Satisfying customers : a study about the influence of service factors on customers’ satisfaction level

Faculty of Economics and Business

Master thesis Business Studies

Satisfying Customers

A study about the influence of service factors

on customers’ satisfaction level

Inge van Amstel Student Number: 10400753 Supervisor: Dr. D. Dekker 30th _{of June, 2014}

Abstract

Service is an important unique selling point which enables companies to differentiate from competitors, and it largely determines the level of customer satisfaction. Personality traits also play a significant role in the satisfaction level of customers, whereby it has been found that being extravert positively influences satisfaction and being neurotic negatively influences satisfaction (Costa & McCrae, 1980). This study focuses on the relationships between different service behavior regarding the service factors of Parasuraman, Zeithaml & Berry (1985) and customer satisfaction, and explores how the personality traits extraversion and neuroticism moderate these relationships. Furthermore, it is explored how a state of delight can be achieved by customers, which has been found to be a higher emotional state then satisfaction. The relationship between the service factors, personality traits and delight is therefore explored.

Analyzing the measurement-scales of this study resulted in the finding that delight should not be measured through a different measurement-scale than satisfaction, a finding which impeaches earlier findings of Berman (2005). Results indicate that delight does not differ from satisfaction in emotional versus cognitive features, nor by surprising elements. The contrary feeling of delight, outrage, has been found to be a separate factor from satisfaction and delight, and therefore has to be measured differently. Therefore, it is explored how outrage is influenced by service behavior and personality traits.

Outcomes of this study show that extraversion and neuroticism have no moderating effect on the relationship between the service factors and satisfaction and outrage. Furthermore it was found that the knowledge and courtesy of employees and their ability to convey trust and confidence, together with their individualized attention for customers most strongly influence satisfaction and outrage. The factor tangibles shows to influence satisfaction, but has no influence on feelings of outrage, because satisfaction and outrage differ in their antecedents.

Key words: Service Factors, Customer Satisfaction, Customer Delight, Outrage, Personality

Table of Contents

1. Introduction

4

2. Literature review

7

2.1 Hospitality 7

2.2 Employee service behavior 9

2.3 Guest Satisfaction 12

2.4 Customer Delight 13

2.5 The Big Five personality traits 15

2.6 Research Gap 16

3. Data and Method

19

3.1 Sample 19

3.2 Research instruments 20

3.3 Control variables 22

4. Results

24

4.1 Data preparation 24

4.2 Factor Analysis SERVQUAL variables 28

4.3 Factor analysis Satisfaction and Delight 32

4.4 New research models 34

4.5 Results Hypotheses 1 37 4.6 Results Hypotheses 2 37 4.7 Results Hypotheses 3 39 4.8 Results Hypotheses 4 40

5. Conclusion

44

7. Appendix

61

1. Introduction

Nowadays, as part of a differentiation strategy, companies have started to add more service to their total offering. Service is seen as a strategic opportunity and a sustainable source of competitive advantage, because it creates value which is based on the competencies of the company, and therefore it is a unique resource which is hard to imitate (Wernerfelt, 1984). It has been shown that companies who focus on the service part of their business achieve better return on sales and thereby improve their value (Fang, Palmatier, & Steenkamp, 2008). So instead of services being add-ons to the product, services are becoming the center of the total offering, with products as add-ons to the services.

Although the success of services does not only rely on the human resources of a company, it has been examined that employees are the key to success at the ‘moments of truth’ (Rogelberg, Barnes-Farrell, & Creamer, 1999). Especially employees who are in direct contact with the customers, have great influence on how the customers perceive the service of the company. By providing excellent customer service, a high level of customer or guest satisfaction can be achieved, which enables a company to differentiate itself from its competitors. It is also found by Rogelberg, Barnes-Farrell & Creamer (1999) that excellent customer service leads to customer satisfaction which, in turn, increases the customer's desire to use the supplier's services in the future and spread a positive word of mouth (WOM).

Companies can go even further in providing service, by fulfilling services which are non-expected by their customers and create a state of delight (Berman, 2005; Kano, Seraku, Takahashi, & Tsuji, 1984). Customers may feel delighted when they experience surprises and feelings of joy, and it has been found that delight has an even greater influence on repurchase intentions and positive (e)WOM then satisfaction (Magnini, Crotts, & Zehrer, 2011).

An industry where the main product of companies has always been providing service to customers is the hospitality industry (King, 1995). “Emphasizing on developing exceptional quality guest relationships helps hotels to build valuable, long-term relationships with their customers. This has resulted in hoteliers becoming much more aware of the quality and value of services being provided by them and desired by potential customers. Therefore, the players in hotel industry like other service industries attempt to overtake their competitors and provide higher product and service quality to attract repeat customers and achieve long term success (Douglas & Connor, 2003)” (in: Ariffin & Maghzi, 2012, p.191).

According to customer loyalty, research has found that people are unlikely to revisit a hotel a second occasion, due to their variety seeking behavior (Ariffin, 2008). An extraordinary

level of service is a factor which can influence the guest in its decision to come back. A high level of satisfaction creates a strong bonding between the guest and the hotel, and is difficult to be imitated by other competing hotels (Ariffin & Maghzi, 2012). A development which has forced hotels to focus even more on providing excellent services, is the trend of electronic word-of-mouth (eWOM) (Sun, Youn, Wu, & Kuntaraporn, 2006). The advances in information technology and the introduction of new methods of communication have led to the development of this trend, also known as online reviews, online recommendations, or online opinions. Compared to traditional WOM, eWOM is more influential for the purchase decisions of consumers due to its convenience, speed, one-to-many reach, and its absence of face-to-face human pressure (Sun et al., 2006). “This breadth of eWOM scope and ease in accessing reviews can deeply affect a company’s performance. The tourism industry is strongly affected by eWOM and, within the tourism industry, hotels are probably the most affected.” (Serra Cantallops & Salvi, 2014).

Recent studies (Ariffin & Aziz, 2012; Ariffin, Nameghi, & Zakaria, 2013) have examined which service factors influence the level of guest satisfaction. These studies tested the factors personalization, warm welcoming, special relationship, straight from the heart and comfort, whereby personalization was found too most strongly influence guest satisfaction, and comfort too less strongly influence guest satisfaction. In these studies, the participants were generalized as one group with the same preferences for service factors. Other studies (Jani & Han, 2013; Jani & Han, 2014) have explored how differences in personalities of hotel guests affect their level of satisfaction. Hotel guests were tested on how they scored on the Big Five personality traits, and these results were related to the level of satisfaction. Results of these studies show that the personality trait neuroticism influences satisfaction negatively, in comparison to the four other traits which influence satisfaction positively. Delight is a relatively new subject in research, which has been found to be related to satisfaction, but measured in a different way.

