MINING THE RELATIONSHIP BETWEEN DEMOGRAPHIC VARIABLES AND BRAND ASSOCIATIONS

International Journal of Managing Value and Supply Chains (IJMVSC) Vol.4, No. 3, September 2013
DOI: 10.5121/ijmvsc.2013.4301 1
MINING THE RELATIONSHIP BETWEEN
DEMOGRAPHIC VARIABLES AND BRAND
ASSOCIATIONS
Ajayeb Abu Daabes1
and Faten Kharbat2
1
Emirates College of Technology, Abu Dhabi, UAE
2
Al Ain University of Science and Technology, Abu Dhabi, UAE
ABSTRACT
This research aims to mine the relationship between demographic variables and brand associations, and
study the relative importance of these variables. The study is conducted on fast-food restaurant brands
chains in Jordan. The result ranks and evaluates the demographic variables in relation with the brand
associations for the selected sample. Discovering brand associations according to demographic variables
reveals many facts and linkages in the context of Jordanian culture. Suggestions are given accordingly for
marketers to benefits from to build their strategies and direct their decisions. Also, data mining technique
used in this study reflects a new trend for studying and analyzing marketing samples.
KEYWORDS
Brand Associations, Branding, Service Marketing, Demographic Variables, X-Means Algorithm, Data
Mining, Ranking, ReilefF.
1. INTRODUCTION
Exploring and analyzing characteristics of consumers can help marketers to target the right
customers in the right way by the right tools. Understanding the relationship between
demographic characteristics and brand associations has many indicators that reflect the customer
behaviour from one side, and how to tailor marketing mix to fit and satisfy customers‟ needs and
preferences from the other side.
According to Aaker [1], brand associations are "the category of a brand's assets and liabilities that
include anything ``linked'' in memory to a brand". Keller [2] defines brand associations as
"informational nodes linked to the brand node in memory and contain the meaning of the brand
for consumers. Associations come in all forms and may reflect characteristics of the product or
aspects independent of the product".
Scholars variously described brand image as “the perceptions and beliefs held by consumers, as
reflected in the associations held in consumer memory” [3]. Several efforts have studied specific
elements of brand associations; for example, [1], [4], [5], [6], [7]. Brand associations form brand
image which is composed of anything deeply placed in customer‟s mind toward the brand.
Associations may be implicit or explicit perceptions with a brand name through information from
commercial/ non-commercial sources (e.g., word of mouth, media, country of origin, channel of
distribution, and many other sources). Therefore, the marketers should manage well these sources
to influence the consumers‟ image to be associated strongly with something positive and unique.
The huge amount of data with complicated relations necessitates using other techniques, such as
data mining, that are concerned with formulating the process of generalization as a search through

2
possible hypothesis instead of focusing on testing exist hypothesis which is applied by
conventional techniques. Data mining is defined to be:" the process of discovering meaningful
new correlations, patterns and trends by sifting through large amounts of data stored in
repositories, using pat-tern recognition technologies as well as statistical and mathematical
techniques" [9]. Accordingly, this study aims to achieve the following goals:
• Study the relationship between brand associations and demographic variables through
data mining techniques.
• Study demographic variables according to their weight on brand associations.
• Recommend some suggestions for marketers to direct marketing strategies for their target
markets.
This study is organized into five sections. Section 2 overviews some of the recent related work
for the brand associations and customer behaviours. Section 3 describes the methodology used in
this study. Sections 4 and 5 illustrate the results and conclusions for the study along with some
future directions.
2. RELATED WORK
The relationship between brand associations and customer characteristics could be measured by
different conventional tools. Keller [2] defines quantitative research techniques to be "a means to
better approximate the breadth and depth of brand awareness, the strength, favourability, and
uniqueness of brand associations, the favourability of brand responses, and the nature of brand
relationships". Quantities trend is very useful in concluding summaries which represent various
types of scale questions. They have been used as "a primary ingredient in tracking studies that
monitor brand knowledge structures of consumers over time" [2]. Qualitative researches
techniques [10], in the other hand, "are a means of identify possible brand association" [2]. They
have unstructured nature and used to provide an in-depth of what brands and products mean to
consumers [11].
Understanding the relationship between customer characteristics and brand associations has been
an interesting area of research [15, 16]. Several studies conducted recently aimed to study many
elements of consumer brand associations throughout qualitative and quantitative researches using
conventional tools to analyze data.
Srivastava [13] conducted a descriptive study to ascertain the perception change in the brand
identity of a well known soap brand, brand personality, awareness levels, usage pattern,
promotional campaigns and the factors influencing the buying behaviour of the consumers. The
study observed that most of the respondents identified the soap clearly as a family/male soap. The
study focused on the main issues affecting the soap‟s brand identity, the markets it has entered,
the positioning strategies it has adopted, the celebrities used in endorsing the product, the pricing
techniques used.
Another study was conducted by Ménde, Oubiña, & Rubio [14] aims to analyze the relative
importance of brand-packaging, price and taste in the formation of brand preference for
manufacturer and store brands in food product categories. The study used a blind taste test of the
product using three brands (two manufacturer brands and one store brand). It used con-joint
analysis to analyze the influence of the intrinsic cue (taste) and the extrinsic cues (price and
brand-packaging) on consumers‟ preference for manufacturer and store brands. The results show
that not knowing the brand to which the taste tested belongs, leads consumers in general to order

