International Journal of Market Research

International Journal
of Market Research
Contents
Editorial . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
Viewpoint
Learn to love procurement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
Louise Cretton
Forum
Perspectives on data mining . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
Niall M. Adams
Main papers
Do institutions really influence political participation? Contextual influences . . . . . . . . . . . . . . 21
on turnout and participation in the worlds democracies
Paul Whiteley, Marianne Stewart, David Sanders and Harold Clarke
Consumer-based brand equity conceptualisation and measurement: . . . . . . . . . . . . . . . . . . . . . . . . . 43
a literature review
George Christodoulides and Leslie de Chernatony
The importance of social motives for watching and interacting with
digital television
Steven Bellman, Anika Schweda and Duane Varan
. . . . . . . . . . . . . . . . . . . . . . . . . .
67
Whose design is it anyway? Priming designer and shifting preferences . . . . . . . . . . . . . . . . . . . . . . . 89

Gorm Gabrielsen, Tore Kristensen and Judith Lynne Zaichkowsky
Purchasing behaviour in an online supermarket: the applicability of E-S-QUAL . . . . . . . . . . . 111
Frederic Marimon, Richard Vidgen, Stuart Barnes and Eduard Cristbal
Conference notes131
IJMR Research Methods Forum: Start listening, stop asking, London, 4 November 2009
Co-creating the future Roy Langmaid
Get real: from the viewing facility to the real world Philly Desai
Book reviews139
Don Tapscott Grown up digital: how the net generation is changing your world
Alan Wilson
Iain Ellwood with Sheila Shekar Wonder woman: marketing secrets for
thetrilliondollar company
Eleanor Shaw
International Journal of Market Research Vol. 52 Issue 1
Editorial
Peter Mouncey
IJMR Research Methods Forum:

Start listening, stop asking
The second Research Methods Forum
was held at the Royal Society in London
on 4 November 2009. As in 2008, it
proved to be a very informative and
lively day, with delegates enthusiastically adding their views and challenging the speakers during the two
Speaker Panel sessions. Towards the
end of this issue you will find the first
two Conference Notes from this event.
The remainder will appear in the next
issue (plus a full paper for one session
at some future date). You will also find
reports on the event at www.warc.com
and www.research-live.com. However,
as Editor, and chair of the day, Id like
to share with you my thoughts on the
key messages from the day.
Learning to listen objectively. While
using question-based methods preshapes the agenda for responses,
humans are also not particularly
good at listening objectively. We
have our own agenda when we listen, just as we do when we speak.
Listening does not just apply to
the needs, views and opinions of
consumers and citizens, it applies to
those of clients, other stakeholders
in research projects and all available
evidence. It also means listening
before asking the questions, and
truly listening to the answers when
DOI: 10.2501/S1470785310201016
interpreting their meaning in the

context of the clients goals.
Ethnography and observing methodologies have come of age in
todays world think laterally to
identify new applications. Technology has opened new doors in this
field, but the principle of informed
consent creates new challenges for
researchers to contend with.
Learn to integrate methods and
skills more effectively in meeting
the real needs of clients, and create
intelligence. The days of a qual/
quant divide are over. Researchers
must also learn to work with other
data sources (see the paper by Niall
Adams in this issue) and evidence.
Researchers need to provide clients with a consultancy added-value
service where appropriate to fend
off new low-cost competitors seeking to eat our lunch (see the new
Calls for Papers described below in
this Editorial), and in doing so they
also must demonstrate why their
expertise, more time and budget
delivers real advantage to clients
(see the Viewpoint in this issue).
And its not just new direct competitors we need to fear leading
international brands are harnessing
the people power within social
networks to design and research
advertising without any real input
from marketing services specialists.
Editorial
Some developing fields, such as

deliberative
projects,
expose
researchers to new challenges and
high levels of risk and uncertainty.
These require: a new skill set to
manage very large projects; dealing
with powerful stakeholders who
have opposing vested interests;
grappling with complex and often
contradictory evidence; working
under the media spotlight; planning
for unforeseen eventualities.
Continually strive to find new ways
to more effectively understand
human behaviour, while recognising the rise of people power. This
means learning to relinquish control in the research process and
accept that the researcher is no
longer in a position to manage
the conversation with consumers.
Think participants, rather than
respondents.
The need to develop new methods
to research communications with
consumers. Maybe the days of the
controlled pre-test and orchestrated
campaign are ending. Helping clients understand and leverage word
of mouth will be a field of increasing importance to the research
sector.
Recognise that ethical boundaries
have moved and that the research
sector will be subject to greater
scrutiny by legislators and regulators. Codes and guidance need to
recognise that the definitions of
interview and data are radically
changing in a world of social networks. Dont adopt methods bordering on the illegal, or be pushed
by clients into doing so.
Work together to tackle todays
challenges and share knowledge
to develop new methods and

solutions, being honest about the
issues and ensuring that the process
and debate is transparent.
Debate what we mean by representiveness and the relevance of this in
meeting the needs of clients.
The debate about listening continues
apace elsewhere. For evidence, see Ray
Poynters review of the Online Research
2009 conference (www.esomar.org),
held in Chicago in late October: The
single biggest message from Chicago
was that researchers need to listen,
which in this context means observing.
As Simon Chadwick commented in his
Editors Note in the December issue of
ESOMARs Research World, qualitative
research (collaborative, ethnography
and co-creation) is innovating ahead
of quantitative, concluding: And if
qualitative leads the way, so be it.
Finally, in a recent address at Cranfield
University, Professor Don Schultz
(Northwestern University) argued that
marketers are still primarily outbound
focused, whereas consumers are saying that they no longer need to listen
to choreographed marketing messages,
taking much more notice of other
sources of information. Calls of help
me are readily addressed from elsewhere. Schultz contended that organisations have no processes in place to
listen to or monitor conversations, asking the audience: Who is in charge of
listening within your organisation? He
cited the backlash in Australia to Krafts
attempt to launch an iSnack2.0 brand
extension of Vegemite and the eventual recourse to a co-created solution.
Schultz called for organisations to integrate their functional silos in facing the
market not dissimilar to my fourth
key message above from the Research

Forum. He concluded by warning delegates that the impact of the current
recession is likely to be transformational there is unlikely to be a return
to normality. The research sector also
needs to take note of this message.
And a date for your diary the 2010
IJMR Research Methods Forum will
be held on 2 November at the Royal
Society, London.
Calls for papers

You will see on the outside of the back
cover that were launching four new
Calls for Papers. As in 2009, successful submissions will be published as
and when accepted for publication,
rather than in special themed issues,
as the Board feels that this maintains
a balance of papers within each issue,
hopefully something of interest for all
readers. Two of the themes reflect the
trend towards accountability in the
field of marketing.
The first new theme is the role
of market research in evaluating the
value added by marketing. In terms
of marketing, market research should
be playing a leading role in evaluating
the contribution played by marketing
in helping an enterprise achieve its
key business goals. However, as Binet
and Field point out in their report,
Marketing in the age of accountability (IPA/WARC, 2006), market
research often delivers only intermediary measures, rather than ones related
to those that matter to the finance
director and the board. Also, in this
issue we publish an updated literature
review of research on brand equity
and measurement methods, but will

evidence from market research methods meet the standards required by the
International Standards Organisation
currently finalising ISO 10668, covering brand valuation methods?
This leads naturally on to the value
added to the business by investing
in market research itself, the second
theme. The Viewpoint in this issue
touches on this challenge, and the
author of the recent Morgan Stanley
report (Advertising agencies: cost
relation dilutes recovery prospects,
Morgan Stanley, October 2009) comments that: We are concerned that
market research pricing may remain
under significant pressure even
in a recovery scenario due to pressure from online competition that
is driving down prices (Researchlive, 19 October 2009) what Simon
Chadwick (Client-driven change: the
impact of changes in client needs on
the research industry, IJMR, 48(4))
described as the impact of disruptive
technologies and his call for a deeper
understanding of clients needs. How
does the market research industry
defend its position and demonstrate
the value provided to clients?
The third new theme, Research in
the public sector, reflects the importance of this sector to the market
research industry it is some time since
IJMR focused on this vitally important
field.
Fourthly, we would be interested in
submissions that might demonstrate
how market research could aid content
owners in developing strategies to fight
the pressure for free access in the digital world.
Full details of these new Calls for
Papers can be found at www.ijmr.com.
Editorial
2009 award winners

Finally, Id like to take this opportunity
to publically congratulate John Cromie
and Michael Ewing for winning both
the Silver Medal and David Winton
award for their paper Squatting at the
digital campfire: researching the open
source software community, published
in our Web 2.0 Special Issue (IJMR,
50(5), 2008), and Greg Rogers and
Didier Soopramanien who won the
IJMR Collaborative Award for The
truth is out there! How external validity
can lead to better marketing decisions
(IJMR, 51(2), 2009). The winners were
presented with their awards at Research
Awards 2009, held at the Lancaster
London hotel on 14 December. Full
details are on www.mrs.org.uk.
Viewpoint
There has been a lot of concern
expressed in recent years as procurement departments are becoming
involved in the purchasing of market research services. Louise Crettons
Viewpoint is based on research conducted for the Incorporated Society
of British Advertisers (ISBA). Her key
message is that this trend is inevitable, and researchers need to learn to
live with it. Cretton outlines suggested
strategies for success in this environment, and counsels procurement managers to reciprocate professionally.
Papers
The papers in this issue provide a further perspective on researching voting
intentions, an update on the literature
covering brand equity, why viewers
want to interact with television, the
impact of designer labelling on con-
sumer preferences, and a model measuring service quality in e-commerce.

The Forum article covers data mining.
Influences on political participation

The first paper is a further submission
from the Call for Papers Researching
Voting Intentions. Whiteley (University
of Essex), Stewart (University of Texas),
Sanderson (University of Essex) and
Clarke (University of Texas) explore
the influence of institutional variables
and other contextual variables on electoral turnout; institutional variables in
this context being the electoral system,
size of electoral district and structure of the state. The authors describe
two aims for their research: first, to
extend the understanding of the relationships between institutions and voting to other types of participation
in democracies; second, to compare
and contrast the role of institutions
with other contextual variables that
might plausibly influence participation. They believe that the influence
of institutional factors is overstated as
contextual variables are missing from
estimating models. The data set used
for the analysis was drawn from the
ISSP Citizenship Survey covering 36
democracies and 50,000 respondents.
The authors describe the range of political participation covered in the data,
the national variations, and how voting
emerges as a distinct form of political
participation only weakly related to
other forms of citizen involvement. The
authors also describe the three models
used as the basis for the composite
model used in their analysis of participatory influences: civic voluntarism,
cognitive engagement and social capital. .
The authors conclude that, while insti-
tutions themselves are important, the

resources and motivations they generate among the electorate should not be
ignored. Institutions measure what the
authors call opportunity structures,
but the roles played in these structures by social structure variables and
community-based measures need to be
taken into account as these have important influences on participation.
Defining and measuring brand equity

In 1996 Paul Feldwick published in
JMRS (IJMRs former title) What is
brand equity anyway, and how do
we measure it? (JMRS, 38(2)), which
included a review of the then current
literature on this important topic. Since
then, the brand has become even more
important. Intangible assets comprise
an increasing proportion of company
value, with the brand a key constituent, and (as mentioned above) a new
ISO covering brand valuation is being
developed, which might include the
need for evidence drawn from market research, so this is a very timely
paper. In the call for greater marketing accountability, measures of brand
health provide important indicators
of success, or otherwise, of marketing
strategy. Finally, neuroscience is changing our thinking about how consumers see brands, leading to suggestions
for new measures and measurement
methods. Christodoulides (University
of Birmingham) and De Chernatony
(Universit della Svizzera italiana,
Aston Business School) have updated
Feldwicks literature review, focusing on consumer-based brand equity
(CBBE) and taking into account the
other factors referred to above. One
thing has not changed there is still
no widely accepted definition of brand

equity, although there are commonalities within the main ones in use. As the
authors point out, this leads in turn to
the range of models available to measure brand equity. Within their very
extensive review, the authors discuss
firm-based and consumer-based brand
equity, the former primarily being a
financial value, while the latter covers
the strength of attachment to a brand
and the descriptions of associations
and beliefs consumers have about a
brand. The authors explore the basis
for the latter in detail, together with
the ways to measure CBBE. They argue
that a universal measure is not feasible and that the complexity of brand
equity means that a single measure
is not adequate. Finally, the authors
provide six key conclusions from their
review that should be considered when
measuring brand equity, including a
warning about the danger in a recession
of focusing purely on financial measures. As Jeremy Bullmore points out
in Apples, Insights and Mad Inventors
(Wiley, 2006): Of course, in a legal
sense, the company owns the brand.
But for a company to feel that it owns
its brands is to tempt it to believe that
it has total control over them: and it
does not.
Motivations for interacting with digital

TV
As described by the authors of this
paper Bellman, Schweda and Varan
(Interactive
Television
Research
Institute, Murdoch University, Perth,
Australia) digital (or i) TV provides
viewers with further opportunities to
engage with media channels, adding to
all those available through the inter-
Editorial
net. However, the authors contend

that the television industrys assumptions as to why consumers wish to
interact in this way remain rooted
in traditional attitudes to TV viewership and ignore the impact of social
motivations, already established as key
factors in the growth of media such as
Facebook and MySpace. They argue
that the new opportunities provided
by iTV require a rethink of the factors measured in viewership research.
The authors discuss the development
of iTV and why the UK provided the
most appropriate market for undertaking their research. They describe the
limitations of viewer segments derived
in earlier research that ignored social
motivations. In order to address this
deficiency, the authors describe why
the motivations identified that underpin consumer need and product use
could be aligned with those relating to
TV viewership. They describe in detail
the development of the measures used
and how they were applied in a CATI
survey of 867 digital TV households
in the UK. Their findings demonstrate
the importance of social motivations,
similar to the characteristics for other
digital channels, and identify why these
matter to consumers. The authors conclude with an example of how their
findings might be applied in engaging
with different types of consumer, leading to a call for further research exploring the application in the design of iTV
programmes and advertising.
Does designer labelling influence

consumer choice?
The third paper, by Zaichkowsky
(Simon Fraser University, Vancouver),
Tore Kristensen and Gorm Gabrielsen
(Copenhagen Business School), explores

the influence of design, designer name
and brand on consumer preference.
Their research is based on conducting
a conjoint-based survey among shoppers to investigate attitudes towards
three designs of rice paper table lamps:
a cheap Chinese import (price 10),
an IKEA product (price 20) and one
designed by a leading internationally
recognised sculptor (price 185). The
key research question addressed by
the authors was whether, and how,
consumer preferences might be influenced by information about the brand
and designer. The research method
and questionnaire content is described
in detail, including the three studies
within the overall programme covering
blind product evaluations: without
any cues; adding in brand; adding in
designer information. The findings
show that, in the blind tests, the
designer version was preferred, and
information about the design origin
anchored that preference. The authors
include a useful literature review covering this field of research, and conclude with a discussion on what might
constitute good design, and whether
other research methods, such as neuroscience, might identify other triggers
that influence preferences related to
design.
Measuring the online shopping

experience
Models for measuring service quality in e-commerce transactions will
become increasingly important as
consumers use this channel for an
increasingly diverse range of purchases,
volumes of purchases grow and the
increasingly competitive environ-
ment leads organisations to actively

look for ways to gain advantage over
competitors. The authors of the final
paper Marimon, Vidgen, Barnes and
Cristobel (Universitat Internacional
de Catalunya) have applied the
ESQUAL model to investigate factors
contributing to customer satisfaction
when using a supermarket e-shop. As
the authors describe, the E-S-QUAL
model was developed from the established SERVQUAL model, specifically
to measure service in online transactions, but as they point out in their
literature review, this variation of the
model has attracted criticisms in some
papers. The authors also contend that
at the time of their research, few if any
studies had established an empirical
link between service quality and actual
sales in the online environment. Six
hypotheses were developed to test in
the research covering efficiency, system
availability, fulfilment (reliability), privacy, perceived value, loyalty relating
to website design, purchase levels and
value (sales). The authors describe the
modified form of E-S-QUAL developed for the survey, conducted among
online customers of a leading Spanish
supermarket. In their conclusions, the
authors believe that they have, firstly,
confirmed that the model works in
this particular online environment and,
secondly, established for the first time
a link between service quality and
actual purchases in online retailing,
using actual data (provided at customer
level by the supermarket) rather than

customer-reported data.
Forum
Data mining: discovering unknown and
exploitable information
The Forum article is based on the
opening overview presented by
Adams (Imperial College, University
of London) as the mornings chair at
the Association for Survey Computing
conference, Data Mining and its
Applications in Market Research and
Statistics, held at Imperial College in
April 2009. This conference covered an
increasingly important topic to market
researchers, and Adams article provides
a useful guide to the general principles
and methods in use today, including the
latest definition of these methodologies, and sources of more information
on the topic. The author describes the
evolution of data mining as a secondary analysis of large data sets and the
concerns about data dredging, which
Adams believes is the responsibility of
analysts, not the software. Within the
article, readers will find the concept of
data streams described, plus the two
main modalities of global models and
local pattern detection, together with
the data-mining process and commonly
used software packages. The author
concludes with a discussion about the
relationship with market research, and
its application to analysing internetsourced data.
www.ijmr.com
International Journal of Market
Research Online
Subscribers to IJMR benefit from instant
access at www.ijmr.com
Key features:
Access to all medal-winning MRS papers
IJMR.com archive
All articles from the previous 12 months
Searchable abstracts of all IJMR articles since 1990.
Market research abstracts
View developments in research from around the world. This includes

over 14,000 market research abstracts published since 1963.
Sources monitored include:
Journal of Consumer Research, Journal of Marketing, Journal of Marketing Research,
Journal of the American Statistical Association, Marketing Research, British Journal of
Sociology, Canadian Journal of Marketing Research, International Journal of Public
Opinion Research, Marketing Intelligence and Planning and many more
Full members of the Market Research Society can access a

12-month archive at www.ijmr.com for only 102 plus vat.
For subscription enquiries please contact:
subscriptions@marston.co.uk or call +44 (0) 1235 465 574
IJMR is published by Warc
Viewpoint
Learn to love procurement
Louise Cretton
DOI: 10.2501/S1470785310201028
The growing attention paid by procurement professionals to the buying

of market research is, I would suggest,
a force for good. It may be the catalyst
we need to get our house in order and
to achieve some of the objectives we
have discussed for many years now,
such as contributing at the top table.
I know that procurement is a dirty
word in some quarters, that it signals a price-driven market, reducing
creativity, demoting our services to a
commodity offering. Cries of we are
different, they dont understand the
complexity of what we do are heard.
They are the same cries that were heard
in the 1980s and 1990s in advertising
and media agencies. Magic and Logic
by Marilyn Baxter is a great little book
produced by ISBA, the advertising and
media agency association the IPA, and
the procurement trade body CIPS, in
response to those concerns, which seeks
to explain the conflicting interests.
Media and advertising agencies
learned not just to accept the attentions of the procurement function, but
to rethink their business model around
the notion of representing and justifying their value to clients in a confident
way. Market research is one of the last
areas of marketing services to begin to
be scrutinised by procurement. However we feel about it, procurement is
on the march and we cannot resist
2010 The Market Research Society
its influence. It is surely far better to

understand what it is it wants and to
engage with it in a productive manner.
As a result of a piece of industry
research initiated by ISBA, I have heard
at first hand the agenda of a sample of
procurement professionals and been
able to assess its impact. Some large
companies have had a market research
procurement function for many years,
but many others have introduced it
only relatively recently. In almost all
companies this function reports to
finance and to the FD. So, yes, there is
bound to be interest in managing costs.
However, I would suggest that, with
the finance director at the head of the
reporting structure, procurement personnel are tapped in to the achievement
of corporate strategic goals and have
broader influence perhaps with the
corporate structure than the marketing
and market research/insight functions.
Cultivating rather than trying to ignore
or avoid procurement may help us to
achieve the greater strategic influence
we believe we should have.
At present, it appears that procurement in the area of market research is
in its infancy in many companies. Procurement professionals are focused on
two main targets: managing cost/value
to the business and establishing an
appropriate roster of agencies to meet
business information needs. The fact
that they have measurable targets does
not understate the determination of
these procurement personnel to achieve
them for the good of their companies.
If you are reading this sitting in an
agency, it is worth thinking about what it
Viewpoint
is that justifies your place on a company

roster (many report the desire to reduce
the size of their rosters to a manageable number with whom they can build
stronger relationships), and whether you
are as transparent and honest as you
might be about your costs. Indeed, could
you break down your costs and explain
them to someone outside of the MR
function in such a way that they understand the value you offer?
Procurement wants to understand
your business and you can help this
process. Procurement professionals
know that they need to learn about our
world, how it is structured, what constitutes quality, where the parameters of
our standards lie. We can help them by
engaging with them, not by pretending
we wished they did not exist.
This is where the client-side researchers and insight people can help, too.
They can facilitate the relationship with
procurement. There are three parties
involved in the buying of research now:
the client-side researcher, the procurement person and the agency. We are
being offered the chance to offer our
services on a professional commercial
footing, where trust and reciprocity of
interest characterise the relationship.
Before I conclude, I want to make
one particular plea: that we learn to
10
value our people. By that I mean the

intellectual contribution they make to
a project. That means being honest
about the time they are likely to spend
thinking on a project, and the skills
and experience they bring to make a
difference to the clients business. This
honesty and transparency will help
procurement focus on value not costs,
difference not commodity. Longer term
it may affect recruitment; shorter term
we need to look at our business models.
The reality of life is that procurement
will only grow in influence. We must
learn to be more effective in our commercial engagement; we must embrace
concepts such as service-level agreements, rate cards and fully transparent
costing models.
Procurement professionals must
play their part, too. Auctions, endless
bureaucracy, even aggression, have not
helped their cause. We are fond of
stating that ours is a creative industry, disciplined by standards. Procurement departments will believe us, and
respect us, only if we give them the
help and the tools to fully appreciate
the value we are able to bring to solving business problems.
Louise Cretton is an Independent
Market Research Consultant.
fORUM
Perspectives on data mining
Niall M. Adams
Imperial College London
As a data analysis technology, data mining has matured to the extent that there
are now a number of sophisticated commercial software packages available. The
purpose of this article is to explore what data mining has become, its relationship
to statistics and its relevance in market research.
Introduction
To begin a discussion of data mining (DM) it is instructive to provide
a couple of textbook definitions. According to Hand et al. (2001), data
mining is the analysis of (often large) observational data sets to find
unsuspected relationships and to summarize the data in novel ways that
are both understandable and useful to the data owner. Witten and Frank
(2005), authors of the public domain data mining software WEKA, say
that DM is the extraction of implicit, previously unknown, and potentially
useful information from data. These are modern definitions, identifying
discovery of unknown and exploitable information as the key element.
In the early days, DM was frequently pitched as a collection of tools
for secondary analysis of large data sets. This presentation emphasised the
benefits of making extra use of data collected for some primary analysis
purpose. For example, the primary purpose of recording supermarket
transactions is accounting and stock control. However, the large collection
of data so acquired may, when suitably analysed, provide new insights
into customer behaviour and preferences that can be profitably exploited.
During this formative period there was some accusation that DM
Received (in revised form): 29 June 2009

DOI: 10.2501/S147078531020103X
11
Forum: Perspectives on data mining
implemented, or encouraged, data dredging repeated searches for

statistically significant results in large data sets. While such concerns were
(and remain) justified, responsibility for such abuse lies with the analyst,
not the tools.
This secondary data analysis view is still relevant, in diverse areas
such as analytic customer relationship management and analysis of large
astronomical databases. However, DM is now a mature technology,
suitable for a variety of primary and secondary analysis tasks on
large data sets. Applications of DM are diverse, including science (e.g.
bioinformatics, particle physics, remote sensing) and commerce (consumer
finance, customer relationship management, fraud detection).
Since modern data mining is essentially a type of data analysis, an
obvious question presents itself: Is data mining simply statistics? In
fact, these can be seen as intersecting disciplines (Hand 1999), with
differences of emphasis, application and interpretation. Statistics and DM
will be compared and contrasted throughout the following discussion.
Interestingly, a recent book on data mining and market intelligence (Chiu &
Tavella 2008) includes separate chapters on statistics and data mining. The
statistics chapter is predominantly concerned with a review of probability
concepts, while the introductory DM chapter includes a description of
the DM process, followed by a list of data mining techniques, including
linear regression, principal component analysis, correspondence analysis
and analysis of variance. Most classically trained statisticians would take
issue with the provenance of this material.
The origins of DM can be found in computer science and the study of
databases. As such, much of the terminology and jargon of data mining has
the flavour of computer science. Wasserman (2004) gives a lookup table
relating data mining and statistical terminology. DM sometimes remains
notionally close to the data storage infrastructure. For example, many
DM textbooks include information about data warehouses, databases and
query languages.
DM can be conceived of as a process of applying algorithms to large
data sets. First, let us consider the data aspect.
Data
DM is fundamentally concerned with large data sets. Hand et al. (2000) cite
a number of large data sources, including credit card transactions, retailer
customer transactions, telecommunications records and oil exploration.
More recent examples include data collected by the Large Hadron collider
12
at CERN and the giant data sets handled daily by major search websites
such as Yahoo! and Google. The numbers associated with such data are
impressively vast. The pertinent observation here is that the data are
acquired and stored in a fully automatic manner.
DM is concerned with automatically obtained observational data.
Sometimes these data might refer to a whole population all customers
of a bank, say. Many organisations retain a data warehouse for all their
electronic data. Such warehouses contain multiple databases, which
provide data for mining exercises. Analysing population data means
that DM is not concerned with inference in the usual statistical sense.
Moreover, statistics has wider concerns related to data, including the
design of experiments and data-collection issues arising in the form of
survey methodology.
An interesting characteristic of many modern DM applications is that the
data can be conceived of as a data stream a potentially unending sequence
of data arriving at high frequency, generated by a stochastic mechanism
subject to unknown and abrupt change. Plastic card transactions provide
a compelling example of a data stream cards are swiped through a
point-of-sale terminal or ATM followed by an authorisation process.
This authorisation process involves some fraud filters, often predictive
models. A bank may process thousands of transactions per second so
there is no chance to update the filter as data arrive. Moreover, the nature
of the populations is subject to change since fraudsters adapt to defences.
The usual approach to handle streaming data is simply to use as small
a window of historical data as needed. Some modern research attempts
to construct streaming (incremental updating and adaptive) versions of
conventional data analysis tools.
The observation that populations can change over time is really about
data quality, an issue we now address. Like any data analysis, data quality
is a critical issue. Despite automatic collection, data used in DM analysis
can be prone to a variety of distortions: missing values, erroneous values,
temporal aspects, and so on. As with any data analysis exercise, the
analyst ignores data quality issues at their peril. Perhaps a more interesting
aspect of data quality in DM concerns selection bias related to entry to
the population. One example is given by reject inference in credit scoring
(e.g. Mays 2005), in which entry to the database is contingent on being
awarded a loan, in this case inducing empty regions in the data space.
Now, let us consider the modalities of data-mining methods. We can
identify two: global models, and pattern detection and discovery.
13
Data mining: global models

Much DM analysis is concerned with providing a global description of
a data set. This includes multivariate data activities such as regression,
classification and clustering. In the guise of segmentation, clustering
should be especially familiar to market researchers. In part, this global
model aspect of DM can be viewed as statistics writ large, with changes in
emphasis and interpretation. As noted earlier, DM has its own distinctive
terminology that betrays this emphasis. For example, DM may refer to
a neural network algorithm. In contrast, a statistical view might regard
a neural network as a model (a mapping from one set of variables to
others, with parameters to be determined from data). Within this view,
the statistician may consider various principles of estimation (frequentist,
Bayesian, etc.) to consider how to obtain these parameters. An algorithm
may then be deployed to instantiate this estimation. The emphasis on
algorithms in DM is natural: DM requires algorithms capable of handling
very large data sets so scalability and other computational aspects
become a central issue. Much data-mining research is concerned with
probing convergence properties of algorithms, for exactly this reason.
Many DM global modelling activities are concerned with prediction, and
a wide variety of linear and non-linear prediction methods are available
in DM software, including neural networks, tree induction, bagging, and
ensemble methods of model combination. The pre-eminence of non-linear
prediction means the model criticism central to much statistical work
involving interpretable models is infrequent in DM. Many non-linear
models may give near equivalent predictive accuracy via very different
mappings, but the mapping itself is of no importance. There have been
some efforts to provide tools that describe the structure of these non-linear
mappings for example, using a decision tree to approximate a neural
network. The merits and reliability of such procedures are unclear. Some
caution is needed in interpreting the predictive results of DM modelling
if the data make up a population, the generalisation of predictive models
beyond this population may questionable. This general lack of inferential
aspects in DM results primarily from the structure of the data.
Data mining: local patterns

Hand et al. (2000) refer to the second modality of data mining as pattern
detection and discovery (PDD). This activity refers to a collection
of approaches whose objective is to identity unusual structures or
relationships in local neighbourhoods of the data. These structures or
14
relationships are generically patterns and we can contrast detection for

patterns of known type (cf. supervised learning) or discovery of unknown
structures (cf. unsupervised learning). The distinction of PDD from global
models is not total. For example, perhaps the simplest version of PDD is
examination of model outliers, as in regression diagnostics, to identify
unusual observations. However, in PDD, the object of interest is exactly
the unusual observations in the hope that their unusualness is useful and
informative. Some DM examples include: finding low probability objects,
finding shapes in time series, and structures in sequences and association
rules.
Association rule analysis, a non-statistical approach for pattern detection,
is concerned with finding the most common combinations of categorical
variables in the database. Agrawal et al. (1993) introduced the problem
of finding association rules and it has since spawned a huge literature. In
the special case of binary variables, association rule analysis is referred to
as market-basket analysis (Hastie et al. 2001). Association rule analysis is
very characteristic of data mining, where ideas from theoretical computer
science have led to many sophisticated and novel algorithms.
Collaborative filtering (going back to Goldberg et al. 1992) represents
another successful example of what might be deemed non-statistical
DM algorithms. Collaborative filtering refers to the activity of searching
multiple data sources for useful patterns. The recommender systems
deployed by many commercial websites, such as Amazon, are a good
example of collaborative filtering algorithms, where the collaborative
aspect relates to the use of data from similar customers to produce the
recommendation.
The challenging part of PDD is that, by its very nature, all of the data set
has to be used. This is in contrast to the DM model building activity, where
random sampling from the data set, to facilitate manageable computations,
can be defended using statistical arguments. In pattern search, sampling
the data set may mean that the local structure being sought is missed.
The fundamental problem with PDD is that we can always find patterns.
This problem is familiar in statistics, with the problem of multiple
comparisons. Conduct sufficient hypothesis tests and we will always get
a significant result. Ignoring computational difficulties, PDD can provide
procedures for generating candidate patterns, but there will be many,
and ranking procedures are crucial to manage the demand on the human
analyst.
Some diverse examples of PDD that we have been involved in include:
finding distinct types of repayment behaviour in unsecured personal loan
15
accounts (Adams et al. 2001), detecting cheating in student continuous

assessments (Hand et al. 2005) and detecting fraud in plastic card
transactions (Weston et al. 2008).
The data mining process