Because it has been found that personality traits influence a person’s satisfaction level, and it has been found that there are several service factors which influence the level of satisfaction, this study will empirically examine how service factors influence the level of satisfaction and delight, and thereby explore how the personality of the guest moderates this relationship. The results of this study will contribute to the research gap of how service oriented employees could adjust their service behavior to the different preferences of hotel guests, and thereby enable hotel management to train their employees to perform the appropriate service behavior for the diverse personalities of hotel guests. Because by performing the right service

behavior, the level of the guest’s satisfaction and delight is expected to increase which will lead to loyal guest behavior and positive €WOM, and thus generate positive results for the company.

2. Literature review

This literature review provides insight into the literature of the variables which are part of this study. The arise and development of variables will be discussed, together with their antecedents. Besides, more insight will be given into the relationship between the different variables.

2.1 Hospitality

In 1999, Brotherton studied the term hospitality and defined it as follows: “A contemporaneous human exchange, which is voluntarily entered into, and designed to enhance the mutual wellbeing of the parties concerned through the provision of accommodation and food or drink” (p.167). He created a model (figure 1) which describes the dimensions of hospitality.

Figure 1: The dimensions of hospitality by Brotherton (1999)

The study of King (1995) examines hospitality in a more extensive way, and describes the history and development of hospitality. The research of King (1995) gives an overview of the definitions of hospitality which are conducted by Hepple, Kipps & Thomson (1990). They state that hospitality is conferred by a host on a guest who is away from home, and that it is interactive because it involves the coming together of a provider and receiver. Furthermore, they state that hospitality is comprised of a blend of tangible and intangible factors, and that hospitality means that a host provides for the guest’s security, psychological and physiological comfort. The following figure 2 provides an overview of the key elements of commercial hospitality according to King (1995).

Figure 2: Key elements of commercial hospitality according to King (1999)

Figure 2 shows that the organization supports the employee by providing resources which are needed to deliver service to the guest. Furthermore, the organization creates a hospitable work environment, and has in-depth knowledge of the guest’s desires and expectations through ongoing market research, which enables the employee to meet the needs and expectations of the guest. The figure also shows an interactive relationship between the employee and the guest. The employee understands the guest’s desires and expectations, and the social skills of the employee enables him to interact with the guest with the right social distance. Furthermore, the interactive relationship between the employee and guest is affected by the social rituals which are associated with arrival and departure of the guest. “The employee has the social skills to guide the guest so that the guest receives the desired level of service, service to other guests is not affected, and the organization is not disrupted by such guests. In this respect, the employee maintains control over the process.”(King, 1995, p.231).

Lashley (2008) also examined hospitality, and developed a model which consists of three hospitality domains. These domains are acknowledged as cultural/social, private/domestic and commercial domains. The first domain is about providing hospitality to others without the obligation for guests to reward the host. This absolute hospitable behavior involves the hosts to permit their guests to respond and act as they like. The second domain, hospitality at private/domestic settings, refers to the hospitality offered by individuals towards others in private setting such as at their homes. The last domain of commercial hospitality

involves services provided by café, restaurant, hotel and catering businesses through which foods, drinks and accommodations are provided in return for payments. The domain of hospitality at private/domestic settings is normally used by people as a benchmark to evaluate the level of hospitality offered by the service providers in the commercial setting.

Like the theories of Brotherton (1999), King (1995) and Lashley (2008), other recent studies about hospitality also describe the relationship between the employee and guest as base for hospitality. Brotherton & Wood (2008) refer to employees treating guests in a friendly, warm and generous way, and the study of Lynch, Molz, Mcintosh, Lugosi, & Lashley, (2011) describes hospitality as “a process that includes arrival, which involves greeting and making the guests feel welcome, providing comfort and fulfilment of the guest’s wishes, and, at the time of departure, thanking guests and extending an invitation to return” (in: Arrifin et al., 2013, p.128).

To conclude, according to the literature, hospitality is about the interaction between the employee and the guest, whereby the employee is supported by the organization to be able to provide in the services which the guest expects to receive. From the moment of arrival until the departure of the guest, the employee should fulfil the guest’s wishes and needs. This study will further explore how the relation between the guest and employee can be improved, by focusing on how employees in the hospitality industry can provide the best service behavior according to the wishes and needs of the guests.

2.2 Employee service behavior

According to the study of Martin (1986), the employee’s attitude, behavior and verbal skills affect the level of fulfillment of the guest’s needs. Hospitality consists of both tangible and intangible elements, and according to the study of Ariffin & Maghzi (2012), the intangible host-guest interpersonal relationship is the core differentiating factor between hospitality and service. “Hospitality in a commercial or organizational setting is a specific kind of relationship between individuals-a host and a guest. In this relationship, the host understands what would give pleasure to the guest and enhance his or her comfort and well-being, and delivers it generously and flawlessly in face to face interactions, with deference, tactfulness and the process of social ritual. The objective is to enhance guest satisfaction and develop repeat business.”(King, 1995, p.229).

Several studies have examined the effects of different service factors on guest satisfaction. Ariffin & Maghzi (2012) designed a five-factor structure to explain the

dimensionality of hospitality specifically in the context of hotel services. Outcomes of the study show that the factor personalization is the most important factor, followed by warm welcoming, special relationship, straight from the heart and comfort. Earlier findings of Suprenant & Solomon (1987) and Smith (1994) support these outcomes. According to the service factors, hospitable behavior of employees focusses on the emotional aspect rather than the functional aspect of the service. “The hospitality businesses must strive to offer memorable services by focusing on guest experiences that stimulate their senses. Firms must act like hosts at private settings and find ways to create surprising “moments of truth” for their guests” (Ariffin & Maghzi, 2012, p.192).

There is also a negative effect related to the expected service behavior of employees. Research has examined the so called “role conflict” which employees may experience during service behavior (Shamir, 1980). While it is important that the standards and requirements of the company are achieved, the emotional equilibrium and self-esteem of the employee should also stay intact (Shamir, 1978; Shamir, 1980). It is found that employees feel satisfied when their guests experience a high level of guest satisfaction, but the required service behavior which is needed from employees to achieve a high level of guest satisfaction may affect the feelings of the employee in a negative way. Service employees are expected to not show negative feelings such as frustration or anger towards their guests. They are expected to show emotions which they may not feel, but which help the business to achieve higher levels of customer satisfaction. A study of Hochschild (1983) examined customer contact jobs which required employees to manage their emotions to create the desired emotion for the customer, also known as emotional labour. Hochschild (1983) found that emotional labour can negatively affect the employees’ wellbeing, because it alienates the employees’ own feelings.

