1 Diversity, work outcomes and its influences: the role of communication explored
Master’s Thesis Tom van Lonkhuijzen 10587780
With an ever more culturally diverse workforce on the rise, several research gaps still exist in the academic field surrounding diversity. This research focuses on the question “To what extent does communication mediate the relationship between diversity and work outcomes?” to fill some of these gaps. Specifically, focus was given to the content of communication, in what way diversity influences it and how it influences affective work outcomes. A survey was distributed among members of the Dutch workforce through social media, garnering over seven hundred responses. Analysis showed that not diversity, but diversity climate was the most significant predictor for communication types and work outcomes, changing
expectations in the ‘diversity-equation’. Communication itself was proven to be a significant predictor for affective work outcomes, but was not found to mediate the already weak relationship between diversity and those same outcomes. The research shows that not the level of diversity in a workgroup, but the climate inside of that workgroup is what influences its communication and outcomes.
June 29th 2018 dr. P.G.A. (Pernill) van der Rijt Graduate School of Communication Master’s programme Communication Science University of Amsterdam 2017-2018
2 Introduction
As workforces around the world grow evermore culturally diverse, the challenge of managing a group of employees with different backgrounds has proven a significant one. Employees from all over the globe bring in new experience, knowledge and skills, but also differ in their opinions, preferences and ways of working (e.g. Knippenberg & Schippers, 2007; Williams & O’Reilly, 1998). This diversity can lead to both positive and negative results, as proven countless times in academic literature. On the one hand, a clashing set of ideals, morals and opinions can lead to a decrease in effective communication, firm performance and team cohesion (Sacco & Schmitt, 2005), or even an increase in dissatisfaction, team conflict and resignation (Fiske, 1998; Hofhuis, van der Zee & Otten, 2014; Feldman, 1981). On the flip side, however, the mix of different types of experience, thoughts and skills also has the ability to improve workgroup knowledge sharing (De Dreu & West, 2001), quality of work (Nakui, Paulus & van der Zee, 2011) and flexibility (Van Knippenberg, de Dreu & Homan, 2004).
Communication has been found as an important foundation for many of the work outcomes that are being researched in diversity research. For example, communication enriches employees’ social networks, improves upon job satisfaction and heightens employee retention rates (Chao, O'Leary-Kelly, Wolf, Klein & Gardner, 1994; Feij, 2003; Katz, 1988). Even more so, negative effects such as stress, negative work attitudes and increased turnover rates have been found to stem from poor communication within the workforce (e.g. Louis, Posner & Powell, 1983; Nelson, 1987). Outcomes like these, combined with their own research, caused Dinsbach, Feij and de Vries (2007) to state that employees can only
become effective organizational members through proper communication with their peers. To do this, they have to maintain a healthy social network at work, as well as constantly update themselves on information about their work, colleagues and the company. Later on, when discussing the relationship between diversity and communication, they add that employees from minority backgrounds are likely to find it more difficult to remain up to date with the different facets of their work – and thus to properly function in the company.
This, however, still does not tell us much about the actual role of diversity in the workforce when it comes to communication. Being a minority member inside of a highly diverse workgroup could have a far different outcome on communication, when compared to that same minority member in a highly homogenous workgroup. Research into this topic is scarce, with many academics shying away from zooming in on communication to find out what influence diversity has on it – or what influence it has on work outcomes.
3 As such, the current paper will look at the effect of diversity on communication in workgroups - not just looking at communication as a simple concept, but actually zooming in on the content of it. This follows up on the aforementioned research by Dinsbach et al. (2007) and also on work by Hofhuis, van der Rijt and Vlug (2016), who found that a diversity climate influences communication. However, in that research, they stayed short of researching the actual diversity in the workgroup and of the content of communication. Combining the work of these two researches will fill a gap in the academic field, where an in-depth look into the communicative results of diversity has been absent thus far.
Moreover, the present research will go further than just filling this gap, by expanding on previous literature in the academic field surrounding diversity. Not only will the effects of diversity on the content of communication be researched, but the effects this has on work outcomes will also be researched. By doing this, it will be possible to see what effects diversity has on communication and how the effect diversity has on communication
influences these work outcomes. Furthermore, it will be analyzed in what way managers can affect the influence of diversity on communication, by including diversity climate. This pro-diversity milieu has been proven to not only be influenceable by managers, but also to improve upon work outcomes (e.g. Hofhuis et al., 2016; Gonzalez & Denisi, 2009).
All of the previously discussed concepts lead to the following research question: “To what extent does communication mediate the relationship between diversity and work outcomes?”. The concepts will be further explored in the theoretical background and methodology, where a conceptual model of the proposed relationships will also be presented.
Filling this gap provides great value to both the managerial practice of managing diverse workgroups, as well as the academic field researching diversity. For the first time, an answer can be given as to what effect diversity has on communication in the workgroup and what work outcomes this creates. By including diversity climate into this equation, an answer can be given as to whether managers are able to influence the communication in their
workgroups by fostering such a climate. This gives managers practical tools and knowledge to improve upon work outcomes in diverse workgroups.
Theoretical framework
In the following section, the previous research into the field of diversity management will be discussed in depth. First, the state of the art in diversity research will be discussed, before transitioning to communication in the workgroup and its relationship with affective work
4 outcomes. Finally, diversity climate will be discussed. Hypotheses based on the prior
academic work will be presented, as well as the conceptual model that will be leading in the present research.
State of the art: Diversity research
Over the years, diversity in the workforce has been defined and operationalized in various ways by a wide variety of scholars, stemming from multiple schools of thought. After all, any and all differences between workgroup members can fall under the denominator of diversity. One of the more focused definitions of diversity in the workforce comes from a 2007 study by van Knippenberg and Schippers, stating simply that diversity is “a characteristic of social grouping that reflects the degree to which objective or subjective differences exist between group members” (p. 516).
They go on to specify objective differences as, for example, age, gender and ethnicity. These objective differences are, thus, differences that we can observe fairly easy. They are surface level traits that all of us possess and are aware of. As they are very observable and closely linked to our identity, these are studied far more often than subjective differences (Van Knippenberg & Schippers, 2007). These subjective differences are less easily observed and are often more latent than the objective differences. For example, one can think of things like personality types, attitudes, norms and values (Bowers, Pharmer & Salas, 2000). Compared to their ‘objective’ counterparts, they are more complicated and implicit nature (Harrison, Price & Bell, 1998; Jehn, Northcraft & Neale, 1999). In the present paper, the tradition of focusing on objective differences over subjective differences will be followed.
Specifically, the focus in this research will be on the ethnic background of the participants. This choice was made as it is in line with the research that the present paper is based on. In their 2007 research, Dinsbach, Feij and de Vries specifically look at the differences between minority and majority members, based on their ethnicity. Additionally, in the research by Hofhuis, van der Rijt and Vlug (2016), the focus in terms of diversity climate is also on cultural differences, rather than on other objective differences.