Modern DM often involves more than specific algorithms and dataanalytic procedures. Like any data-orientated activity, there will be a
need for pre-processing. Moreover, results of one analysis often suggest
refinements, which make the DM analysis an iterative process. This is
reminiscent of the iterative character of statistical modelling exercises.
The whole process of data selection, processing and analysis, and
iteration, is sometimes called knowledge discovery in databases (KDD).
Commercial DM software attempts to provide graphical interfaces for
managing this whole process within which modular steps involving data
transformations, algorithms, and so on, can be chained together. This is
very useful functionality for the complexity of many modern DM studies.
However, without careful consideration of the validity of each step, such
capabilities raise the possibility of garbage in, garbage out. There is a
slight tendency to suggest that it is not important to fully understand the
DM algorithms deployed during the analysis process. The merits of this
can be argued both ways. For example, one does not need to know how
the components of a car engine work in order to drive to work. On the
other hand, being able to diagnose and jury-rig a repair is useful on a dark
night. This really illustrates the attitude of DM practitioners towards risk,
in contrast to the generally risk-averse statistical community.
DM is often an activity of two steps: model construction followed by
deployment. In this view, a data set is selected for the specific purpose of
model building. For global models, often used for prediction, random subsampling may be adequate for constructing a data set. This has significant
merit in reducing the computational burden of model building. The final
model may then be repeatedly deployed on the whole database, as well
as new data. As noted above, however, PDD must operate on the whole
database directly.
There have been attempts to properly formalise the data-mining process
into models, procedures and standards. Useful outputs of these attempts
are simple diagrams that convey the steps in the process. Perhaps the most
widely distributed is CRISP-DM (Cross Industry Standard Process for
Data Mining) (Shearer 2000), which has been developed by a consortium
of businesses for the entire data-mining community. At the highest level,
16
the CRISP-DM reference model has six interacting tasks, including data
preparation, modelling and deployment. The process model associated
with the commercial SAS Enterprise Miner is called SEMMA, an acronym
for Sample, Explore, Modify, Model, Assess. Unsurprisingly, DM process
models and standards tend to be very similar. While such models provide
a guide and checklist, like process models for statistical analysis, they are
no substitute for experience and creativity. Moreover, when DM is framed
as an exploratory activity with the objective of finding unexpected and
valuable structures, it will not readily admit to formalisation.
Finally, it is worth noting that there are always humans in the DM
process loop, including the ultimate recipient of the discovered knowledge.
Like statistical modelling, involvement of domain experts is crucial for
success.
Software
There are both commercial and public-domain DM software packages. The
big commercial players include SAS Enterprise Miner and SPSS Clementine.
Important public domain software includes the R statistical package (R
Development Core Team 2006) and WEKA (Witten & Frank 2005).
The public-domain software often includes more up-to-date algorithms,
and provides more flexibility in their deployment and combination.
Commercial software usually provides a greatly enhanced graphical user
interface with functionality for DM process management, very flexible
data interoperability and connectivity to database management systems.
In this latter regard, certain DM software vendors are moving towards
in-database data mining. This endeavour, which involves executing DM
algorithms direct on the database, rather than the extraction of data
mentioned above, is a response to a perceived customer need for more
dynamic deployment of DM. However, the utility of such in-place analysis
will depend very much on the application.
Marketing
Much traditional market research involves carefully designed surveys. It is
not clear that DM can add much to the primary analysis of such surveys,
beyond the statistical tools presently in use. Prospects for secondary analysis
of large collections of survey data are also unclear, for a number of reasons
including the problems of data integration (different variables across
surveys) and selection effects (different target populations across surveys).
17
The better prospect for data mining in market research relates to new
technology, specifically the internet. This includes analysing WWW traffic
and click-stream data, e-commerce and search marketing (e.g. Chiu &
Tavella 2008, Chapter 12). These areas all involve large amounts of
complex data, collected in an unstructured and automatic manner. DM is
well suited to such areas.
Conclusion
Data mining is a mature data analysis technology that has considerable
intersection with statistics. Characteristic of data mining is a necessary
stress on algorithmic aspects and a preference for prediction. Data mining
encourages novel exploration of data much more than statistics. Tools to
automate the whole data-mining analysis process are useful, but, like any
tool, should always be used with caution. Opportunities provided by the
World Wide Web offer good prospects for the application of data mining
in market research.
Acknowledgements
I am grateful to an anonymous reviewer for useful comments, and to
David Hand for interesting feedback on this paper.
References
Adams, N.M., Hand, D.J. & Till, R.J. (2001) Mining for classes and patterns in behavioural
data. Journal of the Operational Research Society, 52, 9, pp. 10171024.
Agrawal, R., Imielinski, T. & Swami, A. (1993) Mining association rules between sets of items
in large databases. Proceedings of the ACM SIGMOD Conference on Management of Data
(SIGMOD 93), ACM Press, pp. 307328.
Chiu, S. & Tavella, D. (2008) Data Mining and Market Intelligence for Optimal Marketing
Returns, Burlington, MA: Butterworth-Heinemann.
Goldberg, D., Nichols, D., Oki, B.M. & Terry, D. (1992) Using collaborative filtering to
weave an information tapestry. Communications of the ACM, 35, 12, pp. 6170.
Hand, D.J. (1999) Statistics and data mining: intersecting disciplines. SIGKDD Explorations,
1, pp. 1619.
Hand, D.J., Adams, N.M. & Heard, N.A. (2005) Pattern discovery tools for detecting cheating
in student coursework. In K. Morik, J.-F. Boulicault & A. Siebes (eds) Local Pattern
Detection. Lecture notes in Computer Science 3539 (Springer), pp. 3950.
Hand, D.J., Mannila, H. & Smyth, P. (2001) Principles of Data Mining. Cambridge, MA:
MITPress.
Hand, D.J., Kelly, M.J., Blunt, G. & Adams, N.M. (2000) Data mining for fun and profit.
Statistical Science, 15, 2, pp. 111126.
18
Hastie, T., Tibshirani, R. & Friedman, J. (2001) The Elements of Statistical Learning. New
York: Springer.
Mays, E. (2005) Handbook of Credit Scoring. Chicago, IL: Financial World Press.
R Development Core Team (2006) R: a language and environment for statistical computing.
RFoundation for Statistical Computing, Vienna, Austria (http://www.R-project.org).
Shearer, C. (2000) The CRISP-DM model: the new blueprint for data mining. Journal of Data
Warehousing, 5, 4, pp. 1322.
Wasserman, L. (2004) All of Statistics: A Concise Course in Statistical Inference. New York:
Springer.
Weston, D.J., Hand, D.J., Adams, N.M., Whitrow, C. & Juszczak, P. (2008) Plastic card fraud
detection using peer group analysis. Advances in Data Analysis and Classification, 2, 1,
pp.4562.
Witten, I.H. & Frank, E. (2005) Data Mining: Practical Machine Learning Tools and
Techniques with Java Implementations (2nd edn). San Francisco, CA: Morgan Kaufmann.
About the author

Niall M. Adams is Reader in Statistics in the Department of Mathematics,
Imperial College London. He has methodological research interests in data
mining and pattern recognition, and application interests in consumer
finance and cell biology.
Address correspondence to: Niall M. Adams, Department of Mathematics,
South Kensington Campus, Imperial College London, London SW7 2AZ,
United Kingdom.
Email: n.adams@imperial.ac.uk
19
The pre-eminent source for

compiling and disseminating
advertising knowledge.
Les Carlson, Clemson University
NEW
ANNUAL SPECIAL
ISSUE GIVING
YOU OVER 150
MORE PAGES
Enhance your
understanding of
advertising and
marketing
communications
Gain vital insight

into the latest
theories, advances
and development in
the field
Pinpoint gaps in
your knowledge
and identify areas
for further research
To subscribe visit
www.internationaljournalofadvertising.com/358
Do institutions really influence political

participation?
Contextual influences on turnout and
participation in the worlds democracies
Paul Whiteley
University of Essex
Marianne Stewart
University of Texas at Dallas
David Sanders
University of Essex
Harold Clarke
University of Texas at Dallas
This paper examines the influence of institutions and other contextual variables
in a set of individual-level models of political participation, using a multi-level
modelling strategy. It uses data from Citizenship Survey of the International
Social Survey Programme conducted in 2004, to model relationships in many
of the worlds democracies. It examines the effects of variables that have been
shown to be important in aggregate-level models of turnout, such as the effective
number of parties, the distortions in representation associated with the electoral
system, and the size of districts. It compares the institutional measures with other
contextual variables that arise from rival models of individual-level political
participation. The institutional variables have a modest impact on individuallevel turnout, but their impact is much less important in relation to other types
of participation. For the latter, non-institutional contextual variables arising from
models of political participation appear to be more important.
Introduction
There has been a good deal of research into the determinants of electoral
turnout in a variety of different countries and this has identified a number
of important factors that explain why people vote in aggregate models.
Received (in revised form): 2 November 2009

DOI: 10.2501/S1470785310201041
21
Do institutions really influence political participation?
Several of these analyses have emphasised the importance of institutions,

such as the electoral system, the size of electoral district and the federal
structure of the state for explaining turnout (Owen & Grofman 1984;
Jackman 1987; Jackman & Miller 1995; Gray & Caul 2000; Farrell
2001; Mueller 2003). Institutional variables are not the whole story,
however, since the characteristic of specific elections, such as the closeness
of the contest and the mobilising activities of parties, appear to play an
important role (Cox & Munger 1989; Matsusaka & Palda 1993; Seyd &
Whiteley 1992; Gray & Caul 2000).
There is a good theoretical basis for explaining why these institutions
might influence voting behaviour. They help to constrain the range of
possible outcomes in voting models and thereby make the calculus of
voting more tractable. One of the oldest established results in the spatial
theory of elections is that political parties will try to get ideologically close
to the voter located in the median position on the leftright ideological
scale (Downs 1957). This is because most voters can be found at this point
of the scale, and this result imparts stability and predictability to electoral
politics, in effect creating an equilibrium. Unfortunately, it does not apply
if the ideological space is larger than two dimensions for example, if there
is a leftright scale, a second authoritarianlibertarian scale and, finally,
a materialistpost-materialist scale. In this case formal theory shows that
no stable equilibrium will necessarily exist, and so parties will constantly
change issue positions in order to try to outflank each other, and voters
will get confused about where parties are placed (Black 1958; Arrow
1951; McKelvey 1976; Schofield 1978). Thus a plausible explanation
for the absence of excessive instability in elections is the existence of
institutions, including practical restrictions on the range of issues that are
important in elections to constrain outcomes, promote stability and make
individual calculations of the costs and benefits of voting possible (Shepsle
1979; Shepsle & Weingast 1987). Thus there are good reasons for thinking
that institutions are important for sustaining democratic participation.
There are two aims to this paper. The first is to extend the analysis of the
relationship between institutions and voting to other forms of participation
that occur in contemporary democracies. If electoral systems influence
voting behaviour then it is a plausible hypothesis that they can affect other
forms of participation as well. A second aim is to compare and contrast
the role of institutions with other contextual variables that might plausibly
influence participation. These are the contextual-level counterparts of
individual-level variables known to be important in participation models.
While institutions may influence political participation, their role may be
22
relatively minor in comparison with rival contextual variables. In fact it is

possible that institutions appear to be important in this literature because
other contextual variables have been omitted from estimating equations.
As always, contextual factors have to be judged in relation to a wellspecified individual-level model of political participation. So the analysis
is developed with the assistance of three rival models that have been
extensively discussed in the participation literature. These are the civic
voluntarism, the cognitive engagement and the social capital models. This
paper uses survey data from the ISSP Citizenship Survey conducted in 36
advanced industrial democracies with more than 50,000 respondents.1 The
analysis starts with an examination of the structure of political participation
in these countries. This is followed by a discussion of the rival models
and an analysis of the effects of these models on political participation.
Aggregate-level covariates are then introduced into the individual-level
analysis, based on institutional and theoretical considerations relating
to the individual-level models. The methodological approach involves
estimating multi-level models of participation (Snijders & Bosker 1999;
Raudenbusch & Bryk 2002).
Political participation in the worlds democracies

Figure 1 includes the percentage of respondents who participated in nine
different political activities over the previous 12 months, in the countries
covered in the Citizenship survey. With the single exception of voting,
these activities all involve a minority of citizens, and most only small
minorities. Figure 1 provides a fairly conservative estimate of participation
since it counts only respondents reporting these activities in the previous
12 months. The ISSP survey identifies individuals who report participating
at some point in the past and, not surprisingly, this identifies a larger
number of individuals.2 This list does not of course exhaust the range of
political activities that is undertaken by citizens. There are no data on
electoral campaigning or standing for public office, for example, and so it
does not cover all forms of political participation. However, the list does
capture a broad range of political activities that underpin civil society and
modern democratic politics.
There is considerable cross-national variation in the participation
measures. To illustrate this point, the variations in consumer participation
The ISSP survey was conducted in 37 countries, but missing data prevent the inclusion of South Africa in the
analysis.
2
For example, 26% of respondents report boycotting products at some point in the past and 24% report taking
part in a demonstration.
1
23
80
70
60
50
%
40
Key:
Media: contacted or appeared in the media to express views
Protest: participated in a demonstration
Contact: contacted a politician or civil servant to express views
Meeting: attended a political meeting or rally
Party: been a member of a political party
Boycott: boycotted or bought goods for political, ethical or environmental reasons
Donate: donated money or raised funds for a political activity
Petition: signed a petition
Vote: voted in the last national election
67.4
30
20
10
3.7
0
19.6
19.9
Donate
Petition
15.9
Media
6.4
7.7
8.3
Protest
Contact
Meeting
9.9
Party
Boycott
Vote
Note: All activities relate to the previous year.

Source: ISSP Citizenship Survey, 2004 (N = 51,046)
Figure 1 Participation in 36 countries in 2004
that is, boycotting and buycotting goods appear in Figure 2. This

form of participation is growing across the democratic world, but it
remains very marginal in ex-communist countries like Bulgaria, Poland
and Russia. At the other end of the scale, large numbers of individuals
get involved in these consumer-related forms of participation in the rich
long-established democracies like Canada, Australia and Switzerland.
Another example that is particularly important for the social capital
model relates to variations in voluntary activity across the democratic
world. This appears in Figure 3, and confirms the pattern observed in
Figure 2 that former communist countries are much less likely to have
volunteers than the long-established democracies like the US, France and
Canada.
The underlying latent structure of participation in these countries is
identified in Table 1, which contains a principal components analysis of
the nine items. This analysis shows that there are two underlying factors.3
There is a general participation factor associated with the first principal
component, which loads highly on all the measures, except party
membership and voting, and it explains about one-third of the variance.
Thus individuals who sign petitions are also quite likely to boycott goods,
The table contains only those factor loadings that exceed 0.49 in magnitude.
24
0
Bulgaria
Philippines
Hungary
Russia
Poland
South Africa
Brazil
Chile
Cyprus
Latvia
Czech Republic
Slovenia
Portugal
Slovakia
Mexico
Japan
Taiwan
Israel
Uruguay
Venezuela
Spain
Ireland
South Korea
Netherlands
Britain
United States
Norway
Finland
Germany
France
Denmark
New Zealand
Sweden
Australia
Canada
Austria
Switzerland
Bulgaria
Russia
Hungary
Poland
Japan
Brazil
Cyprus
Portugal
Mexico
Czech Republic
Latvia
Uruguay
Chile
Taiwan
Austria
Germany
Spain
Slovakia
Venezuela
Philippines
Sweden
Britain
Slovenia
Israel
Finland
Switzerland
South Korea
Netherlands
Ireland
Norway
United States
Australia
France
New Zealand
Denmark
Canada
50
40
30
20
10
Figure 2 Cross-national variations in boycotting of goods for political reasons
40
30
20
10
Figure 3 Cross-national variations in voluntary activities
25
Table 1 Principal components analysis of the participation indicators

Indicator
General participation
Party related participation
Petition
Boycott
Protest
Meeting
Contact
Money
Media
Party
Vote
0.77
0.76
0.60
0.50
0.60
0.66
0.56
__
__
__
__
__
0.59
__
__
__
0.86
__
Eigenvalues
Percent of variance explained
3.4
32.9
1.1
17.3
to contact the media and to donate money to political organisations.

The second factor is much more specialist in character, and really.
relates to party membership. Not surprisingly this is associated with
attending political meetings, but it is much narrower in scope than the
first principal component and it explains just over 17% of the variance
in the data.
One notable feature of Table 1 is that voting appears to be unrelated to
both of these factors, since none of the loadings on the voting measure was
large. This finding is consistent with a good deal of evidence that voting is
a distinct form of political participation that stands on its own and is only
weakly related to other forms of citizen involvement (Verba & Nie 1972;
Barnes & Kaase 1979; Pattie et al. 2004).
Rival models of participation

There is a wide variety of alternative models that can be used to analyse
political participation and that have appeared in the literature over the
years (Whiteley & Seyd 2002). They all provide different, though in some
cases overlapping, mechanisms for explaining participation, and they
can be categorised into broadly two types. First, there are sociological
accounts that stress the importance of social structures operating largely
outside the control of the individual, and, second, there are choice-based
models that place the individual actor at the centre of the stage (Pattie
et al. 2004). Indicators available in the ISSP Citizenship Survey make it
possible to test three of these models.4 They are the civic voluntarism and
Other models of participation include the relative deprivation model, the mobilisation model and the general
incentives model.
26
social capital models, which, broadly speaking, are examples of the first
type, and the cognitive engagement model, which is an example of the
second. We consider each of these in turn.
The civic voluntarism model

The civic voluntarism model is perhaps the best-known model of political
participation, and is associated with the work of Sidney Verba and his
various collaborators (Verba & Nie 1972; Verba et al. 1978; Parry et al.
1992; Verba et al. 1995). The core idea of the model is that participation is
largely determined by the individuals resources. Verba and Nie underlined
this point in their original study of participation in America: According
to this model, the social status of an individual his job, education, and
income determines to a large extent how much he participates (1972,
p. 13).
Subsequent work has defined these resources as time, money and
civic skills (Verba et al. 1995, p. 271), and has incorporated additional
motivational variables into the analysis. The latter measure the individuals
psychological engagement with politics as well as their sense of political
efficacy in influencing the political process. The motivational variables are
seen as being largely derivative of the individuals resources, and they play
a linking role between social characteristics on the one hand and political
participation on the other. So the model provides a largely sociologically
based account of political participation.
For estimation purposes, variables such as the respondents social status,
their educational attainment and the time they have available for political
involvement are included as indicators of individual resources. In addition,
political efficacy captures motivational aspects of civic engagement, and
voluntary activity and religious attendance are included as indicators of
civic skills. The indicators in the model are examined in more detail in the
Appendix.
According to the civic voluntarism model, highly educated individuals
with high social status should be more likely to participate. Similarly,
individuals with spare time available, perhaps because of working parttime, should also be more active. Equally, a strong sense of political efficacy
should also boost their participation. Voluntary activity, which measures
the individuals involvement in a variety of voluntary organisations, such
as trades unions, sports clubs and cultural organisations, should also help
to promote the civic skills needed to sustain political participation. By
the same reasoning, active engagement in a religious organisation should
27
have a similar effect. Verba and his collaborators make a similar argument
about partisanship, since individuals who think of themselves as party
supporters should be more motivated to join and become active in politics
(Verba et al. 1995).
The social capital model

James Coleman (1988) introduced the term social capital into modern
social theory. He defined it as a set of obligations and expectations
that bind individuals together and promote social solidarity. In his
interpretation social interactions generate credit slips of obligations and
norms of reciprocation, and in an environment in which individuals trust
each other, these can be utilised by third parties to solve collective action
problems. The core idea here is that networks of voluntary relationships
between individuals foster mutual trust and engagement, and encourage
people to work together to solve common problems. In a similar way,
Robert Putnam defined social capital as features of social organization,
such as trust, norms and networks that can improve the efficiency of
society by facilitating co-ordinated actions (1993, p. 167). From this
perspective, social capital is like other types of capital that can be invested
to make society more efficient.
If individuals are embedded in relatively strong networks of communitybased relationships, are active volunteers and trust their fellow citizens they
are more likely to participate in politics as a consequence (Putnam 2000).
For most researchers trust is the key indicator of social capital (Putnam
1993, 2000; Fukuyama 1995; Brehm & Rahn 1997; Van Deth et al.
1999). Interpersonal trust allows individuals to move beyond their own
immediate circle of family members and friends and engage in cooperative
behaviour with strangers. In the social capital model, interactions between
individuals in voluntary associations generate interpersonal trust and in
turn this produces social capital (Whiteley 1999).
The key signature variables associated with the social capital model are
voluntary activity and interpersonal trust. Other indicators are individual
ties, captured by the individuals marital status and whether or not they
live in a single-person household. The type of community they live in is
also relevant. Large cities or conurbations are more anonymous and less
likely to build social capital than small towns or rural communities where
individuals are more likely to know each other (Oliver 2001). The key
measure, however, is inter-personal trust, which is based on a question
about trusting fellow citizens. The social capital model predicts that
28
individuals with many social ties, who live in relatively small homogenous
communities, live in multiple person households, volunteer in the
community and who trust their fellow citizens should be more politically
active than individuals who lack these characteristics.
The cognitive engagement model

The core idea of the cognitive engagement model is that participation
is motivated by the individuals ability and willingness to process and
understand information about politics and society (Norris 2000; Dalton
2002; Clarke et al. 2004, 2009). This account stresses the importance of
the individuals educational attainment, their knowledge of politics and
engagement with the political process as explanations of their involvement.
In common with the civic voluntarism, this model attaches particular
importance to education as a factor that promotes individual engagement.
But it does so for different reasons from the civic voluntarism model. In the
cognitive engagement model, education is an indicator of the individuals
ability to make sense of the political world, and an index of their ability
to process and understand political information. In the civic voluntarism
model it is primarily a measure of resources and social status. Viewed
from the perspective of classical political theory, cognitively engaged
citizens are close to the Greek conception of the good citizen, who is an
informed member of the polis and who fully participates in the process of
government.
In the cognitive engagement model education, media consumption of
politics, interest in politics, discussion of politics with friends and fellow
workers, and the civic norms that support participation should all be
positive predictors of participation. As the Appendix shows, the media
usage measure combines indicators of television, radio, newspapers and
internet usage into a cumulative scale. Similarly, civic norms are scaled by
means of a principal components analysis of four items.
Individual-level models
The three models can be estimated separately and compared with each
other to determine which one provides the best account of political
participation. However, research suggests that each of them contributes
to a general understanding of participation, and none of them formally
encompasses the others (Pattie et al. 2004). For this reason they are
estimated as a composite model in Table 2. This table contains models
29
Table 2The individual level participation models (heterogeneity corrected with robust
standard errors)
Predictors
Vote
Participation
Highest educational attainment

Social status
Political efficacy
Religiosity
Partisanship
Voluntary activity index
Interest in politics
Discuss politics
Media consumption of politics
Participatory norms
Interpersonal trust
Type of community
Married
Single-person household
Male
Age
0.17***
0.00
0.04***
0.01
1.92***
0.07***
0.11***
0.08***
0.02
0.19***
0.03***
0.09***
0.20***
0.08
0.05
0.03***
0.07***
0.00
0.03***
0.02***
0.07***
0.08***
0.06***
0.13***
0.01***
0.07***
0.05***
0.02***
0.00
0.01
0.08***
0.006***
Chi-squared test of between country

intercept effects
3573.6***
11392.3***
R-squared
Party membership
0.07***
0.00
0.03***
0.00
1.04***
0.19***
0.49***
0.23***
0.03***
0.21***
0.01
0.11***
0.01
0.02
0.20***
0.01***
3857.1***
0.32
Logistic regression models; *p < 0.10; **p < 0.05; ***p < 0.01
of voting, general participation and party membership. The voting and

party membership models use dummy dependent variables and a logistic
regression specification for their estimation. The participation model uses
the first principal component from Table 1 as the dependent variable and
an OLS estimation strategy. The three models are estimated using a multilevel procedure with robust standard errors, so that between-country
differences in the 36 nations are controlled for with a random intercept
specification. This corrects for any homogeneities that might exist between
respondents within countries. A chi-square test of the variance of betweencountry intercepts appears at the bottom of each table. This identifies the
existence of country-related differences in the data.5
To examine the voting model first, the civic voluntarism model in
the form of educational attainment, political efficacy, partisanship and
voluntary activity, plays an important role in explaining voting behaviour.
The one exception to this relates to religiosity, which has no impact
on voting, a finding inconsistent with Verba and his teams analysis
of participation in the US (Verba et al. 1995, p. 358). Alongside civic
voluntarism, the social capital model is also important, since interpersonal
See Raudenbusch and Bryk (2002, p.102) for a further discussion of this random intercepts model.
30
trust, community type and marital status are all positive predictors
of voting. Third, the cognitive engagement model is important, too,
since interest in politics, discussion of politics with friends and fellow
workers, and participatory norms are all significant. Variables from.
all three models make a contribution to explaining turnout. In addition,
the estimates show that individuals are more likely to vote as they get
older.
The general participation model is rather similar to the voting model,
in that all three theoretical accounts contribute towards explaining
participation with the expected signs. The differences between voting and
generalised participation relate to religiosity, which has a negative impact on
participation, and community type, with individuals in small communities
being more involved than individuals in large ones. In addition, females
are more likely to participate than males, and the age effect is negative
rather than positive. Apart from these differences, resources, measures of
engagement and indicators of social capital such as interpersonal trust all
have a positive impact on generalised participation. So the model has much
in common with the voting model.
The party membership model also demonstrates that resources,
motivations and engagement are important factors in influencing
participation. However, in this case men are significantly more likely to
get involved than women, as was true of voting, and older respondents are
more likely to join parties than younger ones. Voluntary activity in nonpolitical organisations plays a particularly important role in explaining
party membership, although interpersonal trust does not.
One interesting feature of Table 2 is that the chi-square tests of betweencountry variances are all highly significant. This indicates, not surprisingly,
that it is important to take into account differences between countries
when explaining individual-level participation in a comparative setting.
Furthermore, there is considerable scope for aggregate or country-level
covariates to assist in explaining the between-country variations in the
different forms of participation. We examine this issue next.
Country-level effects
As the earlier discussion indicated, much of the aggregate analysis of
electoral participation demonstrates the importance of institutional
variables such as the electoral system and the party systems as factors
that help to explain turnout. A recent meta-analysis of some 83 studies of
aggregate turnout concluded that institutions and demographics play the
31
key roles in determining cross-national variations in turnout (Geys 2006).

The analysis concluded:
Based on our results, we argue for the construction of a core model of turnout.
In this model, those variables that have proved their worth in previous research
and are clearly linked to theoretical work ought to be considered as cornerstones.
This would imply that, say, variables measuring the closeness of the election,
voting habits and the size of the population are indispensable to any future
analysis of turnout. (Geys 2006, p. 653)
A number of researchers have compared explanations of turnout based

on values and beliefs with those based on institutions. This work suggests
that institutions are far more important than values in explaining turnout
(Jackman 1987; Jackman & Miller 1995; Gray & Caul 2000). However,
more recent work challenges this conclusion, arguing that, while institutions
matter, they do not matter very much when it comes to explaining electoral
participation (Klingemann 2009).
Comparing and contrasting values with institutions is a slightly odd
thing to do if particular values have not been shown to play a prominent
role in explaining electoral participation at the individual level of analysis.
It is more interesting to compare institutions with contextual variables
that are the aggregate-level counterparts of measures that are important
in individual-level models. For example, if individual resources such as
education are important to participation, as the civic voluntarism model
suggests, then country-level resources, such as the number of graduates
in the workforce, should be important contextual influences. Equally,
the cognitive engagement model suggests that political discussion is a
significant predictor in participation models, so we might expect national
levels of political debate to be important too. Finally, the social capital
model shows that voluntary activity helps to drive all forms of political
participation, so country-level differences in volunteering should have an
independent impact on individual participation.
This discussion raises the possibility that institutional effects identified in
aggregate voting models may be spurious, and merely reflect specification
errors in these models. In particular a failure to take into account crossnational differences in resources, motivations, social ties, volunteering
and levels of citizen cognitive engagement might explain the institutional
effects. To investigate this further we need to specify which institutional
variables might be relevant for understanding wider forms of participation
in addition to voting. Clearly, there are many institutional variables
that might be important, but three have played a prominent role in the
32
literature. They are the electoral system, the party system and the size of
the relevant political community (see Gallagher & Mitchell 2005).
To consider the electoral system first, an extensive amount of research
has been devoted to the task of understanding how voting systems
influence turnout (Lijphart 1984; Cox 1997; Gallagher & Mitchell 2005).
A good deal of this work centres on Duvergers Law, the proposition that
single member district plurality electoral systems are likely to produce two
effective political parties (Duverger 1954). Though the evidence for this
proposition is mixed (Gallagher & Mitchell 2005, pp. 3033), it clearly
has implications for political participation beyond turnout.
There is a mechanical-institutional effect associated with Duvergers Law,
the fact that the single member plurality system distorts the relationship
between votes and seats much more than proportional systems. This
serves to discourage third parties by making it difficult for them to win
representation in the legislature. In addition, there is a psychological
effect arising from restrictions imposed on the opportunity structures
for participation generated by the electoral system. Tarrow defines
opportunity structures as dimensions of the political environment which
either encourage or discourage people from using collective action (1994,
p. 18). It is known that distortions in the relationship between votes and
seats serve to inhibit electoral participation, since proportional electoral
systems experience higher turnouts than plurality systems (Blais & Carty
1990; Ladner & Milner 1999; Franklin 2004). Given this, it seems likely
that such distortion will inhibit party membership and activism as well,
since it restricts the variety and effectiveness of parties, and so reduces the
incentives for individuals to get involved.
The impact of electoral distortion on other types of participation is
ambiguous, however, since it can be argued that a narrow party system
will serve to inhibit all forms of participation. On the other hand,
restrictions on parties might stimulate alternative forms of participation
like protesting or boycotting if they provide substitutes for political action.
These possibilities can be explored by means of a well-known measure of
the electoral distortion, the Gallagher Least-Squares index, which relates
vote shares to seat shares.6 A high score on this index denotes a large
electoral distortion.
The second institutional effect arises from the fragmentation of the party
system. It is known that a highly fragmented party system can inhibit
electoral participation because this imposes extra information processing
The least-squares index is defined as: (((Si Vi)2)/2), where Si is the seat share for party i and Vi is the vote
share for party i.
33
costs on individuals (Blais & Dobrzynska 1998). They have to know

more about the parties on offer in order to make a choice, and this raises
the costs of voting. The effect of a large number of parties on other types
of participation, however, is likely to be different. In the case of party
membership, an increase in the number and range of political parties is
likely to stimulate party membership since it improves the opportunity
structures and raises individual incentives to get involved. The more
political parties represented in the legislature the greater the opportunities
for ambitious proto-politicians and the closer the elected representatives
are to sharing power. This enhances incentives for individuals to join
political parties and to become active (Whiteley & Seyd 2002).
In relation to participation in general, again the impact of party
fragmentation is ambiguous. On the one hand it increases the opportunity
structures for participation within parties and this might divert activity
from other forms of involvement. On the other hand, a highly fragmented
party system shares power widely and a large number of party members
means that there are many potential recruits for holding political meetings,
organising petitions and planning protest demonstrations. The effects of
party fragmentation can be estimated using the effective number of parties
index, which is an entropy-based measure of party fragmentation7 (Laakso
& Taagepera 1979).
A third measure that has been discussed in the voting literature relates to
the size of electoral districts. Since Downs (1957) first examined this issue,
it has been suggested that large electoral districts inhibit turnout, since they
reduce the probability of an individual making a difference to the outcome
of the election. As is well known, this probability is vanishingly small
even in modest-sized constituencies, but this does not prevent individuals
from believing that they can make a difference in marginal seats (Clarke
et al. 2004). More generally, this is linked to wider debates about the
relationship between size and democracy, which suggest that collective
action of all types is more difficult to achieve in large polities (Olson
1965; Dahl & Tufte 1973). The implication is that large constituencies can
inhibit all types of participation, an idea that can be examined using data
on the average size of legislative constituencies across different countries.
Alongside the three aggregate institutional variables we can include
country-level effects associated with the individual-level models of
participation described earlier. The percentage of graduates and the mean
efficacy score in each country capture two key variables from the civic
The effective number of elective parties is defined as: 1/(Pv)2, where Pv is the proportion of votes obtained by
party P.
34
voluntarism model. The former is a contextual indicator of resources, and

the latter a measure of the perceived effectiveness of political participation.
The mean levels of interest in politics and the percentage of individuals
in each country who often discuss politics are contextual indicators
associated with the cognitive engagement model. Finally, mean levels of
voluntary activity and the national perceptions of trust in others provide
country-level indicators of the social capital model.
Aggregate-level results
Table 3 contains the aggregate-level estimates of the participation and
institutional variables derived from re-estimating the models in Table 2,
but this time incorporating the aggregate covariates using a random
intercepts specification. Thus the effects in Table 3 are measures of shifts in
the intercepts of the individual-level models in Table 2, with positive scores
indicating that a contextual variable boosts individual-level participation,
and negative scores indicating the opposite. The individual-level models
are not shown since they do not differ significantly from those in Table 2.
To consider the voting model first, the estimates confirm that all
three institutional variables have a significant impact on voting with the
expected signs. Thus large electoral districts, high levels of distortion in
the voteseat relationship and a highly fragmented party system all serve
to inhibit individuals from voting. In addition, it appears that a high
percentage of graduates in a country stimulates turnout, as does a strong
sense of political efficacy. This suggests that the civic voluntarism model
has an impact on voting at the country level, over and above any effects
associated with institutions. In contrast the aggregate-level variables from
Table 3 The country-level participation models (random intercepts specification)

District size
Least-squares index
Effective number of parties
Percentage of graduates
Mean efficacy score
Mean interest in politics
Percentage discuss politics often
Percentage always or usually can be trusted
Mean voluntary activity
Aggregate R square
Voting
Participation
Party membership
1.0E-5***
0.09*
0.61**
0.07*
0.49*
0.03
0.01
0.05**
0.16
1.0E-6
0.01
0.03
0.00
0.02
0.01
0.00
0.01**
0.07*
1.0E-6
0.08**
0.08
0.05***
0.42***
0.02**
0.06***
0.01
0.19
0.39
0.34
0.32
*p < 0.10; **p < 0.05; ***p < 0.01.
35
the cognitive engagement model appear to have no effect on individuallevel participation, although their individual-level counterparts are still
important, as Table 2 shows. In the case of the aggregate measures
from the social capital model it appears that interpersonal trust inhibits
voting turnout and voluntary activity has no effect. The former effect is
anomalous and inconsistent with findings from the social capital literature
(see, for example, Putnam 2000).
The aggregate participation model is very different from the aggregate
voting model. In this case none of the institutional variables has an
impact on participation, and nor do the aggregate variables from the
civic voluntarism and cognitive engagement models. However, both
mean interpersonal trust and aggregate voluntary activity appear to
stimulate participation, indicating that the social capital model does
play an important role in explaining generalised participation across the
democratic world at both the individual and aggregate levels.
Finally, the party membership model shows a different pattern yet again
from the other two models. In this case neither district size nor party
fragmentation have an independent influence on party membership, but
the least-squares index does have an important effect. The effect is similar
to that of voting in that high levels of electoral distortion inhibit party
membership in the same way as they inhibit voting. As is well known, high
levels of electoral distortion are associated with the single member plurality
electoral system commonly found in the Anglo-American democracies, and
this type of electoral system serves to undermine the incentives to join and
be active in political parties. In addition in the party membership model,
measures from the civic voluntarism and cognitive engagement models
appear to have contextual leverage, although the sign on the interest in
politics variable is negative. Apart from this anomalous finding, all the
other effects are consistent with expectations. Finally, in sharp contrast to
its importance in explaining generalised participation, the social capital
model has no contextual impact on party membership.
Discussion and conclusions

This analysis shows that institutions that have frequently been discussed
in aggregate models of turnout do appear to have an important influence
on individual-level electoral participation. But it is also true that variables
associated with the civic voluntarism model have an important influence
as well, something generally ignored in the aggregate voting literature.
Institutions matter, but so do resources and the motivations they generate
36
among voters. In contrast, institutions appear to have little or no effect

on other forms of political participation, with the single exception of the
electoral system having a clear influence on party membership in much
the same way as it influences voting. Thus the distortion produced by
the first-past-the-post electoral system undermines both voting and party
involvement. With that single exception, other forms of participation and
party membership are best explained at the aggregate level by variables
from the participation models, notably the social capital model in the
case of participation, and the civic voluntarism and cognitive engagement
models in the case of party membership.
More generally, institutions measure opportunity structures and so
have a legitimate claim to being important influences on participation. But
this evidence points to the importance of social structural variables and
community-based measures as other aspects of the opportunity structures
that can influence participation. There are of course other institutional
variables such as constitutional rules and federalist arrangements, which
could be taken into account in a broader model, and these may well play
a role in explaining generalised participation. But this should be done in
a context that comprehensively models opportunity structures rather than
focuses just on the narrow range of variables that have dominated the
literature up to now.
There has been a vast amount of aggregate-level research explaining
electoral participation, as the meta-analysis referred to earlier indicates
(Geys 2006). Such comparative aggregate analysis facilitates the
investigation of the effects of institutions, since the most important of these,
such as the electoral system, vary only at the national level in the case of
primary elections. The present analysis supports such investigations, since
it is clear that key institutions influence electoral participation in expected
directions. However, these models may also be misleading by failing to
take into account other variables at the aggregate level that influence the
choices that individuals make about their rates of participation. These
other variables derive from the resources available in a country, and
the attitudes that are shared widely throughout particular societies have
been relatively neglected because they are much harder to measure at the
national level than institutions. However, further research points to the
need to combine well-specified individual-level models with important
contextual effects that go beyond examining institutions if cross-national
variations in electoral participation are to be understood.
37
Appendix: Measurement of variables

