Professional Documents
Culture Documents
Section 10.1 σ 12 σ 22
b. σ x1 − x2 = +
n1 n2
10.1 Two samples are independent if they are 2
(5.22) (6.80) 2
drawn from two different populations and = +
the elements of the two samples are not 650 675
= .33230155
related. Example 10−1 in the text provides
an example of independent samples. In two The 95% confidence interval for
dependent samples, the elements of one μ1 − μ 2 is
sample are related to the elements of the ( x1 − x 2 ) ± zσ x1 − x2
second sample. Example 10−2 in the text = –.49 ± 1.96 (.33230155)
provides an example of dependent samples. = –.49 ± .65 = –1.14 to .16
E = zσ x1 − x2 = 1.96(.33230155) = .65
10.2 The sampling distribution of x1 − x2 for
large and independent samples is
10.5 Step 1: H0: μ1 − μ 2 = 0, H1: μ1 − μ 2 ≠ 0
approximately normal. The mean of x1 − x2
Step 2: Since σ 1 and σ 2 are known, the
is μ1 − μ 2 , where μ1 and μ 2 are the
samples are independent, and the
population means. The standard deviation
populations are normally distributed, use
σ 12 σ 22 the normal distribution.
of x1 − x2 is σ x1 − x2 = + , where
n1 n2 Step 3: For α = .05, the critical values of z
are –1.96 and 1.96.
σ 1 and σ 2 are the population standard Step 4:
deviations and n1 and n2 are the sample
σ 12 σ 22
sizes. σ x1 − x2 = +
n1 n2
10.3 a. x1 − x2 = 7.82 – 5.99 = 1.83 (2.35) 2
(3.17) 2
= +
18 15
σ 12 σ 22 (2.35) 2 (3.17) 2 = .98829764
b. σ x1 − x2 = + = +
n1 n2 18 15 ( x1 − x 2 ) − ( μ1 − μ 2 )
z=
= .98829764 σ x1 − x2
The 99% confidence interval for μ1 − μ 2 is (7.82 − 5.99) − 0
= = 1.85
( x1 − x2 ) ± zσ x1 − x2 .98829764
= 1.83 ± 2.58 (.98829764) Step 5: Do not reject H0 since 1.85 < 1.96.
= 1.83 ± 2.55 = –.72 to 4.38
E = zσ x1 − x2 = 2.58(.98829764) = 2.55
225
226 Chapter 10 Estimation and Hypothesis Testing: Two Populations
(∑ x )
2
(560) 2
∑x 2
n
− 26, 999.8 −
12 = 8.87522940
s2 = =
n −1 12 − 1
x1 − x2 = 49.28 – 46.67 = 2.61
(∑ x )
2
(22.57) 2
∑x 2
n
− 51.6873 −
10
s1 = = = .28806056
n −1 10 − 1
From Sample 2: n2 = 12, ∑x = 22.37, ∑x2 = 42.9519
x 2 = ∑x/ n2 = 22.37/12 = 1.864
(∑ x )
2
(22.37) 2
∑ x2 − n
42.9519 −
12
s2 = = = .33716622
n −1 12 − 1
x1 − x2 = 2.257 – 1.864 = .393
232 Chapter 10 Estimation and Hypothesis Testing: Two Populations
Section 10.3
10.47 The paired samples procedure is used when two samples are drawn in such a way that for each data value
collected from one sample there is a corresponding data value collected from the second sample, and both
of these data values are collected from the same source.
b. s d = s d / n = 4.8 / 23 = 1.00086919
df = n – 1 = 23 – 1 = 22
The 95% confidence interval for μd is
d ± ts d = 13.2 ± 2.074(1.00086919) = 13.2 ± 2.08 = 11.12 to 15.28
c. s d = s d / n = 11.7 / 18 = 2.75771645
df = n – 1 = 18 – 1 = 17
The 90% confidence interval for μd is
d ± ts d = 34.6 ± 1.740(2.75771645) = 34.6 ± 4.80 = 29.80 to 39.40
b. s d = s d / n = 14.7 / 27 = 2.82901632
df = n – 1 = 27 – 1 = 26
The 95% confidence interval for μd is
d ± ts d = 55.9 ± 2.056(2.82901632) = 55.9 ± 5.82 = 50.08 to 61.72
246 Chapter 10 Estimation and Hypothesis Testing: Two Populations
c. s d = s d / n = 8.3 / 16 = 2.075
df = n – 1 = 16 – 1 = 15
The 90% confidence interval for μd is
d ± ts d = 29.3 ± 1.753(2.075) = 29.3 ± 3.64 = 25.66 to 32.94
sd = n = 7 = 1.98805959
n −1 7 −1
s d = s d / n = 1.98805959 / 7 = .75141590
df = n – 1 = 7 – 1 = 6
a. The 95% confidence interval for μd is
d ± ts d = –1.43 ± 2.447(.75141590) = –1.43 ± 1.84 = –3.27 to .41
sd = n = 7 = 4.57737708 minutes
n −1 7 −1
s d = s d / n = 4.57737708 / 7 = 1.73008592 minutes
df = n – 1 = 7 – 1 = 6
a. The 99% confidence interval for μd is
d ± ts d = 3.43 ± 3.707(1.73008592) = 3.43 ± 6.41 = –2.98 to 9.84 minutes
10.54
Non- Non-
