Professional Documents
Culture Documents
Chapter 09 - Analysis of Variance
Chapter 09 - Analysis of Variance
CHAPTER 9
ANALYSIS OF VARIANCE
9-1. H0: X X X X 1 2 3 4
H1: X X All 4 are different
X X
X X X 2 equal; 2 different
X
X X X X 3 equal; 1 different
X X X X 2 equal; other 2 equal but different from first 2
9-2. ANOVA assumptions: normal populations with equal variance. Independent random sampling from
the r populations.
9-3. Series of paired t-test are dependent on each other. There is no control over the probability of a
Type I error for the joint series of tests.
F Distribution
10% 5% 1% 0.50%
(1-Tail) F-Critical 2.0019 2.4626 3.5127 3.9634
9-5. r = 4 n1 = 52 n2 = 38 n3 = 43 n4 = 47
Computed F = 12.53. Reject H0. The average price per lot is not equal at all 4 cities. Feel very
strongly about rejecting the null hypothesis as the critical point of F (3,176) for = .01 is
approximately 3.8.
F Distribution
10% 5% 1% 0.50%
(1-Tail) F-Critical 2.1152 2.6559 3.8948 4.4264
9-6. Originally, treatments referred to the different types of agricultural experiments being performed on
a crop; today it is used interchangeably to refer to the different populations in the study.Errors are
the differences between the data points and their sample means.
9.7. Because the sum of all the deviations from a mean is equal to 0.
9-1
Chapter 09 - Analysis of Variance
9-10. An error is any deviation from a sample menu that is not explained by differences among
populations. An error may be due to a host of factors not studied in the experiment.
9.11. Both MSTR and MSE are sample statistics given to natural variation about their own means.
(If x > 0 we cannot immediately reject H0 in a single-sample case either.)
9-12. The main principle of ANOVA is that if the r population means are not all equal then it is likely
that the variation of the data points about their sample means will be small compared to the
variation of the sample means about the grand mean.
9-13. Distances among populations means manifest themselves in treatment deviations that are large
relative to error deviations. When these deviations are squared, added, and then divided by df’s,
they give two variances. When the treatment variance is (significantly) greater than the error
variance, population mean differences are likely to exist.
9.15 SST = SSTR + SSE, but this does not equal MSTR + MSE. A counterexample:
Let n = 21 r=6 SST = 100 SSTR = 85 SSE = 15
Then SST = SSTR + SSE = 85 + 15 = 100.
SSTR SSE 85 15 SST
But = MSTR MSE 18
r 1 n r 5 15 n 1
9-16. When the null hypothesis of ANOVA is false, the ratio MSTR/MSE is not the ratio of two
independent, unbiased estimators of the common population variance 2 , hence this ratio does not
follow an F distribution.
9-2
Chapter 09 - Analysis of Variance
Now sum this over all observations (all treatments i = 1, . . . , r; and within treatment i, all
observations j = 1, . . . , ni:
r ni r ni r ni r ni
i 1 j 1
( xij – 2
x) = ( x
i 1 j 1
i – 2
x) +
i 1 j 1
2( x i –
x )( xij – xi ) +
i 1 j 1
( xij –
x i )2
r
Notice that the first sum of the R.H.S. here equals
i 1
ni( x i – 2
x ) since for each i the
summand doesn’t vary over each of the ni) values of j. Similarly the second sum is
r ni ni
2 x ) ( xij –
[( x i – xi )]. But for each fixed i, ( xij – x i ) = 0 since this is just
i 1 j 1 j 1
the sum
of all deviations from the mean within treatment i. Thus the whole second sum in the long R.H.S.
above is 0, and the equation is now
r ni r r ni
i 1 j 1
( xij – 2
x) = ni( x i – 2
x) +
i 1 j 1
( xij – x i )2
i 1
ANOVA Table 5%
Source SS df MS F Fcritical p-value
Between 381127 2 190563.33 20.7084038 3.3541312 0.0000 Reject
Within 248460 27 9202.2222
Total 629587 29
9-3
Chapter 09 - Analysis of Variance
MINITAB output
One-way ANOVA: UK, Mex, UAE, Oman
Source DF SS MS F P
Factor 3 187.70 62.57 11.49 0.000
Error 28 152.41 5.44
Total 31 340.11
Critical point F (3,28) for = 0.05 is 2.9467. Therefore we reject H0. There is evidence of
differences in the average price per barrel of oil from the four sources. The Rotterdam oil market
may not be efficient. The conclusion is valid only for Rotterdam, and only for Arabian Light. We
need to assume independent random samples from these populations, normal populations with
equal population variance. Observations are time-dependent (days during February), thus the
assumptions could be violated. This is a limitation of the study. Another limitation is that February
may be different from other months.
9-20. An F(.05,2,101) = 3.61 result, relative to a critical value of 3.08637, indicates a significant difference
in their perceptions on the roles played by African American models in commercials.
