“Advertising Effectiveness of Pre-Roll Advertisements on Desktops and Smartphones”

(1)

“Advertising Effectiveness of Pre-Roll

Advertisements on Desktops and

Smartphones”

(2)

“Advertising Effectiveness of Pre-Roll Advertisements on

Desktops and Smartphones”

M. T. (Marco) Lubbers University of Groningen Faculty of Economics and Business

Master Marketing Management

June 23, 2014

Eerste Jacob van Campenstraat 47-iii 1072 BD Amsterdam

Email: marco.lubbers@gmail.com Tel: +31 (0) 6 29 19 23 73

Student number: 1715364

(3)

Abstract

Pre-roll video advertising has become very popular in the past years and spreading from desktops to mobile devices. Due to its newness, research on the effectiveness of pre-roll advertisements is scarce, especially in comparing desktops and mobile devices. This study contributes to the body of research by examining the effectiveness of the pre-roll ads on laptops and smartphones, and moderating the screen size relationship with intrusiveness. Advertising effectiveness is measured by recall, as this tool is considered very useful for video ads. The experiment uses data from a large randomized experiment with 179 respondents. Unfortunately no significant main effects have been found in this study. However, recall was very high both on smartphones and desktops, indicating the pre-roll is very effective. Lastly, contrary to existing research, it was found that the skip ad-option did not make a difference in the perceived degree of intrusiveness.

(4)

Preface

The future of video is online, not only interesting content, films, documentaries or series but also the future of advertising. The industry has been growing at an incredible rate and quality standards for online series often surpass those of traditional media.

As a filmmaker and marketing student these new grounds are interesting as they set the pace of online advertising for the coming years. I believe it is interesting to see what the driving forces of video services like YouTube are and how marketing departments of companies can benefit from these insights.

This report examines the effectiveness of the pre-roll video using recall in order to fill a research gap for online advertising effectiveness. It provides directions for future researchers and insights for companies so that they can allocate their advertising budgets more efficiently.

(5)

Introduction

With the rise of online videos a new advertising possibility has emerged that started to become big in 2008 (YouTube 2014), namely the online pre-roll video advertisement. Nowadays most people who watch videos online are exposed to pre-roll ads, as most sites that host videos use these ads to generate revenue (ComScore 2014). In essence the pre-roll advertisement is a video or type of banner or display ad with moving images (Manchanda, Dube, and Goh 2006). This ad is played prior to the content the viewer intended to watch. Online pre-roll video advertising has been growing significantly since its inception, and accounted for a total of $4,1 billion dollars in 2013 and is expected to grow to $8 Billion dollars by 2016 (Emarketer 2013) in the US alone. Pre-roll ads had a market share of 23,4% of total digital display ad spending in the US in 2013 (Emarketer 2013). This number is forecasted to grow to nearly 32% by 2016. These online pre-roll ads are played on all devices that are able to play online videos like personal computers, tablets and mobile devices (YouTube 2014).

(8)

It is argued that the communication dynamics and relationships from brand to recipient have changed due to social media and video sharing websites like YouTube (Paek, Hove, Jeong, and Kim 2011). Several researchers have found differences between advertising using traditional media, the television, and online media, in which the online media experience is more interactive and participatory (Calder, Malthouse, and Schaedel 2009; Nambisan and Baron 2007; Dwyer 2007; Varan, Murphy, Hofacker, Robinson, Potter, and Bellman 2013). Furthermore online media tends to foster lean forward viewing behavior while traditional media have viewing behavior which is more leaning backward (Calder et al. 2009; Barwise 2001).

(9)

advertising campaign, one should know to what extent screen size influences advertising effectiveness.

When advertising campaigns are used for branding purposes, they often consist of a long-term strategy that is for instance focused on improving brand awareness, customer relationships, and attracting new customers (Hollis 2005). Rich media formats tend to have stronger branding effects compared to simple animated banner ads, of which video has the strongest positive effect on branding (Spalding, Cole, and Fayer 2009). According to Bhat, Bevans, and Sengupta (2002), the total amount of impressions is a good indicator of exposure of an ad. The exposure to ads can lead to two types of brand attention, namely, directed and non-directed attention (Yoo 2008). In non-directed attention people can engage in subconscious processing that potentially has an effect on creating memories and attitudes towards the advertised brand without clicking on the ad. Furthermore, it is argued pre-attentive processing can occur when viewers are exposed to banner ads, which indicates that both recall and brand awareness, are good indicators for the assessment of advertising effectiveness (Dreze and Hussherr 2003). Yoon and Lee (2007) found that the exposure effects of ads that received no clicks had the same effect on brand awareness, attitudes, and implicit memory as the ads that had been clicked on. This shows that even though click through rates are important in today’s advertising landscape, other factors still need to be included so that advertising campaigns are not undervalued (Yoon, Lee, 2007). Solely focussing on the CTR and conversion rates might leave the advertiser with a wrongly valued campaign as not everything is measured and accounted for.

(10)

ads or animated banner ads (Li, Edwards and Lee 2002; McCoy et al. 2008; Goldfarb and Tucker 2011). A recent report on mobile video advertisements, including smartphones and tablets, demonstrates that click-through rates for skippable pre-roll videos are higher 2.3% compared to the non-skippable version 1.5% (MMA 2014). On the other hand, completion rates of skippable ads 8% are much lower, compared to non-skippable ads 93% (MMA 2014), indicating that many viewers skip the ad. Furthermore, half of the viewers who skip an ad watch 25% of the advertisement, and 30% of the viewers watch half of the ad (MMA 2014). When viewers skip an ad, they take action by clicking a button in the advertisement. Thus a behavioral response has been elicited and cognitive processing of the ad may have occurred. This however is not reflected in a click-through rate.

Current online advertising metrics like the click-through rate and conversion rate are widely used and provide insight into the amount of people who visit the advertised website or purchase products. However, these metrics do not provide insight into the memory effects of viewers, as this cannot be measured with a simple click-through rate. In one study it is even argued that effective features of ads are based on guesses instead of empirical research (Sigel, Braun, and Sena 2008). Furthermore, click-through rates have declined steadily over the past years, and (Yoo 2008). Moreover, Yoo (2008) argues online ad measurement should focus on reach rather than direct responses, as direct responses do not reflect unconscious processing and thus does not accurately reflect the overall value of the ad. Spalding et al. (2009) state that format choices for ads have little to do with branding goals, and that advertisers are more focused on click-through rates, media budgets, and deadlines for creative units. This indicates the possibility of a misalignment between branding goals and advertising format choice. Nottorf (2014) states that evaluating online marketing activities through standalone metrics does provide insight in the changing consumer behavior and is thus not completely accurate. Moreover, it might result in inefficient allocation of budgets for advertising campaigns, as only behavioural response (CTR) and viewing duration are analysed, but not on memory effects.

