U
NIVERSITY OF
A
MSTERDAM
Pre-App Advertising
A different approach to mobile application
advertising
Master Thesis
University of Amsterdam
Tom W.T. Peeters
10475575
Master thesis
June, 2014
Final version
Prof. Dr. Ed Peelen
Executive Programme in Management studies
Marketing Track
Amsterdam, the Netherlands
Pre-App Advertising
A different approach to mobile application
advertising
Abstract
World leading company Apple realized over 40 billion mobile application downloads in 2013 (Apple, 2014). In addition, consumers spend over 2 hours per day within mobile applications (Khalaf, 2013). Not without a reason that companies start targeting the mobile application business to engage with its customers. To illustrate, mobile app advertising has become the fastest growing sector in mobile advertising reaching 17 billion dollars in expenditures by 2018 (Juniper, 2013). Despite the golden era of mobile app advertising has been intensively verified, research concerning this type of advertising business is still very limited. Even more so, research findings indicate that current app advertising strategies just don’t work (Gupta, 2013). Therefore, this research was aimed to enhance research in the field of mobile application marketing. Besides, it was intended to optimize current mobile application scheduling strategies by introducing a nearly untouched and novel approach to mobile application advertising, known as Pre-App Advertising.
To investigate its potential, it was decided to measure the difference in advertising effectiveness between this novel advertising strategy and the most frequently used in-app advertising strategy. Herewith, the variables recall, recognition and attitude were investigated as the leading antecedents for advertising effectiveness. Besides, skip option and ad size as two influencing factors were elaborated upon. The method consisted of an online WordFeud app experiment accompanied with a subsequent survey. In total 871 respondents from a multinational engineering and consultancy firm were collected for participation purposes. Based on the tested hypotheses, it was found that the pre-app advertising strategy entails strong opportunities to either outcompete or support current in-app advertising strategies. And not only that, results revealed that the difference in potential between both advertising strategies was far from exploited. In other words, the added value of pre-app advertising seems to be much higher than this research illustrated. In addition, the impact of skip option and ad size on advertising effectiveness was found to be significant as well. Due to the little empirical work conducted within this field accompanied with the strong significance of the research findings, this study will raise the awareness of scientific researchers as well as practitioners concerning the golden opportunities of mobile application advertising like pre-app advertising. As a consequence, the strong acceleration in scientific research concerning mobile application advertising as well as the embracement of pre-app advertising in practice seems inevitable.
Keywords: Advertising, E-Commerce, Communication, Mobile Advertising, Mobile applications, Apps, Mobile Marketing, Pre-App Advertising, In-App Advertising, Marketing Strategy, Innovation.
Table of Contents
Abstract ...3
1. Introduction ...5
2. Literature review...7
2.1 Pre-app advertising vs In-app advertising ...7
2.2 Measuring advertising effectiveness ...8
2.3 Which type of pre-app advertising strategy works best? ... 13
3. Data and method ... 16
3.1 Research strategy ... 16
3.2 Sample and procedure ... 17
4. Results and discussion ... 22
4.1 Pre-app advertising vs In-app advertising ... 22
4.2 Which type of pre-app advertising strategy works best? ... 29
5. General discussion ... 39
5.1 Managerial and scientific implications ... 40
5.2 Limitations and future directions ... 41
6. Conclusions ... 43
References ... 45
Appendix 1: Invitation letter ... 50
Appendix 2: Experiment ... 51
1. Introduction
“Flappy Bird app racks up 50.000 dollars in revenues per day due to mobile application advertising (The Verge, 2014).”
Since the rise of the internet in the 1990s, the traditional advertising media like billboard, print and TV advertising is slowly losing ground. In the next few years we are approaching a big milestone in which online advertising will take over TV advertising as the number 1 spot in advertising expenditures (eMarketer, 2013). A logical consequence, since the online world is rapidly taken over traditional offline services like purchasing, trading, watching TV and videos, playing games and chatting (New York Times, 2012). As with every new introduction, the introduction of online advertising is facing new issues and dilemmas and therefore additional research is required to enhance the effectiveness of online advertising.
Up to date different variables have been used to investigate the effectiveness of online advertising. Some researchers focused on creating advertising awareness or enhancing purchase behavior (Dreze & Hussherr, 2003; Ilfeld & Winer 2002; Sherman & Deighton, 2001; Manchanda et al., 2006) while others focused on knowledge, liking, preference and conviction of advertising (Brackett & Carr, 2001; Tsang, Ho, & Liang, 2004; Li, Edwards, & Lee, 2002). Although conventional wisdom implies that current online advertising strategies are very effective, more and more insights challenge this thought. Customers generally don’t like advertisements (McCoy et al., 2007) and even start to avoid, ignore and block advertising messages (Baek & Morimoto, 2012). Therefore, it is time to change our mindset and opt for better online advertising strategies. Advertising strategies that customers are willing to accept.
Customers are using different devices for online purposes. Herewith, smartphones are becoming increasingly popular compared to the traditional desktops and laptops to fulfill this job (eMarketer, 2013). As a consequence, mobile advertising expenditures are growing more quickly than any other advertising expenses. By 2017, U.S. mobile advertising is expected to account over 15% of all ad expenses, equaling over 31 billion dollars in total expenditures (eMarketer, 2013). Combined with the fact that advertising strategies and effectiveness can be totally different in the smartphone era compared to the traditional device era, online advertising research is more and more focusing on mobile advertising (Gao et al., 2013; Watson, McCarthy, & Rowley, 2013). This trend will further accelerate since the dramatic increase in mobile applications, so called apps (Raines et al., 2013). Apps enable customers to continuously reconfigure their smartphones to their preferred conditions and are nowadays very popular (Raines et al., 2013; Persaud & Azhar, 2012). To illustrate, the Apple store realized over 40 billion application downloads in 2013 (Apple, 2013). Besides, recent research studies illustrated that people are spending more times on apps than anything else on their
smartphone. On average U.S. consumers spend over two hours per day within apps (Khalaf, 2013). Since applications are introduced just recently, the amount of research related to advertising in mobile applications, so called in-app advertising is still very limited (Raines et al., 2013). Even more so, current in-app advertising strategies just don't work and success depends on how marketers place their messages and whether customers feel comfortable with it (Gupta, 2013). Although this advertising strategy is still under discussion, it is growing exorbitantly and has become the fastest growing sector in mobile advertising. To illustrate, the in-app advertising spending is expected to increase from 3.5 billion dollars in 2013 up to 17 billion dollars by 2018 (Juniper, 2013). Therefore, we cannot deny the desperate need for additional research aimed at tackling the specific dilemmas and issues of mobile application advertising as well as exploiting its golden opportunities.
