Вы находитесь на странице: 1из 21

This article was downloaded by: [Institute of Professional Studies]

On: 12 March 2010


Access details: Access Details: [subscription number 918851505]
Publisher Routledge
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-
41 Mortimer Street, London W1T 3JH, UK

Journal of Marketing Communications


Publication details, including instructions for authors and subscription information:
http://www.informaworld.com/smpp/title~content=t713704530

Brand equity: extending brand awareness and liking with Signal Detection
Theory
Gewei Ye a; W. Fred Van Raaij b
a
Department of Marketing and e-Business, Towson University, Baltimore, MD 21252-0001, USA b
Tilburg University, 5000 LE Tilburg, The Netherlands

To cite this Article Ye, Gewei and Van Raaij, W. Fred(2004) 'Brand equity: extending brand awareness and liking with
Signal Detection Theory', Journal of Marketing Communications, 10: 2, 95 — 114
To link to this Article: DOI: 10.1080/13527260410001693794
URL: http://dx.doi.org/10.1080/13527260410001693794

PLEASE SCROLL DOWN FOR ARTICLE

Full terms and conditions of use: http://www.informaworld.com/terms-and-conditions-of-access.pdf

This article may be used for research, teaching and private study purposes. Any substantial or
systematic reproduction, re-distribution, re-selling, loan or sub-licensing, systematic supply or
distribution in any form to anyone is expressly forbidden.

The publisher does not give any warranty express or implied or make any representation that the contents
will be complete or accurate or up to date. The accuracy of any instructions, formulae and drug doses
should be independently verified with primary sources. The publisher shall not be liable for any loss,
actions, claims, proceedings, demand or costs or damages whatsoever or howsoever caused arising directly
or indirectly in connection with or arising out of the use of this material.
JOURNAL OF MARKETING COMMUNICATIONS 10 95–114 (June 2004)

Brand equity: extending brand awareness and


liking with Signal Detection Theory
GEWEI YE
Towson University, Department of Marketing and e-Business, 800 York Road, Baltimore, MD
21252-0001, USA

W. FRED VAN RAAIJ


Downloaded By: [Institute of Professional Studies] At: 14:53 12 March 2010

Tilburg University, PO Box 90153, 5000 LE Tilburg, The Netherlands

Brand equity, which is a central topic in modern marketing, may be assessed from
three perspectives: customer mind set, product market outcomes and financial market
outcomes. Brand awareness (memory) and brand liking are elements of customer
mind set brand equity. The factors determining brand awareness and likeability are
also determinants of the change in financial brand equity. In order to understand
these factors, Signal Detection Theory is employed for finding the components of
brand awareness and likeability. Signal Detection Theory has a strong tradition in
psychology, but is under-represented in marketing and consumer behaviour. This
study extended the concept of brand awareness to ‘awareness sensitivity and bias’ and
the concept of ‘brand likeability’ to ‘liking sensitivity and bias’ using Signal Detection
Theory. The effect of divided attention on the extended components was investigated
in three laboratory experiments. It was found that, in the attended mode compared
with the unattended mode, consumers perform better in preserving a favourable brand
awareness and have a conservative reaction tendency. This effect of attention occurs in
building brand awareness for short presentations, but not for long presentations. These
findings may serve as guidelines for a strategy formulation for enhancing customer
mind set brand equity.

KEYWORDS: Brand equity; brand awareness; brand liking; Signal Detection Theory;
attention; response bias

INTRODUCTION

When shoppers say they don’t like Coca-Cola, should the negative brand liking be attributed to the
failure of branding campaigns, or to the conservative tendency of consumers to say: ‘I don’t like the
brand’? Further, if they say that they are conservative in responding, do they really report the objec-
tive tendency? These questions cannot be answered with conventional measures, but with Signal
Detection Theory.
A strong brand is a very valuable asset of a firm (Aaker, 1991, 1996; Keller, 1998; Aaker and
Jacobson, 2001). According to one estimate the brand value of Coca-Cola is $84 billion, that of
Journal of Marketing Communications ISSN 1352–7266 print/ISSN 1466–4445 online
© 2004 Taylor & Francis Ltd http://www.tandf.co.uk/journals
DOI: 10.1080/13527260410001693794
96 YE AND VAN RAAIJ

Microsoft is $57 billion and that of IBM is $44 billion (Morris, 1999). In order to build a strong
brand a comprehensive understanding of brand-related consumer behaviour will improve
marketing productivity (Keller, 1993). Hence, a deep understanding of the components of brand
equity from a customer-based perspective is vital to the success of brand management.
A variety of definitions of brand equity are offered in the literature (Leuthesser, 1988; Keller,
1993, 1998; Aaker, 1996). For example, Aaker (1996) defined brand equity as a set of four cate-
gories of brand assets linked to a brand’s name or symbol that add to (or subtract from) the value
provided by a product or service to a firm and/or to that firm’s customers. These four categories
are (1) brand awareness, (2) perceived quality, (3) brand associations and (4) brand loyalty
(Keller, 2002). Aaker (1996) adopts a definition under Keller’s (1998) framework from a
consumer-based and operational perspective.
Brand equity may be assessed from three aspects: customer mind set, product market out-
comes and financial market outcomes (Ailawadi et al., 2003). The measures of customer mind set
include the awareness, attitude, association, attachment and loyalty that customers have towards
Downloaded By: [Institute of Professional Studies] At: 14:53 12 March 2010

a brand (Aaker, 1991; Keller, 1993, 2003). The product market measures reflect the brand’s
performance in the marketplace. For example, price premium is such a measure that captures the
ability of a brand for charging a higher price than an unbranded equivalent (Aaker, 1991). The
financial market measures assess the value of a brand as a financial asset, including the purchase
price at the time a brand is sold or acquired and the discounted cash flow valuation of licensing
fees and royalties (Ailawadi et al., 2003).
Customer mind set brand equity is the source of the associated financial brand equity (Keller,
2003). Keller (1993) defined customer-based brand equity as the differential effect of brand
knowledge on the consumer response to the marketing of the brand. He also defined brand
knowledge in terms of two core components, brand awareness and brand image. Brand aware-
ness relates to the brand recall and recognition performance by consumers. Brand image refers to
the set of associations linked to the brand that consumers hold in their memory. The
favourability, strength and uniqueness of brand associations (image) are the factors in brand
knowledge that determine the differential response that makes up brand equity (Keller, 1993,
1998, 2003).
Brand awareness and brand image can be operationalized for the current study. This paper
uses ‘brand recognition memory’ for indicating brand awareness and a measure of likeability for
indicating the brand image that represents the favourability and strength of brand associations.
With favourable brand associations consumers tend to like the brand. In the current study,
customer-based brand equity can be operationally defined as a combination of brand awareness,
indicated by recognition memory and brand likeability.
Due to the emergent interest in branding, different kinds of information are linked to a brand,
including awareness, attributes, benefits, images, thoughts, feelings, attitudes and experiences
(Keller, 2003). However, brand awareness and likeability are the core components at the heart
of the various brand equity models (Aaker, 1991, 1996; Keller, 1993, 1998). The current study
starts with the two core components of brand equity using Signal Detection Theory. Other
components of brand equity may be investigated in a similar way in future research.
Much attention has been devoted to understanding recognition memory with Signal Detec-
tion Theory in psychology (Macmillan and Creelman, 1991; Ratcliff et al., 1994; Yonelinas,
1994, 2002; DeCarlo, 2002). Signal Detection Theory provides an unbiased strength (sensitivity)
measure of recognition memory and a cut-off point measure for indicating people’s tendency
to retrieve information from memory. However, little research has been conducted on the
recognition memory of elements of brand equity using Signal Detection Theory. To the best of
SIGNAL DETECTION THEORY 97