3
their preferences fundamentally by taste. However, the results differ by product category and
consumer segment analyzed. Consumers who evaluate the taste of store brands as better change
their preferences more when they know which brand belongs to which taste. Further, the change
in preference when consumers know the brand-taste correspondence is clearly greater in the most
differentiated category. Store brands are only evaluated more favourably than manufacturer
brands in price.
Alimen & Cerit [15] analyzed via exploratory study the impacts of gender, field of education, and
having consumed the brand, on consumers‟ brand knowledge. The survey was con-ducted by
using convenience sampling over Turkish university students. The study discovered significant
differences with respect to usage, gender, and departments. The study revealed that students in
departments more related to fashion and female students have more knowledge about the tested
brands. Also, "the findings demonstrate that consumption of a brand increases both brand
awareness and brand image" [15].
Alexandris et al. [17] focused on the exploratory factor analysis and revealed eight brand
association factors: popularity, management, logo, escape, vicarious achievement, nostalgia, pride
and affect. The study was conducted over 165 participants of a managed-owned fitness club. Two
questionnaires were designed of 25 questions each, one to measure brand associations, and the
other to measure service quality. It was found that five of the eight brand associations (escape,
nostalgia, pride, logo, and affect) significantly contributed to the prediction of loyalty. Also, it
was revealed that the service quality dimensions predicted significant amount of variances in all
the eight brand associations.
3. RESEARCH METHODOLOGY
This study applies some of the well known data mining techniques to study the relative
importance of demographic variables and the favourable bonds with brands associations for fast
food restaurants‟ brand. In this research we aim to study some of restaurants‟ brand associations
in relation with customers‟ demographic variables. The following restaurants‟ brand associations
were studied:
1. Brand Elements which are defined as the name or logo or any other symbols represent the
restaurant.
2. Emotions that reflect feelings that are carried toward the brand.
3. Trust that means the level of credibility the brand stands for.
4. Distribution Channels that reflect availability and convenience for the customer.
5. Brand Quality that reflects perceived quality from customer.
6. Country of Origin that represents the perceived image the customer has toward the country of
origin.
7. Price that indicates the level of affordability for the customer.
8. Loyalty that represents positive feelings toward the brand that lead to a commitment to re-
purchase the brand and continue using it.
Experiments were performed using two well-known data mining techniques via Weka software
[18]. Weka is a collection of machine learning algorithms for solving real-world data mining
problems [18]. The Weka workbench [19] contains a collection of visualization tools and
algorithms for data analysis and predictive modelling, together with graphical user inter-faces for
easy access to this functionality. The first data mining technique is the X-Mean Clustering
Algorithm which is an extended K-Means Clustering algorithm [20] by an improve-structure part
in which the centres are challenged to be split in its region. The decision between the records of
each centre and itself is done comparing the BIC-values of the two structures [21]. The second

