Mid-rolls and pre-rolls: the effects of different commercial
placement positions on online switching behavior, brand recall
and brand attitude on YouTube
Master’s Thesis
By: Floris Hessing, 10878203
Supervisor: P.C. Neijens
Graduate School of Communication
Persuasive Communication
29-01-2016
Wordcount: 10723
Abstract
This study investigates the effects of different ad or commercial placement positions on YouTube (pre-roll ads in front of a video vs. multiple types of mid-roll ads in a video) on in browser tab switching
behavior, brand recall and brand attitude. In addition to this, this article also studies the importance of perceived advertisement intrusiveness and irritation as possible factors that intermediate these effects. Furthermore, video liking is discussed as an interaction variable that influences the effects of commercial placement position on switching behavior and perceived intrusiveness. To test the proposed expectations an experiment with 176 participants has been performed. Results indicate that intrusive mid-roll ads can enhance brand recall and that mid-roll ads can lead to stronger perceptions of advertisement intrusiveness. Tab switching behavior was not influenced by commercial placement position or video liking and neither was perceived intrusiveness. To conclude, feelings of irritation were found to have negative effects on brand recall and brand attitude.
___________________________________________________________________
Introduction
Since Google took it over in 2006, the digital, international video platform of YouTube which was founded one year before, started commercializing at a rapid pace. Where
advertisements were nowhere to be found in the first few years of the website’s existence, these days intensive advertising has become the video website’s second nature. In terms of the types of ads that YouTube uses, three types of ads can be distinguished: 1) banner ads on the website, next to or above the videos, 2) small banner ads at the bottom of the video screen during videos and 3) video ads during or in front of videos.
It is this last type of ads that this study will focus on. A number of questions arise when Google’s current use of so-called “rolled” ads is analyzed. First of all for example, why does YouTube almost exclusively use pre-roll video ads (ads in front of videos)? And why is it that YouTube only chooses too use mid-roll ads (ads that start playing during a video, pausing the video for the duration of the ad) when videos are longer than about fifteen minutes? Furthermore, when a mid-roll video ad is used, how does YouTube decide at what time during a video this ad should be showed?
Direct scientific literature on what the answers to these questions may be,
theoretically anwering what the possible benefits of pre-roll ads - compared to mid-roll ads - are, is relatively scarce. In addition to this, sources that do inform about the (positive) effects
of certain types of commercial position placement seldomly discuss the effects of those placements in an online context. Taking this into consideration, this study seeks to analyze these effects. Specifically, one behavioral, one cognitive and one affective effect will be examined. Based on existing (in)direct literature on commercial placement positioning (on television for example) - this literature will be discussed in depth in the following section - it is argued that in browser tab switching behavior (clicking away to other internet tabs when an ad starts playing; behavioral), brand attitude (affective) and brand recall (cognitive) may be influenced when different types of such placement are used.
Moreover, this study also seeks to analyze the role of the extent to which someone likes a video or not, influences these effects. Although the number of educational and and informational videos on YouTube has grown - and it remains to do so - over the years, YouTube is mainly used as an entertainment platform. Connecting this idea of the phenomenon of video liking to our central interest of the effects of different commercial placements, the following question arises: does video liking influence the described possible effects of different commercial placement positions?
Based on the here above, from here on this study tries to answer and analyze the following research question:
RQ: What are the effects of different commercial placement conditions (mid-roll vs. pre-roll) on in
browser tab switching behavior, brand recall and brand attitude and which interaction role does the extent to which a person likes a video or not play in predicting those effects?
Theory
Browser tab switching behavior
As mentioned above, the general aim of this study is to investigate the (behavioral, cognitive and affective) effects of different advertisement timings or placement positions in online (YouTube) videos. In order to be able to do this, it is important at first to bundle what is known about the effects of different times of showing ads so far. This bundling however, is not so easy. The amount of literature about the specific topic of advertisement placement position is relatively thin, and the available literature mainly focuses on the effects of
Ruster, 1992). So far, only one relevant article can be found that studies the effects of
advertisement placement position in an online context (Kim, 2015). The results from both the older television based studies, as well as from the newer, more specific study however, are generally in line with each other in terms of what they mention about the effects of different timings or placements of ad/commercial breaks. To understand those effects, it is necessary at first to highlight which different moments or types of advertisement placement position are mentioned in the literature.
In his dissertation on switching or zapping in commercial breaks on television, Lex van Meurs (1999) mentions that there are two types of advertising breaks. These can be categorized in the so-called “centre breaks”, which interrupt a programme for a certain amount of time to show a number of ads, only to pick up the programme where it left off again once the ads are shown, and the “beginning (or end) breaks” (sometimes called “head or shoulder breaks”) which take place in between different (television) programmes. In addition to this simple categorization, more recent literature – mainly professional and often specialized on the YouTube domain in which people usually only come to watch relatively short videos, instead of entire programmes – has renamed the digital equivalents of the ad types mentioned above into pre-roll, mid-roll and post-roll (video) ads. The only true practical difference between this newer categorization and the older one is – next to the naming – the fact that the newer “roll” ad breaks usually only contain one ad, whereas the old ad “breaks”, practically always consist of multiple ads. In general however, television ads and YouTube ads are pretty much similar in terms of how they are placed in, or in front a video or program.
Considering the facts that the behavioral interest of this study lies in the effects of differently positioned ads on online switching behavior, and that - as discussed - there are no studies yet that focus on online, in browsers “tab” switching behavior, it is important at this point that the similarities between online tab switching and traditional TV-switching or “zapping”, are explored as well. In other words: in which ways are television switching and switching to other so-called website tabs in a browser comparable to each other? Although there are probably more, this study argues that the answer to this question has at least four dimensions. First all both types of switching require personal action in the form of a mouse click or a click on the television or a remote control. Second, in relation to the
aforementioned point, switching to other tabs or channels is extremely easy and can be done in a matter of seconds. Third, both types of switching offer an immense assortment of alternatives (either other websites, or other channels) people can switch to. And fourth, after having switched, both media - this is the most important point - prevent people from seeing the page that they have switched away from. In the case of online switching, the audio of other internet pages can still be heard after having switched.
tab switching, it is argued that the existing literature on the effects of different ad placements on television switching behavior form a good theoretical basis to design hypotheses by on the effects of differently placed ads on digital switching behavior. Continuing, based on the relatively univocal literary observations concerning the effects of differently placed ads on television, generally reading that people are a lot less eager to switch during a centre break than in a beginning or end break (for example: ITV Network Centre, 1993; Van den Berg & Ruster, 1992), the following is subsequently hypothesized:
H1: People who see a mid-roll video ad during a YouTube video are less likely to switch to another internet tab than people who see a pre-roll video ad.
Brand recall
As the research question that is posed above already suggested, more effects of different YouTube ad placement positions than only those in terms of changes in switching behavior are expected. As previously noted, two other conceptual fields of interest also form the basis of this article. Starting off with the cognitive side of this interest, the central concept of brand recall thus needs to be elaborated.
