Master thesis Business Administration-Strategy
The Relationship between Corporate Culture and Ethical Decision Making
Considering the role of proximity and individual value orientation
Student Marije Pronk
Student number 11095695
Qualification MSc. in Business Administration – Strategy Track First supervisor mw. dr. F.M. Bridoux
Second supervisor dhr. Dr. J. Stoelhorst
Version Final
2 This document is written by Student Marije Pronk who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.
The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
3
Contents
Abstract ... 5
1. Introduction ... 6
2. Literature review ... 8
2.1 Ethical decision making ... 8
2.1.1. A definition of ethical decision making ... 8
2.1.2. An ethical decision making framework ... 9
2.2. Corporate culture ... 10
2.2.1. Defining corporate culture ... 10
2.2.2. Corporate culture and ethical decision making ... 11
2.3. Proximity ... 12
2.4. Personal value orientation ... 14
2.5. Research gap and research question ... 15
3. Conceptual framework ... 16
3.1. Corporate culture ... 17
3.1.1. Different types of corporate culture ... 17
3.1.2. Different culture types and ethical decision making ... 20
3.2. Proximity as a moderator ... 22
3.3. Personal value orientation as a moderator ... 24
4. Research design ... 28
4.1. Research method ... 28
4.2. Sample ... 29
4.3. Measures ... 29
4.3.1. Ethical decision making ... 30
4.3.2. Proximity ... 30
4.3.3. Personal value orientation ... 31
4.3.4. Corporate culture ... 32
4.4. Control variables ... 32
5. Results ... 33
5.1. Internal reliability of the scales and manipulation checks ... 33
5.2. Normality check ... 34
5.3. Descriptive statistics ... 35
5.3.1. Sample descriptive ... 35
4
5.4. Testing of hypotheses ... 41
5.4.1. Linear regression analysis ... 41
5.4.2. Moderation analyses ... 43
5.5. Additional results ... 50
6. Discussion ... 51
6.1. Interpretation of results ... 51
6.1.1. Direct effect of cultures on EDM ... 51
6.1.2. Interaction effect of proximity ... 52
6.1.3. Interaction effect of personal value orientations ... 54
6.1.4. Additional results ... 57
6.2. Implications ... 58
6.2.1. Implications for theory ... 58
6.2.2. Implications for practise ... 59
6.3. Limitations and suggestions for future research... 59
7. Conclusion ... 62
References ... 63
5
Abstract
When trying to improve ethical behavior within an organization, managers often turn to the corporate culture as this is perceived as an important predictor of ethical decision making (EDM). This study contributes to the understanding of EDM by investigating its relationship with corporate culture. The direct effects of market culture and clan culture on EDM are examined as well as the moderating effects of proximity (a characteristic of the ethical dilemma), and of self-enhancement and self-transcendence (an individual characteristic). Employees with a self-enhancing value orientation are mostly concerned with their own success and gratification. Employees with a self-transcending value orientation are mostly focused on the welfare of others and universalism. Conducting an online questionnaire enabled to collect a sample of 282 responses from respondents who are working in the Dutch labour market. Results show no support for the direct effects in the model and for most moderating effects. However, a marginally significant interaction between market culture and personal value orientation indicates that self-enhancing employees are less likely to make ethical decisions than self-transcendent employees within a market culture. An important implication for managers that follows from this study is that a focus on corporate culture is unlikely to be the key to stimulate ethical behavior. Instead, managers should focus on recruiting employees with a self-transcending value orientation. This study contributes to the understanding of the concept of ethical decision making as well as providing some guidance regarding the methodological pitfalls to avoid in future research in this field.
Key words: ethical decision making, corporate culture, clan culture, market culture, personal
6
1. Introduction
Ethical behavior has been an important topic in many industries for some time now and the study of ethical decision making has made good progress (Tenbrunsel & Smith-Crowe, 2008). Unethical decisions are brought to light regularly even among organizations that were perceived as trustworthy by society, such as Volkswagen and multiple accountancy firms. The costs for organizations and society that result from unethical and often illegal decisions, made at all levels in organizations, create a serious need for further research on the topic of ethical decision making (Schwartz, 2015). In particular, it is important that organizations are designed in a way that encourages ethical decision making and discourages unethical behavior as this will reduce costs and increase performance (Wu, 2002).
Whereas the amount of research has been growing on the topic of ethical behavior, moral awareness, moral identity, moral judgement, and ethical decision making, there is still much ground to cover on these subjects (Treviño, Weaver, & Reynolds, 2006). In a review of the literature, Treviño et al. (2006) concluded that most people’s thinking about what is right and what is wrong is influenced to a large extent by other people’s behavior, rules, and laws. For ethical decision making within organizations, these contextual influences point to a potentially important role for corporate culture as the major source of beliefs, values, norms, and behaviors shared in the organization (Hartnell, Ou, & Kinikci, 2011). Yet, the influence of corporate culture on ethical decision making has not been studied sufficiently. Multiple literature reviews indicate that results have been mixed in studies that focus on the effect of corporate culture on ethical decision making (e.g. Craft, 2013; Treviño et al., 2006). Moreover, it seems likely that other factors play a role in the relationship between corporate culture and ethical decision making, as suggested by Lehnert, Park, and Signh (2015). Thus, to understand the impact of corporate culture on ethical decision making, it is crucial that different moderating variables are explored, which is done in this study.
7 According to Jones (1991), it is important to not only look at situational factors when trying to predict ethical decision making, but to also consider issue-specific characteristics. One of these characteristics is the proximity of the actor making the decision to the one(s) affected by the ethical decision that the individual makes (e.g. Mencl & May, 2009; Jones, 1991). This variable can have a considerable influence on the decisions people make in moral situations and on the way in which they are affected by situational factors. Because we know that physical distance to the people that are affected by the decision can make a substantial difference in our feelings about the subject (Milgram, 1963). Individuals can easily relate to others who are close to them, such as family or neighbouring countries, whereas they find it hard to relate to people that are more distant (Trope & Liberman, 2010). Taking into account this psychological mechanism, it will be of value to the existing literature to investigate the impact of proximity on the relationship between corporate culture and ethical decision making, as this could be part of the answer to the mixed results that are found.
