Graduate School of Communication
When online advertising works and when it does not:
A comparative analysis of the distractor devaluation effect
and the mere exposure effect
Master thesis
Faculty of Social and Behavioral Science
Persuasive Communication track
Supervisor: dr. Sandra Zwier
Date of completion: February 1
st2017
Student: Bart van Teutem
1
Abstract
Most online advertising business models, including those of industry giants such as Facebook and Google AdWords, require advertisers to pay per thousand ad impressions. This is based on the idea that exposure to stimuli such as online ads suffices to enhance consumers’ responses, also known as the mere exposure effect (MEE). The extant research in support of the MEE, however, generally encourages people to browse freely without providing clear and specific search goals, while recent studies in experimental psychology suggest that negative effects of ad exposure may actually result when people are driven by specific goals. In the latter case, the disruption of goal-driven attention is likely to result in devaluation of the distractor stimulus such as an ad, a phenomenon referred to as the distractor devaluation
effect (DEE). The present study critically tested the MEE versus the DEE in an experimental
study where people browsed websites with different mind-sets: goal-directed versus
exploratory. The results showed that online browsing with a goal-directed mind-set leads to more negative evaluations of brands in online advertisements compared to browsing with an exploratory mind-set. Additionally, brand evaluations of people browsing online with an exploratory mind-set only increased after multiple exposures to online ads. The proximity of an ad to a website’s primary content did not impact subsequent brand evaluations. All in all, the present study importantly concludes that online ads only increase brand evaluations when they do not distract people from completing their goals while browsing the web. When validated in future research, these findings do not only constrain the popular concept of the
mere exposure effect in advertising research but also have important implications for
advertising business models, since not all exposure proves to be good exposure.
Keywords: online advertising, selective attention, mere exposure, distractor devaluation, goal-driven attention, exploratory search.
2
Introduction
Consumers’ selective attention makes it very difficult for marketers to appeal to their target audience. Because of consumers’ limited ability to process information they are unable and unwilling to pay attention to all the ads they encounter (Just & Carpenter, 1992; Kahneman, 1973). Throughout the years consumers develop persuasion knowledge (Friestad & Wright, 1994) and as they get older they get better and better at consciously, and subconsciously, selecting the persuasive stimuli to which they do, and do not attend. Especially online consumers have become adept at ignoring ads (Resnick & Albert, 2014). Often they avoid paying attention to areas where ads are likely to occur (Owens & Chaparro, 2011; Resnick & Albert, 2014). Disrupting this selective attention process proves to be a difficult challenge for online marketers nowadays. But they try hard.
Traditionally, marketers tend to believe in the notion: ‘there’s no harm in trying’. Either an ad is consciously or subconsciously attended to and it will have positive effects on consumers’ cognitive, affective, or behavioural responses or an ad is not processed and it simply does not produce any effect. This line of thought has led online marketers to experiment with an abundance of advertising formats ranging from highly disruptive pop-up ads to highly embedded native advertisements, all in order to grasp a glimpse of the attention of their elusive target audience. These assertive strategies of online marketers are based on the idea of the mere exposure effect (MEE). Supporters of the MEE argue that people develop a positive affect towards the things they grow familiar to (Zajonc, 1968). Simply being exposed to online ads without thoroughly processing them may, according to the MEE, already be sufficient to enhance consumers’ brand responses because it increases familiarity (Mantonakis, Whittlesea, & Yoon, 2008). The MEE has been a popular concept within
3
advertising research for decades that, according to meta-analysis results, finds support in hundreds of studies (Bornstein, 1989).
A more recently developed theory, labelled the distractor devaluation effect (DDE), shines a different light on some of the claims posited by the MEE. The DDE suggests that exposure to advertising, in some situations, may result in negative affective ad and brand evaluations. The DDE finds its origin in the field of cognitive psychology and suggests that disrupting goal-driven attention is likely to result in a negative affective evaluation of the distractor stimulus (Raymond, Fenske, & Tavassoli, 2003). According to the DDE, the ignoring of distracting stimuli requires inhibition, and these inhibitory states become associated with the distractor stimulus (Raymond, Fenske, & Westoby, 2005). During a subsequent encounter, the distractor stimulus will trigger a negative affective evaluation due to a reinstated inhibitory state
(Fenske & Raymond, 2006).
The DDE has mostly been studied in the field of cognitive- and neuropsychology. It was not until recently that the DDE has been applied to the field of marketing. For advertising, the DDE posits that people may develop negative affect towards ads that distract them from their immediate goals. In future encounters, the brands from these ads might, therefore, be ignored. DDE research in marketing is scarce, but the few studies that have been conducted so far found support for DDE’s assumptions (Duff & Faber, 2011; Janiszewski, Kuo, & Tavassoli, 2013). Duff and Faber (2011) demonstrated that people develop more negative attitudes towards brands that appear as distracting banner ads on a web page. In addition, Janiszewski et al. (2013) found that products that are initially ignored by consumers as distractors during a selection task will also be ignored at subsequent encounters. Although the few studies that have been conducted so far support the generalizability of the DDE to the field of advertising
4
effects, many questions remain. The contradicting claims of the DDE with regard to the MEE bring about a need for more research that can identify important moderators and possible boundary conditions for both effects.
The present study, therefore, aims to provide a better understanding of the DDE in relation to the MEE by trying to identify those conditions under which each effect is more likely to occur in an online advertising context. In the present study, the effects of being distracted by an online ad on the affective evaluation of the brand will be compared between consumers that are engaged in a goal-driven task, and consumers that are engaged in exploratory search. Additionally, this study will look at the moderating role of the number of exposures to the ad and the location of the ad on a webpage.
RQ: ‘What are the effects of being distracted by an online ad on the affective evaluation of the
brand displayed in the ad? How do these effects differ during a goal-driven task versus exploratory search? How are these effects moderated by number of exposures to the ad and the location of the ad?’
The DDE can have important implications for advertising practice. Many advertising business models, including those of industry giants such as Facebook and Google AdWords, are fully or partly based on costs per mille (CPM). CPM-models require advertisers to pay per
thousand ad impressions. This business model relies on the assumed positive effects of ad impressions that are suggested by the MEE. The DDE, on the other hand, suggests that mere exposure to ads may also bring about negative effects for the advertiser. Possible identified boundary conditions in this study may therefore help marketers decide whether it is sensible to rely on MEE and accept a CPM strategy, or choose a cost per click (CPC) strategy instead.
5
CPC-models require advertisers to pay per click, thereby ensuring that money is only paid for consumers that show interest in the ad. The results from the present study may prevent
marketers from making suboptimal advertising budget decisions and aid them in the justification for the choices they make during their campaign development.