Subject Dominant dominant d d2 Subject Dominant dominant d d2
1 59 53 6 36 10 47 38 9 81
2 32 30 2 4 11 40 35 5 25
3 27 24 3 9 12 36 36 0 0
4 18 20 –2 4 13 21 25 –4 16
5 42 40 2 4 14 51 48 3 9
6 12 12 0 0 15 30 30 0 0
7 29 24 5 25 16 32 31 1 1
8 33 34 –1 1 17 14 14 0 0
9 22 22 0 0 18 26 27 –1 1
∑d = 28 ∑d = 216
2
sd = n = 18 = 3.18493184 pounds
n −1 18 − 1
s d = s d / n = 3.18493184 / 18 = .75069563 pounds
df = n – 1 = 18 – 1 = 17
a. The 99% confidence interval for μd is
d ± ts d = 1.56 ± 2.898(.75069563) = 1.56 ± 2.18 = –.62 to 3.74 pounds
Section 10.4 Inferences About the Difference Between Two Population Means for Paired Samples 249
sd = n = 13 = 13.88967519 seconds
n −1 13 − 1
s d = s d / n = 13.88967519 / 13 = 3.85230278 seconds
df = n – 1 = 13 – 1 = 12
a. The 95% confidence interval for μd is
d ± ts d = 21.62 ± 2.179(3.85230278) = 21.62 ± 8.39 = 13.23 to 30.01 seconds
sd = n = 6 = 1.42746862 mpg
n −1 6 −1
s d = s d / n = 1.42746862 / 6 = .58276162 mpg
df = n – 1 = 6 – 1 = 5
a. The 99% confidence interval for μd is
d ± ts d = –1.72 ± 4.032(.58276162) = –1.72 ± 2.35 = –4.07 to .63 mpg
10.57
Time Filter I Filter II d d2 Time Filter I Filter II d d2
1 24 26 –2 4 11 11 9 2 4
2 31 30 1 1 12 8 10 –2 4
3 35 33 2 4 13 14 17 –3 9
4 32 28 4 16 14 17 16 1 1
5 25 23 2 4 15 19 16 3 9
6 25 28 –3 9 16 19 18 1 1
7 29 24 5 25 17 25 27 –2 4
8 30 33 –3 9 18 20 22 –2 4
9 26 22 4 16 19 23 27 –4 16
10 18 18 0 0 20 32 31 1 1
∑d = 5 ∑d2 = 141
d = ∑d /n = 5/20 = .25 ppm
(∑ d ) 2 (5) 2
∑d − 141 −
2
sd = n = 20 = 2.71205884 ppm
n −1 20 − 1
s d = s d / n = 2.71205884 / 20 = .60643479 ppm
df = n – 1 = 20 – 1 = 19
a. The 95% confidence interval for μd is
d ± ts d = .25 ± 2.093(.60643479) = .25 ± 1.27 = –1.02 to 1.52 ppm
Section 10.5 Inferences About the Difference Between Two Population Proportions For Large and 251
Independent Samples
b. Step 1: H0: μd = 0, H1: μd ≠ 0
Step 2: Since the samples are dependent, σ d is unknown, the sample is small, and the population of
paired differences is approximately normally distributed, use the t distribution.
Step 3: For α = .05 with df = 19, the critical values of t are –2.093 and 2.093.
Step 4: t = (d − μ d ) / s d = (.25– 0)/.60643479 = .412
Step 5: Do not reject H0 since .412 < 2.093.
Conclude that the average paired difference for concentration levels is not different from zero.
Section 10.5
10.58 The sampling distribution of pˆ1 − pˆ 2 for two large samples is approximately normal, with its mean equal
p1q1 p 2 q 2
to p1 − p 2 , and standard deviation equal to + , where p1 and p2 are the population
n1 n2
proportions, q1 = 1 − p1 , q 2 = 1 − p 2 , and n1 and n2 are the sample sizes.
10.59 Two samples are considered large enough for the sampling distribution of the difference between two
sample proportions to be approximately normal when n1 p1 , n1q1 , n2 p 2 , and n2 q 2 are all greater than 5.
x1 + x2 290 + 396
d. p= = = .312 and q = 1 – p = 1 – .312 = .688
n1 + n2 1000 + 1200
⎛1 1 ⎞ ⎛ 1 1 ⎞
s pˆ 1 − pˆ 2 = p q ⎜ + ⎟ = (.312)(.688) ⎜ + ⎟ = .01983774
⎝ n1 n2 ⎠ ⎝ 1000 1200 ⎠
( pˆ 1 − pˆ 2 ) − ( p1 − p2 ) −.04 − 0
z= = = –2.02
s pˆ 1 − pˆ 2 .01983774
b. The first two steps and the calculation of the observed value of the test statistic are the same for both
the p-value approach and critical value approach.
Step 1: H0: p1 − p 2 = 0, H1: p1 − p 2 ≠ 0
Step 2: Since n1 pˆ1 , n1qˆ1 , n2 pˆ 2 , and n2 qˆ 2 are all greater than 5, use the normal distribution.
n1 pˆ 1 + n2 pˆ 2 1200 (.29) + 1300 (.27 )
p= = = .28 and q = 1 − p = 1 − .28 = .72
n1 + n2 1200 + 1300
⎛1 1 ⎞ ⎛ 1 1 ⎞
s pˆ 1 − pˆ 2 = p q ⎜ + ⎟ = .28 (.72) ⎜ + ⎟ = .0179743407
⎝ n1 n2 ⎠ ⎝ 1200 1300 ⎠
( pˆ 1 − pˆ 2 ) − ( p1 − p 2 ) .02 − 0
z= = = 1.11
s pˆ 1 − pˆ 2 .0179743407
p-value approach
Step 3: From above, z = 1.11.
