Antecedents of E-Loyalty in the Context of Online Fashion Retailers

(1)

Bachelor Thesis

Antecedents of E-Loyalty in the Context of Online Fashion Retailers

Authors:

Beatra Berisha Mohammed Zaki Ghazal Miranda Okello Supervisor: Viktor Magnusson Examiner: Åsa Devine Semester: Spring 2020 Course Code: 2FE21E

(2)

Acknowledgements

The authors of this paper would like to express their appreciation to a number of people who were crucial in the development of this paper.

First and foremost, we would like to extend our thanks to our supervisor, Viktor Magnusson who was supportive and dedicated his time in order to give us valuable feedback and

guidance during the research process.

Secondly, we would like to thank Åsa Devine and our classmates who provided invaluable feedback during our seminars which helped guide our paper and research focus.

Thirdly, we would like to thank Magnus Willeson for lending us his expertise in statistics in order to guide us with our data processing techniques.

Finally, we would like to express our gratitude for all of those who participated in the study.

Beatra Berisha Mohammed Zaki Ghazal Miranda Okello

(3)

Abstract

With recent developments in the internet and the rise of e-commerce, the concept of e-loyalty has gained more traction as a result of its growing importance. Due to this, there is a

continuous shift away from brick and mortar fashion retailing towards online fashion retailers. As a result, it is evident that future and current fashion retailers must have some form of online presence in order to succeed in today’s landscape. The purpose of this paper was to explain the effect of e-trust, e-satisfaction and e-service quality on e-loyalty in the context of online fashion retailers. In order to fulfil this purpose, a deductive explanatory approach was taken. Firstly, the authors conducted research on the subject of e-loyalty and what factors influence it. Furthermore, a conceptual model applicable to the context of online fashion retailers was developed. Moreover, an operationalization was created which was the basis of a self-completion questionnaire that was designed in order to collect primary data.

After the primary data was collected through convenience sampling and partly through expert sampling, it was cleaned and then placed into SPSS in order to be analyzed. Multiple

statistical tests were conducted in order to examine the significance of the data and the relationship between the independent variables (e-trust, e-satisfaction and e-service quality) and the dependent variable (e-loyalty). The results from the analysis showed that all the aforementioned independent variables have a positive effect on e-loyalty in the context of online fashion retailers which meant that the hypotheses from this study were found to be accepted. As a result, e-trust, e-satisfaction and e-service quality were all found to have a positive relationship with e-loyalty.

(4)

1. Introduction

1.1 Background

Early definitions of brand loyalty have been characterized by six distinct factors. These are:

1) Biased 2) Behavioural response 3) Express over time 4) Measured by a decision-making unit 5) one or more brands in a set of brands 6) Part of a customer’s decision making process”

(Jacoby and Kyner, 1973 p.2)”. If these criteria are fulfilled, then a customer is considered brand loyal as they are making repeat purchases with a behavioural motive behind them.

However, with recent developments in the internet and the rise of e-commerce, the concept of e-loyalty has gained more traction as a result of its growing importance. E-loyalty can be classified as an extension of brand loyalty to an online context of consumer behaviour (Igi- global.com, 2020). It is considered essential to e-commerce retailers since just like brand loyalty, e-loyalty allows companies to reap multiple advantageous benefits. However, this becomes a more difficult task for online retailers since they are forced to compete with both similar online stores and similar offline stores (Cristobal, Flavián and Guinalíu, 2007).

Nurturing and maintaining e-loyalty in customers of online retailers is essential since new customers generally cost 20-40% more for online retailers in comparison to offline retailers (Reichheld and Schefter, 2000). Furthermore, research shows that online customers tend to exhibit a clear tendency towards loyalty. This means that if online retailers fail to initially convert new customers into loyal customers, they risk losing them permanently to

competitors (Reichheld and Schefter, 2000). As a result of this, it becomes crucial to understand how E-loyalty is formed. To do this, its antecedents must be examined.

When examining the research on E-loyalty, multiple sources recognize E-satisfaction, (Christodoulides and Michaelidou, 2010; Al-dweeri et al., 2019; Kim, Jin and Swinney, 2009), E-trust (Al-dweeri et al., 2019; Kim, Jin and Swinney, 2009; Cyr, Hassanein, Head and Ivanov, 2007 and Parasuraman and Malhotra, 2000) and E-Service Quality, sometimes referred to as Etail Quality, (Al-dweeri et al., 2019; Cristobal, Flavián and Guinalíu, 2007;

Wolfinbarger and Gilly, 2003 and Parasuraman, Zeithaml, & Malhotra, 2005) as antecedents of E-loyalty. As a result of this, it becomes important to clearly understand the

aforementioned antecedents and explain how they impact e-loyalty towards online retailers.

(7)

Corritore, Kracher and Weidenbeck (2003) explain that trust analyzed in an online context is defined as e-trust. E-trust is defined as “the attitude of confident expectation in an online situation of risk that one's vulnerabilities will not be exploited” (Corritore, Kracher and Wiedenbeck, 2003, p.740). Positive attitudes of confidence and positive expectations towards a site could be achieved if there is no reason to have a lack of trust. The benefits from

forming trust may increase positive emotions of customers towards a site which can make them more willing to establish a long-term relationship with the website, hence creating e- loyalty (Lin and Lee, 2012).

As previously mentioned, e-satisfaction is also a crucial antecedent of e-loyalty since it is unlikely that a customer will exhibit loyalty if they are dissatisfied with an online retailer (Christodoulides and Michaelidou, 2010). Customers tend to shop online as it is generally more convenient for them than visiting a traditional brick and mortar store. Therefore, if an online retailer fails to provide convenience, customers will be unsatisfied and less likely to exhibit e-loyalty (Christodoulides and Michaelidou, 2010). Furthermore, variety or product range is an important part of e-satisfaction since customers may gravitate and be loyal to an online retailer that provides a satisfactory choice of products (Christodoulides and

Michaelidou, 2010).

Finally, e-service quality also plays a significant role in affecting e-loyalty. E-service quality is important as it is a major point of interaction between the customer and online retailer and is a crucial part of the relationship between them (Wolfinbarger and Gilly, 2003). Reliability is a key part of e-service quality and is important when it comes to retaining customers in an online context (Wolfinbarger and Gilly, 2003). This is because if e-service quality is

unreliable (e.g. poor delivery times, customer service etc.) then it is unlikely that a customer will stay with the same online retailer, they are more likely to switch over to a competitor.