Another important set of factors that influences ethical decision making is individual differences such as personality, education, and gender (O’Fallon & Butterfield, 2005). Among the many sources of individual differences, this work investigates the interactive effect of personal value orientations. Value orientation refers to human values that express motivational goals that supersede the context-specificity of a situation (Schwartz, 1994). Value orientations have already been shown to be related to ethical decision making (e.g., Cohen & Morse, 2014). However, how these individual differences interact with the environment is still unclear. Researching the impact of value orientations on the relationship between organizational culture and ethical decision making addresses this gap in the literature. Whereas previous research has focused on both contextual and individual differences as separate antecedents of ethical decision making, the interactions among
8 contextual factors and individual differences have not yet been explored extensively (Treviño et al., 2006).
By focusing on the interactive effect of organizational culture, personal value orientations, and proximity, this research makes three contributions to the literature on ethical decision making. First, it complements existing literature by providing an interactional model of situational, contextual, and personal factors to predict ethical decision making. Second, it shows what kind of methodological difficulties may arise when researching ethical behavior and it provides directions for future research on this topic. Third, it gives managers insight in the type of people that is most likely to lead to ethical decision making within their organization.
2. Literature review
In this section a short overview of the literature will be provided. First, the current state of research on ethical decision making is discussed and a definition of the concept is provided. Second, corporate culture is explained and the relationship between corporate culture and ethical decision making is investigated. Third, the concept of proximity is described and lastly the concept of personal value orientation is presented. I conclude with the gap in the literature concerning these subjects and corresponding research questions.
2.1 Ethical decision making
2.1.1. A definition of ethical decision making
Most researchers find it difficult to define ethical decision making (EDM), as it hard to describe the construct in a descriptive and not in a prescriptive way (Tenbrunsel & Smith-Crowe, 2008). Even the researchers who have defined the concept note that it is imprecise. Jones (1991, p. 367) defined an ethical decision as ‘a decision that is both legally and morally
9 acceptable to the larger community’. An alternative definition that is offered by Treviño et al. (2006) is that it consists of individual behavior that is judged according to moral norms that are generally accepted in a larger social context. It is clear that the social context of a larger group is important when looking at the concept of EDM. Treviño et al. (2006) explain that there are different approaches in defining ethical and unethical behavior. A possible approach is to call ethical any behavior that reaches a minimum norm of moral behavior. Another possible approach is to call behavior ethical only when it largely exceeds this boundary of moral behavior (i.e. whistle-blowing). A last possible approach used in the literature is to focus only on unethical behavior, such as lying, stealing and cheating (Treviño et al., 2006). In this work, the focus will be on ethical decision making as defined by Jones (1991). Furthermore, moral decision making and ethical decision making will be used interchangeably in this work, as this is often assumed to represent the same phenomenon (e.g. Cohen & Morse, 2014).
2.1.2. An ethical decision making framework
There are several approaches in the literature to the phenomenon of EDM. However, most research is based on Rest’s (1986) four stages model of EDM, which consists of the following phases: (1) awareness, (2) judgement, (3) intention, and (4) behavior. The first phase of moral awareness describes the interpretation and recognition of a moral problem in a situation. The second phase of moral judgement is an important one in the process because this is where people reason about the moral situation (Schwartz, 2015). According to Rest (1986), this phase is followed by moral intention, which comprises motivational and intentional aspects. People develop the intention to respond in a certain way to the moral situation. The last phase is concerned with the actual moral or immoral behavior of the individual. This is the actual action that follows after the intention is set. Rest (1986) argues that individuals go through
10 these four phases when making an (un)ethical decision. There is some critique on this four-phased model however, as it does not include a distinction between conscious and unconscious awareness and judgement about a moral issue (Schwartz, 2015). In an attempt to combine multiple models, Schwartz (2015) proposes an integrated ethical decision making model. This model combines the rationalist-based models (EDM based on reason) and the non-rationalist-based models (EDM based on emotion and intuition) and uses the stages model which is developed by Rest (1986). In this work I will focus on the last stage of the model, as this is the stage where one carries out the ethical or unethical behavior (Kish-Gephart, Harrison, & Treviño, 2010). So this is the stage where the actual decision making is executed.
Although much research has been conducted on the direct influences of contextual factors, issue-related factors, and individual factors on EDM (i.e. Kish-Gephart, Harrison, & Treviño, 2010), the amount of research on the interactive effects of these factors on EDM is limited (Lehnert, Park, & Signh, 2015; O’Fallon & Butterfield, 2005). While there have been some studies that investigate the interactive effect of situational factors and personal factors (e.g. Treviño, 1986; Hunt & Vitell, 1986), issue-related factors are not often included in these studies (Lehnert, Park, & Signh, 2015). Lehnert, Park, and Signh (2015) addressed this gap before by trying to combine situation, personal as well as issue-related factors in a literature review. However, empirical research is needed to investigate these interactive effects on EDM.
2.2. Corporate culture
2.2.1. Defining corporate culture
Corporate culture and its effect on organizational behavior has been investigated abundantly (Hartnell et al., 2011; Douglas et al., 2001; Butterfield, Trevin, & Weaver, 2000). It is known
11 that it is an important and fundamental characteristic of organizations that influences organizational, group and individual behavior (Hartnell et al., 2011). According to Treviño (1986, p.611) culture can be defined as ‘the common set of assumptions, values, and beliefs shared by organizational members’. Cameron and Quinn (2006, p.17) see culture as ‘a reflection of what is valued, the dominant leadership styles, the language and symbols, the procedures and routines, and the definitions of success that make an organization unique’. Moreover, Ravasi and Schultz (2006) added that corporate culture contains an element for guidance of interpretation and action, because it defines what is appropriate for different situations. This entails that culture has a predictive element of organizational behavior and specifically seems to influence ethical behavior in an organizational context. In the process of hiring new employees, organizations will hire and attract people with values that match the corporate culture (O’Reilly, Chatman, & Caldwell, 1991). This is what enforces the culture as it is and how it becomes an important ‘force’ in many organizations (O’Reilly, Chatman, & Caldwell, 1991).