9.21.(From Minitab):
Source df SS MS F
Treatment 2 91.0426 45.5213 12.31
Error 38 140.529 3.69812
Total 40 231.571
9-4
Chapter 09 - Analysis of Variance
p-value = .0001. Critical point for F (2,38) at = .05 is 3.245. Therefore, reject H0. There is a
difference in the length of time it takes to make a decision.
ANOVA Table 5%
Source SS df MS F Fcritical p-value
Between 91.0426 2 45.521302 12.3093042 3.2448213 0.0001 Reject
Within 140.529 38 3.6981215
Total 231.571 40
9-22. An F(.05,2,55) = 52.787 result, relative to a critical value of 3.165, indicates a significant difference in
the monetary-economic reaction to the three inflation fighting policies.
9-23. The test results exceed the critical value of F(.01,3,236) = 3.866. The results indicate that the
performances of the four different portfolios are significantly different.
9-25. Where do differences exist in the circle-square-triangle populations from Table 9-1, using Tukey?
From the text: MSE = 2.125
triangles: n1 = 4 x1 = 6
squares:n2 = 4 x 2 = 11.5
circles: n3 = 3 x3 = 2
For = .01, q (r,nr) = q 0.01(3,8) = 5.63 Smallest ni is 3:
T = q MSE / 3 = 5.63 2.125 / 3 = 4.738
| x1 x 2 | = 5.5 > 4.738 sig.
| x 2 x 3 | = 9.5 > 4.738 sig.
| x1 x 3 | = 4.0 < 4.738 n.s.
Thus: “1 = 3”; “2 > 1”; “2 > 3”
9-5
Chapter 09 - Analysis of Variance
9.27. Since H0 was rejected in Problem 9-19, there are significant differences.
T = q0.05(4,28) 5.4433 / 8 = 3.332
|UK – MEX| = |60.16 – 58.39| = 1.77
|UK – UAE| = |60.16 – 55.19| = 4.97
|UK – OMAN| = |60.16 – 54.1238| = 6.0362
|MEX – UAE| = |58.39 – 55.19| = 3.2
|MEX – OMAN| = |58.39 – 54.1238|= 4.2662
|UAE – OMAN| = |55.19 – 54.1238| = 1.0662
All are < 0.22, thus not significantas expected.
Tukey test for pairwise comparison of group means
UK
r 4 Mex Mex
n-r 28 UAE Sig UAE
q0 4.04 Oman Sig Sig
T 3.33248
9-6
Chapter 09 - Analysis of Variance
9-31. We cannot extend the results to planes built after the analysis. We used fixed effects here, not
random effects. The 3 prototypes were not randomly chosen from a population of levels as would
be required for the random effects model.
9-32. A randomized complete block design is a design with restricted randomization. Each block of
experimental units is assigned to treatments with randomization of treatments within the block.
9-33. Fly all 3 planes on the same route every time. The route (flown by the 3 planes) is the block.
9-34. Look at the residuals. If the spread of the residuals is not equal, we probably have unequal 2 ,
the assumption of equal variances is violated. A histogram of the residuals will reveal normality
violations.
9-35. Otherwise you are not randomly sampling from a population of treatments, and inference is not
valid for the entire “population.”
9-36. No. Rotterdam (and Arabian Light) was not randomly chosen.
9-37. If the locations and the artists are chosen randomly, we have a random effects model.
9-39. Limitations and problems: (1) We don’t know the overall significance level of the 3 tests; (2) If we
have 1 observation per cell then there are 0 degrees of freedom for error. Also, for a fixed sample
size there is a reduction of the df for error.
9.41. Since there are interactions, there are differences in emotions averaged over all levels of
advertisements.
9-7
Chapter 09 - Analysis of Variance
ANOVA Table 5%
Source SS df MS F Fcritical p-value
Location 2520.988 2 1260.49 50.645 3.1239 0.0000Reject
Job Type 2499.432 2 1249.72 50.212 3.1239 0.0000Reject
Interaction 212.716 4 53.179 2.1367 2.4989 0.0850
Error 1792 72 24.8889
Total 7025.136 80
9-8
Chapter 09 - Analysis of Variance
9-46. Since there are interactions but neither of the main factors have significant F-tests, a likely
conclusion is that the two factors work in opposite directions, i.e., inverse to each other.
9-47. Advantages: reduced experimental errors (the effects of extraneous factors) and greater economy of
sample sizes.
9-48. Use blocking by firm, to reduce the error contributions arising from differences between firms.
9-49. Could use a randomized blocking design: 4 observations, UK, Mexico, UAE, Oman at 4 locations
and 4 different dates.
9-50. A good blocking variable would be size of firm in terms of total assets or total sales, etc.
9-51. Yes. Have people of the same occupation/age/demographics use sweaters of the 3 kinds under
study. Each group of 3 people are a block.
