Presentation given to the European Advertising Association's (EAA) International Conference on Research in Advertising (ICORIA) in Berlin, Germany, June 23-25, 2011.
25+ years’ experience (310) 882-6330 Love Spells in Wilmington, DE | black ma...
An Exploratory Investigation of a Brand Equity Model for an Internet Portal Website
1. Eastern Michigan University
Department of Marketing
An Exploratory Investigation of a Brand Equity
Model for an Internet Portal Website
G. Russell Merz, Ph.D., Professor
Department of Marketing
Eastern Michigan University
EMU
1
2. Eastern Michigan University
Department of Marketing
Presentation Agenda
•
Introduction and Literature Review
– Website User Experiences
– Website Brand Development
– Building Branding Theories
•
Theoretical Framework
– Brand Equity Model
– Research Questions
– Hypotheses
•
Methodology/Findings
–
–
–
–
•
Data Collection and Measurement
Analysis Methods
Sample Profile
Structural Equations Modeling (SEM)
Results
Discussion
– Contributions and Implications
– Limitations and Directions for Future
Research
EMU
2
3. Eastern Michigan University
Department of Marketing
Introduction to the Study
•
•
The value of branded websites in ecommerce is without dispute.
Website brand research has evolved
from:
– Descriptions of user experiences
(perceptions) and reactions (attitudes) to
attributes of branded websites; to,
– Prescriptions about how website brand
strategies should be formulated; to,
– Theoretical understanding of how branding
outcomes are developed by websites.
•
This paper contributes to this research
stream in two ways by:
– Proposing a novel approach to measuring
brand constructs such as image and
preference; and then,
– Modeling the constructs within a
comprehensive consumer based brand
equity framework (Keller 2003, 2008).
EMU
Branding
Theories
Prescriptions for
Branding
Strategies
Descriptions of User
Reactions
3
4. Eastern Michigan University
Department of Marketing
Previous Research:
Descriptions of User Reactions to Branded Websites
•
Much of the earlier research described the experiences and
reactions by users to various website design characteristics ,
such as:
– Atmospherics of the website (Eröglu et al 2001, Mummalaneni 2005
and Richard 2005).
– Attitudes toward the website resulting from the brand (Ward and Lee
2000, Balabanis and Reynolds 2001), website attributes (Chen and
Wells 1999, Ho and Wu 1999, Newman et al 2004) and, services
and functions (Van Riel et al 2003).
– Perceived service or website quality (Janda et al 2002, Trocchia and
Janda 2003, Zhang and van Dran 2001).
– In addition, some studies have treated the measurement of website
characteristics as predictors of user satisfaction (Mummalaneni
2005, Trocchia and Janda 2003, Yen and Gwinner 2003 and Zviran
et al 2006) and shareholder value (Rajgopal et al 2001).
EMU
4
5. Eastern Michigan University
Department of Marketing
Previous Research:
Prescriptions for Website Branding Strategies
•
Other studies proposed brand strategy guidelines for websites
based on:
– Generalizations from branding practices based on the Brand
Pyramid (Clarke 2001), benchmark databases (Flores 2004), and
comparative case studies (Ibeh et al 2005, Bergstrom 2000,
Simmons 2007);
– Empirical investigations examining the interplay between brand
image, trust and personality and factors such as website design
(Chang et al 2002, 2003); shopping features (Ha 2004);
customization (Thorbjornsen et al 2002) and levels of website
usage (Thorbjornsen et al 2004).
– Relationships between the summative constructs of brand
attitudes, knowledge and experience and other attitudinal and
behavioral outcomes such as perceived risk, search and
intentions to adopt (Chen and He 2003); website involvement
(Dahlen et al 2003); website satisfaction (Ha and Perks 2005; Na
and Marshall 2005); and revisit and recommendation intentions
(Toms and Taves 2004).
EMU
5
6. Eastern Michigan University
Department of Marketing
Previous Research:
Website Brand Performance Theories
•
Finally, a few studies provide theory-based models for understanding
and explaining website brand performance. For instance:
– The relationships between on-line brand experiences, familiarity and
levels of brand trust (Ha and Perks 2005),
– Website brand trust building features and on-line/off-line brand purchases
(Lee et al 2005),
– The relationship between branded communications tools (newsletters)
and subsequent brand image (Müller et al 2008),
– Brand image dimensions and brand loyalty (Alwi 2009), and
– Brand personality congruence and brand loyalty (Magin et al 2003).
•
However, despite the breadth of the existing literature, a key
component of brand strength missing from the many previous studies
is the concept of brand equity.
