Diversity in research teams: its effect on performance
and the role of transformational leadership
Thesis author: Burt van Dinter Student number: s1017330
First supervisor: Prof. dr. S.M. Groeneveld Second reader: Dr. I.C.M. van der Weijden Master program: Public Administration Track: Public Management
Abstract
The purpose of this study was to examine the moderating role of transformational leadership on the relationship between gender and nationality diversity in research teams and the performance of PhD students during their PhD trajectory. Dutch
universities had a performance-oriented organizational culture for the last 25 years. To explore when and why diversity in research teams yield positive results on performance in this particular organizational setting, a theoretical framework was developed based on theories regarding diversity, transformational leadership and performance. To test the hypotheses deriving from the theoretical framework, survey data from 135 PhD students affiliated with a Dutch university were used for regression analyses. It was expected that gender and nationality diversity in research teams enhances the performance of PhD students, and that transformational leadership reinforces this relationship between diversity and performance through affecting motivation on the individual level. The results did not support the hypotheses. This study further discusses the theoretical and practical implications of these results and provides suggestions to further advance understanding of diversity effects.
1. Introduction
To deal with globalizing markets, stiffer competitions and rapid innovations, the
workforce of public organizations has becoming more diverse in terms of demographic background (e.g. race and gender) and will continue to do so (Oberfield, 2016; van Veelen & Ufkes, 2017). Universities are an example of public organizations pursuing diversity among their research teams to take advantage of the benefits a diverse workforce, such as increased innovation and creativity, which can lead to enhanced performance (Cheruvelil et al., 2014; Hicks, 2012). Universities and their research teams have become even more diverse due to the increased volume of foreign and female students (Barjak & Robinson, 2008; Goede, Belder & Jonge, 2013; Snowball & McKenna, 2017; Testa & Egan, 2014). Students and other researchers in diverse research teams have the potential to perform better compared to students and researchers in
homogeneous research teams, due to the various skills, knowledge and viewpoints diversity among team members can offer (Cheruvelil et al., 2014; Snowball & McKenna, 2017). It is however by no means certain that diversity in teams leads to positive performance outcomes (Ilgen, Hollenbeck, Johnson & Jundt, 2005, van Knippenberg, de Dreu & Homan).
With the introduction of performance-based funding in 1993 and performance agreements in 2012, the organizational culture in Dutch universities has become
performance-oriented (De Boer et al., 2015). The emphasis of this performance-oriented organizational culture lies on the outcome of universities to obtain funding from the government, such as graduated Master students and PhD students (De Boer et al., 2015). Scarcity in academic jobs, increased competitiveness between researchers for making new discoveries and publications and more project-based research funding has led to a further increase of the performance-oriented organizational culture in Dutch
universities (De Boer et al., 2015; Waaijer, Teelken, Wouters & van der Weijden, 2017). In the Netherlands, the contest for academic jobs has intensified over the last decade. The number of academic jobs did not rise as fast as the number of PhD graduates. PhD students will be even more dependent on their individual merits, as a consequence of the performance-oriented organizational culture (Waaijer et al., 2017).
The emphasis on individual merits and interests of researchers, as a result of the
performance-oriented organizational culture, can impede elaboration between research team members in Dutch universities (Nederveen Pieterse, van Knippenberg & van Dierendonck, 2013). Moreover, more diverse teams can further obstruct collaboration between researchers, as people are less likely to collaborate with others dissimilar to the self (Knippenberg, de Dreu & Homan, 2004). To utilize the added value diversity can offer and foster elaboration between team members, the role of leadership has to be taken into account (Kearney & Gebert, 2009). The most studied form of leadership in the past three decades is transformational leadership (Banks, McCauley, Gardner & Guler, 2016). Transformational leadership is associated with different positive outcomes in organizations, such as higher motivation and satisfaction among employees, but also an improved performance (Banks et al., 2016). In relation to diversity, transformational leadership seems a viable form of leadership to benefit from the opportunities a diverse workforce can offer (Ashikali & Groeneveld, 2015, Kearney & Gebert, 2009). A
transformational leader will encourage elaboration and will reduce intergroup biases by shifting attention to commonalities and shared goals, which will lead to an improvement of performance (van Knippenberg, de Dreu & Homan, 2004; Kearney & Gebert, 2009). Although both diversity and transformational leadership are popular research topics, only a few studies researched the moderating role of how transformational leadership affects the relation between diversity and performance (Ashikali & Groeneveld, 2015; Kearney & Gebert, 2009). However, diversity in research teams can yield different performance outcomes in organizational cultures that are performance-oriented. The question in this research therefore is:
Does transformational leadership moderate the relationship between diversity in research teams and performance of PhD students in a performance-oriented organizational culture?