(8)

1.2 Problem Discussion

One particular industry that has been revolutionized due to the advent of the internet and the rise of e-commerce is the fashion industry (Hannah Daly, 2020). As a result of these massive shifts, multiple fashion retailers have either shut down as they failed to accommodate to the changes in the market landscape or opted to move online in order to adapt to the rapidly changing market demands (Hannah Daly, 2020). Helm, Kim and Van Riper (2020) state that there is a so-called “retail apocalypse” occurring in the United States where many traditional brick and mortar stores have shut down. This is attributed to customer perception of physical stores failing to fulfil evolving demands from customers and their subsequent shift towards favouring online retailers (Helm, Kim and Van Riper, 2020). This issue is not limited to the United States and has begun to spread elsewhere. For example, in the United Kingdom, approximately 6000 brick and mortar stores closed down in 2018 due to a shift away from physical retail stores (Butler, 2020). Therefore, with this continuous shift away from brick and mortar fashion retailing towards online fashion retailers, it is evident that future and current fashion retailers must have some form of online presence in order to succeed in today’s landscape. In order to effectively have a successful online shopping platform, it is important to foster e-loyalty (Hannah Daly, 2020).

As e-loyalty is reliant on customer behaviour and perception, there are a multitude of factors that can affect the extent of e-loyalty in a customer. In the e-commerce space, it is generally thought that consumers are fickle and quick to change shopping platforms since they are disloyal. However, the opposite is more likely with online consumers showing a tendency to be more loyal (Reichheld and Schefter, 2000). Additionally, it is crucial to clearly understand the antecedents of e-loyalty and how they interact with each other in order to reap the benefits of e-loyalty. For example, the cost of gaining a new customer on an online only store is 20- 40% more costly than a store with both a physical and online presence (Reichheld and Schefter, 2000). Furthermore, since online retailers lack the traditional aspects of offline retailers such as being able to try and feel the product, see and interact with the store staff or enter a physical store, a customer’s uncertainties are much higher than normal (Reichheld and Schefter, 2000). Therefore, in order to mitigate the aforementioned uncertainties, it is

important to foster e-loyalty and understand how it is formed.

(9)

When examining the research surrounding the antecedents of e-loyalty, most of it focuses on particular industries. For example, studies conducted on large e-tailers such as Amazon and Ebay and online concert ticket retailers may not capture the nuance needed to explain the relationship of certain factors and their effect on e-loyalty in an online fashion retailer context (Al-dweeri et al., 2019; Cyr, Hassanein, Head and Ivanov, 2007; Kim, Jin and Swinney, 2009). Additionally, the studies that do not particularly focus on a certain industry and explore a variety of them may miss the nuance required to explain the relationship between the antecedents of e-loyalty and e-loyalty itself in that particular context. (Swaminathan, Anderson and Song, 2018; Sai Vijay, Prashar and Sahay, 2019). As a result of this, a

common theme within existing research is the recommendation to examine other e-tailers that operate within different industries or product categories in order to explain the relationship between the antecedents of e-loyalty and e-loyalty itself. (Al-dweeri et al., 2019; Cyr, Hassanein, Head and Ivanov, 2007; Kim, Jin and Swinney, 2009; Swaminathan, Anderson and Song, 2018; Sai Vijay, Prashar and Sahay, 2019).

The case for examining online fashion retailers becomes more apparent when taking into account how a hedonic product category (fashion products) may cause customers to exhibit different tendencies towards e-loyalty at different strengths. This notion is discussed by Arnold and Reynolds (2003) who state that customers may have a more intense involvement and effective response when purchasing a hedonic product. Furthermore, Arnold and

Reynolds (2003) state that customers tend to act more impulsively when purchasing hedonic products which might also have an impact on their tendency to be loyal that is not present in other non-hedonic contexts. Therefore, this warrants a further examination of this context in order to explain the chosen antecedents (e-satisfaction, e-trust, e-service quality) and their relationship with e-loyalty.

As a result of the growing importance of online fashion retailing and the lack of studies conducted on the antecedents of e-loyalty within that context, the authors have chosen to pursue this context. By doing this, the authors will contribute to the literature on antecedents of e-loyalty in an online fashion retailer context. By analyzing additional product categories (in this case online fashion retailers), the scope of the literature will be more diverse which should allow a possible trend to emerge which can improve generalizability.

(10)

1.3 Purpose

The purpose of this paper is to explain the effect of e-trust, e-satisfaction and e-service quality on e-loyalty in the context of online fashion retailers.

2. Theoretical Framework

The chosen concepts of e-loyalty, e-satisfaction, e-trust and e-service quality all have their roots in the traditional definitions of loyalty, satisfaction, trust and service quality which means there is a significant overlap between them in content. This is because these concepts were developed and initially applicable to brick and mortar stores before evolving to be applicable to online retailers (Ponirin, Scott and Heidt, 2015). Since the authors of this paper are investigating a new context (online fashion retailers) the authors of this paper utilized existing research from both the traditional definitions of the aforementioned concepts and their “e-definitions”. By doing this, the authors hoped to capture a wider scope of the concept, provide nuance and ensure nothing of importance was left out.

2.1 Loyalty and its Relationship with E-Loyalty

Jacoby and Kyner (1973) coined the definition of loyalty by defining the conditions to loyalty. Loyalty was described as being the result of a specific psychological process that should fulfill six conditions: (1) a purchase needs to be ‘nonrandom’ (2) have some sort of behavioural response (3) be expressed over time (4) and be made by some decision-making unit (5) with respect to a set of alternative brands to choose from (6) and is a function of the psychological decision-making process. Loyalty towards a brand is not a random occurrence, it is biased purchasing behaviour (Jacoby and Kyner, 1973). Furthermore, it is an evaluative process where one makes biased decisions which then leads to forming commitment to a brand. Commitment is arguably what separates repeat purchasing behaviour from loyalty.

Effective marketing strategies may elicit repeat purchasing behaviour, but the underlying reasons for the repeat purchasing behaviour may differ from loyalty. Oliver (1999) defined loyalty as a deeply held commitment to repurchase a preferred product or service

consistently, causing repetitive same-brand or same brand-set purchasing, despite

(11)

and Kyner (1973) explain loyalty as a condition that must occur under some duration. A relationship must be formed where behavioural responses are a result of choosing a brand over the other, otherwise it may not be loyalty but instead repeat purchasing behaviour.

Main contributions from early research on marketing constructs such as loyalty generally used psychological components as theoretical foundations to build on concepts. Distinctions between repeat purchasing behaviour and loyalty has been established in attempts to

revitalize previous frameworks with scattered definitions by incorporating psychological theories of cognitive, affective and conative natures (Jacoby and Kyner, 1973; Churchill, 1979; Dick and Basu, 1994; Oliver, 1999, Aaker, 1997).

Oliver (1999) defined behavioral and attitudinal loyalty through cognitive, affective, conative and action dimensions. Cognitive loyalty is a stage belonging to attitudinal loyalty which entails that the customer in question is indicating that one brand is prefered over the other.