To start off with the basis, brand recall as a concept is as old as the entire field of commercial communication itself. After all: commercial (marketing) communication is strongly related to the concept of brand recall as it simply expresses one of the important things that companies desire: being (well) remembered with their brand(s). To make things more clear, the concept of brand recall is nothing more than the ability of a person to name a certain brand in a specific product category. Together with brand recognition, which is the ability to “tell a brand correctly if they ever saw or heard it” (Chi, Yeh & Yang, 2009), brand recall is a part of the bigger (bipartial) umbrella concept of brand memory, which in turn is an important - the only, according to some authors (Aaker, 1996) - indicator of the even bigger umbrella concept of brand awareness (for example: Chi et al. 2006).
In past studies, brand awareness and the ability to remember a brand (as a part of brand awareness) have proven to be good indicators of a person’s intention to buy a certain product, or simplified, purchase intention (PI). Chi, Yeh and Yang, (2009) for example find that brands that are better known are often accompanied by higher levels of purchase intention than brands that are less well known. In addition to this, Aaker (2009) found that brand awareness can lead to loyalty towards a brand. This so-called brand loyalty can subsequently lead to higher levels of purchase intention. These studies however are only examples which prove the existence of the interconnectedness between brand recall and purchase behavior. Furthermore, it can be argued that it is because of this
interconnectedness with purchase intention that brand recall is such an important literary concept in the field of marketing.
Now that the importance of brand recall as a concept is grasped, the literary sources discussing the effects of different commercial placements on brand recall can be discussed. In contrast to the extensive amount of articles on factors that influence brand recall in general, only a hand full of articles has been written about the possible recall effects of different moments of advertisement placement position specifically. Moreover, most of these articles are relatively outdated and television centered, meaning that they are focussing more on switching behavior (which has been discussed in the previous paragraph), than on brand recall. As previously noted, only one article so far has studied the influence of
advertisement placement position in the YouTube domain. This article however does not directly relate to brand recall and is thus only marginally applicable (Kim, 2015).
Taking the above into account, it is relatively difficult to voice clear expectations and to build hypotheses on the relationship between YouTube ad placement position and brand recall solely on direct preexisting literature. Literature on the effects of (disclosures of) product placement prominence on brand recall however, provides a lot of information on the ways through which brand recall might be influenced by commercial placement positions. In this sense, both Gupta and Lord (1998), as well as Van Reijmersdal (2009) for example argue that prominent product placements are better remembered than subtle product placements. One of the most important features of prominence that these authors mention, is the fact that (placement) prominence, in the widest sense of the term, is about noticability.
Noticability in turn it can be argued, is strongly related to the simple concept of attention. After all: it is easier to attent to something that is easily noticable, than to
something that is not. Continuing with this concept, attention is directly linked to brand recall. Logically, a person can only remember a brand that is shown in a commercial or in a product placement if he or she (actively) attends to it (Anderson, 1995). Based on this relationship between noticability and attention, it can be argued that the use of an ad that scores high in terms of how noticable it is, is a better enhancer of brand recall than an ad that is less noticable. This brings us to the introduction of the concept of intrusiveness. This study argues that high levels of perceived ad intrusiveness, which is defined by Ha (1996) as “the degree to which advertisements in a media vehicle interrupt the flow of an editorial unit”, are intrinsically linked with better noticability. Subsequently, this better noticability goes hand in hand with the increase of the number of opportunities of an ad to grab someone’s attention. Indirectly the use of intrusive ads might thus lead to enhanced brand recall. In accordance with this trail of assumptions Acquisti and Spiekermann (2011) have also found direct brand recall enhancing effects of intrusiveness on brand recall. In addition to this the use of
of brand recall as a goal (Stafford and Faber, 2005).
Based on the line of argumentation above and in light of the plausible position that a mid-roll ad is more intrusive in terms of interrupting an editorial unit - in this case, a YouTube video - than a pre-roll ad, as it actually divides a video in two parts, the following hypothesis about the relationship between commercial placement and brand recall has been developed:
H2: People who see a mid-roll ad in a YouTube video will be able to recall a presented brand better
than people who see a pre-roll ad in front of a YouTube video.
The hypothesis as it is posed above, strongly leans on a series of assumptions, not all of which are based on direct literature. The most important of those assumptions reads that mid-roll ads are likely to be perceived as more intrusive than pre-roll ads. Although the line of argumentation on which this assumption is based, is clear and also plausible, the following hypothesis will be tested as well because of the importance of this assumption:
H2a: Mid-roll ads are perceived as more intrusive than pre-roll ads.
Brand attitude
Brand recall is important because it helps customers remember a brand or not. Equally important is brand attitude. Brand attitude as a concept is defined the easiest as “an individual’s internal evaluation of [a] brand” (Mitchell & Olson, 2000). As for brand recall, previous studies on the concept have often elicited the tight relationship between brand attitude and purchase intention. Spears and Singh even point out that the concepts of brand attitude and purchase intention may be so intertwined, that they actually overlap concept wise. In an additional definition they define brand attitude as an “attitude toward [a] brand is a relatively enduring, unidimensional summary evaluation of the brand that presumably energizes behavior” (Spears & Singh, 2004). This last presumption clearly underlines why brand attitude is such an interesting and important topic.
In light of the scientifically agreed upon interconnectedness between brand attitude and purchase intention mentioned above, a multitude of factors that were thought to influence a person’s brand attitude has been studied in the past decades. One factor that plays an important umbrella role in this search for information, is attitude towards the ad or ad attitude. According to multiple studies (for example: Mitchell & Olson, 1981; Brown & Stayman, 1992) the attitude that someone has towards an ad influences the attitude that
someone has towards a brand. The attitudes that someone forms towards an advertisement are so to speak transfered from an ad, to a brand (Muehling, 1987). In many cases
influencing a person’s internal evaluation of a brand can thus be achieved by influencing a person’s evaluation of the ad in which a brand is showcased.
The list of tools - or subfactors - that have been studied as possible influencers of either ad attitude indirectly or brand attitude directly, is long.
In terms of the effects of the subfactor of advertisement placement on brand attitude however, not much direct literature is at hand. As mentioned before, advertisement placement position is discussed almost exclusively in terms of its consequential switching behavior. Two indirect lines of argumentation however help predict the possible effects of different advertisement placement positions on YouTube. First of all it can be noted that brand attitude and brand recall often react on advertisement elements in inverse manners. In other words: some advertisement choices that are beneficial in terms of improving a
consumer’s levels of brand recall can have negative effects on a consumer’s brand attitude and vice versa. In line with this, Gelb and Zinkhan (1986) indeed find that the use of humor commercials leads to more positive brand attitudes, but also to less ad remembrance. Severn, Belch and Belch (1990) find that the use of sexual cues can lead to more positive brand attitudes, but also to lower levels of copy point recall. In terms of product placement, Russell (2002) finds that placements that are incongruent with the plot and the context in which they are placed are remembered better than placements that are more congruent with a plot. This incongruent nature however can raise awareness and reactance, which can lead to more negative brand attitudes. Related to this, Boerman, Van Reijmersdal and Neijens (2014) more recently found that the inclusion of product placement disclosures in videos with sponsored content, can lead to a better recall of the sponsored brands, but also to more critical evaluations of those brands.