The aim of this research is to contribute to the existing literature in two ways. First, this research will add to the literature regarding diversity by examining how diversity influences performance in a performance-oriented organizational culture. It is important to identify which underlying processes and mechanisms of diversity are beneficial or detrimental to the performance of teams operating in a
performance-oriented organizational culture. Second, this research adds to the transformational leadership literature by exploring through which mechanism transformational
leadership exerts influence on the relationship between diversity and performance. The practical importance of the research lies in examining how leadership is a means to meet individual goals and still engender full elaboration of task-relevant information between PhD students and researchers in research teams to yield better academic performances at the collective level.
The outline of this thesis is as follows. In the next section, a theoretical framework is developed of how diversity in teams affects performance in a performance-oriented organizational culture and how this is influenced by transformational leadership style by touching the motivation on the individual level. The third section presents the method and data used. In the fourth section the results are described. In the fifth and final section the results are discussed and conclusions were drawn.
2. Theoretical framework
2.1 Diversity and performance
Diversity refers to the actual or perceived differences on any attribute that differentiate one person from another (van Knippenberg et al., 2004). People can differ on various dimensions, which are almost infinite, like gender, race, age, skill level and functional background. A distinction can be made between these dimensions underlying diversity. Social category diversity contains the dimensions which are easy detectable, such as gender and nationality. Less visible dimensions, such as job-related differences (e.g. functional and educational background), lead to informational/functional diversity (van Knippenberg et al., 2004).
Diversity in work-groups can have different effects on performance. The effects are mostly explained in the literature by two perspectives: the social categorization
perspective and the information/decision-making perspective (van Knippenberg et al., 2004). From the social categorization perspective, people use similarities and
eventually leads to one’s own in-group and one or more out-groups. This social categorization process leads to intergroup bias. Intergroup bias is the more favorable behavior and attitudes toward people in the own in-group. The categorization processes and intergroup bias can lead to problematic relations among subgroups. It can lead to detrimental effects on group performance, due to a lowered group cohesion and group commitment, more conflicts within teams and a higher turnover rate (van Knippenberg et al., 2004).
From the information/decision-perspective, diversity can also have positive effects on group performance (van Knippenberg et al., 2004). More heterogeneous groups would hold a larger pool of resources than homogeneous groups. This pool of resources consists of a broader knowledge, skill and ability regarding the task at hand. Other beneficial effects of a more diverse team are the different perspectives and opinions group members own. Different perspectives and opinions can lead to reconcile
conflicting viewpoints and a more thoroughly elaboration of the task at hand. This will prevent to reach consensus on a solution while better courses of action are available. The more thoroughly elaboration of task-relevant information can also lead to more innovative ideas and solutions to complete the task. In this way diversity can have positive effects on performance (van Knippenberg et al., 2004).
