The Relationship between Customer Satisfaction and Repurchase Behavior in Online Shopping Environments

(1)

The Relationship between Customer Satisfaction and

Repurchase Behavior in Online Shopping Environments

An empirical analysis of online product reviews in the undergarment market

By

Mark van Dijl

S2738422

9741AH, Groningen

vdijlm@gmail.com

+31640197767

University of Groningen

Faculty of Economics and Business

Msc Marketing

Intelligence track

June 2019

(2)

2

Acknowledgement

(3)

3

Abstract

In recent years, online product reviews have become an important source of information for both customers and managers. This master thesis focuses on customer satisfaction, measured through online product reviews consisting of a textual part and a product rating, and its relationship with repurchase behavior. The study provides clarity and nuance in the satisfaction-repurchase relationship in online retail environments by investigating the overall effect of customer satisfaction on repurchase behavior, as well as taking into account the reasons for the satisfaction level based on product quality dimensions. The reasons for satisfaction levels are classified into three distinct categories, namely Design Quality, Product Life Elements and Product Conformance. The results of this thesis study show a positive relationship between customer satisfaction and repurchase behavior, when satisfaction is measured through the numerical product rating, while this relationship does not exist when satisfaction is measured through the textual part of an online product review. Furthermore, the results of this thesis study show that the reasons for satisfaction do not have an influence on the satisfaction-repurchase relationship. The outcomes indicate that managers should try to improve the online product rating in order to increase the repurchase behavior of their customers. Based on the findings of this thesis study, it is proposed that managers should not discriminate in their approach when responding to customer satisfaction levels expressed in online product reviews in their attempt to influence repurchase behavior.

(4)

4

1. Introduction

Classical research on the influence of word-of-mouth shows that “happy customers tell 3 friends about their experiences with a company, while angry customers tell about their experiences to 3000 people” (Chatterjee, 2001). The rise of online social communities has profoundly changed the way consumers’ word-of-mouth messages are disseminated, overcoming the classical limitations of word-of-mouth. The previous short-lived word-of-mouth, usually targeting a small group of friends, has changed into lasting messages, which remain visible to the entire world (Duan, Gu & Whinston, 2008). This change in the nature of word-of-mouth has led to the emergence of internet-mediated communities, where consumers express their opinions about services and products. In contrast to traditional word-of-mouth communities, where tales about product or service experiences rapidly “disappear into thin air”, internet-mediated communities endlessly document this customer-authored information, which can later easily be retrieved by almost every interested party (Dellarocas, Zhang & Awad, 2007). As a result, the seemingly far-fetched situation in which a customer is telling 3000 people about his experience with a company’s products or services quickly became reality in recent years.

Nowadays, electronic word-of-mouth and online user reviews are influential sources of information for consumers, substituting and complementing other forms of word-of-mouth communication concerning the quality of various products (Chevalier & Mayzlin, 2006; Goes, Lin & Au Yeung, 2014). According to a survey by Deloitte’s Consumer Products group (Deloitte, 2007), almost two-thirds of consumers read consumer-written product reviews on the Internet. Among those consumers who read reviews, 82 percent states that their purchase decision was affected by the content present in the review and 69 percent communicates about the content of the review with family, friends and colleagues, thus amplifying their impact. A survey by ACNielsen (2007) showed that most customers regard customer-authored opinions posted online to be just as trustworthy as information provided through a brand’s website, and Rowley (2001) even proposed that commercial enterprises should try organizing online communities rather than to simply advertise on the Internet.

(6)

6

in order to understand the underlying concepts expressed in online product reviews: a study by Engler, Winter & Schulz (2015) shows that online product ratings in fact represent the customer’s satisfaction with the product, while a study by Xiang, Schwartz, Gerdes & Uysal (2015) based on online product review text analytics showed that customers “talk about” their experiences in online product reviews, and that these expressions to a great extent relate to expressions of customer satisfaction. Furthermore, Li, Ye & Law (2013) state that the textual part of online product reviews may capture customer satisfaction more precisely than surveys, which leads to a more precise measure of the construct. Therefore, it can be concluded that online product reviews are a representation of customer satisfaction.

Customer satisfaction is considered to be the most important predictor of repurchase behavior (Jones & Suh, 2000), and overall customer satisfaction has been identified as a prominent offline driver of customer retention in the marketing literature (Gustafsson, Johnson & Roos, 2005). The rise of electronic word-of-mouth expressed customer satisfaction also has an impact on customer repurchase behavior: previous research has identified electronic word-of-mouth (and therefore online product reviews) to be an antecedent of the customers’ intention to revisit online retailers in order to perform repurchases (Gruen, Osmonbekov & Czaplewski, 2006). It is therefore plausible that in online environments customer satisfaction, measured through online product reviews, drives customer repurchase behavior.

It is common knowledge in the field of marketing that retaining customers creates more value for a company than acquiring new customers (Bitran & Mondschein, 1997). Previous research has shown that the costs of attracting one extra customer are approximately five-fold higher than those of retaining an existing customer (Verhoef & Donkers, 2001). Therefore, it is important for a brand to create loyal customers, because they are loyal consumers of the brand, who will perform repurchases and recommend the brand to people in their personal network (Ercis, Ünal, Candan & Yildirim, 2012).

(7)

7

insights into customers’ affective responses to the purchased product. Because of the voluntary nature and open structure of online product reviews, in combination with the anonymity of respondents, big data samples are easily accessible. Therefore, measuring customer satisfaction through online product reviews overcomes the classical limitations of measuring the construct through surveys (Xu, Wang, & Haghghi, 2017), while also probably being more precise (Li et al, 2013).

Not all markets offer the opportunity to observe repeat-purchases, which may be one of the reasons why customer satisfaction does not always seem to lead to repurchase behavior (Chitturi, Raghunathan & Mahajan, 2008; Reichheld, 1996). Peyrot & Van Doren (1994) state in their research that the majority of purchases made by customers are potential repeat purchases, which can repeatedly involve the same seller. Common repeat-purchase products are products that are chosen based on simple heuristics, like brand awareness, pricing or packaging, and where the product is evaluated after the purchase (Ray, Sawyer, Rothschild, Heeler, Strong & Reed, 1973). An exception to this claim is formed by one-time purchases (Peyrot & Van Doren, 1994). To be able to observe consumers repeatedly buying (similar) products from the same seller, this thesis study is conducted at a Dutch lingerie retailer.

