• No results found

The Cross-Channel Effect of Mobile Advertising

N/A
N/A
Protected

Academic year: 2021

Share "The Cross-Channel Effect of Mobile Advertising"

Copied!
66
0
0

Bezig met laden.... (Bekijk nu de volledige tekst)

Hele tekst

(1)

The Cross-Channel Effect of

Mobile Advertising

Does the Content Design impact

Customers’ Online and Offline Store Visit Intentions?

(2)

The Cross-Channel Effect of Mobile Advertising

Does the Content Design impact Customers’ Online and Offline Store Visit Intentions?

Master Thesis

Faculty of Economics and Business MSc Marketing Management

22nd June 2015

(3)

Management Summary

In today’s retail environment, the customer can choose between various options to conduct purchases. Online shops, as well as traditional brick and mortar stores, are commonly visited channels. Alongside this increasing multichannel shopping, marketers are able to advertise across the different channels. Consequently, managers are confronted with the challenge to allocate their marketing budgets across channels, design the advertisement congruently and drive consumers to the intended channel. While existing research started investigating the impact of traditional and online media on channel usage, this study focuses on one of the newest advertising media, namely mobile devices. In particular, the study examines the effect of the content design (brand or promotional content) of mobile advertising on store visit intentions across channels (online and offline stores), and whether this relation is influenced by the type of mobile advertising (SMS or MMS).

The study was conducted by setting up an online experiment. Data was collected from a 2x2x2 between-subjects full-factorial design and additionally a baseline condition measuring the overall advertising effectiveness was included. To test the hypotheses depicting the main and interaction effects, multiple regression analysis was performed.

The main effect of mobile advertising in form of SMS and MMS on the intention to visit either an online shop or a traditional retail store could not be revealed in this research. Nevertheless, mobile advertising with relevant message content is able to influence store visit intentions. In fact, advertising a familiar and known brand name affects the intention to visit a brick and mortar store negatively; and therefore, mobile advertising should include unknown brands to drive customers to the offline store. In contrast, the combination of a MMS and a known brand leads to positive offline store visit intentions. Moreover, broadcasting promotional content in a MMS enhances higher store visit intentions for traditional stores. Therefore, managers, who want to achieve a cross-channel effect of mobile advertising, should advertise via MMS which include either promotions or a known brand.

(4)

Preface

“One’s destination is never a place, but a new way of seeing things.” – Henry Miller Taking a look at this quote by Henry Miller, it describes my experiences during my studies at the University of Groningen perfectly. When I came to Groningen to start my master studies, Groningen has been a place for me. I decided to move to a city which seemed nice to live in and to study at a university with a very good reputation and an interesting master program which suits my preference for a future career. Now, having almost completed my studies, I can definitely say, that it was not choosing just a place. Living in another country helps you to reflect on your own thoughts and offers you opportunities to extend your knowledge. Studying in a foreign country with a different educational system and different expectations from the teaching staff, opened new perspectives and deepened my knowledge which I gathered during my bachelor studies in Germany. Especially, writing the master thesis showed me how to work independently and how to overcome drawbacks. This process would not have been that instructive without the valuable input of my supervisors Peter C. Verhoef and Evert de Haan. To overcome the drawbacks, I was very thankful to have my family, who gave me the chance to study abroad, my boyfriend, who motivated me when I felt like giving up, and my friends and flatmate, who were a welcome change from writing the thesis.

(5)

Table of Content

Management Summary ... I Preface ... II Table of Content ... III

1 Introduction ... 1

2 Literature Review and Hypotheses ... 4

2.1 Mobile Marketing in the Multichannel Environment ... 4

2.2 Mobile Marketing ... 6

2.2.1 Mobile Advertising – A Tool of Mobile Marketing ... 6

2.2.2 Types of Mobile Advertising ... 8

2.2.3 SMS and MMS Mobile Advertising ... 8

2.3 The Impact of Mobile Advertising’s Content Design on Store Visit Intentions ... 10

2.3.1 Brand Name ... 11

2.3.2 Promotion ... 12

2.3.3 Text-based or Image- & Text-based ... 13

2.4 Conceptual Model ... 16

3 Methodology ... 17

3.1 Research Design and Data Collection ... 17

3.2 Procedure ... 18

3.3 Measurement and Manipulation ... 19

3.3.1 Independent Variables – Brand Name and Promotion ... 19

3.3.2 Dependent Variables – Offline and Online Store Visit Intentions ... 20

3.3.3 Moderating Variable – Type of Mobile Advertising ... 21

3.3.4 Control Variables ... 21

3.4 Plan of Analysis and Model Specification ... 22

3.4.1 Plan of Analysis ... 22

3.4.2 Model specification ... 23

4 Results ... 24

4.1 Data ... 24

4.1.1 Data Editing and Cleaning ... 24

4.1.2 Sample Description ... 25

4.1.3 Random Assignment ... 27

4.2 Preparation for the Analysis ... 28

4.2.1 Reliability Analysis ... 28

4.2.2 Manipulation Check ... 30

4.2.3 Normality Test ... 30

4.2.4 Homogeneity of Variance Test ... 31

4.2.5 Multicollinearity Test ... 32 4.3 Hypotheses Testing ... 32 4.3.1 Main Effects ... 32 4.3.2 Moderating Effects ... 37 5 Discussion ... 41 5.1 Conclusion ... 42 5.2 Managerial Implications ... 44

5.3 Limitations and Further Research ... 45

References ... 47

(6)

1

Introduction

Consumers in today’s retail environment are presented with various options to purchase services and goods (Fulgoni, 2014). Traditional brick and mortar stores and online stores are the main channels addressed by consumers. Besides these options, an emerging possibility to conduct shopping is to visit mobile stores (Rigby, 2011). The availability of different channels enhances the transformation of the customer journey (Fulgoni, 2014) and aggravates the consumer’s decision making process to choose the final real or virtual purchase location (Kollmann, Kuckertz and Kayser, 2012). As a result, managers face the challenge to understand consumers’ intentions to visit either an online or an offline environment on their path-to-purchase in order to make strategic decisions, allocate their marketing budgets and drive consumers to the intended channel. While research on channel, consumer and product category characteristics, as drivers of channel choice, have been studied extensively (Chocarro, Cortiñas and Villanueva, 2013; Neslin et al., 2006), research on the impact of advertising as a potential choice driver across the different channels is in its early stages (Dinner, van Heerde and Neslin, 2014).

(7)

promotional activities be designed to drive customers to specific channels, or should they be channel neutral?”.

The present study will extend the research on the cross-channel effect of advertising, which started to investigate the impact of traditional and online media (Ansari, Mela and Neslin, 2008), by focusing on one of the newest advertising media, namely mobile devices. Especially, in the context of multichannel shopping, mobile devices become highly relevant as they offer the opportunity to communicate and interact with the consumer independently from time and place (Shankar and Balasubramanian, 2009). Therefore, mobile devices reveal the ability for companies to accompany consumers throughout the day and during their path-to-purchase (ibid). This new form of communication is called mobile marketing. Retailers are empowered to broadcast real-time and location-specific, personalized information on mobile devices promoting goods and services through different types of mobile marketing (Park, Shenoy and Salvendy, 2008).

