1 Traditional brick-and-mortar retailers face increasing pressure from a changing consumer demographic and incredible growth of e-commerce (Martin, 2018). The impact has been large and has been felt by many kinds and sizes of retailers. Sears, once the largest retailer in the US, filed for bankruptcy in 2018, while Macy’s and J.C. Penney have each closed more than 100 stores. Victoria’s Secrets, Starbucks, Walgreens, and Best Buy are also closing stores. Several retailers such as Toys ‘R’ Us and Bon-Ton liquidated and closed all their stores. Payless ShoeSource, one of the largest and most successful family-owned shoe businesses in the US, is closing all 2100 of its US stores in 2019. In the US, the large numbers of store closures appear to be a trend. In 2017, traditional retailers closed more than 5000 stores; nearly 5000 more stores shut their doors in 2018, and 2019 has already surpassed that number (Clark.com, 2019). The trend of struggling retail organizations is not limited to the US. The store closure trend is also happening in UK retail with more than 1200 retailers collapsing in 2018 (Milligan, 2019). The rest of Europe has seen less impact so far, but as e-commerce grows it is expected that these same pressures will impact the retail sector in every developed economy (Martin, 2018).
2 This retail crisis has driven many retailers to consider the adoption of innovative technologies aimed at reducing costs, increasing transaction throughput, and improving consumer satisfaction (Demirci Orel, Kara, 2014). One such technology that retailers have embraced to address the growing competitive pressure is self-checkout systems (SCOS). These systems originally took the form of self-service kiosks where consumers would do their shopping, take their items to the kiosk, self-scan the items, and pay. Another self-checkout technology that has evolved is mobile self-checkout systems (MSCOS). In MSCOS consumers, utilizing a handheld device or their mobile phone, scan items as they move through the store and complete the transaction perhaps through a mobile payment service app (Andriulo et al., 2015). For the retailer, the goals of these systems are reducing labor costs, increasing transaction throughput, streamlining the checkout processes, and increasing consumer satisfaction and loyalty. For the consumer, these systems promised more control of the buying process, a reduction in the burden associated with traditional checkout lines, and time savings (Demirci Orel, Kara, 2014).
3 Self-checkout systems are growing but have growth challenges. The global self-checkout systems market was valued at $3.7 billion in 2018, and is projected to grow to $6.5 billion by 2024 at a compound annual growth rate of over 10%. According to analysts, the key factors driving the high growth rate are the need to enhance customer experience, reducing waiting time, and low-operational cost (ReportLinker, 2019). MSCOS is increasingly being adopted in the UK, Germany, and other countries of Europe, but its growth is stagnant in the US (Inman, Nikolova, 2017; Kaushik, Rahman, 2015).
4 MSCOS has created some challenges for retailers. Retailers have found that a single bad experience with the technology can cause a customer to shy away from using self-checkout in the future or, worse case, become an ex-customer. This technology has not been as effective at building customer loyalty as was hoped (Hogben, 2018). There have been increased losses due to shrinkage (i.e., items taken without payment), which some analysts have estimated to have increased by more than 100% (Chun, 2018). In order to reduce shrinkage, many retailers have increased surveillance by installing security cameras and facial recognition systems. This increased surveillance is raising privacy concerns that are rising to the level of consumer class action lawsuits (Bender, Dori, 2018). Retailers have not always understood the importance of employee interaction to create a safety net, the importance of transaction speed on customer satisfaction, or the impact of convenience on the perceived accuracy of the transaction (Kimes, Collier, 2015). In addition, mobile technologies with a financial transaction risk like MSCOS, such as mobile banking (Al-Jabri, Sohail, 2012; Alalwan et al., 2016) and mobile payment systems (Johnson et al., 2018), have found evidence to support that perceived security risks related to financial, performance, social, psychological, information, and physical loss are of great concern and can negatively impact adoption.
5 Given the incredible pressure that e-commerce is placing on retailers, their need to innovate with technologies such as MSCOS, and the costs of implementation inherent in any new technology, it is critical that the factors impacting MSCOS adoption be better understood so that retailers better achieve their goals of reducing costs, increasing transaction throughput, and improving consumer satisfaction. This study aims at gaining a better understanding of the factors that affect consumer adoption of MSCOS. This study expands the theoretical lens of diffusion of innovation (DOI) from relative advantage, trialability, compatibility, and ease of use to include other important factors, namely, perceived privacy, accuracy, and security. We develop and test a theoretical model to explore the impact of these factors on MSCOS adoption based on a crowdsourcing data collection. Our findings not only enrich the body of knowledge related to mobile technology adoption, but also provide important insights and implications for retail practice.
