Is Relevance the Key? The Effect of Personalization onAdvertisement Attention, Recognition and Recall in a MultiScreening Context

(1)

Is Relevance the Key? The Effect of Personalization on

Advertisement Attention, Recognition and Recall in a

Multi-Screening Context

Graduate School of Communication

Master’s programme Communication Science Master’s Thesis Supervisor Ivana Bušljeta Banks

Handed in by Mareike Weiss (ID: 12117366) Submission date 26-06-2020

(2)

Abstract

Multi-screening, as a special form of media multitasking, is on the rise and its mostly negative impact on recognition and recall has recently been studied increasingly. However, it is

unclear, how the cognitive performance of multi-screeners can be improved. An online experiment was conducted (a) to investigate the impact of multi-screening on advertising recognition and recall compared to single screening, (b) proposing attention paid to the advertisement as a mediator and (c) manipulating personalization of the advertisement as a moderator for the relationship between multi-screening and attention. The results indicate that multi-screeners recall significantly fewer details than single screeners from the advertisement they were exposed to. Ad recognition and ad attention were affected by interest in the video in such a way that multi-screeners performed equally well as single screeners when they were interested in the content. There was no moderated mediation effect of personalization found, however, this might be due to the operationalization of personalization. The question of how personalization should be manipulated in future studies is raised. Overall, findings suggest that cognitive performance when multi-screening depends largely on rather subjective variables that differ between individuals like interest in the media content and the feeling of being personally addressed.

“Consumer attention [is] a scarce and precious resource that is difficult to obtain and easily lost” (Angell, Gorton, Sauer, Bottomley & White, 2016). This is even more the case with media multitasking “becoming the default method of media consumption” (Brasel & Gips, 2017). Such statements summarize what recent statistics show: according to the Nielsen Company report (2018) almost two thirds (64%) of adult media users use digital devices such as smartphones while watching TV at least sometimes. Recipients state looking up

(3)

information about what is on TV, checking emails or messages, and online shopping for products being advertised among the most frequent activities while watching TV. Media multitasking which involves the simultaneous use of two or more screens, e.g. television and tablet or smartphone, is referred to as multi-screening (Segijn, 2016). An investigation in the Netherlands showed that most participants spent on average 80 minutes per day multi-screening on top of the time they engaged in other media multitasking activities (e.g.

combining a screen device with other media like newspapers or radio). Moreover, the younger people are the longer they multi-screen (Segijn, Voorveld, Vandeberg, Pennekamp, & Smit, 2017). With the rise of digital media environments, multi-screening as a specific form of media multitasking is increasing as well.

Multi-screening was described as one of the ten biggest challenges for marketing in the WARC report of 2013 (Segijn, Voorveld, & Smit, 2017) because it raises the question of how effective advertising is when viewers split their visual attention distributing their cognitive capacity on two channels (Segijn, Voorveld, Vandeberg & Smit, 2017; Mayer & Moreno, 2003). Additionally, the more competitive the viewing environment, i.e. the more distractions there are around the viewer, for example other screens, the more visual attention decreases (Segijn & Eisend, 2019). Paying less attention to either of the two screens results in poorer memory of the content (Jeong & Hwang, 2016). This might be of great concern for advertisers since most of the multitasking occurs in ad breaks (Garaus, 2019). The less

attention viewers pay to the advertisements the lower their persuasion, recall, and recognition, three elements of ad effectiveness (Garaus, Wagner, & Bäck, 2017). Thus, brands need to find a way to gain media users’ attention for their advertising to be effective.

So far it has been suggested to advertisers to engage consumers in related tasks to increase attention to the TV program, which made multi-screeners more attentive to

(4)

However, it might not always be possible for advertisers to control in which context or during which task the audience is exposed to the ad. Thus, another way to gain consumers’ attention might be the personalization of advertisements. Personalized advertising aims at increasing attention paid to the ad, as well as improving ad recall by implying the advertisement is specifically directed at the individual, therefore, making it more relevant to the receiver (Bang & Wojdynski, 2016; Maslowska, Smit & van den Putte, 2016). Thus, personalization is expected to result in higher ad effectiveness (Köster, Rüth, Hamborg & Kaspar, 2015). Until now, little attention has been paid to studying the moderating influence of a personalized advertisement on ad attention, ad recognition, and ad recall in a multi-screening situation. Therefore, building on the findings of previous research on multi-screening, an experiment is conducted to investigate how personalized advertising affects recognition and recall.

The current study contributes to the understanding of moderating factors influencing multi-screening effects, both theoretically and practically, by showing if and how

personalization increases ad attention while multi-screening and improves ad recognition and recall. In the following, based on previous literature, important concepts are defined before three hypotheses are formulated. Subsequently, the method to test the hypotheses is described and results are presented. This work ends with a thorough discussion, limitations of the study, and ideas for future research.

Theoretical Framework

Media multitasking was described differently in various studies over time. Some define it rather broadly as doing two simultaneous tasks of which one is a media activity (Jeong & Fishbein, 2007), others refer to media multitasking when two media are used at the same time (Srivastava, 2013). In this study, the latter definition is narrowed further down to

(5)

define multi-screening as the simultaneous use of two screen media, hence forcing the

consumer to split his or her attention between two visual tasks (Segijn, Voorveld, Vandeberg, & Smit, 2017; Segijn & Eisend, 2019). With the rise of digital media environments, multi-screening as a specific form of media multitasking is increasing. Such behaviors can have an impact on the attention paid to the media content and cognitive outcomes like recognition and recall of information (Jeong & Hwang, 2016).

Ad Recognition and Recall when Multi-Screening

Generally, media multitasking negatively affects attention to and memory of a message (Jeong & Hwang, 2016). For example, reading a text while listening to a podcast or listening to the radio while surfing the internet decreases recall and recognition of the media content (Srivastava, 2013; Voorveld, 2011). This can be explained with the limited capacity theory by Lang (2000). Recipients only have a certain amount of cognitive resources to process (i.e. encode, store and retrieve) media content, which has to be split to more than one medium in a multitasking situation. Additionally, the Elaboration Likelihood Model (ELM, Cacioppo & Petty, 1984) delivers a similar explanation: The distraction of attention from one medium to the other decreases the opportunity and ability to process the content on the central route and therefore makes peripheral processing more likely. Although this may have a positive effect on persuasion (Jeong & Hwang, 2012), it decreases cognitive performance.

As switching between media costs time and cognitive capacity (Segijn, Voorveld, & Smit, 2017), multitasking can cause capacity interference, a competition for cognitive

resources between the multiple media, which is limiting the recipients’ opportunity to process the content of both media thoroughly (Armstrong & Chung, 2000; Angell et al., 2016). In addition to that, multi-screening causes interference on a second level (Garaus, 2019). Since both media require the same modality (visual channel), sensory interference occurs (Mayer &

(6)

Moreno, 2003; Jeong & Hwang, 2016). To allocate attention to both screens, one task always needs to be suspended temporarily while attending the other (Segijn, Voorveld, Vandeberg, et al., 2017). Switching behaviors and their consequences for memory have been particularly looked at regarding advertising effectiveness while multi-screening. Several studies showed that the distraction of using multiple screens impairs recall and recognition of the brand or advertisement (Garaus et al., 2017; Segijn, Voorveld, Vandeberg, & Smit, 2017; Segijn & Eisend, 2019). Thus, recipients’ opportunity to encode the advertisement, and the ability to retrieve its content successfully are expected to be limited, and Hypothesis 1 predicts the following:

H1. Participants in a multi-screening condition will a) have lower ad recognition and b) lower ad recall rates than participants in a single screening condition.