Although some research has already been done to link the concepts of customer satisfaction, online product reviews and customer repurchase behavior, most of these studies only investigated the relationship between online product reviews and their relation to the repurchase behavior of other customers. By examining the textual and numerical parts of online product reviews at a big lingerie retailer and the possible repurchases by the customers, it is investigated in this thesis study whether customer satisfaction measured through individual online customer reviews influences customer repurchase behavior. Further, this study is based on the notion that, by classifying reviews into categories with different reasons for customer (dis)satisfaction and by examining the influence of each category on repurchase behavior, a more thorough and detailed understanding of the influence of customer satisfaction on repurchase behavior can be obtained.

(8)

8

reluctant in repurchasing and tend to switch to another retailer, even though they are satisfied (Chitturi et al, 2008; Reichheld, 1996). The results of the comparative analysis of categorized online product reviews and repurchase behavior identify whether the reasons for customer satisfaction levels can alter the effect of customer satisfaction on repurchase behavior. It is possible that the categorized online product reviews influence the satisfaction-repurchase relationship, thereby causing the ambivalent outcomes.

Accordingly, it is anticipated that the results of this thesis study will enable managers to better predict customer repurchase behavior based on customer satisfaction, while at the same time guidelines can be defined for companies in handling customer reviews with differing valence and reasoning. Furthermore, the results will help companies to better understand why their satisfied customers (do not) repurchase products, which can help them to better shape their marketing strategies. To achieve these objectives, the following research questions are addressed:

- What is the influence of customer satisfaction, measured through online product reviews, on the repurchase behavior of the customer who wrote a review in an online shopping environment?

- Do the antecedents of customer satisfaction have an influence on the determined relationship between customer satisfaction and customer repurchase behavior?

Chapter 2 of this thesis presents a theoretical framework of customer satisfaction and the product quality dimensions determining the level of customer satisfaction, online product reviews and repurchase behavior. It also explains the links between the concepts. Chapter 3 details the applied methodology and explains the data collection and analysis. In chapter 4, the results of the research are presented, while chapter 5 consists of a discussion of the main results. In chapter 6, several recommendations are given. In the final chapter 7, limitations are discussed and possible areas for further research are presented.

2. Conceptual Framework and Hypotheses

(9)

9

product review components, which reflect customer satisfaction, and their impact on repurchase behavior. Also, the way in which customer satisfaction can be measured through online product reviews is explained. Product quality dimensions are major drivers of customer satisfaction that influence the repurchase probability of products belonging to particular categories, and therefore a distinct section of this chapter is devoted to them. These quality dimensions form the basis for the present analysis of the relationships between the different reasons for customer satisfaction levels and repurchase behavior. Lastly, based on the consulted literature, five research hypotheses are proposed and a conceptual model is presented.

2.1 Customer Decision Journey

A consumer decision journey is a model of the different stages or steps customers might go through in their evolving relationship with a particular brand, consisting of cognitive (thinking) stages, affective (feeling) stages and conative (doing) stages (Wijaya, 2015). It also includes steps that happen before, during and after a purchase (Batra & Keller, 2016). Identifying the different stages in the customer decision journey can be very insightful (Batra & Keller, 2016), because this provides an overview of how consumers systematically transform their needs/wants into a purchase decision, by narrowing their consideration set and weighing options, while at the same time showing that the post-sale period is extremely important in determining a consumer’s loyalty and their likelihood of buying products again (Court, Elzinga, Mulder & Vetvik, 2009). By starting this literature review with an overview of the (relevant) stages in the consumer decision process, the context in which the present research is done is clarified and easier to comprehend. Furthermore, the customer decision journey has been subject to a lot of changes over the past years. These changes instigated a desire for new insights, which are partly provided in this research.

(10)

10

hopefully, over time, (8) frequently utilize the product and start repurchasing the brand; (9) engage in post-purchase interactions with the brand; and (10) become a devoted customer who is willing to promote the brand. Figure 1 shows a traditional customer decision journey, consisting of the various stages that customers go through when developing a relationship with a brand.

Figure 1. Traditional customer decision journey (Batra & Keller, 2016).

In recent years, the consumer decision journey has changed a lot, becoming shorter in length, less hierarchical and more complex (Court et al, 2009). Because of the emergence of the World Wide Web, through which customers nowadays have the opportunity to voice their opinions, complaints and recommendations on products and firms (Chatterjee, 2001), the volume and the manner of communication between consumers and firms has fundamentally changed. Where previously consumers only passively received information about brands through mass media, consumers nowadays actively engage in information searches into interesting brands, by examining the brand’s website and complementing this information with information found through other sources, like blogs and product reviews (Batra & Keller, 2016; Zhang, Craciun & Shin, 2010).

(11)

11

and post-purchase phases. Compared to the traditional customer decision journey (Batra & Keller, 2016), the pre-purchase phase encompasses all aspects of the customer’s interaction with the brand, category and environment before a purchase transaction. It thus represents the “Needs/Wants”, “Is Aware/Knows”, “Considers/Examines”, “Searches/Learns”, “Likes/Trusts”, “Sees Value/Is Willing to Pay” and “Commits/Plans” stages. The purchase phase covers all customer interactions with the brand and its environment during the purchase event itself, thereby representing the “Consumes” stage. The post-purchase phase encompasses the customer interactions with the brand after the actual purchase, representing the “Is Satisfied”, “Is Loyal/Repeat Buyer”, “Is Engaged/Interacts” and “Actively Advocates” stages. The pre- and post-purchase phases are the phases that have undergone the most drastic changes due to recent technological developments.

In the pre-purchase phase, companies are no longer the only expert on the performance and quality of their product or services. Through online product reviews, customers either voice their complaints or freely advertise the product or service based on their experiences (Hudson & Thal, 2013). Other customers can easily turn to this new product/service authority to gain information in the evaluation stage.

In the post-purchase phase, the customer nowadays has more opportunities to interact with the brand. In the traditional customer decision journey, post-purchase relationships were typically focused on the use of the product or service. However, this has changed because of the rise of social media websites, which enable customers to engage with brands beyond the traditional interactions (Hudson & Thal, 2013). The change in the nature of post-purchase interactions is illustrated by the acts of customers of a certain facial skin product, where 60% of them conducted an information search on the product after the purchase was made (Court et al, 2009).

Because of this change in dynamics, more research into this new situation is needed. Therefore, the present research focuses on the transition from the ‘is satisfied’ to the ‘is loyal/repeat buyer’ phase in the consumer decision journey. In the following section of this chapter, the concepts influencing customer behavior in these phases will be further explained.