Information/decision-making and social categorization
As previously discussed, both negative and positive outcomes are found when researching the effects of diversity in the workforce. This can be traced back to two schools of thought: ‘information/decision-making’ and ‘social categorization’. These schools of thought were
5 already prominently present in the 1998 academic work by Williams and O’Reilly, forming a large part of their forty year literature overview of diversity research.
The former viewpoint, information/decision-making, states that a more heterogeneous
workgroup will provide better performance than a more homogeneous workgroup. According to Williams and O’Reilly (1998) this is due to the fact that “variance in group composition can have a direct positive impact through the increase in the skills, abilities, information and knowledge that diversity brings, independent of what happens in the group process” (p. 87). As for evidence supporting this theory, many scholars have found these positive effects of diversity on performance outcomes. For example, a more diverse team has been linked to more effective decision making and knowledge sharing (De Dreu & West, 2001), quality of work (Nakui et al., 2011) and flexibility (Van Knippenberg et al. 2004).
The latter viewpoint, social categorization, is based on research from the mid 80’s (Turner, 1985; Turner, Hogg, Oakes, Reicher & Wetherell, 1987), pertaining to the way we connect with people who are from different backgrounds. Simply put, we will categorize others based on our own traits, such as ethnicity or age, and attempt to connect to those that are most similar to us. This quickly creates an ‘in-group and an ‘out’-group - ‘us’ and ‘them’. This links back to the Social Identity Theory by Tajfel and Turner (1986) as well: we want to identify with this in-group and categorizing ourselves in such a group makes us feel good. As Ashforth, Harrison and Corley (2008) put it, social identity is “that part of an individual’s self-concept which derives from his knowledge of his membership of a social group (or groups) together with the value and emotional significance attached to that membership” (p. 327).
This social categorization and sense of belonging to an ‘in-group’ (and thus seeing others that are different as an ‘out-group’) is a challenge in diverse workgroups. For example, research by Goldberg (2005) showed that recruiters showed favoritism towards ‘racially similar’ applicants, over their ‘racially dissimilar’ peers. This same result was found at the end of the 20th century by many researchers (e.g. Judge & Ferris, 1993; Tsui & O’Reilly, 1989), with a paper by Tsui, Xin and Egan (1995) labeling ethnic dissimilarity ‘disruptive’ for the work relationship. In more recent research, ethnic diversity has been found to lower feelings of inclusion, job satisfaction and sense of well-being (Jansen et al., 2014; Barak & Levin, 2002).
It is clear, then, that diversity can have both positive and negative outcomes all the same. Both points of view have provided a significant amount of empirical evidence that diversity in the workforce can have both beneficial and harmful consequences. In their 2004 article, van
6 Knippenberg et al. observe this same contradiction and ultimately propose that as both outcomes can occur, focus should be on the contingencies of categorization, intergroup bias and elaboration to predict the effect diversity will have on the work outcomes. This is another reason why in the present research, focus will not just be on a possible effect diversity, directly, has on a work outcomes, but more so on what happens in between that process.
The work outcomes that will be studied will be so called ‘affective’ work outcomes. This means they aren’t easily captured in numbers, such as profit, revenue and work efficiency, but are intangibles the employees possess and develop. These are a foundation for the performance of workers and, consequently, that of the company. In line with previous research, focus will be on job satisfaction and workgroup identification.
Job satisfaction is a concept that describes the level to which the employee is happy at his current work, summarizing happiness with the work itself, as well as with colleagues and the company at large. In diversity literature, it is quite a common variable to come across, as it is a popular way to measure happiness of the employee (e.g. Jansen et al., 2014; Chao et al., 1994). It is most often found that when measuring a direct effect, diversity will have a
negative influence on job satisfaction (e.g. Barak & Levin, 2002; Fiske, 1998). This direct effect is often later explained through other variables, but in line with the standard in research, a direct effect will also be studied here. The predicted effect is as follows:
H1a: A higher diversity will predict a lower job satisfaction.
Very similar results are found when looking at workgroup identification, the second affective work outcome that will be researched. Whereas job satisfaction looks at the employee’s happiness with his job as a whole, workgroup identification looks at his/her relationship with peers in the same workgroup as the employee. This way, the relationship with close
colleagues can be measured and assessed. Similarly to job satisfaction, most research finds a negative direct effect of diversity on workgroup identification (e.g. Sacco & Schmitt, 2005). As such, the following is predicted:
H1b: A higher diversity will predict a lower workgroup identification.
In the present research, we will be looking at two factors that have proven to be pivotal for the outcomes job satisfaction and workgroup identification: work-related and personal-related communication.
7 Diversity and communication
Over the past decades, the study of communication in the workplace has taken an important role in the social sciences. From the pivotal Feldman studies in the late 70’s and early 80’s (e.g. Feldman 1976; Feldman, 1981) and the late 20th century research into the importance of social networks in the workplace (e.g. Chao et al., 1994; Fisher, 1986; Thomas &
Anderson, 1999), to the still developing body of work into online communication between co-workers (e.g. Haas, Criscuolo & George, 2015; Treem, Dailey & Pierce, 2015). It is all the more grievous too then, with this hefty body of literature surrounding communication, that it is largely ignored in research into diversity. Communication is often mentioned as an off-hand effect that can, in potential, influence the relationship people have with work, but is seldom put on the forefront as a relevant variable in the diversity-equation.
This is surprising, as the relationship between diversity and communication has been proven on many occasions before. Dreher & Cox (1996) proved that in a diverse workgroup, minority employees had a lower chance of developing mentoring relationships, because of a lower quality of communication. This resulted in lower career opportunities and lower job
satisfaction later in life. The cause the researchers attribute this to stems from the similarity-attraction theory (Berscheid & Walster, 1969; Byrne, 1971) – ‘birds of a feather, flock together’. The theory simply holds that being alike with someone makes us automatically more attracted to them, similarly to the social categorization theory that has been discussed earlier.
This is a trend that returns throughout most of the research into the relationship between diversity and communication in the workplace. Empirical evidence has been found for the fact that in diverse workgroups, insufficient communication between the minority and majority groups is a (partial) cause for minority group members to have lower wages for similar jobs (Stroh, Brett & Reilly, 1992; Parks-Yancy, 2006), lower chances of promotion (McGuire, 2000) and lower job satisfaction (e.g. Morrison & Von Glinow, 1990).
The content of communication
The most relevant findings for the present research pertaining to communication, stem from research by Dinsbach, Feij and De Vries (2007). Instead of using communication as a concept in passing by, they completely focused on it and its relationship with diversity. The content of the communication was of particular essence in their research and of great value here. They describe it as: “the content aspect of communication refers to the transmission of
8 literal information, i.e., the factual words that are being pronounced in a conversation, speech or text” (p. 726). Put simply: the content of communication refers to the actual things said, instead of the way or context they are said in. In terms of this content, they make a distinction between work-related communication (consisting of role-, task- and organization-related communication) and person-related communication. Work-related communication thus pertains to any and all forms of communication at work about work, whether it be about the company at large or specific tasks an employee has. The person-related communication here, then, pertains to all communication at work that the employee has that is about life outside of work. This can be about a broad range of topics, from ‘the soccer matches over the weekend’ to family situations, hobbies or other topics.
