Business Statistics: Module 7.

ANOVA Page 1 of 13

Module 7. Analysis of Variance

Analysis of Variance (ANOVA)

 Compare and determine significant differences between the means of three or more
independent groups.

 An extension of t-test since it analyzes the differences between the means of more
than two independent groups.

 Analyzes variances to draw inferences about the means of the groups understudy.

 Helps recognize the relationship of the independent and dependent variables when
we know the presence of significant difference of each independent group’s mean
from the others.

 One-way ANOVA – pertains to “ANOVA procedure used for completely randomized

design” (Berenson et al., 2015, p. 396).

Components of ANOVA and its respective formulas

 Between and among groups variation (SSA) – or between treatments (SSTR), is the
differences from group to group cause the variation among groups; also known as
treatment effect

SSA = Ʃknj(xbarj – grand mean)2

where k = number of groups xbarj = mean of each group

nj = number of samples in each group Ʃ = sum of, or total

 Within groups variation (SSW) – variation is attributed to within groups differences;

considered as random error

SSW = ƩƩ(xij – xbarj)2 shorter version of the SSW formula

where xij = any value in each group xbarj = mean of each group

SSW = Ʃ(x – xbar1)2 + Ʃ(x – xbar2)2 + Ʃ(x – xbar3)2 + …

The above formula is the longer version of SSW formula; and ellipses (…) means to
continue adding the sum of the squared differences of the additional groups.

Please note that you need to use the longer SSW formula if there is no given
standard deviation (S) or variance (S2).
 Business Statistics: Module 7. ANOVA Page 2 of 13

If either the standard deviation (S) or variance (S2) is given, use the following
formula:

SSW = (n1 – 1)(S21) + (n2 – 1)(S22) + (n3 – 1)(S23) + … (depending on number of

groups)

 Total variation (SST) – sum of among-groups and within groups variations

SST = SSA + SSW

 Mean square among groups (MSA) – average sum of square among groups.

MSA = SSA k = number of groups

k–1
 Mean square within groups (MSW) – average sum of square within groups.

MSW = SSW N = number of all values in the data set

N–k k = number of groups

 ANOVA (F) = MSA F is the ANOVA symbol

MSW

 Degree of freedom (df) = k – 1 numerator__

N–k denominator

Like the previous hypothesis testing tools, level of significance and degree of freedom
are required to identify the F-critical value from the F-distribution tables which is
attached right after the references section of Module 7. The F-table is comprised of
column and row headings – column headings are the numerator and row headings are
the denominator in the degree of freedom. Also, there is a separate table for every level
of significance. Only F-critical value tables for .05 and .01 level of significance can be
found in this module.

Also, you need to solve for the mean of every group as well as the grand mean first
before you can start solving for ANOVA.

Mean (xbar) = Ʃx Grand mean = Ʃxbar n = number of values in a group

n k k = number of groups

ANOVA Summary Table

In ANOVA summary table, the first column represents the sources of variation, which
are the among groups and within groups variations. Next is the column for degree of
freedom column for among group variation (k – 1), within group variation (n – k), and
total variation (N – 1), wherein N represents the total number of values in the set of
data. The sum of square column contains the SSA, SSW, and SST values; mean
 Business Statistics: Module 7. ANOVA Page 3 of 13

square column shows the MSA and MSA values; and F and Sig. F columns display the
computed F value and p-value respectively.

Sources degree of freedom Sum of Square Mean square F Sig. F.

MSA = SSA MSA
Among groups k-1 SSA k-1 MSW p-value
MSW = SSW
Within groups N-k SSW N–k
Total N -1 SST

