Is the combination of personalization and repeated exposure a
fruitful way to circumvent advertising ineffectiveness?
An experimental study to the effect of personalisation and repeated ad exposures on attitude to and recall and recognition of advertised products
Narin Esmaeel | 10206027 | Persuasive Communication | M.s. Dr. Stephanie C. M. Welten | December 21st , 2018 | Graduate School of Communication University of Amsterdam | 6778 words
2
Abstract
This study investigated the effects of personalization and repeated exposure on the recall and recognition of the advertisement and attitude to the brand. Repeated exposure and
personalization are popular strategies amongst marketers, but little is known about their underlying mechanisms when they are combined. Marketers but also health specialists could use these insights to improve the persuasiveness of their messages.
The study has been conducted amongst 218 respondents and consisted out of a 2 (personalization: present or absent) x 2 (repeated exposure – two of four exposures to the ad) experimental design. The results showed that personalization had a large effect on attitude. This was not supported for the effect of repeated exposures or the interaction-effect between personalization and repeated exposure. Personalization is also shown to be a significant predictor of recall. This is not the case for repeated exposures to the advertisement. No significant predictions could be made from the model that included the interaction-effect of repeated exposures and personalization. Finally, neither personalization, nor repeated exposures, nor the interaction between personalization and repeated exposure were found to be a significant predictor of correct recognition of the ad. Therefore, support was not found for either the widely discussed mere-exposure effect, or the expected interaction between personalization and repeated exposure were found. Still, this study contributes to the scientific knowledge as it is one of the first to look into the increasingly used combination of
3
Introduction
As soon as we visit any website, we are asked to accept the ‘cookies’. Cookies are a small text file that websites set on our computer (Cofone, 2017; Englehardt, Reisman, Eubank, Zimmerman, Mayer, Narayanan & Felten, 2015). They function as your personal online identification card by keeping track of your subsequent behaviour on the web. The information that these cookies collect is automatically shared with the server that first gave you the cookie. The knowledge stored within these cookies enables improving the fit of the advertisements that are shown to you, which is important as we are flooded with
advertisements everyday (Palmer, 2005; Ayenson, Wambach, Soltani, Good & Hoofnagle, 2011).
Research by Wedel and Pieters (2015) suggests that on an average day we are exposed to at least 1000 advertisements that, both online and offline, are competing for our attention. However, due to this large number and our limited cognitive capacity (Hoeken, Hornikx & Hustinx, 2009) these advertisements often do not reach their goal of ensuring the advertised brand or product is recognized and recalled by the viewer. This is where online cookies can offer a great advantage by enabling the server to personalize the content that you are shown to your preferences. Through adapting the presented content to the receivers’ preferences, it enlarges the relevance and appeal of the advertisement for its viewer (Nasraoui, 2005). Therefore, it has a higher chance to be recalled and recognized and can lead to more positive evaluations of the advertisement.
Additionally, research has investigated repeated exposure as another way of circumventing ad ineffectiveness. However, insufficient research has related exposure to online personalized advertisements. Classic research findings (Anand & Sternthal, 1990; Bornstein & D'agostino, 1992) suggest that the more often one is exposed to an ad, the more likely it is that they will recall it and evaluate it positively. Although it is established that repeated ad exposure can elicit a significant increase of evaluation of the ad, findings also revealed that too much exposure could lead to a decline in the evaluations, therefore creating an inverted u-shape relation between exposure to an ad and its evaluation. Nonetheless, it remains ambiguous when that downturn takes place as research shows opposing results (Schmidt & Eisend, 2015; Montoya, Horton, Vevea Citkowicz & Lauber, 2017). Thus, personalized content can lead to more positive evaluations of that content because the personalization makes it more interesting for the receiver. However, when personalized content is repeated too often, a negative effect also appears to be at play. This negative effect can be caused by boredom with the stimuli because receivers are saturated
4
with its content. This negative effect is likely to appear earlier in personalized advertisements than in non-personalized advertisements because of the former demanding more attention. Due to the increase in attention to the advertisement, receivers will sooner become familiar with the message and understand what it’s trying to accomplish (Hoeken et al., 2009).
Additionally, the ad might be conceived by them as intrusive (Baek & Morimoto, 2013). Such intrusiveness can cause feelings of disturbance (Martin, 1992) and annoyance (McAuley & Leskovec, 2013). In turn, these feelings can be used as coping methods, which are likely to influence recall, recognition and evaluation of the ad negatively. In other words,
personalization could not only lead to positive effects through increasing personal relevance. But when combined with repeated exposure, it could also lead to negative effects instead. The importance of the combined effect of personalization and repeated exposure to advertisements becomes apparent when considering the ample number of papers that have looked into these separate effects. Especially studies on repeated exposure to advertisement have resulted in a large body of knowledge. Relatively recently, research has also taken an interest in the impact of personalization, as it has become an increasingly used tool to enhance engagement and to persuade consumers. A recent study suggests 75% of customers are
displeased with receiving general offers, and 81% even indicates that personal relevance plays the most leading role in purchases (Eagle Eye, 2018). And even though personalization and repeated exposure are increasingly being used and also combined as advertising techniques, knowledge is still limited about their underlying mechanisms. Studies on the combination of personalization and repeated exposure are limited. This study is one of the first to look into this interplay of techniques to increase understanding of the effect of personalization of, and repeated exposure to an advertisement on attitude to and recall and recognition of the
advertisement. Next to contributing to the aforementioned research gap, such findings can form an essential cornerstone for advertising but also health professionals, as they provide more information about how to most effectively influence and steer consumers or patients. More comprehensive knowledge about personalization, repeated exposure and their
interaction can improve the efficiency of persuasive messages as it can help give insights on what approaches are or are not effective. The research question therefore constitutes:
“To what extent is there an effect of personalized versus non-personalized advertisements on consumer’s recognition, recall of, and attitude to the advertised product? And is this effect moderated by the number of exposures to the ad?”
