The effects of customer-perceived service, communication and brand usage encounter quality on relationship marketing outcomes:

The effects of customer-perceived service, communication

and brand usage encounter quality on

relationship marketing outcomes:

An empirical investigation in the German automobile industry

by

Insa Heim

University of Groningen

Newcastle University Business School

Dual Award Master of Science

Advanced International Business Management and Marketing

Master Thesis

The effects of customer-perceived service, communication and brand

usage encounter quality on relationship marketing outcomes:

An empirical investigation in the German automobile industry

Dual Award Master of Science

Advanced International Business Management and Marketing

Supervisors:

University of Groningen,

Faculty of Economics and Business: Drs. Ad Visscher Newcastle University, Business School: Dr. Nima Heirati Insa Heim

Student Numbers: S 253 790 7 B 140 173 806 Sven-Hedin-Str. 27

26389 Wilhelmshaven, Germany

Submission Date: 08 January 2015

A

The aims of this research are to test relationships between car brand customers’ experience encounter quality perceptions (service, communication, brand usage) and relationship outcomes (brand loyalty). In the conceptual and structural model, customer satisfaction and commitment as mediators of the relationships were the-orized to be influential. Customer experience is a multifaceted phenomenon and a meta-concept. Therefore this research puts forward the understanding of the sub-ject and the influence of its facets on relationship outcomes.

For the objective testing of relationships and hypothesis with primary data, a non-experimental survey research has been conducted. 214 customers in the German market answered a questionnaire. Subsequently, partial least squares structural equation modeling (PLS-SEM) method was used to test the model.

Key findings include that high service encounter, communication encounter and brand usage encounter quality positively influence brand loyalty by driving satis-faction. However, only brand usage encounter quality was found to lead to cus-tomer commitment, and subsequent brand loyalty. Finally, this research contrib-utes to existing studies in carrying forward empirical testing of experience con-structs. For practitioners these insights might help in designing strategic market-ing actions.

A

It's all about the experience.

T

C

List of Abbreviations ... VII!

1.! Introduction ... 1! 1.1! Introduction,...,1! 1.2! Background,...,1! 1.3! Research,Objective,and,Questions,...,3! 1.4! Research,Rationale,...,4! 1.5! Research,Design,...,5! 1.6! Structure,of,Thesis,...,6! 1.7! Conclusion,...,6!

2.! Literature Review and Conceptual Development ... 7!

L

A

AVA: Average Variance Accounted For AVE: Average Variance Extracted B2B: Business to Business

B2C: Business to Consumer

CBSEM: Covariance-Based Structural Equation Modeling CEM: Customer Experience Management

CR: Composite Reliability

CRM: Customer Relationship Management EM: Expectation Maximization

e.g.: exempli gratia (for example) et al.: et alii (and others)

H: Hypothesis

HOC: Higher-Order Construct ibid.: ibidem

i.e.: id est

LOC: Lower-Order Construct

MCAR: Missing Completely At Random

PLS-SEM: Partial Least Squares Structural Equation Modeling SD: Standard Deviation

SEM: Structural Equation Modeling

SPSS: Statistical Package for the Social Sciences VIF: Variance of Inflation Factor

1. I

1.1 Introduction

The concept of customer experiences has received increasingly attention from scholars and practitioners, because it is regarded as means to foster relationships towards customers and potentially builds customer loyalty (Addis & Holbrook, 2001; Fournier, 1998; Holbrook & Hirschman, 1982; Lemke, Clark, & Wilson, 2011; Verhoef et al., 2009). Customers’ loyalty towards a brand ensures sustaina-ble business success in a highly competitive marketplace. Therefore, the purpose of this research is to understand the potential of experience aspects to positively influence relationship outcomes, specifically brand loyalty, in an automobile con-text. This introductory chapter (Chapter 1) first presents the background of the study. Afterwards, the research objective and questions as well as the research rationale are outlined. The study’s research design is presented before the struc-ture of this research is exposed.

1.2 Background

companies face high competitive pressures (Bügel, Buunk, & Verhoef, 2010; Seidel, Loch, & Chahil, 2005) and cannot rely on customer retention based on contractual relationships. Thus switching costs are relatively low (Jones & Sasser, 1995) and long purchasing cycles procreate a challenging phase for customer re-tention initiatives (Pan, Sheng, & Xie, 2012). Considering these aspects, under-standing mechanisms that help automobile companies achieve to create brand loyal customers, thereby reaping related economic benefits (e.g. through repur-chases, cross-buying, and word-of-mouth (WOM)) appears critical.

It is a common notion that for business success the creation of customer satisfac-tion and loyalty are crucial (Dick & Basu, 1994; Kumar & Shah, 2004; Oliver, 1999) and the latter “a company’s most enduring asset” (Pan, Sheng, & Xie, 2012, p.150). As means to create loyal customers, researchers and practitioners devote attention to customer orientation (Vargo & Lusch, 2004; Woodruff, 1997), cus-tomer relationship management (CRM) (Buttle, 2009; Chen & Popovich, 2003; Richards & Jones, 2008; Sheth & Parvatiyar, 1995), and the design of superior customer experiences (Devaraj, Matta, & Conlon, 2009; Verhoef et al., 2009). Thereby it becomes apparent that with the development of the service-dominant logic (Grönroos, 1984; Vargo & Lusch, 2004) and its application to the goods sector (Godlevskaja, Iwaarden, & Wiele, 2011), differentiation augmented to cus-tomer servicing, interaction, the building of relationships and the perspective that customer-perceived value is created by experiences (Prahalad & Ramaswamy, 2004). Designing several unique contact points between firm, brand and consumer in each phase of the customer lifecycle gains in importance to create experiences that foster brand loyalty (Hättich, 2009). This comprises managing the customers’ total experience including a company’s service, products, brand and its ambassa-dors and was identified as driver of relational outcomes that bear competitive ad-vantages (Brakus, Schmitt, & Zarantonello, 2009; Day, 2003; Lemke et al., 2011; Löffler, 2013; Mascarenhas, Kesavan, & Bernacchi, 2006; Meyer & Schwager, 2007; Schmitt & Zarantonello, 2013).

turer Porsche aims at increasing customer delight by providing and designing su-perior experiences and customer contact points pre, during and post purchase of the car (Löffler, 2013). Thereby it follows the objective to retain customers by augmenting the product (ibid.). To conclude, the creation of experiences and de-sign of the ‘customer journey’ constitute a means for customer retention. In the automobile industry, experiences can be deemed especially important after the purchase of the car. Then they help to build and enhance relationships with exist-ing customers and drive brand loyalty.

1.3 Research Objective and Questions

The purpose of this research is to investigate German car brand customers’ expe-rience encounter quality perceptions. Furthermore encounter quality constructs’ potential to drive relational outcomes will be examined. Thus the research objec-tive reads as follows:

The main research objective is to understand the extent that an automobile firm is able to drive brand loyalty by offering customer experience

encoun-ters of superior quality.

To address the aforementioned research objective, this study aims to answer spe-cific research questions that are outlined in Table 1.1.