Different studies investigated the effectiveness of various types of mobile marketing like Short Messaging Services (SMS), mobile banner advertisement and mobile promotions (Bakar and Bidin, 2013; Hanley and Becker, 2008; Scharl, Dickinger, and Murphy, 2005) on purchase intentions; however, there is still room to further strengthen and deepen this research stream (Bart, Stephen, and Sarvary, 2014; Vatanparast and Butt, 2010); especially, in the multichannel environment to reveal cross-channel effects of advertising (Dinner, van Heerde and Neslin, 2014).

(8)

The following research question has been developed and will build the basis of this research: What impact does mobile advertising have on store visit intentions across channels (offline and online stores)? Does the content of the message (brand name and promotion) influence the intention to visit one particular purchase channel? And does the type of the medium (SMS or MMS) further strengthen customers’ intentions?

In order to understand these relationships in more detail, this research is divided into three main variables that may have effects on each other: The message content of mobile advertising (brand name and promotion), the type of mobile advertising in form of SMS and MMS, and the intention to either visit a store online or offline. The following three effects will be researched:

1. The differential effect of exposure to branded message content of mobile advertisement, varying between a familiar and an unfamiliar brand name, on the intention to visit an offline or online store.

2. The differential effect of exposure to promotional message content of mobile advertisement, varying between a promotion and no promotion, on the intention to visit an offline or online store.

3. The differential effect of exposure to different types of mobile advertising, varying in text- or text and image-based design, on the relationship of the message content and the intention to visit an offline or online store.

(9)

2

Literature Review and Hypotheses

This literature review will give a theoretical background of the concepts considered in this research and will substantiate the formulated hypotheses. First, mobile marketing in the multichannel environment will be presented. Then, the concept of mobile marketing and mobile advertising in particular will be discussed. Thereupon, the different types of message content, as well as two types of mobile advertising, will be explained. The relationship between differential mobile advertising message contents and store visit intentions in the multichannel environment will be drawn to develop the conceptual model and present the assumed moderating role of type of mobile advertising.

2.1 Mobile Marketing in the Multichannel Environment

(10)
(11)

2.2 Mobile Marketing

In order to understand the impact of mobile marketing on consumers’ store visit intentions for web shops and brick and mortar retailers, the related marketing environment of this communication channel will be studied. Marketers’ attention to mobile devices as an advertising tool is increasing, as the penetration of mobile devices has reached almost 100% in western countries (OECD, 2015). A mobile device is “a portable computing device such as a smartphone or tablet computer” (Oxford Dictionaries, 2015). The majority of the population is using a mobile device because it allows communicating, collecting information or provides entertainment independent of time and location (Laszlo, 2009). As consumers are always in a hurry, become busier and more difficult to reach and their mobiles represent “always-on-always-with-you” devices, direct communication opportunities become interesting from a manager’s point of view (Leek and Christodoulides, 2009). This opportunity is captured by the emerging marketing channel “mobile marketing”. Many studies already investigated mobile marketing, but still no common definition for the term exists (Varnali and Toker, 2010). Leppäniemi, Sinisalo and Karjaluoto (2006, p. 38) developed the following basic definition of mobile marketing: “the use of the mobile medium as a means of marketing communications”, which focuses on the function of communication. According to the Mobile Marketing Association (MMA) (MMA a, 2015), mobile marketing is defined as "a set of practices that enables organizations to communicate and engage with their audience in an interactive and relevant manner through and with any mobile device or network". This definition extends the function of communication by the opportunity for retailers to engage with their customers. Shankar and Balasubramanian (2009, p. 118) further specify the nature of engagement in mobile marketing as “the two-way or multi-way communication and promotion of an offer between a firm and its customers using a mobile medium, device or technology.” The two-way or multi-way communication underlines the opportunity for retailers to interact with their customers (Rigby, 2011). Different mobile marketing practices put emphasis on the interaction with the customer: mobile website creation, mobile emailing and messaging, mobile advertising, mobile couponing, mobile customer service and mobile social network management (Shankar et al., 2010).

2.2.1 Mobile Advertising – A Tool of Mobile Marketing

(12)
(13)

2.2.2 Types of Mobile Advertising

The MMA developed a categorization of mobile advertising tools and came up with five different categories of mobile advertising technologies: mobile messaging (SMS and MMS), mobile internet, downloads and other types of content and applications, search and location based technologies (MMA b, 2013). Within these categories the distinction between pull and push technologies is made. Push advertising sends relevant marketing messages to consumers without the explicit request of the message but with customers’ permissions to receive those messages (Shankar et al., 2010). In contrast, pull advertising contains information consumers accessed and requested independently on their mobile devices (Xu et al., 2010). Watson, McCarthy and Rowley (2013) found that the customers’ interests for pull advertising can be attained through trust building, leaving the customer their control and entertaining content. However, Quah and Lim (2002) mention the dominance in mobile advertising of the push model as it offers benefits like time and money saving compared to accessing information via the browser. Therefore, in the following the emphasis will be put on mobile advertising push technologies.

In the past years, research investigated the effect of mobile marketing on consumer behavior in general (Richard, 2013; Persaud and Azhar, 2012; Vatanparast and Butt, 2010) or studied one specific advertising tool like SMS, mobile display banners or Quick Response-codes (Bart, Stephen, and Sarvary, 2014; Drossos et al., 2014; Watson, McCarthy and Rowley, 2013). Regarding the comparison of different mobile marketing tools, research is still in its early stages (Gavilan, Avello and Abril, 2014; Xu, Oh and Teo, 2009). However, the distinction and the type specifics between various formats provide a valuable guideline for marketing managers (Xu, Oh and Teo, 2009). In the present study, two types of push mobile advertising, SMS and MMS, and the variation of their message content design will be observed to reveal whether they have a differential impact on consumers’ intentions to either visit an offline or online store. This insight will be helpful knowledge for managers to understand, if marketing tools have to be designed differently to attract customers for different purchase channels (Neslin et al., 2006).

2.2.3 SMS and MMS Mobile Advertising

(14)

investigate store visit intentions rather than purchase intentions in the offline and online channel, because customers’ intentions are often good predictors of whether they actually will perform the specified behavior in the future (Davis and Warshaw, 1991). For the review of the existing literature on the effect of SMS and MMS mobile advertising the deviating outcome variable can be neglected as purchase intentions in the specific channel are representatives of the store visit intentions (Fishbein and Azjen, 1975).