5
Theoretical Background
The Effect of Personalization, Repeated Exposure and their interaction on Attitude The goal of advertising is to steer consumers towards certain products or brands. Especially their affective response towards an advertisement is considered an important determinant of advertising effectiveness (MacKenzie, Lutz & Belch, 1986), as it mediates the effect of brand attitude and purchase intention. Personalization is increasingly used as an advertising strategy, in order to influence attitude towards the brand (Aguirre, Mahr, Grewal, Ruyter & Wetzels, 2015). Specifically, personalization entails adapting an advertisement or message to the specific likes, wishes and needs of its receivers, which results in more relevant content (López-Nores, Blanco-Fernández & Pazos-Arias, 2012). Personalization can enhance consumers’ relationship with the brand and the product in two ways: it reduces irrelevant messages and it enables marketers to reach each consumer in a unique and personal way, which in turn will likely influence their relationship positively (Robins, 2003).
Marketers therefore increasingly try to establish these favourable outcomes by personalizing their content (Evergage, 2018). There are a variety of ways in which
advertisements can be personalized for the receiver; for instance, it can include somebody’s name or the advertisement can be adapted based on their previous online behaviour (Smit, Van Noort & Voorveld, 2014). For example, if a person would search for a car rental service in Utrecht on Thursday, the following day that same person can be shown an advertisement for such a service while reading the online news. Because personalization can make
advertising more efficient due to its increased relevance for the receiver, most marketers are repeatedly implementing this approach (Smit, Van Noort & Voorveld, 2014; Evergage, 2018). Furthermore, personalized content could also igniite emotional activation of the
viewer in relation to the presented message (Bas & Grabe, 2013). In other words, by adapting a message to the likes and needs of the receiver, this can affect how they feel. As a
consequence of the personalization, the addressee can feel emotionally and personally addressed. Correspondingly, they can feel more connected to the message which can even lead to an increase of identification with the personalized message. This occurs because they believe the advertisement appeals to their wishes and needs which means they are likely to identify with it, adopt the message presented in it and evaluate the ad more positively than without the personalized character. Thus, based on this information, the following hypothesis is proposed:
6
H1a: A personalized advertisement will lead to a stronger positive attitude to the brand, than a non-personalized advertisement.
As mentioned before, next to personalization, repeated exposure is also an often used marketing strategy. Furthermore, the repeated exposure paradigm is an often researched topic that refers to a person repeatedly being shown a certain stimulus (Zajonc, 2001). Solely by repeatedly exposing the individual to this stimulus – whether or not perceived consciously – he or she is increasingly prone to evaluate the stimulus more positively (i.e., the mere exposure effect, Zajonc, 2001; Campbell & Keller, 2003; Cacioppo & Petty, 1985). Within the field of advertising, the repeated exposure paradigm has also vastly been researched, as various studies have explored which effects are established by repetition. Amongst others, they have looked into how many repetitions of particular features of an advertisement improve or decrease sales (Pechmann & Stewart, 1988).
Earlier research suggests a non-linear relationship exists between the liking and attitude towards an advertisement and the exposure to it (Sawyer, 1981; Simon & Arndt, 1980). These studies suggest that when someone is exposed to an advertisement multiple times, they are more likely to have a positive reaction towards the ad. This is the result of being able to give more thought and attention to the advertisement (Tellis, 1988). By revisiting an advertisement or brand, the different aspects that jointly make up the attitude towards that ad or brand are reinforced. Previous research into the effects of repition on attitude, first report a positive impact on attitude, by means of increased familiarity and habituation (Haugtvedt, Schumann, Schneier & Warren, 1994; Pechmann & Stewart, 1988). However, later repetitions, are assumed to result in adverse effects on attitude due to weariness and annoyance with the presented ad (Schindler, Reinhard & Stahlberg, 2011).
The optimal number of exposures for the maximum persuasive effects on attitude has been a widely discussed topic (Zajonc, 2001; Campbell & Keller, 2003; Cacioppo & Petty, 1985). Tellis (1997) divides the conducted research into two groups: the minimalists, that argue that maximum evaluation is achieved by only a few (one to three) exposures (Montoya, Horton, Vevea Citkowicz and Lauber, 2017) and the repetitionists, who are firm believers that repeated exposure is imperative and that there is no limit to the evaluations (Schmidt & Eisend, 2015; Tellis, 1997). Even though there is a consensus about the existence of the inverted U-shaped relation between repeated exposure and attitude, extensive research does not give a conclusive answer to what the limit is of exposures for the maximum effect. A meta-analysis conducted by Schmidt and Eisend (2015) indicates that the optimal number of
7
exposures is different per dependent variable. They suggest that the effect of repetition on attitude deteriorates from the 11th exposure onwards. According to Tellis (1997), Schmidt and Eisends’ (2015) findings would be categorized as in conjunction with the repetitionists viewpoint. The findings inferred in a meta-analysis by Montoya et al. (2017) would, on the contrary, be categorized as in line with the minimalists’ ideology. They suggest that the optimal number of exposures for attitude consists of less than 4. From the 4th exposure onwards, especially simple stimuli are more likely to lead to saturation and can cause feelings of boredom.
Nonetheless, in line with the review by Tellis (1997), this paper does not aim to settle this ongoing debate, as it is believed that both perspectives are likely to be true under the circumstances that they were studied. The context of the advertising repetition forms the main factor which can determine how many repetitions are required for optimal effects. While ensuring ecological validity, participants were shown an actual online news article with two or four advertisements implemented in them. As the amount of space next to such an article is limited, exposures are expected to lead to a quicker optimum. Following the repetitionists’ view would include at least ten repetitions of the advertisement on the same news page. Repeating the same advertisement so often would not be in line with what one would come across in the real world and thus harm ecological validity. Additionally, such a large number of exposures also might lead the participant bias as they become aware that the advertisement is part of the goal of the study. Therefore, further inferences will be made in line with the minimalists’ view.