Voting Self-reported turnout in the previous national election
Participation Factor scores from the principal components analysis of
the items in Table 2
Party membership Self-reported membership or activism in a political party
Education
0 No formal qualification
1 Lowest formal qualification
2 Above lowest qualification
3 Higher secondary completed
4 Above higher sec level, other educ
5 University degree completed
Social status
Self-placement on a 10-point scale varying from 1 (low status) to 10 (high
status)
Efficacy scale sum of responses to three statements:
People like me dont have any say about what the government does
I dont think the government cares much what people like me think
I think most people in (Britain) are better informed about politics and
government than I am
(coded so that a high score denotes high levels of efficacy).
Religiosity attendance at religious services
7 Several times a week + every day
6 Once a week
5 2 or 3 times a month
4 Once a month
3 Sev times a year
2 Once a year
1 Less frequently
0 Never
Partisanship
1 Supports a political party
0 Does not support a party
38
Voluntary activity
Sum of responses to questions asking about membership of a trades union,
a church organisation or sports organisation, where:
3 Belong and participate
2 Belong not participate
1 Used to belong
0 Never belonged to
Interest in politics
4 Very interested
3 Fairly interested
2 Not very interested
1 Not at all interested
Discuss politics
4 Often
3 Sometimes
2 Rarely
1 Never
Media consumption of politics
Sum of responses to questions about newspaper readership, television
watching, radio listening and internet usage
4 Every day
3 34 days a week
2 12 days a week
1 Fewer than CODE 3
0 Never
Participatory norms
Principal components analysis of statements about what a good citizen
should do:
always to vote in elections
keep watch on the actions of government
to be active in social or political associations
to choose products for political, ethical or environmental reasons, even if
they cost a bit more
(measured with 10-point scales).
39
Interpersonal trust
4 Always trust
3 Usually trust
2 Usually careful
1 Always careful
Type of community
1 Urban, a big city
2 Suburb, outskirt of a big city
3 Town or small city
4 Country village
5 Farm or home in the country
Marital status, Single-person household and Sex are all dummy variables.
Age is in years.
References
Arrow, K. (1951) Social Choice and Individual Values. New York: Wiley.
Barnes, S. & Kaase, M. (1979) Political Action. London: Sage.
Black, D. (1958) Theory of Committees and Elections. Cambridge: Cambridge University
Press.
Blais, A. & Carty, R.K. (1990) Does proportional representation foster voter turnout?
European Journal of Political Research, 18, pp. 16781.
Blais, A. & Dobrzynska, A. (1998) Turnout in electoral democracies. European Journal of
Political Research, 33, pp. 239261.
Brehm, J. & Rahn, W. (1997) Individual-level evidence for the causes and consequences of
social capital. American Journal of Political Science, 41, pp. 8881023.
Clarke, H.D., Sanders, D., Stewart, M.C. & Whiteley, P. (2004) Political Choice in Britain.
Oxford: Oxford University Press.
Clarke, H.D., Sanders, D., Stewart, M.C. & Whiteley, P. (2009) Performance Politics and the
British Voter. Cambridge: Cambridge University Press.
Clarke, H.D., Sanders, D., Stewart, M.C. & Whiteley, P. (forthcoming) Performance Politics:
Party Choice in Britain. Cambridge: Cambridge University Press.
Coleman, J. (1988) Social capital in the creation of human capital. American Journal of
Sociology, 94, supplement S95-S119.
Cox, G.W. (1997) Making Votes Count. Cambridge: Cambridge University Press.
Cox, G.W. & Munger, M.C. (1989) Closeness, expenditures and turnout in the 1982 US house
elections. American Political Science Review, 83, pp. 217230.
Dahl, R. & Tufte, E. (1973) Size and Democracy. Stanford CA: Stanford University Press.
Dalton, R.J. (2002) Citizen Politics: Public Opinion and Political Parties in Advanced
Industrial Democracies (3rd edn). Chatham, NJ: Chatham House Publications.
Downs, A. (1957) An Economic Theory of Democracy. New York: Harper & Row.
Duverger, M. (1954) Political Parties. New York: John Wiley & Sons.
Farrell, D.M. (2001) Electoral Systems: A Comparative Introduction. London: PalgraveMacmillan.
40
Franklin, M. (2004) Voter Turnout and the Dynamics of Electoral Competition in Established
Democracies since 1945. Cambridge: Cambridge University Press.
Fukuyama, F. (1995) Trust: Human Nature and The Reconstitution of Social Order. New
York: Touchstone Books.
Gallagher, M. & Mitchell, P. (2005) The Politics of Electoral Systems. Oxford: Oxford
University Press.
Geys, B. (2006) Explaining voter turnout: a review of aggregate-level research. Electoral
Studies, 25, pp. 637663.
Gray, M. & Caul, M. (2000) Declining voter turnout in advanced industrial democracies
19501997. Comparative Political Studies, 33, pp. 10911122.
Jackman, R.W. (1987) Political institutions and voter turnout in the industrial democracies.
American Political Science Review, 81, 2, pp. 405423.
Jackman, R.W. & Miller, R.A. (1995) Voter turnout in the industrial democracies during the
1980s. Comparative Political Studies, 27, 4, pp. 467492.
Klingemann, H.D. (2009) The Comparative Study of Electoral Systems. Oxford: Oxford
University Press.
Laakso, M. & Taagepera, R. (1979) Effective number of political parties: a measure with
applications to Western Europe. Comparative Political Studies, 12, pp. 327.
Ladner, A. & Milner, H. (1999) Do voters turn out more under proportional than majoritarian
systems? The evidence from Swiss communal elections. Electoral Studies, 18, pp. 235250.
Lijphart, A. (1984) Democracies. New Haven: Yale University Press.
Matsusaka, J.G. & Palda, F. (1993) The Downsian voter meets the ecological fallacy. Public
Choice, 77, pp. 855878.
McKelvey, R.D. (1976) Intransitivities in multidimensional voting models and some
implications for agenda control. Journal of Economic Theory, 12, pp. 472482.
Mueller, D.C. (2003) Public Choice III. Cambridge: Cambridge University Press.
Norris, P. (2000) A Virtuous Circle: Political Communications in Postindustrial Societies.
Cambridge: Cambridge University Press.
Oliver, E. (2001) Democracy in Suburbia. Princeton, NJ: Princeton University Press.
Olson, M. (1965) The Logic of Collective Action. Cambridge, MA: Harvard University Press.
Owen, G. & Grofman, B. (1984) To vote or not to vote: the paradox of non-voting. Public
Choice, 42, pp. 311325.
Parry, G., Moyser, G. & Day, N. (1992) Political Participation and Democracy in Britain.
Cambridge: Cambridge University Press.
Pattie, C., Seyd, P. & Whiteley, P. (2004) Citizenship in Britain: Values, Participation and
Democracy. Cambridge: Cambridge University Press.
Putnam, R. (1993) Making Democracy Work: Civic Traditions in Modern Italy. Princeton, NJ:
Princeton University Press.
Putnam, R. (2000) Bowling Alone: The Collapse and Revival of American Community. New
York: Simon & Schuster.
Raudenbusch, S.W. & Bryk, A.S. (2002) Hierarchical Linear Models: Applications and Data
Analysis Methods (2nd edn). Thousand Oaks, CA: Sage Publications.
Schofield, N. (1978) Instability of simple dynamic games. Review of Economic Studies, 45,
pp.575594.
Seyd, P. & Whiteley, P. (1992) Labours Grassroots: The Politics of Party Membership.
Oxford: Oxford University Press.
Shepsle, K.A. (1979) Institutional arrangements and equilibrium in multidimensional voting
models. American Journal of Political Science, 23, pp. 2759.
Shepsle, K.A. & Weingast, B.R. (1987) The institutional foundations of committee power.
American Political Science Review, 81, pp. 86108.
41
Snijders, T. & Bosker, R. (1999) Multilevel Analysis: An Introduction to Basic and Advanced
Multilevel Modelling. London: Sage.
Tarrow, S. (1994) Power in Movement. Cambridge: Cambridge University Press.
Van Deth, J., Maraffi, M., Newton, K. & Whiteley, P.F. (1999) Social Capital and European
Democracy. London: Routledge.
Verba, S. & Nie, N. (1972) Participation in America. Chicago: University of Chicago Press.
Verba, S., Nie, N. & Kim, J.-O. (1978) Participation and Political Equality: A Seven Nation
Comparison. Cambridge: Cambridge University Press.
Verba, S., Schlozman, K. & Brady, H. (1995) Voice and Equality: Civic Voluntarism in
American Politics. Cambridge, MA: Harvard University Press.
Whiteley, P. (1999) The origins of social capital. In J. Van Deth, M. Maraffi, K. Newton & P.F.
Whiteley (eds) Social Capital and European Democracy. London: Routledge.
Whiteley, P. & Seyd, P. (2002) High Intensity Participation The Dynamics of Party Activism
in Britain. Ann Arbor, MI: University of Michigan Press.
About the authors

Paul Whiteley is Professor of Government at the University of Essex.
David Sanders is Professor of Government at the University of Essex.
Marianne Stewart is Professor of Government, Politics and Political
Economy at the University of Texas at Dallas.
Harold Clarke is Ashbel Smith Professor of Political Science at the
University of Texas at Dallas and Adjunct Professor of Government at the
University of Essex.
Address correspondence to: Paul Whiteley, Department of Government,
University of Essex, Wivenhoe Park, Colchester, CO4 3SQ, UK.
Email: whiteley@essex.ac.uk
42
Consumer-based brand equity

conceptualisation and measurement
A literature review
George Christodoulides
University of Birmingham
Leslie de Chernatony
Universit della Svizzera italiana, Lugano and Aston Business School
Although there is a large body of research on brand equity, little in terms of a

literature review has been published on this since Feldwicks (1996) paper. To
address this gap, this paper brings together the scattered literature on consumerbased brand equitys conceptualisation and measurement. Measures of consumerbased brand equity are classified as either direct or indirect. Indirect measures
assess consumer-based brand equity through its demonstrable dimensions and are
superior from a diagnostic level. The paper concludes with directions for future
research and managerial pointers for setting up a brand equity measurement
system.
Introduction
Marketers are continually under pressure to justify the impact of
marketing activities, and this has renewed interest in measures of
marketing performance (OSullivan & Abela 2007). The Marketing
Science Institute (MSI) indicatively placed accountability and return on
investment of marketing expenditure at the top of its research priorities
for 200810 (MSI 2008). Financial measures such as sales and profit
provide only partial indicators of marketing performance due to their
historical orientation and typically short-term horizon (Mizik & Jacobson
2008). Intangible, market-based assets, on the other hand, provide a richer
understanding of marketing performance, reconciling short- and long-term
performance (Ambler 2003) as well as bridging marketing and shareholder
Received (in revised form): 20 July 2009

DOI: 10.2501/S1470785310201053
43
Consumer-based brand equity conceptualisation and measurement
value (Srivastava et al. 1998). While competitors can emulate financial and
physical assets, intangible assets represent a more sustainable competitive
advantage (Hunt & Morgan 1995).
Brand equity is a key marketing asset (Davis 2000; Ambler 2003),
which can engender a unique and welcomed relationship differentiating
the bonds between the firm and its stakeholders (Hunt & Morgan 1995;
Capron & Hulland 1999), and nurturing long-term buying behaviour.
Understanding the dimensions of brand equity, then investing to grow this
intangible asset raises competitive barriers and drives brand wealth (Yoo
et al. 2000). For firms, growing brand equity is a key objective achieved
through gaining more favourable associations and feelings among target
consumers (Falkenberg 1996). Previous research established a positive
effect of brand equity on: consumer preference and purchase intention
(Cobb-Walgren et al. 1995); market share (Agarwal & Rao 1996);
consumer perceptions of product quality (Dodds et al. 1991); shareholder
value (Kerin & Sethuraman 1998); consumer evaluations of brand
extensions (Aaker & Keller 1990; Rangaswamy et al. 1993; Bottomley
& Doyle 1996); consumer price insensitivity (Erdem et al. 2002); and
resilience to product-harm crisis (Dawar & Pillutla 2000).
Over the last 15 years, brand equity has become more important as the
key to understanding the objectives, mechanisms and net impact of the
holistic impact of marketing (Reynolds & Phillips 2005). In this context,
it is not surprising that measures capturing aspects of brand equity have
become part of a set of marketing performance indicators (Ambler 2003).
The discussion of brand equity and its measurement has a broad range
of adherents, both academic and practitioner, that collectively share what
can be described as a black box orientation (Reynolds & Phillips 2005).
Evidence of the importance of brand equity for the business world is the
fact that there is currently a significant number of consulting firms (e.g.
Interbrand, WPP, Young & Rubicam and Research International), each
with their own proprietary methods for measuring brand equity (Haigh
1999). In setting up the future research agenda for brand management,
Keller and Lehman (2006) unsurprisingly identified brand equity and its
measurement as a significant research topic.
The literature on brand equity, although substantial, is largely fragmented
and inconclusive. As Berthon et al. (2001) put it, perhaps the only thing
that has not been reached with regard to brand equity is a conclusion.
This paper provides a systematic review of the literature on brand
equity conceptualisation and measurement, and concludes with some
directions for future research. Figure 1 shows the structure of the paper
44
Brand equity
CBBE
FBBE
Direct
Multiattribute
Indirect
Other
Intermediate
Outcome
Figure 1 Brand equity methodologies
and introduces the broad categories of methodologies to measure CBBE.

Although the whole purpose of setting up a brand equity monitor is to
enable marketers to appreciate its key drivers, it is beyond the scope of this
paper to review research on antecedents and consequences of brand equity
as exemplified by Yoo et al. (2000).
Firm-based versus consumer-based brand equity

As Winters (1991) states, if you ask ten people to define brand equity,
you are likely to get ten (maybe 11) different answers as to what it means
(p. 70). Since then, many studies have been published on brand equity but
Winters statement is even more relevant today than it was in 1991. Brand
equity is such a complex concept that the diversity of its conceptualisations
in the literature may be due to the blind men and elephant syndrome
(Ambler 2003); different studies describing different aspects of this
intangible asset. The lack of an agreed definition of brand equity has
in turn spawned various methodologies for measuring the construct.
Although there is no universally accepted definition of brand equity, there
is at least some consensus in that brand equity denotes the added value
endowed by the brand to the product (Farquhar 1989, p. RC7). This value
can serve as a bridge that links what happened to the brand in the past
and what should happen to the brand in the future (Keller 2003); hence
Amblers (2003) characterisation of brand equity as a repository of future
profits or cash flows that results from past marketing investment.
Firms, however, are not the only recipients of brand value. According
to the literature the main recipients of brand value are either firms or
customers, and such a view is clearly presented in Aakers (1991) definition
45
of brand equity as a set of assets and liabilities linked to a brand, its name
and symbol, that add to or subtract from the value provided by a product
or service to a firm and/or that firms customers (p. 15). A similar yet more
output-orientated definition is that of Srivastava and Shocker (1991), who
define brand equity as a set of associations and behaviors on the part of a
brands consumers, channel members and parent corporation that enables
a brand to earn greater volume or greater margins than it could without
the brand name and, in addition, provides a strong, sustainable and
differential advantage. So far, the brand equity construct has been viewed
from two major perspectives in the literature. Some authors focused on
the financial perspective of brand equity (Farquhar et al. 1991; Simon &
Sullivan 1993; Haigh 1999) and others on the customer-based perspective
(Aaker 1991; Keller 1993; Yoo & Donthu 2001; Vazquez etal. 2002; de
Chernatony et al. 2004; Pappu et al. 2005; Christodoulides et al. 2006).
The first perspective discusses the financial value brand equity creates to
the business and is often referred to as firm-based brand equity (FBBE).
However, the financial value of brand equity is only the outcome of
consumer response to a brand name. The latter is considered the driving
force of increased market share and profitability of the brand, and is based
on the markets perceptions (consumer-based brand equity).
In a related study trying to understand interpretations of brand equity,
Feldwick (1996) identified three different ways in which the term brand
equity has been used: first to signify the total value of a brand as a
separate asset when it is sold or included on a balance sheet; second, as a
measure of the strength of consumers attachment to the brand; and, third,
as a description of the associations and beliefs the consumer has about the
brand. While the first sense of the term is associated with firm-based brand
equity, the other two senses reflect consumer-based brand equity. Kapferer
(2004) attempted to link consumer-based brand equity dimensions (i.e.
brand assets) to brand value (net discounted cash flow attributable to
the brand after paying the cost of capital invested to produce and run the
business and the cost of marketing) (e.g. price premium) through CBBE
consequences such as price premium. For Kapferer it is essential for brands
to yield financial benefits if they are to claim high levels of equity.
Consumer-based brand equity

The conceptualisations of consumer-based brand equity have mainly
derived from cognitive psychology and information economics. The
dominant stream of research has been grounded in cognitive psychology,
46
focusing on memory structure (Aaker 1991; Keller 1993). Aaker (1991)

identified the conceptual dimensions of brand equity as brand awareness,
brand associations, perceived quality, brand loyalty, and other proprietary
brand assets such as patents, trademarks and channel relationships. The
former four dimensions of brand equity represent consumer perceptions
and reactions to the brand, while proprietary brand assets are not pertinent
to consumer-based brand equity. Keller (1993) looked at consumer-based
brand equity strictly from a consumer psychology perspective and defined
it as the differential effect of brand knowledge on consumer response to
the marketing of the brand (p. 2). According to this conceptualisation, a
brand has a positive (or negative) value if the consumer reacts more (or
less) favourably to the marketing mix of a product of which he/she knows
the brand name than to the marketing mix of an identical yet unbranded
product. Consumer response to the marketing mix of a brand can be
translated at various stages of the purchase decision-making sequence,
such as preference, choice intentions and actual choice. According to
Keller (1993), brand knowledge is a key antecedent of consumer-based
brand equity and is in turn conceptualised as a brand node in memory to
which a variety of associations have been linked. Brand knowledge is then
decomposed into two separate constructs: brand awareness and brand
image (associations). The majority of conceptual studies summarised in
Table 1 agree that awareness and associations are important components
of consumer-based brand equity. The majority of conceptual studies on
CBBE took place in the early/mid-1990s with subsequent research being
mostly empirical. It may also be argued that the emphasis of conceptual
research has shifted from the relational intangible asset (i.e. brand equity)
per se to the consumerbrand relationship (e.g. Fournier 1998) and
customer equity (e.g. Rust et al. 2000, 2004).
In parallel, brand equity research rooted in information economics draws
on the imperfect and asymmetrical nature of markets (Erdem & Swait
1998). In this context, economic agents are required to transmit information
about their specific characteristics by means of signals. According to Erdem
et al. (2006), brand names act as signals to consumers. A brand signal
becomes the sum of that brands past and present marketing activities.
Imperfect and asymmetrical market information produces uncertainty in
consumers minds. A credible brand signal generates consumer value by:
(1) reducing perceived risk; (2) reducing information search costs; and (3)
creating favourable attribute perceptions (Erdem & Swait 1998). Under
this approach, CBBE is defined as the value of a brand signal to consumers
(Erdem & Swait 1998). From our review of the literature we regard the
47
Table 1 Conceptual research on CBBE

Study
Dimensions of CBBE
Aaker (1991, 1996) brand awareness

brand associations
perceived quality
brand loyalty
Blackston (1992) brand relationship
(trust, customer satisfaction with the brand)
Keller (1993) brand knowledge
(brand awareness, brand associations)
Sharp (1995) company/brand awareness
brand image
relationships with customers/existing customer franchise
Berry (2000) brand awareness
brand meaning
Burmann et al. (2009) brand benefit clarity
perceived brand quality
brand benefit uniqueness
brand sympathy
brand trust
Source: the authors
two research streams (cognitive psychology and information economics)

as complementary, and we propose a definition of CBBE that contains
elements from both, i.e. a set of perceptions, attitudes, knowledge, and
behaviors on the part of consumers that results in increased utility and
allows a brand to earn greater volume or greater margins than it could
without the brand name.
Measurement of CBBE
Although Aaker (1991) and Keller (1993), among others, conceptualised
brand equity, they never operationalised a scale for its measurement. This
spawned a number of methodologies to quantify this highly regarded
intangible asset, most of which employ complex statistical procedures (e.g.
Park & Srinivasan 1994; Leuthesser et al. 1995), making them difficult to
comprehend and use among practising marketers. Empirical endeavours
to operationalise consumer-based brand equity can be classified based on
their approach to measurement (i.e. direct or indirect). Direct approaches
to brand equity measurement attempt to measure the phenomenon
directly by focusing on consumers preferences (e.g. Srinivasan 1979; Park
& Srinivasan 1994) or utilities (e.g. Kamakura & Russell 1993; Swait
48
etal. 1993), while indirect approaches measure brand equity through its
demonstrable manifestations (e.g. Yoo & Donthu 2001; Pappu et al. 2005).
Table 2 summarises the main studies on consumer-based brand equity
measurement.
Table 2 Research on CBBE measurement
Measurement
Dimensions of CBBE
Measurement
level
Context
Product category
Direct approach
Srinivasan (1979)
n.a.
aggregate
US
health care
Kamakura &
Russell (1993)
perceived quality
brand intangible value
aggregate
US
detergents
Swait et al. (1993)
n.a.
individual
US
deodorants,
trainers,
jeans
Park & Srinivasan

(1994)
attribute-based brand equity

non-attribute-based brand
equity
individual
US
toothpaste,
mouthwash
Leuthesser et al.
(1995)
n.a.
individual
Austria
detergents
Shankar et al.
(2008)
Offering value, relative brand

importance
aggregate
US
insurance
Indirect approach via intermediate measures

Lassar et al.
(1995)
performance
social image
value
trustworthiness
attachment
individual
US
televisions
watches
Yoo & Donthu

(2001)
brand awareness
brand associations
perceived quality
brand loyalty
individual
US,
Korea
athletic shoes,
film, colour
television sets
Vazquez et al.
(2002)
product functional utility

product symbolic utility
brand name functional utility
brand name symbolic utility
individual
Spain
sports shoes
Washburn &
Plank (2002)
brand awareness
brand associations
perceived quality
brand loyalty
individual
US
crisps
paper towels
de Chernatony
etal. (2004)
brand loyalty
satisfaction
reputation
individual
UK
financial services
(continued)
49
Table 2 Research on CBBE measurement (continued)

Measurement
level
Measurement
Dimensions of CBBE
Netemeyer et al.
(2004)
individual
perceived quality
perceived value for the cost
uniqueness
willingness to pay a premium
Context
Product category
US
colas, toothpaste,
athletic shoes,
jeans
Pappu et al.
(2005)
brand awareness
brand associations
perceived quality
brand loyalty
individual
Australia
cars,
televisions
Christodoulides
etal. (2006)
emotional connection
online experience
responsive service nature
trust
fulfilment
individual
UK
e-tailers
Kocak et al. (2007) product functional utility

product symbolic utility
brand name functional utility
brand name symbolic utility
individual
Turkey
sports shoes
Buil et al. (2008)
individual
UK,
Spain
soft drinks,
sportswear,
electronics, cars
aggregate
US
consumer
packaged goods,
groceries
brand awareness
perceived quality
brand loyalty
brand associations
(perceived value, brand
personality, organisational
associations)
Indirect approach via behaviour-based measures

Ailawadi et al.
(2003)
n.a.
Source: the authors
Direct approaches
Studies falling into the former group neglect the theoretical dimensions
of the construct that, if properly operationalised, can provide actionable
insights into the drivers of equity. Ultimately, what these studies are trying
to achieve is a separation of the value of the brand from the value of the
product (e.g. by using the multi-attribute model). Over the years, this has
proved to be conceptually and methodologically problematic as brands
supervene on products, much as the mental has been claimed to supervene
on non-aesthetic properties (Grassl 1999, p. 323).
50
Multi-attribute approaches
Srinivasan (1979), Park and Srinivasan (1994) and Jourdan (2002) use the
multi-attribute model as a common departure point to measure consumerbased brand equity. Srinivasan (1979) defines brand equity, which back
then he called brand-specific effect, as the component of a brands
overall preference that is not explained by the multiattribute model
(p. 12). In line with this definition, Srinivasan (1979) measures brand
equity by comparing observed preferences based on actual choice with
consumer preferences derived from a multi-attribute conjoint analysis. The
difference between overall preference and the preference estimated by the
multi-attribute model is subsequently quantified by means of a monetary
scale (dollar-metric scale). Estimates of brand equity that result from this
method occur, at best, at the segment level. As with every attempt to
measure brand equity directly, this approach does not shed light on the
sources of brand value. Fifteen years later, Park and Srinivasan (1994)
achieved measurement of brand equity at the individual level, which they
operationally defined as the difference between an individual consumers
overall brand preference and his or her multiattributed preference based in
objectively measured attribute levels (p. 273). Objective preferences can
be obtained by laboratory tests, blind tests or surveys with experts. Park
and Srinivasan (1994) also disaggregate consumer-based brand equity into
two parts: an attribute component, based on consumers evaluations of the
brands physical characteristics; and a non-(product) attribute component,
based on symbolic associations attached to the brand. Although it provides
important insights into the perceptual distortions caused by a specific
product attribute, this method does not break down the non-attributebased component of brand equity. It is also naive to assume that experts
(or dentists in the context of the study) are immune from the brand equity
effect and are able to provide objective attribute scores. Jourdan (2002)
notes that the difference of utility implied in the Park and Srinivasan
(1994) definition of brand equity may not entirely be imputable to the
brand as part of it is due to measurement error. Aside from the brand
name effect, overall preference may not coincide with the preference based
on objectively measured product attributes for two other reasons. First, a
consumer may positively evaluate all product attributes yet choose another
brand due to unobservable variables that affect preferences and may
even be random to the individual consumer (i.e. random error). Second,
preference based on objective evaluations of a products attribute levels is
estimated by means of the multi-attribute model, the arbitrary choice of
which, as well as the number and nature of the attributes retained, may
51
be potential sources of systematic error. As such, Jourdan (2002) argues

for an error component, which data from a repeated measures experiment
show is not negligible. His amendment resulted in improved levels of
reliability and validity of brand equity measurement. Also, the choice of a
single sample provides better control of distortional factors. Despite this
methods advantages the complexity of its inherent experimental design
translates into little managerial value.
Other direct approaches

Not deviating substantially from the underlying logic of the studies in the
previous section, Leuthesser et al. (1995) begin with the assumption that
personal evaluation of a given brand on a number of attributes is always
biased. This bias is caused by the fact that consumers are predisposed
towards brands they know. At the level of subjective attribute-by-attribute
evaluations, this predisposition is manifest through the statistical halo
effect (or error): the correlations of inter-attributes in the multi-attribute
model would be higher than if consumers held no a priori overall attitude
(global effect) towards the brand being rated. According to the authors,
it is this perceptual distortion that forms the basis of brand equity. In the
same line of thinking, they postulate that the halo effect corresponds to
the aggregate value of the brand. In order to isolate the effect of this halo,
Leuthesser et al. (1995) describe two statistical procedures: partialling
out and double centering. This method does not provide any indication
of the sources of consumer-based brand equity and is therefore of little
value to brand managers. More importantly, this method does not take
into account the part of consumer-based brand equity that hinges on
associations attached to the brand name. As a result, this method can be
applied only in product categories where the positioning of competitive
brands is functional or experiential (Park et al. 1986). Another weakness
of this method is that equity is at best measured at the aggregate level
rather than the individual level. Finally, the method does not overcome
the shortcomings of previous methods, which draw heavily on statistics,
making it hard to use by brand managers.
Unlike previous studies, which focus on preference, Kamakura and
Russell (1993) examine consumers actual purchase behaviour by means of
a segmentwise logit model. The empirical estimation of the model is based
on real purchase data from supermarket checkout scanners. Consumerbased brand equity is measured as the implied utility or value assigned
to a brand by consumers (p. 10). By removing the effects of short-term
52
advertising and price promotions, Kamakura and Russell (1993) came

up with a proxy of Brand Value. Two major sources of brand equity
were subsequently identified as a result of decomposing Brand Value
into tangible and intangible components. The authors go on to propose
Brand Value and Intangible Brand Value as two alternative measures
of brand equity. While Brand Value provides a fairly good diagnostic
for a brands competitive position, Intangible Brand Value isolates the
utility associated with intangible factors such as brand associations and
perceptual distortions. While providing a preliminary breakdown of
Brand Value, Intangible Brand Value is not decomposed further to enable
brand managers to manage the sources of that value. Similar to Srinivasan
(1979), this approach does not allow evaluating consumer-based brand
equity at the individual consumer level. The method offers the advantage
of reflecting actual consumer behaviour as opposed to preferences, but at
the same time is confined to contexts where scanner data are available.
Finally, the method assumes that brand separability is possible, a position
challenged severely by researchers who adopt an inclusive as opposed
to an additive approach to branding (Barwise et al. 1990; Ambler &
Barwise 1998).
Building on the information economics paradigm, the approach adopted
by Swait et al. (1993) uses the entire utility value attached to a brand
rather than isolating specific parameters thereof. Their argument is that
the effect of brand equity occurs throughout the components of the utility
function and therefore any measure of brand equity should reflect total
utility. Based on this, they propose a new measure of consumer-based
brand equity called Equalisation Price (EP), which encompasses the
monetary expression of the utility a consumer attributes to a bundle
consisting of a brand name, product attributes and price (p. 30). Based
on a hypothetical choice task and additional information relating to
consumers purchases and product usage, image and socio-demographics,
EP is then calculated by means of a multinomial logit model. This is the
hypothetical price at which each brand would have the same market share
in that consumers purchases (Barwise 1993). One of the advantages of
this measuring instrument is that it incorporates a series of qualitative
variables related to symbolic associations. The instrument developed by
Swait et al. (1993) allows identifying the sources of brand associations
and determining importance weights in the function of consumer utility.
Another advantage is that it permits the calculation of consumer-based
brand equity at the individual level. Nevertheless, the specification of the
model postulates that all consumers have identical preferences, rendering
53
the method inappropriate for markets characterised by inhomogeneous

consumer choice.
Shankar et al. (2008) developed a model of brand equity that combines
financial and consumer survey data. The two multiplicative components
of brand equity are offering value and relative brand importance. Offering
value is the net present value of a product or product range carrying a brand
name, and can be estimated through financial measures such as forecast
revenues and margin ratios. Relative brand importance is a measure that
seeks to isolate the effect of brand image on consumer utility relative to
the effect of other factors that also affect consumer choice. Shankar et
al. (2008) identify brand reputation, brand uniqueness, brand fit, brand
associations, brand trust, brand innovation, brand regard and brand
fame as drivers of brand image that may be captured through a consumer
survey. An advantage of this method is that it allows estimating brand
equity for multi-category brands. While this method is beneficial in terms
of combining both financial and consumer data, it makes comparisons
with rival brands difficult due to competitor financial measures often
being unavailable at the brand level. Moreover, this approach produces
an aggregate estimate of brand equity as only relative brand importance is
measured at the individual level.
Indirect approaches
Compared to direct approaches, indirect approaches to CBBE measurement
adopt a more holistic view of the brand and seek to measure brand equity
either through its manifest dimensions or through an outcome variable
such as a price premium.
Lassar et al. (1995) defined consumer-based brand equity as the
enhancement in the perceived utility and desirability a band name confers
on a product (p. 10). Based on a previous study conducted by Martin
and Brown (1990), the authors suggested five CBBE dimensions, namely
performance, value, social image, trustworthiness and commitment. The
hypothesised structure of brand equity was supported by survey data
collected from consumers in two product categories: TV monitors and
watches. Lassar et al. (1995) also reported adequate levels of internal
consistency and discriminant validity for the resultant 17-item Likerttype CBBE scale. Understanding the complexity of previous brand equity
measurement techniques, Lassar et al. (1995) paved the way for a simple
paper-and-pencil instrument that enables managers to easily monitor their
brand equity through its constituent dimensions. Furthermore, the metric
54
can also be applied to various product fields as individual scale items

measure consumer perceptions at a rather abstract, non-product classspecific level. Despite its merits, Lassar et al.s (1995) CBBE scale focuses
solely on associations and excludes significant behavioural components
of brand equity (e.g. behavioural loyalty). Also, the scale was developed
and validated with a convenience sample of 113 consumers, which is
considered inadequate for confirmatory factor analysis (Hinkin 1995).
Finally, the authors do not report any tests on the scales external validity.
Similar to the previous authors holistic definition of brand equity,
Vzquez et al. (2002) define consumer-based brand equity as the overall
utility that the consumer associates to the use and consumption of the
brand; including associations expressing both functional and symbolic
utilities (p. 28). It is notable that the above definition highlights ex-post
(i.e. utilities obtained by consumers following a brands purchase) as
opposed to ex-ante utilities (i.e. utilities obtained prior to purchase), the
latter being the focus of investigation under the information economics
paradigm. Their empirical study involving consumer evaluations of
athletic shoe brands verified the existence of four dimensions of brand
utilities: product functional utility, product symbolic utility, brand name
functional utility and brand name symbolic utility. Further tests including
confirmatory factor analysis showed that their proposed 22-item scale
has strong psychometric properties. An additional conclusion of the
study is that neither the additive nor the inclusive conception of a brand
is supported. Results showed that product and brand utilities maintain
discriminant validity, suggesting that consumers do not view the two
entities (i.e. product and brand) as identical. Nevertheless, the strong intercorrelations between the dimensions require brand managers not to regard
the product and brand name as entirely independent entities. The resulting
scale has a number of advantages over preceding methods of brand
equity measurement. In the first place, unlike previous methods which
involve complex statistical modelling, the Vzquez et al. (2002) method
is relatively easy to administer. Second, the developed scale sheds light
on the sources of consumer-based brand equity through four dimensions.
Third, the scale allows measurement at the individual level. Nevertheless,
the scale was calibrated solely in the athletic shoes sector and therefore
certain adaptations are required to administer the scale in other contexts.
Finally, this method focused on ex-post brand utilities, thus neglecting
significant ex-ante brand utilities. In a follow-up study, Koak et al. (2007)
sought to replicate the results of Vzquez et al. (2002) and to determine
whether their scale could be applied to a different cultural context, i.e.
55
Turkey. Koak et al. (2007) therefore used the same dimensionality of