Experimental research has explored whether the guest’s reaction to inauthentic service behavior is less positive than to authentic service behavior. Results have shown that guests are less impressed with acting performances of service employees who tried to hide their own emotions in the service delivery (Ekman, 1992; Grandey, Fisk, Mattila, Jansen, & Sideman, 2005; Langhorn, 2004).

To decrease the chance that employees experience a “role conflict”, attention should be given to the level of emotional intelligence of the employees. The study of Langhorn (2004) examined the effect of the Emotional Intelligence Quotient on service behavior and guest satisfaction. It is found that employees who score high on emotional intelligence positively affect guest satisfaction and business performance. According to Langhorn (2004), “employees

who are emotionally intelligent are said to have abilities in five domains. They recognize their own emotions and express them to others, recognize and understand the emotions of others, use emotions with reason and emotional information in thought, regulate and manage their own and the emotions of others, and they control strong emotional states like anger, frustration, excitement, anxiety, etc.”(p.79).

In 1985, Parasuraman, Zeithaml and Berry developed a model which enables companies to measure how guests perceive the quality of the provided services, also known as the SERVQUAL model. This model is founded on the gap between what the customer expects from a service, and the evaluation of the performance of the service. The original SERVQUAL model consisted of ten dimensions, but in 1988 these were reduced to the following five dimensions: tangibles, reliability, responsiveness, assurance and empathy (Buttle, 1996a). The dimension tangibles relates to the appearance of physical facilities, equipment, personnel and communication materials. The second dimension, reliability, refers to the employees’ ability to perform the promised service dependably and accurately. The dimension responsiveness is about the employees’ willingness to help customers and provide prompt service. The dimension assurance links to the knowledge and courtesy of employees and their ability to convey trust and confidence. The last dimension, empathy, refers to the provision of caring and individualized attention to customers. Research on the SERVQUAL model has found evidence that these dimensions include all factors which are related to service behaviour in organizations, because the service factors which are used in other studies can all be related to the service factors of the SERVQUAL model (Ariffin & Maghzi, 2012; King, 1995; Parasuraman et al., 1985; S. L. Smith, 1994; Surprenant & Solomon, 1987). Personalization, together with Straight from the heart, can be linked to the factor Empathy. Comfort and Special relationship match with the factor Assurance, and the factor Warm welcoming is related to the same values as the factor Responsiveness.

In summary, the literature has found main service factors which influence guest satisfaction, which all can be linked to the dimensions of the SERVQUAL model. Employees are the key to success at the moments of truth, and they largely influence the level of guest satisfaction. Attention need to be given to the so called role-conflict, which may occur when employees do not fully support the service behavior which the organization and guest expect them to perform. Training the emotional intelligence of employees may decrease the negative effects of service behavior for employees. Therefore, it could be of importance for hotel management to find out more about the emotions of guests and the factors on which they base their level of satisfaction.

This knowledge enables employees to perform the expected service behavior which satisfies the guest. This study will use the SERVQUAL model to examine how hotel guests perceive provided service factors.

2.3 Guest Satisfaction

“Guests expect a greeting and a welcome, as well as thanks and an acknowledgement upon departure. When these rituals are not performed, hospitality is lacking, and guests may be dissatisfied” (King, 1995, p.229). It has been examined that the level of guest satisfaction depends on the extent to which the received services match the guests’ expectations (Austin, 1992; Westbrook & Oliver, 1981). This finding is supported by the World Tourism Organization (1985), which defines guest satisfaction as “a psychological concept that involves the feeling of well-being and pleasure that results from obtaining what one hopes for and expects from an appealing product and/or service”.

Literature research shows that the level of guest satisfaction in hotel services is primarily influenced by how the guest’s requirements, wants and needs are fulfilled by both tangible and intangible elements of the services (Ahmad, M Ariffin, & Ahmad, 2008; Greathouse, Gregoire, Shanklin, & Tripp, 1996; Lam & Zhang, 1999). Herzberg (1966) and Balmer & Baum (1993) examined the effect of tangible elements on the satisfaction level of guests. Results of these studies showed that when expectations of the tangible elements were not met, this created dissatisfaction. But exceeding the expectations of the tangible elements did not produce satisfaction.

It has also been examined that the level of guest satisfaction is influenced by the comparison of experiences in private hospitality settings to experiences in commercial settings (Lashley, 2008). Guests evaluate hospitality experiences primarily in emotional terms. Therefore, service employees should be aware of the emotional dimensions of guests and how to meet these emotional needs. The study of Lashley (2008) confirms the earlier finding that employee service behavior affects the guests’ emotional experience which influences guest satisfaction and thereby positive word of mouth and guest loyalty.

Concluding, to achieve guest satisfaction, provided services should meet the expectations of the guest. This can be achieved by different factors which fulfill both tangible and intangible elements of service. Guests evaluate hospitality experiences primarily in emotional terms, and compare experiences in commercial settings to experiences in private settings. This study will measure the level of guest satisfaction according to the measurement scale of Westbrook &

Oliver (1981), which focusses on the match between expected and perceived services.

2.4 Customer Delight

In 1980, the customer delight construct was developed by the study of Plutchik. He found that a combination of the emotions joy and surprise resulted in a feeling of delight. Nowadays, researchers are still trying to find a measure scale which examines the antecedents of the state of delight. According to Berman (2005), “Delight is a construct related to but separate from satisfaction as it is based on different things (in the same way that dissatisfaction is related to but distinct from satisfaction). While customer satisfaction is generally based on exceeding one's expectations, customer delight requires that customers receive a positive surprise that is beyond their expectations” (p.129). This statement is confirmed by the recent study of Magnini, Crotts and Zehrer (2011), who explain that there is no lineair relationship between dissatisfaction, satisfaction and delight.

The model of Kano, Seraku, Takahashi & Tsuji (1984) explains the difference between satisfaction and delight. Kano et al (1984) discussed the conventional model of customer satisfaction, which proposes that higher satisfaction occurs regardless of the nature of the attributes. “Based on observations of nonlinear relationships between product attribute performance and customer satisfaction,Kano et al. (1984) put forth a proposition—that not all attributes are equalin the consumer’s mind—and an understanding of the inequities of attribute performance and their impact on consumer satisfaction is the key to increasingoverall customer satisfaction, as well as remaining competitive in the market place (Gregory & Parsa, 2013, p.26). Kano et al. (1984) developed five categories which are related to the expected effect of product attributes on satisfaction, from which the following three categories are mostly used. The first category is attractive requirements, whereby product attributes result in customer delight if these attributes are present, but they do not result in dissatisfaction when they are absent. Because these attributes are not expected, they cause delight when perceived. As found in the study of King (1995), hotels can differ from their competitors by creating an unexpected service level with surprising factors, and therefore, companies in the hospitality industry focus on achieving a state of delight by their customers. The second category contains one-dimension requirements, which are attributes that customers usually explicitly demand for. The higher the level of fulfilment of these requirements, the higher the level of customer satisfaction. The third category is for the must-be requirements, which can lead to an extreme state of dissatisfaction when not fulfilled. A fulfillment of the requirements does not lead to satisfaction, it only causes a state of dissatisfaction when not fulfilled. The following figure 3 shows the Kano model.