the authors’ knowledge, no research has been conducted on brand likeability, which is a vital
element of brand equity, with Signal Detection Theory.
The advantages of using Signal Detection Theory have been well documented in psychology
in the sense that the sensitivity and bias of responses can be measured while conventional
percentage measures just provide an overall indicator (Green and Swets, 1966; Macmillan and
Creelman, 1991; Yonelinas, 1994).
Percentages of correctly retrieved (preferred) items among all presented (marketed) items are
the conventional measures of brand awareness (recognition) and brand likeability. The disadvan-
tages of the conventional measures are that (1) they are overall indicators that do not differen-
tiate between the strength and tendency of consumer reactions to brands and (2) different
cutting scores (cut-off points) may lead to different results. Thus, the conventional measures can
be biased by the (liberal or conservative) response tendency of individuals (Green and Swets,
1966; Snodgrass and Corwin, 1988). Misinterpretations of results may occur using conventional
measures if the bias of consumer response tendencies is not considered. Hence, a deeper under-
Downloaded By: [Institute of Professional Studies] At: 14:53 12 March 2010

standing of the underlying components of brand awareness and likeability may be reached with
Signal Detection Theory than with a percentages method.
In marketing practice using the conventional measures may lead to inaccurate conclusions.
On which aspect (strength of consumer memory or feeling or response tendency) should
research be focused? Lacking a comprehensive understanding of brand awareness and likeability,
marketing programmes may not be focused on the real issue. For example, conventional mea-
sures only tell marketers that the brand likeability may be weak, if a survey shows a poor score
of consumers’ favourability judgement of the brand associations. In which aspects (sensitivity or
bias) is the brand weak? Are consumers unhappy about the brand (product) itself? Or is it their
conservative response tendency (e.g. to think twice) to say that they ‘like’ the brand? These
questions cannot be answered in a straightforward manner with conventional measures.
Brand awareness and likeability are vital and central to brand equity. In order to fit with psy-
chological theory better and to advance the understanding of consumer-based brand equity, this
paper extends brand awareness and likeability to four components: the sensitivity and bias of
brand awareness and the sensitivity and bias of brand likeability. A divided-attention procedure
is used for investigating the effect of attention on the extended components of brand awareness
and likeability. Principles of branding based on a deeper understanding of brand equity are then
discussed.

THEORETICAL FRAMEWORK
This section introduces Signal Detection Theory. It then conceptually extends Keller’s (1993,
1998) framework of brand awareness and likeability to four core components based on Signal
Detection Theory. These components are the sensitivity and bias of brand awareness and the
sensitivity and bias of brand likeability, as illustrated in Table 1. Two theoretical models are

TABLE 1. The four components of brand equity

Sensitivity (A) (strength) Response bias (B)

Brand awareness (recognition) Awareness sensitivity Awareness bias


Brand likeability Likeability sensitivity Likeability bias
98 YE AND VAN RAAIJ

described along with the four components. One is the global memory model (Gillund and
Shiffrin, 1984; Ratcliff et al., 1994) for brand awareness (recognition memory). The other is the
proposed Signal Detection Theory brand likeability model.

Signal Detection Theory


Signal Detection Theory evolved from the development of communications and radar equipment
in the first half of the twentieth century. It moved to psychology (Green and Swets, 1966),
initially as part of sensation and perception, in the 1950s and 1960s as an attempt to understand
some of the features of human behaviour when detecting very faint stimuli (signals) among other
stimuli (noise) that were not being explained by traditional theories of thresholds (Snodgrass
and Corwin, 1988; Yonelinas, 1994, 2002; DeCarlo, 2002). Recently, Signal Detection Theory
migrated to marketing and consumer research (Tashchian et al., 1988; Cradit et al., 1994).
However, besides the methodology introduction, little research with Signal Detection Theory in
Downloaded By: [Institute of Professional Studies] At: 14:53 12 March 2010

marketing and consumer behaviour has been reported.


A typical Signal Detection Theory paradigm usually encompasses a detection task. With the
detection task, a mix of stimuli contains signals and noise. Signals are stimuli to be detected,
while noise is background, which is not to be detected. For example, in a recognition task
signals are exposed items in a study phase, whereas noise consists of new items never exposed
to before. The detection task of a test phase is to find the signals in a mix of signals and noise.
When subjects respond to a signal as a ‘signal’ it is called a ‘hit’, otherwise it is a ‘miss’. When
subjects respond to noise (distracters or signal not present) as a ‘signal’ it is called a ‘false alarm’,
otherwise it is a ‘correct rejection’ (Fig. 1).
The sensitivity and bias of an operator’s reactions are measured in signal detection. Sensitivity
is specified by the signal-to-noise ratio. More specifically, the sensitivity of strength is the
standardized mean difference of the strength between the signal and noise distributions (Fig. 2).
If the sensitivity is high (i.e. the signal is strong relative to the noise), the operator is effective in
detecting the presence of signals.
On the other hand, the operator’s response tendency (bias) is distinct from its sensitivity. It
specifies how certain one must be before one is willing to say ‘yes, a signal is present’. In other

FIGURE 1. The 2 × 2 matrix of Signal Detection Theory.


SIGNAL DETECTION THEORY 99
Downloaded By: [Institute of Professional Studies] At: 14:53 12 March 2010

FIGURE 2. Signal Detection Theory. Xc and Xc' are the cut-off points of awareness/liking
strength separating the reported positivity and negativity on the continuum of strength.
Non-parametric Signal Detection Theory was used in the current study. According to
Snodgrass and Corwin (1988), normal distributions are not required in this case, although
this figure uses normal-like distributions for display. The actual distribution of the data may
not necessarily be normal.

words, the bias is the cut-off criterion for reporting a signal (e.g. the response bias in Fig. 2).
The response bias can change while the sensitivity remains the same. For example, a radiologist
may decide to accept weaker indications of abnormality of X-rays in order to refer patients to
treatment.