According to Drossos et al. (2013), SMS advertising has the highest use frequency and growth in prospects is still expected. Shankar and Balasubramanian (2009) found the following main advantages of text messages: the simplicity for managers and consumers, measurability and high response and conversion. SMS advertising as a form of push advertising predominantly includes promotions which are strategically designed to enhance sales in the short term; and therefore, inducing immediate purchases (Okazaki and Taylor, 2008). In 2014, Drossos et al. revealed that mobile text advertising influences purchase intentions and that the cognitive dimensions of product involvement and impulsiveness moderate this relationship. In 2012, the American multichannel retailer Macy’s has launched a SMS advertisement campaign to drive consumers to its e-commerce site (Mobile Commerce Daily a, 2015). Besides the case of Macy, this tactic of SMS advertising to consumers has been adopted by many other retailers to direct consumers to their online-presence (PayPal, 2015). Furthermore, retailers exploit the benefit of customization and location-based targeting by using SMS advertising to increase store-traffic to their brick and mortar presence (ibid). For example, Kiehl’s, a high-end skincare brand, has broadcasted a SMS campaign that sends commercial messages in form of in-store promotions to attract customers and increase purchase intentions and sales in their offline stores (Mobile Commerce Daily b, 2015).

(15)

Research on the impact of the two communication tools on general purchase intentions is nascent. Furthermore, the marketing practice examples demonstrated that marketers have not chosen a dominant communication tool (SMS or MMS) to drive sales in a certain distribution channel. As there exist evidence that both forms of mobile advertising have a positive impact on consumer behavior, the overall impact of mobile advertising is expected to be positive and described by the following hypothesis:

H1: Mobile advertising in form of SMS and MMS has a positive impact on offline and online store visit intentions.

2.3 The Impact of Mobile Advertising’s Content Design on Store Visit Intentions

In the following, the impact of the message content of SMS and MMS on store visit intentions will be reviewed to investigate the effectiveness of mobile advertising to drive consumers’ store visit intentions for the offline or online purchase channel.

(16)

or online, the present study includes the trade-off and regards the intention to visit offline or online stores. In the following, the two possibilities of designing the message content – brand name and promotion - of SMS and MMS advertisements are considered to alter consumers’ store visit intentions.

2.3.1 Brand Name

The function and outcome of brand names in advertising has been studied extensively by many researchers. Landes and Posner (1987) noted that the brand name supports consumers searching process and lowers the needed cognitive effort for product evaluations. Moreover, the risk of an undesired product quality is reduced by including the product name (Rao and Monroe, 1987). Previous studies show that brand names are a significant driver of purchase intentions, especially for high knowledge customers (Grewal et al., 1998) and represent a signal of credibility to reduce customers’ perceived risks (Erdem and Swait, 1998). Given the function of brand names, the present thesis will examine the effect of brand names on store visit intentions as a mean to design the content of mobile advertisement messages congruent with target channels.

(17)

advertising effectiveness with the repeated exposure for TV and online advertisements. In contrast, ads depicting unfamiliar brands are processed more extensively (ibid).

As the present study investigates the effect of different mobile advertisements in the multichannel environment and the different store visit intentions, the impact of brand familiarity on purchase intentions for the offline and online environment is taken into account. Especially in the online environment recent studies focused on the power of brand familiarity to drive purchase intentions. Park and Stoel (2005) concluded that brand familiarity is used as an information source to make purchase decisions on the internet. Therefore the higher the brand familiarity is, the higher the purchase intentions will be (ibid). Furthermore, Park and Lennon (2009) confirmed the previous finding and found that known brands are positively evaluated and will result in online purchases. Regarding the presented studies, this research assumes that online advertising for familiar brands compared to less familiar brands will lead to online rather than offline store visit intentions. As the literature studying the cross channel effect of advertising did not focus on the inclusion of familiar versus unfamiliar brand names yet, the following hypotheses are drawn from the existing literature on online advertising:

H2:Mobile advertising including a familiar brand name has a positive impact on online store visit intentions.

H3: Mobile advertising including an unfamiliar brand name has a positive impact on offline store visit intentions.

2.3.2 Promotion

The inclusion of price promotions in advertising represents an economic incentive to purchase a product (Honea and Dahl, 2005) and is commonly used by marketers as a tool to directly impact purchase intentions and sales (Park, Shenoy and Salvendy, 2008). Consumers make use of promotions to evaluate a product and derive purchase decisions (Raghubir, 2004). Promotions inform consumers about the availability of a product, generate awareness of retailer’s marketing activities, encourage continuous visits of the stores and help to increase customer loyalty (Park and Lennon, 2009). Therefore, the present thesis will test the impact of mobile price promotions on store visit intentions as a means to adapt the content design of mobile advertisement messages.

(18)

by a study by Pauwels et al. (2011), who found out that online promotion in form of an informational website drives in-store sales for sensory products. According to Grewal et al. (2011), who studied innovations in retail pricing and promotions, mobile devices represent an opportunity for marketers to send coupons and promotions to customers’ mobile phones in order to drive brick and mortar sales. Moreover, Park, Shenoy and Salvendy (2008) compared different case studies of mobile marketing and concluded that most of the promotions applied in the mobile marketing environment consist of rewards between $0 and $10, and free coupons which can be redeemed immediately in the store. Applying such a promotion Dunkin Donuts was able to increase their customer’s in-store visits by 21% (Tatango b, 2015).

Whereas the previous studies made no comparison between the impact of advertising on different purchase channels and focused on one channel or purchase intentions in general, the following studies compared both environments. Dinner, van Heerde and Neslin (2014) conducted a study in cooperation with a multichannel fashion retailer and ascertained that all promotions had a greater influence on offline than online sales and explained this with the better performance of the price promotional element of promotions in brick and mortar stores. This finding is consistent with the finding of Zhang and Wedel (2009) who observed that undifferentiated promotions are more profitable in the offline environment, whereas customized promotions exert a higher performance in the online environment. Therefore, it can be assumed that general price promotions, which will be observed in this research, will lead to store visit intentions for traditional brick and mortar retail stores. Thus, the inclusion of a promotional content in mobile advertising is expected to drive offline purchase intentions and the following hypotheses for both types of mobile advertising can be derived:

H4: Mobile advertising including a promotion has a positive impact on offline store visit intentions.

H5: Mobile advertising excluding a promotion has a positive impact on online store visit intentions.

2.3.3 Text-based or Image- & Text-based

(19)

advertisement by not using the technical capabilities of mobile phones (ibid). Whereas SMS have a limited amount of characters (160 per message) and do not include images, MMS consist of graphics, photos videos or audio clips (Park, Shenoy and Salvendy, 2008). MMS offer the possibility to present additional content by displaying an image of the product. These design factors of an advertisement – in this case either solely text-based or a combination of image- and text-based – and the content of an advertisement are expected to interact (Park, Shenoy and Salvendy, 2008). Thus, the impact of the difference in the type of message, SMS or MMS, on the relationship between the message content (brand name or promotion) and online and offline store visit intentions will be examined in this study.