4
data mining algorithm used in this research is the Ranker Algorithm which “ranks attributes by
their individual evaluations.” [19]. The algorithm is used in conjunction with the Relief algorithm
(ReliefF [23]) to estimate and evaluate the variables. It is also used to detect conditional
dependencies between data features. Many researchers have demonstrated that ReliefF algorithm
is very useful in different applications and domains, for example [24] and [25]. It is usually used
to deal with data that is noisy, incomplete, or/and has more than two classes.
The questionnaire was developed to collect the primary data needed for this study; it consists of
two main parts: the first part includes questions about demographic data of the respondent, which
are: gender, marital status, educational degree, age, and income level. Table 1 illustrates the
variables and their categories along with their frequencies.
The second part of the questionnaire includes eight variables. 49 statements were developed to
cover all the variables. Likert scale in five levels was applied. The variables: (1) Brand elements
were measured in paragraphs 1-5. (2) Brand quality was measured in paragraphs 6-18, (3) Brand
trust was measured in paragraphs 19-22, (4) Emotions were measured in paragraphs 23-27, (5)
Distribution channels were measured in paragraphs 28-33, (6) Price was measured in paragraphs
34-38, (7) Country of origin was measured in paragraphs 38-43, and (8) Loyalty was measured in
paragraphs 44-49.
Table 1: Summary of Demographic Variables
A total of 1067 survey questionnaires were distributed via convenience sample covering fast food
restaurant brand chains‟ clients in the capital of Jordan, Amman. From this, 889 questionnaires
were returned and 77 questionnaires were dropped from the analysis due to incomplete answers,
with a final sample of 812 respondents representing a response rate of 76%. Briefly, the
respondents consisted of 61% (n=493) males, about 54% the age group of 20-30 years old, about
78% were single, and more than half had university education. All in-come ranges were covered
%
Frequency
Category
Demographic
Variables
16
394
Male
Gender
49
469
Female
52
504
Less than 20 years
Age
23
342
50
-
40
9
43
46
-
30
2
34
36
-
20
4
24
More than 50 years
47
144
Single
Marital status
64
642
Married
2
30
Divorcee
1
39
Less than High School
Educational degree
67
635
High School
3
40
Diploma
27
343
Bachelor
63
664
Graduate
62
667
Less than 500 JD
Income level
45
515
200
-
6000
56
619
6006
-
6200
66
90
6206
-
5000
56
644
More than 2000JD

5
in the questionnaire, but about 32% of the respondents had an estimated monthly income between
500-1000 JD. A summary of demographic variables is presented above in Table 1.
4. RESULTS
Our work has two main results. First, we present the findings for the X-Means algorithm over
each demographic variable separately. Next, we evaluate the variables to discover their ranking
which gives us indication about their importance to the brand associations.
After applying X-Means algorithm over the sample, five figures are produced to reflect each
demographic variable separately in relation with brand associations using excel line chart type as
shown in Figure 1.
a) Clusters based on Gender b) Clusters based on Marital Status
c) Clusters based on Degree Level d) Clusters based on Age
e) Clusters based on Income
Figure 1: The Clusters Based on the Demographic Variables
Figure 1(a) reveals that there are positive associations between the customers‟ gender and the fast
food restaurants‟ brand. The figure shows that the sample is divided into three groups; the first
and the second groups are very similar in their trend. They show neutral and favourable relations
with brand associations between customers and fast food restaurants brands; that is, most of the
associations have a centre around or above 2. The first group is for the males customers and the
second group is for part of the females‟ customers. The third group is for the other part of the