(12)

12 2.2 Customer Satisfaction

In the ‘is satisfied’ phase of the customer decision journey, the consumer makes an assessment of her/his satisfaction with the purchased brand. Therefore, this section is focused on the concept of customer satisfaction.

Although the concept of customer satisfaction has been subject to change during the past few decades, researchers generally agree that customer satisfaction is an assessment of the overall experience of consumption (Johnson, Anderson & Fornell, 1995; Oliver, 1997). Customer (dis)satisfaction is defined as “the result of a post-consumption or post-usage evaluation, containing both cognitive and affective elements” (Homburg, Koschate & Hoyer, 2005, p. 3). According to the expectancy-disconfirmation paradigm (Oliver, 1980), satisfaction levels are based on the customers’ comparison of previously held expectations with the observed performance of a product or service. In addition, the negative or positive affect that originates from the cognitive process of disconfirmation/confirmation, contributes to the customer’s satisfaction level (Oliver, 1993; Oliver, Rust & Varki, 1997).

Customer satisfaction is considered to be one of the most desired results of all marketing endeavors in market-oriented firms (Kandampully & Suhartanto, 2000). Therefore, the present research also addresses satisfaction with the “performance” of products or services, which is a post-consumption assessment of the perceived quality compared to the initially held beliefs and expectations about the quality of the product or service (Anderson, 1994; Anderson & Sullivan, 1993; Bitner, 1990; Churchill & Surprenant, 1982; Oliver, 1980; Oliver & DeSarbo, 1988; Tse & Wilton, 1988).

2.2.1 Relationship Customer Satisfaction and Repurchase Behavior

(13)

13

retention (Gustafsson et al, 2005), and satisfaction was also reported to be a strong single predictor of repurchase behavior, sometimes explaining more than 50% of the repurchase behavior (Jones & Suh, 2000). Several studies also found a positive relationship between customer satisfaction, measured through surveys, and repurchase behavior in online shopping environments (Anderson & Srinivasan, 2003; Atchariyachanvanich, Okada & Sonehara, 2007; Tsai & Huang, 2007). Based on the findings of the presented literature, customer satisfaction is taken as an antecedent of repurchase behavior.

Nevertheless, explaining repurchase intention and behavior using only customer satisfaction is questionable (Capraro, Broniarczyk & Srivastava, 2003). A study by Reichheld (1996) reported that only 15%-35% of satisfied customers actually came back, while Chitturi, Raghunathan & Mahajan (2008) also state that it might surprise companies that 60% of the customer that switch retailers still classify themselves as ‘satisfied’.

The strength of the effect of customer satisfaction on customer retention also varies and depends on elements inherent to the relevant industry (Jones & Sasser, 1995), and on other social factors (Oliver, 1999). Mittal & Kamakura (2001) show in their study that customer characteristics influence the satisfaction-repurchase relationship. Their results show that at the same customer satisfaction level, older woman are more likely perform repurchases than younger woman. It is possible that younger woman are still eager to search for information on differing brands, while older woman already accumulated a large amount of brand information, which leads to a saturation effect. Because older woman are less interested in alternatives, they are less likely to switch.

Most interestingly, the vast majority of the studies showing a relationship between customer satisfaction and customer repurchase behavior were conducted in a traditional shopping context. Whether the results of these studies can be extrapolated to an online shopping setting (Khalifa & Liu, 2007), and whether these results are the same when customer satisfaction is measured through online product reviews, is presently not clear.

2.3 Repurchase Behavior

(14)

14

their relationships, commonalities and differences. Furthermore, a description of the concept of commitment, which often drives repurchase intention and behavior (Ercis et al, 2012), is presented.

The repurchase behavior stage is the final stage in a simplified consumer decision process, when the customer assesses whether to continue purchasing a particular product (Wu, Tao & Lin, 2017). Repurchase behavior is defined as “the objectively observed level of repurchase activity” (Seiders, Voss, Grewal & Godfrey, 2005, p. 27). In general, the vast majority of consumer purchases represent a series of events rather than a single isolated event (Peyrot & Van Doren, 1994).

Keaveney (1995) found that customers switching behavior is based on pricing, inconvenience, core service failures, failed service encounters, response to failed service encounters, competition and ethical problems. On the other hand, research by Bolton & Lemon (1999) shows that customers continue their purchasing based on previous satisfaction levels, their assessment of payment levels, and prices, where customer satisfaction is found to be the main determinant of repurchase behavior (Gilly & Gelb, 1982; Solnick & Hemenway, 1992). If a consumer’s experience with a product is pleasant, the reduction of post-choice dissonance should lead her or him to develop more positive feelings for the selected brand, thereby increasing the chance of repurchase (Mittelstaedt, 1969).

A concept that is very similar to repurchase behavior is the concept of loyalty. Initially, studies measured loyalty in terms of repurchase behavior (Brown, 1953), or in terms of the probability of a product repurchase (Kuehn, 1962; Lipstein, 1959). However, later studies mostly agree that brand loyalty consists of an attitudinal part next to its behavioral component (Jacoby & Keyner, 1973). Therefore, Dick & Basu (1994, p. 102) defined customer loyalty as “the relationship between relative attitude and repeat patronage”. In a more recent study by Law, Hui & Zhao (2004, pp. 547-548), loyalty is defined as “a deeply held commitment to rebuy or patronize a preferred product/service consistently in the future, thereby causing repetitive same-brand or same brand-set purchasing, despite situational influences and marketing efforts having the potential to cause switching behavior”. Loyalty is not limited to brands, as customers can also be loyal to a particular store (Corstjens & Lal, 2000).

(15)

15

cognitive sense, then proceeds to become loyal in an affective sense and, still later, in a behavioral manner. Loyalty leads to reduced marketing costs, more new customers, greater trade leverage, favorable word-of mouth and greater resistance among loyal customers to rival marketing efforts (Aaker & Equity, 1991; Dick & Basu, 1994). The behavioral part of loyalty is often referred to as the repurchase behavior (Gommans, Krishnan & Scheffold, 2001). Furthermore, Chaudhuri & Holbrook (2001) find that the behavioral component of loyalty leads to a higher market share, while the attitudinal component leads to a higher relative price for the product. Because in this research the focus is not on the attitudinal part of consumer behavior, repurchase behavior is taken as the focal concept.