In addition to this inverse behavior notion as a first line of argumentation, other grounds for assumptions can be found when looking past advertisement placement position as a variable itself. As elaborated in the previous section on brand recall, different moments of advertisement placement position are likely to be perceived differently in terms of
intrusiveness. Continuing, in a study on product placement in advergames, intrusiveness was found to be a strong indicator of brand attitude (Hernandez, Chapa, Minor, Maldonado & Barranzuela, 2004). Another study on product placement in advergames furthermore found a significant relationship between the perceived intrusiveness of a placed brand and irritation. This irritation can subsequently lead to more negative brand attitudes (Martí-Parreño, Aldás-Manzano, Currás-Pérez & Sánchez-Garcia, 2012; Stafford & Faber, 2005). To conclude, all of the above is also in line with the statements of Ritter and Cho (2009) who found that ads in the beginning of an audio podcast are usually perceived as less intrusive and less irritating
than ads that are placed in the middle of such a podcast.
Based on the previous propositions the following hypothesis is formulated in terms of the expected relationship between YouTube commercial placement position and brand attitude:
H3: People who see a YouTube video containing a mid-roll video ad will have a more negative
brand attitude towards the brand in that ad than people who see a pre-roll video ad.
Intrusiveness and irritation
Since both intrusiveness as well as - to a lesser extent - irritation, have been
mentioned as important factors that are expected to intermediate and shape the relationship between commercial placement position, brand recall and brand attitude, this study deems it necessary to also investigate the direct effects of these factors. In addition to our earlier hypotheses, this study thus also seeks to explore whether high levels of perceived
advertisement intrusiveness directly lead to enhanced brand recall as found by authors as Acquisti and Spiekermann (2011) and Stafford and Faber (2005). Also in terms of
intrusiveness, this study seeks to find out whether higher levels of perceived intrusiveness lead to more negative brand evaluations as found by Hernandez et al. (2004), Martí-Parreño et al. (2012) and Stafford and Faber (2005). In light of all this literature, the following
hypothesis will be tested:
H4: High levels of perceived intrusiveness will lead to higher levels of brand recall, but also to more
negative brand attitudes.
As mentioned above, the importance of advertisement irritation as a possible
intermediating predictor between commercial placement position and brand recall and brand attidude will be studied as well. In terms of what is known about advertisement irritation so far, a multitude of authors describe a certain interrelatedness between perceived irritation and perceived intrusiveness. In addition to what the authors in the above paragraph on brand attitude already said, both Li et al. (2002) as well as Edwards et al. (2002) find that popup ads on the internet, which have a very interruptive nature - comparable to mid-roll ads - can directly lead to advertisement irritation. When this ad is bigger - and thus more
intrusive - this ad is likely to be perceived as more irritating (Chatterjee, 2008). Based on these sources which suggest that perceived intrusiveness and feelings of irritation are interrelated indeed, the next mediation hypothesis is developed:
H4a: Irritation with an ad will mediate the relationship between perceived advertisement intrusiveness
and brand recall and brand attitude.
Video liking
As mentioned in the introduction of this piece, this study also seeks to shed light on the importance of the extent to which the liking of the video in or in front of which an ad is placed, in terms of influencing switching behavior, brand recall and brand attitude. Before those alleged effects are discussed, it is first of all important to understand what video liking as a concept actually is. To facilitate this understanding this study will adopt an amended version of the definition Murry, Lastovicka and Singh (1992) use of program liking. According to Murry, Lastovicka and Singh, program liking is a “summary evaluation of the experience of viewing a television program”. Based thereon, this study will define video liking as “the summary evaluation of the experience of viewing a (YouTube) video”.
Continuing with the effects of video liking on our first factor of interest of tab switching behavior, it has to be noted that literature again is rather thin. As for the relationships
between television switching behavior and commercial placement position, two interesting findings however can be cited. Lex van Meurs (1999) for example argues that the reason why people are less likely to switch during a centre break, might stem from the fact that such breaks make people more afraid of missing a part of the show they are watching, than beginning or end breaks. Building furthter on this, this study argues that people will be more afraid to miss a part of something they like, than of something that they do not like.
Secondly, it is also found that someone is less likely to switch channels, when something informative or enjoyable is actively keeping him or her busy (Woltman Elpers, Wedel & Pieters (2003). Assuming that a video that a person has already started watching meets this criterion better than a video that has not yet started, the following interaction hypothesis is developed:
H5a: People who see a mid-roll video ad and like the video in which this ad is placed are less likely to
switch away to another browser tab than people who see a mid-roll ad who do not like this video.
To conclude this theory section and to add even more to the existing literature on the effects of commercial placement position, our final hypothesis will not focus on the
interaction effects of video liking on the relationship between commercial placement position and brand recall and brand attitude directly. Instead however this study will try to seek whether a person’s video liking levels influence the extent to which this person perceives a
certain ad as less or more intrusiveness. Moreover, one literary source seems extremely useful to predict the possible interaction effects of video liking at this point. Edwards, Li and Lee (2002) argue that ads that are more conflicting in terms of how strongly they interfere with a person’s “goals” - these can be informational as well as hedonic - are perceived as more intrusive. In other words: people who feel that their goals - for example of enjoying a video - are interrupted more, are likely to perceive an ad as more intrusive than people who think that their goals are not interrupted at all. Based on this, the following is assumed:
H5b: People who see a mid-roll video ad and like the video in which this ad is placed will perceive
this ad as more intrusive than people wo see a mid-roll video ad who do not like this video.
Methods
Design
To test the hypotheses that have been posed above an experimental 3 (commercial placement position; commercial placement before a video vs. commercial placement during a silent moment in a video vs. commercial placement halfway through a sentence during a video) x 2 (video liking; low video liking vs. high video liking) between subjects factorial design was employed. In the experiment the indepentent variable of commercial placement position was manipulated (for details, see the stimuli section below). The moderator variable of video liking was selected by performing a median split across the video liking scores of our participants. This led to a division where all the participants who had video liking scores of 5.50 and lower, were marked with a 1 (low video liking), whereas people who had scores higher than 5.50 were marked with a 2 (high video liking).
Procedure
Pretests
Before the definitive survey was launched, two small pre-tests, both with different goals, have been carried out. The first of these pretests was used to verify whether the assumption was true that the brand Sonos was relatively unknown amongst the Dutch population. This was done by asking (a total of) 19 students from the Dutch city of Leiden if
they had ever heard of the brand Sonos. Based on the results of this test (only three (15.9%) students recognized the brand) Sonos was selected to be this study’s target brand
The second pretest that was performed was merely a test run of the survey. Based on the unanymous observations of the three testers that were consulted it was decided to shorten the created video stimuli by 1 minute from 5:30 to 4:30 to make it less monotonous. No other changes had to be made based on the observations of these three testers.