Research objectives
This research is aimed at contributing to literature and practice by elaborating a nearly untouched and novel approach to mobile application advertising, known as Pre-App Advertising. A specific advertising strategy for mobile applications in which the provision of content and advertising is disconnected. In this way, it is intended to open up the gates for additional research to enhance its theoretical understandings as well as its potential in practice. Besides introducing the concept of pre-app advertising, it is decided to introduce the first research bricks related to this novel advertising strategy. Herewith, the following research question is proposed:
The next chapter is aimed at the provision and synthesis of existing literature as well as the elaboration and justification of the proposed hypotheses. Subsequently, the research design, data collection, analysis methods as well as the justification of the methodology is provided. After conducting the experiment and subsequent survey, the empirical results are presented. This will fuel the discussion section in which the defined hypotheses and acquired data are discussed. This paper ends with the conclusive findings and highlighting its huge practical implications as well as its suggestions for future research.
Research question:
To what extent is the advertising effectiveness in mobile applications affected by changing the mobile advertising strategy from in-app advertising to pre-app advertising?
2. Literature review
2.1 Pre-app advertising vs In-app advertising
Since the rise of YouTube in 2005, the interest in online video streaming exploded. This opened up golden opportunities for the subsequent introduction of advertising. Therefore, in recent years different advertising strategies for online video content have been investigated. In terms of ad location, four main strategies can be distinguished: (1) banner ad, (2) pre-roll, (3) mid-roll and (4) post-roll advertising (Mei et al., 2010). One of the main differences between banner ad advertising and the other three strategies is related to its advertising scheduling. Within banner advertising, advertisements are introduced while watching video (content), while for pre-, mid- and post-roll advertising, the ads are integrated prior, during or after viewing content (Mei et al., 2010; Bellman et al., 2012). For clarity purposes, the difference between the four advertising scheduling strategies is illustrated in table 2-1.
Table 2-1: Difference in advertising scheduling strategy between banner-ad, pre-roll, mid-roll and post-roll advertising
For online videos, especially the pre-roll and mid-roll advertising strategies are gaining extreme popularity. Most probably, because these strategies are more effective and less intrusive than banner ad advertising (Mei et al., 2010). It is quite remarkable that these widely introduced pre-, mid- and post-roll types of advertising strategies for online video streaming have not gained much scientific attention for other online advertising purposes like mobile application advertising (Kusse, 2013). Even more so, the practical implementation of disconnecting advertising from content provision in mobile applications is very limited. Fortunately, it seems that companies are becoming more and more aware of its business potential. To illustrate, some mobile advertising companies started with developing software to provide video advertising during app launch, between game levels or during screen change (Bremer, 2014; Vdopia, 2010).
The mobile application process can be divided into three main phases 1) the ‘pre-app’ phase of the application in which the application is launched or loaded, 2) the ‘in-app’ phase in which the application is used and 3) the ‘close-app’ phase in which the application is closed. Up to date, research as well as the practical applicability has been intensively focused on the introduction of advertising during phase 2. In contrary, advertising during the other phases has gained very limited
attention for mobile application advertising. More specifically, no scientific research aimed at investigating the impact of advertising during either phase 1 or 3 has been identified. Therefore, we cannot deny the strong research gap this paper is aimed to bridge. Due to the extreme popularity of pre-roll advertising in practice (YouTube, 2014; Kusse, 2013) and the development of advertising during app launch (Vdopia, 2010), it is decided to focus on advertising during phase 1, the start-up phase of mobile applications. This novel and high potential mobile application advertising strategy is known as ‘Pre-App Advertising’ (Bremer, 2014). An approach in line with the pre-roll advertising strategy for online videos.
2.2 Measuring advertising effectiveness
Since it is nearly impossible to quantify the return of advertising, antecedents of advertising effectiveness have been used to reflect on its subsequent impact (Lewis & Rao, 2013). Within this group of antecedents, advertising recall, recognition and attitude are among the most extensively used and accepted parameters to fulfill this purpose (Bruin, 2010). Moreover, brand recall and recognition are confirmed to be the core foundation in the development of a long term relationship with customers (Keller, 2008). As a consequence, intensive research has been conducted on those influencing factors to enhance its understanding as well as its resulting impact on advertising effectiveness.
2.2.1 Recall and recognition
The definition as well as the difference between brand recall and recognition is explained within the famous developed conceptual model of brand equity (Keller, 1993). This model conveys an overview of the most important components influencing the differential impact of brand knowledge on the consumers’ responses to brand marketing as illustrated in Figure 2-1.
Pre-App Advertising
Includes advertisements in mobile applications, solely during the start-up phase
Disconnects the provision of content and advertising, in which advertising is provided prior to in-app use (viewing content)
As can be derived from this figure, brand knowledge constitutes of both brand image and brand awareness. The former is related to the types, favorability, strength and uniqueness of the brand association while the latter is related to both brand recall and brand recognition. Herewith, brand recall is defined as “the consumers’ ability to retrieve the brand when given the product category, the
needs fulfilled by the category or some other type of probe as a cue” while brand recognitions is
defined as “the consumers’ ability to confirm a prior exposure to the brand when given the brand as
the cue” (Keller, 1993, p. 3). In other words, brand recall requires the consumer to generate the
brand from memory while brand recognition is realized when the exposed brand is recognized from a list of brands. Consequently, it can be argued that both variables have to be researched separately since brands that can be recalled can be recognized while the other way around is not always the case (Lerman & Garbarino, 2002).
Advertising repetition is a well known factor influencing recall and recognition (Lee & Cho, 2010;
Dreze & Hussherr, 2003). This theory builds on the traditional advertising theory, like TV advertising in which the positive relationship between repetition and both recall and recognition has been extensively verified (Rethans, Swasy, & Marks, 1986). Besides, it is widely accepted that the motivation and ability to process the message strongly affects the recall and recognition of advertisements (Cacioppo & Petty, 1983; Schumann, Petty, & Clemons, 1990). Motivation and ability are the well known antecedents of elaboration likehood as described in the famous elaboration likehood model (ELM) (Cacioppo & Petty, 1983). Besides, product involvement plays an eminent role in recall and recognition (Petty, Cacioppo, & Schumann, 1983). In case of high product involvement, advertising processing generally takes place via the central route which increases the advertising recall and recognition. This in contrary to low product involvement in which advertising processing
takes place via the peripheral route which subsequently reduces brand recall and recognition (Petty, Cacioppo, & Schumann, 1983). Another, more specific influencing factor for online advertising recall and recognition is the mental mode of the web user. Research has illustrated that goal-directed web users are much less likely to recall and recognize online advertisements compared to users who are freely surfing a website (Danaher & Mullarkey, 2003). The latter research finding is in line with the relationship found between task orientation, advertising avoidance and advertising recall (Hershberger & Costea, 2009). This specific research concluded that the higher the task orientation the higher the subsequent advertising avoidance. As a consequence, the advertising recall will be reduced as depicted in Figure 2-2.