(20)

which in turn influence purchase intentions, SMS is expected to be an effective advertising tool when purchase intentions are not measured by the mediating influence of mental imagery. Drossos et al. (2014) found a positive effect of SMS advertising on purchase intentions moderated by product involvement and consumers’ impulse buying tendency. Moreover, Okazaki and Taylor (2008) indicated the positive outcome of SMS marketing perceived by managers, as they hold favorable attitudes towards the capability of SMS advertising to build a brand. Xu, Oh and Teo (2009) revealed a positive effect of SMS and MMS advertising on purchase intentions of movie tickets and apparel. Furthermore, Xu, Oh and Teo (2009) found out that consumers’ attitudes towards MMS are more favorable and lead to higher purchase intentions for MMS advertised products. The present study expects a positive impact of MMS compared to SMS mobile advertising on the relationship between the message content (brand name and promotion) and store visit intentions in the online and offline environment. MMS advertising offers additional image content which is valued by customers and positively influences purchase intentions; thus, the following hypotheses depict the expected moderating effect of MMS advertising compared to SMS advertising:

H6a: Mobile advertising in form of MMS, providing additional image content, strengthens the positive effect of the inclusion of a familiar brand in a mobile advertising on online store visit intentions.

H6b: Mobile advertising in form of MMS, providing additional image content, strengthens the positive effect of the inclusion of an unfamiliar brand in a mobile advertising on offline store visit intentions.

H6c: Mobile advertising in form of MMS, providing additional image content, strengthens the positive effect of the inclusion of a promotion in a mobile advertising on offline store visit intentions.

(21)

2.4 Conceptual Model

The previous overview of the relevant variables provides the basis for this research and their hypothesized relationships result in the conceptual model (see Figure 1). The present thesis will measure the effect of mobile advertising on store visit intentions for brick and mortar and online stores. A distinction hereby is made between the two types of mobile advertising, SMS and MMS. These messages will be altered by testing two types of message content in form of a known versus unknown brand name and the use of a price promotion versus no use of a promotion will be researched. Moreover, the moderating effects of SMS advertising as a proxy of text-based communication and MMS advertising providing the possibility to communicate a combination of text and pictures will be taken into account. The distinct variations of the message content are implemented to answer the research question:

What impact does mobile advertising have on store visit intentions across channels (offline and online stores)? Does the content of the message (brand name and promotion) influence the intention to visit one particular purchase channel? And does the type of the medium (SMS or MMS) further strengthen customers’ intentions?

(22)

3

Methodology

The previous chapter provided an overview over the existing literature and came up with several hypotheses which have been combined in a conceptual model. In the present chapter, the conceptual model will be tested by performing an empirical research. Therefore, the chosen research methodology will be explained beginning with the research design and the method of data collection. Thereupon, the measurement and manipulation of variables and the operationalization of scales for the dependent variables will be presented. Finally, the plan of the analysis will be elaborated.

3.1 Research Design and Data Collection

In the present study, an experiment will be conducted in order to test the variation of the message content (brand name and promotion), the different types of mobile advertising (SMS and MMS) and their impact on offline and online store visit intentions.

An experiment allows the manipulation of one or more independent variables to measure their effect on one or more dependent variables (Aronson, Wilson and Brewer, 1998). By manipulating the dimensions of the message content (brand name: known or unknown brand name; promotion: yes or no) and type of mobile advertising (SMS or MMS), a 2 x 2 x 2 between-subject, full factorial design is set up. The mobile advertising content brand name and promotion depict the independent variables. The dependent or outcome variables are offline and online store visit intentions. As a full factorial design is used, which implies the crossing of both levels of each independent variable and both levels of the moderating variable, the experiment consists of eight conditions. Moreover, a baseline condition is included which presents no advertising at all to enable a comparison with the other eight conditions and to assess the overall effectiveness of mobile advertising. Participants are randomly assigned to one of the nine conditions; and therefore, have the equal chance to be allocated to each experiment condition (Aronson, Wilson and Brewer, 1998). A random assignment ensures that participants do not vary significantly before the experimental manipulations and observed differences between participants afterwards result from the experimental manipulation (ibid).

(23)

they have to rate the importance of several characteristics of offline and online stores to control for general preferences in store choice. The experiment ends with a set of questions to include further control variables and collect some general descriptives of the participants. The questionnaire will be distributed online among people living in Europe. This group is expected to be confronted regularly with the choice between visiting an offline or online store and is able to make independent decisions. An online experiment offers the possibility of a fast and flexible distribution via e-mail and social media platforms, such as Facebook or Twitter. Furthermore, the collected responses will be digital and available for the further analysis.

3.2 Procedure

(24)

to answer a question representing a manipulation check to see whether the brand names are perceived as known or unknown. Finally, participants are requested to answer several control questions regarding their gender, age, education and channel experience in the offline and online environment. The full questionnaire is provided in Appendix 1.

3.3 Measurement and Manipulation

3.3.1 Independent Variables – Brand Name and Promotion

The independent variables need to be manipulated in order to test the effect of different content types of mobile advertising on the decision to visit an offline or online store. For the sake of this research, two variations of the mobile advertising content, namely brand name and promotion, are chosen which result in two categorical independent variables.

Brand name

As elaborated in the theoretical part of the thesis, the brand name will be manipulated by promoting a known brand versus an unknown brand. The promoted product in both forms of mobile advertising will be jeans as jeans are perceived as a common garment for women and men and the majority of the population is familiar with this clothing item. The known jeans brand is represented by the brand “Levi’s”. As Levi’s is one of the leading apparel retailers for jeans and the first company which started selling jeans, the brand familiarity is rated as high and every participant is expected to know the brand. This is confirmed by a study in the Indian apparel industry which found that branded jeans are known and the highest preference for branded jeans was assigned to Levi’s (Pandya and Pandya, 2013). An unknown brand name “BlueStar” was created to serve for the “no brand familiarity” condition. In order to control the brand familiarity of both brands, participants have to indicate their familiarity with the brand after the exposure to the ad. The study expects that participants will not know the brand BlueStar and will be familiar with the brand Levi’s.

Promotion

(25)

voucher for the next purchase of a new pair of jeans. The promotion includes the information that the discount can be redeemed either in an online store or in a brick and mortar store to not predetermine customers’ offline or online store visit intention.

3.3.2 Dependent Variables – Offline and Online Store Visit Intentions

(26)

3.3.3 Moderating Variable – Type of Mobile Advertising

The considered moderating variable “type of mobile advertising” is a categorical variable and consists of two dimensions, SMS and MMS. SMS and MMS possess the same characteristics and usage prerequisites (Xu, Oh and Teo, 2009), such as the overall visual appearance, the text-limit and customers’ opt-ins to receive the advertising on their mobile phones. The only difference between the two dimensions represents

the picture used in MMS advertising, which depicts the manipulation and enables the researcher to further specify the effect of different designs of mobile advertising. The additional image (see Figure 2) (Simplyshellie, 2015) shows four adults (two men and two women) wearing jeans.

3.3.4 Control Variables

The final section of the research consists of several questions which are included to measure the control variables. The control variables are used to assess descriptions of the participants concerning their age, gender and education. Moreover, control variables, which measure the importance of functional characteristics and channel experience, in both types of stores are regarded. These variables are intended to reveal possible influences on consumers’ channel choice after being exposed to the mobile advertisement. Prior research found effects of both variables on channel choice. The channel experience as an influencing variable has been tested by Gensler, Verhoef and Boehm (2012). They asked participants where they have made their last purchase which results in a categorical variable with the dimensions online shop and offline store (Gensler, Verhoef and Boehm, 2012). This variable is included to measure the last location for a jeans purchase of the participants. Additionally, Frambach et al. (2007) measured the functional importance of several store characteristics on a 7-point Likert scale (1 = not important at all; 7 = very important). Store characteristics include the “accessibility” (3 items), “ease of use” (3 items), “usefulness” (5 items), and “social presence” (2 items) perceived in a store. The importance of these store characteristics is measured for offline and online search and purchase resulting in 4 scales of this type. “Usefulness” is excluded from the present research as a pre-test (N = 12) revealed that the rating of 13 items over 4 situations takes too long; and therefore will lead to high drop-out rates in the online survey.