Theoretical framework
The mere exposure effect
The ‘mere exposure effect’ (MEE) was initially coined by Zajonc (1968) and suggests that exposure to stimuli increases positive affective responses towards these stimuli through familiarity (Bornstein, 1989). It is assumed that these increased positive affective responses are not guided by rational processes and may also arise subliminally (Zajonc, 2001). MEE’s have been demonstrated in a wide array of different domains, such as art (Cutting, 2003; Zajonc, Shaver, Tavris, & Van Kreveld, 1972), sound (Heingartner & Hall, 1974; Obermiller, 1985), Chinese characters (Zajonc, 1968; Zizak & Reber, 2004), voting behaviour (Verhulst, Lodge, & Lavine, 2010; Verrier, 2012), faces (Rhodes, Halberstadt, & Brajkovich, 2001; Zebrowitz, White, & Wieneke, 2008), and marketing (Janiszewski, 1993; Mantonakis et al., 2008; Matthes, Schemer, & Wirth, 2007). Although hundreds of studies have been conducted, and the MEE has been found to be robust and reliable (Bornstein, 1989), the precise
mechanisms and boundary conditions regarding the effect have remained largely unclear (Maio & Haddock, 2009; Ruggieri & Boca, 2013).
During the last few decades several explanations for the MEE have been developed (Lee, 1991; Reber, Winkielman, & Schwarz; 1998; Zajonc, 2001). Although some of these explanations are still being considered (Ruggieri & Boca, 2013), the processing fluency hypothesis (PFH) is currently dominating the field (Leynes & Addante, 2016; Moreland &
6
Topolinski, 2010). One of the strengths of the PFH is its ability to explain subliminal MEE’s in the absence of recognition (Hansen & Wänke, 2009). The PFH is based on two premises that both find strong support in research results (Moreland & Topolinski, 2010). The first premise is that the repeated processing of a stimulus results in an increase in processing fluency. Additionally, the PFH posits that this increased processing fluency results in more positive affect. However, how exactly this increased fluency results in more positive affect remains a matter of debate (Leynes & Addante, 2016). Some authors argue that people tend to misattribute the positive experience of processing fluency to the target stimulus (Nordhielm, 2002; Whittlesea, 1993), while others suggest that the processing fluency itself is being experienced as affectively positive (Fang, Singh, & Ahluwalia, 2007; Winkielman & Huber, 2009).
Distractor devaluation effect
The ‘distractor devaluation effect’ (DDE), an effect first described by Raymond et al. (2003), may counteract the evaluative consequences of the MEE. Raymond et al. were the first to demonstrate that selective attention may negatively affect evaluations. They exposed their participants to abstract circular or square Mondrian patterns that had to be identified as targets or ignored as distractors. After the search task, the participants evaluated the targets and distractors they were exposed to as well as some additional novel stimuli. What Raymond et al. found was that their participants rated the distractors more negatively compared to the targets and novel stimuli. Their findings imply that a person’s attentional focus during
exposure to a stimulus can shape the affective evaluation of this stimulus during a subsequent encounter. It appears that previously ignored stimuli are likely to be rated more negatively compared to previously attended or novel stimuli (Raymond et al., 2003). The DDE hereby suggests that a MEE does not occur for stimuli that were previously ignored, but opposite
7
effect takes place instead. The DDE has, in line with the MEE, been replicated across a broad range of different stimuli (Elliot, 2014). DEE’s have been found after trying to ignore faces (Goolsby et al., 2009; Kiss et al., 2007), abstract shapes (Kihara, Yagi, Takeda, & Kawahara, 2011), Roman letters (Kihara et al., 2011; Veling, Holland, & Van Knippenberg, 2007), Chinese characters (Martiny-Huenger, Gollwitzer, & Oettingen, 2014), consumer products (Janiszweski et al., 2013), and online banner ads (Duff & Faber, 2011).
Among the most prominent explanations regarding the DEE’s underlying mechanisms is the devaluation-by-inhibition hypothesis (DIH), developed by Fenske and Raymond (2006). The DIH posits that ignoring distracting stimuli requires attentional inhibition. As a consequence, these inhibitory states become associated with the distractor stimulus and will be reinstated during a subsequent encounter. The reinstated attentional inhibition will in turn negatively affect the stimulus’ evaluations. Several neurophysiological studies have found support for the DIH (Doallo et al., 2012; Fragopanagos et al., 2009; Kiss et al., 2007) by demonstrating that brain responses (i.e. event related potentials) related to inhibition measured during the exposure to distractors covary with the subsequent affective evaluations of these distractors.
Selective attention
A possible explanation for the discrepancy in findings between MEE and DDE research results can be found in the domain of selective attention. Selective attention involves people’s tendency to consciously, and subconsciously, choose which parts of their current sensory input will be used for further processing (Smith & Kosslyn, 2014). Processing information requires mental resources, and people only have a limited pool of these mental resources (Lang, 2001). Therefore, people at any given moment are presented with too many cues to include in coherent perceptions (Smith & Kosslyn, 2014). People deal with this excessive
8
input via selective attention (Treisman, 1964). It is this selective attention process that enables consumers to filter out (Lo, Hsieh, & Chiu, 2014; Owens & Chaparro, 2011; Resnick & Albert, 2014) the majority of the thousands of ads they are exposed to on a daily basis (Belch & Belch, 2015).
Differences in processing modes
The active ignoring of ads is likely to occur during top-down processing (TDP), a processing mode that has largely been neglected in marketing research so far (Duff & Faber, 2011). People that are engaged in TDP try to match the features of what they perceive to something they either know or are expecting (Carrasco, 2011). TDP is intentional and driven by a person’s knowledge, beliefs, expectations, and goals (Carrasco, 2011). For online consumers, TDP takes place when they are browsing the web with a specific goal in mind, such as booking an affordable hotel for a weekend trip or finding a good recipe for a diner with friends. In these situations, online consumers are motivated to focus their attention on the information that will help them accomplish their goals and ignore stimuli that distract them from their goals (Duff & Faber, 2011; Janiszewski et al., 2013). According to the DDE, a negative affective response to these ignored distracting stimuli may be expected.
To demonstrate this phenomenon in the domain of marketing, Duff and Faber (2011) investigated how people with clear and specific goals respond to online ads. In their experiment, the participants were given the task of finding specific information from news articles. Duff and Faber (2011) hypothesised that participants with such clear and specific goals would respond negatively to online ads because they would find the ads distracting and, therefore, try to actively ignore them. As a consequence of ignoring the ads, the participants were expected to develop more negative attitudes toward the brands that appeared in the ads.
9
Their expectations were supported by the results. Additionally, Resnick and Albert (2014) found in an eye-tracking study that goal-directed online consumers pay less attention to banner ads compared to online consumers in exploratory search. The reduced ad dwell time for goal-directed participants in the study of Resnick and Albert (2014) supports the notion of Duff and Faber (2011) that goal-directed online consumers actively try to ignore ads.
