Professional Documents
Culture Documents
Effects of Outdoor Advertising: Does Location Matter?: Rick T. Wilson
Effects of Outdoor Advertising: Does Location Matter?: Rick T. Wilson
Brian D. Till
Saint Louis University
ABSTRACT
Outdoor Advertising
Despite its smaller contribution to the overall spending within the advertising
industry, outdoor advertising is becoming of greater importance to practitioners
due to its ability to reach consumers who are very mobile and exposed less fre-
quently to traditional forms of media (Francese, 2003). Within academia, how-
ever, outdoor advertising has not received as much attention as more “mainstream”
media such as TV. Those studies that have focused on outdoor advertising tend
to use measures of advertising effectiveness such as recall and recognition (Donthu,
Cherian, & Bhargava, 1993; Fitts & Hewett, 1977; King & Tinkham, 1989;
Wilson & Till, 2008), attitudes (Shavitt, Vargas, & Lowrey, 2004), prevalence of alco-
hol and tobacco (Lee & Callcott, 1994), and audience measurement (Bloom, 2000).
As a growing medium, outdoor advertising has faced considerable public
scrutiny. The major issues surrounding outdoor advertising are that it is said to
be aesthetically displeasing, that it acts as an unsafe distraction for motorists,
and that it displays objectionable products (Taylor & Taylor, 1994). As a result
of these criticisms, laws such as the 1965 Highway Beautification Act and the
Bonus Act—part of the Federal Aid Highway Act of 1958—have attempted to
address the issue of roadside clutter, while the industry has pursued self-
regulation to curb the prevalence of advertising harmful products such as alco-
hol and tobacco (Taylor & Chang, 1995). Regarding safety, however, there has not
been any credible research linking outdoor advertising to traffic accidents
(Taylor, 1997).
Associative Learning
Associative learning, and more specifically classical conditioning, is a potentially
useful theoretical platform for understanding how the environment in which
outdoor advertising appears may influence consumer attitudes. Classical condi-
tioning involves the association between a conditioned stimulus (CS; i.e., brands)
and an unconditioned stimulus (US; i.e., celebrity endorsers), and, through their
association, the CS will bring about a conditioned response (CR) that is typi-
cally, but not necessarily, similar to the unconditioned response (UR) stemming
from the US itself (Shimp, 1991). While initially used in a non-advertising con-
text (e.g., Staats & Staats, 1958), classical conditioning has been successfully
used within the realm of advertising covering such topics as cognition or beliefs
toward the brand (Kim, Allen, & Kardes, 1996), affect toward the ad and/or brand
(Gresham & Shimp, 1985; Mitchell, 1986), attention to the brand (Janiszewski & War-
lop, 1993), purchase intent (Allen & Madden, 1985; Gorn, 1982), private label effects
(Till & Priluck, 2000), brand gender perceptions (Till & Priluck, 2001), celebrity
endorsers (Till, Stanley, & Priluck, 2008), and the effectiveness of product place-
ments (Schemer et al., 2008). Within these advertising studies, a variety of uncon-
ditioned stimuli have been used to elicit a conditioned response, including music
(Gorn, 1982), humor (Allen & Madden, 1985), pictures of scenic vistas (Priluck &
Till, 1998; Shimp, Stuart, & Engle, 1991), pictures of kittens (Kim, Lim, &
Bhargava, 1998), scenes of people having fun (Janiszewski & Warlop, 1993),
celebrities (Till, Stanley, & Priluck, 2008), affectively valenced advertising (Gre-
sham & Shimp, 1985), and rap music artists (Schemer et al., 2008).
H1a: Subjects who are exposed to outdoor advertising that is embedded within
positively valenced (negatively valenced) environments will have more
(less) favorable attitudes toward the advertised brand.
H1b: Subjects who are exposed to outdoor advertising that is embedded within
positively valenced (negatively valenced) environments will have higher
(lower) levels of purchase intent toward the advertised brand.
Method
Pretesting. Prior to conducting the main experiment, two pretests were per-
formed. First, the names of the fictitious brands—both target and filler—were
selected such that they did not evoke any unusual associations and were relatively
neutral. Similar to Stuart, Shimp, and Engle (1987), attitudes toward the ficti-
tious brands were measured by a convenience sample of upper-level business
students using a seven-item, seven-point semantic differential scale (good/bad,
high quality/poor quality, like very much/dislike very much, superior/inferior,
attractive/unattractive, pleasant/unpleasant, and interesting/boring). Potential
backgrounds were also pretested. Attitudes toward the positive and negative back-
grounds were measured in the same manner using a five-item, seven-point seman-
tic differential scale (good/bad, like very much/dislike very much, attractive/
unattractive, pleasant/unpleasant, and interesting/boring).