6 The remainder of this paper is organized as follows. First, a review of the literature associated with MSCOS is discussed. Next, the theoretical foundation for the study is established and a conceptual model is presented. Then, the methodological approach is discussed, details of the study are described, and results of the data analysis are presented. Finally, implications for both research and practice are explored.
7 Most of the research in the area of SCOS has focused on general self-checkout technology and has not differentiated between the two types: kiosk-based and mobile-based. Research in the SCOS area has explored factors associated with adoption (Aloysius et al., 2016; Inman, Nikolova, 2017), service quality (Demirci Orel, Kara, 2014; Fernandes, Pedroso, 2017), and consumer satisfaction and retention (Fernandes, Pedroso, 2017).
8 SCOS studies have used a variety of theoretical frameworks, but few MSCOS studies have utilized diffusion of innovation (DOI) (Andriulo et al., 2015). For example, a study conducted by Kaushik and Rahman (2015), using the technology acceptance model (TAM) (Davis, 1989), considered the impact of ease of use, usefulness, subjective norm, and trust on SCOS adoption. Lin and Hsieh (2011) developed and validated a self-service technology quality scale (SSTQUAL) which has since been used to explore factors impacting customer satisfaction (Demirci Orel, Kara, 2014) and SCOS deployment options (Considine, Cormican, 2016).
9 Very few studies have explored the impact of perceived security, privacy, and accuracy on SCOS adoption. Venkatesh et al. (2017) designed a study that considered the impact of security beliefs, ease of use, and usefulness on adoption of different types of auto-ID enabled artifacts such as barcode scanners and RFID. Aloysius et al. (2018) conducted a study that explored factors impacting mobile point-of-sale (POS) services. Their study considered the impact of technology enablers (i.e., ease of use and usefulness) and privacy concerns (i.e., collection, errors, secondary use, and unauthorized access) on shopping outcomes (i.e., repatronage, store image, and usage intention). The study by Collier and Kimes (2015) explored the impact of accuracy, trust, speed of transaction, exploration, convenience, and satisfaction on SCOS adoption.
10 Several studies have considered privacy and security concerns in some areas related to MSCOS. For example, a study by Featherman et al. (2010) explored the impact of privacy risk, security, and reliability concerns along with perceived usefulness and ease of use on e-service adoption, and Arvidsson (2014) studied the impact of security risk, trust, relative advantage, ease of use, and compatibility on adoption of mobile payment services. The majority of studies looking at accuracy have done so in the context of data and system accuracy in areas such as healthcare (Dillon, Lending, 2010), information aggregation (Oechslein et al., 2015), and location based services (Zhu et al., 2017).
11 There is a need for additional studies concerning MSCOS adoption. There are few studies that specifically study MSCOS. Of those that do study MSCOS, few have used DOI theory and combined it with the financial transaction risk factors of accuracy, privacy, and security to study MSCOS adoption. This study fills that gap.
12 According to diffusion of innovation (DOI) theory (Rogers, 1995), information about an innovation flows throughout a society from individual to individual enabling them to form a perception of that innovation. Rogers (1995) proposed five factors resulting from dissemination of information that ultimately drives innovation adoption. These five factors include: relative advantage of the new innovation over the precursor, compatibility with the individual’s values, needs, and past experiences, trialability, observability, and complexity.
13 Moore and Benbasat (1991) refined these factors within the information systems (IS) context by expanding on Rogers’ original model and validated scales to measure the constructs. The intent of this work was to develop scales that would be useful in a wide range of IS contexts. Observability from Rogers’ (1995) original DOI theory was thought to capture both the notion of observability as well as communicability. As a result, observability was removed and replaced with results demonstrability and visibility. These two constructs capture the demonstrability of benefits and the visibility of those benefits to the individual adopter and others. They also added voluntariness to the model to capture the freedom with which the individual has to choose whether or not to adopt. Finally, image, the ability to enhance one’s status within a given social system, was determined to be significantly different from relative advantage and thus warranted inclusion in the model.
