714 Chapter 12: Chi-Square Tests and Nonparametric Tests

CHAPTER 12

2
12.1 (a) For df = 1 and α = 0.01, χ = 6.6349.
2
(b) For df = 1 and α = 0.025, χ = 5.0239.
2
(c) For df = 1 and α = 0.05, χ = 3.8415.

2
12.2 (a) For df = 1 and α = 0.05, χ = 3.8415.
2
(b) For df = 1 and α = 0.025, χ = 5.0239.
2
(c) For df = 1 and α = 0.01, χ = 6.6349.

12.3 (a)-(b)
Observed Freq Expected Freq Observed Freq Expected Freq Total Obs, Row 1
20 20 30 30 50
chi-sq contrib= 0 chi-sq contrib= 0
Observed Freq Expected Freq Observed Freq Expected Freq Total Obs, Row 2
30 30 45 45 75
chi-sq contrib= 0 chi-sq contrib= 0
Total Obs, Col 1 Total Obs, Col 2 GRAND TOTAL
50 75 125
( f0 – fe )2
(c) χ STAT
2
= ∑
All Cells fe
= 0 + 0 + 0 + 0 = 0. Since χ STAT
2
< 3.841, it is not significant

at the 5% level of significance.

12.4 (a)
Observed Freq Expected Freq Observed Freq Expected Freq Total Obs, Row 1
20 25 30 25 50
chi-sq contrib= 1.00 chi-sq contrib= 1.00
Observed Freq Expected Freq Observed Freq Expected Freq Total Obs, Row 2
30 25 20 25 50
chi-sq contrib= 1.00 chi-sq contrib= 1.00
Total Obs, Col 1 Total Obs, Col 2 GRAND TOTAL
50 50 100
(b) Decision rule: If χ 2 > 3.841, reject H0.
( f0 – fe )2
Test statistic: χ STAT
2
= ∑
All Cells fe
= 1.00 + 1.00 + 1.00 + 1.00 = 4

( f0 – fe )2
Decision: Since χ STAT
2
= ∑
All Cells fe
= 4 is greater than the critical value of 3.841,

it is significant at the 5% level of significance.

Copyright © 2009 Pearson Education, Inc. Publishing as Prentice Hall.

 715

12.5 (a) Excel output:

Observed Frequencies
Gender
Enjoy Shopping Male Female Total
Yes 136 224 360
No 104 36 140
Total 240 260 500

Expected Frequencies
Gender
Enjoy Shopping Male Female Total
Yes 172.8 187.2 360
No 67.2 72.8 140
Total 240 260 500
Level of Significance 0.01
Number of Rows 2
Number of Columns 2
Degrees of Freedom 1
Results
Critical Value 6.634891
Chi-Square Test Statistic 53.8258
p-Value 2.19E-13
Reject the null hypothesis
H0: π 1 = π 2 H1: π 1 ≠ π 2 where population: 1 = males, 2 = females
Decision rule: df = 1. If χ 2
STAT > 6.635, reject H0.
Test statistic: χ 2
STAT = 53.8258
Decision: Since χ STAT
2
= 53.8258 is greater than the upper critical bound of 6.6349,
reject H0. There is enough evidence to conclude that there is significant difference
between the proportions of males and females who enjoy shopping for clothing at the
0.01 level of significance.
(b) p-value = virtually zero. The probability of obtaining a test statistic of 53.8258 or
larger when the null hypothesis is true is virtually zero.
(c) (a) H0: π 1 = π 2 H1: π 1 ≠ π 2 where Populations: 1 = males, 2 = females
Decision: Since χ STAT
2
= 0.0106 is less than the upper critical bound of
6.635, do not reject H0. There is not enough evidence to conclude that the
proportion of males and females who enjoy shopping for clothing are
different.
(b) p-value = 0.9180. The probability of obtaining a test statistic of 0.0106 or
larger when the null hypothesis is true is 0.9180.
(d) The results of (a) – (c) are exactly the same as those of Problem 10.31. The χ STAT2
in
(a) and the Z STAT in Problem 10.31 (a) satisfy the relationship that
= 53.8258 = (Z STAT ) = −7.342 and the p-value obtained in (b) is exactly the
2
χ STAT
2

same as the p-value in Problem 10.31 (b). Similarly, the χ STAT

2
in (c) and the Z STAT
= 0.0106 = (Z STAT )
2
in Problem 10.31 (d) satisfy the relationship that χ STAT
2

= −0.1032 and the p-value obtained in (c) is exactly the same as the p-value in
Problem 10.31 (d).

Copyright © 2009 Pearson Education, Inc. Publishing as Prentice Hall.

716 Chapter 12: Chi-Square Tests and Nonparametric Tests

12.6 PHStat output:

Chi-Square Test
Observed Frequencies
Column variable
Row variable Parents Teachers Total
Very Confident 401 162 563
Not Very Confident 684 648 1332
Total 1085 810 1895
Expected Frequencies
Column variable
Row variable Parents Teachers Total
Very Confident 322.3509 240.6491 563
Not Very Confident 762.6491 569.3509 1332
Total 1085 810 1895
Data
Level of Significance 0.01
Number of Rows 2
Number of Columns 2
Degrees of Freedom 1
Results
Critical Value 3.8415
Chi-Square Test Statistic 63.86862
p-Value 1.33E-15
Reject the null hypothesis

Expected frequency assumption

is met.
(a) H0: π 1 = π 2 H1: π 1 ≠ π 2 where population: 1 = parents, 2 = teachers
Decision rule: df = 1. If χ 2
STAT > 3.8415, reject H0.
Test statistic: χ 2
STAT = 63.8686
2
Decision: Since χ STAT = 63.8686 is greater than the upper critical bound of 3.8415,
reject H0. There is enough evidence to conclude that there is significant difference
between the proportion of parents and teachers who enjoy shopping for clothing at the
0.05 level of significance.
(b) p-value is virtually 0. The probability of obtaining a test statistic of 63.8686 or larger
when the null hypothesis is true is virtually zero.
(c) You should not compare the results in (a) to those of Problem 10.32 part (b) because
Problem 10.32 was a one-tail test.

12.7 (a) p =22/50 = 0.44

(b) p = 3/50 = 0.06
(c) H0: π 1 = π 2 H1: π 1 ≠ π 2
Decision rule: df = 1. If χ STAT
2
> 3.841, reject H0. Test statistic: χ STAT
2
= 19.2533
Decision: Since χ STAT
2
= 19.2533 is greater than the upper critical bound of 3.841, reject
H0. There is enough evidence of a significance difference in the proportion willing to
delay the date of the concert and the proportion willing to delay receipt of a new PDA.

Copyright © 2009 Pearson Education, Inc. Publishing as Prentice Hall.

 717

12.8 (a) H0: π 1 = π 2 H1: π 1 ≠ π 2

PHStat output with computation:
Observed Frequencies
Column variable
Row variable Less Than More Than Total
$50,000 $50,000
Fair 225 180 405
Unfair 280 320 600
Total 505 500 1005

Expected Frequencies
Column variable
Row variable Less Than More Than Total
$50,000 $50,000
Fair 203.5075 201.4925 405
Unfair 301.4925 298.5075 600
Total 505 500 1005

Data
Level of Significance 0.05
Number of Rows 2
Number of Columns 2
Degrees of Freedom 1

Results
Critical Value 3.841459
Chi-Square Test Statistic 7.64198
p-Value 0.005703
Reject the null hypothesis

Decision rule: df = 1. If χ STAT

2
> 3.841, reject H0.
( f0 – fe )2
Test statistic: χ STAT
2
= ∑
All Cells fe
= 7.642

Decision: Since χ STAT

2
= 7.642 is greater than the upper critical bound of 3.841, reject
H0. There is enough evidence to conclude that there is a significant difference in the
proportion of adults who think the U. S. tax code is unfair between the two income
groups.
(b) p-value is 0.0057. The probability of obtaining a test statistic of 7.642 or larger when
the null hypothesis is true is 0.0057.
(c) The results of (a) and (b) are exactly the same as those of Problem 10.34. The χ STAT2

in (a) and the ZSTAT in Problem 10.34 (a) satisfy the relationship that χ STAT
2
= 7.642 =
(Z STAT )2 = (- 2.7644)2 and the p-value in Problem 10.34 (b) is exactly the same as
the p-value obtained in (b).

Copyright © 2009 Pearson Education, Inc. Publishing as Prentice Hall.

718 Chapter 12: Chi-Square Tests and Nonparametric Tests

12.9 (a)
Age
Source of News 36-50 Above 50 Total
Newspapers 82 104 186
Other 118 96 214
Total 200 200 400
(b) H0: π 1 = π 2 H1: π 1 ≠ π 2
PHStat output with computation:

Expected Frequencies
Age
Source of News 36-50 Above 50 Total
Newspapers 93 93 186
Other 107 107 214
Total 200 200 400

Data
Level of Significance 0.05
Number of Rows 2
Number of Columns 2
Degrees of Freedom 1

Results
Critical Value 3.841459
Chi-Square Test Statistic 4.863833
p-Value 0.027425
Reject the null hypothesis
Decision rule: df = 1. If χ STAT
2
> 3.841, reject H0.
( f0 – fe )2
Test statistic: χ 2
STAT = ∑ = 4.8638
All Cells fe
Decision: Since χ STAT
2
= 4.638 is greater than the upper critical bound of 3.841, reject
H0. There is enough evidence to conclude that there is a difference in the proportion who
get their news primarily from newspapers between those respondents 36 to 50 years old
and those respondents above 50 years old.
(b) p-value is 0.0275. The probability of obtaining a test statistic of 4.8638 or larger
when the null hypothesis is true is 0.0275.
(c) The results of (a) and (b) are exactly the same as those of Problem 10.37. The χ STAT2

in (a) and the ZSTAT in Problem 10.37 (a) satisfy the relationship that χ STAT
2
= 4.8638
= (Z STAT ) = (- 2.2054) and the p-value in Problem 10.37 (b) is exactly the same
2 2

as the p-value obtained in (b).

Copyright © 2009 Pearson Education, Inc. Publishing as Prentice Hall.

 719

12.10 (a) H0: π 1 = π 2 H1: π 1 ≠ π 2

PHStat output with computation:
Observed Frequencies
Column variable
Row variable Undergraduate MBA Total
Highest cost fund 27 18 45
Not Highest cost fund 73 82 155
Total 100 100 200

Expected Frequencies
Column variable
Row variable Undergraduate MBA Total
Highest cost fund 22.5 22.5 45
Not Highest cost fund 77.5 77.5 155
Total 100 100 200

Data
Level of Significance 0.05
Number of Rows 2
Number of Columns 2
Degrees of Freedom 1

Results
Critical Value 3.841459
Chi-Square Test Statistic 2.322581
p-Value 0.127508
Do not reject the null hypothesis

Decision rule: df = 1. If χ STAT

2
> 3.841, reject H0.
( f0 – fe )2
Test statistic: χ STAT
2
= ∑
All Cells fe
= 2.3226

Decision: Since χ STAT

2
= 2.3226 is smaller than the upper critical bound of 3.841, do not
reject H0. There is not enough evidence to conclude that there is a significant difference
between undergraduate and M.B.A. students in the proportion who selected the highest
cost fund.
(b) p-value is 0.1275. The probability of obtaining a test statistic of 2.3226 or larger
when the null hypothesis is true is 0.1275.
(c) The results of (a) and (b) are exactly the same as those of Problem 10.36. The χ STAT2

in (a) and the ZSTAT in Problem 10.36 (a) satisfy the relationship that χ STAT
2
= 2.3226
= (Z STAT ) = (1.5240) and the p-value in Problem 10.36 (b) is exactly the same as
2 2

the p-value obtained in (b).

12.11 (a) df = (r – 1)(c – 1) = (2 – 1)(5 – 1) = 4

(b) χ 2 = 9.488 (c) χ 2 = 13.277

Copyright © 2009 Pearson Education, Inc. Publishing as Prentice Hall.

720 Chapter 12: Chi-Square Tests and Nonparametric Tests

12.12 (a) The expected frequencies in the first row are 20, 30, and 40.
The expected frequencies in the second row are 30, 45, and 60.
(b) χ STAT
2
= 12.500. The critical value with 2 degrees of freedom and α = 0.05 is 5.991.
The result is deemed significant.
(c) Pairs of proportions that differ at the 0.05 level are marked with * below:
Pairwise Critical
Comparisons Range p −p
j j'

A to B 0.19582 0.2*
A to C 0.1848 0.3*
B to C 0.1848 0.1
There are two (2) pairs of proportions that differ significantly.

12.13 (a) The expected frequencies in the first row are 25, 25, and 25.
The expected frequencies in the second row are 25, 25, and 25.
(b) χ STAT
2
= 4.000. The critical value with 2 degrees of freedom and α = .05 is 5.991. The
result is not deemed significant.
(c) The Marascuilo procedure is not appropriate.

12.14 PHStat output:

Observed Frequencies
Column variable
Row variable Germany France UK Greece US Total
Yes 100 120 280 390 570 1460
No 900 880 720 610 430 3540
Total 1000 1000 1000 1000 1000 5000

Expected Frequencies
Column variable
Row variable Germany France UK Greece US Total
Yes 292 292 292 292 292 1460
No 708 708 708 708 708 3540
Total 1000 1000 1000 1000 1000 5000

Data
Level of Significance 0.05
Number of Rows 2
Number of Columns 5
Degrees of Freedom 4

Results
Critical Value 9.487728
Chi-Square Test 742.3961
Statistic
p-Value 2.3E-159
Reject the null
hypothesis

Copyright © 2009 Pearson Education, Inc. Publishing as Prentice Hall.

 721

12.14 (a) H 0 : π1 = π 2 = π 3 = π 4 = π 5 H1 : Not all π j are equal.

cont. where population 1 = Germany, 2 = France, 3 = UK, 4 = Greece, 5 = US
( f0 – fe )2
Test statistic: χ STAT
2
= ∑
All Cells fe
= 742.3961

Decision: Since the calculated test statistic 742.3961 is greater than the critical value of
9.4877, you reject H 0 and conclude that there is a difference in the proportion of people
who eat out at least once a week in the various countries.
(b) p-value is virtually zero. The probability of obtaining a data set which gives rise to a
test statistic of 742.3961 or more is virtually zero if there is no difference in the
proportion of people who eat out at least once a week in the various countries.
(c) Excel output of the Marascuilo procedure:

Sample Sample
Group Proportion Size
1 0.10 1000
2 0.12 1000
3 0.28 1000
4 0.39 1000
5 0.57 1000

Other Data
Level of significance 0.05
d.f 4
Sqrt(chi-square) 3.0802

Absolute Std. Error Critical

Comparison Difference of Difference Range Results
Group 1 to Group 2 0.02 0.01399 0.04308 Means are not
different
Group 1 to Group 3 0.18 0.01708 0.05260 Means are different
Group 1 to Group 4 0.29 0.01811 0.05578 Means are different
Group 1 to Group 5 0.47 0.01831 0.05639 Means are different
Group 2 to Group 3 0.16 0.01753 0.05399 Means are different
Group 2 to Group 4 0.27 0.01853 0.05709 Means are different
Group 2 to Group 5 0.45 0.01873 0.05768 Means are different
Group 3 to Group 4 0.11 0.02096 0.06457 Means are different
Group 3 to Group 5 0.29 0.02114 0.06510 Means are different
Group 4 to Group 5 0.18 0.02198 0.06769 Means are different
At the 5% level of significance, there is no significant difference between the
proportions of Germany and France while there is a significant difference between all
the remaining pair of countries.