Whether it is actual or imagined competitive features, the consumer is looking for

differentiation such as price or competitive features through personal experiences or through advertising (Oliver, 1999). The affective loyalty dimension refers to having an enhanced liking or attitude towards a brand. The customer may buy the product simply because they like it, meaning that this attitudinal loyalty stems from accumulated satisfied usage occasions (Oliver, 1999). Conative loyalty is the third step in the process developing loyalty and is also of behavioural intent according to Oliver (1999). Although, previous stages of attitudinal loyalty requires some form of commitment, conative loyalty requires as set

intention/commitment, not collected information or by simply liking the product. Conative loyalty is brand-specific, it is intentional enough to perhaps make the commitment to repurchase, but with that said it does not guarantee that the actual purchase will be made (Oliver, 1999). The purchase may be anticipated but not realized, which leads into the final stage of loyalty: action loyalty. Action loyalty is described as movement from intentions to actions. The customer must overcome any obstacles that may prevent the act in order for any loyalty to occur, causing a different form of commitment: committing to repurchasing. The fourth action phase connects the attitude-based loyalty with the behavioural interests with the action of repurchasing (Oliver, 1999).

More recent research has built upon the topic of repeat purchasing behaviour and attitudes by incorporating statistical elements to loyalty in order to align it with the birth and growth of

(12)

traditional brand loyalty defined according to the online experience, mediated by technology.

Gommans et al., (2001) defines e-loyalty as the intention to revisit a site which is based upon the trust and security factors as well as ease of navigating the website, the customer service, the brand image- and community building involvement as well as the value propositions such as the product quality, the ability to customize, the price and the promised guarantees.

Anderson and Srinivasan (2003) continues to argue that the current definition of loyalty may be built upon past psychological concepts about attitudes and behaviour. E-loyalty is

described to consist of those attitudes and behaviours and argues that they result in repeat purchasing behavior towards an electronic business (Anderson and Srinivasan, 2003;

Gommans et al., 2001).

2.2 Trust and its Relationship with E-trust

E-trust is an important factor part of the Web site attributes that were identified by Zeithaml, Parasuraman and Malhotra (2000). Trust may originally be described as the assurance the customer in this case needs or the confidence the customer may feel when visiting a site. This may be dependent on the reputation or the offering part of the site and how clear and truthful information may be presented to the customer. Singh and Sirdeshmukh (2000) define trust in two different ways where the primary definition of trust relates to the expectancy

conceptualization of trust. This refers to the expectations of the intention and/or behaviours on the other party, i.e. focusing more on the beliefs and expectation on the other party to act responsibly and according to the set expectations. The secondary definition refers to the behavioural conceptualization of trust which emphasizes the actions exchanged between the parties. The actions are held responsible and are used as evidence to the consumers intentions to rely on both the exchange and the other parties ‘motives’ (Sing and Sirdeshmukh, 2000).

Moreover, e-trust may be one of the most contributing factors when discussed in a e-retail context since it comes with a perceived risk (Merrilees and Fry, 2003). The feeling of having lack of e-trust may arise from sensing a lack of control regarding access to personal

information during the navigation process and perceived good reputation. Consumers may not trust online providers enough to engage, which may lead to a lack of relationships exchanges (Hoffman, Novak and Peralta, 1999; Chen and Barnes, 2007). Furthermore,

Corritore, Kracher and Wiedenbeck (2003) explain the concept of trust in the form of a model

(13)

context: credibility, ease of use and risk. These three factors originate from external factors from the environment (physical and psychological) part of the online situation at hand which may be the navigation architecture on the site, the design elements or the information

accuracy. Once these factors are experienced by the visitor on the site, the credibility factor which may be explained from a multidimensional approach as honesty, predictability, expertise and reputation, is taken into account to secondly be judged on its ease of use.

The ease of using the website is in other words explained as how easy it is to navigate the site, for example the ease of searching, if there are any broken links or the transaction interaction quality, which collectively contributes to both the aforementioned concept of credibility and the third concept part of the model: risk (Corritore, Kracher and Wiedenbeck, 2003; Hoffman Novak and Peralta, 1999; Chen and Barnes, 2007). Risk may arise when there is a lack of control in avoiding non-credible outcomes, making risk and credibility a direct link to trust since they may be the main prohibiting factors to trust (Corritore, Kracher and Wiedenbeck, 2003; Merrilees and Fry, 2003). If trust is established, buyer and seller relationships may be formed even in an online context according to Hoffman, Novak and Peralta (2017). Trust is a key element if the main goal is to maintain a relationship, especially in an online context since confidential information belonging to customers are being traded in exchange for customers committing to a purchase (Al-Dweeri et al., 2017; Cyr, Hassanein, Head and Ivanov, 2007). The willingness to even engage in a relationship in an online context may be affected if there is a lack of e-trust (Hoffman, Novak and Peralta, 1999). In both an online- as well as an offline context there needs to be a willingness to rely and have confidence in the seller. Trust is therefore an essential element to customer's purchase intentions both offline and online. Although the circumstances differ, the consumer in

question is still being asked to be vulnerable to the seller to a certain extent, which directs the trust in committing to a purchase towards having a direct influence on purchase intentions (Cyr, Hassanein, Head and Ivanov, 2007; Chen and Barnes, 2007; Kim, Jin and Swinney, 2009)

(14)

2.3 Satisfaction and its Relationship with E-satisfaction

According to Oliver (1981, p.27 cited in Swaminathan, Anderson and Song, 2018),

satisfaction is determined by the “summary psychological state resulting when the emotion surrounding disconfirmed expectations is coupled with a consumer’s prior feelings about the consumer experience”. Satisfaction has been found to be one of the main antecedents when forming brand loyalty which does not differ when forming e-loyalty considering e-

satisfaction is granted as a natural antecedent to e-loyalty (Gupta, Schiviniski and Brzozowska-Woś, 2017). The prior drivers of e-satisfaction are found to be consumer’s perceived value, choice and care (Swaminathan, Anderson and Song, 2018). The authors further examine that the level of e-satisfaction that customers experience when purchasing products online is shown to be influenced by their perceived value. If the product purchased meets the needs and expectations of the customers, i.e. perceived value, they will feel satisfaction which will hopefully lead to repetitive purchasing (Swaminathan, Anderson and Song, 2018).

Price is also an established factor to consumer’s satisfaction which affects the perceived value given from the product. Customer care is proved by the level of care that a company is approaching its customers with. There are multiple ways for e-commerce companies to show care for its customers.Swaminathan, Anderson and Song, (2018) have determined that the performance of customer service, anticipating downtime, accurate billing as well as the time of the delivery are included in customer care.