Table 1.1 Research questions

1. To what extent do an automobile firm’s customer experience encounters affect brand loyalty through driving the level of customer satisfaction, and customer commitment?

1a) To what extent does a firm’s service encounter quality affect brand loyalty through

driving the level of customer satisfaction, and customer commitment?

1b) To what extent does a firm’s communication encounter quality affect brand loyalty

through driving the level of customer satisfaction, and customer commitment?

1c) To what extent does brand usage encounter quality affect brand loyalty through

2. To what extent does customer satisfaction affect customer commitment and brand loyalty?

3. _{To what extent does customer commitment affect brand loyalty?}

1.4 Research Rationale

Many researchers and practitioners assert that CRM (Buttle, 2009; Chen & Popovich, 2003; Richards & Jones, 2008; Sheth & Parvatiyar, 1995), and the de-sign of superior customer experiences (Berry & Carbone, 2007; Devaraj et al., 2009; Verhoef et al., 2009) help in achieving customer retention. Customer expe-riences have attracted attention of researchers more than 30 years ago (Holbrook & Hirschman, 1982). The relationship marketing paradigm which transcends pure economic exchange can be traced back to the last century (Morgan & Hunt, 1994; Palmer, 2002; Sheth & Parvatiyar, 1995). However, there still remain “underde-veloped” (Schmitt & Zarantonello, 2013, p.26) research areas: Rather limited is the investigation of the voice of the customer and the perceptions of companies’ relational actions (Verhoef et al., 2009). Research has not been advanced eminent-ly, thus suggests that firms need to ask their analytical departments to identify antecedents of customer loyalty (Verhoef & Lemon, 2013). However, companies need to “take the voice of the customer fully into account, since customer input is critical to ensuring richer experiences” (Payne et al., 2009, p. 383). Furthermore, previously the direction of customer perceptions has been questioned and dark-sides of relational marketing are outlined (O’Malley & Prothero, 2004). Thus a deeper understanding of the customer’s view is needed. Therefore, linking experi-ential clues with relational outcomes serves as valuable frame for the investiga-tion.

litera-ture (Addis & Holbrook, 2001; Lemke et al., 2011), yet have not been investigat-ed with quantitative approaches and empirical testing to a great extent (Klaus & Maklan, 2012). In the automobile industry, quantitative studies with the focus on brand loyalty investigate loyalty drivers such as product and dealer satisfaction (e.g. Mittal & Kamakura, 2001; Verhoef, Langerak, & Donkers, 2007). For expe-rience marketing insights, qualitative methods were applied and resulted in con-ceptual foundations (Lemke et al., 2011). With similar results, case studies of au-tomobile brands sketch firm created customer experiences (Löffler, 2013; Payne et al., 2009). Conclusively, as a valuable research contributes something new to existing studies (Whetten, 1989, p. 494), the rationale of this research lies in con-ceptually linking and empirically testing the influence of customers’ experience encounter quality perceptions on relational outcomes in an automobile setting. Outcomes broaden the paradigmatic perspective of the experience-marketing phe-nomenon and are useful for understanding customer decision-making. Practically, insights are important for the design of automobile companies’ relationship mar-keting strategies.

1.5 Research Design

Modeling (PLS-SEM) was deemed the appropriate technique to assess the meas-urement as well as structural model (Hair, Sarstedt, Ringle, & Mena, 2012; Henseler, Ringle, & Sinkovics, 2009; O’Cass, Heirati, & Ngo, 2014; Wold, 1980).

1.6 Structure of Thesis

After Chapter 1, a review of current literature and theories related to the con-structs relevant for this study will be provided in Chapter 2. It justifies the re-search frame chosen. Basing on this framework, relationships between the con-structs will be specified and hypotheses developed. In Chapter 3, research design and methodology are stated. Afterwards, Chapter 4 presents the results of the em-pirical investigation. Finally, Chapter 5 provides the discussion of the findings, theoretical contributions, managerial implications, limitations of the study, and suggestions for future research.

1.7 Conclusion

2. L

R

C

D

2.1 Introduction

The present chapter examines concepts in existing literature as well as theoretical aspects relevant to the research aims and questions introduced beforehand. First, Section 2.2 critically reviews the existing work of the relationship and experience marketing domain. Subsequently, the experience encounters investigated in this research are defined. Then, relationship outcomes that build the endogenous con-structs of the conceptual model are outlined (Section 2.3). Section 2.4 specifies relationships among the theoretical constructs under investigation. Finally, the theoretically derived conceptual model that shows the relationships among con-structs of interest and addresses the research questions outlined in Chapter 1 is presented graphically.

2.2 Relationship Marketing and Customer Experience Encounters

but that a competitive and successful company needs to compete on the augment-ed product level (Grönroos, 2009; Payne & Holt, 2001). Customer experiences are a means to augment the product and deliver value that is not only utilitarian (Holbrook & Hirschman, 1982), and thus represent a constituting constructs of relationship marketing (Agariya & Singh, 2011).

conceptual-ized in this study as a higher level of abstraction. Because “not all customer en-counters contribute equally to the overall assessment of experience” (Buttle, 2009, p. 172), this study directly compares the encounters’ individual links towards rela-tionship outcome constructs.

Table 2.1 Definitions and conceptualizations of customer experience Source Definitions and Concepts

Chen & Chen, 2010, p. 30

Experience quality is subjective in terms of measurement while service quality is objective. The evaluation of experience quality tends to be holistic/ gestalt rather than attribute-based, and the focus of evaluation is on self (internal) but not on service environment (external). In addition, the scope of experience is more general than specific, the nature of benefit is experiential/hedonic/symbolic rather than functional/utilitarian, and the psychological representa-tion is affective instead of cognitive/attitudinal.

Gentile et al., 2007, p.397 The Customer Experience originates from a set of interactions be-tween a customer and a product, a company, or part of its organiza-tion, which provoke a reaction. This experience is strictly personal and implies the customer’s involvement at different levels (rational, emotional, sensorial physical and spiritual). Its evaluation depends on the comparison between a customer’s expectations and the stim-uli coming from the interaction with the company and its offering in correspondence of the different moments of contact or touch-points.

LaSalle & Britton, 2003, p. 30

An interaction, or series of interactions, between a customer and a product, a company or its representative that lead to a reaction. Lemke et al., 2011, p. 849

We define customer experience as the customer’s subjective re-sponse to the holistic direct and indirect encounter with the firm, including but not necessarily limited to the communication encoun-ter, the service encounter and the consumption encounter.

Meyer and Schwager

2007, p. 118 Customer Experience is the internal and subjective response cus-tomers have to any direct or indirect contact with a company. Direct contact generally occurs in the course of purchase, use, and service and is usually initiated by the customer. Indirect contact most often involves unplanned encounters with representatives of a company’s products, service or brands and takes the form of word-of-mouth recommendations or criticisms, advertising, news reports, reviews and so forth.

those elements which the retailer can control (e.g., service interface, retail atmosphere, assortment, price), but also by elements that are outside of the retailer’s control (e.g., influence of others, purpose of shopping). Additionally, we submit that the customer experience encompasses the total experience, including the search, purchase, consumption, and after-sale phases of the experience, and may involve multiple retail channels.