Copyright © 2009 Pearson Education, Inc. Publishing as Prentice Hall.

722 Chapter 12: Chi-Square Tests and Nonparametric Tests

12.15 (a)
Chi-Square Test

Observed Frequencies
Column variable
Row variable Insurance Pharmacies Research Total
Yes 40 80 90 210
No 160 120 110 390
Total 200 200 200 600

Expected Frequencies
Column variable
Row variable Insurance Pharmacies Research Total
Yes 70 70 70 210
No 130 130 130 390
Total 200 200 200 600

Data
Level of Significance 0.05
Number of Rows 2
Number of Columns 3
Degrees of Freedom 2

Results
Critical Value 5.991465
Chi-Square Test Statistic 30.76923
p-Value 2.08E-07
Reject the null hypothesis

Expected frequency assumption

is met.
H0 : π1 = π 2 = π 3 H1 : Not all π j are equal.
where population 1 = Insurance, 2 = Pharmacies, 3 = Research
( f0 – fe )2
Test statistic: χ STAT
2
= ∑
All Cells fe
= 30.7692

Decision: Since the calculated test statistic 30.7692 is greater than the critical value of
5.9915, you reject H 0 and conclude that there is a difference in attitudes on sharing of a
patient’s medical records without the patient’s consent to the different organizational
groups.

Copyright © 2009 Pearson Education, Inc. Publishing as Prentice Hall.

 723

12.15 (b) p-value is virtually zero. The probability of obtaining a data set which gives rise to a
cont. test statistic of 30.7692 or more is virtually zero if there is no difference in attitudes on
sharing of a patient’s medical records without the patient’s consent to the different
organizational groups.
(c) Excel output of the Marascuilo procedure:

Marascuilo Procedure

Level of Significance 0.05

Square Root of Critical 2.447746831
Value

Sample Proportions
Group 1 0.2
Group 2 0.4
Group 3 0.45

MARASCUILO TABLE
Proportions Absolute Differences Critical Range
| Group 1 - Group 2 | 0.2 0.109466566 Significant
| Group 1 - Group 3 | 0.25 0.110488049 Significant

| Group 2 - Group 3 | 0.05 0.120847817 Not significant

At the 5% level of significance, there is no significant difference in attitude on sharing
information to pharmacy group and research group while there is a significantly lower
portion of patients object to sharing patient’s medical records without the patient’s
consent to insurance group.

Copyright © 2009 Pearson Education, Inc. Publishing as Prentice Hall.

724 Chapter 12: Chi-Square Tests and Nonparametric Tests

12.16 (a) H 0 : π1 = π 2 = π 3 H1 : at least one proportion differs

where population 1 = under 35, 2 = 35-54, 3 = over 54
PHStat output:
Observed
Frequencies
Column variable Calculations
Row variable Under 35 35-54 Over 54 Total fo-fe
Saturday 48 56 24 128 5.333 13.333 -18.667
A Day other than 152 144 176 472 -5.333 -13.333 18.667
Saturday
Total 200 200 200 600

Expected Frequencies
Column variable
Row variable Under 35 35-54 Over 54 Total (fo-fe)^2/fe
Saturday 42.667 42.667 42.667 128 0.667 4.167 8.167
A Day other than 157.333 157.333 157.333 472 0.181 1.130 2.215
Saturday
Total 200 200 200 600

Data
Level of Significance 0.05
Number of Rows 2
Number of Columns 3
Degrees of Freedom 2

Results
Critical Value 5.9915
Chi-Square Test 16.5254
Statistic
p-Value 0.0003
Reject the null
hypothesis
Decision rule: df = (c – 1) = (3 – 1) = 2. If χ STAT
2
> 5.9915, reject H0.
( f0 – fe )2
Test statistic: χ 2
STAT = ∑ = 16.5254
All Cells fe
Decision: Since χ STAT
2
= 16.5254 is greater than the upper critical bound of 5.9915,
reject H0. There is enough evidence to conclude that there is a significant relationship
between age and major grocery shopping day.
(b) p-value = 0.0003. The probability of obtaining a sample that gives rise to a test statistic
that is equal to or more than 16.5254 is 0.03% if the null hypothesis is true.

Copyright © 2009 Pearson Education, Inc. Publishing as Prentice Hall.

 725

12.16 (c)
cont.
Pairwise Critical pj − p j'
Comparisons Range
1 to 2 0.1073 0.04
2 to 3 0.0959 0.16*
1 to 3 0.0929 0.12*
There is a significance difference between the 35-54 and over 54 groups, and between
the under 35 and over 54 groups.
(d) The stores can use this information to target their marketing on the specific group of
shoppers on Saturday and the days other than Saturday.

12.17 (a) H 0 : π1 = π 2 = π 3 H1 : at least one proportion differs

where population 1 = under 35, 2 = 35-54, 3 = over 54
Decision rule: df = (c – 1) = (3 – 1) = 2. If χ STAT
2
> 5.9915, reject H0.
Test statistic: χ STAT
2
= 4.1314
Decision: Since χ STAT
2
= 4.1314 is less than the upper critical bound of 5.9915, do not
reject H0. There is not enough evidence to show that there is a significant relationship
between age and major grocery shopping day.
(b) p-value = 0.1267. The probability of obtaining a sample that gives rise to a test statistic
that is equal to or more than 4.1314 is 12.67% if the null hypothesis is true.
The larger the sample size, the more power the χ 2 test has and, hence, there is a higher
likelihood of rejecting a false null hypothesis.

Copyright © 2009 Pearson Education, Inc. Publishing as Prentice Hall.

726 Chapter 12: Chi-Square Tests and Nonparametric Tests

12.18 PHStat output:

Chi-Square Test

Observed Frequencies
Column variable
Row variable men age 62 – women age 62 – men age 65 – women age 65 – Total
64 64 69 69
Still Working 52 41 31 23 147
Not Working 48 59 69 77 253
Total 100 100 100 100 400

Expected Frequencies
Column variable
Row variable men age 62 – women age 62 – men age 65 – women age 65 – Total
64 64 69 69
Still Working 36.75 36.75 36.75 36.75 147
Not Working 63.25 63.25 63.25 63.25 253
Total 100 100 100 100 400

Data
Level of Significance 0.05
Number of Rows 2
Number of Columns 4
Degrees of Freedom 3

Results
Critical Value 7.814728
Chi-Square Test 20.33825
Statistic
p-Value 0.000144
Reject the null hypothesis

Expected frequency assumption

is met.
(a) Ho: H1 Not all are equal.
where population 1 = men age 62 – 64, 2 = women age 62 – 64, 3 = men age 65 – 69, 4
= women age 65 – 69
( f0 – fe )2
Test statistic: χ STAT
2
= ∑
All Cells fe
= 20.3383

Decision: Since the calculated test statistic 20.3383 is greater than the critical value of
7.8147, you reject H 0 and conclude that there is enough evidence of a significant
difference among the groups with respect to the proportion who are still working.
(b) p-value is virtually zero. The probability of obtaining a data set which gives rise to a
test statistic of 20.3383 or more is virtually zero if there is no difference among the
groups with respect to the proportion who are still working.

Copyright © 2009 Pearson Education, Inc. Publishing as Prentice Hall.

 727

12.18 (c) PHStat output of the Marascuilo procedure:

cont.
Marascuilo Procedure
Level of Significance 0.05
Square Root of Critical Value 2.795483458

Sample Proportions
Group 1 0.52
Group 2 0.41
Group 3 0.31
Group 4 0.23
MARASCUILO TABLE
Proportions Absolute Critical Range
Differences
| Group 1 - Group 2 | 0.11 0.195983129 Not significant
| Group 1 - Group 3 | 0.21 0.190318846 Significant
| Group 1 - Group 4 | 0.29 0.182607348 Significant
| Group 2 - Group 3 | 0.1 0.188731368 Not significant
| Group 2 - Group 4 | 0.18 0.180952229 Not significant
| Group 3 - Group 4 | 0.08 0.174801561 Not significant
The proportion of men age 62 – 64 who are still working is significantly different
from the other groups at 5% level of significance. There is no significant difference
in proportions among the remaining groups.

12.19 (a) PHStat output:

Chi-Square Test
Observed Frequencies
Column variable
Row variable 1995 2005 2007 Total
Paid with Plastic 90 125 155 370
Did Not Pay with Plastic 410 375 345 1130
Total 500 500 500 1500

Expected Frequencies
Column variable
Row variable 1995 2005 2007 Total
Paid with Plastic 123.3333 123.3333 123.3333 370
Did Not Pay with Plastic 376.6667 376.6667 376.6667 1130
Total 500 500 500 1500

Data
Level of Significance 0.05
Number of Rows 2
Number of Columns 3
Degrees of Freedom 2
Results
Critical Value 5.991465
Chi-Square Test Statistic 22.78163
p-Value 1.13E-05
Reject the null hypothesis
Expected frequency assumption is met.

Copyright © 2009 Pearson Education, Inc. Publishing as Prentice Hall.

728 Chapter 12: Chi-Square Tests and Nonparametric Tests

12.19 (a) Ho: H1 Not all are equal.

where population 1 = 1995, 2 = 2005, 3 = 2007
( f0 – fe )2
Test statistic: χ 2
STAT = ∑ = 22.7816
All Cells fe
Decision: Since the calculated test statistic 22.7816 is greater than the critical value of
5.9915, you reject H 0 and conclude that there is enough evidence of a significant
difference between the years in the proportion of respondents who paid for a specific
type of purchase with plastic.
(b) p-value is virtually zero. The probability of obtaining a data set which gives rise to a
test statistic of 22.7816 or more is virtually zero if there is no significant difference
between the years in the proportion of respondents who paid for a specific type of
purchase with plastic.
(c) PHStat output of the Marascuilo procedure:
Marascuilo Procedure

Level of Significance 0.05

Square Root of Critical Value 2.447746831

Sample Proportions
Group 1 0.18
Group 2 0.25
Group 3 0.31

MARASCUILO TABLE
Proportions Absolute Critical Range
Differences
| Group 1 - Group 2 | 0.07 0.063367812 Significant
| Group 1 - Group 3 | 0.13 0.065816631 Significant

| Group 2 - Group 3 | 0.06 0.069353787 Not significant

Year 1995 is significantly different from 2005 and 2007 in terms of the proportion of
respondents who paid for a specific type of purchase with plastic at a 5% level of
significance. There is no significant difference between 2005 and 2007.

12.20 df = (r – 1)(c – 1) = (3 – 1)(4 – 1) = 6

12.21 (a) χ 2 = 21.026 2

(d) χ = 23.209
(b) χ 2 = 26.217 2
(e) χ = 23.209
(c) χ 2 = 30.578

Copyright © 2009 Pearson Education, Inc. Publishing as Prentice Hall.

 729

12.22 (a) The answer to this part will depend on individual’s view.
(b) PHStat output:
Chi-Square Test

Observed Frequencies
Column variable
Row variable Yes No Neutral Total
Women 61 11 28 100
Men 49 27 24 100
Total 110 38 52 200

Expected Frequencies
Column variable
Row variable Yes No Neutral Total
Women 55 19 26 100
Men 55 19 26 100
Total 110 38 52 200

Data
Level of Significance 0.01
Number of Rows 2
Number of Columns 3
Degrees of Freedom 2

Results
Critical Value 9.21034
Chi-Square Test Statistic 8.353625
p-Value 0.015347
Do not reject the null hypothesis

Expected frequency assumption

is met.
H0: There is no relationship between genders and how they feel about executive pay.
H1: There is a relationship between genders and how they feel about executive pay.
( f0 – fe )2
Test statistic: χ STAT
2
= ∑
All Cells fe
= 8.3536

Decision: Since the calculated test statistic 8.3536 is smaller than the critical value of
9.2103, you do not reject H 0 and conclude that there is not enough evidence that men
and women feel differently about executive pay.

Copyright © 2009 Pearson Education, Inc. Publishing as Prentice Hall.

730 Chapter 12: Chi-Square Tests and Nonparametric Tests

12.22 (c) PHStat output:

cont.
Chi-Square Test
Observed Frequencies
Column variable
Row variable Yes No Neutral Total
Women 122 22 56 200
Men 98 54 48 200
Total 220 76 104 400
Expected Frequencies
Column variable
Row variable Yes No Neutral Total
Women 110 38 52 200
Men 110 38 52 200
Total 220 76 104 400
Data
Level of Significance 0.01
Number of Rows 2
Number of Columns 3
Degrees of Freedom 2
Results
Critical Value 9.21034
Chi-Square Test Statistic 16.70725
p-Value 0.000236
Reject the null hypothesis

is met.
H0: There is no relationship between genders and how they feel about executive pay.
H1: There is a relationship between genders and how they feel about executive pay.
( f0 – fe )2
Test statistic: χ 2
STAT = ∑ = 16.7073
All Cells fe
Decision: Since the calculated test statistic 16.7073 is larger than the critical value of
9.2103, you reject H 0 and conclude that there is enough evidence that men and women
feel differently about executive pay.
(d) As you can see from the results in (b)-(c), the same proportions will result in different
test statistic and, hence, potentially totally different decision and conclusion on the test.
Therefore, it is important to always report the sample size in survey results.

12.23 H0: There is no relationship between age group and type of communication preferred.
H1: There is a relationship between age group and type of communication preferred.
Decision rule: If χ STAT
2
> 16.9190, reject H0.
( f0 – fe )2
Test statistic: χ STAT
2
= ∑
All Cells fe
=52.6607

Decision: Since χ STAT

2
= 52.6607 is greater than the critical bound of 16.9190, reject H0. There
is enough evidence to conclude there is a relationship between age group and type of
communication preferred.

Copyright © 2009 Pearson Education, Inc. Publishing as Prentice Hall.

 731

12.24 (a) H0: There is no relationship between the commuting time of company employees and
the level of stress-related problems observed on the job.
H1: There is a relationship between the commuting time of company employees and
the level of stress-related problems observed on the job.
PHStat output:
Observed Frequencies
Stress Calculations
Commuting Time High Moderate Low Tota fo-fe
l
Under 15 min. 9 5 18 32 -3.1379 -0.2414 3.3793
15-45 min. 17 8 28 53 -3.1034 -0.6810 3.7845
Over 45 min. 18 6 7 31 6.2414 0.9224 -7.1638
Total 44 19 53 116
Expected Frequencies
Stress
Commuting Time High Moderate Low Tota (fo-fe)^2/fe
l
Under 15 min. 12.1379 5.2414 14.6207 32 0.8112 0.0111 0.7811
15-45 min. 20.1034 8.6810 24.2155 53 0.4791 0.0534 0.5915
Over 45 min. 11.7586 5.0776 14.1638 31 3.3129 0.1676 3.6233
Total 44 19 53 116
Level of Significance 0.01
Number of Rows 3
Number of Columns 3
Degrees of Freedom 4
Results
Critical Value 13.2767
Chi-Square Test 9.8311
Statistic
p-Value 0.04337
Do not reject the null hypothesis
(a) Decision rule: If χ STAT
2
> 13.277, reject H0.
( f0 – fe )2
Test statistic: χ STAT
2
= ∑
All Cells fe
= 9.831

Decision: Since χ STAT

2
= 9.831 is less than the critical bound of 13.277, do not reject H0.
There is not enough evidence to conclude there is any relationship between the
commuting time of company employees and the level of stress-related problems
observed on the job.
(b) Decision rule: If χ STAT
2
> 9.488, reject H0.
Decision: Since the χ STAT
2
= 9.831 is greater than the critical bound of 9.488, reject
H0. There is enough evidence at the 0.05 level to conclude there is a relationship
between the commuting time of company employees and the level of stress-related
problems observed on the job.

