Content uploaded by Rick T. Wilson
Author content
All content in this area was uploaded by Rick T. Wilson on May 19, 2018
Content may be subject to copyright.
Creativity, attention and the memory for brands: an outdoor
advertising field study
Rick T. Wilson
a
*, Daniel W. Baack
b
and Brian D. Till
c
a
Marketing, Texas State University, San Marcos, TX, USA;
b
Marketing, University of Denver,
Denver, CO, USA;
c
Marketing, Xavier University, Cincinnati, OH, USA
(Received 3 January 2014; accepted 15 October 2014)
This study investigates whether creativity is sufficient by itself to attract attention to the ad
space or whether the ad must also be conspicuous. Attention to the ad is an important
driver of message processing that leads to favourable advertising effectiveness outcomes,
such as improved memory for the brand. To provide insight on this, we conducted a field
study of billboard advertising along an urban expressway to explore the variables that
affect recognition of outdoor advertising. Using a computational neuroscience software
program, we find evidence for the presence of an attention capture threshold. That is,
creativity’s message processing promoting strategy only improves memory when
billboards cross a threshold, i.e., when the ads are at a sufficient level of conspicuity
within an individual’s visual field. This threshold represents a boundary condition for
creativity and provides evidence that attention must first be captured in some contexts
before creativity increases the memory for brands in advertising. Results also show that
billboard size, visual saliency, and brand familiarity increase recognition rates.
Keywords: creativity; out-of-home advertising; billboards; attention; visual saliency;
recognition
It seems intuitive that creativity will only influence the memory for outdoor advertising
after an ad has first been seen or attended to. Yet researchers continually suggest a dual
influence of creativity on attention in that creativity both attracts initial attention to adver-
tising and increases the amount of attention already directed to it (e.g., Sasser and Koslow
2008; Smith and Yang 2004). We contend, however, that in some situations, creativity by
itself may not be sufficient to attract initial consumer attention, which promotes deeper
message processing and leads to improved memory for the brand. Our observation builds
on extant literature, including Baack, Wilson, and Till (2008), who suggest that creativity
by itself fails to attract consumer attention to advertising in certain contexts, particularly
when consumers face scarce cognitive resources. Baack, Wilson, and Till (2008) note
that an advertisement must also be conspicuous for creativity to have an effect on mes-
sage processing, and term this sufficiency requirement an attention capture threshold.
Within the creativity literature, the assumed dual influence of creativity on attention to
advertising likely stems from assumptions in the existing research. These studies typically
presuppose focal attention on the media vehicle where advertising is found. The methods
used rely on media where target advertising is embedded within the vehicle itself, and
where attention to the media is already assumed and set (e.g., television, magazine, etc.;
Ang, Lee, and Leong 2007; Jeong, Kim, and Zhao 2011; Lehnert, Till, and Carlson 2013;
Pieters and Wedel 2004; Till and Baack 2005). This research also typically uses laboratory
*Corresponding author. Email: rick.t.wilson@txstate.edu
Ó2015 Advertising Association
International Journal of Advertising, 2015
Vol. 34, No. 2, 232261, http://dx.doi.org/10.1080/02650487.2014.996117
Downloaded by [Texas State University, San Marcos] at 14:03 01 May 2015
experiments, which tend to be free of many of the distractions found in more ecologically
valid environments (Pieters and Wedel 2004; Till and Baack 2005). It is therefore not sur-
prising that previous research assumes the double role of creativity in both attracting initial
attention to advertising and increasing attention already directed at the advertising.
The purpose of our study is to increase our understanding of attention and advertising
creativity and how this affects message processing and the memory for brands. To this
end, we consider a proposed attention capture threshold in an out-of-home advertising
context and hope to demonstrate that creativity does not always initially attract attention
to advertising. Specifically, we investigate roadside billboard advertising, which is typi-
cally an environment with incidental exposure to advertising. Consumers in these envi-
ronments are highly distracted (Wilson and Till 2012). Using a roadside advertising
context permits us to demonstrate that advertising in some contexts must first be conspic-
uous to capture attention, before any positive attentional effects, such as memory for the
brand, accrue from creative messaging.
This study contributes to the marketing literature in several ways. First, this research
offers a deeper discussion of attention and message processing, and how it works for crea-
tive advertising. We tie the effectiveness of creativity to the underlying constructs of
attention. Prior creativity research in advertising merely mentions attention, making little
to no attempt to describe how attention works in advertising environments, and assuming
that creativity draws consumer attention to advertising in all circumstances (Ang, Lee,
and Leong 2007; Heiser, Sierra, and Torres 2008; Till and Baack 2005). We offer a more
comprehensive discussion of attention, drawing a distinction between perceptual features,
such as size, form, colour, and luminance, which draw attention to an ad in a context, and
cognitive features, which draw attention into an ad.
Second, and much like creativity, out-of-home advertising research is characterized
by limited discussion of attention and a lack of systematic investigation into its effective-
ness including underlying memory (Taylor 2012; Taylor, Franke, and Bang 2006; Wilson
and Till 2012). As the specific context of this study, the out-of-home research stream will
also be linked more strongly with the basic building blocks of visual attention that lead to
improvements in the memory for brands.
Third, we use an innovative software tool, derived from cognitive neuroscience, to
assess the prominence of individual outdoor ads within their environment and then relate
the output of this software to the likelihood that a particular outdoor ad will be recognized
based on its perceived level of creativity. The tool, namely the computational neurosci-
ence program developed by Itti, Koch, and Niebur (1998), analyses the physical charac-
teristics of a visual image and identifies the image components that receive the most
viewer attention, plus the order in which these items are likely to be viewed. This technol-
ogy potentially offers marketing and advertising researchers another method, a priori, to
identify which elements are likely to best capture attention. The neuroimaging we use in
this study allows us to directly track the factors influencing attention capture without
cumbersome, and at times impractical, eye-tracking equipment.
Finally, we conduct our study in the field, permitting us to more faithfully replicate
the actual outdoor advertising exposure experience and to more clearly assess the real-
world impact of medium with respect to creativity. Doing so incorporates the surrounding
environment and its associated distracters, which will aid in the identification of the atten-
tion capture threshold.
Literature review and hypothesis development
Creativity is assumed to have a dual influence on attention during message processing
(Smith and Yang 2004). It is thought to both attract initial attention to the ad space and
International Journal of Advertising 233
Downloaded by [Texas State University, San Marcos] at 14:03 01 May 2015
intensify the processing of brand information within that space. In our search for the
attention capture threshold, we use the message response involvement (MRI) framework
and visual attention theory to conceptualize how drivers might process billboard advertis-
ing, and to demonstrate that creativity’s dual influence may not function as previously
thought in an out-of-home advertising context. We use visual attention theory to under-
stand how selective attention works, and use the MRI framework to explain how the
memory for brand information is enhanced due to the allocation of attention during ad
exposure.
Message response involvement framework
The MRI framework was first introduced in 1989 by MacInnis and Jaworski as an integra-
tive and parsimonious framework to understand attitude formation due to message proc-
essing. It was later refined to include other measures of advertising effectiveness such as
brand learning and memory (MacInnis, Moorman, and Jaworski 1991). The model is
based on the pioneering works of Batra and Ray (1985), Greenwald and Leavitt (1984),
Kahneman (1973), Mitchell (1981), Petty and Cacioppo (1986), and Smith and Swinyard
(1982). Like models before it, it includes the three primary antecedents of message proc-
essing, namely motivation, opportunity, and ability. However, where MRI differs from its
predecessors is in its prescribed levels of message processing. It provides for six levels of
processing, which better enables it to highlight the varying levels of implicit and explicit
measures of memory for brands. The model is frequently used to understand how atten-
tion to advertising influences message processing within the larger advertising environ-
ment and in more specific advertising contexts, such as creativity (e.g., Smith and Yang
2004) and out-of-home advertising (e.g., Wilson and Till 2012).
The MRI model contains three sections: antecedents, message processing, and conse-
quences. Antecedents are comprised of an individual’s motivation, opportunity, and abil-
ity to process an ad. These antecedents moderate the relationship between ad exposure
and message processing, and collectively influence the extent to which an individual’s
mental activities analyse and encode ad information. Greater levels of message process-
ing evoke more enduring brand attitudes and information into memory, which are the
consequences of message processing.
Motivation is defined as the drive, urge, wish, or desire to process brand information
(Bayton 1958). Motivation affects message processing in two ways through the allocation
of attention. First, it influences what objects receive cognitive processing resources, or
what MacInnis and Jaworski (1989) refer to as the direction of attention. Second, it influ-
ences the intensity, or capacity, of working memory that is directed toward the attended
object. The more motivated an individual is to process the ad, the greater is his or her
understanding about the brand, its benefits, and its implications for the self.
Moderating motivation are the other antecedents of message processing opportunity
and ability. Opportunity is defined as the extent to which conditions are favourable for
message processing. In a roadside advertising context, billboards that are visible for lon-
ger periods of time, located closer to the road, or not eclipsed by vegetation or buildings
have a greater opportunity to be seen and processed (Wilson and Till 2012). Ability is
defined as the proficiency or skill in interpreting brand information. Individuals who are
knowledgeable or have more experience with an advertised brand or product category are
more likely to successfully and more quickly process advertising claims. Similarly, ads
that utilize simple layouts and less complex messaging are also more likely to be easily
processed.
234 R.T. Wilson et al.
Downloaded by [Texas State University, San Marcos] at 14:03 01 May 2015
As the directed attention to the ad increases due to motivation, opportunity, and abil-
ity, more cognitive capacity is allocated to brand processing. At the lowest levels of proc-
essing, individuals are likely to only identify basic features of the ad, such as colours or
shapes, or perform basic categorization of ad information, such as viewing an image in a
positive or negative manner. At these low levels, only recognition of salient ad features is
likely possible (e.g., predominance of the colour red, etc.). With moderate levels of brand
processing, individuals may begin to formulate the meaning of the ad (e.g., an ad for a
Coca-Cola product) or begin to integrate ad information (e.g., combine words and images
to conclude the ad is for a new Coca-Cola line extension). At this moderate level, recogni-
tion of the ad is more likely to occur. Finally at high levels of brand processing, individu-
als engage in role taking (e.g., envision oneself consuming the product) or constructive
processing (e.g., use information not contained within the ad to imagine new scenarios
involving the brand). At these highest levels, deeper semantic processing increases the
possibility that aided and unaided brand recall will be possible. In sum, more attention to
the ad (direction) leads to a greater depth of processing (capacity), which in turn evokes
greater memory for the brand.
Visual attention theory
The MRI framework is used to explain how memory for brands develops from successful
message processing due to the allocation of directed attention and attention capacity dur-
ing ad exposure. We now use visual attention theory to supplement MRI by articulating
the exact mechanisms by which directed attention is distributed across objects inside and
outside the ad space during message processing. Visual attention theory is based on the
seminal works of Kahneman (1973), Treisman and Gelade (1980), Wolfe (1994,1998),
and many others. According to visual attention theory, directed attention to objects in any
visual scene is driven by an object’s salience, called bottom-up factors, and an observer’s
knowledge of the object, called top-down factors. These two sets of factors explain how
visual search occurs not only in broader environmental contexts, but also in more specific
advertising contexts (Milosavljevic and Cerf 2008; Pieters and Wedel 2004).