From the normal distribution table, area to the right of z = 1.11 is 1 – .8665 = .1335, so the
p-value = 2(.1335) = .2670
Step 4: Do not reject H0 since .2670 > .01.
(continued on next page)
Section 10.5 Inferences About the Difference Between Two Population Proportions For Large and 255
Independent Samples
(continued)
critical value approach
Step 3: For α = .01, the critical values of z are –2.58 and 2.58.
Step 4: From above, z = 1.11.
Step 5: Do not reject H0 since 1.11 < 2.58.
Conclude that the percentages of Independents and Democrats who rate the healthcare system as being
excellent or good are the same.
b. The first two steps and the calculation of the observed value of the test statistic are the same for both
the p-value approach and critical value approach.
Step 1: H0: p1 − p 2 = 0, H1: p1 − p 2 < 0
Step 2: Since n1 pˆ1 , n1qˆ1 , n2 pˆ 2 , and n2 qˆ 2 are all greater than 5, use the normal distribution.
n1 pˆ 1 + n2 pˆ 2 1800 (.79) + 1900 (.83)
p= = = .811 and q = 1 − p = 1 − .811 = .189
n1 + n2 1800 + 1900
⎛1 1 ⎞ ⎛ 1 1 ⎞
s pˆ 1 − pˆ 2 = p q ⎜ + ⎟ = .811(.189) ⎜ + = .0128774282
⎝ n1 n2 ⎠ ⎝ 1800 1900 ⎟⎠
( pˆ 1 − pˆ 2 ) − ( p1 − p 2 ) −.04 − 0
z= = = −3.11
s pˆ 1 − pˆ 2 .0128774282
p-value approach
Step 3: From above, z = –3.11.
From the normal distribution table, area to the left of z = –3.11 is .0009, so the p-value = .0009
Step 4: Reject H0 since .0009 < .01.
critical value approach
Step 3: For α = .01, the critical value of z is –2.33.
Step 4: From above, z = –3.11.
Step 5: Reject H0 since –3.11 < –2.33.
Conclude that the percentage of all U.S. adults who were using the internet in January 2011 is less than
the percentage of all U.S. adults who used the internet in January 2012.
10.71 a. pˆ 1 = x1 / n1 = 53/400 = .133 and pˆ 2 = x 2 / n2 = 51/470 = .109
pˆ 1 − pˆ 2 = .133 − .109 = .024
c. The first two steps and the calculation of the observed value of the test statistic are the same for both
the p-value approach and critical value approach.
Step 1: H0: p1 − p 2 = 0, H1: p1 − p 2 ≠ 0
Step 2: Since n1 pˆ1 , n1qˆ1 , n2 pˆ 2 , and n2 qˆ 2 are all greater than 5, use the normal distribution.
x1 + x2 53 + 51
p= = = .120 and q = 1 – p = 1 – .120 = .880
n1 + n2 400 + 470
⎛1 1 ⎞ ⎛ 1 1 ⎞
p q ⎜ + ⎟ = (.120)(.880) ⎜ +
⎝ 400 470 ⎟⎠
s pˆ 1 − pˆ 2 = = .02210613
⎝ n1 n2 ⎠
( pˆ 1 − pˆ 2 ) − ( p1 − p 2 ) .024 − 0
z= = = 1.09
s pˆ 1 − pˆ 2 .02210613
p-value approach
Step 3: From above, z = 1.09.
From the normal distribution table, area to the right of z = 1.09 is 1 – .8621 = .1379.
p-value = 2(.1379) = .2758
Step 4: Do not reject H0 since .2758 > .05.
critical value approach
Step 3: For α = .05, the critical values of z are –1.96 and 1.96.
Step 4: From above, z = 1.09.
Step 5: Do not reject H0 since 1.09 < 1.96.
Conclude that the proportions of cars that fail the emissions test at the two stations are not different.
Supplementary Exercises
σ 12 σ 22 (1.80) 2 (1.35) 2
10.76 a. σ x1 − x2 = + = + = .48466483 policy
n1 n2 20 25
The 99% confidence interval for μ1 − μ 2 is
( x1 − x2 ) ± zσ x1 − x2 = (11 − 9) ± 2.58(.48466483) = 2 ± 1.25 = .75 to 3.25 policies
σ 12 σ 22 70 2 90 2
a. σ x1 − x2 = + = + = $2.704933764
n1 n2 1500 2000
The 95% confidence interval for μ1 − μ 2 is
( x1 − x2 ) ± zσ x1 − x2 = (909 − 1035) ± 1.96 (2.704933764) = −126 ± 5.30 = −131.30 to − 120.70
(continued)
Since df = 82 is very large, we use the normal distribution to approximate the t distribution.
The 90% confidence interval for μ1 − μ 2 is
( x1 − x2 ) ± zs x1 − x2 = (31.47 – 35.28) ± 1.65(2.47815017) = –3.81 ± 4.09 = –$7.90 to $0.28
Step 1: H0: μ1 − μ 2 = 0, H1: μ1 − μ 2 < 0
Step 2: Since σ 1 and σ 2 are unknown but assumed to be unequal, the samples are independent, and
both samples are large ( n1 ≥ 30 and n2 ≥ 30), use the t distribution. However, since df = 82 is
very large, we use the normal distribution to approximate the t distribution.
