Advertising targeted to the wrist : are consumers ready for smartwatch advertising?

(1)

Advertising targeted to the wrist: are consumers ready for

smartwatch advertising?

Name of author: Margot Haring

Student number: 11363800

Date of submission: 21 June 2018

Name of institution: University of Amsterdam

Field of study: MSc. in Business Administration – Digital Business Track

(2)

Statement of Originality

This document is written by student Margot Haring who declares to take full responsibility for the contents of this document.

I declare that the text and the work presented in this document are original and that no sources other than those mentioned in the text and its references have been used in creating it.

The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.

(3)

__________________________________________________________________________

To all that have stimulated my creativity. To my caring family and beloved friend Kirsten.

-

I would like to express my profound gratitude to my supervisor Adriana who has guided me in this journey and has sparkled my interest in research.

(4)

Abstract

In recent years technological developments have rapidly picked up and new forms of advertising have emerged. This study investigates a new concept in the marketing field, namely location-based smartwatch advertising (LBSA). LBSA enables advertisers to target consumers on a smartwatch device based on consumers’ geo-location. This research sheds light on how consumer innovativeness in IT has an indirect effect on purchase intention through attitude towards LBSA and reviews the moderating effect of consumption goal. An online between-subjects experiment was conducted among 245 Dutch consumers whereby participants were asked for their likelihood to purchase products via LBSA. Participants were randomly assigned to one of the two experimental conditions (i.e., hedonic vs. utilitarian consumption goal) or to the control condition (i.e., no specified consumption goal). Through a scenario-based manipulation, participants were presented with an example of a location-based smartwatch advertisement. Purchase intention was measured before and after the manipulation was made. The results of this study reveal that innovative consumers display more favourable attitudes towards LBSA and subsequently have higher purchase intention scores. A remarkable finding is that the relationship between attitude towards LBSA and purchase intention is contingent on consumers’ consumption goal. More specifically, this relationship is moderated by consumption goal, such that the relationship becomes stronger for hedonic shoppers and weaker for utilitarian shoppers. Implications of these findings are discussed together with areas for future research in the remaining body of this paper.

Key words: location-based advertising, smartwatch technology, hedonic and utilitarian

(6)

List of Tables and Figures

Tables

Table 1 Kolmogorov-Smirnov and Shapiro-Wilk Tests for Normality 39

Table 2 Skewness and Kurtosis Tests for Normality 40

Table 3 Cronbach's Test for Reliability 41

Table 4 Principal Component Analysis 42

Table 5 Coding Scheme in PROCESS 45

Table 6 Means, Standard Deviations, Correlations and Reliabilities 48 Table 7 Regression Results Attitude towards LBSA and Purchase Intention 49

Table 8 Recoded Scheme in PROCESS 50

Table 9 Overview Regression Results 53

Table 10 Related Samples Sign Test 55

Table 11 Two-step Cluster Analysis 56

Figures

Figure 1 Theory Acceptance Model 18

Figure 2 Theoretical Framework 30

Figure 3 Interaction After Manipulation 51

Figure 4 Interaction Before Manipulation 52

Figure 5 Mean Scores Purchase Intention After Manipulation 54

(7)

List of Abbreviations

AI Artificial Intelligence

ARM Acorn RISC Machine

ESM Experience Sampling Method

GDPR General Data Protection Regulation

GPS Global Positioning System

IoT Internet of Things

IPI Information-Possessing Innovativeness

IT Information Technology

KMO Kaiser-Meyer-Olkin

KPI Key Performance Indicator

LBA Location-Based Advertising

LBMA Location-Based Mobile Advertising

LBSA Location-Based Smartwatch Advertising

MDA Mobile Display Advertising

OS Operating System

PCA Principal Component Analysis

PPI Product-Possessing Innovativeness

RAM Random Access Memory

TAM Technology Acceptance Model

TRA Theory of Reasoned Action

(8)

1. Introduction

In recent years, technological developments have rapidly changed the business landscape. Digital technologies such as smartphones, wearable devices (e.g., smartwatches, smart glasses), the Internet of Things (IoT), Artificial Intelligence (AI), Virtual Reality (VR) and deep learning promise to shape the future of consumers’ lives and businesses alike (Kannan & Li, 2016). Today’s digital environment has transformed the business landscape into an omnichannel world where organisations’ marketing strategies and investments are largely impacted by technological advancements (Kannan & Li, 2016; Kannan, Reinartz &Verhoef, 2016).

Unsurprisingly, online advertising has become an important focus for marketers. In the Netherlands alone, online advertising expenditures have surged to reach €954 million in H1 2017, a 13% growth compared to €843 million in the first half of 2016 (IAB & Deloitte, 2017). The biggest growth is notable in the mobile search and display advertisement

segments, counting an ad revenue increase of +32% compared to H1 2016. For the first time in history, spending on mobile advertisement has surpassed spending on desktop

advertisement (IAB & Deloitte, 2017). In line with this trend, ad spending is expected to make up a 77.1% share of digital investments worldwide by 2021 (eMarketer, 2017a). This tremendous increase in digital and mobile ad spending is driven by the widespread adoption and usage of mobile devices. The smartphone society in the Netherlands alone has reached a penetration of 87% (vs. 81% globally) (Deloitte, 2016). Furthermore, a recent study conducted by Ipsos (2017) has revealed that consumers have become increasingly engaged with their mobile devices for a variety of everyday activities: in 2017, 92% of the surveyed mobile users indicated to use their smartphone while out shopping, 90% of all respondents used their smartphone during their leisure time and 81% of all respondents accessed their smartphone while eating in a restaurant (eMarketer, 2017b; Ipsos, 2017).

(9)

As people tend to be connected throughout the day and on-the-go, mobile devices have become an important channel in customer journeys (Lemon & Verhoef, 2016). In the past years, marketers have been able to tailor advertisements to consumers’ precise location and situation using consumers’ mobile devices (via their Global Positioning System (GPS), Wi-Fi connection, Bluetooth or via beacon application) (Grewal et al., 2016; Kannan and Li, 2016). This phenomenon, also known as ‘location-based advertising’ (LBA), bundles the strengths of online and direct marketing. LBA is particularly useful in situations where demand is sensitive to an advertisement’s timing (e.g., sending an Apple promotion in proximity of the Apple Store) (Fong, Fang & Luo, 2015).

With mobile devices becoming an essential component of everyday life, there has been a growing stream of research focused on the implications of mobile as a channel. Hence, many research projects have attempted to disentangle the implications that location-based mobile advertising (LBMA) has on various effectiveness outcomes. To date, the growing stream of research has reviewed the effect of LBA on outcome variables such as coupon redemption behaviour (Danaher et al., 2015), purchase rate (Fong, Fang & Luo, 2015) and attitude toward and intention to purchase product(s) (Bart, Stephen & Sarvary, 2014). Research has also looked into the positive relationship between consumers’ attitude towards LBA and their acceptance of LBA (Limpf & Voorveld, 2015). LBA has been found to be effective in competitive settings (Chen, Li & Sun, 2017; Danaher et al., 2015; Fong, Fang & Luo, 2015) and in different contexts such as physically crowded environments (Andrews et al., 2016). Furthermore, Kivetz & Zheng (2017) underpinned that location-based promotions are more effective for consumers with a hedonic consumption goal than a utilitarian

consumption goal.