Retention is another term that is commonly used for repurchasing (Curtis, Abratt, Rhoades & Dion, 2011), and the concept also focuses on repeated patronage of a marketer (Hennig-Thurau & Klee, 1997). Keiningham, Cooil, Aksoy, Andreassen and Weiner (2007, p. 364) define customer retention as “a customer’s continuation of business relationship with a firm”. A more extensive description is provided by Hansemark & Albinsson (2004, p. 42), who define customer retention as “the customer’s liking, identification, commitment, trust, willingness to recommend, and repurchase intentions, with the first four being emotional-cognitive retention constructs and the last two being behavioral intentions”.

Many firms invest significant amounts of money into customer retention strategies, based on the potential to possibly capture their long-term value (McDougall, 2001), which is logical when examining the promised profits: When customer defection is decreased by 5%, profits can increase to as much as 95%, depending on the industry (Reichheld, 1996). It is shown that ‘old’ customers of a company pay less attention to marketing efforts by competing companies, are less price-sensitive and spread positive messages about their experiences with the focal company (Desai & Mahajan, 1998). Furthermore, the company already understands the needs and wants of the customer and the initial attraction costs are already incurred (Davidow & Uttal, 1989).

(16)

16

of customer retention focuses on the managerial aspects of keeping customers (Jacoby & Chestnut, 1978).

The present study investigates whether there is an effect of (the reasons for) customer satisfaction levels, measured through online product reviews, on the repeated patronage of a customer. This link in online retail environments has to be established first, before managerial aspects in the satisfaction-repurchase relationship can be addressed. Therefore, repurchase behavior is taken as the focal concept in this thesis study. By understanding the relationships between customer satisfaction levels as reflected in online product reviews and repurchase behavior, managers can better shape their future retention strategies.

In the literature dealing with relationship management, next to customer satisfaction, commitment is often identified as a driver of customer retention (Gounaris, 2005; Gustafsson et al, 2005), loyalty (Fullerton, 2003) and repurchase intentions (Ercis et al, 2012). Several definitions of commitment exist in the literature: commitment is defined as “a desire to maintain a relationship” (Moorman, Deshpande & Zaltman, 1992, p. 316), as “a desire to develop a stable relationship, a willingness to make short-term sacrifices to maintain the relationship and a confidence in the stability of the relationship” (Anderson & Weitz, 1992, p. 19), and as “a pledge of continuity between parties” (Dwyer, Schurr & Oh, 1987, p. 19). From these definitions, two dimensions of commitment can be obtained, namely the affective commitment and the calculative commitment (Fullerton, 2003; Hansen, Sandvik & Selnes, 2003; Johnson, Gustafsson, Andreassen, Lervik & Cha, 2001).

(17)

17 2.3.1 Product (Category) Characteristics Influencing Repurchase Behavior

In this section, product characteristics influencing the level of repurchase behavior are explained. A distinction is made between utilitarian and hedonic product characteristics and the concept of interpurchase time is introduced. Afterwards, a description of the possible influence of switching barriers is presented, which can underlie a situation where consumers are forced into continued patronage of a certain product or service.

Consumers usually repurchase products based on the experienced post-consumption emotions (Oliver, 1997), which determine the level of customer satisfaction (Chitturi et al, 2008). In their research into the reasons that explain why high customer satisfaction levels do not by definition lead to loyal customers, Chitturi, Raghunathan & Mahajan (2008) observed that customers differ in their behavioral response to post-consumption emotions triggered by hedonic product characteristics compared to utilitarian product characteristics.

Utilitarian product characteristics are cognitively driven, instrumental, goal-oriented and accomplish a functional or practical task (Strahilevitz & Myers, 1998). Hedonic product characteristics provide an affective or sensory experience of aesthetic or sensual pleasure, fantasy and fun (Hirschman & Holbrook, 1982). From a customer’s perspective, hedonic characteristics are related to affective preferences (“wants”), while utilitarian characteristics are related to cognitive and reasoned preferences (“shoulds”) (Bazerman, Tenbrunsel & Wade-Bezoni, 1998).

Based on their study, Chitturi, Raghunathan & Mahajan (2008) state that the positive emotional response to utilitarian product characteristics consists of emotions of confidence and security. On the contrary, the positive emotional response caused by hedonic product characteristics consists of emotions of cheerfulness and excitement. The results of their research show that the emotional response caused by hedonic product characteristics, as well as the emotional response caused by utilitarian product characteristics positively influence customer loyalty and repurchase intentions, but the former causes a stronger positive effect than the latter. Summarized, their theory implies that the characteristics of the products sold by a retailer determine the probability that a customer’s satisfaction level leads to repurchase behavior (Chitturi et al, 2008).

(18)

18

shopping once every specific period of time (Chiang, Chung & Cremers, 2001; Kahn & Schmittlein, 1989), and that within product categories, brand interpurchase times only vary slightly (Ehrenberg, 1972; Uncles, Ehrenberg & Hammond, 1995). Based on these findings, it can be concluded that interpurchase times remain fairly constant within product categories.

Previous studies have identified interpurchase time to be an important component of customer (re)purchase decisions, because customers need to decide what, when, and how much to buy (Gupta, 1988). Oliveira-Castro, James & Foxall (2011) reported that, on average, the time between purchases becomes longer when customers have made large purchases. This finding resembles the conclusion from an earlier study by Peyrot & Van Doren (1994), who stated that most purchases made by customers are potential repurchases, with the exception of one-time purchases. Based on these studies, it can be concluded that most product categories provide the opportunity to observe repurchase behavior, and that the goodness-of-fit to examine repurchase behavior depends on the characteristics of the product (category). It goes without saying that customers purchase durables, such as houses, at a lower frequency than food products and clothing. Therefore, to examine a possible relationship between customer satisfaction, measured through online product reviews, and repurchase behavior, the product category examined in the present study is women’s undergarment.

For a full understanding of repurchase behavior, it is important to also include an explanation of the concept of switching barriers (Bendapudi & Berry, 1997). Switching barriers can be defined as “the consumer’s assessment of the resources and opportunities needed to perform the switching act, or alternatively, the constraints that prevent the switching act” (Ranaweera & Prabhu, 2003, p. 379). Switching constraints (costs) are the investment of time, money and effort that, in a consumers’ perception, make it difficult to switch.

(19)

19

in these settings usually doesn’t take considerable time, effort or thought (Ranaweera & Prabhu, 2003), and therefore switching barriers are left out of the conceptual model.