6
females‟ customer which shows different trend towards brand associations; it reflects weak
relationship with brand associations.
Figure 1(b) reveals that there are positive associations between the customers‟ marital status and
the fast food restaurants' brand associations. The figure shows that the sample is divided into
three groups; married customer has good and favourable relationship with fast food restaurants
brand along with part of the singles customers. On the other hand there is a segment of single
customers that have below average relationship with brand associations.
Figure 1(c) reveals that there are positive associations between the customers‟ educational degree
and the fast food restaurants' brand associations. Although the figure shows four different groups,
the sample trend seems to have three different tendencies. Graduate and bachelors‟ educational
degree have positive and favourable relationship with fast food restaurants brand associations
along with segment of high secondary schools customers. The remaining two groups are for
diploma and segment of high secondary schools customers which have below average
relationship with brand associations.
Figure 1(d) reveals that there are positive associations between the customers‟ age and the fast
food restaurants' brand associations. The figure shows that one group of the sample has
significant relationship, which is the customers between 20-30 years old, and brand association.
This group is divided into two different clusters. One of them has positive and favourable
relationship with fast food restaurants brand and the other one has weak relationship.
Figure 1(e) reveals that there are positive associations between the customers‟ income and the fast
food restaurants' brand associations. The figure shows two groups that have significant trend;
customers‟ incomes‟ between 500 and 1000JD, and between 1501 and 2000JD. Customers from
the two categories reflect positive and favourable relationship with fast food restaurants' brand.
The second part for our result is the output of the Ranking and Relief algorithm which is a weight
between −1 and 1 for each variable; the more positive weights indicate more predictive attributes.
Table 2 illustrates the demographic variables and their weight based on the used algorithm. The
result reveals that the Income variable is the most important indicator for the current sample on
the brand associations. On the other hand, the age variable has shown to have the least effect.
Table 2: The Demographic Variables‟ Weights Using the Ranking and Relief algorithm
Weight DV
-0.01017 Age
-0.0075 Sex
-0.0036 Degree
-0.00268 Material
0.00895 Income
5. DISCUSSION
The conclusions of this study split into two parts; the first one is that using the X-Means
algorithm over all the demographic variables help discovering brand associations and suggest
some relations between the variables and brand associations. Findings reflect neutral and
favourable relationship between demographic variables and brand associations. These results
support the "variety seeking buying behaviour [22]" type: which occurs in situations "of low
consumer involvement but significant perceived brand differences. In such cases consumers often
do a lot of brand switching" [22]. This indicates that there is no real brand loyalty; hence, further
effort should be conducted in future research to distinguish between associations. However, some
More
predictive

7
demographic variables show to have a unique trend toward brand associations; for example,
married customers seem to have positive trend towards brand associations, some of the females‟
customers tend to have negative brand associations, and the customers from the middle age seem
to have two different trends with the brand associations.
All males group and a segment of females have favourable relationship with brand associations
while the remaining segment of females has weak relationship. This result refers to the culture
prevailed in Jordanian society which allows males to move more freely, and that definitely helps
them to build some associations with their brands.
Being single has strong effect on the brand associations. This result seems to match the Jordanian
social customs which encourage family relations and home cooking meals. Singles studying or
working in Amman may be from another city and hence their bonds with the restaurants are
strong. In general, singles have less responsibility towards their families and strong bonds with
their friends which encourage them to use the restaurant services more often.
Graduate and bachelor‟s customers have significant relationship with brand associations; this
result aligns with the level of awareness within these categories that help to build and create
positive relationships with brands.
Both married and singles customers have significant relationship with brand association except a
segment of singles customers. This result seems to be logic since married customers have a lot of
standards to make purchasing decisions. While singles have different trends, some of them has
positive relations with brand associations, and the other one has weak relations which indicates
that this segment don‟t care of about fast food brand associations.
For the income element, there are no differences between the two clusters, but the figure shows
that both clusters should be in high income range to build such an associations. This result is logic
since the income level in Jordan (Amman in particular) is low, and the visits to restaurants may
be considered a complementary expense for low incomes. For that the output of the Relief
algorithm reflects that the income is the most effective factor in comparison with other
demographic variables in this study.
Customers (20-30) years old seemed to have significant relationships rather than other categories.
This result aligns with nature of Jordanian society in which the majority of this category is
unmarried therefore they visit fast food restaurants frequently and building favourable
relationships with these brands.
It recommends a practical technique that can help marketer to forming their strategies and
decisions using well known data mining techniques instead of pure traditional statistical
techniques. Data mining techniques are used to discover implicit interesting patterns and relations
within the collected data using machine learning, pattern recognition, statistics, databases, and
algorithms. This study opens the door for using such techniques in the field of branding. Further
investigations may illustrate a comparison study between the results using pure statistical
methods and data mining techniques.
The second part is highlighting the brand associations between customers and fast food
restaurants brands chains in Jordan. The result reveals that there are neutral and favourable
associations, and this result is reasonable since the switching cost is low and the customers want
to check variety of services. The study shows that some demographic variables have significant
impact on brand associations, such as, income, marital status, and educational degree. Therefore,
further efforts have taken place to examine the strengths of the suggested associations via ranking
technique to rank the variables according to their impact on brand associations as shown above.