(11)

manipulate the relationship between screen size and ad effectiveness. It is argued that audience surveys are the best way to measure advertising effectiveness, but are less frequently implemented due to the complexity and high cost (Pashkevich et al. 2012). This study will attempt to fill part of this research gap by conveying an experiment with consumers who watch online videos, by measuring the effectiveness of the pre-roll ads. This is valuable data for marketers so that they can make informed decisions on how to allocate their advertising budgets in the most efficient way.

The main research question for this paper is:

How effective are online pre-roll ads on desktop and mobile devices in terms of recall?

The remainder of this study consists of four parts. First a theoretical framework is created that consists of current literature on the effectiveness of advertisements, differences due to screen sizes, and intrusiveness of ads. In the second part experiment is constructed and conducted. The third part consists of the analysis of results. The fourth and last part consists of a discussion of the findings, combined with, limitations, and directions for future research and managerial implications.

Theoretical framework

(12)

a higher click-through rate compared to static ads, as they are more likely to capture viewers’ attention (Cho, Lee, and Tharp 2001; Hong et al. 2004). Moreover, auditory cues in advertisements can positively affect ad recall (Gardner, 1985, Gorn, 1982).

Advertising effectiveness

Many researchers in the field of online advertising effectiveness had their focus click through rates (Luo et al. 2011), leaving out the other measures of like brand attention, for instance in the form of recall. Several researchers do propose evaluating advertising effectiveness using the cognitive component is advisable (Vakratsas and Ambler 1999; Gutpa and Gould 2007; Jeong, Kim, and Zhao 2011). According to Atkinson and Shiffrin’s (1968) model of memory, three types of memory exist. The three types are long-term memory, short-term or working memory, and sensory memory. In this study the most important form of memory is working memory, as this is where the information is stored during the experiment. Advertising effectiveness increases when people encode ads and remember them, this helps the consumer find or recognize the right product he or she wants to buy (Howard and Sheth 1969). The limited capacity model of attention allows examination of the processing of messages (Lang 2006). Two principal factors provide a basis for the model. The first is that people frequently process information, and information processing is an important task, consisting of encoding, storing and retrieving the message. Secondly, people are limited in their cognitive resources for information processing. The model shows that not all information can be processed while cognitive resources are used for competing tasks of information processing. This can occur on a subconscious level via the automatic processing mechanisms, or actively when people use their cognitive resources to achieve individual goals. All stimuli have an effect on the processing of the advertisement, although priority is given to controlled tasks, which are the goals the person actively wants to pursue. When people do not click on an ad when exposed to it, cognitive processing may still occur, either conscious or unconscious, and the advertised brand can enter the consideration set (Yoo 2008). The consideration set is the total subset of brands brought to a consumer’s mind during a choice occasion (Shocker, Ben-Akiva, Boccaro, and Nedungadi 1991).

(13)

dominated by right hemisphere processing. So while watching videos, the part of the brain where recognition stimuli are processed is activated (Krugman 1971). This is in line with advertising metrics in the television industry, where recall has been an often-used tool to measure ad effectiveness (Brown and Rothschild 1993; Rothschild and Churchill 1988). Accordingly, recall and recognition are appropriate measures to assess pre-roll advertising effectiveness.

Screen size

(14)

recall (Hendon 1973). Furthermore, older research showed that the relationship between recall and ad size is exponential so that when recall is to be doubled, ad sizes needs to be quadrupled (Adams 1926; Franken 1925).

The empirical findings of print advertising research are similar to the effects of differences in screen size in that larger ads yield higher recall (Hou, Nam, Peng and Lee 2012; Detenber and Reeves 1996; Li and Bukovac 1999). Viewers exposed to a larger screen had better scores on memory and experienced higher emotional arousal (Lombard, Reich, Grabe, Bracken and Ditton 2000; Detenber and Reeves 1996). This corresponds to a study on video stimuli, where a positive relationship was found between memory and arousing images (Lang, Bolls, Potter and Kawahara 1999). Lombard and Ditton (1997) show that larger screen size has a positive influence on the audience’s viewing experience on levels of involvement and arousal. In a comparison between commercials shown on large and small TVs, large screen viewers pay more attention to commercials than standard screen television viewers and they also had more positive attitudes towards the ads (Mcnive, Krugman, Tinkham, 2012). This matches the results of Vakratsas and Amber (1999) that found emotionally involved viewers have an increased chance to process the message of the ad. Furthermore, research on games show that larger screens increase the positive mood compared to smaller screens (Hou, et al. 2012). In a virtual learning environment experiment with children, those who used bigger screens had better recall of facts than those who were using smaller screens (Fassbender, Richards, Bilgin, Thompson and Heiden 2012). When it comes to online advertising, large banner ads have higher click-through rates and are superior in attract the viewer’s attention compared to smaller ads (Li and Bukovac 1999). According to Tucker and Goldfarb (2010), high visibility leads to higher recall on online advertising, but only when they the ad is not targeted.

Hypothesis 1

(15)

Bellman et al. 2009), have a positive effect on involvement and arousal (Lombard et al. 2000; Lang et al. 1999), and this involvement and arousal leads to higher recall as well (Vakratsas and Ambler 1999). Since laptops and desktops have larger screens than smartphones, the pre-roll video advertisement will be larger on a laptop than on a smartphone. From this it can be assumed that screen size positively influences recall of pre-roll advertisements, leading to the fist hypothesis of this study:

H1: Pre-roll ads shown on large screens have a higher advertising effectiveness in

terms of recall, than pre-roll ads shown on small screens.

Intrusiveness

As a pre-roll advertisement is a video that plays before the intended video starts, it blocks the viewer from watching the content immediately. In this case their media goal is interrupted when advertisements are shown, as the viewer has to wait for the ad to be completed, or when the viewer can skip the ad and go straight to the content. According to Goldfarb and Tucker (2011), an advertisement is intrusive when it has one or more of the following characteristics that relate to this study. First, consisting of in-stream video and audio, in which the ad is implemented in a video stream either at the beginning, middle or end of the video. Secondly, when a takeover occurs where, the ad uses part of the screen normally used for the content of the website. Thirdly, the ad is an interstitial or pop-up in which the ad is played before the intended content loads. Fourthly, when it is interactive and requires a two-way interaction with the viewer. Fifthly when a floating ad is used where the ad is on top of a website and becomes unobtrusive or invisible after a specific time period. In essence, every pre-roll ad is intrusive as the ad takes over the page or part of the page, blocks the viewer to watch the content he or she wanted to watch, it is played before the video loads, and is interactive in some cases as the user can sometimes skip the ad by clicking on it.