Step 3: For α = .05, the critical value of z is –1.65.
( x − x 2 ) − ( μ1 − μ 2 ) −3.81 − 0
Step 4: z = 1 = = –1.54
s x1 − x2 2.47815017
Step 5: Do not reject H0 since –1.54 > –1.65.
Conclude that the average weekly spending reduction for all families of four with discretionary
incomes between $300 and $400 per week is not less than the average weekly spending reduction for
all families of five with discretionary incomes between $300 and $400 per week.
(continued)
Step 1: H0: μ1 − μ 2 = 0, H1: μ1 − μ 2 > 0
Step 2: Since σ 1 and σ 2 are unknown but assumed to be unequal, the samples are independent, and
both samples are large ( n1 ≥ 30 and n2 ≥ 30), use the t distribution. However, since
n1 = 1650 and n2 = 1820 are very large, we use the normal distribution to approximate the
t distribution.
Step 3: For α = .01, the critical value of z is 2.33.
( x − x 2 ) − ( μ1 − μ 2 ) 2170
Step 4: z = 1 = = 4.55
s x1 − x2 476.7589547
Step 5: Reject H0 since 4.55 > 2.33.
Conclude that the average salary of accountants and auditors is higher than that of loan officers.
b. accountants/auditors: n1 = 1650, x1 = $70,130, s1 = $16, 700
loan officers: n2 = 1820, x 2 = $67, 960, s 2 = $7900
s12 s 22 16, 700 2 7900 2
s x1 − x2 = + = + = 450.9051466
n1 n2 1650 1820
Since n1 = 1650 and n2 = 1820 are very large, we use the normal distribution to approximate the t
distribution. The 98% confidence interval for μ1 − μ 2 is
( x1 − x 2 ) ± zs x1 − x2 = (70,130 − 67,960) ± 2.33 (450.9051466)
= 2170 ± 1050.61 = $1119.39 to $3220.61
Step 1: H0: μ1 − μ 2 = 0, H1: μ1 − μ 2 > 0
Step 2: Since σ 1 and σ 2 are unknown but assumed to be unequal, the samples are independent, and
both samples are large ( n1 ≥ 30 and n2 ≥ 30), use the t distribution. However, since
n1 = 1650 and n2 = 1820 are very large, we use the normal distribution to approximate the
t distribution.
Step 3: For α = .01, the critical value of z is 2.33.
( x − x 2 ) − ( μ1 − μ 2 ) 2170
Step 4: z = 1 = = 4.81
s x1 − x2 450.9051466
Step 5: Reject H0 since 4.81 > 2.33.
Conclude that the average salary of accountants and auditors is higher than that of loan officers.
(continued)
Step 1: H0: μ1 − μ 2 = 0, H1: μ1 − μ 2 > 0
Step 2: Since σ 1 and σ 2 are unknown but assumed to be unequal, the samples are independent, and
both samples are large ( n1 ≥ 30 and n2 ≥ 30), use the t distribution.
Step 3: For α = .05 with df = 69, the critical value of t is 1.667.
( x − x2 ) − ( μ1 − μ 2 ) .9 − 0
Step 4: t= 1 = = 3.655
s x1 − x2 .24624812
Step 5: Reject H0 since 3.655 > 1.667.
Conclude that the mean evacuation time is smaller at closing time than during fire drills.
(continued)
Step 1: H0: μ1 − μ 2 = 0, H1: μ1 − μ 2 ≠ 0
Step 2: Since σ 1 and σ 2 are unknown but assumed to be unequal, the samples are independent, and
both samples are large ( n1 ≥ 30 and n2 ≥ 30), use the t distribution. However, since df = 131
is very large, we use the normal distribution to approximate the t distribution.
Step 3: For α = .10, the critical values of z are –1.65 and 1.65.
( x − x 2 ) − ( μ1 − μ 2 ) −5.12 − 0
Step 4: z = 1 = = –3.11
s x1 − x2 1.64723462
Step 5: Reject H0 since –3.11 < –1.65.
Conclude that the average number Christmas cards mailed out by all households in Town A is different
from the corresponding average for Town B.
sd = n = 8 = 4.138236339 bites
n −1 8 −1
s d = s d / n = 4.13823634 / 8 = 1.46308749 bites
df = n – 1 = 8 – 1 = 7
a. The 95% confidence interval for μd is
d ± ts d = 4.63 ± 2.365(1.46308749) = 4.63 ± 3.46 = 1.17 to 8.09 bites
d = ∑d /n = –44/9 = –4.89
(∑ d ) 2 (−44) 2
∑d − 508 −
2
sd = n = 9 = 6.05071162
n −1 9 −1
s d = s d / n = 6.05071162 / 9 = 2.01690387
df = n – 1 = 9 – 1 = 8
a. The 95% confidence interval for μd is
d ± ts d = –4.89 ± 2.306(2.01690387) = –4.89 ± 4.65 = –9.54 to –.24
266 Chapter 10 Estimation and Hypothesis Testing: Two Populations
a. Let p̂1 = the proportion of social media users in the 30−49 age group who cite connecting with others
with common hobbies or interests as a major reason for using social networking sites. Let p̂ 2 = the
proportion of social media users in the 50−64 age group who cite connecting with others with common
hobbies or interests as a major reason for using social networking sites.
pˆ 1qˆ1 pˆ 2 qˆ 2 (.16)(.84) (.18)(.82)
s pˆ 1 − pˆ 2 = + = + = .0218101896
n1 n2 562 624
The 95% confidence interval for p1 − p 2 is
( pˆ1 − pˆ 2 ) ± zs pˆ − pˆ
1 2
= (.16 − .18) ± 1.96 (.0218101896) = −.02 ± .043 = −.063 to .023
Chapter 10 Supplementary Exercises 267
b. The first two steps and the calculation of the observed value of the test statistic are the same for both
the p-value approach and critical value approach.
