Hypothesis tests for population proportion, p, large samples

The same six steps on page 2 of the document Hypothesis or Significance Testing will be used to test
the population proportion p for large samples, since you recall the procedure was described as standard.
The only major difference is the expression for the test statistic z for the sample proportion ps,
which is computed as follows:

1
ps p 0
2n
z= .
p0 q 0

n

Determining the continuity correction

1 1

Upper-tail 2n Lower-tail 2n

1

Two-tail 2n if ps > p0
1

2n if ps < p0

Example 1

According to a Nationwide Mutual Insurance Company Driving While Distracted Survey conducted in
2008, 81% of the drivers interviewed said they have talked on their cell phones while driving (The New
York Times, July 19, 2009). The survey included drivers aged 16 to 61 years selected from 48 states.
Assume that this result holds true for the 2008 population of all such drivers in the United States. In a
recent random sample of 1600 drivers aged 16 to 61 years selected from the United States, 83% said
they have talked on their cell phones while driving. Using the 5% significance level, can you conclude
that the current percentage of such drivers who have talked on their cell phones while driving is
different from 81%.

Solution

Let p be the current proportion of all US drivers aged 16 to 61 years who have talked on their cell
phones while driving, and let ps be the corresponding sample proportion. Then from the given
information:
n=1600, ps =0.83, =0.05

In 2008, 81% of US drivers aged 16 to 61 years said they have talked on their cell phones while driving.
Hence,
p0=0.81, q 0=10.81=0.19

Page 1 of 4
Steps for the hypothesis test, (refer to page 2 of the document, Hypothesis or Significance Testing)

1. Ho: p=0.81 (The current percentage is not different from 81%)

H 1: p 0.81 (The current percentage is different from 81%)

2. np=1600 0.81=1296> 5,nq=1600 0.19=304>5

Therefore the sample size is large and so the normal distribution can be used to carry out
the test, since the sampling distribution of p0 is approximately normal; in notation form:

pq 0.81 0.19
(
ps N p ,
n ) (
, i. e . ^p N 0.81,
1600 ) .

3. Remember from previous work the sign in the alternative hypothesis means the test
is two-tailed and the significance level is 5%, so the rejection region in each tail of
0.05
sampling distribution ofps is= =0.025 . Critical values of z= 1.96 . So
2 2
reject Ho if z<1.96z> 1.96 . Accept it otherwise.

4. Test statistic z is calculated as follows:

1
0.83 0.81
3200
z= =2.007
(0.81)(0.19)

1600

5. 2.007>1.96 , so reject Ho.

6. We conclude that the current percentage of all US drivers aged 16 to 61 years who have
talked on their cell phones while driving is different from 81%. Consequently, we can
state that the difference between the hypothesized population proportion of 0.81 and the
sample proportion of 0.83 is too large to be attributed to sampling error alone when
=0.05 .

Example 2

Direct Mailing Company sells computers and computer parts by mail. The company claims that at least
90% of all orders are mailed within 72 hours after they are received. The quality control department at
the company often takes samples to check if claim is valid. A recently taken sample of 150 orders
showed that 129 of them were mailed within 72 hours. Do think the companys claim is true? Use a
2.5% significance test.

Solution

Let p be the proportion of all orders that are mailed by the company within 72 hours after they are
received. Let ps be the corresponding sample proportion. From the given information:

129
n=150, ps = =0.86, =0.025
150

Page 2 of 4
The company claims that at least 90% of all orders are mailed within72 hours. Assuming this claim is
true:
p=0.9q=10.9=0.1

1. Ho: p 0.9 (The companys claim is true.)

H1: p<0.9 (The companys claim is false.)

2. np=150 0.9>135>5nq=150 0.1=15> 5

0.9 0.1
Therefore, the sample is large and so ps N 0.9,( 150 ) .

3. A one-tailed left test (or lower-tailed test) and a significance level of 0.025 indicates that we
reject Ho, if z<1.96 .

1 1
ps+ p 0.86+ 0.9
2n 300
4. The test statistic z= = =1.497
^p ( 0.9)( 0.1)

150

5. 1.497>1.96 ; do not reject Ho.

6. We can state that the difference between the sample proportion and the hypothesized value of
the population proportion is small, and this difference may have occurred owing to chance
alone. Therefore, the proportion of all orders that are mailed within 72 hours is at least 90%,
and the companys claim is true.

Questions

1. In each of the following cases, do you think the sample size is large enough to use the normal
distribution to make a test of hypothesis about the population proportion? Explain why or why
not.
a. n=40p=0.11 , c. n=100 p=0.73
b. n=8 p=0.05 , d. n=50 p=0.14 .

2. A random sample of 500 observations produced a sample proportion of 0.38. Find the critical
and observed values of z for each of the following tests of hypotheses using =0.05 .
a. Ho: p=0.30 versus H1: p>0.30 .
b. Ho: p=0.30 versus H1: p 0.30 .

Page 3 of 4
3. Consider Ho: p=0.70 versus H1: p 0.70 .
a. A random sample of 600 observations produced a sample proportion equal to 0.68.
Using =0.01 , would you reject the null hypothesis?
b. Another random sample of 600 observations taken from the same population produced
a sample proportion equal to 0.76. Using =0.01 , would you reject the null
hypothesis?

4. According to a 2008 survey by the Royal Society of Chemistry, 30% of adults in Great Britain
stated that they typically run the water for a period of 6 to 10 minutes while taking the shower.
(http://www.rsc.org/AboutUs/News/PressReleases/2008/EuropeanShowerHabits.asp). Suppose
that in a recent survey of 400 adults in Great Britain, 104 stated that they typically run the water
for a period of 6 to 10 minutes when they take a shower. At the 5% significance level, can you
conclude that the proportion of all adults in Great Britain who typically run the water for a
period of 6 to 10 minutes when they take a shower is less than 0.30.

5. In a 2009 non-scientific poll on www.ESPN.com, 67% of the respondents believed that Roger
Federer was going to defeat Andy Roddick in the 2009 Wimbledon Gentlemens singles
championship. Suppose that a survey of 150 tennis fans conducted in Europe at the same time
resulted in 118 who believed that Federer was going to win. Perform a hypothesis test to
determine if it is reasonable to conclude that the percentage of all European tennis fans who
believed that Federer was going to win the 2009 championship was higher than 67%, the result
in the ESPN.com poll. Use a 2% significance level.

6. As noted in U.S. Senate Resolution 28, 9.3% of Americans speak their native language and
another language fluently. (Source: www.actfl.org/i4a/pages/index.cfm?pageid=3782).
Suppose that in a recent sample of 880 Americans, 69 speak their native language and another
language fluently. Is there significant evidence at the 10% significance level that the percentage
of all Americans who speak their native language and another language fluently is different
from 9.3%?

Page 4 of 4