(16)

the ads when their threshold is met (Krugman and Johnson 1991). Due to the auditory and visual stimuli of pre-roll ads, they are very salient, making the ads difficult to ignore and seen as intrusive (Goldfarb and Tucker 2011b). This can result in further negative consequences, as people tend to dislike ads that are highly visible (Chan, Dodd and Stevens 2004). Research on traditional media like television and radio, has shown that intrusive ads are sometimes avoided by zapping (Abernethy 1991) or engaging in other activities (Krugman and Johnson 1991). When viewers stop watching an ad, recall decreases as viewers do see the complete ad, as they have devoted their cognitive resources to other activities. Many researchers have found that cognitive avoidance, which is characterized by like looking away from the screen or engaging in other activities (Speck and Elliot 1997) reduces ad recall and ad recognition (Thorson and Zhao 1997; Ehrenberg and Twyman 1967). Furthermore, when faced with increased ad loading or advertising clutter people purposely try to avoid these ads (Dreze and Hussherr 2003). Advertising intrusiveness can evoke a psychological reaction to advertisements that disturbs the cognitive process of a consumer (Li, Edwards, and Lee 2002), leading to lower recall of the ads (Li et al. 2002). Furthermore, irritation of the advertisement is found to negatively affect advertising (Okazaki 2004; Kargoankar and Wolin 1999).

Ad Skipping

(17)

type of pre-roll advertisement is now skippable after the ad has been playing for five seconds, and is perceived to be less intrusive (Pashkevich al. 2012). This is in line with research on banner ads that shows people might feel irritation when their browsing activity is interrupted, and irritation increases when there is no possibility to skip or close the ad (McCoy et al. 2008).

Duration prompt

Another factor that can influence the intrusiveness of advertisement is the duration prompt (Yu, Chan, Zhao and Gao 2012). The duration prompt indicates how long it will take for the commercial to be completed. When viewers are shown how long the pre-roll will take, they can make an accurate and informed decision on whether they want to keep waiting or stop watching and going elsewhere. Which in the case of online advertising could lead to leaving the page on which the ad is shown. Viewers can experience anxiety when not knowing how long an advertisement will take (Rachman 1998). According to researchers in the field of psychology, anxiety can decrease working memory capacity (Eysenck and Calvo 1992), increase unpleasant feelings and reduce cognitive processing capacity and memory (Leight and Ellis 1981). However, when viewers are exposed to the duration prompt in the form of a countdown, they experience lower levels of anxiety, leading to higher recall of the ads (Yu et al. 2012).

Hypothesis 2

(18)

relationship of screen size on advertising effectiveness, leading to the second hypothesis of this paper:

H2: Intrusiveness pre-roll advertisements negatively affect the relationship of screen size on advertising effectiveness.

Conceptual model

The constructs and relationships discussed in the theoretical framework are depicted in the conceptual model of figure 1. Where screen size positively influences the advertising effectiveness, and where intrusiveness has a negative effect on this relationship.

Figure 1: Conceptual model.

Research design

In this section of the paper, the design of the experiment will be discussed as well as the means of data collection, description of dependent, independent, and moderating variable, and statistical techniques to analyse the data. This study measures the effectiveness of pre-roll advertisements on desktop and mobile devices and the moderating influence of intrusiveness. Respondents will be exposed to a popular video that has a pre-roll advertisement, after which they will fill out the questions of the experiment.

Data collection technique

(19)

computerized experiment administration. It will be spread out through the two social media platforms Facebook and Twitter, as well as with the use of Enalyzer, a company specialized in online experiments. This combination of allows for a quick response rate, and a heterogeneous sample. Furthermore, the snowball sampling technique is applied to further increase the amount of respondents, where respondents are asked to send the questionnaire to their friends and people from their network. This type of referral sampling increases the potential sample size significantly and furthermore increases the heterogeneity of the sample. A more heterogeneous sample tends to be a better reflection of society (Goodman, 1961). Respondents fill out three demographic questions, namely the age, gender and the level of education. In this way, groups can be compared in order to make sure variances among the four groups are equal.

Groups

As differences across devices are measured, the experiment will start with a question on whether they use their smartphones or desktop to complete the experiment. Respondents are asked to do the experiment using a desktop or laptop computer, or a mobile phone, as tablet users are not part of the research. Since the experiment needs to distinguishes on screen size and intrusiveness, four groups are necessary, with a minimal sample size of n=30 in order to increase the chances of finding a significant effect (Mandeville 1969; Cochran 1947). The four groups complete the experiment on a desktop or laptop with an intrusive ad, on a desktop or laptop with a non-intrusive ad, on a smartphone with an intrusive ad, and on a smartphone with a non-intrusive ad.

Experiment design

(20)

view duration of 89 seconds (Catchyourmoment 2014). In order to increase the chances of the respondent completing the experiment, after the pre-roll advertisement, the video will be around 60 seconds in length. Once the video is shown, four questions will be asked about the content of the video. The video will feature a pre-roll video ad that is edited into the real video. A directed attention condition is used to mask the actual research, this setup recreates a realistic situation for a viewer of videos who searches the web for interesting content without paying extra attention to advertisements (Fennis and Stroebe 2010). All participants were tested immediately on advertising recall after viewing the viral video and filling out three questions on the video. The entire experiment is attached in the appendix.

Variables

The variables depicted in Figure 1 will be tested and each variable will be discussed separately.

Independent variables: Screen size

This study attempts to find differences in advertising effectiveness of pre-roll advertisements on mobile devices and desktops. The independent variable consists of screen size, either large or small. The distinction between large and small is created by laptops and desktops as large screens, and smartphones for the small screens. As the ads are identical both on desktops and laptops, and smartphones, the results of the experiment can be attributed to the varying screen sizes and degree of intrusiveness. The data is coded into a new variable, so that the desktop receives a value of 0, and the smartphone a value of 1.

Dependent variable: Advertising effectiveness

(21)

higher recall than when only text-based questions are shown (Brown and Rothschild 1993; Childers et al. 1986). The dual coding theory (Pavio 1986) suggests that people use two cognitive systems to process information, either trough a system that processes the information verbally, and the other for processing pictorial or graphic cues. When both systems are triggered, using both text and images, cognitive processing increases even further and memory is enhanced (Mayer 2001; Goodrich 2011). The four questions will be recoded into dichotomous variables, with the right answer receiving a score of one, and the wrong answer a score of zero. The four scores on the individual questions are aggregated to create sum of all scores and creating the recall score. Since the dependent variable is a count variable, a regression analysis will be used to find differences in advertising effectiveness on different screen sizes. When the assumptions for linear regression are not met, the one-way ANOVA test is performed.

Moderating variable: Intrusiveness

(22)

Attitude towards the ads

In order to measure the attitude towards the ad, and check whether manipulation of intrusiveness is successful, in the end of the experiments are asked to fill out six questions in a Likert scale format. Due to the central tendency bias of the Likert scale, where some respondents may not opt for the extreme options, it is useful to take the Likert scale design that yields the highest mean scores. Therefore, a 5-point Likert scale is used in the questionnaire despite the fact that many questionnaires are using seven or nine levels. However, an empirical study (Dawes 2008) found that a 5- or 7- point scale produces slightly higher mean scores relative to the highest possible attainable score, compared to those produced from a 9-point scale, where the difference is statistically different. The format of the five-level Likert scale ranges from strongly disagree (1), somewhat disagree (2), neutral (3), somewhat agree (4), strongly agree (5).