Brand awareness
Using Signal Detection Theory for measuring recognition memory is well established in psycho-
logy (Snodgrass and Corwin, 1988; Yonelinas, 1994, 2002; DeCarlo, 2002). Recognition
memory is indicated with two measures in experimental psychology: the strength difference
(sensitivity) of the standardized means of the signal and noise distributions and the response
tendency (bias) for detecting a signal.
In branding, the widely accepted conceptualization of a memory structure includes brand/
node associations and spreading activation (Collins and Loftus, 1975; Wyer and Srull, 1986;
Keller, 1993). For example, the associative network memory model views brand knowledge as
consisting of a set of nodes and links. Nodes are stored information such as brands connected by
links that vary in strength. A spreading activation process from node to node determines the
extent of retrieval in memory such as free recall and recognition.
More specifically, the global memory model has been successfully applied to various experi-
mental settings for recognition memory (Gillund and Shiffrin, 1984; Yonelinas, 2002). The
model assumes that a test item presented for recognition will be compared to all items
in the memory in order to determine the degree of match (familiarity) between the test item
and items in the memory. The familiarity value in turn determines the old–new (presented or
100 YE AND VAN RAAIJ

non-presented) judgement. The higher the familiarity of the item, the more likely an ‘old’
response will occur (Ratcliff et al., 1994).
Brand awareness can be assessed with brand recognition memory, which in turn is categorized
into recognition sensitivity and recognition bias. Hence, we can further categorize brand
awareness into awareness sensitivity and awareness bias.
The strength (sensitivity) of brand awareness refers to the standardized means difference
between the signal and noise distributions. In other words, the sensitivity is an objective (unbias-
ed) measure of consumer ability for detecting ‘seen before’ brands from ‘unseen’ brands. It is not
biased by consumer motivation or expectancy to respond. On the other hand, the bias of brand
awareness refers to the consumer tendency to respond to a mix of seen before and unseen
brands. When the bias is large, consumers are more conservative (e.g. think twice) about report-
ing a signal, whereas a small bias indicates that consumers are liberal with responding. As
mentioned, this categorization of brand awareness is based on recognition memory. A similar
categorization may be implemented for recall memory in future research.
Downloaded By: [Institute of Professional Studies] At: 14:53 12 March 2010

The Signal Detection Theory brand likeability model


The use of Signal Detection Theory for modelling brand likeability is not yet documented in
marketing and consumer research. Therefore, this paper first conceptualizes signals and noise of
a liking task on brand names as part of a newly proposed Signal Detection Theory likeability
model, which is followed in the next section by defining the extended two components of
likeability.
The strength of brand liking in the Signal Detection Theory likeability model is assumed to
be in a continuum of strength of favourability of brand nodes, analogous to the continuum of
strengths of familiarity of brands for recognition memory in the global memory model (Gillund
and Shiffrin, 1984; Ratcliff et al., 1994). This assumption is quantitatively valid in the sense that
the liking response data to various brand nodes can be continuous, regardless of whether one or
two factors or processes are underlying positive and negative liking.
An experimental setting has been designed that simulates marketing presentations and testing.
After being exposed to pairs of brand names, subjects are asked whether they like the presented
and non-presented brands, suggesting that subjects are motivated (signalled) to look for brands
with a positive valence. In general, presented brands should be more likely to be liked than non-
presented brands because of the mere-exposure effect, which suggests that presented stimuli are
preferred over non-presented stimuli, even when the presented stimuli are not recognizable
(Zajonc, 1968; Bornstein, 1989). Thus, liking for presented (attended and unattended) brands
constitutes the signal distribution, whereas liking for non-presented brands constitutes the noise
distribution. Hence, a non-presented brand distribution and a presented brand distribution may
be constructed, analogous to the noise and signal-plus-noise distributions of Fig. 2.
The following experiments are designed so that subjects make a yes/no choice for likeability
of the stimuli (brands). Thus, they are motivated to identify positively valenced stimuli. The
reason underlying this is the mere-exposure effect (Zajonc, 1968; Zajonc and Markus, 1982;
Nordhielm, 2002). Hence, presented brands are supposed to evoke a positive liking, whereas
non-presented brands evoke a neutral or baseline liking.
The overlapping area (the noise and signal-plus-noise distributions in Fig. 2) means that the
liking of presented and non-presented brands may be overlapping. In other words, the area
indicates that it is possible for a presented and non-presented brand to evoke the same level
of liking.
SIGNAL DETECTION THEORY 101

Extending brand likeability with Signal Detection Theory


Drawing on the Signal Detection Theory model, the likeability of brands is split into two new
components: the sensitivity (strength) and bias of likeability. The sensitivity may be the standard-
ized means difference between the strength of likeability of the signal and noise distributions. It
is an objective (unbiased) measure of the distance between the positive and neutral favourability
distributions. Internal factors such as the consumers’ ability for detecting or situational factors
such as vividness and baseline familiarity of brands (e.g. famous brands can be easily detected
from non-famous brands) may affect the strength of likeability (Loewenstein et al., 2001).
In other words, the sensitivity for brand liking is the strength difference between the two
distributions of neutral (noise) and positive likeability (signal).
However, another component of brand likeability, likeability bias, has received little attention
in consumer research. It is the cut-off criterion for concluding (identifying) a positive liking of
the brand. For example, after consuming some brands of wine, consumers are asked to say
Downloaded By: [Institute of Professional Studies] At: 14:53 12 March 2010

whether they like the brands. The response bias for reporting positively (saying yes) is the
likeability bias. If the bias (criterion) is conservative, consumers think twice before reacting. If a
liberal criterion is adopted, it is easy (e.g. think less) to respond.

Summary
This paper has extended ‘brand awareness’ to awareness sensitivity (strength difference) and
awareness bias and ‘brand likeability’ to likeability sensitivity (strength) and likeability bias based
on the two measures of Signal Detection Theory (see Table 1). The two underlying models are
the global memory model on brand awareness (recognition) and the proposed Signal Detection
Theory likeability model.

HYPOTHESES
Based on Signal Detection Theory and the core components of customer-based brand equity,
brand awareness and brand likeability, this paper extends the two core components to four com-
ponents. This section introduces a divided-attention procedure for investigating the effect of
attention on the four components and deriving hypotheses.
In a divided-attention procedure that simulates marketing, subjects participate in two phases
of the procedure: a study phase and a test phase. Subjects first study pairs of brand names pre-
sented on a computer screen, with a pair of two brands at a time. One of the two brands is
indicated with colour or an asterisk below. The subjects are instructed to pay attention to the
coloured or pointed brand. The other brand of the pair is ignored (unattended). In the test phase
the subjects make recognition judgements on their memory of the brand and liking judgements
on their favourability of the brand. In the meantime, the differences in recognition and liking
between the attended (or unattended) and non-presented brands are measured.
The depth of processing consumers engage in during encoding (study phase) may be indicated
with various controls, such as repetition (Zajonc, 1968; Bornstein, 1989; Nordhielm, 2002), self-
related generation (Craik and Tulving, 1975), stimulus presentation duration (Bornstein, 1989)
and level of attention (Ye and Van Raaij, 1997). Here the paper defines two modes of brand
information processing: deep and shallow information processing. When consumers attend to
brands or are exposed long enough to the brand, a deep processing model is employed, whereas
a shallow processing mode is employed when consumers do not attend or are exposed only
102 YE AND VAN RAAIJ

briefly to the brand. According to the global memory model (Gillund and Shiffrin, 1984;
Ratcliff et al., 1994), in the deep processing mode the strength of familiarity of the brand node
in consumers’ memory is strong, so the match between the brand as a test item and the brand
node memory is facilitated. It is more likely that an ‘old’ response is given to the presented
(marketed) brands with higher familiarity. Therefore, subjects in the deep (attended) mode may
perform better in unbiased retrieving of the encoded brand information (brand awareness) than
in the shallow mode.
In addition to the global memory model, empirical data on divided attention in psychological
literature have found it to disrupt recognition memory. It has been demonstrated that dividing
attention disrupted the recognition (explicit) memory of studied words in experiments (Shapiro
and Krishnan, 2001; Wallace et al., 2001). Other supportive evidence is that an attended picture
was recognized more frequently than an unattended one (Goldstein and Fink, 1981). Although
the result is obtained with conventional measures (hit rates in the Signal Detection Theory) of
recognition memory, there is no indication to show that it is biased by respondents’ internal
Downloaded By: [Institute of Professional Studies] At: 14:53 12 March 2010