Although one would expect a vast amount of online behaviour to be driven by clear and specific goals and thereby guided by TDP, most online marketing research that studied the MEE has focussed on bottom-up processing (BUP). BUP is, in opposite to the goal-directed TDP, stimulus-driven, which means that a person’s attention is spontaneously directed towards an oncoming stimulus (Hahn, Ross, & Stein, 2006). For online consumers, BUP takes place when they are browsing the web without a specific goal in mind, a state called exploratory search (Janiszewski, 1998). Exploratory search appears to be highly encouraged in advertising experiments (Duff & Faber, 2011). Participants in online advertising studies are often instructed to ‘browse as they normally would’ or ‘look at whatever catches their
attention’ (Choi & Rifon, 2002; Lee & Ahn, 2012; Rifon, Choi, Trimble, & Li, 2004; Perkins & Forehand, 2012; Wang, Shih, & Peracchio, 2013).Others that have tried to imitate a goal-directed environment did so by asking their participants to evaluate a specific website as a whole (Chatterjee, 2008; Moore, Stammerjohan, & Coulter, 2005; Yoo, 2008).
The scarcity of studies that measured the effects of online ads during clear and specific goal-directed searches may have brought about a bias in MEE research results because perceived goal impediment has been identified as the most important reason for ad avoidance (Cho & Cheon, 2004; Speck & Elliot, 1997). According to Janiszewski (1998), online consumers in exploratory search are short of motivation to search efficiently. Since online consumers in
10
exploratory search do not have a clear goal in mind (Janiszewski, 1998), they also do not feel the need to actively inhibit certain stimuli on a website because these do not serve as
distractors. In these situations, the ads will not receive a negative affective evaluation, and a MEE is likely to take place through familiarity effects (Zajonc, 1968). These findings may, however, only apply to people that are in exploratory search. On the other hand, when people are engaged in a more goal-directed search task they may try to actively ignore the ads because they serve as distractors. As a consequence, a DEE may take place, and the brands from these ads may receive a more negative affective evaluation. In current marketing practice, no distinction is being made between consumers’ processing modes, while negative effects of exposures to advertisements may be expected for people that are driven by clear and specific goals.
The following hypothesis is proposed:
H1: Exposure to an online banner ad during goal-directed tasks results in a more negative
affective evaluation of the brand than exposure to an online banner ad during exploratory search.
The moderating role of ad location
A factor that might moderate the effects of search mode on brand attitudes is ad location. In earlier DDE-research, the location of a distractor stimulus has been found to affect inhibition and subsequent evaluation (Duff & Faber, 2011; Martiny-Huenger et al., 2014; Raymond et al., 2005). Raymond et al. (2005) found that the proximity of the distractor to the target affects distractor evaluations. They found across three different experiments that distractors observed near a target are rated more negatively compared to distracters observed further
11
away from a target. They argued that distractors located close to the target produce stronger interference and, therefore, require stronger inhibition. Martiny-Huenger et al. (2014) further corroborated the findings of Raymond et al. (2005) by demonstrating that stronger distractor interference indeed increases DEE’s. In addition, Duff and Faber (2011) could demonstrate that the magnitude of a DDE can also be influenced by target-distractor proximity in the domain of advertising. In their study, the participants produced stronger DDE’s for distractor ads that were located near the target content. For the present study, these findings suggest that online ads that are located at the centre of a website’s primary content are likely to be rated more negatively compared to ads further away from a website’s primary content by people that are engaged in a goal-directed task. However, an opposite effect may be expected for people that are engaged in exploratory search. Online ads that are located at a more prominent place, such as at the centre of a website’s primary content, are likely to receive more attention and, thereby, processing compared to ads located further away from a website’s primary content (Resnick & Albert, 2013). This additional processing is likely to result in more positively affective evaluations because it results in an increase in processing fluency.
The following hypothesis is proposed:
H2: Location of the ad will moderate the effects of search mode on brand attitudes, such that
an ad placed in the centre of a website’s primary content will lead to lower brand attitudes compared to an ad placed further away from the primary content during goal-directed tasks, while during exploratory search an ad placed in the centre will lead to higher brand attitudes compared to an ad placed further away from the primary content.
12
The moderating role of repetition
A second factor that may moderate the effect of search mode on brand attitude is repetition. Repeated exposure has been found to increase MEE’s (Bornstein, 1989; Zajonc, 2001). This is in line with the processing fluency hypothesis (PFH), which suggests that repetition increases the processing fluency of a stimulus. However, the effects of repetition on DDE’s remain more unclear. Most studies investigating the DDE have been focussed on presenting unique stimuli, thereby ignoring the potential role of repetition. The studies by Panagiotidou and Vivas (2009) and Janiszewski et al. (2013) are among the very few that have investigated the effects of repetition in the context of the DDE. The results of both studies indicate that repetition results in a more negative affective evaluation of distractors. Panagiotidou and Vivas (2009) found that repeated distractor faces were evaluated more negatively compared to unique distractor faces, while Janiszewski et al. (2013) found that repeatedly ignored products were more often ignored at a subsequent encounter compared to products that were ignored only once. A possible explanation for these findings is that repeated distractors result in stronger inhibition and, consequently, result in greater devaluation compared to unique distracters. However, this has not been investigated yet.
Based on the findings described above the following hypotheses are proposed:
H3: Repetition of the distractor stimulus will moderate the effects of search mode on brand
attitude, such that a high repetition will lead to lower attitudes towards the brand during goal-directed tasks and higher attitudes during exploratory search.
13
Study overview
All in all, the present study endeavours to get a better understanding of how being exposed to online banners ads affects the evaluation of the brands from these ads under a variety of conditions. First, the present study looks at how differences in search mode may bring about different effects of online banners ads on brand attitudes, such that exposure to online banner ads leads to higher brand attitudes through the occurrence of a MEE during exploratory search, but to lower brand attitudes through the occurrence of a DDE during goal-directed search tasks (H1). Secondly, the interaction between search mode and the location of the ad was investigated by predicting that both DDE’s and MEE’s will be the strongest for ads placed in the centre of the primary content (H2). Finally, the present study addressed the moderating role of ad repetition by hypothesizing that more exposures to the same ad will lead to stronger DDE’s and MEE’s (H3 & H4).
Method
Design
To test the hypotheses an online experiment was conducted. The experiment was created and conducted with the online tool Qualtrics. The design of the study was a 2 (Search mode: goal-directed vs. exploratory search, between-subjects) x 2 (Ad location: centre of primary content vs. away from primary content, between-subjects) x 3 (Ad repetition: 0 vs. 1 vs. 6, within-subjects) post-test only design. See Table 1 on the next page for an outline of the design:
Sample
The link to the online experiment with a request for participation was disseminated
throughout the personal network of the researcher via email and Facebook. In the requests (see Appendix A) was explained that the online experiment would involve seven 12-second
14 Table 1:
Outline of the experimental design with Search mode and Ad location as between-subjects variables and Ad repetition as within-subjects variable.