Control variable

Participants of the study are asked whether they had seen an ad in the video.

Familiarity with the advertisement enhances the degree of recall as the viewer has already seen and processed the ad before the experiment Recognizing is oftentimes more easy and requires fewer cognitive resources than processing new impulses. Therefore, the ad that is chosen is more than two months old, and has been aired only for a brief period of time whilst not becoming very popular by receiving awards or extra attention.

Results

In this chapter the results of the experiment will be discussed. The goal of this experiment is to find differences in recall among the devices and what the effect of intrusiveness is on this relationship. I will start with a description of the characteristics of the sample. Secondly, a check for manipulation is conducted. Thirdly, the results of the main effects between the independent, dependent variables and the moderator will be discussed. Fourthly, additional findings will be explained. Finally the major findings are discussed.

Sample characteristics.

(23)

needed. All respondents completed the experiment and there were no missing values for the participants of the experiment. In total, 13 respondents did not use a smartphone or desktop when completing the experiment. These respondents are not included in the dataset, as they have no added value in this experiment. To check whether the deleted respondents were different from the rest of the sample, their age was tested using an independent samples t-test. The results of the independent samples t-test, t(179) = -1.63, p = 0.106, show that the removed respondents (M= 3.08 SD = 1.24) are not significantly different from the others in the sample (M = 2.68, SD =.96, p=.06) when it comes to age. The groups that respondents could choose from were six different age groups, with group two being between 18-25, and group three being between 25-35 years old respectively. Explaining the means being 3.08 and 2.68 rather than specific ages. Although the deleted group is rather small (n=13), a t-test can still be used as it keeps its predictive power in sample sizes that are very small or unequal (de Winter, 2013). Since the respondents that completed the questionnaire with another device did not differ from the sample, it can be concluded they can be deleted safely. Having removed the 13 respondents from the data resulted in a dataset of 179 respondents, distributed over the four groups. Each group has been exposed to the same pre-roll advertisement and video, with the only difference being the degree of intrusiveness and the device the participant used to watch the video. The participants were randomly assigned to either the intrusive pre-roll or to the non-intrusive pre-roll. The smartphone non-non-intrusive group consisted of n=47, Smartphone intrusive n=34, laptop non-intrusive n=47, laptop intrusive n=51, filling all groups with sufficient participants for enough statistical power to make data analysis possible. All respondents had completed the experiment, as they could not continue without providing an answer, therefore there is no missing data. Multiple studies have shown analysis of variance can be performed when the sample size per group is n>30 (Mandeville, 1969, Cochran, 1947). Therefore, the sample size meets the criteria to perform a robust analysis.

Device Type of Ad Number of participants

Smartphone Intrusive n= 34

(24)

Laptop Intrusive n= 51

Laptop Non-intrusive n= 47

Table 1. Sample characteristics

Frequencies

Of the respondents 55% (SD=.50) of them completed the experiment on a smartphone, while the other 45% (SD=.50) completed the experiment on a laptop or desktop. Of all respondents, 47% (SD=.50) were shown an non-intrusive advertisement, in that they were shown the duration prompt along with the skip-ad button. Recall of the ad is M=3.86 (SD=.49), showing that on average respondents scored 3.86 on a scale of 0-4 on recall of the content of the advertisement.

Device Intrusiveness Recall

N 179 179 179 Mean .55 .47 3.86 Median 1.0 .00 4.00 Standard deviation .50 .50 .49 Skewness -.19 .10 -4.81 Kurtosis -1.99 -2.01 28.14 Table 2, Frequencies.

Representativeness of the sample

(25)

Age Gender Education level

Category Frequency Category Frequency Category Frequency <18

years

3 (1.7%) Female 63 (35.2 %) Junior general secondary (VMBO)

6 (3.4%)

18-25 years

72 (40.2%) Male 116 (64.8%) Senior general secondary (HAVO) 7 (3.9%) 25-35 years 93 (52%) Pre-university (VWO) 6 (3.4) 35-45 years 4 (2.2%) Senior secondary vocational (MBO) 14 (7.8%) 45-55 years 4 (2.2%) Higher professional (HBO) 63 (35.2%) >55 years 3 (1.7%) University (WO) 83 (46.4%)

Table 3, sample demographics.

Comparing groups on demographics.

In order to make sure the four different groups have similar variances on the three demographic constructs, a series of independent samples t-tests is executed. All results are depicted in table four, five, and six. As all groups have equal variances except for the group education, where smartphone users are significantly different from than respondents using the laptop as the former has enjoyed higher education. However, as this is the only group that is different, and both groups score in the upper segment of education, it is assumed this variance is not problematic. Furthermore,

Gender

(26)

independent samples t-test, t(96) =-1.21, p=.231 shows that respondents seeing the non-intrusive ad on their smartphones (M=1.28 SD=.45) are not significantly different from respondents seeing the intrusive ad on their smartphones (M=1.39 SD=.49) when it comes to gender.

AGE

The results of the independent samples t-test, t(177) =.71, p =.476 show that the respondents using laptops (M=2.73, SD=.81) are not significantly different from the participants using smartphones (M=2.64, SD=.79) when it comes to age. Furthermore, t(79) = .49, p=.625 shows that respondents seeing the non-intrusive ad on their laptops (M=2.77, SD=.81) are not significantly different from respondents seeing the intrusive ad on their laptops (M=2.68, SD=.81) when it comes to age. Moreover, t(96) = -1.08, p=.283 shows that respondents seeing the non-intrusive ad on their smartphones (M=2.55, SD=.75) are not significantly different from respondents seeing the intrusive ad on their smartphones (M=2.73, SD=.83) when it comes to their age.

Education

The results of the independent samples t-test, t(177) =-1.99, p =.049 show that the respondents using laptops (M=4.86, SD=.1.46) are significantly different from the participants using smartphones (M=5.23, SD=.1.03) when it comes to education. Furthermore, t(79) = .83, p=.409 shows that respondents seeing the non-intrusive ad on their laptops (M=4.98, SD=1.31) are not significantly differently educated from respondents seeing the intrusive ad on their laptops (M=4.71, SD=1.64). Moreover, t(96) = -.20, p=.841 shows that respondents seeing the non-intrusive ad on their smartphones (M=5.21, SD=1.14) are not significantly differently educated from respondents seeing the intrusive ad on their smartphones (M=5.25, SD=.94).