2.2.2. Corporate culture and ethical decision making
Corporate culture is an important factor to investigate in relation to EDM, because it can not only encourage unethical behavior, but also legitimize unethical behavior (Treviño, Butterfield, & McCabe, 1998). Findings on corporate culture often support managerial beliefs that managing corporate culture contributes to managing organizational ethics (Loe, Farrel, & Mansfield, 2000). However, other researchers state that more research is needed to investigate the influence of corporate culture on individual ethical behavior (Treviño et al., 2006). Elm and Nichols (1993) found no relationship between ethical climate and moral reasoning of managers in their study. In contrast, Moberg and Caldwell (2007) found a relationship between corporate culture and moral imagination of employees, with a salient ethic themed
12 corporate culture leading to more moral imagination of employees. However, they also found that this relationship was not straightforward, but influenced by the moral identity of employees. Specifically, they found that employees who had a stronger moral identity, were less affected by the corporate culture. Moreover Kish-Gephart et al. (2010) found a positive relationship between unethical culture and unethical behavior in their meta-analysis, but this effect dissapeared when other corporate culture characteristics, such as code of conduct and code enforcement, where studied simultaneously. Results about the influence of corporate culture on EDM are thus mixed (Hartnell et al., 2011; O’Fallen & Butterfield, 2005), which
makes this an interesting subject for further investigation. In a review, Lehnert et al. (2015) identified corporate culture as one of the factors that need further investigation in relation to EDM. They furthermore state that not only the direct relationship between the two should be investigated, but also possible interaction effects of factors that may play a role in organizations (e.g. Lehnert al al., 2015; Treviño, den Nieuwenboer, & Kish-Gephart, 2014).
To sum up, it is unclear how corporate culture relates to ethical decision making and how this relationship is influenced by other moderating variables such as proximity of the one(s) affected by the decision and personal value orientations of the decision maker. These interaction effects may explain why results of situational factors, such as corporate culture are mixed or not significant. The following part of this literature review introduces the moderating variables this study investigates.
2.3. Proximity
The concept of proximity is interesting to study in relation to ethical decision making as this could be an issue-related characteristic that influences the relationship between corporate culture and EDM. Jones (1991) introduced proximity as one of the components of moral intensity of an issue. He proposed that components of moral intensity affect the relationships
13 between the different steps of Rest’s (1986) process model of ethical decision making. Jones (1991) suggested that these components have an interactional effect on each other and are therefore often investigated as one issue-related construct (moral intensity) (Mencl & May, 2009). However, there is a need for consideration of the effects of the separate components, as these can have different (interactive) effects with situational and individual factors on EDM (Mecl & May, 2009). Although there is significant theoretical support for the influence of proximity on EDM, the concept is often ignored in research (Hunt & Vitell, 1986).
Proximity of the moral issue is defined by Jones (1991, p. 376) as ‘the feeling of nearness (social, cultural, psychological, or physical) that the moral agent has for victims (or beneficiaries) of the act in question’. It has been showed that social, cultural, psychological, as well as physical proximity play an important role in the perceived intensity of a moral issue (Jones, 1991). Ethical decision making is a complex process in which decisions are based on evaluations about what is ‘right’ and what is ‘wrong’ (moral judgement), followed by moral motivation or intention of an individual, that can lead to ethical decisions. To make judgements about whether a decision is right or wrong and to what extent this is true, people look at different factors (O’Fallon & Butterfield, 2005).
Jones (1991) and Barnett (2001) both argue that proximity is an important predictor of ethical decision making, as people care more about people who are close to them than about people who are distant. Barnett (2001) found that a high proximity score leads to less intention to engage in unethical behavior. This seems in accordance with the literature about people’s feelings and imaginative capacity for others that are close to them (Milgram, 1963; Trope & Liberman, 2010). It is unclear however how individuals make their decisions when proximity is low. Moreover, results of studies that examined the relationship between proximity and ethical decision making are mixed (O’Fallon & Butterfield, 2005; Mencl & May, 2009). Barnett and Valentine (2004) for example found no relationship between
14 proximity and ethical decision making. In a study of Barnett (2001) relationships where found between proximity and ethical judgement and behavioral intentions but not in the direction that was expected. In contrast, Watley and May (2004) found support for a positive relationship between proximity and EDM in their vignette study.
Lastly, it is known from research in social psychology that cues from the environment become more important when a situation is perceived as ambiguous (Latane & Darley, 1968; Chekroun & Brauer, 2002). In the case of low proximity, people can be uncertain about the morality of a decision and will look for cues in their environment to make the decision. This seems likely, as Jones (1991) argues that people will perceive ethical decisions as less salient and vivid when proximity is low. Also Ravasi and Schultz (2006) found that corporate culture has a large role in the process of sensemaking of employees. This study will investigate whether individuals use corporate culture as a source of such environmental cues when proximity is low by investigating an interactive effect of corporate culture and proximity on EDM.
2.4. Personal value orientation
Because there are many differences between individuals who work for organizations, it is not surprising that they respond differently to culture. To capture these differences and classify the things that people find important, personal value orientations can be used. According to Schwartz (1992) people use values to justify behavior and to judge people, events and actions. The concept of values is often used to provide a better understanding of human attitudes and behavior within organizations (Schwartz, 1994). We now know that people have different value orientations and that these individual differences produce different motivations for different behavior (Schwartz, 1994). A definition of values is given by Schwartz (1994, p. 21), who states that values are ‘desirable trans-situational goals, varying in importance, that
15 serve as guiding principles in the life of a person or other social entity’. Schwartz (1994, p. 21) moreover states that ‘(1) values serve the interests of some social entity, (2) values can motivate action, giving it direction and emotional intensity, (3) values function as standards for judging and justifying action, and (4) values are acquired both through socialization to dominant group values and through the unique learning experiences of individuals’. From this definition it is clear that value orientation is a concept that is closely related to ethical behavior and more specifically to ethical decision making. Although it is confirmed by Weaver and Trevino (2001) that an ethics program is most effective when it focuses on compliance and values, the role of individual value orientation on EDM is not investigated sufficiently (Cohen & Morse, 2014).
As morality is concerned with interpersonal relations (Cohen & Morse, 2014), it is clear that not only the environment of an individual should be considered in the form of corporate culture. Personal characteristics of the individual should also be taken into account when trying to explain EDM (Douglas, Davidson, & Schwartz, 2001). That is not to say that personal characteristics or motivations can predict EDM without considering the environment. A good example is the research of Cohen, Montoya, and Insko (2006). They found that people are prone to behave in certain ways, but the environment evokes certain (different) behaviors of these individuals. This suggests that there is an interactive relationship between situational factors and personal factors in explaining ethical decision making and that personal factors should not be ignored when trying to explain EDM (Cohen & Morse, 2014).