Recent pre-roll advertising research for online videos found a significant higher recall rate compared to the standard schedule of advertising, banner ad advertising (Yahoo, 2011). Most likely, this is related to the fact that within pre-roll advertising, video advertisements are provided before viewing video content while for standard banner-advertising, content and advertisements are provided simultaneously. For the latter situation, customers are enabled to focus on the content while advertisements are provided. As a consequence, task orientation is increased which adversely affects the advertising recall (Figure 2-2). As aforementioned, the advertising scheduling of pre-app advertising is similar to pre-roll advertising while in-app advertising is related to the standard banner advertising strategy. This has contributed to the introduction of the following hypotheses:
2.2.2 Advertising attitude
In the early stages of research related to advertising attitude, scientific studies were mainly focusing on brand attitude (Ab) and its subsequent impact on brand choice (Shimp, 1981). Herewith, brand
attitude can be defined as the overall evaluation of the brand influenced by its salient believes (Keller, 1993; Fishbein & Ajzen, 1975). However, as research and insights related to advertising attitude accumulated in time, the attitude towards the advertisement (Aad) gained more and more attention
H1a: The pre-app advertising strategy would result in a higher ad recall compared to the
in-app advertising strategy
H1b: The pre-app advertising strategy would result in a higher ad recognition compared to the
in-app advertising strategy
as a positive mediator for brand attitude (MacKenzie, Lutz, & Belch, 1986; Gardner, 1985; Mitchell & Olson, 1981; Shimp, 1981). Even more so, it was suggested to target advertisement attitude (Aad) and
brand attitude (Ab) as two distinctive variables (Mitchell & Olson, 1981). Logically, since investigating
the attitude towards the ad as separate variable is of extreme importance in case the added value of a brand is neglectable compared to its competitive brands (Mitchell & Olson, 1981). The consecutive research related to the mediating effect of attitude towards the advertisement on brand attitude contributed to the development of the well known Dual Mediation Hypothesis (DMH) model (MacKenzie, Lutz, & Belch, 1986). This DMH model can be labeled as an extension of the famous Elaboration Likehood Model (ELM) and is illustrated in the figure below (MacKenzie, Lutz, & Belch, 1986; Petty, Cacioppo, & Schumann, 1983).
The attitude of consumers towards the advertisement (Aad) and its subsequent attitude towards the
brand (Ab) are dependent on many factors. In the past decade research studies have been conducted
to enhance the understandings of advertising attitude. Most research findings concluded that ad perception, ad credibility, mood, nature of exposure, attitude towards the advertiser (Aad) and
advertising in general (Ag) are influential factors affecting advertising attitude (MacKenzie & Lutz,
1989). Studies specifically focusing on online advertising found that the entertainment level, skip option, perception, personal relevance, credibility, irritation and intrusiveness level are strong antecedents of advertising attitude (Parreno et al., 2013; Kusse, 2013; McCoy et al., 2008; Tsang, Ho, & Liang, 2004; Brackett & Carr, 2001). Other influencing factors include information, relaxation, escape possibilities, advertising duration time, social interaction and the inclusion of personalized advertising (Tsang, Ho, & Liang, 2004; Banerjee & Dholakia, 2008). Personalized advertising is the tailoring of advertising messages to individual profiles and preferences. This type of advertising is expected to grow exorbitantly in upcoming years (WPP, 2014). Location based advertising (LBA) is such an example in which marketers align their advertising messages to the specific locations of customers. Although its potential is extensively verified, additional research is required to ensure appropriate applicability in practice. This since it has been found that LBA can decrease advertising attitude as well depending on its perceived intrusiveness (Banerjee & Dholakia, 2008).
In line with the above developed hypotheses for advertising recognition and recall, hypothesis for advertising attitude can be derived from research related to online video advertising. This since useful research related to advertising scheduling for mobile applications has not been identified. Based on the difference in advertising scheduling, it was found that banner ad advertising is perceived as more intrusiveness compared to pre-roll advertising (Mei et al., 2010). Most probably, this will be the case as well for in-app advertising compared to pre-app advertising. Besides, recent research for online videos found out that decreasing the commercial duration time enhances advertising attitude significantly (Kusse, 2013). Logically, since customers generally don’t like advertisements and therefore prefer limited advertising time spans (Baek & Morimoto, 2012; McCoy et al., 2007). Therefore, it can be expected that the attitude towards pre-app advertising is more positive compared to in-app advertising due to its very limited advertising duration time. To illustrate, advertisements are only provided during the start-up phase lasting maximum 2-5 seconds for pre-app advertising while for in-pre-app advertising the advertisements are incorporated during the in-use phase of the application. For game applications this period may last over 1 hour. Even more so, the pre-app advertising strategy ideally integrates the Nobile prize winner Kahnemans’ developed Peak
End Rule which is extensively verified over the years (Kahneman et al., 1993;Diener, Wirtz, & Oishi, 2001). This rule states that customers’ evaluations of experiences including attitudes are not mainly affected by the overall experience but by its peak and final end (Kahneman et al., 1993). For pre-app advertising, the pleasant peak experience will last the longest as well as it is scheduled at the tailor end of the app experience. In contrary, for in-app advertising the pleasant peak experience will last just 2-5 seconds and is scheduled at the start of the application experience. Therefore, the overall experience for in-app advertising can be expected to be less pleasant. Based on the above stated arguments, the following hypotheses concerning the differential impact between pre-app and in-app advertising on attitude have been proposed:
Since this research is aimed at investigating the novel advertising (scheduling) strategy whereof customers have limited experience with, it is of importance to extend knowledge concerning the attitude towards the advertising strategy as well. In other words, is this novel advertising strategy more preferable compared to the standard in-app advertising strategy? By doing so, its potential can be elaborated upon more effectively. Therefore, an additional hypothesis was introduced. Due to the
H1d: The pre-app advertising strategy would result in a more positive attitude towards the
advertising brand (Ab) compared to the in-app advertising strategy
H1c: The pre-app advertising strategy would result in a more positive attitude towards the ad (Aad)
aforementioned differences in ad exposure duration and app experience concerning both advertising strategies, this includes the following statement:
2.3 Which type of pre-app advertising strategy works best?
The pre-roll advertising strategy for online YouTube videos includes full screen coverage for advertising purposes. Besides, most of the advertisements are accompanied with a skip option (YouTube, 2014). Since pre-app advertising is strongly related to pre-roll advertising, it is decided to integrate skip option and increased ad size for additional pre-app advertising research purposes. The consequential research findings will fuel existing knowledge by its gained insights concerning the question “Which type of pre-app advertising strategy works best?”
