THE EFFECT OF RELATION-‐SPECIFIC INVESTMENTS ON BUYER-‐SUPPLIER RELATIONSHIP
PERFORMANCE IN INDUSTRIAL MARKETS
WITH RELATIONAL TRUST AND RELATIONAL COMMITMENT AS MODERATING VARIABLES
By
THE EFFECT OF RELATION-‐SPECIFIC INVESTMENTS ON BUYER-‐SUPPLIER RELATIONSHIP
PERFORMANCE IN INDUSTRIAL MARKETS
WITH RELATIONAL TRUST AND RELATIONAL COMMITMENT AS MODERATING VARIABLES
By
CARIEN DE RAAD
University of Groningen
Faculty of Economics and Business
Master of Business Administration
Management summary
Background: the use of relation-‐specific investments has been demonstrated to be effective in improving buyer-‐supplier relationship performance. For this research, we carry out a two-‐step analysis. First, we investigate the main effect: “Is there a positive effect of relation-‐specific investments on buyer-‐supplier relationship performance?” Second, this study focuses on the moderating effect of relational trust and relational commitment: “Is there a positive influence of relational trust on the effect relation-‐specific investments have on buyer-‐supplier relationship performance?” and “Is there a positive influence of relational commitment on the effect relation-‐specific investments have on buyer-‐supplier relationship performance?”
Methods: the research made use of a database, which consists of 230 suppliers. This database describes the characteristics of the parties involved (the buyer, the supplier and in some cases the intermediary) and the characteristics of the focal relationship (its development, the characteristics of the exchange, the organizational setting and the relationship atmosphere). A univariate linear regression is used to investigate the main effect and a multiple regression (moderator analyses) is used to examine the moderating effect.
Results and conclusions: the results of this research show that relation-‐specific investments significantly positively affect buyer-‐supplier relationship performance. This research also makes clear that relational trust significant positively moderates the effect relation-‐specific investments have on buyer-‐supplier relationship performance. The moderating effect of relational commitment on the effect of relation-‐ specific investments on buyer-‐supplier relationship performance is not significant. Therefore can be concluded that this study collaborates for the biggest part to the evidence in which buyer-‐supplier relationship is positively affected by relation-‐specific investments and is moderated by relational trust.
Key words: Buyer-‐supplier relationship performance, relation-‐specific investments,
relational trust, relational commitment
Research theme: Relationship atmosphere
Table of content
Management summary ... 3
Chapter 1: Introduction ... 6
1.1 Background problem ... 6
1.2 Theoretical and social relevance ... 7
1.3 Research questions ... 8
1.4 Structure of thesis ... 8
Chapter 2: Theoretical framework ... 9
2.1 Literature review ... 9
2.1.1 Relation-‐specific investments ... 9
2.1.2 Buyer-‐supplier relationship performance ... 10
2.1.3 Relational trust ... 12
2.1.4 Relational commitment ... 14
2.2 Conceptual model ... 15
Chapter 3: Research design ... 16
3.1 Research method ... 16
3.2 Sampling design ... 16
3.3 Measures ... 16
3.4 Plan of analysis ... 17
3.4.1. Reliability analysis ... 17
3.4.2 Data test for linear regression ... 18
3.4.3 Regression analysis ... 19
3.4.4 Data test for multiple regression (moderator analysis) ... 19
3.4.5 Moderator analysis ... 20
Chapter 4: Results ... 22
4.1 Reliability analysis ... 22
4.1.1 Reliability relation-‐specific investments ... 23
4.1.2 Reliability buyer-‐supplier relationship performance ... 24
4.1.3 Reliability relational trust ... 24
4.1.4 Reliability relational commitment ... 25
4.1.5. Reliability of the variables ... 26
4.3 Hypothesis 1 ... 27
4.4 Data test for multiple regression (moderator analysis) ... 29
4.5 Hypothesis 2 ... 32
4.6 Hypothesis 3 ... 34
Chapter 5: Conclusions, discussion and recommendations ... 37
5.1 Conclusion ... 37
5.1.1 Main effect ... 37
5.1.2 Moderating effects ... 38
5.2 Discussion ... 38
5.3 Limitations ... 39
5.4 Implications and future research directions ... 39
References ... 41
Appendixes ... 47
Appendix 1: Questions ... 47
Appendix 2: Output reliability analysis ... 50
2.1 Relation-‐specific investments ... 50
2.2 Buyer-‐supplier relationship performance ... 52
2.3 Relational trust ... 55
2.4 Relational commitment ... 57
2.5 Definitive questions ... 59
Appendix 3 Data test linear regression ... 62
Appendix 4 Output hypothesis 1 ... 65
Appendix 5 Data test multiple regression (moderator analysis) ... 66
Appendix 6 Output hypothesis 2 ... 77
Appendix 7 Output hypothesis 3 ... 78
Chapter 1:
Introduction
1.1 Background problem
Today's business environment becomes a more and more highly competitive and fast-‐paced business environment (Pazos, Chung and Micari, 2013). In international industrial businesses the role of personal interaction becomes more important (Mainlea and Ulkuniemi, 2013). It involves personal interaction between people with different cultural backgrounds, practices, rules, etc. (Wang and Nayir, 2006). Different cultures imply different mental programming, which governs activities, motivations and values (Yeh, 1988). To be successful, these different activities, motivations and values need to be neutralized by personal interaction (Hofstede, 1983). There is a shifting from standardized and anonymous market relations to specific customer-‐supplier relationships, the interaction between business people no longer takes place in a social vacuum; this setting is called ‘the relationship atmosphere’ (Williamson, 1985). For example, in a wholesaling environment, firms with close relationships with suppliers can achieve a competitive advantage by receiving merchandise in short supply and information on new and best selling products (Stern, 1996). Therefor within this relationship atmosphere research theme it is important to research how the buyer-‐supplier relationship performance can be better.