2.5. Research gap and research question
Following the literature review, it becomes clear that there is a gap in the literature concerning ethical decision making. First of all, the relationship between corporate culture and EDM is not sufficiently investigated, as theory predicts a strong influence of corporate culture on
16 EDM, but empirical results are mixed. This asks for more research that tests this relationship by including possible moderating variables that might have an effect. Once research has identified the moderating factors influencing the impact of corporate culture on EDM, the actual effect of corporate culture on EDM will become clear and the mixed results found in previous studies can be explained by the other variables that play a role as well.
This study investigates the impact of corporate culture on ethical decision making as well as possible effect of value orientation and proximity on this relationship. In other words, the research questions driving this work are:
What is the relationship between corporate culture and ethical decision making? Is this relationship influenced by proximity and individual value orientation?
This research will add value to current ethical decision making literature by including personal and issue-related factors that may be important to consider when trying to design an organization where ethical decision making is the norm.
3. Conceptual framework
In this study, a conceptual model is developed to test the relationship between corporate culture and EDM. The type of dilemma that is being investigated in this study is one in which individuals need to compare their own individual benefits and needs with those of others. Ethical dilemmas always consist of a situation in which there is conflict and tension (Aroskar, 1980). In this study I look at a dilemma in which there is a conflict between the individual and another party; namely other individuals within the organization. Figure 1 is a representation of the conceptual model. In the following section I will elaborate on this model and provide some arguments for the proposed relationships. This will lead to the presentation of the hypotheses that are tested in this study.
17
Figure 1. Conceptual model
3.1. Corporate culture
3.1.1. Different types of corporate culture
Corporate culture is a very broad concept that can include many components of an organization. There have been many researchers trying to organize the concept by proposing different dimensions (Cameron & Quinn, 2006). Although I acknowledge that there is no perfect approach to corporate culture that includes all relevant elements and excludes all irrelevant elements, it is good to use a framework that is based on empirical research. While no framework can be comprehensive, it will provide a way to put some boundaries on and give clarity to the concept of corporate culture. A framework that is extensively investigated and supported in the literature is the competing values framework, developed by Quinn and Rohrbaugh (1983).
In this framework Quinn and Rohrbaugh (1983) make a distinction between two dimensions. These dimensions set out different effectiveness criteria that are perceived as
18 valuable by organizations. The first dimension is the dimension which focuses on flexibility versus stability as preferred effectiveness criteria by the organization. On one side of the continuum, there are criteria that focus on flexibility, discretion, and dynamism and on the other side there is a focus on stability, order, and control (Cameron & Quinn, 2006). This means that some organizations are viewed as effective when they are changing, adaptable, and organic and some are viewed as effective when they are predictable and mechanistic. The second dimension uses focus as effectiveness criteria. On one side are effectiveness criteria of internal integration, orientation, and unity and on the other side are external integration, differentiation, and rivalry. This means that some firms are viewed as effective when having an internal focus and organizational cohesion and other firms are viewed as effective when focused on interacting or competing with other entities outside their boundaries (Cameron & Quinn, 2006). Both dimensions are based on dilemmas that organizations struggle with and that are described in multiple frameworks in the literature (Quinn & Rohrbaugh, 1983).
These two dimensions form four quadrants in the form of values that people within an organization have about organizational performance (Cameron & Quinn, 2006). The four quadrants are as following: (1) clan, (2) adhocracy, (3) hierarchy, and (4) market. Because the quadrants have competing values and can be the exact opposite on the diagonals, this framework is called the competing values framework. The different quadrants explain different orientations and values of human behavior within organizations and every organization has one of these four culture types as the dominant culture (Cameron & Quinn, 2006). The clan culture has an internal focus and is flexible. The adhocracy culture has an external focus and is flexible. The hierarchy culture has an internal focus and is stable. Lastly, the market culture has an external focus and is stable.
In this research I will only study the clan and the market culture as these culture types seem to be very much in opposition to each other, as one is focused on collaboration and the
19 other on competition (e.g. Cameron & Quinn, 2006; Hartnell et al., 2011). Furthermore, these two culture types focus more on the values that are dominant within the organization, whereas the adhocracy culture and the hierarchy culture are more focused on the structural side of corporate culture, such as means to increase productivity and profits (Hartnell, Ou, & Kinicki, 2011). As such we can expect the clan and market cultures to be more closely related to employees’ decisions on ethical dilemmas than the adhocracy and hierarchy cultures.
The clan culture has many characteristics of a family type organization (Cameron & Quinn, 2006). With the underlying assumption that the organization will be most effective when employees have positive affect towards the organization, there is large investment in human resources in clan organizations (Hartnell et al., 2011). It is expected that this type of culture will lead to effective teamwork, trust, open communication, and employee commitment (Cameron & Quinn, 2006). The values that are shared within clan cultures are attachment, affiliation, collaboration, trust, and support (Hartnell et al., 2011). Furthermore, management wants to empower employees and facilitate participation, commitment and loyalty (Cameron & Quinn, 2006). The organization is held together by tradition and loyalty and people share individual problems with one and other (Cameron & Quinn, 2006).
The market culture, on the other hand, is focused on transactions in the external market (Hartnell et al., 2011). The internal organization is primarily managed through market mechanisms by using monetary incentives, which creates a strong focus on profitability (Cameron & Quinn, 2006). The values that are most important in market organizations are communication, competition, competence, and achievement (Hartnell et al., 2011). This culture is focused on results and management facilitates aggression and competition to achieve these results (Hartnell et al., 2011). It is clear that there are many oppositions between the clan culture and the market culture (Hartnell et al., 2011). This is why I propose different relationships between these culture types and the likelihood of EDM.
20
3.1.2. Different culture types and ethical decision making
Whereas an organization with a clan culture is focused on internal maintenance, flexibility, concern for people and sensitivity to customers, an organization with a market culture is focused on external positioning with a need for stability and control (Cameron & Quinn, 2006). According to Cameron and Quinn (2006), this will make the clan culture an environment that fosters morale, by empowering the employees. It can be argued that a market culture will create an environment that is opposite to this moral enhancing environment, which will in effect reduce moral behavior.