In their research, Dinsbach et al. (2007), set out to find differences in the content of
communication between minority and majority employees. They find that relatively, minority employees have a higher amount of work-related communication and a lower amount of person-related communication. These results tell us that minority members a) do not
experience large problems in the interaction with colleagues about work-related topics, but b) do have trouble maintaining personal relations at work through person-related
communication.
Drawing from this work by Dinsbach, Feij and De Vries (2007), it is expected that in a diverse workgroup, the communication content will mainly be focused on work-related
communication and less so on person-related communication, when compared to a less diverse workgroup. This is expected as less diverse workgroups exist out of a larger number of majority employees, meaning that they are likely to have a higher amount of person-related communication. More heterogeneous workgroups naturally contain a multitude of minority members, who have more problems communicating about person-related topics. As such, more diverse workgroups are expected to increase their amount of communication about work and see a decrease in communication about personal topics.
H2a: A higher diversity will predict a higher amount of work-related communication. H2b: A higher diversity will predict a lower amount of person-related communication.
The present field of literature, as previously discussed, lacks proper research about the role specific communication content can have on affective work outcomes, such as job
satisfaction and workgroup identification. Thus, expectations can only be based on research that looks at the role communication at large plays in the development of employees’
9 Looking at this wide field of research, it’s clear that communication in any but all cases has a positive influence on work outcomes, especially on affective ones. For example, it has often been found that proper communication leads to both mentoring relationships and friendship on the workfloor, as well as increased opinions of others (McGuire, 2000; Stroh et al., 1992; Parks-Yancy, 2006, Dreher & Cox, 1996; Reohr, 1991). When looking specifically at the affective work outcomes researched in this paper, one finds that higher amounts of
communication lead to both higher job satisfaction and higher workgroup identification. For example, Morrison and Von Glinow (1990) find that through a decrease in communication, minority members (operationalized as gender) have lower perceived workgroup identification, while majority members do have higher workgroup identification. For job satisfaction, in research by both Amos, Hu & Herrick (2005) and de Vries, van den Hooff & de Ridder (2006), it was found that improved communication within workgroups led to a higher job satisfaction. The empirical evidence found for communication positively influencing job satisfaction and workgroup identification, provides the prediction that both types of communication content will positively influence the affective work outcomes.
H3a: A higher amount of work-related communication will predict a higher job satisfaction. H3b: A higher amount of work-related communication will predict a higher workgroup identification.
H4a: A higher amount of person -related communication will predict a higher job satisfaction. H4b: A higher amount of person-related communication will predict a higher workgroup identification.
Finally, it is important to put these hypotheses in perspective in the broader diversity-equation. First, it has been predicted that a higher diversity will lead to a higher amount of work-related communication and a lower amount of person-related communication. Second, both these types of communication are expected to lead to a higher job satisfaction and a higher workgroup identification. This then creates two separate paths, where there is both a negative and a positive path leading from diversity to the affective work outcomes. First, the negative path would predict that a higher diversity negatively influences person-related communication, which in turn would have positively influenced the work outcomes.
H5a: The negative relationship between diversity and job satisfaction will be mediated by person-related communication.
H5b: The negative relationship between diversity and workgroup identification will be mediated by person-related communication.
10 Second, the positive path would predict that, since diversity increases the amount of work-related communication, a positive effect will be shown on both job satisfaction and workgroup identification. This way, diversity can both stimulate these affective work outcomes, as well as halt them.
H6a: The negative relationship between diversity and job satisfaction will be mediated by work-related communication.
H6b: The negative relationship between diversity and workgroup identification will be mediated by work-related communication.
Diversity Climate
Most of the previously described research seems to leave little room for influence by
managers, but prior research does provide a practical hope for managers aspiring to build a better performing diverse team: diversity climate. Defined by Hofhuis, van der Rijt and Vlug (2016): “Firstly, a strong diversity climate is characterized by the possibility of employees to freely discuss their cultural heritage and display cultural behaviors in the workplace.
Secondly, diversity climate encompasses the belief that cultural differences provide added value to the team or organization, and that diversity is actively promoted” (p. 2).
In other words, a diversity climate stimulates the open discussion and expression of culture, meanwhile inspiring the belief that cultural differences benefit the group as a whole. This is much akin to the information-decision making perspective that has been discussed earlier, in which the diverse skills and backgrounds of a diverse team would cause them to produce better results. Moreover, in literature, diversity climate has not only been proven to link to higher customer satisfaction and better sales performance (McKay, Avery & Morris, 2008; McKay, Avery, Liao & Morris, 2011; Gonzalez & Denisi, 2009), but also to better
communication and decreased diversity-related conflict (e.g. Drach-Zahavy & Trogan, 2013; Gonzalez & Denisi, 2009; Hofhuis, van der Zee & Otten, 2012).
Hofhuis et al. (2016) found that a diversity climate fosters openness and trust within teams, leading to improved communication within workgroups. This effect was also found prior to their research through workgroup involvement (Hobman, Bordia & Gallois, 2004) and through team identification (Luijters, van der Zee & Otten, 2008). This means that managers
fostering a diversity climate are able to improve upon the bonds and communication between employees in diverse teams. In the present research, this is seen as a moderating variable in the relationship between diversity and the communication variables.
11 H7a: The predicted effect of a higher diversity lowering work-related communication will be weaker when the diversity climate is stronger.
H7b: The predicted effect of a higher diversity lowering person-related communication will be weaker when the diversity climate is stronger.
Presentation of conceptual model
The conceptual model presented below (Figure 1) is a tool to help visualize the ‘diversity-equation’ as discussed throughout the theoretical background. In it, all variables and hypotheses used in the present research can be found, as well as a visualization of the predicted relationships.
Figure 1. Conceptual model of the present research.
Methods
In the following section, the methods of conducting the research will be outlined, including the research design, sample and operationalization of the utilized variables.
12 To research the relationship between diversity, diversity climate, communication and
affective work outcomes, a cross-sectional survey was conducted. The choice for this
method was made because of the ability to gather information from a wide variety of people - resulting in a sample with several levels of diversity. This provides a higher external validity for the research, meaning wider generalization is possible. Additionally, benefits of the survey method are the low cost of distribution and the relatively small amount of time needed to gather data, as well as the relative ease of analyzing it.
The modality used was an online survey, meaning that participants clicked on a link and were then able to fill out the questionnaire online – through both computers and
smartphones. Distribution of the survey happened through the social media account of a national Dutch radio host, who distributed the survey at no additional cost. To ensure a higher amount of respondents, a small incentive was given out in terms of ‘goodies’ (e.g. a baseball cap and a t-shirt) to five randomly selected participants.