Another thing found to play an important role is breakdown in services which is shown to have a negative effect on e-satisfaction. The level of satisfaction is also affected by choice which concerns the amount of product lines offered by an e-commerce company. A higher number of choices on a website increases the possibility for customers to find products that match their needs and wants. Hence, the quality perception increases as well which has a positive effect on e-satisfaction (Swaminathan, Anderson and Song, 2018). Beside these three drivers, other studies have found other important drivers of e-satisfaction. One of the most important drivers is found to be convenience (Schaupp and Belanger, 2005; Szymanski and Hise, 2000; Evanschitzky, Iyer, Hesse and Ahlert, 2004). An initial study of e-satisfaction made by (Szymanski and Hise, 2000) also found website design as the essential drivers of e- satisfaction. In the same study, product information was also established to have a significant

(15)

information of the products offered are needed in order to achieve the most effective outcomes (Szymanski and Hise, 2000).

Another reexamination of the aforementioned study made by Evanschitzky, Iyer, Hesse and Ahlert (2004) confirms site design is an important aspect of convenience which is

subsequently important to e-satisfaction. In order for a company to achieve a good site design, easy-to-navigate sites should be designed along with being fast and uncluttered.

Customers’ wish to order products online fast and easy, hence saving time and making browsing easy is included in convenience (Szymanski and Hise, 2000). Beside convenience, Schaupp and Belanger (2005) established merchandising, i.e. the product itself and the range of products offered, as a strong driver.

2.4 Service Quality and its Relationship with E-service quality

Grönroos (1984) explained that the original concept of service quality was an unexplored concept that had not yet been explained in the form of a model. By providing a model about how the quality of the services is perceived by customers, Grönroos (1984) explained that the concept of service quality is a multidimensional concept both from a service aspect as well as the quality aspect.

Apart from claiming that there is a divide in what is perceived or expected from a service and what is functional or technical in the quality, it was suggested that service quality stems from comparison. The actual service performance is held accountable through customers

expectations, something that is still highly relevant even though the experiences and interactions were not tested in an online context (Grönroos 1984; Lewis and Booms, 1983, cited in Parasuraman, Zeithaml and Berry, 1985; Smith and Huston, 1982 cited in

Parasuraman, Zeithaml and Berry, 1985; Churchill and Surprenant, 1982; Parasuraman, Zeithaml and Malhotra, 2005). The aforementioned building blocks to services and quality are explained to collectively contribute to the judgements made of the overall superiority or excellence regarding the object in question (Berry, Parasuraman and Zeithaml,1988). It is also argued that the perceptions and expectations towards the object or entity are part of service quality in an online context. Along with the growth of the e-commerce business there has been a shift in e-service quality due to the importance of differentiation. By tying in the

(16)

be defined as the extent to which a website facilitates efficient and effective shopping, purchasing and delivering of the services or goods in question. The functional and technical aspects as well as the perceptions and expectations of the good or service is still highly relevant in an online context and is arguably still a recipe for financial gain (Parasuraman, Zeithaml and Malhotra, 2005; Zeithaml, Parasuraman and Malhotra, 2002; Berry,

Parasuraman and Zeithaml, 1988).

Wolfinbarger and Gilly (2003) argue that web design, customer service, security and fulfillment/reliability are contributing factors to e-service quality and that they collectively contribute to a customers overall perception of quality. The webdesign entails what

experiences visitors to a website may encounter which may be navigating the site, searching for information, processing an order, appropriate personalization attempts or selecting products. The customer service aspects are part of experiencing the website but with aims to be helpful and responsive to customer inquiries, while the security factor focuses on

maintaining privacy of shared information such as credit card information. The fulfillment- or reliability factor focuses on the accuracy regarding how a product or service is displayed, described and delivered to ensure that customers are confident that they will receive what they thought they ordered (Wolfinbarger and Gilly, 2003). Furthermore, Parasuraman, Zeithaml and Malhotra (2005) build upon the aforementioned factors part of the e-service quality modeling provided by Wolfinger and Gilly (2003) by explaining that e-service quality is heavily reliant on being reliable and valid.

(17)

3. Conceptualization

The purpose of this section is to utilize existing research in order to present a

conceptualization of the antecedents of e-loyalty and how they affect it. Through this, a number of hypotheses on the relation between these antecedents can be drawn and tested later on.

3.1 E-Trust and its effect on E-Loyalty

Research indicates that since online transactions are perceived as high risk by customers, e- trust plays a fundamental role as an antecedent of e-loyalty (Kim, Jin and Swinney, 2009).

Corritore, Kracher and Wiedenbeck (2003) state that one factor that can make customers perceive a website as untrustworthy is if the website is difficult to navigate and use.

Therefore, if the website is well optimized, customers are more likely to perceive it as trustworthy.

Additionally, Bartikowski and Singh (2014) state that customers are usually less satisfied and hesitant with online purchases due to issues with the trustworthiness of a site. Therefore, if a website is perceived as credible, it is more likely that this risk element will be gone and the website will be perceived as trustworthy which means that e-loyalty will be positively affected (Chen and Barnes, 2007). This perception of credibility is usually influenced by multiple factors such as reputation, expertise, predictability and honesty. For the sake of this paper, the term “credibility” was used as an umbrella term for the aforementioned concepts since they are similar in meaning or are influenced by one another. In the context of online fashion retailing, credibility can be influenced by word of mouth, the reputation and history of the online fashion retailer, mentions in the media and the size of the company.

Additionally, it is important for an online fashion retailer to ensure the confidentiality of the customer’s personal information after a transaction is made. If customers feel that their information is being sold to other companies or is being exposed through data leaks, it is unlikely that the online fashion retailer will be seen as trustworthy. Therefore, there will be no chance of a relationship (which is essential for e-loyalty) between the customer and the retailer being maintained (Al-Dweeri et al., 2017; Cyr, Hassanein, Head and Ivanov, 2007).

(18)

Finally, when approaching an online fashion retailer, a customer normally has certain expectations of how an online fashion retailer will behave (order processing, customer service etc.). If there is a match between the expectations of the customer and the actual behaviour of the online fashion retailer, online trust will most likely be positively influenced (Singh and Sirdeshmukh, 2000).

Therefore, based on the aforementioned points, it can be seen that e-trust can have a significant effect on online fashion retailers. Therefore, the first hypothesis is:

H1: E-trust has a positive effect on e-loyalty in an online fashion retailer context.

3.2 E-Satisfaction and its effect on E-Loyalty

Research indicates that perceived value, choice and care are significant drivers of e- satisfaction (Swaminathan, Anderson and Song, 2018). They in turn affect business credibility which influences business loyalty.

Additionally, a well designed website that communicates accurate information about the online retailer’s products through an optimized user interface system leads to less time spent searching for and processing information on the customer’s end. This will cause them to be more satisfied and likely to return thereby facilitating e-loyalty (Kim, Jin and Swinney, 2009). If a fashion item is poorly described (e.g. no accurate sizing measurements are provided), it is less likely that a customer will be satisfied as they lack information about the product.

Furthermore, perceived value plays an important role in influencing e-satisfaction.