Since the introduction of this model in 1984, it has been widely used in a variety of contexts within the academic literature, including hospitality research.

Figure 3: Kano’s model of customer satisfaction (Berger et al., 1993)

According to Magnini et al., (2011), repurchase intentions and willingness to recommend are the most commonly reported outcomes of customer delight, and therefore companies are focused on achieving levels of delight by their customers. Berman (2005) describes several important differences between satisfaction and delight. First of all, satisfaction is considered to be cognitive and based on perceptions, while delight is considered to be affective and based on emotions. Furthermore, satisfaction is based on meeting or exceeding expectations, while delight is based on fulfilling unexpected features. Another important difference is that satisfaction has a weaker memory trace than delight. The SERVQUAl model measures customer satisfaction by comparing customer’s expectations to customer’s received experiences, which is influenced by the one-dimensional and must-be requirements. Customer delight can be measured by the presence of attractive requirements.

For this study, a self-contructed measurement scale is developed which is based on the findings of Berman (2005) and Dekker (2014), and the construct of this measurement-scale will be examined during this study. Furthermore, this research explores if the participants feel delighted during their hotel experience and examines how delight differs from satisfaction in its reaction to the service factors as antecedents and personality traits as moderators.

2.5 The Big Five personality traits

Personality reflects an individual’s psychological characteristics, and determine individual’s behavior, thinking and feeling. It is found to have a significant role in consumer behavior (Mowen, Park, & Zablah, 2007). The most commonly used theory for categorizing different personalities is that of the Big Five personality traits, which consists of the factors openness to experience, conscientiousness, extraversion, agreeableness, and neuroticism, and is based on studies about personality structure (Goldberg, 1990; Norman, 1963; Saucier & Goldberg, 1996). Openness to experience refers to the individuals’ proactiveness in seeking and appreciating innovation, conscientiousness relates to the individuals’ achievement propensity, extraversion factors show the quantity and intensity of the individuals’ interpersonal interaction, agreeableness relates to the individuals’ orientation toward being empathetic with others, and neuroticism is about the individuals’ proneness to psychological distress (Goldberg, 1990).

Studies have shown that the Big Five personality traits influence consumers’ affective responses (Orth, Limon, & Rose, 2010), satisfaction (Faullant, Matzler, & Mooradian, 2011; Lin & Worthley, 2012; Matzler, Faullant, Renzl, & Leiter, 2005), post-purchase behavior (Mooradian & Olver, 1997) and loyalty (Durukan & Bozaci, ). The study of Costa & McCrae (1980) examined the influence of extraversion and neuroticism on satisfaction and well-being, and the outcomes can be seen in figure 4. They found that extraversion causes satisfaction, whereas neuroticism causes dissatisfaction, which both influence the state of happiness.

Figure 4: A model of personality influences on positive and negative affect on subjective well-being (Costa & McCrae, 1980)

The study of Costa & McCrae (1980) focused on the effect of satisfaction on the level of happiness. Other studies have examined the impact of personality on customer satisfaction, and

their results confirm the findings of Costa & McCrae (1980). The study of Gountas & Gountas (2007) noted a significant relationship between personality factors and satisfaction. Faullant et al. (2011) and Mooradian & Olver (1997) found that extraversion had a positive effect on customer satisfaction, and neuroticism had a negative effect on customer satisfaction. The study of Orth et al. (2010) found that the factor neuroticism was negatively related to satisfaction. Furthermore, the study of Jani & Han (2014) found a significant impact from extraversion, agreeableness and neuroticism on guest satisfaction. All findings support the outcome of the study of Baudin, Aluja, Rolland, & Blanch (2011), namely that extraversion and neuroticism are the universal factors for satisfaction in general.

Reviewing the existing literature on the effect of the Big Five personality traits on guest satisfaction, it is found that extraversion and neuroticism most strongly affect guest satisfaction. Therefore, in this study, the moderating effect of these personality traits on the relationship between the SERVQUAL service factors and satisfaction and delight will be examined and explored.

2.6 Research Gap

Regarding the findings in the literature about the influence of personality traits on satisfaction, and the relation between the SERVQUAL service factors and satisfaction, this study will examine how extraversion and neuroticism moderate the relationship between the SERVQUAL service factors and satisfaction. Furthermore, the effect of the service factors and personality traits on delight will be explored. The research question of this study is:

“How do the personality traits Extraversion and Neuroticism moderate the relationship between Service Factors and Satisfaction? And how does this model differ when Satisfaction is replaced by Delight?”

Studies have found that service factors cause different effects on customer satisfaction (Kano et al., 1984; Parasuraman et al., 1985), whereby empathy is found to most strongly influence satisfaction (S. L. Smith, 1994; Surprenant & Solomon, 1987). Therefore, the first hypotheses is stated as follow:

H1: The service factors of the SERVQUAL model influence the level of satisfaction, whereby it is expected that the service factor empathy has the strongest effect on satisfaction.

Furthermore, it is found that the personality traits of the Big Five Factor theory influence the guests’ level of satisfaction (Baudin et al., 2011; Faullant et al., 2011; Gountas & Gountas, 2007; Jani & Han, 2014; Mooradian & Olver, 1997; Orth et al., 2010). Results of these studies show that the personality trait extraversion has a strong positive effect on guest satisfaction, and that neuroticism has a strong negative effect on guest satisfaction. To examine these findings, the following hypotheses are developed:

H2: Extraversion moderates the relationship between service factors and guest satisfaction, such that the relation is stronger when extraversion is high.

H3: Neuroticism moderates the relationship between service factors and guest satisfaction, such that the relation is weaker when neuroticism is high.

The model of the first three hypotheses is illustrated in the following figure 5.

Figure 5: Conceptual model hypotheses 1,2 and 3

During the last decades, the literature has started to focus on the new concept called Customer Delight (Berman, 2005; Kano et al., 1984; Magnini et al., 2011; Plutchik, 1980). Therefore, this study will also explore the level of customer delight of the participants of this study. It has been found that customer delight is related to satisfaction, but it is affected by unexpected features instead of expected features, and it is measured by emotional instead of cognitive attributes (Berman, 2005). To find out more about how satisfaction differs from delight, the following explorative hypotheses are developed.

H4: Explore how the model of Hypotheses 1,2 and 3 differ when Satisfaction is replaced by Delight.