The Mediating Role of Attention

Attention defines the amount of cognitive resources that the viewer allocates to the medium and is a necessary condition for content to be processed (Kahneman, 1973; Garaus 2019). Processing information begins with encoding it, i.e. paying attention to the content, to be able to store it in memory, and retrieve it (Lang, 2000). Hwang and Jeong (2018) found that memory performance depends on the attention paid to the medium. In their multitasking experiment, participants remembered the content of the so-called primary medium (the one to which more attention was paid) better than the content of the secondary medium. Similar results were found by Segijn, Voorveld, Vandeberg, and Smit (2017): here, multi-screening participants remembered ads on the screen they primarily paid attention to equally well as single screeners, but not if the main focus lay on the other screen. Such findings were confirmed in an eye-tracking study as well, where participants who looked at and paid

(7)

attention to the ad recognized the brand faster than those who did not pay attention due to multitasking (Guitart, Hervet, & Hildebrand, 2019). Two recent meta-analyses provide a summary of those findings and prove attention to be an underlying factor for decreased cognitive outcomes (Jeong & Hwang, 2016; Segijn & Eisend, 2019). When attention is divided between two screens, it can, therefore, be argued that the more attention is paid to one screen, the more likely the content displayed on that screen will be remembered and vice versa. Attention, thus, is proposed as a mediator for recognition and recall:

H2. a) Participants in a multi-screening condition will pay less attention to the ad than participants in a single screening condition, and b) this ad attention will mediate the effect of multi-screening on ad recognition and ad recall with lower ad attention leading to lower ad recognition and ad recall rates than high ad attention.

Personalization as a Motivational Factor

The eye-tracking study by Guitart et al. (2019) further showed that, as long as participants paid attention to the ad, it did not make a significant difference regarding brand memory whether they were multitasking or not. Thus, assuming that attention plays a key role in advertisement memory, advertisers need to find ways to gain viewers’ attention, even when they are distracted, as is the case in a multi-screening environment. One way to gain viewers’ attention to the ad, even when they are using two or more devices, could be to heighten their motivation (MacInnis, Moorman, & Jaworski, 1991), e.g. through personalization of the advertisement.

Besides the opportunity and the ability to process information, motivation to process plays an important role in both Lang’s limited capacity theory (2000) and the ELM (Cacioppo & Petty, 1984). Relevance of the message is an essential factor regarding motivation as the probability for irrelevant information to be processed decreases when people only have

(8)

limited cognitive capacity (Bang & Wojdynski, 2016). Furthermore, since media

multitasking, and multi-screening in particular, heightens the cognitive load, consumers are less likely to process the content centrally (Garaus, 2019). In contrast, if a recipient is highly motivated to process information, e.g. because the message is personally relevant, it is more likely that attention is directed towards the content or, to speak in the terms of the ELM, central processing is more likely to occur. Regarding media multitasking, there is evidence that recognition and recall do not decline when media users are motivated enough to process the content, e.g. an advertisement (Angell et al., 2016; Srivastava, 2013), and personalization of the content can motivate participants to allocate sufficient attention (Kazakova, Cauberghe, Hudders & Labyt, 2016). The goal of personalization is to increase attention and enhance motivation to process by giving the receiver the feeling that the message is meant for them personally, hence, making it more meaningful (Hawkins, Kreuter, Resnicow, Fishbein, & Dijkstra, 2008). To personalize a message, Hawkins et al. (2008) suggest three strategies: identification, which clearly refers to the receiver by picture or name, raising expectations of customization, e.g. by explicitly stating the individual suitability of a product and

contextualization, meaning the adjustment of the message to contextual variables of the receiver. These three means of personalization can induce a feeling of “me-ness” in viewers’ minds, i.e. the feeling of being personally addressed by the message (Maslowska et al., 2016, p. 77). As a result, the message becomes more relevant to the receiver, which, for example, increases click-rates and memory of the content of personalized banner ads significantly compared to generic banner ads (Bragge, Sunikka, & Kallio, 2012; Köster et al., 2015)

Bang and Wojdynski (2016) compared personalized to non-personalized

advertisements in a high and low cognitive demand condition and found that personalized advertising attracted significantly more attention especially under high cognitive demand. Transferred to the current study, a multi-screening situation would represent a highly

(9)

cognitive demanding task compared to the single screening situation because processing information coming from two channels sharing the same modality requires more resources (Mayer & Moreno, 2003).

With regard to multi-screening, according to Bang and Wojdynski (2016), ad

personalization “may play a role in attracting consumers’ attention to advertisements that they might otherwise miss” (p. 868), e.g. because they are distracted by another screen. Taking into account that viewers allocate more attention to personalized advertising, H3 is

formulated. Due to a higher cognitive demand or distraction, an increase in motivation through personalization is especially relevant in a multi-screening context for the advertisement to be processed. Therefore, we posit that:

H3. The effect of multi-screening on ad attention will be moderated by

personalization, in such a way that a personalized advertisement will lead to increased ad attention compared to a non-personalized advertisement, thus increasing ad

recognition and ad recall.

All proposed variables and the formulated relationships are summarized in a conceptual model (see Figure 1).

Methodology Participants and Design

An online-experiment was conducted employing a 2×2 between-subjects factorial design with multi-/single screening as an independent variable and personalization vs no personalization as a moderator, resulting in four groups: two multi-screening conditions, one with a personalized advertisement shown to participants, one with a non-personalized ad, and

(10)

Figure 1

Conceptual model: Effect of multi-screening on ad recall through ad attention, moderated by personalized advertisement

two single screening conditions, either with a personalized or non-personalized advertisement. Previous studies in the field were conducted with about 30 people per condition (e.g. Jeong & Hwang, 2012; Segijn, Voorveld, Vandeberg, & Smit, 2017; Voorveld, 2011), the minimum group size recommended for sufficient statistical power (Geuens & De Pelsmacker, 2017). For this experiment, 205 people were recruited and randomly assigned to one of the four groups, of which 127 completed the questionnaire and watched the entire stimulus, resulting in, on average, 31.75 participants per group. Participants were invited to the study via an online link distributed on Facebook, Instagram, WhatsApp, surveycircle.com, surveyswap.io, and e-mail. The sample consisted of 70.6% women and 29.4% men, mostly university

students (70.9%) and employees (23.6%), a few postgraduate students (3.9%), and trainees (1.6%). All participants were between 19 to 36 years old (M = 25.13, SD = 3.55), resulting in a slightly right-skewed, almost normal distribution. People that age are considered very likely to involve in multi-screening behavior on a regular basis (Segijn, Voorveld, Vandeberg, et al.,

Multi- vs single screening Personalized vs non-personalized ad _{Ad attention} Ad recall H3 H1 Ad recognition

(11)

1_{For links to the original video material see Appendix B.}

2017), sometimes even without noticing it, which makes it a natural behavior, not resulting in too much discomfort during the experiment. They adopted the internet and mobile digital devices at such an early stage of their life to make it an essential part of it. According to the Nielsen company report (2018), 18 to 34-year-olds spend more time on digital devices (44% of media use) than other screen media like TV (39% of media use). Thus, the setting at home with their own and most used devices is intended to be as natural as possible within the scope of this study, which increases the external validity concerning the performance of the target behavior.