Copyright © 2009 Pearson Education, Inc. Publishing as Prentice Hall.

732 Chapter 12: Chi-Square Tests and Nonparametric Tests

12.25 H0: There is no relationship between the age group and where people primarily get
their news.
H1: There is a relationship between the age group and where people primarily get
their news.
Excel output:
Observed
Frequencies
Age Group
Media Under 36 36-50 50+ Total
Local TV 107 119 133 359
National TV 73 102 127 302
Radio 75 97 109 281
Local Newspaper 52 79 107 238
Internet 95 83 76 254
Total 402 480 552 1434

Expected Frequencies
Age Group
Media Under 36 36-50 50+ Total
Local TV 100.64 120.17 138.19 359
National TV 84.66 101.09 116.25 302
Radio 78.77 94.06 108.17 281
Local Newspaper 66.72 79.67 91.62 238
Internet 71.21 85.02 97.77 254
Total 402 480 552 1434

Data
Level of Significance 0.05
Number of Rows 5
Number of Columns 3
Degrees of Freedom 8

Results
Critical Value 15.50731
Chi-Square Test 22.18122
Statistic
p-Value 0.004591
Reject the null
hypothesis
Decision rule: If χ STAT
2
> 15.5073, reject H0.
Test statistic: χ STAT
2
= 22.1812
Decision: Since χ STAT
2
= 22.1812 is greater than the critical bound of 15.5073, reject H0.
There is evidence of a significant relationship between the age group and where people
primarily get their news. The “50+” group has a lower than expected frequency of
getting their news through the Internet while the “under 36” group has a higher than
expected frequency of getting their news through the Internet.

Copyright © 2009 Pearson Education, Inc. Publishing as Prentice Hall.

 733

12.26 H0: There is no relationship between when the decision is made of what to have for dinner
and the type of household.
H1: There is a relationship between when the decision is made of what to have for dinner and
the type of household.
Decision rule: d.f. = 12. If χ STAT
2
> 21.026, reject H0. Test statistic: χ STAT
2
= 129.520
Decision: Since the χ STAT
2
= 129.520 is greater than the critical bound of 21.026, reject H0.
There is enough evidence to conclude there is a relationship between when the decision is
made of what to have for dinner and the type of household.

12.27 (a) H 0 : π1 = π 2 H1 : π 1 ≠ π 2 where 1 = group1, 2 = group2

Decision rule: If ZSTAT < – 1.96 or ZSTAT > 1.96, reject H0.
B −C 25 − 16
Test statistic: Z STAT = = = 1.4056
B+C 25 + 16
Decision: Since ZSTAT = 1.4056 is between the local critical bound of – 1.96 and the
upper critical bound of 1.96, do not reject H0. There is not enough evidence of a
difference between group 1 and group 2.

12.28 (a) H 0 : π1 ≥ π 2 H1 : π 1 < π 2 where 1 = beginning, 2 = end

Decision rule: If ZSTAT < –1.645, reject H0.
B −C 9 − 22
Test statistic: Z STAT = = = – 2.3349
B+C 9 + 22
Decision: Since ZSTAT = – 2.3349 < – 1.645, reject H0. There is enough evidence to
conclude that the proportion of coffee drinkers who prefer Brand A is lower at the
beginning of the advertising campaign than at the end of the advertising campaign.
(b) p-value = 0.0098. The probability of obtaining a data set which gives rise to a test
statistic smaller than – 2.3349 is 0.98% if the proportion of coffee drinkers who prefer
Brand A is not lower at the beginning of the advertising campaign than at the end of the
advertising campaign.

12.29 (a) H 0 : π1 = π 2 H1 : π 1 ≠ π 2where 1 = beginning, 2 = end

Decision rule: If Z < – 2.5758 or Z > 2.5758, reject H0.
B −C 21 − 36
Test statistic: Z STAT = = = – 1.9868
B+C 21 + 36
Decision: Since ZSTAT = – 1.9868 is in between the two critical bounds, do not reject H0.
There is not enough evidence to conclude there is a difference in the proportion of voters
who favored Candidate A prior to and after the debate.
(b) p-value = 0.0469. The probability of obtaining a sample which gives rise to a test
statistic that differs from 0 by – 1.9868 or more in either direction is 4.69% if there is
not a difference in the proportion of voters who favor Candidate A prior to and after the
debate.

Copyright © 2009 Pearson Education, Inc. Publishing as Prentice Hall.

734 Chapter 12: Chi-Square Tests and Nonparametric Tests

12.30 (a) H 0 : π1 ≥ π 2 H1 : π 1 < π 2

where 1 = before, 2 = after
Decision rule: If ZSTAT < –1.645, reject H0.
B −C 5 − 15
Test statistic: Z STAT = = = – 2.2361
B+C 5 + 15
Decision: Since ZSTAT = – 2.2361 < – 1.645, reject H0. There is enough evidence to
conclude that the proportion who prefer Brand A is lower before the advertising than
after the advertising.
(b) p-value = 0.0127. The probability of obtaining a data set which gives rise to a test
statistic smaller than – 2.2361 is 1.27% if the proportion who prefer Brand A is not
lower before the advertising than after the advertising.

12.31 (a) H 0 : π1 ≥ π 2 H1 : π 1 < π 2

where 1 = last year, 2 = now
Decision rule: If ZSTAT < –1.645, reject H0.
B −C 5 − 20
Test statistic: Z STAT = = =–3
B+C 5 + 20
Decision: Since ZSTAT = – 3 < – 1.645, reject H0. There is enough evidence to conclude
that satisfaction was lower last year prior to introduction of Six Sigma management.
(b) p-value = 0.0014. The probability of obtaining a data set which gives rise to a test
statistic smaller than – 3 is 0.14% if the satisfaction was not lower last year prior to
introduction of Six Sigma management.

12.32 (a) H 0 : π1 ≥ π 2 H1 : π 1 < π 2

where 1 = year 1, 2 = year 2
Decision rule: If ZSTAT < –1.645, reject H0.
B −C 4 − 25
Test statistic: Z STAT = = = – 3.8996
B+C 4 + 25
Decision: Since ZSTAT = – 3.8996 < – 1.645, reject H0. There is enough evidence to
conclude that the proportion of employees absent less than 5 days was lower in year 1
than in year 2.
(b) p-value is virtually zero. The probability of obtaining a data set which gives rise to a
test statistic smaller than –3.8996 is virtually zero if the proportion of employees absent
less than 5 years was not lower in year 1 than in year 2.

12.33 (a) For df = 25 and α = 0.01, χ α2 / 2 = 10.520 and χ 12−α / 2 = 46.928.

(b) For df = 16 and α = 0.05, χ α2 / 2 = 6.908 and χ 12−α / 2 = 28.845.
(c) For df = 13 and α = 0.10, χ α2 / 2 = 5.892 and χ 12−α / 2 = 22.362.

12.34 (a) For df = 23 and α = 0.01, χ α2 / 2 = 9.2604 and χ 12−α / 2 = 44.1814.

(b) For df = 19 and α = 0.05, χ α2 / 2 = 8.9065 and χ 12−α / 2 = 32.8523.
(c) For df = 15 and α = 0.10, χ α2 / 2 = 7.2609 and χ 12−α / 2 = 24.9958.

(n – 1) ⋅ S 2 24 ⋅ 150 2
12.35 χ STAT
2
= = = 54
σ2 100 2

Copyright © 2009 Pearson Education, Inc. Publishing as Prentice Hall.

 735

(n – 1) ⋅ S 2 15 ⋅10 2
12.36 χ STAT
2
= = = 10.417
σ2 12 2

12.37 df = n – 1 = 16 – 1 = 15

12.38 (a) For df = 15 and α = 0.05, χ α2 / 2 = 6.262 and χ 12−α / 2 = 27.488.

(b) For df = 15 and α = 0.05, χ α2 / 2 = 7.261.

2
12.39 (a) If H1 : σ ≠ 12 , do not reject H0 since the test statistic χ = 10.417 falls between the
two critical bounds, χ α2 / 2 = 6.262 and χ 12−α / 2 = 27.488.
2
(b) If H1 : σ < 12 , do not reject H0 since the test statistic χ = 10.417 is greater than the
critical bound 7.261.

12.40 You must assume that the data in the population are normally distributed to be able to use the
chi-square test of a population variance or standard deviation. If the data selected do not
come from an approximately normally distributed population, particularly for small sample
sizes, the accuracy of the test can be seriously affected.
H0: σ ≤ 1.2 F. The standard deviation of the oven temperature has not increased
o
12.41 (a)
above 1.20F.
H1: σ > 1.2 F. The standard deviation of the oven temperature has increased above
o
0
1.2 F.
Decision rule: df = 29. If χ STAT
2
> 42.557, reject H0.
(n – 1) ⋅ S 2 29 ⋅ 2.12
Test statistic: χ STAT
2
= = = 88.813
σ2 1.2 2
Decision: Since the test statistic of χ STAT
2
= 88.813 is greater than the critical
boundary of 42.557, reject H0. There is sufficient evidence to conclude that the
standard deviation of the oven temperature has increased above 1.20F.
(b) You must assume that the data in the population are normally distributed to be able to
use the chi-square test of a population variance or standard deviation.
(c) p-value = 5.53 x 10–8 or 0.00000005. The probability that a sample is obtained whose
standard deviation is equal to or larger than 2.10F. when the null hypothesis is true is
5.53 x 10–8, a very small probability.
Note: The p-value was found using Excel.

12.42 (a) H0: σ = $200. The standard deviation of the amount of auto repairs is equal to $200.
H1: σ ≠ $200. The standard deviation of the amount of auto repairs is not equal to
$200.
Decision rule: df = 24. If χ STAT
2
< 12.401 or χ STAT
2
> 39.364, reject H0.
(n – 1) ⋅ S 2 24 ⋅ 237.52 2
Test statistic: χ STAT
2
= = = 33.849
σ2 200 2
Decision: Since the test statistic of χ STAT
2
= 33.849 is between the critical boundaries
of 12.401 and 39.364, do not reject H0. There is insufficient evidence to conclude that
the standard deviation of the amount of auto repairs is not equal to $200.

Copyright © 2009 Pearson Education, Inc. Publishing as Prentice Hall.

736 Chapter 12: Chi-Square Tests and Nonparametric Tests

12.42 (b) You must assume that the data in the population are normally distributed to be able
cont. to use the chi-square test of a population variance or standard deviation.
(c) p-value = 2(0.0874) = 0.1748. The probability of obtaining a sample whose standard
deviation will give rise to a test statistic equal to or more extreme than 33.849 is
0.1748 when the null hypothesis is true.
Note: The p-value was found using Excel.

12.43 (a) H0: σ = $12. The standard deviation of the monthly cost of calls within the local
calling region is $12.
H1: σ ≠ $12. The standard deviation of the monthly cost of calls within the local
calling region differs from $12.
Decision rule: df = 14. If χ STAT
2
< 6.571 or χ STAT
2
> 23.685, reject H0.
(n – 1) ⋅ S 2 14 ⋅ 9.25 2
Test statistic: χ STAT
2
= = = 8.319
σ2 12 2
Decision: Since the test statistic of χ STAT
2
= 8.319 is between the critical boundaries
of 6.571 and 23.685, do not reject H0. There is insufficient evidence to conclude that
the standard deviation of the monthly cost of calls within the local calling region
differs from $12.
(b) You must assume that the data in the population are normally distributed to be able to
use the chi-square test of a population variance or standard deviation.
(c) p-value = 2(1 – 0.8721) = 0.2558. The probability of obtaining a test statistic equal to
or more extreme than the result obtained from this sample data is 0.2558 if the
standard deviation of the monthly cost of calls within the local calling region is $12.
Note: Excel returns an upper-tail area of 0.8721 for χ STAT
2
= 8.319. But since the
sample standard deviation is smaller than the hypothesized value, the amount
of area in the lower tail is (1 – 0.8721). That value is doubled to accommodate
the two-tail hypotheses.

12.44 (a) H0: σ ≥ 0.035 inch. The standard deviation of the diameter of doorknobs is greater
than or equal to 0.035 inch in the redesigned production process.
H1: σ < 0.035 inch. The standard deviation of the diameter of doorknobs is less than
0.035 inch in the redesigned production process.
Decision rule: df = 24. If χ STAT
2
< 13.848, reject H0.
(n – 1) ⋅ S 2 24 ⋅ 0.025 2
Test statistic: χ 2
STAT = = = 12.245
σ2 0.035 2
Decision: Since the test statistic of χ STAT
2
= 12.245 is less than the critical boundary
of 13.848, reject H0. There is sufficient evidence to conclude that the standard
deviation of the diameter of doorknobs is less than 0.035 inch in the redesigned
production process.
(b) You must assume that the data in the population are normally distributed to
be able to use the chi-square test of a population variance or standard deviation.
(c) p-value = (1 – 0.9770) = 0.0230. The probability of obtaining a test statistic equal to
or more extreme than the result obtained from this sample data is 0.0230 if the
population standard deviation is indeed no less than 0.035 inch.

Copyright © 2009 Pearson Education, Inc. Publishing as Prentice Hall.

 737

12.45 (a) H0: σ = 0.25 ounce. The standard deviation in the weight of raisins packaged per
box is equal to 0.25 ounce.
H1: σ ≠ 0.25 ounce. The standard deviation in the weight of raisins packaged per
box differs from 0.25 ounce.
Decision rule: df = 29. If χ STAT
2
< 16.047 or χ STAT
2
> 45.722, reject H0.
(n – 1) ⋅ S 2 29 ⋅ 0.40576 2
Test statistic: χ STAT
2
= = = 76.395
σ2 0.25 2
Decision: Since the test statistic of χ STAT
2
= 76.395 is greater than the critical
boundary of 45.722, reject H0. There is sufficient evidence to conclude that the
standard deviation in the weight of raisins packaged per box differs from 0.25 ounce.
(b) You must assume that the data in the population are normally distributed to be able to
use the chi-square test of a population variance or standard deviation.
(c) p-value = 2(3.8538 x 10–6) = 7.7076 x 10–6 or 0.0000077. The probability of
obtaining a test statistic equal to or more extreme than the result obtained from this
sample data is 0.0000077 if the population standard deviation is indeed 0.25 ounce.

12.46 (a) H0: σ ≤ 2.5 ampere-hours. The standard deviation in the capacity of the battery is
equal to 2.5 ampere-hours.
H1: σ > 2.5 ampere-hours. The standard deviation in the capacity of the battery
differs from 2.5 ampere-hours.
Decision rule: df = 19. If χ STAT
2
> 30.144, reject H0.
(n – 1) ⋅ S 2 19 ⋅ 2.6589 2
Test statistic: χ STAT
2
= = = 21.492
σ2 2.5 2
Decision: Since the test statistic of χ STAT
2
= 21.492 is less than the critical
boundary of 30.144, do not reject H0. There is not sufficient evidence to conclude
that the standard deviation in the capacity of a certain type of battery differs from 2.5
ampere-hours.
(b) You must assume that the data in the population are normally distributed to be able to
use the chi-square test of a population variance or standard deviation.
(c) p-value = 0.3103. The probability of obtaining a test statistic equal to or more
extreme than the result obtained from this sample data is 0.3103 if the population
standard deviation is indeed no greater than 2.5 ampere-hours.