14 DOI has gained wide popularity in diffusion research across many disciplines (Wikipedia), such as communication technology adoption (Stuart et al., 2010), Internet banking (Tan, Teo, 2000), software methodologies (Hardgrave et al., 2003), tourism (Ganglmair-Wooliscroft, Wooliscroft, 2016), and healthcare (Zhang et al., 2015). Like any scientific theory, there are potential downsides and limitations to DOI. It falls short of some theoretical constructs that account for complex networked technology diffusion and collective adoption behaviours (Lyytinen, Damsgaard, 2001). Another weakness of DOI exists in its one-way information flow (Giesler, 2012). In complex environments, multiple communication flows need to be examined.
15 Despite of some concerns and criticisms, DOI is still one of the most popular and effective diffusion models, and widely used for technology adoption research. Al-Jabri and Sohail (2012) used DOI as their theoretical framework to investigate factors impacting mobile banking adoption. In their study, they found that relative advantage, compatibility and observability had a positive effect on adoption. However, trialability and complexity were not significant. Also, they found that perceived risk negatively impacted mobile banking adoption. Arvidsson (2014) used a combination of the technology adoption model (TAM) (Davis, 1989) and DOI to investigate adoption of mobile payment systems. In this study, relative advantage, ease of use, and trust in the banks were found had a strong positive impact on mobile payment service adoption. Moreover, they found that perceived security risks had a direct effect on mobile payment system adoption intention. Oliveira et al. (2016) combined aspects from the extended unified theory of acceptance and use of technology (UTAUT) (Venkatesh et al., 2012) and DOI to identify the main factors influencing mobile payment system adoption. In this study, they found that compatibility, perceived security, performance expectations, innovativeness, and social influence to have a significant positive impact on mobile payment adoption. Finally, Johnson et al. (2018) in another study exploring factors impacting mobile-payment service adoption, used DOI as their theoretical foundation. In their study, ease of use, relative advantage, and visibility were found to positively impact mobile payment system adoption. Also, perceived security was found to positively impact adoption while perceived privacy risk negatively impacted perceived security.
16 As discussed earlier, over the past decade, a variety of studies considering factors impacting mobile banking and mobile payment adoption have used DOI as their theoretical basis. Moreover, since mobile banking and payment platforms include financial transactions, factors impacting adoption have also considered issues related to perceived risk, privacy, and security of those transactions (Al-Jabri, Sohail, 2012; Arvidsson, 2014; Johnson et al., 2018; Oliveira et al., 2016). Therefore, consistent with prior research, this study uses DOI theory as its theoretical lens and explores the impact of relative advantage, trialability, compatibility, and ease of use on MSCOS adoption. Moreover, since MSCOS involves financial transactions, the impact of perceived security, accuracy and privacy is also explored. The conceptual model is shown in Figure 1.
17 Relative advantage is defined as the degree to which an innovation is perceived as being superior to its precursor (Moore, Benbasat, 1991). Relative advantage has been found to be an important factor affecting technology adoption in a variety of areas. Using DOI, Carter and Campbell (2011) found evidence to support that relative advantage, along with institutional-based trust and e-government information, had a positive impact on Internet voting adoption. In healthcare, Chen and Zhang (2016) found that relative advantage and perceived credibility positively impacted mobile health service adoption, and Emani et al. (2012) found that relative advantage positively influenced patient perception of personal health record systems.
Figure 1 – Conceptual model
Figure 1 – Conceptual model
18 In the area of mobile technology adoption, Al-Jabri and Sohail (2012) used DOI theory to study the factors impacting mobile banking adoption. Their study found evidence that relative advantage, compatibility, observability, trialability, and perceived risk impacted adoption. In education and learning, Joo et al. (2014) found that relative advantage and complexity positively impacted individuals’ intention to use mobile learning platforms. Also, as discussed earlier, Johnson et al. (2018) found evidence to support the positive impact of relative advantage on individuals’ intention to use mobile payment services, and Arvidsson (2014) found that relative advantage positively influenced mobile banking service adoption.
19 In the context of the current study, it is reasonable to suggest that if the consumer perceives MSCOS as better than the traditional checkout line and checkout process, then this improvement would positively impact MSCOS adoption. Hence, consistent with prior studies it is proposed:
20 H1: Relative advantage positively impacts usage intention.
21 Trialability is defined as the degree to which an individual has the opportunity to experiment with an innovation prior to making an adoption decision (Moore, Benbasat, 1991). Trialability has been used in a wide variety of information technology studies. A study by Emani et al. (2012), using DOI as their theoretical foundation, found that trialability positively impacted the perceived value of personal health records when used to facilitate communication with the physician. Folorunso et al. (2010) used DOI to explore factors impacting technology adoption related to social networking. They found evidence to support that trialability together with compatibility positively impacted attitudes toward using technology. Wang (2014) found that trialability and perceived process control positively affected individual’s intention to play online games. Also, a study by Abdekhoda et al. (2016), using TAM and DOI, found that perceived usefulness, ease of use, relative advantage, compatibility, and trialability were all significant in predicting physician attitudes toward electronic medical records.