Ad location Ad repetition Search mode 0 1 6 Goal-directed Exploratory search Centre of primary content Away from primary
content Centre of primary
content Away from primary
content C1 (n = 38) C2 (n = 38) C3 (n = 40) C4 (n = 41) C1 (n = 38) C2 (n = 38) C3 (n = 40) C4 (n = 41) C1 (n = 38) C2 (n = 38) C3 (n = 40) C4 (n = 41)
Note: n refers to the number of participants in each condition.
tasks followed by a few questions, and that people could only participate on a laptop or desktop computer (no mobiles or tablets). The study was conducted in English, and the participants gave their consent before participating. A total of 322 people were contacted of which 209 participated in the study. People that did not fully complete the study (n = 50) were excluded from the final sample. Additionally, two participants were removed from the sample due to missing values.
The final sample consisted of 157 participants aged 17 to 73 (M = 28.58; SD = 10.46). It has to be noted though that 79 per cent of them were between the ages of 21 and 29. Males (59%;
n = 93) were slightly overrepresented in the sample compared to females (41%; n = 64). Most
of the participants (80%; n = 125) were currently residing in the Netherlands. The most common other countries where participants indicated to live in were Austria (3%; n = 4), Germany (3%; n = 4), Australia (2%; n =3), and the United Kingdom (2%; n = 3). The
15
majority of the sample was highly educated, since 81 per cent (n = 127) indicated to have completed a Bachelor’s degree or higher. Nearly everyone had at least completed high school (99%; n = 156). The precise frequencies of the age, countries of residence, and education levels of the participants are outlined in Appendix B.
Stimulus materials
The stimulus materials in the present study consisted of manipulated versions of seven web pages, which each contained information related to travelling to Colombia and an ad.
Participants were asked to imagine they were planning a trip to Colombia and had decided to look up information about the country online. The exact instructions for the participants read as follows:
Please read these instructions carefully.
You are participating in a study that investigates people's online reading comprehension ability and website navigability.
Scenario:
Imagine you are planning a trip to Colombia and decided to look up some information about the country online.
Task:
After continuing, you will be presented seven web pages containing information related to travelling to Colombia. During each web page visit, you will have to look up some specific information (to participants in goal-directed condition)/Please browse freely as you normally would (to participants in exploratory condition). You have twelve seconds to explore each page.
Below it is described how the web pages and the ads were selected for the present study.
Selection of stimulus materials: Web pages
The researcher selected the web pages during several topic-relevant search tasks on Google. To increase the study’s external validity, only web pages that were displayed on the first page
16
of Google’s search results were used. This way it was highly likely that these websites would also be used in real life by people searching for the same kind of information. All of the web pages contained information in English. The topics of the web pages included information about visas, vaccinations, safety issues, travel costs, Spanish lessons, and two travel
destinations in Colombia. In order to mimic a real browsing experience, all of the web pages were extracted from a different website and, therefore, had different layouts, complexity levels, and lengths. A full overview of the seven websites eventually used in the study can be found in Appendix C.
The web pages were extracted at full size via the Full Page Screen Capture plug-in on Google
Chrome. They were then converted to a width of 1100px to ensure they would fit on all
desktop and laptop computer screens while maintaining clear readability. In Qualtrics, custom CSS was added to the ‘Faculty of Social and Behavioural Sciences Theme’ in order to
increase the width of the survey to 1100px so that the web pages would be fully visible.
Selection of stimulus materials: Ads
In order to develop ads for the study, eight existing brand logos were initially chosen via The
Branding Source1, a blog featuring the logo changes of brands from all over the world ranging from vintage to very new brands. To maximise the likelihood of the brands being unknown to the participants, the brands that were chosen originated from countries that were highly likely to be unrepresented in the sample. This was confirmed by the results from the main study, where none of the participants indicated to be currently residing in a country where one of the stimulus brands originated from.
17
A pilot study (see Appendix D) was conducted amongst 24 undergraduates aged 19 to 28 (M = 23.10; SD = 1.95; 58 % females) to select the final brands. The participants were first asked if they knew any of the nine brands and their logos presented to them. If they answered ‘yes’, they were asked if they could indicate what kind of product the brand represented. Only two of the participants indicated to know one of the brands. One participant indicated that she knew the brand Oi, although she could not tell which product the brand represented. Another participant confused the brand Seta, a Finnish LGBT rights organisation, with the multi-vitamin brand Centrum. The other six brands and their logos remained unrecognised by all of the participants. Three of these were picked as the final brands, namely: 1) Kaktus: a mobile telecommunications brand from the Czech Republic, 2) Aleris: a private healthcare provider from Sweden, and 3) Orkla: a holding company that owns a portfolio of consumer food and hygiene brands from Norway.
Adobe Photoshop CC 2017 was used to design the ads. In order to ensure that possible
findings couldn’t be attributed to differences between ads, each brand was used for each ad type. A total of nine ads were thus developed and converted to 336x280px, a format that is used regularly and is considered to be amongst the most effective ad sizes by Google (Google AdSense, 2016). An overview of the nine ads used in the study can be found in Appendix E.
Manipulated variables
Independent variable: Search mode
The manipulation of ‘Search mode’ concerns whether participants would be involved in goal-directed versus exploratory online search. Each participant read the same cover story where they were asked to imagine they were planning a trip to Colombia and had decided to look up information about the country online. Depending on the search mode condition they were in,
18
participants were either told they had to look up specific information during each website visit (goal-directed condition) or instructed to browse freely as they normally would (exploratory condition). Each participant was also told they had twelve seconds to explore each page.
The participants in the goal-directed condition were presented with a question before each web page visit. This question asked the participants to answer a specific question such as: “What are the costs of a visa-extension of 90 days in Colombia?” The answer to this question could be found on the web page that was subsequently shown. This web page would be visible for twelve seconds before automatically continuing to an answer page where the related question was displayed accompanied by an open-answer box. Upon clicking continue, a new question appeared on screen to which the answer could be found on the next web page to be shown. This procedure was followed for all of the seven web pages. All questions and answer options can be found in Appendix F.
The participants in the exploratory condition saw no specific questions, but were merely asked to browse each web site freely as they normally would when planning a trip abroad. Just as in the goal-directed condition, participants in the exploratory condition were also shown each web page during twelve seconds in total.
A second pilot study was conducted amongst 30 participants aged 21 to 30 (M = 24.07; SD = 1.95; 63% females) to test whether the participants in the goal-directed condition experienced their behaviour during the tasks as more goal-directed compared to the participants in the exploratory condition. This was measured with the question: ‘To what extent did you feel that your behaviour while browsing through the websites was goal-directed?’ on a 5-point Likert scale ranging from 1 (not goal-directed) to 5 (extremely goal-directed). The results from an
19
independent-samples t-test suggest that the participants in the goal-directed condition (M =
4.53; SD = 0.64) indeed experienced their behaviour as more goal-directed compared to the participants in the exploratory condition (M = 3.07; SD = 1.22), t (28) = 4.12, p < 0.001, 95% CI [0.74, 2.20]. Additionally, the pilot study aimed to test whether fifteen seconds was an appropriate amount of time for the completion of each task. Ideally, the participants in the goal-directed condition should have enough time to answer the questions without being left with a lot of time to study the page afterwards. Because most of the participants in the goal-directed condition were able to answer the questions correctly and rated the tasks as easy (M = 2.36; SD = 0.84) on a scale ranging from 1 (extremely easy) to 7 (extremely difficult), it was decided to lower the response time to twelve seconds for the main study.