12.47 (a) The lower and upper critical values are 31 and 59, respectively.
(b) The lower and upper critical values are 29 and 61, respectively.
(c) The lower and upper critical values are 25 and 65, respectively.
(d) As the level of significance α gets smaller, the width of the nonrejection region gets
wider.

12.48 (a) The lower critical value is 31.

(b) The lower critical value is 29.
(c) The lower critical value is 27.
(d) The lower critical value is 25.

12.49 T1 = 4 + 1 + 8 + 2 + 5 + 10 + 11 = 41

12.50 The lower and upper critical values are 40 and 79, respectively.

Copyright © 2009 Pearson Education, Inc. Publishing as Prentice Hall.

738 Chapter 12: Chi-Square Tests and Nonparametric Tests

12.51 Decision: Since T1 = 41 is between the critical bounds of 40 and 79, do not reject H0.

12.52 (a) The ranks for Sample 1 are 1, 2, 4, 5, and 10, respectively.
The ranks for Sample 2 are 3, 6.5, 6.5, 8, 9, and 11, respectively.
(b) T1 = 1 + 2 + 4 + 5 + 10 = 22
(c) T2 = 3 + 6.5 + 6.5 + 8 + 9 + 11 = 44
n(n + 1) 11(12)
(d) T1 + T2 = = = 66 T1 + T2 = 22 + 44 = 66
2 2
12.53 The lower critical value is 20.

12.54 Decision: Since T1 = 22 is greater than the lower critical bound of 20, do not reject H0.

12.55 H0: M1 = M2 where Populations: 1 = traditional, 2 = experimental

There is no difference in performance between the traditional and the experimental
training methods.
H1: M1 ≠ M2 There is a difference in performance between the traditional and the
experimental training methods.
Decision rule: If T1 < 78 or T1 > 132, reject H0.
Test statistic: T1 = 1 + 2 + 3 + 5 + 9 + 10 + 12 + 13 + 14 + 15 = 84
Decision: Since T1 = 84 is between the critical bounds of 78 and 132, do not reject H0. There
is not enough evidence to conclude that there is a difference in median performance between
the traditional and the experimental training methods.

12.56 (a) The data are ordinal.

(b) The two-sample t-test is inappropriate because the data are ordinal, the sample size is
small and the distribution of the ordinal data is not normally distributed.
(c) H0: M1 = M2 where Populations: 1 = California, 2 = Washington
H1: M1 ≠ M2
PHStat output:
Data
Level of Significance 0.05
Population 1 Sample
Sample Size 8
Sum of Ranks 47
Population 2 Sample
Sample Size 8
Sum of Ranks 89
Intermediate Calculations
Total Sample Size n 16
T1 Test Statistic 47
T1 Mean 68
Standard Error of T1 9.521905
Z Test Statistic -2.20544
Two-Tailed Test
Lower Critical Value -1.95996
Upper Critical Value 1.959964
p-value 0.027423
Reject the null hypothesis

Copyright © 2009 Pearson Education, Inc. Publishing as Prentice Hall.

 739

n1 (n + 1) 8(16 + 1)
12.56 (c) μT 1 = = = 68
2 2
n n (n + 1) 8(8)(16 + 1)
cont. σ T1 = 1 2 = = 9.5219
12 12
T1 − μT1
Z STAT = = −2.2054
σT 1

Decision: Since ZSTAT = −2.2054 is lower than the lower critical bounds of −1.96,
reject H0. There is enough evidence of a significant difference in the median rating of
California Cabernets and Washington Cabernets.

12.57 (a) H0: M1 = M2 where Populations: 1 = Untreated, 2 = Treated

H1: M1 ≠ M2
PHStat output:
Wilcoxon Rank Sum Test
Data
Level of Significance 0.05

Population 1 Sample
Sample Size 20
Sum of Ranks 535
Population 2 Sample
Sample Size 20
Sum of Ranks 285

Intermediate Calculations
Total Sample Size n 40
T1 Test Statistic 535
T1 Mean 410
Standard Error of T1 36.96846
Z Test Statistic 3.381261
Two-Tailed Test
Lower Critical Value -1.95996
Upper Critical Value 1.959964
p-value 0.000722
Reject the null hypothesis

n1 (n + 1) 20(40 + 1)
μT 1 = = = 410
2 2
n n (n + 1) 20(20 )(40 + 1)
σ T1 = 1 2 = = 36.9685
12 12
T1 − μT1
Z STAT = = 3.3813
σT 1

Decision: Since ZSTAT = 3.3813 is greater than the upper critical bounds of 1.96, reject
H0. There is enough evidence of a difference in the median surface hardness between
untreated and treated steel plates.
(b) You do not need any additional assumption except that the data from the two groups
have to be independently drawn.

Copyright © 2009 Pearson Education, Inc. Publishing as Prentice Hall.

740 Chapter 12: Chi-Square Tests and Nonparametric Tests

12.57 (c) The results of (a) conclude that there is enough evidence of a difference in the
cont. median surface hardness between untreated and treated steel plates while the results
of Problem 10.16 conclude that there is enough evidence of a difference in the mean
surface hardness between untreated and treated steel plates.

12.58 (a) H0: M1 = M2 where Populations: 1 = Wing A, 2 = Wing B

H1: M1 ≠ M2
PHStat output:
Data
Level of Significance 0.05

Population 1 Sample
Sample Size 20
Sum of Ranks 561
Population 2 Sample
Sample Size 20
Sum of Ranks 259

Intermediate Calculations
Total Sample Size n 40
T1 Test Statistic 561
T1 Mean 410
Standard Error of T1 36.96846
Z Test Statistic 4.084563

Two-Tailed Test
Lower Critical Value -1.95996
Upper Critical Value 1.959964
p-value 4.42E-05
Reject the null hypothesis
n1 (n + 1) 20(40 + 1)
μT = = = 410
1
2 2
n n (n + 1) 20(20 )(40 + 1)
σ T1 = 1 2 = = 36.9685
12 12
T1 − μT1
Z STAT = = 4.0846
σT 1

Decision: Since ZSTAT = 4.0846 is greater than the upper critical bounds of 1.96, reject
H0. There is enough evidence of a difference in the median delivery time in the two
wings of the hotel.
(b) The results of (a) conclude that there is enough evidence of a difference in the
median delivery time in the two wings of the hotel while the results of Problem 10.69
conclude that there is enough evidence of a difference in the mean delivery time in
the two wings of the hotel.

Copyright © 2009 Pearson Education, Inc. Publishing as Prentice Hall.

 741

12.59 (a) H0: M1 = M2 where Populations: 1 = Computer-assisted individual-based

2 = Team-based program
Median assembly time in seconds is the same for employees trained in a
computer-assisted, individual-based program and those trained in a team-based
program.
H1: M1 ≠ M2 Median assembly time in seconds is different for employees trained in a
computer-assisted, individual-based program and those trained in a team-based
program.
Decision rule: If ZSTAT < –1.96 or ZSTAT > 1.96, reject H0.
n1 ⋅ (n + 1) 21⋅ (43)
Test statistic: T1 = 379 μT = = = 451.5,
1
2 2
n1 ⋅n2 ⋅(n + 1) 21⋅21⋅ 43
σT = = = 39.7524
1
12 12
T1 − μT1 379 − 451.5
Z STAT = = = – 1.82
σ T1 39.7524
Decision: Since ZSTAT = – 1.82 is between the critical bounds of ± 1.96, do not reject
H0. There is not enough evidence to conclude that the median assembly time in
seconds is different for employees trained in a computer-assisted, individual-based
program and those trained in a team-based program.
(b) You must assume approximately equal variability in the two populations.
(c) Using the pooled-variance t-test allowed you to reject the null hypothesis and
conclude that mean assembly time in seconds is different for the two sets of trainees
in Problem 10.18. Using the separate-variance t-test in Problem 10.18, however, the
test statistic fell short of the critical bound and you were unable to conclude there was
any difference in the mean assembly times. In this test, the test statistic using the
Wilcoxon rank sum test with large-sample Z-approximation also fell short of the
critical bound and you failed to reject the null hypothesis here as well.

Copyright © 2009 Pearson Education, Inc. Publishing as Prentice Hall.

742 Chapter 12: Chi-Square Tests and Nonparametric Tests

12.60 (a) H0: M1 ≥ M2 where Populations: 1 = unflawed

2 = flawed
H1: M1 < M2
Decision rule: If ZSTAT < -1.645, reject H0.
Test statistic: T1 = 286
n1 ⋅ (n + 1) 18 ⋅ (59) n1 ⋅ n2 ⋅ (n + 1) 18 ⋅ 40 ⋅ 59
μT = = = 531; σ T1 = = = 59.498
1
2 2 12 12
T1 − μT1 286 − 531
Z STAT = = = -4.118
σ T1 59.498
Decision: Since ZSTAT = −4.118 is less than the lower critical bound of −1.645, reject
H0. There is enough evidence to conclude that the median crack size is less for the
unflawed sample than for the flawed sample.
(b) You must assume approximately equal variability in the two populations.
(c) Using both the pooled-variance t-test and the separate-variance t-test
allowed you to reject the null hypothesis and conclude in Problem 10.19 that the
mean crack size is less for the unflawed sample than for the flawed sample. In this
test using the Wilcoxon rank sum test with large-sample Z-approximation also
allowed you to reject the null hypothesis and conclude that the median crack size is
less for the unflawed sample than for the flawed sample.

12.61 (a) H0: M1 = M2 where Populations: 1 = Bank 1 2 = Bank 2

H1: M1 ≠ M2
PHStat Output:
Wilcoxon Rank Sum Test

Data
Level of Significance 0.05

Population 1 Sample
Sample Size 15
Sum of Ranks 153
Population 2 Sample
Sample Size 15
Sum of Ranks 312

Intermediate Calculations
Total Sample Size n 30
T1 Test Statistic 153
T1 Mean 232.5
Standard Error of T1 24.10913
Z Test Statistic -3.29751

Two-Tailed Test
Lower Critical Value -1.95996
Upper Critical Value 1.959961
p-value 0.000976
Reject the null hypothesis
(a) Decision rule: If ZSTAT < –1.96 or ZSTAT > 1.96, reject H0.
Decision: Since ZSTAT = –3.2975 is less than the lower critical bound of –1.96, reject
H0. There is enough evidence to conclude that the median waiting time between the
two branches is different.

Copyright © 2009 Pearson Education, Inc. Publishing as Prentice Hall.

 743

12.61 (b) You must assume approximately equal variability in the two populations.
cont. (c) Using both the pooled-variance t-test and the separate-variance t-test allowed you to
reject the null hypothesis and conclude in Problem 10.14 that the mean waiting time
between the two branches is different. In this test using the Wilcoxon rank sum test
with large-sample Z-approximation also allowed you to reject the null hypothesis and
conclude that the median waiting time between the two branches is different.

12.62 (a) H0: M1 = M2 where Populations: 1 = subcompact, 2 = compact

H1: M1 ≠ M2
PHStat output:

Wilcoxon Rank Sum Test

Data
Level of Significance 0.05

Population 1 Sample
Sample Size 31
Sum of Ranks 678
Population 2 Sample
Sample Size 15
Sum of Ranks 403

Intermediate Calculations
Total Sample Size n 46
T1 Test Statistic 403
T1 Mean 352.5
Standard Error of T1 42.67611
Z Test Statistic 1.183332

Two-Tail Test
Lower Critical Value -1.95996
Upper Critical Value 1.959964
p-Value 0.236678
Do not reject the null hypothesis
T1 − μT1
Z STAT = = 1.1833
σT 1

Decision: Since ZSTAT = 1.1833 is between the critical bounds of −1.96 and 1.96, do
not reject H0. There is not enough evidence to conclude that there is a difference in
the median battery life between subcompact cameras and compact cameras.
(b) You must assume approximately equal variability in the two populations.
(c) Using the pooled-variance t-test, you fail to reject the null hypothesis and conclude
that there is not enough evidence of a difference in the mean battery life between the
two types of digital cameras in Problem 10.11(a). Using the Wilcoxon rank sum test
with large-sample Z-approximation, you also fail to reject the null hypothesis and
conclude that there is not enough evidence of a difference in the median battery life
between subcompact cameras and compact cameras.

12.63 (a) WL = 13, WU = 53 (b) WL = 10, WU = 56 (c) WL = 7, WU = 59

(d) WL = 5, WU = 61

12.64 (a) WU = 53 (b) WU = 56 (c) WU = 59 (d) WU = 61

Copyright © 2009 Pearson Education, Inc. Publishing as Prentice Hall.

744 Chapter 12: Chi-Square Tests and Nonparametric Tests

12.65 (a) WL = 13 (b) WL = 10 (c) WL = 7

(d) WL = 5

12.66
Observation Di abs(Di) Sign of Di R signed R R(+)
1 3.2 3.2 + 6 6 6
2 1.7 1.7 + 2.5 2.5 2.5
3 4.5 4.5 + 7 7 7
4 0 0 Discard - - -
5 11.1 11.1 + 9 9 9
6 -0.8 0.8 - 1 -1 0
7 2.3 2.3 + 5 5 5
8 -2 2- 4 -4 0
9 0 0 Discard - - -
10 14.8 14.8 + 10 10 10
11 5.6 5.6 + 8 8 8
12 1.7 1.7 + 2.5 2.5 2.5
'
W= Σin=1 Ri( + ) = 50

12.67 n ' = 10, α = 0.05, WL = 8, WU = 47

12.68 Since W = 50 > WU = 47, reject H 0 .

W = Σin=1 Ri(
' +)
12.69 = 67.5

12.70 n ' = 12, α = 0.05, WU = 61

12.71 Since W = 67.5 > WU = 61, reject H 0 .

12.72 (a) H0: MD = 0 H1: MD ≠ 0

where Populations: 1 = A, 2 = B
n'
n = 9 , W = ∑ Ri
' (+ )
=2
i =1
Decision rule: Reject H0 if W < 3 or > 33. Since W = 2 is smaller than 3, reject H0.
There is enough evidence of a difference in the median summated rating between
brand A and Brand B.
(b) In Problem 10.22, you conclude that there is enough evidence of a difference in the
mean summated ratings between the two brands. Here, you conclude that there is
enough evidence of a difference in the median summated rating between brand A and
Brand B.

Copyright © 2009 Pearson Education, Inc. Publishing as Prentice Hall.

 745

12.73 (a) H0: MD ≤ 0 H1: MD > 0

where Populations: 1 = Whole Foods, 2 = Fairway
Minitab output:
Wilcoxon Signed Rank Test: Differences
Test of median = 0.000000 versus median > 0.000000

N for Wilcoxon Estimated

N Test Statistic P Median
Differen 10 10 51.0 0.010 0.6900

Since the p-value = 0.010 equals the level of significance, do not reject H0. There is not
sufficient evidence that the median price is higher at Whole Foods Market than at the
Fairway Supermarket.
(b) The t test for the mean difference in Problem 10.25 concludes that the mean price is
higher at Whole Foods Market than at the Fairway Supermarket, and the Wilcoxon
signed-ranks test concludes that there is not sufficient evidence that the median price is
higher at Whole Foods Market than at the Fairway Supermarket.