22 In the area of mobile technology adoption, Odumeru (2013) conducted a study of mobile banking adoption using DOI as its theoretical foundation. In this study, evidence was found to support that trialability along with relative advantage, complexity, compatibility, and observability were significant predictors of mobile banking adoption. A study conducted by Chung and Holdsworth (2012) found that trialabilty was a significant predictor of mobile commerce adoption. Moreover, based on studies in mobile banking, mobile payment, and mobile commerce, Slade et al. (2014) suggested an extension of UTAUT2 to included trialability, self-efficacy, innovativeness, and perceived risk and trust.
23 Therefore, consistent with prior adoption studies, it is reasonable to maintain that if the consumer has the ability to try MSCOS prior to committing to its use, then this experience would positively impact their willingness to adopt the new technology. Hence, it is proposed:
24 H2: Trialability positively impacts usage intention.
25 In order for an individual to adopt a new innovation, it is important that the innovation be compatible with the individual’s lifestyle. The concept of compatibility has been a very important predictor of individual acceptance in a variety of studies ranging from mobile payment systems (Oliveira et al., 2016) to healthcare (Abdekhoda et al., 2016). In the context of this study, compatibility is defined as the degree to which an innovation is perceived as being consistent with the individual’s values, needs, and past experience (Moore, Benbasat, 1991).
26 In the area of mobile technology adoption, Kang et al. (2015) found that compatibility was a significant predictor of both affective and cognitive involvement which impacted the intention to download and use mobile retail apps. Wu and Wang (2005), using TAM as their theoretical lens, found that compatibility in conjunction with perceived risk, cost, and perceived usefulness were significant predictors of mobile-commerce adoption. The importance of compatibility is consistent with the findings of studies discussed earlier by Odumeru (2013) and Al-Jabri and Sohail (2012) in mobile banking and Oliveira et al. (2016) in mobile payment systems.
27 Therefore, consistent with prior mobile technology adoption studies, it is reasonable to maintain that the more compatible MSCOS is with the individual’s lifestyle the more that new technology would be viewed as having advantages over the traditional checkout systems and in turn the more likely it would be adopted by the consumer. Hence, the following is proposed:
28 H3: Compatibility positively impacts relative advantage.
Ease of Use
29 In addition to compatibility, ease of use can be perceived as an advantage by the consumers and leads to system adoption. Ease of use is defined as the degree to which the individual perceives the usage of the innovation as being free from mental and physical effort (Davis, 1989). Ease of use has been used in a variety of adoption studies and has been shown to be a good predictor of individual adoption.
30 In mobile technology adoption, Arvidsson (2014) found that ease of use, trust, and perceived security were important in predicting mobile payment service adoption, and Kang et al. (2015) found that ease of use along with convenience, interactivity, compatibility, and effort expectancy were significant in predicting location-based service adoption in a retail environment. Also, a study by Ozturk et al. (2016) found evidence to suggest that ease of use along with compatibility and loyalty intentions were important predictors of mobile hotel booking technology adoption. As discussed earlier, a study by Johnson et al. (2018) found that ease of use was an important predictor of mobile payment service adoption.
31 Therefore, consistent with prior mobile technology adoption studies, it reasonable to propose that ease of use can be viewed by the consumer as an important factor enhancing the perceived relative advantage of the new technology over the precursor solution. Therefore, consistent with prior research the following is proposed:
32 H4: Ease of use positively impacts relative advantage.
33 As mobile technologies have grown more ubiquitous, so have the threats to the enterprise, the device, and the individual using those devices (Potter, 2007). The risks to mobile platforms are numerous. They can include physical attacks through loss of a device, which can include theft and disposal, logical attacks through network interfaces, external interface attacks, insider usage and data leakage, and logical attacks through malware and malicious software (Hevesi, 2019). Mobile devices are capable of storing and processing a considerable amount of potentially sensitive information about their users ranging from location data to credentials for mobile banking and enterprise virtual private networks (Dmitrienko, 2015). Given the amount of information accessible on mobile devices, breaches can severely compromise the individual’s security and privacy potentially resulting in financial losses (Delac et al., 2011).
