Professional Documents
Culture Documents
Chapter 12
Analysis of Variance
1. 9.01 from Appendix B.6 (LO12-1)
2. 9.78 (LO12-1)
5. Ho: 12 = 22 H1: 12 22
Reject Ho where F > 3.10 (3.10 is about halfway between 3.14 and 3.07)
(12) 2
F = 1.44, found by F = = 1.44
(10) 2
Do not reject Ho. It is reasonable to conclude variations of the two populations could be the
same. (LO12-1)
6. Ho: 12 22 H1: 12 22
(3.9)2
Reject Ho when F > 3.68 F = 1.24, found by F = = 1.24
(3.5)2
Do not reject Ho. There is no evidence the variations in stock 1 is larger than stock 2.
(LO12-1)
12-1
Chapter 12 - Analysis of Variance
9. Step 1: Ho: Southwyck = Franklin = Old Orchard H1: Treatment means are not all the same
Step 2: The 0.05 significance level was chosen
Step 3: Use a F-statistics
Sep 4: Reject Ho if F > 4.26
Sep 5: Source SS df MS F
Treatment 276.50 2 138.25 14.18
Error 87.75 9 9.75
Reject Ho.
Step 6: The mean incomes are not all the same for the three tracks of land. (LO12-2)
10. Step 1: Ho: Banking = Retail = Insurance H1: Treatment means are not all the same
Step 2: The 0.05 significance level was chosen
Step 3: Use a F-statistics
Sep 4: Reject Ho if F > 3.89
Sep 5: Source SS df MS F
Treatment 22.93 2 11.45605.7325
Error 24.00 12 2.0
Reject Ho as 5.7325 > 3.89.
Step 6: The mean number of hours spent at the computer are not equal by industry. (LO12-2)
11. a. Ho: 1 = 2 = 3 H1: Treatment means are not all the same
b. Reject Ho if F > 4.26
c. SST = 107.20 SSE = 9.47 SS total = 116.67
d. Source SS df MS F
Treatment 107.20 2 53.600 50.96
Error 9.47 9 1.052
Total 116.67 11
e. Since 50.96 > 4.26, Ho is rejected. At least one of the means differ.
f. ( X 1 − X 2 ) t MSE (1/ n1 + 1/ n2 )
(9.667 − 2.20) 2.262 1.052(1/ 3 + 1/ 5)
7.467 1.69
[5.777, 9.157] Yes, we can conclude that the treatments 1 and 2 have different means.
(LO12-3)
12. a. Ho: 1 = 2 = 3 H1: Treatment means are not all the same
b. Reject Ho if F > 3.47
c. SST = 46.96 SSE = 53.00 SS total = 99.96
d. Source SS df MS F
Treatment 46.96 2 23.48 9.30
Error 53.00 21 2.52
Total 99.96 23
e. Since 9.30 > 3.47, Ho is rejected. At least one of the means differ.
f. ( X 1 − X 2 ) t MSE (1 / n1 + 1 / n2 ) or (6.0 − 4.25) 2.080 2.52 (1 / 101 + 1 / 8)
which is 1.75±1.57 or [0.18, 3.32]
Yes, we can conclude that the treatments 2 and 3 are different. (LO12-3)
13. There are two methods to answer this question. Using the six-step method we have the
following.
Step 1: Ho: 1 = 2 = 3 = 4 H1: Treatment means are not all equal
12-2
Chapter 12 - Analysis of Variance
12-3
Chapter 12 - Analysis of Variance
12-4
Chapter 12 - Analysis of Variance
42
40
Mean
38
36
34
32
30
D-320 J-1000 UV-57
Machine
21. a. There appears to be an
interaction. Sales using the UV-57 are better inside and the D-320 does better outside.
b. Two-way ANOVA: Sales versus Machine, Position
Source SS DF MS F P
Machine 16.333 2 8.167 0.72 0.502
Position 104.167 1 104.167 9.12 0.007
Interaction 457.333 2 228.667 20.03 0.000
Error 205.500 18 11.417
Total 783.333 23
Since the machine p-value (0.502) is greater than 0.05, it appears there is no
difference in the mean sales by machine. The p-value for position (0.007) is less than
0.05 indicating there is evidence that mean sales inside and outside are different.
However, there is a considerable interaction between those factors. Further study is
recommended.
c. T-statistics for the null hypothesis that sales inside are equal to sales outside for each
of the machines are: -6.39, -2.41, and 3.22 for the D-320, J-1000, and UV-57,
respectively. We would not reject the null hypothesis of equal means if the t value is
between -2.571 and 2.571 for the J-1000 and D-320 and if the t-value is between
-2.776 and 2.776 for UV-57. Thus, at the 0.05 level of significance, you can conclude
that D-320 does better outside and the UV-57 does better inside. (LO12-5)
12-5
Chapter 12 - Analysis of Variance
950
Mean
900
850
800
12-6
Chapter 12 - Analysis of Variance
(22.95) 2
F= = 2.41
(14.79) 2
Fail to reject Ho. There is no evidence the variation of the monthly sales for Sharkey is less
than White. (LO12-1)
28. Source SS df MS F
Treatment 320 2 160 8.00
Error 180 9 20
Total 500 11
a. 3
b. 12
c. 4.26
d. Ho: 1 = 2 = 3 H1: Not all means are equal
e. Ho is rejected. The treatment means differ. (LO12-2)
29. Ho: Discount = Variety = Department H1: Not all means are equal
Ho is rejected if F > 3.89
Source SS df MS F
Treatment 63.33 2 31.667 13.38
Error 28.40 12 2.367
Total 91.73 14
Ho is rejected. There is a difference in the treatment means. (LO12-2)
30. a. Step 1: Ho: Goust = Jet Red = Cloudtran H1: Not all means are equal
Step 2: 5% level chosen for us.