H4a: Explore how the service factors of the SERVQUAL model influence the level of Delight.

H4b: Explore how Extraversion moderates the relationship between the service factors and delight.

H4c: Explore how Neuroticism moderates the relationship between the service factors and delight.

3. Data and Method

The following part describes the sample of this study, and illustrates how the survey will be designed and distributed. Furthermore, attention will be given to the research instruments of this study.

3.1 Sample

The population of the sample of this study consists of all Dutch people who have visited a hotel in the last year. There is no clear overview of this population, but an estimation can be made from studies of the “Centraal Bureau voor de Statistiek” (CBS). One research found that in 2012, 24% of the Dutch population stayed in a hotel for leisure goals (Veerkamp, Boskamp, van der Meulen, 2012). Other studies of the CBS show that 40% of the hotel bookings are business related, which means that there are also people of the Dutch population who visit hotels for business goals (CBS, 2011). Therefore, an assumption is made that the total percentage of Dutch people who has visited a hotel in the past year and thus can participate in this study is 35%. The total Dutch population counts 16.8 million people. Due to the large sample size of 35% of this population, the number of respondents that are needed for this study is 385 (Smith, 2013). This number is based on a confidence level of 95%, together with an estimated standard deviation of 0.5.

The participants of this study are found by a non-probability self-selection sampling technique. The survey is designed as a questionnaire, which is spread online by using the social networks Facebook and LinkedIn, and the questionnaire is sent by e-mail to family, friends and relatives. The questionnaire can be found as appendix in chapter 7. Furthermore, the receivers will be asked to distribute the questionnaire in their network. This means that this study makes use of convenience sampling and the snowball sampling technique. Because of the large sample size, it will be most efficient to spread the survey online because this way it can reach the highest number of people in a short time.

Although this study contains a very large population, the time limit of the research and the number of people of the population who will actually receive the survey should be taken into account when estimating the response rate. Counting the connections in the social networks, an estimation can be made that around 600 people will receive the survey. To achieve the highest possible response rate, the survey will be designed and distributed in a user-friendly way (Fan & Yan, 2010). Considering results of studies which have examined the average response rates from online questionnaires, it is expected that 40% of the people who receive

the survey will actually participate in this study, which should give a result of 240 respondents (Deutskens, De Ruyter, Wetzels, & Oosterveld, 2004; Kaplowitz, Hadlock, & Levine, 2004).

3.2 Research instruments

To ensure the validity of this study, validated measures from the literature are used. The independent service factor variables are measured by the five service dimensions of the SERVQUAL model of Parasuraman et al. (1985), which include the dimensions tangibles, reliability, responsiveness, assurance and empathy. The reliability scores has been found to be respectively .72, .83, .82, .81 and .86 (Parasuraman et al., 1985). Evidence is found that these five dimensions include all factors which are related to service behaviour in organizations (Parasuraman et al., 1988). Participants are asked about the experience of their last hotel visit, and they rate this visit on 22 items distributed over the five dimensions, according to a five-point Likert scale which ranges from 1 (strong disagree) to 5 (strongly agree). An example item of the SERVQUAL model for the dimension empathy is “The employees of the hotel understand the needs of the guests”.

The moderating variables extraversion and neuroticism are measured by using the model of Goldberg (1992). This model provides a five-point Likert scale, ranging from 1

(strong disagree) to 5 (strongly agree), and conducts 10 items for each personality trait, of

which “I am the life of the party” is an example item for extraversion and “I am easily disturbed” an example item for neuroticism. By measuring 10 items for both extraversion and neuroticism, a high reliability level will be ensured. The internal consistency of both factors is measured .86 (Goldberg, 1990).

The dependent variable Guest Satisfaction is measured by the well-established scale developed by Westbrook & Oliver (1981). This model consists of four items which are measured through a five-point Likert scale, ranging from 1 (strong disagree) to 5 (strongly

agree). An example item is “My choice to stay at this hotel was a wise one”. The reliability

score of this model has been found to be .93 (Westbrook & Oliver, 1981).

The dependent variable delight is measured by a self-established measurement scale, based on the study of Berman (2005) and Dekker (2014). Berman (2005) states that there is no linear relationship among dissatisfaction, satisfaction and delight, and “extremely satisfied” does not equate “delight”. According to Berman (2005), customer delight is frequently measured on a scale which ranges from outrage to delight. He defines outrage and delight as follows. “Outrage and pain occur when a customer experiences a poor and unanticipated scenario, and delight occurs as a result of fulfilment of unexpected, valuable, memorable, and

positive reproducible events” (p.132). Compared to satisfaction, delight is a more positive and emotional response, which commits a customer to a product. “At the far right end of the customer satisfaction continuum is the zone of delight. A number of academics and consultants view positive surprise and joy as the key ingredients in customer delight (Oliver, Rust, & Varki, 1997). According to two management consultants, "the key to creating a memorable service…is to create conditions and do things that are unexpected, unpredictable, valuable, memorable, and reproducible (Zemke & Bell, 2003)”(Berman, 2005, p.133). Another study mentions the term "positively outrageous service" for customer delight, and describes this state as unexpected, random, extraordinary, and disproportionately positive (Gross, 1991). According to Berman (2005), outrage and delight occur when performance is not related to a prior expectation, and are influenced by the element of surprise. “For example, performance that is a positive surprise will result in delight, whereas a positive expected level of performance yields satisfaction. Likewise, a negative surprise results in outrage and pain, and a negative expected performance results in dissatisfaction.”(Berman, 2005, p.135). On the basis of the described findings about customer delight, five items are developed to measure the level of customer delight. According to the literature, delight is described and defined by the words joy, positively surprising elements and exceeding expectations. Therefore, the first three items to measure customer delight are based on these three factors. The opposite of delight has been found to be described by the words outrage and negatively surprising elements. Therefore, two reversed items are developed on the base of these factors. All items are measured on a five-point Likert scale which ranges from 1 (strong disagree) to 5 (strongly agree).

1. I experienced a feeling of joy during my stay at the hotel 2. I am positively surprised during my stay at the hotel 3. My stay at the hotel has largely exceeded my expectations 4. I experienced a feeling of outrage during my stay at the hotel 5. I am negatively surprised during my stay at the hotel

Furthermore, the emotional state of guests will be measured by using a measurement scale of Dekker (2014), which demands the guest to choose an emoticon that expresses the emotion which fits the experience of the hotel visit. Figure 7 shows the four emoticons from which the participant may choose.

Figure 7: Measure-scale Delight by Dekker (2014)

This measurement scale refers to the findings of Berman(2005), which states that delight is most often measured on a scale which ranges from outrage/pain to delight, whereby these two emotions outdrive the statuses of dissatisfaction and satisfaction.