H1b: Two exposures to the advertised stimuli will lead to a more positive attitude toward the brand, than four exposures.
Both personalization of and repeated exposure to advertisements are possible solutions to stand out in the current overload of advertisements and hence the often ineffectiveness of advertising around us. As argued for the first hypothesis, through personalization the
relevance of the advertisement is increased for the one being addressed by the ad (Hoeken et al., 2009; Neuwirth, Frederick & Mayo, 2002; Nordhielm, 2002). This means that the advertisement has a higher interest to the recipient and thus is likely to increase involvement (Hoeken et al., 2009). It is argued that the positive effects of personalized advertising on attitude compared to non-personalized ads are weaker after four exposures than after two exposures. Earlier findings suggest the reason for this is the existence of wear-in and wear-out effects. After consumers have been exposed to an ad for a limited number of times, the ad is
8
said to wear-in, and thus have a significant positive effect on their evaluation of the ad. If any further exposures no longer maintain or reinforce those positive effects on the evaluation, the ad is said to have worn out. Due to the increase in relevance for receivers, personalized advertisements are expected to require fewer exposures than non-personalized advertisements until the wear-out effect is established. Because personalization increases the involvement of the recipient with the message, only two exposures to the ad are required before weariness takes place, which negatively affects attitude (Van Doorn & Hoekstra, 2013). Consequently, this adverse effect quickly outweighs the positive effect (Anand & Sternhal, 1990).
H1c: The effects of advertisement personalization on attitude are moderated by exposure in that
a. The attitude towards personalized advertisement is more positive after two exposures than after four exposures.
b. The attitude towards non-personalized advertisement is more positive after four exposures than after two exposures
The Effect of Personalization, Repeated Exposure and their interaction on Recall In general, advertising effectiveness is often measured via the processing of the message, such as being able to recall the advertisement. On a daily basis we only have limited
cognitive capacity, and as a result, we are not able to process every message, ad or pamphlet we encounter thoroughly with our complete attention. Therefore, we need to be economical with the use of our cognitive energy and are inclined to process all information presented to us primarily in a shallow manner. More specifically, this is called called the heuristic process route (Bohner, Moskowitz & Chaiken, 1995; Chaiken, Liberman & Eagly, 1989). This heuristic or shallow processing, however, also leads to superficial ad effectiveness, which is not in line with the aim of advertisers. By personalizing the advertisement to receivers, the presented information will become more relevant to them. It will increase the receiver’s involvement with the advertised product or message as it increases the extent to which the receiver believes the presented message or product is important to them (Hoeken et al., 2009; Neuwirth, Frederick & Mayo, 2002; Nordhielm, 2002). Personalization, therefore, can by-pass the unwanted result of ad ineffectiveness, as it ensures individuals to be more inclined to acquire comprehensive and critical information about the advertised product or presented message. In other words, the personalized content stimulates the receiver to use more
9
in higher recall. This is the result of being able to give more thought and attention to the advertisement and to have better recall and understanding of the message of the ad (Tellis, 1988).
H2a: A personalized advertisement will lead to a stronger recall of the advertisement, than a non-personalized advertisement.
Additionally, repetition of an advertisement is also shown to strengthen the recall of the advertised brand, and can even lead to significant suppression of recalling competitive brands (Alba & Chattopadhyay, 1986). Repeatedly being exposed to an ad is said to enhance recall of both the advertisement as the brand. This enhancement in recall is the result of an increase of processing of the ad. The more one is exposed to an ad, the more time they can spend on processing the ad, and the more they will be able to recall it (Krugman, 1972). More repetition of an advertisement gives the receiver more time to process the ad thoroughly. Therefore, sole repetition can be an effective method to maintain recall of an advertisement (Krugman, 1972). Thus, the following hypothesis is proposed:
H2b: Four exposures to the advertised stimuli will lead to a stronger recall of the advertisement, than two exposures.
However, when considering the mechanism that is caused by personalization, a different trend becomes apparent. The heightened attention to the ad, as a consequence of the increased relevance of the personalized ad, is expected to subside recall over time (Lehnert, Till & Carlson, 2013). This is the result of habituation with the message, which takes place when one is increasingly exposed to repeated exposures of an ad (Pechmann & Stewart, 1988). This habituation causes a decrease in what can be learned or obtained from the advertisement after repeated exposures, and thus the positive effect of personalization on recall levels off. Although the positive increase in recall diminishes, individuals will still be able to correctly recall it. In other words, repeating a personalized advertisement two or four times does not affect recall. The positive effect of personalization on recall remains, but is not reinforced with more exposures as individuals’ knowledge is expected to already be saturated. No interaction effect is therefore expected for personalization and repeated exposure.
The Effect of Personalization, Repeated Exposure and their interaction on Recognition The threshold theory suggests that recall and recognition measure the same construct, but that recognition requires a less vivid memory than recall (Singh, Rothschild & Churchill, 1988). Other findings also indicate that both recognition and recall are associated scales of memory.
10
Dawson & Reardon (1973) imply that recall and recognition covary. They advocate that the difference is that recognition is a more sensitive scale – the ad is either wrongly or correctly identified – whereas recall demands the person to think back and describe what they have encountered. Due to its simple nature and less stringent character, recognition may allow for false positives. Given their different nature, it is argued that they measure different constructs that play a role in different settings. When looking around a certain store, a simple recognition of a product might be sufficient to stimulate purchase, while in high involvement situations it is likely that recall is preferred.
Personalization of an ad prompts individuals to link the message to themselves (Burnkrant & Unnava, 1995; Warnick, Xenos, Endres & Gastil, 2005). This self-reference facilitates a personal connection with the message which stimulates the individual to process the message more systematically. Additionally, illustrating personal appeal increases the interest of the receiver by adding more vividness and relevance. This is, however, not the case for non-personalized messages (Brosius & Bathelt, 1994).