2.3.1 Skip option
The commercial skipping option in TV programs increases its advertising attitude significantly (McCoy et al., 2008). Logically, since only customers who accept advertisements will watch them. Moreover, a skip option will allow customers to decide for themselves whether or not it is the appropriate moment to watch advertisements. In this way, the freedom and control is neither threatened nor eliminated which will reduce reactance probability (Brehm, 1966). As a consequence, the attitude towards advertisements is increased (McCoy et al., 2008). The integration of an advertisement skip option is recently researched for pre-roll advertising in online videos (Kusse, 2013). Herewith, it was found that the attitude for this type of advertising may be enhanced by integrating a skip button as well as by reducing the forced commercial advertisement time prior to viewing content. For mobile applications this skip option ideally suits pre-app advertising (iVidopia, 2010). This since in comparison with pre-roll advertising, pre-app advertising includes advertisements prior to viewing content. In line with the research findings for pre-roll advertising, it can be expected that the attitude towards pre-app advertising compared to in-app advertising becomes more positive in case the skip option is included (Kusse, 2013). Besides, the introduction of a skip option forces customers to click on the skip button underneath the advertisement for continuing purposes. In this way, the attention towards the advertisement will be enhanced. As a consequence, both recognition and recall may be beneficially affected as well.
H1e: The pre-app advertising strategy is more preferable compared to the in-app advertising
2.3.2 Ad size
Pre-roll advertising integrates maximum flexibility concerning the ad size or advertisement screen area coverage. In other words, the relative screen area filled with advertising can be adjusted between 1 and 100%. This since for pre-roll advertising, the provision of advertising and content is disconnected. In contrary, banner-ad advertising includes the provision of content and advertising simultaneously which limits the advertising coverage flexibility significantly. Since research has illustrated that ad size beneficially influences memory, this may be another important reason that pre-roll advertising for online videos constitutes a significant higher recall rate compared to the standard schedule of advertising, banner ad advertising (Yahoo, 2011; Homer, 1995). As a derivative from pre-roll advertising, it can be expected that the increased ad size for pre-app advertising has positive consequences concerning its advertising recall and recognition as well.
Besides memory, research demonstrated that a large ad generates a more favorable attitude compared to a small ad (Percy & Rossiter, 1983). However, other research found that increased ad size enhances intrusiveness (Tsang, Ho, & Liang, 2004). Presumably, the latter is the case when the attitude towards the advertisement itself is low (McCoy et al., 2007). Based on the above described influencing factors, two additional pre-advertising strategies are defined for research purposes. First of all, the pre-app advertising strategy with integration of a skip option. Secondly, the pre-app advertising strategy with integration of a skip option as well as an increased ad size. Since customers generally don’t like ads, an increased ad size is expected to decrease its attitude for this research (McCoy et al., 2007). Therefore, it is decided to increase the ad size slightly (from 10% to 20% screen area coverage) for the pre-app advertising strategy with the inclusion of (A) skip option and (B) increased ad size. In this way, the provided ad is expected to further increase its recall and recognition while not reducing its attitude.
2.3.3 Research variables
Based on the above stated arguments the following hypotheses concerning the advertising recall and recognition are defined:
H2a: The pre-app advertising strategy with (A) skip option and (B) increased ad size would result in
a higher ad recall compared to the other pre-app advertising strategies
H2b: The pre-app advertising strategy with (A) skip option and (B) increased ad size would result in
With respect to advertising attitude, the following hypotheses are defined:
2.3.4 Additional research
As aforementioned, it is of strong added value to enhance understandings concerning the attitude towards the pre-app advertising strategy as well. Herewith, the following hypothesis is proposed.
The aforementioned variables and hypotheses are combined in a theoretical framework which is depicted in the figure below. The research method for testing the defined hypotheses is further elaborated in the following section.
H2d: The pre-app advertising strategy with (A) skip option and (B) increased ad size would result in
either the same or a more positive attitude towards the advertising brand (Ab) compared to the
other pre-app advertising strategies
H2c: The pre-app advertising strategy with (A) skip option and (B) increased ad size would result in
either the same or a more positive attitude towards the ad (Aad) compared to the other pre-app
advertising strategies
H2e: The preference for the pre-app advertising strategy with (A) skip option and (B) increased ad
size is either the same or higher compared to the other pre-app advertising strategies
Fig 2-4: Theoretical framework in which the advertising scheduling strategy is expected to influence recall, recognition and attitude for mobile application advertising
Advertising scheduling strategy (H1e, H2e) Recall (H1a, H2a) Recognition (H1b, H2b) Attitude (H1c, H1d, H2c, H2d)
3. Data and method
For this research, 3.200 Dutch employees working for a multinational engineering and consultancy firm have been motivated for research participation purposes. The research consisted of an experiment in which the extremely popular and widely used WordFeud app experience was simulated. The subsequent survey was used to measure the differential impact of the defined advertising strategies on recall, recognition and attitude. To test the aforementioned hypotheses, 4 experimental conditions were introduced. In this chapter the research design including data collection, analysis methods and justification of methods is outlined. Herewith, it is decided to start with the defined research strategy. Subsequently, the sample and procedure and measurement instruments are described more specifically.
3.1 Research strategy
Research can be conducted in different experimental settings. The development of a new mobile application which includes the required research conditions and subsequent surveys can be regarded as a preferable experimental condition. By doing so, a realistic app experience on a smartphone can be introduced. Besides, the real behavior of participants can be investigated without bias from conscious thought of experimental participation. However, this option would lead to strong time constraints. Not only due to its required development time but also because of its time required to make it available from the App Store for downloading purposes. Another option incorporates the use of an existing mobile application for research purposes. However, this method includes some major limitations. First of all, the advertisements provided in existing mobile applications change over time and cannot be controlled effectively (Nath et al., 2013). Second of all, there were no apps found within the App store which includes the option for selecting either pre-app or in-app advertising. Finally, it is difficult to prevent time delay between conducting the experiment and subsequent survey examining. Therefore, it was decided to simulate an app experience in another online environment. Since a mobile application is nothing more than a tool which enables easy and fast access to linked online content, it is a viable option to provide the content outside a mobile application (Gupta, 2013). By doing so, the required research conditions and subsequent surveys can be incorporated very effectively and within the available research time span.
Based on the above statements, an online research tool (Qualtrics, 2014) was used to develop the experiment as well as the subsequent survey. Qualtrics is an industry-leading provider of online survey content. To illustrate, in the year 2012, over 1 billion surveys were conducted using the online Qualtrics survey tool (Qualtrics, 2014). Besides surveys, Qualtrics enables the inclusion of videos, presentations and so on.