12.74 (a) H0: MD = 0 where Populations: 1 = In-line 2 = Analytical lab

H1: MD ≠ 0
Minitab Output:
Wilcoxon Signed Rank Test: Differences

Test of median = 0.000000 versus median not = 0.000000

N for Wilcoxon Estimated

N Test Statistic P Median
Differen 24 23 128.5 0.784 -0.005000

Since the p-value = 0.784 is greater than the 0.05 level of significance, do not reject
H 0 . There is insufficient evidence of a difference in the median measurements in-
line and from an analytical lab.
(b) Using the paired-sample t-test in Problem 10.23, you do not reject the null
hypothesis; you conclude that there is not enough evidence of a difference in the
mean measurements in-line and from an analytical lab. Using the Wilcoxon signed
rank test, you do not reject the null hypothesis; you conclude that there is not enough
evidence of a difference in the median measurements in-line and from an analytical
lab.

12.75 (a) H0: MD = 0 where Populations: 1 = Book Store 2 = Amazon

H1: MD ≠ 0
Minitab Output:
Wilcoxon Signed Rank Test: Differences

Test of median = 0.000000 versus median not = 0.000000

N for Wilcoxon Estimated

N Test Statistic P Median
Differen 15 15 54.0 0.755 -1.500

Copyright © 2009 Pearson Education, Inc. Publishing as Prentice Hall.

746 Chapter 12: Chi-Square Tests and Nonparametric Tests

12.75
cont.
Book Store Amazon Di |Di| Ri Ri(+)
120 101.22 18.78 18.78 10 10
137.5 115.33 22.17 22.17 11 11
65 37.05 27.95 27.95 12 12
82.75 71.36 11.39 11.39 6 6
90 83 7 7 2 2
39.95 26.37 13.58 13.58 7 7
40 26.4 13.6 13.6 8.5 8.5
120 108.99 11.01 11.01 5 5
25 14.99 10.01 10.01 3 3
133.75 102.3 31.45 31.45 15 15
40 26.4 13.6 13.6 8.5 8.5
79.5 68.76 10.74 10.74 4 4
154.75 126.15 28.6 28.6 13 13
30.95 31.95 -1 1 1 0
155.75 126.97 28.78 28.78 14 14
n'
'
n = 15 WL = 16 , WU = 104 W = ∑ Ri(+ ) = 119
i =1

Since W = 118 is greater than WU = 104 , reject H 0 . There is sufficient evidence of

a difference in the median price of textbooks between the local bookstore and
Amazon.com.
(b) Using the paired-sample t-test in Problem 10.24, you do not reject the null
hypothesis; you conclude that there is not enough evidence of a difference in the
mean price of textbooks between the local bookstore and Amazon.com. Using the
Wilcoxon signed rank test, you reject the null hypothesis; you conclude that there is
enough evidence of a difference in the median price of textbooks between the local
bookstore and Amazon.com.
Normal Probability Plot
35
30
25
20
i 15
D
10
5
0
-5
-2 -1 0 1 2
Z Value
The normal probability plot suggests that the distribution of the differences does not
appear to be normal. Since the paired t test in Problem 10.24 requires the differences
to be normally distributed while the Wilcoxon signed rank test does not, the
conclusion drawn from the Wilcoxon signed rank test should be more reliable.

Copyright © 2009 Pearson Education, Inc. Publishing as Prentice Hall.

 747

12.76 (a) H0: MD = 0 where Populations: 1 = Before 2 = After

H1: MD ≠ 0
Minitab Output:
Wilcoxon Signed Rank Test: Differences
Test of median = 0.000000 versus median not = 0.000000

N for Wilcoxon Estimated

N Test Statistic P Median
Differen 35 35 156.0 0.009 -5.500

Since the p-value = 0.009 is smaller than the 0.05 level of significance, reject H 0 .
There is sufficient evidence of a difference in the median performance ratings
between the two programs.
(b) Using the paired-sample t-test in Problem 10.27, you reject the null hypothesis and
conclude that there is enough evidence of a difference in the mean performance
ratings between the two programs. Using the Wilcoxon signed rank test, you reject
the null hypothesis and conclude that there is enough evidence of a difference in the
median performance ratings between the two programs.
12.77 (a) H0: MD ≥ 0 where Populations: 1 = two days 2 = seven days
H1: MD < 0
Minitab Output:
Wilcoxon Signed Rank Test: Differences

Test of median = 0.000000 versus median < 0.000000

N for Wilcoxon Estimated

N Test Statistic P Median
Differen 40 40 0.0 0.000 -0.5100

Since the p-value = 0.000 is smaller than the 0.01 level of significance, reject H 0 .
There is sufficient evidence that the median strength is less at two days than at seven
days.
(b) Using the paired-sample t-test in Problem 10.28, you reject the null hypothesis and
conclude that there is enough evidence that the mean strength is less at two days than
at seven days. Using the Wilcoxon signed rank test, you reject the null hypothesis
and conclude that there is enough evidence that the median strength is less at two
days than at seven days.
2
12.78 For the 0.01 level of significance and 5 degrees of freedom, χ U = 15.086 .

2
12.79 (a) Decision rule: If H > χ U = 15.086 , reject H0.
(b) Decision: Since Hcalc = 13.77 is less than the critical bound of 15.086, do not reject
H0.

Copyright © 2009 Pearson Education, Inc. Publishing as Prentice Hall.

748 Chapter 12: Chi-Square Tests and Nonparametric Tests

12. 80 PHStat output:

Kruskal-Wallis Rank Test for Differences in Medians

Data
Level of Significance 0.05

Intermediate Calculations
Sum of Squared Ranks/Sample Size 1960.583
Sum of Sample Sizes 18
Number of Groups 3

Test Result
H Test Statistic 11.7924
Critical Value 5.991465
p-Value 0.00275
Reject the null hypothesis
(a) H0: M1 = M2 = M3 where population 1 = front, 2 = middle, 3 = rear
H1: At least one of the medians differs.
Since the p-value = 0.00275 < 0.05, reject H0. There is enough evidence of a
difference in the median sales for the three locations.
(b) In (a), you conclude that there is enough evidence of a difference in the median sales
for the three locations, while in problem 11.13(a) you conclude that the mean sales
are different across the three store aisle locations.
(c) The normal probability plots suggest that middle aisle and rear aisle data appear to
deviate from the normal distribution. Hence, the Kruskal-Wallis rank test is more
appropriate.
Front
10
8
t 6
n
o
rF
4
2
0
-1.5 -1 -0.5 0 0.5 1 1.5
Z Value

Middle
3.5
3
2.5
e
l 2
d
d
i 1.5
M
1
0.5
0
-1.5 -1 -0.5 0 0.5 1 1.5
Z Value

Copyright © 2009 Pearson Education, Inc. Publishing as Prentice Hall.

 749

12.80 (c)
cont.
Rear
7
6
5
r4
a
e
R3
2
1
0
-1.5 -1 -0.5 0 0.5 1 1.5
Z Value

12.81 PHStat output of Kruskal-Wallis rank test:

Data
Level of Significance 0.05

Group 1
Sum of Ranks 30
Sample Size 4
Group 2
Sum of Ranks 10
Sample Size 4
Group 3
Sum of Ranks 38
Sample Size 4

Intermediate Calculations
Sum of Squared Ranks/Sample Size 611
Sum of Sample Sizes 12
Number of groups 3
H Test Statistic 8

Test Result
Critical Value 5.991465
p-Value 0.018316
Reject the null hypothesis
(a) H0: Mexperts = Mreaders = Mdarts H1: At least one of the medians differs.
Since the p-value = 0.0183 is lower than 0.05, reject H0. There is sufficient evidence
of a significant difference in the median return for the three categories.
(b) In (a), you conclude that there is evidence of a significant difference in the median
return for the three categories, while in problem 11.8 (a), you conclude that there is
evidence of a significant difference in the mean return for the three categories.
(c) The samples are randomly and independently obtained and problem 11.8(d) shows
that there is not enough evidence to conclude there is a significant difference in the
variation in the return for the three categories. However, stock returns are known to
be not normally distributed. So the assumptions of part (a) of this problem are more
appropriate.

Copyright © 2009 Pearson Education, Inc. Publishing as Prentice Hall.

750 Chapter 12: Chi-Square Tests and Nonparametric Tests

12.82 PHStat output of Kruskal-Wallis rank test:

Data
Level of Significance 0.05
Group 1
Sum of Ranks 640
Sample Size 15
Group 2
Sum of Ranks 291
Sample Size 15
Group 3
Sum of Ranks 468
Sample Size 15
Group 4
Sum of Ranks 431
Sample Size 15
Intermediate Calculations
Sum of Squared Ranks/Sample Size 59937.73
Sum of Sample Sizes 60
Number of groups 4
H Test Statistic 13.51716
Test Result
Critical Value 7.814728
p-Value 0.003642
Reject the null hypothesis
(a) H0: Mmain = MSat1 = MSat2y = MSat3 H1: At least one of the medians differs.
Since the p-value = 0.0036 is lower than 0.05, reject H0. There is sufficient evidence
of a difference in the median waiting time in the four locations.
(b) In (a), you conclude that there is evidence of a difference in the median waiting time
among the four locations, while in problem 11.9 (a), you conclude that there is
evidence of a difference in the mean waiting time among the four locations.

12.83 PHStat output:

Kruskal-Wallis Rank Test for Differences in Medians

Data
Level of Significance 0.05

Intermediate Calculations
Sum of Squared Ranks/Sample Size 5299.4
Sum of Sample Sizes 25
Number of Groups 5

Test Result
H Test Statistic 19.83508
Critical Value 9.487729
p-Value 0.000538
Reject the null hypothesis
(a) H0: M1 = M2 = M3 = M4 = M5 where population 1 = money market, 2 = six-month
CD, 3 = one-year CD, 4 = 2.5-year CD, 5 = five-year CD
H1: At least one of the medians differs.
Since the p-value = 0.0005 < 0.05, reject H0. There is enough evidence of a
difference in the median yields of the different accounts.
(b) In (a), you conclude that there is enough evidence of a difference in the median
yields of the different accounts while in problem 11.11(a) you conclude that there is
enough evidence of a difference in the mean yields of the different accounts.

Copyright © 2009 Pearson Education, Inc. Publishing as Prentice Hall.

 751

12.84 PHStat output of Kruskal-Wallis rank test:

Data
Level of Significance 0.05
Intermediate Calculations
Sum of Squared Ranks/Sample Size 8705.333
Sum of Sample Sizes 30
Number of Groups 5
Test Result
H Test Statistic 19.32688
Critical Value 9.487729
p-Value 0.000678
Reject the null hypothesis

(a) H0: MA = MB = MC = MD = ME H1: At least one of the medians differs.

Since the p-value = 0.0007 < 0.05, reject H0. There is sufficient evidence of a
difference in the median rating of the five advertisements.
(b) In (a), you conclude that there is evidence of a difference in the median rating of the
five advertisements, while in problem 11.12 (a), you conclude that there is evidence
of a difference in the mean rating of the five advertisements.
(c) Since the combined scores are not true continuous variables, the nonparametric
Kruskal-Wallis rank test is more appropriate because it does not require the scores to
be normally distributed.

12.85 PHStat output of Kruskal-Wallis rank test:

Data
Level of Significance 0.05
Group 1
Sum of Ranks 55
Sample Size 10
Group 2
Sum of Ranks 174
Sample Size 10
Group 3
Sum of Ranks 277
Sample Size 10
Group 4
Sum of Ranks 314
Sample Size 10
Intermediate Calculations
Sum of Squared Ranks/Sample Size 20862.6
Sum of Sample Sizes 40
Number of groups 4
H Test Statistic 29.65317
Test Result
Critical Value 7.814728
p-Value 1.63E-06
Reject the null hypothesis
(a) H0: M1 = M2 = M3 = M4 H1: At least one of the medians differs.
Since the p-value is virtually zero, reject H0. There is sufficient evidence of a
difference in the median distance traveled by the golf balls with different designs.

Copyright © 2009 Pearson Education, Inc. Publishing as Prentice Hall.

752 Chapter 12: Chi-Square Tests and Nonparametric Tests

12.85 (b) In (a), you conclude that there is evidence of a difference in the median distance
cont. traveled by the golf balls with different designs, while in problem 11.14 (a), you
conclude that there is evidence of a difference in the mean distance traveled by the
golf balls with different designs.
Since the combined scores are not true continuous variables, the nonparametric
Kruskal-Wallis rank test is more appropriate since it does not require the scores to be
normally distributed.

12.86 PHStat output:

Kruskal-Wallis Rank Test
Data
Level of Significance 0.05
Group 1
Sum of Ranks 254
Sample Size 10
Group 2
Sum of Ranks 270
Sample Size 10
Group 3
Sum of Ranks 241
Sample Size 10
Group 4
Sum of Ranks 55
Sample Size 10
Intermediate Calculations
Sum of Squared Ranks/Sample Size 19852.2
Sum of Sample Sizes 40
Number of groups 4
H Test Statistic 22.26
Test Result
Critical Value 7.814725
p-Value 5.76E-05
Reject the null hypothesis
(a) H0: MKroger = MGlad = MHefty = MTuffstuff H1: At least one of the medians differs.
Since the p-value is virtually zero, reject H0. There is sufficient evidence of a
difference in the median strength of the four brands of trash bags.
(b) In (a), you conclude that there is evidence of a difference in the median strength of
the four brands of trash bags, while in problem 11.10, you conclude that there is
evidence of a difference in the mean strength of the four brands.

12.87 d.f. = 5, α = 0.1, χU2 = 9.2363

12.88 (a) H0: M1 = M2 = M3 = M4 = M5 = M6 H1: At least one of the medians differs.

Reject H0 if FR > 9.2363.
(b) Since FR = 11.56 > 9.2363, reject H0. There is enough evidence that the medians are
different.

Copyright © 2009 Pearson Education, Inc. Publishing as Prentice Hall.

 753

12.89 Minitab output:

Friedman Test: Rating versus Brand, Expert

Friedman test for Rating by Brand blocked by Expert

S = 20.03 DF = 3 P = 0.000
S = 20.72 DF = 3 P = 0.000 (adjusted for ties)

Est Sum of
Brand N Median Ranks
A 9 25.000 25.0
B 9 26.750 34.5
C 9 24.000 20.0
D 9 22.250 10.5

Grand median = 24.500

(a) H0 : M A = M B = MC = M D H1 : Not all medians are the equal.
Since the p-value is virtually zero, reject H0 at 0.05 level of significance. There is
evidence of a difference in the median summated ratings of the four brands of
Colombian coffee.
(b) In (a), you conclude that there is evidence of a difference in the median summated
ratings of the four brands of Colombian coffee while in problem 11.23, you conclude
that there is evidence of a difference in the mean summated ratings of the four brands
of Colombian coffee.

12.90 (a) H0: M 1 = M 2 = M 3 = M 4 = M 5 H1 : Not all medians are equal.