The results of the pretest indicated that the best names for the fictitious
brands were Grumet, Fique, and Degrau. The target brand for the study was
selected as Fique casual footwear, while the filler brands were Grumet bottled
water and Degrau shampoo. Pretesting also indicated that the best background
for the positively valenced background was a rural setting of gently rolling hills
with a small white farmhouse in the distance, while the negatively valenced
background was a rusted train car in front of a dilapidated warehouse.
Dependent Variables. As with the pretest, attitude toward the brand was
measured by a seven-item, seven-point semantic differential scale. Purchase
STUDY 2
Associative learning has been shown to affect product attributes and beliefs
about products based on the visual elements found within an ad (Mitchell &
Olson, 1981). However, a common concern with outdoor advertisers is not sim-
ply what elements within the ad affect product beliefs but also what factors
external to the ad (e.g., the location of the ad) might impact advertising effec-
tiveness. The purpose of Study 2 is to understand how the location of outdoor
ads may affect consumers’ beliefs about the product.
H2a: Subjects who are exposed to outdoor advertising that is embedded within
environments that project an image similar (dissimilar) to the salient
product attributes of an advertised brand will express stronger (weaker)
beliefs about those salient product attributes.
H2b: Subjects who are exposed to outdoor advertising that is embedded within
environments that project an image similar (dissimilar) to the salient
product attributes of an advertised brand will express higher (lower) lev-
els of purchase intent for that brand.
Method
Pretesting. Prior to conducting the main experiment, two different pretests were
performed. The first test ensured that the backgrounds were neutrally valenced. This
was accomplished using a three-item, seven-point semantic differential scale
(good/bad, like very much/dislike very much, and pleasant/unpleasant).
The second test ensured that the background possessed the desirable prod-
uct attributes/beliefs that were to be tested. In order to ensure that the brand
beliefs were appropriate for the product category being advertised (i.e., bottled
water, mobile phone), it was necessary to first to develop a list of salient prod-
uct attributes and then a scale for measurement. To do this, Ahtola’s (1975) vec-
tor model is used, which is based on the Fishbein model. Ahtola’s model has
been used in prior associative learning studies and improves on the Fishbein
model in that is more detailed, allowing subtle aspects of consumer belief struc-
tures to be captured (Mitchell & Olson, 1981). The tests indicated that the most
salient product attributes for bottled water are clean, fresh, and refreshing. The
most salient product attributes for a mobile phone (that is, just the phone and
not the attached wireless service) are stylish and cutting edge.
These five attributes were then tested against two dozen backgrounds. Two
backgrounds were selected for the mobile phone—one that projected a stylish
and cutting-edge image and one that did not. The most stylish and cutting edge
visual was a highly urban New York City background, while the least stylish and
cutting edge was a rural farm scene. For bottled water, again two backgrounds
were chosen—one that projected a clean, fresh, and refreshing image and one
that did not. The background rated as most clean, fresh, and refreshing was a
scenic snow-capped mountain background, and the image that was low on these
attributes was an urban New York City background.
Dependent Variables. The dependent variables for this experiment were the
brand beliefs (i.e., clean, fresh, refreshing, stylish, and cutting edge) associated with
the outdoor advertising locations. Each attribute was measured on a disagree/agree
scale of 1 to 7. As with Study 1, purchase intent was measured on a scale of 1 to 10.
Subjects. The subjects used in this experiment were upper-level business stu-
dents (different students from those used in Study 1), and the same cover story
from Study 1 was employed. With respect to sample size, the total sample was
122, with 36 subjects in the matching-attribute condition, 31 subjects in the
non–matching-attribute condition, and 55 subjects in the control condition.
Cutting edge 51 5.29 1.19 34 5.59 1.05 31 5.10 1.54 2, 113 1.279
Stylish 52 5.42a 1.11 31 6.06a 0.63 29 5.55 0.91 2, 109 4.603
Purchase intentb 52 0.772 0.114 30 0.836 0.095 29 0.783 0.154 2, 108 2.725
a
p ⭐ 0.05.
b
Log(10) transformation.
Table 3. Means and Standard Deviations by Condition (Study 2): Inferred Product Attributes and Purchase Intent from Backgrounds
(Bottled Water).