General procedure
As denoted before, this study’s data has been gathered by the means of an experimental online survey created by the digital survey building tool Qualtrics. If an individual would click on the hyperlink leading to the survey, he or she would first of all be informed about the survey’s ethical code in terms of anonymity of data use. The participant would also be informed about the goal of the study, said to be the acquiring of information on the effects and consequences of funny online (YouTube) videos. After agreeing with the terms of participation, participants were told that they were going to watch a fragment of a Dutch comedian, which they would be questioned about afterward. After being exposed to one of the stimulus videos (see stimuli), the participants were asked a number of questions. The order in which the different variables of our interest were asked to our participants is generally the same as the order that is used in the operationalization section below. Video liking however was asked before all other questions were asked, just after the video. Also tab switching behavior was measured after the questions that were set up to measure irritation were asked. On average it took a participant 13 minutes to finish the survey.
Participants
Initially 504 participants started with our survey, all of which have been gathered through convenience sampling, using distribution through and across Facebook, online internet forums and personal contacts. The data of 328 of those 504 participants was not used in our analyses because the respective participants either did not finish the survey (N = 308), were too young (N = 11) or because they filled in the survey incorrectly or after the final collection date of 1-4-2016 (N = 9). In total, this left 176 participants of which the data could be used. Our study’s relatively high dropout rate might be explained by our survey’s relatively high participation threshold, including the watching of a 5 minute video with audio. Of the 176 participants that were not dropped from our survey 74 were male, whereas 102 were female. The age of the participants ranged from 18 years old to 68 years old with an
average age of 31.74 (SD = 13.29). In terms of level of education, 2 participants had completed a LBO level education, 12 participants had completed either a VMBO or MAVO level education, 28 participants had completed a VWO or a HAVO level education, 40 had completed a HBO level education, 79 people had completed a WO level education and 15 had completed a different type of education. Most of our participants had a Dutch nationality (N = 173). Only 3 of our participants had other nationalities (Belgian, N = 2; Romanian, N = 1). All of our participants were randomly appointed to one of our three experimental groups. After the the removal of the mentioned participants of our sample, our experimental groups contained 63 (condition 1; pre-roll placement), 54 (condition 2; mid-roll “in silence”
placement) and 59 (condition 3; mid-roll “in sentence” placement) participants.
Operationalization
Stimuli
Three videos have been developed to test this study’s hypotheses. All of the videos that have been developed exist out of two parts: a (normal) video part and a commercial part. The only differences between the three videos that were created, are - as the foregoing already makes clear - the placements of the commercials in relationship to the (normal) videos. The participants that got to see video 1 (experimental group 1) had to watch the selected commercial (more on this commercial later) before the “normal” video started playing. Experimental group 2 had to watch the same commercial, but this time it started playing during the video. This is also what happened for group 3. Where the video ad for group 2 was placed in a silent - inbetween two sentences - moment in the video however, in condition 3 the video ad popped up halfway through a sentence, just before the punchline of a joke. By placing the commercials in the videos as described above it was attempted to manipulate the perceived feeling of intrusiveness across our three experimental groups.
The video that all our participants had to watch, was a 4-minute fragment of an act by a Dutch comedian. This fragment, featuring jokes about the behavior of people when they are at a party, was selected because of two reasons. First of all it was argued that comedy is a very subjective field. It is not uncommon that one group of people likes a certain style of humour, whereas another group completely dislikes this style. By selecting a fragment of a comedian it was thus hoped that it would become easier to divide our total pool of
participants into two (clear) groups, either consisting of those who liked the video, or of those who did not like the video. The second reason for selecting a “humorous” fragment leaned on the fact that it was believed that placing a commercial halfway through a joke would be one of the most effective (and easiest) ways to manipulate perceived intrusiveness. In other words: a piece of comedy was thought to be ideal for creating our third, “very intrusive”
experimental condition.
The commercial that was inserted in this fragment then was a 30-second Superbowl commercial featuring the home audiosystem brand Sonos.There are several reasons for picking this brand and this commercial. First of all, Sonos as a brand is relatively unknown in the Dutch audio market. By picking a relatively unknown brand it was attempted to eliminate or at least minimize the effects of existing brand familiarity and maybe even preexisting brand attitude - both variables which might dillute the results of our study - on brand attitude and brand recall. In addition to this, it was also believed that audio systems - product
category wise - are not necessarily, age or gender bound. Enjoying high quality music after all is something that anyone can enjoy, irrespectively of age or gender. Selecting a brand from this specific sector was thus also believed to minimize the possible effects of those control variables on our outcome variables. At last this specific Sonos commercial was chosen because it only mentioned the brand name on screen. This way people who only listened to the ad - people who clicked away to other tabs - would not be able to remember the brand.
To prevent people from skipping the ad, or discovering that the ad was a part of the video instead of a regular ad, the typical red YouTube timeline was removed from the YouTube player. This is a technical option Google offers when people want to embed a certain video somewhere online. Furthermore, to make the video as realistic as possible, the message that the ad would only take 30 seconds could be seen at the bottom of the screen when the Sonos commercial was playing.
Switching behavior
During the ad switching behavior was measured by asking a series of questions. First of all, the participants were asked to truthfully answer the question if they had clicked away to another tab, or if they had looked on their phone at some point during their particpation till then and if so, how many times. If they had not done so, people could continue with the rest of the survey. If a participant had done so however, he or she was asked two more
questions: “where did you click away to?” (this could be answered by “another tab”, “my phone” or “another computer programme”; multiple options could be selected) and “at what time during your participation did you click away?” (this could be answered by “when filling in the survey”, “during the video, when Micha Wertheim was talking” and “during the video, when the ad was playing”; multiple options could be selected). For every participant that marked down that he or she clicked away when the ad was playing, a “1” was recorded. For every other participant, a “0” was recorded.
Brand recall was measured both by asking the participants the open question if they remembered the Sonos brand they had seen in the video (unaided brand recall), as well as by letting them select Sonos from a list of seven brands (Bose, JBL, Sony, Sonos, Samsung, LG, Yamaha) that operate in the same sector (aided brand recall). For unaided brand recall, which was recorded first, a “1” was marked down as when a participant was able to
succesfully recall the Sonos brand, whereas a “0” was marked down when a participant was not able to do this. For aided brand recall, where the respondents were deliberately asked to tick all the brands they saw in the video, people could only score a “1” (“remembered”) if they only ticked the Sonos box only; people who ticked more than one brand were always marked with a “0” even if Sonos was among those brands. This way people who guessed the brand right by chance were filtered out.
Brand attitude
To measure brand attitude a shortened five item scale that is based on a longer - but very repetitive - 31-item scale from Spears and Singh (2004) was adapted and used in this research. On a seven-point semantic scale ranging from the most negative answer (in this case a “1” would be marked down) to the most positive one (in this case a “7” would be marked down), the participants were asked to indicate if they thought Sonos was either: unappealing or appealing, bad or good, unpleasant or pleasant, unfavorable or favorable and if they thought Sonos was unlikable or likeable. A principal component analysis (PCA) reported that the five items together indeed formed a 1-dimensional scale. Only one
Eigenvalue higher than 1 was observed (3.67, explaining 73.43% of the variance). Also, the scale has a Cronbach’s alpha of .89, thus indicating that it is a very reliable (>.80) scale.