Experimental Procedure and Stimuli

At the beginning of the study, participants gave their informed consent and were then asked to watch a video about the Top 5 Caribbean beaches1_{. In the multi-screening condition,} participants were additionally asked to search the internet for new sunglasses on their

smartphone while watching the video. The approximately four-minute video was disrupted after two minutes by a 30-second commercial for sunglasses1 in both multi- and single screening conditions. After having seen the video and the commercial, participants filled in a 10-minutes questionnaire to measure attention, free recall, and recognition of the advertised product and brand, the extent to which they felt personally addressed as well as some demographic information. Then they were debriefed and thanked for their participation.

The operationalization of the independent variable with the video clip disrupted by a short commercial follows a multi-screening study conducted by Segijn, Voorveld, Vandeberg, and Smit (2017). A four-minute video clip was regarded long enough to give participants in the multi-screening condition some time to search for sunglasses online before they are exposed to the advertisement, while still not being exhaustively long for participants to drop out of the experiment. The beach-ranking video was chosen because it was considered a

(12)

relatively neutral content, easy to follow and nice to watch. Additionally, it matched the commercial and the search task thematically, contributing to the contextual factors of

personalization. The sunglasses were thought to be a gender-neutral, often used product most people are familiar with, making it suitable for this experiment.

To operationalize multi-screening a computer or laptop and a smartphone were used. Smartphones, alongside TV, are the most often combined screens (40% of cases; Segijn, Voorveld, Vandeberg, et al., 2017). Furthermore, they are often used for information search, which is a prominent reason for media multitasking (Hwang, Kim & Jeong, 2014), and matches the task to search for sunglasses. Although most people engage in multi-screening while watching TV (Angell et al., 2016), the laptop or computer was chosen because it seemed more suitable than a TV to conduct an online experiment, while at the same time, with video streaming providers on the rise, computers are increasingly used as a substitute for television (ARD/ZDF Onlinestudie, 2019) and therefore, the computer was considered an adequate device for the purpose of this study. By letting participants use their own devices and giving them a natural online search task, multi-screening was operationalized with the most possible authenticity.

To manipulate the personalization of the advertisement, two different versions of a sunglasses commercial were created and displayed randomly across conditions. The “Your face is your money maker” campaign by the brand William Painter was chosen and the original commercial was cut in a non-personalized and personalized version using Adobe Premiere Pro. The difference between the two was that the speaker in the personalized version addressed the viewer with “you” throughout the advertisement, while he described only product and brand attributes in the non-personalized ad barely addressing the viewer. Product attributes, for example, were state-of-the-art materials, highest quality polarized lenses, 100 days trial, and a lifetime guarantee. Both commercials were held as similar as possible in all

(13)

other aspects, i.e. they were cut out of the same original commercial, were equally long, and overlapped regarding product attributes.

For this experiment, to respect respondents’ anonymity, the advertisement was not personalized using identifiable information like demographics, location-based, or online behavioral data. Instead, the second and third strategy by Hawkins et al. (2008), raising expectations for customization and contextualization, were employed: personalization was operationalized through an advertisement addressing the viewer directly to raise expectations of customization. Second-person pronouns instead of third-person pronouns were used, which has been proven to increase personal involvement and message recall (Burnkrant & Unnava, 1989) and should foster the feeling of “me-ness” (Maslowska et al., 2016).

Regarding contextualization, the ad was chosen to match the content participants were viewing and the task they were asked to engage in to provide a meaningful context for the receiver and thus to imply that the message is directed at the consumer (Hawkins et al., 2008). The intention of matching task and commercial was to serve as contextual cues that are

expected to increase personalization.

To test the manipulation of personalization, two pretests were conducted. In each case, 18 participants were asked to watch either the personalized or the non-personalized version of the stimulus and rate to what extent they felt personally addressed by the advertisement. There was no additional task given. A perceived-personalization scale was adapted from Maslowska et al. (2016) using three items measured on a 1 (not at all) to 7 (totally) Likert scale: “I think the advertisement was targeted at me”, “I recognize myself in the group the advertisement was targeted at” and “I felt personally addressed by the advertisement”. After the first versions of the stimulus did not bring the desired result, F(1, 17) = 0.05, p = .818, they were modified to create a sharper contrast between the personalized and

(14)

it in such a way that whenever the protagonist spoke, he addressed the viewer with “you”. On the other hand, all parts of the video directed at the viewer were cut out for the

non-personalized version, so the speaker was never addressing the viewer directly and was only talking about the sunglasses instead. The goal was to create a noticeable difference in how the viewer was approached. A second pretest was conducted. Although there were no significant differences between means of perceived personalization either, F(1, 17) = 1.25, p = .281, the changes led to a shift in the right direction (Mpersonalized = 3.62, Mnon-personalized = 2.64) and the stimulus was thus accepted for the main study.

According to Sigall and Mills (1998), non-significant results of a pretest do not automatically impede the interpretation. The authors argue that a manipulation check is especially recommended if the influence of the stimuli variations on the manipulated variable lacks plausibility (Sigall & Mills, 1998). In this study, the use of second-person pronouns was used to manipulate personalization, which is proven to be an effective means by previous research (Burnkrant & Unnava, 1989, Maslowska et al., 2016). O’Keefe (2003) argues that it does not matter if participants perceive a message manipulation, because if the variation of the stimuli is or is not present is not a matter of perception. What is important is the psychological state that results from the manipulation, here the feeling of being personally addressed.

Asking the participants specific questions about their perception of the message might not assess the adequacy of the manipulation appropriately (O’Keefe, 2003) because

personalization might have been perceived unconsciously as well.

Multi-screening conditions were not included in the pretest since twice as many people would have been necessary, which would have limited the number of people available for the main study. Furthermore, if someone is supposed to multi-screen or not was

considered a rather obvious manipulation, for which a pretest is not strictly required (Sigall & Mills, 1998; O’Keefe, 2003). Hence, there was no task related to the ad in the pretest, only the

(15)

message characteristics themselves were tested in a single screening situation. It is, thus, necessary to conduct the main study and to observe how participants respond when presented to the whole operationalization before making definite conclusions about the success of the manipulation.