Rank
Itunes Wal-Mart MusicNow Musicmatch Napster
3 1 4 5 2
4.5 1 4.5 2 3
2.5 5 4 2.5 1
4 1 2 3 5
1.5 4 3 1.5 5
Rj 15.5 12 17.5 14 16
(Rj)^2 240.25 144 306.25 196 256
12 c
12
Test statistic: FR = ∑ R.2j − 3r (c + 1) = (1142.5) − 3(5)(6) = 1.4
rc(c + 1) j =1 (5)(5)(6)
Upper critical value: χ α2 = 9.4877 p-value = 0.8442
Since the p-value = 0.8442 > 0.05 and FR = 1.4 < 9.4877, do not reject H0. There is
insufficient evidence of a difference in the median prices at the digital music services.
(b) In (a), you conclude that there is insufficient evidence of a difference in the median
prices at the digital music services while in problem 11.26(a), you conclude that there
is not evidence of a difference in the mean prices for albums at the five digital music
services.

Copyright © 2009 Pearson Education, Inc. Publishing as Prentice Hall.

754 Chapter 12: Chi-Square Tests and Nonparametric Tests

12.91 Minitab output:

Friedman Test: price versus kitchen, items
Friedman test for price by kitchen blocked by items

S = 10.65 DF = 3 P = 0.014
S = 11.09 DF = 3 P = 0.011 (adjusted for ties)

Est Sum of
kitchen N Median Ranks
Fairway 10 1.5763 20.5
Gristede’s 10 1.5288 21.0
Stop&Sho 10 1.7138 22.0
Whole Fo 10 2.2363 36.5

Grand median = 1.7638

(a) H 0 : M1 = M 2 = M 3 H1 : Not all medians are equal.
Since the p-value is smaller than 0.05, reject H0 at 0.05 level of significance. There is
evidence of a difference in the median prices for these kitchen staples at the four
stores.
(b) In (a), you conclude that there is evidence of a difference in the median prices for
these kitchen staples at the four stores, and in problem 11.25, you conclude that there
is evidence of a difference in the mean prices for these kitchen staples at the four
stores.

12.92 Minitab output:

Friedman Test: Thickness versus Position, Batch1

Friedman test for Thickness by Position blocked by Batch1

S = 97.97 DF = 4 P = 0.000
S = 99.63 DF = 4 P = 0.000 (adjusted for ties)

Est Sum of
Position N Median Ranks
1 30 240.45 32.0
2 30 242.55 64.0
18 30 245.25 97.5
19 30 249.15 141.0
28 30 246.85 115.5

Grand median = 244.85

(a) H 0 : M 1 = M 2 = M 18 = M 19 = M 28 H1 : Not all medians are equal.
Since the p-value is virtually zero, reject H0 at 0.05 level of significance. There is
evidence of a difference in the median thickness of the wafers for the five positions.
(b) In (a), you conclude that there is evidence of a difference in the median thickness of
the wafers for the five positions, and in problem 11.27, you conclude that there is
evidence of a difference in the mean thickness of the wafers for the five positions.

Copyright © 2009 Pearson Education, Inc. Publishing as Prentice Hall.

 755

12.93 Minitab output:

Friedman Test: Strength versus Days, Samples

Friedman test for Strength by Days blocked by Samples

S = 80.00 DF = 2 P = 0.000

Est Sum of
Days N Median Ranks
2 40 3.0863 40.0
7 40 3.5888 80.0
28 40 4.5838 120.0

Grand median = 3.7529

(a) H 0 : M 2 = M 7 = M 28 H1 : Not all medians are equal.

Since the p-value is virtually zero, reject H0 at 0.05 level of significance. There is
evidence of a difference in the median compressive strength after 2, 7 and 28 days.
(b) In (a), you conclude that there is evidence of a difference in the median compressive
strength after 2, 7 and 28 days, and in problem 11.28, you conclude that there is
evidence of a difference in the mean compressive strength after 2, 7 and 28 days.

12.94 The Chi-square test for the difference between two proportions can be used only when the
alternative hypothesis is two-tailed.

12.95 The Chi-square test can be used for c populations as long as all expected frequencies are at
least one.

12.96 The Chi-square test for independence can be used as long as all expected frequencies are at
least one.

12.97 The McNemar test should be used when determining whether there is evidence of a difference
between the proportions of two related samples.

12.98 The Wilcoxon rank sum test should be used when you are unable to assume that each of two
independent populations are normally distributed.

12.99 The Wilcoxon signed ranks test should be use when comparing the median difference between
matched items or repeated measurements of the same item.

12.100 The Kruskal-Wallis test should be used if you cannot assume that the populations are
normally distributed.

12.101 The Friedman rank test should be used to test for the difference in c medians in a randomized
block design when data collected are only in rank form within each block or when normality
cannot be assumed.

Copyright © 2009 Pearson Education, Inc. Publishing as Prentice Hall.

756 Chapter 12: Chi-Square Tests and Nonparametric Tests

12.102 (a) H0: There is no relationship between a student's gender and his/her pizzeria selection.
H1: There is a relationship between a student's gender and his/her pizzeria selection.
Decision rule: d.f. = 1. If χ STAT
2
> 3.841, reject H0. Test statistic: χ STAT
2
= 0.412
Decision: Since the χ STAT
2
= 0.412 is smaller than the critical bound of 3.841, do not
reject H0. There is not enough evidence to conclude that there is a relationship
between a student's gender and his/her pizzeria selection.
(b) Test statistic: χ STAT
2
= 2.624
Decision: Since the χ STAT
2
= 2.624 is less than the critical bound of 3.841, do not
reject H0. There is not enough evidence to conclude that there is a relationship
between a student's gender and his/her pizzeria selection.
(c) H0: There is no relationship between price and pizzeria selection.
H1: There is a relationship between price and pizzeria selection.
Decision rule: d.f. = 2. If χ STAT
2
> 5.991, reject H0. Test statistic: χ STAT
2
= 4.956
Decision: Since the χ STAT
2
= 4.956 is smaller than the critical bound of 5.991, do not
reject H0. There is not enough evidence to conclude that there is a relationship
between price and pizzeria selection.
(d) p-value = 0.0839. The probability of obtaining a sample that gives a test statistic
equal to or greater than 4.956 is 8.39% if the null hypothesis of no relationship
between price and pizzeria selection is true.
(e) Since there is no evidence that price and pizzeria selection are related, it is
inappropriate to determine which prices are different in terms of pizzeria preference.

Copyright © 2009 Pearson Education, Inc. Publishing as Prentice Hall.

 757

12.103 Customer product ratings/reviews

(a) Ho: H1 Not all are equal.
where population 1 = older adults, 2 = younger adults, 3 = youth
Chi-Square Test

Observed Frequencies
Column variable
Row variable Older Adults Younger Adults Youth Total
Use 71 78 81 230
Do Not Use 29 22 19 70
Total 100 100 100 300

Expected Frequencies
Column variable
Row variable Older Adults Younger Adults Youth Total
Use 76.66667 76.66667 76.66667 230
Do Not Use 23.33333 23.33333 23.33333 70
Total 100 100 100 300

Data
Level of Significance 0.05
Number of Rows 2
Number of Columns 3
Degrees of Freedom 2

Results
Critical Value 5.991465
Chi-Square Test Statistic 2.944099
p-Value 0.229455
Do not reject the null hypothesis

Expected frequency assumption

is met.

( f0 – fe )2
Test statistic: χ 2
STAT = ∑ = 2.9441
All Cells fe
Decision: Since the calculated test statistic 2.9441 is smaller than the critical value of
5.9915, you do not reject H 0 and conclude that there is not enough evidence of a
significant difference among the age groups in the proportion who use customer product
ratings/reviews.
(b) Since you do not reject H 0 , it is inappropriate to use the Marascuilo procedure to (a).

Copyright © 2009 Pearson Education, Inc. Publishing as Prentice Hall.

758 Chapter 12: Chi-Square Tests and Nonparametric Tests

12.103 For sale listings with seller ratings

cont.
(a) Ho: H1 Not all are equal.
where population 1 = older adults, 2 = younger adults, 3 = youth
Chi-Square Test

Observed Frequencies
Column variable
Row variable Older Adults Younger Adults Youth Total
Use 69 72 77 218
Do Not Use 31 28 23 82
Total 100 100 100 300

Expected Frequencies
Column variable
Row variable Older Adults Younger Adults Youth Total
Use 72.66667 72.66667 72.66667 218
Do Not Use 27.33333 27.33333 27.33333 82
Total 100 100 100 300

Data
Level of Significance 0.05
Number of Rows 2
Number of Columns 3
Degrees of Freedom 2

Results
Critical Value 5.991465
Chi-Square Test Statistic 1.644663
p-Value 0.439406
Do not reject the null hypothesis

Expected frequency assumption

is met.

( f0 – fe )2
Test statistic: χ 2
STAT = ∑ = 1.6447
All Cells fe
Decision: Since the calculated test statistic 1.6447 is smaller than the critical value of
5.9915, you do not reject H 0 and conclude that there is not enough evidence of a
significant difference among the age groups in the proportion who use for sale listings
with seller ratings.
(b) Since you do not reject H 0 , it is inappropriate to use the Marascuilo procedure to (a).

Copyright © 2009 Pearson Education, Inc. Publishing as Prentice Hall.

 759

12.103 For sale listings without seller ratings

cont. (a) Ho: H1 Not all are equal.
where population 1 = older adults, 2 = younger adults, 3 = youth
Chi-Square Test

Observed Frequencies
Column variable
Row variable Older Adults Younger Adults Youth Total
Use 58 60 65 183
Do Not Use 42 40 35 117
Total 100 100 100 300

Expected Frequencies
Column variable
Row variable Older Adults Younger Adults Youth Total
Use 61 61 61 183
Do Not Use 39 39 39 117
Total 100 100 100 300

Data
Level of Significance 0.05
Number of Rows 2
Number of Columns 3
Degrees of Freedom 2

Results
Critical Value 5.991465
Chi-Square Test Statistic 1.092896
p-Value 0.579003
Do not reject the null hypothesis

Expected frequency assumption

is met.

( f0 – fe )2
Test statistic: χ STAT
2
= ∑
All Cells fe
= 1.0929

Decision: Since the calculated test statistic 1.0929 is smaller than the critical value of
5.9915, you do not reject H 0 and conclude that there is not enough evidence of a
significant difference among the age groups in the proportion who use for sale listings
without seller ratings.
(b) Since you do not reject H 0 , it is inappropriate to use the Marascuilo procedure to (a).

Copyright © 2009 Pearson Education, Inc. Publishing as Prentice Hall.

760 Chapter 12: Chi-Square Tests and Nonparametric Tests

12.103 Online classified ads

cont. (a) Ho: H1 Not all are equal.
where population 1 = older adults, 2 = younger adults, 3 = youth
Chi-Square Test

Observed Frequencies
Column variable
Row variable Older Adults Younger Adults Youth Total
Use 57 59 66 182
Do Not Use 43 41 34 118
Total 100 100 100 300

Expected Frequencies
Column variable
Row variable Older Adults Younger Adults Youth Total
Use 60.66667 60.66667 60.66667 182
Do Not Use 39.33333 39.33333 39.33333 118
Total 100 100 100 300

Data
Level of Significance 0.05
Number of Rows 2
Number of Columns 3
Degrees of Freedom 2

Results
Critical Value 5.991465
Chi-Square Test Statistic 1.871857
p-Value 0.392222
Do not reject the null hypothesis

Expected frequency assumption

is met.

( f0 – fe )2
Test statistic: χ STAT
2
= ∑
All Cells fe
= 1.8719

Decision: Since the calculated test statistic 1.8719 is smaller than the critical value of
5.9915, you do not reject H 0 and conclude that there is not enough evidence of a
significant difference among the age groups in the proportion who use online classified
ads.
(b) Since you do not reject H 0 , it is inappropriate to use the Marascuilo procedure to (a).

Copyright © 2009 Pearson Education, Inc. Publishing as Prentice Hall.

 761

12.103 Message-board posts

cont. (a) Ho: H1 Not all are equal.
where population 1 = older adults, 2 = younger adults, 3 = youth
Chi-Square Test

Observed Frequencies
Column variable
Row variable Older Adults Younger Adults Youth Total
Use 57 65 71 193
Do Not Use 43 35 29 107
Total 100 100 100 300

Expected Frequencies
Column variable
Row variable Older Adults Younger Adults Youth Total
Use 64.33333 64.33333 64.33333 193
Do Not Use 35.66667 35.66667 35.66667 107
Total 100 100 100 300

Data
Level of Significance 0.05
Number of Rows 2
Number of Columns 3
Degrees of Freedom 2

Results
Critical Value 5.991465
Chi-Square Test Statistic 4.300034
p-Value 0.116482
Do not reject the null hypothesis

Expected frequency assumption

is met.

( f0 – fe )2
Test statistic: χ STAT
2
= ∑
All Cells fe
= 4.3000

Decision: Since the calculated test statistic 4.3000 is smaller than the critical value of
5.9915, you do not reject H 0 and conclude that there is not enough evidence of a
significant difference among the age groups in the proportion who use message-board
posts.
(b) Since you do not reject H 0 , it is inappropriate to use the Marascuilo procedure to (a).

Copyright © 2009 Pearson Education, Inc. Publishing as Prentice Hall.

762 Chapter 12: Chi-Square Tests and Nonparametric Tests

12.103 Webblogs
cont. (a) Ho: H1 Not all are equal.
where population 1 = older adults, 2 = younger adults, 3 = youth
Chi-Square Test

Observed Frequencies
Column variable
Row variable Older Adults Younger Adults Youth Total
Use 55 61 67 183
Do Not Use 45 39 33 117
Total 100 100 100 300

Expected Frequencies
Column variable
Row variable Older Adults Younger Adults Youth Total
Use 61 61 61 183
Do Not Use 39 39 39 117
Total 100 100 100 300

Data
Level of Significance 0.05
Number of Rows 2
Number of Columns 3
Degrees of Freedom 2

Results
Critical Value 5.991465
Chi-Square Test Statistic 3.026482
p-Value 0.220195
Do not reject the null hypothesis

Expected frequency assumption

is met.

( f0 – fe )2
Test statistic: χ STAT
2
= ∑
All Cells fe
= 3.0265

Decision: Since the calculated test statistic 3.0265 is smaller than the critical value of
5.9915, you do not reject H 0 and conclude that there is not enough evidence of a
significant difference among the age groups in the proportion who use webblogs.
(b) Since you do not reject H 0 , it is inappropriate to use the Marascuilo procedure to (a).

Copyright © 2009 Pearson Education, Inc. Publishing as Prentice Hall.

 763

12.103 Dating site profiles/personals

cont. (a) Ho: H1 Not all are equal.
where population 1 = older adults, 2 = younger adults, 3 = youth
Chi-Square Test

Observed Frequencies
Column variable
Row variable Older Adults Younger Adults Youth Total
Use 49 52 59 160
Do Not Use 51 48 41 140
Total 100 100 100 300

Expected Frequencies
Column variable
Row variable Older Adults Younger Adults Youth Total
Use 53.33333 53.33333 53.33333 160
Do Not Use 46.66667 46.66667 46.66667 140
Total 100 100 100 300

Data
Level of Significance 0.05
Number of Rows 2
Number of Columns 3
Degrees of Freedom 2

Results
Critical Value 5.991465
Chi-Square Test Statistic 2.116071
p-Value 0.347137
Do not reject the null hypothesis

Expected frequency assumption

is met.