Even though LBMA offers meaningful opportunities for businesses, the adoption rate of wearable devices is expected to take place much faster than the adoption rate of mobile

(10)

devices have in the past (Park & Skoric, 2017). The emergence and adoption of smartwatch devices has created a new territory for marketers: one allowing advertisements to be

increasingly targeted and instant (Chuah et al., 2016; Kumar & Gupta, 2016). In 2017, 452 million connected wearable devices were registered worldwide and this number is expected to double, connecting 929 million users by 2021 (Statista, 2017a). Smartwatch sales accounts for 52% of the global wearable market and the implications hereof are likely to impact the marketing field in the near future (Kannan & Li, 2016; Statista, 2017a). In Western Europe, smartwatch sales grew 33% in 2017 compared to 2016 and projections suggest that

smartwatch sales will continue to grow in 2018 (GFK, 2017; IDC, 2017). A recent study by Deloitte (2017) in the Netherlands found that 68% of the respondents owned at least one sort of connected device: a 21% increase compared to 2016. More specifically, 14% indicated to possess a smartwatch, which doubles the figures of 2016 (Deloitte, 2017). A research by Kentico (2016a) showed that 30% of the Dutch consumers expect to purchase a smartwatch device in the upcoming two years and that 71% of users (N = 547) indicate that they would opt-in to advertising on their smartwatch (Kentico, 2016b).

However, to the best of our knowledge, prior academic research has not adequately reviewed the impact that smartwatches and LBA could have when combined: a potential scenario in which consumers could be able to receive (personalised) advertisements in real-time via their smartwatch. In contrast to mobile devices that need to be handheld for the advertisement to be viewed by the recipient (Lee, Kim & Sundar, 2015), location-based smartwatch advertisement (LBSA) has the advantage of being delivered to a visible, wrist-worn device. Recent research by Chuah et al. (2016) underpins this, as their research shows that visibility is an important antecedent of smartwatch adoption. Furthermore, statistics show that 52% of the smartwatch users, use their smartwatch for notifications and texts (Statista, 2017b), implying the smartwatch can be seen as a visible extension of the smartphone.

(11)

The current state of the literature on smartwatches is scare yet very recent. Prior research in the field has mainly focused on the intention to adopt smartwatches (Chuah et al., 2016; Hong, Lin & Hsieh, 2017), usage motives (Lyons, 2015), usability patterns (Pizza et al., 2016; Schirra & Bentley, 2015) and user experience (Jeong H. et al., 2017). Across studies, consumer innovativeness is seen as an important personality trait that is related to the intention to adopt new IT products such as smartwatches (Agrawal & Prasad, 1998; Boateng, Okoe & Omane, 2015; Hong, Lin & Hsieh, 2017; Jeong C. et al., 2017) or new media sources (Midgley & Dowling, 1978; Rogers & Shoemaker, 1995). Furthermore, a longitudinal study found that moments on smartwatches were perceived as less disturbing than micro-moments on smartphones (Jeong H. et al., 2017). This argument gives ground to further explore the concept of LBSA.

The objective of this paper is to address the lack of research with regards to

advertising on wearable devices and more specifically on smartwatches. Given the nascent nature of LBMA and smartwatch research, there are no studies to date that have reviewed the potential of displaying location-based ads on smartwatch devices. This study thus reviews the effect of LBSA on consumers’ purchase intention and the determinants that lead towards purchase intention. Furthermore, previous studies do not explore if all types of shoppers (i.e., hedonic vs. utilitarian shoppers) are ready to receive advertisements on their writs. As such, this research paper addresses the following research question: “How does consumption goal

moderate the indirect effect of consumer innovativeness in IT on purchase intention through attitude towards LBSA?”

It is noteworthy that given the still relatively small population of smartwatch users (14%) (Deloitte, 2017; Kentico, 2016a) in the Netherlands, this study will focus on the Dutch consumer that is in the possession of a smartphone. A justification for this sampling frame is the exploratory nature of this study in which we aim to provide an overview of the potential

(12)

of LBSA. An important assumption is that the possession of a smartphone is needed in order to connect a smartwatch to mobile applications that are capable of displaying ads. In a scenario-based between-subject experimental design, we manipulate consumers’

consumption goal in order to research if hedonic or utilitarian shoppers are more likely to purchase products that were presented to them by means of LBSA.

To the best of our knowledge, this is the first study to combine smartwatch

technology and LBA into one study. Hence, this paper has several theoretical and practical contributions. First, this study deepens the understanding of LBA by reviewing the impact that LBA could potentially have if displayed on a wrist-worn device. Second, it provides useful managerial insights for devising new forms of advertisements. Our findings give marketers an all-encompassing view of the potential of LBSA. More specifically, this study helps marketers understand who is interested in LBSA (i.e., the corresponding level of consumer innovativeness in IT) and which type of shoppers to target (i.e., utilitarian or hedonic shoppers).

This study is structured as follows: in chapter two, we first thoroughly review the remainder of the work with regards to LBMA, smartwatch adoption and behaviour. We aggregate these two streams of research into one construct which we call LBSA. In chapter three, we provide an overview of the research method, the sample, the measurement of variables, the data preparation and data collection procedures. We devote chapter four to our results and we discuss how they relate to previous findings in chapter five. In chapter six, we review the implications of our results from a theoretical and practical perspective. Finally, we devote the final chapter to the limitations of this study and we thereby give several directions for future research.

(13)

2. Literature Review

This chapter provides a detailed theoretical background for the concept of LBSA and the accompanying hypotheses. In order to do so, we merge two streams of research that have not been combined in previous studies. We commence with the current literature about LBMA and review how consumers’ consumption goal moderates the relationship between attitude towards LBMA and the intention to purchase advertised products. We then discuss the most important literature regarding smartwatch technology and the role of consumer

innovativeness in the area of Information Technology (IT). Finally, we conclude this chapter by combining the two streams of research into one theoretical construct.

2.1 Location-Based Mobile Advertising 2.1.1 Definition and Evolution of LBMA

Technological developments have changed the evolution of advertising and the focus of marketers. In recent years, we have seen advertising become increasingly personalised, data-driven and real-time (Kumar & Gupta, 2016). As this shift continues, marketers are

increasingly using location-based applications to reach and target consumers (Andrews et al., 2016; Grewal et al., 2016). Location-based applications are driven by sensors that can define a consumer’s precise location using GPS, Wi-Fi, Bluetooth or via beacon application. This contextual information then in turn triggers the advertisement to be sent. Although academic definitions of LBMA are limited, the concept can be defined as “a marketing effort that uses location-tracking technologies to deliver geo-precise advertisements on a consumer’s mobile device” (Bauer & Strauss, 2016; Hühn et al., 2017; Ketelaar et al., 2017; Lee, Kim & Sundar, 2015; Van ‘t Riet et al., 2016).

Whereas location-based advertising (LBA) is not a new phenomenon per se, it has made room for new forms of location-based advertising such as LBMA. In the past, LBA was

(14)

(e.g., ‘turn left to MacDonald’s’). More recently, static billboards have disappeared as more dynamic signage has emerged, allowing marketers to change the content of their

advertisements (Bauer & Strauss, 2016). As technology has evolved and consumers have become increasingly connected to their mobile devices (Deloitte, 2016; eMarketer, 2017b; Ipsos, 2017), the nature of LBA has also changed. Firms can now decide on the format and the content of their advertisement. Furthermore, firms can decide if their advertisement is ‘static (e.g., push ad notification, SMS or display banner), dynamic (e.g., animated banner), and interactive (e.g., based on an iAd platform) and whether it contains embedded video elements’ (Grewal et al., 2016, p. 9). Depending on the organisation’s goal, marketers may present consumers with discounts (%), coupons or promotions such as ‘buy one, get one free’ (Grewal et al., 2016). These developments have given marketers the tools at hand to deliver ‘the right content in the right format to the right person at the right time’ (Tam & Ho, 2005, p. 271). We thus distinguish LBMA from LBA in two important manners:

§ LBMA focuses on micro moments where timing is key (Fong, Fang & Luo, 2015), whereas traditional LBA takes a more macro perspective that is less sensitive to timing (e.g., roadside billboards) (Bauer & Strauss, 2016; Fong, Fang & Luo, 2015). § LBMA has different format options (i.e., push and pull advertisements), whereas

traditional LBA methods are static (Grewal et al., 2016; Limpf & Voorveld, 2015).