2.2.1 Service Encounter

sophis-ticated manner. Furthermore, this perspective reflects the findings of Lemke et al. (2011) who outline “Caring – Procedures and Processes” (p.854) as one category of experience quality and driver of relationship outcomes. Facilitating service depicts a firm’s level three relationship marketing aspect and fosters a “structural” (Berry, 1995, p. 241) bond towards the brand.

In general, the service encounter is a relevant aspect because it augments a pure market exchange to a social interaction (Czepiel, 1990). Going beyond this nar-row conceptualization, product price is an important marketing and service aspect (Herrmann et al., 2007; Mugge, Schifferstein, & Schoormans, 2010; Zeithaml, 1988). A firm can directly influence the price which is an extrinsic, product-related cue to quality (Zeithaml, 1988). Price also manifests itself in customers’ product quality expectations (Herrmann et al., 2007). Post purchase product price equity can be positioned within the theory on distributive justice, which argues that in relation to their investments, customers expect to get what they deserve (Oliver & Swan, 1989). Furthermore, for durable goods customer’s attention paid to prices is greater than for non-durables (Zeithaml, 1988). As a consequence, price negotiations including service person interactions are common when pur-chasing an automobile (Herrmann et al., 2007). For these reasons, customer-perceived product price equity reflects the service encounter quality construct in terms of fairness and customer’s evaluation of a car’s value for money.

2.2.2 Communication Encounter

research the communication encounter is reflected in formal, and personal com-munication as well as preferential treatment.

Formal communication is a form of direct marketing and represents the informa-tive efforts, the car brand undertakes towards its customers including the amount, frequency, diverse channels of information delivery (Keller, 2009; Palmatier et al., 2006). In this thesis the term, personal communication will be used to refer to a more personalized mode of marketing and relational communication. It includes listening to and learning about the customer what resembles a more dialogical way of communicating and results in customized messages or “incentives” (Christy, Oliver, & Penn, 1996, p. 182ff) such as personalized gifts.

Moreover, the communication encounter reflects the perception of customers about the extent to which regular and established customers are offered special treatment compared to non-regular customers (e.g. relational benefits based on economic and customization advantages) (Chen & Hu, 2010; Kristof De Wulf, Odekerken-Schröder, & Iacobucci, 2001; Gwinner, Gremler, & Bitner, 1998). Therefore, the delineation preferential treatment is used in this research. This as-pect addresses the incentive to remain with a provider because of the more cogni-tive expected benefits and additional customer-perceived value that arise from the long-lasting relationship. Furthermore, from a firm perspective it represents one main aspect of customer relationship management which manifests itself in cus-tomer selectivity and targeting (Reutterer, Mild, Natter, & Taudes, 2006).

2.2.3 Brand Usage Encounter

draws upon customers’ consumption experience and feelings arising from con-sumption (Holbrook & Hirschman, 1982).

Veloutsou (2009) concludes by declaring that there are two ways in which brands “contribute to the relationships that consumers experience in their daily life” (p.128). These are brand-consumer and consumer-consumer relationships. Firstly, because brands are used as status symbol, their “social impact” (Lemke et al., 2011, p. 856) as construct of the brand usage encounter and the deeply held con-nection between a brand and a consumer’s life theme (Fournier & Yao, 1997) becomes apparent. Value is derived from the possibility of self-expression, which goes back to the concept of the extended self (Belk, 1988). This theory suggests that the customer regards its possession, the product that is labelled with a brand, as part of his/her identity (Belk, 1988; Dimitriadis & Papista, 2010). Also for cars, which have “become a metaphor of self-expression and a place for socializing” (Seidel et al., 2005, p. 440) the concept of the extended self is recognized. In addi-tion, socializing and other user connection will refer to connecting to other users of the brand, building a brand community and personal relationships with other customers (Lemke et al., 2011). Because the feeling of community has the poten-tial to foster the emotional bond towards the brand (Palmer, 2010; Tsai, 2005), it can be deemed influential for the establishment of relationship outcomes, such as brand loyalty.

2.3 Relationship Outcomes

2.3.1 Customer Satisfaction

2.3.2 Customer Commitment

Compared to the construct satisfaction, customer commitment shows the attach-ment towards and ‘stickiness’ with which a customer is bound towards a brand and wants to maintain the relationship (Gupta & Zeithaml, 2006; Gustafsson, Johnson, & Roos, 2005; Morgan & Hunt, 1994). Customer commitment has been found as driver of customer retention (Adjei & Clark, 2010; Anders Gustafsson et al., 2005; Verhoef, 2003) and customer loyalty (Dagger et al., 2011; Hennig-Thurau et al., 2002; Odekerken-Schröder et al., 2003; Oh, Lee, & Kim, 2011; Palmatier et al., 2006). Thereby its notion as attitude and distinctness from loyalty becomes apparent (Bettencourt, 1997). Commitment can be differentiated into sub constructs (Fullerton, 2011; Anders Gustafsson et al., 2005; Oh et al., 2011). However, particularly these sub-components show “inconsistent conceptualiza-tions” (Gupta & Zeithaml, 2006, p. 5). With due regard to the reason for the ‘stickiness’, components can be defined and distinguished subsequently: Calcula-tive commitment includes cogniCalcula-tive switching barriers such as costs to change the brand and represents the “dark side of marketing relationships” (Fullerton, 2011, p. 95). It refers to customers staying with the brand for rational considerations and being locked-in (Zineldin, 2006). Affective commitment can be viewed as emo-tional attachment and a person’s “root psychological state” (Fullerton, 2011, p. 93) towards a brand (Moliner, Sánchez, Rodríguez, & Callarisa, 2007). Normative commitment represents that customers feel morally obliged to maintain the rela-tionship and is more prevalent in B2B contexts (Oh et al., 2011). Therefore in this research, customer commitment encompasses emotional and cognitive attachment towards the car brand.

2.3.3 Brand Loyalty

pay-offs for the company. Researchers conceptualize several ways in which customers demonstrate brand loyalty. For instance repeat purchases (Hellier, Geursen, Carr, & Rickard, 2003; Zeithaml et al., 1996) and positive word of mouth (Gounaris & Stathakopoulos, 2004; Swan & Oliver, 1989) are considered. Furthermore, in the literature a differentiation between two approaches, behavioural and attitudinal loyalty, is suggested (Dick & Basu, 1994). While behavioural consequences of loyalty include actual repurchases and repeat patronage behaviours, attitudinal loyalty refers to positive behavioural intentions towards a brand (Kuenzel & Halliday, 2008; Kumar et al., 2013)). The latter is taken into consideration for this investigation due to a lack of objective data and because intentions approximate actual behaviour (Fishbein & Ajzen, 1975).