Bottom-up factors
Directed attention to objects due to bottom-up factors is feature-driven and reflects sen-
sory stimulation. These objects ‘pop out’ by being prominent over the entire visual field.
Consumers involuntarily pay attention to these objects. Because of their ability to attract
attention reflexively, bottom-up factors are often termed ‘preattentive’ (Treisman and
Gelade 1980). Consumers process preattentive information quickly and only those objects
possessing stimulus-rich features receive focal attention (Treisman 1986). During this ini-
tial stage, objects are processed in parallel, meaning people are able to preattentively pro-
cess multiple objects simultaneously (Wolfe 1994).
A limited number of bottom-up features guide visual attention (Wolfe 1994,1998).
Those most often studied include size, motion, curvature, orientation, colour, and lumi-
nance. Less obvious features include surface reflectance and three-dimensional layout
(Wolfe et al. 2003). In the context of this study, the most relevant bottom-up factors likely
to affect roadside advertising are the size, colour, intensity, and orientation of a billboard.
Within most visual attention research, size is typically discussed separately from colour,
intensity, and orientation. These latter three bottom-up factors are traditionally grouped
together because these items represent the best approximation of the visual features that
International Journal of Advertising 235
Downloaded by [Texas State University, San Marcos] at 14:03 01 May 2015
are detected early in human visual search, and humans tend to respond collectively to
these factors rather than individually (Borji, Sihite, and Itti 2013; Itti and Koch 2000;Le
Meur and Chevet 2010). Collectively, colour, luminance, and orientation are referred to
as visual saliency. Below, we discuss visual saliency and size in more detail.
Visual saliency. Each component of visual saliency acts to attract attention to objects
within a visual field. For orientation, objects that are horizontally, vertically, or diago-
nally aligned become easier to identify when they deviate from the predominant orienta-
tion of the other objects in the visual field (Treisman and Gormican 1988). The same is
true for colour. It is easier to find objects of a particular colour when this colour deviates
from the predominant colour in the visual field, especially when complementary (Wolfe
1994). Finally, luminance, which compares bright and dark areas in the visual field, is a
colour-blind assessment of contrast between objects (Hubel and Livingstone 1990;
L€
uschow and Nothdurft 1993).
Research in cognitive neuroscience finds visual saliency to be strongly correlated with
the objects that humans preattentively process. In particular, subjects are more likely to
identify objects as the most visually salient item in a picture when these objects have con-
trasting colours, have stark differences in brightness or darkness, and/or are aligned along
a real or imaginary axis. This research compares human behaviour, typically through eye
tracking (Le Meur et al. 2006; Parkhurst, Law, and Niebur 2002; Peters et al. 2005)or
subjects circling an object in a photo (Borji, Sihite, and Itti 2013; Elazary and Itti 2008),
to computerized visual attention models able to identify the bottom-up features of colour,
luminance, and orientation. Objects having contrasting colours, areas of bright and dark
contrast, and/or elements aligned along an axis consistently garnered early visual
attention.
Based on the previous discussion, billboard advertisements that are more visually
salient are more likely to be noticed and subsequently processed. Out-of-home advertis-
ing research has found that using attractive colours and bright and dark contrasts in the ad
design lead to more favourable memory for the brand (Donthu, Cherian, and Bhargava
1993; Prendergast and Hang 1999). Businesses that regularly use billboards as part of
their marketing efforts reaffirm these claims. This research found that businesses believe,
among other claims, that using strong colours and clear contrasts in billboard ad design
was a critical factor to billboard success (Taylor, Franke, and Bang 2006). Extensive
research of on-premise signs also supports the importance of visual saliency in attracting
attention (Taylor, Claus, and Claus 2005). This research suggests that on-premise signs,
as well as billboards, help to promote top-of-mind awareness by creating more associa-
tions in memory. Sign conspicuity increases the opportunity to be noticed, and subsequent
message processing. The MRI framework suggests that greater attention toward advertis-
ing increases the likelihood for deeper message processing, which leads to improved
memory for brands. Ad characteristics that in great part drive attention to billboards are
bottom-up factors, or visual saliency, which leads to our first hypothesis:
H1: Recognition of billboard advertising is greater for those ads possessing a higher level of
visual saliency.
Size. Typically, consumers find it easier to attend to small objects when contrasted with
large objects, or vice versa, especially in laboratory research (Treisman and Gelade 1980;
Wolfe and Bose 1991). However, in more dynamic environments, such as driving, where
the visual field contains a great variety of objects with some being relevant to the driving
236 R.T. Wilson et al.
Downloaded by [Texas State University, San Marcos] at 14:03 01 May 2015
task and many others irrelevant, objects larger in size tend to be more visually salient
(Hughes and Cole 1984).
In the out-of-home advertising environment, billboards that are larger in size are more
conspicuous and therefore have a greater likelihood of being seen. Indeed, a study track-
ing subjects’ eye movements while driving found that larger billboards received glances
averaging 1.08 seconds compared to .53 seconds for smaller billboards (Beijer 2002).
Another eye-tracking study of advertising in airports and shopping malls finds that larger
ads, as compared to smaller ads, are more likely to be noticed and, when noticed, are also
more likely to be viewed multiple times and for longer periods of time (Thomas-Smith
and Barnett 2010). Research associated with on-premise signs also points to the impor-
tance of size in attracting attention (Taylor, Claus, and Claus 2005).
In another study, researchers instructed participants to verbally report all items that
caught their attention as they drove a predetermined route through suburban roads (Hughes
and Cole 1984). The authors found that larger road signs were noticed more frequently
than smaller or medium-sized signs, due to their greater conspicuity. Tying size more
directly to memory, research in a transit-advertising context found that larger ads garnered
higher levels of recall and recognition than did smaller ads (Wilson and Till 2008). Based
on this research, we predict that larger outdoor ads will more effectively capture attention,
leading to deeper levels of message processing and greater memory for advertised brands.
H2: Recognition of billboard advertising is greater for those ads larger in size.
Top-down factors
Researchers find that bottom-up factors have the greatest influence on attention immedi-
ately after their presentation as people orient themselves to the situation or task (Donk
and Soesman 2010; Le Meur et al. 2006). As time progresses, however, directed attention
to objects due to bottom-up factors decreases as task-driven, top-down factors take hold
(Parkhurst, Law, and Niebur 2002). Top-down factors relate to cognition and account for
an observer’s existing knowledge and expectations about a visual scene (Corbetta and
Shulman 2002). Top-down factors are extremely goal-directed and task-driven
(Theeuwes 2004; Yarbus 1967), and consumers allocate their directed attention to objects
that possess task-relevant features. Unlike processing resulting from bottom-up attention,
which is done in parallel, processing from top-down attention is serial (Wolfe 1994).
In a driving context, top-down factors pertain directly to the driving task. For exam-
ple, drivers actively search the visual field for pedestrians, vehicles, and other potential
traffic hazards (Chapman and Underwood 1998). Drivers can also use their existing
knowledge about the driving environment to refocus their attention. For example, drivers
often look to one side of the road for directional and exit information, or to intersections
for stop signs and traffic lights (Shinoda, Hayhoe, and Shrivastava 2001). Alternatively,
drivers may choose to redirect their attention towards task-irrelevant objects, such as veg-
etation and advertising, to fend off boredom (Trick et al. 2006). Consequently, attention
in these situations is shifted to one area or another at the discretion of the driver.
Three top-down factors that affect message processing are brand familiarity, product
involvement, and product motivation. Their role in attention to advertising is generally
well understood (cf., MacInnis and Jaworski 1989; Pieters and Wedel 2004).
Brand familiarity. Familiarity with a brand tends to decrease the amount of attention to
advertisements, as consumers believe there is less new information to process. Indeed,
International Journal of Advertising 237
Downloaded by [Texas State University, San Marcos] at 14:03 01 May 2015
Pieters and Wedel (2004) found that consumers spent considerably less time on ads for
familiar brands than for unfamiliar brands. Despite the lack of attention, brand familiarity
does help consumers tap into existing knowledge and brand structures therefore enabling
stronger brand associations to be formed in memory (Kent and Allen 1994). In an outdoor
advertising context, where the opportunity to process advertising is limited, brand famili-
arity’s influence on information processing can be especially beneficial as it promotes
more complete processing of the ad, leading to increased memory for the brand (Meurs
and Aristoff 2009; Osborne and Coleman 2008). This leads to the next hypothesis:
H3: Recognition of billboard advertising is greater for those ads containing highly familiar
brands as compared to less familiar brands.
Product involvement. Product involvement refers to the extent to which consumers
believe a product is relevant to their needs (Pratkanis and Greenwald 1993). In other
words, when consumers are involved in purchasing a product, they will invest time and
effort in evaluating product and brand options. When consumers are uninvolved, they
spend little time evaluating products and will often purchase products or brands out of
habit or with little thought. Pieters and Wedel (2004) found high-involvement products to
be more likely to generate increased levels of attention to advertisements for those prod-
ucts. With greater attention to the ad, we expect higher recognition rates (MacInnis,
Moorman, and Jaworski 1991), resulting in the following hypothesis:
H4: Recognition of billboard advertising is greater for high-involvement product categories.
Product motivation. Product motivation refers to the reasons why consumers purchase
products (e.g., think/functional versus feel/hedonic). In their 2004 study, Pieters and
Wedel found that product motivation did not affect the overall amount of attention given
to magazine ads, but it did influence the amount of attention given to the ad’s brand and
text components. In this latter case, think/functional products received more attention
than did feel/hedonic products. A study by Rosbergen, Pieters, and Wedel (1997) found
that product motivation did influence the amount of attention given to ad components
(e.g., brand, text, etc.) and the overall ad, but it did not influence recognition scores. In
accordance with this last study, we hypothesize that product purchase motivations will
not influence recognition in an outdoor advertising context:
H5: Recognition of billboard advertising is not influenced by product motivation.
Creativity and the attention capture threshold
Creativity in advertising
Creativity is an important issue for advertising practitioners and academics. A great
amount of advertising and creativity research focuses on linking creativity to advertising
effectiveness, such as increased incidences of recall (Lehnert, Till, and Carlson 2013; Till
and Baack 2005) and recognition (Baack, Wilson, and Till 2008). This is thought to occur
through increased consumer involvement with the ad and brand (MacInnis, Moorman,
and Jaworski 1991).
Creativity is typically defined by a factor termed originality, novelty, or divergence.
Within the advertising literature, this factor has been defined as a single construct (Ang
and Low 2000; Sheinin, Varki, and Ashley 2011) or as a multidimensional construct
238 R.T. Wilson et al.
Downloaded by [Texas State University, San Marcos] at 14:03 01 May 2015
(Smith, Chen, and Yang 2008; Smith et al. 2007). As a multidimensional construct, its
components are suggested to include originality (uniqueness), flexibility (multiple ideas),
synthesis (able to connect unrelated ideas), elaboration (contains numerous details), and
artistic value (visually distinctive).