Intrusiveness
To measure (perceived) intrusiveness, a translated version of an existing 7-item Likert scale introduced by Li, Edwards and Lee (2002) was used. Participants were asked to answer to what extent they found that the Sonos ad was distracting, disturbing, forced, interfering, intrusive, invasive and obtrusive at the moment that it was shown. For each individual question, participants could indicate whether they perceived the ad as “not at all ...” (1), as “very much …” (7) or any full number inbetween. As for the previous measures, low scores for this measure indicate low levels of perceived intrusiveness (with a minimum score of 1), whereas high measures indicate high levels of intrusiveness (with a maximum score of 7). Furthermore, a principal component analysis (PCA) has been performed to test the expected one-dimensionality of this scale. This analysis showed that our measure of intrusiveness was indeed one-dimensional. Only one component with an Eigenvalue higher than 1 (4.64, explaining 66.24% of the variance) was found. A reliability test of this construct
has measured a Cronbach’s alpha score of .90.
Irritation
Irritation towards the ad was measured by using a slightly altered and translated version of an existing 3-item scale that was used by Chakrabarty and Yelkur (2006). Instead of using 6-point Likert scales however, 7-point Likert scales (1 = completely disagree; 7 = completely agree) were used to offer respondents a neutral option, as well as to improve the overall consistency of the survey. The statements the participants had to give their opinion about were “I like the ad”, “This ad was irritating” and “This ad was stupid”. Of those three items the first had to be reversed. This way low scores for this construct (with a minimum score of 1) indicate low levels of irritation with the ad, versus high scores (with a maximum of 7) indicate high levels of ad irritation. A principal component analysis, testing the presumed one-dimensionality of this scale, confirmed that our irritation measure was one-dimensional. Only one component with an Eigenvalue higher than 1 (2.28, explaining 76.00% of the variance) was found. Furthermore a reliability test of this construct measured a reliable Cronbach’s alpha score of .84.
Video liking
Due to the nature of the video that was chosen, video liking was measured by asking six questions that were originally used to measure perceived humour in an advertisement (Chattopadhyay and Basu, 1990). On semantic scales ranging from 1 (the most negative option) to 7 (the most positive option) the participants were asked if they found the Micha Wertheim fragment they had to watch either humorous or not humorous, funny or not funny, playful or not playful, amusing or not amusing, dull or not dull and boring or not boring or anywhere in between. The last two items of this scale had to be reversed. This way, all six items together make up a video liking scale in which low scores indicate low levels of video liking, whereas high scores indicate high levels of video liking. A PCA of the six items above shows that this scale is one-dimensional. One Eigenvalue above 1 (4.68, explaining 78.00% of the variance) was found.The individual factor loadings that have been found in all
performed PCAs can be found in appendix 1. The reliability Cronbach’s alpha score of this construct is .94.
Control variables: age, level of education and brand familiarity
Three (control) variables were expected to possibly influence our dependent
variables next to our independent variables. In terms of age and education level, the first of those control variables, it was assumed that that older people and people with higher levels of education, might be more sceptical towards advertising in general. This might influence
brand attitude scores. Furthermore in terms of brand familiarity, it was assumed that people who had heard of the Sonos brand before, would have a better brand recall than people who had never heard of the brand before.
Results
Preliminary analyses
To test if the aforementioned control variables of age, education level and
brand familiarity had to be included in the upcoming analyses, a correlation analysis
with our independent variables was conducted to determine if our possible control
variables were equally distributed across our three manipulated commercial
placement conditions and our two selected video liking groups. Based on this
analysis it was concluded that this was the case: brand familiarity was distributed
equally across both our independent variables r(174) = -.02, p = .848 (commercial
placement), r(174) = .01, p = .912 (video liking), as were education level r(174) = .11,
p = .165 (commercial placement), r(174) = -.07, p = .372 (video liking) and age r(174)
= -.09, p = .262 (commercial placement), r(174) = .10, p = .182 (video liking). None
of our selected control variables thus had to be included in our main analyses.
Hypotheses testing
In the paragraphs below the results of this study’s main analyses will be displayed. The order in which this is done, is slightly different than the order in which our hypotheses were have been introduced. Instead the results of our different hypotheses are discussed in the order in which effects are conceptually expected to take place.
H1 & H5a
To test the hypotheses that relate to switching behavior a two-way ANOVA has been conducted. In this analysis commercial placement position was included as an independent variable, video liking as an interaction (moderator) variable and switching behavior as a dependent variable. Results show an insignificant main effect of commercial placement position on switching behavior F(2, 170) = .47, p = .629. This means that the placement of a commercial (either before or during a video) does not significantly influence a person his or
her switching behavior, or in the words of our first hypothesis (H1): people who got to see a mid-roll video ad (condition 2: M = .11, SD = .31; condition 3: M = .15, SD = .36) were not necessarily more likely to stay on the same browser page when the ad was playing, than people who got to see a pre-roll video ad (M = .17, SD = .38). Based on these observations H1 thus has to be rejected.
In terms of the other hypothesis about switching behavior, the ANOVA conducted above also leads to conclusions of insignificance: no interaction effect between commercial placement position and video liking on switching behavior was found F(2, 170) = .49, p = .615. H5a, which stated that the above described behavior of not switching amongst mid-roll ad viewers would be even more probable amongst people who also liked the video in which the respective ad was placed, thus has to be rejected as well. Based on the observations of this study there is no ground to state that people who like a video (condition 2: M = .10, SD = .30; condition 3: M = .12, SD = .33) are less likely to switch away during a mid-roll video ad, than people who do not like this video (condition 2: M = .13, SD = .34; condition 3: M = .18, SD = .39).
H2 & H3
To test the direct effects of commercial placement position on the two core dependent variables of brand recall and brand attitude a one-way MANOVA, in which commercial placement position was entered as an independent variable and brand attitude, aided brand recall and unaided brand recall were added as dependent variables, was conducted. The results of this analysis show that there is no significant effect of commercial placement position on brand attitude F(2, 172) = .07, p = .931 and neither is there on aided brand recall F(2, 172) = 1.54, p = .218. A significant effect of commercial placement position on unaided brand recall is measured however F(2, 172) = 3.30, p = .039. Based on the results of a post hoc Tukey test, it can be concluded that there is a significant difference between our pre-roll group (37.1% brand recall) and our second mid-roll group (condition 3; 59.3% brand recall) p = .038. There was no significant difference between our two mid-roll groups (condition 2; 53.9% brand recall) p = .819. Neither was there between our first mid-roll group and our pre-mid-roll group p = .172. Furthermore an ANOVA testing the direct effects of commercial placement position on unaided brand recall in which all participants with high (≥ 4,) brand familiarity scores were excluded (remaining N = 110), also showed a significant effect F(2, 107) = 4.32, p = .016. A Tukey test showed again that there are significant
differences between our pre-roll group scores (N = 41, 19.5% brand recall) and our condition 3 group (N = 36, 50% brand recall) p = .012. No significant effects were found between other group combinations (p = .573 between groups 1 and 2; p = .179 between group 2 and 3).
People in group 2 had a brand recall score of 30.3% (N = 33).