2.1.2 Overview of the LBMA Literature

LBMA is a young field of research that has mainly been analysed from a technological perspective in refereed conference proceedings (Bauer & Strauss, 2016). A systematic literature review by Bauer & Strauss (2016) (N = 66) found twenty-nine (29) publications in refereed journals and thirty-seven (37) publications in refereed conference proceedings. Interestingly, the literature regarding LBMA has mainly focused on the exploration of

(15)

technological capabilities (32 publications), followed by user acceptance (11 publications) and privacy issues (8 publications). Only six (6) publications reviewed the effectiveness of LBMA by including the user in the loop (Bauer & Strauss, 2016).

In a field panel (N = 8,543) in a Western country, Danaher et al. (2015) researched factors that influence mobile coupon effectiveness. Amongst others, Danaher et al. (2015) present several findings that are particularly relevant to advancing the field of LBMA. First of all, prior redemption history was found to have a strong effect on redemption rate (b = .35, p < .01). This implies that consumers with prior experience with mobile promotions (“promo prone consumers”) have higher redemption rates. Furthermore, horizontal distance (< 100m) between the retailer and the consumer has proven to have a strong effect on redemption rate (b = .11, p < .01). An interesting finding is that if the distance to the

advertised retailer increases by 50 meters, the promotion discount will need to increase with 5% in order to obtain the same effect. Danaher et al. (2015) experimented with different products (i.e., food and beverages, menswear and shoes) and found that promotions for low-value products such as food and beverages have stronger effects on redemption rates (b = .62, p < .01). This implies that the effect of LBA might also vary, depending on the product category.

Fong, Fang & Luo (2015) conducted two field panel experiments (N = 50,600) in focal and competitive locations with varying promotional discounts (low: 20%, mid: 40%, high: 60%). Fong, Fang & Luo (2015) found promotions to be more effective close to the advertised retailer. For promotions close to the retailer, promotional increase from low discount to medium discount (20% - 40%) substantially increased purchase intention (b = .51, p < .01). For promotions near competitors, promotional increase from medium discount to high discount (40% - 60%) increased purchase intention (b = .32, p < .01). Consistent with earlier findings (Danaher et al., 2015), this entails that LBMA needs to be of

(16)

higher promotional value when the distance to the advertised retailer increases, in order to yield similar effect sizes.

A valuable addition to the literature by Andrews et al. (2016) shows that physical crowdedness does not negatively impact the effect of LBMA on purchase intention. In a field experiment (N = 10,690) in a subway in Asia, consumers were presented with SMS

advertisements at different moments of the day. A score to crowdedness was given based on the time of the day and the number of people present in the subway. Physical crowdedness of 4.97 passengers/m2 (weekday 17:30 – 18:30) had the biggest positive effect on purchase intention. When crowdedness peaks to 4.97 passengers/m2 (baseline: 1.96 passengers/m2), purchase increases with 46.9%. This finding supports the work of Danaher et al. (2015) who also find that time of the day has an effect on coupon effectiveness (b = .10, p < .01).

However, caution must be given to the results as perceptions of physical crowdedness may differ. Whereas Andrews et al. (2016) give a score to crowdedness based on actual

crowdedness in a subway in Tokyo, cultural differences may have an effect on how crowdedness is perceived (Lewis, 2005).

In a large field experiment (N = 39,946), Bart, Stephen & Sarvary (2014) researched which product types were most suited for Mobile Display Advertising (MDA). In order to do so, the researchers made a distinction between hedonic vs. utilitarian products. The former consists of products that stimulate a sensory experience, whereas the latter involves the consumer’s cognition. Bart, Stephen & Sarvary (2014) also made a distinction between low vs. high involvement products. An analysis of fifty-four (54) MDA campaigns showed that high-involvement utilitarian campaigns (e.g., technology or health) had positive and significant effects on attitude and purchase intention (attitude: b = .24, p < .01; purchase intention: b = .34, p < .01). The researchers state that this explanation may be due to this product category involving the central peripheral processing route. Caution must be given to

(17)

the results of this study as sample sizes varied considerably per product category (min: 3,513 and max: 14,160) and the researchers did not review what consumers were seeking for (e.g., coincidently, consumers may have been interested to buy high involvement utilitarian products before being presented with the MDA). Furthermore, the covariate distance in relation to the advertised product was not explored. Nevertheless, the findings help the research field in understanding that the type of advertised products can have effects on attitudes towards promotions and purchase intention.

We summarise the key points from the LBMA literature as follows:

§ The field of LBMA is in need of more consumer-centred research (Bauer & Strauss, 2016).

§ Prior experience with a particular form of advertising is an important antecedent for the usage of it (Danaher et al., 2015).

§ Promotional timing, distance to the advertised retailer and advertised product type are covariates that influence the effect of LBMA (Andrews et al., 2016; Bart, Stephen & Sarvary, 2014; Danaher et al. 2015; Fong, Fang & Luo, 2015).

2.1.3 Attitude towards LB(M)A as a Determinant of Purchase Intention

In the literature related to LBMA, the effectiveness of location-based mobile ads is often measured in terms of an increase in consumers’ purchase intention rather than the actual purchase behaviour itself (Agrebi & Jallais, 2015; Bart, Stephen & Sarvary, 2014; Goldfarb & Tucker, 2011; Grewal et al., 2016; Martins et al., 2018). Furthermore, ‘increase purchase intention’ and ‘increase favourable attitude’ are often seen as objectives for measuring the effectiveness of advertising campaigns (Bart, Stephen & Sarvary, 2014; Kumar & Gupta, 2016).

As the success of (a new) advertising format largely depends on users’ acceptance of it, the Technology Acceptance Model (TAM) (Davis, Bagozzi & Warshaw, 1989) and the

(18)

Theory of Reasoned Action (TRA) (Ajzen & Fishbein, 1980) have often been used to examine the relationship between attitude and behaviour (Agrebi & Jallais, 2015; Limpf & Voorveld, 2015; Lin & Kim, 2016; Muk, 2003; Taylor & Todd, 1995; Van der Heijden, Verhagen & Creemers, 2003).

Introduced by Davis in 1986, the TAM explains users’ adoption behaviour of computer information systems. According to the TAM, perceived usefulness and perceived ease of use are two cognitive belief dimensions that determine a user’s attitude toward using the IT product, the user’s behavioural intention to use the IT product and the user’s actual use of it (Chuah et al., 2016; Davis, Bagozzi & Warshaw,1989) (see figure 1).

Figure 1: Technology Acceptance Model (Davis, Bagozzi & Warshaw 1989, p. 985)

Similarly, the TRA also posits that behavior follows from a favourable attitude towards the product (Limpf & Voorveld, 2015; Ajzen & Fishbein, 1980).

Due to ‘its robustness, flexibility and explanatory strength’ (Agrebi & Jallais, 2015, p. 16), various components of the TAM and the TRA have been used to test the effects of different types and formats of advertising. Limpf & Voorveld (2015) reviewed consumers’ attitude towards LBA and their intention to accept it. The researchers found that the more consumers have a positive attitude towards LBA, the more likely they are to accept it (b = .69, p < .01). Ha, Park & Lee (2014) found similar results: attitude towards advertising has a positive effect on purchase intention (b = .34, p < .01) (Ha, Park & Lee, 2014).