response tendency.
In the literature related to brand name recall, it has been found that a brand name explicitly
conveying a product benefit leads to a higher recall of an advertised benefit claim consistent in
meaning with the brand name compared with a non-suggestive brand name (Keller et al., 1998).
A suggestive brand name may invoke more associations for the brand nodes in the memory
according to the associative network memory model (Keller, 1993) and the global memory
model, thus leading to deep processing, which may be similar to the effect of attending to the
brand name.
To summarize, the memory models and empirical data in psychology and consumer research
predict that subjects perform better on brand awareness measures in the deep mode than in the
shallow mode. Hence, this paper proposes the following hypothesis.
H1: There is an effect of attention on brand recognition sensitivity. The more attended, the more
sensitive the operator’s recognition memory is between seen before and unseen brand names.
There is also an effect of attention on brand recognition bias. When subjects are attending to
items (words or Chinese characters) they tend to adopt a conservative criterion for reporting
seen before items (Yang and Ye, 1995). In addition, the sensitivity and bias measures in the
recognition memory literature, particularly with the non-parametric Signal Detection Theory
that is used in the current study, seem to be congruent most of the time (Snodgrass and Corwin,
1988). Therefore, this paper proposes the second hypothesis.
H2: There is an effect of attention on brand recognition bias. The more attended, the more
conservative the response criterion will be.
The effect of attention may occur for the newly constructed brand liking sensitivity based on
the Signal Detection Theory likeability model. In the continuum of favourability of brand
nodes, a deep mode of brand processing (e.g. repetition) is associated with a more favourable
strength of liking, according to the mere-exposure effect (Zajonc, 1968; Bornstein, 1989). In
other words, the signal (marketed brands) distribution is shifted further away from the noise
(non-presented brands) distribution (Fig. 2). Thus, the mean difference between the two distri-
butions (strength sensitivity of brand liking) is larger.
To elaborate, this theoretical conceptualization is supported with empirical data on the mere-
exposure effect suggesting that, the more repetition of a subliminally exposed stimuli, the more
the stimuli will be liked (Zajonc, 1968; Zajonc and Markus, 1982). Although the conventional
SIGNAL DETECTION THEORY 103

measure (percentage of correct identification of preferred stimuli) is used for finding the result,
it is believed that the result is not biased by respondents’ internal response tendency. Hence, this
paper proposes the first of the second set of hypotheses.
H3: There is an effect of attention on brand liking sensitivity. The more attended, the more liked the
brands will be.
Since the likeability bias is a new construct in the Signal Detection Theory model, little is
known about it. Thus, a similar effect of attention on the liking bias, analogous to the effect on
the bias of recognition memory, is expected. In this case, it is assumed that the level of brand
processing (attention) at the encoding (marketing) phase influences the internal response
tendency of consumers to the brand.
H4: There is an effect of attention on the brand liking bias. The more attended, the more
conservative the liking bias will be.
Downloaded By: [Institute of Professional Studies] At: 14:53 12 March 2010

GENERAL METHOD
Three non-parametric Signal Detection Theory experiments for testing the four hypotheses on
the effect of attention on the four new components are reported. This section first introduces
non-parametric Signal Detection Theory followed by the generic method used in all three
experiments.

Non-parametric Signal Detection Theory


Tashchian et al. (1988) and Cradit et al. (1994) first introduced Signal Detection Theory to
marketing and consumer research. In this study’s brand name experiments the liking bias indi-
cator is the response bias index (Bⴖ ) of a non-parametric Signal Detection Theory (Grier, 1971;
Snodgrass and Corwin, 1988).
Non-parametric Signal Detection Theory analysis is used here rather than parametric Signal
Detection Theory because a normal distribution is not required for the data (Snodgrass and
Corwin, 1988). Therefore, with a liberal restriction on the data distribution, non-parametric
signal detection analysis can calculate the sensitivity (A) and response (liking and recognition)
bias (B). The response bias and the sensitivity are computed with non-parametric formulas.
(For H ³ ≥ FA, Aⴕ = 0.5 + [(H – FA)(1 + H – FA)]/[4H(1 – FA)] and for H < FA, Aⴕ = 0.5 –
[(FA – H)(1 + FA – H)]/[4FA(1 – H)]. For H ≥ FA, B ⴖ = [H(1 – H) – FA(1 – FA)]/[H(1 – H)
+ FA(1 – FA)] and for H < FA, Bⴖ = [FA(1 – FA) – H(1 – H)]/[FA(1 – FA) + H(1 – H)].
These are non-parametric formulas. In parametric Signal Detection Theory the sensitivity d is
the separation (mean difference) of the two distributions over the variance.) The hit rate (H) is
defined as the conditional probability of responding yes to a signal (e.g. an attended or unat-
tended ‘old’ brand) in recognition and likeability judgement tasks. The false alarm rate (FA) is
defined as the conditional probability of responding yes to a noise (e.g. a new brand).
The response bias Bⴖ represents an individual’s response bias when they say yes to a brand
name in a recognition or liking judgement task of the test phase. A high Bⴖ indicates a more
conservative decision criterion, whereas a low B ⴖ represents a more liberal decision criterion
(Snodgrass and Corwin, 1988). As is known, B ⴖ is affected by an individual’s motivation
to respond. A high motivation for doing a good job (more hits) is accompanied by a liberal
decision criterion B ⴖ. A low motivation is accompanied by a conservative decision criterion.
104 YE AND VAN RAAIJ

Stimulus materials
Subjects participated in four tests (tests A, B, C and D) after the brands were presented for 0.5,
1 or 2 s in the three experiments, respectively. Test A was a recognition judgement for the
attended brand names, test B was a liking judgement for the attended brand names, test C was
a recognition judgement for the unattended brand names and test D was a liking judgement for
the unattended brand names. Each subject participated in all four tests.
Pilot studies were carried out in order to select randomly a list of 80 novel or unfamiliar and
moderately attractive brand names from a book entitled Levensmiddelen Almanak ’93, which con-
tains the brand names of the fast-moving consumer goods sold in supermarkets in The
Netherlands. First, 101 brand names were randomly selected from this book according to alpha-
betical order. Twenty undergraduate students were asked to rate the familiarity and attractive-
ness of each brand name on a four-point scale. With regard to familiarity, the top 16 most
familiar brand names were excluded from the list, because over half of the students rated these
Downloaded By: [Institute of Professional Studies] At: 14:53 12 March 2010