3.2 Sample and procedure
To ensure proper testing of recall, recognition and attitude, it was decided to execute the research in two consecutive steps and divide the participants in 4 groups. Each group was linked to a dedicated experimental condition. After completion, a survey was introduced in which the impact of the specific experimental condition on recall, recognition as well as attitude was examined. The difference in outcome between the 4 defined groups was used for hypotheses testing purposes. Logically, this is a much stronger research method compared to the integration of one single group allocated to multiple experimental conditions in consecutive order. This since the latter option would include strong spillover bias. To illustrate, the know-how obtained from the first experimental condition would influence the respondent’s ability to recall and recognize brands in subsequent experimental conditions. Besides, the attitude towards the ad (Aad) as well as the brand (Ab) can be
influenced due to spillover effects. Concerning the measurement of advertising strategy preference, this spillover bias was expected to be neglectable. Therefore, this variable was measured by asking each respondent their preferred advertising strategy. Prior to research, 50 relatives and direct colleagues conducted the pretest. In this way the viability of the proposed experiment, survey and measurement instruments was verified. The outcome of this pretest was used for research improvement purposes. A schematic overview of the applied research strategy is clarified in the table below.
Table 3-1: Schematic overview of the research strategy for testing the impact of advertising strategy on recall, recognition and attitude
Group A Group B Group C Group D
Condition In-App Advertising
Standard Pre-App Advertising Standard Pre-App Advertising skip option Pre-App Advertising skip option increased ad size Experiment Survey - Recall - Recognition - Attitude - Recall - Recognition - Attitude - Recall - Recognition - Attitude - Recall - Recognition - Attitude Advertising + Content Content Advertising Content Advertising skip option Content Advertising skip option increased ad size
3.2.1 Subjects and Invitation
Concerning the subjects, it was decided to target employees working for a multinational engineering and consultancy firm. This firm is headquartered in the Netherlands and encompasses over 7.000 employees and 150 offices within 35 countries. For conformity purposes, it was decided to focus on the 3.600 Dutch employees working within the Netherlands. Besides, it was decided to exclude 400 related employees from participation. This since the pretest among relatives and friends demonstrated that the existence of personal relationships biases behaviour in real life. As a consequence, the target group consisted of 3.200 employees listed within the email database of the targeted company. At first, the target group was randomly divided in 4 equal groups (A, B, C and D). Subsequently, each group was asked for participation by email. The email consisted of an invitation letter accompanied with a link towards the dedicated online experiment (see Table 3-1). At the end of the experiment, respondents were directed to the subsequent online survey. The invitation letter for all groups was exactly the same and included the supportive statement of the company as well as the University of Amsterdam. Herewith, the strength of the persons’ commitment and involvement with an organization and its willingness to put in additional effort and respond to its invitation is used effectively (Mowday, Steers, & Porter, 1979). The detailed research questions, testing conditions and variables of interest were not made salient to the respondents. This to prevent potential bias on the outcome of this study. The invitation letter was personalized by incorporating the employee’s personal name. In addition, the time span of the experiment including subsequent survey was mentioned and maximized to 5 minutes. After all the data was acquired, a subsequent email was sent to all employees to clarify the full aim of this study.
Based on the 3.200 employees invited for participation purposes, it was found that 244 invitations could not be delivered due to out of office replies and expired email addresses. The 2.956 successful email invitations resulted in 871 respondents. This corresponds to a response rate of 29.5%. The outline of the invitation letter is depicted in Appendix 1.
3.2.2 Experiment
For this research it was decided to incorporate the WordFeud app experience. Game applications are among the most frequent used applications in smartphones (Flurry, 2013; Infographic, 2013). In addition, WordFeud can be regarded as one of the most popular game apps available (Böhmer et al., 2011). To illustrate the app has been downloaded over 15 million times and is still growing in popularity (Norway Post, 2012). Besides, WordFeud can be regarded as a timeless game and used by people of different ages. Since a higher audience size and user rate generally coincides with a higher effectiveness of mobile application advertising, the practical value of incorporating WordFeud was significantly enhanced.
To make the WordFeud experience as realistic as possible, a smartphone experience simulation on the computer screen was introduced. During the start, a real smartphone image including its app visualizations popped up. Subsequently, the respondents were asked to press the WordFeud App button on the screen to start the game. After the WordFeud app was loaded, the respondents were asked to find a word. In line with the real WordFeud game, respondents were enabled to take as much time as required to fulfill the job. Concerning the advertisement, it was decided to integrate a copy of a realistic in-app advertisement found within the App store with integration of the famous retail brand Adidas. More specifically, the advertisement consisted of the Adidas brand sign accompanied with the text “20% off, check the app now!”. After the respondents clicked on the WordFeud app button, a loading phase of 5 seconds was introduced. This is in line with the average loading time of mobile application (Compuware, 2013). Subsequently, the WordFeud game was provided.
For the in-app advertising experimental condition (Group A), one advertisement was provided at the top of the page solely during the in-app phase. This ad encompassed around 10% screen area coverage. A strategy used by many companies for in-app advertising in mobile applications (Gupta, 2013). For the pre-app advertising conditions (Group B, C and D), the same ad was provided solely during the loading phase of the application. Experimental condition 3 (Group C) was accompanied with a skip option. Herewith, a skip option appeared after 5 seconds of loading time. For continuing purposes, it was required to click the skip button. Experimental condition 4 (Group D) was a further extension of condition 3 in which the advertisement size was enlarged from 10% to 20% screen area coverage. An overview of all consecutive screen views for each condition is depicted in Appendix 2.
3.2.3 Survey
Following up the WordFeud app experience, the participants completed the online survey (Appendix 3). This survey was aimed to effectively examine the difference between pre-app advertising and in-app advertising on advertising recall, recognition and attitude. In addition, it was researched which pre-app advertising strategy works best: (1) standard pre-app advertising, (2) pre-app advertising including (A) skip option or (3) pre-app advertising including (A) skip option and (B) increased ad size.
Recall
The aim of investigating recall is to “capture ‘top-of mind’ accessibility of brand in memory measured
by a correct identification of a brand by listing a product category” (Keller, 1993). This definition is in
line with the so called ‘aided recall’ as extensively used in literature (Baack, Wilson, & Till, 2008; Zinkhan, Locander, & Leigh, 1986). Aided recall can be measured by questioning the brand names remembered during the experiment while given multiple related product categories as a cue. Another option is to provide the censored version of the advertisements in which the brand names
are blocked out. Subsequently, it is asked to write down the corresponding brand names (Dreze & Hussherr, 2003; Dubow, 1995). This in contrary to the measurement of unaided recall in which it is asked measured to write down the brand names remembered during the experiment without any further support (Baack, Wilson, & Till, 2008). Since both aided and unaided recall is measured frequently in literature, it was decided to integrate one aided question as well as one unaided question as being used by Hershberger (2009) and Dreze (2003). Herewith, the unaided question was introduced prior to the aided question. This to ensure no bias due to information spillover from the aided question to the subsequent unaided question.