However, early writers, such as Young (1985) and Mumford and Gustafson (1988),
noted that creativity means far more than just being original or unique. Originality is cer-
tainly at the core of creativity, but these early writers suggested that creativity depends on
situational and individual attributes (Mumford and Gustafson 1988). The work of Mark
Runco moved the conversation forward, noting the importance of both message appropri-
ateness and the popularity of the ideas in the message, in any conceptualization of creativ-
ity (Runco and Charles 1993; Runco and Smith 1992). This in turn prompted more recent
work to propose an additional creativity factor termed ‘relevance,’ meaningfulness or
connectedness, which refers to a creative ad’s ability to connect to its audience and be rel-
evant to the product (Ang, Lee, and Leong 2007; Smith et al. (2007).
Creativity definitional work has not solely occurred within the advertising literature.
Consider the work of Albert Rothenberg and Carl Hausman in the 1970s, about creativity
in psychology. These authors, who introduced important concepts such as a Janusian per-
spective on creativity (Rothenberg 1976), which led to the seminal advertising work of
Blasko and Mokwa (1986), take a broad, holistic view of the creative process. Drawing
on Aristotle, Kant, Galton, Freud, and many others, Rothenberg and Hausman (1976,3)
emphasize the ‘paradoxical and complex’ nature of creativity. Sternberg and Lubart
(1996) also take this perspective, talking about various approaches to creativity research
far different from those used in much advertising research (including ‘mystical
approaches’).
Creativity and context
Within the existing writings on creativity and advertising effectiveness, the role of crea-
tivity in increasing attention to advertising remains a fundamental assumption. Even in
studies where attention is an afterthought, the underlying supposition that creativity
increases attention underpins the work. Some studies have directly linked advertising cre-
ativity to increased attention, but typically through survey questions about ‘Which ad did
you pay attention to?’ rather than explicit measures of attention such as eye tracking
(e.g., Smith et al. 2007). This lack of measurement is surprising, to at least some degree,
because the dominant theoretical models applied to the question all assume some atten-
tional advantage, including potentially in terms of preattention (e.g., Smith and Yang
2004). Often, authors link these proposed attentional benefits to the divergence or origi-
nality component of creative ads. In simple terms, the contrast or divergence found in
original ads increases attention (Smith and Yang 2004).
When reviewing the large literature examining creativity and advertising (see Sasser
and Koslow 2008 for a comprehensive review), it appears that studies focus almost exclu-
sively on traditional media. Television and print media dominate the investigations. This
raises important potential questions regarding the role of media context. Traditional
media already assumes a high level of processing and focus. Add the corresponding ten-
dency for forced exposure experiments, and questions regarding the ecological validity of
the results, particularly in contrast to cases of cognitive scarcity, are merited. The one
study, to our knowledge, that examined creativity in a more cognitively scarce environ-
ment (Baack, Wilson, and Till 2008) found weaker effects than those found in previous
studies using forced exposure to traditional media. Together, the studies highlight a need
International Journal of Advertising 239
Downloaded by [Texas State University, San Marcos] at 14:03 01 May 2015
to study creativity’s effects in a media plagued by inattentiveness, such as an out-of-home
context. Considering the scarcity of cognitive resources available in typical out-of-home
viewing, this medium represents an interesting context in which to explore a potential
attentional threshold.
In this context of cognitive scarcity, it is possible that if the ad does not effectively
capture viewer attention, the creative message may never motivate the viewer to more
deeply process the ad. The inability for some process-promoting strategies, such as crea-
tivity, to have an effect on consumer memory for brands has been referred to as an atten-
tion capture threshold (Baack, Wilson, and Till 2008). In identifying the attention capture
threshold, Baack, Wilson, and Till (2008) assessed the role of creativity in two out-of-
home advertising contexts. Their study found that in environments where consumers are
in a highly distracted state, such as in airports, creativity had no influence on the memory
for advertisements. In a captive space, such as the cinema, creativity did influence mem-
ory for preshow advertising. Baack, Wilson, and Till (2008) did not explicitly test for an
attention capture threshold in their study, but hypothesized its existence after discovering
that creativity had no effect in the transit advertising environment but did in the cinema
environment.
In a roadside advertising context, which is also a highly distracting environment (Wil-
son and Till 2012), we similarly expect that creativity will only have an impact on ad rec-
ognition when billboards are larger in size and visually salient, which are factors that
correspond to their conspicuity. Research on roadside advertising appears to supports this
claim. Specifically, an eye-tracking study found that increasing the number of driver dis-
tractions, such as traffic flow, decreases the amount of attention and the number of glan-
ces towards billboards (Beijer 2002).
Consequently, we expect that creativity will have no effect on ad recognition by itself
in situations where a billboard is not highly salient or large in size. That is, if the billboard
is not conspicuously placed in a driver’s field of vision, in terms of size and visual
saliency, then creativity will not aid in message processing and therefore will not influ-
ence ad recognition. However, where a billboard is larger in size or visually salient, crea-
tivity will influence ad recognition. This leads to the last hypothesis:
H6: Ad creativity will have a positive effect on the recognition of billboard advertising only
when billboards are (a) large in size or (b) visually salient within their environment.
Method
To test our hypotheses, we administered a field study in a large Midwestern city. We
deemed a field study, in which subjects drive a predetermined route, to be the most appro-
priate research design as it increases ecological validity and provides a context where
consumers are naturally distracted and where cognitive scarcity is present. This increases
the importance of attention capture before executional factors, such as creativity, affect
viewer ad processing.
Participants
We recruited participants for the study from the undergraduate population of a large, pri-
vate Midwestern university, and gave each individual US $30 as a participation incentive
and to cover gas expenses. Using a student sample allows us to control for age and driver
experience. Additionally, the use of a homogeneous sample allows for more powerful
240 R.T. Wilson et al.
Downloaded by [Texas State University, San Marcos] at 14:03 01 May 2015
testing of effects by controlling for various extraneous individual-difference variables
(Calder, Phillips, and Tybout 1981). Eighty participants drove the route and 78 success-
fully finished the survey.
Procedure
We told participants that they were taking part in a driving study sponsored by city offi-
cials and university researchers wishing to evaluate alternate routes into the city in
response to a major highway construction project. Upon arrival, participants were briefed
on the study and handed a map detailing the route they were to drive. A researcher
instructed participants to obey all traffic laws including speed limits. Due to safety and
liability concerns, participants did not wear eye tracking or other recording devices.
Attention to advertising (visual saliency) and memory for brands (recognition) were mea-
sured post-hoc and are discussed in their respective sections.
All participants drove the same route through the city. We chose the route because it
had a wide assortment of billboards covering all the variables of interest. The route is
also a highly traveled highway and, as such, the brands advertised tend to have broad
appeal and tend to be as applicable to students as other population groups. Indeed, 65%
of our participants indicated that they travel the route at least once per month. The route
started at the university and followed a 25-mile path through urban streets and urban
expressways. All billboards were found on a 9-mile stretch of limited-access highway
three lanes in each direction for a total 18 miles of billboard-laden expressway. This 18-
mile stretch of highway did not begin until nearly 4 miles into the drive, which allowed
the participants to orient themselves to the driving task (Cole and Hughes 1984). Upon
their return to the testing centre, participants completed a short questionnaire and received
their $30 stipend. The questionnaire included ruse-consistent questions, route familiarity,
demographic variables of interest, and a measure of advertising recognition.
A total of 67 billboards were along the route. All billboards were static, standard-sized
bulletins. No digitals or spectaculars were found along the route. The design of billboards
found in the sample varied greatly with respect to the percent of ad surface space dedi-
cated to text, pictorials/graphics, and brand identifiers (e.g., logos, web addresses, etc.).
To control for potential issues associated with field research, we incorporated several
controls into our methods and survey design. With respect to outdoor advertising, we
held constant the advertising format (billboards), street type (limited-access highway),
and billboard orientation (angled toward the road). Research finds that these items influ-
ence the attention to roadside advertising and, thus, our method limited the number of
confounding variables (Traffic Audit Bureau 2010).
For our participants, we held constant their age and driving experience (college stu-
dents) and limited their in-car driving distractions (driving alone and no cell phone
usage). Additionally, all participants drove the route during off-peak hours (10 a.m. to
3 p.m., Monday through Thursday). These items not only helped to limit confounding
effects, but also provided a safer testing environment (Osborne and Coleman 2008). We
also asked a number of self-reported items on our survey: weather conditions; listening to
the radio or other devices; frequency in driving the route; perception of traffic density;
and prevalence of accidents, police, and disabled vehicles. None of these items was sig-
nificantly correlated with recognition.
Several precautions were included to diminish or test for any potential order effects.
First, we split our billboard sample into three equal groups based upon those viewed at
the beginning, middle, and end of the drive. Recognition rates among all groups were not
International Journal of Advertising 241
Downloaded by [Texas State University, San Marcos] at 14:03 01 May 2015
significantly different from one another (beginning, 30.2% recognition; middle, 25.0%;
and end, 26.8%; p D0.386). Second, the sample billboards were all located along an 18-
mile stretch of highway with nearly 4 miles of driving occurring both before and after the
billboards, which acted as a mental clearing task, thereby reducing the impact of any pos-
sible order effects. Third, participants underwent an explicit memory-clearing task before
completing the recognition portion of the survey (i.e., a series of mathematical calcula-
tions). Finally, the order of the billboards in the recognition test booklet was randomized.
Variables
Dependent variable
The dependent variable for all tests was a dichotomous measure of ad recognition (no,
yes) in which all ‘yes’ responses for each ad were summed across all participants, and
then divided by the total number of participants, to produce a percent recognition rate.
This represents an ad’s raw recognition score. We analysed our data on an advertisement
level rather than a participant level because we seek to understand what ad characteristics
lead to improved measures of memory, rather than how differences among participants
lead to improvements in memory. We deemed recognition an appropriate outcome mea-
sure resulting from attention and message processing, as increases in overall attention
have been linked to increases in recognition (Navalpakkam and Itti 2005), and recogni-
tion has been used as an operationalization of attention in past research (Hartmann, Apao-
laza, and Alija 2013; Jeong, Kim, and Zhao 2011). Recognition, and more broadly brand
identification, is a common measure of advertising effectiveness within the out-of-home
industry (Taylor, Claus, and Claus 2005; Taylor, Franke, and Bang 2006).
The recognition task involved participants looking through a booklet of 73 randomly
ordered 8.5 £11-inch colour photographs of billboards. Photos consisted of a close-up
image of the billboard panel with a minimal half-inch surrounding border. The booklet
contained 67 target billboards plus six distracter billboard ads. The distracter ads assessed
the extent of false positives and were actual billboards, taken from a Western US market,
not appearing in the Midwest city used for data collection.
In order to take into account respondent error associated with the recognition task
(i.e., false positives), the recognition scores for each ad were adjusted in accordance with
signal detection theory (Singh and Churchill 1986; Tashchian, White, and Pak 1988). Sig-
nal detection theory allows for recognition scores to be adjusted in two ways. The first
adjustment removes any potential bias based on a respondent’s propensity to answer ‘yes’
or ‘no’ independent of his or her memory for an ad (Singh and Churchill 1987). This is
termed the biased hit rate or B’
H
. This adjustment corrects the raw recognition score
downward when a respondent has a tendency to say yes on recognized ads, and upward
when a respondent has a tendency to say no.