as familiar brand names. Regarding the attractiveness rating, five extremely attractive and un-
attractive brand names were also excluded from the list (85 brand names) in order to eliminate
the interference of pre-experimental attractiveness towards brand names. This interference was
further eliminated using the sensitivity index of the non-parametric Signal Detection Theory
in later analysis (Snodgrass and Corwin, 1988). Thus, a list of 80 unfamiliar and moderately
attractive brand names was obtained.
To summarize, the homogeneity of baseline attractiveness and familiarity was obtained in the
pilot study so that prior brand attractiveness and familiarity could be controlled for. The prior
brand attractiveness might have been the subjects’ individual differences. Although it may be
controlled as a covariate in later analysis, it is thought that a clean homogeneous baseline is more
desirable for investigating the two experimental controls: attention and duration.
The moderately attractive and familiar baselines shift only the means of the noise distributions
(non-presented brands), as compared with a non-attractive baseline. The effects of attention and
duration are not affected by the choice of the baseline, because the ‘signal’ distribution is
assumed to be a signal-plus-noise distribution (Fig. 2).
The subjects were exposed to brand names, first in a study phase and then in four test phases.
Twenty pairs of brand names for the study phase were randomly selected from the list of 80
brand names. One brand name of each pair of brand names, printed in red, was presented in the
centre of the monitor screen and the other brand name, printed in blue, was presented to the
left of the centre of the screen. The subjects were instructed to memorize the red brand names.
Thus, the red brand names were attended to and the blue brand names were not attended to.
All brand names presented in the tests were in black, which is consistent with neither the red
nor the blue brand names in the study phase. The context effect of colour was thus controlled
for (Tulving and Thompson, 1973; Murnane and Mathews, 1995). Half of the attended brand
names were randomly assigned to the right side of the screen and half were on the left side of
the screen in order to eliminate spatial interference.
Each type of test (tests A–D) consisted of 20 brand names with ten ‘old’ (prior exposure)
brand names randomly selected from the studied set and ten new brand names randomly selected
from the remainder of the list of 80 brand names. With regard to the attended condition, half
of the 20 test brand names were selected randomly from the 20 attented (‘red’) brand names
presented in the study phase. The other half was selected from the rest of the list, to which the
subjects were not exposed during the study phase. Regarding the non-attended condition, half
of the 20 test brand names were randomly selected from the 20 non-attended (‘blue’) brand
SIGNAL DETECTION THEORY 105

names of the study phase and the other half were new brand names from the list that the subjects
did not contact in the study phase.

EXPERIMENT 1
Subjects and design
Sixteen students of a management school were recruited. In experiment 1 the brands were pre-
sented for 0.5 s in the study phase. The experimental design was a 2 × 2 × 2 (attended versus
non-attended × test: recognition versus liking × measure: sensitivity versus bias) mixed factorial
design with attention, test and measure as within-subject factors.

Procedure
Downloaded By: [Institute of Professional Studies] At: 14:53 12 March 2010

Terminals with colour monitors were used for presenting the stimuli. One to five subjects
attended the experiments at the same time and their responses were transferred from the termi-
nals to the central computer after the experiment. The central computer controlled the starting
time of the study phase and each test (tests A–D) for all terminals. The terminals were set in a
large laboratory room and were separated by facing opposite directions. Thus, the interference
between subjects was controlled for and minimized. The four tests lasted 10–15 min for each
participant.
In the study phase of the experiment 20 pairs of brand names, one printed in red (attended
stimulus) and the other in blue (unattended stimulus) for each pair, were presented for 0.5 s.
The subjects were instructed to memorize the brand names printed in red.
In the test phase of the experiment the subjects attended to four tests following a Latin-square
design sequence. Twenty brand names were presented in each test. Each brand name was pre-
sented for 1000 ms. The subjects were instructed to make a judgement immediately after each
brand name disappeared. When the participant pressed a key, another brand name was immedi-
ately presented. After the four tests were completed the responses were automatically transferred
to the central computer.
The counterbalance was performed using a Latin-square design with test sequences. This is
equivalent to a different random selection of test items for each subject. Both designs achieve the
same aim of controlling for the differences in the test items, although the individual difference
was addressed once in the pilot study with the baseline homogeneity.
There were four test sequences numbered from 1 to 4. The difference between the test
sequences was the order of the four tests presented to a subject. A test sequence included all four
tests. The aforementioned four tests (tests A–D) were constructed with attention (attended and
unattended) and test type (recognition and liking). Each test contained ten presented and ten
non-presented brands.
A group of 16 subjects was randomly assigned to a test sequence in each of the three experi-
ments and went through a test sequence of the four tests, but with the same duration. One test
sequence was used just four times because a total of 16 subjects were allocated to the group. The
four test sequences (1–4) with a Latin-square design (sequence 1 was ADCB, sequence 2
was BCAD, sequence 3 was DABC and sequence 4 was CBDA) guaranteed that the test items
(presented and non-presented) were counterbalanced.
The instructions for the study phase were identical for all subjects, but the instructions for the
different tests varied. In the recognition test the subjects were instructed to make old/new
judgements about the test items. Old items were those that they thought they had seen in the
106 YE AND VAN RAAIJ

study phase, while new items were those that they did not think they had seen in the study
phase. In the liking judgement test the subjects were instructed to make a quick liking
judgement (e.g. answer to ‘Do you like the brand?’) about the stimulus according to their first
impression.
All subjects were instructed to press 1 on the small keyboard if their judgements were yes
(‘old’ or ‘like’) or to press 2 if their judgements were no (‘new’ or ‘dislike’). They were told to
respond as fast as possible, although their response times were not recorded.

Results for brand awareness (recognition)


A 2 × 2 analysis of variance (ANOVA) with attention (attended versus unattended) and measure
(Aⴕ versus Bⴖ) as factors for the recognition data revealed a significant main effect of the
attention factor (F1,60 = 8.249 and p = 0.006 < 0.05). The main effect of measure yielded
F1,60 = 50.839 and p = 0.001 < 0.05. No significant effect was found for the interaction between
Downloaded By: [Institute of Professional Studies] At: 14:53 12 March 2010

attention and measure (F1,60 = 0.226, p = 0.636 > 0.05, R2 = 0.497 and adjusted R2 = 0.472).
As a result, both hypotheses 1 and 2 were supported in experiment 1. Table 2 displays the
means of the hit rate, false alarm rate, Aⴕ and B ⴖ for the brand awareness (recognition) data. The
effect of attention occurred for both brand recognition sensitivity and recognition bias. In the
deep processing mode (brands are attended to), consumers perform better in retrieving pre-
viously encoded brand information than in the shallow processing mode (brands are unattended
to). In the meantime, consumers adopt a more conservative criterion (Bⴖ ) for reporting brand
awareness in the condition of deep as compared to shallow processing.

Results for brand liking


A 2 × 2 ANOVA with attention (attended versus unattended) and measure (Aⴕ versus B ⴖ) as
factors for the recognition data revealed a significant main effect of the attention factor
(F1,60 = 19.006 and p = 0.001 < 0.05). The main effect of measure was also significant
(F1,60 = 198.104 and p = 0.001 < 0.05). No significant effect was found for the interaction
between attention and measure (F1,60 = 2.865, p = 0.096 > 0.05, R2 = 0.786 and adjusted
R2 = 0.775). Table 2 displays the means of the hit rate, false alarm rate, Aⴕ and Bⴖ for the brand
liking data.

TABLE 2. Experiment 1: brand awareness and liking (presentation duration of


0.5 s)

Awareness Liking

Attended A: Unattended C: Attended B: Unattended D:


deep mode shallow mode deep mode shallow mode

Hit rate 0.385 0.144 0.235 0.191


False alarm rate 0.122 0.085 0.138 0.231
Aⴕ 0.700 0.595 0.610 0.455
Bⴖ 0.408 0.262 0.205 0.137
SIGNAL DETECTION THEORY 107

Both hypotheses 3 and 4 were supported in experiment 1. The subjects performed better
in the deep processing mode in preserving the strength of the liking judgement for the brand
(high Aⴕ) than in the shallow processing mode. The subjects adopted a conservative criterion
(high Bⴖ) for reporting brand liking in the deep processing mode and a liberal criterion (low Bⴖ)
in the shallow processing mode.