In order to account for brand loyalty’s multidimensionality, besides repurchase intentions, recommendation intentions are contemplated. First repurchase inten-tions are defined. If the customer as decision-making unit has a repurchase pattern that is systematically biased to purchase one specific brand out of a brand set, he/she can be regarded as brand loyal (Kuenzel & Halliday, 2008). Regarding the automobile context, a brand loyal customer has the intention to buy the same car brand again. Secondly, recommendation intentions can be viewed as propensity to spread positive WOM about the car brand (Lobschat, Zinnbauer, Pallas, & Joachimsthaler, 2013). Thereby customers “tend to be supportive and make posi-tive recommendations about the brand” (Kuenzel & Halliday, 2008, p. 295). Rec-ommendations translate indirectly into economic benefits for the firm (e.g. acqui-sition of new customers due to trustworthy sources of information, multiplication of marketing message) (Swan & Oliver, 1989). Both, repurchase and recommen-dation intentions share that customers have a systematic tendency of behaving favourably towards a brand, thus are defined to be brand loyal.

2.4 Development of Hypotheses

Companies try to utilize diverse encounters in order to build favourable customer relationships and deliver customer value after a product has been sold. In turn, customers’ perceptions of these encounters will directly influence customer’s (1) affective states as well as (2) cognitive states that lead customers to build and maintain their relationship with the brand. Both, satisfaction and commitment constructs then affect (3) behavioural brand loyalty intentions and act as media-tors of customer experience attributes and brand loyalty. Even though direct posi-tive correlations between experience encounter quality and brand loyalty can be expected (Duncan & Moriarty, 1998; Kim, Han, & Park, 2001; Veloutsou, 2009; Zeithaml et al., 1996), a mediating impact of satisfaction and commitment on brand loyalty is justifiable. Theoretical underpinnings of the cognition-affect-behaviour hierarchy (Eagly & Chaiken, 1993) and Oliver’s (1999) framework that includes stages of loyalty development support this conceptualization. Further-more, qualitatively developed means-end frameworks (Klaus & Maklan, 2012; Parasuraman, Zeithaml, & Berry, 1988) and empirical studies encourage that cus-tomers’ quality perceptions are mediated by customers affective states and at-tachment towards the brand (Olsen, 2002; Tsai, 2011).

2.4.1 Effects of Service Encounter Quality Perceptions

a greater extent than product quality, price, reputation and personal relationship quality (Rapp, 1995). Secondly, in a car insurance and personal superannuation service context, perceived justice and fairness in a customer’s complaint handling process and in general the perceived process quality of the service were found to drive customer satisfaction and indirectly repurchase intentions (Hellier et al., 2003). Thus,

H1: Customer-perceived service encounter quality is positively related to (a) customer satisfaction and (b) customer commitment.

2.4.2 Effects of Communication Encounter Quality Perceptions

cus-tomers may regard personalization to be of higher quality than pure information processing and thus derive a greater emotional value from it, personal communi-cation is assumed to especially influence affective commitment components and relationship satisfaction positively.

In a previous study, the relationship aspects of preferential treatment and reward-ing have been found to drive relationship commitment and indirectly buyreward-ing be-haviour (Odekerken-Schröder, et al. 2003). This is also in line with the argumen-tation of Buttle (2009) who proposes that customers in a business to customer (B2C) context may value benefits of a relationship such as recognition (Buttle, 2009, pp. 41f.). Therefore, it is assumed that regularly informing customers about e.g. new products or offers, communicating in a personal way and delivering spe-cial treatment benefits will have a positive effect on satisfaction and espespe-cially commitment that further influences brand loyalty. Thus,

H2: Customer-perceived communication encounter quality is positively related to (a) customer satisfaction and (b) customer commitment.

2.4.3 Effects of Brand Usage Encounter Quality Perceptions

Using the brand, either for self-expression purposes or in order to connect with other users, underlies cognitive customer considerations. Therefore, according to the cognition-affect-behaviour hierarchy (Eagly & Chaiken, 1993), brand usage encounter quality is supposed to drive customer satisfaction and commitment, thereby indirectly leading to brand loyalty. As part of brand attachment, self-expression is conceptually linked to relationship quality constructs (Dimitriadis & Papista, 2010). Empirically, self-expression and identification with the brand are found to be a antecedents of brand loyalty (Kim, Han, & Park, 2001; Kuenzel & Halliday, 2010; Park, Macinnis, Priester, Eisingerich, & Iacobucci, 2010).

from the utilization of the brand during the consumption stage which fosters rela-tionships towards the brand. Thus,

H3: Customer-perceived brand usage encounter quality is positively related to (a) customer satisfaction and (b) customer commitment.

2.5 Effects of Intermediate Relationship Outcomes

This section presents the theorized effects of intermediate relationship outcomes (customer satisfaction and customer commitment). Customer satisfaction and commitment are proposed to mediate the link between (1) customer-perceived encounter quality and (2) behavioural intentions as ultimate relationship outcome.

2.5.1 Effects of Customer Satisfaction

Image 1 Satisfaction-Loyalty Relationship (Jones & Sasser, 1995, p.8)

However, it is theorized that high contentment with the automobile and with the relationship towards the car brand lead to repeat purchase and recommendation intentions. In support of this reasoning, numerous studies revealed that the effect of satisfaction on loyalty is positive in general (Fornell, Johnson, Anderson, Cha, & Bryant, 1996; Kumar et al., 2013; Szymanski & Henard, 2001). In an automo-bile context Gustafsson and Johnson (2002) show that satisfaction influences loyalty positively. Moreover, automobile satisfaction is found to foster customer’s intention to recommend the manufacturer (Mittal, Kumar, & Tsiros, 1999). Spe-cifically the satisfaction-loyalty linkage is greater for products that are bought irregularly and have long purchase cycles (such as cars) than for regularly pur-chased goods (Pan et al., 2012). In addition, customer satisfaction is found to in-fluence customer commitment positively (Bügel et al., 2010; Fullerton, 2011; Hennig-Thurau et al., 2002). Consequently, it is also posited that customer satis-faction increases emotional as well as cognitive attachment towards the car brand and drives customer commitment. Thus,

2.5.2 Effects of Commitment

Researchers find that affective commitment towards a brand in the categories, cars, sneakers and laptops influences brand loyalty positively (Iglesias, Singh, & Batista-Foguet, 2011). In a retail context, Fullerton (2005) identifies differences among calculative and affective commitment as drivers of loyalty. Thus, whereas affective commitment resulted to influence loyalty positively, calculative com-mitment showed negative effects (Fullerton, 2005). In the insurance service set-ting Verhoef, Franses, and Hoekstra (2002) find only affective commitment to have significant effects on payment equity and the number of services purchased. Therefore, the emotional component seems to drive loyalty to a greater extent than calculative commitment. Because in the automobile industry, the quality of alternatives is low (Bügel et al., 2010), thus customers rather remain brand loyal. Therefore also calculative commitment is hypothesized to influence brand loyalty positively. Thus,

H5: Customer commitment is positively related to brand loyalty.

2.6 Conceptual Model

Subsequently, in Figure 2.1 the conceptual model with the postulated hypotheses among constructs is presented.