I expect the norms conveyed by the organizational culture to affect employees’ behavior. Social identity theory (Tajfel & Turner, 1979) explains individual and group behavior in an organizational context by stating that (1) individuals strive for a positive self-concept and (2) an individual’s identity partly consists of membership of social groups (Van Dick, Grojean, Christ, & Wieseke, 2006). This means that the norms and values of the group in which one has membership have an influence on the behavior and decisions that an individual makes (Penner, Dovidio, Piliavin, & Schroeder, 2005). In the context of organizations, it is clear that employees have membership in their organization. This in turn implies that the norms and values of the organization (the corporate culture) have a direct effect on the behavior of its employees. Ethical decision making is a good example of such behavior. This means that the degree of identification and feelings of membership are important when looking at ethical decision making of employees. Identification of the norms and values of the group can even lead to internalization of these values (Ashforth & Mael, 1989).
Moreover, ethical decision making, as defined in this study, can be seen as a form of cooperative behavior according to the large amount of literature on cooperative behavior which describes group members will show more cooperative behavior when they have a
21 strong identification with their group (De Cremer & Van Vugt, Social identification effects in social dilemmas, 1999). Members of a clan culture show stronger identification with their organization than members of a market culture because of the socialization process, fraternity, and pride (Kerr & Slocum, 1987). This interdependence of employees within a clan culture will thus lead to more identification, which will lead to more cooperative behavior. So it is expected that a clan culture will also lead to more EDM.
Furthermore, it is found that overt communication leads to more cooperation among members of an organization (Dawes, Van de Kragt, & Orbell, 1988). According to Dawes et al. (1988), this can be explained by the enhancement of group identity through overt discussion among the members. Another source of cooperative behavior is the amount of trust between employees (Yamagishi & Kiyonari, 2000). Trust leads to expectations of reciprocity, which enhances cooperative behavior. Because high trust between employees is seen as an important characteristic of the clan culture (Cameron & Quinn, 2006), it seems clear that there will be more cooperation in a clan culture, which means EDM is more likely to occur.
This is also found in an empirical study of Nill and Schibrowsky (2005), in which a relationship was found between corporate culture and EDM. These researchers found that participants were less likely to make unethical decisions when a culture was primed that focused on collaboration and trust. In contrast, when they primed a culture that focused on winning and performance the participants were more likely to make unethical decisions. This study provides strong indications that employees who are working in a clan culture are more likely to show ethical decision making, whereas employees working in a market culture are less likely to show ethical decision making. Following this research, it is expected that a clan culture will lead to more ethical decision making and that a market culture will lead to less ethical decision making. This leads to the following hypotheses:
22
H1a: The more clan-oriented a corporate culture is, the more likely employees are to make ethical decisions.
H1b: The more market-oriented a corporate culture is, the less likely employees are to make ethical decisions.
3.2. Proximity as a moderator
It became clear from the literature on proximity that the social, cultural, psychological, and physical distance of the people affected by a decision have a strong effect on the choice of the decision maker (e.g. Barnett, 2001; Watley & May, 2004). I argue that the higher the proximity of the decision maker to the people affected by the decision, the less the decision maker will look at the environment for information or cues to make the decision. For EDM in a working environment this means that an individual, who has to make a decision in an ethical dilemma, will look at the corporate culture for cues when proximity is low en will not ‘need’ these cues when proximity is high. When proximity is high it is likely that decision makers will empathize with the beneficiary or victim of the ethical decision (Mencl & May, 2009).
Social identity theory (Tajfel & Turner, 1979) provides a theoretical background for this argumentation. This theory from social psychology explains behavior and cognition by looking at group processes. It states that people categorize individuals and themselves into different groups (Trepte, 2006). Others can be perceived as part of the in-group (from which the perceiver is part of) or the out-group (from which the perceiver is not part of) (Trepte, 2006). It is a way to define others and to position oneself in relation to others and thus form one’s own identity (Ashforth & Mael, 1989). Individuals tend to discriminate the out-group and favour the in-group (Hogg & Terry, 2000).
23 I argue that people will be perceived as part of the in-group when proximity is high. As a result, the decision will be made in favour of the one(s) affected by the (un)ethical decision. People tend to look at the environment or culture in the context of an organization, when proximity is low, but not when proximity is high. In the case of high proximity, when the decision maker knows the one affected by the decision personally, the co-worker will be perceived as part of the in-group because they are likely to identify themselves as being part of the same (sub)group (Ashforth & Mael, 1989). This in-group perception will lead to higher consideration of the co-worker’s interests and will thus diminish the effects of environmental cues, like the culture of the organization. When the co-worker is part of the out-group, this will lead to all kinds of negative cognitions about the co-worker, as it is found that individuals have a negative view on members of an out-group in comparison to members of their in-group (Ashforth & Mael, 1989). They hold negative stereotypes and even depersonalize out-group members (Ashforth & Mael, 1989).
In the case of low proximity, when the decision maker does not know the one affected by the decision personally, the co-worker can be perceives as part of an out-group. It seems likely that individuals will focus on environmental cues when proximity is low. This means that they will behave more in line with the corporate culture of the organization. For the clan culture there will be a stronger positive relationship between clan culture and EDM when proximity is low and a weaker positive relationship when proximity is high, because corporate culture matters less in case of high proximity. The same effect will take place in organizations with a market culture. This means that the expected negative relationship between market culture and EDM will be weaker when proximity is high and stronger when proximity is low.
H2a: The positive relationship between clan culture and ethical decision making is moderated by proximity. This relationship is weaker when proximity is high.
24
H2b: The negative relationship between market culture and ethical decision making is moderated by proximity. This relationship is weaker when proximity is high.
3.3. Personal value orientation as a moderator
Schwartz (1994) developed a value theory to classify ten types of values on the basis of their motivational goals. These values are shared by people of different countries and cultures and have four higher-level values, namely (1) enhancement, (2) openness to change, (3) self-transcendence, and (4) conservation. The motivational differences between the different values are more continuous than discrete, with the opposing values being on the ends of a bipolar scale (Schwartz, 1994). Openness to change and conservation represent the first opposing scale of higher-order values. People who score high on the value of openness to change will focus on values that represent their own thoughts, action and values favouring change (Schwartz, 1994). In contrast, people who score high on the value of conservation are more focussed on stability, self-restriction, and preservation of traditional actions (Schwartz, 1994). The second scale of higher order values is the one opposing enhancement and self-transcendence, which will be used in this study.