( f0 – fe )2
Test statistic: χ STAT
2
= ∑
All Cells fe
= 2.1161

Decision: Since the calculated test statistic 2.1161 is smaller than the critical value of
5.9915, you do not reject H 0 and conclude that there is not enough evidence of a
significant difference among the age groups in the proportion who use dating site
profiles/personals.
(b) Since you do not reject H 0 , it is inappropriate to use the Marascuilo procedure to (a).

Copyright © 2009 Pearson Education, Inc. Publishing as Prentice Hall.

764 Chapter 12: Chi-Square Tests and Nonparametric Tests

12.103 Peer-generated and peer-reference information

cont. (a) Ho: H1 Not all are equal.
where population 1 = older adults, 2 = younger adults, 3 = youth
Chi-Square Test
Observed Frequencies
Column variable
Row variable Older Adults Younger Adults Youth Total
Use 49 55 68 172
Do Not Use 51 45 32 128
Total 100 100 100 300

Expected Frequencies
Column variable
Row variable Older Adults Younger Adults Youth Total
Use 57.33333 57.33333 57.33333 172
Do Not Use 42.66667 42.66667 42.66667 128
Total 100 100 100 300
Data
Level of Significance 0.05
Number of Rows 2
Number of Columns 3
Degrees of Freedom 2
Results
Critical Value 5.991465
Chi-Square Test Statistic 7.712573
p-Value 0.021146
Reject the null hypothesis
Expected frequency assumption
is met.
( f0 – fe )2
Test statistic: χ STAT
2
= ∑
All Cells fe
= 7.7126

Decision: Since the calculated test statistic 7.7126 is larger than the critical value of
5.9915, you reject H 0 and conclude that there is enough evidence of a significant
difference among the age groups in the proportion who use peer-generated and peer-
reference information.
(b) PHstat output for the Marascuilo procedure:
Marascuilo Procedure
Level of Significance 0.05
Square Root of Critical Value 2.447746831
Sample Proportions
Group 1 0.49
Group 2 0.55
Group 3 0.68
MARASCUILO TABLE
Proportions Absolute Differences Critical Range
| Group 1 - Group 2 | 0.06 0.172631239 Not significant
| Group 1 - Group 3 | 0.19 0.167362172 Significant
| Group 2 - Group 3 | 0.13 0.166932027 Not significant
There is a significant difference between older adults and youth in the proportion who
use peer-generated and peer-reference information at a 5% level of significance.

Copyright © 2009 Pearson Education, Inc. Publishing as Prentice Hall.

 765

12.103 Peer-posted event listings

cont. (a) Ho: H1 Not all are equal.
where population 1 = older adults, 2 = younger adults, 3 = youth
Chi-Square Test
Observed Frequencies
Column variable
Row variable Older Adults Younger Adults Youth Total
Use 46 53 71 170
Do Not Use 54 47 29 130
Total 100 100 100 300

Expected Frequencies
Column variable
Row variable Older Adults Younger Adults Youth Total
Use 56.66667 56.66667 56.66667 170
Do Not Use 43.33333 43.33333 43.33333 130
Total 100 100 100 300
Data
Level of Significance 0.05
Number of Rows 2
Number of Columns 3
Degrees of Freedom 2
Results
Critical Value 5.991465
Chi-Square Test Statistic 13.54751
p-Value 0.001143
Reject the null hypothesis
Expected frequency assumption is met.
( f0 – fe )2
Test statistic: χ 2
STAT = ∑ = 13.5475
All Cells fe
Decision: Since the calculated test statistic 13.5475 is larger than the critical value of
5.9915, you reject H 0 and conclude that there is enough evidence of a significant
difference among the age groups in the proportion who use peer-posted event listings.
(b) PHstat output for the Marascuilo procedure:
Marascuilo Procedure
Level of Significance 0.05
Square Root of Critical Value 2.447746831
Sample Proportions
Group 1 0.46
Group 2 0.53
Group 3 0.71
MARASCUILO TABLE
Proportions Absolute Differences Critical Range
| Group 1 - Group 2 | 0.07 0.172648591 Not significant
| Group 1 - Group 3 | 0.25 0.164982494 Significant
| Group 2 - Group 3 | 0.18 0.165109551 Significant
There is a significant difference between older adults and youth, and younger adults and
youth in the proportion who use peer-generated and peer-reference information at a 5%
level of significance. There is no significant difference between older adults and
younger adults.

Copyright © 2009 Pearson Education, Inc. Publishing as Prentice Hall.

766 Chapter 12: Chi-Square Tests and Nonparametric Tests

12.104 (a) H0: There is no relationship between the attitudes of employees toward the use of
self-managed work teams and employee job classification.
H1: There is a relationship between the attitudes of employees toward the use of self-
managed work teams and employee job classification.
Decision rule: If χ STAT
2
> 12.592, reject H0.
Test statistic: χ STAT
2
= 11.895
Decision: Since χ STAT
2
= 11.895 is less than the critical bound 12.592, do not
reject H0. There is not enough evidence to conclude that there is a relationship
between the attitudes of employees toward the use of self-managed work teams and
employee job classification.
(b) H0: There is no relationship between the attitudes of employees toward vacation time
without pay and employee job classification.
H1: There is a relationship between the attitudes of employees toward vacation time
without pay and employee job classification.
Decision rule: If χ STAT
2
> 12.592, reject H0.
Test statistic: χ STAT
2
= 3.294
Decision: Since χ STAT
2
= 3.294 is less than the critical bound 12.592, do not
reject H0. There is not enough evidence to conclude that there is a relationship
between the attitudes of employees toward vacation time without pay and employee
job classification.
12.105 (a) H0: π 1 = π 2 H1: π 1 ≠ π 2 where Populations: 1 = sample tape, 2 = full tape
Decision rule: If χ STAT
2
> 3.841, reject H0. Test statistic: χ STAT
2
= 4.267
Decision: Since χ STAT
2
= 4.267 is greater than the critical bound 3.841, reject H0. There
is enough evidence to conclude that the proportions of tapes purchased is different
depending on whether the prospective customer received a sample tape or the full
educational program.
(b) The sales representative should consider the option of distributing a full tape to the
prospective customers. A program of on-going monitoring of purchase action should
be maintained since the possibility exists that the full tape could be copied and
distributed without purchase.
(c) H0: π 1 = π 2 = π 3 H1: Not all π j are the same.
where population 1 = videotape, 2 = personal sales call, 3 = telephone

Decision rule: If χ STAT

2
> 5.991, reject H0. Test statistic: χ STAT
2
= 5.375
Decision: Since χ STAT
2
= 5.375 is less than the critical bound 5.991, do not reject H0.
There is not enough evidence to conclude that there is a difference in the proportion
of tapes purchased on the basis of the sales strategy used.
(d) Since the null hypothesis was not rejected in part (c) above, you do not expect to find
any significant differences across the three proportions, and it is not appropriate to
use the Marascuilo procedure.
(e) No specific recommendation for sales strategy can be made based on the analysis of
these data.

Copyright © 2009 Pearson Education, Inc. Publishing as Prentice Hall.

 767

12.106 (a) H 0 : π1 = π 2 H1 : π 1 ≠ π 2
where 1 = before, 2 = after
Decision rule: If Z < – 1.645 or Z > 1.645, reject H0.
B −C 6 − 14
Test statistic: Z = = = – 1.7889
B+C 6 + 14
Decision: Since Z = – 1.7889 is smaller than the lower critical bound of – 1.645, reject
H0. There is enough evidence of a difference in the proportion of respondents who prefer
Coca-Cola before and after viewing the ads.
(b) p-value = 0.0736. The probability of obtaining a data set which gives rise to a test
statistic equal to or more extreme than – 1.7889 is 7.36% if there is not a difference in
the proportion of respondents who prefer Coca-Cola before and after viewing the ads.
(c) The numbers in the second table are obtained from the row and column totals of the first
table.
(d) H 0 : π1 = π 2 H1 : π 1 ≠ π 2 where 1 = before, 2 = after
( f0 – fe )2
Test statistic: χ STAT
2
= ∑
All Cells fe
=0.6528

Decision: Since the calculated test statistic 0.6528 is smaller than the critical value of
3.8415, you do not reject H 0 ; you conclude that there is not a significant difference in
preference for Coca-Cola before and after viewing the ads.
(e) p-value = 0.4191. The probability of obtaining a data set which gives rise to a test
statistic larger than 0.6528 is 41.91% if there is not a significant difference in preference
for Coca-Cola before and after viewing the ads.
(f) The McNemar test performed using the information in the first table takes into
consideration the fact that the same set of respondents are surveyed before and after
viewing the ads while the chi-square test performed using the information in the
second table ignores this fact. The McNemar test should be used because of the
related samples (before-after comparison).

12.107 (a) H 0 : π1 = π 2 H1 : π 1 ≠ π 2where 1 = before, 2 = after

Decision rule: If ZSTAT < – 1.96 or ZSTAT > 1.96, reject H0.
B −C 3 − 11
Test statistic: Z STAT = = = – 2.1381
B+C 3 + 11
Decision: Since ZSTAT = – 2.1381 is smaller than the lower critical bound of – 1.96, reject
H0. There is enough evidence of a difference in the proportion of respondents who prefer
Toyota before and after viewing the ads.
(b) p-value = 0.0325. The probability of obtaining a data set which gives rise to a test
statistic that differs from 0 by – 2.1381 or more in either direction is 3.25% if there is
not a difference in the proportion of respondents who prefer Toyota before and after
viewing the ads.
(c) The numbers in the second table are obtained from the row and column totals of the first
table.

Copyright © 2009 Pearson Education, Inc. Publishing as Prentice Hall.

768 Chapter 12: Chi-Square Tests and Nonparametric Tests

12.107 (d) H 0 : π1 = π 2 H1 : π 1 ≠ π 2 where 1 = before, 2 = after

( f0 – fe )2
cont. Test statistic: χ STAT
2
= ∑
All Cells fe
=0.6410

Decision: Since the calculated test statistic 0.6410 is less than the critical value of
3.8415, you do not reject H 0 ; you conclude that there is not a significant difference in
preference for Toyota before and after viewing the ads.
(e) p-value = 0.4233. The probability of obtaining a data set which gives rise to a test
statistic larger than 0.6410 is 42.33% if there is not a significant difference in preference
for Toyota before and after viewing the ads.
(f) The McNemar test performed using the information in the first table takes into
consideration the fact that the same set of prospective purchasers of new cars are
surveyed before and after viewing the ads while the chi-square test performed using the
information in the second table ignores this fact. The McNemar test should be used
because of the related samples (before-after comparison).

12.108 (a) H0: There is no relationship between the presence of environmental goals and the
type of manufacturing process.
H1: There is a relationship between the presence of environmental goals and the type
of manufacturing process.
( f0 – fe )2
Test statistic: χ STAT
2
= ∑
All Cells fe
=11.635

Decision rule: If χ STAT

2
> 7.815, reject H0.
Decision: Since χ STAT
2
= 11.635 is greater than the critical bound 7.815, reject H0. There
is enough evidence to conclude that there is a relationship between the presence of
environmental goals and the type of manufacturing process.
(b) The p-value is 0.00874. The probability of obtaining a data set which gives rise to a test
statistic of 11.635 or more is 0.00874 if there is no relationship between the presence of
environmental goals and the type of manufacturing process.
(c) H0: π 1 = π 2 H1: π 1 ≠ π 2
where Populations: 1 = with cost-cutting goal, 2 = without cost cutting goal
( f0 – fe )2
Test statistic: χ STAT
2
= ∑
All Cells fe
=10.94

Decision rule: If χ STAT

2
> 3.841, reject H0.
Decision: Since χ STAT
2
= 10.94 is greater than the critical bound 3.841, reject H0. There
is enough evidence to conclude that there is a difference in improved environmental
performance for teams with a specified goal of cutting costs.
(d) The p-value is 0.000941. The probability of obtaining a data set which gives rise to a
test statistic of 10.94 or more is 0.000941 if there is no difference in improved
environmental performance for teams with a specified goal of cutting costs.

Copyright © 2009 Pearson Education, Inc. Publishing as Prentice Hall.

 769

12.108 (e) H0: π 1 = π 2 H1: π 1 ≠ π 2

cont. where Populations: 1 = with cost-cutting goal, 2 = without cost cutting goal
( f0 – fe )2
Test statistic: χ STAT
2
= ∑
All Cells fe
=0.612

Decision rule: If χ STAT

2
> 3.841, reject H0.
Decision: Since χ STAT
2
= 0.612 is less than the critical bound 3.841, do not reject H0.
There is not enough evidence to conclude that there is a difference in improved
profitability for teams with a specified goal of cutting costs.
(f) The p-value is 0.4341. The probability of obtaining a data set which gives rise to a test
statistic of 0.612 or more is 0.4341 if there is no difference in improved profitability for
teams with a specified goal of cutting costs.
(g) H0: π 1 = π 2 H1: π 1 ≠ π 2
where Populations: 1 = with cost-cutting goal, 2 = without cost cutting goal
( f0 – fe )2
Test statistic: χ 2
STAT = ∑ =3.454
All Cells fe
Decision rule: If χ STAT
2
> 3.841, reject H0.
Decision: Since χ STAT
2
= 3.454 is less than the critical bound 3.841, do not reject H0.
There is not enough evidence to conclude that there is a difference in improved morale
for teams with a specified goal of cutting costs.
(h) The p-value is 0.063. The probability of obtaining a data set which gives rise to a test
statistic of 3.454 or more is 0.063 if there is no difference in improved morale for teams
with a specified goal of cutting costs.

12.109 (a)
Fees Growth Value Grand Total
No 266 234 500
Yes 198 170 368
Grand Total 464 404 868
(b) H0: There is no relationship between the objective of a mutual fund and whether or
not there is a sales charge.
H1: There is a relationship between the objective of a mutual fund and whether or not
there is a sales charge.
Decision rule: If χ STAT
2
> 3.8415, reject H0.
( f0 – fe )2
Test statistic: χ STAT
2
= ∑
All Cells fe
= 0.0311

Decision: Since χ STAT

2
= 0.0311 is less than the critical bound of 3.8415, do not reject
H0. There is not enough evidence to conclude there is a relationship between the
objective of a mutual fund and whether or not there is a sales charge.

Copyright © 2009 Pearson Education, Inc. Publishing as Prentice Hall.

770 Chapter 12: Chi-Square Tests and Nonparametric Tests

12.110 (a)
Fees Average High Low Grand Total
No 180 201 119 500
Yes 131 154 83 368
Grand Total 311 355 202 868
(b) H0: There is no relationship between the perceived risk of a mutual fund and whether
or not there is a sales charge.
H1: There is a relationship between the perceived risk of a mutual fund and whether
or not there is a sales charge.
Decision rule: If χ STAT
2
> 5.9915, reject H0.
( f0 – fe )2
Test statistic: χ STAT
2
= ∑
All Cells fe
= 0.2916

Decision: Since χ STAT

2
= 0.2916 is less than the critical bound of 5.9915, do not reject
H0. There is not enough evidence to conclude there is a relationship between the
perceived risk of a mutual fund and whether or not there is a sales charge.

12.111 (a)
Risk Growth Value Grand Total
Average 140 171 311
High 302 53 355
Low 22 180 202
Grand Total 464 404 868
(b) H0: There is no relationship between the perceived risk of a mutual fund and its
objective.
H1: There is a relationship between the perceived risk of a mutual fund and its
objective.
Decision rule: If χ STAT
2
> 5.9915, reject H0.
( f 0 − f e )2
Test statistic: χ 2
STAT = ∑
All Cells fe
= 298.6042

Decision: Since χ STAT

2
= 298.6042 is greater than the critical bound of 5.9915, reject H0.
There is enough evidence to conclude there is a relationship between the perceived risk
of a mutual fund and its objective.