Step 1: H0: p1 − p 2 = 0, H1: p1 − p 2 ≠ 0
Step 2: Since n1 pˆ1 , n1qˆ1 , n2 pˆ 2 , and n2 qˆ 2 are all greater than 5, use the normal distribution.
n1 pˆ 1 + n2 pˆ 2 562 (.16) + 624 (.18)
p= = = .171 and q = 1 − p = 1 − .171 = .829
n1 + n2 562 + 624
⎛1 1 ⎞ ⎛ 1 1 ⎞
s pˆ 1 − pˆ 2 = p q ⎜ + ⎟ = .171(.829) ⎜ + ⎟ = .0218956182
⎝ n1 n2 ⎠ ⎝ 562 624 ⎠
( pˆ 1 − pˆ 2 ) − ( p1 − p 2 ) −.02 − 0
z= = = −.91
s pˆ 1 − pˆ 2 .0218956182
p-value approach
Step 3: From above, z = –.91.
From the normal distribution table, area to the left of z = –3.11 is .1814, so the
p-value = 2(.1814) = .3628
Step 4: Do not reject H0 since .3628 > .01.
critical value approach
Step 3: For α = .01, the critical values of z are –2.58 and 2.58.
Step 4: From above, z = –.91.
Step 5: Do not reject H0 since –2.58 < –.91.
Conclude that the percentage of social media users in the 30−49 age group who cite connecting with
others with common hobbies or interests as a major reason for using social networking sites was not
significantly different from the corresponding percentage of users in the 50−64 age group.
a. Let p̂1 = the proportion of women in the 40−49 age group who say that women of their age with no
special risk factors for breast cancer should have an annual mammogram. Let p̂2 = the proportion of
women in the 50 or over age group who say that women of their age with no special risk factors for
breast cancer should have an annual mammogram.
pˆ 1qˆ1 pˆ 2 qˆ 2 (.50)(.50) (.56)(.44)
s pˆ 1 − pˆ 2 = + = + = .0208728652
n1 n2 1055 1240
The 98% confidence interval for p1 − p 2 is
( pˆ1 − pˆ 2 ) ± zs pˆ − pˆ
1 2
= (.50 − .56) ± 2.33 (.0208728652) = −.06 ± .049 = −.109 to − .011
b. The first two steps and the calculation of the observed value of the test statistic are the same for both
the p-value approach and critical value approach.
Step 1: H0: p1 − p 2 = 0, H1: p1 − p 2 < 0
This is a left-tailed test.
Step 2: Since n1 pˆ1 , n1qˆ1 , n2 pˆ 2 , and n2 qˆ 2 are all greater than 5, use the normal distribution.
n1 pˆ 1 + n2 pˆ 2 1055 (.50 ) + 1240 (.56)
p= = = .532 and q = 1 − p = 1 − .532 = .468
n1 + n2 1055 + 1240
⎛1 1 ⎞ ⎛ 1 1 ⎞
s pˆ 1 − pˆ 2 = p q ⎜ + ⎟ = .532 (.468) ⎜ + = .0208993614
⎝ n1 n2 ⎠ ⎝ 1055 1240 ⎟⎠
( pˆ 1 − pˆ 2 ) − ( p1 − p 2 ) −.06 − 0
z= = = −2.87
s pˆ 1 − pˆ 2 .0208993614
(continued on next page)
268 Chapter 10 Estimation and Hypothesis Testing: Two Populations
(continued)
p-value approach
Step 3: From above, z = –2.87.
From the normal distribution table, area to the left of z = –2.87 is .0021, so the
p-value = .0021
Step 4: Reject H0 since .0021 < .01.
critical value approach
Step 3: For α/2 = .01, the critical value of z is –2.33.
Step 4: From above, z = −2.87.
Step 5: Reject H0 since –2.87 < –2.33.
The proportion of women in the 40−49 age group who say that women of their age with no special risk
factors for breast cancer should have an annual mammogram is less than the corresponding percentage
of women in the 50 and over age group.
a. Let p̂1 = the proportion of men who say that working part time hinders their career opportunities. Let
p̂2 = the proportion of women who say that working part time hinders their career opportunities.
pˆ 1qˆ1 pˆ 2 qˆ 2 (.52)(.48) (.43)(.57 )
s pˆ 1 − pˆ 2 = + = + = .018721565
n1 n2 1350 1480
The 95% confidence interval for p1 − p 2 is
( pˆ1 − pˆ 2 ) ± zs pˆ − pˆ
1 2
= (.52 − .43) ± 1.96 (.018721565) = .09 ± .037 = .053 to .127
b. The first two steps and the calculation of the observed value of the test statistic are the same for both
the p-value approach and critical value approach.
Step 1: H0: p1 − p 2 = 0, H1: p1 − p 2 ≠ 0
This is a two-tailed test.