H3a: A personalized advertisement will lead to a stronger recognition of the advertisement, than a non-personalized advertisement.
It is widely suggested that repetition of a message is required for that message to be remembered. More exposures to an ad are likely to lead to increased recognition. Both recall and recognition are considered to be measures of memory (Ley & Karker, 1982). The latter, however, has widely been discussed as people have called into question whether recognition is as good of a scale as recall. They argue that it is a less sensitive measure as results to recognition are significantly higher than to recall. Because recognition is a less stringent measure for the respondent, repeated exposures are said to quickly result in ceiling effects after which recognition is said to not show any decline (Haber, 1970; Krugman, 1979). Singh et al. (1988) refute that. They infer that once an individual is no longer exposed to an
advertisement, memory measured by recall strongly deteriorates. Memory measured by recognition, however, is said to show a slower decay which will not immediately be present (Ley & Karker, 1982). Next to that, for correct recall, more repetitions are required than for accurate recognition.
H3b: Exposure does not affect recognition of the advertised stimuli
The increase in relevance that comes with personalization and repeatedly being exposed to an ad both enhance recognition. Although implementing both strategies of personalization and repeated exposure will influence how quickly recognition is established,
11
as long as individuals are still exposed to an ad, their recognition will not level off (Haber, 1970; Krugman, 1979). If individuals already know an advertisement and also come across it, they will not forget about it and they will be able to correctly recognize it. Consequently, no interaction effect is expected between personalization and repeated exposure on recognition of the ad. No interaction effect is therefore expected for personalization and repeated exposure on recognition.
Method Design and Sample
The current study has employed an online experimental 2 (personalization:
personalized advertisement vs non- personalized advertisement)× 2 (repeated ad exposure: two repeated exposures vs four repeated exposures) design, with personalized or
non-personalized advertisement and two or four exposures to the advertisement as the factors. 230 respondents (64.3% female, 34.4% male, 1.2% non-binary, Mage = 27, SD =7.39) were
recruited via the University of Amsterdam, the Dutch language school ‘Koentact’, the podcast ‘Onder Mediadoctoren’ and survey share sites such as ‘Poll Pool’, ‘SurveyCircle’ and
‘SurveySwap’. Although it is preferred to have a homogeneous sample that reflects the “real” population for research findings to be generalizable, a study conducted by Lynch (1999) indicated that results derived from a “real” population sample do not hold an advantage in generalizability to results derived from a student-sample. Furthermore, Collie, Sparks & Bradley (2000) argue that student samples are justifiable because students are also part of the
12
“real” population and next to that are acquainted with online advertisements. Considering we have 4 different groups within our experimental design (non-personalized vs two exposures; non- personalized vs four exposures; personalized vs two exposures; personalized vs four exposures) we passed the (4x50) 200 participants that were required to properly be able to make inferences from this research (VanVoorhis & Morgan, 2007). Respondents have been randomly assigned to one of the four conditions. The experiment and short surveys have been conducted via Qualtrics.
Research Conduit
In this experiment, participants were first asked to agree to their data being used for research endings while their anonymity was safeguarded. To prevent participant bias, participants were told that the actual interest of the study lies in how they experience online content. Afterwards they were asked to indicate to which European country they would travel to if they could choose one at the moment of filling in the survey by their preferred travel method. The countries they could choose of consisted of 9 of the 10 most visited European countries from the Netherlands (CBS, 2017) and their favourite means of getting there could be by car, by train or by plane. From the top 10 most visited countries it was chosen to omit Belgium as due to its proximity and country size, it is uncommon to visit it by plane.
Followings, multiple filler tasks 1were included to prevent any bias in respondents’ answers. Manipulation
Next participants were instructed to read an actual online news article obtained from the BBC that was either manipulated by personalisation on individual preferences or a control condition. The personalization condition was operationalised by including an advertisement for their earlier chosen favourite holiday destination and way of transport to that destination. This could be an advertisement to their preferred destination with KLM, Interrail or Sixt. The control group included an advertisement for Samsonite luggage. Following the rationale of Bleier & Eisenbeiss (2015) for the control group an arbitrary product was shown which is of a similar, yet not the same product category. This consisted of the high end luggage brand 'Samsonite', as high end luggage is something that one would not buy ‘on a whim’ but rather think about consciously similar to booking a holiday and which also similarly to booking a holiday, can infer feelings of relaxation (Milan & Howard, 2007).
1 Participants had to fill out a “I am not a robot” – CAPTCHA; indicate how often they read, listened to or
watched online news, newspapers, television or radio and they were asked to order seven adjectives to their preference. A comprehensive overview can be found in Appendix B, section 1.
13
An overview of the different stimuli can be found in Appendix C. Participants were then again asked to respond to some filler tasks2, to prevent them from becoming aware of the goal of the research.
Dependent Variables
The measurement of all dependent variables was acquired from earlier research. The scale for attitude was implemented as proposed by Spears & Singh (2004). The questions ‘I find the brand of the advertised product appealing/ good/ pleasant/ favorable/ likeable’ were rated on a 10-point Likertscale from certainly not (1) to certainly yes (10). A principal component analysis (PCA) showed that the 5 items that measure attitude towards the brand form a single uni-dimensional scale: only one component has an eigenvalue above 1
(eigenvalue 4.21) and there is a clear point of inflexion after this component in the scree plot. Reliability of the scale is good, Cronbach's alpha = .95.