2.4 Online Product Reviews

2.4.1 Online Product Reviews and their Impact on Customer Decisions

Online product reviews can be defined as “peer-generated product evaluations posted on company or third-party websites” (Mudambi & Schuff, 2010, p. 186). An online product review usually consists of two parts: a textual review, that describes the characteristics (e.g. advantages and disadvantages) of a product, and a product rating (usually ranging from 1 to 5 stars), that represents a customer’s opinion on a specified scale (Lackermair, Kailer & Kanmaz, 2013; Mudambi & Schuff, 2010)

There has already been quite some research done in the field of online product reviews (Liu & Park, 2015). Dellarocas, Zhang & Awad (2007) demonstrated that adding online product reviews to their benchmark model improved its forecasting accuracy. Previous research also showed that in their relationship with product sales, online product reviews function as both influencer and predictor (Dellarocas et al, 2007). Further, Chevalier & Mayzlin (2006) and Senecal & Nantel (2004) independently reported that online product reviews affect consumer purchase decisions.

When evaluating online product reviews, customers do not only take the content of the review into account. In particular, Hu, Liu & Zhang (2008) found that customers also pay attention to contextual information, such as the reviewer’s reputation and reviewer exposure.

(20)

20

Nonetheless, Ghose & Ipeirotis (2011) pointed out that product ratings might not fully capture the polarity of information in online product reviews. Although focusing on the textual part of the online product reviews, for example by using the expressed customer experiences and feelings to come to a decision, is a more effortful strategy (Shah & Oppenheimer, 2008; Zhang & Markman, 2001), Chevalier & Mayzlin (2006) obtained evidence that customers often read the textual part of online product reviews, rather than basing their decision solely on simple summary statistics.

2.4.2 Online Product Reviews as a Measure of Customer Satisfaction

Based on observations presented in the previous section, it can be concluded that a substantial part of the previous research into online product reviews was focused on the relationship between (online) product reviews and the repurchase behavior of other customers. However, online product reviews can also present valuable information for a company, because the reviews offer a unique, real-time opportunity to monitor customer attitudes towards the company and its offerings. The obtained insights can then be used in better shaping the company’s manufacturing, distribution and marketing strategies (Dellarocas et al, 2007). Interestingly, however, to my knowledge no previous study has examined the relationship between customer satisfaction expressed through an online product review and the same customer’s repurchase behavior.

As outlined in the previous section, customers use the numerical ratings and textual parts of online product reviews in different ways when considering them as sources of information. However, there is also a difference in what is expressed by these distinct components: Sahoo, Dellarocas & Srinivasan (2018) state that online product reviews with comparable numerical ratings can contain significantly different amounts of information in the textual part of the review. Sometimes, the textual part merely comprises an explanation supporting the numerical rating, while in other reviews the textual part covers rich details about experiences with different product characteristics (Sahoo et al, 2018). Therefore, Ullah, Amblee, Kim & Lee (2016) describe online numerical product ratings as a summary statistic, while the textual part can be described as a context-specific, more detailed explanation of specific product aspects (Hu et al, 2014).

(21)

21

numerical product ratings can reflect customer satisfaction, while a later study by Engler, Winter & Schulz (2015) indicated that such online product ratings in fact represent the customers’ satisfaction with the product. Their conclusions were drawn, based on a customer satisfaction model of online product rating, which showed improved explanatory results compared to traditional quality-centric ratings in describing rating scores. A study by Xiang, Schwartz, Gerdes & Uysal (2015) based on online product review text analytics showed that customers “talk about” their experiences in online product reviews and that these expressions to a great extent compare to expressions of customer satisfaction. Moreover, Li, Ye & Law (2013) also argue that online product review text is a good representation of customer satisfaction.

It can thus be concluded that the numerical ratings and the textual parts of the online product reviews are representations of customer satisfaction, but that the information provided by both components can differ. The summarizing nature of the numerical online product rating provides a static, clear and easy to comprehend overview of the level of satisfaction, but by only looking at this rating the reasons for the expressed customer satisfaction levels remain unidentifiable. On the contrary, the textual part of the online product review provides a detailed overview of the feelings of the customer (Hu et al, 2014). The satisfaction level expressed through the text sentiment is a more volatile representation of customer satisfaction, which can contain both positive and negative expressions.

Hu, Zhang & Pavlou (2009) state in their research that submitted online product reviews have a J-shaped distribution, indicating that most online product reviews are very positive, some are negative, and almost no online product reviews express a moderate customer satisfaction level. According to their research, people with moderate assessments of the purchased product are less passionate to spend time and effort on voicing their opinion. Therefore, they state that customers are more likely to write reviews when they are extremely satisfied or extremely dissatisfied. Because the textual part of the online product review consists of a detailed evaluation of specific product characteristics (Hu et al, 2014), it is possible that the textual part of the online product review is used by customers to express their satisfaction with parts of the product that are really liked or disliked, while the online product rating might be more a representation of the overall satisfaction with the product.

(22)

22

managers, who are trying to come up with a tailored response to received online product reviews, in order to increase the chances for a repurchase by the customer who wrote the review.

Therefore, to truly understand the influence of customer satisfaction on customer repurchase behavior in online environments, and to understand the influence of the quality dimensions that determine the level of customer satisfaction, the effect of customer satisfaction levels measured through online product ratings, as well as the effect of customer satisfaction levels expressed through sentiments in the text, on repurchase behavior needs to be examined.

2.4.3 Numerical Ratings

The reviewers’ online product rating on a scale from 1 to 5 stars (Ullah et al, 2016) can be defined as “the summary online evaluation of the product and a parsimonious word-of-mouth metric that influences other consumers’ purchase behaviors and their willingness to pay” (Sridhar & Srinivasan, 2012, pp. 71-72).

The main reason for customers to make use of online product ratings, is to simplify their decision-making process (Dabholkar, 2006). Chevalier & Mayzlin (2006) investigated the impact of the average product rating on sales, and they found that one-star ratings have a larger influence on sales than five-star ratings. Sun (2012) reported that the variance in the product ratings is an indication of the product type, where a high variance indicates a niche product that some love and others hate.

Studies examining the relationship between online product ratings and consumers’ purchasing decisions mostly find a significantly positive influence (Lin, Lee & Horng, 2011; Mauri & Minazzi, 2013; Park, Lee & Han, 2007; Sun, 2012). Because online product ratings reflect customer satisfaction (Engler et al, 2015; Moon, Bergey & Iacobucci, 2010), it is expected that the given numerical rating positively influences the repurchase behavior of the customer who gave the rating.