EXPERIMENTS 2 AND 3
The study investigated whether hypotheses 1–4 were supported for longer presentations in
experiments 2 and 3. The durations of presentation in the study phases in experiments 2 and 3
were 1 and 2 s, respectively. Two groups of 16 students from a management school were
recruited as the participants for experiments 2 and 3. The remainder of the design and procedure
was similar to experiment 1. The data for the two experiments are presented in Tables 3 and 4.
Downloaded By: [Institute of Professional Studies] At: 14:53 12 March 2010

Results for brand awareness


A 2 × 2 ANOVA on the recognition data with attention and measure as the two independent
variables was performed for experiment 2. The main effect of attention was significant
(F1,60 = 25.282 and p < 0.05). The main effect of measure was significant (F1,60 = 38.636 and
p < 0.05). No significant effect was found for the attention and measure interaction
(F1,60 = 0.238, p = 0.627 > 0.05, R2 = 0.517 and adjusted R2 = 0.493).

TABLE 3. Experiment 2: brand awareness and liking (presentation duration of 1 s)

Awareness Liking

Attended A: Unattended C: Attended B: Unattended D:


deep mode shallow mode deep mode shallow mode

Hit rate 0.407 0.135 0.250 0.194


False alarm rate 0.113 0.122 0.157 0.235
Aⴕ 0.721 0.511 0.582 0.464
Bⴖ 0.456 0.201 0.186 0.135

TABLE 4. Experiment 3: brand awareness and liking (presentation duration of 2 s)

Awareness Liking

Attended A: Unattended C: Attended B: Unattended D:


deep mode shallow mode deep mode shallow mode

Hit rate 0.344 0.203 0.260 0.207


False alarm rate 0.125 0.088 0.172 0.235
Aⴕ 0.684 0.624 0.584 0.469
Bⴖ 0.370 0.322 0.197 0.127
108 YE AND VAN RAAIJ

A 2 × 2 ANOVA on the recognition data with attention and measure as the two independent
variables was performed for experiment 3. The main effect of attention was not significant
(F1,60 = 1.243 and p = 0.269 > 0.05). The main effect of measure was significant (F1,60 = 40.271
and p < 0.05). No significant effect was found for the interaction between attention and measure
(F1,60 = 0.015, p = 0.902 > 0.05, R2 = 0.409 and adjusted R2 = 0.379).
The interesting finding of experiment 3 was that divided attention did not influence brand
awareness when a long presentation duration was used (2 s) (deep processing). In other words,
hypotheses 1 and 2 are not supported in this condition.

Results for brand liking


A 2 × 2 ANOVA on the liking data with attention and measure as the two independent vari-
ables was performed for experiment 2. The main effect of attention was significant (F1,60 = 6.644
and p = 0.012 < 0.05). The main effect of measure was significant (F1,60 = 121.672 and p ≤ 0.05).
Downloaded By: [Institute of Professional Studies] At: 14:53 12 March 2010

No significant effect was found for the interaction between attention and measure (F1,60 = 1.033,
p = 0.314 > 0.05, R2 = 0.683 and adjusted R2 = 0.667).
A 2 × 2 ANOVA on the liking data with attention and measure as the two independent
variables was performed for experiment 3. The main effect of attention was significant
(F1,60 = 10.118 and p < 0.05). The main effect of measure was significant (F1,60 = 155.498 and
p < 0.05). No significant effect was found for the interaction between attention and measure
(F1,60 = 0.581, p = 0.449 > 0.05, R2 = 0.735 and adjusted R2 = 0.721).

Summary
The main effects of attention for the 12 conditions × three durations (0.5, 1 and 2 s) × four com-
ponents of brand equity (awareness sensitivity and bias and likeability sensitivity and bias) are
summarized in Table 5. Hypotheses 3 and 4 on brand liking were supported across the three
presentation durations, while hypotheses 1 and 2 on brand awareness were supported for the
short (0.5 s) and middle (1 s) presentation durations, but not for the long presentation (2 s).

Conclusions
It was found that attention had a strong effect on the sensitivity (strength) and bias of brand
awareness and brand liking in most cases in experiments 2 and 3. In general, the strength and
bias of recognition and liking of the brands were stronger in the deep processing mode (e.g.
attending to brands) than in the shallow processing mode. However, in the presentation dura-
tion of 2 s the effect of attention on the strength and bias of brand awareness was attenuated,
although the effect on the strength and bias of brand liking remained the same.

TABLE 5. The effect of attention on the four components

Short (0.5 s) Middle (1 s) Long (2 s)

Awareness sensitivity (H1) Yes Yes No


Awareness bias (H2) Yes Yes No
Likeability sensitivity (H3) Yes Yes Yes
Likeability bias (H4) Yes Yes Yes
SIGNAL DETECTION THEORY 109

With deep processing due to the long presentation duration, the subjects seemed to
remember unattended brands as well as attended brands, as compared with non-presented
brands. In other words, the shallow processing of brand awareness of unattended brands was
enhanced with deep processing due to the longer presentation. Thus, due to attention, deep
processing did not generate a difference of brand awareness between attended and unattended
brands for a longer duration (2 s). However, due to the presentation duration, deep processing
did not affect the effect of attention on brand liking.

DISCUSSION
Customer-based brand equity is defined as consisting of two core components: brand awareness
and brand liking (Keller, 1993, 1998). In the experimental settings of the current study, market-
ing management was simulated by a study phase with manipulations of attention and presentation
of brands. Extending the two core components, brand equity can be understood as a mix of
Downloaded By: [Institute of Professional Studies] At: 14:53 12 March 2010

the strength and bias of brand awareness and brand liking. The factors that promote the
four components contribute to building the strength of overall brand equity from a consumer
perspective. Theoretically, extending brand awareness and liking fits with psychological theory
because Signal Detection Theory is a better and more comprehensive measurement than the
percentage measures (Green and Swets, 1966; Yonelinas, 1994). For example, the sensitivity and
bias measures can indicate the objective response tendency for a consumer’s brand disliking.
Consumers may say they dislike Coca-Cola and are conservative in making this judgement. The
Signal Detection Theory measures can identify the real factors underlying the negative response:
a failure of the branding campaigns (using the sensitivity measure) or the consumer is conservative
(using the bias measure).
In marketing practice a comprehensive understanding of the core components of brand equity
is obtained with the four new components. Hence, marketing programmes can be more focused
on specific components in order to improve marketing productivity. In the meantime, the stra-
tegic goals of marketing programmes for building a strong brand may be set, that is to facilitate
consumers (1) in retrieving brand information, (2) in using a liberal criterion for retrieving brand
information, (3) in liking the brand and (4) in using a liberal criterion for liking the brand. These
goals can be viewed as the four principles of strategic branding.