(19)

In a similar study about consumers’ intentions to opt in to wireless advertising, Muk (2003) found that favourable attitudes towards wireless ads were positively related to the intention to opt in to wireless ads. More specifically, Muk researched cross-national differences and found stronger relationships between attitude and intentions to opt in to wireless ads for Korean consumers (β = .55, p < .01) than for American consumers (β = .46, p < .01). In a related study, Lin and Kim (2016) reviewed consumers’ attitude

towards LB(M)A and their intention to buy (rather than to accept) the advertised product. The study by Lin and Kim (2016) on user response to sponsored Facebook advertising also used the TAM to find that favourable attitude towards Facebook ads had a positive relationship with product purchase intention (β = .38, p < .01).

The TAM and the TRA have the ability to explain the adoption of various types of advertisements (Limpf & Voorveld, 2015; Lin & Kim, 2016; Muk, 2003) and the adoption of smartwatch products (Chuah et al., 2016; Kim & Shin, 2015). Therefore, we consider the TAM to be a valid typology to explain the relationship between attitude towards LBSA and a behavioural outcome (i.e., purchase intention). The more consumers have a favourable attitude towards LBSA, the more likely they are to accept the LBSA and thereby purchase from it. We thus hypothesise:

H1: Attitude towards LBSA is directly and positively related to purchase intention.

2.1.4 Consumer Consumption Goal as a Moderator

Advertising has become more personal, instant and location-based. Furthermore, mobile adoption has given marketers new means by which they can display advertisements (Grewal et al., 2016; Kumar & Gupta, 2016). This together has caused advertisements to become increasingly invasive (e.g., by means of display banners or push notifications).

(20)

In a factorial design (N = 379), Edwards, Lee & Li (2002) found that ads were perceived as less intrusive when they were related to the participant’s task than when they weren’t (t = 4.04, p < .01). This finding suggests that the intended effect of advertisements will largely depend on the consumer’s task or goal. A side-note must be made to the findings of Edwards, Lee & Li (2002) as participant task was extrinsically insinuated in a hypothetical situation. However, a more recent study by Bleier & Eisenbeiss (2015) revealed consistent findings. In a field experiment (N = 38,501), Bleier & Eisenbeiss (2015) researched what drives consumers to click on online advertisement banners. The researchers found that the displayed ad needs to match with the consumer’s goal. As such, the results show that goal-directed consumers have higher tendencies to click away from display advertisements that do not match their goal, even when advertisements are personalised (b = -.05, p < .05).

The above-mentioned findings align with Cho & Cheon (2004) who claim that ’when ads interrupt a consumer’s goal, it may result in undesirable outcomes, such as aggravation, negative attitudes, ad avoidance and lower purchase intentions’ (Cho & Cheon, 2004, p. 90). Kumar & Gupta (2016) also suggest that advertisements need to be aligned with consumers’ consumption goal.

Kivetz & Zheng (2017) had a closer look at consumers’ consumption goal and identified two types of shoppers: hedonic shoppers (i.e., fun and sensations seekers) and utilitarian shoppers (i.e., functional and practical seekers). In six (6) different experimental studies (N = 1,056), Kivetz & Zheng (2017) manipulated consumption goal (hedonic vs utilitarian), product framing (more vs. less hedonic) and promotion (price promotion vs. no price promotion). Two (2) studies (study 2 and study 3) focused on the interaction between consumption goal and promotion on purchase intention. As expected, promotions had a positive effect on purchase intention for hedonic consumption goals (M = 2.89, SD = 1.71,

(21)

F(1,111) = 9.91, p = .002). This effect was not observed for utilitarian consumption goals (M = 3.77, SD = 2.01, F(1,111) = 0.094, p = .759).

The arguments and logic above also apply to LBSA as this form of advertising will disturb consumers from what they are doing during their shopping experience. Hence, for LBSA to be effective, marketers will rigorously need to think about which consumers to target (Bleier & Eisenbeiss, 2015; Cho & Cheon, 2004; Edwards, Li & Lee, 2002; Kivetz & Zheng, 2017). In line with the findings presented by Kivetz & Zheng (2017), we hypothesise:

H2: The relationship between attitude towards LBSA and purchase intention is moderated by

consumption goal, such that this relationship is stronger for advertisements that are presented to consumers that have a hedonic shopping rather than a utilitarian consumption goal.

2.1.5 Regulations & Privacy Concerns

LBMA is subject to legal restrictions in most European countries and more specifically in the Netherlands. ‘In the EU, the two major governing pieces of legislation on personal data protection are the Data Protection Directive (95/ 46/EC) and the Privacy Directive

(2002/58/EC, as revised in Directive 2009/136/EC) of the European Parliament and Council’ (Cheung, 2014, p. 48). These legislations require advertisers to have prior consent from consumers before sending targeted advertisements (i.e., so-called ‘push advertisements’). In addition, consumers must be able to change or withdraw their consent at any point in time (Cheung, 2014; European Commission, 2018; Grewal et al., 2016). As of May 2018, the European Union's new data protection law has given consumers more control over their personal data and the ability ‘to be forgotten’ (i.e., requiring organisations to delete all digital traces). Furthermore, significant penalties and fines have been established for those

businesses that do not process consumer data according to the latest legislations (European Commission, 2018).

(22)

Unsurprisingly, concerns with regards to consumer privacy have become important topics that are related to LBMA (Grewal et al. 2016; Hühn et al. 2017; Ng & Wakenshaw 2017). Hühn et al. (2017) more specifically researched the impact of location congruence on ad intrusiveness. Although Hühn et al. (2017) did not find a difference between the impact of congruent (β = .03, p > .05) and non-congruent advertisements (β = .09, p > .05) on ad intrusiveness, congruent ads were found to be more valuable (β = .45, p < .01) and relevant (β = .46, p < .01) to consumers (Hühn et al., 2017). This finding stands in conflict with earlier findings presented by Hühn et al. (2012) and Lee et al. (2015) that do show that location-incongruent ads have a negative effect on perceived ad intrusiveness. A possible explanation for these different findings comes from the chosen research design and how ad intrusiveness was measured. Whereas Hühn et al. (2012) and Lee et al. (2015) conducted an experiment with a simulated smartphone and location (in)congruent advertisement, Hühn et al. (2017) made use of the Experience Sampling Method (ESM). In their ESM, Hühn et al. (2017) did not control for extraneous variables that could have influenced the results (e.g., their sample of students may have been busy chatting to friends when the location incongruent ad was displayed to them) (Hühn et al., 2017).

Despite the fact that data privacy is a topic of growing importance, privacy issues lie beyond the scope of this study for two reasons:

§ New changes in legislations have only been effective since May 2018, which was after the data collection period of this study.

§ We suggest that businesses and consumers require time to embrace the new legislation. Furthermore, we believe consumers’ attitude toward their privacy will change over time and we therefore strongly suggest future research in this area (see chapter 7 for areas of future research).

(23)

2.2 Smartwatch Technology

2.2.1 Definition and Evolution of Smartwatch Technology

Understanding technological developments and their implications has been an important subject discussed by previous streams of marketing research (Kannan & Li, 2016; Lamberton & Stephen, 2016). As digital developments have affected consumers’ lives, so have they changed how business is done. Technology and digital platforms have created new customer journeys and have enabled businesses to reach, inform, engage, sell and retain customers in new ways (Lamberton & Stephen, 2016; Lemon & Verhoef, 2016). Furthermore, as digital technologies become more focused on the individual and wearable, location and geography have become important predictors of consumer behaviour (Kannan & Li, 2016).