The scale for free recall as proposed by Precourt (2016) was implemented and
participants were instructed to write down all that came to mind when thinking of the different brands, products and ads they had encountered. This was coded by the researcher while blind to the experimental condition to the example of Precourt (2016). Initial coding of recall was scored a “0” when they had no recollection of the ad, a “1”when respondents remembered seeing an ad for an airline, train, car or luggage or the brand of the product and a “2” when they remembered seeing an ad for an airline, train, car or luggage and the correlating brand. However, an exploration of the answers seemed to only show either no recall (0) or partial recall (1). Therefore the third variable (2) was omitted. A comprehensive overview of the coding scheme can be found in Appendix D. Consequently, the collected data on recall was recoded into dichotomous variables.
The scale for recognition as proposed by Norris & Colman (1992) was used to measure recognition. Participants were shown the manipulated advertisement that they were exposed to together with the 3 other advertisements – e.g. they had to choose between the advertisement for KLM, Interrail, Sixt or Samsonite – and indicate which advertisement they were shown. Their answers were then recoded to ‘1’ if they correctly identified the
advertisement that they had been exposed to on the newspage, or a ‘0’ if they failed to do so.
2 Participant were asked about the topic of the article, how they would score the webpage (1-10), how readable
they considered the article (1-10), how relevant they found the article to be for them and how they would rate the article (1-10). A detailed overview can be found in Appendix B, section 2.
14
Manipulation checks
To verify the difference between the personalized and non-personalized condition, participants were asked to indicate how much they agreed with two statement on a ten-point Likert scale (1- certainly not; 10- certainly yes). These statements were ‘Do you think any part of the webpage was created especially for you?’, M = 5.22, SD = 3.35, Min = 1, Max = 10, and ‘Did you have an impression of being personally addressed anywhere on the webpage?’, M = 4.55, SD = 3.10, Min = 1, Max = 10.
To check whether participants were aware of how many repetitions of the advertisement they were exposed to, they were asked to rate the number of the same advertisements they saw, M = 3.18, SD = 2.07, Min = 0, Max = 9.
Results Randomization Checks
To test whether the participants were indeed randomly assigned to a condition, three one way ANOVA’s were conducted with birthyear, political position and education level as outcome variables. These showed that per condition, there was no significant difference in birthyear, F(3, 214) = 1.73 , p = .162, no significant difference in political position F(3, 211) = .20 , p = .897 and no significant difference in education level F(3, 214) = .06 , p = .983. As gender was measured at a nominal level with three outcomes; male, female or non-binary, a chi-square test indicated that there were 4 cells with an expected count less than 5. These consisted of the spread of the non-binary participants over the 4 conditions. Although these cells with the low expected count violate one of the assumptions, it should not result in a problem for our inferences because our data contains 214 participants of which only 3 participants indicated to be non-binary3. Followings, the Chi-square outcomes were
interpreted and suggested that there was no significant difference of condition in gender, X2 (6) = 7.56, p = .272.
Controlling for Covariates
Due to the violations of normal distribution, a spearman’s correlation was performed for birthyear, gender, political position and education level to control for a correlation with recall, recognition and attitude towards the brand. While most correlations were found non-significant, a significant correlation was found for gender and recognition (rs = .13, N = 218, p
3 Although reports lack for the Netherlands, in a 2014 research conducted in the United Kingdom, 0.4% of
people defined as non-binary (Titman, 2014). Therefore, it is expected that the number of people defining as non-binary are representative of the Dutch population.
15
< .049), birthyear and recognition (rs = -.17, N = 218, p < .012) and a marginal significant
correlation for education level and recognition (rs = .13, N = 218, p < .065). No significant
covariates were found for recall or attitude towards the brand, p > .05. An overview of all correlational tests can be found in Appendix F.Because the randomization check showed no aspect to be significantly different amongst the conditions, no variables were controlled for in the testing of the hypotheses.
Manipulation Checks
An independent samples t-test showed that personalization was successfully
manipulated, as personalized advertisements led to a significant higher belief that the content was created especially for them (M = 5.44, SD = 3.41) than non-personalized advertisements (M = 3.67, SD = 2.48). This was significant, Mdifference = -1.76, t(216) = -4.377, p < .001 and represents a medium sized effect d = .528. The same holds for the manipulation of repeated exposures. The participants who were exposed to four exposures (M = 3.63, SD = 2.08) also indicated to have encountered significantly more advertisements, than those who were
exposed to two exposures (M = 2.69, SD = 1.96). This Mdifference = -.94, was significant, t(213) = -3.387, p < .001 and also represents a medium sized effect d = .452.
Effects on Attitude
To test the effects of personalization, repeated exposure and their possible interaction effect on attitude towards the brand, a two-way ANOVA was conducted. It should be noted that the assumption of normality was not met, although those of no outliers and homogeneity of variances were met. Both personalization (non-personalized versus personalized) and repeated exposure (two exposures versus four exposures) consisted of two levels. There was a marginal significant main effect of personalization on attitude, F(1, 215) = 3.36, p = .068, ω2 = 0.15, that consisted of a large effect. Personalized advertisements (M = 5.52, SD = 2.44) had a marginal significant effect on attitude towards the brand, compared to non-personalized ones (M = 4.86, SD = 2.61). There was no main effect of repeated exposure on attitude, F(1, 215) = .02, p = .902, ω2 < -.00 , nor of the interaction of personalization and repeated exposure, (1,
16
215) = .27, p = .602, ω2 = -0.02.
Effects on Recall
The effects of personalization, repeated exposure and their interaction on recall was tested via a two-way loglinear analysis. This produced a final model which retained all the effects. The likelihood ratio of this model was X2 (4) = 5.15, p = .272. The results indicate that the highest-order interaction (personalization x repeated exposure x recall interaction), X2 (7) = 37.61, p < .001 was significant a significant predictor of the data. As the hypotheses also evolve around personalization and recall, and repeated exposure and recall separately, it is important to look into these interactions. The interaction between personalization and recall appears to significantly predict the model fit, X2 (1) = 13.98, p < .001, but this is not the case for the interaction of repeated exposure and recall, X2 (1) = 2.52, p = .113. Thus while H2a: ‘A personalized advertisement will lead to a stronger recall of the advertisement, than a non-personalized advertisement’ is supported, this is not the case for H2b: ‘Four exposures to the advertised stimuli will lead to a stronger recall of the advertisement, than two exposures’. Tabel 3
Recall in association with personalization and the number of exposures
Recall Personalization Exposure Count % of total
Not recalled correctly
Not personalized Two exposures 35 16,4%
Four exposures 49 22,8% 4,99 5,45 4,77 5,59 4,2 4,4 4,6 4,8 5 5,2 5,4 5,6 5,8 Two exposures Four exposures A T T IIT UD E T OW A RDS T HE BRA ND
Figure 2. The effects of personalization and repeated exposure on attitude
towards the brand.