(27)

Furthermore, the items measuring the “usefulness” were perceived as similar to “ease of use” items.

3.4 Plan of Analysis and Model Specification

3.4.1 Plan of Analysis

The plan of the analysis covers three steps to derive the experiment’s results. Before deep diving in the analysis of the hypotheses, the demographics will be analyzed. This is done by a frequency analysis and computing averages to describe the participants’ age, gender, and education. Moreover, the participants’ channel experiences (online versus offline) and the perceived importance of different store characteristics will be described. If there exists a significant difference between the online and offline channel experience and the importance of store characteristics, these variables will be considered as confound and included in the subsequent analysis.

After the analysis of the control variables, the reliability of the dependent variables and the manipulation of the independent variables will be tested. The reliability analysis should confirm the consistency of the developed scale for online and offline store visit intentions. This is necessary to make sure that the dependent variables produce consistent results. Cronbach’s alpha is chosen as the reliability test. Furthermore, a factor analysis for the constructs of store characteristic’s importance will be performed to reduce the set of variables and make it more manageable. These factors are also checked for internal consistency by using Cronbach’s alpha. The manipulation of the independent variable brand familiarity needs also be confirmed as significant. Therefore the results of the control question (“Have you heard of the advertised brand before?”) will be analyzed with cross-tabulation to see if participants indicated Levi’s as the known brand and BlueStar as the unknown brand.

(28)

variables, this analysis is the appropriate way to derive valuable results (Malhotra, 2010). To be able to use the regression models, several assumptions have to be met. This study will check for the existence of a normal distribution, the homogeneity of variances and no multicollinearity. These assumptions will be tested before the regression analyses can be performed. As the study does not deal with time series data, other assumptions of linear regression will not be taken into account.

3.4.2 Model specification

The regression models consist of the independent variables examined in the literature review and the methodology part. The regressed models are represented by the following equations:

- Model 1 and 4: Offline Store Visit Intention = α0 + α1×NO_AD + α2×PROMO + α3×BRAND + α4×MMS + α5×[MMS×PROMO] + α6×[MMS×BRAND] + ε, - Model 2 and 5: Online Store Visit Intention = β0 + β1×NO_AD + β2×PROMO +

β3×BRAND + β4×MMS + β5×[MMS×PROMO] + β6×[MMS×BRAND] + ε, - Model 3 and 6: Difference between Online and Offline Store Visit Intention = γ0 +

γ1×NO_AD + γ2×PROMO + γ3×BRAND + γ4×MMS + γ5×[MMS×PROMO] + γ6×[MMS×BRAND] + ε,

with αs, βs and γs as the coefficients, NO_AD as the baseline condition, PROMO as promotional content, BRAND as a known brand name, MMS as the type of mobile advertising (MMS), and ε as the residual.

(29)

4

Results

The following chapter presents the collected data and covers the analysis of the data. Firstly, the cleaning of the gathered data will be described. Secondly, the descriptives of the respondents will be presented to gain an overview of the characteristics of the respondents. Thereupon, the necessary preparation of the data for the final analysis will be conducted and the assumptions of the used statistic model will be tested. Lastly, multiple regression analysis will be performed to validate the hypotheses depicting the main and moderating effects of the independent variables on the dependent variables.

4.1 Data

4.1.1 Data Editing and Cleaning

Before the actual data sample can be described and analyzed, the data has to be edited and cleaned. In the first step, the data is edited by reviewing the survey answers with regard to accuracy and precision (Malhotra, 2010). In total, the survey has been started by 278 people of whom 212 finished the survey. Out of the 278 people, 44 did not fill in any question and solely opened the survey which implies a dropout rate of 15.8%. Therefore, these 44 responses were removed from the data set. Furthermore, the remaining 22 respondents, who did not complete the entire survey, dropped out during the rating of the importance of the store characteristics on 7-point Likert scales. Five of the 22 people stopped the questionnaire before answering the first block, assessing the importance of the online store characteristics for the purpose of searching a new pair of jeans. Moreover, two of the 22 respondents dropped out after this first block. 15 respondents closed the survey after the second block asking for the importance of the offline store characteristics for the purpose of searching new jeans. As the pre-test also indicated that the rating of many similar items might deter respondents, this depicts a possible explanation for this particular dropout point. The 22 respondents were removed from the data set as they account for almost 10% (N = 23.4) of the total sample size; and therefore, the assignment of missing values might pose problems (Malhotra, 2010).

(30)

will be remained in the sample. Secondly, the responses assessing the importance of the store characteristics are tested for outliers. Two participants showed a response pattern not deviating between the items of the four blocks assessing the store characteristic importance. The participants indicated that all store characteristics are not important at all to them (1 on a 7-point Likert scale) and also answered all other control questions with the first option. This shows an automatic pilot behavior in answering the survey which can be confirmed by looking at their response time for the whole survey which covers 3 minutes; hence, these responses were excluded from the data set. Thereupon, the recall of the advertisement and the advertised brand has been retrieved. This question was included to test, whether participants filled in the survey attentively. To access the correctness of recall, cross-tabulation is used. The results in the crosstab (see Appendix 2) show that 13 people saw a brand advertisement (Levi’s = 5; BlueStar = 8) and indicated that they have not seen an advertisement. Moreover, four people (Levi’s = 1; BlueStar = 3) indicated that they forgot which brand has been advertised. One respondent answered that he saw the brand Levi’s, although he had been exposed to a BlueStar ad. These respondents will be retained in the data set because it is a common phenomenon that people forget about advertisements or confuse brands due to the advertising clutter (Fennis and Stroebe, 2010).

Consequently, the data editing and cleaning result in a total sample size of 210 valid response sets. As the present study examines nine different experimental sets, a total of 180 respondents (at least 20 per condition) are necessary to obtain reliable estimates for a condition (Hair et al., 2009). Therefore, 210 responses are sufficient for the further analysis.

4.1.2 Sample Description

(31)

degree. As most of these academics are aged between 22 and 24, a possible explanation for their high online usage might be that they just finished their bachelor studies and have enough free time to use the internet. Regarding the respondents’ online and offline shopping behaviors, the most people use traditional and online stores one to two times a month (online store = 46.2% and offline store = 58.1%). Only one person indicated that he never shops at offline stores (0.5%) and five people (2.4%) do not use online stores at all.