(23)

23

In their research, Mittal & Kamakura (2001) show that, although customers with different characteristics have different repurchase probabilities, customer satisfaction leads to repurchase behavior in an offline retail environment. Furthermore, several studies (Anderson & Srinivasan, 2003; Atchariyachanvanich et al, 2007; Tsai & Huang, 2007) found that customer satisfaction, measured through surveys in online environments does also lead to repurchase behavior. Therefore, in the present study the following H1 hypothesis is proposed:

H1: The more positive the numerical rating part of the online product review, the higher the probability that the customer will perform a repurchase

2.4.4 Valence of Text

While numerical ratings can be described as a summarized statistic on a standardized scale, sentiments expressed in the text can be defined as “tactic, context-specific explanations of the reviewer’s feelings, experiences, and emotions about the product or service” (Hu et al, 2014, p. 4). Given that numerical ratings are representations of the product value, or of the product quality, the complementary information captured by the textual part of the online product review is primarily useful for customers to evaluate to what extent product characteristics and features match their own unique preferences. Hence, the text in the review tells something about the fit with a customer’s personal needs.

Results from studies on the explanatory power of the valence of online social media conversations are interesting, but they do show some inconsistencies: Godes & Mayzlin (2004) examined the relationship between Usenet conversations and Nielsen (viewership) ratings. Their results demonstrate that the distribution of conversations has significant explanatory power, while the valence does not seem to have an influence. Liu (2006) studied the impact of Yahoo! Movies prerelease message board discussions on motion picture box office revenues, which showed that the volume of the present dialogs has explanatory power, but their valence doesn’t.

(24)

24

(Anderson & Srinivasan, 2003; Atchariyachanvanich et al, 2007; LaBarbera & Mazursky, 1983; Mittal & Kamakura, 2001; Tsai & Huang, 2007; Zeithaml et al, 1996). Therefore, the second hypothesis (H2) for this thesis study is defined as follows:

H2: The more positive the customer-authored part of the online product review, the higher the probability that the customer will perform a repurchase

Although both the numerical rating and the valence of the customer-authored part of the online product review are expected to have a positive influence on the repurchase behavior of a customer, it is still important to measure the effect of customer satisfaction on repurchase behavior through both components. It could be possible that the numerical rating expresses the overall satisfaction, while the textual part of the review express the satisfaction with certain product aspects. By testing both the H1 and H2 hypotheses, it can be clarified whether or not both components should be employed when investigating the relationship between customer satisfaction levels expressed in online product reviews and repurchase behavior.

Utz, Kerkhof & van den Bos (2012) state that customers usually focus their evaluations in online product reviews on the quality dimensions of a product. According to their study, the quality dimensions of customer service are difficult to judge in online product reviews. After examining the online product reviews used for the present study, it becomes clear that almost all the online product reviews are indeed focused on satisfaction with the product quality aspects. In the rare case that customer’s express their satisfaction with service quality aspects, this only comprises a minor part of the content in the online product review. Therefore, this thesis study focuses on product quality dimensions as the main drivers for customer satisfaction levels expressed in the examined online product reviews.

(25)

25 2.5 Product Quality

Product quality can be defined and measured as “belief statements or attribute performance” (Olsen, 2002, p. 241). Initial research on the concepts of product quality and customer satisfaction led to uncertainty about the distinctiveness of the two concepts (Dabholkar, 1993; Oliver, 1993). For example, a study by Spreng & Singh (1993) did not succeed in discriminating between the two concepts. However, later studies did report discriminant validity, although they also found a very high correlation (Bansal & Taylor, 1997; Dabholkar, 1995). After determining that product quality and customer satisfaction can be interpreted as two separate constructs, the sequence of occurrence in the mind of the consumer has become relevant (Dabholkar, Shepherd & Thorpe, 2000). Oliver (1993) proposed that product quality is an antecedent of customer satisfaction. The resulting model was validated by various other investigations, who all supported this notion (Anderson & Sullivan, 1993; Spreng & Mackoy, 1996). Therefore, I assume in my conceptual model that product quality influences the level of customer satisfaction.

The relationship between product quality and behavioral intentions was investigated by different researchers, which led to differential findings, conditional on the industry in which the research was conducted: Gotlieb, Grewal & Brown (1994) found that customer satisfaction mediates the relationship between quality and behavioral intentions in a study conducted in a medical environment, while Bansal & Taylor (1997) could not identify such a relationship in their research conducted in the banking sector.

The link between product quality, customer satisfaction and repurchase intentions was investigated by several researchers, who found a positive relationship, although this relationship tends to vary per industry, product or situation (Fornell, Johnson, Anderson, Cha & Bryant, 1996; Johnson, et al, 2001). Olsen (2002) addressed the relationship between product quality, customer satisfaction and repurchase behavior, and reported that there exists a direct positive effect between product quality and repurchase behavior, while customer satisfaction also acts as a mediator in the relationship between product quality and repurchase behavior.

(26)

26

to customer satisfaction and repurchase behavior. In turn, this leads to a more thorough understanding of the relationships between product quality, customer satisfaction and repurchase behavior in online retailing environments. Furthermore, based on the results, firms can decide which kinds of product reviews pose threats or opportunities, and which reviews should not receive any attention.

Garvin (1988) identified eight quality dimensions, on which customers base their level of satisfaction with a product. These dimensions are: performance, features, reliability, durability, serviceability, conformance, perceived quality and aesthetics. In the next parts of this thesis, each of the quality dimensions is defined. Furthermore, some of the quality dimensions are grouped, in order to answer the second research question, namely whether antecedents of customer satisfaction have an influence on the determined relationship between customer satisfaction and customer repurchase behavior.

2.5.1 Design Quality

According to Garvin (1988), the performance dimension deals with the primary purpose of the product and how well the product is achieving its objective. In an earlier study, Garvin (1984) defined performance as the primary operating characteristics of a product. For a lingerie product, for example a bra, a performance trait would be the offered body support.

According to Garvin (1988), features are the added touches, bells and whistles, secondary characteristics that the product possesses and/or extra features present in the product. Product features are usually added to products to ‘spice them up’ (Garvin, 1984). Clothing features can be attributes like ‘collar presence’, ‘skin exposure’ and ‘sleeve length’ (Chen, Gallagher & Girod, 2012).

Aesthetics represent one of the two subjective dimensions of quality, because how good a product looks, feels, sounds, tastes or smells is clearly a matter of personal judgement (Garvin, 1984). According to Garvin (1988), aesthetics deal with the sensory characteristics and outward appearance of a product. Aesthetics can therefore be defined as the pleasure obtained from sensory perception (Hekkert, 2006).

(27)

27

on their research, and the notion that performance, aesthetics and features are all related to the design of the product, this categorization is adopted in the present study.