Recognition
With recognition, it is aimed to “capture potential retrievability or availability of brand in memory
measured by the correct confirmation of a prior exposure to a brand” (Keller, 1993). In literature
different methods are used to measure recognition. They all come back to the provision of multiple brands whereof some have been shown before and some not. Subsequently, the participants have to indicate whether or not a particular brand has been seen before or not (Baack, Wilson, & Till, 2008; Shapiro & Krishnan, 2001.; Dubow, 1995). The condition under which recognition is questioned can differ slightly. Some research studies integrate unaided recognition measures, like the forced choice method (Baack, Wilson, & Till, 2008; Shapiro & Krishnan, 2001). Herewith, it is questioned for multiple brands whether or not they have been seen before or not (Baack, Wilson, & Till, 2008; Shapiro & Krishnan, 2001). Other researchers used the so called aided recognition measures (Dreze & Hussherr, 2003; Dubow, 1995). This can be obtained by providing pictures of the advertisement in advance, in which the brand names are blocked out (Dreze & Hussherr, 2003; Dubow, 1995). Subsequently, it is asked if a particular brand has been seen before or not (Baack, Wilson, & Till, 2008; Shapiro & Krishnan, 2001). Since the integrated aided recall question already includes the censored version of the ad as support (Dreze & Hussherr, 2003), a subsequent unaided recognition question will be biased due to the spillover information from the aided recall question in advance. Therefore, it was decided to integrate the aided recognition measure in this research.
Attitude
As aforementioned, the attitude towards the ad (Aad) as well as the brand attitude (Ab) will be
investigated. In literature, different scales have been used to measure both attitude levels. For example, some researchers used 2-point scales to measure both brand attitude and attitude towards the advertisement (Hanson & Biehal, 1995; Homer, 1990; MacKenzie & Lutz, 1989) while others used 5-point scales (Mitchell & Olson, 1981). Also the amount and type of items for both brand attitude and attitude towards the advertisements differ among existing literature. Items frequently used include good-bad, like-dislike, pleasant-unpleasant, irritating-not irritating, favorable-unfavorable,
good quality-poor quality (MacKenzie & Lutz, 1989; Hanson & Biehal, 1995; Mitchell & Olson, 1981). For this study, it was decided to opt for the 7-point Likert scale. First of all, because this scale is extensively used in literature. Secondly, the higher the point scale the more flexibility is included for answering purposes. The items used for measuring both variables were derived from recognized researches by Mitchell (1981) and MacKenzie (1989). To optimize the data analysis process significantly, it was decided to integrate addition questions for clarification purposes. In other words, all respondents were asked to clarify their answers.
4. Results and discussion
Within this chapter, the results obtained from analyzing the data set of 871 respondents are presented and discussed. In addition, the conveyed hypotheses are tested. Concerning the analyses, multiple chi-square tests, t-tests and MANOVA tests were conducted whereof a different statistical test method was used for verification purposes in case the obtained significance was close to relevant (P < .10). Herewith, it was found that the variance in results between the different test methods did not influence the significance of the outcome. It is decided to present the findings in two consecutive steps. The first part is aimed at analyzing the difference between pre-app advertising and in-app advertising. In the second part, the findings concerning the question: “Which type of pre-app advertising strategy works best?” are presented and discussed.
4.1 Pre-app advertising vs In-app advertising
4.1.1 Results and hypotheses testing
The differential impact between pre-app advertising and in-app advertising was investigated by analyzing the differential outcomes for Group A and Group B (see Table 1). For Group A, in total 233 responses were collected of which 194 participants completed the test successfully. This resulted in an effective test completion rate of 85%. For Group B, the completion rate amounted 78% (based on 183 successive responses out of 239 participants). Reasons for not conducting the test successfully can be diverse. Lengthy questionnaires are well known to adversely affect response rates (Edwards et al., 2002). The obtained email replies and results of the survey questions revealed no adverse effect for this particular test. To illustrate, over 95% of the respondents labeled the test duration as (more than) satisfactory (Appendix 3-13). Most probably, because this test lasted just 5-10 minutes. In contrary, the required smartphone experience was mentioned as less satisfactory. This since respondents with limited smartphone experience found difficulties completing the test. As a consequence, the test completion rate for Group A and Group B amounted 85% and 78%, respectively.
Sample
The difference in respondent distribution concerning gender and age for both groups was not significant. For both groups around 75% consisted of male respondents while 25% included female respondents. This so called gender inequality is quite normal for the company of interest. The mean (M) and standard deviation (SD) concerning the age of the respondents amounted M=40.1; SD =10.3. The youngest and oldest respondent was aged 17 and 66 years, respectively (see Appendix 3-1).
Recall
Group A was exposed to the standard in-app advertising experimental condition (see Table 3-1). Within this condition, the respondents were provided with the advertisement solely during the in-app phase. In other words, during the time the participant was asked to lay down a WordFeud word (see Appendix 2-1c). In contrary, Group B was exposed to the standard pre-app advertising condition. Herewith, the advertisement was shown solely during the start-up phase of the application (see Appendix 3-1c). This period was fixed and lasted 5 seconds. In contrary, the advertisement provision period for Group A was variable and decided by the participants time frame aimed at finding a word for participation purposes. It was found that this particular period lasted on average 1 minute 54 seconds.
The calculated recall rate was based on the percentage of participants who were able to correctly recall the provided advertising brand. For Group A, 21 out of 194 participants were able to recall the brand Adidas successfully whereas 2 participants were able to recall the brand solely with provision of the censored version of the advertisement. This coincided with a total recall rate of 11.9% whereof the unaided and aided recall fraction amounted 10.9 % and 1.0%, respectively. For Group B, 7 out of 183 participants recalled the brand successfully. Herewith, 2 participants were able to retrieve the brand’s name solely by providing the ad’s censored version. Consequently, the total recall rate for Group B equaled 3.8% of which 2.7% was based on the unaided recall fraction. An overview of the difference in recall rate for Group A and B is depicted in Figure 4-1.
In-App Standard Pre-App Standard 0 5 10 15 20 25 0 R e ca ll ( % )
The following hypothesis was proposed concerning the difference in recall between both advertising strategies:
To investigate this hypothesis a chi-square test was conducted. The outcome of this test (χ2 (1, N = 377) = 8.29, p < .01, φ = .15) revealed that the recall rate of Group A (11.9%) was significant higher compared to Group B (3.8%). In other words, the findings were exactly opposite to prior predictions. As a consequence, it can be stated that H1a is not supported under the applied experimental conditions.