‘Wearable technology’ is a broad concept that has often been researched beyond the scope of this study (e.g., in fitness or in healthcare) (Wright & Keith, 2014). Therefore, we refine the scope of this study to smartwatch technology. By doing so, we create a clear distinction from related technologies that will not be considered within this study. The academic literature lacks a clear definition of smartwatches. Hence, we build and extend upon the definition presented by Chuah et al. (2016): ‘a smartwatch is a small, wearable device that is worn like a traditional watch but allows for GPS tracking and the installation and use of applications’ (Chuah et al., 2016, p. 277).

Developments related to smartwatch technology date back to 1972. Back then, the launch of the Hamilton Pulsar P1 marked the first digital watch in history (Charlton, 2013). However, it wasn’t until the early 1980’s that Seiko launched the Pulsar NL C01, capable of doing more than simply telling the user the date and time (Charlton, 2013; Kim & Shin, 2015). From there, Seiko continued developing and updating its product range and launched the Seiko UC-2000 in 1984, a smartwatch that was able to store and display text that was transferred via a keyboard. The Seiko UC-2000 was able to store up to 2,000 characters on a

(24)

4x10-character display. Launched in the same year, the Seiko RC-1000 was able to connect to a number of computers and store data such as telephone numbers. This watch marked the beginning of a series of smartwatch advances, including the launch of WatchPad by IBM and Citizen. The WatchPad ran on the Linux operating system, a 32-bit ARM processor and had eight megabytes of RAM and 16 megabytes of flash memory, along with a speaker and microphone (Charlton, 2013; Kim & Shin, 2015). From thereon, the number of smartwatch functionalities picked-up at an even faster pace: in 2003 IBM enabled smartwatches to play simple games, change displays and host applications. By then, smartwatches ran on wireless connectivity by using FM broadcast signals. Flash-forward to 2018 and smartwatches are now capable of just about anything including developments such as glucose monitoring features (Acharya, 2017). Today, there is a growing list of smartwatches that is compatible with mobile devices. These devices are equipped with a larger touchscreen than traditional wearable devices, have an Operating System (OS), GPS signal and an application store. Examples of popular smartwatches include the Apple Watch, LG Watch Style, Samsung Gear or Fitbit Ionic (Techradar, 2018).

Although the gap in terms of functionalities between smartphones and smartwatches minimises, smartwatches are not expected to replace smartphones. Rather smartwatches are expected to serve as an extension of paired smartphones. Smartwatches and smartphones can connect via Bluetooth whereby smartwatches can display information from a smartphone in a faster and more condensed and visible manner (Chuah et al., 2016; Kim & Shin, 2015). There are two distinct features that make smartwatches particularly suitable for LBSA:

§ Smartwatches are wrist worn devices that figure as an extension of the smartphone. Unlike smartphones, smartwatches do not have to be handheld in order to view a notification or advertisement.

(25)

2.2.2 Overview of the Smartwatch Literature

The literature related to smartwatches is still in its infancy, yet the body of available research has vastly expanded. Recent studies and conference proceedings are mainly related to

adoption intention, usage motives, usability patterns and user experience. Given the favourable statistics and projections related to smartwatch sales (GFK, 2017; IDC, 2017; Kentico, 2016a), researchers have attempted to disentangle what drives people to adopt smartwatches (Chuah et al., 2016; Hong, Lin & Hsieh, 2017). In a survey design study (N = 226), Chuah et al. (2016) found a main effect of perceived usefulness (β = .46, p < .01) and visibility (β = .29, p < .05) on attitude towards smartwatches and adoption intention (β = .50, p < .01). This entails that consumers adopt smartwatches for two reasons: (1) due to technological abilities and (2) as a fashion statement. Hong, Lin and Hsieh (2017) had a closer look at who intend to adopt smartwatches. In a survey design study (N = 276), the researchers reviewed how smartwatch adoption hinges on a consumer’s hedonic or utilitarian value and personal innovativeness. Hong, Lin and Hsieh (2016) made a distinction between two types of consumers: hedonic value seekers (i.e., those who seek for fun and enjoyment) and utilitarian value seekers (i.e., those who seek for learning). Their findings show that consumer innovativeness has a positive effect on hedonic value (β = .49, p < .01) and on utilitarian value (β = .38, p < .01), which in turn has a positive effect on adoption intention (hedonic: β = .38, p < .01; utilitarian β = .50, p < .01). The results suggest that ‘the more innovative the consumers, the stronger their hedonic/utilitarian attitudes become with regard to the latest products or services’ (Hong, Lin & Hsieh, 2017, p. 270). Although Chuah et al. (2016) and Hong, Lin and Hsieh (2017) use different determinants of adoption intention, a parallel between both studies can be drawn. Consumers seeking for smartwatch usability (Chuah et al., 2016) closely relate to utilitarian value seekers (Hong, Lin and Hsieh, 2016), whereas consumers seeking for smartwatch visibility (Chuah et al., 2016) more closely relate

(26)

to hedonic value seekers (Hong, Lin & Hsieh, 2017). These findings suggest that there are important differences between consumers that affect the success of smartwatch sales.

In recent years, many conference proceedings have looked into the behavioural aspect of smartwatch technology. Although results are often based on small sample sizes, the

findings help understand what drives consumers to use smartwatches. A closer look into usage motives (Lyons, 2015) shows that consumers (N = 50) value the ability of

smartwatches to provide quicker access to notifications, text messages and incoming calls. Twenty-one (21) respondents mentioned that smartwatches would save them time as they wouldn’t have to get their phone out and unlock it. This finding was also found in an experimental study by Pizza et al. (2016), in which usage patterns were analysed (N = 22). The results of their experiment show that checking notifications was the second most used smartwatch feature after checking the time. 16.8% of all daily smartwatch interactions were devoted to checking notifications (49.6% for checking the time). Across studies we see that the smartwatch often figures as ‘an extension to the smartphone’ (Schirra & Bentley, 2015).

In a longitudinal study using the Apple Watch (N = 50), Jeong H. et al. (2017) quantified wearing behaviours. The results show that on average, participants wore the smartwatch 10.48 hours per day (weekdays: 8.66 hours/day; weekend: 11.32 hours/day). Based on wearing behaviour, Jeong H. et al. (2017) classified users into three groups: (1) the work-hour wearer group (N = 29, 9AM – 6 PM), (2) the active-hour wearer group (N = 15, 9AM – 9 PM) and (3) the all-day wearer group (N = 6) that also wears their smartwatch during their sleep. Workplace and classroom (4.57) followed by restaurant and café (4.51) had the highest scores in terms of wearing frequency (1 = never wear; 5 = always wear). In line with Lyons (2015), Pizza et al. (2016) and Schirra & Bentley (2015), notification checking was mentioned as one of the most important smartwatch functionalities.

(27)

Furthermore, Jeong C. et al. (2017) found that respondents perceived

micro-interactions less disturbing on their smartwatch than on their smartphone. We summarise the key points from the smartwatch literature and conference proceedings as follows:

§ There is a distinction to be made between two types of consumers: hedonic value seekers and utilitarian value seekers (Chuah et al., 2016; Hong, Lin & Hsieh, 2017). § Consumer innovativeness is seen as an important trait in relation to smartwatch

adoption (Hong, Lin & Hsieh, 2017).

§ The smartwatch figures as an extension of the smartphone, allowing quicker access to notifications (Jeong C. et al, 2017; Lyons, 2015; Pizza et al., 2016; Schirra & Bentley, 2015).