17
Personalized Two exposures 26 12,2%
Four exposures 30 14,1% Recalled correctly Not personalized Two exposures 17 8,1% Four exposures 10 4,4%
Personalized Two exposures 25 11,8%
Four exposures 26 12,2%
Total 218 100%
Effects on Recognition
The effects of personalization, repeated exposure and their interaction on recognition was tested via a two-way loglinear analysis. This produced a final model which excluded all the effects. The likelihood ratio of this model was X2 (6) = 3.51, p = .743. The results indicate that neither the highest-order interaction (personalization x repeated exposure x recognition interaction), X2 (1) = .02, p = .904, nor the interaction of personalization x recognition, X2 (1) = 2.79, p = .095, nor the interaction of repeated exposure x recognition, X2 (1) = .23, p = .635, nor the interaction of personalization x repeated exposure, X2 (1) = .03, p = .856, were
significant. Additionally, as main effects both personalization, p = .892, and repeated exposure, p = 498, do not significantly affect the fit of the model. Consequently, neither hypothesis about recognition is supported.
Tabel 4
Recognition in association with personalization and the number of exposures
Recognition Personalization Exposure Count % of total
Not recognized correctly
Not personalized Two exposures 20 7,8%
Four exposures 20 7,8%
Personalized Two exposures 14 6,4%
Four exposures 14 6,4% Recognized correctly Not personalized Two exposures 32 14,7% Four exposures 38 17,4%
18
Personalized Two exposures 38 17,4%
Four exposures 42 19,3%
Total 218 100%
Conclusion and Discussion
In this study, a 2 (personalized vs non-personalized) by 2 (two repeated exposures vs four repeated exposures) experimental design was set up to see to what extent (non-)
personalization of an advertisement influences recall and recognition of the ad, and attitude towards the brand. At the same time, the possibly moderating role of two or four repeated exposures to the advertisement was considered. Data from 218 participants, ranged from 18 to 53 years old, was analyzed.
The findings show that while personalization, with marginal significance, had a large effect on attitude, this is not supported for the effect of repeated exposures or the interaction-effect between personalization and repeated exposure. On attitude, only hypothesis 1a about personalization was supported. Personalization is also shown to be a significant predictor of recall. This is not the case for repeated exposures to the advertisement. No significant predictions could be made from the model that included the interaction-effect of repeated exposures and personalization, which confirms the hypothesis that scores would not be different. Findings therefore support H2a. Finally, neither personalization, nor repeated exposures, nor the interaction between personalization and repeated exposure were found to be a significant predictor of correct recognition of the ad. Therefore, support was not found for either the widely discussed mere-exposure effect, or the expected interaction between personalization and repeated exposure were found. Amongst others, this contradicts findings of the meta-analysis by Schmidt & Eisend (2015). Still, this study contributes to the scientific knowledge as it is one of the first to look into the increasingly used combination of
personalization and repeated exposures.
The answer to the research question ‘to what extent is there an effect of personalised versus non-personalised advertisements on consumer’s recognition, recall of and attitude to the advertised product? And is this effect moderated by the number of exposures to the ad’, is as a consequence limited. This might have been caused by the ecological setting in which participants were shown the advertisements. As a means to ensure ecological validity,
respondents were exposed to an actual online news article that for means of this research was manipulated with extra advertisements on the side. This way of coming across advertisements
19
corresponds with how (non-)personalized advertisements are often displayed on news websites such as NU.nl, theguardian.co.uk, and NOS.nl. However, ensuring the ecological validity of this study is likely to have interfered with the receivers’ selective attention to the advertisement. A recent study by Yagi & Inoue (2018) confirms that selective attention to an ad is an important characteristic of the mere-exposure effect. They infer that the expected positive effect of repeatedly being exposed to an advertisement is moderated by attention to the ad. Participants within this study were told that the study was aimed at obtaining greater insight in how they experienced news pages and asked them to look at the webpage and read the article as they would normally do when reading an online article. Although results did indicate that repeated exposure was successfully manipulated, on the basis of the open suggestions 4participants could include at the end of the survey, it is presumed respondents mainly focused on the article and quickly scanned through the webpage without paying proper attention to the advertisements. The positive effect that repeated exposure generates on
attitude and recall is modulated by attention to the exposure, in a way that when no attention is paid at the repeated advertisements, the repeated exposure will not positively influence participants’ recall of the ad or their attitude towards the brand and therefore is ineffective in reaching its persuasive goal. Yagi & Inoue (2018) inferred from their study that the
inattention towards the products they were advertising was the result of accompanying it with good-looking female models who took up most of the attention. In the current study it is likely that participants focussed their attention towards the article, as they were told that they would be asked several questions about the full webpage afterwards. Still, in this paper it is not argued that the mere-exposure effect cannot occur for disregarded items (Montoya et al, 2017; Bornstein & D'agostino, 1992). Rather, it is suggested that attention modulates the effect of repeated exposure on recall and attitude.