There is research done about relationship atmospheres in industrial markets. Increasing evidence suggests that business relationships are of paramount importance for firms, because such relationships can create value for both parties involved (Anderson and Weitz, 1992). Relationship atmospheres among firms involved in distribution has changed from one of confrontation to that of collaboration (Hoyt and Huq, 2000). Furthermore, there is argued that to come to that collaboration, substantial investments in relationships are necessary (Zhoa and Wang, 2011). Zhao and Wang (2011) say that using relation-‐specific investments can cause value creation on the buyer-‐supplier relationship performance. Different literature also suggests that value creation on the buyer-‐supplier relationship performance depends on special relationship characteristics. Hallen and Sandstorm (1991) uses six dimensions to show that value creation depends on special relationship characteristics: (1) power/ dependence balance; (2) cooperativeness/ competitiveness; (3) trust/opportunism; (4) understanding; (5) closeness; and (6) commitment and Zaheer, McEvily Perronne (1998) emphasize trust lead to increased relationship success, Meyer and Allen (1991) makes clear commitment contributes to a positive business performance and Morgan and Hunt (1994) say that value creation includes trust and commitment.
Within the research theme relationship atmosphere there are many different topics that can be researched. When looking at the literature above, there has been much research done into the various recurring topics that are interesting to investigate. The importance of a good buyer-‐supplier relationship performance emerges, but also the influence of relation-‐specific investments, commitment and trust. The effect of relation-‐specific investments on buyer-‐supplier relationship performance is mainly investigated. For example Williamson (1985) emphasizes that relation-‐specific investments are a key to relationship success. Also trust and commitment are important. The effect of trust and commitment on relationship performance is independently of each other often researched, but only one time the effect of trust and commitment on relationship performance is tested together, but this was already in 1994 by Morgan and Hunt. Morgan and Hunt (1994) theorized that a good relationship performance depends on successful relationship marketing, which requires relational trust and relational commitment. In the research of Morgan and Hunt (1994), relational trust and relational commitment were seen a mediators. This research will provide a new insight by researching if relation-‐specific investments further improve the buyer-‐ supplier relationship performance and because of the return of commitment en trust from the literature above it is also interesting to research if commitment and trust also influence the process in which relation-‐specific investments influence the buyer-‐supplier relationship performance. Here relational trust and relational commitment will be seen as moderators, how they moderate the effect relation-‐specific investments have on buyer-‐supplier relationship performance.
The aim of this research within the research theme relationship atmosphere is therefor to find out whether relation-‐specific investments affect the buyer-‐supplier relationship performance in industrial markets and whether this effect is influenced by relational trust and relational commitment.
1.2 Theoretical and social relevance
provide new insights: the database is very wide, because the focus is international and there is no distinction between different branches.
Both buyer and supplier are affected by today’s highly competitive business environment. If it becomes clear which must be held in order to get the best the buyer-‐supplier relationship performance, companies in industrial markets can use this information. If businesses in industrial markets know how the buyer-‐ supplier relationship performance is influenced, these businesses can use this information to respond to changes and will become stronger players in this highly competitive market. But for businesses that want to use strong relationships, the businesses first have to know how to achieve these strong relationships.
The main question that can be asked and derives from the aim of this research will be: “What is within the relationship atmosphere concept the effect of relation-‐specific investments on the buyer-‐supplier relationship performance in industrial markets?”