consumer-based brand equity (i.e. product, functional utility, product,
symbolic utility) and, to facilitate comparability of results, tested the scale
in the same product category (i.e. sport shoes). The results showed that
the original 22-item scale developed by Vzquez et al. (2002) in Spain was
not appropriate for the Turkish sample. Instead a 16-item scale that was
similar but not identical to the original scale was supported by the data.
This led Koak et al. (2007) to conclude that the differences between the
original and the replication study may be due to cultural diversity. In other
words, consumers may arrive at different evaluations of brands as a result
of different cultural conditions.
In a separate endeavour, Yoo and Donthu (2001) sought to develop an
individual-level measure of consumer-based brand equity that is reliable,
valid, parsimonious, and draws on the theoretical dimensions put forward
by Aaker (1991) and Keller (1993). Data to calibrate and validate the
scale were collected from three independent samples of American, Korean
American and Korean consumers. The resultant battery measuring
multi-dimensional brand equity consists of ten items reflecting the three
dimensions of brand loyalty, perceived quality and brand awareness/
associations. To assess MBEs convergent validity, Yoo and Donthu
(2001) further developed a four-item unidimensional (direct) measure of
brand equity, which they labelled as overall brand equity. A strong and
significant correlation was found between the two measures.
Among the indirect approaches to consumer-based brand equity
measurement, the Yoo and Donthu (2001) study arguably has the most
strengths and fewest weaknesses. First, Yoo and Donthus (2001) adoption
of an etic approach to scale development, which refers to the simultaneous
use of samples from multiple cultures, suggests that the scale is culturally
valid. Second, the scale is applicable to various product categories without
requiring further adjustments such as in the case of Vzquez et al. (2002).
Third, the instrument is parsimonious and easy to administer, making it
simple for brand managers to regularly assess the equity of their brands.
Fourth, measurement of brand equity is made at the individual consumer
level. Fifth, the authors carried out a rigorous multi-step validation process.
Nevertheless, the studys limitations, as outlined below, necessitate
further scale development to allow researchers and practitioners to arrive
closer to a universally accepted measure of consumer-based brand equity
(Washburn & Plank 2002).
The major limitation of Yoo and Donthus (2001) three-factor consumerbased brand equity scale is that brand awareness and brand associations,
56
two theoretically distinct underlying constructs of brand equity, collapsed

into one dimension. The question of whether or not brand awareness
and brand associations should be collapsed is critical. Although the two
constructs are clearly correlated, both Aaker (1991) and Keller (1993)
distinguish between brand awareness and associations. According to
Aakers (1991) conceptualisation, brand awareness must precede brand
associations. Nonetheless, the two dimensions are not synonymous
since one can be aware of a brand without having a strong set of brand
associations linked in memory. Pappu et al. (2005) achieved a distinction
between the dimensions of brand awareness and brand associations;
however, their confirmatory factor model suffers from a serious limitation.
Two of brand equitys dimensions brand awareness and brand loyalty
are operationalised by one and two indicators respectively, making
the psychometric properties of their scale questionable (confirmatory
factor analysis requires a minimum of three indicator variables for each
exogenous construct). Another limitation is related to the exclusive
reliance on student samples to develop and validate their brand equity
scale. Students are generally not effective surrogates for consumers (James
& Sonner 2001). A third limitation is concerned with Yoo and Donthus
(2001) selection of product categories. In an era when branding has
become critical for services (van Riel et al. 2001; Brodie et al. 2006), and
several service brands enjoy prominent positions in Interbrands annual
top brands ranking, it is a serious omission that Yoo and Donthu (2001)
chose only product brands for their survey (films, jeans and athletic shoes).
Furthermore, while attentive to cultural variations, their study was based
on specific country cultures. Further evidence about the dimensionality of
brand equity and the constructs invariance among cultures was presented
by Buil et al. (2008), who collected and compared data from consumers
in the UK and Spain. The hypothesised structure of consumer-based brand
equity consisting of brand awareness, perceived quality, brand loyalty and
brand associations (decomposed into perceived value, brand personality
and organisational associations) was supported in both countries. Further
research may look further into the conceptual and metric equivalence of
brand equity such as in individualist vs collectivist cultures, and also in
developed vs developing markets.
Except for measures of consumer-based brand equity intended to
be applicable across product categories, the marketing literature also
reports studies developing category/industry specific measures of brand
equity (e.g. de Chernatony et al. 2004; Christodoulides et al. 2006). For
instance, de Chernatony et al. (2004) identified three dimensions of CBBE
57
specific to financial services brands, namely brand loyalty, satisfaction

and reputation. Similarly, Christodoulides et al. (2006) focused on brand
equity measurement in an online context and through interviews with
experts identified five dimensions of e-tail brand equity, i.e. emotional
connection, online experience, responsive service nature, trust and
fulfilment. Consumer surveys are in both cases used to then assess
consumer-based brand equity, each time through its manifest dimensions.
Price premium
Another way to indirectly measure consumer-based brand equity is
through an outcome variable of consumer-based brand equity, i.e. price
premium. This method calculates the additional income or profit that is
generated as a result of the differential selling price between a branded
and a generic (non-branded) product (Barwise et al. 1989). Ailawadi etal.
(2003) proposed a revenue premium measure as a proxy for measuring
consumer-based brand equity. This is defined as the difference in revenue
(i.e. net price volume) between a branded good and a corresponding
private label (p. 3). Two of the advantages of this measure are the reliance
on actual market data (as opposed to subjective judgements) and the
relative ease of calculation. On the negative side, price premium does
not provide insights into the sources of brand equity. Also, brand equity
building usually implies one of two generic strategies: a price premium
strategy or a market share strategy (Park & Srinivasan 1994). In the former
case, revenue premium provides satisfactory results. However, in the case
where the brand in question strives to increase its market share, the price
premium method fails to deliver accurate results of brand value. Third,
often no equivalent generic product is available and, even when available,
it is likely to be extremely difficult to obtain a breakdown of competitors
profitability figures by individual product line.
In parallel with the academic research on brand equity measurement,
various consultancies and market research firms have also developed
their own methodologies, which cannot be neglected as they occasionally
appear in scholarly research (e.g. Chu & Keh 2006; Mizik & Jacobson
2008). The best-known methodologies are summarised in Table 3.
Comparing the measures used by different consultancies (see Table 3) as
well as the measures used by consultancies and academics (see Tables 1
and 3) shows little common ground in terms of the constituent dimensions
of brand equity.
58
Table 3 Consultancy-based measures of CBBE

Name
Measures of CBBE
Interbrand Brand Strength Market

Stability
Brand leadership
Trend
Brand support
Diversification
Protection
Y&R Brand Asset Valuator Knowledge
Esteem
Relevance
Differentiation
WPP Brand Dynamics Presence
Relevance
Performance
Advantage
Bonding
Research International Equity Engine Affinity
Perceived functional performance
The interaction between the brands equity and its price
Source: the authors
Conclusions and directions for future research

Our extensive literature review identified two streams of research with
regard to CBBEs conceptualisation: the dominant stream derives from
cognitive psychology, while a secondary stream draws on signalling theory
from information economics. Although there is some agreement with
regard to the definition of brand equity as the added value endowed by
the brand to the product, additive approaches to measuring brand equity
have recently given way to more holistic metrics. Our literature review
identifies two main classes of CBBE measurement methods. First, methods
that seek to quantify brand equity directly; and, second, methods that
seek to measure brand equity either through its demonstrable dimensions
or through price premium (outcome variable). It is mostly earlier studies
that attempted to measure brand equity directly (e.g. Srinivasan 1979)
and faced serious problems such as brand separability. Moreover, direct
techniques have limited managerial value as they usually rely on complex
statistical models and provide no insights into the sources of brand value.
Indirect approaches, by contrast, use simple pen and pencil instruments
to tap consumer-based brand equity through its individual dimensions.
59
Despite their managerial usefulness as a diagnostic tool, indirect measures

of brand equity still have limitations, some of which derive from the lack
of agreement on what dimensions constitute consumer-based brand equity,
although a wave of studies (e.g. Yoo and Donthu 2001; Washburn & Plank
2002; Pappu et al. 2005; Buil et al. 2008) endorse Aakers (1991, 1996)
dimensionality. We believe that there is no such thing as a universal measure
for brand equity, and that the market sector and life-stage of the brand need
to be taken into account when selecting an appropriate set of measures to
assess brand equity (Baker et al. 2005). It is notable from our literature
review that while some CBBE facets may be product or category specific
(e.g. car performance), others are softer and likely to be more generic in
their applicability and scope (e.g. trust) (cf. Morgan 2000). Tables 1 and 3
suggest that not only is there diversity in the views of academics on CBBEs
dimensionality, but also there seems to be a gap between their views and
consultants views. This may be because consultants have a business model
based on which they generate an income stream through their proprietary
methodologies. Future research should investigate the relevance of CBBE
measures for practising managers. Furthermore, our literature review shows
a bias to reporting measures developed and validated in the US. Future
research should apply these scales to other contexts to assess the conceptual
and psychometric equivalence of CBBE measures across cultures. Finally,
existing CBBE measures relied heavily on evaluations of product (especially
fmcg) brands. Branding is essential not just for products but also for
services. As such, CBBE measures should be tested with service brands, and
if necessary new service-specific brand equity scales should be developed.
Furthermore, we recognise that a market consists of clusters of individuals
with different levels of loyalty (Dick & Basu 1994; Morgan 2000). Research
also shows that each of these clusters may differ significantly on brand
equity, suggesting that aggregate brand equity scores may be misleading
(Rust et al. 2004). Future research needs to address the relative strength of
brand equity by type of user. For instance, are committed, loyal customers
typically the most valued? Why should habitual customers not be just as
valued (Knox 2001)?
Finally, the largest contribution (in terms of volume) to the existing
body of research knowledge on brand equity comes from studies
adopting a psycho-cognitive perspective, which is largely underpinned
by a linear consumer thinking process (see, for example, hierarchy of
effects). Emerging advances in neural psychology allude to much more
complex brain activity when consumers process a brand (e.g. Quartz &
Asp 2005; Yoon et al. 2006; Moutinho & Santos 2009). For instance,
60
Yoon et al. (2006) have recently challenged the view that the processing
of products and brands is akin to that of humans. Further developments
in neurosciences are expected to provide a better understanding of how
consumers feel about, resonate and value a brand.
The following pointers are intended to assist practising marketers and
market researchers in setting up their own system to measure brand equity.
1. Brand equity is a complex and multi-faceted concept and, as such, it
needs to be captured through a set of measures rather than a single
measure. The measures selected should be aligned with the brand
vision (Davis 2000). For instance, a supermarket like Waitrose is
likely to be more concerned about perceived quality than a discounter,
which would focus more on value for money. Measures should also
be attentive to the organisational culture of the corporation since
externally focused organisations with flexible structures would be
more attentive to particular facets than internally focused organisations
with stable structures.
2. Understanding brand equity is about understanding customer value
within a particular situational context and level of co-producing value.
It is therefore important for brand managers and market researchers to
know how their brand contributes to the overall product experience.
3. Brand category is an important variable in brand equity measurement.
The usefulness of different dimensions of brand equity is not uniform
across diverse industries. A brand equity monitor should incorporate
dimensions that drive value within the specific industry (e.g. satisfaction
for financial services (de Chernatony et al. 2004) and online customer
experience for e-brands (Christodoulides et al. 2006). The holy grail
of universal measures is akin to fools gold a shining example of
statistical depth with little that drives significant growth.
4. Brand equity monitor systems should consist of perceptual and
motivational factors that can be modelled against consequential
behavioural (e.g. purchase recency/frequency) measures.
5. Functional (e.g. performance), emotional (e.g. affinity) and experiential
facets should be considered for inclusion in a brand equity measurement
system to truly appreciate the evolving nature of brands.
6. In recessionary periods, managers may be excessively focused on the
short-term fiscal outcomes of brand strategies. Insightful consultancy
advice needs to exhort the importance of appreciating the facets of
brand equity that have given rise to monetary gain and how this can
be sustained.
61
References
Aaker, D.A. (1991) Managing Brand Equity. New York: Free Press.
Aaker, D.A. (1996) Building Strong Brands. New York: Free Press.
Aaker, D.A. & Keller, K.L. (1990) Consumer evaluations of brand extensions. Journal of
Marketing, 54, 1, pp. 2741.
Agarwal, M.K. & Rao, V.R. (1996) An empirical comparison of consumer-based measures of
brand equity. Marketing Letters, 7, 3, pp. 237247.
Ailawadi, K.L., Lehmann, D.R. & Neslin, S.A. (2003) Revenue premium as an outcome
measure of brand equity. Journal of Marketing, 67, 4, pp. 117.
Ambler, T. (2003) Marketing and the Bottom Line: Creating the Measures of Success. London:
Financial Times/Prentice Hall.
Ambler, T. & Barwise, P. (1998) The trouble with brand valuation. Journal of Brand
Management, 5, 6, pp. 367377.
Baker, C., Nancarrow, C. & Tinson, J. (2005) The mind versus market share guide to brand
equity. International Journal of Market Research, 47, 5, pp. 523540.
Barwise, P. (1993) Introduction to the special issue on brand equity. International Journal of
Research in Marketing, 10, 1, pp. 38.
Barwise, P., Higson, C., Likierman, A. & Marsh, P. (1989) Accounting for Brands. London:
London Business School.
Barwise, P., Higson, C., Likierman, A. & Marsh, P. (1990) Brands as separable assets. Business
Strategy Review, 1, 2, pp. 4359.
Berry, L. (2000) Cultivating service brand equity. Journal of the Academy of Marketing
Science, 28, 1, pp. 128137.
Berthon, J.P., Capon, N., Hulbert, J., Murgolo-Poore, N.J., Pitt, L. & Keating, S. (2001)
Organizational and Customer Perspectives on Brand Equity: Issues for Managers and
Researchers. Auckland: ANZMAC, Massey University.
Blackston, M. (1992) Building brand equity by managing the brands relationships. Journal of
Advertising Research, 32, 3, pp. 7983.
Bottomley, P.A. & Doyle, J.R. (1996) The formation of attitudes towards brand extensions:
testing and generalising Aaker and Kellers model. International Journal of Research in
Marketing, 13, 4, pp. 365377.
Brodie, R.J., Glynn, M.S. & Little, V. (2006) The service brand and the service-dominant
logic: missing fundamental premise or the need for stronger theory? Marketing Theory, 6, 3,
pp. 363379.
Buil, I., de Chernatony, L. & Martinez, E. (2008) A cross-national validation of the consumerbased brand equity scale. Journal of Product and Brand Management, 17, 6, pp. 384392.
Burmann, C., Jost-Benz, M. & Riley, N. (2009) Towards an identity-based brand equity
model. Journal of Business Research, 62, pp. 390397.
Capron, L. & Hulland, J. (1999) Redeployment of brands, sales forces, and general marketing
expertise following horizontal acquisitions: a resource-based view. Journal of Marketing, 63,
2, pp. 4154.
Christodoulides, G., de Chernatony, L., Furrer, O. & Abimbola, T. (2006) Conceptualising
and measuring the equity of online brands. Journal of Marketing Management, 22, 7/8,
pp.799825.
Chu, S. & Keh, H.T. (2006) Brand value creation: analysis of the InterbrandBusiness Week
brand value rankings. Marketing Letters, 17, 4, pp. 323331.
Cobb-Walgren, C.J., Ruble, C.A. & Donthu N. (1995) Brand equity, brand preference, and
purchase intent. Journal of Advertising, 24, 3, pp. 2540.
Davis, S.M. (2000) Brand Asset Management. San Francisco, CA: Jossey-Bass.
62
Dawar, N. & Pillutla, M.M. (2000) Impact of product-harm crises on brand equity: the
moderating role of consumer expectations. Journal of Marketing Research, 37, 2, pp.215
226.
de Chernatony, L., Harris, F.J. & Christodoulides, G. (2004) Developing a brand performance
measure for financial services brands. Services Industries Journal, 24, 2, pp. 1533.
Dick, A.S. & Basu, K. (1994) Customer loyalty: towards an integrated framework, Journal of
the Academy of Marketing Science, 22, 2, pp. 99113.
Dodds, W.B., Monroe, K.B. & Grewal, D. (1991) Effects of price, brand, and store
information on buyers product evaluations. Journal of Marketing Research, 28, 3, pp.307
319.
Erdem, T. & Swait, J. (1998) Brand equity as a signaling phenomenon. Journal of Consumer
Psychology, 7, 2, pp. 131157.
Erdem, T., Swait, J. & Louviere, J. (2002) The impact of brand credibility on consumer price
sensitivity. International Journal of Research in Marketing, 19, 1, pp. 119.
Erdem, T., Swait, J. & Valenzuela, A. (2006) Brands as signals: a cross country validation
study. Journal of Marketing, 70, 1, pp. 3449.
Falkenberg, A.W. (1996) Marketing and the wealth of firms. Journal of Macromarketing, 16,
4, pp. 424.
Farquhar, P. (1989) Managing brand equity. Journal of Advertising Research, 30, 4, pp. RC7
RC12.
Farquhar, P.H., Han, J.Y. & Ijiri, Y. (1991) Recognizing and Measuring Brand Assets.
Report# 91-119. MA: Marketing Science Institute.
Feldwick, P. (1996) What is brand equity anyway, and how do you measure it? Journal of the
Market Research Society, 38, 2, pp. 85104.
Fournier, S. (1998) Consumers and their brands: developing relationship theory in consumer
research. Journal of Consumer Research, 24, 4, pp. 343373.
Grassl, W. (1999) The reality of brands: towards an ontology of marketing. American Journal
of Economics and Sociology, 58, 2, pp. 313359.
Haigh, D. (1999) Understanding the Financial Value of Brands. Brussels: European
Association of Advertising Agencies.
Hinkin, T.R. (1995) A review of scale development practices in the study of organizations.
Journal of Management, 21, 5, pp. 967988.
Hunt, S.D. & Morgan, R.M. (1995) The comparative advantage theory of competition.
Journal of Marketing, 59, 2, pp. 115.
James, W.L. & Sonner, B.S. (2001) Just say no to traditional student samples. Journal of
Jourdan, P. (2002) Measuring brand equity: proposal for conceptual and methodological
improvements. Advances in Consumer Research, 29, 1, pp. 290298.
Kamakura, W.A. & Russell, G.J. (1993) Measuring brand value with scanner data.
International Journal of Research in Marketing, 10, 1, pp. 922.
Kapferer, J.-N. (2004) The New Strategic Brand Management. London: Kogan Page.
Keller, K.L. (1993) Conceptualizing, measuring and managing customer-based brand equity.
Keller, K.L. (2003) Strategic Brand Management: Building, Measuring, and Managing Brand
Equity. Upper Saddle River, NJ.
Keller, K.L. & Lehmann, D.R. (2006) Brands and branding: research findings and future
priorities. Marketing Science, 25, 6, pp. 740759.
Kerin, R.A. & Sethuraman, R. (1998) Exploring the brand valueshareholder value nexus for
consumer goods companies. Journal of the Academy of Marketing Science, 26, 4, pp.260
273.
63
Koak, A., Abimbola, T. & Ozer, A. (2007) Consumer brand equity in a cross-cultural
replication: an evaluation of a scale. Journal of Marketing Management, 23, 1/2, pp. 157
173.
Knox, S. (2001) Measuring and managing brand loyalty. Journal of Strategic Marketing, 9,
pp.111128.
Lassar, W., Mittal, B. & Sharma, A. (1995) Measuring customer-based brand equity. Journal
of Consumer Marketing, 12, 4, pp. 1119.
Leuthesser, L., Kohli, C.S. & Harich, K.R. (1995) Brand equity: the halo effect measure.
European Journal of Marketing, 29, 4, pp. 5766.
Martin, G.S. & Brown, T.J. (1990) In search of brand equity: the conceptualization and
measurement of the brand impression construct. In T.L. Childers et al. (eds) Marketing
Theory and Applications, Vol. 2. Chicago, IL: American Marketing Association, pp. 431
438.
Mizik, N. & Jacobson, R. (2008) The financial value impact of perceptual brand attributes.
Journal of Marketing Research, 45, 1, pp. 1532.
Morgan, R.P. (2000) A consumer-orientated framework of brand equity and loyalty.
International Journal of Market Research, 42, 1, pp. 6578.
Moutinho, L. & Santos, J.P. (2009) Neuroscience in marketing: empirical evidence of social
and emotional meanings conveyed by brands. Presentation at the 5th Thought Leaders
International Conference on Brand Management, Athens.
MSI (2008) Research Priorities: Guide to MSI Research Programs and Procedures. MA:
Marketing Science Institute. Available at: www.msi.org/pdf/MSI_RP08-10.pdf (accessed
2October 2008).
Netemeyer, R.G., Krishnan, B., Pullig, C., Wang, G., Yagci, M., Dean, D., Ricks, J. & Wirth, F.
(2004) Developing and validating measures of facets of customer-based brand equity. Journal
of Business Research, 57, l.2, pp. 209224.
OSullivan, D. & Abela, A.V. (2007) Marketing performance measurement ability and firm
performance. Journal of Marketing, 71, 2, pp. 7993.
Pappu, R., Quester, P.G. & Cooksey, R.W. (2005) Consumer-based brand equity: improving
the measurement empirical evidence. Journal of Product and Brand Management, 14, 3,
pp. 143154.
Park, C.S. & Srinivasan, V. (1994) A survey-based method for measuring and understanding
brand equity and its extendibility. Journal of Marketing Research, 31, 2, pp. 271288.
Park, W.C., Jaworski, B.J. & MacInnis, D.J. (1986) Strategic brand concept-image
management. Journal of Marketing, 50, 4, pp. 135145.
Quartz, S. & Asp, A. (2005) Brain branding: brands on the brain. ESOMAR Annual Congress,
Cannes.
Rangaswamy, A., Burke, R.R. & Oliva, T.A. (1993) Brand equity and the extendibility of
brand names. International Journal of Research in Marketing, 10, 1, pp. 6175.
Reynolds, T.J. & Phillips, C.B. (2005) In search of true brand equity metrics: all market share
aint created equally. Journal of Advertising Research, 45, 2, pp. 171186.
Rust, R.T., Zeithaml, V.A. & Lemon, K.N. (2000) Driving Customer Equity. New York: The
Free Press.
Rust, R.T., Zeithaml, V.A. & Lemon, K.N. (2004) Customer-centered brand management.
Harvard Business Review, 82, 9, pp. 110118.
Shankar, V., Azar, P. & Fuller, M. (2008) BRAN*EQT: a multicategory brand equity model
and its application at Allstate. Marketing Science, 27, 4, pp. 567584.
Sharp, B. (1995) Brand equity and market-based assets of professional service firms. Journal of
Professional Services Marketing, 13, 1, pp. 313.
Simon, C.J. & Sullivan, M.V. (1993) The measurement of determinants of brand equity: a
financial approach. Marketing Science, 12, 1, pp. 2852.
64
Srinivasan, V. (1979) Network models for estimating brand-specific effects in multi-attribute

marketing models. Management Science, 25, 1, pp. 1121.
Srivastava, R.K. & Shocker, A. (1991) Brand equity: a perspective on its meaning and
measurement, Working Paper 91-124. Cambridge, MA Marketing Science Institute.
Srivastava, R.K., Shervani, T.A. & Fahey, L. (1998) Market-based assets and shareholder value:
a framework for analysis. Journal of Marketing, 62, 1, pp. 218.
Swait, J., Erdem, T., Louvire, J. & Dubelaar, C. (1993) The equalization price: a measure of
consumer-perceived brand equity. International Journal of Research in Marketing, 10, 1,
pp.2345.
Van Riel, A.C.R., Lemmink, J. & Ouwersloot, H. (2001) Consumer evaluations of service
brand extensions. Journal of Service Research, 3, 3, pp. 220231.
Vzquez, R., Del Rio, A.B. & Iglesias, V. (2002) Consumer-based brand equity: development
and validation of a measurement instrument. Journal of Marketing Management, 18, 1/2,
pp. 2748.
Washburn, J.H. & Plank, R.E. (2002) Measuring brand equity: an evaluation of a consumerbased brand equity scale. Journal of Marketing Theory and Practice, 10, 1, pp. 4661.
Winters, L.C. (1991) Brand equity measures: some recent advances. Marketing Research, 3, 4,
pp. 7073.
Yoo, B. & Donthu, N. (2001) Developing and validating a multidimensional consumer-based
brand equity scale. Journal of Business Research, 52, 1, pp. 114.
Yoo, B., Donthu, N. & Lee, S. (2000) An examination of selected marketing mix elements and
brand equity. Journal of the Academy of Marketing Science, 28, 2, pp. 195211.
Yoon, C., Gutchess, A.H., Feinberg, F. & Polk, T.A. (2006) A functional magnetic resonance
imaging study of neural dissociations between brand and person judgments, Journal of
Consumer Research, 33, 1, pp. 3140.
About the authors

George Christodoulides is Lecturer in Marketing at the University of
Birmingham Business School. His research focuses on branding and
e-marketing, particularly the way the internet and other interactive
technologies affect brands. Christodoulides is a regular presenter at
national and international conferences and his research has appeared
in highly ranked journals including the Journal of Advertising Research,
Journal of Marketing Management, Marketing Theory, Service Industries
Journal and Journal of Product and Brand Management. He is the principal
investigator of a 100,000 grant funded by Economic and Social Research
Council (ESRC) examining brand equity measurement in Europe.
Leslie de Chernatony is Professor of Brand Marketing at the Universit
della Svizzera italiana, Lugano, Switzerland, Honorary Professor at Aston
Business School, UK and Managing Partner at Brands Box Marketing
& Research Consultancy. His research is globally disseminated through
books, international conference presentations, TV and radio broadcasts
and numerous journal articles, some of which won best paper prizes. He
has run many highly acclaimed brand strategy workshops throughout the
65
world. His advice about brand management has been sought by numerous
organisations. He has acted as an expert witness in court cases over
branding issues.
Address correspondence to: George Christodoulides, Lecturer in
Marketing, Birmingham Business School, The University of Birmingham,
University House, Edgbaston, Birmingham B15 2TT, United Kingdom
Email: G.Christodoulides@bham.ac.uk
66
The importance of social motives for

watching and interacting with digital
television
Steven Bellman, Anika Schweda and Duane Varan
Interactive Television Research Institute
Contrary to a key assumption of the TV industry, interaction with digital

interactive TV (iTV) programmes and ads is driven as much by social motivations
as it is by information seeking. This insight was revealed by a survey of a
representative sample of 867 digital TV households in the UK, which has one
of the largest and most experienced digital iTV audiences in the world. This
new survey used a comprehensive but efficient set of motivation items, so
that no important motivations were left out, which may explain why social
motivations emerged as important in this study, whereas they have not been in
studies of traditional TV watching. Suggestions are made for how marketers and
programme producers can make iTV content that appeals to viewers who are
motivated by social needs.
Introduction
Marketers can use differences in motivations to craft appeals that are
attractive only to the most profitable segments of consumers. These
motivations are perhaps best revealed in in-depth, personal interviews (e.g.
Valentine & Evans 1993), but can also be uncovered using quantitative
surveys, which are needed to measure the relative sizes of potential
segments. A survey of consumer motivations needs to be efficient, asking
the minimum possible number of questions, but at the same time it needs to
be comprehensive, so that no important motivations are left out. This study
illustrates the use of one such survey, which uncovered the motivations for
watching and also interacting with television programmes and ads in the
United Kingdom (UK) digital TV audience.
Received (in revised form): 30 October 2008

DOI: 10.2501/S1470785310201065
67
The importance of social motives for watching and interacting with digital television
Digital transmission has brought new capabilities to TV, such as

interaction with programmes and ads. Viewers interact with their TV
by pressing coloured buttons on their remote controls in response to
text or symbols superimposed on the picture. For example, British Sky
Digital (Sky) in the UK offers advertisers an interactive television (iTV)
ad format called the dedicated advertiser location (DAL), which offers
an interactive experience similar to a small internet website. In the United
States, analogue television was switched off on 12 June 2009 (www.fcc.
gov), but the United Kingdom is also one of the worlds leading digital
markets, with 80% of households receiving digital television, and the
longest experience worldwide with digital iTV advertising. The first iTV
ad appeared on Sky in March 2000, and five years later the Sky platform
had broadcast over 600 interactive campaigns for a variety of advertisers
(Zip 2005). The experience of digital television viewers in the UK provides
a glimpse into the future, and many lessons for potential iTV advertisers
around the world.
We generated a new survey for this study because we suspected that
existing surveys of the motivations of TV audiences, which have come out
of the uses and gratifications tradition in audience research, under-sample
social motivations for TV watching (e.g., to increase self-assurance, or selfpride). It is possible that social motivations were not important for watching
analogue TV, but with the switchover to digital and the introduction of
interactive programmes and ads, we thought it was time to reassess whether
the motivations of the audience have also changed. While it is well known
that social influence has a powerful effect on the adoption of innovations
(e.g. Rogers 1995), such as digital iTV, digital iTV advertisers apparently
assume that, after it has been purchased, this technology is mainly used to
satisfy the same old needs satisfied by watching analogue TV: information
seeking and entertainment. For example, the advice Sky gives digital iTV
advertisers is: dont forget that iA [iTV advertising] is still advertising, so
keep it entertaining as well as informing (www.skymedia.co.uk). Recent
research into the uses and gratifications of other interactive technologies,
such as the internet (e.g. Papacharissi & Rubin 2000), have highlighted
the association of social motivations with usage of these technologies, and
we thought it was likely that these motivations would be important for
interacting with digital TV as well.
This paper documents a survey of a representative sample of 867 UK
digital TV viewers, which asked questions about eight basic motivations
for consumer behaviour, including informational, entertainment and social
68
motivations, to uncover their motivations for watching and interacting

with digital TV.
Background
Interactive television in the United Kingdom
Figure 1 shows the penetration of digital television in the UK since 1992.
By January 2008, eight in ten (82%, 21.1 million) of the 25.6 million TV
households in the UK had access to digital television (BARB 2008). Most
accessed digital television over-the-air, via Freeview Digital Terrestrial
Television (DTT: 56%, 12.0 million), or via satellite (42%, 8.9 million)
or cable (3.4 million, 16%), or combinations of these three (which is why
the percentages sum to over 100%). In the UK, however, the Sky network
which had nearly 9 million customers in June 2008, i.e. 43% of digital
households (www.sky.com) is the only network that offers a return path
for digital advertising, via a phone line connected to the set-top box on
the TV.
Interactive television advertising viewer segments

The UK digital iTV research firm, Zip Television, identified five segments
of digital TV viewers in the UK, using a cluster analysis of 33 items
measuring attitudes towards TV ads, interactive TV and iTV ads (Zip 2005).
100%
90%
80%
Terrestrial
Cable
Satellite
70%
60%
50%
40%
30%
20%
10%
0%
1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007
Figure 1 The growth of digital TV in the United Kingdom
69
Only one of these segments consisted of current interactors. Dedicated

Interactors love interactive TV, and also have a favourable attitude
towards advertising. They talk about ads with their friends, and interact
with ads that offer free samples, provided they are interested in the product
or service being advertised.
The Zip survey also identified some specific appeals advertisers could
use to tempt the other segments to interact. For example, Apprehensive
Stargazers are concerned about interacting with ads. They think interaction
with an ad might be boring, or may mean they miss part of the programme,
and also find it hard to tell if ads are interactive. But, like Dedicated
Interactors, they talk about ads with their friends and are enticed by free
samples and giveaways. If a trusted celebrity told them how to interact
with an ad, Apprehensive Stargazers would probably do it. Two other
segments, however, Unimpressed Pragmatists and Unengaged Passives,
are unresponsive to freebies and competitions. And Bitter Ad Haters
have a negative attitude towards advertising (they do not talk about ads
with their friends), and have not interacted with an ad so far because they
think it would be boring and they would miss part of the programme.
They are unresponsive to offers of free samples or advice from trusted
celebrities. They also find it hard to tell if an ad is interactive, so some of
their negative opinion may be based on ignorance.
Limitations of the Zip report

While Zips (2005) pioneering study identified these five useful segments
in the digital iTV audience in the UK, it also had two major limitations.
First, Zip used purposive sampling of customers from one digital platform
only (Sky), rather than representative sampling of the entire UK digital
TV audience. It also deliberately over-sampled interactors, which meant it
was unable to estimate the number of households in the UK that were in
each segment.
Another limitation of the Zip study was that it did not survey the
motivations of these segments. It suggested some useful appeals that
advertisers could use to target the five segments, but it did not help them
to create these appeals, because it did not investigate what would drive
the different segments to act on these appeals (Rossiter & Bellman 2005).
The main purpose of this study was to uncover these motivations, using
a comprehensive list of motivations for consumer behaviour, such as the
consumption of TV programmes.
70
Eight basic motivations for consumer behaviour

Fennell (1978) analysed consumers wants and product-use situations
and suggested that all product purchases can be explained by just eight
basic motivations. Rossiter and Bellman (2005) have categorised these
eight motives into two types: negatively originated motives (informational
motives arising from a problem with the current state) and positiveending motives (transformational motives arising from the opportunity to
achieve a better-than-normal state). They argue that their drive-reduction
(informational) versus drive-increasing (transformational) classification is
more logically consistent, and therefore more managerially useful, than
similar classifications, such as utilitarian versus hedonic (Katz 1960;
Locander & Spivey 1978; Batra & Ahtola 1990). They also argue that
extensions to this basic list of eight are redundant restatements of one or
more of the basic eight. For example, the two additional motives proposed
by Donovan and Henley (2003), consistency with personal values and
social conformity, can be subsumed back into the motive of social
approval. The five negatively originated (informational) motives are as
follows.
1. Problem Removal. The consumer is currently experiencing a problem
and seeks a product or service that will solve that problem. In the
context of TV watching, an example might be relaxing in front of the
set to unwind after a hard days work.
2. Problem Avoidance. The consumer seeks a product or service that
will prevent a future problem. For example, a viewer might check the
weather on the TV to see if it is likely to rain.
3. Incomplete Satisfaction. A current product or service is not a problem,
but then again, it is not completely satisfying either, and the consumer
seeks a better one. This would be the typical motivation for switching
channels.
4. Mixed Approach-Avoidance. A product or service generates mixed
emotions: there are some things the consumer likes and others that
are disliked, and again this prompts the search for something better.
For example, viewers might appreciate the variety offered by a pay-TV
service, but wish the subscription fees were cheaper.
5. Normal Depletion. This empty larder motivation drives a lot of
habitual consumer behaviour. An example is subscribing to a cable
network to ensure a regular supply of the type of TV content the
consumer prefers.
71
As Rossiter and Bellman (2005) highlight, life would be pretty dull if all
we were motivated to do was to return to a normal state from a problem
state. The three positive-ending (transformational) motives offer the
possibility of achieving, perhaps only briefly, a better-than-normal state,
as follows.
1. Sensory Gratification. The consumer seeks enjoyment from the
product or service, through stimulation of the senses. This would be
the typical reason for choosing to watch an entertaining TV show.
2. Intellectual Stimulation and Mastery. In the Rossiter and Bellman
(2005) classification, seeking information out of curiosity rather than
to solve a problem is a transformational motive. This would be the
typical motivation for watching documentaries.
3. Social Approval. The consumer seeks recognition from social reference
groups through the use of the product or service. An example is
watching a TV show to share in the next days conversation around
the water cooler.
We used items from previous studies of TV watching to measure these
eight basic motives. This allowed us to compare our results with those of
the previous studies.
Motivations for watching television