Earlier research showed inconsistent findings describing the direct effect of diversity on different group outcomes, such as performance, innovation and creativity (Horwitz, 2005; Jehn & Bezrukova, 2004; Wang, Rode, Shi, Luo & Chen, 2013). Even meta-analyses were not able to give a decisive answer on the effect diversity among work-groups has on group outcomes (Bell, Villado, Lukasik, Belau & Briggs, 2011; Horwitz & Horwitz, 2007). One could argue that the positive effects of diversity lies in functional and informational diversity as described above and negative effects are linked to social category diversity (van Knippenberg et al., 2004). To explain these equivocal results of diversity on performance, van Knippenberg et al. (2004) developed the Categorization-Elaboration Model (CEM). The CEM suggests that all dimensions of diversity elicit information/decision-making processes as well as social categorization processes. Therefore, the social categorization perspective and the information/decision-making perspective are not separated entities, but interact with each other. This results in
diversity leading to intergroup biases due to social categorization processes, impeding the elaboration of task-relevant information, viewpoints and ideas (van Knippenberg et al., 2004).
To identify the conditions under which diversity exerts effect on performance, various moderating variables were examined in earlier research (Bell et al., 2011; Choi & Rainey, 2010; Joshi & Roh, 2009; van Knippenberg & Schippers, 2007).
Task-complexity is a moderating variable influencing the relationship between diversity and performance (Horwitz & Horwitz, 2007). A more complex task requires more
innovative and creative solutions than routine tasks. It is necessary for team members to combine their diverse expertise and different viewpoints and ideas to generate a
sufficient solution for the task at hand (Horwitz & Horwitz, 2007).
Task interdependence is the degree of interaction required between team members to complete a task (Stewart & Barrick, 2000). When task interdependence is high, team members need to interact more with each other to complete the task at hand. In diverse teams, high task interdependence will therefore lead to a more thoroughly elaboration of task-relevant information and sharing different viewpoints and ideas to meet the requirements of the task (Horwitz & Horwitz, 2007; van Knippenberg & Schippers, 2007).
The effects of diversity in teams on performance are also moderated by the size of the team (Horwitz & Horwitz, 2007; Roberge & van Dick, 2010). Large teams, in contrast to smaller teams, will find more difficulties regarding interaction and coordination
between team members due to a decreasing cohesion among team members. This will lead to more social categorization processes and intergroup biases, which will impede the elaboration of task-relevant information (Horwitz & Horwitz; 2007; Roberge & van Dick, 2010).
The organizational culture in Dutch universities has been performance-oriented for the last 25 years (de Boer et al., 2015; Waaijer, Teelken, Wouters & van der Weijden, 2018). Although the organizational culture of Dutch universities could be seen as a context in
which research teams conduct their research, the performance-oriented organizational could influence the relationship between diversity in research teams and performance (Guillaume, Dawson, Woods, Sacramento & West, 2013; Jehn & Bezrukova; 2004). Organizational culture as defined by Reicher and Schneiders (1990) is “a common set of shared meanings or understandings about an organization”. These shared meanings and understandings can result in strongly held values and norms about patterns of
behaviors of work groups and the organization itself. These patterns of behavior can be strongly performance-oriented (Chatman & Jehn, 1994). A performance-oriented organizational culture emphasizes individual accomplishments above collective
accomplishments (Choi & Rainey, 2010; Sarros, Cooper & Santora, 2008). The emphasis on individual accomplishments will foster social categorization by creating a greater concern about team member’s self-image. This can be detrimental to the efficient
functioning of team members, because team members putting more effort in retaining a positive self-image than the task itself (Nederveen Pieterse et al., 2013). Beside this possible detrimental effect, a performance-oriented organizational culture can also impair the possible beneficial effects of diversity by impeding collaboration. This will lead to individual produced outcomes instead of synergetic and superior cooperative outcomes (Beersma & De Dreu, 1999).
Conducting scientific research is seen as highly complex and high task interdependent (Colbeck, 2008). Over the years, PhD students at Dutch universities had to conduct their PhD research in a more diverse research team in terms of gender and nationality
(Cheruvelil et al., 2014). PhD students and their research team members will often held a degree in higher education and perform their research in the same research field. Therefore, the main focus in this research lies on the effects of social category diversity on performance (i.e. gender and nationality). As the PhD research is an ongoing process and the final product is not graded, the expected and experienced delay in months during the PhD trajectory and consideration of stopping with the PhD trajectory are performance indicators of the performance of PhD students (Castelló, Pardo, Sala-Bubaré & Suñe-Soler, 2017; Gardner, 2009).