EMU
6
7. Eastern Michigan University
Department of Marketing
Theoretical Framework:
Research Objective
•
•
The objective of this study is to build and
test a brand equity model based on the
customer-based brand equity (CBBE)
framework proposed by Keller (2003)
using brand related survey data collected
from an internet portal website.
The intent is to develop a three stage
model that relates two indices of brand
meaning (brand performance and brand
image) to a single index of brand judgment
(brand preferences), and then to two
indicators of brand resonance (brand
loyalty and brand advocacy).
EMU
Keller’s Consumer Based Brand Equity Model
7
8. Eastern Michigan University
Department of Marketing
Research Questions and Hypotheses
Given the findings from the previously reviewed literature, this study
investigates the following three research questions:
•
First, is there evidence that the website brand image is differentially affected
by the perceived brand performance of the website design?
– H1: The index of website brand image is positively explained by an index of
website brand performance.
•
Second, is there evidence that website brand image and perceived brand
performance affects brand preference?
– H2a, b: The index of website brand preference is positively explained by indices of
website (a) brand performance and (b) brand image.
•
Third, as postulated in the CBBE theory is there evidence that brand
judgment affects brand resonance?
– H3a, b: The index of website brand preference positively explains the variation in
two indicators of brand resonance (a) website brand loyalty and (b) website
brand advocacy.
EMU
8
9. Eastern Michigan University
Department of Marketing
Brand Judgment
Brand Meaning
Brand Resonance
Information Seeking
Content
Tangible or
Experiential
Entertainment
Functionality
Security
Socializing
Brand
Performance
Music and Gaming
Personalization
Brand
Loyalty
H2a
Advertising
H1
Brand
Preference
H2b
Customer Care
Trustworthy
Innovative
Unapproachable
EMU
H3b
Brand
Advocacy
Brand
Image
Quality
Distinctive
H3a
Intangible or
Inferred
The proposed model is
constructed from three second
order latent variables and two
first order variables.
10. Eastern Michigan University
Department of Marketing
Methodology:
Measurement and Data Collection
•
•
•
The website that was the subject of this study is a branded internet portal
that provides a full range of services to users such as shopping assistance,
access to information sources such as news and sports, online games, email, etc.
The use of one website was considered sufficient for an exploratory
examination of the hypothesized relationships.
The data was collected during a two week period from website visitors who
were randomly selected and asked to participate through on-network
invitations in a web survey hosted by a data collection firm.
– There were usable 3,127 responses to the invitation.
– All participants were required to be at least 13+ years and have visited the
website in past 30 days.
– No check for non-response bias was done, but because of the randomness of
the survey trigger, it is not believed that the responses are substantially biased.
EMU
10
11. Eastern Michigan University
Department of Marketing
Findings:
Profile of the Sample
•
•
Examination of the demographics
shows that most respondents are
female (54.4%), middle aged (3554, 42%), with household incomes
of between $35K to $110K
(51.2%), and education that
includes some college education
through the bachelor’s level
(59.9%).
The usage characteristics
revealed that 56.5% of users have
used the brand for 5 years or
more, and 55.5% were classified
as heavy users of the website.
EMU
11
12. Eastern Michigan University
Department of Marketing
Methodology:
Analysis and Modeling
First, since brand preferences were measured
using a set of ordinal ranking measures, the
preference dimensions were defined by using
principal components analysis with varimax
rotation.
Second, the entire first and second order
theoretical brand equity model was estimated in a
single step using a latent variable partial least
squares (LV-PLS) to test the hypothesized causal
structure.
Third, the measurement model quality is assessed
by examination of the construct and discriminant
validity.
Fourth, the significance of the hypothesized paths
is evaluated by a bootstrapping routine.
EMU
12
13. Eastern Michigan University
Department of Marketing
Measurement:
EMU
Index of Brand performance: A composite second order
latent measure comprised of five first order latent variables
that capture the perceived performance of the website.
13
14. Eastern Michigan University
Department of Marketing
Measurement:
EMU
Index of Brand image: A composite second order latent
measure comprised of six BAV brand image factors specified by
Lebar et al (2005) developed from twenty four multiple
response items.
14
15. Eastern Michigan University
Department of Marketing
Measurement:
Index of Brand preference: A composite second order latent
measure comprised of four dimensions of brand preferences
derived using factor analysis from ordinal preference
rankings of the branded website compared to other websites.
Brand loyalty: A first order latent containing three items measuring the likelihood that the
respondent will: (1) use the website again, (2) choose the website as primary internet
portal in future, and (3) whether the website brand has earned the respondent’s loyalty.
Brand advocacy: A single item scale asking respondents to rate the likelihood that they
will recommend the website to someone else.