Self-transcendence consists of values of universalism and benevolence, which are concerned with welfare of others (Schwartz, 1994). Furthermore, people who are self-transcendent are focused on reciprocity and equality. This focus on others will be stimulated in an environment where a clan culture is prevalent (Cameron & Quinn, 2006). In contrast, self-enhancement consists of values of power, achievement, and hedonism, which are focused on status, control, dominance, personal success, and one’s own gratification (Schwartz, 1994). I expect these values to interact and reinforce the effects of a market culture on ethical
25 decision making, as a market culture will legitimize the unethical decision making that will lead to personal success and gratification.
The self-activation theory (Verplanken & Holland, 2002) states that priming of environmental values leads to value-congruent behavior when these values are congruent with the self-concept. This means that attention will go to these values when they are cognitively primed and are similar to the value orientation of the individual. Thus, personal values will have an even stronger effect on behavior when they are activated by the environment. Utz (2004) provides support for this theory with her studies on the interactive effect between priming and personal values. She found that self-transcendent individuals who were primed to focus on themselves behaved more cooperatively than self-transcendent people who were not primed to focus on themselves. This theory can be applied to the priming effect of a corporate culture on an individual within an organization. I would then expect self-transcendent individuals within a clan culture to be even more likely to show EDM than self-enhancing individuals in a clan culture, as self-transcendent individuals are primed in accordance with their self-concept. Moreover, self-enhancing individuals working in a market culture organization would be even less likely to show EDM than self-transcendent individuals, as this type of culture will prime values that are in accordance with a self-enhancing value orientation making them more sensitive to go after their own (unethical) benefits. These arguments are supported by the empirical literatire on cooperation in social dilemmas.
An ethical dilemma can be considered as a form of a social or public good dilemma, as these dilemmas represent a situation where (1) the pay-off for each individual is higher when they do not cooperate than when they do, regardless of the behavior of other group members, and (2) all individuals get a lower pay-off if all members defect than when all members cooperate (Dawes, 1980). The resemblance to the ethical dilemma, as considered by this study, is obvious as there is a conflict between the individual pay-off and the pay-off for
26 others in both descriptions. This is referred to as a mixed-motive social dilemma in which the self-interest is in conflict with what is best for their relationships, others and communities (Balliet, Parks, & Joireman, 2009). The literature on social dilemmas shows much interest in the subject of individual value orientation, as much research is conducted on the influence of personal value orientation on behavior in a social dilemma situation (e.g. Dawes, 1980; Dawes, Van de Kragt, & Orbell, 1988; Balliet, Parks, & Joireman, 2009). The value orientation of an individual seems to explain individual behavior in these types of situations as it represents the weight people assign to their own outcome and to the outcome of others (Balliet, Parks, & Joireman, 2009).
Van Lange, Vugt, Meertens, and Ruiter (1998) found that the cooperative behavior of individuals with a pro-self orientation depends more on the perceived personal increase in well-being than the cooperative behavior of pro-social individuals. Pro-self individuals can be considered as self-enhancing individuals as defines by Schwartz (1994). Pro-socials have the same value orientation as self-transcendent individuals as defined by Schwartz (1994) and pro-others can be seen as self-enhancing individuals as defined by Schwartz (1994). Van Lange et al. (1998) also found that pro-social individuals, who scored high on trust in other group members, showed more cooperative behavior than pro-social individuals with low trust in others. Pro-self individuals with high and low trust showed even less cooperation. When we take into account that individuals who work in a clan culture will be likely to trust each other, as this is a characteristic of this culture type, it seems likely that self-transcendent individuals will behave even more cooperatively in this type of culture than self-enhancing individuals.
Furthermore, De Cremer and Van Vugt (1999) found support for their goal-transformation theory which states that the salience of group-membership can lead to more cooperative behavior of self-enhancing individuals. This has some implications for
27 organizations with a market culture where there is no emphasis on group membership, but rather on competition, aggression, and the external market (Cameron & Quinn, 2006). De Cremer and Van Vugt (1999) also found that group identification does not make a significant difference in cooperation for self-transcending individuals, but it does for self-enhancing individuals. Without the salience of group membership in market cultures, individuals with self-enhancing value orientations will thus be likely to show even less cooperative behavior than self-transcendent individuals. Therefore, it can be argued that the negative effect of a market culture on ethical decision making will be even stronger for self-enhancing employees than for self-transcendent employees.
Thus, I expect the influence of a clan culture on the ethical decision making of an employee to be stronger when the employee is already inclined to focus on others and wants to enhance their well-being. Moreover, I expect the negative effect of a market culture on EDM to become stronger when individuals are already inclined to make decisions to benefit themselves. Following these arguments, the hypotheses are formulated:
H3a: The relationship between clan culture and ethical decision making is moderated by personal values, in such a way that the more self-transcendent the individual is, the stronger the relationship between clan culture and ethical decision making and the more self-enhancing the individual is, the weaker the relationship.
H3b: The relationship between market culture and ethical decision making is moderated by personal values, in such a way that the more self-enhancing the individual is, the stronger the relationship between market culture and ethical decision making and the more self-transcendent the individual is, the weaker the relationship.
28
4. Research design
4.1. Research method
This research is of an explanatory nature, as the purpose of this work is to establish whether there is a relationship between the different variables of the conceptual model (Saunders, Lewis, & Thornhill, 2009). Moreover, a quantitative method and a survey strategy in combination with a vignette strategy are used in this study. According to Saunders et al. (2009), the survey strategy is a popular method as it is economical, large sample sizes are relatively easy to obtain, and information is easy to compare and understand. This leads to the possibility to infer relationships between different variables and to test a model with multiple constructs and relationships. Data will be collected by using an online questionnaire, as this fits the explanatory nature of this research (Saunders et al., 2009). A disadvantage of online questionnaires is that the response rates are often low (Fan & Yan, 2010). Advantages of online questionnaires are that populations can be reached that are difficult to reach in another way, it saves researchers much time in collecting data, and the costs are low (Wright, 2005).
A vignette study with a between-subject experimental design is used in the online survey to manipulate proximity and to measure ethical decision making. A situation is described in which the respondents need to make an ethical decision. There are two situations: one in which proximity is low and one in which proximity is high. The participants were randomly assigned to one condition or the other. All other dimensions of the issue were the same in both conditions. The random assignment of participants is an important characteristic of the vignette design as this produces experimental characteristics (Saunders et al., 2009). Another advantage of this methodology is the realism that a vignette evokes and the possibility to manipulate the construct of proximity (Aguinis & Bradley, 2014). This type of vignette design and manipulation of the levels of an independent variable (in this case a
29 moderating variable) is called a paper people study and is often used to measure intentions and behavior like organizational citizenship behavior and ethics (Aguinis & Bradley, 2014).