8
6. CONCLUSIONS AND FUTURE WORK
The conclusions of this study split into two parts; the first one is that using some data mining
techniques, such as X-Means algorithm, over the demographic variables help discovering brand
associations and suggest relations between these variables. The findings reflect neutral and
favourable relationship between demographic variables and brand associations. These results
support the "variety seeking buying behaviour" type: which occurs in situations "of low consumer
involvement but significant perceived brand differences. In such cases consumers often do a lot
of brand switching" [22]. This indicates that there is no real brand loyalty for the restaurants‟
brands; hence, further effort should be conducted in future research to distinguish between
associations. However, some demographic variables show to have a unique trend toward brand
associations; for example, income, marital status, and educational degree have significant
differences between customers and brand associations.
The second part of the conclusion relies on predicting significant targets for marketers that is
derived from the previous results. In general all customers hold neutral associations with
restaurants‟ brand. Hence, this fact creates a lot of pressure on marketers to switch customer form
neutral status to loyalty status.
Segments of the female group have shown weak associations with restaurants‟ brand; this
requires marketers to attract this segment through their brand strategies. Also, category of
customers between 20 and 30 years old presents significant relation with the associations, which
indicates positive attitudes towards fast food brands. Marketers are advised to focus on this
category to make them more loyal and to attract other ages‟ segments.
High income customers seem to have a good trend towards brand associations; however, low
income levels have negative attitude toward fast food brands. Marketers should persuade other
categories with their pricing strategies.
High educational customers who have significant relations with restaurants‟ brand reflect that
lower educational level customers may have an ignorance relation, which means that customers
from this segment does not care about any associations!
Married customers and singles both are targeted for fast food restaurants; marketers shouldn‟t
ignore any segment unless they scan the culture and their trends.
Finally, this research recommends a practical technique that can help marketer to form their
strategies and decisions using well known data mining techniques instead of pure conventional
techniques. This study opens the door for using such techniques in the field of branding. Further
investigations may take place to compare this study with others from different culture and
countries.
REFERENCES
1. Aaker, D.: Managing Brand Equity. Capitalizing on the Value of a Brand Name. Free Press: New
York (1991).
2. Keller, K.: Strategic Brand Management, 3rd edition, Upper Saddle River, NJ: Prentice-Hall
(2008).
3. Kotler, P., Keller, K.: Marketing Management, Upper Saddle River, NJ: Prentice-Hall, 13th
edition (2009)
4. Aaker, D.: Building Strong Brands, Free Press: New York (1996)
5. Aaker, D., Jacobson, R.: Study Shows Brand-building Pays Off For Stockholders, Advertising
Age, vol. 65 (1994).
6. Aaker, J.: Dimensions of Brand Personality, Journal of Marketing, Research, vol. 34: 3, 347-356
(1997).