1.3 Research questions
• Is there a positive effect of relation-‐specific investments on buyer-‐supplier relationship performance within the relationship atmosphere concept?
• Is there a positive influence of relational trust on the effect relation-‐specific investments have on buyer-‐ supplier relationship performance within the relationship atmosphere concept?
• Is there a positive influence of relational commitment on the effect relation-‐specific investments have on buyer-‐supplier relationship performance within the relationship atmosphere concept?
1.4 Structure of thesis
This thesis consists of five chapters. The first chapter consists of an introduction of the topic, the aim, relevance, contribution and research question(s) of this thesis. The second chapter gives an overview of the existing literature that is relevant for this thesis. Important definitions, common views, assumptions are addressed and the conceptual model is given with the accompanying explanation and the hypothesis. The third chapter explains the methodology part of this thesis. Chapter four is an overview of the results
Chapter 2:
Theoretical framework
2.1 Literature review
This section discusses the existing literature on the research theme relationship atmosphere and the related components that will be examined in this research: relation-‐specific investments, buyer-‐supplier relationship performance, relational commitment, relational trust and the relation between these components.
Where previously the buyer and supplier were considered separately, recent trends in industrial markets indicate that buyers and suppliers go for the common benefits by working more closely together (Heide and John, 1990). Providing a good buyer-‐supplier relationship performance becomes more and more important to use as a competitive advantage. In order to maximize the value creation in the supply chain, management of buyer-‐supplier relationships are central to the success (Chen, Paulraj and Lado, 2004). Studies have shown that successful management of these relationships contributes to the performances of the firms (Tan, Kannan, Handfield and Ghosh, 1999). In addition, there are more and more firms which make use of their network to support the buyer-‐supplier relationship performance (Gulati, 2000).
In order to ensure that the buyer and supplier are better able to work together, investments will be made to enhance this cooperation (Heide and John, 1990). The challenge for marketers and corporations is to understand how relation-‐specific investments affect buyer-‐supplier relationship performance. Dimensions such as trust and commitment are shown to play an important role in high-‐value strategic relationships, where specific investments are high (Morgan and Hunt, 1994). To contribute to the existing literature this paper will discuss the value of relation-‐specific investments, moderated by the effect of relational commitment and relational trust for the level of relationship performance between buyers and suppliers in industrial markets.
2.1.1 Relation-‐specific investments
Relation-‐specific investments serve as a solid foundation for efficient knowledge sharing between channel members (Hussain, Lucas and Asif Ali, 2003). More chances of working together enable both channel members to acquire valuable managerial experience, knowledge of operation process and working culture of each other (Hussain, Lucas and Asif Ali, 2003). Relationship learning in the channels relies on the interaction between people from both sites (Kale, Singh, Perlmutter, 2000). A close and intense interaction between individual members of the concerned organizations acts as an effective mechanism. Channel members tend to generate more relationship learning after they make relation-‐specific investments in channel relationships (Kale, Singh, Perlmutter, 2000).
The transaction cost theory states that management of relationships will be predicted by the degree of specific investments involved (Williamson, 1985). The theory determines adaptation (or relationship-‐ specific investments) and reduction in uncertainty as a key to relationship success (Williamson, 1985). For instance, if one party makes relationship-‐specific investments, this will only be done when the other party reduces the risk of opportunism by also making relationship-‐specific investments or by offering contractual guarantees (Rokkan, Heide and Wathe, 2003).
Economics, such as Williamson (1985) examined the costs of asset specificity in inter-‐firm relationships. They point out that a firm making specific investments increases its reliance on its transactional partner and is subject to the partner’s opportunistic behavior. Economists have long recognized that investments in specialized assets increase a firm's performance (Williamson, 1985). Relation-‐specific investments signal the desire to invest in an endured relationship (Anderson & Weitz, 1992). Although investments in specialization stimulate performance, the motive to make relation-‐specific investments is tempered by the fact that the more specialized a resource becomes, the lower is its value in alternative use. Strong forms of asset specificity are relatively rare because compared to general resources the contingent value of relation-‐specific investments exposes their owners to a greater risk of opportunism (Klein, 1978).
2.1.2 Buyer-‐supplier relationship performance
Over the years, the natures of the buyer-‐supplier relationships undergo some major changes. Supply chain researchers frequently describe these relationships as becoming closer and terms such as partnerships and alliances are being used as a contrast with the traditional spot market exchange. It is not unusual to read that buyer firms are looking to their suppliers to help them achieve a stronger competitive position.
buyer-‐supplier relationships can be a source of competitive advantage for manufacturing firms (Carr and Pearson, 1999). For most companies, knowledge of technologies, markets and customers are the key to maintaining competitive advantage. A relationship is a joint activity between channel members in which two partners share information (Selness and Sallis, 2003). A well-‐performing relationship exists if both partners are satisfied with the relationship’s effectiveness and efficiency (Selness and Sallis, 2003).