34 Perceived security, in the context of this study, is defined as the perception that the vendor will take actions appropriate to ensure that technology usage is risk free (Shin, 2010). The impact of perceived security has been studied in a wide range of areas related to technology implementation and adoption. Venkatesh et al. (2017) considered six different shopping assistant artifacts by manipulating hardware configurations, content design, and shopping outcomes in a retail environment. They found that while RFID was the preferred with respect to technology adoption it was least preferred with respect to security beliefs. Arpaci et al. (2015) in their study of the educational use of cloud based services found that security and privacy concerns had a direct impact on attitude, which in turn, affected adoption intention. In healthcare, Alexandrou and Chen (2019) explored the impact of perceived security risk, ease of use, and usefulness on intention to use mobile devices. In this study, they found that perceived security risks had a negative impact on intention to use mobile technology within a healthcare environment.
35 Security and privacy issues appear to be of utmost importance in mobile technology that involves financial transactions. Al-Jabri and Sohail (2012) found that perceived risks negatively impacted mobile banking adoption. Studies by Johnson et al. (2018) and Oliveira (2016) found evidence to support perceived security impacted the intention to adopt mobile-payment services. Moreover, Liebana-Cabanillas et al. (2014) found evidence to support that consumer concerns related to financial losses could negatively impact mobile-payment system usage.
36 Similarly, in the area of MSCOS it follows that the consumer will be concerned that they are executing transactions within a secure environment. Hence, consistent with prior research the following is proposed:
37 H5: Perceived security positively impacts usage intention.
38 According to Westin (1967), information privacy refers to the ability of individuals to maintain control of their personal information. Within the context of this study perceived privacy is defined as the individual’s perception that their personal information is safe from potential compromise. Perceived privacy has been studied in a variety of areas in technology adoption. Almadhoun et al. (2011), in the context of higher education, explored the impact of privacy, security, and trust on willingness to share information and develop new relationships on social networking sites (SNS). Their findings suggested that perceived privacy and perceived trust had a significant impact on willingness to share information. Cases et al. (2010) explored the impact of perceived privacy on website trust within the context of email marketing campaigns. Results of their study provided evidence to suggest that perceived privacy led to trust in the website. In addition, a study conducted by Featherman et al. (2010), exploring e-service adoption, found that perceived privacy risk negatively impacted intention to use and perceived usefulness. A study conducted by Emani et al. (2012) found that perceived privacy could have a positive impact on personal health record system adoption.
39 Similar to security, privacy concerns tend to be a concern for consumers with respect to mobile technology adoption. For example, when investigating mobile payment system adoption, Johnson et al. (2018) found that perceived privacy risk had a negative impact on perceived security which in turn impacted intention to adopt mobile payment system. Other studies employed a much broader definition of the risk associated with mobile technology usage. In Alalwan et al. (2016) study of mobile banking adoption, perceived risk was defined as the consumer’s expectation of a potential loss. This loss could come in the form of a performance, social, financial, or psychological loss. Moreover, they related the risk in financial transactions to include a source of privacy risk. In this study they found perceived risk to have a negative impact on mobile banking adoption. Similarly, studies conducted by Al-Jabri and Sohail (2012) and Martins et al. (2014) found evidence to support the impact of perceived risk on mobile banking adoption.
40 MSCOS represents a service that involves a financial transaction and thus carries many of the risks that have been shown to exist in mobile banking and mobile payment systems. Hence, it is argued that, consistent with prior research, if the individual feels that their privacy is protected, then their perception of the MSCOS security will be enhanced. Therefore, the following is proposed:
41 H6: Perceived privacy positively impacts perceived security.
42 In the current study, perceived accuracy is defined as the individual’s perception that the technology will operate properly and in a flawless manner (Lankton et al., 2014). While only a few studies have considered the issue of perceived accuracy, they have covered a variety of areas. A study by Oechslein et al. (2015) investigated the impact of perceived accuracy along with trusting beliefs, perceived usefulness and ease of use on use of news aggregators. They found that perceived accuracy positively impacted both perceived usefulness and usage intention. In healthcare, Dillon and Lending (2010) considered the impact of perceived accuracy of both systems and data together with self-efficacy on provider attitudes toward patient care information systems. They found that both perceived system and data accuracy had a significant positive impact on both self-efficacy and attitude toward the patient health care system.
