Constructing and
Interpreting
Confidence Intervals
Confidence Intervals
In this power point, you will learn:
 Why confidence intervals are important in evaluation
research
 How to interpret a confidence interval
 How to construct a confidence interval yourself
Confidence Intervals
• Purpose
• The purpose of a confidence interval is to estimate a
population mean or proportion based on a sample mean or
proportion.
• Relationship with Margin of Error
• When estimating the population mean or proportion, we do
not try to estimate the exact value. We construct an
interval within which we are 95% sure the population
means or proportion lies. The width of the interval on each
side of the mean or proportion is referred to as the margin
of error.
• Example…….If you estimate that the percent of farmers
who will increase crop yield using the new fertilizer is 75%
and the margin of error is 5%, more technically you are
stating that you are 95% sure the population percent is
somewhere between 70% and 80%.
• Example using a Mean…..If you estimate the mean
number of bushels of corn to be 400 for farmers who used
the new fertilizer and our margin of error is 10, then you
are stating you are 95% sure the mean is somewhere
between 390 to 410.
Constructing a Confidence Interval
• Constructing a confidence interval by hand
• Before you can construct a confidence interval by hand,
you must have data from a random sample. You need:
• The sample size
• The proportion or mean you want to estimate
• The standard deviation of the mean you want to
estimate
• A critical value associated with the probability
• For a proportion this is a Z value
• The value associated with a 95% confidence interval
is l.96
• Other values can be found in a Z table
• For a proportion this is a T value
• The value associated with a 95% confidence interval
is l1.98
• Other values can be found in a T table
Constructing a Confidence Interval
• It is relatively easy to construct a confidence interval.
The following slides tell you how to do it.
 Construct a confidence interval using a proportion or
percent
 Statistical formula (most precise)
 Formula used by Opinion-Polls (most commonly used)
 Simplified formula used by Opinion-Polls (easiest and most
conservative)
 Construct a confidence interval using a mean
 Examples of Constructing Confidence Intervals
EXAMPLE
50% of our sample of 36 individuals indicated they would vote
for Obama in this election.
Research Question: What percent of population will vote for
Obama?
Relevant Information
p = .50
z = 1.96 for 95% confidence interval
n = 36
Simple Instructions
1.96 times the square root of .50 (1 - .50) divided by 36
(number of class members who voted)
Take this number and add it to .50 and then subtract it from .50
(proportion who voted for Obama in sample)
34 to 66% of the
Or Step by Step
population will vote
Step 1
1-.50 = .50
for Obama. NOTE,
Step 2
.50 times .50 = .25
our confidence
interval is so wide
Step 3
.25 divided by 36 (n) = 6.944
primarily because
Step 4
Square root of 6.944 = .08333
our sample is so
Step 5
1.96 times .08333 = .16333
small.
Step 6
.50 plus .16333 = .66333
Step 7
.50 minus .16333 = .33667
Step 8
Confidence interval is .33667 to .66333
We are 95% sure that the mean score is between
73.1 and 78.5.
Example
Research Question
What is the mean GPA of participants after
participating in our tutoring program?
Relevant Information
Sample GPA = 3.224
Sample Size = 40
T Value = 1.98 (value for a 95% confidence
interval.)
Standard Deviation of sample = .1864
Step by Step Instructions
Step 1 1.98 times .1864=.3691
Step 2 Square root of 40 = 6.3246
Step 3 .3691/6.3246=.0584
Step 4 3.224 plus .0584=3.2824
Step 5 3.224 minus .0584=3.1656
Conclusion: We are 95% sure that the mean GPA
for our entire group is between 3.1656-3.2824
Width of Confidence Intervals
Factors that influence the width of the interval.
 Standard Error
 Standard error estimates the dispersion of the data in the
population. The more dispersed your data the wider is your
confidence interval*
 Sample Size
 The smaller the sample size that you used to construct a
confidence interval then the wider the confidence interval*. In
layman’s terms, you can estimate a population mean or
proportion more accurately if you have a larger sample size.
 Probability
 The greater the probability, the wider will be your confidence
interval*. Thus a 99% confidence interval is wider than a
95%** confidence interval. In layman’s terms, the more certain
you want to be that the proportion or mean actually does lie
within the interval the wider the interval must be.
 Proportion as a Special Case
 When estimating a proportion then the closer to the middle (.5)
the proportion is the wider will be your confidence interval.
*The margin or error is directly related to the width of the confidence interval. The wider
the interval, the greater the margin of error.
**Rule of thumb is to use a 95% confidence interval, but you can construct an interval
with any degree of certainty.
Contact Information
• Dr. Carol Albrecht
• USU Extension
• Assessment Specialist
• Carol.albrecht@usu.edu
• 979-777-2421