Recognition
Recognition is measured based on the ability to confirm a prior exposure to a brand (Keller, 1993). For this research it was decided to measure this variable by providing the respondents a list of 4 brands (Adidas, Nike, Reebok and Asics). Subsequently, it was asked if they could select the brand provided within the advertisement (see Appendix 3-4a). For Group A, in total 12 respondents out of 194 participants were able to recognize the Adidas brand while not recall it. Logically, the participants who were able to recall the brand were able to recognize it as well. Consequently, 35 successive recognition values were registered. This coincides with a total recognition rate of 18.0%. For Group B, an additional 6 respondents were able to solely recognize the brand. Therefore, the total recognition rate for this group equaled 7.1%.
In-App Standard Pre-App Standard 0 5 10 15 20 25 0 R e co g n it io n ( % )
Figure 4-2: Recognition as function of the total participants
H1a: The pre-app advertising strategy would result in a higher ad recall compared to the in-app
The following hypothesis was proposed concerning the difference in recognition between both advertising strategies:
In line with the recall test, a chi-square test was conducted to test the proposed recognition hypothesis. In line with the recall findings as described above, the chi-square test (χ2 (1, N = 377) = 10.14, p < .01, φ = .16) concluded that the recognition rate of Group A (18.0%) was significant higher compared to Group B (7.1%). These findings are in line with the expectations in which the recall rate was expected to be positively correlated with the recognition rate. Since H1a did not hold under the applied condition, the conclusions concerning H1b were similar. Therefore, H1b is not supported.
Attitude
The attitude towards the ad (Aad) was measured by the mean of eight 7-point evaluation scales (not
curious–curious, not interesting-interesting, not pleasant-pleasant, annoying-not annoying, no added value-added value, dislike-like, not nice-nice, bad-good). The attitude towards the brand (Ab) was
measured by the mean of eight 7-point evaluation scales (bad quality–good quality, not innovative-innovative, not loving-loving, not stylish-stylish, no added value-added value, dislike-like, bad-good, not interesting-interesting). A reliability analyses was conducted to ensure that all evaluation scales contributed to the same attitude construct. Based on the obtained Cronbach alpha values for Group A (Aad, α = .91 and Ab, α = .94) and Group B (Aad, α = .90 and Ab, α = .93) together with α -values’ if
item deleted’, no adjustments in measurement scales were required. An overview of the mean, standard deviation and scale reliability coefficients for both attitude constructs are outlined in the table below.
The following hypothesis was proposed concerning the difference in ad attitude between both advertising strategies:
Ad attitude Brand attitude
M(x̄) SD (σ) (α) M(x̄) SD (σ) (α)
Group A 2.36 .50 .91 4.60 .31 .94
Group B 2.55 .65 .90 4.57 .34 .93
Table 4-1: Means, standard deviations and scale reliability coefficients for both advertising constructs (Aad and Ab)
H1b: The pre-app advertising strategy would result in a higher ad recognition compared to the
in-app advertising strategy
H1c: The pre-app advertising strategy would result in a more positive attitude towards the ad (Aad)
To test the difference between both groups for ad attitude (incorporating 8 subscales), a one-sided multivariate analysis of variances (MANOVA) was conducted. Herewith, it was found that the difference was marginally significant (F(1,8, N = 377) = 1.84, p = .07, ΛPillai = .04). In other words, the conveyed hypothesis can neither be fully accepted nor rejected. Conclusively, H1c is partially
supported.
The following hypothesis was proposed concerning the difference in brand attitude between both advertising strategies:
Herewith, the same test was conducted in comparison with the investigation of the ad attitude. These results revealed no significance in brand attitude (F(1,8, N = 377) = .39 , p = .93, ΛPillai = .01). As a consequence, H1d is not supported under the applied experimental conditions.
Preference
The introduced attitude towards the advertising strategy is a derivative from the frequently measured variable named attitude towards advertising in general (MacKenzie & Lutz, 1989; Durvasula et al., 1990). This was measured by the participants’ preference for either in-app or pre-app advertising (Lavidge & Steiner, 1961). Herewith, the participants were provided with the schematic representations of both advertising strategies including textual functionality descriptions (see Appendix 3-7). Subsequently, they were asked to select their preferred advertising strategy, either in-app advertising (χ = 1), pre-app advertising (χ = 2). In total 377 participants (Ʃ Group A, B) faced this particular question. It was found that 39 respondents preferred the in-app advertising strategy (10%) while 338 respondents preferred the pre-app advertising strategy (90%). The results are depicted in the figure below.
H1d: The pre-app advertising strategy would result in a more positive attitude towards the
The following hypothesis was proposed concerning the difference in preference between in-app and pre-app advertising:
A one-sample t-test was used to investigate the preference hypothesis. Herewith, it was found that the overall difference in preference rate between both advertising strategies yielded a significant difference (t(N = 377) = 25.25, p < .005, df = 376). In other words, the attitude towards the pre-app advertising strategy was significant higher compared to the in-app advertising strategy. Therefore, H1e is supported.
4.1.2 Discussion
The first part of the experiment was aimed to investigate the difference between the standard in-app advertising strategy and the standard pre-app advertising strategy. Herewith, the conducted chi-square test (χ2 (1, N = 377) = 8.29, p < .01, φ = .15) revealed a higher recall rate for the in-app advertising condition compared to the pre-app advertising condition. In line with these findings, a higher recognition value for the in-app advertising condition was found as well (χ2 (1, N = 377) = 10.14, p < .01, φ = .16). The accompanied significance was contradictory to the defined hypotheses (H1a, H1b) in which the disconnection between advertising provision and task orientation was supposed to enhance advertising recall and recognition (Hershberger, & Costea, 2009). Additional
H1e: The pre-app advertising strategy is more preferable compared to the in-app advertising
strategy In-App Standard Pre-App Standard 0 20 40 60 80 100 0 P re fe re n ce ( % )
analytical statistics have been conducted to clarify its cause. Herewith, it was found that the difference in advertising duration time most probably biased the outcome. To illustrate, the respondents for the in-app advertising condition were exposed to the advertisement for about 1 minute 54 seconds while for the pre-app advertising condition this time lasted just 5 seconds. The factor 20 longer advertising provision period was due to the additional time respondents used to fulfill the job for finding a word. The consequential impact on memory seems verified since a longer ad exposure time is well known to enhance recall as well as recognition (Goldstein, McAfee, & Suri, 2011; Danaher & Mullarkey, 2003). One of the strong added values concerning the experimental set-up was related to its focus on optimizing standardization. As a consequence, the impact of other potential biasing factors on the obtained findings can be labeled as neglectable. This since the difference in exposure duration time was the only legitimate difference besides the advertising scheduling between both advertising strategies. To ensure no bias from sample differences, additional statistical analyses have been conducted. Herewith, no significant difference in gender, age, and mobile game experiences between both groups was found.