9-52. As stated in 9-23, a good blocking variable would be some measure of diversity in the portfolio.
9-53. We could group the executives into blocks according to some choice of common characteristics
such as age, sex, years employed at current firm, etc. The different blocks for the chosen attribute
would then form a third variable beyond Location and Type to use in a 3-way ANOVA.
9-56. n = 70 r=4
SSTR = 9,875 SSBL = 1,445 SST = 22,364
SSE = 22,364 – 1,445 – 9,875 = 11, 044
11,044 9,875
MSE = = 53.35 MSTR = = 3,291.67
(69)(3) 3
F (3,207) = MSTR/MSE = 61.7
Reject H0. p-value is very small. Not all of the four methods are equally effective.
9-9
Chapter 09 - Analysis of Variance
9-58. n1 = 32 n2 = 30 n3 = 38 n4 = 41 n =141
MSTR = SSTR/(r – 1) = 4,537/3 = 1,512.33
F (3,137) = MSTR/MSE = 1,512.33/412 = 3.67
(at = 0.05) 2.67 < 3.67 < 3.92 (at = 0.01)
We can reject H0 at = 0.05. There is some evidence that the four names are not all equally well
liked.
Source SS df MS F
software 77,645 2 38,822.5 63.25
computer 54,521 3 18,173.667 29.60
interaction 88,699 6 14,783.167 24.09
error 434,557 708 613.78
Total 655,422 719
Critical value of F(.05, 2, 148) = 3.0572, which is less than F = 13.65. The results are significant.
9-10
Chapter 09 - Analysis of Variance
9-61.
Source SS df MS F
pet 22,245 3 7,415 1.93
location 34,551 3 11,517 2.99
interaction 31,778 9 3,530.89 0.92
error 554,398 144 3,849.99
Total 642,972 159
9-62. F-ratio = 4.5471 p-value = .0138 (using a computer). At = 0.05, only groups 1 and 3 are
significantly different from each other. Drug group is significantly different from the No. Treatment
group.
ANOVA Table 5%
Source SS df MS F Fcritical p-value
Between 3203.12 2 1601.56 4.54708749 3.123901138 0.0138 Reject
Within 25359.6 72 352.21667
Total 28562.7 74
9-63. a. Blocking (repeated measures) is more efficient as every person is his/her own control.
Reductions in errors. Limitations? Maybe carryover effects from trial to trial.
9-11
Chapter 09 - Analysis of Variance
ANOVA Table 5%
Source SS df MS F Fcritical p-value
Between 2137.78 2 1068.8889 22.2082976 3.219938094 0.0000 Reject
Within 2021.47 42 48.130159
Total 4159.24 44
9-65. n = 50 r =3
SSTR = 128,889 SSE = 42,223.987
128,899 / 2
F (2,98) = = 0.14958
42,223,987 / 98
Do not reject the null hypothesis
9-12
Chapter 09 - Analysis of Variance
9-67. Rents are equal on average. There is no evidence of differences among the four cities.
9.69. A one-way ANOVA strongly rejecting H0. For the three levels of Store, 95% confidence intervals
are calculated for means, as shown, which do not overlap at all.
Case 11: Rating Wines
(Template: ANOVA.xls, sheet: 1-Way)
data:
n 11 10 13 11
Chard Merlot C.Blanc C.Sauv
1 89 91 81 92
2 88 88 81 89
3 89 99 81 89
4 78 90 82 9
5 80 91 81 92
6 86 88 78 90
7 87 88 79 91
8 88 89 80 93
9 88 90 83 91
10 89 87 81 97
11 88 88 88
85
86
1) Do not reject the null hypothesis, there is no difference in the average ratings due to the type of grape.
ANOVA Table 5%
Source SS df MS F Fcritical p-value
Between 411.617 3 137.21 0.8594 2.8327 0.4698
Within 6545.63 41 159.65
Total 6957.24 44
9-13
Chapter 09 - Analysis of Variance
1.
ANOVA
n 10 10 10
Scan1 Scan2 Scan3
1 16 13 18
2 15 18 19
3 12 13 15
4 15 15 14
5 16 18 19
6 15 14 16
7 15 15 17
8 14 15 14
9 12 14 15
10 14 16 17
ANOVA Table 5%
Source SS df MS F Fcritical p-value
Between 20.6 2 10.3 3.4893 3.3541 0.0449 Reject
Within 79.7 27 2.9519
Total 100.3 29
ANOVA Table 5%
Source SS df MS F Fcritical p-value
Row 20.76667 4 5.19167 2.1239 2.5787 0.0934
Column 90.7 2 45.35 18.552 3.2043 0.0000Reject
Interaction 14.13333 8 1.76667 0.7227 2.1521 0.6705
Error 110 45 2.44444
Total 235.6 59
Reject the null hypothesis of equal number of scans per minute (columns)
Do not reject the null hypothesis that the clerks are equally efficient.
There are no interaction effects present.
9-14