Sensitivity and strength of brand awareness


As found in experiments 1 and 2, when the brand names are being attended to in the deep
processing mode, the sensitivity and strength of the brand in memory is stronger than in the
shallow processing mode. This is consistent with the prediction (hypothesis 1) from the global
memory models in the sense that the strength of familiarity of the brand nodes in memory is
enhanced in the deep processing mode. Thus, the match between the test brand and the
memory is facilitated, which in turn generates more sensitivity in detecting the presented
(marketed) brand.
In general, it is suggested that marketers induce consumers into a deep processing mode on
branding and marketing by attracting their attention. In this way, a strong brand will be built
in the memory that can be easily retrieved when a buying or repurchasing decision has to be
made. This suggestion has one exception, as implied by the result of experiment 3. Attracting
consumers’ attention works best when branding material is exposed to consumers only briefly.
110 YE AND VAN RAAIJ

When the exposition is long (2 s) the difference between the deep and shallow mode disappears.
Then it is not needed to attract consumer’s attention explicitly.

Bias of brand awareness


Attracting consumers’ attention in order to build strong brand awareness in their memory is
important for building a strong brand equity. However, the minus side of the deep processing
mode is the conservative criterion associated with it. With Signal Detection Theory (Fig. 2), a
conservative bias reduces the hit and false alarm rates (Fig. 1). In other words, consumers tend
to be less willing to retrieve encoded (marketed) brand information in the deep processing mode.
They may think twice before retrieving the brand information (e.g. remembering the brand).
An optimal treatment of balancing the sensitivity (strength) and the bias of brand awareness is
enhancing the sensitivity (strength) of recognition memory while keeping the bias unchanged.
Theoretically, this can be done. Signal Detection Theory suggests that the bias remains
Downloaded By: [Institute of Professional Studies] At: 14:53 12 March 2010

unchanged if the expectancy and motivation of consumers can be controlled. The expectancy
can be represented by the consumers’ perceived percentage of the presented (branded) material
in the mix of test brands. The motivation can be represented by the weights (e.g. rewards) for
reporting a signal (branded material). It can be inferred from the current study that attention has
influenced either expectancy or motivation or both in the sense that the bias of recognition
judgement is affected. Future research may search for a method of manipulating the level of
processing (e.g. using famous versus non-famous brand names such as the strategy of brand
extension), but leaving the expectancy or motivation unchanged. In this way, the sensitivity
strength of brand awareness may be increased with a stable liberal bias.

Strength of brand liking


It was found that hypothesis 3 was supported in all three experiments in the sense that the sen-
sitivity of brand liking is higher in the deep processing mode. Why hypothesis 3 was supported
may be explained by employing the Signal Detection Theory likeability model. In a nutshell,
the strength of liking of presented (marketed) brands was enhanced in the deep processing mode
during the study (marketing) phase, according to the mere-exposure effect. The strength loca-
tions of the marketed brands (signals) move to the right of the continuum (Fig. 2). Thus, the
distance between the signal (marketed brand) distribution and the noise (non-marketed brand)
distribution in the deep processing mode was larger than in the shallow processing mode.
Note that the strength sensitivity is the mean distance (difference) between the strength of the
presented brands (in deep or shallow processing modes) and non-presented brands. Thus,
hypothesis 3 was supported.
The goal of treating the sensitivity of brand liking is to facilitate consumers in liking the
brand. This goal is associated with building a strong sensitivity of brand liking that is usually the
objective of conventional marketing programmes. The current research suggests that marketers
should introduce consumers to a deep information-processing mode, the deeper the better.
Attracting consumers’ attention always leads to a sensitivity improvement of brand liking across
the three levels of short to long presentations of brand names.
As a result of the current study, it was found that the effect of attracting attention in order to
build a strong brand liking is not attenuated by long exposures of the brand. This suggests that
marketers may introduce consumers to a deep processing mode as much or as long as possible in
order to increase brand liking.
SIGNAL DETECTION THEORY 111

Bias of brand liking


Similar to the bias of brand awareness, a conservative bias of brand liking was associated with the
deep processing mode. An optimal treatment of brand liking in marketing programmes will
facilitate consumers (1) in liking the brand and (2) in using a liberal criterion for liking the
brand. In this way, the sensitivity of liking will be increased, whereas the bias is unchanged or
the bias will be shifted to be liberal, whereas the sensitivity is unchanged or both are improved.
This is particularly useful in studying consumer repurchasing decisions.
Consumer repurchasing decisions may be determined by the strength and bias of brand liking.
After a satisfactory consumption of the branded product, the strength of liking may be high,
leading to a strong brand loyalty for repurchasing. However, the repurchasing decision may also
be affected by the criterion for differentiating favourably from a baseline (neutral) liking for the
brands. In other words, the brand liking bias may be a determinant of the brand repurchasing
decisions. For example, according to the Signal Detection Theory likeability model (Fig. 2),
Downloaded By: [Institute of Professional Studies] At: 14:53 12 March 2010

given the same liking strength to the brand, the cut-off point for identifying liking may be set
differently based on the level of brand processing, as is found in the study. With different
cut-off points (from Xc to Xc' in Fig. 2), the brand with the same liking strength (point P) may
be judged as either ‘liked’ or ‘disliked’, which would lead to a positive or negative repurchasing
decision for the branded product. Hence, brand equity, the most valuable asset of the branded
product, may be influenced by the liking bias.
As is now known, brand liking consists of two components, brand likeability bias and brand
likeability sensitivity. Marketers should employ a different strategy for enhancing the liking bias
compared with the conventional approach to enhancing liking sensitivity. For example, promo-
ting Coca-Cola should build a liberal criterion for consumers to like the brand, in addition to
branding and packaging for enhancing brand liking sensitivity. Hence, having built a strong
brand liking in consumer’s memory, attracting consumers’ attention for the brand may lead to a
liberal liking bias in order to achieve brand purchase. In other words, facilitating the brand
attention may lead consumers to use a liberal criterion in choosing the brand if the brand
memory is strong. For example, when consumers are making a decision to buy a soft drink in
Atlanta (the headquarter city of the Coca-Cola company), the brand of Coca-Cola is more likely
to be chosen with less attention (liberal brand liking bias), compared with scenarios with more
attention to the same brand.
In summary, the level of brand attention becomes a ‘double-edged sword’ in (1) promoting
a strong sensitivity of brand liking and (2) facilitating a liberal criterion for consumers to like the
brand. Thus, searching for strategies that only influence the bias but not the sensitivity of brand
liking is needed for future research. According to the Signal Detection Theory likeability model,
controlling the level of expectancy and motivation is likely to affect the bias only. Hence, in
future research varying the rewards for liking a presented (marketed) brand or telling consumers
the different percentages of presented (marketed) brands in the mix of test brands may be useful
in changing the bias of brand liking only. In this way, both the sensitivity and bias of brand
liking can be promoted.

Future research
The advantages of Signal Detection Theory measures over percentages (hit rates) are well
documented in the psychological literature (Macmillan and Creelman, 1991). Experiments may
be designed for observing the differences between the bias (analogous to consumers’ actual
112 YE AND VAN RAAIJ

tendency to make judgements) and hit rates (analogous to consumers’ negative response).
As a result, the causes of consumers’ negative responses may be objectively identified and
attributed to either bias or sensitivity, whereas the conventional percentage measures (hit rates)
cannot do so.