Measures

Ad attention, ad recall, and ad recognition were measured with a questionnaire

following the video.The first variable measured was attention. Participants were asked to rate on a scale from 1 (not at all) to 7 (fully) how much attention they paid to the video. A second question asked them if they noticed an advertisement disrupting the video and if yes, to rate how much attention they paid to the advertising on a scale from 1 to 7. Self-report measures are adequate since participants seem to be able to estimate their attention level accurately, even correlating with eye-tracking data, when asked immediately after the exposure (Segijn, Voorveld, Vandeberg, & Smit, 2017). Secondly, free recall was measured with an open question, where participants could write down anything, they remembered about the ad. For every detail they remembered correctly, respondents scored one point (Kazakova et al., 2016; Segijn, Voorveld, Vandeberg, & Smit, 2017). Thirdly, there were two multiple-choice

questions asked to measure recognition: “Which of the following brands do you remember having seen in the advertisement?” with the answer options of four brands, including the correct brand and three others serving as distractors (Köster et al., 2015), and “Which of the following product attributes were mentioned in the ad?” followed by a list of nine attributes such as style, free shipping, 100 days trial, or highest-quality polarized lenses, of which only four were mentioned in the ad. Again, every right answer was granted with a score of 1. Free recall was measured before recognition to ensure that the cues given in the multiple-choice questions were not confounding free measures (Voorveld, 2011).

(16)

Covariates

Besides demographics like gender, age, and education, participant’s frequency of multi-screening might influence their memory performance in the experiment. Frequent multi-screeners might be trained and used to the situation, which could result in better

cognitive performance (Garaus et al., 2017). Hence, multi-screening behavior was added as a control variable. The questionnaire was based on an adapted scale for measuring the

frequency of multitasking during television viewing by Collins (2008), which was also used by Srivastava (2013). Participants were asked how often they engage in different multi-screening activities like “viewing television while using the Internet”, “searching the Internet for information while watching television” and “chatting on the Internet about television shows while viewing” on a scale from 1 (= “never”) to 7 (=”always”). An exploratory factor analysis was performed with principal axis factoring (PAF), resulting in one eigenvalue above 1, indicating that all items load on the same latent construct, explaining 73.28% of the

variance. Hence, a scale was created and proven reliable, Cronbach’s alpha = .81 (M = 4.14, SD = 1.33), showing that most participants multi-screen “about half the time” and 81.7% involved in multi-screening at least sometimes, while 70% watch television on their laptop at least sometimes (M = 4.10, SD = 2.00).

Another influential factor concerning the motivation to watch the video and ad might be interest in the topic. People who are interested in the topic probably pay more attention to the video, which might also influence the attention they pay to the advertisement (Srivastava, 2013). Thus, using a 5-point scale, it was asked to what extent participants found the video interesting. On average, respondents were moderately interest in the video (M = 2.83, SD = .94) and only slightly interested in the advertisement (M = 2.35, SD = 1.02).

(17)

Randomization Check

There were no differences between multi- and single screeners regarding gender, χ2(1) = 0.00, p = .991, age, F(1, 105) = 1.10, p = .299, or education, χ2(2) = 0.06, p = .972.

Although the high education level might not represent the mean education level in society, multi-screening prevalence is stronger among people with a higher than average education (Segijn, Voorveld, Vandeberg, et al., 2017), hence the sample can be considered relevant in the context of the study (Geuens & De Pelsmacker, 2017). Furthermore, as long as the education level is held constant across conditions through randomization, it can be controlled for. The same holds for gender. Although there were far more women than men among the respondents, this proportion was equally distributed across conditions and, therefore, held constant. Looking at the two conditions of personalization, neither were there any differences with regards to gender, χ2_{(1) = 0.13, p = .720, age, F(1, 105) = 0.14, p = .708, nor education,} χ2_{(2) = 0.16, p = .925.}

Moreover, the four groups did not differ in the above-defined covariate multi-screening frequency, with F(1, 105) = 0.41, p = .523 for screen conditions, and F(1, 105) = 0.03, p = .855 for personalization conditions. Nevertheless, multi-screening frequency was included as a covariate in every analysis but was never found to influence the results, which is why it will not be further discussed.

In contrast, the randomization checks for interest in the video showed significant differences between multi- and single screeners, F(1, 105) = 5.25, p = .024, with single screeners being more interested in the video (M = 3.02, SD = 0.93) than multi-screeners (M = 2.61, SD = 0.92). For interest in the ad itself, no differences were found (pscreen conditions = .951, ppersonalization = .170). Consequently, interest in the video was included as a covariate in all analyses.

(18)

Manipulation Check

To find out if participants used the device(s) they were told to use in each

experimental group, it was asked how many and which media devices they used during the study. If respondents in the single screening condition reported having used an additional device or if those in the multi-screening condition did not use two devices like they were supposed to, these cases were removed from the analysis. Eight “single screeners” and ten “multi-screeners” did not use the devices as instructed and were, thus, removed, resulting in a sample of 58 single screeners and 51 multi-screeners (n = 109). Although the exclusion of participants led to group sizes smaller than the aspired 30 (25 participants in the smallest group), this decision was made, since those who did not follow the instructions did not act according to the purpose of the experiment and were, therefore, not relevant in the context of the study (Geuens & De Pelsmacker, 2017).

To check if the manipulation of personalization was successful, i.e. if participants’ perceived degree of personalization differed between the non-personalized advertisement and the personalized one (De Keyzer, Dens & De Pelsmacker, 2015), the same three items by Maslowska et al. (2016) were used, like in the pretest. A scale was created after an

exploratory factor analysis (PAF) resulted in one eigenvalue above 1, indicating that all items load on the same latent construct, explaining a variance of 74.56%. The 3-item scale was proven to be reliable with a Cronbach’s alpha of .83 (M = 3.02, SD = 1.18). Hence, means between the groups with a personalized (M = 2.90, SD = 1.09) and non-personalized (M = 3.09, SD = 1.22) stimulus were compared in an independent samples t-test and showed no significant difference, t(107) = -0.88, p = .380. Thus, like in the pretest, the manipulation check failed to prove a successful variation of personalization levels in the advertisement versions even in the presence of the contextual factors that were missing in the pretest. Nevertheless, as stated above, what participants perceived and reported might not mirror the

(19)

actual manipulation of personalization (O’Keefe, 2003). Additionally, in the open question measuring recall, some participants left comments which prove that the personalization was noticed. With regards to contextualization, for example, one respondent stated that “the setting of the advertisement kind of fitted the original video I was watching; beach, hut, surfing, summer sunglasses, palm trees”. Another two commented that “[the ad] fitted well in the video since the speaker had a beach look and the video was mainly outdoor in the sun” and the ad “was interesting because the video was about beaches, and you would probably need sunglasses when going to a beach.” These statements show that the match between video and commercial was successful. The contextual factors might give the viewers the impression that they are shown a sunglasses commercial because they are watching a video about

beaches, increasing the personal relevance (Hawkins et al., 2008). Other participants left remarks about the direct addressing of the viewer, remembering “very direct communication”, that the sunglasses were “made especially for you”, and that “a man was speaking directly to the people”, indicating a successful manipulation.

Therefore, the analysis will show if differences between the groups concerning cognitive outcomes reach significance although the manipulation has not been perceived consciously nor been reported in the manipulation check.

