1
Service Recovery and its Effect on Customer Satisfaction:
The Role of Transparency in the Complaint Handling Process
MSc Business Administration University of Amsterdam Specialization: Marketing
Name: Olga Pak
Student Number: 10599509 Supervisor: Adriana Krawczyk
2
Statement of Originality
This document is written by Student Olga Pak who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.
The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
3
Abstract
In the last decade, airline companies have been experiencing an exponentially increasing amount of service failures due to the high demand for international flights. Although service failures are impossible to always preemptively predict, the airline industry is infamous for their lack of customer care and transparency in the complaint handling process. This research therefore asks two fundamental questions. First of all, does the quality of a service recovery significantly influence the level of customer satisfaction following a service failure? Secondly, does the level of informational transparency following a service failure positively moderate the hypothetical effect of service recovery quality on customer satisfaction?
As previous academic literature had also pointed towards, service recovery quality was found to have a positive effect on Net Promoter Score, which implies that a higher quality of recovery leads to higher levels of customer satisfaction. Although transparency was also found to have a positive direct effect on customer satisfaction, transparency was not found to be a moderator on the relationship between recovery quality and satisfaction of customers. Therefore, varying degrees of informational transparency did not have a sufficient influence on the service recovery process to be statistically significant. Although the second hypothesis was not proved, this research still provides some fundamental findings both for future research and for airline companies. Academics should investigate transparency as a moderator in different industries and transparency as a moderator on another direct effect, whereas airline companies should become customer-focused and more transparent to have more favorable ratings and retain customers.
4
Table of Contents
1. INTRODUCTION ... 5
2. LITERATURE REVIEW ... 8
2.1CUSTOMER SATISFACTION ... 9
2.2SERVICE FAILURES &RECOVERIES ... 11
2.3TRANSPARENCY ... 13
2.4MOTIVATION FOR THE FOCUS ON THE AIRLINE INDUSTRY ... 16
3. THEORETICAL FRAMEWORK ... 17
3.1SERVICE RECOVERY QUALITY ... 18
3.2TRANSPARENCY ... 19 4. CONCEPTUAL MODEL ... 20 5. METHODOLOGY ... 21 5.1RESEARCH DESIGN ... 21 5.2SCENARIOS ... 22 5.3SAMPLE ... 23 5.4MEASURES ... 23 5.5LIMITATIONS ... 24 6. RESULTS ... 25 6.1SAMPLE CHARACTERISTICS ... 25
6.2DATA CLEANING,MISSING VALUES AND RECODING ... 25
6.3RELIABILITY ... 26
6.4FACTOR ANALYSIS... 29
6.5MANIPULATION CHECK ... 30
6.6DESCRIPTIVE STATISTICS &CORRELATION ANALYSIS ... 31
6.7HYPOTHESIS TESTING ... 34
6.8HYPOTHESIS TESTING -PROCESS ... 38
7. DISCUSSION ... 39
7.1FINDINGS ... 39
7.2THEORETICAL &PRACTICAL IMPLICATIONS ... 40
8. CONCLUSION ... 42
9. REFERENCES ... 44
10. APPENDIX ... 53
10.1SURVEY DESIGN ... 53
10.2SPSSOUTPUT:FACTOR ANALYSIS ... 58
10.3SPSSOUTPUT:RELIABILITY ... 59
10.4SPSSOUTPUT:FACTORIAL ANOVA ... 62
5
1. Introduction
One of the primary goals of business organizations is to provide effective customer service to ensure that they retain their existing customers and attract new customers to the business. Retaining the existing customers shows customer loyalty, and achieving this loyalty with the customers is critical to the success of organizations because it helps them to maintain their share of the market in the industry. However, the imperfect nature of organizations means that service failures occur at times, and this has a negative effect on customer loyalty. Service failure refers to the occurrence of errors in the organization, which hinder the effective delivery of services to the customer. Incidences of service failure are usually followed by customer complaints. Failure to address customer complaints effectively can lead to customer dissatisfaction, which has a negative impact on customer loyalty (Davidow, 2003). Problems with customer loyalty will therefore logically lead to issues with customer retention, which businesses can not afford to have if they want to remain competitive in the market. This is because dissatisfied customers are likely to seek services elsewhere, which reduces the market share of the organization. Therefore, organizations need to have effective complaint handling procedures when a service failure does arise as to address the concerns raised by their customers.
Furthermore, the effective handling of customer complaints promotes faster service recovery. Service recovery refers to the restoration of services to a level that is satisfactory to the customers (Grönroos, 1988). Quick service recovery is important because it ensures that other customers do not experience similar problems, which can be damaging to the reputation of the organization. In other words, the effective handling
6 of customer complaints helps organizations to address problems quickly to mitigate the negative consequences of such problems.
Service recovery literature points towards the fact that the airline industry seems to suffer from regularly occurring service failures (Keiningham, Morgeson, Aksoy & Williams, 2014; Nikbin & Hyun, 2015; Lorenzoni & Lewis, 2004; Bamford & Xystouri, 2005; Lee, 2017). Perhaps the most recent example of such a service failure is seen in the scandal that happened to United Airlines in April 2017. The accident, which went viral on social media and the Internet in general, showed a man, bleeding, who is violently being dragged off of a United Airlines flight. The reason for this accident was that that particular flight was overbooked, and therefore, after attempting to convince passengers to give up their seats by offering them money, the staff decided to choose four passengers to leave the plane according to ticket class, frequent flier status and check-in time (Lartey, 2017). For an airline company, a service failure can lead to public backlash, a drop in the stock price of the company and a general decrease in the attractiveness of the company. Although there was a general opinion of disgust surrounding the accident that happened with United Airlines, the reality is that these types of service failures occur much too frequently in the airline industry (Lee, 2017). This year alone there have been similar accidents happening with other airlines (BBC, 2017; Zhang, 2017), and these accidents only cover the overbooked examples of failures in the airline industry. Given the fact that there is such a prominence of service failures in the airline industry, it is essential for companies to properly recover from these accidents. More importantly, does a correct and adequate response to a service failure significantly affect customer satisfaction ratings?