The questionnaire was built and distributed in a Qualtrics environment with clear University of Amsterdam-branding, so that participants were aware that the research was in no way
affiliated with the radio show. All participants were greeted with a brief explanation of the goals of the research and an informed consent message upon clicking the link to the survey. All surveys were self-reports and filled out individually in Dutch. Data was handled
anonymously and respondents who wished to participate in the ‘give-away’ for the goodies were sent to a separate survey to leave their contact details.
Sample
The population for the research entails Dutch citizens working 20 hours or more a week. Thus, only people of 18 years and older, working at least 20 hours a week, will be eligible to partake in the study. The sampling utilized for this study is a convenience sample, gathered from people who clicked the link on the social media page of the Dutch radio host.
A total of 756 respondents completely filled out the survey, of which 58 surveys were halted early automatically as the participant did not comply with the age limit or the minimum of working twenty hours a week. This left a total of 698 participants (N=698) for the data analysis.
The sample consisted of 495 men and 199 women (with 4 respondents defining themselves as ‘other’), which is skewed very heavily in favor of the men compared to a report by the
13 Dutch Central Bureau for Statistics (or for short: CBS) (Central Bureau for Statistics, 2017). This report states that the man/woman division in the workforce in The Nederlands is 52% to 48%, respectively. However, in the present research, women only represent around 28% of the participants. The average age of participants was 45 (M = 45.67, SD = 9.40) with the median very close at 47 (Mdn = 47). Most participants listed the MBO as their highest completed level of education (N=311), followed by an HBO Bachelor (N=131) and
MAVO/VMBO (N=112) – signaling that lower education than average was prevalent in the sample (Central Bureau for Statistics, 2017).
The sample consisted of almost all Dutch-born respondents: 98,3% (N=686), with some of the exceptions being Belgian (N=2), Belizean (N=1), Canadian (N=1) and Romanian (N=1). This is heavily skewed in favor of Dutch participants, as CBS notes that 21% of the workforce in The Netherlands has a non-Dutch background (Central Bureau for Statistics, 2018).
Participants had worked an average of almost twelve years for their current employer (M = 11.45, SD = 10.12), though this number was skewed by outliers; the mean tenure was 10 years (Mdn = 10). Respondents worked an average of 38 hours per week (M = 37.28, SD = 12.26), though it is to be noted here that these were not the contractual hours but the self-reported hours of work – some respondents listed up to 70 hours of work per week.
Variables
In this section, the definitions and operationalizations of the variables utilized in the present research are explained. Several examples of survey questions are given in this section – for presentation purposes these were translated to English. For the complete, Dutch, version of the questionnaire, please refer to Appendix I.
Diversity. For diversity, the previously mentioned definition by Van Knippenberg and Schippers (2007) will be utilized: “a characteristic of social grouping that reflects the degree to which objective or subjective differences exist between group members” (p. 516). In line with popular research in the field and the most relevant studies to the present one, only ethnic diversity will be measured. This will be done by letting participants select the amount of co-workers in their team from other countries (a list of all countries is provided to them for easy selection). Afterwards, Blau’s Index (Blau, 1977) will be calculated, providing a
continuous variable for the data analysis. This index provides a score between 0 and 1, indicating the chance that two people in the workgroup are different from each other – in this particular case, meaning that they will have a different ethnicity. Thus, a higher score
14 Variables measured using composite scales
The following five variables (work- and person-related communication, job satisfaction, workgroup identification and diversity climate) were all measured using composite scales – meaning that the items measuring them were added together and divided by the amount of questions to create a mean response score to the measured construct. All variables were measured using five point Likert scales, ranging from -2 (e.g. totally disagree or never) to +2 (e.g. fully agree or very often), with a neutral variable in the middle. A higher score always indicated a higher level of the measured concept for the participant.
A factor analysis using principal axis factoring (PAF) extraction was used to identify the constructs in the data. One over-arching factor analysis was used, instead of separate factor analyses. This was done to ensure that all variables measured individual constructs, ruling out overlap in the measured constructs. A direct oblimin rotation was utilized, as a correlation between the items was expected. All five variables were visible as factors in the factor
analysis, as they all had an Eigenvalue of greater than 1 (Eigenvalue factor 1 (‘Workgroup Identification’) is 8.43, factor 2 (‘Diversity Climate’) is 2.89, factor 3 (‘Job Satisfaction’) is 2.23, factor 4 (‘Work-related Communication’) is 1.86 and factor 5 (‘Person-related
Communication’) is 1.21). The specifics of each variable will be discussed in the following section, while the pattern matrix can be found in Appendix I – A (Table 1).
The Keiser-Meyer-Olkin measure of sampling adequacy verified the adequacy of the sample for the measurement of these scales, KMO = .91. Bartlett’s test of sphericity provided the confirmation for the significance of the scale, with p < 0,001. Finally, the five factors together explained 69.5% of the total variance in the original variables. This means that almost 70% of the differences we see within the constructs can be explained through the variables that were measured, which is quite high.
Work- and person-related communication. For communication types, the definition and operationalization of Dinsbach et al. (2007) was followed, in which a distinction is made between ‘work-related communication’ and ‘person-related communication’. Work-related communication is simply the communication that is actually work related – a further distinction can be made between task-, role- and organization-related content, but as an analysis of the specific effects of these is beyond the scope of this study, all three will be included and seen as a single variable.
15 Work-related communication was measured with five statements, which participants scored on the aforementioned Likert-scale. Statements were based on the research by Dinsbach et al. (2007) and research by Vora and Markoczy from 2011. An example of a statement for work-related communication is “Within my team, we advise each other about tasks we share”.
For work-related communication, all five items loaded above 0.69 on a single dimension, meaning the items all have a strong association to the concept. The reliability was high (α = 0.89) and could not be improved by excluding items from the scale.
Person-related Communication. In the definition of Dinsbach et al. (2007), person-related communication is the specific type of communication related to one’s personal issues. To measure this construct, four statements were used, which were based on research by Dinsbach et al. (2007) and Vora and Markoczy (2011). An example of one of these
statements is: “Within my team, I ask people about their private lives”. Much like work-related communication, person-related comunication proved to be a highly reliable scale (α = 0.94). All four items of this scale loaded above 0.85 on one dimension, meaning all items were entered into the composite variable.
Job Satisfaction. For job satisfaction, the ‘golden standard’ definition by Hoppock (1935) is used as the leading definition: “Any combination of psychological, physiological and
environmental circumstances that cause a person truthfully to say I am satisfied with my job”.
An example of a statement here is “I am very satisfied with my current work”. The statements utilized are based on research by Hofhuis et al. (2016). All items in the scale loaded above 0.59 on a single dimension and the reliability for the scale was good (α = 0.82), with no items that could be removed to better the scale reliability – thus, all items were used.
Workgroup Identifcation. Workgroup Identification reveals in how far the employee relates to his peers at work. For this concept, the definition by Allen & Meyer (1990) will be followed, which states that workgroup identification is the level in which the employee identifies with, is involved in, and enjoys membership in said workgroup. This is a central definition that is seen as a standard for research around affective commitment towards organization. Four statements based on research of Hofhuis et al. (2016) were used, with a five-point Likert scale for the respondents to answer on. Examples of statements are “I feel strong ties with my team” and “I experience a strong sense of belonging to my team”.