The eight motivations identified by Rossiter and Bellman can be aligned
with many of the motivations identified in uses and gratifications studies
of TV watching. While debate continues over the precise meaning of the
term uses and gratifications (Anderson & Meyer 1975; OGuinn &
Faber 1991), the uses and gratifications perspective was the first to apply
the concept of segmentation based on motives to audiences for radio
and television. Herzog (1941) observed that the same radio soap opera
could be listened to for very different reasons (e.g. a good cry, escape,
excitement or advice), with presumably different effects. In the 1970s
the uses and gratifications tradition was criticised for its vagueness and
profusion of constructs, but since Rubin (1983), researchers have generally
used the same core set of items (developed originally by Greenberg 1974)
in order to stabilise the field and replicate findings.
Rubin (1983) used factor analysis to reduce the number of potential
motivations in this list of 27 items down to five factors: (1) entertainment
(e.g. I watch TV because it entertains me); (2) pass time/habit (e.g.
72
when I have nothing better to do); (3) information (e.g. so I can learn
how to do things that I havent done before); (4) escape (e.g. so I can
get away from the rest of the family or others); and (5) companionship
(e.g. so I wont have to be alone). However, Rubins companionship
factor tapped parasocial relationships only (Horton & Wohl 1956); the
items he used to measure social interaction e.g. so I can talk with other
people about whats on (McQuail et al. 1972) did not form a reliable
factor, and for that reason this motive was dropped from his analysis.
Companionship was not substantially associated with TV usage. His
conclusion was that TV watching is not socially motivated, and this has
been a key assumption ever since. In this study, we add to Rubins core set
of items, in an attempt to carry out a theoretically exhaustive sample of
the motivations associated with usage of digital iTV, which hopefully will
resurrect social motivations as important for interacting and even simply
watching digital iTV.
Motivations for using other media

More recent research into the motivations for using new interactive media
such as the internet have found that social motivations can be as important
as needs for information and entertainment (Eighmey 1997; Eighmey
& McCord 1998; Flaherty et al. 1998; Korgaonkar & Wolin 1999;
Papacharissi & Rubin 2000; Kaye & Johnson 2002; Luo 2002; Stafford
et al. 2004; Pew Internet & American Life Project 2006). This suggested
to us that social motivations may also be important for watching and
interacting with digital iTV, perhaps as important as entertainment and
information seeking, or even more important. For example, in the most
recent study closely related to this one, Livaditi et al. (2003) identified
entertainment/companionship as the primary motivation for watching
interactive TV, with information-seeking as secondary.
Selection of motivation items

In this section, we describe how we arrived at our final set of motivation
items. This selection was not straightforward, as Rubins (1983) five
factors appeared to confound motivations that are logically distinct,
according to the Rossiter and Bellman (2005) framework. For example,
entertainment (sensory gratification) is a positive-ending (transformational)
motive, while passing time (problem removal) is a negatively originated
(informational) motive. Yet both were correlated with the same canonical
73
root (entertainment) in Rubins analysis. It is a relatively easy matter,

however, to treat these separately, a priori, in our analysis. A more difficult
problem is created by Rubins information-seeking motivation. Watching
for information could be driven by a negatively originated motivation
(problem avoidance) or a positive-ending motivation (intellectual
stimulation and mastery). We therefore used two different items to
measure these two motivations. The items used to measure the uses and
gratifications associated with the internet (Papacharissi & Rubin 2000;
Stafford et al. 2004) provided us with an additional pool from which to
select measures of the eight basic motives. We also selected items from
studies of the uses and gratifications associated with two technologies
closely related to digital iTV: remote control devices (RCDs; Walker &
Bellamy 1991; Walker et al. 1993), and videocassette recorders (VCRs;
Rubin & Bantz 1987; Rubin & Rubin 1989). Table 1 lists the eight basic
motivations and the items we used to measure them, citing the source(s)
for each item, and the name of the factor they loaded on in their source
study.
The following section describes the other measures included in this
survey, and the sample used, which consisted of 867 viewers, representative
of the audience for digital TV in the UK. We then report the results of this
survey and discuss the implications of these results for practitioners and
researchers.
Method
A computer-aided telephone interview (CATI) survey was carried out by
the same firm that conducted the previous Zip (2005) survey, over the
period 24 November to 1 December 2005.
Sample
The total sample included 1010 participants, but, of these, only 867 were
asked the full set of questions, including the 18 motivation items. Quotas
ensured that the sample was highly representative of the 16.5 million digital
satellite (DSAT) homes in the UK in late 2005 (Ofcom 2005). Table 2
lists the demographics of the sample. Also shown in Table 2 are profiles
of the UK population as a whole and the profile of the UK population
with access to digital TV (Continental Research 2005). Our sample was
not statistically different from either of these two populations, which
are very similar because of the high penetration of digital TV in the UK.
74
Table 1 Uses and gratifications items and factors from previous studies
Motivations
Items (I watch digital iTV because )
Negatively-originated motives
1. Problem removal
I can find out whats on without looking at the TV listings in
thepaper or TV guide (GETTING MORE FROM TV; 4)*
It is a more convenient way to see movies (MOVIE RENTAL; 2)*
2. Problem avoidance
I dont have to watch as many adverts (AVOIDING COMMERCIALS; 4, 5)*

Its good for information (INFORMATION; 6)*
3. Incomplete satisfaction
I have enough choices to change channel when something I dont

like comes on (SELECTIVE AVOIDANCE; 4, 5)*
I like the freedom to set my own schedule (TIME SHIFTING; 2)
4. Mixed approach-avoidance Its good for when I have nothing better to do (PASS TIME/HABIT; 1, 6)*
It is relatively cheap (CONVENIENCE; 6)*
5. Normal depletion
Positive-ending motives
6. Sensory gratification
It gives me choice over what I watch (FREEDOM OF CHOICE; 3)*

I like to have something available for the kids to watch (CHILDRENS
VIEWING; 2)*
It is entertaining (ENTERTAINMENT; 1, 2, 6)
Its enjoyable (ENTERTAINMENT; 1)
I just like to use it (ENTERTAINMENT; 1, 6)
7. Intellectual stimulation
I can learn something (INFORMATION; 3)

I get to see more news (INFORMATION; 4)
8. Social approval
I want to entertain people who come over (SOCIAL INTERACTION; 1, 3)

I like to participate in programmes (INTERPERSONAL UTILITY; 6)*
Its good company when Im on my own (COMPANIONSHIP; 1)*
Notes: All items measured on a five-point scale (match your reasons for watching [PLATFORM]?: 1 = do
not match at all to 5 = match exactly). Numbers in brackets indicate the source(s) of the item: (1) Rubin
(1983); (2) Rubin & Bantz (1987); (3) Rubin & Rubin (1989); (4) Walker & Bellamy (1991); (5) Walker et al.
(1993); (6)Papacharissi & Rubin (2000).
* Item wording modified to measure iTV usage in the UK.
However, since our sample is slightly skewed towards older viewers, we

controlled for age in our analyses, as well as the other two demographics
in Table 2, gender and children. Since many homes in the UK subscribe to
multiple systems, potential respondents were asked: If you have access to
multiple platforms, which TV platform do you use most often? in order
to fill quotas based on the proportions of households associated with the
three main platforms in the UK at the time of the survey (Sky 47% (n =
486), Freeview 33% (n = 373), and Digital Cable 16% (n = 185); other
platforms 4% (n = 65): Continental Research 2005).
75
Table 2 Demographics of sample

Total

This study
UK digital TV audience
UK population
867 households
867 individuals
16.5 million households

39,430,804 individualsb
25 million householdsa
59,834,300 individualsb
Gender:

Male
49%
50%c
Female
51
50

(1) = 0.04

n.s.
48%c
52
(1) = 0.04
n.s.
Age:

1624 years
10
15
2534 years
15
18
3544 years
23
23
4554 years
17
16
5564 years
16
13
65 years or more
19
15

(5) = 3.99

n.s.
14
16
20
15
14
20
(5) = 2.26
n.s.
Children:

Kids 015
34
37
No kids 015
66
63

(1) = 0.30

n.s.
32
68
(1) = 0.26
n.s.
Notes: a Ofcom (Office of Communications, United Kingdom) (2005) Digital Television Update: Q3 2005.
b
National Statistics (Office for National Statistics, United Kingdom) (2005) Population Estimates (available
at: www.statistics.gov.uk).
c
Continental Research (2005) The Autumn 2005 Digital TV Report.
Measures
The full survey questionnaire consisted of 30 questions, many of which
had multiple sub-items (e.g. the motivations question). The questions fell
into three categories: (1) digital platform and demographics questions;
(2) digital television usage questions, including interactivity and ad
avoidance; and (3) motivations for watching digital television (uses and
gratifications). In some cases the wording of a previously used uses
and gratifications item was changed to be more understandable for UK
residents, in consultation with the marketing research firm that carried
out the survey.
Table 3 lists the results of a factor analysis of the 18 uses and gratifications
items, showing that these items loaded on three motivation factors, which
we identified as Entertainment, Information and Social, based on the
highest-loading items on each factor. In other words, our strategy of adding
76
0.22
0.19
0.46
0.46
0.32
0.31
0.28
Its good company when Im on my own (COMPANIONSHIP [Social approval])
I just like to use it (ENTERTAINMENT [Sensory gratification])
It is a more convenient way to see movies (MOVIE RENTAL [Problem removal])
I like the freedom to set my own schedule (TIME SHIFTING [Incomplete satisfaction])
I can find out whats on without looking at the TV listings in the paper or TV guide
(GETTING MORE FROM TV [Problem removal])
0.21
0.22
Notes: Factor that the item loaded on in its source study is indicated in CAPITALS. Consumer motive that the item was intended to measure is indicated in
italics. Items used to measure the three motive scales are indicated in bold. Extraction method: principal axis factoring. Rotation method: oblimin with Kaiser
normalisation.
0.20
0.18
It is relatively cheap (CONVENIENCE [Mixed approach-avoidance])
I dont have to watch as many adverts (AVOIDING COMMERCIALS [Problem avoidance])
0.19
I can learn something (INFORMATION [Intellectual stimulation])
0.58
0.61
0.18
0.21
Its good for information (INFORMATION [Problem avoidance])
0.41
I like to participate in programmes (INTERPERSONAL UTILITY [Social approval])
0.62
0.46
I like to have something available for the kids to watch (CHILDRENS VIEWING [Normal depletion])
0.31
0.18
0.15
0.15
Information
I get to see more news (INFORMATION [Intellectual stimulation])
0.54
I want to entertain people who come over (SOCIAL INTERACTION [Social approval])
0.26
0.23

0.22
0.56
0.70
It gives me choice over what I watch (FREEDOM OF CHOICE [Normal depletion])
0.55

0.13
0.75
Its enjoyable (ENTERTAINMENT [Sensory gratification])
Its good for when I have nothing better to do (PASS TIME/HABIT [Mixed approach-avoidance])
0.81
It is entertaining (ENTERTAINMENT [Sensory gratification])
I have enough choices to change channel when something I dont like comes on
(SELECTIVE AVOIDANCE [Incomplete satisfaction])
Social
Entertainment
Items (I watch digital iTV because )
Table 3 Factor analysis of uses and gratifications items
77
items to the core set used by Rubin (1983) was successful in detecting a
Social motive for watching digital TV, as well as the two primary motives
he had discovered: Entertainment and Information. Although our set
of items potentially measured eight separate consumption motives, five
were spread across these three main ones for the consumption of digital
television. Because we used an oblique rotation, following Rubin (1983),
these factors were highly correlated (see Table 3); but since this rotation
method minimises cross-loadings (all were less than 0.30), the factors
were more interpretable than when they were using a simple orthogonal
transformation (we got the same factors using a varimax rotation). In most
cases, the items loaded on factors consistent with the motive they were
intended to measure. All three Sensory Gratifications items loaded on
Entertainment. Both Intellectual Stimulation items loaded on Information,
as well as one of the Problem Avoidance items that had loaded on Rubins
Information factor. Most importantly, two of the three items intended to
measure the Social Approval motivation loaded on the Social motivation
factor. The other, which had loaded on Rubins Companionship factor,
loaded on Entertainment in our data, perhaps reflecting the sense that TV
viewing can be a solution to the problem of not having companionship.
The items with loadings greater than 0.40 were used to construct three
scales measuring these Entertainment, Information and Social motivations.
Ideally, the items used should have had loadings greater than 0.70, which
would have guaranteed very high levels of reliability for these scales
(Nunnally 1978). Because the items used to measure the Social motivation
factor had low loadings, the reliability of this scale was barely acceptable,
even for exploratory research (Cronbachs = 0.51). The factor analysis
demonstrates that this scale is a valid measure of Social motives (as we
said above, cross-loadings are low), but because its reliability is less than
0.70, more than half of its variance is random error variance, which
makes it very unlikely that we will find significant associations between
this variable and other variables. In other words, the low reliability of this
scale means any significant findings we report in relation to it are very
conservative estimates of its true relationships with other variables. The
other two scales, Entertainment ( = 0.85) and Information ( = 0.72),
both had acceptable levels of reliability.
To test the discriminant validity of these three scales, we used 1000
bootstrap samples (generated using AMOS) to estimate the distributions of
their correlations. The 95% confidence intervals for all three correlations
did not include 1.0, which suggests that these three scales measure
different things, even though they are highly correlated (Entertainment
78
Information: 0.55 to 0.64, M = 0.59; EntertainmentSocial: 0.37 to 0.48,

M = 0.43; InformationSocial: 0.31 to 0.43, M = 0.37).
Results
Sample descriptives
One-fifth (20%) of the sample had a very long history of experience with
digital TV, compared to audiences in other countries, dating back seven
years or more (M = 5.63 years, SD = 2.44). Over four-fifths of the sample
(82%) reported watching television seven days a week (M = 6.5 days), for
an average of two hours a day. Two-fifths (40%) said they interacted with
iTV programmes, but only one-sixth (17%) interacted with iTV ads (only
Sky viewers could interact with ads: 32% of Sky viewers had interacted
with a TV ad).
Motivations for watching digital television

Participants were read out the list of 18 reasons (shown in Table 1) people
might give for watching digital iTV. As each reason was read out (the order
of presentation was randomised across participants), the participants
indicated whether that reason matched their own on a five-point scale (1
= does not match at all to 5 = matches exactly).
Table 4 lists the means and standard deviations for the three motivation
scales (identified in Table 3), and the correlations between them and with
three measures of digital iTV usage, and attitude towards iTV: I love
interactive TV Im always pressing red! All three were significantly
correlated with overall usage (hours per week), interaction with iTV
programmes, interaction with iTV ads, and also attitude towards iTV. This
meant that Social motivations had a significant relationship with all three
measures of digital TV usage, in contrast to previous studies that have
found no relationship between social motivations and usage of analogue
TV (e.g. Rubin 1983). Controlling for demographics using regression
analysis (see Table 4), we also used AMOS to estimate the 95% confidence
intervals for these relationships, to determine the relative importance of
these three motivations for each dependent variable. Although the three
motivations were highly correlated, multicollinearity was not a serious
problem in these analyses, as the maximum condition index (15.48) for all
four regressions was less than 32, and the variable inflation factor (VIF)
79
Table 4 Correlations between motivation factors and usage of digital iTV

Variable
A. TV hours per week
SD
13.70 10.15
B. Interaction with
TV programmesa 1.64 0.96 0.11**
C. Interaction with TV adsa 1.47 0.84 0.02
0.28***
D. I love interactive TV
Im always pressing red! 1.51 0.88 0.11**
0.30*** 0.33***
1. Entertainment 3.50 0.93 0.19*** 0.14*** 0.14** 0.23*** .

2. Social 2.16 0.92 0.09** 0.19*** 0.23*** 0.42*** 0.43***
3. Information 3.16 1.03 0.09** 0.19*** 0.16*** 0.32*** 0.59*** 0.37***

a
Measured on a five-point scale (1 = never to 5 = very often). *** p < 0.001, ** p < 0.01, * p < 0.05.
for each individual predictor was always less than 10 i.e. tolerance was
always greater than 0.10 (DeMaris 2004).
The Entertainment motive was the most important predictor of simply
watching digital iTV, as the 95% confidence interval for its regression
coefficient (0.14 to 0.31) was above the absolute values of the means for
the Social motive (M = 0.13, confidence interval: 0.05 to 0.20) and the
Information motive, which had a negative relationship with digital TV
watching (M = 0.12, confidence interval: 0.20 to 0.04). The Social and
Information motives were equally important, using the same yardstick. All
three had a significant relationship with watching digital iTV, as none of
their confidence intervals included zero.
The Social motive was also equal to the Information motive for
predicting interaction with digital iTV programming. The confidence
interval for each of these two motives included the mean for the other
(Social: 0.04 to 0.20, M = 0.12; Information: 0.08 to 0.25, M = 0.17). The
confidence interval for Entertainment (0.11 to 0.06) included zero, so it
cant be ruled out that this motive has no relationship with this dependent
variable.
Only the 459 (53% of 867) Sky subscribers in the sample were asked
about their interaction with iTV ads. Again, for this dependent variable,
the Social motive (confidence interval: 0.06 to 0.28, M = 0.17) was equally
as important as the Information motive (confidence interval: 0.01 to 0.24,
M = 0.12), and Entertainment was not a significant predictor (confidence
interval: 0.12 to 0.11, M = 0.01). When attitude towards interaction
with digital iTV was the dependent variable, however, all the sample
were included in the analysis, and the Social motive was the primary
80
predictor, as its confidence interval (0.31 to 0.45, M = 0.39) was above

the mean for the Information motive (M = 0.20, confidence interval: 0.12
to 0.28). Once again, it cant be ruled out that the Entertainment motive
has no relationship with this dependent variable, as its confidence interval
included zero (0.13 to 0.03, M = 0.05).
Although not a focus of this study, the results for the demographics
variables in Table 5 are consistent with previous studies of watching and
interacting with iTV, such as the Zip (2005) study. Older people tend to
watch more TV, but are less likely to interact with it. Males watch less
than females, but are more likely to interact. Although the Zip study found
an association between having children in the house and a positive attitude
towards interaction with iTV, in our data we found that the presence
of children under 15 tends to reduce enthusiasm for interaction, after
controlling for age and gender.
Table 5 Regression results: predicting TV watching and interaction from motivation factors
Dependent variable

Predictor

Interaction Interaction I love interactive
TV hours
with TV
with
TV Im always
per week programmes
TV ads
pressing red!
1. Age group (1619, 2024, 2534,

3544, 4554, 5564, 65 and over)
2. Gender (1 = male, 1 = female)
3. Children 015
0.28***
0.07*
0.07
0.15***
0.07*
0.04
0.19***
0.02
0.08
0.06
0.09**
0.09**
4. Entertainment
5. Social
6. Information
0.22***
0.13**
0.12**
0.02
0.12**
0.17***
0.01
0.17**
0.12*
0.05
0.39***
0.20***
11.6%
6.9%
R (adjusted)
7.9%
21.6%
Notes: Standardised betas; ***p < 0.001, **p < 0.01, *p < 0.05.
Discussion
Motivations for interacting with digital television
The surprising finding of this study was the disconnection between
industry expectations about viewer motivations for interaction with
digital TV and their actual motivations. These industry expectations seem
to be based on previous research in the uses and gratifications tradition,
which identified two needs, entertainment and information-seeking, as the
main motivations for watching analogue TV. Currently, the majority of
81
digital iTV ads (57% of Sky Interactive campaigns) have the objective of
satisfying just one of these needs: more information (the next highest is
lead generation (call back), 24%).
In contrast to these expectations, we found that interaction with digital
iTV programmes and ads was prompted as much by social motives as it
was by information-seeking motives. Furthermore, entertainment motives
were not a significant predictor of either type of interaction (with iTV
programmes or iTV ads). Social motives were nearly twice as important
as information-seeking for predicting a favourable attitude towards
interacting with digital iTV, and again entertainment had no relationship
with this attitude at all. Entertainment was over twice as important as the
other two motives for watching digital TV, which is more like watching
traditional TV, but in contrast to studies of traditional TV usage (e.g. Rubin
1983), social motives were also significantly important for TV viewing
as well as for TV innteraction, and equally as important as informationseeking motives.
Our results show that digital iTV is very similar to other interactive
technologies that are used for social motivations, such as the internet
(e.g. Papacharissi & Rubin 2000; Stafford et al. 2004; Pew Internet &
American Life Project 2006). Participation in iTV programmes may
generate similar feelings of belonging and appreciation that participation
in a real community does, or participation in virtual communities can,
through networking sites (e.g. Facebook), role-playing games, chat, blogs
and email. Interaction with ads may also give socially motivated viewers
more to talk about with their networks. As the convergence between
television, the internet and portable devices such as mobile phones and
iPods accelerates, social motivations for interaction with television could
become even more important. One of the main areas of study for iTV
researchers in future should be trying to understand more completely all
the aspects of the social motivation for interacting with TV.
Implications for practitioners

The main implication for practitioners from this study is that they need to
consider how to appeal to social needs, as well as needs for information
and entertainment, when constructing programmes and commercials for
digital iTV. The items we used for our social motivation scale suggest some
ideas about how to do this.
First, people with a social motive want to have something to share with
people who come over. These people like to watch programmes and ads
82
in co-viewing situations (according to our data, solitary watching satisfies

an entertainment motive). Co-viewing can take two forms: first, the silent
passive mode of the traditional cinema, and now in-home theatres, which
content producers have always done a good job of catering towards,
but, second, co-viewing can also be loud and socially interactive. In such
situations, the content on the screen needs to supply a constant stream
of things to talk about. The structure and content of these programmes
needs to be simple enough so that people can follow the action despite the
distractions in the room. Interactive options might be very attractive for
socially motivated viewers, encouraging discussion about which option to
take.
As well as participation with others in the room, socially motivated
viewers want to participate in programmes. Recent years have seen a
growth in the phenomenon of viewer-controlled content, in shows like
Big Brother and Pop Idol. Quiz shows that encourage viewers to play
along at home would also be popular with these viewers, but in the UK
they need to shake off the stigma of recent scandals associated with the
faking of results. The growth of YouTube and other outlets for usergenerated content provides another avenue for socially motivated viewers
to participate in the generation of screen programmes and advertising.
In our Introduction, we said that one of the main purposes of this study
was to uncover the motivations that would prompt viewers to respond to
the specific appeals that the Zip (2005) study had identified for different
segments of non-interactors in the UK digital iTV audience. For example,
the segment it labelled Apprehensive Stargazers could be enticed to
interact if a trusted celebrity told them how to do it, and reassured them
that they would not be bored or miss any part of the programme. Since
Apprehensive Stargazers seem to be socially motivated (they talk about
ads with their friends), this appeal would work best using an execution
specifically targeted for such positive-ending motivations like social
approval: a transformational execution (Rossiter & Bellman 2005). The
right transformational execution would suggest that interaction will
generate a feeling of self-pride not vanity but a confidence that others
will approve their new status as an interactor. Care needs to be taken in the
selection of the celebrity as their trustworthiness will depend critically on
how much Apprehensive Stargazers identify with that celebrity. Positiveending transformational executions should also be useful for other
segments with primarily information-seeking motives. For example, Bitter
Ad Haters are unresponsive to offers of freebies or advice from trusted
celebrities, but might be responsive to appeals which claim that interaction
83
with programmes or ads will generate positive feelings of intellectual

stimulation and mastery.
Limitations and suggestions for future research

More research is needed, however, to understand what kinds of iTV
programme and ad strategies work best at satisfying social motivations.
The literature on social influence suggests many intriguing possibilities
(see, e.g., Cialdini 2001). It would be interesting to test whether classic
social influence tactics, such as the foot-in-the-door technique (Freedman
& Fraser 1966), peer pressure, praise or similarity (all of which can be
augmented using computers: Fogg 2003) can be effective iTV strategies.
A major limitation of this study is that we surveyed only the audience
for digital TV in the UK, instead of all TV viewers, so we are unable to
substantiate our claim that social motives are more important for watching
digital TV than they are for watching analogue TV. We suspect that these
motives have always been important, but in previous studies, such as
Rubins (1983), they were imperfectly measured, and so didnt form a
factor strong enough to use in regressions predicting usage of analogue
TV. In our sample of digital TV viewers, social motives were equally as
important as information motives for watching digital TV, which is more
like watching traditional TV than interaction with digital TV.
Our measure of the social motivation for watching and interacting with
digital iTV had low reliability, but the highly significant results we obtained
despite this suggest that a useful direction for future research would be the
development of a more reliable scale to measure this motivation. Also,
while we argued that TV watching is a form of consumer behaviour, and
therefore the eight basic motivations we measured were sufficient, future
research may find that additional motivations drive usage of digital iTV
(e.g. the ego-defensive motivation proposed by Katz 1960).
We deliberately chose to survey the UK digital TV audience because it
has the longest experience with digital iTV advertising. Nevertheless, we
acknowledge the need for other studies to be conducted in the US and
other countries, because our results may have been affected by peculiarities
of the UK market. For example, nearly 60% of our sample interacted with
digital text services, which are not a widely used feature of US television.
The TV landscape is undergoing rapid change, and future studies are
needed, even in the same UK marketplace, to test whether social motives
for watching and interacting with digital iTV remain as important as
information and entertainment motives, or are even more important.
84
Conclusion
This study surveyed digital households in the most mature digital iTV
marketplace in the world, the UK, using an efficient but comprehensive
survey of basic consumption motivations, adapted to watching and
interacting with digital TV. We found that social motivations were
one of the main reasons for interacting with television programmes
and advertising. Further research is needed to fully understand the
ramifications of this finding for the design of iTV ads and programmes.
But strategies that address social motives might be very successful, since
a large proportion of the population have still not yet responded to the
current strategies of addressing needs for information or entertainment.
Acknowledgements
The authors thank John R. Rossiter for his insightful comments on drafts
of this paper, and especially Zip Television and Continental Research
for their collaboration in collecting the data. This research was funded
by the sponsors of the Beyond :30 project (www.beyond30.org) and the
Australian CRC for Interactive Design.
References
Anderson, J.A. & Meyer T.P. (1975) Functionalism and mass media. Journal of Broadcasting,
19, pp. 1122.
BARB (Broadcasters Audience Research Board Ltd, United Kingdom) (2008) Multi-channel
development 20022008. Web page, available at: www.barb.co.uk/TVFACTS.cfm.
Batra, R. & Ahtola, O.T. (1990) Measuring the hedonic and utilitarian sources of consumer
attitudes. Marketing Letters, 2, 2, pp. 159170.
Cialdini, R.B. (2001) Influence: Science and Practice (4th edn). Boston, MA: Allyn & Bacon.
Continental Research (2005) The Autumn 2005 Digital TV Report. London: Continental
Research.
DeMaris, A. (2004) Regression with Social Data: Modeling Continuous and Limited Response
Variables. Hoboken, NJ: John Wiley & Sons.
Donovan, R.J. & Henley, N. (2003) Social Marketing: Principles and Practice. Melbourne,
Australia: IP Communications.
Eighmey, J. (1997) Profiling user responses to commercial web sites. Journal of Advertising
Research, 37, 3, pp. 5966.
Eighmey, J. & McCord, L. (1998) Adding value in the information age: uses and gratifications
of sites on the World Wide Web. Journal of Business Research, 41, 3, pp. 187194.
Fennell, G. (1978) Consumers perceptions of the product-use situation: a conceptual
framework for identifying consumer wants and formulating positioning options. Journal of
Marketing, 42, 2, pp. 3847.
Flaherty, L.M., Pearce, K.J. & Rubin, R.B. (1998) Internet and face-to-face communication:
not functional alternatives. Communication Quarterly, 46, pp. 250268.
85
Fogg, B.J. (2003) Persuasive Technology: Using Computers to Change What We Think and
Do. San Francisco, CA: Morgan Kaufmann.
Freedman, J.L. & Fraser, S.C. (1966) Compliance without pressure: the foot-in-the-door
technique. Journal of Personality and Social Psychology, 4, pp. 195202.
Greenberg, B.S. (1974) Gratifications of television viewing and their correlates for British
children. In: J.G. Blumler & E. Katz (eds) The Uses of Mass Communications: Current
Perspectives on Gratifications Research. Beverly Hills, CA: Sage Publications, pp. 7192.
Herzog, H. (1941) On borrowed experience: an analysis of listening to daytime sketches.
Studies in Philosophy and Social Science, 9, pp. 6595.
Horton, D. & Wohl, R.R. (1956) Mass communication and para-social interaction:
observations on intimacy at a distance. Psychiatry, 19, pp. 215229.
Katz, D. (1960) The functional approach to the study of attitudes. Public Opinion Quarterly,
24, 2, pp. 163204.
Kaye, B.K. & Johnson, T.J. (2002) Online and in the know: uses and gratifications of the web
for political information. Journal of Broadcasting & Electronic Media, 46, 1, pp. 5471.
Korgaonkar, P.K. & Wolin, L.D. (1999) A multivariate analysis of web usage. Journal of
Livaditi, J., Vassilopoulou, K., Lougos, C. & Chorianopoulos, K. (2003) Needs and
gratifications for interactive TV applications: implications for designers. Proceedings of the
36th Hawaii International Conference on System Sciences (HICSS03), pp. 100b, available
at: doi.ieeecomputersociety.org/10.1109/HICSS.2003.1174237.
Locander, W.B. & Spivey, W.A. (1978) A functional approach to attitude measurement.
Journal of Marketing Research, 15, 4, pp. 576587.
Luo, X. (2002) Uses and gratifications theory and e-consumer behaviors: a structural equation
modeling study. Journal of Interactive Advertising, 2, 2, pp. 111.
McQuail, D., Blumler, J.G. & Brown, J.R. (1972) The television audience: a revised
perspective. In: D. McQuail (ed.) Sociology of Mass Communications: Selected Readings.
Harmondsworth, Middlesex, UK: Penguin, pp. 135165.
Nunnally, J.C. (1978) Psychometric Theory (2nd edn). New York: McGraw-Hill.
Ofcom (Office of Communications, United Kingdom) (2005) Digital Television Update: Q3
2005. Online report, 9 December, available at: www.ofcom.org.uk/research/tv/reports/dtv/
dtu_2005_q3/.
OGuinn, T.C. & Faber, R.J. (1991) Mass communication and consumer behavior. In: T.S.
Robertson & H.H. Kassarjian (eds) Handbook of Consumer Behavior. Englewood Cliffs,
NJ: Prentice-Hall, pp. 349400.
Papacharissi, Z. & Rubin, A.M. (2000) Predictors of internet use. Journal of Broadcasting &
Electronic Media, 44, 2, pp. 175196.
Pew Internet & American Life Project (2006) The Strength of Internet Ties. Online Report,
January 25, available at: www.pewinternet.org/pdfs/PIP_Internet_ties.pdf.
Rogers, E.M. (1995) Diffusion of Innovations (4th edn). New York, NY: Free Press.
Rossiter, J.R. & Bellman, S. (2005) Marketing Communications: Theory and Applications.
Sydney: Pearson Prentice-Hall.
Rubin, A.M. (1983) Television uses and gratifications: the interactions of viewing patterns and
motivations. Journal of Broadcasting, 27, 1, pp. 3751.
Rubin, A.M. & Bantz, C.R. (1987) Utility of videocassette recorders. American Behavioral
Scientist, 30, 5, pp. 471485.
Rubin, A.M. & Rubin, R.B. (1989) Social and psychological antecedents of VCR use. In: M.R.
Levy & B. Gunter (eds) Home Video and the Changing Nature of the Television Audience.
London: J. Libbey/IBA, pp. 92111.
Stafford, T.F., Stafford, M.R. & Schkade, L.L. (2004) Determining uses and gratifications for
the internet. Decision Sciences, 35, 2, pp. 259288.
86
Valentine, V. & Evans, M. (1993) The dark side of the onion: rethinking the meanings
of rational and emotional responses. Journal of the Market Research Society, 35, 2,
pp.125144.
Walker, J.R. & Bellamy, R.V. Jr (1991) Gratifications of grazing: an exploratory study of
remote control use. Journalism Quarterly, 68, 3, pp. 422431.
Walker, J.R., Bellamy, R.V. Jr & Traudt, P.J. (1993) Gratifications derived from remote control
devices: a survey of adult RCD use. In: J.R. Walker & R.V. Bellamy Jr (eds) The Remote
Control in the New Age of Television. Westport, CT: Praeger, pp. 103112.
Zip (Zip Television) (2005) Passive 2 Active: Extending the Reach of Interactivity on TV.
Report. London: Zip Television.
About the authors

Steven Bellman is Associate Professor and Deputy Director of the
Interactive Television Research Institute at Murdoch University, Perth,
Australia.
Anika Schweda is a former post-doctoral research fellow at the Interactive
Television Research Institute at Murdoch University, Perth, Australia.
Duane Varan is Professor of New Media and Director of the Interactive
Television Research Institute at Murdoch University, Perth, Australia, and
also the Executive Director and Chief Research Officer for The Disney
Media & Advertising Lab in Austin, Texas.
Address correspondence to: Steven Bellman, Interactive Television
Research Institute, Murdoch University, South Street, Murdoch WA, 6150,
Australia.
E-mail: S.Bellman@Murdoch.edu.au
87
Whose design is it anyway?