7 One important attribute that organizations ought to embrace when handling customer complaints is transparency due to the importance of justice and fairness in the customer’s eyes that satisfies their expectations (Chang & Chang, 2010). Whilst recovering from a service failure, being honest with the customer should logically help the success of that recovery. Since the customer will already be relatively dissatisfied with the company as a whole, being honest and forthright with the customer might make the difference between a satisfied and dissatisfied customer. Being transparent, which refers to the complete disclosure of information in a timely manner (Peek & Rosengren 2000), is the most honest form of business and is known to ameliorate consumer welfare, which significantly affects the attractiveness of the firm (Carter & Curry, 2010). It is found to be a positive force for a company’s image and often leads to more favorable ratings of the firm as well as more loyal customers (Singh, 2015).
This thesis would be the first to analyse how transparency affects customers in the service recovery process in the airline industry. The studies of service recoveries in the airline industry are prevalent throughout academic literature. Chang & Chang (2010) found that interactional and procedural justice in the service recovery process has an effect on recovery satisfaction, and that overall satisfaction mediates the relationship between satisfaction and loyalty. Nikbin, Marimuthu, Hyun & Ismail (2014) studied the effects of recovery satisfaction in terms of justice on customer loyalty and found that failure attributions of stability and controllability moderated this relationship. Although many academics have studied service failures and the effect of recoveries on customer satisfaction in the airline industry (Bamford & Xystouri, 2005; Bejou & Palmer, 1998; Edvardsson, 1992; Keiningham et. al, 2014; Lorenzoni & Lewis, 2004; Chang & Chang, 2010), there is a real gap in the academic literature relating to the effect of transparency
8 in the service recovery process in the airline industry. Therefore, the main research question arises: how does transparency in the complaint handling process affect customer satisfaction in the airline industry?
The purpose of this study is to examine the role of transparency in the complaint handling process. Specifically, this study seeks to find out how transparency helps organizations to maintain customer loyalty after the occurrence of a service failure. This is an interesting topic of study because we are living in an age where information about customers and organizations is readily available from a variety of sources (Craven, 2015). Consequently, modern organizations are expected to be transparent because all information about them eventually becomes a matter of public record. The subsequent chapters of this thesis are organized as follows. First, an in depth analysis of the academic literature on service failures and recoveries, customer satisfaction and transparency will be undertaken. Following the literature review, the theoretical framework will be presented complemented by a proposal of the hypothesized relationships between the variables. Third, the sampling method and general method of the research paper will be laid out. Following the method section, the results will be shown and an analysis of the results and the conclusions regarding the results will be made. Lastly, the results and conclusion of the results will be analyzed after which the limitations and future research directions will be mentioned.
2. Literature Review
In the literature review, the themes and idea of the proposed research question and topic of this research paper will be analysed. The review will start by analysing the literature regarding customer satisfaction, explaining what customer satisfaction is and
9 why it is important for the survival of an organization. Following customer satisfaction, the literature review will analyse service failures and recoveries, specifically what service failures are and how companies recover from these failures. Lastly, this literature review will focus on transparency as a concept and the effects of transparency in the workplace. Following the literature review, a theoretical framework will be presented in which the concepts will be merged and hypotheses will be created according to past literature and research gaps.
2.1 Customer Satisfaction
Customer satisfaction is essentially the measure of how services and products sold by a company meet customer expectations or how they exceed or fall short of customer expectations. In Marketing Metrics: The Definitive Guide to Measuring Marketing Performance, Farris, Bendle, Pfeifer & Reibstein (2010) define customer satisfaction as the number of customers, or percentage of total customers, whose reported experience with a firm, its products, or its services exceeds, specified satisfaction goals. The academic literature has identified various different ways of measuring customer satisfaction numerically. To measure customer satisfaction, many academic articles tend to ask respondents to qualitatively explain how satisfied they were in different situations through a survey (Cronin, Brady & Hult, 2000; Anderson, Fornell & Lehmann, 1994; Gloor, Fronzetti, Colladon, Giacomelli, Saran & Grippa, 2017). These academic articles tend to not have a fixed, determined and renowned metric to measure customer satisfaction but rather directly ask respondents about their level of satisfaction in an organization’s decision through basic survey questions (Cronin, Brady & Hult, 2000; Anderson, Fornell & Lehmann, 1994; Gloor, Fronzetti, Colladon, Giacomelli, Saran & Grippa, 2017). To measure customer satisfaction, Dixon,
10 Freeman & Toman (2010) took another approach and developed the Customer Effort Score. This metric is defined as the customers’ intention to keep doing business, increase the amount they spend on the organizations goods and services or spread positive (and not negative) word of mouth regarding the organization (Dixon, Freeman & Toman, 2010). On the other hand, Reichheld (2003) believes that customer satisfaction is most effectively measured through the Net Promoter Score – one of the most used metrics for customer satisfaction and overall perception of an organization - and he stresses the importance for organizations to focus on this figure. In this view, if companies want to grow their business, he believes that they won’t learn from complex measurements of customer satisfaction and retention (Reichheld, 2003). He believes that an organization should simply need to know what their customers tell their friends and acquaintances about the company and its operations, which is the true indication of a customer that is satisfied with a company (Reicheld, 2003). Customer satisfaction can also be measured through the Net Promoter Score (NPS) which measures the willingness of customers to recommend a product or service to others (Gloor et al., 2017).
Customer satisfaction is of the utmost importance for service-providing organizations since satisfaction affects future consumer choices, which in turn lead to improved consumer loyalty and consumer retention (Abratt, Curtis, Dion & Rhoades, 2011). When customers are satisfied with the service that a certain organization provides, they stay loyal because they want to continue their mutually beneficial relationship with the company – exchanging a monetary amount for a service that is worth the customer’s money (Abratt, Curtis, Dion & Rhoades, 2011). When an organization manages to satisfy its customers’ needs and wants, they have ultimately
11 achieved the key to gaining loyalty from their customers (Oliver, 1997). Tsiotsou (2006) and Chiou and Pan (2009) have found that customer satisfaction indeed has a direct and positive effect on purchase intentions of customers as well as the repetitive nature of this behavior. This entails that when customers are satisfied they will be more likely to buy purchase a good or service from a specific organization and they will be more likely to repeat this purchase behavior.
2.2 Service Failures & Recoveries
For companies that provide services to their customers, a service failure is a problem related to the service offered to the customer, where an issue occurred with the organization providing the service (Maxham, 2001). Simply put, a service failure is when a customer was under the impression that a service would be provided to them and within a certain timeframe and the company has not fulfilled their promise due to a variety of reasons (Chan & Wan, 2008). Failures in the delivery of a service are almost impossible to avoid and will inevitably happen at some point during the operations of a service-providing business (Sengupta, Balaji & Krishnan, 2015). When such a failure does take place, customers who are involved in the service failure will be displeased and will experience disconfirmation (Smith, Bolton & Wagner, 1999).