Independent variable: Ad location
The manipulation of ‘Ad location’ concerns whether the ad on the web page was placed at the centre of the web page’s content where the answer to the question could be found (primary content) or away from the primary content. Depending on the condition the participants were assigned to, the ads appeared either in the centre of the web page content or further away at the right of the web page content. An overview of where the ads were located on each web page can be found in Appendix C.
Independent variable: Ad repetition
The manipulation of the variable ‘Ad repetition’ concerns the number of times the
participants were exposed to the same branded ad during the web page visits. Each web page included one ad, so every participant was exposed to seven ads in total. For each participant, one brand was never shown, one brand was shown once, and one brand was shown six times during the web page visits. Further, as outlined in Appendix G, the assignment of the brand
20
types and ad types to repetition levels was randomised so that possible effects of the
experimental manipulations could not be attributed to a-priori differences between repetition levels in the type of brands or ads the participants were exposed to.
Control variables
Control variable: Prior visits to Colombia
Two variables were included as control variables. The first variable, Prior visits to Colombia, measured whether the participants had visited Colombia before via a yes or no question. Having visited the country already might result in a different experience of the web pages and was, therefore, controlled for. Thirteen participants (8.3%) answered ‘yes’ to this question.
Control variable: Use of ad blockers
The second control variable, Personal use of an ad-blocker, concerned the participants’
personal use of ad blockers. An ad blocker is a filtering software that automatically blocks ads from websites, which is becoming increasingly popular. According to eMarketer (2016), 29 per cent of the Internet users in the US will use ad blockers in 2017. People that are used to seeing ads on websites regularly might respond differently to ads compared to people that generally block them. Therefore, in order to control for this possible bias, the participants were asked to indicate if they have an ad blocker installed on their browsers. The majority of the participants (64 %; n = 101) indicated to make use of ad blockers when browsing online.
Measures
Brand attitude
The Brand attitudes (BA) of the participants were measured by asking the participants to describe their overall feelings about the brands ‘Aleris, ‘Orkla’, and ‘Kaktus’. The order in
21
which the brands were rated was randomised. The participants rated the brands on three items that have been identified by Spears and Singh (2004) to be the most valid for measuring brand attitudes. The items were measured via 9-point semantic differentials, ranging from 1
(unappealing/unfavourable/unpleasant) to 9 (appealing/favourable/pleasant). Before rating the brands the participants were asked to simply rate the brands based on their ‘initial gut reaction’. The scale proved to be very reliable (α = .92). The general BA of the participants was slightly positive (M = 5.00; SD = 1.00).
Manipulation checks
The manipulation check item ‘Perception of goal-directedness of own behaviour’ measured whether the participants actually experienced their behaviour during the goal-directed tasks as more goal-directed compared to the exploratory search tasks. People were asked to rate to which extent they felt their behaviour during the tasks was goal-directed on a 5-point Likert scale ranging from 1 (extremely goal-directed) to 5 (not goal-directed). Also, in order to verify whether the manipulation for Ad location was successful, the participants were told that during each web page visit they were exposed to an ad and asked to indicate whether they thought that the ads were located more around the centre of the webpage and among the primary content (answer option 1) or placed on the right side of the web page and further away from the primary content (answer option 2).
Procedure
After the participants clicked on the participation link, they were first asked to confirm their participation consent. Next, they were given instructions. They were told again that the study could only be completed on a laptop or desktop computer (no phones or tablets). Also, they were urged to enlarge the window of their browsers to maximum size. After confirming this
22
they were able to continue to be randomly assigned to one of the conditions (see Appendix G). Each participant read the same cover story asking them to imagine they were planning a holiday trip to Colombia and had decided to look up information about the country online. They were told they would be presented seven web pages containing information related to travelling to Colombia. Depending on the Search mode condition they were assigned to, they were asked to look up specific information during each website visit (goal-directed condition) or asked to browse freely as they normally would (exploratory search condition). Each
participant was told they had twelve seconds to explore each page. After finishing the tasks the brand attitudes were measured, followed by the manipulation check items, the control questions, and the items measuring demographics. After completing the questionnaire, the participants were thanked for their participation and made aware that they could leave their email addresses in order to receive information about the study’s real purpose and its results.
Results
Randomization
To ensure that differences between conditions in their attitudes towards the brands cannot be attributed to a priori differences in the demographic composition or the type of ads and brands shown to experimental groups, randomization checks were conducted for the following demographic properties: Age, Gender, and Level of education, as well as for the different types of brands and ads used for the brand that was not shown, the brand that was shown once, and the brand shown six times during the web page visits. A detailed report of the results of the randomization checks can be found in Appendix H. In summary, the results suggested that Gender and the types of brands and ads used for each repetition level were equally distributed across conditions, so that any effects of the experimental manipulations can’t be attributed to a-priori differences in the gender composition of the groups or the type
23
of brands and type of ads the participants were exposed to. However, the results also showed an unequal distribution for Age and Level of education across conditions. Therefore, Age and Level of education were included as control variables for the main analyses.
Manipulation checks
To check whether the participants in the goal-directed conditions actually perceived their behaviour during the web page visits as more goal-directed compared to the participants in the exploratory conditions, an independent-samples t-test was conducted with Search mode as independent and Perception of goal-directedness of own behaviour as dependent variable. A significant effect of Search mode was found, t (155) = 8.87, p < 0.001, 95% CI [1.05, 1.65]. The participants in the goal-directed condition (M = 1.72; SD = 0.83) perceived their
behaviour during the web page visits as significantly more goal-directed (MD = 1.35) compared to the participants in the exploratory condition (M = 3.07; SD = 1.06). This result confirms that the manipulation for Search mode was successful.
To check whether the manipulation for Ad location was also successful, a Chi-square test was conducted with Ad location and Perceived location of the ads as variables. The results showed that there was a significant difference between the two groups regarding the perceived
location of the ads they were exposed to, χ2
(1, N = 157) = 24.90, p = <0.001, V = 0.40. As expected, the group that was exposed to ads in the centre of the web pages indicated more often that the ads were located in the centre, while the group that was exposed to ads at the right side of the web pages more often indicated the opposite (see Table 2). Thereby, the results suggest that the manipulation for Ad location was successful.
24 Table 2:
Results of the manipulation check for Ad location.
Ad location
Perceived location of ads
Centre Right
Centre 45 33
Right 15 64
Main analyses
The effects of search mode and ad location on Brand attitude
To test both H1 about the effect of Search mode on Brand attitude and H2 about the interaction between Search mode and Ad location on Brand attitude, a MANOVA was conducted with Search mode (goal-directed vs. exploratory) and Ad location (centre of primary content vs. away from primary content) as between-subjects factors. Mind that because Ad repetition was not included in this analysis, no distinction between repetition levels was made. The level with zero repetition was, therefore, renamed Brand attitude for ad that was not shown (BAN), while the levels with one and six repetition were computed into a new variable called Brand attitude for ads that were shown (BAS). As such, Brand attitude for ads that were shown (BAS) and Brand attitude for the ad that was not shown (BAN) were the dependent variables in the MANOVA, and Age, Level of education, Prior visits to Colombia, and the Personal use of an ad-blocker were control variables.