A tendency to say yes is a negative bias correction and is determined when the false
alarm rate (the percentage of distracter ads claimed to be recognized) is greater than miss
rate (the percentage of target ads not recognized). In this case, B’
H
is calculated as:
B0
HD1¡x.1¡x/
y.1¡y/(1)
where yis the hit rate (percentage of target ads recognized) and xis the false alarm rate
(percentage of distracter ads recognized).
242 R.T. Wilson et al.
Downloaded by [Texas State University, San Marcos] at 14:03 01 May 2015
When the false alarm rate is less than the miss rate, this represents a respondent’s ten-
dency to say ‘no’ and is a positive bias correction. In this instance, B’
H
is calculated as:
B0
HDy.1¡y/
x.1¡x/¡1 (2)
The second adjustment to the raw recognition rate is based on a respondent’s ability to
discriminate between target and distracter ads (Singh and Churchill 1987). This is termed
memory sensitivity or A’ and is calculated as:
A0
D1
2C.y¡x/.1Cy¡x/
4y.1¡x/(3)
where yis the hit rate and xis the false alarm rate. A’ varies from 0.5 to 1.0, where 0.5
represents chance and 1.0 perfect recognition memory (Singh and Churchill 1987).
Once we calculated B’
H
and A’ for each respondent, we calculated a bias adjustment
by multiplying each respondent’s raw recognition score for each ad by B’
H
and A’.Table 1
demonstrates how these calculations are made for a portion of the study’s sample. Next,
the bias adjustments are summed across all participants for each ad and then divided by
the total number of participants, to produce an average bias adjustment score for each ad.
Thus, an average score is now available by ad for the bias adjustment and the original
raw recognition score. Finally, multiplying the bias adjustment and raw recognition
scores together and then adding the result to or subtracting it from the raw recognition
score creates the response bias-adjusted recognition score, which is the dependent vari-
able used in the analysis.
Bottom-up factors and the computational model of visual attention
Two bottom-up factors are used in the analysis visual saliency (i.e., colour, intensity,
and orientation) and billboard size. Visual saliency is operationalized using the computa-
tional model of visual attention, and so we first discuss how the model works before artic-
ulating its specific operationalization. Both items are used as a proxy for attention (Wolfe
1994,1998).
Computational model of visual attention. The visual attention model is a computational
software program that successfully mimics human behaviour with respect to visual atten-
tion. The method, based on the biologically plausible model of bottom-up visual attention
proposed by Koch and Ullman (1985), relies on computer algorithms to emulate human
visual attention. The model fits within the confluence of cognitive and computational neu-
roscience (Milosavljevic and Cerf 2010).
While a number of computational models have been developed, the most widely used
is the one developed by Itti and colleagues (Itti and Koch 2000; Itti, Koch, and Niebur
1998). Their model analyses an image for three bottom-up features: colour, intensity
(luminance), and orientation. Other models often include only a subset of these features
(cf., Le Meur and Chevet 2010). Itti and colleagues’ computational model of visual atten-
tion (hereafter referred to as simply the visual saliency model) analyses a static image for
the presence of contrasting colours, areas of brightness and darkness, and variations in
object orientation. The analysis occurs separately for each of the three bottom-up features
International Journal of Advertising 243
Downloaded by [Texas State University, San Marcos] at 14:03 01 May 2015
Table 1. Calculation of response bias-adjusted recognition scores.
Raw recognition score
a
Adjustments Bias-adjusted score
Respondent Ad 1 Ad 2 ... Ad 67 B’
H
A’ Ad 1 Ad 2 ... Ad 67
101... 00.081 0.559 0 0.045 ... 0
200... 0 0.118 0.408 0 0 ... 0
... ... ... ... ... ... ... ... ... ... ...
78 1 0 ... 1 0.228 0.312 0.071 0 ... 0.071
Avg. raw recognition
score
b
0.241 0.218 ... 0.241
Avg. bias-adjusted score
b
0.015 0.026 ... 0.012
Response bias-adjusted
recognition score
0.241 C
(0.241
0.015) D0.237
0.218 C
(0.218
0.026) D0.212
... 0.241 C
(0.241
0.012) D0.238
Note:
a
Recognition is a dichotomous measure where 0 refers to an unrecognized ad and 1 refers to a recognized ad.
b
The average raw and bias-adjusted recognition scores for each of
the 67 ads are averaged across 78 respondents.
244 R.T. Wilson et al.
Downloaded by [Texas State University, San Marcos] at 14:03 01 May 2015
by generating a conspicuity map for each feature (see Figure 1). For each conspicuity
map, the software program analyses the pixelated image using a center-surround method-
ology. That is, it analyses a pixel, or group of pixels, and compares it to its neighbouring
pixels. Areas with significant differences in colour, intensity, and orientation are thus
extracted from the image as potential objects that are likely to be attended to for stimu-
lus-driven reasons.
Early visual features within the colour conspicuity map are based on contrasting col-
our combinations of redgreen and blueyellow. For the intensity conspicuity map,
objects that are expected to attract attention are identified through a group of pixels with
dark centres and bright surrounds, or vice versa. The orientation conspicuity map is cre-
ated by searching for pixels that are aligned along 0,45
,90
, and 135axes. To deter-
mine which object (i.e., group of pixels) is the most salient across the entire image, the
saliency model linearly sums the three conspicuity maps into one saliency map. The
group of pixels with the highest value within the saliency map is selected as the object
that is most likely to be attended.
To determine the next object likely to receive attention, the saliency model inhibits
Figure 1. Conspicuity and saliency maps (Color online).
International Journal of Advertising 245
Downloaded by [Texas State University, San Marcos] at 14:03 01 May 2015
the first attended-to item, a process that has been demonstrated to occur naturally in
human behaviour (Posner and Cohen 1984). This prevents the first object from immedi-
ately being reselected as salient. The object is inhibited for approximately 500900 ms
(simulated time), depending on the static image’s complexity. Once inhibited, the
saliency model recalculates the conspicuity and saliency maps to identify the second
most salient object, a process that takes about 3070 ms. The program continues this pro-
cess until all salient objects are identified and the first attended-to object is reselected.
Through the selection process of salient objects, the saliency model creates a scan-path
plot, which identifies all objects in the order of their selection (see Figure 2). The number
of items in a scan-path plot depends on the image’s complexity. For a more detailed
description of the visual saliency model, see Itti, Koch, and Neibur (1998) or Koch and
Ullman (1985).
The visual saliency model, as designed by Itti and colleagues (Itti and Koch 2000; Itti,
Koch, and Niebur 1998), has been shown to be both reliable and valid in predicting actual
human eye fixations. In what is probably the most widely cited study with respect to the
model’s validation, Parkhurst, Law, and Niebur (2002) equipped four students with eye-
tracking equipment and had them view images from four different image databases:
home interiors, natural landscapes, building/city scenes, and geometric shapes. Compari-
sons between the students’ eye fixations and the saliency map produced by the visual
saliency model were significantly correlated and above chance. Correlations were 40%
stronger for early fixations. Of particular interest for our study is that correlations were
also greater for natural landscapes and building/city scenes, which are environments
where out-of-home advertising is typically found. Later, Peters et al. (2005) replicated
Parkhurst, Law, and Niebur’s study using an entirely different image database and found
the saliency map and eye fixations to be significantly correlated above chance.
Figure 2. Scan-path plot (Color online).
246 R.T. Wilson et al.
Downloaded by [Texas State University, San Marcos] at 14:03 01 May 2015
Le Meur and Chevet (2010) also validated Itti and colleagues’ model using eye track-
ing and found both methods to be highly correlated. However, these authors additionally
validated the model against another form of human behaviour beyond eye tracking. They
used a database of 5000 images in which coders had identified the most visually interest-
ing objects in a scene by drawing a rectangle around these objects. These images were
fed into the visual saliency model, and results confirmed the correlation above chance.
This additional test not only demonstrated that the saliency map is highly correlated with
objects people find interesting, but also showed that people pick objects as interesting
based on bottom-up factors. An earlier study by Elazary and Itti (2005) used a similar but
different database of images to show that objects identified as interesting correlate
strongly with the saliency map.
Visual saliency. To calculate visual saliency (hypothesis 1), we extracted a still photo-
graph of each billboard from video taken along the 18 miles of outdoor advertising-laden
highway. Researchers took the video from inside a car and from the perspective of the
driver. To ensure that billboards were comparable across photos, we attempted to hold
constant the distance each billboard was from the driver. We used on-screen billboard
height as a proxy measure for distance. Therefore, billboards on the right-hand side of the
road, which are inherently closer to the driver, were three fourths of an inch in height.
Billboards on the left-hand side of the road, which are inherently further from the driver,
were a half-inch in height.
The photos were then fed into the visual saliency software, which uses a Matlab algo-
rithm from Walther and Koch (2006) to analyse an image for the group of pixels (i.e., an
object) most likely to receive attention based on bottom-up factors. The visual saliency
model was set to identify salient objects in an image with a focal area (circle) having a
radius of 1/16 the image width (Elazary and Itti 2008). In other words, visual saliency is
determined by looking for objects that represent approximately 3% of the pixels in any
given image (see Figure 1).
For each photograph, the visual saliency model identified the number of salient
objects in the image. Any salient objects resulting from auto traffic were manually
removed from the series to ensure that we only included those items permanently part of
the viewing and therefore available to be seen by all participants. If the targeted billboard
was found within the series of salient objects, its position within the series was converted
into a visual saliency score, ranging between 0 and 1, by dividing its position by the total
number of prominent objects. This is represented by the equation:
VS D1¡.p¡1/=x(4)
where VS refers to the visual saliency score of the targeted billboard, prefers to the posi-
tion of the targeted billboard within the series of salient objects, and xrefers to the total
number of salient objects in the series. A target billboard not identified within the series
of salient objects received a score of 0.
The visual saliency score was then converted to a dichotomous variable, with 1 repre-
senting billboards with a high level of visual saliency and 0 for billboards with a low level
of visual saliency (hypothesis 1). A low/high variable for visual saliency was deemed the
most appropriate measure, as previous research using the computational neuroscience
model for visual attention found that, generally, the first set of objects identified as salient
approximate more closely with actual eye movements due to bottom-up factors (Elazary
and Itti 2008; Parkhurst, Law, and Niebur 2002). Therefore, billboards with higher visual
International Journal of Advertising 247
Downloaded by [Texas State University, San Marcos] at 14:03 01 May 2015
saliency scores represent objects with features more closely aligning with bottom-up fac-
tors. Also, other studies have used a similar approximation (c.f., Underwood, Humphrey
and van Loon 2011; Underwood and Foulsham 2006). Consequently, we median-split the
sample such that billboards with high saliency, thereby being more likely to capture atten-
tion to lead to message processing, were separated from ads with low saliency or low lev-
els of attention capture. The dichotomous view of saliency best matches the notion of the
attention capture threshold.
Size. The size of the billboard (hypothesis 2) is measured as the square feet of the bill-
board’s ad display area, i.e., 14’ £48’ billboards are 672 square feet and 19’ £48’ bill-
boards are 972 square feet.