Hypothesis 1 – The Effect of Multi-Screening on Recognition and Recall

To test Hypothesis 1a, i.e. to find out if multi- and single screeners differ in their ad recognition, a sum of all correctly recognized answers per condition was created and each participant was granted a point for every right answer, i.e. ticking the correct product

attributes among the 9 multiple-choice options (maximum of 4 points per person, see question about ad recognition in Appendix A). Subsequently, an ANCOVA was conducted with an alpha level of p < .05, including screen condition as a factor and ad recognition as a

(20)

dependent variable as well as video interest as a covariate to compare the ad recognition score means for multi- (M = 1.98, SD = 1.14) and single screeners (M = 2.41, SD = 1.06). Levene’s test of equality of error variances was not significant, Levene’s F(1, 107) = .03, p =.868, indicating that results can be interpreted without restrictions. The ANCOVA showed that multi- and single screeners did not differ significantly in recognition of the advertisement, F(1, 105) = 2.61, p = .109, when adding video interest as a covariate. To investigate what results would have looked like without controlling for video interest, all analyses were conducted without the covariate as well. Hence, a univariate ANOVA was conducted, Levene’s F(1, 107) = .14, p = .706, with screen condition as a factor and ad recognition as a dependent variable. Results were significant, F(1, 107) = 4.23, p = .042, implying that single screeners recognized more details from the ad than multi-screeners. These results are reported here because it was noticed that including or excluding video interest appears to be critical for the correlation between screen condition and ad recognition to be insignificant or significant. It, thus, seemed interesting to mention and will be further discussed later. Nevertheless, as video interest was not equally distributed across conditions, it has to be included in the analysis and only the results of the ANCOVA can be considered useful regarding hypothesis testing. Consequently, H1a is rejected (see Table 1).

For H1b, all answers to the open question measuring free recall were read and participants were given one point for every detail they remembered correctly using different categories, e.g. product details (product type and attributes), brand name (whole name or half of it, either William or Painter was accepted), details about the speaker (gender, looks,

attitude), the setting, tone or storyline of the advertisement. A maximum of 12 points could be reached and score means were calculated for multi- (M = 2.67, SD = 1.28, maximum of 6) and single screeners (M = 3.55, SD = 1.69, maximum of 7). A univariate ANCOVA, meeting the criteria for an analysis of variance with Levene’s F(1, 107) = 3.22, p = .08, included video

(21)

interest as a covariate and recall as a dependent variable. Here, the recall scores between the groups differed significantly, F(1, 105) = 7.62, p = .007, η2 = .07, meaning that single screeners recalled more details of the commercial than multi-screeners. Thus, H1b is confirmed (see Table 1).

Table 1

ANCOVA Results for the Effect of Multi-Screening on Ad Recognition and Ad Recall, Controlling for Frequency of Multi-Screening and Video Interest

Sum of Squares df Mean Square F p η2 Video interest Ad recognition Ad recall 4.69 2.10 1 1 4.69 2.10 3.96 0.91 0.049 0.341 0.04 0.01 MS-Frequency Ad recognition Ad recall 0.28 0.02 1 1 0.28 0.02 0.23 0.01 0.630 0.926 0.00 0.00 Screen condition Ad recognition Ad recall 3.10 17.53 1 1 3.10 17.53 2.61 7.62 0.109 0.007* 0.02 0.07 Error Ad recognition Ad recall 124.30 241.57 105 105 1.18 2.30 Total Ad recognition Ad recall 667.00 1338.00 109 109

Note. N = 109. MS-Frequency = frequency of multi-screening. Values marked with * are significant (p < .05).

Hypothesis 2 – The Mediating Effect of Ad Attention

H2a assumed differences in ad attention levels between conditions and was tested conducting a univariate ANOVA and ANCOVA with screen condition as the independent variable, ad attention as the dependent variable, and in the latter case, including video interest

(22)

as a covariate. The results of both analyses of variance are reported because there was a similar pattern observed as with H1a. For the ANOVA, Levene’s F(1, 107) = .13, p = .716, results seemed to be significant, F(1, 107) = 4.30, p = .041, but only until video interest was added as a covariate. The ANCOVA results, Levene’s F(1, 107) = .32, p = .572, showed that means of ad attention between multi- (M = 4.65, SD = 1.45) and single screeners (M = 5.26, SD = 1.61) did not differ significantly, F(1, 105) = 2.94, p = .089. Therefore, H2a cannot be accepted.

H2b proposed that a difference in ad attention affects ad recognition and ad recall and, thus, mediates the effect of multi-screening on memory. Model 4 in the PROCESS macro (Hayes, 2013) was used to estimate the indirect effect of multi-screening on ad recognition and ad recall mediated by ad attention with 10,000 bootstrap samples and a bootstrap confidence interval of 95% (Boerman, Willemsen, & Van Der Aa, 2017). The model contained screen condition as the independent variable, ad recognition as the dependent variable, and ad attention as a mediator. Without controlling for video interest, a full mediation effect occurred for ad recognition (indirect effect = -0.11, SE = 0.07, BCBCI [-0.260, -0.003]; direct effect = -0.33, SE = 0.21, p = .120, BCBCI [-0.74, 0.09]). However, like before, this effect disappeared when including video interest as a covariate. In the controlled model, the only significant relationship was the one between ad attention and ad recognition (b = 0.15, p = .031). The mediation became insignificant (indirect effect = -0.08, SE = 0.06, BCBCI [-0.22, 0.01]), with a confidence interval reaching from negative to slightly positive and, thus, including 0.

Repeating the model with ad recall as a dependent variable, the indirect effect was not significant either (indirect effect = -.08, SE = .08, BCBCI [-0.29, 0.03]). The direct effect of multi-screening on ad recall, however, stayed significant as seen in H1b (direct effect = -0.74,

(23)

Direct effect b = -0.27, p = .210

Indirect effect b = -0.08, 95% BCBCI [-0.22, -0.01]

SE = 0.30, p = .015, BCBCI [-1.33, -.15]). Consequently, H2b is not supported for both recognition and recall (see Figure 2).

Figure 2

Mediation Model with Indirect and Direct Effects of X = Screen Condition on Y = Ad Recognition and Ad Recall Mediated by M = Ad Attention, Controlling for Video Interest

Hypothesis 3 – The Moderating Effect of Personalization

The moderated mediation proposed in H3 was investigated using Model 7 in the PROCESS macro (Hayes, 2013) with 10,000 bootstrap samples and a bootstrap confidence interval of 95% (Boerman et al., 2017b). The model was run with screen condition as an independent variable, ad recognition as a dependent variable, ad attention as a mediator, personalization as a moderator and video interest and multi-screening frequency as covariates. The results showed neither a significant effect of multi-screening (b = -0.52, p = .223) nor personalization (b = 0.05, p = .906) on ad attention, neither was there an interaction effect between screen condition and personalization on ad attention (b = 0.01, p = .985). Again, there was no significant direct effect of screen condition on ad recognition (b = -0.27, p =

Ad attention Multi- vs single screening _{Ad recall} Ad recognition Direct effect b = -0.74, p = .015*

(24)

.210). Only the significant relationship between ad attention and ad recognition found above was confirmed (b = 0.15, p = .031). The index of moderated mediation was insignificant (index of moderated mediation = 0.002, SE = 0.10, BCBCI [-0.20, 0.21]).