2.2.3 Consumer Innovativeness in IT as a Determinant of Attitude towards LBSA In recent studies related to technology, personal innovativeness has been seen as an important determinant of attitude towards advertisements. According to Agrawal and Prasad (1998), consumer innovativeness in IT can be defined as ‘the willingness of an individual to try out any new information technology’ (Agrawal & Prasad, 1998, p. 206). Agrawal and Prasad (1998) argue that consumer innovativeness is an important individual difference variable that helps understand how attitude and behavioural intention is formed. Although there is

theoretical and empirical support for its role in innovation adoption (Midgley & Dowling, 1978; Rogers & Shoemaker, 1995), consumer innovativeness has not yet been included in technology acceptance models (Agrawal & Prasad, 1998). As such, with the introduction of personal innovativeness in IT, Agrawal and Prasad (1998) have added a new construct to the TAM and TRA (Ajzen & Fishbein, 1980; Davis, Bagozzi & Warshaw, 1979). Ever since, multiple researchers have used consumer innovativeness in advertising and in technology studies (Boateng, Okoe & Omane, 2015; Hong, Lin & Hsieh, 2017; Jeong C. et al., 2017). In

(28)

the next paragraphs, we highlight two studies in particular that are related to advertising (Boateng, Okoe & Omane, 2015) and to wearable technology (Jeong C. et al., 2017).

In a survey research design (N = 473), Boateng, Okoe and Omane (2015) found a ‘positive relation between personal innovativeness and attitudes towards mobile advertising’ (β = .37, p < .01)(Boateng, Okoe & Omane, 2015, p. 2017). The findings suggest that marketers must be selective in deciding who to target for their (mobile) advertising campaigns. Innovative consumers will develop more positive attitudes towards (mobile) advertising whereas less innovative consumers will have less favourable attitudes towards (mobile) advertising.

Jeong C. et al. (2017) make a distinction between two types of consumer innovativeness: Product-Possessing Innovativeness (PPI) and Information-Possessing Innovativeness (IPI). Whereas the former relates to the possession of new IT products, the latter relates to a consumer’s willingness to seek out for information on new products earlier than others (Jeong C. et al., 2017). In a survey design among Korean consumers (N = 312), Jeong C. et al. (2017) researched the concepts of PPI and IPI in relation to wearable devices. The researchers found a stronger effect of IPI (t = .45, p < .01) on adoption factors than PPI (t = .01, p < .05). Consumers that scored high on information processing innovativeness, had a more favourable image of wearable devices (t = .45, p < .01) which in turn lead to higher purchase intention (t = .21, p < .01).

The above-mentioned examples give ground to using consumer innovativeness in the area of IT as a determinant of attitude towards LBSA. We therefore hypothesise that:

H3: Consumer innovativeness in the area of IT is directly and positively related to attitude

(29)

2.3 A New Concept: Location-Based Smartwatch Advertising 2.3.1 Definition of LBSA

The previous section has introduced the concept of LBMA and the opportunities smartwatch devices have brought along. By bundling the strengths of both concepts, we propose a new view to the marketing field, namely “location-based smartwatch advertising”. As the concept has never been reviewed in academic literature before, we propose the following definition: “Location-based smartwatch advertising makes use of the Bluetooth connection between smartwatch and smartphone in order to deliver geo-precise advertisements to consumers’ wrists.” LBSA is particularly suitable to the field of advertising due to its geographic accuracy (Chuah et al., 2016; Fong, Fang & Luo, 2015; Kim & Shin, 2015), instant nature (Bauer & Strauss, 2016) and daily usage patterns of smartwatches users (Jeong C. et al, 2017; Lyons, 2015; Pizza et al., 2016; Schirra & Bentley, 2015).

2.3.2 Combining Constructs from Two Streams of Research

In order to advance the field of advertising, we deem it particularly useful to explore the potential of LBSA. Hence, (1) we question if there is a market for LBSA, and (2) if there is, who is willing to buy from smartwatch advertising?

We combine the following two constructs in order to provide an answer:

§ Consumer innovativeness in the area of IT – Attitude towards LBSA: derived from the smartwatch literature (see section 2.2).

§ Attitude towards LBSA – Purchase intention and the moderating role of

(30)

2.4 Theoretical Framework

The theoretical framework of this study was devised by combining constructs from the field of LBMA and the field of smartwatch technology:

Figure 2: Theoretical Framework

§ Consumer innovativeness in the area of IT – Attitude towards LBSA. This hypothesised relationship is based on the importance of having an individual difference variable. This difference variable adds to the ‘attitude – behaviour’ construct as depicted by the TAM and TRA (Ajzen & Fishbein, 1980; Davis,

Bagozzi & Warshaw, 1979). The proposed construct is particularly useful as it helps explain individual differences towards advertisements (Agrawal & Prasad, 1998; Boateng Okoe & Omane, 2015; Jeong C. et al. 2017; Rogers & Schoemaker, 1995; Midgley & Dowling, 1978).

§ Attitude towards LBSA – Purchase intention. This construct is based on the TAM (Davis, Bagozzi & Warshaw, 1979) and the TRA (Ajzen & Fishbein, 1980) and is particularly useful due to the applicability of this construct in studies related to new (online and mobile) advertising formats (Limpf & Voorveld, 2015; Lin & Kim, 2016; Muk, 2003) and in studies related to smartwatch adoption (Chuah et al., 2016; Kim & Shin, 2015). Consumer innovativeness in the area of IT Attitude towards LBSA Purchase intention Consumption goal (hedonic vs. utilitarian) H3 (+) H1 (+) H2 (+)

(31)

§ Consumption goal as a moderator. This construct is based on earlier findings that suggest that advertisements need to match a consumer’s goal and that hedonic shoppers have more favourable behaviours when presented with advertisements (Bleier & Eisenbeiss, 2015; Cho & Cheon, 2004; Edwards, Lee & Li, 2002; Kivetz & Zheng, 2017; Kumar & Gupta, 2016).

(32)

3. Method

This chapter provides a description of the research design, the chosen sample and the data collection method used in this study. In the following section, we provide a detailed overview of the preparatory steps that were undertaken for our data analyses, the measurement

variables and the statistical procedures that were used. 3.1 Experimental Design

The aim of this study is to explore the concept of LBSA and the moderating role of

consumption goal. In order to test the hypotheses, an online between-subjects experiment was conducted. This method was chosen as it is particularly suitable for exploratory research in which the manipulation of an independent variable (i.e., consumption goal) causes a change in another dependent variable (i.e., purchase intention) (Field, 2015; Saunders, Lewis & Thornhill, 2016). In the chosen research design, at t = 1, all participants were presented with the items measuring consumer innovativeness in IT, attitude towards LBSA and purchase intention. At t = 2, participants were randomly assigned to one of the two experimental conditions (hedonic condition N = 81; utilitarian condition N = 82) or to the control condition (N = 82). Participants were then presented with one of the three shopping scenarios (see appendix 9.1), followed by a smartwatch advertisement. The use of scenarios has been favoured by numerous researchers as it allows ‘researchers to study the emerging phenomenon without being constrained by timing of the study or the state-of-the art technology’ (Sheng, Nah, & Siau, 2008, p. 63). Given the limited financial resources and time constraints in this study, a scenario-based method was also applied to this research design. A manipulation check was included in the study to test whether the chosen hedonic and utilitarian scenarios were effective in manipulating the experimental condition (Limpf & Voorveld, 2015). After being presented with one of the three scenarios, participants were presented with the same purchase intention construct as in t =1. However, participants were

(33)

asked to answer the five items based on the scenario that had been presented to the participant (see appendix 9.2 for the full questionnaire).