Top-down factors
The measurements for all of the top-down factors were replicated based on the procedures
outlined by Pieters and Wedel (2004), which utilize a separate group of respondents from
the same population to assess each of the three top-down factors: brand familiarity, prod-
uct involvement, and product motivation. Respondents who categorized the top-down
factors were not the same individuals as those who participated in the main study.
Brand familiarity. Brand familiarity (hypothesis 3) was assessed across a separate group
of 24 undergraduate students at the same institution where the study took place. A brand
is defined as an ‘unknown brand’ if it is not known, a ‘known brand’ if it is known but lit-
tle else is known about the brand, and a ‘well-known brand’ if the brand is familiar and
more is known about the brand than just its name. Each product was assigned a modal
familiarity category (Pieters and Wedel 2004). A two-way random effects model for con-
sistency between judges produced an average interclass coefficient correlation (ICC) of
0.974, demonstrating a high reliability (McGraw and Wong 1996). Prior to placing brand
familiarity into the regression, it was converted into two dummy variables. A dummy var-
iable was created for ‘known brand’ and ‘well-known brand,’ leaving ‘unknown brand’ as
the variable to which the other two dummy variables are compared.
Product involvement. Product involvement (hypothesis 4) was assessed across a sepa-
rate group of 19 undergraduate students. Product involvement is defined as the extent to
which consumers believe a product category is relevant to their needs (Pratkanis and
Greenwald 1993). This variable was operationalized as uninvolved (0) if consumers are
not motivated to purchase a product or if the decision to purchase is not very important,
and involved (1) if consumers are motivated to purchase a product or if the decision to
purchase is very important (Pieters and Wedel 2004). A two-way random effects model
for consistency between judges produced an ICC of 0.861. Each product was assigned a
modal involvement category.
Product motivation. Product motivation (hypothesis 5) was assessed across a separate
group of 19 undergraduate students. Motivation is defined as the drive in or reason behind
purchasing products (MacInnis and Jaworski 1989). We operationalized this variable as
‘solving a problem’ (0) if a consumer purchases the product because it prevents or solves
a problem and/or because they need it (i.e., utilitarian), or as ‘makes me feel good’ (1) if
a consumer purchases the product because it makes them feel good and/or it helps their
personal growth (i.e., hedonic; Pieters and Wedel 2004). A two-way random effects
248 R.T. Wilson et al.
Downloaded by [Texas State University, San Marcos] at 14:03 01 May 2015
model for consistency between judges produced an ICC D.959. Each product was
assigned a modal motivation category.
Creativity
Creativity is assessed by four outdoor advertising professionals on a scale of 0 to 100,
with 100 representing a highly creative ad (Till and Baack 2005). Using professionals to
rate advertising as compared to message recipients to rate advertising creativity has a
strong precedent (cf., Baack, Wilson, and Till 2008; Till and Baack 2005) with profes-
sional judges exhibiting less variance (Reid, King, and DeLorme 1998), having greater
face validity (Kover, Goldberg, and James 1995), and reflecting more closely the advertis-
ing industry more broadly (Haberland and Dacin 1992).
Creativity was defined as an ad that is unique, unusual, or attention-grabbing in some
way, such as through its use of words or visuals. This measure represents the universally
accepted creativity dimension of divergence and excludes the less-agreed-upon dimen-
sion of relevance. We focused solely on divergence, as our professional judges are quali-
fied to evaluate originality and uniqueness but cannot possibly evaluate relevance without
knowing who the target audience is for all billboards in the sample. A two-way random
effects model for consistency between judges produced an average ICC of 0.818, demon-
strating a high reliability.
Highly creative ads in our study tended to use a variety of tactics, such as humor and
rhyming, as in a regional convenience store’s claim that ‘paninis are for weenies,’ in
referring to its oversized sandwiches. Other highly rated creative ads utilized embellish-
ments (add-ons that extend beyond the standard advertising structure) or made references
to commonly known rhyming proverbs that tie easily and humorously into product bene-
fits, such as the McDonald’s ‘You snooze. You lose.’ advertisement for its new breakfast
burritos.
Attention capture threshold
To assess the interaction of size and visual saliency with creativity, the creativity variable
was multiplied against each of the bottom-up factors (i.e., size and visual saliency) to cre-
ate two new interaction variables (hypotheses 6).
Controls
Beyond the bottom-up and top-down factors mentioned previously, research finds several
other factors that influence attention capture in a roadside advertising context. These fac-
tors include distance and the side of the road where the ad appears (Beijer 2002; Cole and
Hughes 1984; Donthu, Cherian, and Bhargava 1993; Taylor, Claus, and Claus 2005; Traf-
fic Audit Bureau 2010). A billboard was categorized as being on the right- (0) or left-hand
(1) side of the road. Distance from the road was calculated, using Google satellite images,
as the distance, in centimeters, from the base of the billboard’s structure to the center lane
of the side of the highway having an opportunity to view the ad.
As mentioned previously, we also considered a number of other controls, but none
was significantly correlated with recognition. These variables included a number of self-
reported items on our survey: weather conditions; listening to the radio or other devices;
frequency in driving the route; perception of traffic density; and prevalence of accidents,
police, and disabled vehicles. To minimize the number of extraneous variables in the
model, we do not include them in the analysis.
International Journal of Advertising 249
Downloaded by [Texas State University, San Marcos] at 14:03 01 May 2015
Results
The primary purpose of this research is to identify whether an attention capture threshold
exists for creativity. To test for its presence, we utilized a hierarchical regression analysis,
which permits us to control for the effects of the bottom-up and top-down factors on ad
recognition. A review of QQ plots of the residuals and scatter plots of the errors indi-
cates there are no normality or homoscedasticity issues.
Regressions were created in five steps by forcibly entering each group of variables
within each block. The first block contained the factors known to influence the visibility of
billboards, namely side of the road and distance from the road (controls). The second block
contained the bottom-up factors of visual saliency and billboard size. The third block con-
tained the top-down factors of brand familiarity, product involvement, and product motiva-
tion. The fourth block contained creativity, and the final block consisted of creativity’s
interaction with each of the two bottom-up factors (i.e., size and visual saliency).
Prior to running the regressions, a review of the Pearson correlation matrix indicated
some potential collinearity issues with the interaction variables and their components (see
Table 2). This issue is evident in regression model 5 in Table 3, and will be discussed shortly.
The first model, which contains the factors known to influence the visibility of bill-
boards (side of road, distance from road), is not significant (adj R
2
D0.004, F(2,65) D
0.855, p D0.430; see Table 3). Neither of the predictor variables is significant.
The second model, which adds the bottom-up factors, is significant (adj R
2
D0.170,
F(4,63) D4.380, p <0.01), as is its change in R
2
(p <0.01). At this point, size (p <
0.01) and visual saliency (p <0.01) are significant, indicating that visually salient and
larger billboards are more frequently recognized.
The third model adds the top-down factors. This model is significant (adj R
2
D0.426,
F(8,59) D7.116, p <0.000), and the model shows considerable improvement with a
change in R
2
of 0.275 (p <0.000). Size (p <0.01) and visual saliency (p <0.05) remain
significant, and the well-known brands variable is also significant (p <0.000).
In model 4, creativity is added. The model shows no improvement and adj R
2
actually
declines slightly from the previous model (adj R
2
D0.416, F(9,58) D6.219, p <0.000).
Billboard size (p <0.01), visual saliency (p <0.05), and well-known brands (p <0.000)
remain significant.
In model 5, the interaction variables for the identification of the attention capture
threshold are added. The model is significant, but its F-value and adj R
2
drop slightly
from the previous two models (adj R
2
D0.402, F(11,56) D5.040, p <0.000), suggesting
a poorly fitting model. Indeed, a closer inspection of the model reveals that the variance
inflation factors (VIFs) for the interaction variables and their components are in excess of
10. Marketing scholars suggest that VIFs above 10 are problematic (Mason and Perreault
1991). Considering that the collinearity issue directly pertains to the variables associated
with the study’s primary hypothesis (H6), additional models are developed to identify the
best possible fitting model.
As shown in Table 4, four additional models are developed that systematically remove
one or both of the interaction variables or their components. Doing so reduces VIFs below
2.0. Other models were possible, but in the interest of space, only these four models are
shown. The additional but non-displayed models had fit statistics that were not as good as
the four presented here. Mason and Perreault (1991) suggest that dropping one of the col-
linear variables is a possible solution, but this runs the risk of misspecifying the model
and biasing some of the coefficient estimates. The Akaike information criterion (AIC)
can be used, along with Adj R
2
and the F statistic, to choose the best fitting model. Among
250 R.T. Wilson et al.
Downloaded by [Texas State University, San Marcos] at 14:03 01 May 2015
Table 2. Pearson correlation matrix (N D67).
123456789101112
1. Recognition 1.00 0.15 0.01 0.40
0.22
0.20 0.03 0.19 0.53
0.31
0.41
0.29
2. Side of road ¡0.15 1.00 0.48
¡0.26
¡0.25
0.17 ¡0.01 ¡0.17 0.16 0.04 ¡0.25
¡0.27
3. Distance ¡0.01 0.48
1.00 ¡0.18 ¡0.26
0.09 0.06 ¡0.18 0.34
0.24
¡0.17 ¡0.26
4. Size 0.40
¡0.26
¡0.18 1.00 0.01 ¡0.08 0.21
¡0.11 0.11 0.00 0.97
0.06
5. Visual saliency (VS) 0.22
¡0.25
¡0.26
0.01 1.00 0.00 0.10 0.02 ¡0.01 0.18 0.07 0.97
6. Product involvement ¡0.20 0.17 0.09 ¡0.08 0.00 1.00 0.02 ¡0.05 0.02 ¡0.14 ¡0.08 ¡0.05
7. Product motivation 0.03 ¡0.01 0.06 0.21
0.10 0.02 1.00 ¡0.14 0.13 0.16 0.17 0.13
8. Known brand ¡0.19 ¡0.17 ¡0.18 ¡0.11 0.02 ¡0.05 ¡0.14 1.00 ¡0.45
¡0.01 ¡0.10 0.04
9. Well-known brand 0.53
0.16 0.34
0.11 ¡0.01 0.02 0.13 ¡0.45
1.00 0.49
0.13 0.08
10. Creativity 0.31
0.04 0.24
0.00 0.18 ¡0.14 0.16 ¡0.01 0.49
1.00 0.07 0.32
11. Size £creativity 0.45
¡0.08 0.14 0.43
0.19 ¡0.16 0.21
¡0.05 0.48
0.90
1.00 0.34
12. VS £creativity 0.29
¡0.27
¡0.26
0.06 0.97
¡0.05 0.13 0.04 0.08 0.32
0.12 1.00
Note:
p<0.05,
p<0.01,
p<0.001.
International Journal of Advertising 251
Downloaded by [Texas State University, San Marcos] at 14:03 01 May 2015
Table 3. Hierarchical regression results.