The model was repeated with ad recall as a dependent variable. Here, the only significant correlation was the direct effect of screen condition on ad recall (b = -0.74, p = .015), confirming what has been found in H1b and H2b. The index of moderated mediation was insignificant (index of moderated mediation = 0.002, SE = 0.11, BCBCI [-0.26, 0.21]). Consequently, H3 is rejected (see Figure 3).

Figure 3

Moderated Mediation Model with Indirect and Direct Effects of X = Screen Condition, W = Personalization and X*W = Interaction Effect on Y = Ad Recognition and Ad Recall Mediated by M = Ad Attention, Controlling for Video Interest

See Figure 2 Ad attention Multi- vs single screening _{Ad recall} Ad recognition See Figure 2 Personalized vs. non-personalized ad Interaction of screen

(25)

Discussion

The present study investigated the influence of multi-screening (vs single screening) on ad recognition and ad recall, with ad attention as an underlying factor and examined if

personalized (vs non-personalized) advertising affects the attention paid to the ad, changing cognitive outcomes. Based on Lang’s (2000) Limited Capacity Theory and the Elaboration Likelihood Model by Cacioppo and Petty (1984), it was proposed that multi-screening impedes ad recognition and ad recall because it poses a distraction and requires more cognitive resources than single screening, leading to a less thorough content processing and worse memory (H1). It was further assumed that ad attention mediates this effect, meaning that less attention to the ad in a multi-screening situation leads to lower ad recognition and recall rates (H2). This is unless the motivation to process the ad is high. Personalization was identified as a means to increase motivation due to the individual relevance of the message for the receiver (Bang & Wojdynski, 2016) and to improve content processing, leading to better cognitive performance (H3).

To test these hypotheses, an online experiment employed a natural approach to operationalize multi-screening with participants’ own laptops and mobile devices. Personalization was operationalized by creating two versions of an actual commercial

embedded in a video, one addressing the viewer directly and using contextual cues (Hawkins et al., 2008), one without any personal appeal to the viewer.

Results showed a different outcome than expected. While there was no significant difference between the effects of multi-screening vs single screening on ad recognition, the influence was confirmed for ad recall: multi-screeners recalled significantly fewer details from the ad than single screeners. However, there was no mediation effect of ad attention, so the significant negative effect of multi-screening on ad recall was not explained by the model. The only significant influence found in the mediation model was the one of ad attention on ad

(26)

recognition, confirming the connection between the two, despite the lack of a significant relationship between screen condition and ad attention. Moreover, personalization did not moderate this relationship. Apparently, whether the advertisement was personalized or not had no impact on the attention participants paid to it in either screen condition and, therefore, did not increase memory performance. Thus, out of the three hypotheses, only H1b was confirmed, while H1a, H2, and H3 could not be accepted.

Interestingly, in the analyses it became clear that the level of interest participants had in the video they were watching, influenced ad recognition scores as well as attention paid to the ad. While a significant main effect of the screen condition on recognition and a full mediation effect were found when excluding video interest, this effect disappeared both times as soon as video interest was included as a covariate. This raises the question, whether interest might be the driver to pay attention to and process media content, even when multi-screening. In the context of ad recognition, video interest seemed to compensate for the fact that

participants were multi-screening and led to an insignificant difference in recognition scores compared to single screeners.

In contrast, video interest did not change the significant correlation between screen condition and ad recall. Here, multi-screening affected message processing negatively, even when viewers were interested in the video, impeding the retrieval of information. Since free recall measures function as an index of how well the viewer can retrieve a piece of

information without any cues at all, they require more cognitive resources than mere recognition, where several cues are given for the respondent to choose from (Lang, 2000). With that in mind, one explanation for ad recall depending on the screen condition in contrast to ad recognition depending on video interest could be that for the ability to freely recall the ad, the viewer needs to pay attention to the content without distractions, which is more likely in a single screening situation. However, regarding ad recognition, it might be sufficient to

(27)

occasionally switch between two screens to spend just enough resources to the media content to process essential cues. Therefore, interest in the video might serve as some compensation for the distraction by the mobile device, because it increases the relevance and motivation to process the content presented in the second medium as well (Srivastava, 2013). It would be of great interest to investigate the possible influence of interest in the content on attention and memory in multi-screening situations, especially because in this study video interest was not randomly distributed across conditions. Hence, a larger sample would be needed to control for possible individual differences between participants and groups to examine if the above-formulated assumption could be true or if it is a result of the small sample size.

Another limitation of this study is its operationalization of personalized advertising. Although the two strategies implemented, raising expectations of customization and

contextualization (Hawkins et al., 2008), were proven effective in previous studies (Burnkrant & Unnava, 1989, Maslowska et al., 2016), here, the manipulation did not reach significance. However, since the stimulus variation is not a matter of perception and rather a question of altering participants’ psychological state (O’Keefe, 2003) and the fact that some respondents recalled being directly addressed in the open question, it seems that manipulation was noticed, even if it was not measurable. Raising expectations of customization and contextualization might have been too weak for the majority of respondents to recognize or to be reported in the manipulation check. Participants’ personal data could have been used to further tailor the commercial individually. Nevertheless, using real information on participants’ previous online search behavior or demographics would have been beyond the scope of this study.

Secondly, an explanation for the unsuccessful manipulation check could be that respondents perceived the “feeling of me-ness” (Maslowska et al., 2016) unconsciously only. Respondents’ psychological state could have been manipulated without them being able to report it (O’Keefe, 2003). In such a case, however, there should have been differences in the

(28)

results between groups regarding memory performance and ad attention levels. Therefore, future studies should personalize the stimuli in a stronger way with personal data from

participants, making a successful manipulation more likely. Taking individual preferences and behavioral patterns into account might also heighten the personal relevance and interest in the ad, which seems to be a requirement for attention and memory. Nevertheless, if customized tailoring is the only effective way of personalization, the use of personal data raises ethical issues and the question of how much and what kind of data should be used for

personalization. Some studies found that a low or moderate level of personalization is generally appreciated by consumers as it increases relevance, but the more data is used and the more personalized the ad, the more likely are negative feelings of intrusiveness to arise (personalization paradox, e.g. Boerman, Kruikemeier, & Zuiderveen Borgesius, 2017; Zarouali, Ponnet, Walrave, & Poels, 2017). Consumers’ privacy concerns must, therefore, be taken into consideration when putting such approaches into practice.

Consequently, to find out if personalization is an effective means to reach the goal of raising interest and gaining attention in multi-screening situations, more research is needed. This study took a step in this direction in a threefold way. Firstly, it confirmed that the goal of advertisers must be to gain consumers’ attention in order for the ad to be remembered since ad attention and ad recognition are strongly connected. Secondly, how often participants multi-screen did not seem to play a role, but interest in the content surrounding the advertisement was identified as an important factor for viewers to be attentive. Thirdly, in the context of multi-screening, personalization methods that do not identify the individual by using personal data but rather take a general approach to induce a “feeling of me-ness” (Maslowska et al., 2016) might not be enough to gain viewers’ attention when used on their own.