16 The factor analysis confirmed that all four items correlated strongly with the concept; the lowest item loading on a single dimension was 0.79. With a Cronbach’s Alpha of 0.93 (α = 0.93), the reliability of the scale was excellent, leaving no reason not to include all items in the scale.
Diversity Climate. The final variable, diversity climate, was defined and operationalized by Hofhuis, van der Rijt and Vlug (2016): “Firstly, a strong diversity climate is characterized by the possibility of employees to freely discuss their cultural heritage and display cultural behaviors in the workplace. Secondly, diversity climate encompasses the belief that cultural differences provide added value to the team or organization, and that diversity is actively promoted” (2016, p. 2).
Six statements were given to participants to measure to what extent they applied to their workplace. An example is “Within my team it is seen as an advantage to work with people of different cultural backgrounds”.
All but one of the statements loaded highly upon this scale; ‘There is room within my team to live and work according to my own culture’ loaded at a meager 0.25, with all others loading at 0.59 and above on one dimension. As the statement with the low factor loading was proven to be valid and reliable in previous research (Hofhuis et al., 2016), it was left in the composite scale. The Cronbach’s Alpha of 0.84 (α = 0.84) confirmed that the scale had a good
reliability, which could only be heightened slightly (α = 0.87) if the previously mentioned statement with a lower factor loading was deleted. For reasons mentioned earlier, a decision was made not to do so.
Demographic Variables Finally, all participants were asked for some demographics to get a better image of the sample, as well as to be able to use them in a correlation analysis. The measured variables were age (in years), gender, educational level, tenure (in years) and hours worked per week. A description of the sample can be found in the Methods-section and the results of the correlation analysis will be discussed in the Results-section.
Results
Correlation analysis
After all composite scales had been constructed, a correlation analysis was used to
17 means and standard deviations (where relevant) for all variables can be found, as well as the correlations and reliability scores (measured using Cronbach’s Alpha). The mean scores for job satisfaction and workgroup identification were high above the middle point, indicating high affective work outcomes for the participants. Work-related communication (M = 3.60, SD = 0.80) was more prevalent in the sample than person-related communication (M = 2.99, SD = 0.86) and on average, the diversity climate was strong (M = 3.76, SD = 0.61). Finally, the average diversity in teams the participants worked in was low (M = 0.22, SD = 0.26), with almost half of the participants (N=343) reporting a fully homogenous team.
Notably, in the correlation analysis, even though all associations between diversity and the dependent variables were significant, almost none proved to have an actual association of interest – most loaded below r = 0.10. Only workgroup identification (r = 0.11, p < 0.05) and person-related communication (r = -0.18, p < 0.01) showed weak associations with diversity, with the others falling short. More interesting were the strong associations between the communication variables, diversity climate, job satisfaction and workgroup identification. This suggests that the hypothesized relationships between them can exist.
The measured demographic variables were also entered into the correlation analysis to see whether they could be of significant influence on the proposed relationships. The variables that were entered, were age, tenure, educational level, gender and working hours (per week). Very few significant correlations were found between the demographic variables and dependent variables. Working hours per week, educational level and age all correlated significantly (p < 0.01) with person-related communication, but all of these were found to be weak correlations, loading below r = 0.20. Thus, the demographic variables were not entered into any of the further analyses.
Testing of hypotheses
The first two hypotheses stated that diversity has a negative influence on both job
satisfaction (H1a) and workgroup identification (H1b) – both hypotheses were tested using separate linear regression models. The first model, testing the influence of diversity on job satisfaction, showed diversity to indeed have a significant effect on job satisfaction, with F(1, 700) = 4.54, p < .05. However, the effect is very weak, explaining a meager 1% (R² = 0.01), b* = -0.08, t = -2.13, p < .05, 95% CI [-0.37, -0.02]. Every point the diversity increases using Blau’s index (Blau, 1977), job satisfaction decreases by 0.19 on the composite scale. In light of the result, hypothesis 1a is thus accepted. However, it should be noted that this is a very
18 weak effect. Results for all analyses have been placed in the appendix of the research; for this analysis, they are found in Appendix I – C.
The second lineair regression model, testing the influence of diversity on workgroup
identification, showed similar results, with F(1, 700) = 8.72, p < .005, predicting 1.2% in the variance of the workgroup identification scale (R² = 0.01). Once again, a weak effect was shown to exist, this time between diversity and workgroup identification, b* = -0.11, t = -2.95, p < .005, 95% CI [-0.50, -0.10]. For every increase on the diversity scale, a decrease of 0.3 was found on the workgroup identification scale – leading H1b to be accepted as well. Full results for this analysis can be found in Appendix I – D.
In the second set of hypotheses, it was predicted that diversity would predict a higher amount of work-related communication (H2a) and a lower amount of person-related communication (H2b) in the population. Once again, both were tested using linear regression models. The model for the effect of diversity on work-related communication turned out to be significant, F(1, 700) = 5.09, p < .05, but weak, with R² = 0.01, b* = -0.09, t = -2.26, p < .05, 95% CI [-0.49, -0.03]. An opposite effect of what was expected is shown to exist, with work-related communication decreasing by 0.26 on the composite scale for every point diversity went up – hypothesis 2a is thus rejected. For the results of this analysis a table is presented in
Appendix I – E.
Hypothesis 2b was tested using a linear regression, measuring the influence of diversity on person-related communication, also providing a significant model, F(1, 700) = 22.75, p < .001. The model could predict 3.1% of the variance in the amount of person-related
communication (R² = 0.03), showing a weak association between that variable and diversity, b* = -0.18, t = -4.77, p < .001, 95% CI [-0.84, -0.35]. This time, however, the predicted effect was found, as the person-related communication decreased by 0.59 for every point on the diversity scale that was added. This leads to the acceptance of H2b. All results for this analysis can be found in Appendix I – F.
Next, it was hypothesized that both work-related (H3a) and person-related (H4a)
communication would have a positive influence on job satisfaction. To test these hypotheses, two separate regression models were used, after which a single multi-linear regression model was used to determine which of the two effects was stronger.
The first model, used to test the effect of work-related communication on job satisfaction, was significant at F(1,707) = 105.75, p < .001. With R² = 0.13, the model was able to predict
19 13% of the variance in job satisfaction. There was a moderate association between the two variables, with b* = 0.36, t = 10.28, p < .001, 95% CI [0.22, 0.33]. Results showed that for every point work-related communication went up on the scale, job satisfaction rose by 0.28, meaning that H3a is accepted. The same pattern was followed for H4a, utilizing a model to measure the influence of person-related communication on job satisfaction. This model was also proven to be significant, with F(1,707) = 50.62, p < .001. This model was able to predict less of the variance in job satisfaction; 6.8% (R² = 0.07). The association between the two variables was moderate, with b* = 0.26, t = 7.11, p < .001, 95% CI [0.13, 0.24]. Similarly to the relationship between work-related communication and job satisfaction, every point that person-related communication rose with, also heightened job satisfaction – this time by 0.18. This leads us to also accept H4a. Full results for this analysis can be found in Appendix I – G and Appendix I – H, respectively.