3.2 Sample

Given the relatively small population of smartwatch users in the Netherlands (14%) (Kentico, 2016a), we recruited Dutch consumers in the possession of a smartphone (N = 275). A

justification for this sampling frame is the assumption that a smartphone is the prerequisite needed in order to connect to a smartwatch capable of displaying advertisements. The Dutch population is particularly relevant to study as previous findings by Kentico (2016a) have shown that Dutch consumers are likely to adopt smartwatches in the upcoming two years. More specifically, 30% of the Dutch consumers expect to purchase a smartwatch device in the upcoming two years (Kentico, 2016a) and 71% of smartphone users (including users from the Netherlands, the UK, France and Germany) indicate that they would opt-in to smartwatch ads (Kentico, 2016b). Hence a survey was administered amongst smartphone users in the Netherlands. The survey was administered in both English and Dutch, enabling participants to choose the language of their preference (see appendix 9.2). The original scale items were translated from English to Dutch and checked by a professional translator.

The survey was filled out by 275 participants and fully completed by 245 participants. Missing cases were excluded listwise (Field, 2015). This sample size seems sufficient to yield generalisable results (Field, 2015; Saunders, Lewis & Thornhill, 2016). Furthermore, the sample size falls into the sample size range of similar advertising studies. To be more

specific, an analysis of 41 between-subjects designs with 3 treatments shows that on average, the sample size consisted of 77 participants per treatment (min: 18.7, max: 420.3) (Bellemare, Bissonnette & Kröger, 2014). Hence the sample size of this study exceeds the one of similar studies (N = 245 > N = 231) and thus seems appropriate.

(34)

Participants were recruited through non-probability convenience sampling. This method was chosen due to its ease, low costs and time effectiveness (Saunders, Lewis & Thornhill, 2016). The link to the online questionnaire was shared via LinkedIn (2 posts) and Facebook (1 post).

A further analysis of the sample (N = 245) shows that 51.4% of the participants were male and 48.6% were female. The majority of participants were between the age of 25 and 34 years (M = 2.54, SD = 1.28). Furthermore, the sample included a broad range of educational backgrounds. 9.9% of respondents had completed primary and/or secondary education (primary education: 0.5%, secondary education: 9.5%). 38.8% of respondents had completed an educational programme at a university of applied sciences (secondary vocational

education: 7.7%, higher professional education: 31.1%). The largest majority of respondents, namely 51.3%, had completed an education at a research university (university bachelor: 27.1%, master or doctorate: 24.2%). A control question (“do you live in the Netherlands?”) was included in the survey to monitor the chosen sampling frame. An analysis of the sample shows that indeed 100% of the respondents lived in the Netherlands.

A closer analysis of the respondents’ smartphone and smartwatch behaviour shows that 98.2% of the respondents owned a smartphone and 99.6% hereof regularly used their smartphone to access the Internet. More specifically, 69.4% of respondents indicated to be heavy smartphone users and accessed their smartphone more than 10 times a day (less than once a day: 0.7%, between 3-4 times: 1.9%, between 5-6 times: 3.0%, between 7-8 times: 9.3%, between 9-10 times: 15.7%). Furthermore, 59.7% of the respondents indicated to have been using a smartphone for more than 8 years (less than one year: 0.4%, between 2-3 years: 0.7%, between 3-5 years: 5.6%, between 5-8 years: 33.6%). In terms of smartwatch

behaviour, 26.7% of the respondents indicated to own a smartwatch. With 61.1% of the smartwatch users owning an Apple Watch, this watch was the most preferred choice by

(35)

respondents. 13.9% of the smartwatch users owned a Fossil Q and 12.5% a Samsung Gear. Intentions to buy a smartwatch varied: 10.5% of the respondents indicated to intend to buy a smartwatch in the upcoming year, 37.0% were undecided (“maybe”) and 52.5% of the respondents did not intend to buy a smartwatch in the upcoming year.

(36)

3.3 Measurement of Variables

All items used in this study were derived from validated scales (a > .80) in the literature related to location-based smartwatch or mobile advertising. The scales were adapted where needed to suit the particular context of LBSA.

Consumer innovativeness in IT. Consumer innovativeness in IT was measured by

combining the original scale from Agrawal & Prasad (1998) (a = .84) and the original scale from Jeong C. et al. (2017) (a = .93). Although the scale by Agrawal & Prasad dates back ten (10) years, the scale is particularly relevant as it has been used in recent publications related to new forms of (mobile) advertising (Boateng, Okoe & Omane, 2015). In line with the research by Jeong C. et al. (2017), consumer innovativeness in IT was reviewed by combining items that relate to Product Possession Innovativeness (PPI) and Information Possessing Innovativeness (IPI). This choice was made as the combination of PPI and IPI gives a well-rounded view of respondents’ level of innovativeness in IT (Jeong C. et al., 2017). Four (4) original scale items from Agrawal & Prasad (1998) and six (6) original items from Jeong C. et al. (2017) were used. One (1) item was reverse coded, implying that a low score represented a high level of consumer innovativeness: “in general, I am hesitant to try out new information technologies” (Agrawal & Prasad, 1998, p. 210). The item was reverse coded in order to detect acquiescence bias (Saunders, Lewis & Thornhill, 2016). Respondents were asked to indicate their agreement with regards to the ten (10) statements on a 7-point Likert scale ranging from 1: strongly disagree to 7: strongly agree.

Attitude towards location-based smartwatch advertising. Attitude towards LBSA was

measured by adapting the original scale from Taylor & Todd (1995) (a = .90). This scale builds on the TAM (Davis, Bagozzi & Warshaw, 1989) and has been used in recent research related to location-based mobile advertising (Limpf & Voorveld, 2015). The four (4) items were adapted to suit the context of LBSA. Respondents were asked to indicate their

(37)

agreement with regards to the four (4) statements on a 7-point Likert scale ranging from 1:

strongly disagree to 7: strongly agree. In addition, respondents were asked for their

likelihood to opt-out of LBSA statements on a 7-point Likert scale ranging from 1: strongly

disagree to 7: strongly agree. As such, respondents were asked to indicate their level of

agreement with “I would opt-out of location-based smartwatch advertising”. The item was reverse coded, implying that a high level of agreement with the statement indicated a low level of attitude towards LBSA. The item was reverse coded in order to detect acquiescence bias (Saunders, Lewis & Thornhill, 2016).

Purchase intention. Purchase intention was measured using scales from Ha, Park &

Lee (2014) (a = .89) and Ko, Cho & Roberts (2005) (a = .95). These scales were chosen as both scales were used in advertising studies that measured the ‘attitude towards

advertisement – purchase intention’ construct. The scales were adapted to suit the context of LBSA. Purchase intention was measured using the same construct and formulation before and after the experimental manipulation. This was done in order to allocate any variance in the dependent variable to the manipulation (i.e., consumption goal: hedonic/utilitarian). Respondents were asked to indicate their agreement with regards to the five (5) statements on a 7-point Likert scale ranging from 1: strongly disagree to 7: strongly agree.

Manipulation Check. Two (2) manipulation checks were included in the survey in

order to measure if the scenarios were effective in manipulating the moderator (i.e., consumption goal) (Limpf & Voorveld, 2015). Respondents presented with the hedonic experimental condition were asked to indicate their level of agreement with regards to the following statement: “the goal of my shopping trip was to enjoy myself with my friends”. Respondents presented with the utilitarian experimental condition were asked to indicate their level of agreement with regards to the following statement: “the goal of my shopping trip was to accomplish a practical task”. Respondents were asked to indicate their agreement with

(38)

regards to the statement on a 7-point Likert scale ranging from 1: strongly disagree to 7:

strongly agree. Respondents that did not undergo an experimental treatment (i.e., the control

group) weren’t presented with the manipulation check question.