Problems illustration

1. A manufacturer of a certain brand (Brand A) of light bulbs would like to determine if

it’s the mean lifetime (in hours) of its own produced light bulbs is at par or even
better than competitors’ light bulbs. He did an experiment among its competitors’
light bulbs and its own products. He tested an equal number of randomly select light
bulbs of each brand understudy, turn the light bulbs on, and left it on until it burnt
out. The results of the experiment are shown on the table. At a .01 level of
significance, is there significant difference on the mean lifetime of the different
brands of lightbulbs?
No. Brand A (x – xbar)2 Brand B (x – xbar)2 Brand C (x – xbar)2 Brand D (x – xbar)2
1 1046 21.16 1028 0.04 1036 1 1059 6.76
2 1062 424.36 1032 17.64 1041 16 1061 21.16
3 1031 108.16 1040 148.84 1034 9 1043 179.56
4 1015 696.96 1020 60.84 1040 9 1049 54.76
5 1053 134.56 1019 77.44 1034 9 1070 184.96
Total 5207 1385.20 5139 304.80 5185 44 5280 447.20
Mean 1041.4 1027.80 1037 1056.4
Grand Mean = (1041.4 + 1027.80 + 1037 + 1056.4) = 1040.65
4

Before we proceed with analyzing the scenario, let us solve the totals, means, and
squared differences of the x and the mean.

The focus of the problem is the mean lifetime of the light bulbs produced by a
manufacturer and its competitors. He compared four brands of light bulbs including
his own product.

As previously mentioned, your guide in writing your hypothesis is the given question
in the problem, which is your alternative hypothesis; and you need to state it in
declarative form.

Also, please note that no value should appear in the null and alternative hypothesis.

Null hypothesis: There is no significant difference on the mean lifetime of the four
brands of light bulbs.

Alternative hypothesis: There is significant difference on the mean lifetime of the

four brands of light bulbs.
Business Statistics: Module 7. ANOVA Page 4 of 13

Or you can state in another way the alternative hypothesis.

Alternative: Not all mean lifetime of the four brands of light bulbs are equal.

Based on the ANOVA formulas, there is no way you will get a negative answer in
any of its components; thus, there is no need to divide the alpha by 2, take it as is.

Level of significance: .01

Then, determine the degree of freedom

Degree of freedom = k – 1 = 4 – 1 = 3 numerator

N – k 20 – 4 16 denominator

Do not divide the numerator and denominator; nor convert it to decimal point. Take
the degree of freedom as it in order to determine the F critical value @ 0.01 alpha
from F-table after the references section.

F-test critical = 5.29

Draw the normal curve and plot the critical t-values on the curve. Let’s focus on the
upper tail since there is no negative F-value.

Normal curve
5.29

Write your decision rule.

Decision rule: Reject null hypothesis if F > +5.29; otherwise, do not reject the null
hypothesis

Now we can start with our computation

SSA = Ʃknj(xbarj – grand mean)2

= 5(1041.4–1040.65)2 + 5(1027.8–1040.65)2 + 5(1037–1040.65)2 + 5(1056.4–1040.65)2
=2135.35

Since no standard deviation nor variance is given, we will use the long SSW formula

SSW = Ʃ(x – xbar1)2 + Ʃ(x – xbar2)2 + Ʃ(x – xbar3)2 + Ʃ(x – xbar4)2

= 1385.2 + 304.8 + 44 + 447.2 (these values came from the table above written in red)
= 2181.20

SST = SSA + SSW = 2135.35 +2181.20 = 4316.55

 Business Statistics: Module 7. ANOVA Page 5 of 13

MSA = SSA = 2135.35 = 2135.35 = 711.78

k – 1 (4 – 1) 3

MSW = SSW = 2181.20 = 2181.20 = 136.33

N – k (20 – 4) 16

F = MSA = 711.78 = 5.22

MSW 136.33

Next, state your conclusions; and give your recommendations.

Conclusion: Do not reject the null hypothesis, 5.22 < 5.29; which means there is no
enough evidence to support that at there is significant difference on the mean
lifetime of the four brands of light bulbs.

Recommendation: Based on the result, all brands of bulbs have relatively the same
mean lifetime. This means the lifetime of the brand A bulb is comparable with the its
competitors. Still, brand A manufacturer may need to do something to improve the
lifetime of its light bulbs. He may look into the process of producing the light bulb as
it may need to improve it. He also needs to check the quality of the necessary
materials as well as the equipment used. (You can give more recommendations
aside from the given example.)