Finally, a multi-linear regression model was used to see which of the effects proved to be stronger when measured in a single model. A model was used that had job satisfaction as a dependent variable and both work- and person-related communication as independent variables. This turned out to be significant at F(2, 706) = 58.72, p < .001. The model predicts a total of 14.3% (R² = 0.14) in the variance in job satisfaction, with both predictors providing a significant influence. Work-related communication has the strongest influence of the two, b* = 0.31, t = 7.90, p < .001, 95% CI [0.18, 0.29], and person-related communication following closely behind, b* = 0.12, t = 3.21, p < .005, 95% CI [0.03, 0.14]. With every point the work-related communication scale went up, job satisfaction rose by 0.24, and for every point the respondents answered higher on the person-related scale, job satisfaction went up by 0.09. This leads us to conclude that work-related communication has the stronger influence on job satisfaction, when compared to person-related communication in a single model. The results of this analysis can be found in Appendix I – I.
In the fourth set of hypothesis, predictions were made that workgroup identification would, much like job satisfaction, be influenced positively by work-related (H3b) and person-related (H4b) communication. Again, two single linear regression models were used, before
combining both variables into a single multi linear regression model.
First, a model measuring the effect of work-related communication on workgroup identification, was used to test H3b; results can be found in Appendix I - J. This model proved to be significant with F(1, 707) = 284.50, p < .001, predicting 29% (R² = 0.29) of the variance in workgroup identification. A moderate to strong association was found between work-related communication and workgroup identification, with b* = 0.54, t = 16.87, p < .001,
20 95% CI [0.42, 0.52]. For every point that a participant went up on the work-related
communication composite scale, they went up 0.47 on the workgroup identification composite scale - H3b is thus accepted.
Second, another model was created, this time to test the effects of person-related
communication on workgroup identification, to test H4b. This model was also significant, at F(1, 707) = 191.39, p < .001. It predicted 21% (R² = 0.21) of the variance in workgroup identification, with a moderate association between person-related communication and workgroup identification, b* = 0.46, t = 13.83, p < .001, 95% CI [0.32, 0.43]. An increase of 0.38 in the score for workgroup identification was shown for every point increase on the person-related communication scale, proving H4b. Once more, full results can be found in the appendix; Appendix I – K.
Finally, a multi-linear regression model was used, measuring the effects of work- and person-related communication on workgroup identification (for the full results, please refer to
Appendix I – L). The model turned out to be significant (F(2, 706) = 190.64, p < .001). This model predict a high amount of the variance in the workgroup identification: a total of 35.1% (R² = 0.35). Work-related communication had the strongest association with workgroup identification, b* = 0.41, t = 12.23, p < .001, 95% CI [0.30, 0.42], when compared to person-related communication, b* = 0.28, t = 8.33, p < .001, 95% CI [0.18, 0.28]. For every point added to the work-related communication scale, a total of 0.36 was added to the workgroup identification scale. For every point that the person-related communication scale went up, participants raised their score on the workgroup identification by 0.23.
The previous results show that diversity, whether weak or not, has a negative influence on job satisfaction and workgroup identification, while both work- and person-related
communication have a positive influence on these affective work outcomes. In the next hypotheses it was predicted that the negative effect of diversity on both job satisfaction and workgroup identification could be predicted through the lowered amount of person-related communication (H5a and H5b). Additionally, it was predicted that a positive path would also exist, with a positive influence of diversity on work-related communication, leading to a higher amount of job satisfaction and workgroup identification (H6a and H6b).
To test H5a and H5b, two separate regression models were tested using the PROCESS macro (Hayes, 2012) – the first for the effect of diversity on job satisfaction with person-related communication as a mediator, the second switching the outcome variable to
21 workgroup identification. Full results of these regression models can be found under
Appendix I – M and Appendix I – N.
The first model tested the effect of diversity on job satisfaction, with person-related
communication as a mediator. This model was significant (F(2, 699) = 27.38, p < .001) and predicted 7.3% (R² = 0.07) of the variance in job satisfaction. A significant interaction effect on job satisfaction was also found, b = -0.11, BCa CI [-0.17, -0.06], but this was very small, κ²=-0.05, 95% BCa CI [-0.07,-0.03]. The small size of the effect, causes a rejection of H5a. In the second model, the effect of diversity on workgroup identification was tested, using
person-related communication as a mediator. This provided another significant model (F(2, 699) = 96.73, p < .001), explaining 22% (R² = 0.22) of the variance in workgroup
identification. The interaction effect on workgroup identification was significant (b = -0.22, BCa CI [-0.33, -0.13]) and even though the effect was small (κ²=-0.08, 95% BCa CI [-0.12,-0.05], it still proves that hypothesis 5b was correct. The effect of diversity on workgroup identification is indeed mediated by person-related communication, albeit a very small negative effect.
Another set of regression models using the PROCESS macro (Hayes, 2012) was used to test H6a and H6b – results of both of these can be found under Appendix I – O and Appendix I – P, respectively. The first tested the effect of diversity on job satisfaction with work-related communication as a mediator, the second tested the effect of diversity on workgroup
identification, once again with work-related communication as a mediator.
The first model tested H6a by testing the influence of diversity on job satisfaction, with work-related communication as a mediating variable. The model was significant (F(2, 699) = 57.02, p < .05) and predicted 9% (R² = 0.09) of the variance in job satisfaction. The indirect effect was also significant, b = 0.07, BCa CI [-0.14, -0.01], but the effect was very small, with κ²=-0.03, 95% BCa CI [-0.06,-0.00]. The extremely small effect, along with the fact that the effect would in fact have been negative, causes H6a to be rejected. A second model, using workgroup identification as dependent variable, diversity as independent variable and work-related communication as a moderator, was used to test H6b. This model was also
significant, with F(2, 699) = 57.02, p < .001. It was able to predict 29% of the variance in workgroup identification (R² = 0.29). A significant interaction effect was found once more, with b = -0.12, BCa CI [-0.23, -0.02]. However, once again, the effect was very small and different from the predicted positive effect, with κ²=-0.05, 95% BCa CI [-0.09,-0.01]. This also caused a rejection of H6b.
22 The final two hypothesis pertained to the relationship between diversity, diversity climate, work-related communication and person-related communication. It was predicted that a stronger diversity climate would strengthen the positive relationship between diversity and work-related communication (H7a) and would lower the predicted negative relationship between diversity and person-related communication (H7b). Since H2a was rejected (as diversity predicted a lower amount of work-related communication) hypothesis H7a is automatically rejected, but the analysis was still used to test whether an effect existed.
