An Introduction To Statistics-4

SECTION 7.
1
Hypothesis Test for a Single Population Mean (Large Samples)
A college statistics instructor claims that the mean age of college statistics students is
24. Do we just assume that the instructor is right, or is there a way we can find out
whether the claim is true? We will develop a five-step procedure to determine the
validity of such a claim. The procedure is called a hypothesis test. The test is based
on the confidence intervals that we developed in Chapter 6, and in many ways mir-
rors our legal system. Lets take a look at a confidence interval example to develop
some intuition for this process.
E XA MP L E A college statistics instructor claims that the mean age of college

7.1 statistics students is 24. A random sample of 116 college statistics
students revealed a mean age of 22.7 years with a standard devia-
tion of 5.68 years. Construct a 95% confidence interval for the true
mean age of all college statistics students. Use the interval to
determine the validity of the instructors claim.
First, we find the critical value z/2 for a 95% confidence interval ( = 0.05).
1.96 0 1.96
We see that z/2 = 1.96. Now we determine the margin of error.

s
E = z/2
n
5.68
= 1.96
116
= 1.03
Next we construct the confidence interval.
x E x + E
22.7 1.03 22.7 + 1.03
21.67 23.73
312
7.1 Hypothesis Test for a Single Population Mean (Large Samples) 313
The confidence interval tells us that we are 95% sure that the true mean age for
all college statistics students is between 21.67 and 23.73 years old. There is a 5%
chance that the mean is not in this interval, which tells us that there is less than a 5%
chance that the mean is actually 24. The instructors claim appears to be invalid.
Now we develop the five-step procedure for a hypothesis test. We will come back
to this example several times to show how a hypothesis test goes hand in hand with
a confidence interval. The five steps are as follows.
State the claim and the hypothesis

Choose a level of significance
Choose a test statistic
State a decision rule
Do the calculations and draw conclusions
Step 1: State the Claim and the Hypotheses

The claim is the statement that we are interested in investigating. Each hypothesis test
should begin by identifying the claim to be tested, and writing it in words. For the pre-
vious example, the claim would be The mean age of college statistics students is 24.
Once the claim has been identified, we need to write it in symbolic form. In our
claim, we are comparing a mean of a population to the number 24. We let repre-
sent the mean age of all college statistics students, and our claim becomes = 24.
The next thing that we do in Step 1 is to write the complement of the claim in
symbolic form. What is the complement of = 24? It is 24.
Suppose the claim had been < 24? What is the complement of less than 24?
At first, we might be tempted to say greater than 24, but this would not be correct.
The mean could be less than 24, greater than 24, or equal to 24. If the claim only uses
one of these, the complement combines the other two. Therefore, the complement of
< 24 is 24.
E XA MP L E Here are the six different claims that could be made comparing the
mean age of all college statistics students to 24.
1. The mean age of all college statistics students is 24.

Claim: = 24 Complement: 24
2. The mean age of all college statistics students is not 24.
Claim: 24 Complement: = 24
3. The mean age of all college statistics students is less than 24.
Claim: < 24 Complement: 24
4. The mean age of all college statistics students is at least 24.
Claim: 24 Complement: < 24
5. The mean age of all college statistics students is greater than 24.
Claim: > 24 Complement: 24
6. The mean age of all college statistics students is at most 24.
Claim: 24 Complement: > 24
314 CHAPTER 7 One-Sample Hypothesis Tests
Once the claim and its complement have been established, we must decide which
of those two is the null hypothesis (H0) and which is the alternate hypothesis (HA).
Of the claim and its complement, the null hypothesis is the statement containing
equality. In other words, we are looking for the symbols =, , or . We assume that
the null hypothesis is true, and look for sample evidence to suggest that it is false. In
our example, we are looking for a sample mean that is significantly lower than 24 or
significantly higher than 24 to suggest that the true mean is not 24.
The assumption that the null hypothesis is true is equivalent to assuming that a
person on trial is innocent (until proven guilty). If there is sufficient evidence to prove
that the assumption of the persons innocence is false beyond a reasonable doubt,
then we reject our assumption that the person is innocent and declare him guilty. The
evidence in a hypothesis test comes from a sample. If there is enough sample evidence
to suggest that the null hypothesis is false, we reject the null hypothesis and accept
the alternate hypothesis.
When we assume that the null hypothesis is true, we are actually assuming that
the population mean is equal to the value in the claim. We assume that = 24
whether the null hypothesis is = 24, 24, or 24. We say that the claimed
value of the mean is 24.
After the null hypothesis has been determined, the alternate hypothesis is the
statement that does not contain equality. If we reject the null hypothesis as false, then
we accept the alternate hypothesis as true. If we fail to reject the null hypothesis, we
are not saying that the null hypothesis is true. We can only say that there is not
enough sample evidence to convince ourselves that it is false. In this case, we fail to
support the alternate hypothesis as true. When there is not enough evidence to find a
defendant guilty, is the defendant then found innocent? No, the defendant is found
not guilty. This means that it was not proven that the person was innocent. There
was simply not enough evidence to conclude that the person was guilty.
Reject null hypothesis: support alternate hypothesis

Fail to reject null hypothesis: fail to support alternate hypothesis
E XA MP L E Lets take a look at three claims, and try to determine what the null
and alternate hypotheses are.
1. The mean number of hours spent studying per week by college students is less
than 10 hours.
Claim: < 10 H0: 10
Complement: 10 HA: < 10
2. The mean price for a home sold in Visalia, California is $100,000.
Claim: = 100,000 H0: = 100,000
Complement: 100,000 HA: 100,000
3. The mean starting salary for a college graduate is at most $30,000.
Claim: 30,000 H0: 30,000
Complement: > 30,000 HA: > 30,000
There are two types of errors that we can make in a hypothesis test. A Type I error
is rejecting a null hypothesis that is actually true. This is equivalent to convicting an
innocent man. A Type II error is failing to reject a null hypothesis that is actually false.
This is equivalent to letting a guilty man go free. We try to minimize Type I errors in
hypothesis testing. We control the chances of making a Type I error by establishing a
level of significance.
Step 2: Choose a Level of Significance ()

We want to reject the null hypothesis only if the probability that it is true is very small.
By very small we usually use 5% or 1%. In the case of 5%, we reject the null hypoth-
esis if there is less than a 5% chance that the null hypothesis is true. In other words,
if the chances that the null hypothesis is true are low enough, then we will declare it
false. This 5%, or 1%, is referred to as the level of significance. We use the Greek
letter alpha () to stand for the level of significance. This is equivalent to the phrase
reasonable doubt in a trial. We will discuss the use of in confidence intervals and
hypothesis tests when we get to Step 4.
The level of significance that is used most frequently is 0.05. However, if the con-
sequences of rejecting a null hypothesis that is true are serious, it is common to use
1.1 instead. Suppose a company is testing a new drug that they believe lowers high
blood pressure. A lower level of significance is appropriate here, because of the seri-
ous health implications.
Step 3: Choose a Test Statistic

The test statistic is a formula that will be used to determine the probability that the
null hypothesis is true based on the sample evidence. For the large-sample, single-
mean test the formula is
Test Statistic for a Large-Sample, Single-Mean Test

(Population Standard Deviation Known)
x
z =
/ n
where x = sample mean
= claimed population mean
= population standard deviation
n = sample size
As mentioned earlier in the text, the population standard deviation is often unknown.
If the sample size is 30 or greater, the sample standard deviation is an acceptable esti-
mate of the population standard deviation.
Test Statistic for a Large-Sample, Single-Mean Test

(Population Standard Deviation Unknown)
x
z =
s/ n
where x = sample mean
= claimed population mean
s = population standard deviation
n = sample size
This test statistic is the same formula we used in the previous chapter when try-
ing to find probabilities involving a sample mean. If we are assuming that the value
of is correct, this test statistic will allow us to calculate the probability of obtaining
a sample mean that is as extreme as the mean of our sample.
The test statistic will change for different types of hypothesis tests. The above for-
mula applies only to a test for a single population mean when the sample size is at
least 30. In this chapter there will be a second test statistic for a small-sample, single-
mean test, and a third for the large-sample, single-proportion test.
Step 4: State the Decision Rule

When we state the decision rule, we are establishing the guidelines for rejecting the
null hypothesis. We will be finding the critical value(s) for our test statistic. If the value
of the test statistic is beyond the critical value for the test, we reject the null hypothe-
sis. The decision rule for this test depends on two things: the alternate hypothesis and
the level of significance.
We return to our example to explore the establishment of a decision rule.

statistics students is 24. Test his claim at the 0.05 level of signifi-
cance. A random sample of 116 college statistics students revealed
a mean age of 22.7 years with a standard deviation of 5.68 years.
Step 1
Claim in words: The mean age of all college statistics students is 24.
Claim: = 24
Complement: 24
H0: = 24
HA: 24
Step 2
Level of significance: = 0.05
Step 3
x
Test statistic: z =
s/ n
To establish our decision rule, we must first decide what types of sample means
suggest that the null hypothesis is false and, more important, that the alternate
hypothesis is true. There are two ways that the population mean could be not equal
to 24: it could be less than 24, and it could be greater than 24. This is an example
of a two-tailed test. It is called a two-tailed test because the sample means that
suggest the null hypothesis is false are found in the right tail of the z-distribution
( > 24) and the left tail of the z-distribution ( < 24).
To set up our decision rule, we look for the critical values that separate the outer
5% of the z-distribution from the middle 95%. We do this by dividing the level of
significance in half ( /2) and placing that amount in each of the tails. We then look
for the z-values that go with such a graph. We need to look for the z-value that is
associated with an area of 0.0250 (z = 1.96), and the z-value that is associated
with an area of 0.9750 (z = 1.96).
0.0250 0.0250
1.96 1 0 1 1.96
The shaded regions on the outside are called the regions of rejection. If the
value of the test statistic falls into one of these regions, we reject the null hypothe-
sis. Otherwise, we fail to reject the null hypothesis. All that remains to be done is to
formally state our decision rule.
Decision rule: Reject H0 if z < 1.96 or if z > 1.96.
Two-Tailed Tests A test is two-tailed if our alternate hypothesis involves the sym-
bol . To find our decision rule for a two-tailed test, divide the level of significance and
place that amount in the left and right tails. Find the critical values of z for the graph
that we have, and reject the null hypothesis if the value of the test statistic is outside
of these critical values.
We used the symbol z/2 for the critical value in a confidence interval in the pre-
vious chapter. Here is why we did that. Suppose we were going to construct a 95% con-
fidence interval for a population mean. The middle 95% leaves 5% of the area
under the curve in the two outer tails.
0.0250 0.0250
1.96 0 1.96
That 5% we call alpha (). If we divided in half, that would put /2 in each tail.
Thus, our critical values go along with the critical values that we would obtain by put-
ting /2 in each tail to begin with.
E XA MP L E Find the decision rule for a two-tailed test at the 0.01 level of
significance.
First we divide by 2, which gives us 0.005. We put 0.005 in the left tail and right
tail, which produces the graph shown here.
0.005 0.005
2 1 0 1 2
2.575 2.575
We look for the z-values associated with areas of 0.0050 (left tail) and 0.9950 (right
tail). From the graph, the decision rule is
Reject H0 if z < 2.575 or if z > 2.575
Right-Tailed Tests Suppose we had the following hypotheses.