A basic requirement for relationship performance is the reduction in uncertainty for both parties (Morris and Carter, 2005). In buyer-‐supplier relationships it is important that both parties perceive that they are gaining value from the relationship if it is to continue and the relationship is to be considered a success (Narayandas and Rangan, 2004). Most researchers on buyer-‐supplier relationships agree that the perception from both buyers and suppliers should be studied in order to gain insights into their relationships (John and Reve, 1982). To test buyer-‐supplier performance, there can be looked at performance characteristics of supply management orientation (Shin, Collier and Wilson, 1999).
There is research done about the influence that investments have in de context of industrial buyer-‐ supplier relationships (Joshi & Stump, 1999; Nielson, 1996). Relation-‐specific investments in a supplier-‐ buyer relationship can determine the relationship (Cook, 1977). It is difficult to replace a partner, for example when a supplier makes dedicated investments, the supplier creates dedicated assets, which in return increases his switching costs and makes the supplier more dependent (Williamson, 1975). The literature shows that relation-‐specific investments have much impact on businesses. Relation-‐specific investments in buyer-‐supplier relationships increase the supplier’s switching costs, decrease the supplier’s intention to terminate the relationship with the buyer, and thus create a mutual dependence between the buyer and the supplier (Weiss and Kurland, 1997).
There are different dynamics of relationship performance for the buyers and suppliers; there are clear differences in drivers of relationship success, for example, buyers want their suppliers to adapt their products, services, procedures and processes and to make relationship-‐specific investments for the buying company (Anderson and Weitz, 1992). But suppliers want a good buyer-‐supplier relationship to accomplish the channel goal, to continue to work together in the long term and improve their profitability
Empirical studies from for example Liu, Liu and Luo (2009), Selnes and Sallis (2003) and Anderson and Weitz (1992) support the positive relationship between relation-‐specific investments and relationship performance. Relation-‐specific investments can promote relationship learning between partners, and thus improve relationship performance. According to Ganesan (1994) developing a long-‐term relationship requires substantial sacrifices, such as increased specific investments in the relationship between buyers and suppliers.
Because of studies mentioned above which support the positive relationship between relation-‐specific investments and buyer-‐supplier relationship performance, this will also be tested in this research. To research this, a hypothesis will be drawn that will be tested by the use of the database. The hypothesis that is made here is:
H1: Buyer-‐supplier relationship performance is positively affected by relation-‐specific investments.
To have a good buyer-‐supplier relationship performance not only relation-‐specific investments are important, a long-‐term partnership is also important. If we look at long-‐term partnerships, we can think of trust and commitment (Morgan and Hunt, 1994). Literature from the supply chain, marketing and strategy field identify predictors of relationship success. There is a general agreement that communication between partners’ leads to increased trust and commitment (Morgan and Hunt, 1994).
2.1.3 Relational trust
Relational trust is defined as the confidence or belief that the exchange partner possesses about the honesty and benevolence of other partners (Kumar, Scheer and Steenkamp, 1995). Trust is the willingness you have as a party, to perform actions by another party based on an expectation that the other party can do better (Mayer, Davis and Schoorman, 1995). Trust refers to ‘‘confidence in an exchange partner’s reliability and integrity’’ (Morgan and Hunt, 1994). Trusting other parties provides the basis for assessing predictability of future behavior based on past interaction and promises and reducing uncertainty (Crosby, Evans and Cowles, 1990).
effective planning. Once trust is established, firms learn that coordinated, joint efforts will lead to outcomes that exceed what the firm would achieve if it acted solely in its own best interest (Anderson and Narus, 1990).
By studying buyer-‐supplier relationships, hypothesis are supported that buyers that trust the counterpart is likely to be engaged in collaborative joint efforts (Moorman, 1992). This suggests that the buyer that trusts its supplier will exchange relevant, comprehensive, accurate and timely information, and thereby contribute to problem-‐solving and planning effort (Zand, 1972). Thus, trust will form the relational basis for the development and maintenance of cooperation between for example buyer and supplier (Zand, 1972). Lending support to previous studies, the strongest antecedent for relationship performance for buyers is trust from the supplier (Zaheer, McWvily and Perrone, 1998). Furthermore, buyers want their suppliers to adapt their products, services, procedures and processes and to make relationship-‐specific investments for the buying company (Anderson and Weitz, 1992).