Gender

(1=male, 2 = female)

Age Education

Smartphone Laptop Smartphone Laptop Smartphone Laptop

Sample size N=81 N=98 N=81 N=98 N=81 N=98

(27)

Standard deviation

.48 .49 .79 .81 1.03 1.46

p-level .641 .476 .049

Table 4. Comparisons of groups – Device size Gender (1=male, 2 = female) Age Education Intrusive - smartphone Non-intrusive smartphone Intrusive - smartphone Non-intrusive – smartphone Intrusive smartphone Non-smartphone Sample size n= 34 n= 47 n= 34 n= 47 n= 34 n= 47 Mean 1.39 1.28 2.73 2.55 5.25 5.21 Standard deviation .49 .45 .83 .75 .94 1.14 p-level .231 .283 .841

Table 5. Comparison of groups – Intrusiveness on smartphones Gender (1=male, 2 = female) Age Education Intrusive - Laptop Non-intrusive Laptop Intrusive - Laptop Non-intrusive – Laptop Intrusive Laptop Non-intrusive Laptop Sample size n= 51 n= 47 n= 51 n= 47 n= 51 n= 47 Mean 1.38 1.35 2.68 2.77 4.71 4.98 Standard deviation .49 .49 .75 .81 1.64 1.31 p-level .786 .625 .409

(28)

Scale reliability analysis

The reliability of the recall scale is tested by with the use of Cronbach’s Alpha. The composited scale of recall yields an alpha of .60. Although an alpha <. 70 is usually regarded as low reliability, multiple researchers argue a coefficient of .60 is acceptable (Dowing 2004; Nunally 1967). As Cronbach’s Alpha increases when the amount of items increases (Cortina 1993), the recall scale can therefore become relatively low as it only consisted of four times. Furthermore, the scale “if item deleted” does not increase the internal reliability by a big improvement, as removal of this item would lead to a Cronbach’s Alpha of .71, which does not improve from an acceptable score to a good score. Since the scale consists of four questions, and the pictorial question is important for measuring recall, I have decided to keep the variable in the analysis.

Linear regression

A linear regression was conducted with recall as a dependent variable and device size as the independent variable. The results show a positive relationship as F (1.177)= .67, p= .416, but the relationship device size and recall is not significant. When the moderator intrusiveness is included in the regression analysis, the effect of the screen size on recall is not significant as F (3.175) =1.48, =. 222, so it can be concluded the moderator does not influence the relationship between screen size and recall in a significant way.

Violations of assumptions

(29)

assumptions as a value <3.0 is desired. This indicates the probability lies in more in its centre and compared to a normal distribution the central peak is higher (Balanda, MacGillivray, 1988). The data does not meet requirements after the log-transformation has been performed, therefore no linear regression should have been performed.

One-way ANOVA

A 2 (device: smartphone vs. desktop,) x 2 (intrusive vs. non-intrusive) full factorial ANOVA on the sum of correctly answering the different items of recognition (product category, brand, content, picture) was conducted. Equal variances are assumed in the population, this is shown in the Levene’s F-test, F (1, 177) = 3.01, p= .084, which is not significant, so the ANOVA can be used. The null hypothesis that is being tested is that the mean score of recall is equal for respondents using a laptop or a smartphone. In this case the null hypothesis is accepted as the significance is larger than p = .05. The one way ANOVA shows no significant difference has been discovered between the respondents who used a smartphone and those who used a laptop F (1, 177)= .67,

p= .416. Inspection of the means show that respondents who used a smartphone did

have higher recall of the advertisement (M= 3.89, SD= .35) compared to those who used a laptop (M= 3.83, SD= .62), although the differences are minimal and not significant.

Manipulation Type of ad Range Mean SD

Smartphone Intrusive 0-4 3.98 .14

Smartphone Non-intrusive 0-4 3.79 .46

Desktop Intrusive 0-4 3.82 .76

Desktop Non-intrusive 0-4 3.83 .52 Table 7, Means of dependent variable.

Hypothesis 1:

The results show that the independent variable device size did not have an effect on advertising effectiveness. Therefore, Hypothesis H1: Pre-roll ads shown on large

(30)

Moderator – Two-way ANOVA

To test the moderator effect of intrusiveness a two way ANOVA was performed. Due to the fact that the sample sizes are unequal, which violates one of the assumptions of a two-way ANOVA, homogeneity can be affected. Levene’s test is significant at p= .001, this shows the variances across the groups are not equal. For this reason the assumptions for two-way ANOVA are violated. Meaning the results of the interaction effect cannot be accounted for. The test of between subjects shows that the interaction effect between screen size-intrusiveness on recall is not significant, F (1, 175) = 1.79 p = .183.

Hypothesis 2: Moderating variable: intrusiveness and ad effectiveness.

The results demonstrated that the moderator intrusiveness did not have an effect on the relationship between a screen size and advertising effectiveness. With these results, hypothesis H2: Intrusiveness pre-roll advertisements negatively affect the

relationship of screen size on advertising effectiveness, cannot be accepted.

Since the data does not meet criteria for a decent analysis and the fact that there are no differences between the groups in recall regarding screen size and intrusiveness, the two hypotheses are rejected. No other significant main effects were demonstrated.

Moderator manipulation check

There was no significant effect for the manipulation intrusiveness of the ad t (177) = 1.43 p= .154. Respondents were asked how much they disliked having to wait for the ad in the experiment. The respondents that had been exposed to the intrusive had a mean of M= 3.89 (SD= 1.13), and the respondents exposed to the non-intrusive ad had mean of M= 4.12 (SD = .95). This shows that respondents that were exposed to the non-intrusive had a higher dislike, although this result is not significant. This shows that people did not experience the ad as different, meaning the manipulation did not work, contrary to findings from previous research (Pashkevich et al. 2012). Therefore, the results of the moderator cannot be generalized.

Duration prompt

(31)

Perceived intrusiveness

No significant difference was found between the non-intrusive group as M= 4.35 (SD= .70) and intrusive group as M= 4.42 (SD= .56) on the possibility to skip the ad, t(177) = -.76, p= .449. Therefore, the people who could skip the ad perceived it as intrusive as the group who could not skip the ad. Despite the fact that the results are not significant this finding is in contrast with previous research of Pashkevich et al. 2012.

ANCOVA – Seen ad before

Since they have seen the advertisement before, they already knew what the ad was about. When the variable (seen_before) has an influence on the results, it should be used as a covariate in the analysis. The ANOVA analysis shows a non-significant effect of device size on recall p= .416. After conducting the ANCOVA, by controlling for having seen the ad before, the effect of device size on recall is not significant p= .526, showing this variable (seen_before) does not have an effect on the dependent variable.

ANCOVA – Seen ad in experiment

An ANCOVA was performed to measure whether the variable seen_ad, that measured whether respondents actually encountered the shown ad in the experiment, has an influence on the results. Of the 179 respondents, 94.44% of the respondents, that is 169 of them reported they had seen the pre-roll ad in the experiment. The underlying assumption of homogeneity of variance for the one way ANCOVA has been met, as F(1, 177) = 2.42, p = .121. After conducting the ANCOVA, by controlling for having seen the ad in the experiment, the effect of device size on recall is significant F(1, 176) = 9.49, p = .007. This shows there is a relationship between the covariate and the dependent variable.