H0: 30,000
HA: > 30,000
What type of sample mean would be required to suggest that the null hypothesis was
false and that the alternate hypothesis was true? To conclude that > 30,000, we
would have to have a sample mean that was significantly higher than 30,000, which
is the claimed value of . We find the sample means that are significantly higher than
30,000 in the right tail.
The decision rule for a right-tailed test differs from a two-tailed test because the
entire level of significance is put in the right tail. It is not divided in half, because
there are not two tails to put it in.

0 z
Once is placed in the right tail, we find the critical value in the same fashion as we
did with a two-tail test.
E XA MP L E Find the decision rule for a right-tailed test at the 0.01 level of
significance.
We put 0.01 in the right tail, which produces the graph shown here.
2 1 0 1 2
2.33
We look for the z-value associated with an area of 0.9900. From the graph, the
decision rule is
Reject H0 if z > 2.33
Left-Tailed Tests Suppose we had the following hypotheses.

H0: 10
HA: < 10
What type of sample mean would be required to suggest that the null hypothesis was
false and that the alternate hypothesis was true? To conclude that < 10, we would
have to have a sample mean that was significantly lower than 10, which is the
claimed value of . We find the sample means that are significantly lower than 10 in
the left tail.
The decision rule for a left-tailed test differs from the previous tests because the
entire level of significance is put in the left tail. It is not divided in half, because
there are not two tails to put it in.

z 0
Once is placed in the left tail, we find the critical value in the same fashion
as before.
E XA MP L E Find the decision rule for a left-tailed test at the 0.05 level of
7.7 significance.
We put 0.05 in the left tail, which produces the graph shown here.
2 1 0 1 2
1.645
We look for the z-value associated with an area of 0.0500. From the graph, the
decision rule is
Reject H0 if z < 1.645
Step 5: Do the Calculations and Draw Conclusions

The first task in Step 5 is to calculate the test statistic using the sample information.
Once that has been accomplished, we check our decision rule to make our decision
about the null hypothesis H0. Then we must draw our final conclusion about the
claim being tested. Lets begin with the calculation.
Recall our example: A college statistics instructor claims that the mean age of col-
lege statistics students is 24. Test his claim at the 0.05 level of significance. A random
sample of 116 college statistics students revealed a mean age of 22.7 years with a stan-
dard deviation of 5.68 years.
Lets calculate the test statistic for our example. There are four values that need
to be plugged into this statistic: the sample mean (x), the claimed population mean
(), the sample standard deviation (s), and the sample size (n). Here is what we get
when we plug the correct values into the test statistic.
x
z =
s/ n
22.7 24
=
5.68/ 116
= 2.47
If the value of our test statistic is 2.47, what is our decision about the null hypothe-
sis H0? Recall the decision rule:
0.9500
0.0250 0.0250
1.96 1 0 1 1.96
Reject H0 if z < 1.96 or if z > 1.96
Since 2.47 falls in a rejection region (2.47 < 1.96), our decision is to reject H0.
1.96 1 0 1 1.96
2.47
We still must say something about our claim; after all, we were testing the instruc-
tors claim, not the null hypothesis. Our conclusion about the claim depends on two
things: whether our claim was the null or alternate hypothesis, and whether the null
hypothesis was rejected or not.
If the claim is the null hypothesis, then our conclusion is easy to state. In such a
case, whatever our decision is about the null hypothesis, our conclusion about the
claim is the same. If the claim is the null hypothesis and the null hypothesis is
rejected, then the claim is rejected as well. We have proved the claim to be false. If the
claim is the null hypothesis and we fail to reject the null hypothesis, then we fail to
reject the claim. In this case we have failed to prove the claim to be false. Please note
that this is not the same as proving the claim to be true. When the claim is the null
hypothesis, either the claim is proved to be false or we fail to prove that it is false.
If the claim is the alternate hypothesis, then we either support it as being true or
we fail to support it. If the claim is the alternate hypothesis and the null hypothesis is
rejected, then the claim is supported. In other words, if we throw the null hypothesis
out as being false, then we accept the alternate hypothesis (the claim) as being true.
We have proved the claim to be true. If the claim is the alternate hypothesis and we
fail to reject the null hypothesis, then we fail to support the claim. In this case we have
failed to prove the claim to be true.
The possible conclusions are illustrated in the following chart. You may find it
handy to use as a guide as you work through the homework exercises, but you must
understand why each conclusion is the way it is. If you cannot understand and be able
to draw your own conclusion, then you will never fully understand the process of con-
ducting a hypothesis test.
Fail to Reject H0
Claim is H0 Reject the claim Fail to reject the claim

Claim is HA Support the claim Fail to support the claim
In our example, the claim was the null hypothesis. Since our decision was to reject
the null hypothesis, then our conclusion is that there is sufficient sample evidence to
reject the claim that the mean age of all college statistics students is 24.
P-Values
One more step that can be added to a hypothesis test is the calculation of a p-value.
The p-value for a hypothesis test is the probability of obtaining a sample as extreme
as our sample if the null hypothesis is actually true. We will find the probability of
obtaining a value as extreme as the test statistic, or more extreme, if the null hypoth-
esis is indeed true. When calculating a p-value, we need to know two things: what
tailed test we have and the value of the test statistic.
Two-Tailed P-Values We find a p-value for a two-tailed hypothesis test by putting

the value of the test statistic on a normal distribution graph, finding the area in the
tail that it creates, and multiplying by 2. Multiplying by 2 takes into consideration the
tail on the opposite side.
For our example, we found a test statistic of 2.47. Put it on a normal distribution
graph, and find the area in the left tail.
p-value = 2 P(Z < 2.47)

= 2 (0.0068)
= 0.0136
0.0068
2 1 0 1 2
2.47
This tells us that the probability of obtaining a sample mean as extreme as ours if the
null hypothesis is actually true is only 0.0136. This is a much stronger statement than
saying that there is less than a 0.05 probability that it is true.
The decision rule can be rewritten in terms of the p-value and the level of signifi-
cance. Recall that we want to reject the null hypothesis only if the probability that it
is true is less than the level of significance. This translates to the following:
Reject H0 if p-value <
In our example, since the p-value was less than 0.05, the correct decision is to reject
the null hypothesis. What if the level of significance had been 0.01? We would have
failed to reject the null hypothesis, because our p-value is greater than 0.01.
Left-Tailed P-Values We find a p-value for a left-tailed hypothesis test by putting

the value of the test statistic on a normal distribution graph, and finding the area to
the left of that value. If we had a left-tailed hypothesis test, and the test statistic was
calculated to be 1.57, here is how the p-value would be calculated.
p-value = P(Z < 1.57)

= 0.0582
2 1 0 1 2
1.57
Right-Tailed P-Values We find a p-value for a right-tailed hypothesis test by put-

ting the value of the test statistic on a normal distribution graph, and finding the area
to the right of that value. If we had a right-tailed hypothesis test, and the test statistic
was calculated to be 2.29, here is how the p-value would be calculated.
p-value = P(Z > 2.29)

= 1 0.9890
= 0.0110
2 1 0 1 2
2.29
Our example has been spread out over the entire section. Before moving on, here
it is from start to finish, all in one example.

statistics students is 24. Test his claim at the 0.05 level of signifi-
cance. A random sample of 116 college statistics students revealed
a mean age of 22.7 years with a standard deviation of 5.68 years.
Step 1
Claim in words: The mean age of all college statistics students is 24.
Claim: = 24
Complement: 24
H0: = 24
HA: 24
Step 2
Step 3
x
Test statistic: z =
s/ n
Step 4
Two-tailed test, = 0.05
0.9500
0.0250 0.0250
1.96 1 0 1 1.96

Step 5
x
z =
s/ n
22.7 24
=
5.68/ 116
= 2.47
Decision: Reject H0.

Conclusion: There is sufficient sample evidence to reject the claim that the mean
age of all college statistics students is 24.
p-value
p-value = 2 P (Z < 2.47)

= 2 (0.0068)
= 0.0136
2 1 0 1 2
2.47
Now we will look at three more examples in their entirety. Please use them as a
reference while you do the homework exercises. Although this process may seem for-
eign and difficult, it will be used throughout most of the remainder of the book, so you
need to get comfortable with it now. There will be small changes from section to sec-
tion (different test statistics, different ways of finding the decision rule, . . .) but the
general process will remain the same.
E XA MP L E Test the claim that the mean price for a dozen roses at a California
florist is higher than $45 at the 0.01 level of significance. A random
sample of 30 California florists had a mean price of $48.33 per
dozen with a standard deviation of $6.07.
Step 1
Claim in words: The mean price for a dozen roses at a California florist is higher
than $45.
Claim: > 45
Complement: 45
H0: 45
HA: > 45
Note that the claim is the alternate hypothesis, so our conclusion will be either to
support the claim or fail to support the claim. Also note that the alternate hypothe-
sis involves the symbol >. This tells us that our test is a right-tailed test.
Step 2
Step 3
x
Test statistic: z =
s/ n
Since we are comparing the mean of a population to a certain number ($45) and
the sample size is at least 30, the above test statistic is the appropriate one.
Step 4
Right-tailed test, = 0.01
0.9900
0.01
2 1 0 1 2
2.33
Decision rule: Reject H0 if z > 2.33.

Step 5
x
z =
s/ n
48.33 45
=
6.07/ 30
= 3.00

Conclusion: There is sufficient sample evidence to support the claim that the mean
price for a dozen roses at a California florist is higher than $45.
p-value
p-value = P (Z > 3.00)

= 1 0.9987
0.0013 = 0.0013
0 3.00
E XA MP L E Test the claim that the mean amount spent by a customer at

7.10 Dylans Place, a local fast-food restaurant, is less than $5 at the
0.05 level of significance. A random sample of 34 customers had
spent a mean of $4.43 and a standard deviation of $3.23.
Step 1
Claim in words: The mean amount spent by a customer at Dylans Place is less
than $5.
Claim: < 5
Complement: 5
H0: 5
HA: < 5
sis involves the symbol <. This tells us that our test is a left-tailed test.
Step 2
Step 3
x
Test statistic: z =
s/ n
Since we are comparing the mean of a population to a certain number ($5) and the
sample size is at least 30, the above test statistic is the appropriate one.
Step 4
Left-tailed test, = 0.05
2 1 0 1 2
1.645
Decision rule: Reject H0 if z < 1.645.

Step 5
x
z =
s/ n
4.43 5
=
3.23/ 34
= 1.03
Decision: Fail to reject H0.