To assess the channel experience of respondents, they were asked, where they bought their last pair of jeans. The majority made their last purchase in a traditional store (77.6%), although most of these respondents conduct online purchases one to two times a month (69.9%). This might be an indicator for a preference of the offline channel for jeans purchases and will be taken into account while checking the regression results to see if the channel experience significantly influences the store visit intentions for the offline store.

Moreover, participants had to rate the importance of several characteristics of the online and offline store for the purpose of searching and purchasing a new pair of jeans. The average values can be found in Appendix 3. The mean values indicate that respondents do not differentiate between searching and purchasing while judging the importance of the different store characteristics. However, there exists a difference between the importance of different store characteristics for the online and offline environment. Participants see a higher importance in the “accessibility” (Mon_accessibility = 5.68; SD = 1.43) and “ease of use”

(Mon_easeofuse = 6.12; SD = 1.06) of the online than the offline store (Moff_accessibility = 5.14; SD =

1.40; Moff_easeofuse = 5.30; SD = 1.33). The importance of “social presence” is rated higher in

offline stores (Moff_socialpresence = 4.57; SD = 1.72) compared to online stores (Mon_socialpresence =

3.35; SD = 1.63).

Furthermore, the choice probability scales for the dependent variables online and offline store visit intentions show the following values: The general intention to search for a new pair of jeans in an online store is slightly higher (Mon_search = 60.96%; SD = 28.72) than the intention

to search in an offline store (Moff_search = 57.38%; SD = 26.82). In contrast, the purchase

intention for a new pair of jeans is higher in the offline store (Moff_purchase = 67.64%; SD =

(32)

Table 1 – Overview sample descriptives

4.1.3 Random Assignment

Every participant of the 210 response sets has been exposed to one of the nine different experiment conditions. The distribution of the participants over the conditions can be found in Table 2a. Due to deletion of participants who did not fill any question or had an unacceptable share of missing values (>10%), the distribution of participants to the conditions is slightly varying between conditions. The biggest difference depicts nine people (17 in condition MMS_L_P versus 26 in MMS_L_P and SMS_B_P) and one condition (MMS_L_P) consists of less than 20 participants. Thus, the experimental design is fairly balanced. To assure that participants were randomly assigned over all nine conditions, the random assignment has to be validated by comparing the means of the descriptives across the different conditions. Age is be tested by an ANOVA, whereas the remaining descriptives are reviewed by Pearson’s chi-squared test. The results of the ANOVA and the Pearson’s chi-squared test (see Table 2b) show that there are no significant differences (p > .05) between the control variables age (p = .153), gender (p = .116), education (p = .862), last jeans purchase (p = .423), daily online use (p = .570), offline and online shopping frequency (p = .359; p = .579). Therefore, the random assignment of participants to the experimental conditions has been successful.

Mean Std. Devation Frequency Percent (%)

Age 28.20 9.20 Gender

Online search intention 60.96 28.72 Male 78 37.1

Offline search intention 57.38 26.82 Female 132 62.9

Online purchase intention 44.56 30.67

Offline purchase intention 67.64 25.86 Online shop 47 22.4 Traditional store 163 77.6 Frequency Percent (%) Frequency Percent (%)

Education Daily Online Use

Less than high school 8 3.8 Less than 1 hour 19 9.0

High School 56 26.7 1-2 hours 47 22.4

Bachelor 83 39.5 2-4 hours 81 38.6

Master 57 27.1 4-6 hours 43 20.5

Other 6 2.9 More than 6 hours 20 9.5

Frequency Percent (%) Frequency Percent (%)

Offline Shopping Online Shopping

Never 1 0.5 Never 5 2.4

Less than once a month 65 31.0 Less than once a month 90 42.9 1-2 times a month 122 58.1 1-2 times a month 97 46.2

Once a week 22 0.5 Once a week 14 6.7

Twice or more a week 0 0.0 Twice a week or more 4 1.9 N=210 (100%)

(33)

Condition N ANOVA F Sig.

Baseline 23 Age 1.503 0.153

MMS_L_P* 17 Pearson's χ² test Value Sig.

MMS_B_P* 23 Gender 12.885 0.116

SMS_L_P* 24 Education 23.489 0.862

SMS_B_P* 24 Daily Online Use 29.958 0.570 MMS_L_NP* 26 Offline Shopping Frequency 25.887 0.359 SMS_L_NP* 23 Online Shopping Frequency 29.783 0.579 MMS_B_NP* 24 Last jeans purchase 8.111 0.423 SMS_B_NP* 26 (Channel experience)

Total 210 Table 2b

Table 2a

*L= Levi's; B=BlueStar; P=Promotion; NP= No promotion

Table 2 – Condition distribution and random assignment test

4.2 Preparation for the Analysis

In order to perform the final analysis, it is necessary to run a factor analysis. By performing a factor analysis the correlated variables assessing one construct are reduced into a smaller, more manageable set of uncorrelated factors (Malhotra, 2010). Factor analysis will be run for the following constructs: store visit intentions and store characteristic’s importance. Furthermore, the manipulation of the brand name (known versus unknown) has to be checked. Finally, three assumptions necessary to perform a multiple regression analysis are reviewed with a normality test, a homogeneity of variance test and a multicollinearty test.

4.2.1 Reliability Analysis

Store Visit Intentions

(34)

Store characteristic’s importance

In the present study, the store characteristic’s importance is measured with several items for different underlying concepts (for example, ease of use and social presence). Firstly, an exploratory factor analysis with varimax rotation is conducted. The KMO measure of sampling adequacy and Bartlett’s test of sphericity are used and reveal that factor analysis is appropriate (KMO = 0.806; Bartlett’s test of sphericity= p < .001). The factor analysis leads to eight factors which exceed the number of factors (six factors) proposed in the theoretical construct (Frambach et al., 2007). Therefore, a confirmatory factor analysis, predetermining six factors, is executed. This factor analysis leads to low communalities (< 0.4) for the item “accessible”; thus, this item is removed from the initial response set for all blocks (online and offline search and purchase). The KMO value (0.783) is sufficient and the Bartlett’s test of sphericity is significant (p < .001). Consequently, all items showed communalities over 0.4 and factor loadings over 0.5. All factors derived by the factor analysis are saved on the basis of the respective factor scores and can be used for further analysis. As the item “accessible” was removed from the theoretical construct “accessibility”, this factor is renamed in “availability”, describing the importance of the location- and time-based availability of a store. All other factor names are taken from the store characteristics described in Frambach et al. (2007). The results of the factor analysis can be found in Appendix 4.

After conducting the factor analysis, the resulting dimensions need to be checked whether they are meaningful and therefore reliable. Cronbach’s alpha which assesses the internal consistency is used as a reliability measure. Internal consistency is guaranteed, if Cronbach’s alpha is higher than 0.6. All derived factors show a sufficient Cronbach’s alpha (see Appendix 4). Moreover, the factor “availability” is tested for reliability with the excluded item “accessible” and indeed, the deletion of this item leads to a higher Cronbach’s alpha for both factors (online and offline availability) (online = 0.829 < 0.864; offline = 0.841 < 0.885). Thus, all factors are strong enough to proceed with these dimensions instead of the original items.