ANCOVA - Ad relevance

(32)

Conclusion

The pre-roll advertisement has become a very popular online advertising tool in a relative short period of time. Growth has emerged both on laptops and on smartphones, and nowadays the pre-roll virtually exists on every major website that features videos (Emarketer 2013). However, due to its relative newness and quick growth, research on the effectiveness of the pre-roll is in its early stages. This study makes several contributions to the field of advertising effectiveness of pre-roll advertisements. Due to the fact that a large proportion of this study resulted in statistically non-significant outcomes, most of the findings provide directions for topics for future research. Firstly, contrary to the hypothesis, no significant main effect was found, as advertising effectiveness in terms of recall does not change when screen size changes. Secondly, recall scores for both devices are very high, indicating the pre-roll ad seems to be very effective at leaving a short-term memory trace. This could be one of the reasons why no significant effects have emerged as the mean recall scores of both groups are very close to the maximum score. Thirdly, although also not statistically significant, intrusiveness does not seem to affect the relationship between screen size and advertising effectiveness in terms of recall. Fourthly, in contrast to the literature, it appears intrusiveness does not change when a skip ad button and duration prompt are added, indicated by the manipulation check of the moderator.

Discussion

In order to answer the research question, empirical research has been performed. The research question - How effective are online pre-roll ads on desktop and mobile

devices? – was answered using two hypotheses that acted as a guideline through the

theoretical framework. This study focussed on how screen size affects advertising effectiveness measured by aided recall. It was expected that (H1) Pre-roll ads shown

on large screens have a higher advertising effectiveness than pre-roll ads shown on small screens. In addition a moderator was added; the intrusiveness of pre-roll

advertisements. Intrusiveness (H2) Intrusiveness pre-roll advertisements negatively

affect the relationship of screen size on advertising effectiveness. In the conducted

(33)

Although no significant main effects were found for screen size, respondents from both devices had very high recall scores, indicating a high overall effectiveness of the advertisement. This finding is consistent with previous literature that the video advertisement is very salient and triggers cognitive processing (Chaiken and Eagly 1983; Rieber 1990; Lang 2000) and could therefore be very effective in terms of recall.

For the second hypothesis (H2) no significant main effect was found that intrusiveness of the advertisement had a negative effect on the hypothesized relationship between screen size and advertising effectiveness. However, the results of the manipulation check are remarkable, even though statistically non-significant findings should always be taken with care. In contrast to the theoretical framework, the skip-ad option did not decrease the intrusiveness of the advertisement. The respondents exposed to the non-intrusive ad even disliked the ad more than the intrusive group did. One possible explanation for this occurrence is that by definition all pre-roll advertising is intrusive as it blocks the viewer in its goals (Tucker and Goldfarb 2011). Even though the viewer is empowered to skip the ad, it is still perceived as intrusive. As previous research indicates, intrusiveness can potentially lead to cognitive avoidance and lowered advertising effectiveness (Thorson and Zhao 1997; Ehrenberg and Twyman 1967).

Limitations and future research directions

(34)

Secondly, the advertisement is very salient the experiment and task involvement of the respondent is high. Therefore memory traces of the ads are very accessible and recall is high (Chattopadhyay and Nedungadi, 1992). Normally, ads in real life situations are often processed with low levels of involvement, and do not receive a lot of cognitive resources (Chaiken, 1980; Petty and Cacioppo, 1986). Although this study attempted to mimic a real life situation by priming the respondent to focus on the funny viral video, it still is an experiment where respondents have higher levels of involvement and cognitive processing (Chattopadhyay and Nedungadi, 1992). Furthermore this study has only tested the recall of one advertisement, while in real life people tend to watch more online videos per month resulting in an increased exposure to ads. This heightened exposure to ads can result in ad avoidance and decreased levels of processing (Dreze and Hussher 2003), creating an even larger dissimilarity between the experiment setting and real life situation. Moreover, when ad content is matched to its environment, it appears recall of the commercial is enhanced (Moorman, Neijens, and Smit 2005). In this case this could mean the content of the ad matches the device, as KPN and 4G Internet are services for smartphones and thus result in enhanced processing that decrease the hypothesized difference in effectiveness.

Thirdly, the advertisement features a brand that is well known in the Netherlands (Synpact 2012). Established brands are well known and viewers can process these advertisements more easily due to perceptual fluency (Fennis and Stroebe 2010). The brand featured in the experiment, KPN, might already be associated with telecommunications and 4G Internet. An analysis of covariance could have been conducted to control for brand knowledge. Previous exposure to the ad did not have a significant influence on recall as the results of an analysis of covariance indicated. However, future research could use fake or unknown ads to the respondents so that a previous encounter with the ad is impossible, brand awareness is non-existent, and a better measure of recall is achieved.

(35)

intrusiveness failed. Therefore, an interesting research topic for future research is to find out when a pre-roll ad is perceived as intrusive and if pre-roll ads can be perceived as non-intrusive.

Fifthly, the sample size does not accurately reflect society, making generalizations on the data is rather difficult, although the data could be appropriate for highly educated young adults. It would be suitable to repeat this study with a larger sample size to be able generalize conclusions and provide managers with valuable data for their forthcoming advertising campaigns.

Sixthly, respondents had to fill out the recall questions shortly after being exposed to the advertisement. Recall tends to be higher when the time between the stimulus and question is rather short. A set of follow-up questions could have been sent to respondents a week after the experiment in order to check whether they would still recall what ad they have seen in the experiment. Moreover, it could be that the recall questions were not difficult enough, so that nearly all respondents could remember the advertisement and differences did not occur.

Lastly, since the measure for effectiveness in this study is solely based on recall and recognition, the scope of the results is limited to memory effects. Other variables that could measure effectiveness, like purchase intention or brand attitude, synergy effects, how they are interrelated, or cross-device differences like the environment the viewer is in, matching pre-roll ad content to the desired video, persuasion of the ads. Even a combination of a click-through rate might provide interesting results and paving the way for a more holistic model of pre-roll advertising effectiveness. This however might not be cost effective to measure the effectiveness of ads on all these levels, but when a small proportion is measured using recall, and the rest through click-stream data, advertising campaigns can be measured and valued more accurately.

References

(36)

Aksakalli, V. (2012). Optimizing direct response in Internet display advertising.

Electronic Commerce and Research Applications, 11, 229-240.

Atkinson, R. C., & Shiffrin, R. M. (1968). Human memory: a proposed system and its control processes” In Spence, K. W., & Spence, J. T., The psychology of Learning

and Motivation: Advances in Research and Theory, 89-195. New York: Academic

Press.

Balanda K. P., & MacGillivray H.L. (1988). “Kurtosis: A critical review”. The

American Statistician, 42, 111–119.

Barwise, P. (2001). TV, PC, or mobile? Future media for consumer e-commerce.

Business Strategy Review, 12, 35-42.

Bellman, S., Schweda., A., & Varan, D. (2009). Viewing angle matters, screen type does not. Journal of Communication, 59, 609-634.

Bellman, S., Treleaven-Hassard, S., Robinson, J. A., Rask, A., & Varan, D. (2012). Getting the balance right, Commercial loading in online video programs. Journal of

Advertising, 41, 5-24.