Step 3: Use an F-statistic
Step 4: Ho is rejected if F > 3.29
Step 5: Here is the ANOVA Table
Source SS df MS F
Treatment 127.1 2 63.6 3.04
Error 669.9 32 20.9
Total 797.0 34
Ho is not rejected since 3.04 < 3.29.
Step 6: There is no difference in the mean flight times.
12-7
Chapter 12 - Analysis of Variance
1 1
b. (50.5 − 55.714) 2.037 20.9 + This reduces to -5.21 4.31. So the
14 7
difference is between -0.90 and -9.52 minutes. (LO12-3)
31. Ho: Rec Center = Key Street = Monclova = Whitehouse H1: Not all means are equal
Ho is rejected if F > 3.10
Source SS df MS F
Factor 87.79 3 29.26 9.12
Error 64.17 20 3.21
Total 151.96 23
Since computed F of 9.12 > 3.10, the null hypothesis of no difference is rejected at the 0.05
level there is evidence the number of crimes differs by district. (LO12-2)
Source DF SS MS F P
Factor 2 3182.0 1591.0 35.56 0.000
Error 29 1297.5 44.7
Total 31 4479.5
The hypothesis of identical means can definitely be rejected since the p-value is virtually zero.
As seen from the non-overlapping confidence intervals, clothes have a mean attention span at
least ten minutes below the other groups.
12-8
Chapter 12 - Analysis of Variance
34. The null hypothesis of equal means is rejected because the F statistic (9.61) is greater than the
critical value (3.49). The p−value (0.0016) is also less than the significance level (0.05). The
mean repair times are different. (LO12-2)
35. The null hypothesis of equal means is rejected because the F statistic (8.26) is greater than the
critical value (5.61) using the 0.01 significance level. The p−value (0.0019) is also less than
the significance level (0.01). The mean miles per gallon are different. (LO12-2)
36. a. The null hypothesis of equal means is rejected because the F statistic (11.33) is greater
than the critical value (6.36). The p−value (0.0010) is also less than the significance
level (0.01). The mean production rates are different.
1 1
b. (43.333 − 41.5) 2.947 0.5444 +
6 6
This reduces to 1.833 1.255. So the difference is between 0.578 and 3.088.
(LO12-3)
12-9
Chapter 12 - Analysis of Variance
38 The critical value for a F-statistic with 6 degrees of freedom in the numerator and 48
degrees of freedom in the denominator at a 0.05 significance level is approximately 2.295
(averaging 2.34 and 2.25). Since 6.66 is greater than 2.295 the null is rejected. You can
also get the same conclusion from the p-value of 0.000, which is less than 0.05. At least
one mean is different.
Monday is higher than Tuesday, Thursday and Saturday because zero is not included in
the “Monday subtracted from” intervals. Tuesday is lower than Wednesday and Friday
because zero is not included in the “Tuesday subtracted from” intervals. Similarly
Wednesday is higher than Thursday and Saturday. Thursday is lower than Friday and
Sunday. Friday is higher than Saturday. (LO12-2)
39. For color the critical value of F is 4.76, for size it is 5.14
Source SS df MS F P
Treatment 25.0 3 8.3333 5.88 0.0227
Blocks 21.5 2 10.75 7.59 0.0321
Error 8.5 6 1.4167
Total 55.0 11
Ho for both treatments and blocks (color and size) is rejected. At least one mean differs for
color and at least one mean differs for size. (LO12-4)
12-10
Chapter 12 - Analysis of Variance
Source SS df MS F P
Treat 21.4 3 7.133 0.668 0.5876
Blocks 4278.7 4 1069.675 100.204 0.0000
Error 128.1 12 10.675
Total 4428.2 19
a. Critical value for Ho is 3.49. Computed F is 0.668, do not reject Ho.
b. Critical value for Ho is 3.26. Computed F is 100.204, Reject Ho for block means.
There is a difference in mean home values.