Conclusion: There is not sufficient sample evidence to reject the claim that the
mean amount spent by a customer at Dylans Place is less than $5.
p-value
p-value = P (Z < 1.03)

= 0.1515
0.1515
2 1 0 1 2
1.03
E XA MP L E Test the claim that the mean time Americans spend watching TV
7.11 per day is 4.4 hours at the 0.05 level of significance. A random
sample of 80 people had a mean time of 2.225 hours viewing per
day and a standard deviation of 1.088 hours.
Step 1
Claim in words: The mean time spent watching TV by Americans per day is
4.4 hours.
Claim: = 4.4
Complement: 4.4
H0: = 4.4
HA: 4.4
Note that the claim is the null hypothesis, so our conclusion will be either to reject
the claim or fail to reject the claim. Also note that the alternate hypothesis involves
the symbol . This tells us that our test is a two-tailed test.
Step 2
Step 3
x
Test statistic: z =
s/ n
Since we are comparing the mean of a population to a certain number (4.4 hours)
and the sample size is at least 30, the above test statistic is the appropriate one.
Step 4
Two-tailed test, = 0.05
1.96 1 0 1 1.96

Step 5
x
z =
s/ n
2.225 4.4
=
1.088/ 80
= 17.88

time Americans spend watching TV per day is 4.4 hours.
p-value
Since 17.88 does not come close to being on our z-table, we will use the biggest
value we havenamely, 3.50.
p-value = 2 P (Z < 17.88)

< 2 P (Z < 3.50)
= 2 (0.0002)
0.0002 = 0.0004
3.50 0
The p-value for this test is (much) less than 0.0004.
MICROSOFT EXCEL Hypothesis Tests for a Single

Population Mean
TWO-TAILED CRITICAL VALUES
To use Excel to find two-tailed critical values, we first must find /2. If the level of sig-
nificance () is 0.05, then /2 = 0.025. To find the critical value for the left tail, type
the following in any cell.
=NORMSINV(0.025)
This Excel function returns the z-value for which the area to the left of it is equal to
1.25. Round Excels result to two decimal places. To find the critical value for the right
tail, type the following in any cell.
=NORMSINV(1-0.025)
We must subtract 0.025 from 1 to obtain the area to the left of the right tail. Again,
round the result to two decimal places. For values of /2 other than 0.025, replace
0.025 by the appropriate value.
RIGHT-TAILED CRITICAL VALUES

To use Excel to find a right-tailed critical value, we need . Suppose the level of sig-
nificance was 0.01; in other words, = 0.01. To find the critical value for the right
tail, type the following in any cell.
=NORMSINV(1-0.01)
Round Excels result to two decimal places.
LEFT-TAILED CRITICAL VALUES

To use Excel to find a left-tailed critical value, we need . Suppose that = 0.05. To
find the critical value for the left tail, type the following in any cell.
=NORMSINV(0.05)
Round Excels result to two decimal places.
TWO-TAILED p-VALUES
If the calculated value of the test statistic is negative, we need to multiply the area to
the left of the test statistic by 2. In any cell, we type the following.
=2*NORMSDIST(z-value)
If the test statistic was calculated to be z = 2.47, then we would type

=2*NORMSDIST(-2.47)
Round all p-values to four decimal places.

We can also use Excel to calculate two-tailed p-values if the calculated value of
the test statistic is positive. In such a case, we type the following in any cell.
=2*(1-NORMSDIST(z-value))
For example, if we calculated the test statistic to be z = 1.23, then we would type the
following in any cell.
=2*(1-NORMSDIST(1.23))
We must subtract the result from 1 because Excels NORMDIST function gives us the
area to the left of the z-value, but we need the area in the right tail.
LEFT-TAILED p-VALUES
Since we are looking for the area to the left of a particular z-value, NORMSDIST will
work without any adjustments. In any cell, we type the following.
=NORMSDIST(z-value)

=NORMSDIST(-1.57)
Round that answer to four decimal places.

RIGHT-TAILED p-VALUES
Because we are looking for the area to the right of a particular z-value, we must sub-
tract the value that NORMSDIST gives us from 1. In any cell, we type the following.
=1-NORMSDIST(z-value)

=1-NORMSDIST(2.29)
Round that answer to four decimal places.

Here is an example that uses these tools in performing a hypothesis test.
E XA MP L E An algebra instructor has designed a final exam that he believes will

7.12 produce a mean score of 70. Mr. Thomas, one of his colleagues,
disagrees, claiming that the mean score for all algebra students on
this exam will be below 70. Mr. Thomas randomly selects 38 alge-
bra students, and gives them the exam. Here are their scores.
32 41 43 44 46 46 47 48 49 50
52 54 54 56 57 58 58 60 60 61
65 66 66 67 67 68 68 68 68 71
72 76 83 87 92 92 94 100
Use these sample data to test Mr. Thomas claim at the 0.01 level
of significance.
We will need the sample statistics (mean and standard deviation).
Open a new Excel worksheet, and type the sample values in column A (cells
A1 through A38).
In cell B1, type Sample Mean. In cell B2, type Sample Standard Deviation.
In cell B3, type Sample Size.
Now that the framework is set up, we can begin the calculations.
To calculate the sample mean, type =AVERAGE(A1:A38) in cell C1.

To
C2.
calculate the sample standard deviation, type =STDEV(A1:A38) in cell
Type the sample size (38) in cell C3.

We begin with the hypothesis test by hand, and then will turn to Excel when
needed.
Step 1
Claim in words: The mean score for all algebra students on this exam is below 70.
Claim: < 70
Complement: 70
H0: 70
HA: < 70
sis involves the symbol <. This tells us that our test is a left-tailed test.
In cell B4, type Claimed Mean. Type the claimed value of the mean (70) in
cell C4.
Step 2
Step 3
x
Test statistic: z =
s/ n
Since we are comparing the mean of a population to a certain number (70) and the
sample size is at least 30, the above test statistic is the appropriate one.
Step 4
Left-tailed test, = 0.01
Type Critical Value in cell B6. We will calculate the critical value for this test
in cell C6. Since this is a left-tailed test, type =NORMSINV(0.01).
The value Excel gives us is -2.32634. Our decision rule is then
Reject H0 if z < 2.32634
Here is the critical value using our usual method.
0.01
2 1 0 1 2
2.33

Step 5
In cell B7, type Test Statistic.
Next we calculate the test statistic in cell C7. Type =(C1-C4)/(C2/SQRT(C3)).
The result Excel gives us is 2.7704. Here is the formula worked out.
x
z =
s/ n
62.8 70
=
16.04/ 3 8
= 2.77

score for all algebra students on this exam is below 70.
p-value
Type p-value in cell B9. We will calculate the p-value for this test in cell C9.
Since we are looking for a left-tailed p-value, type =NORMSDIST(C7). The
result that Excel gives us is 0.002799.
p-value = P (Z < 2.77)

= 0.5 0.4972
= 0.0028
2 1 0 1 2
2.77
Here is what your Excel worksheet should look like.

32 Sample Mean 62.78947
41 Sample Standard Deviation 16.04412
43 Sample Size 38
44 Claimed Mean 70
46
46 Critical Value 2.32634
47 Test Statistic 2.7704
48
49 p-value 0.002799
50
52
54
54
56
57
58
58
60
60
61
65
66
66
67
67
68
68
68
68
71
72
76
83
87
92
92
94
100
Hypothesis Tests for a Single Population Mean
The TI-83 can help us to perform a hypothesis test, as shown in the following
example.
E XA MP L E An algebra instructor has designed a final exam that he believes will