Additionally, in this study manipulation of personalization consisted of showing participants an advertisement for travelling to a European destination of choice by their favourite way of transport. Considering the nature of the personalization that was opted for, if further studies include a similar type of personalization, it is desired to also control for
participants’ pre-existent wish to go on a holiday. This pre-existing desire might influence the effects of the manipulation on recall, recognition and attitude towards the advertisement. If
4 Some of these answers were: ‘Did not pay attention to any of the advertisements’, ‘I do not recall the ads, I
more scrolled the text’, ‘I wasn't paying attention to the adverts. I was looking mainly on the left of the webpage’, ‘I did not pay attention at the advertisements, I just skipped them and concentrated to the text’, ‘I ignored them to be honest’.
20
someone already had a strong desire to go on a holiday, it is likely that this pre-existing desire would be the reason for a high score on attitude towards the brand of the advertisement. That is KLM, Sixt, Interrail and Samsonite all can bring them closer to achieving that wish. The opposite is also important to note; if someone is not open to the idea of travelling at all, an ad is not likely to change their mind. Controlling for their pre-existent desire to go on a holiday would therefore be recommended in follow-up studies.
Further research is needed to fully explore the interplay of personalization and
repeated exposure, while also considering the role of attention. Nonetheless, this study forms an important contribution to the existing body of knowledge for advertising practitioners; this study shows that while the mere-exposure effect might be susceptible to attention, this is not the case for the effect of personalization. These findings suggest that personalization is an effective strategy to influence recall of the advertisement and attitude towards the brand, without putting too much strain on participants.
References
Alba, J. W., & Chattopadhyay, A. (1986). Salience effects in brand recall. Journal of
Marketing Research, Vol.23(4), 363-369.
Anand, P. & Sternthal, B. (1990). Ease of message processing as a moderator of repetition effects in advertising. Journal of Marketing Research, 26 (8), 345–353.
Ayenson, M., Wambach, D., Soltani, A., Good, N., & Hoofnagle, C. (2011). Flash cookies and privacy II: Now with HTML5 and ETag respawning. Team for Research in
Ubiquitous Secure Technology, 2011.
Baek, T. H., & Morimoto, M. (2012). Stay away from me. Journal of Advertising, 41(1), 59-76.
Bas, O., & Grabe, M. E. (2015). Emotion-provoking personalization of news: Informing citizens and closing the knowledge gap?. Communication Research, 42(2), 159-185. Bleier, A., & Eisenbeiss, M. (2015). Personalized online advertising effectiveness: The
interplay of what, when, and where. Marketing Science, 34(5), 669-688. Bohner, G., Moskowitz, G. B., & Chaiken, S. (1995). The interplay of heuristic and
systematic processing of social information. European review of social psychology, 6(1), 33-68.
21
Bornstein, R. F., & D'agostino, P. R. (1992). Stimulus recognition and the mere exposure effect. Journal of personality and social psychology, 63(4), 545.
Brosius, H. B., & Bathelt, A. (1994). The utility of exemplars in persuasive communications.
Communication Research, 21(1), 48-78.
Burnkrant, R. E., & Unnava, H. R. (1995). Effects of self-referencing on persuasion. Journal
of Consumer Research, 22(1), 17–26.
Cacioppo, J. T., & Petty, R. E. (1985). Central and peripheral routes to persuasion: The role of message repetition. Psychological processes and advertising effects, 911.
Campbell, M. C., & Keller, K. L. (2003). Brand familiarity and advertising repetition effects.
Journal of consumer research, 30(2), 292-304.
Cauberghe, V., & De Pelsmacker, P. (2010). Advergames. Journal of advertising, 39(1), 5-18. Centraal Bureau voor de Statistiek, CBS. (2017, May 18th). Vakanties; kerncijfers. Retrieved
from
http://statline.cbs.nl/Statweb/publication/?VW=T&DM=SLNL&PA=70232ned&D1=7 4-117&D2=(l-9)-l&HD=150630-1344&HDR=G1&STB=
Chaiken, S. (1980). Heuristic versus systematic information processing and the use of source versus message cues in persuasion. Journal of personality and social psychology, 39(5), 752.
Chaiken, S., Liberman, A., & Eagly, A. H. (1989). Heuristic and systematic information
processing within and beyond the Persuasion Context. In J. S. Uleman, & J. A. Bargh
(Eds.), Unintended thought, 212, 212-252. New York: Guilford Press.
Chen, S., & Chaiken, S. (1999). The heuristic-systematic model in its broader context.
Dual-process theories in social psychology, 15, 73-96.
Cohen, J. (2001). Defining identification: A theoretical look at the identification of audiences with media characters. Mass communication & society, 4(3), 245-264.
Collie, T. A., Sparks, B., & Bradley, G. (2000). Investing in interactional justice: A study of the fair process effect within a hospitality failure context. Hospitality Research
22
Cofone, I. N. (2017). The way the cookie crumbles: online tracking meets behavioural economics. International Journal of Law and Information Technology, 25(1), 38-62. Dawson, M. E., & Reardon, P. (1973). Construct validity of recall and recognition
postconditioning measures of awareness. Journal of experimental psychology, 98(2), 308.
Englehardt, S., Reisman, D., Eubank, C., Zimmerman, P., Mayer, J., Narayanan, A., & Felten, E. W. (2015). Cookies that give you away: The surveillance implications of web tracking. Proceedings of the 24th International Conference on World Wide Web, 289-299.
Evergage. (2018). 2018 Trends in personalization [Data file]. Retrieved from
https://www.evergage.com/wp-content/uploads/2018/04/Evergage-2018-Trends-in-Personalization-Survey.pdf
Haugtvedt, C. P., Schumann, D. W., Schneier, W. L., & Warren, W. L. (1994). Advertising repetition and variation strategies: Implications for understanding attitude strength.
Journal of Consumer Research, 21(1), 176-189.
Havlena, W., Cardarelli, R., & De Montigny, M. (2007). Quantifying the isolated and synergistic effects of exposure frequency for TV, print, and internet advertising.
Journal of Advertising Research, 47(3), 215-221.