Priming designer and shifting preferences
Gorm Gabrielsen, Tore Kristensen and Judith Lynne Zaichkowsky
Copenhagen Business School
A series of studies is presented which investigates preference among similar but

different designs within a product category. The variables of price, brand name
and priming designer are shown to shift preferences. Without brand names,
consumers prefer a well-designed object. When supplied with information about
brand names and designer, they may shift their preferences to designs they believe
are designer brands, even when the actual design is not.
Introduction
Research on design in marketing is heavily represented by investigation of
successful designs or innovative designs for new product development (e.g.
Luo et al. 2008; Srinivasan et al. 1997). Besides new product research, there
is a growing interest in investigating the added value of design elements to
everyday product categories, to understand how non-functional attributes
increase preference (Chitturi et al. 2008; Hagtvedt & Patrick 2008; Orth
& Malkewitz 2008). Design is said to be a major differentiating attribute
in the preference and choice of consumer goods (Zolli 2004).
Consumers seek to buy uniquely designed products that represent
significance and meaning in their lives, perhaps to indicate and reflect
their own uniqueness of identity or to signal an avant-garde role in
society. Distinctive designs have also made everyday goods indicators of
social and cultural status. It has been said that, above all, good design
provides functionality, and hence makes life easier in practical day-to-day
living (Petroski 1994; Heskett 2002). Therefore major manufacturers of
everyday items, from cars to tea bags, are focusing on design to generate
appeal and obtain a competitive edge in the marketplace (Business Week
On-Line 2006; Fabricant 2006).
Received (in revised form): 15 December 2008

DOI: 10.2501/S1470785310201077
89
Many brands employ famous designers to cue the specialness of the

object and reinforce consumers choice through the name of the designer,
rather than the actual design. For example, designer labelled brands can
be 40-dollar Martha Stewart bed sheets in Target, trendy Karl Lagerfeld
clothes at H&M, or 40,000-dollar patchwork handbags from Marc
Jacobs at Louis Vuitton. Consumers are led to rely on the name Martha
Stewart-Target or Marc Jacobs-Louis Vuitton as powerful and proven
indicators of trustworthy designer brand goods.
Sometimes the design forms the actual distinctive identity of goods, and
the brand name or designer name is not obviously or visibly attached to
the product, leaving designer items completely unlabelled as such. In these
cases the consumer may not know the designer when purchasing and
cannot obviously show off the designer brand name to the uninitiated. For
example, the jewellery trade relies on unique materials and design, as does
the furniture category, where the designer is rarely known by those outside
the industry (Louie 2006). In these cases, the object might not be readily
identified as a designer brand by consumers unfamiliar with the product.
Sometimes distinctively designed products are knocked off by low-price
look-a-likes and it is very difficult to tell the originals from the knock-offs,
especially when the label or brand name is missing from the object. For
example, a real Arne Jacobsen egg chair retails for over US$12,000, but
other furniture manufacturers may sell a look-a-like for considerably
less. Copying the distinctive design of consumer goods is a worldwide
problem for original manufacturers, representing estimated losses of over
US$30 billion a year to original companies (McCubbins 2004; IIPA 2007).
Consumers themselves may not know when they are purchasing a knockoff, especially when there are no external cues such as brand name to rely
on. They also may not understand why an object is so expensive if they
do not know it is an original design, and may therefore actually seek out
less expensive look-a-likes. Consumers may rely on price or distribution
channels to assess the uniqueness of their purchase, but they still may not
be knowledgeable as to or aware of the designer aspect that correlates so
highly with price.
An important question for marketing and retailers is: How do you get
consumers to value and identify original designs over copy-cat brands or
knock-offs? To answer that question we first need to know (a) which
consumers can recognize a designer good?, (b) Can such a designer
good decrease price sensitivity?, (c) Can a designer good override a wellknown brand name? and, finally, (d) Is there a design-prone segment? If
preferences can be shifted by providing consumers with information about
90
the designer, purchasers may become less price sensitive and perhaps more
likely to purchase original designer items. Therefore the purpose of this
study is to assess how consumer evaluations of an object change when
they are given information about brand names and information about the
designer.
Selected literature review

Design
Bob Veryzer has researched design and aesthetics, documenting what
makes a consumer recognize and prefer a good design (Veryzer 1995, 1999;
Veryzer & Hutchinson 1998). He finds consumers generally have a gestaltlike aesthetic response to the design principles of unity and proportions.
Furthermore, consumers are not consciously aware of the design principles
or their reaction to them, yet their preferences and choices of products are
definitely influenced by the design of the objects.
Research in Psychological Science explains consumer preferences by
relating the design principles to emotions. Humans are found to prefer
emotionally neutral visual objects that are curved, over those that have
primarily pointed features and sharp angles (Bar & Neta 2006). The
rationale for this finding is that sharp transitions in a contour portray a
sense of threat, on either a conscious or unconscious level, and thus trigger
a negative bias. Other research finds that people tend to prefer highly
prototypical stimuli or objects over unusual exemplars. This is called the
beauty-in-averageness effect (Winkielman et al. 2006). The reason given
for this finding is that prototypes are processed fluently, with greater speed
and efficiency than atypical objects. Because of this fluency in processing,
more positive emotions or reactions are elicited. Therefore it can be said
that demand should be higher for an object with a pleasing design.
Form can override function

Consumers may perceive that aesthetics are associated with an artistic
edge on branded products. Sometimes the form may be so strong that it
overrides the functional benefits of the product. For example the Juicy Salif,
created by the brand Alessi, is wildly successful (see Figure 1). The juice
squeezer costs 20 times the price of a normal juicer, yet it is significantly
deficient for practical purposes. This distinctive identity through design
has enormous commercial importance for the manufacturer, for it adds
91
value especially where margins are low. As

Heskett (2002) observes, the term squeezer
should perhaps be more appropriately applied
to profit leveraging rather than functionality
for users. So the objective for many product
categories, particularly those where consumers
have low involvement and are price sensitive, is
to make the consumer more design sensitive and
perhaps lower price sensitivity.
Why a designer name may override aesthetics:

cultural and social influences as taste
It has been said that goods are indicators of
Figure 1 The Juicy Salif
social and cultural status (McCracken 1988).
squeezer by Alessi
As disposable income increases, the potential for
conspicuous consumption increases, and so then demand for distinctive
products increases. Because of this, designer brands have proved to be
powerful devices particularly in the more expensive sectors of consumer
products.
Purchase of designer objects may be a risky issue, especially for publicly
consumed goods (Loewenstein et al. 2001). Ones taste and perception
of quality may appear to be wrong when styles change quickly or fads
come and go. For example, hundreds of dollars are spent each summer
by consumers on sunglasses. Last years sunglasses work well, but the
style of the frame may be fashionably obsolete. Small-framed sunglasses
are turned in for large sunglasses for those who want to appear socially
or culturally savvy. Furniture and other similar products may hold this
same social stigmatisation (Douglas 1996). To possess an object in the
belief that it is of fine quality or in good taste, only to stand corrected
by an expert (or someone competing in the social game of choice) is a
bad experience (Bourdieu 1984). Ridicule and shame may be the result.
Therefore consumers may be extra cautious when buying items associated
with good taste or designer status.
In this context an important issue is whether a framing mechanism
is operational (Tversky & Kahneman 1981). Framing occurs when a
stimulus, or cue, makes people perceive the object or design differently
than they would without it. For instance, when the product is presented
in the context of designer, it might be evaluated differently, than if it
was presented in another context, or simply stands alone. In addition to
92
the contextual framing, when an evaluation is conducted as a step in a

sequence of many different steps in a decision process, the evaluation step
is affected both by what has preceded it and what is expected to come
afterwards (Langley et al. 1995). The designer prefix is such a contextual
framing in that, when the term is used, the expectations of what the
product is or what it means may change.
Additional information, such as objective knowledge about the choice,
may frame the selection and preferences. The knowledge provided puts
the identification, preferences and buying decision in a particular context,
which will enable a competent person to make the right choice. For
example, one may think that all high-quality automobiles have leather
seats, and conclude that a particular auto is of high quality and desirable
because of that cue. Therefore perhaps when choices are primed by the
cue designer (which is used as a framing cue for evaluation), people will
assign this knowledge to their evaluation and choice.
Sometimes the design is not that aesthetically appealing, yet when
coupled with a designer label or famous brand, the product sells
famously. Handbags are a good example of this especially the Fugly, or
Frankenbag, a limited edition denim patchwork bag from Louis Vuitton,
which sold for over US$40,000 (National Post 2007). Aesthetics are
subjective, and in many circumstances purchasers of such items are known
as fashion victims, because the design is really not that aesthetically
appealing it is purchased because the individual relies on the designer
label as a proxy for good taste.
However, it is not known if they are paying more for the designuniqueness or for the desirable brand name. The brand name and design
are confounded. The two are closely related, but the conscious brand may
be well known and positive, while the other (the actual design) may be
truly negative to the unconscious perception. Here, conceptual fluency
may be low (Reber et al. 2004). So the consumer may be taking their cue
for aesthetics from the brand name, rather than from the inherent design.
Therefore it may be enlightening to separate designer brand name from
the actual design.
Brand name
The amount of information available to cite from on the importance of
brand name to evaluation and choice seems infinite and overwhelming (for
a review see Hoeffler & Keller 2003). Dawar and Parker (1994) found
brand name and price were universal indicators for product evaluation and
93
choice in the product category of electronic equipment. Brand name may

also override information concerning other important product attributes,
and may be the only cue used by consumers in choice. Van Osselaer and
Alba (2000) found brand cues may be so powerful that they may block the
learning of quality-determining attribute cues. Therefore once brand name
is included, the evaluations may change, so the visual design elements or
the gestalt of the product may be diminished by a strong brand name.
Design acumen
People high in design acumen make quicker sensory connections and
exhibit more sophisticated preferences regarding the design of things than
those with little design acumen (Csikszentmihalyi & Robinson 1990).
This design acumen may be acquired in different ways. For example, one
may inherently perceive good design, or one may have extensive training
and experience in design. Its a matter of a persons ability to judge the
nature and quality of the work through salient ratios, symmetry, balance,
perceptual grouping, the use of strong metaphorical cues, and contrast.
Consumer researchers who have looked at product design and consumer
responses are Bloch (1995) and Bloch, Brunel and Arnold (2003). They
find strong individual differences in consumers concern for product
appearance or design through a scale to measure the Centrality of Visual
Product Aesthetics (CVPA). Therefore, there may be individual differences
in the attraction and identification of design closely tied to ones CVPA.
Consumers who score highly on inherent design acumen would rely on the
actual design of a product to form their preferences, while those scoring
low on design aptitude might rely on external cues such as brand name or
designer labels.
Prior experience
Preferences within product categories develop over time with purchasing
experience, changes in income, knowledge and maturation. So people
with more experience in purchasing from a product category may in
fact have cultivated certain tastes that apply to their choices (Johnson &
Lehmann 1997). A person may learn what to look for in a product design
and to assess the important determinants of attractiveness for that product
category. For example in the product category of bed linen, over time,
people may learn that the thread count, ply and type of cotton all combine
to form a different quality of tactile sensations. Any one of these attributes
94
alone may not signal a certain level of quality, but an experienced consumer
might predict the combination of attribute levels that will give the best
sensation. Following on from this, people who have purchased repeatedly
from the product category might have different preferences than those who
have little or no prior purchase experience from the product category.
Summary
From the literature, several factors are reviewed that may influence
peoples preference for choices within a product category. These may be:
(1) an inherent ability to recognise unified design; (2) cues such as the
brand name and price, which may interfere with the evaluation of the
design attribute and override it; (3) the framing of the designer label or
related information; (4) a persons individual design acumen; and, finally,
(5) prior experience with the product category. Each of these factors will
be controlled for or measured in assessing consumers evaluations and
preferences among different choices with similar designs within a product
category.
Methodology
Three experimental conjoint designs were used to test how people evaluate
preferences within a product class when selecting with: (1) only visual
design information; (2) the origin or brand information added; and (3)
the information that one of the choices is a designer or a special choice.
Evaluations first took place without the price information attached to the
choices. The evaluations were then repeated within each study, adding the
actual prices of each choice as a conjoint variable. In this way the weight
given to price within evaluations could be estimated separately. Price was
not of interest per se, but since price is the barrier or facilitator to decision
to purchase for many situations, it was important to include it in the
analyses to further frame the results and see how the evaluation changes
with the price (Lichtenstein et al. 1988).
Procedure
Stimuli
The product category used was rice paper lamps, which are a small type
of lamp dominated by inexpensive products that many people buy to use
in back rooms, storage rooms or childrens rooms. In general this is a
95
mundane product category, but it is possible for some rice paper lamps to
be of high-price design. Three different brands of rice paper lamps were
selected. To represent the designer choice, a lamp designed by the famous
Japanese-American sculptor, Isamu Noguchi, was chosen. This designer
is well represented in major art galleries all over the world and there is a
foundation in his name in New York. Therefore he has some credibility as
a famous designer that may not be common knowledge among consumers
(Micucci 2003).
To represent a knock-off (or copy-cat) lamp, a no-name brand bought
at a local Chinese import assortment shop, which copied a similar shape
and style of the designer Noguchi lamp, was chosen. While the Chinese
lamp had a similar shape, the finishing and materials on this lamp were
inferior to those on the designer lamp. The third lamp was represented
by IKEA, which is a widely distributed, well-known brand for affordable
everyday home furnishings. This lamp was made out of similar materials,
but had a different shape or design than the designer and knock-off lamp.
No attempt was made to determine a priori which lamp followed the best
design principles, but the Noguchi lamp was the original well-crafted lamp
in this product category.
The actual price points of the lamps were 10 (Chinese), 185 (Noguchi)
and 20 (IKEA). The lamps were labelled Q, R and S in the blind
treatment, to give each lamp a meaningless and neutral label to choose
from. The lamps are shown in Figure 2.
Data collection
The experiments were conducted in various shopping malls in a large city.
Computers were set up with a self-directed data collection program and
the actual goods were displayed alongside the computer station so the
respondents could look at, touch and feel the tactility of the goods. Passersby were asked if they would take a few minutes to evaluate the lamps.
Their participation lasted less than 10 minutes. The order of the computer
presentation of the lamps for evaluation was randomised across subjects in
each experiment.
The sample
The respondents were a convenience sample, self-selected randomly, as
they were attracted to the display stand in the malls and lobbies. A total
of 60 men and 75 women participated (135 in total). Their self-reported
income and age categories are as follows: 8% young; 91% middle-aged;
96
Lamp Q: Chinese 10
Lamp R: Isamu Noguchi 185
Lamp S: IKEA 20
Figure 2 The three rice paper lamps
97
1% older; 78% lower income; 18% middle income; 5% upper income.1

Their prior experience in purchasing lamps was: 22% purchased no
lamps; 45% purchased one or two; 33% had purchased three or more
lamps previously. The number of designers recalled was: 32% could name
none; 16% named one; 24% named two; 27% could name three or more
designers.
Questionnaire content
Data were collected from each individual on their aesthetic aptitude (CVPA
scale: Bloch et al. 2003). The questions used to measure the individuals
sensitivity to design were: (1) I enjoy seeing displays of products that have
superior designs; (2) A products design is a source of pleasure for me;
(3) I see things in a products design that other people tend to pass over;
(4) I have the ability to imagine how a product will fit in with designs of
other things I already own; and (5) When I see a product that has a really
great design, I feel a strong urge to buy it. Respondents prior experience
with lamps was measured by the number of lamps they had previously
purchased. Their design knowledge was measured by asking them to recall
the names of any designers of household objects and furnishings they may
be aware of.
Conjoint design
Conjoint design is used extensively in evaluating design issues because it
allows evaluation of preference weights for the price cue separate to the
gestalt or the design issues (Bloch 1995). Data collection was done by
showing pairs of pictures of the lamps on the computer screen and asking
respondents to slide the computer cursor towards the one they preferred.
There was a training screen session of how to use the cursor scale by
sliding the cursor one way or another depending on the strength of ones
preference for a choice. If one lamp was only preferred a little then the
cursor was moved only slightly towards that lamp. If one lamp was greatly
preferred, then the cursor was moved completely to the right or left. They
could easily see the actual lamps while they rated them pairwise on the
computer.
It is likely that these data suffer from social bias in under-reporting ones income to strangers, and regressing
ones age to the average. What is important are the reported experience and knowledge levels, which are well
distributed.
98
The model
The respondents rated their preference for the lamps on a 22cm line on
the computer screen, labelled at the left anchor I prefer the left lamp very
much and at the right anchor I prefer the right lamp very much. Two
new lamps appeared for each judgement. The distance from the middle of
the scale to the preference was registered electronically.
The corresponding values of the scores of subject p, with lamp i to the left
and lamp j to the right, was denoted yijp. It is assumed that the numerical
score increases with the strength of the preference for j over i, and the
equal but opposite preferences (j over i, and i over j) corresponds to equal
but opposite scores. Thus for subject p, yijp is the assessed preferences of
lamp j over lamp i, and yjip is the assessed preference of lamp i over lamp
j. If, for example, yijp is positive, j is preferred over i. If yijp is negative, i is
preferred over j.
The underlying assumptions of the mathematical model are that the yijps
are independent random variables and normally distributed with mean
value E(yijp) = j i, i j and i,j = 1,2,3. As the only difference of
s are of interest, one can add the restriction that the s should sum to
zero. When price is added to the evaluation of the lamps, another variable
is added. The score of subject p with lamp i, with price k to the left and
lamp j with price l to the right was denoted: y(ik)(jl)p. Again y(ik)(jl)p
are assumed independent random variables and normally distributed with
mean value E(y(ik)(jl)p) = (j i) + (l k), i j, i,j = 1, 2, 3 (lamps).
l k; l,k = 1, 2, 3 (prices), where 1 + 2 + 3 = 0 and 1 + 2 + 3
= 0. This is the model for the conjoint layout.
The equation for no preference among the lamps is: 1 = 2 = 3 (= 0)
The equation for no preference among the prices is: 1 = 2 = 3 (= 0)
As the equations are general linear models, they are tested with F tests.
Results
Study 1: Evaluation of lamps without external cues
In the first study, 47 individuals (22 women and 25 men) rated the lamps
labelled with the letters Q, R and S. Respondents were not told the names
or origins of the lamps. After they rated the three pairs of lamps naked,
the respondents were asked to rate them again, this time with the actual
prices of each lamp rotated among the choices. A full factorial was used
with the price variable yielding nine paired comparisons.
99
With no price or contextual information, lamp R (a^ R = 83.33) was

preferred strongly to lamp Q (a^ Q = 6.33), which again was preferred
strongly to lamp S (a^ S = 89.67); F(2,139) = 19.24, p < 0.001. Therefore
without any external cues, respondents did strongly prefer the well-crafted
designer lamp. When the lamps were evaluated with the price information
the scale shrunk, and the Q and R lamps became much closer in preference
^
(a^ Q = 33.67 vs a
R = 43.67, N.S.). Although lamp Q was a lesser-preferred
lamp, the price information decreased the gap to R so that the preferred
lamp (R) was not significantly more attractive at the higher price. Lamp
S (a^ S = 77.33) remained highly undesirable compared to the other two,
even when evaluated at the lower price points (p < 0.001). So a low price
did not make up for the perceived less attractive design in lamp S. In other
words, people preferred both the Q and R lamps at 10 and at 20 over the
S lamp. Even when the S lamp was priced at 10 or 20, it was not always
preferred over the Q and R lamps priced at 185.
When looking at the price trade-off variable alone, it is not surprising
^
that the smallest value of 10 was preferred (b 15 = 94), but 20 was also
^
rated positively (b 30 = 71) and these two price points were not rated
significantly different from one another (t(419) = 1.48, N.S.). Therefore
consumers were quite willing to pay double the lowest price for the most
preferred design, but sought to avoid the design they did not like, S, no
matter how inexpensive the lamp. Generally people sought to avoid the
^
high price of 185 (b 225 = 165), p < 0.001.
After rating the lamps, the respondents were asked whether they would
consider buying any of the lamps. Thirty-nine per cent said they would
not consider buying any of them.2 Thirteen percent said they would buy
Q; 38% said they would buy R; and only 9% said they would buy lamp
S. They did not know which price matched which lamp, so they could not
use price for an identification cue for which lamp to purchase.
At the end of the data collection, respondents were told that one of the
lamps was designed by the famous designer Isamu Noguchi and they
were asked to guess which lamp he designed. Most people chose the IKEA
lamp S (51%) as being the designer lamp; 36% correctly identified the real
designer lamp R, and 13% identified the Chinese knock-off (lamp Q) as
the original designer lamp.
Therefore respondents preferred the designer lamp, especially when
rating the lamps without any information. When price information
was added, their preference became insignificant from the copy-cat
We did not ask why they would not purchase any of the lamps.
100
lamp. However, when asked which was the designer lamp, most people
misidentified the lamp that was least preferred as the designer lamp. There
were no differences in the correct identification of the lamps between men
and women (2 = 0.40, df = 1, N.S.).
The data on previous experience in lamp purchases were classified into
three categories: (1) those that had never purchased a lamp before (N = 9);
(2) those that had purchased one or two lamps (N = 22); and (3) those that
had purchased three or more lamps (N = 15). Again a chi-square analysis
of guess as to which was the designer lamp and previous lamp purchasing
experience was not significant (2 = 3.45, df = 2, N.S.). Respondents were
also asked to recall any designers they were aware of, and we classified
their responses into four groups: (1) those who recalled no designers (N =
14); (2) those who could name one designer (N = 7); (3) those who recalled
two designers (N = 16); and (4) those who recalled three or more designers
(N = 10) (no one named four and one person named five designers).
None of the respondents named the designer, Noguchi, used in this set
of experiments. There was no relationship between recall of number of
designers and ability to identify the correct designer lamp (2=5.26, df =
3, N.S.).
When the correlations were examined, the data showed that there was
no correlation between number of lamps previously purchased and number
of designers recalled (r = 0.05 N.S.). There was also no relationship
between number of designers recalled and the Centrality of Visual Product
Aesthetics Scale (r = 0.16, N.S.).
Study 2: Branded evaluations

Fifty-two new respondents (32 women and 20 men) participated in the
second experiment. In this study, the lamps were labelled or branded as the
Chinese lamp, the Isamu Noguchi lamp and the IKEA lamp. Therefore
the lamps were identified as originating with distinct manufacturers, or
lamps with brand names. The respondents were not told who Noguchi is,
or that he is a famous designer.
^
The results show that the Noguchi lamp was still preferred (a
R =
68) over the other two lamps (F(2,154) = 9.03, p < 0.001). However,
preference for the IKEA lamp increased and preference for the Chinese
lamp decreased when the lamps were labelled as such, and they were
^
^
equally preferred (a
S = 31 vs. a Q = 37: N.S.). The name or label Chinese
lamp may have a negative quality that depresses evaluation beyond the
actual design. Perhaps this cue evokes an image of poor craftsmanship
101
or even a faulty product. The name IKEA may have a very positive
effect on the evaluation due to the familiarity of the brand name in the
marketplace. For IKEA, this may be an example of Zajoncs (1980) theory
that familiarity breeds liking. When we compare scores from the first
experiment, we see the weight for the IKEA lamp shifted from 90 to
31, while the Chinese-labelled lamp shifted from +6 to 37. Therefore a
well-known brand name benefits the perception of the object. The Chinese
brand name may have a past general negative reputation and this could
be influencing perceptions.
When the prices were added to the evaluation, the scales compressed,
showing an effect similar to that seen in the first experiment: Chinese lamp
a^ Q = 13.33; Noguchi lamp a^ R = 30.67; IKEA lamp a^ S = 17.33. There
was no significant difference in the evaluation between the Chinese and
IKEA lamps with the prices attached.
The respondents were then asked whether they would consider buying
any of the lamps. Thirty-nine per cent said they would not buy any (same as
the first experiment); 8% said they would buy the Chinese lamp; 39% said
they would buy the Noguchi lamp; and 15% said they would buy the IKEA
lamp. Therefore with the brand names attached to the lamps, the purchase
preference rates for the Chinese lamp and the IKEA lamp inversed. Most
people still preferred the actual designer lamp. Again respondents stated
their purchase intentions without knowing which price belonged to which
lamp, so price could not be used as an indicator of any one lamp.
These data were analysed by gender and we found that slightly more
women than men indicated they would buy none of the lamps or the
Noguchi lamp (2 = 7.2, p = 0.07). We then looked at the possible relation
between buying intentions and their previous purchase experience (2 =
9.5, df = 6, N.S.) and knowledge of designers (2 = 4.3, df = 9, N.S.). No
significant relationships were found among any of these variables with
choice or evaluation of the lamps.
Study 3: Prime the designer

In the third study, 36 new respondents (21 women and 15 men) evaluated
the three rice paper lamps again, labelled only Q, R and S. This group was
informed at the beginning of the experiment that one lamp, but not which
one, was designed by the Japanese-American sculptor Isamu Noguchi,
and that his sculptures can be seen at famous local and international art
museums. That is, the subjects were primed for designer and given factual
information about the designer and his fame. The very first question posed
102
to the respondents was: Which one is the Noguchi lamp? The responses
were as follows: 6% thought it was the Chinese lamp; 36% correctly
identified the Noguchi lamp; and 58% wrongly identified the IKEA lamp
as the designer Noguchi lamp. Again, the majority of respondents selected
the IKEA lamp as the designer lamp, just as they did in the first experiment.
The respondents then went through the same procedure as for the first
experiment, rating the preference of the lamps labelled Q, R and S, with
and without the price factorial. Here almost opposite results to those
from the first experiment were found for the preference of the IKEA and
the Chinese lamps. The preference weights, when the respondents were
^
^
primed that one was a designer lamp, were as follows: a
Q = 98.67; a R
^
= 107.33; a
S = 8.67 (F(2,106) = 24.63, p < 0.001). Pairwise, each is
significantly different at the p < 0.01 level.
Without and with the price information, the Noguchi lamp, R, is
strongly preferred (107.33 vs 58), the IKEA lamp, S, is about neutral
(8.67 vs 4), and the Chinese lamp, Q, is least preferred (98.67 vs 54).
Finally, respondents were asked about their buying intentions: 33% said
they would buy none of the lamps; 6% said they would buy Q; 44% said
they would buy R; and 17% said they would buy S. Again, there was no
relation between this outcome and any of the other measured variables:
gender, CVPA scale, prior experience in purchasing lamps, or the number
of designers recalled.
There was no relationship between respondents guess as to which was
the designer lamp and the number of designers they were aware of (2 =
0.25, df = 3, N.S.; see Table 1). Among respondents believing that lamp R
was the Noguchi lamp (which it was), 62% expressed a buying intention
for this lamp, and 38% did not want to buy any of the lamps. None
of these respondents expressed any intention to purchase the other two
lamps, IKEA or Chinese. Among respondents who (wrongly) believed the
Noguchi lamp was S, 33% would not buy any of the lamps, 33% would
buy the Noguchi lamp, and 29% wanted to buy the lamp they (wrongly)
believed was the Noguchi lamp. Therefore giving people information
about the designer did seem to increase preference towards the lamp
they thought was the designer lamp. However, not always, as some who
wrongly believed that the IKEA lamp was the designer lamp indicated that
they would rather buy the Noguchi lamp.
This perhaps indicates, although the name (brand) dimension is effective,
that design is still very important in choice. It shows that it may be possible
to design with such virtuosity and quality that it may interfere with the
effect of a strong brand. The results do show the strengths of good design.
103
Table 1 Guesses of which is the designer lamp and buying intentions for Study 3

Indicated selection
Designer guess
R (correct Noguchi)
S (incorrect IKEA)
No purchase intention
38%
33%
Lamp Q 0% 6%
Lamp R
62%
33%
Lamp S 0%
29%
Exact likelihood ratio (2-sided) 2 = 7.9; df = 3; p < 0.1.
No doubt the combination of good design, good brand and good price
would be unbeatable.
CVPA scale and preference evaluation over the three studies

The computed Cronbachs alpha reliability for these five items on CVPA,
over three different samples, was 0.67, 0.85 and 0.84. When consumer
preferences for the lamps were examined by the mean split on CVPA
scores, the data showed that both low and high scorers equally preferred
the Noguchi lamp. No matter what other information was provided,
the Noguchi lamp was positively evaluated and preferred over the other
choices.
Differences were found in experiments two and three with regard to how
low and high CVPA scorers evaluated the choices. However, there does
not seem to be a clear answer on which consumers can identify a good
design using the CVPA scale. Those scoring high on the CVPA evaluated
the IKEA lamp less than the other choices only when it was without any
brand or designer prime. In fact, those scoring high on the CVPA were
also more likely to think the designer lamp was the IKEA one, when it was
not, and evaluated it more positively than those in the first experiment (see
Table2). Respondents who scored relatively lower on the CVPA scale had
a very low preference for the IKEA lamp and a strong negative preference
for the Chinese lamp. The label Chinese lamp might be interpreted as a
sign of a non-designer lamp and this might be interpreted as a switch-off
of the design context framing.
Discussion and conclusions

Over all three studies, the data found that the design of the lamps affects
individual preference, and hence the lamps as objects are indeed different.
In general the lamp that was well crafted by an established artist was
104
Table 2 Differences in preferences between low and high CVPA scores

Lower CVPA
Higher CVPA
Experiment 1: Naked lamps

N = 21
N = 26
Q 13.97 0.41
R 82.11 84.58
S
96.08
84.11

F(2,137) = 0.11; N.S.
Experiment 2: Branded
N = 31
N = 21
Chinese lamp
10.14
63.33
Noguchi lamp 66.51 82.33
IKEA lamp
56.37
19.00

F(2,152) = 2.67; p = 0.07
Experiment 3: Primed by designer
N = 24
N = 12
Q
66.74
162.25
R
101.60 118.11
S
34.86 44.14