12 3 4 5
Side of road ¡0.184 (1.3) ¡0.055 (1.4) ¡0.049 (1.4) ¡0.049 (1.4) ¡0.044 (1.4)
Distance 0.080 (1.3) 0.155 (1.3) ¡0.024 (1.5) ¡0.026 (1.6) ¡0.006 (1.7)
Size 0.409
(1.1) 0.336
(1.2) 0.337
(1.2) 0.549 (21.2)
Visual saliency (VS) 0.238
(1.1) 0.210
(1.1) 0.208
(1.2) ¡0.142 (29.6)
Product involvement ¡0.165 (1.0) ¡0.163 (1.1) ¡0.154 (1.1)
Product motivation ¡0.126 (1.1) ¡0.127 (1.1) ¡0.143 (1.2)
Known brand 0.049 (1.3) 0.045 (1.4) 0.030 (1.5)
Well-known brand 0.551
(1.4) 0.544
(1.9) 0.523
(2.0)
Creativity 0.012 (1.6) ¡0.027 (2.4)
Size £creativity ¡0.512 (10.5)
VS £creativity 0.390 (33.2)
Adj R
2
¡0.004 0.170 0.426 0.416 0.402
F 0.855 4.380
7.116
6.219
5.040
AIC
C
¡120.2 ¡122.7 ¡127.3 ¡125.3 ¡121.7
Observations 67 67 67 67 67
Note: Standardized betas are presented in the models and variance inflation factors (VIFs) in parentheses. AIC refers to the Akaike information criterion, with lower values indicating a
better fitting model.
p<0.05,
p<0.01,
p<0.001.
252 R.T. Wilson et al.
Downloaded by [Texas State University, San Marcos] at 14:03 01 May 2015
candidate models, AIC provides a measure-of-fit value that minimizes the amount of
information lost by adding too many variables (overfitting) or removing too many varia-
bles (underfitting). AIC is an estimate of the KullbackLeibler divergence, and therefore
represents the distance between a given model and the ‘full reality’ distribution of data
(Burnham and Anderson 2004). Consequently, a smaller AIC value indicates a better fit-
ting model. Due to a small sample size, AIC is estimated using the corrected version,
denoted AIC
C
and calculated as:
AICCDnln RSS
n
C2k(5)
where nrefers to the sample size, krefers to the number of parameters in the model, and
RSS refers to the model’s residual sum of squares (Hurvich and Tsai 1989).
Model 6 removes billboard size and maintains the size £creativity interaction vari-
able. Visual saliency is represented as only its main effect. The model is significant (adj
R
2
D0.402, F(9,58) D5.938, p <0.000) but its fit statistic is below those of model
3 (AIC
C
D124.7), which is the best model of the previous hierarchical regression in
Table 3. The set-up for model 7 is the same as model 6 but now removes creativity. Simi-
lar to model 6, the AIC
C
, adj R
2
, and F statistic for model 7 are not better than those of
model 3 (adj R
2
D0.341, F(8,60) D5.266, p <0.000, AIC
C
D123.3). Model 8 now
adds billboard size and creativity back in but removes visual saliency. Only the visual
saliency £creativity interaction is present. The model does not show improvement over
the preceding models (adj R
2
D0.420, F(9,58) D6.306, p <0.000, AIC
C
D125.5).
Finally, the set-up for model 9 is similar to that of model 8 but removes creativity. Of the
variables of interest, only billboard size and visual saliency £creativity variables remain.
This model has the best measure of fit (adj R
2
D0.429, F(8,59) D7.210, p <0.000,
Table 4. Alternative regression results.
6789
Side of road ¡0.060 (1.4) ¡0.109 (1.4) ¡0.047 (1.4) ¡0.047 (1.4)
Distance ¡0.032 (1.5 ¡0.096 (1.5) ¡0.013 (1.6) ¡0.017 (1.5)
Size 0.330
(1.2) 0.331
(1.2)
Visual saliency (VS) 0.188 (1.2) 0.143 (1.2)
Product involvement ¡0.169 (1.1) ¡0.151 (1.0) ¡0.160 (1.1) ¡0.157 (1.0)
Product motivation ¡0.104 (1.1) ¡0.087 (1.1) ¡0.130 (1.1) ¡0.132 (1.1)
Known brand 0.046 (1.4) ¡0.017 (1.3) 0.038 (1.4) 0.033 (1.3)
Well-known brand 0.554
(1.9) 0.498
(1.4) 0.534
(1.9) 0.522
(1.5)
Creativity ¡0.620
(6.1) ¡0.022 (1.8)
Size £creativity 0.700
(5.8) 0.180
(1.6)
VS £creativity 0.230
(1.3) 0.223
(1.2)
Adj R
2
0.402 0.341 0.420 0.429
F 5.938
5.266
6.306
7.210
AIC
C
¡124.7 ¡123.3 ¡125.5 ¡127.5
Observations 67 67 67 67
Note: Standardized betas are presented in the models and variance inflation factors (VIFs) in parentheses. AIC
refers to the Akaike information criterion, with lower values indicating a better fitting model.
p<0.05,
p<0.01,
p<0.001.
International Journal of Advertising 253
Downloaded by [Texas State University, San Marcos] at 14:03 01 May 2015
AIC
C
D127.5). As compared to all other models, model 9 has the highest adj R
2
and F
value, and the lowest AIC
C
. Model 9 is selected as the primary model with which most of
the hypotheses can be tested, except for those variables removed for collinearity issues.
Hypothesis 1 could not be tested, as it was removed from model 9 due to the collinear-
ity issues evidenced by the high VIFs. However, earlier models suggest that visually
salient billboards have higher recognition rates. Hypothesis 2 is supported, indicating that
billboards larger in size have higher levels of recognition (p <0.01). Hypothesis 3 is also
supported, in that advertisements for more well-known brands are recognized at a greater
rate than are ads for unknown brands (p <0.01). Hypothesis 4, for product involvement,
is not significant (p D0.102) and therefore is not supported.
For hypothesis 5, we did not expect to find a relationship between recognition and prod-
uct purchase motivation, and, in fact, our effects testing was not significant (p D0.180).
Because hypothesis 5 was written as a null hypothesis, we undertook an additional test to
ensure that the means for product motivation (i.e., utilitarian and hedonic) were equivalent.
We performed an equivalence test following Weber and Popova’s (2012)procedureforan
independent-sample test. Following Weber and Popova’s (2012) recommendations, we
used a balanced approach in determining our study’s custom delta (D).Drawingonthe
effect size reported for media effects (r D0.52; Table 3; Weber and Popova 2012), coupled
with the effect size from Pieters and Wedel’s (2004) study (r D0.471), the balanced
approach produced a custom Dof 0.35. The test for equivalence was significant (t(66) D
0.03, DD0.35, p
eq
D0.009 (two-tailed)). This indicates that the means for the two prod-
uct motivation samples are equivalent, supporting our initial findings.
Hypothesis 6a could not be directly tested, as the creativity £size interaction was
removed from the final model. However, it is shown as significant in models 6 and 7, sug-
gesting it does have a role in ad recognition. Creativity only improves recognition when
ads are placed on larger billboards. Hypothesis 6b is supported. The interaction of visual
saliency and creativity is significant (p <0.05), indicating that ads viewed as more creative
only influence recognition rates when ads are also visually salient. Taken together, hypothe-
ses 6a and 6b provide evidence for the existence of the attention capture threshold.
Another observation from the final model is that, based on the standardized betas, the
well-known brands variable provides the greatest explanatory power as to why billboards
are later recognized in a roadside advertising context (beta D0.522). Bottom-up factors
also play a role, in that billboard size is the second most important variable (beta D
0.331), followed by the interaction of visual saliency and creativity (beta D0.223).
Discussion and implications
This study draws on the message response involvement framework and visual attention
theory to investigate the boundary conditions for the message-processing advantages of
creativity in advertising. Attention is a multidimensional construct, and we found that cre-
ativity operates differently depending on whether directed attention and attention capacity
are assumed. We also note that the visual saliency software used in our study may be
helpful in identifying one of the underlying constructs of creativity divergence, which
is at the definitional core of creativity. We now explore the implications of these findings
for the creativity literature and also discuss how they might impact the out-of-home
advertising industry, which formed the context of our study.
Creativity boundary condition
Prior research hypothesized that creativity acts to promote deeper message processing and
builds stronger brand associations in memory by both directing attention to the ad and
254 R.T. Wilson et al.
Downloaded by [Texas State University, San Marcos] at 14:03 01 May 2015
allocating greater cognitive capacity to the message-processing task (Smith and Yang
2004). These direction and capacity assumptions for creativity are typically made using
television and print media where attention to advertising has been assumed or forced
through laboratory experiments. Building on the work of Baack, Wilson, and Till (2008),
we have separated the direction and capacity components of attention by using a type of
media where directed attention could not be assumed (Wilson and Till 2012). Using road-
side advertising as our context, we have shown that for creative advertising to improve
message-processing capacity, it must first be conspicuous to capture directed attention.
Our identified boundary condition, or what Baack, Wilson, and Till termed an attention
capture threshold, significantly improves our understanding of how creativity works in
different advertising environments, and provides for a more sophisticated discussion of
the attention that creativity is purported to capture.
In our study, creativity by itself did not improve brand memory, suggesting that creativ-
ity was unable to improve attention capacity, or the depth of processing, as it is claimed to
promote. However, when directed attention was presumed possible through our visual
saliency and size measures, creative ads were recognized at a greater rate than ads deemed
less creative. What our results suggest is that future creativity research must approach the
attention assumption as a multi-dimensional construct. Attention refers to both the direction
and capacity of cognitive resources (MacInnis and Jaworski 1989), and, as our study dem-
onstrates, not all advertising environments can assume directed attention.
As a case in point, television and magazines can assume a certain level of directed
attention as ads here are embedded within the medium and typically replace content in its
entirety when visible. In this manner, attention does not need to be diverted to another
part of the media space to view an ad. For example, TV ads typically fill the entire screen,
and many magazine ads occupy a full page, so exposure to their advertising is quite likely
providing consumers are attending the medium. For these types of media where direct
attention is assumed, research has shown that creativity increases the capacity of attention
allocated to message processing, thereby improving advertising effectiveness.
The other extreme is out-of-home advertising, where attention is typically directed
elsewhere and attention to advertising in these media-consumption environments is often
incidental or related to bottom-up factors. Again, our results suggest that creativity does
not improve message processing unless the media receive directed attention. This demon-
strates that creativity may not always attract directed attention, but once directed attention
is focused on the medium or ad, creativity can improve attention capacity.
Somewhere in between these two media categories are newspapers and the Internet.
Ads here are similar to TV and magazines in that they are embedded within the media,
permitting greater opportunity for directed attention. They are also similar to out-of-
home in that they coexist with content and therefore compete for attentional resources.
However, and unlike most out-of-home placements, newspaper and Internet ads are in
much closer proximity to content, allowing for smaller shifts in directed attention. It is
entirely possible that creativity may function differently in this media consumption envi-
ronment. Future creativity research may wish to explore the effects of direct attention and
attention capacity in these other forms of media.
Visual saliency and divergence
Divergence, or novelty, within the creativity literature has often been defined as a single
construct (Ang and Low 2000; Sheinin, Varki, and Ashley 2011) or as a multidimensional
construct (Smith et al. 2007; Smith, Chen, and Yang 2008). In either case, an important
International Journal of Advertising 255
Downloaded by [Texas State University, San Marcos] at 14:03 01 May 2015
definitional component has been that the ad should be visually interesting, have interest-
ing shapes or colours, or, more generally, have artistic value. Through our exploration of
visual saliency and creativity, it has occurred to us that the visual saliency software used
in our study could be utilized to separate the visual component of divergence (i.e., artistic
value) from the more cognitive components (e.g., originality, fluency, flexibility, etc.), as
outlined by Smith, Chen, and Yang (2008) and Smith et al. (2007). Smith et al. (2007,
2008) describe an ad with artistic value as an ad that is visually distinctive, able to make
ideas come to life graphically, and artistically produced.