EMU
15
16. Eastern Michigan University
Department of Marketing
Findings:
Structural Equations Modeling
•
•
•
•
•
To test hypotheses a structural equations model (SEM) with latent variables
was estimated using a latent variable partial least squares (LV-PLS)
algorithm (Ringle, et al 2005).
The measurement model in PLS is assessed in terms of item loadings and
reliability coefficients (composite reliability), as well as the convergent and
discriminant validity.
Measures with loadings onto underlying latent variables of greater than 0.7
possess acceptable levels of association with a component (Fornell and
Larcker 1981).
Interpreted like a Cronbach’s alpha for internal consistency reliability, a
composite reliability of 0.7 or greater is considered as an acceptable level of
reliability (Fornell and Larcker 1981).
The average variance extracted (AVE) measures the variance captured by
the indicators relative to the measurement error, and it should be greater
than 0.5 to justify using a construct (Barclay, Thompson and Higgins 1995).
EMU
16
17. Eastern Michigan University
Department of Marketing
Findings:
Measurement Model Discriminant Validity
The square roots of AVE values exceed
inter-correlations of the latent variables.
All Cronbach’s Alphas are > 0.7.
All AVEs exceed minimum of 0.5.
EMU
17
18. Eastern Michigan University
Department of Marketing
Findings:
Indicators of Model Quality
Cross Loadings:
If the cross loadings are smaller and the discriminant
validity test is met, then the structure is acceptable for
exploratory analysis.
Bootstrapping of Path Coefficients:
To obtain the standard errors of the path coefficients a
bootstrapping routine was used. The results are based on
500 cases resampled 500 times.
EMU
18
19. Eastern Michigan University
Department of Marketing
Brand Meaning
Brand Judgment
Brand Resonance
Information Seeking
Content
Entertainment
Functionality
Socializing
Security
Brand
Performance
Music and Gaming
Personalization
Brand
Loyalty
.220
Advertising
.377
(4.41)
(10.45)
.504
Brand
Preference
(14.46)
.279
Customer Care
(5.60)
Trustworthy
Innovative
R2=.142
R2=.188
.417
(11.61)
Brand
Advocacy
R2=.174
Brand
Image
Quality
Distinctive
R2=.254
Unapproachable
EMU
Path coefficients are standardized β (t-statistics in parentheses: all are p ≤ 0.01)
20. Eastern Michigan University
Department of Marketing
Discussion:
Contributions
There are three contributions to the website brand development
literature:
• First, this study takes a holistic approach to understanding the brand equity
of websites. It does this by modeling facets of brand meaning together with
brand preference and brand resonance in a predictive cause and effect
theoretically-based framework. This represents an unique empirical addition
to the website brand literature.
• Second, the model was built using a novel mixed measurement (interval,
nominal and ordinal) approach. This method reduces the potential for
common methods bias, and provides a way to incorporate a wider and
richer source of brand measures into a comprehensive modeling
framework.
• Third, several of the findings within the model support the results of earlier
studies reported in the literature; and because this study uses a
comprehensive brand equity framework, the value of the earlier studies is
enhanced by a better understanding of their role in brand theory
development .
EMU
20
21. Eastern Michigan University
Department of Marketing
Discussion:
Implications
The implications for pure-play websites, such as the one used in this
study, are the following:
• The design characteristics of the website can be considered a strategic
activity that builds brand equity (as reflected in the brand resonance
measures) rather than simply as a media or service delivery vehicle.
• Since perceptions of brand performance influence the overall formation of
brand image, website designers and managers of websites need to
consider if the experiences of users with the website enhance or detract
from desired brand images for the website.
• The effects from both the tangible aspects of website performance as well
as intangible aspects of brand imagery onto the brand preferences formed
by users suggest that website managers must be concerned with both
facets of brand meaning to build preferences and positive evaluations for
their website.
EMU
21
22. Eastern Michigan University
Department of Marketing
Discussion:
Limitations and Directions for Future Research
•
•
•
•
This exploratory study utilized limited and rather abstract measures of the
customer-based brand equity framework espoused by Keller (2003). Future
studies need to consider additional brand measures especially regarding
judgment and emotional responses to brand meaning.
In addition, a fuller set of brand resonance indicators including community
and engagement aspects are needed.
Regarding the respondents to the survey, this study did not examine how
user differences might affect the measurement of and the relationships
between the perceptions and judgments of the brand examined. Although
many of the users in the current sample have been using the website for
some time, their objectives or motivations might be quite diverse. Examining
different segments in the sample may reveal alternative brand equity
configurations reflecting variations in usage rates, conditions and frequency
of use.
Finally, only a single pure-play website was used in this study to examine
the hypothesized relationships. Subsequent studies of website brand equity
need to expand the range of websites as study objects.
EMU
22