3.3 Control variables

In order to minimize unintended effects that influence the results of this study, several control variables are included in this study. These variables are held constant in order to make a more adequate investigation of the relationship between the independent and dependent variables (Field, 2009).

It has been found that age is an important variable which influences the level of customer satisfaction, because age affects the information-processing abilities to evaluate a product or service (Homburg & Giering, 2001; J. Smith & Baltes, 1990). Therefore, participants of this study are asked to indicate their age in years.

Studies have also found evidence that the variable gender influences the level of satisfaction, and that gender affects how satisfaction is achieved (Calasanti, 1996). Research has found that women and men differ in their information-processing abilities, and show that women are more sensitive to the relational aspect of service behavior than men (Mattila, 2000). The control variable level of education is included because it has been found that the level of education may be related to the personality of a person (Borg & Shapiro, 1996). Therefore, it could be that people with high education levels rate their hotel stay on different service factors than people with low education levels.

The number of overnight stays is included as control variable, because it is assumed that guests have the chance to experience more service features when there stay is longer, which may influence their level of satisfaction. Furthermore, it is presumed that during a long stay, guests require for more service features then during a short stay.

Another control variable is the hotels star rating, because it has been examined that expectations of guests differ according to the star rating of a hotel. Hotels with high star ratings induce high expectations of hotel guests, which are often not met, whereas hotels with low star ratings induce low expectations of hotel guests which are often exceeded (López Fernández & Serrano Bedia, 2004).

The sixth control variable measures whether a hotel stay was for business or leisure purposes, because research has found that guests differ in their needs and expectations according to the purpose of their stay (Chu & Choi, 2000). The study of Kashyap & Bojanic (2000) found that business guests value the comfort of the room less than leisure guests, because on average, business guests spent less time in their room. Furthermore, the study found that business guests make more use of hotel services, and the service performance of employees influences the level of guest satisfaction of business guests more than it influences the level of satisfaction of leisure guests (Kashyap & Bojanic, 2000; Lewis & McCann, 2004).

The last control variable measures in which country the hotel was located, because studies have examined that service behavior differs according to different countries and cultures. This is caused by differences in environmental, economic and socio-cultural factors (Malhotra, Ulgado, Agarwal, & Baalbaki, 1994). Furthermore, it has been examined that the cultural differences in the dimensions of Hofstede also influence service behavior of employees (Furrer, Liu, & Sudharshan, 2000).

4. Results

The sample of this study consists of 170 participants, from which 37% are males and 63% are females, and who vary in age from 19 to 87 years old. The average age is found to be 34 years. Regarding the education level of the participants, 48% has finished higher professional education, followed by 29% who has finished university and 20% who has finished high school. A small percentage has finished vocational education or a post-university science study. 15% of the participants visited a hotel for business purposes, others described their stay as a leisure activity. Most participants stayed in a four-star classified hotel, followed by three- and two-star classified hotels. Unfortunately, there were too much missing values for the control variables about the country where the hotel was located and the number of nights that a participant stayed in a hotel. Therefore, these control variables cannot be used in the further analysis of this study.

4.1 Data preparation

To ensure a clear overview of the 170 surveys, a codebook was created by using the IBM SPSS Statistics Program version 22. The collected data was transferred from the online questionnaire program Qualtrics into IBM SPSS.

The frequencies of all variables were examined to see whether questions where unanswered. For questions through the dataset which were not answered, the code 999 was created as “discrete missing value”, and empty answers were labeled as 999. While analyzing the data, these items seemed to be missing completely at random, because the missing values did not depend on another variable, or on the variable itself (Myers, 2011). For items which had 10 percent or less missing values, The Hotdeck method was used to replace the missing values, whereby a “donor” fills in the empty values of the “donee” (Myers, 2011). After replacing the missing values, the counter indicative items of the survey were reversed and recoded. Then, new total variables were computed which represent the average score of the independent and dependent variables, and the moderators. With these new variables, further statistical analyzes could be performed.

Statistical analyzes can only be performed when data is normally distributed. This means that the scores of the variable are distributed over a symmetrical, bell-shaped curve, with the greatest frequency of scores in the middle and with smaller frequencies towards the extremes (Pallant, 2013). The normality of the variables was assessed by analyzing the normal distribution and the skewness and kurtosis of the data. The skewness value provides information of the symmetry of the distribution, whereas kurtosis provides information about the extent to which the distribution of the data peaks at one point. Results of the data showed

no highly extreme values of skewness and kurtosis, and therefore, no data had to be transformed. Analyzing the outliers, each variable had several outliers. But looking at the difference between the mean and the trimmed mean, it had almost no effect when the extreme high and extreme low five percent of variables were extracted. Therefore, outliers were not exclude from the data set (Pallant, 2013). All variables can be defined as normally distributed. To find out if the items of a variables measure the same underlying construct, the Cronbach’s alpha coefficient was used, which has been found to be the most common measure of scale reliability (Cortina, 1993). Previous studies have found that a Cronbach’s alpha coefficient above 0.7 shows a reliable construct between items. As can be seen from table 1, the Cronbach’s alpha coefficient shows all results above .7. The SERVQUAL service factor variables score between .728 and .859, and the personality traits extraversion and neuroticism score respectively .798 and .767. The dependent variable satisfaction has the highest reliability score of .938, and the other dependent variable delight scores .842. These scores differ no more than around .1 above or below the earlier found reliability scores which are described in the method section of this study. Because the variable delight is developed around a self-constructed measure scale, the reliability of this variable can not be compared to earlier findings. However, the score of .842 shows that the items of the variable have a strong internal consistency.

Table 1 also provides other descriptive statistics of the variables, including the mean, standard deviation and their inter-correlations. The mean scores of the moderators show that the participants of this study score on average higher on the personality trait extraversion (3,83) than on the personality trait neuroticism (2,11). The SERVQUAL variables can have a mean score between the 1 and 5, and as shown in table 1, all mean scores are relatively high with values varying between 3.84 (empathy) and 4.15 (assurance).

Furthermore, it can be seen that the personality trait extraversion has a negative significant (p<.01) correlation with the service factors reliability and empathy, from which can be indicated that high scores on neuroticism are related to low scores on reliability and empathy. The personality trait extraversion shows no significant correlations with the service factors.

All inter-correlations between the SERVQUAL variables and the variables satisfaction and delight are significant (p≤.01). All independent variables show a correlation score between .469 and .601 with the dependent variable satisfaction, and a correlation score between .410 and .632 with the dependent variable delight. According to the literature, a score above .3 is preferable (Pallant, 2013).