43 In the area of self-service technology (SST), Kimes and Collier (2013) looked at factors impacting consumer perceptions of SST. They found that convenience had a strong impact on perceived accuracy, speed, and exploration intentions of SST. They also found that current user satisfaction could be enhanced by speed and perceived accuracy of the SST and non-user perceptions of accuracy and exploration could enhance trust in SST. In another study, Kimes and Collier (2015) compared management and consumer views on perceived accuracy along with technology anxiety, need of human interaction, convenience, speed of transaction, satisfaction, and trust on attitudes toward SST. They found that allowing individuals’ sufficient time to execute, review, and verify their transactions increased the perception of accuracy.
44 Within the context of this study, MSCOS is a mobile technology that involves financial transactions and as such can be impacted by perceived risk associated with the transaction as discussed earlier. In the current MSCOS context, there is a risk of financial loss due to an inaccurate transaction. Also, consistent with SST research, perceived accuracy is important to mitigate the perceived risks and positively impacts the perceived security of the system (Collier, Kimes, 2013; Kimes, Collier, 2015). Therefore, consistent with existing literature the following is proposed:
45 H7: Perceived accuracy positively impacts perceived security.
46 Data were collected using a crowdsourcing method utilizing Amazon’s Mechanical Turk. Crowdsourcing has been shown to provide access to a diverse pool of respondents (Paolacci, Chandler, 2014) and has been used in a wide range of studies (Huh et al., 2017; Johnson et al., 2018). In addition, Mechanical Turk has been shown to be a source for high quality data (Kees et al., 2017) and tends to deliver a more representative sample of the technology-using community (Paolacci et al., 2010).
47 In the current study, 329 responses were collected from individuals residing in the United States. 27 responses were deleted due to incomplete responses and data irregularities, resulting in 302 usable responses.
48 The survey was developed using validated scales from prior studies (Agarwal, Prasad, 1997; Davis, 1989; Dinev, Hart, 2006; Kim et al., 2008; Lankton et al., 2014; Lu et al., 2011; Moore, Benbasat, 1991; Schierz et al., 2010; Shin, 2010; Tan, Teo, 2000). Some questions were modified in order to better fit the MSCOS context. For example, “Using a PWS is completely compatible with my current situation” (Moore, Benbasat, 1991, p. 216) was modified to “Using mobile self-checkout is completely compatible with how I purchase goods”, and “Using the WWW would help me to accomplish tasks more quickly” (Agarwal, Prasad, 1997, p. 579) was modified to “Using mobile self-checkout would enable me to complete my shopping more quickly”. All non-demographic items were measured using a 5-point Likert scale ranging from strongly disagree to strongly agree. The initial survey was developed and reviewed twice by three independent researchers. For each review cycle, comments and suggestions were collected and incorporated into the instrument.
49 Data analysis was performed using SPSS and SmartPLS (Ringle et al., 2015). The use of partial least squares (PLS) is considered appropriate for studies in the exploratory stage as the one described in this study. Data were analyzed using a two-step process. First, the measurement model was evaluated to determine the reliability and validity of the constructs. Next, the structural model was evaluated to determine the predictive relevance of the model, magnitude of effects, and variance explained (Anderson, Gerbing, 1988).
50 The demographic information in Table 1 shows that 55.4% of respondents were female and 43% were male. Over half of the respondents were under the age of 35. As for the education level, the majority have attended or graduated from college.
Table 1 – Demographics
Table 1 – Demographics
51 In addition to asking demographic questions, each respondent was asked how using mobile self-checkout might impact their shopping behaviour. As shown in Table 2, it does appear that mobile self-checkout would have a positive impact on both shopping frequency (32.9% would shop more) and sizes of purchases (27.5%).