44. Ho: S = R = L H1: There is a difference in mean price between stores Reject Ho if
F > 3.63
Ho: 1 = 2 = … 9 H1: There is a difference in mean price between items Reject Ho if F > 2.59
Here is the two-way ANOVA Table
Source SS df MS F P
Stores 0.1370 2 0.0685 3.10 0.0726
Items 71.6136 8 8.9517 405.62 0.0000
Error 0.3531 16 0.0221
Total 72.103 26
There is a difference in the mean price of items but not in the mean price by stores, as the p-
value for items is less than the significance level but not for stores. (LO12-4)
12-11
Chapter 12 - Analysis of Variance
c. Source SS df MS F
Treat 495.62 3 165.207 9.847
Error 603.99 36 16.778
Total 1099.61
d. 9.847 > 2.89, so reject Ho. There is a difference in the treatment means.
e. (51.32 − 50.85) 2.03 9.725(1/10 + 1/10) = 0.470 2.831
[−2.361, 3.301] We cannot conclude that the number of minutes of music differ
between the hard rock station and the country/western station. (LO12-3)
1200
1150
Mean
1100
1050
1000
Men Women
Gender
47. a.
b. Here is the ANOVA table:
Source DF SS MS F P
Sector 1 156468 156468 40.61 0.000
Gender 1 44086 44086 11.44 0.004
Interaction 1 14851 14851 3.85 0.067
Error 16 61640 3853
Total 19 277046
Fail to reject the null hypothesis of the interaction effect. The interaction p-value
(0.067) is larger than 0.05. There is no significant interaction between the two factors.
c. Since both p-values are less than 5% and very small, means are different by gender
and by sector.
12-12
Chapter 12 - Analysis of Variance
d. Private sector mean wages and male mean wages are higher than the converse of
either public sector wages or female wages. (LO12-5)
48. a.
Here is the MINITAB output: Two-way ANOVA: Dirt versus Cycle, Brand
Source DF SS MS F P
Cycle 3 0.017225 0.0057417 27.02 0.000
Brand 3 0.005725 0.0019083 8.98 0.001
Interaction 9 0.002200 0.0002444 1.15 0.386
Error 16 0.003400 0.0002125
Total 31 0.028550
b. Since the interaction p-value (0.386) is larger than the 0.05 significance level, there is
no evidence of an interaction effect.
12-13
Chapter 12 - Analysis of Variance
0.16
Mean
0.14
0.12
0.10
A B C D
Brand
c. Both the cycle time means and the brand means are significantly different with
p-values of 0.000 and 0.001 respectively.
d. The confidence interval plots indicate that cycle times of 22 and 24 minutes are
superior to shorter cycle times and that detergent Brand C is better than Brands A and
D. Brands B and C are not statistically different. Brands A, B, and D are not
statistically different. (LO12-5)
49. a. Ho: p2 = np
2
H1: p2 np
2
Reject Ho. The p-value is less than 0.05. There is a difference in the variance of
average selling prices between houses with pools and houses without pools.
b. Ho: g2 = ng
2
H1: g2 ng
2
Reject Ho. There is a difference in the variance of average selling prices between
house with garages and houses without garages. The p-value is very small.
12-14
Chapter 12 - Analysis of Variance
Fail to reject Ho. The p-value is much larger than 0.05. There is no statistical evidence
of differences in the mean selling price between the five townships.
Fail to reject Ho. The p-value is much larger than 0.05. There is no statistical evidence
of differences in the mean selling price between the five agents. Is fairness of
assignment based on the overall mean price, or based on the comparison of the means
of the prices assigned to the agents?
12-15
Chapter 12 - Analysis of Variance
e. In this exercise, students should conduct a two-way or two-way with interaction ANOVA.
The results show that the mortgage type is a significant effect on the mean years of
occupancy. Students will be able to explain this result based on the way the mortgage
types work. In addition, the interaction term provides interesting information.
Depending on the statistical software used, the interaction plots will provide data
about the interaction of mortgage type and default. For example, it may be interesting
to see that both adjustable and fixed rate mortgages default, on average, after about 6
years.
12-16
Chapter 12 - Analysis of Variance
Do not reject Ho. There may be no difference in the variances of team salary by
league.
b. Sample 1, 2, and 3 are “Less than 2”, “2 up to 3”, and “3 or more”, respectively, for
remainder of solution.
Ho: 1 = 2 = 3 H1: Not all treatment mean number of wins are equal.
Do reject Ho. The mean numbers of games won is different across the three different
attendance categories. The teams with higher attendance had high mean number of
wins.
c. Ho: 1 = 2 = 3 H1: Not all treatment mean number of home runs are equal
Do not reject Ho. The mean numbers of home runs could be the same across the three
different attendance categories.
12-17
Chapter 12 - Analysis of Variance
Reject Ho. The mean salaries are likely not the same across the three different
attendance categories. It appears that teams with higher attendance have higher
salaries. (LO12-2)
51.
a. Ho: B = K = T H1: Not all treatment (manufacturer) mean maintenance costs, are
equal.
Analysis of Variance
Means
Do not reject Ho. The mean maintenance costs for the by the bus manufacturer is not different.
P-value = 0.7664.
12-18
Chapter 12 - Analysis of Variance
b. Ho: B = K = T H1: Not all treatments have equal mean miles since the last maintenance.
Do not reject Ho. The mean miles since the last maintenance by the bus manufacturer is not different.
P-value = 0.4828. (LO12-3)
12-19