H4a

SERVICE

ENCOUNTER

Facilitating Services Product Price Equity

COMMUNICATION ENCOUNTER BRAND USAGE ENCOUNTER Customer Satisfaction Relationship Satisfaction Automobile Satisfaction Customer Commitment Brand Loyalty Repurchase Intention Recommendation Intention H1a H1b H2b H2a H3a H3b H4b H5 Figure 2.2 Conceptual model

2.7 Conclusion

3. R

D

Chapter 2 discussed theories and literature relating to the research’s background. Subsequently, hypotheses substantiating linkages among the theoretical constructs were developed. In this chapter the appropriate research design, which resembles the strategy of inquiry followed throughout the research, will be presented. It helps to address the research questions and test the hypothesized relationships in the conceptual model. After depicting the underlying research paradigm and relat-edly the logic of the research process, details of construct’s operationalization and empirical study will be elaborated on.

3.2 Part I: Methodology - Quantitative Survey Research

the research process of this study is outlined in Figure 3.1, showing its objective and post-positive base, which led to the choice of a quantitative research para-digm.

Figure 3.1 Research process

3.3 Part II: Methods

This section comprises the methods employed in the research process. Primary data collection rests upon secondary data collection. Thus first, latent constructs derived from the review of theories and findings from the literature were specified (Chapter 2). Details regarding constructs’ operationalization, a description of the data collection instrument, including sampling and ethical consideration, and spe-cifics of data analyses follow.

3.3.1 Construct and Measure Specification

In the empirical analysis, measurement instruments1 help to measure the theoreti-cal latent constructs specified in the research questions and conceptual model (Bagozzi & Yi, 2012; Jarvis, MacKenzie, & Podsakoff, 2003). The development of measurement items and scales can be regarded as own research area making use of e.g. qualitative focus group investigations (Bearden & Netemeyer, 1999; for reflective factors see Churchill, 1979). Consequently, indicators for measuring the constructs were developed by drawing on prior studies where they have prov-en to be reliable and valid (Table 3.1).

Furthermore, measurements and their related constructs need to be specified in their epistemic nature of relationship. Thereby models can be differentiated into formative and reflective while the “specification should primarily be based on theoretical considerations regarding the causal priority between the indicators and the latent variable involved” (Diamantopoulos & Winklhofer, 2001, p. 274). While formative indicators cause the latent variable, reflective measures are caused by the unobservable construct (Diamantopoulos & Winklhofer, 2001; Jarvis et al., 2003). In most of the underlying studies, measures were indicated as reflective (Fullerton, 2011; Klaus & Maklan, 2012; Schouten et al., 2007). In oth-er cases, a reflective nature could be concluded from the description (De Wulf & Odekerken-Schröder, 2003; Kim et al., 2001; Odekerken-Schröder et al., 2003; Verhoef et al., 2007). Model choice has to be made accurately because otherwise,

it would be impossible to draw sound conclusions about the structural relation-ships among the constructs and model misspecification was found to lead to in-verse results (Collier & Bienstock, 2009). Moreover, subsequent analyses tech-niques are determined by the measurements’ nature (Hair et al., 2012).

In general, it has been cautioned to use single-item measures because measure-ment mono-operationalization bias may occur and model quality can be dimin-ished (Anderson & Gerbing, 1988; Baumgartner & Homburg, 1996; Hair, Sarstedt, Ringle, & Mena, 2012). Thus in this research each latent first-order con-struct will be measured indirectly with multiple items, thereby ensuring that the construct is encompassed accurately. In order to stipulate the first-order constructs included in this research, definitions from Section 2.2. build the foundations of “the logical deducibility of observations” (Bagozzi & Yi, 2012, p. 13). This also delivers the basis for the measurement model specification (Bearden & Netemeyer, 1999) which, besides the definition of epistemic nature of measure-ment, includes the development of consistent measurement items.

Table 3.1 Measurement model specification Constructs (HOC and LOC) Items Service Encounter Quality

Facilitating Services

Klaus and Maklan (2012) ⋅

The whole process of buying a car was so easy, the dealer of the car brand took care of everything. ⋅ It is not just about the now; the dealer of the car brand will look after me for a long time.

⋅ I am already a customer; the dealer of the car brand knows me and takes good care of me, so why should I go somewhere else?

⋅ It was important that the car brand was flexible in dealing with me and looking out for my needs.

⋅ It is important that the people I am dealing with are good people; they listen, are polite and make me feel comfortable. ⋅ The way the car brand deal(t) with me when things go(went) wrong will decide if I stay with them.

Product Price Equity

Verhoef et al. (2007) ⋅ _⋅ My car was reasonably priced. The quality/price ratio of my car is good. ⋅ My car gives me my money’s worth.

Communication Encounter Quality

Formal Communication

Odekerken-Schröder et al. (2003) ⋅ _⋅ My car brand often sends letters to customers like me. My car brand keeps customers informed through e-mails.

⋅ My car brand often informs customers through brochures/ magazines.

Personal Communication

Crosby et al. (1990) ⋅

I was contacted by my car brand, which wanted to stay "in touch" and make sure I was still satisfied. ⋅ I was contacted by my car brand, which wanted to keep abreast of changes in my family, and car needs. ⋅ I received something of a personal nature from my car brand (e.g., birthday card, holiday gift, etc.). ⋅ I was contacted by my car brand, which wanted to describe benefits of new car models.

Preferential Treatment

De Wulf and Odekerken-Schröder (2003)

⋅ This car brand treats regular customers differently than non-regular customers.

Brand Usage Encounter Quality

Self-Connection

Kim et al. (2001) ⋅

The car brand helps me to express myself. ⋅ The car brand reflects my personality. ⋅ The car brand enhances myself.

Other User-Connection

Schouten et al. (2007) ⋅

I have met wonderful people because of my car brand. ⋅ I feel a sense of kinship with other [car brand] owners. ⋅ I have an interest in a club for [car brand] owners.

Customer Satisfaction

Relationship Satisfaction

Odekerken-Schröder et al. (2003) ⋅

As a customer, I have a high quality relationship with my car brand.

⋅ I am happy with the efforts my car brand is making towards customers like me. ⋅ I am satisfied with the relationship I have with my car brand.

Automobile Satisfaction

Kuenzel,and Halliday (2008); Oliver (1980)

⋅ I am satisfied with my car.

⋅ Owning this car has been a good experience. ⋅ I am sure it was the right thing to buy this car.

Customer Commitment

Affective Commitment

Fullerton (2011) ⋅

I feel emotionally attached to my car brand.

⋅ My car brand has a great deal of personal meaning for me. ⋅ I feel a strong sense of identification with my car brand.

Calculative Commitment

Fullerton (2011) ⋅

It would be very hard for me to switch away from my car brand right now even if I wanted to. ⋅ My life would be disrupted if I switched away from my car brand.

⋅ It would be too costly for me to switch from my car brand right now.

Brand Loyalty _⋅

Recommendation Intention

Kuenzel and Halliday (2008) ⋅ _⋅ I would recommend my car brand to friends and relatives. I will speak positively about my car brand. ⋅ I intend to encourage other people to buy my car brand.