In light of our two hypotheses on the direct relationships between commercial placement and brand attitude/recall, the following conclusions can be drawn based on the above statistics. First of all, hypothesis (H2) which stated that seeing a mid-roll video ad will yield higher levels of brand recall than will seeing a pre-roll video ad, can be confirmed partially. Although no significant aided brand recall differences were found between our three experimental groups (see table 1 for the corresponding aided brand recall scores), unaided recall was significantly better among people who had seen the condition 3 video, even more so amongst the people who indicated that they were not familiar with the Sonos brand. Based on these observations it can thus be argued that mid-roll video ads can improve a person his or her unaided brand recall, if these respective ads meet certain (intrusiveness) criteria.
Second of all, no significant differences between the brand attitude scores of any of our three experimental groups were found. The brand attitude scores of our pre-roll group (M = 3.95, SD = 1.37) were not necessarily more positive (as was expected) than the brand attitude scores of our two mid-roll groups (condition 2: M = 3.96, SD = 1.41; condition 3: M = 3.87, SD = 1.14). Based on the gathered data, H3 thus cannot be confirmed.
Table 1 Aided brand recall Condition 1, pre-roll Condition 2, in silence mid-roll Condition 3, in sentence mid-roll Total
Low video liking N = 34 M = .68 SD = .47 N = 24 M = .83 SD = .38 N = 33 M = .64 SD = .49 N = 91 M = .70 SD = .46 High video liking N = 29 M = .52 SD = .51 N = 30 M = .70 SD = .47 N = 26 M = .81 SD = .40 N = 85 M = .67 SD = .47 Total N = 63 M = .60 SD = .49 N = 54 M = .76 SD = .43 N = 59 M = .71 SD = .46 N = 176 M = .69 SD = .46
H2a & H5b
To test the hypotheses on the relationship between commercial placement and intrusiveness another ANOVA, comparable to the one that was performed to test the hypotheses on switching behavior, has been conducted. In this ANOVA commercial
placement condition again was entered as an independent variable and so was video liking (as an interaction variable). The dependent variable that was studied in this analysis was perceived intrusiveness.
The results of this ANOVA show a significant main effect of commercial placement on perceived levels of intrusiveness F(2, 170) = 9.95, p = .000. To chart the whereabouts of this significance a post hoc Tukey test was conducted. This test shows that there are significant differences between the pre-roll group (M = 4.74, SD = .18) and both mid-roll groups (condition 2: M = 5.65, SD = .19, p =.001; condition 2: M = 5.80, SD = .18, p =.000), but not between between the two mid-roll groups p = .870. These observations indicate that the use of a mid-roll video ad will lead to stronger perceptions of advertisement intrusiveness than the use of a pre-roll ad. Furthermore, to obtain this effect it makes no difference when exactly this mid-roll ad is shown. Based on the conclusions above H2a can be confirmed.
To evaluate H5b, the interaction effect between commercial placement condition and video liking as it was produced in this second/third ANOVA needs to be assessed. Doing so reveals that a significant interaction effect between commercial placement condition and video liking could not be found F(2, 170) = .15, p = .858 (see table 2 for sub-group
intrusiveness score variations). H5b, which expected that intrusiveness measures would be higher for people who liked the video of the comedian and who saw a mid-roll video ad (condition 2: M = 5.77, SD = 1.04; condition 3: M = 5.83, SD = 1.18), than for people who did not like the video of the comedian and who saw a mid-roll video ad (condition 2: M = 5.52, SD = 1.52; condition 3: M = 5.76, SD = 1.27), thus has to be rejected.
Table 2 Intrusiveness Condition 1, pre-roll Condition 2, in silence mid-roll Condition 3, in sentence mid-roll Total
Low video liking N = 34 M = 4.57 SD = 1.68 N = 24 M = 5.52 SD = 1.52 N = 33 M = 5.76 SD = 1.27 N = 91 M = 5.25 SD = 1.58 High video liking N = 29 M = 4.91 SD = 1.60 N = 30 M = 5.77 SD = 1.04 N = 26 M = 5.83 SD = 1.18 N = 85 M = 5.50 SD = 1.35 Total N = 63 M = 4.73 SD = 1.64 N = 54 M = 5.66 SD = 1.27 N = 59 M = 5.79 SD = 1.22 N = 176 M = 5.37 SD = 1.47 H4 & H4a
In order to test this study’s (final) hypotheses concerning the direct and mediational roles of intrusiveness and irritation in predicting brand recall and brand attitude, a four step analysis plan (Barron & Kenny, 1986) is followed. Due to the nature of all variables of interest - in terms of measurement - the first three steps of this analysis plan can be performed by executing a series of simple bivariate correlations between a number of combinations of these variables. In terms of the first step, which addresses the fact that a certain variable can only possibly fulfill a mediating role if there is a direct relationship too
(H4) between the initial independent and dependent variables (unaided brand recall, aided brand recall and brand attitude), the following is observed: there is no significant effect of perceived intrusiveness as IV on unaided brand recall r(174) = .12, p = .123, aided brand recall r(174) = .10, p = .211, or brand attitude r(173) = -.10, p = .190 as DVs.
Based on these figures H4, which stated expectations about the direct effects of perceived intrusiveness on the above mentioned dependent variables, has to be rejected. Higher perceptions of intrusiveness do not seem to lead to higher levels of brand recall. Neither do those perceptions lead to more negative brand attitudes.
Although the absence of a direct effect of our IV on our DVs implies that that H4a cannot be accepted - this because there is no main effect that can be mediated - H4a, which stated that perceptions of irritation would mediate the above mentioned direct effects of intrusiveness on brand recall and brand attitude, cannot be rejected yet. The above
performed correlation analysis after all shows that there are significant correlations between intrusiveness and irritation r(174) = .46, p = .000 (step 2 of the followed analysis plan) and between irritation and unaided brand recall r(174) = -.19, p = .010, aided brand recall r(174) = -.25, p = .001 and brand attitude r(174) = .54, p = .000 (step 3). Based on these results the following can be concluded: irritation does not function as a complete mediator between intrusiveness and our DVs. If this were to be the case, all four steps of our would analysis plan would have to show significant results. The significant effects of intrusiveness on irritation and of irritation on our DVs however suggest that irritation may be a serial or
sequential mediator - meaning that X can lead to M and M can lead to Y, without there being a direct effect of X on Y - between intrusiveness and our DVs. The significant effects that this study finds at this point however only partially answer to our expectations. Higher levels of irritation - as expected - lead to less positive brand attitudes. Instead of leading to a better brand recall however, higher irritation levels lead to lesser brand recall. Based hereon, H4a can be accepted partially.