F(2,104) = 4.27; p < 0.05
preferred, and hence the designs themselves clearly contain information. In

the first experiment, the interesting result is the Noguchi guess. Very few
people rightly guessed that R was the real designer lamp, despite the fact
that it was preferred. Most people wrongly guessed the odd-shaped lamp,
S, was the designer lamp, but it was least preferred by the respondents
that is, people associated the designer label with the avant-garde shape
they did not really prefer when it was unlabelled or when they were not
primed.
There is, not surprisingly, no correlation between the guesses as to
which is the designer lamp and design aptitude, prior experience in buying
lamps, gender or knowledge of designers. However, most people chose
the Noguchi lamp as the one they would prefer to buy, and this was not
related to their ability to identify the designer lamp. Lamp S, which is
shaped like a conch and lacking a tripod, may take time to perceive and
judge, hence perhaps it was given a lower preference based on the theory
of cognitive consistency that is, the design of the conch as a lamp was
not consistent with how they think a lamp should be formed. The reader
is also reminded that the respondents never knew and were not told the
actual prices of each lamp and therefore could not use price as a cue for
designer.
The results of the second experiment found rather strong effects for
brand name and are perhaps to be expected. Here the Chinese label
resulted in a more negative evaluation than when respondents evaluated
105
the same lamp without the Chinese label. Made-in-China products are
frequently associated with poor quality and are the source of countless
deficient product recalls (BBC News 2007). The cue may be negative, and
hence it interferes with the design elements of the lamp. The lamp that was
evaluated the most negatively (S) now turned positive with the IKEA brand
label. IKEA is a well-known international brand representing value for
money (especially to the middle-class, and most of our subjects were selfdescribed as middle-class). This positive well-known cue overrode what
was perceived as a less desirable design when unbranded. Therefore the
connotations of the brand go far beyond the visual aspects of the design
and people may rely on deep-seated or previously learned beliefs about
names or brands for evaluation (Hoeffler & Keller 2003).
Study 3 showed how priming respondents about a designer and
providing concrete information about the fame of the designer shifted
preferences. Furthermore, the initial guess of which was the designer lamp
appeared to anchor their preferences. The design knowledge had a strong
influence when matched with a real superior design. So the answer to the
question of How do you convert price-sensitive consumers to designsensitive consumers, so they are willing to pay more for their goods? may
lie in giving consumers concrete information about the designer at the
point of purchase. It is likely that the more prestigious the information is
perceived as or valued by the consumer, the more likely the shift will be to
the designer brand name.
In this study respondents were primed, in the narrative, that the sculptor
Isamu Noguchi has his sculptures located in famous art museums. These
are strong institutions that may have an equivalent effect as a strong
brand. The effect of a designer name, known only to a few people, when
associated with a strong institution may increase the desire to own objects
made by the designer. This relationship can be thought of in terms of
Heiders Balance Theory (1958). The individual highly regards the art
museum, and if the designers goods are in the art museum, they must be
valuable and worth the individual owning at a higher price. Therefore this
shift to the designer good keeps the cognitive aspects of the individual
positively in balance.
Research implications
The studies are valid in that they are conducted in situations where real
consumers are conducting their business. Replications with other product
categories and actual purchases are needed to show the robustness of the
106
findings. The links between design, designer and brand name are ripe for
further research. The name of a designer like Alessandro Mendini may
add to the brand, like Alessi, and add recognition and image to a product
(Keller 1993). Future research also needs to look at the differences between
publicly consumed luxury goods versus privately consumed necessities.
While this study found an effect for designer brand using a relatively
privately consumed good, the effects for publicly consumed luxury goods
may be even more pronounced.
A direct extension of this study might be to establish a standard for
evaluating good design. It seems essential to have an expert view of
the particular designs. For instance, a panel of acclaimed design experts
who have academic credentials and/or serve as competition judges might
establish a frame of reference of measurement in connection with good
design. A different approach might be to reduce the emphasis on branding
and to establish design as the major issue. Doing that allows one to
measure how much various forms of framing will add to the pure designs.
The design effects should be reinforced when supported by good
branding, labelling and/or relevant concrete information pertaining to
the design. Market researchers can assist companies by testing specific
information in their communication strategies and measuring the responses
to the brand, gaining a deeper understanding of consumers evaluations
and price trade-offs for design elements. They may use this information as
a key element in creating demand for originals, as opposed to imitations,
knock-offs or retail brand copies.
There are no doubt individual differences in the ability to identify good
design. However, we cannot confidently say that our measures were able
to capture those differences. It could be that cognitively based paper and
pencil measures correlate with learned inferences and not with automatic
reactions to the stimulus. Experiments using eye tracking and response
times might add to the knowledge of important design elements, as they
are used as a proxy for processing fluency, which is a measure for liking
(Lee & Labroo 2004). Market research is now relying on more invasive
measures, such as fMRI, to establish real reactions to different package
designs (Reimann et al. 2008). This integration of brain activity to market
research may soon become the status quo.
107
References
Bar, M. & Neta, M. (2006) Humans prefer curved visual objects. Psychological Science, 17, 8,
pp. 645648.
BBC News (2007) New recall of toys made in China. Business, 27 September.
Bloch, P.H. (1995) Seeking the ideal form: product design and consumer response. Journal of
Marketing, 59 (July), pp. 1629.
Bloch, P.H., Brunel, F.F. & Arnold, T. (2003) Individual differences in the centrality of visual
product aesthetics: concept and measurement. Journal of Consumer Research, 29 (March),
pp. 551565.
Bourdieu, P. (1984) Distinction. London: Routledge.
Business Week On-Line (2006) Hyundais hunt for distinctive design. Business Week On-Line,
2, 23, p. 7.
Chitturi, R., Raghunathan, R. & Mahajan, V. (2008) Delight by design: the role of hedonic
versus utilitarian benefits. Journal of Marketing, 72 (May), pp. 4863.
Csikszentmihalyi, M. & Robinson, R.E. (1990) The Art of Seeing. Malibu, CA: J. Paul Getty
Museum.
Dawar, N. & Parker, P. (1994) Marketing universals: consumers use of brand name, price,
physical appearances, and retailer reputation as signals of product quality. Journal of
Marketing, 58 (April), pp. 8195.
Douglas, M. (1996), Thought Styles. London: Sage.
Fabricant, F. (2006) Teas new bag. Globe and Mail, 16 September, p. L2.
Hagtvedt, H. & Patrick, V.M. (2008) Art infusion: the influence of visual art on the perception
and evaluation of consumer products. Journal of Marketing Research, 45 (June), pp. 379389.
Heider, F. (1958) The Psychology of Interpersonal Relations. New York: Wiley.
Heskett, J. (2002) Toothpicks and Logos: Design in Everyday Life. Oxford: Oxford University
Press.
Hoeffler, S. & Keller, K.L. (2003) The marketing advantages of strong brands. Brand
Management, 10, 6 (August), pp. 421445.
IIPA (2007) The Copyright Industries Note USTR Decisions in its Annual Special 301 Report.
Washington, DC: International Intellectual Property Alliance.
Johnson, M.D. & Lehmann, D.L. (1997) Consumer experience and consideration sets for
brands and product categories. In D. MacInnis & M. Brucks (eds) Advances in Consumer
Research, 24, Association for Consumer Research, pp. 295300.
Keller, K.L. (1993) Conceptualizing, measuring, and managing customer-based brand equity,
Langley, A., Mintzberg, H., Pitcher, P., Posada, E. & Saint-Macary, J. (1995) Opening up
decision making: the view from the black stool. Organization Science, 6, 3 (May/June),
pp.260279.
Lee, A.Y. & Labroo, A.A. (2004) The effect of conceptual and perceptual fluency on brand
evaluation. Journal of Marketing Research, XLI (May), pp. 151165.
Lichtenstein, D.R., Bloch, P.R. & Black, W.C. (1988) Correlates of price acceptability. Journal
of Consumer Research, 15 (September), pp. 243252.
Loewenstein, G.F., Weber, E.U., Hsee, C.K. & Welch, N. (2001) Risk as feelings. Psychological
Bulletin, 127, 2, pp. 267286.
Louie, E. (2006) Bill Stumpf, 70, a designer of the Aeron chair, dies. New York Times,
10September, p. 15.
Luo, L., Kannan, P.K. & Ratchford, B.T. (2008) Incorporating subjective characteristics in
product design and evaluations. Journal of Marketing Research, 45, 2, (May) pp.181194.
McCracken, G. (1988) Culture and consumption: a theoretical account of the structure and
movement of cultural meaning of consumer goods. Journal of Consumer Research, 13 (June),
pp. 7184.
108
McCubbins, T.F. (2004) Product design, trade dress and the law. Business Horizons, 47, 1
(January/February), pp. 36.
Micucci, D. (2003) Dismantling of work by Noguchi sets off protests. International Herald
Tribune, 2526 October, p. 6.
National Post (2007) No Fugly on the floor. National Post, 7 July, p. WP 6.
Orth, U.R. & Malkewitz, K. (2008), Holistic package design and consumer brand impressions.
Journal of Marketing, 72 (May), pp. 6481.
Petroski, H. (1994) The Evolution of Useful Things: How Everyday Artefacts from Forks
and Pins to Paper Clips and Zippers Came to Be as They Are. New York: Vintage Books.
Reber, R., Schwarz, N. & Winkielman, P. (2004) Processing fluency and aesthetic pleasure:
is beauty in the perceivers processing experience? Review of Personality and Social
Psychology, 8, 4, pp. 364382.
Reimann, M., Newhaus, C., Weber, B. & Zaichkowsky, J. (2008) Using the fMRI to
understand consumer perception of differentiation among low involvement goods. Working
paper, Stanford, Department of Psychology.
Srinivasan, V., Lovejoy, W.S. & Beach, D. (1997) Integrated product design for marketability
and manufacturing. Journal of Marketing Research, 34, 1, pp. 154163.
Tversky, A. & Kahneman, D. (1981) The framing of decisions and the psychology of choice.
Science, 211, pp. 453458.
Van Osselaer, S.M.J. & Alba, J.W. (2000) Consumer learning and brand equity. Journal of
Consumer Research, 27 (June), pp. 116.
Veryzer, R.W. Jr (1995) The place of product design and aesthetics in consumer research.
In F. Kardes & M. Sujan (eds) Advances in Consumer Research, Vol. 22, Association for
Consumer Research, Provo, Utah, pp. 641645.
Veryzer, R.W. Jr (1999) A nonconscious processing explanation of consumer response to
product design. Psychology and Marketing, 16 (September), pp. 497522.
Veryzer, R.W. Jr & Hutchinson, J.W. (1998) The influence of unity and prototypicality on
aesthetic responses to new product designs. Journal of Consumer Research, 24 (March),
pp.374394.
Winkielman, P., Halberstadt, J., Fazendeiro, T. & Catty, S. (2006) Prototypes are attractive
because they are easy on the mind. Psychological Science, 17, 9, pp. 799806.
Zajonc, R.B. (1980) Feeling and thinking: preferences need no inferences. American
Psychologist, 35, 2, pp. 151175.
Zolli, A. (2004) Why design matters more. American Demographics, October, pp. 5253.
About the authors

Gorm Gabrielsen holds a Masters in Mathematics and Mathematical
Statistics from the University of Copenhagen (1976). Since 1983 he has
been appointed as Associate Professor of Statistics at Copenhagen Business
School. He has published on theoretical matters and worked as a statistical
consultant for both companies as well as official organizations. His special
areas of interest are general social science, forensic psychiatry, design
management and food research.
Tore Kristensen is Professor of Strategic Design at the Copenhagen
Business School. His published work concerns design of products and
the value of design for a company. His research includes design of
109
environments as seen in a consumer perspective, as well as the use of

physical space to increase creativity. His most recent research involves
sensorics and the value of music.
Judith Lynne Zaichkowsky is Professor of Marketing and Communications
at the Copenhagen Business School. She received her PhD from UCLA in
Marketing, with minors in Psychology and Statistics. Dr Zaichkowskys
1985 Journal of Consumer Research paper on the involvement construct
has been recognised as one of the top cited articles in consumer behaviour
and one of the most influential articles in the field of advertising. She is
also the author of The Psychology behind Trademark Infringement and
Counterfeiting.
Address correspondence to: Judith Lynne Zaichkowsky, Professor of
Marketing and Communications, Department of Marketing, Copenhagen
Business School, Solbjerg Plads 3, 2000 Frederiksberg, Denmark.
Email: zaichkow@sfu.ca
110
Purchasing behaviour in an online

supermarket
The applicability of E-S-QUAL
Frederic Marimon
International University of Catalonia
Richard Vidgen
University of New South Wales
Stuart Barnes
University of East Anglia
Eduard Cristbal
University of Lleida
The purpose of this paper is twofold: first, to assess the applicability of the four
dimensions of online service quality, as proposed in the E-S-QUAL scale, to the
setting of an online supermarket; and, second, to propose and test a model that
links these e-quality dimensions with loyalty and purchasing behaviour in the
setting of an online supermarket.
An online questionnaire was used to survey 131 customers of an online Spanish
supermarket using the E-S-QUAL scale. The data were analysed by exploratory
factor analysis to test the applicability of the E-S-QUAL scale to the setting of
an online supermarket and generate an extended model (including constructs for
perceived value, loyalty and actual purchases). The model was then checked
by structural equation modelling (SEM).
The four dimensions proposed by the E-S-QUAL scale were confirmed in the
setting of an online Spanish supermarket. The influence of these various quality
dimensions on perceived value, loyalty and actual purchases are delineated here.
The study reassures online vendors that E-S-QUAL is an appropriate
instrument by which to measure online service quality. The study also provides
empirical evidence that high levels of e-service quality have a positive influence
on purchasing behaviour.
The study is the first to provide definitive empirical evidence of the commonly
presumed linkage between the quality dimensions proposed in the E-S-QUAL
scale and the constructs of loyalty and actual not self-reported purchase
behaviour.
Received (in revised form): 28 January 2009

DOI: 10.2501/S1470785310201089
111
Purchasing behaviour in an online supermarket
Introduction
The well-known SERVQUAL instrument (Parasuraman et al. 1985, 1988;
Parasuraman et al. 1991) has become one of the most widely used generic
measures of service quality. More recently, Parasuraman et al. (2005) have
developed a new instrument, known as E-S-QUAL, to assess service quality
in the context of electronic commerce (e-commerce). The E-S-QUAL scale
is composed of 22 items arranged in four dimensions for measuring the
service quality delivered by online retail websites. Although this scale has
been in existence for only a few years, it has nevertheless sparked debate
in the literature with regard to its efficacy (Ingle & Connolly 2006; Kim
etal. 2006; Boshoff 2007; Connolly 2007).
The purpose of the present study is to explore and clarify some aspects
of this debate. In general, criticism of E-S-QUAL has focused on the
interrelated issues of its structure and its applicability to various settings.
Questions have been raised about the number and nature of the dimensions
of service quality proposed in the instrument (Collier & Bienstock 2006;
Boshoff 2007), while other authors have noted that there are few reports
of practitioners having actually applied the measure (Ingle & Connolly
2006; Kim et al. 2006; Boshoff 2007).
In addition, the present authors note that, whatever the merits of the
E-S-QUAL scale, it is of interest that there are few (if any) studies that
have established an empirical link between service quality and actual sales
in the online environment. Although some studies have proposed models
linking e-quality with loyalty (Ribbink et al. 2004; Enzmann & Schneider
2005; Chiou & Shen 2006), none appears to have established a definite
link between online service quality and actual purchases.
The objectives of the present study are therefore twofold:
1. to explore the suitability of E-S-QUAL for application in a particular
e-commerce setting (an online Spanish supermarket), and
2. to analyse how website quality impacts upon customer loyalty and
actual e-sales.
The structure of the paper is as follows. In the next section we discuss
the literature and the development of the research model and hypotheses
tested in the study. This is followed by a section outlining the methodology
used in the research. After that we report on the findings of our analysis,
including factor analysis, PLS, and tests for validity and reliability of
the structural model. The penultimate section discusses the findings and
provides juxtaposition with the literature. Finally, the last section rounds
112
off with conclusions, including implications for future research and

practice in this area.
Literature and research model

E-S-QUAL (Parasuraman et al. 2005) was not the first attempt to establish
a measurement scale of various aspects of e-commerce service. According
to several reviews of the literature in this area (Barnes 2006; Cristbal
& Marimon 2007; Li & Suomi 2007), the pioneers in assessing website
quality have been Barnes and Vidgen (2000, 2002), and Loiacono et al.
(2000, 2007). Each of these research teams developed their own scale
separately, and both called it WEBQUAL (although it was the latter who
formally registered the term). However, as Kim et al. (2006) have pointed
out, both of these scales focus on the user experience and technical quality
of websites, rather than on the quality of the entire service provided. Other
significant contributions in this emerging area of research have been made
by Yoo and Donthu (2001), who developed the so-called SITEQUAL
scale, and van Riel et al. (2001), who identified three dimensions that can
be used to evaluate overall customer satisfaction and intention to use a
website again in the future.
This stream of research on e-quality has evolved in parallel with a
marked increase in online sales, which has triggered a growing interest in
the measurement of the quality of online services. Web presence and low
prices were previously believed to be key drivers of success (Yang & Fang
2004); however, it has become apparent more recently that this is not
enough (Kim et al. 2006). In these circumstances, instruments such as E-SQUAL might have been expected to attract interest among e-commerce
vendors that wish to make an accurate assessment of the quality they are
providing to their customers. However, several authors have noted that
there are few reports of practitioners having actually applied the measure
(Ingle & Connolly 2006; Kim et al. 2006; Boshoff 2007).
The E-S-QUAL scale (Parasuraman et al. 2005) is composed of 22 items
arranged in four dimensions for measuring the service quality delivered by
online retail websites. The four dimensions are as follows.
1. Efficiency: ease and speed of accessing the site (eight items).
2. System availability: reliable technical functioning of the site (four
items).
3. Fulfilment: the extent to which the sites promises about order delivery
and product availability are fulfilled (seven items).
113
4. Privacy: the degree to which the site is safe and protects customer
information (three items).
Some academics have expressed concerns about the scale. In particular,
Boshoff (2007) has contended that a six-factor configuration provides a
better explanation of e-commerce service quality than the original fourfactor model of Parasuraman et al. (2005). Four of Boshoff s (2007)
factors were quite close to the original model, but two new ones have also
been postulated by splitting the original fulfilment factor into two factors
(called delivery and reliability). However, despite these misgivings,
Boshoff (2007) has acknowledged that E-S-QUAL is, in general, a useful
tool.
Other authors, such as Connolly (2007), have concluded that the
ESQUAL model is useful in a variety of situations. Collier and Bienstock
(2006) also applauded E-S-QUAL as an important step in conceptualising
e-service quality, although they proposed a wider model with some new
constructs; their main objection was that E-S-QUAL uses only reflective
indicators, whereas they contended that e-service quality is made up of
formative indicators.
Given that the instrument is quite recent, debate about its utility is
to be expected, both within the academic field and among e-business
practitioners as well. It is apparent that further research and practical
testing of this instrument is required.
Whatever the merits of the E-S-QUAL scale, it is of interest that few
(if any) studies have linked e-commerce quality with e-commerce sales.
There are some studies that have proposed models linking e-quality with
loyalty (Ribbink et al. 2004; Enzmann & Schneider 2005; Chiou & Shen
2006), but it is difficult to find any literature that attempts to explain
whether (and how) quality impacts on actual purchases. For example,
Huang (2008) found that loyalty to a web-based travel agency resulted in
an increased intention to purchase from that agency, but the author had
nothing to say about actual sales.
Research model
Drawing on E-S-QUAL and the work of Boshoff (2007), who hypothesised
relationships between the previously identified dimensions of e-service
quality (efficiency, system availability, fulfilment and privacy) and
the constructs perceived value and loyalty, we extend the model to
include actual purchases (Figure 1).
114
Efficiency
H1
System
availability
H2
Perceived
value
H3
H5
Loyalty
H6
Actual
purchases
Fulfilment
H4
Privacy
Figure 1 Hypothesised relationships between the dimensions of e-service quality and the
constructs of perceived value, loyalty, and actual purchases
The hypotheses can be summarised as follows.

Hypothesis H1: Higher levels of efficiency in a website are
positively related to higher levels of perceived value.
Hypothesis H2: Higher levels of system availability in a website
are positively related to higher levels of perceived value.
Hypothesis H3: Higher levels of fulfilment in a website are
Hypothesis H4: Higher levels of privacy in a website are
Hypothesis H5: Higher levels of perceived value in a website are
positively related to higher levels of loyalty with regard to that
website.
Hypothesis H6: Higher levels of loyalty with regard to a website
are positively related to higher levels of actual purchases on that
website.
115
The first five hypotheses are similar to the Boshoff (2007) model. For the
sixth hypothesis, the present study added a new dependent variable (actual
purchases), which has not been reported on previously in e-service research.
Methodology
Data collection and sample
The setting for the research is a Spanish supermarket, Plusfresc, which
provides an online ordering system to its customers. Plusfresc is a
supermarket chain with over 75 years of history, which currently has 69
shops and more than 850 workers. It is the leader in sales in the local
area where it operates, with more than 150,000 customers per week
and a turnover of 113 million in 2007. The supermarket has a long
and substantial history of innovation and customer orientation. In 1996,
it introduced its loyalty card, Plusi, the first customer marketing card
in Catalonia. In 1998, Plusfresc won the Global Electronic Marketing
Award in the United States for the best electronic marketing programme
presented by a non-American company. Following this trajectory of
technology adoption, in 2001 Plusfresc developed its online presence and
by 2007 was getting more than 16,600 visits, resulting in 4284 orders with
an average expenditure of 111.5. Although online sales, at 479,231, are
a relatively small part of total turnover they are recognised as a growth
area for sales and a strategic capability for Plusfresc.
The E-S-QUAL instrument (Parasuraman et al. 2005) was adapted for
application in the case studied here. Table 1 provides a list of the quality
dimensions (and their items) used in the present study. As in the original
E-S-QUAL model, responses were recorded on five-point Likert-type scales
with anchors of 1 = strongly disagree and 5 = strongly agree. Responses to
the items constituting perceived value and loyalty were made on a fivepoint Likert-type scale with anchors of 1 = poor and 5 = excellent.
Only two slight modifications to items (both within the dimension
of fulfilment) were made from the original E-S-QUAL scale. The
supermarket delivers customer orders daily on a fixed delivery schedule
with a failure rate close to zero. To reflect this operating environment,
item FUL3 was removed and item FUL7 was reworded to assess delivery
convenience rather than delivery forecast accuracy.
Table 2 shows the items used for perceived value and loyalty (the
same that were used in Parasuraman et al. 2005), with the same Likert
scale from 1 = strongly disagree to 5 = strongly agree.
116
Table 1 Questionnaire factors and items for service quality (adapted from E-S-QUAL)*
Efficiency
EFF1
EFF2
EFF3
EFF4
EFF5
EFF6
EFF7
EFF8
This site makes it easy to find what I need.

It makes it easy to get anywhere on the site.
It enables me to complete a transaction quickly.
Information at this site is well organised.
It loads its pages fast.
This site is simple to use.
This site enables me to get on to it quickly.
This site is well organised.
System availability
SYA1
This site is always available for business.
SYA2
This site launches and runs right away.
SYA3
This site does not crash.
SYA4
Pages at this site do not freeze after I enter my order information.
Fulfilment
FUL1
It delivers orders when promised.
FUL2
This site makes items available for delivery within a suitable time frame.
FUL4
It sends out the items ordered.
FUL5
It has in stock the items the company claims to have.
FUL6
It is truthful about its offerings.
FUL7
The delivery time offered to me is convenient.
Privacy
PRI1
PRI2
PRI3
It protects information about my web shopping behaviour.

It does not share my personal information with other sites.
This site protects information about my credit card.
*FUL3 in original E-S-QUAL is removed and FUL7 has been reworded to reflect the fixed delivery times of
the supermarket operation.
Table 2 Items for constructs perceived value and loyalty

Perceived value
PEV1
The prices of the products and services available at this site (how economical the site is).
PEV2
The overall convenience of using this site.
PEV3
The extent to which the site gives you a feeling of being in control.
PEV4
The overall value you get from this site for your money and effort.
Loyalty
LOY1
LOY2
LOY3
LOY4
LOY5
Say positive things about this site to other people?

Recommend this site to someone who seeks your advice?
Encourage friends and others to do business with this site?
Consider this site to be your first choice for future transactions?
Do more business with this site in the coming months?
117
Respondents were uniquely Table 3 Items for construct actual purchases

identifiable through their customer
Actual purchases
ID. This allowed each respondent
Number of online orders in 2007:
to be linked to their actual PUR1

1 = one or two orders
purchases at Plusfresc. Data made
2 = three or four orders
available by Plusfresc included
3 = between 5 and 9 orders

4 = between 10 and 19 orders
the number of orders a customer

5 = 20 orders or more
made and the total amount (value)
PUR2
Total value of online orders in 2007:
of those purchases. The purchase

1 = <175
data were categorised into PUR1
2 = between 176 and 500
(number of purchases) and PUR2
(value of purchases), as in Table 3.

5 = >1501
Each item was grouped as: (i) five
possible answers for number of
purchases; and (ii) five possible answers for total monetary amount. This
operationalisation of the actual purchases construct reflects both the value
and frequency of purchases made.
The questionnaire was distributed to all 413 online customers of the
supermarket using a specific web page designed for the purpose. Since the
population was not large, all respondents were encouraged to answer the
questionnaire by the provision of a small reward to each. As a result, 131
valid responses were received, which implies an acceptable response rate of
31.7%. For instance, Boshoff (2007), in his study devoted to assessment of
the E-S-QUAL scale, obtained a response rate of 20.4%. Manfreda et al.
(2008) have specifically analysed the response rates in web surveys. They
compared 45 studies, and the response rates were between 11.3% and
82.1%, with an average of 32.6% and a standard deviation of 17.3%.
Roster et al. (2004) also obtained a similar rate in their study comparing
response rates between web surveys and telephone surveys: 32.6% for web
surveys and 40.5% for telephone surveys. Thus, overall, our response rate
compares favourably with similar studies.
The demographic characteristics of the sample are summarised in
Table4. A gender bias (75.6% female) was detected. Almost half (48.1%)
of the respondents were aged between 35 and 44 years. This represents
a higher percentage than in the total population of the online buyers in
Plusfresc, and consequently older buyers were under-represented. This is
also detected in the Boshoff study. The educational level of the sample
was high, with more than half (57.3%) of the sample having a university
degree. This kind of bias was also detected in the Parasuraman et al. (2005)
study and in Boshoff (2007). To place our study in context, consider the
118
Table 4Demographic characteristics of the response sample and the population of online
shoppers of Plusfresc

Sample
Population
Number
Number
Age category
24 years
Between 25 and 34 years
65 years
Total
0
37
63
26
3
2
131
0
28.2
48.1
19.8
2.3
1.5
100.0
4
99
165
100
34
11
413
1.0
24.0
40.0
24.2
8.2
2.6
100.0
Gender
Male
Female
Total
32
99
131
24.4
75.6
100.0
129
284
413
31.2
68.8
100.0
Education level
Without studies
Primary school
Secondary school
University degree
Others
Total
1
5
46
75
4
131
0.8
3.8
35.1
57.3
3.1
100.0
5
26
112
249
21
413
1.2
6.3
27.1
60.3
5.1
100.0
following population data. The latest study of B2C e-commerce published

by the Spanish Ministry of Industry, Tourism and Trade (2008) shows that
online shoppers in Spain were 40.5% female in 1997, with 40.2% aged
between 25 and 35 years and 36.5% with a university degree.
Findings
An array of exploratory factor analysis taking the items of Table 1 was
conducted to identify the quality dimensions formed from our data.
The scale was analysed in accordance with John and Reves (1982) and
Hair et al.s (1998) recommendations. As expected, the four E-S-QUAL
dimensions appeared, although with some discrepancies from the original
composition.
1.
2.
3.
4.
Efficiency: EFF1, EFF4, EFF8.

System availability: EFF5, SYA2, SYA3, SYA4.
Fulfilment: FUL1, FUL2, FUL4, FUL5, FUL7.
Privacy: PRI1, PRI2, PRI3.
119
The first factor is efficiency, which contains EFF1, EFF4 and EFF8. Item
EFF5 loaded on system availability. However, closer inspection of the
wording of the item (It loads its pages fast) may explain this migration
since it is understandable that this could be interpreted as being closer to
the factor of system availability than to that of efficiency. In system
availability, item SYA1 (availability) loaded almost equally on two factors
and was removed. The remaining efficiency factors loaded on more than
one component with item EFF6 loading equally on three factors. For
fulfilment, all the items loaded with the exception of FUL6. All three
privacy items loaded cleanly.
The next step was to assess the reliability of each of these four quality
factors (see Table 4). Acceptable levels were achieved in all criteria usually
used for this purpose (Hair et al. 1998). Cronbachs and composite
reliability in every case exceeded the threshold value of 0.7 for internal
consistency (Nunnally & Bernstein 1994). In addition, four factorial
exploratory analyses were performed, one for each factor, and all extracted
only one factor. These findings confirmed the unidimensionality of each
item to its first-order dimension.
A first-order confirmatory factorial analysis was performed using EQS
software. In view of the size of the sample, a robust maximum-likelihood
estimation method was chosen. The comparative fit index (CFI) was 0.945
and the root mean-square error of approximation (RMSEA) was 0.051.
The fit indices shown in Table 6 are acceptable (Byrne 1994; Hu & Bentler
1999). The Satorra-Bentler scaled chi-square was 112.7 on 84 degrees of
freedom and its probability value for the chi-square statistic was 0.02. The
loads were all high (at a significance level of 0.05). The model is therefore
an acceptable fit for the data.
Content validity of the scale can be assumed on the basis of the close
similarity between the present scale and the original E-S-QUAL model of
Table 5 Reliability analysis of the E-S-QUAL adapted subscales

Subscale

Items
Efficiency
EFF1, EFF4, EFF8
System availability
SYA2, SYA3, SYA4, EFF5
Fulfilment
FUL1, FUL2, FUL4, FUL5, FUL7
Privacy
PRI1, PRI2, PRI3
120

Range for

Cronbachs

alpha
Cronbachs removing
alpha
one item
0.756
0.887
0.757
0.879
0.5610.811
0.8390.869
0.6820.733
0.8180.835
Range for
correlations
of the items and
the sum of
the subscale
0.4490.659
0.7160.795
0.4320.605
0.7530.778
Table 6Loads on quality factors and goodness fit statistics for E-S-QUAL adapted scale

Efficiency
EFF1
EFF4
EFF8
Loadings*
r2
0.471
0.773
0.902
0.222
0.597
0.814
System availability
SYA2
0.869
SYA3
0.811
SYA4
0.768
EFF5
0.810
0.755
0.658
0.590
0.656
Loadings*
r2
Fulfillment
FUL1
0.691
FUL2
0.638
FUL4
0.631
FUL5
0.529
FUL7
0.623
0.477
0.407
0.398
0.280
0.388
Privacy
PRI1
0.828
PRI2
0.844
PRI3
0.854
0.685
0.712
0.729
* These are standardised loading estimates from confirmatory factor analysis. All parameters significant
at p < 0.05.
Goodness of fit statistics (robust method)

c2
112.698 (p-value = 0.02004)
df
84
CFI
0.945
Bentler-Bonett non-normed fit index
0.931
Bollens (IFI) fit index
0.947
RMSEA
0.051
90% confidence interval for RMSEA
0.21; 0.74
Parasuraman et al. (2005). Convergent validity was confirmed when the

factor loadings of the confirmatory model were found to be statistically
significant (level of 0.05) and greater than 0.5 points (Sanzo et al. 2003);
only one parameter was below this threshold (EFF1 = 0.471).
In summary, the first objective of this study has been realised by
establishing that a scale that is very close to the generic E-S-QUAL scale
is suitable for assessment of e-quality in an online supermarket. The same
dimensional factors have been retained, although some unsuitable items
have been removed from the original E-S-QUAL list.
Relationship between e-quality and purchases

In pursuit of the second objective of this study (an analysis of the extent
to which website quality impacts upon customer loyalty and e-sales) a
new construct was required: actual purchases. This item was created
specifically for this study because, as previously noted, there is apparently
no study in the extant literature that has attempted to correlate actual
purchase information with e-service quality. Two additional constructs
were used: perceived value and loyalty, based on Parasuraman et al.
121
(2005), and modelled following Boshoff (2007). The analysis conducted

on the four E-S-QUAL factors was also conducted on these three factors
and, as a result, one item from perceived value, PEV1 (economy), was
dropped. All the loyalty items were retained, as were items for actual
purchase. Table 7 shows the reliability analysis for these constructs.
Table 7 Reliability analysis of the constructs Perceived value, Loyalty and Purchasing

Subscale
Perceived value
Loyalty
Purchasing

Items
PEV2, PEV3, PEV4
LOY1, LOY2, LOY3, LOY4, LOY5
PUR1, PUR2

Range for

Cronbachs

alpha
Cronbachs removing
alpha
one item
0.826
0.896
0.958
0.7230.783
0.8490.898

Range for
correlations
of the items and
the sum of
the subscale
0.6590.718
0.6280.843
0.920
Structural equation modelling (SEM) was conducted using a partial

least squares (PLS) procedure run in the Smart-PLS software package,
which allows path modelling with latent variables (Ringle et al. 2005).
The PLS procedure is able to model latent constructs under conditions of
non-normality and small-to-medium sample size (Compeau & Higgins
1995). Rather than assuming equal weights for all indicators of a scale,
PLS allows each indicator to vary with regard to how much it contributes
to the composite score of the latent variable. Indicators with weaker
relationships to related indicators and the latent construct are thus given
lower weightings. A power analysis using G*Power 3.0 ( = 0.05, =
0.20) indicates that the sample size is adequate for explaining medium
population effects, where f 2 > 0.11 (Faul et al. 2007).
The analysis in Table 8 shows that four hypotheses in our original
research model are accepted: H2, H3, H5 and H6. The rejection of H1
and H4 suggests that efficiency and privacy have no influence on value (at
least in the context of an online supermarket).
Discussion
The present analysis found no relationship between efficiency and
perceived value (thus rejecting H1). This finding represents the main
difference between the present study and previous studies (Parasuraman
et al. 2005; Boshoff 2007). It would seem that the difference arises from
the specific context. Whereas a generic online buyer is concerned about
122
Table 8 PLS summary

Loadings

EFF
FUL
PRI
SYA
VAL
LOY
PUR
EFF1
0.6744**
EFF4
0.8884**
EFF8
0.8846**
FUL1
0.7264**
FUL2
0.6620**
FUL4
0.7917**
FUL5
0.6081**
FUL7
0.7339**
PRI1
0.8756**
PRI2
0.9241**
PRI3
0.8872**
EFF5
0.8514**
SYA2
0.8850**
SYA3
0.8374**
SYA4
0.8449**
PEV2
0.8657**
PEV3
0.8374**
PEV4
0.8799**
LOY1
0.8791**
LOY2
0.9060**
LOY3
0.8840**
LOY4
0.7695**
LOY5
0.7648**
PUR1
PUR2
0.9818**
0.9778**
AVE
0.6755
0.5002
0.8026
0.7469
Cronbachs a
0.7557
0.7571
0.8792
0.8871
Composite reliability 0.8602
0.8323
0.9242
0.9219
R2
0.9600
0.9584
0.9796
0.0792
0.7416
0.8259
0.8959
0.3645
0.7014
0.8964
0.9243
0.4301
Path parameters estimates