Following a service failure, an organization has to respond or recover from it. In academic literature, a service recovery refers to service providers’ actions to adequately and properly adjust their behaviours to handle customers’ complaints and to recover their satisfaction and loyalty (Miller, Craighead & Karwan, 2000; Jong & De Ruyter, 2004). Academics have identified key elements of service recoveries that ensure its effectiveness and positive effect on the customer who experienced the service failure (Jung & Seock, 2017; Halstead, Dröge & Cooper, 1993; Kelley & Davis, 1994; Weun,
12 Beatty & Jones, 2004). Kelley & Davis (2004), through their literature analysis, found that certain academics discriminated between two key standards of service delivery expectations, namely concepts referred to as ‘will’ expectations and ‘should’ expectations. They therefore identify important service attributes by distinguishing between actions and attributes of the service recovery quality that customers think have to happen (‘should’ expectations) and those that customers would like to ideally see but do not deem absolutely necessary (‘will’ expectations) (Kelley & Davis, 1994). Other academics have distinguished in the same way but have used different terminology, for example Parasuraman, Berry & Zeithaml (1991) who used adequate expectations and desired expectations. Although many distinctions of service recovery characteristics exist in academic literature, perhaps the most useful explanation of key attributes of service recoveries for research comes from Jung & Seock (2017). This construct presents two dimensions of service recovery which both have an effect on the quality of the recovery as well as its consequential effect on the customer’s attitude towards the organization. The first is named tangible recovery, which refers to tangible compensation provided to customers to reduce real damages in the form of gifts, presents, free services, discounts or coupons (Jung & Seock, 2017; Miller, Craighead & Karwan, 2000). The second is named psychological recovery and it refers to an apology, sense of empathy or explanation provided by the company to rectify the problem caused by the service failure and improve the satisfaction and retention of their customers (Jung & Seock, 2017; Kuo & Wu, 2012).
The importance of the service recovery is exemplified by the research done by the Office of Fair Trading (1990). The research showed that when people make a complaint about the service that a certain company provides and that complaint is
13 satisfactorily resolved, three quarters of them will decide to purchase from the same organization (OFT, 1990). On the other hand, when customers make a complaint about a service and that complaint is not resolved, lass than half of those customers will buy from the same organization again (OFT, 1990). As shown through this research, customer satisfaction through a high quality of service recovery is necessary to obtain customer loyalty when a service failure does occur. Although there is a consensus that service recovery quality is somewhat positively associated with customer satisfaction, with respect to the construct mentioned by Jung & Seock (2017) academics found differing results for effects of apology and compensation on customer satisfaction. Apology is traditionally viewed as a valuable reward that redistributes esteem (Hatfield, Walster & Berscheid, 1978) and proved to boost customer satisfaction (Goodwin & Ross, 1992). Nonetheless, the academics found that, of the service recovery strategies, apology has the smallest effect on raising customer satisfaction (Fang, Luo & Jiang, 2012). Compensation was found to have a higher effect on overall reactions by customers and their level of satisfaction with the organization since compensation strongly impacts distributive justice (Fang, Luo & Jiang, 2012). Although apology and compensation both affected customer satisfaction, the strategy that had the most favourable impact on customers was a third introduced by the researchers, namely quality improvement – defined as firm’s improvements in providing services to customers so that similar future service failures won’t happen again in the future (Fang, Luo & Jiang, 2012).
2.3 Transparency
In the literature, academics have often had varying definitions of transparency. These vary from Jordan, Peek & Rosengren (2000) defining it as “the disclosure of
14 timely and accurate information” to Bushman, Piotroski & Smith (2004) defining it as “the availability of firm-specific information to those outside publicly traded firms”. Through an analysis of the academic literature on transparency, Schnackenberg & Tomlinson (2016) concluded that transparency, as a behavioural concept, refers to the perceived quality of intentionally shared information from a sender. These authors also provided a widely accepted and used construct for transparency. In fact, they suggest that there are three primary aspects of transparency: information disclosure, clarity and accuracy. Information disclosure is defined as the perception that relevant information is received in a timely manner (Schnackenberg & Tomlinson, 2016; Bloomfield & O’Hara, 1999). This implies that information must be shared openly for it to be considered transparent. The second aspect, clarity, is the perceived level of lucidity and comprehensibility of information received from a sender (Schnackenberg & Tomlinson, 2016). This idea entails that information must be presented in a clearer way by organizations for the information to be considered transparent (Winkler, 2000; Schnackenberg & Tomlinson, 2016). Finally, the third and last dimension is accuracy. Accuracy is defined as the perception that information is correct to the extent possible given the relationship between the sender and the receiver (Schnackenberg & Tomlinson, 2016). Therefore, for information to be transparent it cannot be considered purposefully biased or unfoundedly contrived (Schnackenberg & Tomlinson, 2016). Companies can also be categorized into three distinct groups relating to their level of transparency. Companies can be transparent, translucent – which is the middle option – or opaque – which are companies who are not transparent at all (Lamming et al., 2001). Finally, Rawlins (2008) presents yet another categorization of transparency. He defines transparency as the deliberate attempt to make available all legally releasable
15 information along four dimensions. Therefore, the release of such information has to be in a manner that is timely, accurate, balanced and unequivocal (Rawlins, 2008).
When transparency is taken as a concept on its own, without applying any contextual information to the concept, transparency has been found to lead to more trust between two parties (Eggert & Helm, 2003; Grimmelikhuijsen & Porumbescu, 2013; Walker, 2016). More specifically, the academic literature has found that transparency and trust are two interconnected concepts in that a company that behaves more transparently in terms of disclosure will gain more trust from its employees and its customers (Walker, 2016). Consumers increasingly lose trust in both companies on a general level and governments alike, and transparency is proposed as the solution to this intangible problem (Grimmelikhuijsen & Porumbescu, 2013). An article by Eggert and Helm (2003) found without a doubt that, according to their model and structural equation estimates, transparency ultimately delivers added value to the customer. Furthermore, it increases customer satisfaction and leads to favourable behavioural intentions, which are all consequently related to customer loyalty and retention (Eggert & Helm, 2003; Eskildsen & Kristensen, 2007). The literature therefore points towards transparency leading to favourable customer perspectives, which should therefore positively influence customers during a service recovery following a service failure.