Bhat, S., Bevan, M., & Sengupta, S. (2002). Measuring users’ web activity to evaluate and enhance advertising effectiveness. Journal of Advertising, 31, 97-106.

Bond, T. G., & Fox, C. M. (2001). Applying the Rasch model: Fundamental

measurement in the human sciences. Mahwah, NK: Lawrence Erlbaum.

Brown, T. J., & Rothschild, M. L. (1993). Reassessing the impact of television advertising clutter. Journal of Consumer Research, 20, 138-146.

Bucklin, R. E. & Catarina S. (2009). “Click here to Internet insight: Advances in clickstream data analysis in marketing, “Journal of Interactive Marketing, 23, 35-48.

(37)

CBS. (2013). Onderwijsniveau bevolking gestegen. Retrieved May 30, from http://www.cbs.nl/nl-NL/menu/themas/onderwijs/publicaties/artikelen/archief/2013/ 2013-3905-wm.htm.

Calder, B., Malthouse, E. C., & Schaedel, U. (2009). An Experiental study of the relationship between online engagement and advertising effectiveness. Journal of

Interactive Marketing, 23, 321-331.

Catch Your Moment. (2014). Average view duration of YouTube videos. Retrieved June 16, 2014 from https://www.youtube.com/analytics ?o=U#dt=nt,fs=1621 5,fe=16242,fr=lw-001;fcr=0,r=retention.

CBS. (2011). Mannen en vrouwen in Nederland. Retrieved May 30, from http://www.cbs.nl/NR/rdonlyres/9A0E2D35-B9B6-4BB0-B6D5-C9727B3F0181/0/ 2011k1b15p37art.pdf

Chaiken, S., & Eagly, A. H. (1983). Communication modality as a determinant of persuasion: The role of communicator salience. Journal of Personality and Social

Psychology, 45, 241-256.

Chaiken, S. (1980). Heuristic versus systematic information processing and the use of source versus message cues in persuasion. Journal of Personality and Social

Psychology, 39, 752-766.

Chan, A., Dodd, J., & Stevens, R. (2004). The efficacy of pop-ups and the resulting effect on brands. A white paper: Bunnyfoot Universality

(38)

Choo, C., Lee, J., & Tharp, M. (2001). The effects of different forced exposure levels to banner advertisements on the WWW. Journal of Advertising Research, 41, 45-57. Cisco (2014), Global mobile data forecast, 2013-2018. Retrieved March 28, 2014 from http://www.cisco.com/c/en/us/solutions/collateral/service-provider/visual-networking-index-vni/white_paper_c11-520862.html

Cochran, W. G., (1947). Some consequences when the assumptions for the analysis of variance are not met. Biometrics, 3, 22-38.

ComScore (2014). US Digital future in focus. Retrieved April 3, 2014 from http://www.comscore.com/Insights/Presentations_and_Whitepapers/2014/2014_US_ Digital_Future_in_Focus.

ComScore. (2013). Video Metrix, U.S. Desktop, December 2013. Retrieved March 4, 2014 from https://www.comscore.com/Insights/Press_Releases/2014/1/comScore _Releases_December_2013_US_Online_Video_Rankings

ComScore. (2012). 2012 US Digital future in focus. Retrieved March 4, 2014 from http://www.comscore.com/Insights/Press_Releases/2012/2/comScore_Releases_the_2 012_U.S._Digital_Future_in_Focus

Cunningham, T., Hall, A. S., & Young, C. (2006). The advertising magnifier effect: An MTV study. Journal of Advertising Research, 4, 46.

Dawes, J. (2008). "Do data characteristics change according to the number of scale points used? An experiment using 5-point, 7-point and 10-point scales". International

Journal of Market Research, 50, 61–77.

De Winter, J. C. F. (2013). Using the student’s t-test with extremely small sample sizes. Practical Assessment, Research and Evaluation, 18.

(39)

Dimmick, J., Kline, S., & Stafford, L. (2000). The gratification of niches of personal email and the telephone. Communication Research 27, 33-50.

Dorai-Raj, S., Liu, H., & Zigmond, D. (2011). “advertising and traffic: Learing from

online video data.” Presentation at the ARF Audience Measurement conference,

2011.

Dreze, X., & Hussher, F. X. (2003). Internet Advertising: Is anybody watching?

Journal of Interactive Marketing, 17, 8-23.

Dwyer, P. (2007). Measuring the value of electronic word of mouth and its impact in consumer communities. Journal of Interactive Marketing, 21, 63-79.

Ehrenberg, A. S. C., & Twyman, W. A. (1967). On measuring television audiences.

Journal of the Royal Statistical Society, 130, 1-60.

Elliot, B. W., Hodge, F. D., & Sedor, L. M. (2012). Using online video to announce a restatement: Influences on investment decisions and the mediating role of trust. The

Accounting Review, 87, 513-535.

Eysenck, M. W., & Calvo, M. G. (1992). Anxiety and performance: The processing efficiency theory. Cognition and Emotion, 6, 409-434.

Fassbender, E., Richards, D., Bilgin, D., Thompson, & A., Heiden, W. (2012). VirSchool: the effect of background music and immersive display systems on memory for facts learned in an educational virtual environment. Computers and

Education, 58, 490-500.

Fennis, B. M., & Stroebe, W. (2011). The psychology of advertising. Hove and New York: Psychology press.

(40)

Franken, R. B. (1925). The attention value of newspaper advertisements. New York: Association of National Advertisers.

Gao, Y., Koufaris, M., & Ducoffe, R. H. (2004). An experimental study of the effects of promotional techniques in web-based commerce. Journal of Electronic Commerce

in Organizations, 2, 1-20.

Gardner, M. P. (1985). Mood states and consumer behaviour: A critical review.

Journal of Consumer Research, 12, 281-300.

Gatarski, R. (2002). Breed better banner: Design automation through online interaction. Journal of Interactive Marketing, 16, 2-13.

Goldfarb, A., & Tucker, C. (2011a) Online Advertising. Advances in Computers, 91, 289-315.

Goldfarb, A., & Tucker, C. (2011b). Online display advertising: Targeting and obtrusiveness. Marketing Science, 30, 389-404.

Goodman, L. A. (1961). Snowball sampling. Annals of Mathematical Statistics, 32, 148-170.

Goodrich, K. (2011). Anarchy of effects? Exploring attention to online advertising and multiple outcomes. Psychology & Marketing, 28, 417-440.

Gorn, G. J. (1982). The effects of music in advertisements on choice behavior: a classical conditioning approach. Journal of Marketing, 46, 94-101.

(41)

Hair, J., Anderson, R. E., Tatham, R. L., & Black, W. C. (1995). Multivariate data

analysis. 4th ed. New Jersey: Prentice-Hall Inc.

Hammer, P., Riebe, E., & Kennedy, R. (2009). How clutter affects advertising effectiveness. Journal of Advertising Research, 49, 159-163.