Separating the divergent construct in this manner could provide creativity researchers
an alternative and interesting method to align the artistic value component with a tool
that actually assesses visual saliency through a visual lens, rather than assessing the same
through a cognitive method such as survey questions. As discussed earlier, the software
identifies objects that stand out or are unique in their visual environment, which is consis-
tent with the artistic value component. Indeed, prior research has found that software-
selected objects strongly correlate with what people identify as visually interesting (Elaz-
ary and Itti 2005; Le Meur and Chevet 2010).
Visual saliency and advertising
In addition to the artistic value-defining benefits mentioned in the preceding section, the
visual saliency software might also be used more broadly in advertising research as a
proxy for directed attention, especially in environments where other measures of atten-
tion, such as eye tracking, are impractical or pose safety concerns. In our study, visual
saliency and its interaction with creativity were found to influence brand recognition,
which is consistent with on-premise sign research that stresses the importance of sign
conspicuity for message processing and consumer memory for brands (Taylor, Claus, and
Claus 2005). Taken together, these results highlight the usefulness of the visual saliency
software in studies of advertising effectiveness.
The visual saliency software is also very likely to be useful in more specific advertis-
ing applications such as the strategic placement of outdoor ads. In fact, the out-of-home
advertising industry is moving in this direction. In 2013, the Outdoor Advertising Associ-
ation of America announced the availability of its Creative Testing Tool, which permits
agencies and clients to preview their creative ad on pre-selected billboards, street furni-
ture, and transit structures to see how it would appear at varying distances to a pedestrian
or driver. This tool helps to determine a sign’s readability. If coupled with the Creative
Testing Tool, the visual saliency software could offer a powerful method to assess the
impact of creativity to both capture attention and be legible, factors which are known to
be critical to out-of-home advertising success (Taylor, Franke, and Bang 2006). We hope
that our use of this tool generates interest in the software for both academic research hand
practitioner sales applications.
Out-of-home advertising
In addition to the results for creativity mentioned earlier, our study also finds that billboard
size, visual saliency, and brand familiarity have a direct effect on brand recognition rates. In
fact, brand familiarity had the greatest impact on recognition, suggesting that familiar brands
likely have an advantage over unfamiliar brands in that familiarity facilitates message proc-
essing, allowing for more complete comprehension of the ad and encoding of information
into memory. Brand familiarity also likely builds on existing brand associations in memory.
256 R.T. Wilson et al.
Downloaded by [Texas State University, San Marcos] at 14:03 01 May 2015
Our results add to the limited number of studies using out-of-home media and better help
researchers to appreciate how this media works under varying conditions. We encourage
future research to validate our field observations in a laboratory environment using technolo-
gies such as driver simulations, as used by van der Zwaag et al. (2012).
A few variables in our study did not influence recognition rates, and we offer some
insight that might be of interest to academics and practitioners alike. Product involvement
and product motivation were not significant in any of our models, and we believe this may
be related to our testing context. Unlike other testing environments where subjects are
typically free to deviate attention from editorial content to ads with few or no consequen-
ces (cf., Pieters and Wedel 2004), our study’s primary task was driving a car a task that
presents severe consequences for extensive attention deviations. It is possible that product
involvement and motivation may have a role in increasing the motivation to process ad
information but not the opportunity to process it. These variables do not appear to enable
message processing during cognitively demanding tasks where little time is dedicated to
reading ad information.
We also controlled for several other factors that are known to influence attention cap-
ture in the roadside advertising context. However, the side of the road and the distance
from the road where billboards are placed had no effect on recognition rates. A reason for
the lack of results may be related to the type of road used in the study. Billboards were
located alongside a 9-mile stretch of urban expressway, which likely limited where most
billboard structures could be located, both legally and for optimal viewing. Most bill-
boards were located close to the highway, reducing much of the range associated with dis-
tance. None was located far enough from the road to hinder attention or message
processing. The close proximity to the highway may have also minimized the detrimental
effects of locating billboards on the left-hand side of the road.
On a broader practitioner level, our research results also confirm conventional wisdom
held within the industry. That is, size, ad design, and location matter. In fact, creative exe-
cutions that are able to ‘pop’ or stand out within their larger environment may have a bet-
ter chance of being attended to and processed. Ad managers may find it beneficial to take
a more active role in selecting billboard locations where the ad’s visual salience may help
it to stand out within its environment. Taking such a proactive role may act to reduce the
total number of billboards required to generate effective reach for out-of-home cam-
paigns. This could then help to reduce advertising clutter, an issue that is frequently dis-
cussed within the advertising industry (Jeong, Kim, and Zhao 2011; Taylor, Franke, and
Bang 2006; Wilson and Till 2011).
Acknowledgements
The authors would like to thank the American Academy of Advertising for the 2009 Research Fel-
lowship Grant, which funded the participant incentive.
Disclosure statement
No potential conflict of interest was reported by the author(s).
References
Ang, S.H., Y.H. Lee, and S.M. Leong. 2007. The ad creativity cube: Conceptualization and initial
validation. Journal of the Academy of Marketing Science 35, no 2: 22032.
International Journal of Advertising 257
Downloaded by [Texas State University, San Marcos] at 14:03 01 May 2015
Ang, S.H. and S.Y.M. L.ow. 2000. Exploring the dimensions of ad creativity. Psychology and
Marketing 17, no. 10: 835584.
Baack, D.W., R.T. Wilson, and B.D. Till. 2008. Creativity and memory effects: Recall, recognition,
and an exploration of non-traditional media. Journal of Advertising 37, no. 4: 8594.
Batra, R. and M.L. Ray. 1985. How advertising works at contact. In Psychological processes and
advertising effects: Theory, research, and applications, ed. L.F. Alwitt and A.A. Mitchell,
1344. Hillsdale, NJ: Lawrence Erlbaum Associates Publishers.
Bayton, J.A. 1958. Motivation, cognition, learning basic factors in consumer behavior. Journal of
Marketing 22, no. 1: 2829.
Beijer, D.D. 2002. Driver distraction due to roadside advertising. Masters thesis, Graduate Depart-
ment of Mechanical and Industrial Engineering, University of Toronto.
Blasko, V.J. and M.P. Mokwa. 1986. Creativity in advertising: A Janusian perspective. Journal of
Advertising 15, no. 4: 4372.
Borji, A., D.N. Sihite, and L. Itti. 2013. What stands out in a scene? A study of human explicit
saliency judgment. Vision Research, http:// dx.doi.org/10.1016/j.visres.2013.07.016. (accessed
August 23, 2013).
Burnham, K.P. and D.R. Anderson. 2004. Multimodel inference: Understanding AIC and BIC in
model selection. Sociological Methods & Research 33, no. 2: 261304.
Calder, B.J., L.W. Phillips, and A.M. Tybout. 1981. Designing research for application. Journal of
Consumer Research 8, no. 2: 197207.
Chapman, P.R., and G. Underwood. 1998. Visual search of dynamic scenes: Event types and the
role of experience in viewing driving situations. In Eye guidance in reading and scene percep-
tion, ed. G. Underwood, 36993. Oxford, England: Elsevier Science Ltd.
Cole, B.L., and P.K. Hughes. 1984. A field trial of attention and search conspicuity. Human Factors
26, no. 3: 299313.
Corbetta, M., and G.L. Shulman. 2002. Control of goal-directed and stimulus-driven attention in the
brain. Nature Reviews: Neuroscience 3, no. 3: 20115.
Donk, M., and L. Soesman. 2010. Science is only briefly represented: Evidence from probe-detec-
tion performance. Journal of Experimental Psychology: Human Perception and Performance
36, no. 2: 286302.
Donthu, N., J. Cherian, and M. Bhargava. 1993. Factors influencing recall of outdoor advertising.
Journal of Advertising Research 33, no. 3: 6473.
Elazary, L., and L. Itti. 2008. Interesting objects are visually salient. Journal of Vision 8, no. 3:
115.
Greenwald, A.G., and C. Leavitt. 1984. Audience involvement in advertising: Four levels. Journal
of Consumer Research 11, no. 1: 58192.
Haberland, G.S., and P.A. Dacin. 1992. The development of a measure to assess viewers’ judgments
of the creativity of an advertisement: A preliminary study. In Advances in consumer research,
Vol. 19, ed. J. Sherry and B. Sternthal, 81725. Provo, UT: Association for Consumer
Research.
Hartmann, P., V. Apaolaza, and P. Alija. 2013. Nature imagery in advertising: Attention restoration
and memory effects. International Journal of Advertising 32, no. 2: 183210.
Heiser, R.S., J.J. Sierra, and I.M. Torres. 2008. Creativity via cartoon spokespeople in print ads.
Journal of Advertising 37, no. 4: 7584.
Hubel, D.H., and M.S. Livingstone. 1990. Colour and contrast sensitivity in the lateral geniculate
body and primary visual cortex of the macaque monkey. Journal of Neuroscience 10, no. 7:
222337.
Hughes, P.K., and B.L. Cole. 1984. Search and attention conspicuity of road traffic control devices.
Australian Road Research Board 14, no. 1: 19.
Hurvich, C.M. and C. Tsai. 1989. Regression and time series model selection in small samples. Bio-
metrika 76, no. 2: 297307.
Itti, L. and C. Koch. 2000. A saliency-based search mechanism for overt and covert shifts of visual
attention. Vision Research 40, no. 1012: 1489506.
Itti, L., C. Koch, and E. Niebur. 1998. A model of saliency-based visual attention for rapid scene
analysis. IEEE Transaction on Pattern Analysis and Machine Intelligence 20, no. 11: 12549.
Jeong, Y., Y. Kim, and X. Zhao. 2011. Competing for consumer memory in television advertising.
International Journal of Advertising 30, no. 4: 61740.
Kahneman, D. 1973. Attention and effort. Englewood Cliffs, NJ: PrenticeHall.
258 R.T. Wilson et al.
Downloaded by [Texas State University, San Marcos] at 14:03 01 May 2015
Kent, R.J., and C.T. Allen. 1994. Competitive interference effects in consumer memory for adver-
tising: The role of brand familiarity. Journal of Marketing 58, no. 3: 97105.
Koch, C., and S. Ullman. 1985. Shifts in selective visual attention: Towards the underlying neural
circuitry. Human Neurobiology 4: 21927.
Kover, A.J., S.M. Goldberg, and W.L. James. 1995. Creativity vs. effectiveness? An integrating
classification for advertising. Journal of Advertising Research 35, no. 6: 2939.
Lehnert, K., B.D. Till, and B.D. Carlson. 2013. Advertising creativity and repetition: Recall, wear-
out and wearin effects. International Journal of Advertising 32, no. 2: 21131.
