Marketing Research: Analysis of Variance and Covariance

MARKETING
RESEARCH
Analysis of Variance and Covariance
Topics to be covered:
 Data Requirement of One-Way ANOVA

 A Marketing Example of One-Way ANOVA
 The Steps of Conducting the One-Way ANOVA
Data Requirement of One-Way ANOVA
 One-way analysis of variance (ANOVA) can be conducted under the

following conditions:
Independent Variable (Non-metric) → Dependent Metric Variable

 Here the non-metric independent variable will be known as factor and
since we have only one factor in this analysis that is why it is known
as the one-way ANOVA. The number of categories in the factor will
be known as treatment and will be mathematically indicated by C.
A Marketing Example of One-Way ANOVA
Problem: A CMO of a company marketing carbonated soft-drinks has

developed three new flavors of cola and decided to measure the customers’
preference for these three flavors from 12 customers (each flavor will be given
to 4 customers) by using 9 point Likert Scale where 1 will indicate least
favored and 9 will indicate most favored. Which test you conduct here?
Solution: Here we can identify the following two variables:
So, one-way ANOVA is the appropriate analysis here.

The Steps of Conducting the One-Way ANOVA
 The procedure for conducting one-way analysis of variance may be

summarized below:
 Identify the Dependent and Independent Variables
 Decompose the Total Variation (Between Variation and Within
Variation: SStotal= SSbetween + SSwithin or mathematically, 𝑆𝑆𝑦 = 𝑆𝑆𝑥 +
𝑆𝑆𝑒𝑟𝑟𝑜𝑟
 Test the Significance (test the associated hypothesis)
 Measure the Effects (measure the strength of association)
 Interpret the Results (interpret the result and report it to the
marketing manager)
Problem: One-Way ANOVA
A CMO of a company marketing carbonated soft-drinks has developed three

new flavors of cola and decided to measure the customers’ preference for
these three flavors from 12 customers (each flavor will be given to 4
customers) by using 9 point Likert Scale where 1 will indicate least favored
and 9 will indicate most favored. The table shows the data collected from all
these customers:
[a] Which analysis would you suggest here? Why?

[b] Conduct your suggested analysis.
Solution (a):
Here we can identify the following two variables:
So, one-way ANOVA is the appropriate analysis here.

Solution (b)
Now the one-way ANOVA can be conducted here by conducting the

following five steps:
Step-1: Identify the Dependent and the Independent Variable: Here
Independent Variable is cola-flavors (non-metric) and the dependent
variable is customers’ preference (metric variable).
Step-2: Decompose the Total Variation: Here the following table can be
constructed for the purpose of decomposing the total variation:
(continued…)
Here n1 = 4, n2 = 4, n3 = 4, and n = n1 + n1 + n1 = 4 + 4 + 4 = 12 = total

sample size
Now the category means will be given by—
𝑌1 12
𝑌1 = = =3
𝑛1 4
𝑌2 20
𝑌2 = = =5
𝑛2 4
𝑌3 28
𝑌3 = = =7
𝑛3 4
3+5+7
And the grand means will be given by 𝑌 = =5
3
(continued…)
Now the between variation, 𝑆𝑆𝑥 = 𝑛(𝑌 − 𝑌)2
= 4 (5 – 3)2 + 4 (5 – 5)2 + 4 (5 – 7)2
= 16 + 0 + 16
= 32
And the within variation will be given by 𝑆𝑆𝑤𝑖𝑡ℎ𝑖𝑛 = y 𝑆𝑆𝑒𝑟𝑟𝑜𝑟 = (𝑌 − 𝑌)2
= (3 – 4)2 + (3 – 2)2 + (3 – 4)2 + (3 – 2)2
+ (5 – 6)2 + (5 – 4)2 + (5 – 5)2 + (5 – 5)2
+ (7 – 8)2 + (7 – 6)2 + (7 – 8)2 + (7 – 6)2
= 1+1+1+1+1+1+0+0+1+1+1+1
=10
(continued…)
Now the total variation can be calculated by adding between variation

and within variation,
SStotal = SSbetween + SSwithin
SSY = SSX + SSerror
SSy = 32 + 10 = 42
(continued…)
Step-3: Test the Significance: Now the following null and alternative
hypothesis can be developed here:
H0 : μ1 = μ2 = μ3
H1 : μ1≠ μ2≠ μ3
Now the following ANOVA table can be constructed for the purpose of
testing the associated hypothesis:
(continued…)
Here, C = number of categories in the factor = 3

Sample size, n = 12
𝑆𝑆𝑋 𝑆𝑆𝑒𝑟𝑟𝑜𝑟
Mean Squares are calculated by 𝑀𝑆𝑋 = and 𝑀𝑆𝑒𝑟𝑟𝑜𝑟 =
𝑑𝑓𝑋 𝑑𝑓𝑒𝑟𝑟𝑜𝑟
𝑀𝑆𝑋
Finally F-value is given by 𝐹 =
𝑀𝑆𝑒𝑟𝑟𝑜𝑟
(continued…)
Here the calculated value is Fcalculated = 14.41 and the value of Fcritical can
be found by consulting the F-Table (can be found at end of your text
book) by considering the appropriate level of confidence, and the
associated degrees of freedom in the following way:
Confidence level is assumed to be 95%, so the associated level of
significance is α = .05
First Degree of Freedom, df = 2
Second Degree of Freedom, df = 9
Now we have Fcritical = 4.26
(continued…)
Here Fcalculated = 14.41 > Fcritical = 4.26. So null hypothesis will be

rejected here. Now it can be concluded that there is a relationship
between cola-flavors and customers’ preference. In other words, cola-
flavors affect customers’ preference.
(continued…)
Step-4: Measure the Effects: Now the strength of association will be

calculated by the following formula:
𝑆𝑆𝑋 32
𝜂2 = = = .7619 or 76.19%
𝑆𝑆𝑌 42
The measure indicates that the relationship between cola-flavors and

customers’ preference is strong (by previously introduced rule of
thumb).
Step-5: Interpret the Results: There is a relationship between cola-
flavors and customers’ preference and this relationship was found to be
strongly held. The changes of customers’ preferences can be explained
by the changes of cola-flavors by 76.19%.
Thank You!

Marketing Research: Analysis of Variance and Covariance

Uploaded by

Copyright:

Available Formats

Marketing Research: Analysis of Variance and Covariance

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Marketing Research: Analysis of Variance and Covariance

Uploaded by

Copyright:

Available Formats

MARKETING

 Data Requirement of One-Way ANOVA

 One-way analysis of variance (ANOVA) can be conducted under the

Independent Variable (Non-metric) → Dependent Metric Variable

Problem: A CMO of a company marketing carbonated soft-drinks has

So, one-way ANOVA is the appropriate analysis here.

 The procedure for conducting one-way analysis of variance may be

A CMO of a company marketing carbonated soft-drinks has developed three

[a] Which analysis would you suggest here? Why?

Here we can identify the following two variables:

So, one-way ANOVA is the appropriate analysis here.

Now the one-way ANOVA can be conducted here by conducting the

Here n1 = 4, n2 = 4, n3 = 4, and n = n1 + n1 + n1 = 4 + 4 + 4 = 12 = total

Now the total variation can be calculated by adding between variation

Here, C = number of categories in the factor = 3

Here Fcalculated = 14.41 > Fcritical = 4.26. So null hypothesis will be

Step-4: Measure the Effects: Now the strength of association will be

The measure indicates that the relationship between cola-flavors and

You might also like