Trust has been a widely studied concept both by itself and, most importantly, as a component of the quality of relationships. Research on trust has shown that it is a multi-‐dimensional concept. Grunig (1983) has identified three dimensions of trust that are measurable: competence, integrity and dependability/reliability. Competence is the belief that an organization has the ability to do what it says it will do, including the extent to which an organization is seen as being effective and that it can compete and survive in the marketplace. Integrity is the belief that an organization is fair and just. And dependability/reliability is the belief that an organization will do what it says it will do, that it acts consistently and dependably. Previous research in channel relationships has emphasized the importance of trust in fostering collaboration (Anderson and Narus, 1990).
research this, a hypothesis will be drawn that will be tested by using the database. The Hypothesis that is made here is:
H2: The higher the level of relational trust, the stronger the positive relationship between relation-‐specific investments and buyer-‐supplier relationship performance will be.
2.1.4 Relational commitment
Commitment has been defined as an enduring desire to develop and maintain exchange relationships characterized by implicit and explicit pledges and sacrifices for the long-‐term benefit of all partners involved (Rylander, Strutton and Pelton, 1997). Affective commitment is the result of emotional bonds that may drive parties to maintain and improve the quality of their relationship (Bendapudi and Berry, 1997). Thus, a social structure is generated through individuals’ desire to be psychologically and emotionally consistent throughout the interaction (Meyer and Allen, 1991). During this process managers identify shared values and goals of their organizations to which they are psychologically attached (Gundlach, 1995). According to this view, committed partners desire to continue their relationship because they like and enjoy the relationship.
Since there are relation-‐specific investments, commitment will be important. If buyers and suppliers are more commitment to each other, they are more willing to share knowledge (Zhao and Wang, 2011). By examining the role of specific investments and commitment during a relationship life-‐cycle is found that the transaction-‐specific investments enhances commitment in the exploration phase and has a positive effect during the decline phase (Sandy and Ganesan, 2000). These inputs or investments into the relationship act as barriers against one party leaving the relationship, as it becomes more costly to terminate the relationship (Morgan and Hunt, 1994).
Studies mentioned above support that relational commitment could act as a moderator or mediator. Like relational trust, relational commitment will be tested in this research as a moderator. The studies mentioned above support that relational commitment influence the effect relation-‐specific investments have on buyer-‐supplier relationship performance. This research will test whether relational commitment will positively influence this main effect. To research this, a hypothesis will be drawn that will be tested by using the database. The Hypothesis that is made here is:
H3: The higher the level of relational commitment, the stronger the positive relationship between relation-‐ specific investments and buyer-‐supplier relationship performance will be.
2.2 Conceptual model
In this conceptual model, there is a direct relationship between relation-‐specific investments and buyer-‐ supplier relationship performance. Relational trust and relational commitment will be tested on moderating. There will be argued how to improve the buyer-‐supplier relationship performance through relation-‐specific investments, moderated by relational trust and relational commitment.
Figure 1. Conceptual Model
Chapter 3:
Research design
3.1 Research method
In order to test the conceptual model and thereby answering the research question: “What is within the relationship atmosphere concept the effect of relation-‐specific investments on the buyer-‐supplier relationship performance in industrial markets?” There is made use of a database. The database describes the characteristics of the parties involved (the buyer, the supplier, and – in some cases – the intermediary) and the characteristics of the focal relationship (its development, the characteristics of the exchange, the organizational setting and the relationship atmosphere).
3.2 Sampling design
The target population consists of N=230. The respondents are suppliers. The focus of the database is international and there is no distinction between different branches.
3.3 Measures
In order to measure all the variables in the conceptual model, there will be looked at different questions from the questionnaire that is used to create the database. The questionnaire asked questions on various topics. In order to research the variable relation-‐specific investments their will be looked at the questions 27 until 30. These questions are about investments buyers and supplier will make. These questions research both the buyer and supplier perspective, there is made use of one construct instead of splitting into two constructs (buyer and suppliers). To check if this is a logical choice, it’s a good idea to see if the averages of the questions 28 (In all, how large is the investment made by the customer in his relationship with your firm?) and 30 (In all, how large is the investment made by your firm in the relationship with this customer?) are close to each other. Otherwise it would be better to see the buyer and supplier independent of each other. The questions can be answered from 1 to 5 and 9. Where 1 is none, 5 is very large and 9 do not know. 9 is omitted in order to calculate the average.
the relationship’s effectiveness and efficiency (Selness and Sallis, 2003). This research uses soft measures to test buyer-‐supplier relationship performance.