CONCLUSIONS
Branding and brand equity are of vital importance to marketing and marketing communications
because they link customer (consumer) behaviour to firms’ financial metrics. Customer mind set
brand equity, including brand awareness and likeability, is the source for the associated financial
brand equity (Keller, 2003). However, what are the determinants of the mind set elements, such
as awareness and likeability? How can the mind set elements be manipulated in a way that may
ultimately affect brand equity? These invoke a further exploration of the customer mind set
elements of brand equity such as brand awareness and liking.
Downloaded By: [Institute of Professional Studies] At: 14:53 12 March 2010

In this paper brand awareness and brand likeability of consumer-based brand equity were
extended to four components with Signal Detection Theory: the sensitivity and bias of brand
awareness and the sensitivity and bias of brand liking. It was found that the brand liking sensi-
tivity and bias were stronger in the attended mode than in the unattended mode. A similar effect
was found for the sensitivity and bias of brand awareness for short presentations, but not for long
presentations. Knowing these determinants (e.g. ‘level of processing’, attention and duration) is
instrumental to marketing programmes endeavouring to enhance brand equity.

ACKNOWLEDGEMENTS
The authors thank Kevin L. Keller and Robert Wyer for their helpful comments on an earlier
draft of this paper.

REFERENCES
Aaker, D.A. (1991) Managing Brand Equity. New York: The Free Press.
Aaker, D.A. (1996) Building Strong Brands. New York: The Free Press.
Aaker, D.A. and Jacobson, R. (2001) The value relevance of brand attitude in high-technology markets.
Journal of Marketing Research 38, 485–93.
Ailawadi, K., Lehmann, D. and Neslin, S. (2003) Revenue premium as an outcome measure of brand
equity. Journal of Marketing 67, 1–17.
Bornstein, R.F. (1989) Exposure and affect: overview and meta-analysis of research. Psychological Bulletin
106, 265–89.
Collins, A.M. and Loftus, E. (1975) A spreading activation theory of semantic processing. Psychological
Review 82, 407–28.
Cradit, D., Tashchian, A. and Hofacker, C. (1994) Signal Detection Theory and single observation designs:
methods and indices for advertising recognition testing. Journal of Marketing Research 31(1), 117–27.
Craik, F.I.M. and Tulving, E. (1975) Depth of processing and the retention of words in episodic memory.
Journal of Experimental Psychology 104(3), 268–94.
DeCarlo, L. (2002) Signal Detection Theory with finite mixture distributions: theoretical developments
with applications to recognition memory. Psychological Review 109(4), 710–22.
Gillund, G. and Shiffrin, R.M. (1984) A retrieval model for both recognition and recall. Psychological
Review 19, 1–65.
Goldstein, E.B. and Fink, S. (1981) Selective attention in vision: recognition memory for superimposed
line drawings. Journal of Experimental Psychology: Human Perception & Performance 7(5), 954–67.
SIGNAL DETECTION THEORY 113

Green, D.M. and Swets, J.A. (1966) Signal Detection Theory and Psychophysics. New York: John Wiley.
Grier, J.B. (1971) Non-parametric indexes for sensitivity and bias: computing formulas. Psychological Bulletin
75, 424–9.
Keller, K.L. (1993) Conceptualizing, measuring, and managing customer-based brand equity. Journal of
Marketing 57, 1–22.
Keller, K.L. (1998) Strategic Brand Management. Upper Saddle River, NJ: Prentice-Hall.
Keller, K.L. (2002) Branding and brand equity. In B. Weitz and R. Wensley (eds), Handbook of Marketing.
London: Sage Publications, pp. 151–78.
Keller, K.L. (2003) Brand synthesis: the multi-dimensionality of brand knowledge. Journal of Consumer
Research 29, 595–600.
Keller, K.L., Heckler, S. and Houston, M.J. (1998) The effects of brand name suggestiveness on advertising
recall. Journal of Marketing 62, 48–57.
Leuthesser, L. (ed.) (1988) Defining, Measuring and Managing Brand Equity: A Conference Summary.
Cambridge, MA: Marketing Science Institute.
Loewenstein, G., Weber, E., Hsee, C. and Welch, N. (2001) Risk as preferences. Psychological Bulletin
Downloaded By: [Institute of Professional Studies] At: 14:53 12 March 2010

127(2), 267–86.
Macmillan, N.A. and Creelman, C.D. (1991) Detection Theory: A User’s Guide. New York: Cambridge
University Press.
Morris, K. (1999) The name’s the thing. Business Week 134, 36–9.
Murnane, K. and Mathews, P.P. (1995) Effects of changes in relative cue strength on context-dependent
recognition. Journal of Experimental Psychology: Learning, Memory, and Cognition 21(1), 158–72.
Nordhielm, C. (2002) The influence of level of processing on advertising repetition effects. Journal of
Consumer Research 29, 371–83.
Ratcliff, R., McKoon, G. and Tindall, M. (1994) Empirical generality of data from recognition memory
receiver-operating characteristics functions and implications for the global memory models. Journal
of Experimental Psychology: Learning, Memory, and Cognition 20(4), 763–85.
Shapiro, S. and Krishnan, S. (2001) Memory-based measures for assessing advertising effects: a comparison
of explicit and implicit memory effects. Journal of Advertising 30(3), 1–13.
Snodgrass, J.G. and Corwin, J. (1988) Pragmatics of measuring recognition memory: application to
dementia and amnesia. Journal of Experimental Psychology: General 117, 34–50.
Tashchian, A., White, D. and Pak, S. (1988) Signal detection analysis and advertising recognition: an
introduction to measurement and interpretation issues. Journal of Marketing Research 25, 397–404.
Tulving, E. and Thompson, M. (1973) Encoding specificity and retrieval processes in episodic memory.
Psychological Review 48, 318–73.
Wallace, W.P., Shaffer, T.R., Amberg, M.D. and Silvers, V.I. (2001) Divided attention and prerecognition
processing of spoken words and nonwords. Memory & Cognition 29(8), 1102–10.
Wyer Jr, R.S. and Srull, T.K. (1986) Human cognition in its social context. Psychological Review 93,
322–59.
Yang, Z.L. and Ye, G. (1995) A study on implicit memory of Chinese characters (II). Acta Psychologica
Sinica 27, 1–7.
Ye, G. and Van Raaij, W.F. (1997) What inhibits the mere-exposure effect: recollection or familiarity?
Journal of Economic Psychology 18, 629–48.
Yonelinas, A.P. (1994) Receiver-operating characteristics in recognition memory: evidence for a
dual-process model. Journal of Experimental Psychology: Learning, Memory, and Cognition 20, 1341–54.
Yonelinas, A.P. (2002) The nature of recollection and familiarity: a review of 30 years of research. Journal
of Memory and Language 46, 441–517.
Zajonc, R.B. (1968) Attitudinal effects of mere exposure. Journal of Personality and Social Psychology 9
(2, Part 2), 1–27.
Zajonc, R.B. and Markus, H. (1982) Affective and cognitive factors in preference. Journal of Consumer
Research 9, 123–31.
114 YE AND VAN RAAIJ

BIOGRAPHIES
Gewei Ye is an assistant professor of marketing and electronic business at Towson University,
Baltimore, USA. His research interests are consumer decision processes, brand and customer
equity, marketing research, e-business and Internet technology.
Fred Van Raaij is a professor of economic psychology at Tilburg University, The
Netherlands. His research interests include consumer decision making, marketing communica-
tion, media research and consumer confidence.
Downloaded By: [Institute of Professional Studies] At: 14:53 12 March 2010

Вам также может понравиться