7.13 produce a mean score of 70. Mr. Thomas, one of his colleagues,
disagrees, claiming that the mean score for all algebra students on
this exam will be below 70. Mr. Thomas randomly selects 38 alge-
bra students, and gives them the exam. Here are their scores.
32 41 43 44 46 46 47 48 49 50
52 54 54 56 57 58 58 60 60 61
65 66 66 67 67 68 68 68 68 71
72 76 83 87 92 92 94 100
Use these sample data to test Mr. Thomas claim at the 0.01 level
of significance.
Put these 38 scores in list L1. We will need sample statistics before continuing on to
the hypothesis test. Press STAT , then use the key to move to the CALC
menu. Select item 1: 1-Var Stats. On the main screen, press 2nd 1 to access
list L1, and press ENTER . Here is the screen of sample information.
We can use 62.8 for the sample mean, and 16.04 for the sample standard devia-
tion. Since the sample size of 38 is at least 30, we may use the sample standard
deviation as an estimate of the population standard deviation. To begin the hypoth-
esis test, Press STAT , then use the key to move to the TESTS menu.
Select item 1: Z-Test. Fill in the screen in the following manner.
Inpt: Highlight Stats. We use this option when we type in the standard devi-
ation, mean, and sample size manually. We use the Data option to take these
statistics from a list.
0: Enter the claimed value of the mean, which is 70.
: Enter the value that will be used for the population standard deviation.
Again, in this example, we will use the sample standard deviation, which is
16.04.
: Enter the value of the sample mean. You may find that it has already been
xentered for you. If it has not, use 62.8.
n: Enter the sample size. Again, you may find that it has already been entered
for you. If it has not, use 38.
: Select the form of the alternate hypothesis. The claim for this test is that the
mean score for all algebra students on this exam is below 70, so we are test-
ing H0: 70 against HA: < 70. Enter the second option, <0.
Highlight Calculate and press ENTER .
You will see the following screen, which lists the alternate hypothesis, calculated
value of the test statistic, p-value, sample mean, and sample size.
EXERCISES 7.1
Explain the following, in your own words.
1. How to write the claim in symbols
2. How to write the complement of the claim in symbols
3. How to identify the null hypothesis
4. How to identify the alternate hypothesis
5. The meaning of level of significance
6. The purpose of a test statistic
7. How to determine whether a test is left-tailed, right-tailed, or two-tailed
8. How to formulate the decision rule for a right-tailed test
9. How to formulate the decision rule for a left-tailed test
10. How to formulate the decision rule for a two-tailed test
11. The difference between the decision about the null hypothesis and the conclusion
about the claim
12. The meaning of a p-value
13. A skeptical student believes that the mean height of adult males is actually greater
than 69 inches. He randomly samples 45 adult males. The sample has a mean
height of 70.4 inches with a standard deviation of 2.44 inches. Test the students
claim at the 0.05 level of significance.
Step 1:
(a) State the claim in words.
(b) State the claim in symbols.
(c) State the complement of the claim in symbols.
(d) State the null hypothesis (H0).
(e) State the alternate hypothesis (HA).
Step 2:
(f) State the level of significance ().
Step 3:
(g) State the appropriate test statistic.
Step 4:
(h) What tailed test is this test?
(i) State the decision rule.
Step 5:
(j) Calculate the test statistic.
(k) State your decision about the null hypothesis (H0).
(l) State your conclusion about the claim.
For the remaining hypothesis tests, use the five-step procedure developed in this section.
Refer to Exercise 13 for help.
14. A counselor at a community college claims that the mean GPA for students who
have transferred to State University is higher than 3.0. A random sample of 45
such students had a mean GPA of 3.12 with a standard deviation of 0.31.
(a) Test the counselors claim at the 0.05 level of significance.
(b) Explain how you would gather the data to test this claim, and list any biases
that may be present using your method.
15. A student is interested in the mean life span of males in his town. He has heard
claims that the mean age at death for males in his town is 70 years. He randomly
selects 88 male deaths from the past year. The mean age at death for these 88 men
was 68.9 years, with a standard deviation of 16.71 years. Test the claim that the
mean age at death for males in his town is 70 years, at the 0.05 level of significance.
16. A student is concerned with the lack of effort among students at her school. She
claims that students spend 5 or fewer hours a week studying. A random sample of
100 students produces a mean study time of 5.73 hours per week, with a standard
deviation of 3.464 hours.
(a) Test the claim that the mean study time per week for these students is 5 or
fewer hours at the 0.05 level.
17. I recently heard someone say Movies are getting too long! It seems like there are
a lot of movies that are over two hours long! I claim that the mean length of
movies is less than 2 hours (120 minutes). A random sample of 90 recent releases
produced a mean length of 108.44 minutes with a standard deviation of 17.952
minutes. Test my claim at the 0.01 level of significance.
18. At the 0.05 level of significance, test the claim that the mean mileage for cars is
less than 20 miles per gallon. A sample of 48 cars had a sample mean of 16.9
miles per gallon, with a standard deviation of 6.83 miles per gallon.
19. A TV commercial claims that the mean cost of a funeral today is above $5000. A
survey of 60 funerals showed that they had a mean cost of $5235, with a standard
deviation of $1739.
(a) Test the commercials claim at the 0.05 level of significance.
20. A random sample of 100 women who smoked while pregnant showed that they
smoked a mean of 7.4 cigarettes per day with a standard deviation of 4.95 ciga-
rettes. (Based on data contained in Centers for Disease Control and Prevention
Monthly Vital Statistics Report.)
(a) Test the claim that women who smoke while pregnant smoke a mean of less
than 10 cigarettes per day, at the 0.01 level of significance.
21. A sample of 212 women who gave birth showed that they gained a mean of 26.2
pounds, with a standard deviation of 10.48 pounds. (Based on data contained in
Centers for Disease Control and Prevention Monthly Vital Statistics Report.)
(a) Test the claim that the mean weight gain while pregnant is higher than 25
pounds, at the 0.05 level of significance.
22. Test the claim that the mean number of prenatal visits for pregnant women is
higher than 10 visits at the 0.05 level of significance. A random sample of 441
women who gave birth showed that they had a mean of 11.2 prenatal doctor
visits, with a standard deviation of 3.76. (Based on data contained in Centers for
Disease Control and Prevention Monthly Vital Statistics Report.)
23. A racing publication reports that the average winning time for a six-furlong race
is 1 minute and 10.8 seconds (70.8 seconds). These are the winning times, in
seconds, for 63 six-furlong races at Santa Anita Park. (Six furlongs are equal to
three-quarters of a mile.)
71.4 72.6 71.0 70.0 71.8 72.0 70.0
71.0 70.8 69.4 70.0 71.0 68.8 71.0
69.2 70.2 72.2 71.0 71.6 69.8 70.4
69.8 70.0 70.2 69.4 69.2 70.6 69.0
70.6 72.2 71.2 71.4 70.2 72.2 72.2
70.6 71.6 69.6 71.2 70.4 69.0 68.6
70.8 69.4 69.6 69.4 70.2 69.4 70.0
68.8 69.8 70.6 71.0 70.8 69.2 70.8
71.0 71.0 70.8 71.0 71.0 69.2 71.6
Test the publications claim at the 0.01 level of significance.
24. A random sample of 120 fillets from farm-raised Atlantic salmon had a mean fat
content of 10.5 grams and a standard deviation of 1.12 grams. Use this sample to
test the claim that the mean fat content of farm-raised Atlantic salmon fillets is
higher than 8.5 grams at the 0.05 level of significance.
25. A random sample of 67 fillets from wild Atlantic salmon had a mean fat content
of 8.1 grams and a standard deviation of 1.48 grams. Use this sample to test the
claim that the mean fat content of wild Atlantic salmon fillets is lower than 8.5
grams at the 0.05 level of significance.
26. A survey of 65 beer drinkers revealed a mean consumption of 25.2 beers per
month, with a standard deviation of 4.96 beers. (Based on the results of a survey
by Maritz AmeriPoll.)
(a) Test the claim that the mean number of beers consumed per month by beer
drinkers is 24 beers (1 case) at the 0.05 level of significance.
27. A random sample of 300 incoming flights at an airport showed that travelers had
to wait in the baggage claim area for a mean of 18.9 minutes, with a standard
deviation of 5.73 minutes. Test the claim that the mean waiting time in the bag-
gage claim area is less than 20 minutes at the 0.01 level of significance.
28. Test the claim that the mean vitamin C content of medium-sized oranges is 66 mg
at the 0.05 level of significance. A sample of 36 medium-sized oranges had a
mean vitamin C content of 64.5 mg, with a standard deviation of 4.14 mg.
29. A politician claims that the mean number of hours Americans worked per week in
a typical week is greater than 40 hours. A sample of 324 American workers pro-
duced a mean of 41.3 hours per week with a standard deviation of 10.63 hours per
week. Test the politicians claim at the 0.05 level of significance. (Based on a study
by the Center for Survey Research and Analysis at University of Connecticut.)
30. A random sample of 74 recently engaged men showed that they had spent a
mean of $2982 on the engagement ring, with a standard deviation of $1691. At
the 0.05 level of significance, test the claim that the mean amount spent on
engagement rings by all men is above $2500. (Based on the results of a survey by
Brides magazine.)
31. How many e-mails does a person get at work per day? A random sample of 47 peo-
ple showed that they receive a mean of 10.7 e-mails at work per day, with a stan-
dard deviation of 10.53 e-mails. At the 0.05 level of significance, test the claim
that the mean number of e-mails received per day is higher than 5. (Based on a
survey by the John J. Heldrich Center for Workforce Development.)
32. Test the claim that the mean cost of a honeymoon is over $3000 at the 0.05 level
of significance. A random sample of 180 couples spent a mean of $3657 on their
honeymoons, with a standard deviation of $2109. (Based on a survey by Brides

magazine.)
33. A sample of 225 parents were asked how much time they spent per week on school
work or school-related activities. The sample produced a mean of 5.6 hours per
week, with a standard deviation of 4.4 hours. At the 0.01 level of significance, test
the claim that the mean number of hours spent by parents on school work or
school-related activities is 5 hours per week. (Based on a study by Yankelovich
Partners for LearningPays.com.)
34. One hundred one cruise ship passengers were selected at random, and they had a
mean age of 50.1 years old, with a standard deviation of 16.41 years. At the 0.05
level of significance, test the claim that the mean age of a passenger on a cruise
ship is below 55 years old. (Based on a study by Polk.)
35. A study of 42 restaurants showed that the mean cost for a sit-down dinner for one
person at these restaurants, not including tip, was $27.05 with a standard devia-
tion of $7.20. At the 0.05 level of significance, test the claim that the mean cost for
a sit-down dinner for one person at all restaurants is $25. (Based on a study by the
National Restaurant Association.)
36. Television networks and movie studios are always interested in the age of the audi-
ence they are reaching. A random sample of 104 people at a showing of Toy Story 2
had a mean age of 26.4 years with a standard deviation of 14.82 years. At the 0.05
level of significance, test the claim that the mean age of people going to see Toy
Story 2 is less than 30 years. (Based on the results of an informal survey by Disney
in six cities.)
37. A television golf analyst claims that the mean drive for professional golfers is 270
yards. Here are the lengths of 63 randomly selected drives from the 100th U.S.
Open at Pebble Beach.
305 269 289 287 293 289 289 256
288 283 300 229 262 312 274 286
268 282 290 270 281 296 306 256
280 273 293 268 284 266 254 242
283 270 277 293 262 270 273 261
275 276 279 263 273 261 281 241
262 258 274 261 264 260 266 234
252 259 284 257 270 281 250
At the 0.05 level of significance, test the television golf analysts claim. (Source:
On-line scoring for U.S. Open at lycos.com.)
What is the appropriate conclusion about the claim under the following conditions, if the
level of significance is 0.05?
38. Claim to be tested: Mean is less than 25.
p-value: 0.0885
39. Claim to be tested: Mean is greater than 17.1.
p-value: 0.0322
40. Claim to be tested: Mean is equal to 129.7.
p-value: 0.0039
41. Claim to be tested: Mean is not 25.
p-value: 0.0668
42. A colleague tells you that he is trying to verify the claim that households that sub-
scribe to a particular magazine have a mean income that is above $45,000. He is
elated that the mean income of 225 randomly selected households that subscribe
to the magazine is $49,245. Explain to him why this may not be enough to sup-
port his claim, and what other information may be necessary.
The following statistical project contains a major error. Potential errors could be choosing
the wrong hypothesis test or not meeting the assumptions required for this test. Find the
error, and explain what could be done to correct it.
M I N I P RO J E C T
I claim that the mean height for adult females is 5 5 (65 inches). You are to
test my claim at the 0.05 level of significance. Randomly sample 50 adult
females, and find out their height in inches. Use these sample data to test the
claim. In addition to your complete hypothesis test, include
Your raw data
A histogram for the heights
5A pie chart showing the percentages that are 5
3 and 5 7 , and 5 8 or taller
2 or shorter, between
Explain your technique of sampling. Which type of sampling did you use? Was
it truly random? What potential biases may show up in your sample?
EXCEL EXPLORATION The Effect of Sample Size
In this exploration, we will use Microsoft Excel to examine how the sample size affects
a hypothesis test. We will be testing the null hypothesis that a population mean is
equal to 50 by randomly generating samples from a population whose mean is 48
and whose standard deviation is 15. We know that the mean is not equal to 50 (it is
equal to 48), but will our hypothesis-testing procedure be able to reach that conclu-
sion? We will see that one factor that may play a part is the sample size.
If we are testing the null hypothesis that = 50 at the 0.05 level of significance,
the two-tailed decision rule is to reject the null hypothesis if z < 1.96 or z > 1.96. We
will use Excel to generate a sample, and then calculate the test statistic for that sample.
Before we begin we must be sure that the Analysis Tool Pak has been added. Pull
down the Tools menu. If you see a choice that says Data Analysis, then the Tool Pak
is added and you may continue. If you do not see Data Analysis, click on Add-ins
under Tools. When the dialog box opens up, click on the box for Analysis Tool Pak,
and click on OK.
We begin by generating 100 samples of size 30. Pull down the Tools menu and
click on Data Analysis. When the dialog box appears, click on Random Number
Generation and OK. When the next dialog box appears, enter 100 in the box labeled
Number of Variables. (This is the number of samples that we are going to take.) Enter
30 in the box labeled Number of Random Numbers. (This is the sample size.) In the
box labeled Distribution click on Normal. For Parameters, enter 48 in the box
labeled Mean, and 15 in the box labeled Standard Deviation.
Now we will calculate the test statistic. In cell A31, type =AVERAGE(A1:A30). This
is the mean of the sample. To calculate the standard deviation of the sample, type
=STDEV(A1:A30) in cell A32. Finally, to calculate the test statistic, type the following
in cell A33.
=(A31-50)/(A32/SQRT(30))
In the above formula, A31 represents the sample mean, 50 is the claimed population
mean, A32 is the sample standard deviation, and 30 is the sample size. Highlight cells
A31 through A33, then select Copy from the Edit menu. Highlight the cells in rows
31, 32, and 33 through column CV, then select Paste from the Edit menu. This will
calculate the test statistic for each of the 100 samples.
In a trial run, using 100 samples, the null hypothesis was not rejected a single
time. The results of your 100 samples will most likely be the same. Repeat this proce-
dure for samples of size 50, 100, 250, and 1000. Summarize your results, stating how
the increase in sample size affects the test statistic.
SECTION 7.2
Hypothesis Test for a Single Population Mean (Small Samples)
Recall that the sampling distribution of sample means follows the t-distribution for
samples of size less than 30. When we are trying to test a claim about a population
mean using a small sample (n < 30) and the population standard deviation is
unknown, it is appropriate to use the t-distribution as long as the sample comes from
a population that is normally distributed. With small samples, outliers suggest that
the population is skewed and not normally distributed. There is a hypothesis test in
Section 9.3 to test whether a set of data follows a normal distribution, but until then
you can use the material presented in ExtraIs a Set of Data Approximately
Normally Distributed? which follows Section 5.3. If it appears that the population is
not normally distributed, we may instead use a hypothesis test from Section 11.1. Here
is the test statistic for the small-sample test for a single population mean.
x
t = degrees of freedom = n 1
s/ n
The test statistic is identical to the one that we use for a large sample test when the
population standard deviation is unknown, but the decision rule will be different.
Before we take a look at a few hypothesis tests, we will review how to obtain the deci-
sion rule.
E XA MP L E Find the decision rule for a right-tailed test at the 0.01 level of signif-
7.14 icance with 22 degrees of freedom.
Since this is a one-tailed test, we look for 0.01 as a one-tailed value. When we
locate that column (the second column from the left), we go down until we reach
22 degrees of freedom.
One-Tailed Value
Degrees of Freedom 0.005 0.01 0.025 0.05 0.1 0.25