Swaminathan, Anderson and Song, (2018) state that if the products of an online retailer match the expectations of a customer, they will be satisfied and likely to conduct repeat purchases and have a positive attitude towards the online retailer. This is especially the case with online fashion retailers since if an item is ordered with the expectation of high quality (e.g. based on the price of the product) fails to meet those expectations, the customer might not feel satisfied, return the product and not conduct purchases repeatedly.

(19)

Every customer has specific needs and wants with regards to fashion items (e.g. size, colours, design, brands etc.). This makes choice a crucial element for customers and their level of e- satisfaction towards an online fashion retailer. If a company has an unsatisfactory selection of items, the customer is unlikely to purchase or return to the online retailer again (Schaupp and Belanger, 2005; Szymanski and Hise, 2000; Evanschitzky, Iyer, Hesse and Ahlert, 2004).

All online retailers eventually experience a breakdown in services. This may cause

dissatisfaction amongst consumers, but its effects can be mitigated through ensuring quality customer care. This can be done through an effective customer service experience that makes the customers feel respected and cared about (Swaminathan, Anderson and Song, 2018). If a company successfully manages a breakdown of services and also shows customers that it cares about them (through transparency, ensuring deliveries on time, resolving issues swiftly, etc.), then customers are likely to remain satisfied and loyal to the online fashion retailer.

Finally, Schaupp and Belanger (2005) state that convenience is crucial for e-satisfaction and consequently e-loyalty. This is because without it, shopping is made more challenging for the customer and they are easily able to switch stores quickly (in comparison to an offline

context). For example, if a website is difficult to navigate or use, a customer is unlikely to continue visiting that online fashion retailer and be loyal to it (Schaupp and Belanger, 2005;

Szymanski and Hise, 2000; Evanschitzky, Iyer, Hesse and Ahlert, 2004).

Therefore, based on the aforementioned points, it can be seen that e-satisfaction can have a significant effect on online fashion retailers. Therefore, the second hypothesis is:

H2: E-satisfaction has a positive effect on e-loyalty in an online fashion retailer context.

3.3 E-Service Quality and its effect on E-Loyalty

Service quality is a crucial part of online fashion retailers. Without it, customers are unlikely to be loyal to a company. This is because perceived service quality (the level of service a customer expects) stems from comparison to other online fashion retailers. Therefore, if they fail to meet the minimum level of expected service quality, the customer will most likely

(20)

switch to a different online fashion retailer thereby negatively affecting e-loyalty (Parasuraman, Zeithaml and Malhotra, 2005).

Order processing also plays an important role in how a customer perceives the e-service quality offered by an online retailer. If an online fashion retailer takes too long to process orders and deliver them, it will hinder the purchasing process and make the online retailer be perceived as unreliable (Zeithaml and Malhotra 2005; Wolfinbarger and Gilly, 2003).

Furthermore, a well-designed website can offer relevant personalized recommendations of fashion items based on browsing/purchasing habits which can positively influence e-loyalty (Wolfinbarger and Gilly, 2003). For example, a customer who purchases a certain brand of shirts might perceive the online fashion retailer positively if they received suggestions of similar shirt brands and models or even matching fashion attire. This could encourage the customer to conduct repeat purchases and visit the site more frequently.

Therefore, based on the aforementioned points, it can be seen that e-service quality can have a significant effect on online fashion retailers. Therefore, the third hypothesis is:

H3: E-service quality has a positive effect on e-loyalty in an online fashion retailer context.

(21)

3.5 Conceptual Model

Figure 1. The proposed conceptual model.

(22)

4. Methodology

4.1 Research Approach

This section aims to present the research the approach the authors of this paper decided to embark upon and provide the reasoning behind their choices.

4.1.1 Deductive Research

Due to the explanatory nature of this paper, the authors utilized a deductive research approach. Deductive theory is the most generally accepted explanation of the link between existing theory and research. When taking a deductive research approach, the researcher normally draws a hypothesis based on existing theory pertaining to a certain topic (Bryman and Bell, 2015). Through this hypothesis, the researcher must extract items that will further be operationalized into researchable entities. By doing this, the researcher clearly

demonstrates how data on a particular topic will be collected (Bryman and Bell, 2015). Hyde (2000) states that deductive research essentially allows the theory to fill in the gaps of

collected empirical data. Additionally, deductive research allows the researcher to fulfill the purpose of being able to make statistical generalizations about a certain phenomenon as they can analyze the collected empirical data with the help of theory.

Therefore, the authors conducted research on the subject of e-loyalty and its antecedents.

After gathering enough information, a literature review was conducted in order to allow the authors to develop a greater understanding of the chosen topic. From this literature review, a theoretical framework was created which was then used to draw out a conceptual model on how the chosen antecedents might affect e-loyalty. To test the relationships between the antecedents and the concept of e-loyalty, the authors developed hypotheses that would allow an empirical investigation. After these hypotheses were drawn out, the authors identified items related to each chosen concept. These items were then operationalized into different statements that would be used for measuring their relationship with e-loyalty. Essentially, the theory was what guided the research and the hypotheses and not vice versa.

(23)

4.1.2 Quantitative Research Approach

There are two different ways to conduct a study but since this research has an already existing theory as its foundation, it is according to Bryman and Bell (2015) regarded as being a

deductive way of conducting research which is of the quantitative nature. Quantitative research deals with numerical data and large samples which is viewed as an objective approach. The theory of this study is being tested in an online fashion retailer context where the independent variables (e-satisfaction, e-service quality, e-trust) are put to test to see what causes or effect it may have on the dependent variable (e-loyalty). The causality is explained by Bryman and Bell (2015) as one of the preoccupations that determines quantitative

research.

In order to offer a fruitful conclusion that relates to the purpose and is relevant for future research, it is ideal that the data generated is generalizable. Generalization implies that the results may be applied to everyone else and not solely the respondents of the questionnaire (Bryman and Bell, 2015). Hence, in order for a study to be “perfectly” quantitative, the tested relationships between the independent variables and dependent variables need to be reliable, generalizable and objective. The authors of this paper took numerous steps in order to attempt to achieve the aforementioned criteria which will be described in great detail throughout this section.

4.2 Research Design

A research design is of use when conducting marketing research projects. It is a necessary procedure that explains how data is collected and analyzed in order to solve marketing research problems. It is therefore important that the selected research design is in line with the marketing research problem at hand since it lays the foundation for conducting a project (Malhotra, 2010).

A research design may take form in three different kinds of frameworks: exploratory, descriptive and causal (explanatory). Although they may be prescribed with different meanings, they are all related to each other. Explanatory research concerns cause-and-effect relationships, something that is common when conducting experiments such as field- or laboratory experiments. Cause-and-effect relations all stem from seeking evidence to the relation at hand which may take different forms such as time order variables, accompanying

(24)

variation or elimination of alternative explanations which are all collectively regarded as some of the most common forms of evidence used in research (Malhotra, 2010; Iacobucci and Churchill, 2015).