Leelakulthanit & Hongcharu (2012) and Filieri & Lin (2017), both reported that there is a positive relationship between the evaluation of the design of a product and the repurchase behavior of the customer. According to Cooper & Kleinschmidt (1987), the design of the product is the main determinant of sales. The design of a product communicates beliefs about the attributes and performance of a product and can evoke positive and negative affective responses (Bloch, 1995). A consumer can ‘fall in love’ with a product (high level of satisfaction) (Dumaine, 1991), or see the product design as an example of bad taste (low satisfaction) (Bloch, 1995). Behavioral reactions to product design can be classified as either approach behavior or avoidance behavior, where approach behavior displays an attraction to a product, avoidance behavior displays the exact opposite (Bitner, 1992; Donovan & Rossiter, 1982).

When making a distinction between utilitarian product characteristics and hedonic product characteristics (Hirschman & Holbrook, 1982; Strahilevitz & Myers, 1998), product quality dimensions that are grouped into the Design Quality category are identified to be more of a hedonic nature, because a customer “wants” the product to have an attractive design, although this is not a trait almost every product “should” have (Bazerman et al, 1998). According to Chitturi, Raghunathan & Mahajan (2008), hedonic product characteristics have a stronger relationship with repurchase behavior than their utilitarian counterparts. These findings, combined with the finding of Leelakulthanit & Hongcharu (2012) and Filieri & Lin (2017) that the evaluation of product design influences repurchase behavior, and the findings of Anderson & Srinivasan (2003), Atchariyachanvanich, Okada & Sonehara (2007), Mittal & Kamakura (2001) and Tsai & Huang (2007) that Design Quality influences customer satisfaction, which is a determinant of repurchase behavior, lead to the following H3 hypothesis:

(28)

28 2.5.2 Product Conformance

According to Garvin (1988), conformance deals with how the product or service satisfies customer expectations. Conformance is a measure of consistency, i.e. how well a product matches up against pre-established specifications (Garvin, 1984). In a study into drivers of customer satisfaction in e-retailing by Trabold, Heim & Field (2006), this concept was called ‘satisfaction with claims’, which deals with the reliability of the advertising and product claims made by the merchant. Through advertising, a firm makes explicit personal and non-personal promises that influence consumer expectations (Clow, Kurtz, Ozment & Soo Ong, 1997).

Perceived quality is the second subjective dimension of quality (Garvin, 1984). According to Garvin (1988), perceived quality is often referred to as reputation, since it is the perceived reputation of the product based on past performance and on other intangibles that may influence its perceived quality. In other words, it is the subjective assessment of quality resulting from image, advertising, or brand names (Zhang, 2001).

Perceived quality and product conformance both deal with the expectations a consumer has, either of a certain product, for example based on the image seen on the retailer’s website, or of a certain brand or retailer, based on the reputation of that entity. Therefore, they are grouped together in the category Product Conformance.

When making a distinction between utilitarian product characteristics and hedonic product characteristics (Hirschman & Holbrook, 1982; Strahilevitz & Myers, 1998), product quality dimensions that are grouped into the Product Conformance category are identified to be of both a hedonic nature, as well as a utilitarian nature. For example, it is conventional wisdom that certain brands are valued a lot by customers, because of their reputation, which refers more to customer “wants”. On the contrary, when dealing with clothing products, a customer expects a certain size to fit as indicated, which is referring more to “shoulds” (Bazerman et al, 1998).

(29)

29 H4: Customer satisfaction levels caused by the Product Conformance product quality dimension have a moderately strong relationship with the repurchase behavior of the customer who wrote the review, compared to the relationships of the customer satisfaction levels in the other categories with repurchase behavior

2.5.3 Product Life Elements

According to Garvin (1988), reliability measures the consistency of performance of the product over time. Product reliability is defined as “the ability of a product to perform as intended (i.e. without failure and within specified performance limits) for a specified time, in its life cycle application environment” (Mishra, Pecht & Goodman, 2002, p. 1).

Durability reflects the economic or physical life of a product. It is commonly measured by the number of hours, years, or miles that a product can be used before replacement is required (Garvin, 1984).

The concept of durability is closely related to the concept of serviceability, because consumers are not only worried about a product’s breakdown, but also about the elapsed time before service is restored after a product breakdown.

Because the concepts of reliability, durability & serviceability are closely related and all deal with product elements over time, they are grouped in the category Product Life Elements. Customers expect the quality in the long-term of a product to be sufficient, and therefore this category refers more to the “shoulds” (Bazerman et al, 1998). When making a distinction between utilitarian product characteristics and hedonic product characteristics (Hirschman & Holbrook, 1982; Strahilevitz & Myers, 1998), product quality dimensions that are grouped into the Product Life Elements category are identified to be more of a utilitarian nature. According to Chitturi, Raghunathan & Mahajan (2008), utilitarian product characteristics have a weaker relationship with repurchase behavior than their hedonic counterparts.

(30)

30

Based on the findings showing that product reliability influences customer satisfaction, which is a determinant of repurchase behavior (Anderson & Srinivasan, 2003; Atchariyachanvanich et al, 2007; Mittal & Kamakura, 2001; Tsai & Huang, 2007), in combination with the results of the study by Chitturi, Raghunathan & Mahajan (2008) displaying that utilitarian product characteristics have a weaker relationship with repurchase behavior than their utilitarian counterparts the H5 hypotheses is proposed in the following way:

H5: Customer satisfaction levels caused by the Product Life Elements product quality dimension have the weakest relationship with the repurchase behavior of the customer who wrote the review, compared to the relationships of the customer satisfaction levels in the other categories with repurchase behavior

The whole theoretical framework for the present study is shown in Figure 2. In summary, there is a body of previous research in which it was shown that customer satisfaction leads to repurchase behavior in a traditional shopping setting. Whether this relationship also exists in online shopping environments is debatable. Furthermore, the (grouped) product quality dimensions determine the level of customer satisfaction with a product. Garvin (1984) suggested that some aspects of quality have an element of subjectivity, and therefore depend on the eye of the beholder. Firms do not need to excel on all dimensions of product quality in order to become successful. In particular, focusing on a quality niche, especially if that market is overlooked by competitors, can lead to better firm performance. This implies that, because customers have different opinions about what determines product quality, their online product reviews will show different reasons for their level of customer satisfaction.