Although the individual values and norms of PhD students can be in line with the organizational performance-oriented culture of Dutch universities, namely completing
the PhD track for individual merits (de Boer et al., 2015; Waaijer et al., 2017), students will see different opinions and viewpoints as a chance to gain a deeper and more thoroughly understanding of the PhD trajectory (Nederveen Pieterse et al., 2013). Besides getting the PhD title, students will have the opportunity to develop their competences in scientific research. To do so, students are likely to embrace different viewpoints and ideas and elaborate task-relevant information. Diversity is hereby a means to accomplish individual goals. As a result, PhD students will experience less obstruction and hurdles completing their PhD research. Therefore they are more likely to expect and experience less delay during the PhD trajectory (Gardner, 2009; Castelló, 2017).
In research teams in which students work more individually, conflict due to social categorization processes such as negative stereotypes and intergroup biases are less likely to arise than working in a larger research team. Students will only collaborate when needed and use different viewpoints and ideas for their own good (Horwitz & Horwitz, 2007). PhD students working in large research teams however are more task interdependent, due to the performance-oriented organizational culture and its performance agreements, on collaboration to complete their own research, but also to make sure their research team publish first about new discoveries (Horwitz & Horwitz, 2007; Mullen & Tuten, 2010; Waaijer et al., 2018). They are therefore to a certain extent obliged to elaborate task-relevant information, which will diminish the influence of social categorization processes. Collaborating with other members of the research team will also lower the feeling of social isolation. This means that the PhD students will experience more emotional and affective help during their PhD trajectory. The PhD students are therefore less likely to consider stopping with the PhD trajectory (Castelló, 2017). In sum, PhD students will profit from diverse research teams, even if they
conduct their research in a performance-oriented organizational culture. Therefore is expected:
H1: Diversity in research teams will have a positive effect on the performance of PhD
2.2 Diversity, transformational leadership and performance
Due to the social categorization processes underlying the effects of diversity in research teams, it is by no means certain that diversity will exert positive effects on performance (van Knippenberg et al., 2004; Kearney & Gebert, 2009). To take advantage of the broadened pool of resources diverse teams can offer, in terms of skills, ability and knowledge, team members need to learn to work together to utilize these resources. Research showed that leadership in an important factor contributing to this process and utilizing the full potential of diversity in research teams (Ashikali & Groeneveld, 2015, Choi & Rainey, 2010, Kearney & Gebert, 2009, Sarros et al., 2008).
The most studied leadership style today is transformational leadership (Wang, Oh, Courtright, Colbert, 2011). Transformational leadership is a leadership style that is based on the assumption that “leaders can change followers’ beliefs, assumptions, and behavior by appealing to the importance of collective or organizational outcomes” (Moynihan, Pandey & Wright , 2011; 147). Transformational leadership is described as an emotional bond between leader and follower and goes beyond purely rational social exchange, as is the case with transactional leadership (Moynihan et al., 2011). This leadership style is a charismatic and inspirational style with four components through which team members’ behavior can be influenced. (Ashikali & Groeneveld, 2015; Bass, Avolio, Jung & Berson, 2003). These four components are inspirational motivation, idealized influence, individualized consideration and intellectual stimulation.
Inspirational motivation is the leader’s ability to create a collective vision and mission among work group members, giving meaning and purpose to their work. Idealized influence refers to the leader being a role model, who can be trusted and respected by his or her team members. This trust and respect will enlarge the pride and confidence in working for the team and organization. By individualized consideration is meant that individual needs, such as personal grow and need to achieve, are recognized and met by the leader. The leader will foster a climate in which is room for individual growth and will act as coach or mentor. At last, intellectual stimulation is referring to a leader that is questioning known approaches, ideas and solutions to complete a task. Team members are being stimulated to come up with new ideas and perspectives to improve
beyond expectations by changing their values, beliefs and sense of belonging (Ashikali & Groeneveld, 2015; Bass et al., 2003).