All in all, transparency is essential to businesses since it ameliorates consumer welfare (Carter & Curry, 2010) and deals with uncertainty avoidance (Vishwanath, 2003), which significantly affect the attractiveness of a company. The articles discussed above are in line with the majority of academic literature that states that transparency is a positive force for a company’s image and will lead to more loyal customers (Singh, 2015). Adopting a transparent approach not only ameliorates a firm’s image, but it also
16 has a positive effect on a firm’s profitability and its financial revenue (Singh, 2015). Nonetheless, to this day, no academic article so far has analysed transparency as a moderator on the effect of service recovery quality on customer satisfaction.
2.4 Motivation for the Focus on the Airline Industry
There are various reasons that led to the decision of choosing the airline industry as the focus of this research paper. First and foremost, service failures are quite common in airline industries (Bejou & Palmer, 1998; Edvardsson, 1992). The academic literature points to this theme due to the numerous articles that examine service failures and recoveries in the airline industry (Keiningham, Morgeson, Aksoy & Williams, 2014; Nikbin & Hyun, 2015; Lorenzoni & Lewis, 2004; Bamford & Xystouri, 2005). One study even distinguished between the severity of a service failure in the airline industry – namely minor incidents (i.e. failures that do not result in physical harm) and major incidents (i.e. failures that result in injury or death), which indicates the repetitive occurrence of failures in this industry (Keiningham, Morgeson, Aksoy & Williams, 2014).
Given the frequency of service failures in the airline industry, it is of the utmost importance to analyse service recoveries in this industry as well as the customer’s perception and behavioural change with respect to the quality of the recovery. In addition, airline companies have been known to have extremely low customer satisfaction ratings compared to companies in other industries due to frequent issues and failures in their services (Satmetrix, 2014). This research will therefore provide airline companies with the tools to ameliorate their already low NPS (Gloor, Colladon, Giacomelli, Saran & Grippa, 2017) and retain their customers’ loyalty in case of a service failure by showing them that the quality of a service recovery, as well as the
17 transparency in the process, have an effect on customer satisfaction. Given the frequency of service failures in the airline industry, exemplified by the latest United Airlines scandal which is just one of the many that happened in the last few years (Lee, 2017), this research will therefore hopefully show airline companies that taking a transparent approach in the service recovery process is the most reliable strategy that will lead to satisfied customers. Airline companies should learn from theory and realize that the most efficient service recovery strategies come from a transparent process since customers can trust the process and have a favourable view of the firm (Eggert & Helm, 2003).
3. Theoretical Framework
Following the literature review, this research paper will now create a theoretical framework and conceptual model relating to the theories that were analysed and the proposed variables introduced in the previous sections of the paper. In the theoretical framework, the theories regarding service recoveries and transparency, as well as customer satisfaction, will be moulded into hypotheses through logic and previous theoretical results. The first part will analyse service recoveries, and a hypothesis will be created with respect to the analysis of service recoveries and customer satisfaction. The second part will analyse transparency and its theoretical impact on customer satisfaction. Transparency will then be linked to the relationship between service recovery quality and customer satisfaction and the second hypothesis will arise. Following the theoretical framework, the conceptual model with all the variables and the hypothesized relationships is presented.
18
3.1 Service Recovery Quality
When a service failure occurs, an organization responds to the failure through a service recovery. As Jung & Seock (2017) pointed out, service recovery can be thought to have two attributes, or two dimensions. The first dimension is a tangible recovery, also named compensation, in which customers are provided with a compensation for the failure (Miller, Craighead & Karwan, 2000). The second is a psychological recovery, also named apology, in which the company provides an explanation to rectify the problem (Kuo & Wu, 2012).
Academic literature has found that these attributes all lead to higher customer satisfaction ratings in users who experience service failures (Fang, Luo & Jiang, 2012). Authors found that apology had the smallest (but still significant) impact on customer ratings followed by compensation and quality improvements – a dimension added by Fang, Luo & Jiang (2012). All in all, studies point towards the fact that when customers complain about a service and that service is satisfactorily resolved, customers will more favorably view the firm and will be more likely to purchase again from that same firm (OFT, 1990).
The literature that has been analysed in the literature review seems to be consensual on the idea that a higher quality of service recovery will lead to a better image in the eyes of customers who experienced the service failure.
Therefore, a hypothesis can be formulated with respect to this idea:
Hypothesis 1: Following a service failure, a higher quality of service recovery leads to higher levels of customer satisfaction.
19
3.2 Transparency
Transparency is said to have four primary aspects: information accuracy, timeliness, balance and unequivocality (Rawlins, 2008). The first aspect refers to how clear the information that is being disclosed is, in terms of comprehensibility and lucidity (Rawlins, 2008). The second aspect of transparency refers to the timely disclosure of information to the party that is concerned with the information (Rawlins, 2008). The third aspect refers to how balanced and fair the information is (Rawlins, 2008). The fourth, and final, aspect refers to the unambiguous nature of the information that is disclosed (Rawlins, 2008). Companies can also be categorized into three distinct groups relating to their level of transparency – namely transparent, translucent or opaque.
Literature on transparency is far from being scarce, but no author has focussed on transparency in the customer complaint handling process following a service failure. Nonetheless, in academic literature, transparency has always been linked to trust – the relationship being that more transparent companies have customers that trust the company more than non-transparent companies (Grimmelikhuijsen & Porumbescu, 2013). In addition to simply creating more trustworthy relationships between two parties, transparency also directly adds value to the service for customers of the firm (Eggert & Helm, 2003). Furthermore, transparency leads to favourable behavioural intentions and increases customer satisfaction and loyalty (Singh, 2015).
There is a visible consensus that transparency leads to a more favourable image and a better relationship with employees and customers (Grimmelikhuijsen & Porumbescu, 2013; Eggert & Helm, 2003; Eskildsen & Kristensen, 2007). If the quality of a service recovery leads to more favourable customer satisfaction ratings, a company
20 that is transparent with their customers during the service recovery should receive even more favourable customer satisfaction ratings. To this day, no academic article so far has analysed transparency as a moderator on the effect of service recovery quality on customer satisfaction. A moderator refers to the variable that determines the relationship between two variables or, in other words, a moderating effect happens when the interaction between two variables depends on another variable (Cohen et al., 2003). Given the positive effects of transparency found in the literature, it is hypothesized that a more transparent firm during the service recovery process in the airline industry would receive better ratings than an opaque company. Due to the relevance and importance of transparency in the modern business realm, it is essential to study this characteristic as a positive force in the complaint handling process. Thus, the second hypothesis is found:
Hypothesis 2: Transparency has a positive moderating effect on the relationship between service recovery quality and customer satisfaction following a service failure in the airline industry.