It would be of great value to conduct another study with a larger sample and

(29)

also be interesting for both researchers and advertisers to find out if there is a threshold for a minimum of personalization to be noticed when studies already found that there is a

maximum level of personalization before it becomes intrusive (Boerman et al., 2017a). Furthermore, in addition to online banner ads that have received a lot of attention regarding personalization already (Bang & Wojdynski, 2016), some light could be shed on TV or video advertising in general, which was attempted in this study. With the rise of smart TVs, it might become more likely that viewers are exposed to personalized ads when watching TV as well. Here, it could be interesting to explore how video commercials as audio-visual stimuli gain multi-screeners’ attention. While it is already known, that peripheral (low-level) visual cues like banners encourage switching behavior (Brasel & Gips, 2017), it could be investigated if a personalized video clip can catch viewers’ attention via audio as well and if personalization discourages switching away from the ad, similar to more engaging high-level visual cues (e.g. faces). Further findings in this field will foster and expand the existing knowledge of how advertising works in digital, multi-screening environments and will, thus, be of great use for practitioners, too.

Additional remark: Although we are aware that the entire analysis could have been conducted with PROCESS Model 7, for the Master’s thesis, it was decided to take a step by step

approach and to analyze each hypothesis separately.

References

Angell, R., Gorton, M., Sauer, J., Bottomley, P., & White, J. (2016). Don’t Distract Me When I’m Media Multitasking: Toward a Theory for Raising Advertising Recall and

(30)

Armstrong, G., & Chung, L. (2000). Background Television and Reading Memory in Context. Assessing TV Interference and Facilitative Context Effects on Enconding Versus Retrieval Processes. Communication Research, 27(3), 327-352.

ARD/ZDF-Forschungskommission (2020). “ARD/ZDF-Onlinestudie 2019.” Retrieved from http://www.ard-zdf-onlinestudie.de/

Bang, H., & Wojdynski, B. (2016). Tracking users’ visual attention and responses to personalized advertising based on task cognitive demand. Computers in Human Behavior, 55, 867–876.

Boerman, S., Kruikemeier, S., & Zuiderveen Borgesius, F. (2017a). Online Behavioral Advertising: A Literature Review and Research Agenda. Journal of Advertising, 46(3), 363-376.

Boerman, S., Willemsen, L., & Van Der Aa, E. (2017b). “This post is sponsored”: Effects of sponsorship disclosure on persuasion knowledge and electronic word of mouth in the context of Facebook. Journal of Interactive Marketing, 38, 82-92.

Bragge, J., Sunikka, A., & Kallio, H. (2012). An exploratory study on customer responses to personalized banner messages in the online banking context. JITTA: Journal of Information Technology Theory and Application, 13(3), 5-18.

Brasel, S., & Gips, J. (2017). Media multitasking: How visual cues affect switching behavior. Computers in Human Behavior, 77, 258-265.

Burnkrant, R. & Unnava, H. (1989). Self-Referencing: A Strategy for Increasing Processing of Message Content. Personality and Social Psychology Bulletin, 15(4), 628-638. Cacioppo, J., & Petty, R. (1984). The Elaboration Likelihood Model of Persuasion. Advances

in Consumer Research, 11. Retrieved from

(31)

Collins, R. (2008). Media Multitasking: Issues Posed in Measuring the Effects of Television Sexual Content Exposure. Communication Methods and Measures, 2 (1-2), 65-79. De Keyzer, F., Dens, N., & De Pelsmacker, P. (2015). Is this for me? How Consumers

Respond to Personalized Advertising on Social Network Sites. Journal of Interactive

Advertising, 15(2), 124-134.

Garaus, M., Wagner, U., & Bäck, A. (2017). The Effect of Media Multitasking on Advertising Message Effectiveness. Psychology & Marketing, 34(2), 138–156. Garaus, M. (2019). The influence of media multitasking on advertising effectiveness. AMS

Review.

Geuens, M., & De Pelsmacker, P. (2017). Planning and Conducting Experimental Advertising Research and Questionnaire Design. Journal of Advertising: Themed Issue -

Methodology in Advertising Research, 46(1), 83–100.

Guitart, I., Hervet, G., & Hildebrand, D. (2019). Using eye-tracking to understand the impact of multitasking on memory for banner ads: the role of attention to the ad. International

Journal of Advertising, 38(1), 154–170.

Hawkins, R., Kreuter, M., Resnicow, K., Fishbein, M., & Dijkstra, A. (2008). Understanding tailoring in communicating about health. Health Education Research, 23(3), 454–466. Hayes, A. (2013). Introduction to Mediation, Moderation, and Conditional Process Analysis:

a Regression-Based Approach. New York, NY: The Guilford Press.

Hwang, Y., Kim, H., & Jeong, S.-H. (2014). Why do media users multitask?: Motives for general, medium-specific, and content-specific types of multitasking. Computers in Human Behavior, 36, 542-548.

Hwang, Y., & Jeong, S. (2018). Multitasking and task performance: Roles of task hierarchy, sensory interference, and behavioral response. Computers in Human Behavior, 81, 161–167.

(32)

Jeong, S.-H., & Hwang, J. (2012). Does Multitasking Increase or Decrease Persuasion? Effects of Multitasking on Comprehension and Counterarguing. Journal of Communication, 62, 571587.

Jeong, S.-H. & Hwang, Y. (2016). Media Multitasking Effects on Cognitive vs. Attitudinal Outcomes: A Meta-Analysis. Human Communication Research, 42, 599–618.

Kahneman, D. (1973). Attention and effort (Vol. 1063). Englewood Cliffs, NJ: Prentice-Hall.

Kazakova, S., Cauberghe, V., Hudders, L., & Labyt, C. (2016). The Impact of Media

Multitasking on the Cognitive and Attitudinal Responses to Television Commercials: The Moderating Role of Type of Advertising Appeal, Journal of Advertising, 45(4), 403-416.

Köster, M., Rüth, M., Hamborg, K.-C., & Kaspar, K. (2015). Effects of Personalized Banner Ads on Visual Attention and Recognition Memory. Applied Cognitive Psychology, 29, 181–192.

Lang, A. (2000). The limited capacity model of mediated message processing. Journal of Communication, 50(1), 46–70.

MacInnis, D., Moorman, C., & Jaworski, B. (1991). Enahncing and measuring consumers motivation, opportunity, and ability to process brand information from ads. Journal of

Marketing, 55(4), 32–53.

Maslowska, E., Smit, E. G., & van den Putte, B. (2016). It Is All in the Name: A Study of Consumers' Responses to Personalized Communication. Journal of Interactive Advertising, 16(1), 74-85.

Mayer, R. E., & Moreno, R. (2003). Nine Ways to Reduce Cognitive Load in Multimedia Learning. Educational Psychologist, 38(1), 43–52.

(33)

Nielsen Company (2018). “The Nielsen Total Audience Report Q2 2018.” Retrieved from https://www.nielsen.com/wp-content/uploads/sites/3/2019/04/q2-2018-total-audience-report.pdf

Segijn, C. M. (2016). Second Screen Advertising: A Typology of Multiscreening. In P. De Pelsmacker (Ed.), Advertising in new formats and media: Current research and implications for marketers (pp 77-96). Bingley: Emerald Books.