Step 2: Since n1 pˆ1 , n1qˆ1 , n2 pˆ 2 , and n2 qˆ 2 are all greater than 5, use the normal distribution.
n1 pˆ 1 + n2 pˆ 2 1350 (.52) + 1480 (.43)
p= = = .473 and q = 1 − p = 1 − .473 = .527
n1 + n2 1350 + 1480
⎛1 1 ⎞ ⎛ 1 1 ⎞
s pˆ 1 − pˆ 2 = p q ⎜ + ⎟ = .473 (.527 ) ⎜ + = .0187901979
⎝ n1 n2 ⎠ ⎝ 1350 1480 ⎟⎠
( pˆ 1 − pˆ 2 ) − ( p1 − p 2 ) .09 − 0
z= = = 4.79
s pˆ 1 − pˆ 2 .0187901979
p-value approach
Step 3: From above, z = 4.79.
From the normal distribution table, area to the right of z = 4.79 is approximately 0, so the
p-value = 2(0) = 0
Step 4: Reject H0 since α = .02 > 0.
critical value approach
Step 3: For α = .02, the critical values of z are –2.33 and 2.33.
Step 4: From above, z = 4.79.
Step 5: Reject H0 since 4.79 > 2.33.
We conclude the the proportions of men and women who say that working part time hinders their
career opportunities are different.
Chapter 10 Supplementary Exercises 269
10.92 We are assuming that the populations are normally distributed and σ 1 = σ 2 = 2 , so we can use the normal
distribution for the following computation.
a. E = zσ x1 − x2 = 1.5, so σ x1 − x2 = 1.5/z = 1.5/1.65 for a 90% confidence level.
σ 12 σ 22 σ 12 + σ 22
σ x1 − x2 = + = , where n = n1 = n2 .
n1 n2 n
σ 12 + σ 22 (2) 2 + (2) 2
Hence, n = = = 9.68 ≈ 10.
(σ x1 − x2 ) 2 ⎛ 1.5 ⎞
2
⎜⎝ ⎟
1.65 ⎠
For this estimate 10 cars are required.
b. The mean of the sampling distribution of x1 − x2 = μ1 − μ 2 = 33–30 = 3 mpg and the standard
σ 12 σ 22 (2) 2 (2) 2
deviation of x1 − x2 is σ x1 − x2 = + = + = 1.26491106 mpg.
n1 n2 5 5
( x1 − x 2 ) − ( μ1 − μ 2 ) 0−3
Then, for x1 − x2 = 0, z = = = –2.37.
σ x1 − x2 1.26491106
Hence, P( x1 − x2 ≥ 0) = P( z ≥ −2.37) = 1 − P( z < −2.37) = 1 − .0089 = .9911 .
s12 s 22 3.5 2 32
a. s x1 − x2 = + = + = .14577380
n1 n2 1000 1000
Since n1 = 1000 and n2 = 1000 are very large, we use the normal distribution to approximate the t
( x1 − x 2 ) − ( μ 1 − μ 2 ) (15.4 − 15.0) − 0
distribution. Then, for x1 − x2 = .4, z = = = 2.74. Hence,
s x1 − x 2 .14577380
P( x1 − x2 ≥ .4) = P( z ≥ 2.74) = 1 − P( z < 2.74) = 1 − .9969 = .0031 and p-value = .0031.
b. Since p-value = .0031 < .025, reject H0. Thus, the result is statistically significant.
c. The 95% confidence interval for μ1 – μ2 is
( x1 − x2 ) ± zs x1 − x2 = (15.4 – 15.0) ± 1.96(.14577380) = .4 ± .29 = .11 to .69 minute
d. Part c implies that in 95% of all cases the insomniacs fall asleep less than a minute earlier with the new
pill than with the old pill. The result is not of great practical significance.
270 Chapter 10 Estimation and Hypothesis Testing: Two Populations
2 2 2
10.95 E = ts d ≤ 2 mpg, so t ≤ = = n . Thus, we look for the smallest value of n for which
sd 3 / n 3
2
t≤ n with 90% confidence.
3
n 2 3 4 5 6 7 8 9
critical values of t 6.314 2.920 2.353 2.132 2.015 1.943 1.895 1.860
sd = n = 7 = 5.19156826 contacts
n −1 7 −1
s d = s d / n = 5.19156826 / 7 = 1.96222836 contacts
df = n – 1 = 7 – 1 = 6
The 90% confidence interval for μd is
d ± ts d = 1.43 ± 1.943(1.96222836) = 1.43 ± 3.81 = –2.38 to 5.24 contacts
Step 1: H0: μd = 0, H1: μd > 0
Step 2: Since the samples are dependent, σ d is unknown, the sample is small, and the population of
paired differences is normally distributed, use the t distribution.
Step 3: For α = .05 with df = 6, the critical value of t is 1.943.
Step 4: t = (d − μ d ) / s d = (1.43 – 0)/1.96222836 = .729
Step 5: Do not reject H0 since .729 < 1.943.
Conclude that the mean number of contacts is not reduced.