STUDY 3
The rate at which associative learning occurs varies depending on the specific
CS and US used (Stuart, Shimp, & Engle, 1987). Therefore, the lack of signifi-
cance in the preceding two studies may be related to the fact that each of the
outdoor advertisements and its corresponding background were only paired
together once. While associative learning has been found to occur with as few
as one pairing (Stuart, Shimp, & Engle, 1987), other research has suggested
that multiple pairings lead to stronger associative learning effects (Herr &
Fazio, 1991; Stuart, Shimp, & Engle, 1987). However, other studies suggest
that increasing the number of pairings has no impact in developing stronger
associative learning effects (Baker, Honea, & Russell, 2004; Mitchell & Olson,
1981). Considering that consumers are often exposed multiple times to bill-
boards, we undertook a third study to test the impact of multiple brand/con-
text pairings on the attitudes, beliefs, and purchase intent associated with
outdoor advertising. The following hypotheses are proposed consistent with
Herr and Fazio (1991) and Stuart, Shimp, and Engle (1987).
H3a: Subjects who are exposed to multiple pairings of outdoor advertising that
is embedded within positively valenced (negatively valenced) environments
will associate more (less) favorable attitudes toward the advertised brand.
H3b: Subjects who are exposed to multiple pairings of outdoor advertising that
is embedded within positively valenced (negatively valenced) environ-
ments will express higher (lower) levels of purchase intent for the adver-
tised brand.
H3c: Subjects who are exposed to multiple pairings of outdoor advertising that
is embedded within environments that project an image similar (dissim-
ilar) to the salient product attributes of an advertised brand will produce
stronger (weaker) beliefs about those salient product attributes.
Method
Experimental Design. The experimental design for the third study paral-
lels that of the first and second studies, except that subjects were exposed to mul-
tiple pairings of the outdoor advertisements and their corresponding
backgrounds. Subjects were exposed to the outdoor ads six times over the course
of eight days. Like the previous studies, this study utilized three groups—two
treatment groups and one control group. Each group contained three target
brands and two filler brands. The brands were fictitious and were determined
in pretesting to be neutrally valenced. The target brands were a Tremo mobile
phone, Vaza bottled water, and Fique athletic wear, while the filler brands were
Degrau farms and Rocha sports center.
In treatment group 1, the target brands of Vaza bottled water and Tremo
mobile phone were paired with backgrounds that were pretested to project an
image that did not match the salient product attributes of the advertised prod-
uct. Vaza was paired with a highly urban New York City background and Tremo
was paired with a rural farm scene. The first treatment group also paired the
target brand of Fique athletic wear with a background that was pretested to be
negatively valenced (a rusted train car in front of a dilapidated warehouse).
The second treatment group paired the target brands of Vaza bottled water
and Tremo mobile phone with backgrounds that were pretested to project an
image that matched the salient product attributes. Vaza was paired with a back-
ground having snow-capped mountains, and Tremo was paired with a highly
urban New York City background. The second treatment group also paired Fique
athletic wear with a background that was pretested to be positively valenced
(a rural setting of gently rolling hills with a small white farmhouse in the dis-
tance). For the control group, all branded billboards were shown against white
backgrounds.
Subjects. The subjects used in this experiment were again upper-level busi-
ness students (but different from the previous two studies) and the same cover
story was employed. Study 3 total sample size was 68, with subjects fairly evenly
divided among two treatment groups and the control group.
STUDY 4
H4a: Subjects who are more highly involved in processing outdoor advertising
that is embedded within positively valenced (negatively valenced) envi-
ronments will associate more (less) favorable attitudes toward the adver-
tised brand.
H4b: Subjects who are more highly involved in processing outdoor advertising
that is embedded within positively valenced (negatively valenced) envi-
ronments will express higher (lower) levels of purchase intent for the
advertised brand.
H4c: Subjects who are more highly involved in processing outdoor advertising
that is embedded within environments that project an image similar (dis-
similar) to the salient product attributes of an advertised brand will pro-
duce stronger (weaker) beliefs about those salient product attributes.
H4d: Subjects who are more highly involved in processing outdoor advertising
that is embedded within environments that project an image similar (dis-
similar) to the salient product attributes of an advertised brand will
express higher (lower) purchase intent for that brand.
Method
Subjects. The subjects used in this experiment were again upper-level business
students (but different from the previous three studies), and the same cover story
was employed. For Study 4, the total sample size was 98, with 40 subjects in the
control group and 58 subjects divided fairly evenly across the two treatment groups.
GENERAL DISCUSSION
The primary conclusion from these four studies is that the locations where
outdoor ads are found do not appear to affect consumer attitudes, beliefs, or
REFERENCES
Correspondence regarding this article should be sent to: Brian D. Till, Steber Professor
of Marketing, Saint Louis University, 3674 Lindell Blvd., St. Louis, MO 63108
(tillbd@slu.edu).