In de questionnaire there is a special focus on the variables relational trust and relational commitment. The questions 56 until 63 are about relational trust en the questions 83 until 89 are about relational commitment. Appendix 1 shows the different questions per variable.
3.4 Plan of analysis
Some basic descriptive statistics in SPSS will be used to explore the data to test the conceptual model and thereby answering the research question.
3.4.1. Reliability analysis
The variables relation-‐specific investments, buyer-‐supplier relationship performance, relational trust and relational commitment will be tested on reliability. The reliability analysis will be used to refer to the extent to which a scale produces consistent results if repeated measurements are made. In this way you can ensure that the questions used per variable are all reliably measure the same latent variable. To test the internal consistency, the used method will be the Cronbach’s Alpha test using the reliability command. To interpret the output, the rule of George and Mallery (2003) can be followed: > ,9 (excellent), > ,8 (good), > ,7 (acceptable), > ,6 (questionable), > ,5(poor), and < ,5 (unacceptable). The Cronbach’s Alpha cut-‐off will be ,7 and should absolutely not be lower than ,6.
Cronbach's alpha reliability coefficient normally ranges between 0 and 1. The closer the coefficient is to 1,0, the greater is the internal consistency of the items (variables) in the scale. Cronbach's Alpha coefficient increases either as the number of items (variables) increases, or as the average inter-‐item correlations increases (Tavakol and Dennick, 2011). Because of the use many items (variables) to research relation-‐specific investments and buyer-‐supplier relationship performance it is according to Tavakol and Dennick (2011) also important to use the Factor-‐Analyses to know for sure that the Cronbach’s Alpha is high because of the good reliability and not because of the high number of items (variables).
After performing the factor analysis, we look at 4 outcome tables:
1. Correlation Matrix: The Correlations Matrix shows the correlation between different items. It is important that there is correlation. But the correlation may not be higher than .9, otherwise there is multicollinearity. If the correlation is higher than ,9 the factor analysis may not be carried out. 2. KMO and Bartlett’s Test: It is common to assume that the KMO should be higher than .5. In that case
the factor analysis may be carried out.
3. Total Variance Explained: In this table we find the factors that have a higher “eigenvalue” than one. The “eigenvalues” which are higher than one are factors that explain more than a single item
4. Component Matrix: This table shows how many items together maybe one factor.
3.4.2 Data test for linear regression
Before really analyzing the data using the linear regression, first there must be checked that the data can actually be analyzed using linear regression. According to Poole and O’Farrell (1971) it is only appropriate to use linear regression if the data "passes" six assumptions that are required for linear regression to give you a valid result.
Assumption 1: The two variables should be measured at the continuous scale: the variables are either an interval or a ratio variable.
Assumption 2: There needs to be a linear relationship between the two variables: creating a scatterplot in which the dependent variable (buyer-‐supplier relationship performance) is plot against the independent variable (relation-‐specific investments) and then visually inspect the scatterplot to check for linearity.
Assumption 3: There should be no significant outliers: check for outliers using "Casewise-‐Diagnostics”. Assumption 4: Should have independence of observations: check using the Durbin-‐Watson statistic. The Durbin-‐Watson statistic can vary between 0 and 4 with a value of 2 meaning that the residuals are uncorrelated. A value greater than 2 indicates a negative correlation between adjacent residuals whereas a value below 2 indicates a positive correlation.
Assumption 5: The data needs to show homoscedasticity: which is where the variances along the line of best fit remain similar as you move along the line.
3.4.3 Regression analysis
After correlation and passing the six assumptions the next step will be linear regression. The regression analysis is used to predict the value of a variable based on the value of another variable. The variable that needs to be predicted is the dependent variable. In this research the dependent variable is buyer-‐supplier relationship performance. The variable that is used to predict the other variable's value is called the independent variable. In this research the independent variable is relation-‐specific investments. In this research the linear regression is used to understand whether buyer-‐supplier performance can be predicted based on relation-‐specific investments.
In this research the univariate linear regression analysis will be used, because of deriving a mathematical relationship, in the form of an equation, between the single dependent variable “buyer-‐supplier relationship performance” and the single independent variable “relation-‐specific investments”. The univariate linear regression analysis is similar in many ways to determining the simple correlation between two variables. The regression analysis will also be used to test the moderating effect. Moderation occurs when the moderators relational trust and relational commitment interact the main effect. To estimate the quality of the univariate linear regression analysis, there will be made use of three linear regression diagnostics:
1. How much of the variance is explained? 2. What is the significance of the total model?
3. Is there a significant relationship between the variables?
3.4.4 Data test for multiple regression (moderator analysis)
Before really analyzing the data using the multiple regression analysis, the moderator analysis, first there must be checked that the data can actually be analyzed using multiple regression. According to Poole and O’Farrell (1971) it is only appropriate to use multiple regression if the data "passes" eight assumptions that are required for multiple regression to give a valid result.