Le Meur, O., and J. Chevet. 2010. Relevance of a feed-forward model of visual attention for goal-
oriented and free-viewing tasks. IEEE Transactions on Image Processing 19, no. 11: 280113.
Le Meur, O., P. le Callet, D. Barba, and D. Thoreau. 2006. A coherent computational approach to
model bottom-up visual attention. IEEE Transactions on Pattern Analysis and Machine Intelli-
gence 28, no. 5: 80217.
L€
uschow, A., and H.C. Nothdurft. 1993. Pop-out of orientation but no pop-out of motion at isolumi-
nance. Vision Research 33, no. 1: 91104.
MacInnis, D.J., and B.J. Jaworski. 1989. Information processing from advertisements: Toward an
integrative framework. Journal of Marketing 53, no. 4: 123.
MacInnis, D.J., C. Moorman, and B.J. Jaworski. 1991. Enhancing and measuring consumers’ moti-
vation, opportunity, and ability to process brand information from ads. Journal of Marketing
55, no. 4: 3253.
Mason, C.H., and W.D. Perreault, Jr. 1991. Collinearity, power, and interpretation of multiple
regression analysis. Journal of Marketing Research 28, no. 3: 26880.
McGraw, K.O., and S.P. Wong. 1996. Forming inferences about some interclass correlation coeffi-
cients. Psychological Methods 1, no. 1: 3046; Correction, 1, no. 4: 390.
Meurs, L. van, and M. Aristoff. 2009. Split-second recognition: What makes outdoor advertising
work? Journal of Advertising Research 49, no. 1: 8292.
Milosavljevic, M., and M. Cerf. 2008. First attention then intention: Insights from computational
neuroscience of vision. International Journal of Advertising 27, no. 3: 38198.
Mitchell, A.A. 1981. The dimensions of advertising involvement. In Advances in consumer
research, Vol. 8, ed. K.B. Monroe, 2530. Ann Arbor, MI: Association for Consumer
Research.
Mumford, M.D., and S.B. Gustafson. 1988. Creativity syndrome: Integration, application, and inno-
vation. Psychological Bulletin 103: 2743.
Navalpakkam, V., and L. Itti. 2005. Modeling the influence of task on attention. Vision Research 45,
no. 2: 20531.
Osborne, A.C., and R. Coleman. 2008. Outdoor advertising recall: A comparison of newer technol-
ogy and traditional billboards. Journal of Current Issues and Research in Advertising 30, no. 1:
1330.
Parkhurst, D., K. Law, and E. Niebur. 2002. Modeling the role of salience in the allocation of overt
visual attention. Vision Research 42, no. 1: 10723.
Peters, R.J., A. Iyer, L. Itti, and C. Koch. 2005. Components of bottom-up gaze allocation in natural
images. Vision Research 45, no.18: 2397416.
Petty, R.E., and J.T. Cacioppo. 1986. The elaboration likelihood model of persuasion. In Advances
in experimental social psychology, Vol. 19, ed. L. Berkowitz, 123205. New York, NY: Aca-
demic Press.
Pieters, R., and M. Wedel. 2004. Attention capture and transfer in advertising: Brand pictorial, and
text-size effects. Journal of Marketing 68, no. 2: 3650.
Posner, M.I., and Y. Cohen. 1984. Components of visual orienting. In Attention and performance X:
Control of language processes, ed. H. Bouma and D.G. Bouwhuis, 53156. Hilldale, NJ:
Erlbaum.
Pratkanis, A.R., and A.G. Greenwald. 1993. Consumer involvement, message attention, and the per-
sistence of persuasive impact in a message-dense environment. Psychology & Marketing 10, no.
4: 32132.
Prendergast, G., and C.C. Hang. 1999. The effectiveness of exterior bus advertising in Hong Kong:
A preliminary investigation. Journal of International Consumer Marketing 11, no. 3: 3350.
Reid, L.N., K.W. King, and D.E. DeLorme. 1998. Top-level agency creatives look at advertising
creativity then and now. Journal of Advertising 27, no. 2: 116.
International Journal of Advertising 259
Downloaded by [Texas State University, San Marcos] at 14:03 01 May 2015
Rosbergen, E., R. Pieters, and M. Wedel. 1997. Visual attention to advertising: A segment-level
analysis. Journal of Consumer Research 24, no. 3: 305314.
Rothenberg, A. 1976. The process of Janusian thinking in creativity. In The creativity question, ed.
A. Rothenberg and C.R. Hausman, 31126. Durham, NC: Duke University Press.
Rothenberg, A., and C.R. Hausman. 1976. The creativity question. Durham, NC: Duke University
Press.
Runco, M.A., and R.E. Charles. 1993. Judgments of originality and appropriateness as predictors of
creativity. Personality and Individual Differences 15, no. 5: 53746.
Runco, M.A., and W.R. Smith. 1992. Interpersonal and intrapersonal evaluations of creative ideas.
Personality and Individual Differences 13, no. 3: 295302.
Sasser, S.L., and S. Koslow. 2008. Desperately seeking advertising creativity. Journal of Advertising
37, no. 4: 519.
Sheinin, D.A., S. Varki, and C. Ashley. 2011. The differential effect of ad novelty and message use-
fulness on brand judgments. Journal of Advertising 40, no. 3: 517.
Shinoda, H., M.M. Hayhoe, and A. Shrivastava. 2001. What controls attention in natural environ-
ments. Visual Research 41, no. 2526: 353545.
Singh, S.N., and G.A. Churchill, Jr. 1986. Using the Theory of signal detection to improve ad recog-
nition testing. Journal of Marketing Research 23, no. 4: 32736.
Singh, S.N., and G.A. Churchill, Jr. 1987. Response-bias-free recognition tests to measure advertis-
ing effects. Journal of Advertising Research 27, no. 3: 2336.
Smith, R.E., J. Chen, and X. Yang. 2008. The impact of advertising creativity on the hierarchy of
effects. Journal of Advertising 37, no. 4: 4761.
Smith, R.E., S.B. MacKenzie, X. Yang, L.M. Buchholz, and W.K. Darley. 2007. Modeling the
determinants and effects of creativity in advertising. Marketing Science 26, no. 6: 81933.
Smith, R.E., and W.R. Swinyard. 1982. Information response models: An integrative approach.
Journal of Marketing 46, no. 1: 8193.
Smith, R.E., and X. Yang. 2004. Toward a general theory of creativity in advertising: Examining
the role of divergence. Marketing Theory 4, no. 31: 3158.
Sternberg, R.J., and T.I. Lubart. 1996. Investing in creativity. American Psychologist 51: 67788.
Tashchian, A., J.D. White, and S. Pak. 1988. Signal detection analysis and advertising recognition:
An introduction to measurement and interpretation issues. Journal of Marketing Research 25,
no. 4: 397404.
Taylor, C.R. 2012. Editorial. Back to the future: Some topics we should not forget about in advertis-
ing research. International Journal of Advertising 31, no. 4: 699702.
Taylor, C.R., T.A. Claus, and S.L. Claus. 2005. On-premise signs as storefront marketing devices
and systems. Washington, DC: US Small Business Administration.
Taylor, C.R., G.R. Franke, and H. Bang. 2006. Use and effectiveness of billboards: Perspectives from
selective perception theory and retail-gravity models. Journal of Advertising 35, no. 4: 2134.
Theeuwes, J. 2004. Top-down search strategies cannot override attention capture. Psychonomic
Bulletin & Review 11, no. 1: 6570.
Thomas-Smith, C. and G. Barnett. 2010. Seeing is believing: Viewing engagement in place-based
media. In Your audience Dmedia consumer Cgenerator, ed. D.S. Fellows, 24762. Amster-
dam: ESOMAR.
Till, B.D., and D.W. Baack. 2005. Recall and persuasion: Does creative advertising matter? Journal
of Advertising 34, no. 3: 4757.
Traffic Audit Bureau. 2010. EYES ON visibility research: VAIs and how they impact EYES ON
audiences. New York, NY: Media Measurement, Inc., http://www.eyesonratings.com/tabbriefs.
aspx (accessed July 23, 2010).
Treisman, A.M. 1986. Features and objects in visual processing. Scientific American 255, no. 5:
11425.
Treisman, A.M., and G. Gelade. 1980. A feature-integration theory of attention. Cognitive Psychology
12, no. 1: 97136.
Treisman, A.M., and S. Gormican. 1988. Feature analysis in early vision: Evidence from search
asymmetries. Psychological Review 95, no. 1: 1548.
Trick, L.M., J.T. Enns, J. Mills, and J. Vavrik. 2006. Paying attention behind the wheel: A frame-
work for studying the role of attention in driving. Theoretical Issues in Ergonomics Science 5,
no. 5: 385424.
260 R.T. Wilson et al.
Downloaded by [Texas State University, San Marcos] at 14:03 01 May 2015
Underwood, G., and T. Foulsham. 2006. Visual saliency and semantic incongruency influence eye
movements when inspecting pictures. The Quarterly Journal of Experimental Psychology 59,
no. 11: 193149.
Underwood, G., K. Humphrey, and E. van Loon. 2011. Decisions about objects in real-world scenes
are influenced by visual saliency before and during their inspection. Vision Research 51:
20318.
van der Zwaag, M.D., C. Dijksterhuis, D. de Waard, B.L. Mulder, J.H. Westerink, and K.A. Broo-
khuis. 2012. The influence of music on mood and performance while driving. Ergonomics 55,
no. 1: 1222.
Walther, D., and C. Koch. 2006. Modeling attention to salient proto-objects, Neural Networks 19,
no. 9: 1395407.
Weber, R., and L. Popova. 2012. Testing equivalence in communications research: Theory and
application. Communication Methods and Measures 6: 190213.
Wilson, R.T., and B.D. Till. 2008. Airport advertising effectiveness: An exploratory field study.
Journal of Advertising 37, no. 1: 5972.
Wilson, R.T., and B.D. Till. 2011. Effects of outdoor advertising: Does location matter? Psychology
& Marketing 28, no. 9: 90933.
Wilson, R.T., and B.D. Till. 2012. Managing non-traditional advertising: A message processing
framework. In Advertising theory, ed. S. Rodgers and E. Thorson, 33754. New York, NY:
Routledge, Taylor & Francis Group.
Wolfe, J.M. 1994. Guided search 2.0: A revised model of visual search. Psychonomic Bulletin &
Review 1, no. 2: 20238.
Wolfe, J.M. 1998. Visual search. In Attention, ed. H. Pashler. 1374. Hove, England: Psychology
Press.
Wolfe, J.M., and M. Bose. 1991. Visual search for the ‘medium’ stimulus. Unpublished research
report. Cambridge, MA: Massachusetts Institute of Technology.
Wolfe, J.M., S.J. Butcher, C. Lee, and M. Hyle. 2003. Changing your mind: On the contributions of
top-down and bottom-up guidance in the visual search for feature singletons. Journal of Experi-
mental Psychology: Human Perception and Performance 29, no. 2: 483502.
Yarbus, A.L. 1967. Eye movements and vision. New York, NY: Plenum.
Young, J.G. 1985. What is creativity? The Journal of Creative Behavior 19, no. 2: 7787.
International Journal of Advertising 261
Downloaded by [Texas State University, San Marcos] at 14:03 01 May 2015