For gas mileage:
∑d = –17, ∑d2 = 65, and d = ∑d /n = –17/7 = –2.43
(∑ d ) 2 (−17) 2
∑d − 65 −
2
sd = n = 7 = 1.98805959 mpg
n −1 7 −1
s d = s d / n = 1.98805959 / 7 = .75141590 mpg
df = n – 1 = 7 – 1 = 6
The 90% confidence interval for μd is
d ± ts d = –2.43 ± 1.943(.75141590) = –2.43 ± 1.46 = –3.89 to –.97 mpg
(continued on next page)
Chapter 10 Supplementary Exercises 271
(continued)
Step 1: H0: μd = 0, H1: μd < 0
Step 2: Since the samples are dependent, σ d is unknown, the sample is small, and the population of
paired differences is normally distributed, use the t distribution.
Step 3: For α = .05 with df = 6, the critical value of t is –1.943.
Step 4: t = (d − μ d ) / s d = (–2.43 – 0)/.75141590 = –3.234
Step 5: Reject H0 since –3.234 < –1.943.
Conclude that the mean gas mileage after the governors are installed is higher than the mean gas mileage
before governors are installed.
Suggestions
The 90% confidence intervals suggest that the number of contacts could be decreasing, increasing or
remaining constant and that gas mileage (in mpg) is increasing. The conclusions from the hypothesis tests
with .05 significance levels indicate that the decrease in number of contacts is not statistically significant
while the increase is gas mileage is statistically significant. The recommendation would then be to install
the governors on all salespersons’ cars.
10.98 a. No, a paired sample test would not be appropriate in this case. Different groups of cars are used in the
two samples, so the samples are independent.
b. From Sample 1: n1 = 6, ∑x = 140.4, ∑x2 = 3432.36
x1 = ∑x/ n1 = 140.4/6 = 23.40
(∑ x )
2
(140.4) 2
∑ x2 − n 6
3432.36 −
s1 = = = 5.42217668
n −1 6 −1
From Sample 2: n2 = 6, ∑x = 150.7, ∑x2 = 3957.61
x 2 = ∑x/ n2 = 150.7/6 = 25.12
(∑ x )
2
(150.7) 2
∑x 2
−
n
3957.61 −
6
s2 = = = 5.87415242
n −1 6 −1
(continued on next page)
272 Chapter 10 Estimation and Hypothesis Testing: Two Populations
(continued)
Step 1: H0: μ1 − μ 2 = 0, H1: μ1 − μ 2 < 0
Step 2: Since σ 1 and σ 2 are unknown but assumed to be equal, the samples are independent,
the sample sizes are small, and the populations approximately are normally distributed, use the
t distribution.
Step 3: df = n1 + n2 − 2 = 6 + 6 – 2 = 10
For α = .025, the critical value of t is –2.228.
(n1 − 1) s12 + (n2 − 1) s 22 (6 − 1)(5.42217668) 2 + (6 − 1)(5.87415242) 2
Step 4: s p = = = 5.65268373
n1 + n2 − 2 6+6−2
1 1 1 1
s x1 − x2 = s p + = 5.65268373 + = 3.26357847
n1 n2 6 6
( x1 − x2 ) − ( μ1 − μ 2 ) (23.40 − 25.12) − 0
t= = = –.527
s x1 − x2 3.26357847
Step 5: Do not reject H0 since –.527 > –2.228.
Conclude that the mean gas mileage is not lower without the additive.
c. The conclusions are different since H0 was rejected in Exercise 10.56 but in part b of this exercise, H0
was not rejected.
10.99 We need to obtain sample information about the incidence of brain tumors in people who do not use
cellular phones and the incidence of brain tumors in people who do use cellular phones. Since the
occurrence of brain tumor is a fairly rare event, very large samples (perhaps tens of thousands) of healthy
people from each group (people who use cellular phones and those who do not) must be used. For each
group, we would count the number of people that develop a brain tumor over some period of time (say a
year or two). The hypotheses would be H0: p1 – p2 = 0 versus H1: p1 – p2 < 0 where
p1 = P(brain tumor if do not use cellular phones) and p2 = P(brain tumor if use cellular phones).
If the concern is about product liability, the worst mistake would be to conclude cellular phones do not
increase the risk of brain tumor when they really do. This would be a Type II error. A higher α level
would be appropriate since a higher α level means a lower level for β, the probability of a Type II error.
The test is based on the standard normal distribution with the following test statistic.
( pˆ 1 − pˆ 2 ) − ( p1 − p2 ) ⎛1 1 ⎞
z= where, s pˆ 1 − pˆ 2 = pq ⎜ + ⎟ .
s pˆ 1 − pˆ 2 ⎝ 1
n n 2⎠
σ 12 σ 22
10.100 a. We require zσ x1 − x2 ≤ 5 where σ x1 − x2 = + .
n1 n2
σ 12 σ 22 (15) 2 (10) 2 325
Let n = n1 = n2 , so σ x1 − x2 = + = + =
n1 n2 n n n
325 1.65 325 (1.65) 2 (325)
For the 90% confidence level, 1.65 ≤ 5 , so n≥ , and n ≥ = 35.39 ≈ 36 .
n 5 (5) 2
Thus, use a sample size of 36 for each class.
Chapter 10 Supplementary Exercises 273
b. H0: μ1 − μ 2 = 0, H1: μ1 − μ 2 ≠ 0
For α = .05, the critical values of z are –1.96 and 1.96.