Table 2 – Impact on shopping habits
|Increase shopping frequency||Definitely yes||8.5%|
|Might or might not||35.2%|
|Buy more when shopping||Definitely yes||5.0%|
|Might or might not||38.3%|
|How much has online shopping impacted your shopping in retail stores||A great deal||16.5%|
|A moderate amount||27.3%|
|Not at all||4.7%|
Table 2 – Impact on shopping habits
52 Factor loadings, Cronbach alphas, composite reliability, and factor correlations are shown in Table 3. All factor loadings are greater than 0.75, which exceeds the 0.70 threshold and indicates good discriminant validity (Henseler et al., 2009). In addition, since the square root of average variance extracted (AVE) for each latent variable, shown on the diagonal of the factor correlation matrix is greater than the correlation of that variable with other latent variables, the Fornell-Larcker criteria is satisfied (Fornell, Larcker, 1981). This is a further indication of good discriminant validity (Henseler et al., 2009). Cronbach alphas all exceed 0.80, greater than the 0.70 threshold, which indicates satisfactory internal reliability. In addition, composite reliability values, a measure of internal consistency, are all larger than 0.87, which further indicates satisfactory internal consistency. All values for AVE are greater than 0.63 indicating good convergent validity (Henseler et al., 2009). In addition, the model was checked for the presence of multicollinearity. VIF (variance inflation factor) was checked and found to be less than 2.04, less than the threshold of 5.0 suggesting that multicollinearity is not an issue (Hair et al., 2009).
53 Having established the reliability and validity of the measurement model, the structural model is then assessed. Figure 2 shows the structural model with path coefficients, significance levels of paths, and R-squared values. All paths in the model are found to be highly significant. The proposed model explains 55% of the variance associated with usage intention, 58% of the variance associated with relative advantage, and 55% of the variance associated with perceived security.
Table 3 – Factor analysis, Cronbach’s Alpha, and factor correlations
Table 3 – Factor analysis, Cronbach’s Alpha, and factor correlationsNote: Square root of AVE is shown on the diagonal of the factor correlation matrix.
Figure 2 – Structural model
Figure 2 – Structural model
55 Results of the analysis indicate that all seven hypotheses are supported (Table 4).
Table 4 – Hypothesis summary
|H1||Relative advantage positively impacts usage intention||Supported||0.277||<0.001|
|H2||Trialability positively impacts usage intention||Supported||0.372||<0.001|
|H3||Perceived security positively impacts usage intention||Supported||0.228||<0.001|
|H4||Compatibility positively impacts relative advantage||Supported||0.457||<0.001|
|H5||Ease of use positively impacts relative advantage||Supported||0.394||<0.001|
|H6||Perceived privacy positively impacts perceived security||Supported||0.423||<0.001|
|H7||Perceived accuracy positively impacts perceived security||Supported||0.385||<0.001|
Table 4 – Hypothesis summary
56 Based on DOI theory, this study proposed a theoretical model to explore factors affecting MSCOS adoption, which included relative advantage, trialability, compatibility, and ease of use. The study also explored the impact of perceived privacy, accuracy, and security. Support was found for all seven hypotheses. Relative advantage, trialability, and perceived security were found to directly impact usage intention. Compatibility and ease of use were also found to impact usage intention through their impact on relative advantage, and perceived privacy and perceived accuracy were found to impact usage intention mediated through their impact on perceived security.
57 This study makes significant contributions to theory and adds to the body of knowledge related to technology adoption. This study supports diffusion of innovation’s usefulness as being effective at exploring factors that impact the adoption of MSCOS technology. The study provides support consistent with prior research in the areas of mobile-payment services (Arvidsson, 2014; Johnson et al., 2018; Oliveira et al., 2016; Schierz et al., 2010), mobile banking (Al-Jabri, Sohail, 2012; Alalwan et al., 2016), healthcare services (Abdekhoda et al., 2016; Chen, Zhang, 2016; Emani et al., 2012), and self-service technology adoption (Fernandes, Pedroso, 2017; Kang et al., 2015; Kaushik, Rahman, 2015) for the importance of relative advantage, trialability, compatibility, and ease of use in predicting usage intention. As such, this study provides additional empirical evidence and, by use of established theory, contributes to the understanding of factors that impact MSCOS adoption.
58 Moreover, this study also highlights the importance of perceived privacy and security on MSCOS adoption. As discussed in previous sections, while there have been studies exploring the impact of perceived privacy and security in related areas such as mobile payment systems (Arvidsson, 2014; Johnson et al., 2018; Oliveira et al., 2016; Schierz et al., 2010) and mobile commerce (Featherman et al., 2010), few studies in the area of MSCOS focused on these two factors. This study fills the gap. Findings in this study are consistent with those found in related areas of interest, and provide supporting evidence of the impact of perceived privacy on perceived security and in turn perceived security on usage intention. In addition, this study adds perceived accuracy, which has received limited attention in the mobile technology adoption area, to the nomological net and finds that it positively influences perceived security.
