval estimate for the population proportion ( π ) can

be calculated using a similar approach where we construct
our interval with the point estimate for the population
proportion, i.e. sample proportion (p), and the standard error
of the point estimate, which is covered in the next slide.

Point Estimate ± (Critical Value) (Standard error of the point estimate)

p ± (Critical Value)
Chap 8-20
 Confidence Intervals for the
Population Proportion, π
The standard error of the sample proportion depends on two
factors: The population proportion () and the sample size (n).

 (1   )
σp 
n
 Obviously, this is problematic since we don’t know the actual
value of the population proportion.
 Hence, we will estimate this using our sample proportion and
sample size as follows:

p(1 p)
n
Chap 8-21
 Confidence Intervals for the
Population Proportion, π
Recall that the distribution of the sample proportion is
approximately normal if the sample size is large.
The formula for constructing the confidence interval for the
population proportion is given below:

where
p ± Z critical
√ 𝑝 ( 1 −𝑝 )
𝑛
 Z is the standardized normal value for the level of
confidence desired
 p is the sample proportion
 n is the sample size

Chap 8-22
 Confidence Intervals for the
Population Proportion, Example
In a random sample of 80 people, 20 of the people are observed to have the
Covid-19 virus. Form a 95% confidence interval for the true proportion of the
population that have the Covid-19 virus.

Note that the sample proportion is p = 20/80 = 0.25, implying that 25% of the
people in our sample has the virus.

= [0.1551]

We can be 95% confident that the proportion of people that have the Covid-19
virus in the population is somewhere between 15.51% and 34.49%.
Chap 8-23
 Determining Sample Size
 The required sample size can be found to reach a desired margin
of error (e) with a specified level of confidence (1 - ).

 The margin of error, denoted by “e”, is also called the sampling

error.

 It measures the amount of imprecision in the estimate of the

population parameter.

Chap 8-24
 Determining sample size to
estimate the population mean
 The margin of error, denoted by “e”, is also called the sampling error.
 If we are estimating the population mean, the margin of error is the
difference between the sample mean and the population mean:
e=
 Recall the standardization formula for the sample mean from Chapter
7:
𝑋 −𝜇 𝑒 𝑒 √ 𝑛 𝑍𝜎 𝑍 2
𝜎 2
𝑍= 𝑍= 𝑍= = √𝑛
𝜎 𝜎 =𝑛
𝜎 𝑒 𝑒
2

√𝑛 √𝑛
The sample size necessary to achieve a given level of
𝑍 2 𝜎2 margin of error for a given level of confidence, both of
𝑛=
𝑒
2 which to be determined by the researcher, is given by
this formula.
Chap 8-25
 Determining sample size to
estimate the population mean
 To determine the required sample size to estimate
the population mean, you must know:
 The desired level of confidence (1 - ), which
determines the critical Z value
 The acceptable sampling error (margin of error), e
 The standard deviation, σ

𝑍 2 𝜎2
𝑛= 2
𝑒

Chap 8-26
 Determining sample size to estimate the
population mean (example)

Solution:
For 95% confidence, use Z = 1.96
Margin of error = e = $0.5
Population standard deviation = = $2.5
𝑍 2 𝜎2 = 96.04
𝑛= 2
𝑒
n = 96.04 = approximately 96 observations needed to ensure that the margin of
error in estimation of the population mean will be at most $0.5.

If the confidence level was 99% instead, then sample size necessary to achieve
this margin of error would have been larger.
Chap 8-27
 Determining sample size to estimate the
population mean when σ is unknown

 If σ is unknown, we can collect a pilot sample

just to estimate the sample standard deviation (s)
first and then replace the σ value in the initial
sample size formula.
 This would also necessitate the use of critical t-
values instead of the Z-scores.

𝑡 2 𝑠2
𝑛= 2
𝑒

Chap 8-28
 Determining sample size to estimate
the population proportion
To determine the required sample size for the proportion, you must know:
 The desired level of confidence (1 - ), which determines the critical
Z value
 The acceptable sampling error (margin of error), e
 The true proportion of “successes”, π
 π can be estimated with a pilot sample, if necessary (or
conservatively use π = .50)

𝑍=

√
𝑝−𝜋
𝜋 (1 − 𝜋 )
𝑛
𝑍=

√
𝑒
𝜋 (1 − 𝜋 )
𝑛
√ 𝜋 ( 1− 𝜋 ) 𝑒
𝑛
=
𝑍
𝜋 ( 1− 𝜋 ) 𝑒 2
𝑛
= 2
𝑍

𝑍 2 𝜋 (1 − 𝜋 ) The sample size necessary to achieve a given level of margin of

𝑛= 2 error for a given level of confidence, both of which to be
𝑒 determined by the researcher, is given by this formula.
Chap 8-29
Determining Sample Size

Chap 8-30
 Determining Sample Size

Solution: The level of confidence is given as 90% hence Zcritical = 1.645

Margin of error is given as e = 0.04
Pilot Study πpilot = 0.235, so use this value to replace π in the formula.

𝑍 2 𝜋 (1 − 𝜋 )
𝑛= 2
=304
𝑒
With 90% level of confidence, a sample of 304 observations will ensure that the
margin of error in our estimation of the population proportion is at most 4%.
If the margin of error were larger (i.e. 0.06), then sample size necessary to
achieve this margin of error would have been smaller. Chap 8-31