22 2.819 2.508 2.074 1.717 1.321 0.686
The critical value is 2.508. This value is positive, because we are dealing with the
right tail. Now for the decision rule. For a right-tailed test, we want to reject the null
hypothesis if the value of the test statistic is greater than the critical value.
Decision rule: Reject H0 if t > 2.508
E XA MP L E Find the decision rule for a left-tailed test at the 0.05 level of signifi-
7.15 cance with 9 degrees of freedom.
Since this is a one-tailed test, we look for 0.05 as a one-tailed value. When we
locate that column (the fourth column from the left), we go down until we reach
One-Tailed Value

9 3.250 2.821 2.262 1.833 1.383 0.703
The value in the chart is 1.833, so the critical value is 1.833. This value is negative,
because we are dealing with the left tail. For a left-tailed test, we want to reject the
null hypothesis if the value of the test statistic is less than the critical value.
Decision rule: Reject H0 if t < 1.833
E XA MP L E Find the decision rule for a two-tailed test at the 0.05 level of signif-
7.16 icance with 19 degrees of freedom.
Since this is a two-tailed test, we look for 0.05 as a two-tailed value. When we
locate that column (the third column from the left), we go down until we reach
Two-Tailed Value

19 2.861 2.539 2.093 1.729 1.328 0.688
The value in the table is 2.093, so the critical values are 2.093 and 2.093. We use
both the positive and negative values because we are dealing with both the right
and left tails. For a two-tailed test, we want to reject the null hypothesis if the value
of the test statistic is greater than the positive critical value or less than the nega-
tive critical value.
Decision rule: Reject H0 if t < 2.093 or if t > 2.093
Now we will perform some actual tests, using the same five-step procedure as in
the last section. The major difference is the decision rule.
7.2 Hypothesis Test for a Single Population Mean (Small Samples) 339
E XA MP L E Test the claim that the mean number of books read per month by
7.17 community college students is less than 2 at the 0.05 level of signif-
icance. A random sample of 28 community college students had
read a mean of 1.2 books with a standard deviation of 2.14 books.
Step 1
Claim in words: The mean number of books read per month by community college
students is less than 2.
Claim: < 2
Complement: 2
H0: 2
HA: < 2
Since the alternate hypothesis involves <, we have a left-tailed test. Also, since our
claim is the alternate hypothesis, we will either support or fail to support our claim.
Step 2
Level of significance: = 0.05.
Step 3
x
Test statistic: t = degrees of freedom = n 1
s/ n
This is the appropriate test statistic to use when testing a claim about a single pop-
ulation mean based on a small sample (n < 30) when the population standard devi-
ation is unknown.
Step 4
This is a left-tailed test, = 0.05, with 27 degrees of freedom.
One-Tailed Value

27 2.771 2.473 2.052 1.703 1.314 0.684
Decision rule: Reject H0 if t < 1.703.

Step 5
x
t =
s/ n
1.2 2
=
2.12/ 28
= 1.997

number of books read per month by community college students is less than 2.
p-value
Finding p-values for t tests are tricky when using the table. We need to go to the
chart, and look at the entire row for 27 degrees of freedom.
One-Tailed Value

27 2.771 2.473 2.052 1.703 1.314 0.684
Our calculated value for the test statistic (1.997) falls between 1.703 ( = 0.05) and
2.52 ( = 0.025). Our p-value is between 0.025 and 0.05. Using technology, we
will be able to obtain a more precise p-value later.
E XA MP L E A financial expert claims that the mean charitable contribution on

7.18 federal tax returns is $1000. Here are the contributions of eight
randomly selected individuals.
$329 $495 $552 $734

$951 $1379 $1394 $2550
At the 0.01 level of significance, test the experts claim. (Data based
on the results of a survey by Gallup for Independent Sector.)
First we need to calculate the mean and standard deviation for this sample. The
mean is $1048.00, and the standard deviation is $723.29.
Step 1
Claim in words: The mean charitable contribution on federal tax returns is $1000.
Claim: = 1000
Complement: 1000
H0: = 1000
HA: 1000
Since the alternate hypothesis involves , we have a two-tailed test. Also, since our
claim is the null hypothesis, we will either reject or fail to reject our claim.
Step 2
Step 3
x
s/ n
Step 4
This is a two-tailed test, = 0.01, with 7 degrees of freedom.
Two-Tailed Value

7 3.499 2.998 2.365 1.895 1.415 0.711
Decision rule: Reject H0 if t < 3.499 or if t > 3.499.

Step 5
x
t =
s/ n
1048 1000
=
723.29/ 8
= 0.188

Conclusion: There is not sufficient sample evidence to reject the claim that the
mean charitable contribution on federal tax returns is $1000.
p-value
We go to the chart, and look at the entire row for 7 degrees of freedom.
Two-Tailed Value

7 3.499 2.998 2.365 1.895 1.415 0.711
Our calculated value for the test statistic (0.188) falls below 0.711 ( = 0.5). Our
p-value is greater than 0.5.
E XA MP L E Mark, an avid golfer, claims that his average drive is greater than
7.19 250 yards. To prove this, he measured his 14 drives from a round
of golf. Here are the distances, in yards.
262 247 255 239 275 257 260

242 256 281 265 248 255 263
Test Marks claim that his average drive is greater than 250 yards
at the 0.05 level.
Again, we begin by calculating the mean and standard deviation for this sample.
The mean is 257.5 yards and the standard deviation is 11.67 yards.
Step 1
Claim in words: The mean length for all of Marks drives is greater than 250 yards.
Claim: > 250
Complement: 250
H0: 250
HA: > 250
Since the alternate hypothesis involves >, we have a right-tailed test. Also, since our
claim is the alternate hypothesis, we will either support or fail to support our claim.
Step 2
Step 3
x
s/ n
Step 4
This is a right-tailed test, = 0.05, with 13 degrees of freedom.
One-Tailed Value

13 3.012 2.650 2.160 1.771 1.350 0.694
Decision rule: Reject H0 if t > 1.771.

Step 5
x
t =
s/ n
257.5 250
=
11.67/ 14
= 2.405

length for all of Marks drives is greater than 250 yards.
p-value
We go to the chart, and look at the entire row for 13 degrees of freedom.
One-Tailed Value

13 3.012 2.650 2.160 1.771 1.350 0.694
Our calculated value for the test statistic (2.405) falls between 2.160 ( = 0.025)
and 2.650 ( = 0.01). Our p-value is therefore between 0.01 and 0.025.

Population Mean (Small Samples)
RIGHT-TAILED CRITICAL VALUES (t-DISTRIBUTION)
As with the standard normal distribution, Excel can help us find critical values for the
t-distribution. With a one-tailed test, we must multiply the level of significance by 2
to use Excels built-in function. This function returns critical values for two-tailed tests,
for which the level of significance is twice that of a one-tailed test. Here is the format.
=TINV(2*level of significance, degrees of freedom)
For example, for a right-tailed test at the 0.01 level of significance with 22 degrees of
freedom, type the following in any cell.
=TINV(2*0.01,22)
Excel returns a value of 2.508323. The decision rule is
Reject H0 if t > 2.508323
LEFT-TAILED CRITICAL VALUES (t-DISTRIBUTION)

The difference that arises with a left-tailed critical value, as compared to a right-tailed
critical value, is that Excel gives us a positive value. It is up to us to remember that
this value must become negative. We can take care of this by including a negative
sign before the functions name. Again, with a one-tailed test, we must multiply the
level of significance by 2 to use Excels built-in function.
For example, for a left-tailed test at the 0.05 level of significance with 9 degrees of
= -TINV(2*0.05,9)
Reject H0 if t < 1.83311

TWO-TAILED CRITICAL VALUES (t-DISTRIBUTION)

With a two-tailed test we only have to supply Excel with the level of significance (not 2
times the level of significance) and the number of degrees of freedom. Here is the format.
=TINV(level of significance, degrees of freedom)
For example, for a two-tailed test at the 0.05 level of significance with 19 degrees of
=TINV(0.05,19)
Reject H0 if t > 2.093025 or if t < 2.093025
Microsoft Excel can help us with the p-value. Excel will give us a precise p-value
instead of a range of values. We need to supply Excel with the value of the test statis-
tic (use positive values only), the degrees of freedom, and the number of tails. If the
test statistic is negative, you need to type it as a positive value. (There is no difference
in the way you enter a left-tailed test or a right-tailed test.) Here is the format.
=TDIST(test statistic, degrees of freedom, number of tails)
If we had a right-tailed test, with 13 degrees of freedom, and the test statistic cal-
culated to be 2.405, we would find the p-value by typing the following in any cell.
=TDIST(2.405,13,1)
The result that we get is 0.0159, which is our p-value. Using our tables, the p-value is
between 0.01 and 0.025, but Excels p-value is more precise and more useful.
We follow the same format as we did for a right-tailed p-value. However, we do not
type a negative sign in front of our test statistic.
If we had a left-tailed test, with 27 degrees of freedom, and the test statistic cal-
=TDIST(1.997,27,1)
between 0.025 and 0.05, but Excels p-value is more precise and more useful.
TWO-TAILED p-VALUES
The major difference for calculating a two-tailed p-value is that we type a 2 instead of
a 1 for the number of tails.
If we had a two-tailed test, with 11 degrees of freedom, and the test statistic cal-
=TDIST(2.038,11,2)
The result that we get is 0.0663, which is our p-value. The p-value is between 0.05 and
0.1, but Excels p-value is more precise and more useful.
We conclude with an example that uses Excel to perform a hypothesis test.