Repurchase Intention

3.3.2 Primary Data Collection Instrument

The design of the questionnaire for the quantitative survey research design was the next step in the research process. Due to the investigation of the German mar-ket, measures needed to be translated. In order to meet linguistic equivalence (Malhotra, Agarwal, & Peterson, 1996) for all items, and to ensure adequate wording two translation and back-translation procedures (Brislin, 1986; Malhotra et al., 1996) of measures were conducted with bilinguals. Subsequently, wording adaptions were made. Afterwards, the questionnaire has been extensively pre-tested (October 2014) with N = 18 respondents in order to assess whether the con-structs are measured satisfactorily (Malhotra & Peterson, 2001) and to prevent item non-response (Hair, Black, Babin, & Anderson, 2014). Besides quantitative testing, qualitative insights were gained from respondents in order to reveal any misunderstandings and difficulties in item wording, complications with question-naire structure and length, and to assess ethical adequacy. Finally, statements that did not meet reliability and validity scores were omitted, and questions that were difficult to understand have been modified. For instance, a ‘do not know’ answer option was added to the items of preferential treatment. Changes and improve-ments were incorporated into the instrument at hand, which can be found in the appendix (Appendix 1). The survey was distributed with the online survey tool ‘Qualtrics’. Major advantages of the online questionnaire procedure include low cost and high convenience for respondents because answering the questionnaire is time- and place-independent (Malhotra & Peterson, 2001).

3.3.3 Sampling Methods

Paralleling Gale and Wood's (1994) suggestions, this study’s target population includes German car brand customers. Initially, the sample was developed through the personal network of the researcher and respondents were targeted through email and social media channels. Afterwards, snowballing followed from the basis sample. Furthermore, the survey’s hyperlink has been published publicly online on an automobile Internet-forum. Thus, self-selected participants were re-cruited to take part in the study by following the questionnaire’s hyperlink. The main advantage of convenience sampling techniques lies in the fact that re-spondents were likely to have desired characteristics (e.g. car owners) (Bryman & Bell, 2011) which accommodates construct validity (Peter & Churchill, 1986). The qualification of respondents was further ensured with a screening and forced-response question, which asked to indicate the car brand owned. Moreover to be able to access possible participants quickly and easily are distinct advantages of convenience and self-selection sampling. Finally, it could be ensured that suffi-cient observations were undertaken (N = 295) that might at least approximate a probabilistic base. This number exceeded the initially set aim of 200 respondents. For PLS-SEM (refer to Section 4. 3) the rule of thumb for sample size reads as follows “ten times the largest number of structural paths directed at a particular latent construct” (Hair, Ringle, & Sarstedt, 2011, p. 144). Therefore for this study a sample site of N = 40 would have been sufficient. However, the actual sample size exceeds the rule for minimal observations and contributes to approximating generalizability of results.

3.3.4 Ethical Considerations

touch in case of further objections or questions. Furthermore, all participants were informed about data confidentiality and the anonymity of their identities (Malhotra & Peterson, 2001). In general, sensitive questions which may cause stress to the respondents (Malhotra & Peterson, 2001) have not been part of the study, only income, age and gender might be subsumed under this heading. There-fore, respondents did have the right to refuse to answer.

3.3.5 Data Analysis

Table 3.2 Data analysis process

Preliminary Data Assessment and Cleaning

Preliminary Data Assessment and Cleaning needs to be performed in order to identify patterns of missing values (Hair et al., 2014). Randomness of missing values has to be checked case- and item-wise. Subsequently, adequate handling of missing data in order to prevent bias has to be performed. As result from this data analysis stage follows the sample that has to be further studied.

Descriptive Data Analysis

Descriptive Data Analysis comprises the assessment of sample properties and character-istics of cases (e.g. gender, age, income, car-related aspects). Furthermore, it includes how the cases under study compare to the characteristics of the population.

Furthermore the measurement items’ properties are outlined and their distributions ana-lysed. Metrics considered include mean, standard deviation (SD), skewness and kurtosis to examine the departures from a normal distribution.

Multivariate Data Analysis:

Structural Equation Modeling (SEM) Technique

To select the appropriate multivariate analysis technique, variables under investigation have been taken into account. The research model at hand includes more than one inde-pendent and deinde-pendent variable and uses interval scales. With SEM the functional rela-tionships (Bagozzi & Yi, 2012) among the latent variables can be tested simultaneously (Anderson & Gerbing, 1988), thereby giving an advantage over ‘first generation multi-variate techniques’ (e.g. linear regression analysis). Additionally, the measurement and structural model are combined in one analysis, delivering metrics to assess the model quality and strength of relationships. In this study, the SEM analysed the relationships between experience encounter quality (exogenous variables), customer satisfaction, commitment (endogenous mediating variables), and brand loyalty (endogenous variable).

3.4 Conclusion

4. R

4.1 Introduction

This chapter first outlines data preparation and preliminary data analysis. Subse-quently, the results are presented in three steps (Figure 4.1). Before Section 4.2.2 presents descriptive results of the study, Section 4.2.1 outlines the characteristics of the sample. Afterwards, the estimation results of the reflective outer model are discussed. Then, the inner model quality and relationships among latent variables are presented and the research hypotheses are tested.

4.2 Preliminary Data Analysis: Data Screening and Sample

After two and a half weeks of data collection, the online-survey was closed. This resulted in a raw data set of N=295 responses. Authors underline the importance of accurate survey design in order to reduce nonresponses (Hair, Black, et al., 2014) which has been accounted for in Chapter 3, Section 3.3.2. However, (item) nonresponse is neither a neglected aspect in the literature nor a seldom phenome-non in mail-survey research (Craig & McCann, 1978). It was also present in the research at hand and can be traced back to respondents omitting answers. There-Figure 4.1 Proceeding of data analysis and presentation of results

Assessment of Data Characteristics

fore, it is important to identify missing data and its extent and to subsequently find remedies of how to deal with this issue.

In general, raw survey data showed that 156 respondents (52.9 %) provided a ful-ly completed questionnaire. With respect to the data anaful-lysis method of PLS-SEM, it is noteworthy that only complete datasets can be processed with SmartPLS software (Ringle, Wende, & Will, 2005), thus cases with missing val-ues would have been omitted. Consequently, in order to preserve cases, remedies needed to be evaluated. To improve the dataset and handle responses with missing values, different strategies are proposed in the literature, ranging from listwise deletion (Babin, Griffin, Borges, & Boles, 2013) and unconditional mean imputa-tion to multiple imputaimputa-tion of missing values (Craig & McCann, 1978; Gilley & Leone, 1991; Kamakura & Wedel, 2000; Kimmel & Smith, 2001). In this re-search, firstly cases with more than ten % missing values were omitted listwise (Hair et al., 2014, p.45). After also checking for abnormal cases (e.g. SD (Stand-ard Deviation) across all items=0), one further response was deleted, resulting in 214 observations. Secondly, items were checked for non-response patterns. Sensi-tive questions of income (9.3 %) and age (3.7 %) most often carried non-response. In addition, all items of preferential treatment showed a noticeably high amount of more than 50.0 % missing values due to respondents ‘not knowing’ the answer. Thus it can be assumed that differentiating between regular and non-regular cus-tomers in the automobile context is rather uncommon from a customer’s perspec-tive.

model-missing values (Enders & Peugh, 2004; Hair et al., 2014, pp.571f). Finally, a cleaned and EM-imputed data set consisting of 214 cases was retained for subse-quent analyses.