Smartphone Usage. In order to get a basic understanding of respondents’ smartphone

usage, four (4) additional questions related to smartphone behaviour were included. This was done as smartwatch usage is closely related to the possession of a smartphone (i.e.,

smartwatches need a smartphone connection) (Chuah et al., 2016; Kim & Shin, 2015). Hence smartphone behaviour can be an important indicator of a user’s (potential) smartwatch usage (Jeong C. et al, 2017; Lyons, 2015; Pizza et al., 2016; Schirra & Bentley, 2015).

Control Variables. In similar studies related to LBA, prior experience with

location-based advertisements has often been used as a control variable (Grewal et al., 2016; Limpf & Voorveld, 2015). The study by Limpf & Voorveld (2015) showed that there is a (weak) positive relationship between prior experience with LBA and attitude towards LBA (r =.15, N = 224, p = .027) and purchase intention (r = .15, N = 224, p = .027). We therefore deem it relevant to include prior experience with LBA as a control variable.

Demographics. Respondents were asked for their gender, age and level of education.

Respondents were asked to answer the following question: “do you live in the Netherlands?”. This binary variable was added in order to rule out any respondents from outside the

Netherlands (i.e., only respondents within the Netherlands were included in subsequent analyses).

(39)

3.4 Data Preparation

The following section elaborates on the preparatory steps that were undertaken before running the analyses and testing the hypotheses. Prior to the preparatory checks, two counter-indicative items in the survey were recoded (1. “in general, I am hesitant to try out new information technologies” and 2. ““I would opt-out of location-based smartwatch

advertising”). The two items were recoded as both statements were coded in the opposite direction: i.e., a high score on both statements represented a low score on consumer innovativeness in IT and a low score on attitude towards LBSA.

3.4.1 Testing for Normality

One of the assumptions related to our chosen statistical methods is that the dependent variable (i.e., purchase intention) should be approximately normally distributed for each group of the independent variable (i.e., hedonic, utilitarian and control group) (Field 2015). In order to satisfy the assumptions needed for the analyses, the dependent variable was tested for normality using the Kolmogorov-Smirnov test and the Shapiro-Wilk test (Laerd Statistics, 2018) (see table 1). The three conditions were reviewed separately as the research design aimed to create variance in purchase intention (t = 2) for the different groups.

N

Statistic Sig. Statistic Sig.

Hedonic group 81 .16 .00*** .92 .00*** Utilitarian group 82 .16 .03* .89 .02* Control group 82 .18 .00*** .90 .00*** Hedonic group 81 .17 .00*** .88 .00*** Utilitarian group 82 .10 .03* .95 .00*** Control group 82 .20 .00*** .88 .00*** Shapiro-Wilk

Table 1: Kolmogorov-Smirnov and Shapiro-Wilk Tests for Normality

Purchase intention ( t = 1)

Statistical significance: *p < .05; **p < .01; ***p < .001

(40)

The above-mentioned results indicate that purchase intention scores were not normally distributed for the hedonic, utilitarian and control groups as assessed by Shapiro-Wilk’s test (p < .05). However, in order to meet the assumptions of more “robust” tests (or non-parametric tests) that can deal with approximate normal distributions, normality was also tested by calculating the skewness z-score and kurtosis z-score at the .01 level (Laerd

Statistics, 2018) (see table 2).

Table 2 shows that the z-scores for skewness and kurtosis fall within the ±2.58 range (at the .01 level), except for the skewness z-score of purchase intention (t = 2) for the hedonic and control condition. In the particular case of this research, ±2.58 was used as an absolute value to determine the level skewness and kurtosis given our small sample size (N < 300) (Field, 2015; Laerd Statistics, 2018). Based on the skewness and kurtosis criteria, we presume our data to be approximately normally distributed. We take this into account in the selection of statistical tests that are “robust” to violation (Laerd Statistics, 2018).

N

Statistic Std. Error z-score Statistic Std. Error z-score

Hedonic group 81 -.23 .27 -.87 -1.32 .53 -2.50 Utilitarian group 82 -.42 .27 -1.59 -1.33 .53 -2.53 Control group 82 -.62 .27 -2.32 -.92 .53 -1.74 Hedonic group 81 -.81 .27 -3.01 -.40 .53 -0.75 Utilitarian group 82 .58 .27 2.17 -.33 .53 -0.63 Control group 82 -.83 .27 -3.12 -.52 .53 -1.74

Table 2: Skewness and Kurtosis Tests for Normality

Variable

Skewness Kurtosis

(41)

3.4.2 Cronbach’s Alpha

The reliability of the used scales was checked by calculating Cronbach’s a (see table 3). As presented in table 3, all scales have high reliability, with Cronbach’s a > .80. Furthermore, all data have item-total correlations above .30 and the output of the reliability test showed that the removal of items would not significantly improve the reliability of the scale (Field, 2015).

3.4.3 Principal Component Analysis

A principal component analysis (PCA) was run on the 20-item questionnaire that measured consumer innovativeness in IT, attitude towards LBSA and purchase intention (N = 245). The suitability of PCA was assessed prior to running the analysis (Laerd Statistics, 2018a). Inspection of the correlation matrix showed that all variables had at least one correlation coefficient greater than .3. The overall Kaiser-Meyer-Olkin (KMO) measure was .942 with individual KMO measures all greater than .866, a classification of ‘meritorious’ to

‘marvellous’ according to Kaiser (1974). Barlett’s Test of Sphericity was statistically significant (p < .01), indicating that the data was likely factorisable. The three-component solution explained 79.3% of the variance. A Varimax orthogonal rotation was employed and the rotated solution exhibited a ‘simple structure’ (i.e., no cross loadings exist) (Thurstone, 1947). The interpretability of the data was consistent with the variables the questionnaire aimed to measure (see table 4).

Cronbach's ⍺ .954 § Information Possessing Innovativeness .932

§ Product Possession Innovativeness .901

.949 Purchase intention .949 § Before manipulation (t = 1) .955 § After manipulation (t = 2) .948 Questionnaire item Consumer innovativeness in IT

Attitude towards LBSA

(42)

3.4.4 Manipulation Check

In order to check if the scenarios were effective in manipulating the moderator, namely a hedonic consumption goal versus a utilitarian consumption goal, the two manipulation check questions were analysed. Participants that were assigned to the hedonic condition were presented with the statement: “the goal of my shopping trip was to enjoy myself with my friends” (M = 6.44, SD = .742). Participants that were assigned to the utilitarian condition were presented with the statement: “the goal of my shopping trip was to accomplish a practical task” (M = 6.33, SD = .917). Based on the mean scores and standard deviations, we

Component 1 Component 2 Component 3

Item 1 .765 Item 2 .783 Item 3 .798 Item 4 .841 Item 5 .826 Item 6 .811 Item 7 .872 Item 8 .452 Item 9 .830 Item 10 .819 Item 1 .805 Item 2 .826 Item 3 .916 Item 4 .895 Item 5 .913 Item 1 .822 Item 2 .782 Item 3 .805 Item 4 .768 Item 5 .718

Table 4: Principal Component Analysis Questionnaire item

Consumer innovativeness in IT

Attitude towards LBSA

Note : N = 245. Extraction Method: Principal Component Analysis. Rotation Method:

Varimax with Kaiser Normalization. Purchase intention

Advertising targeted to the wrist : are consumers ready for smartwatch advertising?