The first model, testing the effect of diversity on work-related communication with diversity climate as a moderator, was significant at F(3, 698) = 24.25, p < .001, explaining 9% (R² = 0.09) of the variance in work-related communication. However, while the direct effects of diversity and diversity climate were significant, the moderation did not occur, with b* = 0.18, t = 0.06, p = 0.96, 95% CI [-0.34, 0.36]. H7a thus remains rejected. For full results of this model, please refer to Appendix I – Q. The second model tested H6b by using diversity as a the independent variable, person-related communication as the independent variable and diversity climate as a moderator. This model was also significant, with F(3, 698) = 18.97, p < .001 and R² = 0.08, explain 8% of the variance in person-related communication. Once more, both direct effects were significant, but the moderation effect was not, with b* = 0.19, t = 1.16, p = 0.25, 95% CI [-0.16, 0.61]. The second moderation hypothesis, H7b, was also rejected. Once more, please refer to Appendix I – R for the full results of the model.
Direct effects of diversity climate
Even though they were not included as hypotheses, the direct effects of diversity climate on both job satisfaction and workgroup identification were tested to see if there was any
significant effect, in line with previous research. Results of both linear regression models used can be found in Appendix I – S and Appendix I – T, respectively.
The first linear regression model was used to test the direct effect of diversity climate on job satisfaction. It was a significant model, with F(1, 707) = 60.09, p < .001, and R² = 0.08, predicting 8% of the variance in job satisfaction. There was a small association between the two variables (b* = 0.28, t = 7.75, p < .001, 95% CI [0.21, 0.36]), showing that for every score the composite scale of diversity climate went up, there was an increase of 0.28 on the scale of job satisfaction. Next, the second linear regression model tested the effect of diversity climate on workgroup identification, which was significant (F(1, 707) = 137.92, p < .001) and predicted 16% (R² = 0.16) of the variance in workgroup identification. A moderate association was found, b* = 0.40, t = 11.74, p < .001, 95% CI [0.39, 0.55], with workgroup identification
23 going up by 0.47 on the composite scale for every point that diversity climate rose. These results show that diversity climates proves a higher amount of the variance in both job
satisfaction and workgroup identification, which will be discussed in the Discussion-section of this article.
Conclusion & Discussion
In this research, several gaps in the academic literature surrounding diversity were attempted to be filled. Previously, it has been made clear that being a minority or majority member influences your communication on the workfloor, and that communication influences affective work outcomes. However, the role of diversity in the workgroup itself was still unclear.
Similarly, it has been shown that diversity climate influences several work outcomes, but in what way this differs for heterogeneous and homogeneous workgroups hasn’t been
discovered yet. Not only were steps taken towards filling these gaps, but additional steps were taken to create a new conceptual model for diversity’s influence on affective work outcomes, through communication types.
First, diversity was found to have a significant negative effect on both job satisfaction and workgroup identification, albeit with very small effect sizes. This was in line with previous research concerning both job satisfaction (Barak & Levin, 2002; Fiske, 1998) and workgroup identification (Sacco & Schmidt, 2005). Interestingly, when measuring the direct effects of a diversity climate on both of these variables, significant results were also found. Moreover, these results showed a higher association with both job satisfaction and workgroup
identification, and diversity climate was able to explain a much higher amount of the variance in the affective work outcomes. Those results are in line with previous research concerning diversity climate (Gonzalez & Denisi, 2009; Hofhuis et al., 2016) which shows diversity climate to have a direct influence on a multitude of work outcomes.
However, the current research takes new steps in this by providing the influence of both diversity and diversity climate in a single model, unearthing the fact that diversity climate clearly has a stronger influence on affective work outcomes. This could be due to the fact that diversity simply tells us how diverse a workgroup is (a simple characteristic) and diversity climate tells us something about the atmosphere within this workgroup. Both
heterogeneous and homogeneous workgroups can foster a diversity climate, which is less so about diversity itself, but more so about the views of group members towards the differences. Hofhuis, van der Rijt & Vlug (2016) were able to empirically prove that such a climate fosters openness and trust, which explains the positive associations with communication in the
24 present research, as well as the positive influence diversity climate has on the affective work outcomes.
Next, it was discussed that in previous literature, diversity has been shown to heighten the amount of work-related communication and lower the amount of person-related
communication in the workforce. However, in this research, it was found that diversity does not only lower person-related communication, but also work-related communication. In short, diversity in the workforce seems to lower the over-all communication within teams. This can be explained by looking at the research by Dinsbach et al. (2007), from which the
operationalization of communication in the present research stems. In their research, Dinsbach et al. don’t look at how diverse the team is, but whether a participant was a
minority or majority member. They found that minority members would not have much trouble discussing work-related topics, but would do so in terms of personal topics. Here, it was hypothesized that because more diverse workgroups exist out of a multitude of minority members, the amount of work-related communication would heighten and the amount of person-related communication would lower. However, the data showed that not only did these diverse workgroup discuss less personal topics, they also discussed less work-related topics ;over-all communication drops in these diverse workgroups.
A moderation test was made to see if diversity climate had an interaction effect on the relationship between diversity and both of the communication types. In prior research, diversity climate has often proven to be a significant moderator in such relationships. The moderation test, however, showed insignificant results for both person- and work-related communication. The discussion can once again be found in the literature surrounding diversity climate, in particular the paper by Hofhuis, van der Rijt and Vlug (2016). The fact that this climate fosters both openness and trust explains why perhaps not diversity, but diversity climate is of more significant influence on communication within a workgroup.
Then, it was established that prior research has shown that these two types of
communication – work-related and person-related communication, positively influence both of the affective work outcomes. Those results were replicated in this research, finding weak to moderate effects for both types of communications on both job satisfaction and workgroup identification. This matched research by Morrison and Von Glinow (1990), Amos, Hu and Herrick (2005) and de Vries, van den Hooff & de Ridder (2006), from whom the hypotheses were drawn. It proves that whether about work- or personal topics, communicating with peers leads a person to feel happier with his/her job and to feel more related to those same peers.
25 Finally a mediation test was utilized to see whether the negative effects of diversity on both job satisfaction and workgroup identification were mediated by the communication types as described earlier. All four tests came back significantly, showing that an interaction effect did exist in the relationship between diversity and the affective work outcomes, as mediated by work- and person-related communication. However, the effects in these tests were
extremely small, with only the influence of person-related communication on the relationship between diversity and workgroup identification being of slight value. Over-all this leads to the rejection of the idea that a mediation occurs, which the present research was a front-runner in researching. As diversity climate was of more significant influence, there is a distinct possibility that when looking at the ‘diversity-equation’, not diversity itself, but the atmosphere surrounding the team is the more distinctive predictor.
Limitations and future research
Some limitations to the present research may be of significant influence on the outcomes however, such as the composition of the sample, the low scores on the diversity scale in the research, the way diversity was measured and the choice of both mediating and outcome variables.