0.0791
H1 : EFFVAL
H2 : SYAVAL
0.3404*
H3 : FULVAL
0.3058**
H4 : PRIVAL
0.0249
H5 : VALLOY
0.6558**
H6 : LOYPUR
0.2814*
Note: Significance levels denoted by * (1%) and ** (0.1%).
efficiency, a buyer in an online supermarket apparently attaches less

importance to this dimension perhaps because these customers expect
that the service will be efficient in an online supermarket. Another possible
explanation for anomalies in the efficiency dimension has been suggested
by Kim et al. (2006). According to these authors, most of the website
123
attributes that facilitate efficient online shopping and purchasing (such

as advanced search functions, comparison shopping, sitemaps, browsing
instructions, order instructions, express checkouts and e-billing) have
usually been overlooked in previous studies. This observation suggests that
the construct of efficiency should (perhaps) be constituted by other items.
The present study found that system availability had a significant
impact on value, thus confirming H2. Parasuraman et al. (2005) also found
that this dimension was a critical contributor to customers perceptions of
value. This finding cannot be compared with Boshoff (2007) because this
dimension was not included in his model.
The present analyses also found that fulfilment had a significant impact
on perceived value, thus confirming H3. Boshoff (2007) concurred with
this conclusion in finding that fulfilment was the strongest predictor
of value perception in his study. Similarly, Parasuraman et al. (2005)
found that efficiency and fulfilment were the most critical (and equally
important) dimensions of website service quality. The present study concurs
with respect to fulfilment; however, as discussed above, the present study
failed to establish a relationship between efficiency and perceived value.
The present study rejected H4 (that higher levels of privacy in a website
are positively related to higher levels of perceived value). This finding
differed from both Boshoff (2007) and Parasuraman et al. (2005), both of
which confirmed this hypothesised relationship (albeit with a lower level
of significance than that found in the other three quality factors). However,
it should be noted that Parasuraman et al. (2005) observed that previous
research had suggested that privacy might not be a critical factor. In
view of these variable findings, it is apparent that the influence of privacy
on perceived value requires further investigation. The relationship might
depend on the sector being analysed; for example, the cases of Amazon
and Walmart, in which Parasuraman et al. (2005) detected a relationship
between privacy and perceived value, might not be typical. In addition,
privacy is perhaps more of a concern for first-time online customers of
the supermarket and less of a concern for returning customers (of the
type seen in our study) who have already built up trust through prior
transactions. Finally, other reasons, including those that are political or
cultural in nature, for instance, could also contribute towards explaining
why privacy is not linked with perceived value.
The present study found that the strongest relationship of the model
was that between value and loyalty. This finding agrees with Boshoff s
(2007) model in concluding that this relationship is very strong and
significant.
124
Finally, this study has detected a significant influence of loyalty on

actual purchases. This finding has important managerial implications. It
provides empirical confirmation of the final link in the chain between the
dimensions of e-quality and financial turnover, which had not previously
been investigated in earlier studies. Although the R-squared for actual
purchases was quite low, which suggests that other factors might have
to be included in a comprehensive model, the existence of a relationship
between loyalty and actual purchases has clearly been demonstrated.
Conclusions
This study of the applicability of the E-S-QUAL scale (Parasuraman
et al. 2005) to an online supermarket has confirmed that (with minor
modifications) the four dimensions of the original scale are applicable to
assessments of quality in this e-commerce setting. The findings of the study
are thus in general accordance with those of previous authors (Collier &
Bienstock 2006; Connolly 2007) who have endorsed the usefulness of this
measurement instrument.
The present study has also demonstrated important relationships
between most of the E-S-QUAL dimensions of e-service and the constructs
of perceived value, loyalty and actual purchases. In particular, the
dimension of fulfilment has been confirmed as having an important
impact on loyalty and purchasing (although the effect is mediated through
perceived value). In this regard it is interesting to note that established
models of service quality (such as SERVQUAL) have allotted a main role
to reliability (understood as the ability to perform the promised service
dependably and accurately) as a means of ensuring high standards of
service quality; indeed, Parasuraman et al. (1985, 1988) contended that
a company is unable to compete unless it is reliable, and Berry (1995)
has argued that service companies that are unable to provide services as
promised will lose customers. The dimension of fulfilment in the model
proposed in the present study is quite close to the notion of reliability;
moreover, in an online service such as that provided by a supermarket,
fulfilment (or reliability) is especially important. In this regard, it is
interesting to note that most of the complaints that were registered in the
present study (in an open box included in the questionnaire) were related
to this dimension, which demonstrates that the issue of fulfilment (or
reliability) is significant for customers of an online supermarket.
The present study also found that privacy had no effect on perceived
value (although it might be important in other respects, such as the ability
125
to compete). A similar finding was made with regard to the relationship

between efficiency and perceived value. It is acknowledged that, in
studies of other sectors, efficiency has been shown to affect perceived
value; however, in the setting of an online supermarket, no such
relationship is apparent.
Finally, the study has definitely established that e-quality has an effect
on actual purchase behaviour. However, it is acknowledged that there are
likely to be other factors (not included in the present model) that also have
an influence on sales. To our knowledge, this is the first time that a link
between e-quality and actual purchases has been reported using actual
purchase data rather than self-reported data.
Acknowledgment
This article was written as part of a research project entitled Improving
customer satisfaction through standardization in Spanish companies
(ECO2009-12754-C02-01) financed by the Ministry of Science and
Technology via funding programmed for research and development (R+D)
projects.
References
Barnes, S. (2006) E-Commerce and V-Business. Digital Enterprise in the Twenty-First Century.
Oxford: Butterworth-Heinemann.
Barnes, S. & Vidgen, R. (2000) WEBQUAL: an exploration of web site quality. Proceedings of
the 8th European Conference on Information Systems. Vienna, July, pp. 35.
Barnes, S. & Vidgen, R. (2002) An integrative approach to the assessment of e-commerce
quality. Journal of Electronic Commerce Research, 3, 3, pp. 114127.
Berry, L.L. (1995) On Great Service: A Framework for Action. New York: The Free Press.
Boshoff, C. (2007) A psychometric assessment of E-S-QUAL: a scale to measure electronic
service quality. Journal of Electronic Commerce Research, 8, 1, pp. 101114.
Byrne, B.M. (1994) Structural Equation Modeling with EQS and EQS/Windows. Basic
Concepts, Applications and Programming. Thousand Oaks, CA: Sage Publications.
Chiou, J. & Shen, C. (2006) The effects of satisfaction, opportunism, and asset specificity on
customers loyalty intention toward internet portal sites. International Journal of Service
Industry Management, 17, 1, pp. 722.
Collier, J.E. & Bienstock, C.C. (2006) Measuring service quality in e-retailing. Journal of
Service Research, 8, 3, pp. 260275.
Compeau, D.H. & Higgins, C.A. (1995) Application of social cognitive theory to training for
computer skills. Information Systems Research, 6, pp. 118143.
Connolly, R. (2007) Factors influencing Irish consumers trust in internet shopping.
Management Research News: Communication of Emergent International Management
Research, 31, 5.
126
Cristbal, E. & Marimon, F. (2007) A study of e-retailing management: analysing the

expectations and perceptions of Spanish consumers. International Journal of Consumer
Studies, 31, 6, pp. 613622.
Enzmann, M. & Schneider, M. (2005) Improving customer retention in e-commerce through a
secure and privacy-enhanced loyalty system. Information Systems Frontiers, 7, 4/5, pp.350
370.
Faul, F., Erdfelder, E., Lang, A. & Buchner, A. (2007) G*Power 3: a flexible statistical power
analysis program for the social, behavioral, and biomedical sciences. Behavior Research
Methods, 39, 2, pp. 175191.
Hair, J.F., Anderson, R.E., Tatham, R.L. & Black, W.C. (1998) Multivariate Data Analysis
(5th edn). Upper Saddle River, NJ: Prentice Hall International, Inc.
Hu, L. & Bentler, P. (1999) Cutoff criteria for fit indexes in covariance structure analysis:
conventional criteria versus new alternatives. Structural Equation Modeling, 6, 1, pp. 155.
Huang, L. (2008) Exploring the determinants of e-loyalty among travel agencies. Service
Industries Journal, 28, 2, pp. 239254.
Ingle, S. & Connolly, R. (2006) Methodological and research issues using E-S-QUAL
to measure online service quality in Irish SMEs. Irish Journal of Management, 26, 2,
pp.2532.
John, G. & Reve, T. (1982) The reliability and validity of key informant data from Dyadic
relationship in marketing channels. Journal of Marketing Research, 19, pp. 517524.
Kim, M., Kim, J. & Lennon, S. (2006) Online service attributes available on apparel retail web
sites: an E-S-QUAL approach. Managing Service Quality, 16, 1, pp. 5177.
Li, H. & Suomi, R. (2007) Evaluating electronic service quality: a transaction process based
evaluation model. Proceedings of the European Conference on Information Management
and Evaluation. Montpellier, France, September, pp. 331340.
Loiacono, E.T., Watson, R.T. & Goodhue, D.L. (2000) WEBQUAL: a website quality
instrument. Working paper 2000-126-0, University of Georgia.
Loiaco, E.T., Watson, R.T. & Goodhue, D.L. (2007) WEBQUAL: an instrument for consumer
evaluation of web sites. International Journal of Electronic Commerce, 11, 3.
Manfreda, K.L., Bosnjak, M., Berzelak, J., Haas, I. & Vehovar, V. (2008) Web surveys versus
other survey modes a meta-analysis comparing response rates. International Journal of
Market Research, 50, 1, pp. 79104.
Nunnally, J.C. & Bernstein, I.H. (1994) Psychometric Theory. New York: McGraw-Hill.
Parasuraman, A., Zeithaml, V. & Berry, L. (1985) A conceptual model of service quality and
its implications for future research. Journal of Marketing, 49, 4, pp. 4150.
Parasuraman, A., Zeithaml, V. & Berry, L. (1988) SERVQUAL: a multiple-item scale for
measuring consumer perceptions of services quality. Journal of Retailing, 64, 1, pp. 1240.
Parasuraman, A., Berry, L.L. & Zeithaml, V.A. (1991) Refinement and reassessment of the
SERVQUAL scale. Journal of Retailing, 67, 4, pp. 420450.
Parasuraman, A., Zeithaml, V.A. & Malhotra, A. (2005) E-S-Qual: a multiple-item scale for
assessing electronic service quality. Journal of Service Research, 7, 3, pp. 213233.
Ribbink, D., van Riel, A. & Streukens, S. (2004) Comfort your online customer: quality, trust
and loyalty on the internet. Managing Service Quality, 14, 6, pp. 446456.
Ringle, C.M., Wende, S. & Will, A. (2005) SmartPLS 2.0 (beta). Retrieved from http://www.
smartpls.de.
Roster, C.A., Rogers, R.D., Albaum, G. & Klein, D. (2004) A comparison of response
characteristics from web and telephone surveys. International Journal of Market Research,
46, 3, pp. 359373.
Sanzo, M.J., Santos, M.L., Vzquez, R. & lvarez, L.I. (2003) The effect of market
orientation on buyerseller relationship satisfaction. Industrial Marketing Management, 32,
4, pp. 327345.
127
Spanish Ministry of Industry, Tourism and Trade (2008) Comercio Electrnico B2C 2008.
Madrid: Red.es.
Van Riel, A.C.R., Liljander, V. & Jurrins, P. (2001) Exploring consumer evaluations of
e-services: a portal site. International Journal of Service Industry Management, 12, 4,
pp.359377.
Yang, Z. & Fang, X. (2004) Online service quality dimensions and their relationship with
satisfaction: a content analysis of customers reviews of security brokerage service.
International Journal of Service Industry Management, 15, 3, pp. 302326.
Yoo, B. & Donthu, N. (2001) Developing a scale to measure the perceived service quality
of internet shopping sites (Sitequal). Quarterly Journal of Electronic Commerce, 2, 1,
pp.3147.
About the authors

Frederic Marimon, PhD in Business Administration, is Professor at the
International University of Catalonia (Spain) and at the University of
Girona (Spain). He studied industrial engineering at the Polytechnic
University of Catalonia in Barcelona, and has a Master of Business and
Administration from IESE in Barcelona. His interests are on the area of
production, mainly in quality issues and production in service companies.
He has published several articles in national and international academic
journals.
Richard Vidgens research focuses on how technology is developed
and used within organisations. He is interested in how software systems
are constructed, e.g. development methods and agile approaches. He
also studies the interplay between people and technology using complex
systems theory, particularly co-evolutionary theory and social network
analysis. He has further interests in evaluating the quality of ecommerce
offerings in the area of web quality.
Stuart J. Barnes is Chair and Professor of Management in the Norwich
Business School at the University of East Anglia. Previously he worked
at Victoria University of Wellington, New Zealand, and the University
of Bath. Stuart has been teaching and researching in the information
systems field for over a decade. His academic background includes a firstclass degree in economics from University College London and a PhD in
Business Administration from Manchester Business School. His primary
research interests centre on the successful utilisation of new information
and communications technologies by businesses, governments and
consumers. He has published five books (one a best-seller for ButterworthHeinemann) and more than a hundred articles including those in journals
such as Communications of the ACM, the International Journal of
128
Electronic Commerce, European Journal of Marketing, Communications of

the AIS and Information & Management.
Eduard Cristbal holds a PhD in economics and is Associate Professor of
Marketing Management at the University of Lleida, Spain.
Address correspondence to: Frederic Marimon, Associate Professor in
the Economics Faculty, International University of Catalonia, Immaculada,
22, 08017 Barcelona, Spain.
Email: fmarimon@cir.uic.es
129
Conference notes
IJMR Research Methods Forum: Start listening, stop
asking, The Royal Society, London, 4 November 2009
These notes describe the nature of listening and some assumptions that lead us to
hear what we want to hear rather than the intentions of the speaker. They go on
to focus on the idea of co-creation and stress the importance of the creative component in that term. It is this element that transforms co-creation into something
unique, original and compelling. The notes conclude with a summary of some
basic principles in designing and facilitating co-creation
Co-creating the future

Roy Langmaid
DOI: 10.2501/S1470785310201090
The circumstances of early life have a

profound impact on our journey. Having been adopted just after the war, this
led me on a particularly intense search
for identity, character and belonging
over my years of being with, and learning from, others. Here are two enduring characteristics of those experiences
that relate to listening:
1. it is of immense value to understand
other people, by learning to listen
to them, and face the risk that you
might be changed by that listening
2. this opened channels through which
others can communicate their feelings, perceptual worlds, hopes and
fears.
I would summarise what I do differently as relational depth (Mearns
& Cooper 2005). This aims to create
a connection between the researcher
and participants, which goes beyond
the superficial and, in a perfectly safe

fashion, allows a glimpse of what is
going on beneath the surface as people
work with us.
Establishing relational depth requires:
establishing an empathic understanding of participants internal frames of
reference doing our best to walk in
their shoes and experience the world
in the way that they do
building a sense of unconditional
positive regard (Rogers 1967) for
participants
striving to be dependably real or
authentic.
A key element to reducing the barriers to communication is the role played
by listening. Listening has character,
just as speaking does, but we seem not
to know it so well. We are always listening from somewhere; we have a tone
of listening just as we have a tone of
voice.
Background commentary
People say they have an open mind.
Experience of my own mind tells me
131
Conference notes
that this is hardly ever true. There is

a constant background commentary
a little voice playing in my mind,
whether I like it or not, and Im often
listening to it. It is difficult to decipher
whether I am creating what this voice
is saying or not. Some facts about this
background commentary are:
it continues whether I want it or not
it seems to be me speaking to myself
but I cant tell me to shut up
in times of danger or uncertainty it
becomes louder, more insistent
I cannot stop it or shut it off; I can
only choose whether to pay attention
to it or not
often even that choice is difficult
because I dont notice its there.
This background commentary
forms the soundtrack of my life
my moment-by-moment experience
of consciousness, made up of memories, opinions, preferences, sensations,
assessments, fantasies, wishes and fears.
Everything else that happens occurs on
top of that.
Its a tape thats always on, waiting
for me to listen in (more accurately, to
fail to tune it out), which determines
my perceptual world. In other words,
it has direct and lasting effects on my
emotions, attitudes and behaviour.
Im talking about it here today using
another tool of consciousness: speech,
my foreground communication.
The job of the expert practitioner
of insightful work is to create circumstances where both foreground and
background can be fully available for
communication for either may drive
behaviour.
Lets explore this background commentary in its everyday-ness, and what
Im listening to.
132
Listening about what I think about

others, my opinion and assessments
of them, rather than what theyre
actually saying. It is absolutely commonplace to listen this way: we
dont really hear what others are
saying, we hear our interpretation of
it!
Listening for closure and confirmation give me the answer, tell me
what to do. Give me some options.
What can I do to be right, get the
right answer? Now its my turn to
speak.
Listening for safety minimise risks,
look normal, avoid embarrassment,
dont make a fuss. Joke about anything uncomfortable. Dont upset
anyone. Much speaking and listening
in focus groups appears to be of this
type.
These norms of listening were driving the agenda in groups more than
anything else. To create insights and
innovation, we needed a new way. To
do this, we needed to step around the
background commentary to find a new
way of listening. To listen differently
you need to notice the way in which
the past is always intruding into our
sense of the present and the future. We
need to put the past behind us to create
room for a new future.
This has become a key element of cocreation sessions to help overcome the
fact that most possibilities are extensions of the past. Our clients arent
paying us to add another feature to
their product. They can already figure
that out for themselves.
So opening ourselves up to a new
distinction in listening listening for
possibility was the first step towards
co-creation. This being:
a collaborative, inventive style of

work, where customers and staff work
together to create breakthroughs in
products and services. In this work we
focus on the creative element and the
engagement of the prospective mind.
Co-creation is not about problem

solving, it is about the creation of possibility that can be thought of as a
future beyond reach.
The next problem we faced in evolving co-creation was the persistence and
penetration of culture. Culture clings
to us like a veil, and shapes everything. Inside culture, words and actions
become inseparable. Thus the request
bring me some water is very likely to
produce cold drinking water in a glass,
rather than radiator water from the car.
Culture is the biggest background commentary of them all (Fox 2004).
In focus groups, people found it
difficult to put the past behind them
especially if the invitation they accepted
and the setting they found themselves
in resembled a suburban living room or
an office, places dominated by cultural
expectations the living room for small
talk, and the office by management
time. In management time it is possible
to have a meeting lasting one hour and
then stop. In creative time you cant
just switch on and off, it might be a
week that it takes for new ideas to take
shape, or a second.
The magic in co-creation does not
lie in getting a large number of people
together in a workshop, or in putting
staff and customers together except
insofar as these things interrupt the
usual way of doing things it resides
in interrupting the routine, the normal,
creating a culture where the usual does
not apply. Here we get to do ourselves
differently, leading to new possibilities.
The following six steps create a

micro-culture where possibility is nurtured and welcomed:
1. invitation
2. creating relationships
3. creating safety
4. putting the past behind us
5. creating the future
6. completion creating a lead-in to
implementation.
The invitation
Here is an example of a powerful
invitation:
Come and help us invent the future
of flying. Over two days we will work
together as a large group of airline staff
and passengers to invent 21st-century flying. How do you want it to be? What are
your unmet needs? Join us for two days of
creativity and invention at the Big Talk on
1 September 1997.
Creating relationships
If you want to create relationships
that are full of possibility, they need to
be less fettered than those our host
culture determines. Yet they must also
be safe, in the sense that people know
they will not be humiliated, mocked
or excluded especially if they take
a risk.
To accomplish this we ensure that we
create personal rather than professional
or public relationships (Langmaid &
Andrews 2003).
The important work is to create
relationships in a manner that moves
us from the professional/public to personal/private domains. It is in these
latter two that the big discoveries are
made.
133
Conference notes
Professional relationships characterised by their transactional nature

and the fact that messy feelings are
hidden.
Public relationships are on show here
now. They reflect our social face.
Popularity, belonging and acceptance
are very important here.
Personal relationships the stuff that
is important about us, personally.
Shared emotions, likes and dislikes.
Private the secrets attached to bigger emotions, love, hate, anger, grief,
sadness. Deemed unsuitable for public showing.
Creating safety
Establishing rules in speaking and listening which ensure that people are not
being put down and that everyone can
be heard.
Putting the past behind us
People telling their own stories and
then, in combining these tales and
experiences, we take a look at the
future that is waiting to happen the
future that is shaped by the past.
Creating the future
With the space we have opened up, we
create a new future, one that moves,
inspires and touches people. These
features are key; to start people believing in a possible future, it must inspire
them.
Completion
Moving towards validating and implementing the creative output. The presence of the client is all-important.
134
Facilitating creativity
The industry hype so far has been very
much about the co in co-creation, a
lot of fuss about changing the sample
and the setting (process factors), just as
you might expect from researchers. No
one hears the creation in co-creation.
Its what weve learned about fostering
creativity that has made the biggest
change in my life and work.
Here are four principles that facilitate creativity with groups of staff
and customers (Langmaid & Andrews
2003).
1. All people are creative: everyone
makes up a complete human personality every day. They tell stories,
fall in love, run away from monsters,
laugh and cry, and tell lies. Dont
look for special creative people. Cocreation is about going beyond your
present boundaries and not about
being special. We do not doubt the
existence of creative specialists; we
would rather rely on diversity and an
inspiring process to generate results
(Leadbetter 2008).
2. To increase the flow of creativity it
is more productive to work at the
level of environments and activities
(fields) than at the level of individuals (Csikszentmihalyi 1999).
3. People must have enough surplus
energy to create. If they are bogged
down in struggles for survival (including the survival of their identity, as
well as physical survival), they will
not create. In facilitation we engage
people in lots of activities that source
physical energy and levels of excitement, and expand their working
identities in an atmosphere of safety
and acknowledgement.
4. The final principle is that you need

to build the receiver Richard
Seymour (see http://www.channel4..
com/programmes/better-by-design/.
4od#2917621), inventor and des
igner, taught me that:
Creativity is essentially meaningless
unless the social value of the creation is
picked up by society. Thus in our co-creation programmes, creating a structure
for implementing the creation including enrolling the client system is of vital
importance.
To summarise:
to invent new futures you must put
the past behind you
this past is a mixture of personal history and cultural norms
there are no open minds
co-creation involves a diverse group,
but is fundamentally a process of
interrupting routines, norms and the
status quo creating a micro-culture
everyone needs to have a sense of the
primary task the possibility were
working on
there are a series of key preparatory
steps invitation, creating relation-
ships, creating safety, putting the past

behind you
these are followed by creative exercises facilitated by people who appreciate the boundaries and needs of an
environment rich in possibility
you need to enrol your sponsor/client
before, during and after the live work!
With thanks to colleagues Nicky
Forsythe and Mac Andrews.
References
Csikszentmihalyi, M. (1999) in Sternberg, R.
(ed.) Handbook of Creativity. Cambridge:
Cambridge University Press.
Fox, K. (2004) Watching the English.
Hodder & Stoughton.
Langmaid, R. & Andrews, M. (2003)
Breakthrough Zone. Wiley & Sons.
Leadbetter, C. (2008) We-Think. London:
Profile Books (see Professor Pages
experiments, p. 72).
Mearns, D. & Cooper, M. (2005) Working
at Relational Depth in Counselling and
Psychotherapy. Sage Publications.
Rogers, C. (1967) On Becoming a Person.
Constable.
Roy Langmaid is the principal of the

Langmaid Practice.
135
Conference notes
Get real: from the viewing

facility to the real world
Philly Desai
DOI: 10.2501/S1470785310201107
The uniqueness of ethnography

Ethnographic approaches are based
on the principle that cultural groups
can be understood by observing what
people do and say. Although ethnographers may ask people questions, this
is not essential and great insights can
be gained through observation alone.
Ethnographers have to go out into the
community and see what is happening
in the real world, not wait for research
participants to turn up at the viewing
facility.
But observation is not just looking.
Observations need to be focused in
order to come up with real insights.
Observations can be sharpened by
thinking about the three aspects of
context and two aspects of content.
The three aspects of context are as
follows.
1. Time: if the activity you are observing usually takes place in the morning, you need to observe it in the
morning.
2. Space: going to where people are and
observing their behaviour in situ is
essential to ethnography.
3. People: observations should be conducted in their natural social context,
with the family, friends or other people who would normally be there.
The two aspects of content are as
follows.
136
1. Cultural meanings: exploring the

meanings that behaviours, relationships, brands or products have for
people.
2. Physical artefacts: looking at the
material stuff people use to accomplish their tasks, such as mobile
phones, mp3 players, handbags and
cooking implements.
Armed with this framework, researchers are equipped to observe behaviours,
and to generate new insights into social
issues and consumer behaviour.
There are five areas in which ethnography and observation can be particularly useful, as described below.
Five areas for ethnographic insight

1. Retail navigation: observing how
people find their way around shops,
how they are influenced by lighting, floor coverings, point-of-sale
displays, other customers and interactions with staff. These are influences that are rarely mentioned in
interviews or focus groups, but that
we know have a strong effect on purchasing behaviour.
2. Product development: watching people use products in their own homes
can provide fascinating ideas for
new product development and for
improving existing products. Maybe
consumers are using products in
a way not anticipated, such as the
extraordinary growth of text messaging, or the fashion for wearing
trainers with the laces undone. Or
perhaps they are consuming them
in ways that invoke surprise, such
as microwaving Mars bars or freezing Flakes (thanks to Siamack Salari.
of Everyday Lives for these examples).

3. Lifestyles and cultures: becoming
immersed in the life of a group of
people for a few hours, a day or a
week, can be an exciting and inspiring way to understand their lifestyles.
This enables what people wear, what
music they listen to, where they go,
what media they are consuming, how
they speak, and what they eat to be
observed. This can help to generate
meaningful advertising and brand
communications, or provide insights
about how to change peoples behaviour. For example, London research
firm 2CV recently spent whole days
with families to explore the factors that led to childhood obesity.
They found that giving children
unhealthy foods was a way for
low-income mothers to express their
love for their children a powerful
insight if the client wants to change
that behaviour.
4. Urban ethnography: theres a long
tradition of researchers using observation in towns and cities to see whats
really going on in the streets, pubs,
bars and back alleys. When researching community safety, its very useful
to accompany participants and ask
them to show you around their estate
or their town. This will identify what
sorts of spaces feel safe or unsafe, how
the quality of the environment affects
feelings of security, and what sorts
of initiatives might make people feel
safer in their own towns. For example, providing better lighting, cutting
down a few bushes and telling people
when the next bus is due can make
people feel safer when waiting for
public transport.
5. Habitual actions: a lot of human

behaviour is a matter of habit rather
than conscious choice. Whether
watching TV, going shopping, shaving or brushing their teeth, people
may not be completely aware of
what they are doing. This means
people find it hard to tell researchers about their motivations. Observation, on the other hand, can reveal
the hidden rituals and routines that
guide human behaviour. For example, many people underestimate the
amount of time they spend in front
of the TV or computer, not because
they want to lie, but because they
are only half engaged in what they
are watching anyway. Observation
can show whats really going on in
the living room.
Conclusions
Even in situations where a research
project is not primarily a large ethnographic study, it can be made more
ethnographic. These are a few suggested ways to achieve this:
pilot or guerrilla ethnographies,
which involve going out and observing the activity of interest before
conducting any focus groups or
interviews
urban tours and use of photography
and video to bring research to life
accompanied activities, such as shopping, web surfing or travelling on
public transport
immersion days spent with the target
audience
observing how products are used in
home, while also conducting depth
interviews
137
Conference notes
watching at a particular location

such as a supermarket fixture and
carrying out intercept interviews
with people on site.
In conclusion, it is not difficult to
bring most research closer to the real
world through the use of ethnographic
methods, and to generate new insights
138
for clients in the process. But, to do so,

researchers need to give up the controlling model of research and adopt an
approach where respondents take the
lead and this, perhaps, is the real challenge of ethnography.
Philly Desai is Founder, Turnstone
Research
Book reviews
Grown up digital: how the net
generation is changing your
world
Don Tapscott
McGraw-Hill, 2009
368pp, hard cover 19.99
ISBN 978-0-07-150863-6
DOI: 10.2501/S1470785310201119
Another book written by Don Tapscott,

the well-known author and co-author
of Wikinomics, Paradigm Shift and the
Digital Economy. This particular book
focuses on the net generation those
people aged between 11 and 30 who
have grown up with the internet. It
has been inspired by a major research
project that the author was involved
in, which undertook interviews with
11,000 young people worldwide,
examining their behaviours in terms
of thinking, interacting, working and
socialising.
Tapscott states that although TV
was the signature medium of the baby
boomer generation, the net generation dont just watch TV, they view
it while theyre chatting with friends
and navigating the web. They dont
just communicate and collaborate with
close friends, they communicate and
collaborate globally.
The book is divided into three parts:
the first has four chapters explaining
how the net generation thinks and
behaves; the second consists of four
chapters looking at the implications
of this behaviour for education, the
workplace, buying behaviour and the

family; the third has three chapters
focusing on the impact on society in
terms of democracy, and the wider
implications for the world.
The most interesting insights for the
market researcher are found in the
chapter on consumers, or as Tapscott
entitles them the prosumers. He highlights the net generations demands
for greater choice and customisation
in their purchases. They also spend
significant time checking out retailers
and products online before purchasing
or visiting conventional stores. They
expect manufacturers and retailers
to have integrity and wont hesitate
to complain online about bad products or unethical corporate behaviour.
They seek to work with companies to
develop better products or services, so
they want companies to invite and welcome their opinions. They also require
things quickly, whether it is responses
to their opinions or the delivery of their
purchases.
Any parent who has offspring in the
11 to 30 age range will recognise many
of these traits, however the book provides a good synthesis of this behaviour and raises awareness of the likely
implications for manufacturers, retailers and service organisations. It also
raises major questions about the role of
the expert and expert advice, as members of the net generation seek more
advice from their peers rather than
salesmen, travel writers, theatre critics
or even their doctors. This raises major
questions as to how organisations can
139
Book reviews
actively influence behaviour within the

webs of individuals who scrutinise a
firms behaviour and try to change it.
Within the chapter relating to the net
generation and democracy, Tapscott
highlights that this group are less tolerant of poor public services and expect
the public sector to be organised in
ways that maximise convenience to the
citizen in terms of both access and service delivery. He talks about the need for
online citizen panels, deliberative polling, virtual question-and-answer sessions with government representatives,
and software-based scenario planning.
For governments, engagement is critical, particularly as online social media
open up a world of possibilities for
anti-government organisations and
activists to identify and communicate
quickly with a large audience.
Although many readers may already
recognise some of the trends highlighted, the book brings these together
in a way that is thought-provoking and
challenging to the way our world has
operated up until now. Tapscotts writing style makes it an easy read, even
though his seven guidelines relating to
each topic at the end of every chapter get a little tedious. In particular,
the seven parenting guidelines were
especially cringeworthy: have a family
vision; interact; customise your parenting; consciously design your family to
balance work and personal lives; collaborate in parenting; let kids be kids;
and play. That aside, I am more than
happy to recommend this book as a
valuable guide to understanding the
behaviours and likely impact of the net
generation on our world.
Alan Wilson
Strathclyde Business School
140
Wonder woman: marketing

secrets for the trillion-dollar
company
Iain Ellwood with Sheila Shekar
Palgrave Macmillan, 2008
256 pp, hard cover 25
ISBN 978-0-23020-160-6
DOI: 10.2501/S1470785310201120
Marketing academics are slowly recognising the pivotal role women play
when it comes to making, taking, leading and directing buying decisions as
various as the weekly supermarket
shop, the broadband supplier offering
best value for money and the mortgage account offering the best fixed
rate all decisions that, in the present
economic climate, are pertinent to the
purse strings of many households. In
this regard Ellwood and Shekars book
does not address a new topic: the
growing, powerful influence women
have over both household buying decisions and household incomes. What
Wonder Woman: Marketing Secrets
for the Trillion-Dollar Company does
do, however, is move discussion of
women as the critical consumers of
the 21st century beyond the pink and
frilly window dressing that so often
undermines advice on how best to
market products even those identified as male products (the growing
number of male grooming products
being an obvious example of this) to
todays growing army of female decision makers.
The books Introduction opens by
quoting, with some well-versed facts
and figures, the growing economic
power of women and the implications
of this for marketing strategies. Simply
put, Ellwood and Shekar believe that

marketing to women is currently the
most effective business growth strategy (p. 1). Addressing this, the authors
make clear that a key premise of their
book is to provide convincing evidence
of why women are the most financially
attractive target audience (p. 1) and
why marketing to them will accelerate
higher shareholder value (p. 1) both
convincing arguments for encouraging
marketing, brand and communications
managers to challenge their traditional
views of how best to communicate and
connect with female buyers. To help,
Ellwood and Shekar continue by dispelling a number of myths including:
women are not a niche market (in the
US they make more than 80% of all
purchases); women do not buy only
feminine products (they buy 61% of
major DIY products worth $70 billion),
and women are interested in sports (in
2005 they bought more than 80% of
NFL products). So far, so convincing.
However, note that these and many of
the other facts and figures presented in
this book are informed by US research.
As such, the extent to which Women
Rule (p. 1) will resonate outwith US
and other western economies is questionable.
This limitation aside, Ellwood and
Shekar use Chapters 1, 2 and 3 of their
book to document scientific (Chapter
1), psychological and behavioural
(Chapter 2), attitudinal and life stage
(Chapter 3) differences between the
sexes, and consider the implications of
these for marketing to female buyers
(girls and women). This latter point is
more fully developed in Chapters 4, 5
and 6, which consider the specific implications of marketing to Generation Y
(Chapter 4), Generation X (Chapter
5) and baby boomer women (Chapter

6), respectively. The research evidence
and examples presented across these
chapters are interesting, relevant, and
will help both marketing students and
practitioners develop a stronger, betterinformed grasp of the reasons why
female purchasers differ, often fundamentally, from their male counterparts.
The implications of this for rethinking
marketing, particularly marketing communications (Chapter 7) and branding
strategies (Chapter 8), are then laid
out and brought to life with a variety
of examples and illustrations not all
of which are taken from the US. The
concluding chapter focuses on what
Ellwood and Shekar refer to as touchpoint improvement how marketers
can use a womans talent for making
friends as the foundation for building
potent branded customer relationships
(p. 192). This chapter discusses the
implications that womens life cycle
of relationships (p. 192) have for
using branding strategies and techniques to develop durable relationships
with female consumers. In many ways
the chapter recommends that marketers develop branding and communication strategies that resonate with the
experiences of women by, for example, contextualising their life within
communications. A recent example of
this might be the Dove Campaign for
Real Beauty, which includes using real
models (not the air-brushed type) in
its marketing communications. In this
way, while much of this final chapter
makes intuitive sense, it is limited in
two ways: first, it is not really providing new insights into how best to
connect with female consumers; second, like much that is written about
women within and outside of the
141
Book reviews
field of marketing the chapter reverts

to regarding women as a collective,
homogeneous group, which we know
is simply not the case.
To conclude, while Wonder Woman:
Marketing Secrets for the Trillion-Dollar
Company has its limitations, not least
its reliance on US data and research
and its failure to extend the early
research chapters to also include work
that has used feminist theories to consider womens roles within society and
question the extent to which women
142
truly do rule, it does go beyond the

pink packaging often associated with
marketing to women. Given the strong
arguments presented in the book and
the discussion of women as a vital
target market, as well as the advice
on how to fully embrace girl power,
it is surprising perhaps that the book
is written by Iain Ellwood with Sheila
Shekar.
Eleanor Shaw
Strathclyde Business School

International Journal of Market Research

Загружено:

Сведения о документе

Авторское право

Доступные форматы

Поделиться этим документом

Поделиться или встроить документ

Параметры публикации

Этот документ был вам полезен?

Это неприемлемый материал?

Авторское право:

Доступные форматы

International Journal of Market Research

Загружено:

Авторское право:

Доступные форматы

International Journal

Whose design is it anyway? Priming designer and shifting preferences . . . . . . . . . . . . . . . . . . . . . . . 89

International Journal of Market Research Vol. 52 Issue 1

IJMR Research Methods Forum:

interpreting their meaning in the

Some developing fields, such as

to develop new methods and

International Journal of Market Research Vol. 52 Issue 1

key message above from the Research

Calls for papers

and measurement methods, but will

2009 award winners

sumer preferences, and a model measuring service quality in e-commerce.

Influences on political participation

International Journal of Market Research Vol. 52 Issue 1

tutions themselves are important, the

Defining and measuring brand equity

no widely accepted definition of brand

Motivations for interacting with digital

net. However, the authors contend

Does designer labelling influence

(Copenhagen Business School), explores

Measuring the online shopping

International Journal of Market Research Vol. 52 Issue 1

ment leads organisations to actively

level by the supermarket) rather than

Access to all medal-winning MRS papers

All articles from the previous 12 months

Searchable abstracts of all IJMR articles since 1990.

Market research abstracts

View developments in research from around the world. This includes

Full members of the Market Research Society can access a

International Journal of Market Research Vol. 52 Issue 1

The growing attention paid by procurement professionals to the buying

2010 The Market Research Society

its influence. It is surely far better to

is that justifies your place on a company

value our people. By that I mean the

International Journal of Market Research Vol. 52 Issue 1

Imperial College London

Received (in revised form): 29 June 2009

2010 The Market Research Society

Forum: Perspectives on data mining

implemented, or encouraged, data dredging repeated searches for

International Journal of Market Research Vol. 52 Issue 1

Forum: Perspectives on data mining

Data mining: global models

Data mining: local patterns

International Journal of Market Research Vol. 52 Issue 1

relationships are generically patterns and we can contrast detection for

Forum: Perspectives on data mining

accounts (Adams et al. 2001), detecting cheating in student continuous

The data mining process

International Journal of Market Research Vol. 52 Issue 1

Forum: Perspectives on data mining

International Journal of Market Research Vol. 52 Issue 1

About the author

The pre-eminent source for

Les Carlson, Clemson University

Gain vital insight

International Journal of Market Research Vol. 52 Issue 1

Do institutions really influence political

University of Texas at Dallas

University of Texas at Dallas

p < 0.10; p < 0.05; p < 0.01.

Aaker (1991, 1996) brand awareness