E XA MP L E A math instructor has written a new placement exam. He is con-

cerned that the exam may be too long. The math department will
use this new exam if the mean time required to finish the exam for
all students who take it is at most 50 minutes. The exam is given to
20 randomly selected students. Here are the times (in minutes)
required to complete the exam for these students.
43 44 45 46 49 50 51 53 54 55
57 57 58 59 59 60 60 60 60 60
Use these sample data to test the claim that the mean length of
time required to finish the exam for all students who take it is at
most 50 minutes at the 0.05 level of significance.
We will need the sample statistics (mean and standard deviation).
Open a new Excel worksheet, and type the above values in column A (cells
A1 through A20).
In cell B1, type Sample Mean. In cell B2, type Sample Standard Deviation.
In cell B3, type Sample Size. In cell B4, type Claimed Mean.
Now that the framework is set up, we can begin the calculations.
To calculate the sample mean, type =AVERAGE(A1:A20) in cell C1.

To
C2.
calculate the sample standard deviation, type =STDEV(A1:A20) in cell
Type the sample size (20) in cell C3.

Type the claimed value of the mean (50) in cell C4.
We begin with the hypothesis test by hand, and then will turn to Excel when
needed.
Step 1
Claim in words: The mean length of time required to finish the exam for all students
who take it is at most 50 minutes.
Claim: 50
Complement: > 50
H0: 50
HA: > 50
Note that the claim is the null hypothesis, so our conclusion will be either to reject
the claim or fail to reject the claim. Also note that the alternate hypothesis involves
the symbol >. This tells us that our test is a right-tailed test.
Step 2
Step 3
x
Test statistic: t =
s/ n
This is the appropriate test statistic, since this is a one-mean test with a sample
size that is smaller than 30.
Step 4
Right-tailed test, = 0.05
Type Critical Value in cell B6.

We will calculate the critical value for this test in cell C6. Since this is a left-
tailed test, type =TINV(2*0.05,19). (Since this is a one-tailed test, we must
multiply the level of significance by 2.)
The value Excel gives us is 1.729131. Our decision rule is then
Reject H0 if t > 1.729131
Here is the critical value using our usual method.
One-Tailed Value

19 2.861 2.539 2.093 1.729 1.328 0.688
Decision rule: Reject H0 if t > 1.729.
Step 5
In cell B7, type Test Statistic.

Next we will calculate the test statistic in cell C7.
Type =(C1-C4)/(C2/SQRT(C3)).
The result Excel gives us is 2.985792.

Here is the formula worked out.
x
t =
s/ n
54 50
=
5.99/ 2 0
= 2.986

length of time required to finish the exam for all students who take it is at most
50 minutes.
p-value
Type p-value in cell B9.

We will calculate the p-value for this test in cell C9. Since we are looking for
a left-tailed p-value, type =TDIST(C7,19,1).
The result that Excel gives us is 0.003798.
Using the t-distribution chart, we can see that the p-value is less than 0.005. This is
because the test statistic (2.986) is greater than 2.861, which is the highest value in
the table for 19 degrees of freedom.
Here is what your Excel worksheet should look like when you are done.
43 Sample Mean 54
44 Sample Standard Deviation 5.991222
45 Sample Size 20
46 Claimed Mean 50
49
50 Critical Value 1.729131
51 Test Statistic 2.985792
53
54 p-value 0.003798
55
57
57
58
59
59
60
60
60
60
60
Hypothesis Tests for a Single Population Mean

(Small Samples)
The TI-83 can help us perform hypothesis tests for a single mean based on a small
sample, and find p-values for these tests.
The TI-83 can help us with calculating the p-value. The calculator will give us a pre-
cise p-value instead of an interval that it falls in. Suppose we have a right-tailed test,
with 13 degrees of freedom, and the test statistic calculated to be 2.405. Press 2ND
VARS to access the DISTR menu. Select option 5: tcdf(. We need to enter three val-
ues. First, enter the value of the test statistic, which is 2.405. After pressing , , press
1, then 2ND and the , key to access EE, and then finally press 99 followed by a
, . Finally, we need to type in the degrees of freedom, 13, followed by ) . Here is
what the screen should look like. To calculate the p-value, press ENTER .
between 0.01 and 0.025, but the TI-83s p-value is more precise and more useful.
We follow the same format as we did for a right-tailed p-value.
tcdf(-1E99, test statistic, degrees of freedom)
Here is what the screen should look like for a left-tailed p-value for which the test
statistic is calculated to be 1.997, with 27 degrees of freedom.
between 0.025 and 0.05, but the TI-83s p-value is more precise and more useful.
TWO-TAILED p-VALUES
To calculate a two-tailed p-value, we calculate the appropriate one-tailed p-value, and
then double that value. If the test statistic is positive, double the right-tailed p-value.
If the test statistic is negative, double the left-tailed p-value.
If we had a two-tailed test, with 11 degrees of freedom, and the test statistic cal-
culated to be 2.038, here is the screen that calculates the p-value.
The result that we get is 0.0663, which is our p-value. The p-value is between 0.05 and
0.1, but the TI-83s p-value is more precise and more useful.
We conclude this section with an example that uses the TI-83 to perform a
hypothesis test.
E XA MP L E A math instructor has written a new placement exam. He is con-

7.21 cerned that the exam may be too long. The math department will
use this new exam if the mean time required to finish the exam for
all students who take it is at most 50 minutes. The exam is given to
20 randomly selected students. Here are the times (in minutes)
required to complete the exam for these students.
43 44 45 46 49 50 51 53 54 55
57 57 58 59 59 60 60 60 60 60
Use these sample data to test the claim that the mean length of
time required to finish the exam for all students who take it is at
most 50 minutes at the 0.05 level of significance.
Put these 20 times in list L1. To begin the hypothesis test, press STAT , then use
the key to move to the TESTS menu. Select item 2: T-Test. Fill in the screen
in the following manner.
Inpt: Highlight Data. We use this option, rather than Stats, when we want
to take the sample statistics from a list.
0: Enter the claimed value of the mean, which for this example is 50.
List: Enter the list that contains the sample data. In this example, we will
use L1.
Freq: Be sure that this says 1.
: Select the form of the alternate hypothesis. The claim for this test is that
the mean length of time required to finish the exam for all students who take
it is at most 50 minutes.
This becomes a test of H0: 50 versus HA: > 50.
Select option 3: >0.
ENTER .
Highlight Calculate and press
You will see the following screen, which lists the alternate hypothesis, calculated
value of the test statistic, p-value, sample mean, sample standard deviation, and
sample size.
EXERCISES 7.2
1. A random sample of 25 households without health insurance had a mean annual
income of $37,850 with a standard deviation of $25,554. At the 0.05 level of sig-
nificance, test the claim that the mean annual income of households without
insurance is less than $50,000. (Based on results of a study by the Census Bureau.)
2. A mortgage broker believes that the typical family that rents their home owns
fewer than 2 cars on average. A random sample of 26 families that rent their
home owned a mean of 1.29 cars with a standard deviation of 0.97 cars. At the
0.05 level of significance, test the claim that the mean number of cars owned by
families that rent their home is less than 2. (Based on the results of the Census
Bureau American Housing Brief.)
3. In testing the effect of a new zinc spray on the common cold, 27 people were ran-
domly selected and given the spray to take home. At the first sign of a cold they
were to begin using the spray. Their first colds lasted a mean of 1.5 days, with a
standard deviation of 2.1 days. At the 0.05 level of significance, test the claim that
the mean length of a cold being treated by this spray is equal to 2 days. (Based on
the results of a study funded by Gel-Tech.)
4. Fifteen households were randomly selected and asked how much they spend per
week on groceries. The mean grocery bill per week was $73.25 with a standard
deviation of $29.30. At the 0.05 level of significance, test the claim that the mean
grocery bill per week for all households is less than $100 per week.
5. A health researcher is interested in the age that males begin smoking. She claims
that the mean age is less than 19 years old. A random sample of 17 male smokers
had a mean starting age of 18.2 years with a standard deviation of 2.60 years.
(a) At the 0.05 level of significance, test the researchers claim.
6. A statistics instructor, when asked how students do on the first statistics exam,
claims that the mean score for all students on the first exam is above 75. To test
this claim, 22 students are selected at random. Their scores from the first exam had
a mean of 78.2 with a standard deviation of 13.95. Test the instructors claim at
the 0.05 level of significance.
7. How many pairs of shoes do female college students own? A random sample of 15
female college students produced a sample mean of 8.7 pairs of shoes, with a stan-
dard deviation of 0.85 pairs. Use these data to test the claim that the mean num-
ber of pairs of shoes owned by female college students is less than 10 at the 0.05
8. An experimental algebra exam is given to 20 randomly selected college algebra
students. Here are their scores.
68 94 50 60 92 14 49 68 41 67
76 71 32 15 54 54 65 92 58 68
At the 0.05 level of significance, test the claim that the mean score for all college
algebra students on this exam is below 60.
9. A random sample of 24 high school students were found to be working a mean of
11.6 hours per week with a standard deviation of 11.54 hours. At the 0.05 level of
significance, test the claim that the mean number of hours worked per week by
high school students is less than 15 hours per week. (Based on results in the
National Assessment of Educational Progress survey for the National Center for
Education Statistics.)
10. How often do teen mothers read to their children per week? A random sample of
22 teenage mothers read a mean of 2.7 days per week with a standard deviation
of 1.64 days per week. At the 0.05 level of significance, test the claim that teenage
mothers read to their children more than twice a week.
11. Here are the heights of 11 randomly selected professional womens basketball
players.
78 69 71 69 75 71 77 71 77 73 68
At the 0.05 level of significance, test the claim that the mean height of profes-
sional womens basketball players is more than 6 feet.
12. Here are the weights of 11 randomly selected professional womens basketball
players.
195 148 168 142 170 165
170 165 180 172 132
At the 0.05 level of significance, test the claim that the mean weight of profes-
sional womens basketball players is above 150 pounds.
13. These are the number of packages handled by a shipping office on 17 randomly
selected days.
1103 1488 1713 1536 1037 1462
1625 1627 1080 1216 1639 1539
1545 907 1307 1387 1547
Test the claim that the shipping office handles more than 1200 packages per day
at the 0.01 level of significance.
14. A random sample of 13 accountants showed that they had a mean salary of
$46,328 and a standard deviation of $17,298.
(a) Use this sample to test the claim that the mean accountant salary is higher
than $40,000 at the 0.05 level of significance.
15. These are the total scores from 15 randomly selected NFL football games.
41 26 51 38 30 46 63 31
45 27 9 37 42 40 37
Use this sample to test the claim that the mean number of points scored in an NFL
game is 34 points, at the 0.05 level of significance.
16. A random sample of 28 symphony musicians were asked how many hours they
practiced their instrument per week. The survey produced a mean practice time of
7.2 hours with a standard deviation of 3.62 hours per week. At the 0.05 level of
significance, test the claim that the mean time that symphony musicians practice
per week is greater than 5 hours per week.
17. A random sample of 20 workers who log on to Yahoo! while at work were asked
how long they stayed logged on while at work. The mean length was 106 minutes,
with a standard deviation of 42 minutes. Use this sample to test the claim that the
mean time spent on Yahoo! is greater than 90 minutes at the 0.05 level of signifi-
cance. (Based on the results of Nielsen/Net Ratings, January 2000.)
18. A math instructor has developed a test to determine a college students math com-
petency. He claims that the test is designed to produce a mean score of 70. The test
is given to 12 randomly selected college students, and here are their scores.
76 73 81 63 71 82
68 75 66 79 90 70
Test the instructors claim that the mean score for all students is 70 at the 0.05 level
of significance.
19. Tiger Woods has been dominating professional golf. One of his greatest assets is
the length of his drives, which makes every hole shorter for him. Here are the
lengths, in yards, of eight of his drives from the 100th U.S. Open at Pebble Beach.
305 290 308 289 302 309 302 289
Use these drives to test the claim that the mean length of all of Tigers drives is
longer than 290 yards at the 0.05 level of significance. (Source: On-line scoring for
U.S. Open at lycos.com.)
20. An agriculture student is interested in the mean length of a cows lactation (num-
ber of days producing milk). She randomly selects 20 of the universitys cows and
looks at their records. Here are the lengths, in days, of each cows first lactation.
263 279 284 249 335 326 399
283 285 293 329 338 266 286
229 387 387 317 294 315
Test the claim that the mean length of a cows first lactation is greater than 290
days at the 0.05 level of significance.
21. Major league baseball managers keep a close eye on the pitch count of their
starting pitcher, because they believe that a pitcher loses his effectiveness after a
certain number of pitches. Here are the pitch counts of 16 randomly selected start-
ing pitchers.
117 94 137 66 103 100 81 98
86 77 94 95 115 79 93 86
At the 0.05 level of significance, test the claim that the mean pitch count for start-
ing pitchers is 100 pitches.
The following statistical project contains a major error. Potential errors could be choosing
the wrong hypothesis test or not meeting the assumptions required for this test. Find the
error, and explain what could be done to correct it.
I claim that the mean height for adult males is 5 9 (69 inches). You are to test
M I N I RP O J E C T
my claim at the 0.05 level of significance. Randomly sample 20 adult males, and
find out their height in inches. Use these sample data to test the claim. In addi-
tion to your complete hypothesis test, include
Your raw data
A histogram for the heights
A5 pie chart showing the percentages that are 5
7 and 5 11 , and 6 or taller
6 or shorter, between
Finally, answer the following questions.
How did you obtain your sample?