Segijn, C., Voorveld, H., & Smit, E. (2017). How Related Multiscreening Could Positively Affect Advertising Outcomes. Journal of Advertising, 46(4), 455–472.

Segijn, C. M., Voorveld, H., Vandeberg, L. & Smit, E. (2017). The Battle of the Screens: Unraveling Attention Allocation and Memory Effects When Multiscreening. Human Communication Research, 43, 295–314.

Segijn, C., Voorveld, H., Vandeberg, L., Pennekamp, S., & Smit, E. (2017). Insight into everyday media use with multiple screens. International Journal of Advertising: International Conference on Research in Advertising (ICORIA) 2016, 36(5), 779–797. Segijn, C., & Eisend, M. (2019). A Meta-Analysis into Multiscreening and Advertising

Effectiveness: Direct Effects, Moderators, and Underlying Mechanisms. Journal of Advertising, 48(3), 313–332.

Segijn, C., Araujo, T., Voorveld, H., & Smit, E. (2020). Related Multiscreening as a Strategy to Retain Audiences and Increase Persuasion During a Commercial Break. Journal of Broadcasting & Electronic Media, 64(1), 41–61.

Sigall, H., & Mills, J. (1998). Measures of independent variables and mediators are useful in social psychology experiments: but are they necessary? Personality and Social Psychology Review: an Official Journal of the Society for Personality and Social Psychology, Inc, 2(3), 218–226.

(34)

formatting cues in online environments. Computers in Human Behavior, 29(3), 888– 895.

Voorveld, H. (2011). Media multitasking and the effectiveness of combining online and radio advertising. Computers in Human Behavior, 27(6), 2200–2206.

Zarouali, B., Ponnet, K., Walrave, M., & Poels, K. (2017). “Do you like cookies?”

Adolescents' skeptical processing of retargeted Facebook-ads and the moderating role of privacy concern and a textual debriefing. Computers in Human Behavior, 69, 157-165.

Appendix

A. Questionnaire

Start of Block: Introduction Hello,

For my Master thesis in Persuasive Communication at the University of Amsterdam I am conducting research on media multitasking and advertising perception and I need your help! So, I would appreciate it very much if you took about 10 minutes of your time to participate in this study.

You are going to need your laptop or computer to watch a short video and answer a few questions. Please use your smartphone only when asked to do so.

As this research is being carried out under the responsibility of the ASCoR, University of Amsterdam, I can guarantee that:

1) Your anonymity will be safeguarded, and that your personal information will not be passed on to third parties under any conditions, unless you first give your express permission for this.

(35)

2) You can refuse to participate in the research or cut short your participation without having to give a reason for doing so. You also have up to 24 hours after participating to withdraw your permission to allow your answers or data to be used in the research.

3) Participating in the research will not entail your being subjected to any appreciable risk or discomfort, the researchers will not deliberately mislead you, and you will not be exposed to any explicitly offensive material.

4) No later than five months after the conclusion of the research, we will be able to provide you with a research report that explains the general results of the research.

Should you have any complaints or comments about the course of the research and the

procedures it involves as a consequence of your participation in this research, you can contact the designated member of the Ethics Committee representing ASCoR, at the following

address: ASCoR Secretariat, Ethics Committee, University of Amsterdam, Postbus 15793, 1001 NG Amsterdam; 020‐525 3680; ascor‐secr‐fmg@uva.nl.

Any complaints or comments will be treated in the strictest confidence. For more information or in case you have any questions or remarks about the research, feel free to email: mareike.weis@student.uva.nl

I hope that I have provided you with sufficient information. I would like to take this opportunity to thank you in advance for your assistance with this research, which I greatly appreciate.

Kind regards, Mareike Weiss

Informed consent

Before we begin, I need you to read the following and give your consent. You can exit the questionnaire here, if you do not agree to participate.

(36)

I hereby declare that I have been informed in a clear manner about the nature and method of the research, as described in the email invitation for this study.

I agree, fully and voluntarily, to participate in this research study. With this, I retain the right to withdraw my consent, without having to give a reason for doing so. I am aware that I may halt my participation in the experiment at any time.

If my research results are used in scientific publications or are made public in another way, this will be done such a way that my anonymity is completely safeguarded. My personal data will not be passed on to third parties without my express permission.

If I wish to receive more information about the research, either now or in future, I can contact Mareike Weiss (mareike.weis@student.uva.nl). Should I have any complaints about this research, I can contact the designated member of the Ethics Committee representing the ASCoR, at the following address: ASCoR secretariat, Ethics Committee, University of Amsterdam, Postbus 15793, 1001 NG Amsterdam; 020‐ 525 3680; ascor‐secr‐fmg@uva.nl.

o

I understand the text presented above, and I agree to participate in the research study. (1) Start of Block: Multi/Single screening Stimulus

Single screeners: In the following you are going to watch a short video. Please turn on

the sound of your computer.

Please do not use your phone during the video.

When the video is finished, you may continue with the questionnaire.

Multi-screeners: In the following you are going to watch a short video. Please turn on

(37)

While watching the video, please take your smartphone and search for sunglasses online like you would do when you are in need of a new pair. You may browse the internet as long as you like.

When the video is finished, you may continue with the questionnaire.

*Video randomly assigned, personalized or non-personalized*

Start of Block: Dependent variables

Video attention Thinking back to the video you just watched, what would you say how

much attention did you pay to the video? Please rate on a scale from 1 (= no attention at all) to 7 (= full attention).

I paid attention to the video...

Not at all (1) Fully (7)

1 2 3 4 5 6 7

Phone attention And how much attention did you pay to your phone? Please rate on a scale

from 1 (= no attention at all) to 7 (= full attention).

1 2 3 4 5 6 7

Filter question Did you notice an advertisement interrupting the video?

(38)

o

No (2)

Ad attention And how much attention would you say did you pay to the

advertisement? Please rate on a scale from 1 (= no attention at all) to 7 (= full attention).

1 2 3 4 5 6 7

Interest video Do you think the video you watched was interesting? Please rate your

o

Extremely (5)

Free recall Now, please think about the advertisement you just saw and write down

anything you remember about it (for example attributes or details about the product, the brand, the speaker, the setting, the tone, the story etc.).

If you cannot think of a word in English, you may also write it down in your mother tongue.

________________________________________________________________

aerospace-grade titanium (8) → correct (non-personalized version)

▢

"buy one, give one"-donation (9)

Personalization Now I am interested in how you perceived the advertisement.

To what extent do you feel the advertisement was targeted at you? Please rate the following statements on a scale from 1 (= not at all) to 7 (= totally).

I think the advertisement was targeted at me:

o

Not at all (1)

Media use In the following, I am interested in your everyday media use. Please answer

the questions as accurately as possible but without thinking too much. There are no right or wrong answers, it is only about your personal habits. Note that streaming videos, movies and series online counts as watching TV as well.

On a scale from 1 (= always) to 7 (= never), how often are you ... Watching television while using the Internet?

Searching the Internet for information while watching television? Chatting on the Internet while viewing television

Watching television on your laptop/computer? Watching television on an actual TV screen?