4. Conceptual Model
The figure below (Figure 1) shows the conceptual model, which contains the hypothesized causal relationships between all the selected variables in this study. It presents a visualization of the hypothesized direct effect of the independent variable – service recovery quality – on the dependent variable – customer satisfaction. The figure shows whether there is a causal relationship between the quality of a service recovery and how satisfied customers will be with the recovery and the organization as a whole. Furthermore, Figure 1 shows the moderating variable – transparency – and its effect on the relationship between the quality of a service recovery and customer satisfaction. As
21 shown in the conceptual model below, service recovery quality is hypothesized to have a positive effect on customer satisfaction. Likewise, transparency of information is hypothesized to have a positive effect on the relationship between service recovery quality and customer satisfaction.
5. Methodology
This section explains the methodology of the research paper. The subsection 5.1 describes the design of the research, subsection 5.2 describes the scenarios, 5.3 explains the sample and the subsections 5.4 and 5.5 respectively discuss the measurements and the limitations.
5.1 Research Design
This research will use a cross-sectional survey design, which collects data to make inferences about a population of interest at one point in time (Lavrakas, 2008). It is a fitting way of analyzing the model since the findings are generalizable to all customer complaint handling. Furthermore, the structure of this design can be categorized as an
Figure 1: Conceptual model showing the interaction between the dependent, independent and moderator variables
Service Recovery Quality Customer Satisfaction Transparency H1 H2 + +
22 experimental vignette design. Vignette studies use descriptions of situations shown to respondents within a survey to elicit their judgments of those scenarios (Atzmüller & Steiner, 2010). Vignettes are ideal for this research since this type of study recognizes the socially situated nature of individual behavior, which is necessary to mimic service failures as a real-life situation (Hughes, 1998).
The purpose of this research is of an explanatory nature since it attempts to explain relationships between various proposed variables. Nonetheless, since the purposes are not mutually exclusive, this thesis can also be said to have a descriptive nature, meaning that it seeks to provide a description of observations of a phenomenon.
To avoid the brand-engagement bias, a fictional airline company, Air Travel, is created and described in the scenarios.
5.2 Scenarios
This research paper employed a scenario-based experimental survey, where service failures in the airline industry were presented to participants to analyse the effect of the quality of service recovery on customer satisfaction. During the collection of the data, four questionnaires with four different scenarios, shown in the Appendix 10.1, were sent to participants, thereby creating a 2 x 2 (good/bad service recovery & good/bad transparency) matrix to obtain the largest variance possible for both the moderator and the independent variable. Participants were randomly assigned to one of the four questionnaire scenarios, in which they were asked to answer the questions related to the variables service recovery quality, transparency and customer satisfaction.
23
5.3 Sample
With respect to the sampling technique, this research will use a convenience sample. This sampling technique is used since convenience samples require minimal selection time cost due to simplicity of finding respondents (Ferber, 1977). Since this research is written during a master program, student samples were chosen due to the ease of access to university students. Nonetheless, to increase the sample size, a snowball and networking technique – which involves respondents sharing the survey with people they know – were used (Lavrakas, 2008). To maintain the same frame, the referrals from other respondents were also mainly students.
The survey, which can be found in the Appendix 10.1, was sent online via social media – more specifically through Facebook and direct contacts – through personal email. Due to the distribution channels chosen for this survey, it is difficult to estimate the response rate of the survey but, given statistics found online (Fryrear, 2015), the estimated response rate of the survey should be around 30%, which is the aim of the research.
5.4 Measures
The different questionnaires presented in this research will ask participants about their demographics like gender, age and income level. The other variables analyzed in this thesis will use validated Likert scales on a 10- or 7-point scale at interval levels.
The dependent variable - customer satisfaction – is measured through the Net Promoter Score (NPS) (Gloor et al., 2017). NPS is based on the simple question: “Would you recommend our company to a friend?” on a scale of 0-10.
The independent variable - service recovery quality - is measured according to the construct by Jung & Seock (2017). This construct presents two dimensions of service
24 recovery - apology and tangible compensation, which are measured on a 7-point Likert scale. The respondents were randomly assigned to ‘low’ (0) and ‘high’ (1) levels of service recovery quality scenarios.
The moderator variable - transparency – is measured according to the construct of Rawlins (2008) in which four items are considered: accuracy, timeliness, balance and unequivocality of the information disclosed on a 7-point Likert scale. The moderation effect investigates the interaction between the relationship of independent and dependent variable (Cohen et al., 2003). All respondents were also randomly assigned to ‘low’ (0) and ‘high’ (1) levels of transparency scenarios.
5.5 Limitations
Although the cross-sectional design has its benefits, it is only studies at one point in time and therefore a causal relationship between two constructs cannot be inferred. Furthermore, a cross-sectional study presents the limitation of causality. With respect to experimental vignette studies, the most predominant limitation is that it presents the respondent with a fictional situation. Although the vignette study seeks to mimic real life, it is still taken in the context of a fictional event and is therefore not entirely generalizable to an actual situation. The use of a non-probability sampling technique also has its negative aspects since it can hinder the generalizability of the results.
In this research, participants might have experienced the social desirability bias, which involves respondents answering questions in a way that they think will lead to being accepted and liked, due to the moral nature of the issue of transparency. Finally, participants could experience the acquiescence bias, in which they simply agree with anything presented to them.
25
6. Results
The following chapter will illustrate the results that were drawn from the analysis of the proposed model. The first section of this chapter will explain the characteristics of the chosen sample, followed by an analysis of data cleaning, missing values and possible recoding in the second section. The third and fourth section respectively discuss the reliability test and the factor analysis, whilst the fifth subchapter uses the manipulation check – which all three are recommended before analysing the hypotheses (Field, 2009). The sixth section will analyse the descriptive statistics and correlations between the variables. The seventh chapter is dedicated to testing the hypotheses proposed in this research’s model and in the final section the statistical model Process will be discussed.