First, a small significant main effect of Search mode on BAS was found, F (1, 149) = 7.11, p = 0.028, η2
= 0.03. As expected in H1, the participants in the goal-directed condition (M = 4.87; SD = 1.19) evaluated the brands they were exposed to significantly more negative compared to the participants in the exploratory condition (M = 5.28; SD = 1.18). This confirms that the brands in ads were evaluated more negatively when the ads interrupted the goal-directed task than when they were encountered during the exploratory search. Further, no
25
significant main effect of Search mode on BAN was found, F (1, 149) = 1.37, p = 0.517. The fact that the difference between the groups was only significant for the exposed ads (BAS) and not for ads not exposed to (BAN) shows that the manipulation of Search mode did not lead the participants to rate ads in general as more positive or negative, but that it was the goal interruption per se that led to a more negative brand evaluation. Further, Table 3 shows that the differences in BAS between conditions are mainly a consequence of an increase of BAS in comparison to BAN in the exploratory condition (MD = 0.51). The difference between the BAS and BAN was only minimal (MD = -0.05) in the goal-directed condition. Because the BAS in the goal-directed condition was indeed significantly lower than the BAS in the exploratory search condition H1 was supported.
Table 3:
Means and standard deviations of the main effect of Search mode.
Search mode Brand attitudes Unexposed Exposed Goal-directed M = 4.92 SD = 1.82 M = 4.87 SD = 1.19 M = 5.28 SD = 1.18 Exploratory search M = 4.77 SD = 1.77
Contrary to the expectations in H2, no significant interaction effect between Search mode and Ad location on BAS was found, F (1, 149) = 1.14, p = 0.287. Although, as predicted by H2, the participants in the goal-directed condition with ads away from the primary content (M = 4.98; SD = 1.22) rated the brands slightly more positive compared to the participants in the goal-directed condition with ads in the centre (M = 4.76; SD = 1.16), while the participants in the exploratory condition with ads away from the primary content (M = 5.22; SD = 1.28) rated the brands slightly more negative compared to the participants in the exploratory condition with ads in the centre (M = 5.35; SD = 1.09), this interaction between Search mode and Ad
26
location was not significant. This means that the brands in the ads were evaluated more negatively overall when they interrupted the goal-directed online task than when they were encountered during an exploratory search, independent from whether the ad was located at the centre of the web page’s primary content or away from the web page’s primary content. No other significant effects were found.
The moderating role of ad repetition
In order to test H3 and H4 an ANOVA with repeated measures was conducted. Search mode (goal-directed vs. exploratory) and Ad location (centre of primary content vs. away from primary content) were between-subjects factors in this ANOVA, Ad repetition (0 vs. 1 vs. 6 repetitions) was a repeated measures factor, and Age, Level of education, Prior visits to Colombia, and the Personal use of an ad-blocker were control variables.
The interaction between Search mode and Ad repetition proved marginally significant, F (2, 149) = 8.59, p = 0.071, η2 = 0.02) and became conventionally significant when tested against a linear contrast model, F (1, 149) = 4.92, p = 0.028, η2 = 0.03. Table 4 shows that one exposure to the ad did not impact the brand attitudes in the goal-directed versus exploratory condition. However, six exposures to the ad did increase the brand attitude for the participants in the exploratory condition but not for the participants in the goal-directed condition. These findings confirm that for the participants in the exploratory condition a mere exposure effect (MEE) occurred in that six exposures led to a higher brand attitude compared to one or zero exposures. For the participants in the goal-directed condition no distractor devaluation effect (DDE) or MEE was observed, since the brand attitudes at one or six exposures did not increase or decrease compared to zero exposures. H3 is thereby partly supported.
27 Table 4:
Means and standard deviations of the interaction between Search mode and Ad repetition.
Search mode Ad repetition 0 1 6 Goal-directed M = 4.92 SD = 1.82 M = 4.84 SD = 1.72 M = 4.90 SD = 1.83 Exploratory search M = 4.77 SD = 1.77 M = 4.95 SD = 1.70 M = 5.61 SD = 1.74
Finally, a marginally significant three-way interaction effect between Search type, Ad
location, and Ad repetition was found, F (2, 149) = 2.53, p = 0.083, η2 = 0.03. Table 5 shows, as predicted by H4, that in the exploratory condition the MEE was strongest for participants exposed to ads in the centre of the primary content after six exposures. Conversely, the participants in the goal-directed condition that saw ads in the centre of the primary content rated the brands most negatively after six exposures, which is also in line with H4. Thus, participants who were merely exploring the web pages evaluated the brands in the ads more positively after multiple exposures to the ad, particularly when the ads were located at the centre of a web page’s primary content. The participants with a specific search goal in mind on the other hand evaluated the brand in the ads more negatively after multiple exposures to the ad, particularly when the ads were located at the centre of a web page’s primary content.
Even though the above results seem in line with H4, it should be noted that the expected effects occurred after six exposures to the ads only, while after a single exposure actually opposite effects seemed to take place in both the goal-directed as the exploratory condition. The interaction effect found here hence was more quadratic than linear. This was confirmed by the contrasts analysis, which showed no significant effect of the interaction when tested against the expected linear model, F (1, 149) = 0.624, p = 0.431, but did show a significant
28
effect when tested against a quadratic model, F (1, 149) = 4.675, p = 0.032, η2 = 0.03. H4 therefore was not fully supported.
Table 5:
Means and standard deviations of interaction between search mode, ad location, and repetition
Search mode Ad location
Repetition
0 1 6
Goal-directed
Centre of primary content
Away from primary content M = 4.71 SD = 1.79 M = 5.12 SD = 1.85 M = 4.93 SD = 1.59 M = 4.75 SD = 1.86 M = 4.59 SD = 1.73 M = 5.21 SD = 1.89 Exploratory search
Centre of primary content
Away from primary content M = 4.88 SD = 1.69 M = 4.67 SD = 1.87 M = 4.77 SD = 1.72 M = 5.12 SD = 1.70 M = 5.93 SD = 1.60 M = 5.32 SD = 1.83
Discussion
The effect of search mode on brand attitude
This study aimed to get a better understanding of the effects that being distracted by online banners ads brings about on the evaluation of the brands from these ads. The study’s main focus was examining whether people engaged in a search task with clear and specific goals respond differently to online banner ads than people in an exploratory state of mind. The first hypothesis, H1, stated that participants stimulated to browse through web pages with a goal-directed mind-set would evaluate the brands they were exposed to significantly more negative compared to the participants stimulated to browse freely. This expectation was supported. The results showed that the participants in the goal-directed condition indeed rated the brands that were shown more negatively than the participants in the exploratory condition.