Which type of sampling did you use?
Was the sample truly random?
What potential biases may show up in your sample?
SECTION 7.3
Hypothesis Test for a Single Population Proportion (Large Samples)
In this section we will learn to test claims about a single population proportion. These
claims will compare the proportion of a population to a certain value. For example,
13.6% of all pregnant women smoke while pregnant compares the proportion of all
pregnant women who smoke during their pregnancy to 13.6%.
In similar fashion to our work in constructing a confidence interval for a popula-
tion proportion (Section 6.4), we must expect at least 5 successes and 5 failures in our
sample. This is due to the fact that we are essentially applying the normal approxi-
mation to the binomial distribution, which can only be done when the expected num-
ber of successes and failures are both at least 5. If n is the size of our sample and is
the claimed population proportion, then both n and n(1 ) must be at least 5,
or we cannot use the z-distribution for this hypothesis test. (We could test the claim
using the binomial distribution, however.)
This test uses the z-distribution. A major difference between this test and the large-
sample test for a single mean is the test statistic. Here it is.
p
z =
(1 )

n
where p = sample proportion
= claimed population proportion
n = sample size
If we examine the numerator, we notice that it is again the difference between the
sample proportion and the claimed population proportion. The denominator is the
standard error of the distribution of sample proportions.
The test for a single population proportion follows the same five-step procedure.
Lets look at some examples.
7.3 Hypothesis Test for a Single Population Proportion (Large Samples) 353
E XA MP L E A physician claims that 13.6% of all pregnant females smoke during

their pregnancy. A random sample of 400 pregnant women
revealed that 60 of them are smoking while pregnant. Test the
physicians claim at the 0.05 level of significance.
Step 1
Claim in words: 13.6% of all pregnant females smoke during their pregnancy.
Claim: = 0.136 (We use to represent the population proportion.)
Complement: 0.136
H0: = 0.136
HA: 0.136
Step 2
Step 3
p
Test statistic: z =
(1 )

n
Step 4
= 0.05, two-tailed test
1.96 1 0 1 1.96
Step 5
p
z =
(1 )

n
= 0.15 0.136
(0.136)(0.86 4)
400
= 0.82

Conclusion: There is not sufficient sample evidence to reject the claim that 13.6% of
all pregnant women smoke while pregnant.
p-value
2 P (Z > 0.82)
= 2 0.2061
= 0.4122
2 1 0 1 2
0.82
E XA MP L E Test the claim that more than 60% of all college statistics students
are female at the 0.05 level of significance. A random sample of 114
college statistics students revealed that 81 of them were female.
Step 1
Claim in words: More than 60% of all college statistics students are female.
Claim: > 0.6
Complement: 0.6
H0: 0.6
HA: > 0.6
Step 2
Step 3
p
Test statistic: z =
(1
)
n
Step 4
= 0.05, right-tailed test
2 1 0 1 2
1.645
Step 5
p
z =
(1
)
n
81
0.6
114
=
(0.6) (0.4)
114
= 2.41

Conclusion: There is sufficient sample evidence to support the claim that more than
60% of all college statistics students are female.
p-value
0.9920 P (Z > 2.41) = 0.0080
2 1 0 1 2
2.41
E XA MP L E Test the claim that less than 20% of all college students use
prepaid phone cards, at the 0.05 level of significance. A random
sample of 250 students revealed that 41 of them used prepaid
phone cards.
Step 1
Claim in words: Less than 20% of all college students use prepaid phone cards.
Claim: < 0.2
Complement: 0.2
H0: 0.2
HA: < 0.2
Step 2
Step 3
p
Test statistic: z =
(1
)
n
Step 4
= 0.05, left-tailed test
2 1 0 1 2
1.645

Step 5
p
z =
(1 )

n
41
0.2
250
=
(0.2)(0.8)
250
= 1.42

Conclusion: There is not sufficient sample evidence to support the claim that less
than 20% of all college students use prepaid phone cards.
p-value
P (Z < 1.42) = 0.0778
2 1 0 1 2
1.42

Population Proportion
Here is an example using Excel.
E XA MP L E While at a dice table in a casino, a gambler comes up with the idea

that the dice may be loaded. Particularly, he believes that they are
set up so that the sum 7 will come up more often than it is sup-
posed to. He recalls from his probability training that the probability
of rolling a sum of 7 is 61 . He writes down the results from the next
50 rolls, and here they are.
5 9 9 6 6 7 8 9 6 7
11 6 10 8 10 5 2 7 6 7
8 9 3 7 6 8 7 3 3 9
5 8 7 7 9 4 9 7 4 12
3 5 9 7 9 10 8 9 8 7
Use these sample data to test the claim that the proportion of 7s is
greater than 16 at the 0.05 level of significance.
Enter the 50 values in column A (cells A1 through A50). Now we will let Excel
determine whether a roll was 7 or not. In cell B1, type
=IF(A1=7,1,0)
If cell A1 is 7, Excel will place a 1 in cell B1. Otherwise, it will place a 0 in cell B1.
Now copy cell B1 and paste it in cells B2 through B50.
In cell C1, type Number of Sevens. Next to that in cell D1, type
=SUM(B1:B50)
This will total column B, and tell us how many rolls were 7s.
In cell C2, type Sample Size. Type the sample size (50) in cell D2.
In cell C3, type Sample Proportion. In cell D3 type the following to calculate the
sample proportion.
=D1/D2
In cell C4, type Claimed Proportion. Next to that, type =1/6 in cell D4.
Now that the sample information is set up, we will begin the hypothesis test and
use Excel where needed.
Step 1 1
Claim in words: The proportion of 7s is greater than 6.

Claim: > 16
Complement: 16
H0: 61
HA: > 1
6
Step 2
Step 3
p
Test statistic: z =
(1 )

n
Step 4
= 0.05, right-tailed test
In cell C6, type Critical Value.
To calculate the critical value for this test, type the following in cell D6.
=NORMSINV(1-0.05)
Excel returns a value of 1.644853.

Here is how the decision rule is formulated without 0.9500
Excel.
Decision rule: Reject H0 if z > 1.645. 0.05
Step 5
2 1 0 1 2
In cell C7, type Test Statistic. 1.645
To calculate the test statistic, type the following
in cell D7.
=(D3-D4)/SQRT(D4*(1-D4)/D2)
Excel returns a value of 1.011929 for the test statistic.

Now, here is the calculation by hand using the formula.

p
z =
(1 )

n
1
0.22
6
=
1 5

6 6
50
= 1.01
Conclusion: There is not sufficient sample evidence to support the claim that the
proportion of 7s is greater than 16.
p-value
In cell C9, type p-value. In cell D9, we calculate the p-value by typing
=1-NORMSDIST(D7)
Excel returns a value of 0.155786.
Here is the non-Excel calculation.
0.8442 P (Z > 1.01) = 1 0.8442

= 0.1558
0.1558
2 1 0 2
1.01
Below is the Excel worksheet for this example.
5 0 Number of Sevens 11 8 0
9 0 Sample Size 50 7 1
9 0 Sample Proportion 0.22 3 0
6 0 Claimed Proportion 0.166667 3 0
6 0 9 0
7 1 Critical Value 1.644853 5 0
8 0 Test Statistic 1.011929 8 0
9 0 7 1
6 0 p-value 0.155786 7 1
7 1 9 0
11 0 4 0
6 0 9 0
10 0 7 1
8 0 4 0
10 0 12 0
5 0 3 0
2 0 5 0
7 1 9 0
6 0 7 1
7 1 9 0
8 0 10 0
9 0 8 0
3 0 9 0
7 1 8 0
6 0 7 1
Hypothesis Tests for a Single Population Proportion
Here is an example for the TI-83.
E XA MP L E While at a dice table in a casino, a gambler comes up with the idea

that the dice may be loaded. Particularly, he believes that they are
set up so that the sum 7 will come up more often than it is sup-
posed to. He recalls from his probability training that the probability
of rolling a sum of 7 is 61 . He writes down the results from the next
50 rolls, and here they are.
5 9 9 6 6 7 8 9 6 7
11 6 10 8 10 5 2 7 6 7
8 9 3 7 6 8 7 3 3 9
5 8 7 7 9 4 9 7 4 12
3 5 9 7 9 10 8 9 8 7
Use these sample data to test the claim that the proportion of 7s is
greater than 16 at the 0.05 level of significance.
Single-proportion tests are also quite easy on the TI-83. Before we begin, we must
know what the claimed population proportion is, what the sample proportion is, and
what form the alternate hypothesis takes. The claim is that the proportion of 7s is
greater than 16. This turns into a test of H0: 16 versus HA: > 61 . Since there were
11 sevens out of 50 rolls, the sample proportion is 11 50.
Press STAT , then use the key to move to the TESTS menu. Select option 5:
1-Prop Ztest. Here is how to fill out the next screen.
p0: Type 1 6 here, as this is the claimed population proportion.