σ 12 σ 22 (15) 2 (10) 2
Using n1 = n2 = 36 (from part a), we obtain σ x1 − x2 = + = + = 3.00462606
n1 n2 36 36
points. To reject H 0 , we require that z is either less than –1.96 or greater than 1.96, which can be
( x1 − x 2 ) − ( μ1 − μ 2 ) x1 − x 2 − 0
written as | z | > 1.96. Then, > 1.96 and > 1.96 . Hence,
σ x1 − x2 3.00462606
x1 − x 2 > 5.89 . Thus, the sample means must differ by at least 5.89 points in order to conclude that
the two population means are different.
c. A paired sample test would be inappropriate in this case because different groups of students are used
in the two samples, so the samples are independent. If the same group of students were taught by both
instructors, then a paired sample test would also be inappropriate because if a person is taught
something by one instructor, that would influence the effectiveness of the instruction by the other
instructor.
σ 12 σ 22 (.4) 2 (.7) 2
10.101 σ x1 − x2 = + = + = .14146063
n1 n2 25 36
( x1 − x 2 ) − ( μ1 − μ 2 ) −.15 − (1.3 − 1.5)
a. For x1 − x2 = −.15 : z = = =.35
σ x1 − x2 .14146063
( x1 − x 2 ) − ( μ1 − μ 2 ) .15 − (1.3 − 1.5)
For x1 − x2 = .15 : z = = = 2.47
σ x1 − x2 .14146063
Then,
P ⎡⎣ −.15 < ( x1 − x 2 ) < .15⎤⎦ = P (.35 < z < 2.47 ) = P ( z < 2.47 ) − P ( z < .35)
= .9932 − .6368 = .3564.
( x1 − x 2 ) − ( μ1 − μ 2 ) 0 − (1.3 − 1.5)
b. For x1 − x2 = 0 : z = = = 1.41
σ x1 − x2 .14146063
P( x1 − x2 > 0) = P( z > 1.41) = 1 − P( z ≤ 1.41) = 1 − .9207 = .0793
c. P ⎡⎣{−.15 < ( x1 − x 2 ) < .15} | x1 = 2.0⎤⎦ = P ⎡⎣ −.15 < (2.0 − x2 ) < .15⎤⎦ = P [1.85 < x2 < 2.15]
σ x2 = σ 2 / n2 = .7 / 36 = .11666667
x2 − μ 2 1.85 − 1.5
For x 2 = 1.85: z = = = 3.00
σ x2 .11666667
x2 − μ 2 2.15 − 1.5
For x 2 = 2.15: z = = = 5.57
σ x2 .11666667
Hence, P(1.85 < x 2 < 2.15) = P(3.00 < z < 5.57) = P(z < 5.57) − P(z < 3.00) = 1 − .9987 = .0013.
d. If the assumption is reasonable, it means that there is not a lot of difference in the average amount of
weight loss between the two diets. Hence, it does not matter which diet is selected.
274 Chapter 10 Estimation and Hypothesis Testing: Two Populations
10.102 Let x1 = the number of male voters in 100 who favor the candidate and x 2 = the number of female voters
in 100 who favor the candidate. Let p̂1 and p̂2 be the corresponding sample proportions. Then,
p1q1 p 2 q 2 (.65)(.35) (.40)(.60)
σ pˆ 1 − pˆ 2 = + = + = .06837397, and
n1 n2 100 100
⎛ x − x 2 10 ⎞ ⎛ x1 x ⎞
P( x1 > x 2 + 10) = P[( x1 − x 2 ) ≥ 10] = P ⎜ 1 ≥ ⎟ = P ⎜⎝ − 2 ≥ .10 ⎟ = P[( pˆ 1 − pˆ 2 ) ≥ .10]
⎝ 100 100 ⎠ 100 100 ⎠
( pˆ − pˆ 2 ) − ( p1 − p2 ) .10 − (.65 − .40)
For p̂1 – p̂2 = .10: z = 1 = = −2.19
σ pˆ 1 − pˆ 2 .06837397
Hence, P( pˆ1 − pˆ 2 ≥ .10) = P( z ≥ −2.19) = 1 − P( z < −2.19) = 1 − .0143 = .9857 .
Self-Review Test
1. a
2. Two samples are independent if they are drawn from two different populations and the elements of the two
samples are not related. Example 10.1 in the text provides an example of independent samples. In two
dependent samples, the elements of one sample are related to the elements of the second sample. Example
10.2 in the text provides an example of dependent samples.
σ 12 σ 22 (.8) 2 (1.3) 2
3. a. σ x1 − x2 = + = + = .22315914
n1 n2 40 50
The 99% confidence interval for μ1 − μ 2 is
( x1 − x2 ) ± zσ x1 − x2 = (7.6 − 5.4) ± 2.58(.22315914) = 2.2 ± .58 = 1.62 to 2.78
6. Zeke’s Elmer’s d d2
1058 995 63 3969
544 540 4 16
1349 1175 174 30,276
1296 1350 –54 2916
676 605 71 5041
998 970 28 784
1698 1520 178 31,684
∑d = 464 ∑d2 = 74,686
d = ∑d /n = 464/7 = $66.29
(∑ d ) 2 (464) 2
∑d − 74, 686 −
2
sd = n = 7 = $85.56618157
n −1 7 −1
s d = s d / n = 85.56618157 / 7 = $32.34097672
df = n – 1 = 7 – 1 = 6
(continued on next page)
276 Chapter 10 Estimation and Hypothesis Testing: Two Populations
(continued)
a. The 99% confidence interval for μd is
d ± ts d = 66.29 ± 3.707(32.34097672) = 66.29 ± 119.89 = –$53.60 to $186.18