Data is collected in cooperation with my fellow student Sabine Silbeek, as our research models have some of the same constucts (corporate culture and EDM). To gather enough people who want to fill out the questionnaire, we combined our different scales and made one questionnaire. This hopefully increases our sample size, as we have a larger network when we combine our individual networks.
4.2. Sample
This study aims to generalize results to the working population in the Netherlands. However, since the entire population cannot participate in the research, we focus on a sample of the population which represents the population as a whole. The sampling frame includes people who are currently working for at least one month at their organization, because we can then assume they have experienced the corporate culture. We used a convenience sampling technique, which is a probability sampling in which the sampling is done in a non-random manner. We distributed the online survey in Qualtrics using the Anonymous Survey Link of Qualtrics and shared this link on social media such as Facebook and LinkedIn, since most of the people in our network are working and thus belong to the target group. Moreover, we emailed the link to (former) peers and managers. Participants only filled in the questionnaire once since this is a cross-sectional study.
4.3. Measures
To measure the different constructs of the conceptual model, scales are adopted from the literature. As these scales have already been used and validated, this produces certainty that the scales are reliable and valid (Saunders et al., 2009). In other words, the scales would
30 produce the same results when used again and they measure what we expect them to measure. Furthermore, all scales are translated in Dutch, as this assures us of the fact that everybody understands the questionnaire completely and reduces the time needed to fill out the questionnaire (see apendix A for the complete questionnaire).
4.3.1. Ethical decision making
For the vignettes that in which a situation is presented and an (un)ethical decision needs to be made, I have consulted multiple articles that have used an ethical dilemma in this manner. No adequate scenario was found in which proximity could be manipulated and that could be a recognizable situation for the entire working population. It became clear however that an ethical dilemma is a situation where an individual needs to reflect upon competing moral standards or stakeholder claims (Gino, Ayal, & Ariely, 2013). In other words, there should be a trade-off between the interest of the indivual and the interest of others/the organization. These situational characteristics are incorporated in the ethical dilemma that is created for this study. This ethical dilemma is based on the validated ethical dilemma of Akaah (1992) where people call in sick to take a day off. The ethical dilemma is presented and then the following question is asked: ‘Hoe waarschijnlijk is het dat jij je ziekmeldt?’. The answer is then given on a 7-point Likert scale (1 = very unlikely, 7 = very likely).
4.3.2. Proximity
Proximity is manipulated by presenting different versions of the ethical dilemma. A version of the vignette in which proximity is high, is presented to half of the participands and another version in which proximity is low, is presented to the other half of the participants. In the vignette of high proximity a situation is described in which the decision affects colleagues that work in the same department and with whom the respondent works together frequently.
31 In the vignette of low proximity the colleagues that are affected by the decision work in another office and the respondent does not frequently interact with them. The distribution of conditions is randomized, to measure the influence of proximity on EDM. After the vignette, a manipulation check is done by asking how close the respondent felt to his or her colleagues that have to do the extra work. With this check, I can test whether proximity was indeed manipulated in the different vignettes (Ghorbani, Liao, Çayköylü, & Chand, 2013). The answer is given on a 7-point Likert scale (1 = not close at al, 7 = very close). Furthermore a situation check is done by presenting the question ‘Hoe reëel is de beschreven situatie voor
jou?’. This is done to see whether the vignette that is presented is realistic for the respondents.
4.3.3. Personal value orientation
The Portrait Value Questionnaire (PVQ) of Schwartz (1994) is used to measure indivual value orientation. This questionnaire is designed to measure the four higher order values, two of which are used in this study, namely self-transcendence and self-enhancement. In total this questionnaire consists of 40 items, but only 17 items are used in this research. Short portraits are given of people who implicitly hold certain value orientations. After every portrait the respondent is asked to what extent he/she recognizes him- or herself in this description. Answers are given on a 6-point Likert scale ranging from ‘not like me at all’ to ‘very much like me’. The gender of the person in the portrait is manipulated in accordance with the gender of the respondent. An example of an item that measures self-enhancement is ‘Het is
voor hem belangrijk om de leiding te hebben en anderen te vertellen wat ze moeten doen. Hij wil dat mensen doen wat hij zegt’.
32
4.3.4. Corporate culture
To measure corporate culture, the Organizational Culture Assessment Instrument (OCAI) of Cameron and Quinn (2006) is used. This scale is used to assess to what extent people feel that one corporate culture is more dominant within an organization in comparison to the other corporate cultures of the competing values framwork (market, clan, hierarchy, bureaucracy). By using short scenarios of the different cultures, people need to divide 100 points among the four different cultures and indicate which statement is more relevant for their organization. There are six items in which the four different cultures are presented. In the end there is a calculation with the different scores on the different culture types to determine which culture type represents the organization best. In this study, no comparison between the four culture types is needed. Therefore, a different way of rating is used, namely a 7-point Likert scale (1 = strongly disagree, 7= strongly agree). This produces results on the different culture types that are independent of each other. This 7-point scale has been used before to measure corporate culture and exhibited acceptable levels of reliability (Dwyer, Richard, & Chadwick, 2003). In this study I only looked at the market culture and the clan culture dimensions. An example of an item that measures the clan culture is ‘De organisatie waar ik werk is zeer
persoonlijk. Het is als een familie. Mensen delen veel van zichzelf met anderen’.
4.4. Control variables
Multiple control variables are added in this study for two reasons. First, it is a way to obtain more demographic knowledge about the sample. Second, it provides the opportunity to test whether these characteristics of the sample have an effect on the relationships of the conceptual model. I took the following control variables into account: gender, age, organizational tenure, seniority, industry, and organization size.
33 Gender is taken into account as Robin and Babin (1997) found a difference between men and women in a student population in terms of their intentions to make ethical decisions. Age is a control variable in this study as it is found that there is a relationship between age category and ethics (Ruegger & King, 1992). Organizational tenure is included because it seems likely that the longer somebody is working at an organization, the more they will show ethical behavior. Ng and Feldman (2010) found this to be true for organizational citizenship behavior, which is often linked to ethical behavior within organizations. Furthermore, I included seniority because this seems to be an interesting construct to look at in relation to ethical behavior and in fact some evidence is found in the early work of Hodgkinson (1970). Also I look at the type of industry the participants work in, as there is a large debate going on whether some sectors are primed to be more ethical than others (e.g. Cohn, Fehr, & Maréchal, 2014; Schlegelmilch & Robertson, 1995). Lastly, I included organizational size in a categorical manner to see whether there is an effect on EDM. As some research shows a postitive effect between size and EDM and some shows a negative effect (Craft, 2013), this is an interesting control variable to consider.