Conclusion and discussion
The goal of this study was to gain new insights in the effects of different video commercialc placements on YouTube (is the ad shown before or halfway through the video?) on a list of (commercially) relevant behavioral, cognitive and effective variables as well as to contribute to the scientific literature in the still emerging field of online video marketing. Moreover, the extent to which people were to like the video during or in front of which an ad was broadcasted was studied as a possible moderator of some of those effects.
switching behavior are negligible. Contrary to our expectations, evidence could not be found that the viewing of a mid-roll ad would decrease the chance that people would click away to another browser tab during an ad (H1). Also contrary to our expectations, different
advertisement placement positions do not seem to alter someone’s brand attitude: it has not been found that viewing a mid-roll ad will lead to more negative brand attitudes (H3). In accordance with our expectations, viewing a mid-roll ad does lead to better (unaided) brand recall (H2). Furthermore, it was also found that mid-roll ads are perceived as more
interruptive than pre-roll ads (H2a). This is in line with our expectations as well. The here observed effect however was not stronger for people who liked the video they had seen (H5b). A moderating effect of video liking on the relationship between commercial placement position and switching behavior was not found. Mid-roll video ad viewers who liked the video they had just seen, were not (at all) less likely to switch to other browser tabs when the Sonos ad was playing than any other participant (H5a). Concerning the expected direct effects of ad intrusiveness on brand recall and brand attitude, it was found that such a direct effect was non-existent: stronger feelings of being interrupted were not associated with enhanced brand recall or less positive brand attitudes (H4). Feelings of irritation - which were asociated with feelings of being interrupted - on the other hand did have some effects on brand recall and brand attitude. High irritation with the ad was - as expected - an indicator of more negative brand attitudes. Also, in contrast with our expectations however, high irritation indicated low(er) levels of both aided as well as unaided brand recall (H4a). To conclude, this study thus found that perceptions of intrusiveness may be sequentially linked with brand attitude and brand recall through irritation. In case of brand recall however our findings indicate a reverse effect of irritation if it is compared to the effect that was expected.
To fully understand the the findings of this study, the conclusions above need elicitation and contextualization. In terms of our attempt to interpret switching behavior, our general conclusion is that neither the exact placement of a YouTube commercial, neither the liking levels of the video in or in front of which this commercial is placed, influence a person’s switching behavior. These findings are not in line with Van Meurs’ observations that centre break advertising is more likely to prevent people from “switching” than is beginning (or end) break advertising. Furthermore, Van Meurs’ speculation that this suppresive effect of centre break advertising on switching behavior may be caused by the possible fear of missing a part of the thing (the show they like; the video they like) they had started to watch, is not echoed by our results.
The rejection of our literary based hypotheses on switching behavior might partially be a result of a number of methodological (stimulus) imperfections or consequences, rather than theoretical ones. When our stimulus materials were developed, it was attempted to create YouTube videos with a “realistic” feel. As mentioned earlier, this was among other
things done by including an in screen timer (showing the length of the ad in seconds) while the Sonos commercial was playing. This timer in itself however might have suppressed switching behavior amongst all experimental groups, by making people aware of the fact that the ad would “only” take 30 seconds. This notion may have been a more important indicator of switching behavior than either commercial placement or video liking. At this point in time however, this assumption cannot be supported by existing literature: so far only one
interesting article - arguing that the use of such timers or promps can enhance brand recall (Yu, Chan, Zhao & Gao, 2012) - seems to have been performed. This article does not directly assess (tab) switching behavior however, hence the suggestion to analyze this phenomenon in future research.
In addition to the above, it should also be noted that the number of people switching to other tabs during the ad in the video in general was really low (only 14.80% (N = 26) of our participants switched to other tabs during the commercial), making it difficult to compare subtle inter group differences in switching behavior. Although our participants were told indirectly that they were allowed to click on other tabs during the video (“Watch the video as if it is a video YouTube video you have clicked on yourself”) the mere fact that our
participants were participating in an experiment, might have blurred our participants’ final switching scores. Focussed respondents after all might have refrained from their normal switching behavior solely because they thought it would be unappropriate or simply “against the experimental regulations” if they would do so. In terms of future research on tab
switching behavior, it is thus proposed that other methods for measuring switching behavior should be investigated.
Continuing, methodological recommendations for future research are also in place when the measurement of the (moderator) variable of video liking is discussed. In this study, the assumption that the use of a humorous fragment would make it easier to create two clearly distinguishable video liking conditions, because of the important role of subjectivity in humour, did not pay off. The performed median split at a video liking level op 5.50 suggests that video liking in general was relatively high. This has made it nearly impossible to draw solid conclusions on the true effects of video liking. After all: no valid, decent sized group of “dislikers” could be distinguished. Future research on the (moderating) effects of video liking on tab switching behavior (H5a) as well as on (the perceptions of) advertisement
intrusiveness, which has also been a victim of the suboptimal group selection that is
described (H5b), should thus always be executed with great(er) attention to the composition of the different video liking groups.
In terms of the discussion of the remaining direct relationships of this study, the direct relationships that are found between brand recall and commercial placement and
Not only does our study confirm that all mid-roll ads are perceived as more interruptive than pre-roll ads, as was assumed, some - condition 3 - mid-roll ads also enhance brand recall directly, as is found in previous literature (for example: Acquisti & Spiekerman, 2011). In addition to this, this study sought to find out if it matters when exactly a mid-roll ad is shown in a video. Only when it comes to influencing brand recall, it was found that it does: condition 3 ads, unlike condition 2 ads significantly led to enhanced brand recall compared to pre-roll ads. As previously discussed, this effect was even more profound when the people who indicated that they had heard about Sonos before were left out of the analysis. This has some interesting practical implications, which will be discussed later on.
Direct effects of commercial placement on brand attitude and of perceived intrusiveness on brand recall and brand attitude have not been found. At this point, especially the absence of the direct effect of intrusiveness on brand recall should be explored further. The foregoing after all concluded that mid-roll ads on YouTube, which are perceived as more interruptive, can enhance brand recall. The absence of a direct
relationship between the variable of intrusiveness and brand recall however suggests that there might be another thing to mid-roll ads than the feelings of intrusiveness that they produce, that explains why such ads improve brand recall. The question what this other possible recall predicting feature of mid-roll YouTube ads is, might be interesting for future research.
Compared to the absence of a direct effect of ad intrusiveness on brand recall, the absence of a direct effect of ad intrusiveness on brand attitude needs less attention. After all, a direct effect of commercial placement on brand attitude was not found either, thus not necessarily indicating contradictory findings. Our findings however do contradict our expectations, this in such a way that assessments of intrusiveness do not predict brand attitude at all. Why this is so could not be derived from the gathered data.
What could derived from the gathered data however, is that strong feelings of perceived intterupiveness can sequentially lead to more negative brand attitudes and lesser brand recall through irritation. The measured (in)direct interconnectedness between
intrusiveness and irritation, irritation and brand recall and brand attitude and commercial placement position and irritation through intrusiveness, complicates both scientific as well as practical implications. In terms of practical implications, the direct effects that are found between commercial placement location, brand recall and brand attitude pave the way for a possible increased used of mid-roll ads on YouTube by companies and marketeers. Our direct analyses after all have shown that the use of a very interruptive mid-roll ad can lead to better brand recall without having a direct negative impact on brand attitude. This recall enhancing effect furthermore is even stronger when the brand which is advertised is not well known yet: compared to a pre-roll ad, the brand in an interruptive mid-roll ad is much more
likely to be remembered by the one watching the commercial. This information can be very useful. For example when this company wants to launch and advertise for a new brand.