Assumption 1: The dependent variable should be measured at the continuous scale: the variable is either an interval or a ratio variable.
Assumption 2: There is one independent variable, which is a ratio or interval variable and one moderator variable that dichotomous or on continuous scale.
uncorrelated. A value greater than 2 indicates a negative correlation between adjacent residuals whereas a value below 2 indicates a positive correlation.
Assumption 4: There needs to be a linear relationship between the dependent variable and the independent variable for each group of the moderator variables: creating a scatterplot and then visually inspect the scatterplot to check for linearity.
Assumption 5: The data needs to show homoscedasticity: which is where the variances along the line of best fit remain similar as you move along the line.
Assumption 6: The data must not show multicollinearity, this occurs when two or more independent variables are highly correlated with each other: check through an inspection of correlation coefficients and Tolerance/VIF values. There is no multicollinearity when the VIF value is between 1 and 10.
Assumption 7: There should be no significant outliers, high leverage points or highly influential points: outliers, leverage and influential points are different terms used to represent observations in the data set that are in some way unusual when you wish to perform a moderator analysis. Detect outliers using "studentized deleted residuals", check for leverage points and check for influential points using Cook's Distance. If Cook’s Distance is less than 1,00, you don’t have to worry.
Assumption 8: Check that the residuals (errors) are approximately normally distributed: to check this, the Normal P-‐P Plot can be used.
3.4.5 Moderator analysis
By using the moderator analysis there will be examined if the relation between the independent variable “relation-‐specific investments” and the dependent variable “buyer-‐supplier relationship performance”, is affected by the moderators “relational trust” and “relational commitments”. There will be examined, if the relationship between relation-‐specific investments and buyer-‐supplier relationship performance is affected by relational trust or relational commitment. The moderator analysis consists of three steps in SPSS:
1. Centralize
2. Calculating new predictor
rsitrust = rsicentra * trustcentra
rsicommitment = rsicentra * commitmentcentra
3. Regression analysis
A regression analyses should be done with the new (independent) variables: trustcentra, rsicentra and rsitrust with the dependent variable buyer-‐supplier relationship performance. Also a regression analysis should be done with the new (independent) variables: commitmentcentra, rsicentra and rsicommitment with the dependent variable buyer-‐supplier relationship performance.
Chapter 4:
Results
4.1 Reliability analysis
Before testing the effect of relation-‐specific investments on buyer-‐supplier relationship performance in industrial markets with relational trust and relational commitment as moderator variables, all variables must be tested on reliability. See table 1 for an overview of their Cronbach’s Alpha’s. The detailed Cronbach’s Alpha for each of the four variables is drawn from section 4.1.1 to 4.1.4.
TABLE 1 OVERVIEW RELIABILIY
Reliability Statistics Relation-‐Specific Investments
Cronbach’s Alpha
Cronbach’s Alpha Based on Standardized Items
N of Items
,868 ,887 19
Reliability Statistics Buyer-‐Supplier Relationship Performance
Cronbach’s
Alpha Cronbach’s Alpha Based on Standardized Items N of Items
,684 ,695 13
Reliability Statistics Relational Trust
Cronbach’s
Alpha Cronbach’s Alpha Based on Standardized Items N of Items
,704 ,718 4
Reliability Statistics Relational Commitment
Cronbach’s
Alpha Cronbach’s Alpha Based on Standardized Items N of Items
,716 ,722 5
The Factor-‐Analyses is used to test to know for sure the Cronbach’s Alpha is high because of the good reliability and not because of the high number of items. See table 2 for an overview of the output of the Factor-‐Analyses. The detailed Factor-‐Analyses for each of the four variables is drawn from section 4.1.1 to 4.1.4.