59 This study, by grounding predictions in DOI theory, introducing new mediator relationships (e.g., perceived privacy → perceived security and compatibility → relative advantage), and examining previously unexplored relationships (e.g., perceived accuracy → perceived security), may be viewed as making a significant theoretical contribution through theory expansion (Colquitt, Zapata-Phelan, 2007).
60 With respect to retail practice, this study provides insights into consumer willingness to adopt MSCOS. Consistent with DOI, this study highlights the need that new innovations show significant advantages over their precursors. Moreover, this study provides evidence to support that relative advantage can be enhanced by making the new innovation easy to use and compatible with the lifestyle, needs, and values of the consumer. Also, this study highlights the value in allowing the consumer to be able to try and get comfortable with the system prior to commitment, and thus suggests that having employees available throughout the store to aid consumers during this trial period could be beneficial for the consumer and the retailer.
61 Security concerns, as is the case with most mobile technology and especially those that enable financial transactions, continue to be important to consumers. This should encourage retailers to take actions to enhance the consumers’ confidence that transactions are being recorded in an accurate manner, their privacy is being protected, and the overall process is secure. The importance of these factors could give rise to issues for retailers. As losses increase due to shrinkage and retailers move to put in additional security measures that will, in many cases, increase the amount of surveillance on the consumer during the shopping process, care must be taken to ensure that these new measures are not perceived as overly intrusive by the consumer. Research has found that the more intrusive a technology becomes, the more the consumer will be concerned about privacy (Inman, Nikolova, 2017), which could discourage MSCOS usage.
62 One potential limitation of this study is the use of Amazon’s Mechanical Turk and crowdsourcing to collect data. On the one hand, samples collected via crowdsourcing have been found to be more representative of the Internet population than those collected from traditional convenience samples (Paolacci et al., 2010). Also, data collected in this manner tends to be diverse and of high quality (Buhrmester et al., 2011). On the other hand, it could be argued that the sample drawn was from a population more technology-oriented than the general population. However, given that the target population was consumers using mobile technology and the study focus was MSCOS, this sampling method was deemed appropriate.
63 Another limitation of this study is the US focus. While the evolution of retail appears to be a world-wide phenomenon, it is impacting some economies more than others (Milligan, 2019; Martin, 2018). Therefore, a potential area of future research would be to take the findings of this study and expand it to focus on different countries in both Europe and Asia.
64 This study proposed and empirically tested a theoretical model based on DOI theory, and included not only relative advantage, trialability, compatibility, and ease of use, but also perceived privacy, accuracy, and security in order to explore individual intention to adopt MSCOS. This study is important for several reasons. First, traditional brick-and-mortar retailers are under significant pressure from a changing consumer demographic and a growing number of online alternatives to innovate in order to survive. Exploring and understanding the impact of these new innovations in the retail sector is important for both the retailer and the consumer. Secondly, technology and social attitudes toward technology evolve at a rapid pace. As such, it is important to continue to monitor factors affecting adoption especially as a technology matures and becomes more commonplace. Finally, the area of MSCOS represents a quickly evolving and interesting area of research. While retailers rush to innovate in order to stay competitive, they must work to ensure that those innovations deliver the desired value to the consumer and produce the desired outcomes. Otherwise, the consumer will not have intention to adopt the innovation. Currently, while mobile self-checkout appears to be increasing consumer satisfaction, increasing transaction rates, and lowering consumer labor costs, there are also increased losses due to shrinkage. How retailers address this in the future will be interesting to watch as increased security measures to protect retailers could negatively impact consumer perceptions of privacy and hence negatively impact adoption and continuance intention.
65 This study makes important contributions to the technology adoption literature. It further illustrates how DOI can be used to explore factors affecting consumer adoption. It is one of few studies to consider the impact of perceived privacy and security on adoption of MSCOS, which could be of growing importance in the future. Also, the study is one of the few to explore the impact of perceived accuracy within the context of MSCOS.
66 For practitioners, this study provides support for the importance of relative advantage when attempting to replace an entrenched legacy technology. It also highlights the importance of ease of use and compatibility with the needs of the consumer as two key factors impacting the consumer’s perception of relative advantage over traditional checkout lines. Moreover, allowing consumers to try the new technology prior to commitment will also positively impact their willingness to adopt MSCOS. Also, this study provides evidence that perceived security and privacy are issues that continue to weigh heavily on the minds of consumers. Finally, as with any technology where a financial transaction is ultimately involved, the retailer must take action to ensure the consumer that the transaction is performed in an accurate manner.