(31)

31

Figure 2. Conceptual model

3. Methodology

3.1 Data Collection

The data used in this study is supplied by a large woman’s undergarment manufacturer from the Netherlands, which specializes in selling luxurious lingerie. There is not much known about consumer motivations and rationales to purchase in this market (Sanchez Torres & Arroyo-Cañada, 2017), which makes conducting this study in this market even more relevant. As a consequence, this study is of interest to both academic researchers and stakeholders in the fashion industry.

(32)

32

review, the clothing manufacturer rewarded one reviewer with a voucher every month. A Dutch data science company already performed part of the pre-processing of the data, by linking the online product reviews to customer transactions.

3.2 Data Preprocessing

In this section of the methodology chapter, the preprocessing of the data done by the author of this thesis study is discussed in chronological sequence. Before the analysis can be conducted, a couple of proceedings have to be performed. The preprocessing of the data is done by making use of R-studio software. The proceedings are further explained in Table 1.

Table 1.

The data preprocessing steps

Step Proceeding

1 The data supplied by the Dutch data science company was checked for missing values. There were no missing values detected. The data was also checked for duplicates. For 9 online product reviews there was an online product review detected which had the same Review ID. For these online product reviews, the first occurring instance was preserved and the other deleted.

2 A dummy variable indicating whether a repurchase took place was added to the data frame. This dummy variable has the value 1 when a purchase was performed after a consumer wrote a review (indicating a repurchase), and this variable has the value 0 when no purchase was performed after a consumer wrote a review (indicating no repurchase).

3 The textual part of the online product reviews was translated to English using Google Translate. This proceeding had to be performed in order to be able to analyze the sentiment of the textual part of the online product reviews, because the sentiment dictionaries which are used only contain English words.

(33)

33 5 The textual part of the online product reviews was cleaned. The cleaning of the textual part of the

online product reviews involved removing non-English letters and punctuation, transforming capital letters to lowercase letters, removing stop-words and removing double white spaces. Removing punctuation is important, because for example “,” will assume a unique numeric identity, which can cause problems in the analysis (Tirunillai & Tellis, 2014). Transforming words to lowercase, removing non-English letters and removing double white spaces is important, because these can lead to misclassifications in the sentiment analysis and the text classification analysis. Stop words are removed, because words like “the”, “for” or “is” do not indicate a certain sentiment or product quality dimension.

After the preprocessing of the data, 549 online product reviews remained. The variables in the dataset used in the analysis are displayed in Table 2.

Table 2.

The variables used in the analysis Type Variable Description

Review Description of the customers’ experience with one or several products that the customer purchased from the clothing manufacturer, consisting of a rating and a textual part

Client ID Number representing a unique customer Review

timestamp

Time stamp indicating the time the review was received

Review ID Number representing a unique online product review User nickname Nickname of the review writer

Overall rating Numerical rating between 1 and 5, where 1 indicates a very low satisfaction with the product and 5 indicates very high satisfaction with the product

(34)

34 the customer with the product.

True-to-size Numerical rating in which 1 means much smaller than its indicated size and 5 means much bigger than its indicated size.

Repurchase dummy

Dummy variable indicating whether a customer performed a repurchase after writing an online product review

Product quality dimension category dummies

Dummy variables indicating to which product quality dimension category an online product review belongs (constructed based on the categorization analysis).

Transaction

An instance of a product purchase performed by a customer at the lingerie retailer

Client ID Number representing a unique customer Transaction

timestamp

Timestamp indicating the time the transaction was performed

Transaction ID Number representing a unique transaction

3.3 Analysis

In this part of the methodology chapter, the actual data analysis is explained. This part consists of three sections, namely an analysis of the level of customer satisfaction, a text categorization analysis, and the eventual comparison of the (reasons underlying) customer satisfaction levels to the repurchase behavior of the consumers, which will provide the answers to the research questions.

(35)

35

The text categorization analysis consists of an evaluation of the words in the online product reviews. The words in the online product reviews are scored according to the product quality category they belong to, which creates a dictionary in which each word belongs more or less to a certain product quality category. By scoring the words in the online product reviews based on the classification dictionary, the online product reviews can be classified into three categories based on the product quality dimension underlying the sentiment in the online product review.

The third part of the analysis consists of two parts: The actual comparison of the constructs representing customer satisfaction levels (the valence of the text & the online product ratings) to the repurchase behavior, through which the first research question is answered, and the investigation of the influence of the three categories of product quality on the determined satisfaction-repurchase relationship, through which the second research question is answered. This analysis is done by conducting a binary logistic regression.

3.3.1 Sentiment Analysis

A considerable number of research papers state that in a sentiment analysis, online product reviews are classified according to their polarity (either positive or negative) (Pang & Lee, 2008). This view on sentiment analyses is adopted in this study.

The used method consists of two central features: the sentiment dictionary and the processing component. The dictionary consists of a collection of words, which all have a positive, negative or neutral semantic orientation. In the processing component, a file is examined word by word, and all words are compared to a sentiment dictionary.

For the quantitative analysis, choosing the appropriate dictionary is crucial in order to obtain a good measurement of customer satisfaction expressed in the sentiments of the text. Cho, Kim, Lee & Lee (2014) compared the functioning of different sentiment dictionaries in their study. Of the readily available sentiment dictionaries, the often used AFINN sentiment dictionary (Nielsen, 2011) performs well and is therefore used in this thesis study.

(36)

36

The BING sentiment dictionary was used for the descriptive statistics, because it provides a clear view of the ratio of positive and negative words in the online product reviews. The AFINN sentiment dictionary was used for the empirical part of the analysis, because the quantifiable sentiment score enables identifying a statistical relationship with the repurchase behavior of the customers. By adding up the sentiment scores of the words in the online product reviews, the total sentiment score per review is obtained. An example of these scores can be found in Table 3.

Table 3.

Example of online product rating and sentiment scores.

Review ID Rating Sentiment score

1265791 5 10

4567520 5 11

1907352 4 8

1825394 5 8

Note: Review ID has been anonymized

3.3.2 Text Categorization Analysis

Through performing the text categorization analysis, the reviews will be classified into three different categories based on the product quality dimensions underlying the sentiment in the online product review (“Design Quality”, “Product Life Elements” & “Product Conformance”).

The classification into the different categories is done based on a self-built dictionary in R-studio. In this dictionary, every word is scored, receiving a score between 0 and 5 points for each of the three categories, based on the association of that word with that particular category. Table 4 gives a description of which type of word belongs to which category. Table 5 shows the scoring of a few words as an example.

The Relationship between Customer Satisfaction and Repurchase Behavior in Online Shopping Environments