Indirectly however, it was also found that mid-roll ads can lead to greater
assessments of intrusiveness, which in turn may lead to feelings of irritation. When those feelings of irritation do arise, this may have negative consequences for a person’s levels of brand recall and brand attitude. In other words: although this study found that the use of very interrupting mid-roll video ads has optimal direct effects in a sense that it improves brand recall without negatively influencing brand attititude at the same time, the use of certain ads may also cause irritation which can lead to opposite (negative) effects. This (possible) opposite effect should not be forgotten when a company or a marketeer considers the use of mid-roll ads.
As a final theoretical implication, the above discussed, especially points out the interestingness of further studies regarding the affective and cognitive effects of irritation with the ad. When does irritation arise? And when does it not? And which underlying logic
explains why it was found that irritation can also have a negative effect on brand recall? Do negative emotions (i.e. irritation) disrupt cognitive processing (Levine & Burgess, 1997)? Future research on the mechanisms through which ad irritation works, hopefully will tell.
References
Aaker, J. L. (1995). Brand personality: conceptualization, measurement and underlying psychological mechanisms. UMI Dissertation Services.
Aaker, D. A. (2009). Managing brand equity. Simon and Schuster.
Anderson, J. R. (1995). Learning and memory: An integrated approach. New York, NY: John Wiley & Sons, Inc..
Acquisti, A., & Spiekermann, S. (2011). Do interruptions pay off? Effects of interruptive ads on consumers' willingness to pay. Journal of Interactive Marketing, 25(4), 226-240.
Baron, R. M., & Kenny, D. A. (1986). The moderator–mediator variable distinction in social psychological research: Conceptual, strategic, and statistical considerations. Journal of personality and social psychology, 51(6), 1173.
Boerman, Sophie C., Eva A. Reijmersdal, and Peter C. Neijens. "Effects of sponsorship disclosure timing on the processing of sponsored content: A study on the effectiveness of European disclosure regulations." Psychology & Marketing 31.3 (2014): 214-224.
Brown, S. P., & Stayman, D. M. (1992). Antecedents and consequences of attitude toward the ad: A meta-analysis. Journal of consumer research, 34-51.
Chatterjee, P. (2008). Are unclicked ads wasted? Enduring effects of banner and pop-up ad exposures on brand memory and attitudes. Journal of Electronic Commerce Research, 9(1), 51-61.
Chattopadhyay, A., & Basu, K. (1990). Humor in advertising: The moderating role of prior brand evaluation. Journal of Marketing Research, 466-476.
Chakrabarty, S., & Yelkur, R. (2006). The effects of ad irritation on brand attitudes. Journal of Promotion Management, 11(2-3), 37-48.
Chi, H. K., Yeh, H. R., & Yang, Y. T. (2009). The impact of brand awareness on consumer purchase intention: The mediating effect of perceived quality and brand loyalty. Journal of International Management Studies, 4(1), 135-144.
Edwards, S. M., Li, H., & Lee, J. H. (2002). Forced exposure and psychological reactance: Antecedents and consequences of the perceived intrusiveness of pop-up ads. Journal of Advertising, 31(3), 83-95.
Faber, R. J., & Stafford, M. R. (2005). The Future of Consumer Decision Making in the Age of a New Media Promotions and Advertising. Advertising, Promotion and New Media, 351-365.
Ha, L. (1996). Observations: advertising clutter in consumer magazines: dimensions and effects. Journal of Advertising Research, 36, 76-84.
Gelb, B. D., & Zinkhan, G. M. (1986). Humor and advertising effectiveness after repeated exposures to a radio commercial. Journal of Advertising, 15(2), 15-34.
and mode on audience recall. Journal of Current Issues & Research in Advertising, 20(1), 47-59.
Hernandez, M. D., Chapa, S., Minor, M. S., Maldonado, C., & Barranzuela, F. (2004). Hispanic attitudes toward advergames: A proposed model of their antecedents. Journal of Interactive Advertising, 5(1), 74-83.
ITV Network Centre. (1993). I can't believe it's clutter: A major study into the effects non-programme editorial items on the audience to commercial breaks.
Horsley, C. (1986). A quantified study on what people do during commercial breaks. In Seminar on new developments in media research, ESOMAR: Amsterdam, New-York (pp. 23-34).
Kim, S. (2015). Effects of Ad-Video Similarity, Ad Location, and User Control Option on Ad Avoidance and Advertiser-Intended Outcomes of Online Video Ads (Doctoral dissertation, UNIVERSITY OF MINNESOTA).
Levine, L. J., & Burgess, S. L. (1997). Beyond general arousal: Effects of specific emotions on memory. Social Cognition, 15(3), 157-181.
Li, H., Daugherty, T., & Biocca, F. (2002). Impact of 3-D advertising on product knowledge, brand attitude, and purchase intention: The mediating role of presence. Journal of Advertising, 31(3), 43-57.
Li, H., Edwards, S. M., & Lee, J. H. (2002). Measuring the intrusiveness of advertisements: Scale development and validation. Journal of advertising, 31(2), 37-47.
Martí-Parreño, J., Aldás-Manzano, J., & Currás-Pérez, R. I. Sánchez-Garcàa (2012), Factors contributing brand attitude in advergames: Entertainment and irritation. Journal of Brand Management, 18, 1-15.
Meurs, L. V. (1999). Switching during commercial breaks.
Mitchell, A. & Olson J.C. (1981) Are Product Beliefs the Only Mediator of Advertising Effect on Brand Attitude? Journal of Marketing Research, 18, 318-332.
Murry Jr, J. P., Lastovicka, J. L., & Singh, S. N. (1992). Feeling and liking responses to television programs: An examination of two explanations for media-context effects. Journal of Consumer Research, 441-451.
Olson, J. C., & Mitchell, A. A. (2000). Are product attribute beliefs the only mediator of advertising effects on brand attitude?. Advertising & Society Review, 1(1).
Ritter, E. A., & Cho, C. H. (2009). Effects of ad placement and type on consumer responses to podcast ads. CyberPsychology & Behavior, 12(5), 533-537.
Russell, C. A. (2002). Investigating the effectiveness of product placements in television shows: The role of modality and plot connection congruence on brand memory and attitude. Journal of consumer research, 29(3), 306-318.
Severn, J., Belch, G. E., & Belch, M. A. (1990). The effects of sexual and non-sexual advertising appeals and information level on cognitive processing and
communication effectiveness. Journal of advertising, 19(1), 14-22.
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.
Van den Berg, H. & Ruster, R. (1992). Zapp …… en Weg waren Ze. Een Onderzoek naar het Vermijden van Televisiereclame [Zapp …… and Then There Were Gone. A Study of the Avoidance of Television Ads], Unpublished Master’s Thesis, Catholic University of Nijmegen, the Netherlands.
Van Reijmersdal, E. (2009). Brand placement prominence: good for memory! Bad for attitudes?. Journal of Advertising Research, 49(2), 151.
Woltman Elpers, J. L., Wedel, M., & Pieters, R. G. (2003). Why do consumers stop viewing television commercials? Two experiments on the influence of moment-to-moment entertainment and information value. Journal of Marketing Research, 40(4), 437-453.
Yu, R., Chan, A. H., Zhao, P., & Gao, Y. (2012). Enhancing memorability: Do remaining duration prompts affect advertising recall?. International Journal of Advertising, 31(4), 861-876.