x: Type 11 here, the number of successes.
n: Type 50 here, the sample size.
prop: Choose the form of the alternate hypothesis here. The correct option
this time is >p0.
ENTER .
Highlight Calculate and press
Here is the screen that you should see. It lists the claimed population proportion,
the test statistic, the p-value, the sample proportion, and the sample size.
EXERCISES 7.3
1. At a campus-wide meeting, the president of a community college claims that more
than 50% of all community college students are female. In a random sample of 155
community college students, 88 were female. Test the presidents claim at the 0.01
2. In a current events survey, college students were asked to name the star of the hit
movie The Water Boy. Twenty-eight out of 35 randomly selected college students
were able to name Adam Sandler as the star of the movie. At the 0.05 level of sig-
nificance, test the claim that more than 70% of all college students can name the
star of the movie The Water Boy.
3. In a current events survey, college students were asked to name the U.S. senator
who went into space in 1998. Twenty-three out of 35 randomly selected college stu-
dents were able to name John Glenn as the senator who went into space. At the
0.05 level of significance, test the claim that less than 70% of all college students
can name the U.S. senator who went into space in 1998.
4. A counselor claims that more than one-third of college students have divorced par-
ents. A random sample of 40 college students revealed that 18 had divorced parents.
(a) At the 0.01 level of significance, test the counselors claim.
5. The author of a blackjack strategy book claims that the dealer wins 52.5% of all
hands dealt. A blackjack player played 100 hands against a computer dealer, and
the computer dealer won 58 of the hands. Use these data to test the authors claim
at the 0.05 level of significance.
6. An algebra instructor believes that a cumulative midterm exam will help his stu-
dents. His colleagues feel otherwise. One colleague claims that, in general, at most
25% of students would pass such a test. The cumulative midterm is given to 40
randomly selected students, and only 9 pass with a C or better. Test the colleagues
claim at the 0.05 level of significance.
7. Pete Rose, who was elected to the All-Century major league baseball team, is
banned for life from baseball. This prevents the all-time major league hit leader
from being inducted into the Baseball Hall of Fame. A random sample of 200 base-
ball fans revealed that 121 want Pete Rose reinstated so that he may be elected
into Baseballs Hall of Fame. Test the claim that more than half of all baseball fans
feel that Pete Rose should be reinstated, at the 0.05 level of significance.
8. A doctor claims that at least 75% of the patients that he operates on for carpal tun-
nel syndrome return to work within a year of the operation. A sample of 55
patients that were operated on by the doctor for carpal tunnel syndrome showed
that only 35 of them returned to work within a year of the surgery. Test the doc-
tors claim at the 0.05 level of significance.
9. It is claimed that 60% of all 1825-year-olds have used alcohol in the past 30 days.
A survey of 125 students on campus who are between the ages of 18 and 25
showed that 83 have used alcohol in the past 30 days. Test the claim at the 0.05
level of significance. (Based on the results of a study by the U.S. Department of
Health and Human Services.)
10. It is claimed that more than 25% of all 1825-year-olds have used illicit drugs in
the past 30 days. A sample of 300 people between the ages of 18 and 25 revealed
that 85 had used illicit drugs in the past 30 days. Test the claim at the 0.05 level of
significance. (Based on the results of a study by the U. S. Department of Health and
Human Services.)
11. A magazine article claimed that more than half of all Americans who work at
home are men. A survey of 1200 Americans who work at home revealed that 635
were men. (Based on the results of a study by the U. S. Bureau of Labor Statistics.)
(a) Test the magazines claim at the 0.05 level of significance.
12. A political pundit claims that at least 55% of all female voters over the age of 60
vote for the Democratic Party candidate. A survey of 445 female voters who were
60 years old or older showed that 241 of them voted for the Republican candidate
in the 1998 House of Representatives election. Test the pundits claim at the 0.01
level of significance. (Based on the results of a poll conducted for the American
Association of Health Plans.)
13. It has been claimed that at least 2 out of every 3 pet owners allow their pets to
sleep in bed with them. A survey of 295 pet owners at a local pet shop showed that
236 of them allowed their pet to sleep in bed with them. At the 0.05 level of
significance, test the claim that at least 23of all pet owners allow their pets to sleep
in bed with them. (Based on the results of a study by CommSciences for PETsMART.)
14. A group of 630 adults were asked whether they take vitamins. Three hundred
eighty-four of the people asked said yes. Test the claim that less than 75% of adults
take vitamins, using the 0.01 level of significance. (Based on the results of a study
by Market Facts for Vitamins.com.)
15. A survey of 595 college students showed that 240 did not use a condom the last time
they had intercourse. Test the claim that less than half of all college students did not
use a condom the last time they had intercourse at the 0.05 level of significance.
16. An NBC News/Wall Street Journal survey conducted of 2011 adult Americans
revealed that 83% felt that parents not paying enough attention to whats going
on in their childrens lives has become a very serious problem. Use this sample
to test the claim that more than 80% of all adult Americans feel that parents not
paying enough attention to whats going on in their childrens lives has become
a very serious problem. Use the 0.05 level of significance.
17. An NBC News/Wall Street Journal survey conducted of 2011 adult Americans
revealed that 63% identified the high rate of divorce and the breakup of fami-
lies as a very serious problem for American society. Use this sample to test the
claim that 60% of all adult Americans believe that the high rate of divorce and
the breakup of families is a very serious problem for American society at the 0.05
18. Test the claim that more than 25% of all people who suffer strokes are smokers, at
the 0.05 level of significance. The New Zealand Auckland Stroke Study compared
521 stroke patients with 1851 healthy people. Of the stroke patients, 164 were
active smokers.
19. Test the claim that more than 20% of all teachers feel that the value of classroom
technology is overrated. Use the 0.05 level of significance. A survey of 190 teach-
ers showed that 49 of them felt that the value of classroom technology is over-
rated. (Based on the results of a survey by USA Today.)
20. A random sample of 611 Californians revealed that 140 did not have health insur-
ance. Use this sample to test the claim that more than 20% of Californians do not
have health insurance. Use the 0.05 level of significance. (Based on the results of
a study by the Census Bureau.)
21. Test the claim that more than 1% of all pregnant women drink alcohol while preg-
nant at the 0.05 level of significance. A random sample of 325 women who gave
birth showed that 6 of them drank alcohol while pregnant. (Based on data contained
in Centers for Disease Control and Prevention Monthly Vital Statistics Report.)
22. Test the claim that more than half of all babies born are boys, at the 0.01 level of
significance. During 1994, there were 2,022,589 male births and 1,930,178 female
births. (Based on data contained in Centers for Disease Control and Prevention
Monthly Vital Statistics Report.)
23. Test the claim that more than 60% of all college graduates use the Internet at the
0.05 level of significance. An alumni survey of 495 college graduates showed that
307 of them currently use the Internet.
24. A recent survey showed that 112 out of 180 two-parent households with children
had a computer at home. Use this survey to test the claim that more than half of
all households that have two parents living in the home have a computer at
home. Use the 0.05 level of significance. (Based on the results of a study by the
Henry J. Kaiser Family Foundation.)
25. A recent survey showed that 23 out of 72 households that were headed by a sin-
gle female had a computer at home. Use this survey to test the claim that less than
half of all households that are headed by a single female have a computer at
home. Use the 0.05 level of significance. (Based on the results of a study by the
Henry J. Kaiser Family Foundation.)
26. A random sample of 1401 women, financed by the Centers for Disease Control
and Prevention, showed that 55% of them had been victims of some type of abuse
(from physical assault to psychological battering) by an intimate male partner. At
the 0.01 level of significance, test the claim that more than half of all women have
been victims of such abuse from an intimate male partner.

An Introduction To Statistics-4

Uploaded by

Copyright:

Available Formats

An Introduction To Statistics-4

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

An Introduction To Statistics-4

Uploaded by

Copyright:

Available Formats

SECTION 7.

E XA MP L E A college statistics instructor claims that the mean age of college

We see that z/2 = 1.96. Now we determine the margin of error.

Next we construct the confidence interval.

State the claim and the hypothesis

Step 1: State the Claim and the Hypotheses

1. The mean age of all college statistics students is 24.

Reject null hypothesis: support alternate hypothesis

Step 2: Choose a Level of Significance ()

Step 3: Choose a Test Statistic

Test Statistic for a Large-Sample, Single-Mean Test

Test Statistic for a Large-Sample, Single-Mean Test

Step 4: State the Decision Rule

E XA MP L E A college statistics instructor claims that the mean age of college

Right-Tailed Tests Suppose we had the following hypotheses.

Left-Tailed Tests Suppose we had the following hypotheses.

Step 5: Do the Calculations and Draw Conclusions

Reject H0 if z < 1.96 or if z > 1.96

Claim is H0 Reject the claim Fail to reject the claim

Two-Tailed P-Values We find a p-value for a two-tailed hypothesis test by putting

p-value = 2 P(Z < 2.47)

Left-Tailed P-Values We find a p-value for a left-tailed hypothesis test by putting

p-value = P(Z < 1.57)

Right-Tailed P-Values We find a p-value for a right-tailed hypothesis test by put-

p-value = P(Z > 2.29)

E XA MP L E A college statistics instructor claims that the mean age of college

Decision rule: Reject H0 if z < 1.96 or if z > 1.96.

Decision: Reject H0.

p-value = 2 P (Z < 2.47)

Decision rule: Reject H0 if z > 2.33.

Decision: Reject H0.

p-value = P (Z > 3.00)

E XA MP L E Test the claim that the mean amount spent by a customer at

Decision rule: Reject H0 if z < 1.645.

Decision: Fail to reject H0.

p-value = P (Z < 1.03)

Decision rule: Reject H0 if z < 1.96 or if z > 1.96.

Decision: Reject H0.

p-value = 2 P (Z < 17.88)

The p-value for this test is (much) less than 0.0004.

MICROSOFT EXCEL Hypothesis Tests for a Single

RIGHT-TAILED CRITICAL VALUES

Round Excels result to two decimal places.

LEFT-TAILED CRITICAL VALUES

Round Excels result to two decimal places.

If the test statistic was calculated to be z = 2.47, then we would type

Round all p-values to four decimal places.

If the test statistic was calculated to be z = 1.57, then we would type

Round that answer to four decimal places.

If the test statistic was calculated to be z = 2.29, then we would type

Round that answer to four decimal places.

E XA MP L E An algebra instructor has designed a final exam that he believes will

We will need the sample statistics (mean and standard deviation).

To calculate the sample mean, type =AVERAGE(A1:A38) in cell C1.

Type the sample size (38) in cell C3.

The value Excel gives us is -2.32634. Our decision rule is then

Reject H0 if z < 2.32634

Here is the critical value using our usual method.