12.112 (a)
Risk Large Cap Mid Cap Small Cap Grand Total
Average 174 55 82 311
High 127 86 142 355
Low 149 33 20 202
Grand Total 450 174 244 868

(b) H0: There is no relationship between the perceived risk of a mutual fund and its
category.
H1: There is a relationship between the perceived risk of a mutual fund and its
category.

Copyright © 2009 Pearson Education, Inc. Publishing as Prentice Hall.

 771

12.112 Decision rule: If χ STAT

2
> 9.4877, reject H0.
( f 0 − f e )2
cont. Test statistic: χ STAT
2
= ∑
All Cells fe
= 84.7719

Decision: Since χ STAT

2
= 84.7719 is greater than the critical bound of 9.4877, reject H0.
There is not enough evidence to conclude there is a relationship between the perceived
risk of a mutual fund and its category.

12.113
Major
Gender a ef ib is m mr o un Total
f 4 3 3 2 2 8 2 24
m 7 6 2 7 2 2 26
Grand Total 11 9 3 4 9 10 2 2 50
The expected frequency assumption for the χ 2 test is violated. The test results below might
not be reliable.
H0: There is no relationship between major and gender.
H1: There is a relationship between major and gender.
Since χ STAT
2
= 15.1402 is greater than the critical bound of 14.0671, reject H0. There is enough
evidence to conclude there is a relationship between major and gender.

Graduate School
Gender n un y Total
f 7 10 7 24
m 8 7 11 26
Total 15 17 18 50
H0: There is no relationship between plan to go to graduate school and gender.
H1: There is a relationship between plan to go to graduate school and gender.
Since χ STAT
2
= 1.4072 is smaller than the critical bound of 5.9915, do not reject H0. There is not
enough evidence to conclude there is a relationship between plan to go to graduate school
and gender.

Employment Status
Gender ft pt un Total
f 1 18 5 24
m 20 6 26
Total 1 38 11 50
The expected frequency assumption for the χ 2 test is violated. The test results below might
not be reliable.
H0: There is no relationship between employment status and gender.
H1: There is a relationship between employment status and gender.
Since χ STAT
2
= 1.1180 is smaller than the critical bound of 5.9915, do not reject H0. There is not
enough evidence to conclude there is a relationship between employment status and gender.

Copyright © 2009 Pearson Education, Inc. Publishing as Prentice Hall.

772 Chapter 12: Chi-Square Tests and Nonparametric Tests

12.113
cont.
Graduate School
Major n un y Total
a 3 6 2 11
ef 2 2 5 9
ib 1 2 3
is 3 1 4
m 2 3 4 9
mr 3 2 5 10
o 1 1 2
un 2 2
Total 15 17 18 50
The expected frequency assumption for the χ test is violated. The test results below might
2

not be reliable.
H0: There is no relationship between plans to go to graduate school and major.
H1: There is a relationship between plans to go to graduate school and major.
Since χ STAT
2
= 16.5146 is smaller than the critical bound of 23.6848, do not reject H0. There is
not enough evidence to conclude there is a relationship between plans to go to graduate
school and major.

Employment Status
Major ft pt un Total
a 1 8 2 11
ef 6 3 9
ib 1 2 3
is 2 2 4
m 9 9
mr 9 1 10
o 1 1 2
un 2 2
Total 1 38 11 50
The expected frequency assumption for the χ 2 test is violated. The test results below might
not be reliable.
H0: There is no relationship between employment status and major.
H1: There is a relationship between employment status and major.
Since χ STAT
2
= 14.5397 is smaller than the critical bound of 23.6848, do not reject H0. There is
not enough evidence to conclude there is a relationship between employment status and
major.

Copyright © 2009 Pearson Education, Inc. Publishing as Prentice Hall.

 773

12.113
cont.
Employment Status
Grad School ft pt un Total
n 1 13 1 15
un 12 5 17
y 13 5 18
Total 1 38 11 50
The expected frequency assumption for the χ 2 test is violated. The test results below might
not be reliable.
H0: There is no relationship between employment status and plans to go to graduate school.
H1: There is a relationship between employment status and plans to go to graduate school.
Since χ STAT
2
= 4.9579 is smaller than the critical bound of 9.4877, do not reject H0. There is not
enough evidence to conclude there is a relationship between employment status and plans to
go to graduate school.

12.115
Gender
Undergraduate Major f m Total
bi 2 2
bu 5 6 11
cm 4 5 9
ed 2 2
en 3 3
hu 1 1 2
o 3 3
pa 1 1
ps 1 1 2
ss 3 2 5
Total 16 24 40
The expected frequency assumption for the χ 2 test is violated. The test results below might
not be reliable.
H0: There is no relationship between undergraduate major and gender.
H1: There is a relationship between undergraduate major and gender.
Since χ STAT
2
= 10.2104 is smaller than the critical bound of 16.9190, do not reject H0. There is
not enough evidence to conclude there is a relationship between undergraduate major and
gender.

Copyright © 2009 Pearson Education, Inc. Publishing as Prentice Hall.

774 Chapter 12: Chi-Square Tests and Nonparametric Tests

12.115
cont.
Gender
Graduate Major f m Total
a 3 7 10
ef 2 7 9
ib 2 2
is 3 1 4
m 2 5 7
mr 5 2 7
un 1 1
Total 16 24 40
The expected frequency assumption for the χ 2 test is violated. The test results below might
not be reliable.
H0: There is no relationship between graduate major and gender.
H1: There is a relationship between graduate major and gender.
Since χ STAT
2
= 9.7388 is smaller than the critical bound of 12.5916, do not reject H0. There is not
enough evidence to conclude there is a relationship between graduate major and gender.

Gender
Employment Status f m Total
ft 13 20 33
pt 1 4 5
un 2 2
Total 16 24 40
The expected frequency assumption for the χ test is violated. The test results below might
2

not be reliable.
H0: There is no relationship between employment status and gender.
H1: There is a relationship between employment status and gender.
Since χ STAT
2
= 3.8384 is smaller than the critical bound of 5.9915, do not reject H0. There is not
enough evidence to conclude there is a relationship between employment status major and
gender.

Undergraduate Major
Graduate bi bu cm ed en hu o pa ps ss Total
Major
a 1 2 2 1 1 3 10
ef 4 2 1 1 1 9
ib 1 1 2
is 4 4
m 1 1 1 1 1 1 1 7
mr 4 1 1 1 7
un 1 1
Total 2 11 9 2 3 2 3 1 2 5 40
The expected frequency assumption for the χ test is violated. The test results below might
2

not be reliable.

Copyright © 2009 Pearson Education, Inc. Publishing as Prentice Hall.

 775

12.115 H0: There is no relationship between undergraduate major and graduate major.
cont. H1: There is a relationship between undergraduate major and graduate major.
Since χ STAT
2
= 58.4442 is smaller than the critical bound of 72.1532, do not reject H0. There is
not enough evidence to conclude there is a relationship between undergraduate major and
graduate major.

Undergraduate Major
Employment Status bi bu cm ed en hu o pa ps ss Total
ft 2 8 8 1 3 2 2 1 2 4 33
pt 2 1 1 1 5
un 1 1 2
Total 2 11 9 2 3 2 3 1 2 5 40
The expected frequency assumption for the χ test is violated. The test results below might
2

not be reliable.
H0: There is no relationship between undergraduate major and employment status.
H1: There is a relationship between undergraduate major and employment status.
Since χ 2 = 13.7769 is smaller than the critical bound of 28.8693, do not reject H0. There is not
enough evidence to conclude there is a relationship between undergraduate major and
employment status.

Graduate Major
Employment Status a ef ib is m mr un Total
ft 9 7 2 3 6 5 1 33
pt 1 2 1 1 5
un 2 2
Total 10 9 2 4 7 7 1 40
The expected frequency assumption for the χ test is violated. The test results below might
2

not be reliable.
H0: There is no relationship between graduate major and employment status.
H1: There is a relationship between graduate major and employment status.
Since χ STAT
2
= 12.2709 is smaller than the critical bound of 21.0261, do not reject H0. There is
not enough evidence to conclude there is a relationship between graduate major and
employment status.

Copyright © 2009 Pearson Education, Inc. Publishing as Prentice Hall.

776 Chapter 12: Chi-Square Tests and Nonparametric Tests

12.117 H0 : The number of service interruptions per day follows a Poison distribution.
H1 : The number of service interruptions per day does not follow a Poison distribution.
7

∑m
j =1
j fj
650
X= = = 1.3
n 500
Use λ = 1.3
Interruptions Per Day f0 mj f j P( X ) fe
0 160 0 0.272532 136.2659
1 175 175 0.354291 177.1457
2 86 172 0.230289 115.1447
3 41 123 0.099792 49.89603
4 18 72 0.032432 16.21621
5 12 60 0.008432 4.216214
6 8 48 0.001827 0.913513
7 or more 0 0 0.000404 0.201789
Total 500 650

Combine the last two classes:

Interruptions Per Day f0 fe ( f0 − fe )
2
/ fe
0 160 136.2659 4.133885899
1 175 177.1457 0.025989235
2 86 115.1447 7.376914873
3 41 49.89603 1.586084811
4 18 16.21621 0.196217784
5 12 4.216214 14.37007484
6 or more 8 1.115302 42.49884118
Total 500 500 70.18800862

( f0 − fe )
2

χ 2
k − p −1 =∑ = 70.188 χ crit
2
= χ 5,0.01
2
= 15.0863
k fe
Since 70.188 > 15.0863, reject H0. There is sufficient evidence to conclude that the distribution
of service interruptions does not follow a Poisson distribution at 1% level of significance.

Copyright © 2009 Pearson Education, Inc. Publishing as Prentice Hall.

 777

12.118 λ = 1.5
H0 : The number of service interruptions per day follows a Poison distribution.
H1 : The number of service interruptions per day does not follow a Poison distribution.

Interruptions Per Day f0 mj f j P( X )

0 160 0 0.223130
1 175 175 0.334695
2 86 172 0.251021
3 41 123 0.125511
4 18 72 0.047067
5 12 60 0.014120
6 8 48 0.003530
7 or more 0 0 0.000926
Total 500 650

Combine the last two categories:

Interruptions Per Day f0 fe ( f0 − fe )
2
/ fe
0 160 111.5651 21.02756048
1 175 167.3476 0.349923817
2 86 125.5107 12.43795484
3 41 62.75536 7.54191515
4 18 23.53326 1.301007902
5 12 7.059978 3.456642648
6 or more 8 2.22799 14.9534285
Total 500 500 61.06843334

( f0 − fe )
2

. χ 2
k − p −1 =∑ = 61.0684 χ crit
2
= χ 6,0.01
2
=16.8119
k fe
Since 61.0684 > 16.8119, reject H0. There is sufficient evidence to conclude that the distribution
of service interruptions does not follow a Poisson distribution with a population mean of 1.5 at
the 1% level of significance.

12.119 H0 : The number of commercial mortgages approved per week follows a Poison distribution.
H1 : The number of commercial mortgages approved per week does not follow a Poison
distribution.
7

∑m
j =1
j fj
219
X= = = 2.1058
n 104
Use λ = 2.1058

Copyright © 2009 Pearson Education, Inc. Publishing as Prentice Hall.

778 Chapter 12: Chi-Square Tests and Nonparametric Tests

12.119
cont.
Number Approved f0 mj f j P( X ) fe
0 13 0 0.121752 12.66221
1 25 25 0.256382 26.66368
2 32 64 0.269940 28.07378
3 17 51 0.189477 19.70564
4 9 36 0.099749 10.37388
5 6 30 0.042010 4.369
6 1 6 0.014744 1.533351
7 1 7 0.004435 0.461269
8 or more 0 0 0.001511 0.15719
Total 104 219 1.000000 104

Combine the last three classes:

Interruptions Per Day f0 fe ( f0 − fe )
2
/ fe
0 13 12.66221 0.009011434
1 25 26.66368 0.103805808
2 32 28.07378 0.549095235
3 17 19.70564 0.371491012
4 9 10.37388 0.181951945
5 6 4.369 0.608872099
6 or more 2 2.15181 0.010710218
Total 104 104 1.834937752
( f0 − fe )
2

χ k2− p −1 = ∑ = 1.8349 χ crit

2
= χ 5,0.01
2
=15.0863
k fe
Since 1.8349 < 15.0863, do not reject H0. At the 1% level of significance, there is not enough
evidence to reject the null hypothesis that the distribution of commercial mortgages approved
per week follows a Poisson distribution.

12.120 H0 : Battery life follows a normal distribution.

H1 : Battery life does not follow a normal distribution.
Life P( X ) fe
under 0 0.001938 0.968776928
0 - under 1 0.029741 14.87047242
1 - under 2 0.172953 86.47668571
2 - under 3 0.377089 188.5444078
3 - under 4 0.310381 155.1906456
4 - under 5 0.096272 48.13607992
5 - under 6 0.011144 5.571843365
6 or more 0.0005 0.241088258
Total 1.000000 500

Copyright © 2009 Pearson Education, Inc. Publishing as Prentice Hall.

 779

12.120 Combine the first two and the last two classes:
cont.
Life f0 fe ( f0 − fe )
2
/ fe
under 1 12 15.83925 0.930589
1 - under 2 94 86.47669 0.654515
2 - under 3 170 188.5444 1.823947
3 - under 4 188 155.1906 6.936331
4 - under 5 28 48.13608 8.423239
5 or more 8 5.812932 0.822867
Total 500 500 19.59149
( f0 − fe )
2

χ 2
k − p −1 =∑ = 19.5915 χ crit
2
= χ 3,0.05
2
= 7.8147
k fe
Since 19.5915 > 7.8147, reject H0. There is sufficient evidence to conclude that the distribution
of battery life does not follow a normal distribution.

12.121
Life mj f0 mj f j (m − X ) fj
2
P( X ) fe ( f0 − fe )
2
/ fe
j

Below 6 2.5 48 120 6129.12 0.0849 42.4547 0.7243

5 - under 10 7.5 84 630 3333.96 0.1917 95.8750 1.4708
10 - under 15 12.5 164 2050 277.16 0.2976 148.7940 1.5540
15 - under 20 17.5 126 2205 1724.94 0.2590 129.5141 0.0953
20 - under 25 22.5 50 1125 3784.5 0.1264 63.2108 2.7610
25 or more 27.5 28 770 5255.32 0.0403 20.1514 3.0569
Total 500 6900 20505 1.0000 500 9.662379
6 6

∑ mj f j ∑(m − X ) fj
2
j
j =1 6900 j =1 20505
(a) X= = = 13.8 S= = = 6.4103
n 500 n −1 499
(b) H0 : Call length follows a normal distribution
H1 : Call length does not follow a normal distribution
( f0 − fe )
2

χ 2
k − p −1 =∑ = 9.6624 χ crit
2
= χ 3,0.05
2
=7.8147
k fe
Since 9.6624 > 7.8147, reject H0. There is sufficient evidence to conclude that the
distribution of call length does not follow a normal distribution.

Copyright © 2009 Pearson Education, Inc. Publishing as Prentice Hall.