29
It has to be noted though that the difference in brand evaluation between experimental
conditions was mainly caused by an increase in brand evaluation in the exploratory condition and not so much by a decrease in evaluations in the goal-directed condition. In line with previous attention studies (Hahn et al., 2006; Janiszewski et al., 1998; Resnick & Albert, 2014), it was expected that giving people instructions to browse freely would increase the attention paid to and reduce the active inhibition of online banner ads by encouraging bottom-up processing. Hereby, encountering the ads was expected to result in more positive brand evaluations as a consequence of an increase in processing fluency (Hansen & Wänke, 2009; Leynes & Adante, 2016; Moreland & Topolinski, 2010). The results confirmed these expectations by demonstrating that a mere exposure effect (MEE) took place in the exploratory search condition in that being exposed to an ad led to a significant increase in brand attitude. Conversely, it was argued that giving people clear and specific tasks would evoke a goal-directed mind-set and encourage top-down processing (Carrasco, 2011). This was expected to increase people’s active inhibition of online banner ads because these would distract them from their immediate goals (Fenske & Raymond, 2006). The act of inhibiting these banners ads would in turn be likely to become associated with the brands from these ads, resulting in a tendency to devalue them during a subsequent encounter: a distractor devaluation effect (DDE) (Doallo et al., 2012; Fragopanagos et al., 2009; Kiss et al., 2007). At a first glance the results in the goal-directed condition seem to suggest that a DDE did not occur, since no significant increase or decrease in brand evaluation was observed. However, the absence of a significantly more negative brand evaluation in the goal-directed condition does not necessarily imply that a distractor devaluation effect (DDE) did not occur. Several studies that observed similar results (Goolsby et al., 2009; Huang & Hsieh, 2013; Kihara et al., 2012) explained their findings with the possibility of a simultaneous occurrence of a MEE and DEE. Fragopanagos et al. (2009) elaborated on this possibility by presenting a neural
30
network model that outlines how a DDE and a MEE may influence affective responses through independent mechanisms. Kihara et al. (2012) took this one step further by arguing that a DDE and a MEE may even influence affective responses interactively. For the present study the explanations provided by Frapanagos et al. (2009) and Kihara et al. (2012) suggest that in the goal-directed condition the positive effects of a MEE might have largely levelled out the negative effects of a DDE. Since the MEE and the DDE seem to be activated through different mechanisms (Fenske & Raymond, 2006; Moreland & Topolinski, 2010) that might not necessarily exclude one another, the possibility of a simultaneous occurrence of both effects seems plausible. Most importantly, this explanation accounts for the significant differences that were found between targets and distractors in the present study as well as in earlier research with similar findings (Goolsby et al., 2009; Huang & Hsieh, 2013; Kihara et al., 2012). Nevertheless, further studies that are able to identify whether and how both effects might occur simultaneously are necessary in order to validate such assumptions. If it turns out that both effects can exert their influences simultaneously, it would be interesting to identify moderators that are likely to strengthen or weaken one effect with respect to the other.
Regardless of the mechanism at hand, it appears that people’s search mode while coming across websites is key to the effectiveness of banner advertisements. The results imply that online banner ads are more effective at increasing consumers’ attitudes towards brands when consumers are not guided by clear and specific goals, but are rather browsing the web more spontaneously. The present study hereby identifies an important boundary condition for the formation of mere exposure effects, namely not being driven by clear and specific goals. This seems to be an extremely valuable addition to existing MEE-theory that may also have important implication for advertisers.
31
The moderating role of ad location
The second aim of this study was investigating the possible moderating role of the location of the ad on a web page. In H2 was hypothesised that participants in the goal-directed search condition would evaluate the ads placed at the centre of a website’s primary content more negatively compared to the ads placed further away, while for participants in the exploratory condition opposite effect was expected. Unfortunately, these expectations were not supported. No significant differences between the experimental conditions were found.
A possible explanation for the absence of a significant decrease in brand attitudes in the goal-directed condition with ads placed in the centre of the primary content may be that the distance between the ad and the primary content was not large enough and the ad, therefore, still rather strongly inhibited attention. According to the selective tuning model (Cutzu & Tsotsos, 2003), attention is actively inhibited in the immediate vicinity of an attended
stimulus. Although, in the present study, the ads in the condition with ads placed further away were placed at the right side of each web page and thus somewhat further away from the primary content compared to the ads placed in the centre, they were still placed at the same height. As a consequence, these ads were visible at all times while looking at the correct information during the information search tasks. It may be that the ads were still located within the participants’ suppressive ring of inhibition as described by the selective tuning model (Cutzu & Tsotsos, 2003). The differences in brand evaluations between both groups might have been larger and potentially significant when the ads placed further away from the primary content were also displayed at a different height, bringing them further away from the target information and out of the immediate ring of inhibition. This is corroborated by earlier findings from Duff and Faber (2011), who manipulated the banner ads in their study to appear
32
either in the centre of a web page or at the upper-right corner and were able to find a stronger DDE for the ads placed in the centre than at the upper-right corner.
The second part of H2 predicted that the people in the exploratory condition would evaluate the ads placed in the centre of a website’s primary content more positively compared to ads placed further away. Resnick and Albert (2013) found that people engaged in exploratory search pay more attention to ads placed at the centre of a website’s primary content than ads placed further away. This supposed increase in attention is likely to result in more positive affective evaluations due to an increase in processing fluency (Moreland & Topolinski, 2010). However, contrary to the expectation, no significant differences in evaluations for Ad location were found in the exploratory condition. These results seem to contradict the conclusions of Resnick and Albert (2013), who argued that banner ads placed on the right side of a web page are subjected more strongly to banner blindness compared to ads placed in the centre. Yet, although the expectations put up by H2 were not confirmed, the results showed some differences between conditions in the predicted directions indicating a trend towards a significant effect. A larger sample size might have led to significant results. Hence, the possible moderating role of Ad location on the effects of Search type on brand attitudes should not be rejected yet but rather investigated more elaborately. For now, however, it must be concluded that the strength of the effects of Search mode on brand evaluations is not unequivocally subject to the location of the ad.
The moderating role of ad repetition
Finally, this study investigated the moderating role of Ad repetition. H3 predicted that a higher number of repetitions in exposures to the ad would increase the strength of both the DEE and MEE hypothesised in H1. These expectations were partly supported in the study in
33
that the difference in brand attitudes between the search mode conditions was only significant after six exposures to the ad, due to an increase in brand attitudes for the participants in the exploratory condition. It appears that in the exploratory search condition being exposed to an online banner ad once is not sufficient for a MEE to occur. This contradicts the earlier findings of Zajonc, Shaver, Travis, and van Kreveld (1972), which suggest that an MEE is observed even for stimuli presented only once. No increase or decrease in brand attitudes was observed for the goal-directed condition after one or six exposures to the ad in the present study. As mentioned in the discussion for H1, here the absence of any effects of repetition in the goal-directed condition might also be explained by a simultaneous occurrence of a DDE and a MEE (Frapanagos et al., 2009; Kihara et al. 2012), where the positive effects of a MEE might have levelled out the negative effects of a DDE.