6.1 Sample Characteristics
The sample used in this research, which consists of all the participants that took part in this study, was made up of 201 participants: 49.7 % were male and 50.3 % were female. Out of the 200 participants that took part in the survey, 194 – which equals 97% of participants – decided to indicate their age. The mean age of the participants was 26 years old, which makes sense since the majority of the participants were students. The minimum age of the participants are 16 years old whereas the maximum age of participants is 57 years old. Finally, out of the 200 participants, over 30% were Dutch, which made up the majority.
6.2 Data Cleaning, Missing Values and Recoding
Frequencies check were made to examine any errors that might have occurred in the data. Out of 201 survey cases, one survey case is removed from the data set because of
26 missing values. In addition to this, the screening question was presented in the beginning of the survey to check participants’ focus. The recoding of one counter-indicative item was applied to Service Recovery Quality item (rSerRec3).
6.3 Reliability
In this section of the results, the reliability of the variables will be examined to ensure that the outcome of the analysis is accurate. Reliability is described by Field (2009) as the degree to which analysis procedures or data collection techniques lead to consistent findings or results. To examine the reliability of the scales, Service Recovery Quality and Transparency scale were analyzed to check for their Cronbach’s alpha. The Cronbach’s Alpha is the most commonly used objective measure of reliability and it provides the internal consistency of a scale, expressed as a number between 0 and 1 (Takavol & Dennick, 2011). The scales are said to be reliable if the Cronbach’s Alpha value is over 0.7 (Takavol & Dennick, 2011). Since both scales showed that the Cronbach’s Alpha score is significantly above 0.70, the two scales can be said to be statistically significant and therefore reliable. Table 1 illustrates the two scales and their respective Cronbach’s Alpha values.
Table 1: Cronbach’s Alpha values for the two proposed scales
Variable Cronbach’s Alpha
Service Recovery Quality Transparency
0.931 0.908
Nonetheless, the reliability analysis for the Transparency scale, which consisted of 4 items, showed that 1 item had a low Cronbach’s Alpha >.60. Therefore, the
27 reliability statistics table.
Table 2: Reliability Statistics
Cronbach's Alpha Cronbach's Alpha Based on Standardized Items
N of Items
.613 .628 4
Table 3 below shows the statistics if certain items were deleted from the model. Table 3: Reliability Statistics by transparency items
Scale Mean if Item Deleted Scale Variance if Item Deleted Corrected Item-Total Correlation Squared Multiple Correlation Cronbach's Alpha if Item Deleted Trans 1 12.56 31.628 -.195 .042 .908 Trans 2 12.46 15.753 .727 .759 .269 Trans 3 12.42 16.549 .721 .764 .292 Trans 4 13.15 16.055 .615 .576 .352
The above illustrated table 3 is worth mentioning since it shows a critical piece of information regarding this research. First and foremost, the 4 versions of Trans listed above refer to the 4 questions used in this research to test transparency on a 7-point Likert scale. The first question, Trans1, asked participants to rate how they felt about the following statement: “I consider that the organization is accountable for what has happened.” The second question, Trans2, stated: “The organization informed me well what is doing and why the service failure has happened.” The third question, Trans3, asked participants to rate the following statement: “The information provided by the company was relevant and reliable.” The fourth and final statement was the following: “The organization disclosed information in a timely manner.” As explained in the section above, a measure or scale can be said to be reliable when its Cronbach’s Alpha value is above 0.7 (Takavol & Dennick, 2011). When looking at Trans 2, 3 and 4 we
28 can see that removing them from the model would be detrimental to the overall internal reliability of the model since the Cronbach’s Alpha would suffer as a result. Removing the first, second or third statements would lead to Cronbach’s Alpha reliability values of 0.269, 0.292 and 0.352 respectively, which are significantly below 0.7. Nonetheless, when looking at the first question – Trans1 – it is clear that taking it away from the model would benefit the reliability of the results. Since removing the first question lead to a Cronbach’s Alpha value of 0.91, the first question was taken away from the model to ensure more reliable results.
Although the scale was taken from a peer-reviewed article, the first item still showed that it lowered the Cronbach’s Alpha and therefore the reliability of the research. The scale was taken from Rawlins (2008), who defines transparency as the deliberate attempt to make available all legally releasable information along four dimensions. Therefore, the release of such information, according to him, has to be in a manner that is timely, accurate, balanced and unequivocal (Rawlins, 2008). Although it seems unlikely that taking a scale from a renowned paper would lead to one item of the scale being unreliable, there are a few reasons that could explain this phenomenon. The article was written in 2008, and as the exponentially evolving society we live in changes, so do cultures, countries and individuals alike. It is therefore possible that human behavior has changed enough in the last decade to make some of the scales unreliable in certain situations. The model taken from Rawlins (2008) is reliable since it has been used in many other academic articles, but in the specific context of this research the first statement renders it unreliable. To double check, this research ran a factor analysis for the transparency scale, which is explained in the following section.
29
6.4 Factor Analysis
In order to examine if the model is reliable and correct to use as four components in one dimension, a Factor Analysis should be run to confirm it. Factor analysis is a statistical method used to describe variability among the chosen variables in a model in terms of a lower number of unobserved variables (Kim & Mueller, 1978). The key in this table is the column titled cumulative %. This shows whether more components fit better as one single dimension or whether adding more components is actually detrimental to the model. If they belong together, they have high convergent validity.
Table 4: Factor Analysis results
Component
Initial Eigenvalues
Extraction Sums of Squared Loadings
Total
% of
Variance Cumulative % Total
% of Variance Cumulative % 1 2.608 65.190 65.190 2.608 65.190 65.190 2 .942 23.538 88.728 3 .311 7.768 96.496 4 .140 3.504 100.000
Since the cumulative % increases as the number of components increases, this indicates that if the research decides to use the 4 different items, the best to test the model is to create 1 dimension from the 4 components. So grouping the components together and making the transparency measure a single measure shows that this is the most effective way of testing the hypotheses in this model.
Following the Factor Analysis that proved that transparency should be taken as a single dimension comprised of four components, a Reliability Analysis was run for the Transparency scale again but with 3 items in it since the Trans1 component was omitted due to its unreliability. The table below presents the Cronbach’s Alpha value for
30 Transparency with only 3 components.
Table 5: Reliability analysis with Trans1 omitted
Cronbach's Alpha Cronbach's Alpha Based on
Standardized Items N of Items
.908 .911 3
As shown above, when Trans1 was taken out of the dimension of transparency, the Cronbach’s Alpha increased to a significantly reliable level. Therefore, it was correct to exclude the first dimension of transparency.