4.2.1 Sample Characteristics

The sample profile includes demographic variables of car brand customers (i.e., age, income) as well as vehicle-related characteristics (i.e., car brand, new or used car). An overview can be found in Table 4.1 and Table 4.2, respectively.

Table 4.1 Sample characteristics: demographics

Variable Category Observed Frequency (N=214) Percentage Gender Male 121 56.5 Female 93 43.5 Age 17 to 25 Years 45 21.0 26 to 35 Years 61 28.5 36 to 45 Years 20 9.4 46 to 55 Years 32 15.0 56 to 65 Years 30 14.0

66 Years and older 26 12.1

Educational Level Hauptschule 13 6.1 Realschule 57 26.6 A-Levels 40 18.7 Bachelor 39 18.3 Master 57 26.6 PhD 3 1.4 Other 5 2.3 Occupation Unemployed 3 1.4 Student 41 19.2 Worker 6 2.8 Employee 87 40.7 Executive 18 8.4 Self-Employed 10 4.7 Retiree 36 16.8 Other 13 6.0 Gross Income per Year Less than 25.000€ 76 35.5 25.001€ to 40.000€ 58 27.1 40.001€ to 55.000€ 42 19.6 55.001€ to 70.000€ 22 10.3 More than 70.000€ 16 7.5

changed car brands (43.5 %). Half of the respondents are only customers of their current car brand for less than five years (50.9 %), one quarter indicates to be a customer of the car brand for six to ten years. However, more than one fifth of the respondents have relatively long ties towards the car brand as being a customer for 11 years and longer. Furthermore, most of the respondents do not plan to pur-chase a car in the near future as only 5.6 % intend to purpur-chase a car within the next two years. In the longer run, 52.8 % of respondents plan to buy the next car (three to five years), and 41.6 % of respondents in more than 5 years. A compari-son of this sample to the adult, car-driving population in Germany shows that the sample’s demographic and car-related profiles do not meet but also are not com-pletely diverging from those of the population.

Table 4.2 Sample characteristics: Vehicle-related aspects Variable Category Observed Frequency

(N=214)

Percentage

Brand Tier Economy (low-tier) 18 8.4

Volume 137 64.0 Prestige (high-tier) 59 27.6 Car’s Condi-tion when bought New 82 38.3 Used 132 61.7 Previous Ties with Car Brand

Same brand rebought 83 38.8

Different brand rebought 93 43.5

First car bought 38 17.7

Length of Customership

Up to 5 years 109 50.9

6-10 years 53 24.8

11-15 years 20 9.3

More than 15 years 32 15.0

Next Car Pur-chase Planned

Up to 2 years 12 5.6

3-5 years 113 52.8

More than 5 years 89 41.6

4.2.2 Descriptive Results

sym-metry and peakedness/flatness of items’ distributions was evaluated accounting for the distributions’ normality and basing on values of univariate skewness and kurtosis. Most results permitted to assume that the distribution follows a Gaussian pattern because their values lie in the range of ±1.96 (Hair, Black, et al., 2014, p. 71). Nevertheless, for variables S8, C5, B4, B6, SF9 and SF11 a high kurtosis value indicates leptokurtic distributions compared to the normal distribution. Var-iable B6 (kurtosis = 8.40) showed extremely non-normal univariate distribution patterns. Thus it was eliminated from further investigations in order to prevent biased results. The balance of the distribution, which was quantitatively measured with the skewness values showed that the distribution of variable B4 extremely shifted to the left. One should keep in mind the slight non-normality of data dis-tribution for the structural model assessment lateron, because skewness might be a reason for “reduce[d] statistical power” (Hair et al., 2012, p. 421). Even though extremely skewed data foster biased results, in a simulation study PLS-SEM was shown to be a robust method for the analysis of non-normally distributed data (Cassel, Hackl, & Westlund, 1999). However, the distribution of item B6 deviated too extremely from normality, justifying its elimination. Besides this elimination, no further transformation of data for the sake of normality was performed.

Table 4.3 Item descriptions and measurement model results for latent constructs

Item Mean SD

Skew-ness Kurtosis Service Encounter

Facilitating Service

S1 _{The whole process of buying a car was so easy, the dealer of the car}

brand took care of everything. 4.99 2.01 -.73 -.72 S2 _{It is not just about the now; the dealer of the car brand will look after}

me for a long time. 3.74 2.10 .07 -1.31

S3 _{I am already a customer; the dealer of the car brand knows me and}

takes good care of me, so why should I go somewhere else? 3.17 2.18 .48 -1.23 S4 _{It was important that the car brand was flexible in dealing with me}

and looking out for my needs. 4.17 2.18 -.23 -1.38 S5 _{It is important that the people I am dealing with are good people; they}

listen, are polite and make me feel comfortable. 5.36 1.77 -1.08 .29 S6 _{The way the car brand deal(t) with me when things go(went) wrong}

will decide if I stay with them 5.30 1.90 -1.08 .07

Product Price Equity

Communication Encounter

Formal Communication

C1 _{[Car brand] often sends letters to customers like me. 3.02 2.16 .61 -1.12} C2 _{[Car brand] keeps customers informed through e-mails. 2.65 2.05 .89 -.65} C3 _{Car brand] often informs customers through brochures/ magazines. 3.16 2.19 .47 -1.29}

Personal Communication

C4 _{I was contacted by my car brand, which wanted to stay "in touch" and}

make sure I was still satisfied. 3.01 2.23 .61 -1.17 C5 _{I was contacted by my car brand, which wanted to keep abreast of}

changes in my family, and car needs. 1.89 1.60 1.83 2.39 C6 _{I received something of a personal nature from my car brand (e.g.,}

birthday card, holiday gift, etc.). 2.70 2.32 .91 -.85 C7 _{I was contacted by my car brand, which wanted to describe benefits}

of new car models. 2.60 2.20 .98 -.65

Brand Usage Encounter

Self-Connection

B1 _{The car brand helps me to express myself}

2.41 1.82 1.03 -.20

B2 _{The car brand reflects my personality}

2.55 1.96 .91 -.54

B3 _{The car brand enhances myself}

2.26 1.74 1.18 .21

Other User-Connection

B4 _{I have met wonderful people because of my car brand.}

1.93 1.65 1.79 2.19

B5 _{I feel a sense of kinship with other [car brand] owners.}

2.00 1.60 1.58 1.30

B6 _{I have an interest in a club for [car brand] owners.}

1.48 1.21 2.95 8.40

Customer Satisfaction

Relationship Satisfaction

SF4 _{As a customer, I have a high quality relationship with my car brand. 4.20 1.95 -.32 -1.12} SF5 _{I am happy with the efforts my car brand is making towards}

custom-ers like me. 3.76 1.75 -.02 -.70

SF6 _{I am satisfied with the relationship I have with my car brand.}

4.79 1.66 -.53 -.30

Automobile Satisfaction

SF7 _{I am satisfied with my car. 6.13 .98 -1.28 1.72} SF8 _{Owning this car has been a good experience 5.72 1.41 -1.25 1.22} SF9 _{I am sure it was the right thing to buy this car. 6.09 1.17 -1.60 2.77} Customer Commitment

Affective Commitment

CO1 _{I feel emotionally attached to [car brand].}

Ich fühle mich mit [Automarke] emotional verbunden.