(35)

4.2.2 Manipulation Check

The brand familiarity is manipulated by depicting a familiar brand (Levi’s) versus an unfamiliar brand (BlueStar). To assess the correctness of the manipulation, cross-tabulation is used (see Table 3 for results). Respondents who remembered that they saw a Levi’s advertisement indicated that they know the brand. Moreover, respondents remembering the exposure to a BlueStar advertisement answered that they do not know the brand. Therefore, the manipulation of brand familiarity has been successful.

Table 3 – Cross-tabulation brand recall*brand familiarity

4.2.3 Normality Test

To be able to draw valid conclusions from a regression analysis, the assumption of normality has to be met (Field, 2009). Therefore, the Kolmogorov-Smirnov test and Shapiro-Wilk test will be executed to check for the normality of the distribution of the unstandardized residuals in all six regression models. Both test showed significant results (p < .05) for the disturbances of the regression models 1 and 4 (see Table 4). Consequently, the null-hypothesis of a normal distribution of the disturbances has to be rejected. However, this test has limitations regarding large sample sizes as it results very easy in significant outcomes although the deviation from normality is small (Field, 2009). Therefore, to gain more insights about the distribution of the data, the skewness and kurtosis statistics are consulted. The results can be also found in Table 4. The skewness statistic indicates whether the data is symmetrical and normally distributed, whereas the kurtosis statistic depicts how peaked or flat the data is distributed (Field, 2009). Skewness values showing a normal distribution range from -0.5 and 0.5 and from -1.96 and 1.96 for the kurtosis statistic. Nevertheless, a value of zero represents an indicator for a normal distribution. The skewness statistic shows that the residuals of the six regression models are located within the acceptable range of values from -.387 to -.174 (see Table 4). All values are concentrated on the right side of the mean. In the kurtosis statistic the residuals of the regression models show values within the acceptable scope of kurtosis-values from .018 to

(36)

Statistic Sig. Statistic Sig. Statistic Std. Error Statistic Std. Error Unstd. Res. Model 1 .095 .000 .965 .000 -.387 .168 -.780 .334 Unstd. Res. Model 2 .046 .200 .992 .321 -.268 .168 -.037 .334 Unstd. Res. Model 3 .053 .200 .989 .118 -.228 .168 -.244 .334 Unstd. Res. Model 4 .091 .000 .967 .000 -.369 .168 -.773 .334

Unstd. Res. Model 5 .044 .200 .991 .200 -.342 .168 .018 .334

Unstd. Res. Model 6 .045 .200 .993 .401 -.174 .168 -.200 .334 Kolmogorov-Smirnov Shapiro-Wilk Skewness Kurtosis

-.780 (see Table 4). A kurtosis value, which is smaller than 0, depicts a flatter distribution with a wider peak; whereas, values greater than 0 show a sharper distribution and are concentrated around the mean. Compared to the Kolmogorov-Smirnov and the Shapiro-Wilk test these results give a better understanding of the distribution of the outcome variables. There are no extreme deviations from the normal distribution; however, the Kolmogorov-Smirnov and Shapiro-Wilk test showed a violation of the normal distribution of the residuals of two regression models (1 and 4). Still, the violation of the assumption is a limitation of this research and will be incorporated in the limitations section.

Table 4 – Kolmogorov-Smirnov test; Shapiro-Wilk test; Skewness and Kurtosis statistics

4.2.4 Homogeneity of Variance Test

Another necessary assumption of a regression analysis, is the homogeneity of error variances across the different experimental conditions. This assumption can be assessed by the Levene’s test (Field, 2009). The results of the test are depicted in Table 5. The Levene’s test for the regression models 1, 2, 4, 5 and 6 is not significant (p > .05), whereas it shows a significant result for model 3 (p = .045). Therefore, the hypothesis, that the error variances are not significantly different across the experimental conditions, has to be rejected for model 3. This significant result is taken into account when interpreting the results and can be seen as a limitation of this research.

To secure for the violation of the assumptions of normally distributed residuals and homogeneity of variances, the dependent variables online and offline store visit intentions, will be log-transformed and further analysis is conducted with the log-transformed values.

Table 5 – Levene’s test of homogeneity of variance

Levene

Statistic df1 df2 Sig.

(37)

4.2.5 Multicollinearity Test

According to Field (2009), multicollinearity is present when two or more of the independent variables exhibit a strong correlation. Multicollinearity may affect b-coefficients and lead to unreliable estimates. When looking at the Variance Inflation Factor (VIF) for the independent variables in the regression models, VIF scores above five are considered as critical (Field, 2009). For the multicollinearity check the control variables will be reviewed, as the independent variables have been manipulated; and therefore, no multicollinearity will be present. All of the independent variables in the remaining regression models showed VIF score values between 1.053 and 1.668 which is below the critical value of five. Therefore, no multicollinearity is detected. The results of the collinearity diagnose can be found in Table 6.

Model 1-3 Model 4-6 VIF scores 1.058-1.668 1.053-1.548

Table 6 - Variance Inflation Factor value range per model

4.3 Hypotheses Testing

In the following section the hypotheses developed from the literature review will be tested by running multiple regression analysis. As indicated in the methodology section, six different models for the offline and online environment and the difference between these environments have been set up. Beginning with testing the main effects (model 1 to 3), a multiple regression including all independent variables is used (see Table 7 for results). Thereupon, the moderating effect of the type of message (SMS versus MMS) will be checked by performing a multiple regression analysis with the inclusion of the interaction effects (model 4 to 6) (see Table 8 for results).

4.3.1 Main Effects

Online Store Visit Intentions (Model 1)

(38)

of the adjusted R². Furthermore, all included independent variables have an assumed impact on online store visit intentions which will be interpreted by looking at the regression results; therefore, they cannot be removed from the model. However, the high adjustment of the R² might be an indicator for the low explanation power of the independent variables. Moreover, a relatively small number of observations results also in low values for the adjusted R².

In fact, the effect of the independent variables BRAND (p = .816) and PROMO (p = .783) on online store visit intentions is not significant (p > .1). Mobile advertising including a familiar brand name has a positive impact on online store visit intentions but the impact is not significant (p = .816). Therefore, hypothesis 2 is not supported. Moreover, hypothesis 5 has to be rejected as the effect of the dummy variable PROMO is insignificant. The absence of a promotion (reference category of the variable PROMO) influences online store visit intentions negatively which is contradicting to hypothesis 5. Furthermore, the effect of the dummy variable NO_AD is also not significant (p = .216), which indicates that being exposed to a mobile ad versus not being exposed to a mobile advertisement has no impact on the intention to visit an online store. Nevertheless, the effect of NO_AD on online store visit intentions has a negative sign. Thus, the direction of the effect is in line with hypothesis 1. Still, hypothesis 1 is not supported due to the insignificance of the effect (p > .1). Moreover, the moderating variable MMS (with SMS as a reference category) shows no significant impact on online store visit intentions (p = .503). The negative sign of the impact shows a positive impact of SMS compared to MMS advertisements; however, this effect is not taken into account at this step of the analysis as it will be regarded in the regression models including the interaction effects.