6.5 Manipulation Check
Manipulation check has been run in order to assess if the scenarios for high (1) low (0) Service recovery Quality and high (1) low (0) Transparency scenarios in the survey have been effectively manipulated. Manipulation checks are measured variables that show what the manipulated variables concurrently affect (Khan, 2011). A manipulation check is usually executed before commencing with the main analysis of the proposed variables and hypotheses since it checks whether participant perceived different levels of service recovery quality and transparency. The 4 scenarios were randomly assigned to different levels of experimental conditions and a t statistic analysis was used for the manipulation check. The manipulation check was used to test whether the two levels of Service Recovery Quality, namely low quality and high quality, and the two levels of transparency, namely a low and high level of transparency, differ on the dependent variable - customer satisfaction. Therefore, the manipulation check was used to assess if the scenarios have effectively manipulated customer satisfaction. The results, shown below in Table 6 and 7, highlight that all items were statistically significant, which means the scenarios were effectively manipulated.
31 Table 6: Manipulation check results for transparency
Transparency N Mean Std. Deviation Std. Error Mean Sig. Trans 0 Low 101 3.4521 *** 1.85088 .18417 .000*** 1 High 98 5.4626*** 1.30786 .13211
Note, means significantly different: *p < .05; **p < .01; ***p < .001 (two tailed)
Table 7: Manipulation check results for service recovery quality Service Recovery Quality N Mean
Std. Deviation Std. Error Mean Sig. SerRec 0 Low 100 2.6233 *** 1.32798 .13280 .000*** 1 High 100 5.1367 1.70303 .17030
Note, means significantly different: *p < .05; **p < .01; ***p < .001 (two tailed)
The manipulation check analysis results for transparency, as shown in Table 6, show that when the level of transparency is low, the mean is around 3.45. On the other hand, when the level of transparency is high, the mean is around 5.46. Since the difference between the low level and high level of transparency means is significant, the transparency scenario can be said to be effectively manipulated. Table 7, which shows the manipulation check results for service recovery quality, shows that with low levels of service recovery quality the mean is 2.63 whereas the mean with high levels of service recovery quality the mean is 5.14. Just as with transparency, since the difference between the means is significant, the scenario is effectively manipulated.
6.6 Descriptive Statistics & Correlation Analysis
In this section, the descriptive statistics and correlations between all the variables will be shown and briefly analyzed before starting with the hypothesis analysis. The first part of this subchapter will analyze the mean and standard deviations of all the proposed
32 variables whereas the second part will analyse the correlations between all proposed variables.
The means and standard deviations of the dependent, independent, moderator and control variables are shown in Table 8, which is presented below. This table is extremely helpful since it can show preliminary results regarding the data and it can give an indication with respect to whether the hypotheses proposed in this research will ultimately be confirmed. By looking at Table 8, it immediately catches the eye that the mean of the gender variable is exactly 1.5, which indicates that the sample was split exactly between 100 males and 100 females. Furthermore, as already explained before, the mean age of respondents was 25.63 years old, which indicates a young sample since this research was conducted mainly with university students. Finally, and possibly of most relevance to this study, the mean NPS before answering the survey was 6.32 whereas after the survey it was 5.09. This shows that, on average, respondents had less favorable views of the company in the scenarios after the service failure and recovery.
Table 8: Means and standard deviations of the dependent, independent and control variables
M SD 1. NPST1 5.09 3.175 2. ServiceRecQ .50 .501 3. Transparency .49 .501 4. NPST0 6.32 1.570 5. Age 25.63 5.688 6. Gender 1.50 .501
33 Following the analysis of the means of the proposed variables, it is helpful to look at the correlations between the variables in the model (Field, 2009). Just as the means can provide an indication with respect to the results of the hypotheses, the correlations between variables can also show whether the proposed hypotheses will be confirmed or not. Table 9, stated above, shows the correlations between the NPS before (NPST0) and after (NPST1) the survey, service recovery quality, transparency, age and gender. The significant correlations are shown in the table above and are marked with two asterisks to show their significance. Although most of the correlations are insignificant, there are still significant correlations that are of indication for the hypothesis testing. First and foremost, the correlation between age and NPS is insignificant, which means that age does not determine the rating given to a company. Likewise, the correlation between gender and NPS is also insignificant, at 0.35 and 0.57, Table 9: Correlations between all the proposed variables in the model
NPST1 ServRecQ Transp NPST0 Age Gender NPST1 Pearson Correlation 1
Sig. (2-tailed)
N 200
ServiceRecoveryQ Pearson Correlation .559** 1 Sig. (2-tailed) .000
N 200 200
Transparency Pearson Correlation .263** .000 1 Sig. (2-tailed) .000 1.000
N 200 200 200
NPST0 Pearson Correlation .205** .050 -.108 1 Sig. (2-tailed) .004 .488 .131
N 198 198 198 198
Age Pearson Correlation -.138 .083 -.080 -.094 1 Sig. (2-tailed) .054 .250 .266 .194
N 194 194 194 192 194
Gender Pearson Correlation .035 .106 .055 .057 -.077 1 Sig. (2-tailed) .626 .138 .437 .423 .286
34 which shows that gender also does not determine the score given to a company before or after the service failure and recovery. The significant correlations that are of use to this research are the correlations between NPS (NPST1) after service recovery and transparency, which is significant at 0.263, and between NPS (NPST1) after the service recovery and service recovery quality, at 0.559. These significant correlations show that transparency and service recovery quality both moderate the NPS after a service recovery. This means that the more transparent companies are and the higher the service recovery quality is, the more highly rated the company will be. Nonetheless, this does not prove statistical significance since this will be proven in the hypothesis testing, but this does mean that higher levels of transparency and service recovery quality are correlated with higher levels of NPS after a service recovery.
6.7 Hypothesis Testing
This subchapter of the results section will analyze and test the hypotheses that were proposed in this research. After the preliminary steps that included exploring and cleaning the data, factorial ANOVA was conducted to compare the main effects of service recovery quality and the interaction effect between transparency on customer satisfaction. ANOVA, which refers to the Analysis of Variance, is used to analyze intragroup mean differences and their variation. As Field (2009) states, an ANOVA, tells us whether three or more means are the same, so it tests the null hypothesis that all group means are equal. It is a manner of comparing the ratio of systematic variance to unsystematic variance in an experimental study (Field, 2009). More specifically, when an experiment has two or more independent variables it is known as a factorial design, which can also be applied to ANOVA, as is done in this research (Field, 2009). Therefore, a factorial ANOVA is used to analyze a situation in which there are two or
35 more independent variables (Field, 2009). A factorial ANOVA can provide a mixture of appropriate and inappropriate inferences about a variable when the statistics are applied to a related observed variable (Davison & Sharma, 1990). Since there are two independent variables in this research, the factorial ANOVA will be used to test the hypotheses.