Since it was decided to integrate additional questions to enhance sample size understandings, multiple addition insights are gained. First of all, it was found that about 75% of the recall and recognition values where obtained from respondents who experienced a game app before. Most probably, this was due to the level of awareness for mobile app advertising. To illustrate, since only 10% of the screen area coverage was used for advertising purposes, the pre-experience with mobile advertising can help in notifying the provided advertisement more naturally. As a consequence, the subsequent recall and recognition is likely to be increased. Besides, the game experience was the highest for the younger respondents. This clarifies the result that the average age of the respondents who recalled or recognized the brand (M=33.1) was significant lower compared to the average age of the total sample size (M=40.1). Finally, the impact of gender on recall and recognition was investigated. Herewith, the results yielded no significant difference. To illustrate, between 75-80% of the recall and recognition values were obtained from male respondents accounting for 75-80% of the total population.
Besides the important findings concerning recall and recognition, this part of the experiment has gained very important insights concerning mobile advertising attitude. To illustrate, it was found that over 90% of the respondents preferred the novel pre-app advertising strategy over the ‘state of the art’ in-app advertising strategy. The reason behind this result was easy to trace due to the integrated clarification question “What is your preferred advertising strategy and can you please clarify your decision?” (Appendix 3-7). Based on the obtained data, it was found that the extreme preference for the pre-app advertising strategy was mainly based on its lower advertising provision period as well as its lower adverse effect on the user experience. Based on the obtained data concerning the question
“What is your opinion on the ad and can you please clarify your opinion?”,it can be concluded that the difference in ad attitude between both strategies was due to the same differential values concerning advertising provision period and user experience (Appendix 3-5). The reason why its obtained significance was not as high compared to the differential preference concerning the advertising strategy was most probably due to the integrated questioning method. This since the latter difference was measured by the variation in obtained results between Group A and Group B. In contrary, the difference in attitude towards the advertising strategy was measured by asking both respondent groups the exact same question in which they were asked to select their preferred advertising strategy. The data analyses of the question “Can you please clarify your opinion about the brand?” was used to discuss the results concerning the difference in brand attitude (Appendix 3-6). Herewith, no significant difference in Ab between both groups was found which is contrary to
evidence (MacKenzie, Lutz, & Belch, 1986; Gardner, 1985; Mitchell & Olson, 1981; Shimp, 1981). The data analyses revealed the main foundations for this observation. Herewith, it was found that over 95% of the registered respondents did not refer to the introduced ad concerning their brand attitude clarification. In other words, the selected brand attitude values were mainly based on the respondents’ pre-experiment experience in daily life. In this way, the yielded differential impact concerning the ad attitude for this experiment was not able to spillover to the brand attitude variable. The latter fuels evidence concerning the neglectable difference found in brand attitude.
4.2 Which type of pre-app advertising strategy works best?
As aforementioned, the pre-app advertising strategy includes unique possibilities concerning increasing its advertising recall and recognition as well as its attitude. To investigate which pre-app advertising works best, an additional test was executed. Herewith, two additional experimental conditions were investigated and compared with the outcomes of the prior test (in-app advertising vs pre-app advertising). Group C experienced the pre-app advertising condition including a skip option and Group D was exposed to the pre-app advertising condition including a skip option and accompanied with an increased ad size (see Table 3-1).
4.2.1 Results and hypotheses testing
For Group C, in total 151 successive responses were collected out of 197 participants. This corresponds to a test completion rate of 77%. Group D yielded an effective completion rate of 90% (181 completed tests out of 202 participants). In comparison, the test completion rate for Group A and B amounted, 78% and 85%, respectively (see §4.1.1). The sample distribution concerning gender and age for Group C and D was similar compared to Group A and Group B. To illustrate, the average age amounted 39 years whereof 71% of the sample consisted of male respondents. The youngest and oldest respondent equaled 18 and 62 years of age, respectively.
Recall
In line with the standard pre-app advertising condition (Group B), the other two pre-app advertising strategies (Group C and D) were provided with the advertisement solely during the start-up phase. For Group C, a skip option button popped up underneath the advertisement at the tail end of the start-up phase (after 4-5 seconds). Subsequently, the participants were allowed to decide by themselves to continue watching the ad or click the skip option for continuing purposes. In latter case, the next screen was provided in which the participants were asked to lay down the WordFeud word. Group D was exposed to exactly the same situation as group C. However, the ad size of Group D was doubled compared to all other experimental conditions. In other words, the ad size was increased from 10% to 20% screen area coverage.
In total 12 out of 151 participants from Group C recalled the brand Adidas successfully without any further support. No additional successful recall responses were registered based on the provision of the censored ad version. Consequently, the total recall rate amounted 7.9 % whereof the unaided fraction equaled 100%. For Group D, 27 out of 181 participants recalled the brand successfully. Herewith, 3 participants were able to retrieve the brand’s name solely by providing the ad’s censored version. Conclusively, the total recall rate equaled 16.6% of which 1.7% was based on the unaided recall fraction. An overview of the difference in recall rate for all experimental conditions is depicted in Figure 4-4. In-App standard Pre-App standard Pre-App skip option Pre-App skip option increased ad size 0 5 10 15 20 25 0 R e ca ll ( % )
The following hypothesis was proposed concerning the difference in recall between the defined pre-app advertising strategies:
The conducted chi-square test yielded a significant difference among the 3 pre-app advertising conditions (χ2 (2, N = 515) = 14.02, p < .01, φ = .17). More specifically, it was found that the recall rate of Group D was significant better compared to Group C (χ2 (1, N = 332) = 3.86, p < .05, φ = .17) as well as Group B (χ2 (1, N = 364) = 13.22, p < .01, φ = .19). Therefore, H2a is supported.
Additional chi-square tests were conducted to investigate the difference between other groups. The sole introduction of a skip option (Group C vs Group B) was marginally significant (χ2 (1, N = 334) = 3.86, p = .11, φ = .17). Moreover, the difference in recall rate between the most effective pre-app advertising condition (Group D) and the standard in-app advertising condition (Group A) was investigated. Herewith, it was found that the pre-app advertising condition yielded a higher recall rate (16.6%) compared to the in-app advertising condition (11.9%). Based on the statistics, this difference was not significant (χ2 (1, N = 375) = .76, p = .38, φ = .05).
Recognition
For Group C, an additional 4 out of 151 participants were able to recognize the brand while not able to recall it. Combined with the 12 respondents who recalled the brand, a total recognition rate of 16.6% was registered. For Group D, an additional 13 respondents were able to recognize the brand while not recall it. The 27 respondents who were able to recall the brand (and therefore recognize it as well) combined with the 13 respondents who solely recognized the brand equaled a total recognition rate of 23.8%. In the figure below, the overview of all yielded recognition rates is provided.
H2a: The pre-app advertising strategy with (A) skip option and (B) increased ad size would result