(39)

online store visit intentions (p = .095). This reveals that people, who value the social presence in an online store more, are less likely to visit an online store as social presence usually does not exist in an online store. Furthermore, the customers online and offline shopping frequency have an impact on online store visit intentions. A medium and high offline shopping frequency impact online store visit intentions negatively (p = .075; p = .041) and a medium online shopping frequency has a positive effect on online store visit intentions (p = .098). Offline Store Visit Intentions (Model 2)

In order to assess the offline store visit intentions, the intention to search for a product offline has been measured and the results of the model are significant (F = 2.083, and p < .05). The model can explain 18.9% of the total variance (R² = .189). The adjusted R² in this model accounts for .098. Again, this value is relative low compared to the initial R² indicating a high penalty for adding variables. Nevertheless, the model cannot be changed as explained for online store visit intentions; it is kept in mind that the high adjustment of the R² might be a reason for the low explanation power of the independent variables.

(40)

Regarding the channel experience, the significance of an effect cannot be confirmed (p = .167). Similar to online store visit intentions, there exist some effects of the control variables “store characteristic’s importance” on offline store visit intentions. The importance of “availability” (p = .055) and “ease of use” (p = .070) in an online store have both positive impacts on offline store visit intentions (p < .1). This is again surprising, as these characteristics should rather support the use of the online than the offline store. Moreover, the importance of “social presence” in an online store positively influences the use of an offline store (p = .064), which indicates that customers who value social presence are more likely to visit an offline store as the degree of social presence in an online store is rather low. Lastly, medium and high offline shopping frequencies have a positive impact on offline store visit intentions (p = .009; p < .011), whereas a medium online shopping frequency negatively impacts offline store visit intentions (p = .016).

Difference between Online and Offline Store Visit Intentions (Model 3)

The comparison of the intention to either search online or offline for a new product is enabled by subtracting the online and offline store visit intentions. The overall model is significant (F = 2.361, and p < .001) and has an explained variance of R² = .209. The adjusted R² is .120, which depicts a high adjustment of the R² due to the inclusion of variables with a low explanatory power. A change of the model is not possible as explained before.

The independent variables of the conceptual model regarded in this research again show no significant relationship with the dependent variable (p > .1). As the third model should enable the comparison between the differential effects of a known versus an unknown brand and the inclusion of a promotion versus no promotion on the intention to visit either an offline or an online store, the comparison cannot be conducted. Hence, it cannot be indicated whether the inclusion of a known brand and a promotion and the use of MMS have higher impact on the offline or online store visit intention.

(41)

B t B t B t Constant 4.283 12.112**** 3.618 11.226**** .664 1.467 Independent Variables NO_AD -.253 -1.128 .207 1.009 -.460 -1.599 PROMO .036 .275 .139 1.151 -.103 -.605 BRAND .031 .233 -.072 -.597 .103 .607 MMS -.089 -.671 -.103 -.856 .015 .086 Control Variables channel_experience (last purchase: offline)

.039 .240 .204 1.387 -.165 -.800 online_availability .057 .936 .108 1.934* -.051 -.646 offline_availability .132 2.124** .074 1.298 .059 .735 online_easeofuse .158 2.552** .103 1.825* .055 .694 offline_easeofuse .070 1.117 .005 .080 .065 .815 online_socialpresence -.108 -1.680* .109 1.860* -.217 -2.636*** offline_socialpresence -.009 -.135 .031 .524 -.039 -.479 internet_use_medium .096 .606 .002 .016 .093 .462 internet_use_high -.149 -.885 -.179 -1.165 .030 .138 on_shop_medium .222 1.663* -.294 -2.421** .515 3.022*** on_shop_high .163 .665 -.215 -.963 .377 1.205 off_shop_medium -.269 -1.792* .359 2.626*** -.628 -3.269**** off_shop_high -.465 -2.059** .529 2.571** -.994 -3.438**** age -.010 -1.232 .002 .318 -.012 -1.188 gender -.123 -.870 -.026 -.199 -.098 -.538 bachelor .158 1.070 -.065 -.485 .223 1.181 master_and_higher -.067 -.396 -.060 -.392 -.006 -.030 R2 R2 adjusted F p .024 .005 .001 .166 .189 .209 .072 .098 .120

Online store visit intentions (Model 1)

Offline store visit intentions (Model 2)

Difference between Online and Offline store visit intentions

(Model 3)

1.776 2.083 2.361

****p < 0.001, ***p < 0.01, **p < 0.05, *p < 0.1

Dummy-coded variables reference categories: no_ad (ad); promo (no promotion), brand (unknown brand), MMS (SMS), gender (male); education (high school and lower); last purchase (online shop); online use (online use low); online/offline shopping (online/offline shopping low)

(42)

4.3.2 Moderating Effects

Online Store Visit Intentions (Model 4)

Including the predictors of the interaction effects of message type and message content in the regression model for online store visit intentions results in a significant model (F = 1.738, and p < .05) which explains a total variance of R² = .177. The adjusted R² accounts for .075, which is .03 higher than the adjusted R² for online store visit intentions without the interaction effects.

There exists no significant effect of the independent variables PROMO (p = .409) and BRAND (p = .434). Hypothesis 2 is rejected and no direct effect of a known brand on online store visit intentions can be revealed. The exclusion of a promotion (reference category of the dummy-variable PROMO) is not significant. Hence, hypothesis 5 needs to be rejected, as well. Moreover, the dummy variable NO_AD is not significant (p = .255); and therefore, exposing a consumer to a mobile ad versus not exposing him to an advertisement results in no significant impact on online store visit intentions. However, the sign of the NO_AD predictor is negative implying a positive effect of having a mobile advertisement on online store visit intentions. Nevertheless, hypothesis 1 is not supported due to the insignificance of the effect. Furthermore, the moderating variable MMS shows a negative effect on online store visit intentions (p = .508). Though, this effect is replaced by the interaction effects of MMS and BRAND and MMS and PROMO.

Referenties

GERELATEERDE DOCUMENTEN

Figures 12 and 13 for the cosine vortex and Figures 14 and 15 for Burgers’ solution, respectively, show the influence of q on the optimal global smoothing length scale and

The problem is at times though that when one talks about countries or cultures (the national and the cultural are often equated) one tends to focus on Ôcultural stereotypesÕ

Domestication theory on one hand is used to explicate the role mothers play in the shaping process, while technology mediation would highlight the role the mobile phone plays

As I have already said, the Committee on Economic, Cultural and Social Rights, which is supervising the signatories compliance with the United Nation’s ICESCR, has in

Z-scores of the PCS, subscale of general health perception and subscale of physical functioning (RAND-36) were significantly lower and scores in the subscale of

Yu, Xiuzhen; Mostovoy, Maxim; Tokunaga, Yusuke; Zhang, Weizhu; Kimoto, Koji; Matsui, Yoshio; Kaneko, Yoshio; Nagaosa, Naoto; Tokura, Yoshinori..

Also, when in the course of electronic monitoring it turns out a partici- pant has committed more offenses than known at the start of electronic monitoring and thus turns out not

[r]