2.68 1.92 .83 -.58

CO2 _{[Car brand] has a great deal of personal meaning for me.} [Automarke] hat sehr viel persönliche Bedeutung für mich.

2.48 1.89 1.06 -.12

CO3 _{I feel a strong sense of identification with [car brand].} Ich fühle eine starke Zugehörigkeit zu [Automarke].

2.47 1.90 1.11 -.01

Calculative Commitment

CO4 _{It would be very hard for me to switch away from my car brand right} now even if I wanted to.

Es wäre sehr hart für mich, gerade jetzt meine Automarke zu wech-seln, auch wenn ich es wollte.

2.30 1.96 1.31 .29

CO6 _{It would be too costly for me to switch from my car brand right now. 3.65 2.41 .16 -1.61} Brand Loyalty

Recommendation Intention

BL1 _{I would recommend my car brand to friends and relatives 5.37 1.64 -.94 .19} BL2 _{I will speak positively about my car brand 5.50 1.49 -1.04 .70} BL3 _{I intend to encourage other people to buy my car brand 3.25 1.97 .47 -.97}

Repurchase Intention

4.3 PLS-SEM

In this section, PLS-SEM is specified and reasons for its choice over regression analysis and Covariance-Based SEM (CBSEM) are outlined. PLS-SEM is a form of SEM. SEM has been put forward in the 1980s, combining prediction orienta-tion with latent variable modelling (Chin, 1998; Fornell & Bookstein, 1982; Wold, 1980). SEM takes into consideration the measurement as well as all the theorized structural relationships among diverse latent constructs simultaneously (Anderson & Gerbing, 1988; Hair, Hult, Ringle, & Sarstedt, 2014), thus giving precedence over regression analysis. For these reasons SEM has been applied with a growing number (Hair et al., 2012) in marketing and consumer research (Baumgartner & Homburg, 1996). Within SEM, two different approaches, CBSEM and PLS-SEM, can be identified (Hair et al., 2012). While CBSEM does not focus on explained variances in endogenous variables but develops a covari-ance matrix to test relationships among constructs, PLS-SEM fosters the ex-plained variance of endogenous constructs (Hair et al., 2012). Thus, the literature proposes that CBSEM is rather suited for confirmatory research . Instead PLS-SEM should be used for exploratory analysis and theory building.

4.4 Reflective Outer Model Evaluation

In order to prepare for the structural model evaluation, which will follow in Sec-tion 4.5 as well as to test the adequate specificaSec-tion of blocks of manifest items for the estimation of each latent construct, the measurement model’s quality was assessed. This proceeding is in line with a two-step approach that suggests first to evaluate the measurement model before second, to calculate the structural model (Anderson & Gerbing, 1982; Hair, Ringle, & Sarstedt, 2011).

Following other researchers’ procedures in the utilization of PLS-SEM (Grace & O’Cass, 2005; Wilson, 2010), initially, unidimensionality of sets of variables, convergent and content validity were checked and confirmed by running a factor analysis for each construct. Due to the fact that the measurement model could be specified as reflective (compare Section 3.3.1), it is suggested to include indicator reliability, internal consistency reliability, convergent validity as well as discrimi-nant validity for the outer model assessment (Becker, Klein, & Wetzels, 2012; Hair et al., 2011, 2012; Table 4.4). SmartPLS 3.0 software (Ringle, Wende, & Becker, 2014) was used for all computations.

Table 4.4 Evaluation criteria for reflective outer model

Criterion Measure Estimate Source

Reliability Indicator Reliability Outer Loadings ≥ 0.70 (≥ 0.40) (Chin, 1998; Hair et al., 2011; Hair, Sarstedt, et al., 2014; Hulland, 1999) Internal Consistency Reliability Composite Reliability (CR) ≥ 0.70 (≥ 0.60)

(Bagozzi & Yi, 1988) (Nunnally & Bernstein, 1994) Validity Convergent Validity Average Variance Extracted (AVE)

≥ 0.50 (Bagozzi & Yi, 1988)

Discriminant Validity

Fornell-Larcker

Cri-terion AVE > Cross Loadings

(Fornell & Larcker, 1981) Cross Loadings Factor Loading >

Cross Loading

(Hair et al., 2011; Hair, Sarstedt, et al.,

4.4.1 Indicator and Internal Consistency Reliability

Indicator reliability tests to what extent a manifest variable is appropriate for the measurement of the related construct by indicating each factor’s loading. For most items, outer loadings exceeded the recommended cut-off value of 0.70 (Hair et al., 2012; Hulland, 1999; Nunnally & Bernstein, 1994). And all items had acceptable bootstrap2 critical ratios (value > 1.96) (Hulland, 1999), which means that t-values ranging from 11.61 to 89.86 indicate significant loadings at a 0.05 level. However, item S6 (0.65) and measure CO5 (0.15) had component loadings below the benchmark value of 0.70. Thus the impact of item elimination on Average Variance Extracted (AVE) and Composite Reliability (CR) values was checked. S6’s construct had already met required AVE and CR values, and item deletion from the scale would not have caused an incremental increase in those evaluation criteria. Furthermore, Bagozzi and Yi (1988) proposed a more liberal cut-off val-ue of 0.50 for indicator reliability and also content validity was ensured by not removing this benchmark-approaching item. Thereby also the paradigm of ‘con-sistency at large’ (Henseler et al., 2009; Wold, 1980), under which PLS-SEM is characterized to work best was followed. While “loadings of .5 or .6 may still be acceptable if there exist additional indicators in the block for comparison basis” (Chin, 1998, p. 325), CO6’s outer loading of 0.15 was also largely below these more liberal values. Therefore this item was removed due to the lack of indicator reliability.

Furthermore, internal consistency reliability of item scales was assessed with CR. Compared to Cronbach’s alpha, CR does not assume that all measures of a con-struct are equally reliable. Instead it prioritizes indicators according to their indi-vidual reliability and is thus less conservative and more adequate for PLS-SEM evaluations (Hair et al., 2012; Henseler et al., 2009). According to Bagozzi and Yi (1988) and Nunnally and Bernstein (1994) CR values should be higher than 0.60 and 0.70, respectively. Drawing on these criteria, all constructs demonstrate satis-factory CR values (0.78 to 0.94).