9
7. Keller, K.: Conceptualizing, measuring, and managing consumer-based brand equity, Journal of
Marketing, vol. 57, 1-22 (1993).
8. Abu Daabes, A.: Determining the Factors that have an Impact on Building Customer-Based Brand
Equity in the Service Sector “Empirical Study on Fast Food Restaurant Chains in Jordan”, PhD
Thesis (2009).
9. Larose, D.: Discovering Knowledge in Data: An Introduction to Data Mining, John Wiley & Sons
Inc (2005).
10. Belk, R. W.:Handbook of Qualitative Research Methods in Marketing, Edward Elgar Publishing
(2006).
11. Bruner, G. C., Hensel, P. J.: Marketing scales handbook: A compilation of multi-item measures.
Chicago: American Marketing Association, vol. 2. (1996).
12. Delassus, V. P., Descotes, R. M.: Brand name substitution and brand equity transfer, Journal of
Product & Brand Management, vol. 21: 2, 117 – 125 (2012)
13. Srivastava, R.K.: Understanding brand identity confusion, Marketing Intelligence & Planning, vol.
29: 4, 340 – 352 (2011)
14. Méndez, J. L., Oubiña, J., Rubio, N.: The relative importance of brand-packaging, price and taste
in affecting brand preferences, British Food Journal, vol. 113: 10, 1229 – 1251 (2011)
15. Alimen, N., Cerit, A.G.: Dimensions of brand knowledge: Turkish university students'
consumption of international fashion brands, Journal of Enterprise Information Management, vol.
23: 4, 538 – 558 (2010).
16. Danes, J. E. , Hess, J. S., Story, J. W., York, J. L.: Brand image associations for large virtual
groups, Qualitative Market Research: An International Journal, vol. 13: 3, 309 – 323 (2010)
17. Alexandris, K., Douka, S., Papadopoulos, P., Kaltsatou, A.: Testing the role of service quality on
the development of brand associations and brand loyalty., Managing service quality, vol. 18: 3, pp.
239-254 (2008)
18. Hall, M., Frank, E., Holmes, G., Pfahringer, B., Reutemann, P., Witten, I.: The WEKA Data
Mining Software: An Update; SIGKDD Explorations, vol. 11: 1 (2009).
19. Witten, I. H., Frank, E., Hall, M. A.: Data Mining: Practical machine learning tools and
techniques, 3rd Edition, Morgan Kaufmann, San Francisco (2011).
20. Hartigan, A.: Clustering Algorithms (Probability & Mathematical Statistics). John Wiley & Sons
Inc (1975).
21. Pelleg, D., Moore, A.: X-means: Extending K-means with Efficient Estimation of the Number of
Clusters. In Proceedings of the Seventeenth International Conference on Machine Learning (ICML
'00), P. Langley (Ed.). Morgan Kaufmann Publishers Inc., San Francisco, CA, USA, 727-734.
(2000).
22. Kotler, P., Armstrong G.: Principles of Marketing, Prentice Hall, 14 edition (2011).
23. Kononenko, I.: 1994, „Estimating attributes: analysis and extensions of Relief‟. In: L. De Raedt
and F. Bergadano (eds.): Machine Learning: ECML-94. pp. 171–182, Springer Verlag.
24. Marko RS, Igor K: Theoretical and empirical analysis of reliefF and RReliefF. Machine Learning
Journal, 53:23-69, 2003.
25. Stokes M, and Visweswaran S.: Application of a spatially-weighted Relief algorithm for ranking
genetic predictors of disease. BioData Mining, 5:20, 2012.
Authors
Ajayeb Abu Daabes is an Assistant Professor in Marketing at Emirates College of Technology, Abu
Dhabi, UAE. She holds PhD degree in Marketing/Branding from Amman Arab University, Jordan, in 2009.
Her Master degree was in MBA/Marketing in 2002 from Jordan University. Her main research interest is
customer behavior, service marketing, brand management, and relationship marketing.
Faten Kharbat is an Assistant Professor in Artificial Intelligence at the Al Ain University, Abu Dhabi
Campus, UAE. She holds PhD degree in computer science from the University of the West of England,
UK, in 2006. Her main research interest is Learning Classifier Systems, knowledge based systems,
applying data mining techniques to marketing, website design, and recently was involved in e-learning and
quality of higher education.

MINING THE RELATIONSHIP BETWEEN DEMOGRAPHIC VARIABLES AND BRAND ASSOCIATIONS

More Related Content

Similar to MINING THE RELATIONSHIP BETWEEN DEMOGRAPHIC VARIABLES AND BRAND ASSOCIATIONS (20)

Recently uploaded (20)

MINING THE RELATIONSHIP BETWEEN DEMOGRAPHIC VARIABLES AND BRAND ASSOCIATIONS