The structure of a research design is also dependent on theory since a research problem may be solved with the help of theory. Theory is also consequently what the generated data will be compared to and tested against. Since theory is of utmost importance when conducting

research, it is also important that it is conceptualized in terms of the research problem at hand, especially when there is evidence that needs to be collected to accept or reject a hypothesized relationship. Thus, requiring thoroughly reviewed theory in order to provide theoretical grounds that support what is being problematized (Saunders, Lewis and Thornhill, 2016; Iacobucci and Churchill, 2015).

When wanting to investigate and explain multiple relations within quantitative research, researchers often consequently conduct cross-sectional studies. A cross-sectional study concerns a specific phenomenon at a particular time and involves collecting information only once from any given sample of a population (Bryman and Bell 2015; Saunders, Lewis and Thornhill, 2016; Malhotra, 2010). It is also of use to researchers dealing with large

quantitative datasets that may require making multiple distinctions and finding variation in the data (Bryman and Bell, 2015).

The aforementioned explanatory research design is regarded as being the most suitable design approach to this research since the purpose of this thesis was formed to investigate a number of cause-and effect relationships. Furthermore, explanatory design allows for further

investigation and explanation of the created hypothesized statements, which helps the process of explaining the possible relationships found between the stated variables. This is also the reasoning behind choosing to conduct a cross-sectional design approach. Since the purpose of this study is to explain how the multiple independent variables may have positive

relationships with a dependent variable and how the quantitative data may bring variation as a result, a cross-sectional design is a fitting design approach.

(25)

4.3 Data Sources

There are two different ways to collect data for a study: primary and secondary data (Bryman and Bell, 2015). For this study, primary data is needed to fulfill the purpose since it cannot be done by collecting secondary data. Primary data is data collected by the researcher with the goal to solve the purpose which may be accomplished by using surveys, observations or interviews which takes more time as well as more resources but it is crucial in order to test relationships nonetheless (Bryman and Bell, 2015). The most suitable way to collect the primary data was found to be through surveys. The primary data collected for this study will be done specifically towards the purpose of the study and test if the independent variables have an effect on the dependent variable.

4.4 Data Collection Method

In order to collect data for a quantitative approach, there are different methods that the researcher should appoint based on the purpose of the study. The different methods include self-completion questionnaires, observations, experiments or interviews (Bryman and Bell, 2015). However, the researchers of this study found a self-completion questionnaire to be the appropriate method to collect the data since it will help solve the purpose by testing the variables. By conducting a self-completion questionnaire, the sample will be larger which will lead to a better generalization for the whole population. To reach the same amount of respondents by interviews would have taken much more time and resources and would not display any meaningful numerical data that can be statistically analyzed.

It is however crucial to be clear with the questions on a self-completion questionnaire since there is no one the respondents can ask or can answer their questions as in interviews

(Bryman and Bell, 2015). Although, this should not be an issue as long as the questionnaire is structured in a clear and understandable way which can be assured by pre-testing. The

convenience for both the respondents and the researchers is larger when conducting self- completion questionnaires rather than interviews, meaning the questionnaire may be

completed based on the respondents schedule without requiring an interviewer (Bryman and Bell, 2015). Furthermore, close-ended questions will be well presented in order to reduce the

(26)

risk of respondents misunderstanding or skipping any question which would negatively affect the results. Based on the benefits of conducting a self-completion questionnaire, the chosen data method collection for this study will be a self-completion questionnaire which may save time, increase the effectiveness of the study as well as fulfil the purpose.

4.5 Data Collection Instrument

Based on the aforementioned benefits in regard to using a self-completion questionnaire as the chosen data collection method, a fitting data collection instrument must be chosen. Rather than designing the questionnaire through archaic methods such as being paper based and sent through traditional mail, the authors utilized Google Forms. The benefit of having the

questionnaire on Google Forms was that the authors were able to design a self-completion questionnaire that could be effectively structured (See Appendix 2) in order to reduce any confusion and errors from the participants’ side. Additionally, there was no risk of any incomplete answers or wrong answers as the questionnaire could be designed as not submittable if particular questions are unfilled. Furthermore, having the questionnaire on a popular online platform allowed an easy distribution of it through online social media groups and by simply sending the link to acquaintances (as described in the sampling method).

Therefore, by utilizing this data collection instrument, the authors were able to save time and resources in collecting the data, an advantage that is described by Bryman and Bell (2015).

4.5.1 Designing the Questionnaire

When designing the questionnaire, the authors of this paper took certain steps to ensure clarity for the participants and to ensure that missing data was avoided.

To do this, the questionnaires first section had an introductory paragraph informing the participants of the purpose of the research. This section also gave instructions on how to fill in the survey and how the authors would keep their identities anonymous in order to avoid any ethical concerns. Furthermore, this section included the expected time it would take and the total number of questions. Finally, the researchers’ contact information was provided for participants to use should any questions arise.

(27)

After this section, general questions were asked about the participants’ traits (e.g. age, gender, occupation etc.) which are known as control questions. One control question asked the participants if they frequently shopped from online fashion retailers. If they had answered no, the participants were redirected to a different page which thanked them for participating in the questionnaire. This was done to avoid wasting the time of participants as those who did not frequently purchase from online fashion retailers were not relevant to the research

purpose of this paper. Those who had answered yes to the aforementioned question were redirected to the second part of the questionnaire.

The second part asked the participants to fill in what online fashion retailer they purchased regularly from. The logic behind this question was to prepare the participants and make them think about this online fashion retailer when moving on to the next section.

The third and final section of the questionnaire explained to the participants that the online fashion retailer they mentioned in the previous question would be referred to as “Online Fashion Retailer X”. Afterwards, statements were provided and the participants had to select a number on a Likert scale (1-6) to signify to what extent they agreed or disagreed with the statement presented (in relation to Online Fashion Retailer X). After the statements were answered by the participants, they were able to submit the questionnaire and were redirected to a page that thanked them for their participation. The participants were unable to submit the questionnaire unless all control questions and statements were answered. This was done to ensure no collection errors in the data.

4.5.1.1 Measurement

In order to effectively measure the response of the questionnaire participants, a Likert scale was used as previously stated. Bryman and Bell (2015) define a Likert scale as a format that measures a participant’s degree of agreement with a presented statement through a given scale.

In the case of this paper, the authors utilized a Likert scale with a scale ranging from 1 to 6.

The number 1 signified the lower end of the scale and was coded to signify “Strongly Disagree” while the number 6 signified the higher end of the scale and was coded to signify

(28)

“Strongly Agree”. The researchers ensured that this distinction of what the scale meant was presented alongside every statement in the questionnaire to ensure the participants understood it. The authors of this paper chose to utilize a 6 item Likert scale in order to force the

participants to take a stance on the statements presented in the questionnaire. The authors thought that this would show a more clear relationship between variables. This notion is supported by Moors (2007) who states that 6 point Likert scales have the advantage that participants are more likely to agree with positively worded items and disagreed with negatively worded items. This also mitigates the issue of participants using the middle point in an odd numbered Likert scale which might not paint a proper picture due to the

participants’ unwillingness to take a stance (Moors, 2007).