The expected linear interaction between Search type, Ad location and Ad repetition in H4 proved to be quadratic. After six exposures, Ad location seemed to moderate the effects of Search type on Brand attitude as expected, in that ads placed in the centre were evaluated more negatively compared to ads placed further away from the centre in the goal-directed search condition, but more positively in the exploratory search condition. Surprisingly at a single exposure opposite effects were observed, although the differences were smaller here. It might be that the unexplainable opposite scores observed at a single exposure are merely random variations. However, before jumping to any conclusions, further research is required. For now, H4 could not be not fully supported.
Limitations
A number of limitations to the present study should be recognised. First, in order to make sure that the participants in the goal-directed condition were exposed to the web pages for an equal
34
amount of time compared to the participants in the exploratory condition, timers were set so that the web pages would be visible for a limited amount of time during the online tasks. This does not precisely simulate a natural browsing experience, where people decide for
themselves how much time they want to spend on a page. Also, it seems probable that the people in the exploratory condition would feel that twelve seconds were not sufficient to get a good impression of the web page’s contents. Because they did not know what to look for, it is likely that they wanted more time to explore the pages, and this could have led to some
frustration and influenced the results. Unfortunately, frustration was not measured in the study and could, therefore, not be controlled for. Finally and most importantly, the effect sizes of the significant main effect of Search mode and the significant interaction between Search mode and Ad repetition were quite small (η2
= 0.03). This emphasises the need for future research with larger sample sizes.
Academic and practical implications
The findings from the present study have important implications for the existing theory. Earlier studies have concluded that the mere exposure effect (MEE) is robust and reliable (Bornstein, 1989; Zajonc, 2001). In online advertising research, the MEE is, therefore, a popular concept that is often referred to in order to explain and make generalizations about the positive effects of exposure to advertising. Most online advertising studies did, however, not consider consumers’ attentional goals during the exposure to advertising. People are often driven by clear goals while browsing online. Highly goal-oriented uses such as information seeking and information distribution have been identified as the most important gratifications the Internet fulfils for its users (Blank & Groselj, 2014; Stafford, Stafford, & Schkade, 2004). Nonetheless, in earlier advertising experiments, participants have been highly encouraged to browse through websites without clear and specific goals (Duff & Faber, 2011). The results
35
from the present study suggest that the positive effects of mere exposure to advertising are eliminated when consumers do have such clear and specific goals. Failing to acknowledge the importance of consumers’ attentional goals during the exposure to advertising in advertising research may result in a biased view of advertising effectiveness. The possibility of negative effects of advertising due to distractor devaluation effects (DDE) for goal-directed consumers should, therefore, be taken into consideration and elaborated upon in future studies. It would be valuable to investigate in which specific online browsing situations DDE’s and/or MEE’s are more likely to occur, and whether they are able to interact. Furthermore, it would be interesting to measure how both effects hold over time. When validated in future research, the present findings have important implications for advertising practice. When advertising campaigns are aimed at increasing consumers’ affective brand responses, it would be wise for advertisers to choose a platform, app, or website that is often visited without a specific goal in mind (e.g. Facebook) or choose a cost-per-click business model instead. Advertisers would have to acknowledge that not all exposure is good exposure.
36
References:
Belch, G. E. & Belch, M. A. (2015). Advertising and promotion: An integrated marketing
communications perspective. New York: McGraw-Hill.
Blank, G. & Groselj, D. (2014). Dimensions of Internet use: amount variety, and types.
Information, Communication & Society, 17(4), 417-435.
Bornstein, R. F. (1989) Exposure and affect: overview and meta-analysis of research, 1968– 1987. Psychological Bulletin, 106, 265–289
Carrasco, M. (2011). Visual attention: The past 25 years. Vision Research, 51(13), 1484-1525.
Chatterjee, P. (2008). Are unclicked ads wasted? Enduring effects of banner ads and pop-up ad exposures on brand memory and attitudes. Journal of Electronic Commerce Research,
9(1), 51-61.
Cho, C-H. & Cheon, H. J. (2004). Why do people avoid advertising on the Internet? Journal
of Advertising, 33(4), 89-97.
Choi, S. M. & Rifon, N. J. (2002). Antecedents and consequences of web advertising credibility: A study of consumer response to banner ads. Journal of Interactive Advertising,
3(1), 12-24.
Cutting, J. E. (2003). Gustave Caillebotte, French Impressionism, and mere exposure.
Psychonomic Bulletin & Review, 10(2), 319-343.
Cutzu, F. & Tsotsos, J. K. (2003). The selective tuning model of attention: psychophysical evidence for a suppressive annulus around an attended item. Vision Research, 43(2), 205-219.
Doallo, S., Raymond, J. E., Shapiro, K. L., Kiss, M., Eimer, M., & Nobre, A. C. (2012). Response inhibition results in the emotional devaluation of faces: neural correlates as revealed by fMRI. Social Cognitive and Affective Neuroscience, 7(6), 649-659.
37
Duff, B. R. L. & Faber, R. J. (2011). Missing the mark. Advertising avoidance and distractor devaluation. Journal of Advertising, 40(2), 51-62.
Elliot, A. J. (2014). Advances in motivation science, volume 1. Cambridge: Academic Press.
eMarketer. (2016). US ad blocking to jump by double digits this year. Retrieved from https://www.emarketer.com/Article/US-Ad-Blocking-Jump-by-Double-Digits-This-Year/1014111?ecid=NL1001
Fenske, J. & Raymond, J. E. (2006). Affective influences of selective attention. Current
Directions in Psychological Science, 15(6), 312-316.
Fragopanagos, N., Cristescu, T., Goolsby, B. A., Kiss, M., Eimer, M., Nobre, A. C., Shapiro, K. L., Taylor, J. G. (2009). Modelling distractor devaluation (DD) and its neurophysiological correlates. Neuropsychologia, 47, 2354-2366.
Friestad, M. & Wright, P. (1994). The persuasion knowledge model: How people cope with persuasion attempts. Journal of Consumer Research, 21(1), 1-31.
Goolsby, B. A., Shapiro, K. L., Silvert, L., Kiss, M., Fragopanagos, N., Taylor, J. G., Eimer, M., Nobre, A. C., & Raymond, J. E. (2009). Feature-based inhibition underlies the affective consequences of attention. Visual Cognition, 17(4), 500-530.
Google AdSense. (2016). Optimization tips. Most successful ad sizes. Retrieved from https://support.google.com/adsense/answer/17955?hl=en.
Hahn, B., Ross, T. J., & Stein, E. A. (2006). Neuroanatomical dissociation between bottom-up and top-down processes of visuospatial selective attention. NeuroImage, 32(2), 842-853.
Hansen, J. & Wanke, M. (2009). Liking what’s familiar: The importance of unconscious familiarity in the mere-exposure effect. Social Cognition, 27, 161-182.