Also is investigated whether multicollinearity exists between the variables. Multicollinearity can influence the results of a multiple regression analysis in such a way that the predictive power of an independent variable on a dependent variable may be inaccurate, because the independent variable could actually be a combination of other independent variables (Pallant, 2013). To check for multicollinearity, the relationship between variables may not be higher than .7 (Pallant, 2013). Looking at table 1, some variables correlate too strong, namely the SERVQUAL variables assurance and empathy (.731), and the dependent variables satisfaction and delight (.763). This means that multicollinearity has occurred. To examine the consequences of this multicollinearity, a factor analysis is performed to find out if some items of the variables could better be removed or a different distribution of the items over the variables is needed.

Table 1: Scale means, SD's, Inter-correlations and Reliabilities

**.Correlation is significant at the 0.01 level (2- tailed) *. Correlation is significant at the 0.05 level (2-tailed)

4.2 Factor Analysis SERVQUAL variables

According to Pallant (2013), there are a few issues regarding the factorability of a data set. One is its sample size, which has been found to have a minimum size of N=150 (Tabachnick & Fidell, 2001). Furthermore, IBM SPSS offers two statistical measures to assess the factorability of a data set, namely Bartlett’s test of sphericity, which is significant at p<.05 (Bartlett, 1954; Tabachnick & Fidell, 2001), and Kaiser-Meyer-Olkin’s measure of sampling adequacy, which should score .6 or higher (Kaiser, 1970; Kaiser, 1974; Tabachnick & Fidell, 2001). For this study, two factor analysis are performed to examine the underlying structure of the highly correlating variables. One factor analysis is performed for the items of the SERVQUAL variables, and a second factor analysis is completed for the items of the dependent variables Satisfaction and Delight.

Looking at the items of the SERVQUAL variables, Bartlett’s test of Sphericity is significant (p<.05), and the Kaiser-Meyer-Olkin measure of sampling adequacy is .916. These outcomes, together with the sample size of 170, prove that the data set is suitable for a factor analysis. Reason for the factor analysis is the high correlation between the variables assurance and empathy. Because the design and distribution of the SERVQUAL items is built on findings of earlier studies that support a model which consists of five variables, it was decided to perform a principal component analysis whereby a fixed number of five factors was chosen (Parasuraman et al., 1985). Table 2 and 3 give an overview of the outcomes of the factor analysis. The outcome of the five-component solution explained a total of 67.25% of the variance, with the five components contributing respectively 40.44%, 10.35%, 7.89%, 4.51% and 4.06%. The screeplot showed a clear break after the fourth component. To find out whether or not items should be removed to higher the explained variance of the items, the communality outcomes are examined (table 3). According to Pallant (2013), scores above .3 are interpreted as high values which confirm that an item fits well with the other items in its component. It can be seen that all items score above .5, so no items have to be removed. Then, it was examined how the items were distributed over the components. Looking at the component correlation matrix (table 2), this shows both low and high correlations between the components varying between -.55 and .444. To aid in the interpretation of the five components, oblimin rotation was performed (table 3). The rotated solution revealed that the items of the variable tangibles load on component 2, the items of the variable reliability load on component 3, and the items of the variable responsiveness load on component 5. Component 4 shows a loading of a few scattered items. The items of the variables assurance and empathy both load on component 1, which can be related to the high correlation of .731 which was found in the correlation matrix

(table 1). The Structure matrix shows the correlation between the variables and factors, and is unique to the Pattern matrix (table 3).

Table 2: Component matrix of unrotated loadings SERVQUAL items

Table 3: Pattern and Structure Matrix for PCA with Oblimin Rotation of Five Factor Solution of SERVQUAL items

Regarding the outcomes of the factor analysis, it could be an option to unite the variables assurance and empathy. All 22 items of the SERVQUAL model contribute to the variance explained by the variables, so no item has to be deleted to improve the scale and increase the total variance explained. But it seems like assurance and empathy measure for a great extent the same underlying structure. Comparing these results to earlier studies, it has been found that the correlations between the SERVQUAL variables are in general relatively high. Parasuraman, Zeithaml and Berry (1988) found correlations between .52 and .84, and the study of Babakus and Boller (1992) found correlations that vary between .67 and .83. According to Buttle (1996), “SERVQUAL’s five dimensions are not universals; the number of dimensions comprising SQ is contextualized; items do not always load on to the factors which one would a priori expect; and there is a high degree of intercorrelation between the five RATER dimensions.” (p.10).

Research on previous studies about the SERVQUAL model confirm the reliability of the model (Ariffin & Maghzi, 2012; Parasuraman et al., 1985; S. L. Smith, 1994; Surprenant & Solomon, 1987), but there are also studies which confirm the high intercorrelations between the variables and a variety in the loading on factors (Asubonteng, McCleary, & Swan, 1996; Babakus & Boller, 1992; Buttle, 1996a). “Most studies imply greater overlap among the SERVQUAL dimensions – especially among responsiveness, assurance, and empathy – than implied in the original study” (Asubonteng et al., 1996, p.67). These variances in findings affect the discriminant’s validity of the model, and may be due to differences in data collection and analysis procedures. Furthermore, studies have also noticed a weak convergent validity of the model (Asubonteng et al., 1996; Peter, Churchill Jr, & Brown, 1993). “First, factor-analysis results relating to the convergent validity of the items representing each dimension are mixed because in several studies the highest loadings for some items were on different dimensions from those in Parasuraman et al. (1988). Second, lack of support for the discriminant validity of SERVQUAL is reflected by the factor-loading patterns, and the number of factors retained is inconsistent across studies.”(Asubonteng et al., 1996, p.75).

Although it has been found that all five variables contribute to the measurement of service factors, the outcome of the factor analysis of this study together with findings of Asubonteng et al (1996), Babakus & Boller (1992), Buttle (1996a) and Peter et al (1993) support the option to unite the variables assurance and empathy as one variable. Therefore, for the further statistical analyzes which will test the hypotheses of this study, the variables will be combined to one variable called “assurance&empathy”.

4.3 Factor analysis Satisfaction and Delight

The second principal components analysis for the highly correlated variables satisfaction and delight also starts with testing the suitability of the data. Bartlett’s test of Sphericity is again significant (p<.05), and the Kaiser-Meyer-Olkin’s measure of sampling adequacy was found to be .890. Together with a sample size of 170, it is proven that these variables are suitable for a factor analysis.