By implementing a Likert scale with closed statements, the authors were able to reduce the chance of getting widely different answers which can occur with open ended questions.

Additionally, implementing a numerical Likert scale allows quantitative statistical

calculations to be conducted such as regression analysis which allows the authors to examine the relationship between the dependent variable and the independent variables.

4.5.2 Operationalization

In order to clarify what measurements and concepts were used, an operationalization table was conducted. An operationalization is often depicted in a table, since what is being measured needs to be presented in an organized and clear way (Bryman and Bell, 2015). An operationalization also assists the writer in connecting certain items to relevant measures in the form of questions or statements (applicable in this case).

An operationalization was applied to this thesis due to the aforementioned statements regarding clear and organized concepts and measurements. In order to make use of the theoretical chapter and the conceptualization of this thesis, an operationalization table was used to facilitate the process of deciding on questions for both the independent- as well as the dependent variables. More specifically, the operationalization consists of the theoretical concept and its source, as well as the connected items and measurements that are connected to the statement.

(29)

Deducted from theory and theoretical items that were retrieved from the theory by the researchers themselves, questions were created to facilitate the chosen purpose. The

theoretical items were created for the sole purpose to facilitate what would be measured and asked out of the respondents. Furthermore, it simplified and narrowed down the large amounts of theory by assigning it with concepts or words that were informative enough to entail what would be fit to ask.

4.5.2.1 Operationalization table

Statement Theoretical Concept

Item Item

Number

Source Measureme

nt I prefer Online

Fashion Retailer X. over other online fashion retailers.

E-loyalty (Cognitive Loyalty)

Preference ELoyalty1 (Kim, Jin and Swinney, 2009;

Oliver 1999)

6 point Likert scale

I have a positive attitude to Online Fashion Retailer X.

E-loyalty (Affective Loyalty)

Positive Attitude

ELoyalty2 Oliver (1999) 6 point Likert scale

I intend to conduct future purchases from Online Fashion Retailer X.

E-loyalty (Conative Loyalty)

Intention for future purchases

I regularly purchase from Online Fashion Retailer X.

E-loyalty (Action Loyalty)

Regular Purchasing

I trust Online Retailer X to be credible.

E-trust Credibility ETrust1 (Corritore, Kracher and Wiedenbeck, 2003;

Merrilees and Fry, 2003)

(30)

I trust that Online Retailer X’s website is easy to use.

E-trust Ease of use ETrust2 (Corritore, Kracher and Wiedenbeck, 2003)

I trust Online Retailer X to meet my expectations.

E-trust Expectations ETrust3 (Singh and Sirdeshmukh 2000)

I trust that Online Retailer X will maintain the privacy of my personal information.

E-trust Personal information processing

ETrust4 (Corritore, Kracher and Wiedenbeck, 2003; Al- dweeri et al.

2019; Cyr, Hassanein, Head and Ivanov, 2007)

I am satisfied with the value of the products offered by Online Retailer X.

E-satisfaction Perceived Value

ESatisf1 (Swaminathan, Anderson and Song, 2018)

I am satisfied with Online Retailer X’s product range.

E-satisfaction Choice ESatisf2 (Swaminathan, Anderson and Song, 2018)

I am satisfied with Online Retailer X’s customer service.

E-satisfaction Customer Care

ESatisf3 (Swaminathan, Anderson and Song, 2018)

I am satisfied with the product

information that Online Retailer X provides about its

E-satisfaction Product Information

ESatisf4

(Szymanski and Hise, 2000)

(31)

I am satisfied with the convenience Online Retailer X provides with its shopping

experience

E-satisfaction Convenience ESatisf5 Schaupp and Belanger (2005)

I perceive that Online Retailer X delivers my orders on time.

E-service quality

Reliability EServiceQ1 Wolfinger and Gilly (2003)

I perceive that Online Retailer X’s service quality meets my

expectations.

E-service quality

Perceived Service Quality

EServiceQ2 (Grönroos 1984; Lewis and Booms, 1983, cited in Parasuraman, Zeithaml and Berry, 1985;

Smith and Huston, 1982 cited in Parasuraman, Zeithaml and Berry, 1985;

Churchill and Surprenant, 1982;

Parasuraman, Zeithaml and Malhotra, 2005)

I perceive that Online Retailer

E-service quality

Relevant Personalized

EServiceQ3 Wolfinger and Gilly (2003)

(32)

X’s provides me with relevant personal

recommendations.

Recommend ations.

Table 1. Operationalization Table

4.5.3 Pre-testing

When designing a questionnaire, it is important to ensure that respondents understand what they are required to do and not feel ambiguity when interpreting the statements presented to them (Hilton, 2015). To avoid these pitfalls, the authors utilized pre-testing of the

questionnaire. Pre-testing is used to check whether questions and statements function as intended and are understood by potential participants (Hilton, 2015).

Therefore, the authors of this paper chose 8 participants through convenience sampling and asked them to complete the questionnaire and make note of any issues they encountered when completing it. The participants were then asked to present the researchers with any feedback regarding the clarity of the questions asked and statements presented. The participants in the pre-test reported no major issues in completing the questionnaire itself and unanimously said they understood the questions and statements presented.

However, the pre-test allowed the researchers to notice and fix some other issues that arose.

Namely, the authors added a redirect action after a participant answered “No” to the control question “Do you regularly purchase from online fashion retailers? (For example: Asos, Zalando, Nelly, NLYman, Nike, etc.)”. This meant that after a no answer was submitted, the participant’s questionnaire was submitted and they were thanked for participating. This was done because they were not part of the relevant sample determined for this study.

Additionally, the question that asked the participants to name the online fashion retailer they frequently purchased from was edited to include a note that asked the participants to only submit one retailer. This was because some participants submitted more than one online fashion retailer which could contaminate the results of this study. Finally, the participants picked up on minor spelling and grammatical errors that were subsequently fixed.

(33)

4.6 Sampling Method

Sampling refers to the method of selecting a subset of a population for a certain investigation according to. The choice of what subset is a choice made by the researcher and should be a choice that fits and defines the study in question (Bryman and Bell, 2015). Therefore it is of importance to the authors of this study to define a suitable population in order to answer the purpose. The population at large for this study are people who are active on the social media channel of choice, Facebook. Since the data collection method of choice is a self-completion questionnaire, Facebook was the distribution channel of choice due to its convenience.

Furthermore, the questionnaire was sent out to the researchers' acquaintances and friends, as well as to Facebook groups where the main topic of discussion is fashion.