2. Ms. Namit, general manager of Manyaman Restaurant, would like to know the
service quality of its four restaurant branches. She hired seven independent
researchers who have expertise in food-service evaluations as raters. The
restaurant staffs were not informed about the said investigation so that they would
not be anxious and they would act in normal and natural way like the regular
business days. The seven raters assessed the food and service of each of the four
restaurant branches using the rating scale from 0 (low) to 100 (high). The table
below shows the raters’ results. At a .05 alpha, is there significant difference on the
mean ratings on service quality of the four restaurant branches?

Restaurant Branches
Raters A B C D
1 75 68 91 77
2 77 75 93 79
3 79 70 94 82
4 80 66 98 76
5 84 65 95 84
6 82 71 98 86
7 79 73 90 85
Total 556 488 659 569
Mean 79.43 69.71 94.12 81.29
Variance (S2) 8.95 13.24 9.81 15.90
Business Statistics: Module 7. ANOVA Page 6 of 13

Grand mean = (79.43 + 69.71. 94.12 + 81.29) = 81.19

4

Null hypothesis: There is no significant difference on the mean ratings of service

quality of the four restaurant branches.

Alternative hypothesis: There is significant difference on the mean ratings of service

quality of the four restaurant branches.

Level of significance: .05

Degree of freedom = k – 1 = 4 – 1 = 3 .
N – k 28 – 4 24

Critical F value: 3.01

Normal curve
3.01

Decision rule: Reject the null hypothesis if F > 3.01; otherwise, do not reject the
null hypothesis.

Computation:

SSA = Ʃknj(xbarj – grand mean)2

= 7(79.43 – 81.19)2 + 7(69.71 – 81.19)2 + 7(94.12 – 81.19)2 + 7(81.29 – 81.19)2
= 2114.58

Since variances are given, we will use the second formula for SSW.

SSW = (n1 – 1)(S21) + (n2 – 1)(S22) + (n3 – 1)(S23) + (n4 – 1)(S24)

= 6(8.95) +6(13.24) + 6(9.81) +6(15.90)
= 287.40

SST = SSA + SSW = 2114.58 + 287.40 = 2401.98

MSA = SSA = 2114.58 = 2114.58 = 704.86

k – 1 (4 – 1) 3

MSW = SSW = 287.40 = 287.40 = 11.98

N – k (28 – 4) 24

F = MSA = 704.86 = 58.86

MSW 11.98
 Business Statistics: Module 7. ANOVA Page 7 of 13

Conclusion: Reject the null hypothesis, 58.86 > 3.01; which means there is significant
difference on the mean ratings on the service quality of the four restaurant branches.

Recommendation: At least one among the four restaurant branches has different mean
rating on its service quality. The highest and lowest mean ratings on service quality
were noted on branches C and B respectively. Ms. Namit needs to thoroughly look into
factors which greatly affect the service quality of the restaurants particularly Branch B
so she can make the necessary improvements on it. She can also look into the best
practices of branch C and adopt it to branches A, B, and D. (You can give more
recommendations aside from the given example.)

End of Module Discussion

Discuss the three ANOVA assumptions.

(You may search the internet about this topic, but you should be able to discuss it in
your own words. Also, you have to cite your source/s to include the author’s name,
copyright year, article title, and the link))

End of Module Exercises

Read and analyze the problems carefully. Work on the problem and follow the
step by step procedures in solving it – from the null hypothesis to the
recommendation. Round off to two decimal places.

1. A group of senior high school students did an experiment on the five brands of
disposable batteries for flashlights. They would like to compare and determine if
there is significant difference on the mean lifetime (in hours) of five brands of
disposable batteries. The results of their experiment are shown on the table below.
At a .05 alpha, is there significant difference on the mean lifetime of the five brands
of disposable batteries?
a Brand A Brand B Brand C Brand D Brand E
45 29 27 26 25
36 38 37 39 35
38 33 34 30 40
30 35 36 32 33
29 28 29 35 30
33 30 32 31 28
Total 211 193 195 193 191
Mean 35.17 32.17 32.50 32.17 31.83
Variance 34.97 14.97 15.50 19.77 28.57
Grand (35.17 + 32.17 + 32.50 + 32.17 + 31.83) = 32.77
Mean 5

2. A faculty researcher conducted a study to determine if teaching methods used,

which were lecture only, hands-on only, and lecture and hands-on, in the teaching-
learning process have bearing on the students’ performances based on their
standardized test scores. The table next page shows the summary results of the
 Business Statistics: Module 7. ANOVA Page 8 of 13

study. At a .01 level of significance, is there significant difference on the average

standardized test scores of students who received the three different teaching
methods?