5. Results
5.1. Internal reliability of the scales and manipulation checks
All statistical analyses are executed with SPSS. Because the scales that are used in this study are also used in previous studies, I expect the scales to be reliable. To be sure, I have checked the internal consistency by using Cronbach’s alpha. Scales are considered to be reliable when Crobach’s alpha is .70 or higher (Field, 2013).
The scale for market culture has a high reliability with a Cronbach’s alpha of .82. None of the items would considerably affect the reliability if they would be deleted and all the items have a good correlation with the total score of the scale. The clan culture scale also has
34 a high reliability. Cronbach’s alpha is .88 and all items correlate with the total score of the scale. The individual value orientation scales also have a high internal consistency with a Cronbach’s alpha of .86 for the self-enhancement scale and .77 for self-transcendence. No items are deleted as this would not contribute to the reliability of the scales.
As the measures for ethical decision making and proximity do not have multiple items, there is no reliability analysis possible using Cronbach’s alpha (Field, 2013). I will use the realism check to see whether this measure is perceived as realistic by respondents. This question was positioned after the ethical dilemma and asked the respondent to indicate to what extend he or she found the situation to be realistic. 46.5% of the respondents found the situation to be real or was neutral about this. 19.9% of the respondents found the situation to be very unreal. This means that these respondents probably found it difficult to imagine themselves in this situation. By using a one-way ANOVA it was found that there is no significant difference in the degree of realism between the two conditions (Mrealism-low = 3.12, Mrealism-high = 3.40, p = .16). This means that the scenario in one proximity condition was perceived as realistic as in the other condition.
To check whether proximity is sufficiently manipulated scores on the question how proximate the respondent felt to his or her colleagues in the dilemma are compared. After conducting a one-way ANOVA it became clear that there is no significant difference between the two conditions (Mlow = 3.64, Mhigh = 3.85, p = .12). This indicates that the manipulation did not work as anticipated. This is a limitation of the study to which I will come back in the discussion.
5.2. Normality check
A normality check has been conducted for all variables that are presented in the model (see table 1). All variables have a normal distribution except for ethical decision making (see
35 figure 2). A rule of thumb for a normal distribution is that skewness and kurtosis should be between -1 and 1. This is the case for all variables but EDM. Therefore, I have conducted a logarithmic and a root squared transformation to make EDM more normally distributed. Unfortunately this did not improve normal distribution of the EDM variable. Also, I have checked whether respondents, who were outliers on some scales, showed variance in their responses. This was the case for all these respondents so no respondents are excluded because of this.
Table 1
Skewness and kurtosis of variables
Variable Skewness Kurtosis
Clan culture -0.583 -0.574
Market culture -0.152 -0.826
Self-enhancing -0.022 -0.161
Self-transcending 0.022 -0.161
Ethical decision making 2.514 6.93
5.3. Descriptive statistics 5.3.1. Sample descriptive
The data has a sample size of 282 respondents who all work in an organization with more than three people and have been working at this organization for at least one month. Among the respondents, 192 (68%) were female and 90 (32%) were male. The average age is 35 with a range of 18 to 65 years. The respondents were randomly divided across the proximity conditions, with 145 respondents in the low proximity condition and 137 in the high proximity condition.
Of the respondents, 54 per cent works less than 36 hours per week and 27 per cent works between 36 and 40 hours per week (see table 2 for more descriptives). The majority of the sample has been working for the organization for more than 36 months (50%). The level
36 of seniority is evenly distributed with 84 Juniors, 67 Mediors, 87 Seniors, 33 very Senior, and 11 respondents of another category. 54 respondents indicated to work in the industry of business services (19%). This is immediately followed by 53 respondents who work in the education industry (19%). Other industries that are well represented by this sample are care (17%) and the ‘other’ category (17%), which respondents choose if they felt like their industry was not in the answer options. The industries of mail and telecommunication (1.1%), construction (.4%), industrial work (1.8%), and real estate (.4%) are underrepresented and therefore taken together with the ‘other’ category. The respondents mostly have positions of scientist or specialist (21%), service personnel (21%), and ‘other’ categories (22%). Positions that are very little represented by this sample are combined with the ‘other’ category, like craftsman (1%) and machine operator (.4%). No respondent is working in the military or agriculture. Mostly large organizations are represented by this sample as 37 per cent of the sample works for an organization with more than 500 employees. Moreover, 24 per cent of the sample works for an organization with 101 to 500 employees.
Table 2
Descriptives of control variables
Variable Level N %
Gender Female 192 68
Male 90 32
Work hours 1-8 hours 7 3
9-16 hours 38 13
17-24 hours 47 17
25-35 hours 59 21
36-40 hours 76 27
More than 40 hours 55 19
Tenure 1-6 months 48 17
7-12 months 32 11
13-36 months 61 22
More than 36 months 141 50
Seniority Junior 84 30
37 Senior 87 31 Very Senior 33 12 Other 11 4 Industry Construction 1 0 Finance 13 5 Trading 9 3 IT 16 6 Industrial industry 5 2 Education 53 19 Real estate 1 0 Government 22 8 Mail and Telecommunication 3 1 Transport 7 3 Business services 54 19 Healthcare 49 17 Other 49 17 Position Entrepreneur 5 2 Self-employed 5 2 Manager 35 12 Administrative position 35 12 Specialist 20 7 Scientist 59 21 Service position 59 21 Craftsman 2 1 Machine operator 1 0 Other 61 22 Amount of employees 3-5 12 4 6-10 12 4 11-20 24 9 21 - 50 30 11 51 - 100 33 12 101 - 500 68 24 More than 500 103 37 5.3.2. Correlations
All categorical variables had to be transformed into dummy variables before I could run analyses with them to test the hypotheses. For the condition of high and low proximity the dummy variable is made for 0 = low proximity and 1 = high proximity. For gender, 0 = female and 1 = male. For other control variables that have more than two categories, I have created multiple dummy variables. As mentioned before, some categories are taken together