A principal component analysis is performed, to find out if the variables satisfaction and delight measure a same or different underlying construct, and whether or not the items are distributed in the right way over the two variables. First is examined how many components score an eigenvalue of 1 or more. The total variance explained table shows that the first component scores an eigenvalue of 6.12, followed by a second component which scores an eigenvalue of 1.11. The total variance explained by the components is 72,28%, with a contribution of 61.19% of component 1 and 11,10% of component 2. Additionally, the screeplot reveals a clear break after the second component, and the component correlation matrix shows a correlation of .45 between the two components. The outcomes of the communalities table show scores of .56 and higher, which means that no items have to be removed to improve the scale and increase the total variance explained. The Pattern and Component matrix are examined to find out how the items are divided over the two components. The items of satisfaction are all positive worded, and measure the level of satisfaction. The items of delight are both positive and negative worded, and measure the level of delight and its contrary factor outrage. As can be seen from the matrixes, all items of the variable satisfaction and the positive worded items of the variable delight load on component 1, whereas the negative worded items of delight load on component 2.

Table 5: Pattern and Structure Matrix for PCA with Oblimin Rotation of Two Factor Solution of Satisfaction and Delight items

Note: major loadings for each item are bolded

Investigating the outcomes of the factor analysis shows that the positive worded items of delight are measured as a superlative degree of satisfaction, and that the items related to the emotional state of outrage are measured as a separate factor. These results suggest that the items should be divided over two variables, but in a different way.

According to the literature, studies have found that satisfaction is mainly based on exceeding customer’s expectations, whereas customer delight requires that a customer receives a positive surprise which is beyond expectations (Berman, 2005). This study has applied the well-established measure scale of Westbrook & Oliver (1981) to examine satisfaction, with items which are based on experiences of customers. The self-constructed measure scale for the examination of delight is built according to the findings of the study of Berman (2005) and Dekker (2014). Berman (2005) states that “Delight as a construct differs significantly from customer satisfaction . While satisfaction is more cognitive, delight is more affective”. (p.148). Although the findings of Berman (2005) about the distinction between satisfaction and delight are confirmed by other researchers like Cadette and Turgeon (1988) and Schneider and Bowen (1999), delight is a relatively new finding in research on satisfaction, and has no well-established measure scale. “Unfortunately, there is no commonly accepted scale to measure customer delight. Some studies have determined the emotional responses associated with delight by asking subjects to suggest an appropriate name for an emotion/feeling produced by a mixture of emotions... Others used in-depth interviews and/or the critical incident technique where respondents were asked to describe absolutely, positively delightful experiences. These

were labeled by the researchers as delight”(Berman, 2005, p.134). The results of this factor analysis show that the positive worded items of delight seem to measure the same underlying construct as the items of satisfaction. Therefore, the items will be divided in a different way over two new variables, whereby the positive worded items will be linked to the variable satisfaction and the negative worded items will be linked to the variable outrage.

4.4 New research models

These changes have consequences for the research models of this study, because the variables included in the former models are changed. The following figures 8 and 9 show the new models with the new variables according to the outcomes of the factor analysis.

Figure 9: New model for hypotheses 4a, 4b and 4c

Because the hypotheses were formulated around the conceptual models which were developed before the factor analysis, the hypotheses will now be adapted to the variables of the new models. Table 6 gives an overview of the old versus new hypotheses.

Table 6: Overview of the changes in the hypotheses according to the outcomes of the factor analyzes

Hypotheses 1 is changed because the factor analysis showed that the items of assurance measure the same underlaying structure as the items of empathy. Therefore, it can be expected that instead of empathy, the new variable assurance&empathy will have the strongest effect on satisfaction. Hypotheses 2 and 3 did not change in the wording. For the development of hypotheses 4, it was assumed that delight was measured differently than satisfaction, but according to the factor analysis, it seems that delight is measured in line with satisfaction and outrage is the different measured factor. Therefore, delight is in the second model replaced by outrage, and therefore also changed in the hypotheses 4, 4a, 4b and 4c.

4.5 Results Hypotheses 1

H1: The service factors of the SERVQUAL model influence the level of Satisfaction, whereby it is expected that the service factor Assurance&Empathy has the strongest effect on Satisfaction.

To test hypotheses 1, a standard multiple regression analysis is performed, from which the results can be found in Table 7. A significance level of p <.05 is used to confirm or reject a hypothesis. Preliminary analyzes were conducted to ensure no violation of the assumptions of normality, linearity, multicollinearity and homoscedasticity. The control variables were entered at Step 1, and the model summary shows that they explain 2.2% of the variance in satisfaction. The total variance explained by the model as a whole was 55.5%. According to the literature, this is known as a ‘respectable result’ (Pallant, 2013).

To examine how each independent variable contributes to the prediction of the dependent variable, the Beta Standardised Coefficients are analysed. The variable assurance&empathy has the greatest significant unique contribution to explaining the dependent variable, with a score of .514. The second highest contribution comes from the variable tangibles, followed by the variable reliability. The variable responsiveness has no significant contribution to the prediction of the dependent variable, with p >.05.

Table 7: Standard Multiple Regression Analysis

4.6 Results Hypotheses 2

H2: Extraversion moderates the relationship between Service factors and Satisfaction, such that the relation is stronger when Extraversion is high.

In order to test the second hypotheses, the moderating effect of extraversion is tested by the PROCESS program in IBM SPSS. The outcomes of de moderation models of the service factors tangibles, reliability, responsiveness and assurance&empathy all show a significant contribution to the explanation of the variance of the dependent variable satisfaction, with scores of respectively 25,16%, 31.72%, 23,36% and 46,96%. Table 8 shows the results of the moderation effects of extraversion. The interaction effects between extraversion and each

service factor are all non-significant, with p-values of p >.05. This indicates that there is no significant moderating effect of extraversion on the relationship between each of the service factors and the dependent variable satisfaction.

Table 8: Moderation outcomes Extraversion on Service factors and Satisfaction

Although no moderation effect has been found, other statistical analyzes were performed to find out more about the influence of high and low scores of extraversion on the dependent variable satisfaction. The Visual Binning option from IBM SPSS was used to divide the participants of the study in high and low scores on extraversion. According to the literature, dividing a continuous variable into different categories is not always supported (Cohen, 1983; Irwin & McClelland, 2003; MacCallum, Zhang, Preacher, & Rucker, 2002). Studies have found that dichotomization of continuous variables has several negative consequences for the outcomes of analyzes, like loss of effect size and power, spurious statistical significance, loss of measurement reliability and a distortion of the true relationships about variables. Still, dichotomization is used in many studies, because it enables researchers to do an ANOVA or t-test and compare group means, and it makes it easier to communicate group differences. There are two most common options for the division of the continuous variable in equal groups. One is the median-split, whereby the group is divided in two equal groups by finding the median. The second option is based on identifying two cut-off points, which divide the continuous variable in three equal groups. The disadvantage of the latter is a reduction of N, because only the high and low scores will be used for further analysis, and the participants in the middle group are thrown away. Therefore, research suggest to apply the median-split. The higher number of N causes a higher statistical power of the results (Aiken & West, 1991). When