First, the composition of the sample was different from the population that was used for this research. As the population, the Dutch workforce was chosen, specifically men and women of 18 years and older, working 20 hours or more per week. The convenience sample was gathered from the audience of a popular Dutch radio show, which gave a skewed view of the population. For example, participants were on average older, lower educated and reported lower diversity in their teams than the average person in the Dutch workforce (Central Bureau for Statistics, 2017; Central Bureau for Statistics, 2018). Moreover, the distribution of men and women was very unequal, with women just representing 28,5% of the sample, compared to almost 50% in the Dutch workforce (Central Bureau for Statistics, 2017). Another important note here is that the sample was made up of almost all Dutch-born participants; 98.3%. This is heavily skewed as the actual Dutch workforce has large
communities of Moroccan, Surinamese, Antillean, Polish and Turkish born workers (Central Bureau for Statistics, 2018). Future research can improve upon this by ensuring the sample is equally distributed among men and women, as well as being representative of the
workforce.
Second, results of the diversity scale were heavily in favor of homogenous workgroups, meaning that diversity was low. Almost half of the participants reported a single nationality
26 making up for the full workforce and the average diversity was barely to a quarter of the scale. This influences the results of the research negatively, as diverse workgroups are underrepresented in the research. In future researches, this should be addressed by actively sampling in participants who work in diverse teams.
Third, diversity was measured using only the ethnical background of participants using Blau’s Index (Blau, 1977). This was in line with previous research (e.g. Van Knippenberg &
Schippers, 2007) and the trend of the globalization of the workforce, but is a limited way of measuring the construct. After all, diversity can pertain to almost any objective and subjective difference between employees. So not only could gender, age and educational level (Van Knippenberg & Schippers, 2007) have been utilized to measure diversity, but intangibles like personality types, attitudes, norms and values (Bowers et al., 2000) as well. This has been done in many papers, but not often enough have these been combined into a single research to compare different types of diversity. Future research should attempt to compare the
effects of different types of diversity in a single research.
Finally, in future research into diversity, different mediation and outcome variables can be taken into account using the model presented in the present research. In this research, affective work outcomes were utilized to represent the feelings employees had about their peers and their work in general. However, more tangible outcomes are also available, such as profit, revenue and efficiency, which can be used to capture the effects of diversity in numbers. Then, it is also possible to work with different mediating variables, which would connect more closely with these ‘tangible’ work outcomes. For example, Nakui et al. (2011) suggested that the quality of ideas would rise in a more heterogeneous workgroup, providing direction for future research.
The research question for the present research was: “To what extent does communication mediate the relationship between diversity and work outcomes?”. To answer this question simply would be to say that communication does in fact not mediate the relationship between diversity and work outcomes. However, when providing a proper answer to this question, one finds that, surprisingly, this answer has little to do with the mediation by the communication types, but has everything to do with the starting point that was chosen: diversity. This research has proven the point by van Knippenberg et al. (2004) that one should not simply divide the outcomes of diversity in workgroups into ‘positive’ and ‘negative’, but that a look should be taken at the contingencies causing the outcomes. Such as shown by Hofhuis et al. (2016), perhaps it’s not the diversity in a workgroup itself, but the climate inside such a workgroup that influences not the type of communication preferred in that group and the
27 affective work outcomes. When looking towards the future, it is apparent that managers should focus their efforts on creating a climate inside of their workgroup that fosters trust and openness and thus stimulates open communication between team members. It is not the place the team members come from that influences outcomes of their cooperation – it is the way they decide to overcome their differences.
References
Allen, N. J., & Meyer, J. P. (1990). The measurement and antecedents of affective, continuance and normative commitment to the organization. Journal of Occupational and Organizational Psychology, 63(1), 1-18. doi:10.1111/j.2044-8325.1990.tb00506.x
Amos, M. A., Hu, J., & Herrick, C. A. (2005). The impact of team building on communication and job satisfaction of nursing staff. Journal for Nurses in Professional Development, 21(1), 10-16. doi:10.1097/00124645-200501000-00003
Ashforth, B. E., Harrison, S. H., & Corley, K. G. (2008). Identification in organizations: An examination of four fundamental questions. Journal of Management, 34(3), 325-374. doi:10.1177/0149206308316059
Barak, M. E. M., & Levin, A. (2002). Outside of the corporate mainstream and excluded from the work community: A study of diversity, job satisfaction and well-being. Community, Work & Family, 5(2), 133-157. doi:10.1080/13668800220146346
Berscheid, E., & Walster, E. H. (1969). Rewards others provide: Similarity. Interpersonal Attraction, 69-91. Reading, MA: Addison-Wesley
Blau, P. M. (1977). Inequality and heterogeneity: A primitive theory of social structure (Vol. 7). New York: Free Press.
Bowers, C. A., Pharmer, J. A., & Salas, E. (2000). When member homogeneity is needed in work teams: A meta-analysis. Small Group Research, 31(3), 305-327.
doi:10.1177/104649640003100303
Byrne, D. E. (1971). The attraction paradigm (Vol. 11). Michigan, United States: Academic Press
28 Central Bureau for Statistics. (2018, May 15). Arbeidsdeelname; migratieachtergrond.
Retrieved May 18, from
https://opendata.cbs.nl/statline/#/CBS/nl/dataset/82809NED/table?dl=844D
Central Bureau for Statistics. (2018, May 15). Arbeidsdeelname; kerncijfers. Retrieved May 18, from
https://opendata.cbs.nl/statline/#/CBS/nl/dataset/82309ned/table?ts=1528135258155
Central Bureau for Statistics. (2017, June 23). Arbeidsvolume naar bedrijfstak en geslacht; nationale rekeningen. Retrieved May 18, from
https://opendata.cbs.nl/statline/#/CBS/nl/dataset/82579NED/table?dl=8467
Chao, G. T., O'Leary-Kelly, A. M., Wolf, S., Klein, H. J., & Gardner, P. D. (1994). Organizational socialization: Its content and consequences. Journal of Applied Psychology, 79(5), 730-743. doi:10.1037/0021-9010.79.5.730
De Dreu, C.K., & West, M. A. (2001). Minority dissent and team innovation: The importance of participation in decision making. Journal of Applied Psychology, 86(6), 1191-1201. doi:10.1037/0021-9010.86.6.1191
De Vries, R. E., Van den Hooff, B., & de Ridder, J. A. (2006). Explaining knowledge sharing: The role of team communication styles, job satisfaction, and performance
beliefs. Communication research, 33(2), 115-135. doi: 10.1177/0093650205285366
Dinsbach, A. A., Feij, J. A., & De Vries, R. E. (2007). The role of communication content in an ethnically diverse organization. International Journal of Intercultural Relations, 31(6), 725-745. doi:10.1016/j.ijintrel.2007.08.001
Drach-Zahavy, A., & Trogan, R. (2013). Opposites attract or attack? The moderating role of diversity climate in the team diversity–interpersonal aggression relationship. Journal of Occupational Health Psychology, 18(4), 449-457. doi:10.1037/a0033989
Dreher, G. F., & Cox Jr, T. H. (1996). Race, gender, and opportunity: a study of compensation attainment and the establishment of mentoring relationships. Journal of Applied Psychology, 81(3), 297-308. doi:10.1037/0021-9010.81.3.297