TABLE 2 OVERVIEW FACTOR-‐ANALYSES Factor-‐Analyses Relation-‐Specific Investments
Correlation
Matrix Kaiser-‐Meyer-‐Olkin KMO Total Variance Explained Component Matrix
All not higher
than ,9 ,812 4 factors higher than “eigenvalue” 1 Chcstock Chc_finp
Factor-‐Analyses Buyer-‐Supplier Relationship Performance
Correlation
Matrix Kaiser-‐Meyer-‐Olkin KMO Total Variance Explained Component Matrix
All not higher than ,9
,668 4 factors higher than
“eigenvalue” 1
At_u_per
Factor-‐Analyses Relational Trust
Correlation
Matrix Kaiser-‐Meyer-‐Olkin KMO Total Variance Explained Component Matrix
All not higher
than ,9 ,689 1 factor higher than “eigenvalue” 1 none
Factor-‐Analyses Relational Commitment
Correlation
Matrix Kaiser-‐Meyer-‐Olkin KMO Total Variance Explained Component Matrix
All not higher than ,9
,718 1 factor higher than
“eigenvalue” 1
none
4.1.1 Reliability relation-‐specific investments
First is looked at the Cronbach’s Alpha of the nineteen questions used for the variable relation-‐specific investments, see appendix 1. Testing the reliability with these nineteen questions gives a Cronbach’s Alpha of ,868, see appendix 2.1. this is above ,8 and according to George and Mallery (2003) this is good. When looking at the column Cronbach's Alpha if item deleted, there is no higher Cronbach’s Alpha possible when removing another question. This means that the Cronbach’s Alpha for relation-‐specific investments is ,868.
nineteen items. This means that it is better to remove the items Chcstock and Chc_finp, because they don’t actually belong to factor 1.
4.1.2 Reliability buyer-‐supplier relationship performance
To see if the reliability for the variable buyer-‐supplier relationship performance is good, the Cronbach’s Alpha of the sixteen questions associated to the buyer-‐supplier relationship performance is tested, see appendix 1. Testing the reliability of these sixteen questions gives a Cronbach’s Alpha of ,646, see appendix 2.2. This is not very high and therefor there is tested to see when removing some questions the Cronbach’s Alpha becomes higher. By looking at the corrected item-‐total correlation, the questions with the lowest corrected item-‐total correlations can be removed, because this means that these questions measure other things than the questions that have a higher correct item-‐total correlation. First the questions under ,150 are removed, this are three: at_imp, at_uhand and at_lack. After removing these three questions the Cronbach’s Alpha is ,684, see appendix 2.2. When looking at the column Cronbach's Alpha if item deleted, there is no higher Cronbach’s Alpha possible when removing another question. This means that the Cronbach’s Alpha for buyer-‐supplier relationship performance is ,684.
The number of items of thirteen is high; therefore a Factor-‐Analysis is performed. See the total output of this Factor-‐Analysis in appendix 2.2. There is correlations between the thirteen items, but none is higher than,9. And therefor the Factor-‐Analysis is carried out. The KMO is ,668, which is higher than ,5. Therefor the Factor-‐Analysis is carried out. There are four factors with an “eigenvalue” higher than 1. These factors explain more than one single item. When looking at factor 1, this factor is good for twelve of the thirteen items. This means that it is better to remove item At_u_per, because this one doesn’t actually belong to factor 1.
4.1.3 Reliability relational trust
at_expl and at_withi. After removing these three questions the Cronbach’s Alpha is ,696, see appendix 2.3. When looking at the column Cronbach's Alpha if item deleted, there will be a higher Cronbach’s Alpha when removing the item at_handi. This means that the Cronbach’s Alpha for relational trust is ,704.
The number of items of four is not high, but for relational trust the Factor-‐Analyses is also performed. See the total output of this Factor-‐Analysis in appendix 2.3. There is correlations between the four items, but none is higher than,9. And therefor the Factor-‐Analysis is carried out. The KMO is ,689, which is higher than ,5. Therefor the Factor-‐Analysis is carried out. There is one factor with an “eigenvalue” higher than 1. This factor explains more than one single item. When looking at factor 1, this factor is good for all four the items.
4.1.4 Reliability relational commitment
There is looked at the Cronbach’s Alpha of the seven questions (see appendix 1) associated with the variable relational commitment to test of the reliability of the questions which are used are good to test the variable relational commitment. Testing reliability with these seven questions gives a Cronbach’s Alpha of ,628, see appendix 2.4. This is not above .7 and therefor there is tested to see when removing some questions the Cronbach’s Alpha becomes higher. By looking at the corrected item-‐total correlation, the questions with the lowest corrected item-‐total correlations can be removed, because this mean that these questions measure other things than the questions which have a higher correct item-‐total correlation. First the questions under ,300 are removed, this are two: at_not and at_w_rel, because the other questions are above ,300. After removing these two questions the Cronbach’s Alpha is ,716, see appendix 2.4. When looking at the column Cronbach's Alpha if item deleted, there is no higher Cronbach’s Alpha possible when removing another question. This means that the Cronbach’s Alpha for relational commitment is ,716.