Once the subset of the population has been chosen, the sampling method may be defined. If a probability sample is the sampling procedure of choice, each unit should be selected at random, giving all units within the population the chance of being selected for investigation.

The non-probability method implies that some units within the population are more probable to be selected than others, meaning that the units are not selected randomly (Bryman and Bell, 2015; Taherdoost, 2016).

Bryman and Bell (2015) explains the non-probability sampling procedure as an umbrella term that holds two main different types of sampling methods, quota sampling and convenience sampling. A convenience sample is a sample that is most applicable to this research due to it being a sample that is chosen based on its accessibility. Furthermore, a convenience sample may be partly chosen based on what may limit researchers when conducting a study, such as time constraints or limited financial means to conduct the study (Bryman and Bell, 2015;

Ackoff, 1952 cited in Taherdoost, 2016).

Since the aforementioned distribution channel Facebook was chosen out of convenience, the sampling method of choice is partly a non-probability convenience sample, but also an expert sample. Considering that the questionnaire was also sent out to Facebook groups where the main topic of discussion was fashion, it is considered to be more purposeful than the aforementioned convenience sampling method. An expert sample is a sampling tool that is

(34)

useful when a specific research field is being investigated since it requires researchers to purposely sample ‘experts’ in the field. This may be of use to researchers who aim to strengthen the collected evidence to a study (Zafar et al., 2015). Due to the aforementioned criteria of strengthening evidence by approaching specific Facebook groups related to the research area of choice, an expert sample will be used in addition to the convenience sampling method.

4.6.1 Sample selection and Data collection Procedure

Once the targeted population and sample have been defined, a decision on how large the sample may be may arise (Bryman and Bell, 2015). A number of factors may be taken into account in order to make this decision, which is often a decision researchers may find struggling (TEMEL and ERDOGAN, 2017). Israel (1992) argue that the primary factors of importance when deciding on sample size is the purpose of the study, the size of the

population and the level of precision or sampling error in the study. The nature of the research topic is of importance in order to be able to estimate how much useful data that could be obtained from a specific amount of participants. Factors such as quality of the data and the scope of the study are additionally contributing to researchers being able to make estimates in regards to how many participants that might be needed (Malhotra, 2010).

Bryman and Bell (2015) argue that there are a number of factors that should be taken into consideration depending on the study. However, there is no definite answer to the question of how many, rather a question regarding time and cost. While a larger sample size may allow for greater precision due to having less sampling errors, it may also hinder researchers due to it being constraining both in a financial sense as in a timely sense. Since the chosen sample is mainly a convenience sample and the data collection method is in the forms of a self-

completion questionnaire that will be distributed on Facebook, it will be hard to make estimations regarding sample size. A useful tool could perhaps be to take similar studies into account in order to estimate a number (Bryman and Bell, 2015).

Existing research on the topic of e-retailing and e-commerce differs depending on what factors are being measured and in what contexts and there are also a variety of different sample sizes used. What was common amongst Kim, Jin and Swinney (2009), Al-dweeri et al., (2019) and Sai Vijay, Prashar and Sahay (2019) was that they all sent out questionnaires.

(35)

Vijay, Prashar and Sahay (2019) had 210 respondents and Kim, Jin and Swinney had collected useful data from 182 respondents.

While acknowledging that there are a number of factors that may be taken into account when deciding on sample size, Green (1991) constructed a formula to help determine sample size in regards to studies aiming to investigate correlation or regression: 𝑁 > 50 + 8𝑚.Data

collection methods with regression or correlation should aim to collect data from no less than 50 participants, hence pointing towards that N, i.e. the total number of respondents should be no less than 50 and m, the number or independent variables part in a study should always be multiplied by eight. Therefore: 𝑁 > 50 + (8 ∗ 3)which generates 𝑁 > 50 + 24 = 74and proves that N should be no less than 74. In order to reach an estimate regarding the actual sample size instead or the bear minimum, the aforementioned studies sample sizes should also be taken into consideration: 316+ 210 + 182 + 74 = 782.By calculating the mean value, the final estimate was approximately 196, which was decided to be rounded up to 200.

Since the authors of this paper have limited resources time to collect and conduct the analysis to reach a conclusion, the number 200 was only calculated to act as an estimate rather than an exact number.

Once the self-completion questionnaire was closed down, the authors of this paper ended up with 315 completed responses. However, 204 out of the 315 responses were regarded as useful since the remaining respondents did not pass the control questions.

4.7 Data analysis method

Once data has been collected, it may be measured through data analysis. However, one must know the data in order to make choices regarding data analysis techniques. Choosing a data analysis method will help put meaning into the collected data for this study. This was done through the statistical program SPSS, which is the program that was used during the entire data analysis method including the descriptive statistics, the correlation and regression analysis and the reliability test. With the help of SPSS, relationships and correlations that may not be expected could also be detected, it may also prevent researchers from making

(36)

critical mistakes when handling a dataset (Bryman and Bell, 2015; Hellerstein, 2008; Hair et al., 2014).

Since it is of importance to understand the data that is going to be analyzed before entering it into SPSS, it should be sorted through or cleaned. Cleaning the collected data may improve the quality of the data. This may be done by detecting any possible outliers by detecting what may be considered as far from what is expected of the data, based on the rest of the data (Hellerstein, 2008). Bryman and Bell (2015) argue that data should be coded before it is entered to a chosen statistical tool. Coding is a tool that may help researchers improve the quality of the data due to the researcher addressing and categorizing the data by numbers.

Data that once was nonmetric may for example be transformed into metric data, which is otherwise referred to as the process of conducting dummy variable coding. By using dummy variable coding, the researcher is using dummy variables such as 1s and 0s and assigning numbers to subjects. For example if a subject is female, she may be assigned with a 0 (Hellerstein, 2008; Bryman and Bell, 2015).

When examining the data, the authors decided to assign dummy variables to the control variables (age, gender and occupation) in order to allow the variables to be tested along with the independent variables in a regression model (in order to determine their effect on e- loyalty). Furthermore, no outliers were found in the data, so removing data was unnecessary.

4.7.1 Descriptive statistics

Descriptive statistics aims to make assumptions about why and how the data came to be. It is a tool that may assist in the process of both describing and comparing the data numerically (Saunders, Lewis and Thornhill, 2016; Hellerstein, 2018). It is also helpful when sorting through large amounts of data, which is also why descriptive data is often depicted in a frequency table, displaying a summary of the information which is also describing the central tendency. The central tendency may be measured by the arithmetic mean, median and mode (Bryman and Bell, 2015; Malhotra, 2010). The arithmetic mean refers to the process of adding all the distributed values and dividing them by the number of values. The median refers to the midpoint of the distributed values, which is why the mean is sensitive to outliers or extreme values. The mode is the value that occurs most frequently in a distribution.

Antecedents of E-Loyalty in the Context of Online Fashion Retailers

Bachelor Thesis