Types of teaching method Sample size Mean Test Scores Standard Deviation
Lecture only 17 82.85 8.89
Hands-on only 20 89.67 7.65
Lecture and hands-on 23 93.20 6.16
Total 60

3. The human resource development (HRD) director of a certain firm wanted to

evaluate its three training programs for employees to perform certain manual tasks.
He randomly assigned 15 newly recruit employees to the training programs – 5
employees per program. After completing the training programs, employees were
tested on how fast (in terms of minutes) they can finish the tasks. The results of the
test are shown on the table below. At a .05 level of significance, is there evidence of
significant difference on the mean completion time of the employees who attended
the three training programs?

Employees Program 1 Program 2 Program 3

1 10 8 9
2 11 10 8
3 12 9 7
4 13 10 8
5 14 7 9
Total
Mean
Grand Mean

4. A rural bank has four branches in a certain city. The bank president was anxious that
employees were taking advantage of the bank’s substantial sick leave policy; and he
alleged that it might be associated with the branch where employees work. He
requested each branch manager to submit the records of sick leave (in days) availed
of randomly select employees the previous year. The table below shows the results.
At a .01 alpha, is there significant difference on the mean number of sick leave
availed of employees from the four branches?

Branch A Branch B Branch C Branch D

12 10 18 10
18 9 16 7
15 11 15 8
10 10 17 9
11 14
Total 55 51 80 34
Mean ? ? ? ?
Variance 12.25 0.70 2.50 1.67
Grand Mean ?
 Business Statistics: Module 7. ANOVA Page 9 of 13

5. A student researcher studied the daily coffee consumptions (in ounces) of people
from Cavite, Laguna, Batangas, and Quezon. The table next page shows the
summary results. At a .05 alpha, is there significant difference on the mean daily
coffee consumptions of the people from the four places?

Cavite Laguna Batangas Quezon

Sample 30 25 45 24
Mean 3.7 3.4 4.0 3.5
Variance 4.2 4.1 3.9 3.3

References

Albright, S. et al. (2015). Business analytics: data analysis and decision making (5th
ed). Cengage Learning.
Anderson, D., Sweeney, D.J., et.al., (2018). Modern business statistics. Australia:
Cengage Learning.
Antivola, H. (2015). Business statistics: a modular approach. Books Atbp. Publishing.
Anywhere Math. (2016). Introduction to Statistics.
https://www.youtube.com/watch?v=LMSyiAJm99g.
Berenson, M.L., Levine, D.M., & Krehbiel, T.C. (2015). Basic business statistics:
concepts and applications. Pearson Education Sou7th Asia Pte. Ltd.
Bowerman, B. (2017). Business statistics in practice: using modeling, data, and
analytics (8th ed.). McGraw-Hill Education.
Jaggia, S. (2019). Business statistics: communicating with numbers (3rd ed.). McGraw-
Hill Education.
Lee, N. (2016). Business statistics: using excel & SPSS. Sage.
Simple Learning Pro. (2015). Mean, median, mode, range, and standard deviation.
https://www.youtube.com/watch?v=mk8tOD0t8M0.
Sharpe, N. (2015). Business statistics 3rd ed. Pearson Education.
Willoughby, D. (2015). An essential guide to business statistics. John Wiley & Sons.
Business Statistics: Module 7. ANOVA Page 10 of 13
Business Statistics: Module 7. ANOVA Page 11 of 13
Business Statistics: Module 7. ANOVA Page 12 of 13
Business Statistics: Module 7. ANOVA Page 13 of 13