Hanssens, D. M., & Weitz, B. A. (1980). The effectiveness of industrial print advertisements across product categories. Journal of Marketing Research, 17, 294-306.

Heads, J. (1968). Ad sizes and one ad recallers. Journal of Advertising Research. 8, 26-30.

Hendon, D. W. (1973). How mechanical factors affect ad perception. Journal of

Advertising Research, 13, 39-45.

Hollis, N. (2005). Ten years of learning on how advertising builds brands. Journal of

Homer, P. M., (1995) Ad size as an indicator of perceived advertising cost and effort: The effects of memory and perceptions. Journal of Advertising. 24, 1-12

Hong, W., Thong, J. Y. L., & Tam K. Y. (2004). Does animation attract online users’ attention? The effects of flash on information search performance and perceptions.

Information Systems, 15, 60-87.

Hou, J., Nam, Y., Peng, W., & Lee, K. (2012). Effects of screen size, viewing angle, and players immersion tendencies on game experience. Computers in Human

Behavior, 28, 617-623

Howard, J. A., & Sheth, J. N. (1969). The theory of buyer behaviour. New York: John Wiley.

(42)

clutter on brand memory in mega-even broadcasts. International Journal of

Advertising, 30, 617-640.

Jones, M., Marsden, G., Mohd-Nasir, N., Boone, K., & Buchanan, G. (1999). Improving web interaction on small displays. Computer Networks 31, 1129-1137. Emarketer (2013). Trends in video advertising and measurement. Retrieved March 8, 2014 from http://www.slideshare.net/eMarketerInc/emarketer-webinar-trends-in-video-advertising-and-measurement-16163295#)

Kamis, A., & Stohr, E. A. (2006). Parametric search engines: what makes them effective when shopping online for differentiated product? Information and

Management, 43, 904-918.

Kim, K. Y., & Lee, B. G. (2014). Marketing insights for mobile advertising and consumer segmentation in the cloud era: A Q-R hybrid methodology and practices.

Technological Forecasting and Social Change 2014.

Kim, J., & McMillan, S. J. (2009). Evaluation of Internet advertising research: A bibliometric analysis of citations of key sources. Journal of Advertising, 37, 99-112. Kim, S., Qin, T., Liu, T. Y., & Yu, H. (2014). Advertiser-centric approach to understand user click behaviour in sponsored search. Information Sciences, 30, 1-13. Korgaonkar, P. K., & Wolin, L. D. (1999). A multivariate analysis of web usage.

Journal of Advertising Research, 39, 53-68.

Krugman, H. E. (1971). Brain wave measures of media involvement. Journal of

(43)

Lang, A. (2006). The limited capacity model of mediated message processing.

Journal of Communication, 5, 46-70.

Lang, A., Bolls, P., Potter, R., & Kawahara, K. (1999). The effects of production pacing and arousing content on the information processing of television messages.

Journal of Broadcasting and Electronic Media, 43, 451-574.

Leight, K. A., & Ellis, H. C. (1981). Emotional mood states, strategies, and state-dependency in memory. Journal of verbal learning and verbal behaviour, 20, 251-275.

Leppaniemie, M., & Karjaluoto, H. (2008). Mobile marketing: From marketing strategy to mobile marketing campaign implementation. International Journal of

Mobile Marketing, 3, 50-61.

Li, H., & Bukovac, J. L. (1999). Cognitive impact of banner ad characteristics: an experimental study. Journalism and Mass Communication Quarterly, 76, 341-453. Li, H., Edwards, S. M., & Lee, J. H., (2002). Measuring the intrusiveness of advertisements: Scale development and validation. Journal of Advertising, 31, 37-47. Lindstrom, M. (2011). You love your iPhone, literally. The New York Times. Retrieved June 15, 2014 from http://www.nytimes.com/2011/10/01/opinion/you-love-your-iphone-literally.html?_r=1&

Liu, C. E., Sinkovics, R. R., Pezderka, N., & Haghirian, P. (2012). Determinants of consumer perceptions toward mobile advertising: A comparison between Japan and Austria. Journal of Interactive Marketing 26, 21-32.

Lombard, M., & Ditton, T., (1997) At the heart of it all: The concept of presence.

Journal of Computer Mediated Communication, 3, 0.

(44)

Luo, W., Cook, D., & Karson, E. (2011). Search advertising placement strategy: Exploring the efficacy of the conventional wisdom. Information & Management, 48, 404-411.

Manchanda, P., Dube J. P., Goh, K. Y. & Chintagunta, P. K. (2006). The effect of banner advertising on Internet purchasing. Journal of Marketing Research, 43, 98-108.

Mandeville, G. K. (1969). A Monte carlo investigation of the adequacy of standard

analysis of variance test procedures for dependent binary variates. University of

Minnesota.

Marketing Charts (2011). Mobile marketing spend to sextuple. Retrieved April 2, 2014 from http://www.marketingcharts.com/wp/uncategorized/mobile-marketing-spend-to-sextuple-2010-2015-17282/

McCoy, S., Everard, A., Polak, P., & Galetta, D. F., (2008). An experimental study of antecedents and consequences of online ad intrusiveness. International Journal of

Human-Computer Interaction, 27, 672-699.

Mcniven, M., Krugman, D., & Tinkham, S., (2012). The big picture for large-screen television viewing. Journal of Advertising Research, 52, 421-432.

McQuail, D. (1983). Mass Communication Theory: An introduction. London: Sage Publication.

Mobile Marketing Association (2014). Mobile video benchmark study. Retrieved June 17, 2014 from http://www.mmaglobal.com/whitepaper/mobile-video-benchmark-study.

(45)

Nambisan, R., Baron, R. (2007). Interactions in virtual customer environments: Implications for product support and customer relationship management. Journal of

Interactive Marketing, 21, 42-62.

Nottorf, F. (2014). Modeling the clickstream across multiple online advertising channels using a binary logit with Bayesian mixture of normals. Electronic

Commerce Research and Applications, 13, 45-55.

Okazaki, S. (2004). How do Japanese consumers perceive wireless ads? A multivariate analysis. International Journal of Advertising. 23, 429 - 454.

Osborne, J. (2002) Notes on the use of data transformations. Practical Assessment,

Research & Evaluation, 8.

Owen, B., & Wildman, S. (1992). Video Economics, Cambridge, MA: Harvard University Press.

Paek, H., Hove, T., Jeong, H., & Kim, M. (2011). Peer or Expert? The persuasive impact of YouTube video producers and their moderating mechanism. International

Journal of Advertising, 30, 161-188.

Pavio, A. (1986). Mental representations: A dual coding approach. New York: Oxford University Press.

Pashkevich, M., Dorai-Raj, S., Kellar, M., & Zigmond, D (2012). Empowering online advertisements by empowering viewers with the right to choose. Journal of

Petty, R. E., & Cacioppo, J. T. (1986). Communication and persuasion: Central and

peripheral routes to attitude change. New York: Springer.

Punyatoya, P. (2011). How effective are Internet banner advertisements in India.