When looking at the correlation statistics in Table 9, it seems that the hypotheses would be confirmed since both service recovery quality and transparency were significantly positively correlated to the dependent variable NPS. The factorial experiment had four treatments combinations in total: it showed a 2x2 between-subjects factorial design since each factor had two levels: informational transparency during the complaint handling process was either high or low and service recovery quality following the service failure was either high or low. A two-way analysis of variance was conducted on the influence of two independent variables, namely service recovery quality and transparency, and their effects on the rating of customer satisfaction through the Net Promoter Score. After running the factorial ANOVA analysis, the results were illustrated on a table and the significance levels were recorded. Table 10, which is titled Tests of Between-Subjects Effects and can be seen below, shows the significance levels for the factorial ANOVA and it can be used to analyse the proposed hypotheses in this research.
Table 10: Tests of Between-Subjects Effects with NPS1 as dependent variable
Source Type III Sum of Squares df Mean Square F Sig. Partial Eta Squared Corrected Model 927.822a 6 154.637 27.033*** .000 .469 Intercept 101.265 1 101.265 17.703*** .000 .088 NPST0 80.246 1 80.246 14.028*** .000 .071 Age 40.235 1 40.235 7.034*** .000 .037
36 Gender 11.158 1 11.158 1.951 .164 .010 ServRecQual 654.988 1 654.988 114.501*** .000 .384 Transparency 159.219 1 159.219 27.834*** .000 .131 ServRecQual*Trans 5.344 1 5.344 .934 .335 .005 Error 1052.544 184 5.720 Total 6866.000 191 Corrected Total 1980.366 190
Note: Statistical significance: *p < .05; **p < .01; ***p <.001 (two tailed)
The above table illustrates the results of the factorial ANOVA. The first hypothesis proposed in this research stated that higher levels of service recovery quality would lead to higher levels of Net Promoter Scores of participants, which would indicate a higher level of customer satisfaction following the service failure. As is shown in the table, the significance level for service recovery quality is 0.00 which indicates that the result is indeed significant at the .05 significance level (Field, 2009). The results have shown that, just as was predicted following the literature review, higher service recovery quality leads to higher customer satisfaction ratings at 95% confidence level.
The second hypothesis proposed through the model in this research stated that higher levels of transparency in the complaint handling process would positively moderate the effect of service recovery quality on customer satisfaction through NPS. Table 9, which shows the correlations between all the dependent, independent, moderator and control variables showed that, as an individual variable, transparency was positively correlated with service recovery quality. This is also shown in Table 10. Since the direct effect of transparency on service recovery quality is significant with a
37 significance level of 0.00. Nonetheless, this is not what hypothesis 2 is testing since it is testing transparency as a moderator variable. Therefore, an interaction variable between service recovery quality and transparency was made to analyse the moderating effect of transparency on the relationship between service recovery quality and customer satisfaction.
In Table 10, that shown above, the interaction variable between service recovery quality and transparency is shown as ServRecQual*Transp. When looking at the table we can see that the significance level of the interaction variable is 0.335, therefore it is not significant. The interaction effect was not statistically significant and, therefore, the hypothesis was not supported. This indicates that although higher levels of transparency lead to higher levels of customer satisfaction after a service recovery, transparency does not significantly positively moderate the effect of service recovery quality on the customer satisfaction ratings. Tables providing raw data such as frequencies and the full results of the analyses can be found in the tables in the Appendix 10.2.
38 Another way of illustrating the results of the hypotheses testing can be seen in the graph above. It is clear from the Graph 1 that higher levels of service recovery quality and transparency individually have a significant effect on customer satisfaction levels. Nonetheless, the lines do not intersect, which means that there is no synergy. This entails that the effect of service recovery quality on customer satisfaction remains unchanged and is not affected by transparency levels to the point that it is statistically significant.
The results show that there is a main effect of service recovery quality on the NPS. This shows statistically that higher levels of service recovery significantly lead to higher levels of customer satisfaction. The results also showed that there a statistical significant relationship between transparency and customer satisfaction. This entails that higher levels of transparency lead to higher levels of NPS. Nonetheless, as the factorial ANOVA showed, there is no statistically significant interaction effect, concluding that transparency does not moderate the effect of service recovery quality on customer satisfaction during a complaint handling process.
6.8 Hypothesis Testing - Process
For hypothesis testing we applied the ANOVA test due to simplicity of the model with two independent variables and one independent variable. The factorial ANOVA can be referred as a robust and powerful analysis when there is a good understanding of the properties in the model (Clifford, Higgins & Sawilowsky, 1989). However there are other more recent statistical tools to analyse and test hypothesis with complicated moderation and mediation effects such as PROCESS (Hayes & Rockwood, 2016). The PROCESS macro has been run for this study as well and resulted in the same statistical conclusions as the factorial ANOVA. Please see the Appendix 10.5 for the
39 statistical output.
7. Discussion
The purpose of this research was to check a direct positive relationship between service recovery quality and customer satisfaction in the complaint handling process. The second and more important purpose was to show that transparency during a service recovery would have an impact on service recovery quality and would therefore have a moderating effect on the relationship between recovery quality and customer satisfaction. The hypotheses that were provided in this research were rooted in theory, but ultimately the results were not entirely conformant with the predictions put forth in the hypothesis development.
7.1 Findings
The first research of this thesis stated that higher levels of service recovery quality would lead to more satisfied participants during the handling of a customer complaint. Through the use of a factorial ANOVA analysis the results indeed showed that these two variables are positively correlated, and that therefore a higher quality of service recovery would lead to more satisfied customers. These findings are in line with the theory that was analysed in the literature review and subsequently used as a basis for the hypotheses development. In fact, Fang, Luo and Jiang (2012) stated that both tangible recovery attributes and psychological recovery attributes lead to higher customer satisfaction ratings in users who experience a service failure. Customers who complain about an improperly delivered service and who experience a satisfactory resolution of that service failure will more favourably view and will more likely stay loyal to the company (OFT, 1990).