Hoeken, J. A. L., Hornikx, J. M. A., & Hustinx, L. G. M. M. (2009). Overtuigende teksten. Onderzoek en ontwerp. Bussum: Coutinho.
Krugman, H. (1972) Why three exposures may be enough. Journal of Advertising Research, 12(6), 11–14.
Lehnert, K., Till, B. D., & Carlson, B. D. (2013). Advertising creativity and repetition: Recall, wearout and wearin effects. International Journal of Advertising, 32(2), 211-231. Ley, R., & Karker, J. (1982). A comparison of three methods of measuring recognition
memory in high and low academic achievers. Bulletin of the Psychonomic Society, 19(1), 11-14.
López-Nores, M., Blanco-Fernández, Y., & Pazos-Arias, J. J. (2012). Cloud-based
personalization of new advertising and e-commerce models for video consumption.
23
Lynch Jr., J.G. (1999). Theory and external validity. The Journal of Academy of Marketing
Science, 27(3), 367-376.
MacKenzie, S. B., Lutz, R. J., & Belch, G. E. (1986). The role of attitude toward the ad as a mediator of advertising effectiveness: A test of competing explanations. Journal of
marketing research, Vol. 23, 130-143.
Martin, J. R. (1992). English text: System and structure. John Benjamins Publishing. McAuley, J., & Leskovec, J. (2013). Hidden factors and hidden topics: understanding rating
dimensions with review text. In Proceedings of the 7th ACM conference on
Recommender systems, pp. 165-172.
Meijers, J. (2016, February 27). We laten ons vrijwillig in de gaten houden. Financieel Dagblad. Retrieved from: https://fd.nl/morgen/1141220/we-laten-ons-vrijwillig-in-de-gaten-houden
Millan, E. S., & Howard, E. (2007). Shopping for pleasure? Shopping experiences of Hungarian consumers. International Journal of Retail & Distribution
Management, 35(6), 474-487.
Montoya, R. M., Horton, R. S., Vevea, J. L., Citkowicz, M., & Lauber, E. A. (2017). A re-examination of the mere exposure effect: The influence of repeated exposure on recognition, familiarity, and liking. Psychological bulletin, 143(5), 459.
Nasraoui, O. (2005). World wide web personalization. Encyclopedia of data warehousing and
mining, pp. 1235-1241, IGI Global.
Neuwirth, K., Frederick, E., & Mayo, C. (2002). Person-effects and heuristic-systematic processing. Communication Research, 29(3), 320-359.
Nordhielm, C. (2002). The influence of level of processing on advertising repetition effects.
The Journal of Consumer Research, 29, 371–382.
Norris, C. E., & Colman, A. M. (1992). Context effects on recall and recognition of magazine advertisements. Journal of Advertising, 21(3), 37-46.
Palmer, D. E. (2005). Pop-ups, cookies, and spam: Toward a deeper analysis of the ethical significance of Internet marketing practices. Journal of business ethics, 58(1-3), 271-280.
24
Pechmann, C., & Stewart, D. W. (1988). Advertising repetition: A critical review of wearin and wearout. Current issues and research in advertising, 11(1-2), 285-329.
Petty, R. E., Priester, J. R., & Wegener, D. T. (1994). Cognitive processes in attitude change.
Handbook of social cognition, 2, 69-142.
Precourt, G. (2016).How Does Recall Work in Advertising?. Journal of Advertising Research, 56, 3 (2016): 229–230.
Rossiter, J. R., & Bellman, S. (1999). A proposed model for explaining and measuring web ad effectiveness. Journal of Current Issues & Research in Advertising, 21(1), 13-31. Sawyer, A. G. (1981). Repetition, cognitive responses, and persuasion. Cognitive responses in
persuasion, 237-261.
Schindler, S., Reinhard, M. A., & Stahlberg, D. (2011). Repetition of educational AIDS advertising affects attitudes. Psychological reports, 108(3), 693-698.
Schmidt, S., & Eisend, M. (2015). Advertising repetition: A meta-analysis on effective frequency in advertising. Journal of Advertising, 44(4), 415-428.
Simon, J. L., & Arndt, J. (1980). The shape of the advertising response function. Journal of
Advertising Research, 11-30.
Singh, S. N., Rothschild, M. L., & Churchill Jr, G. A. (1988). Recognition versus recall as measures of television commercial forgetting. Journal of Marketing Research, Vol. 25, 72-80.
Smit, E. G., Van Noort, G., & Voorveld, H. A. (2014). Understanding online behavioural advertising: User knowledge, privacy concerns and online coping behaviour in Europe. Computers in Human Behavior, 32, 15-22.
Spears, N., & Singh, S. N. (2004). Measuring attitude toward the brand and purchase intentions. Journal of Current Issues & Research in Advertising, 26(2), 53-66.
Tellis, G. J. (1988). Advertising exposure, loyalty, and brand purchase: A two-stage model of choice. Journal of marketing research, 134-144.
Tellis, G. J. (1997). Effective frequency: One exposure or three factors?. Journal of
25
VanVoorhis, C. W., & Morgan, B. L. (2007). Understanding power and rules of thumb for determining sample sizes. Tutorials in Quantitative Methods for Psychology, 3(2), 43-50.
Van Doorn, J., & Hoekstra, J. C. (2013). Customization of online advertising: The role of intrusiveness. Marketing Letters, 24(4), 339-351.
Warnick, B., Xenos, M., Endres, D., & Gastil, J. (2005). Effects of campaign-to-user and text-based interactivity in political candidate campaign web sites. Journal of
Computer-Mediated Communication, 10(3), 10315.
Wedel, M., & Pieters, R. (2015). The buffer effect: The role of color when advertising exposures are brief and blurred. Marketing Science, 34(1), 134-143.
Zajonc, R. B. (2001). Mere exposure: A gateway to the subliminal. Current directions in