Research in the past showed that transformational leadership has positive effects on individual performance (Wang et al., 2011). Characteristics of the task, team and organization will influence the relationship between transformational leadership and performance (Bass et al., 2003). Not only does transformational leadership directly affect work-related outcomes, but also indirect by shaping the organizational culture (Rainey, 2009; Sarros, Cooper & Santora, 2008). Transformational leadership can establish a culture in which development is encouraged above outcome by emphasizing empowerment, problem solving and innovation. In this developmental and learning organizational culture personal growth and diversity are higher valued (Moynihan et al., 2011).
The transformational leadership style seems to be an appropriate leadership style to utilize the positive effects and mitigate the detrimental effects diversity can have on performance (Ashikali & Groeneveld, 2015, Kearney & Gebert, 2009). Transformational leadership will lead to an increased motivation to elaborate task-relevant information. First, by intellectually stimulating team members leaders encourage them to take advantage of the diverse pool of resources and to share their own resources. Beside, a lack of intellectual stimulation turns out to be important motive to drop out of the PhD trajectory (Castelló et al., 2017). Second, by taking individual considerations into account, such as being valued for the input team members can provide and their
uniqueness (Kearney & Gebert, 2009). In line with the CEM, elaboration of task-relevant information is a positive precedent of performance. The detrimental effects of diversity on performance due to social categorization processes in teams will also diminish by fostering collective team identification. Collective team identification is the emotional binding of a person to their respective team. A transformational leader will foster collective team identification by shifting the attention from difficulties deriving from differences among team members to a collective vision and shared goals. As a result, social categorization processes due to diversity in team will be overcome. This is also a positive precedent of performance and in line with the CEM (Kearney & Gebert, 2009; van Knippenberg et al, 2004). The leader can also create a climate in which dissent is
valued and team members are willing to share their ideas (Kearney & Gebert, 2009). PhD students find themselves better socially integrated into the academic world and their research teams when valued for their input. They are therefore less likely to consider stopping with the PhD program (Castelló, 2017). By increasing collective team identification as a leader, the willingness and motivation to elaborate task-relevant information will increase and thereby performance (Kearney & Gebert, 2009).
H2: Transformational leadership reinforces the positive relationship between diversity
in research teams and the performance of PhD students.
One of the possible mediating factors of transformational leadership on the relationship between diversity and performance in a performance-oriented organizational culture is the goal orientation of team members. Goal orientations are goal preferences in
achievement contexts (Dweck, 1986; Dweck & Leggett, 1998). The primary distinction is between a learning goal orientation and a performance goal orientation (Elliot &
Mcgregor, 2001). The learning goal orientation aims at obtaining mastery of a specific task by increasing competence and developing knowledge. The performance evaluation is based on self-improvement in comparison with one’s earlier level. The performance goal orientation is focused on outperforming others to develop a sense of competence. Team members evaluate their progress and development on the task at hand by comparing their own performance with other’s performances. These two goal orientations are two separate entities and can therefore coexist (Elliot & Mcgregor, 2001). It turns out that a performance-oriented organizational culture can foster a performance goal orientation among team members (Dragoni, 2005). A performance goal orientation stimulates social categorization processes linked to diversity. Team members with a high performance goal orientation are concerned about their competence in relationship to others. Therefore they are less likely to cooperate, because they feel threatened by different perspectives. Instead of putting effort in getting a more thoroughly understanding of the task, effort is put in stereotyping others to enhance their self-image. This will lead to intergroup biases, which can be detrimental for performance output (Nederveen Pieterse et al., 2013).
Transformational leadership will however foster a learning goal orientation by research team members, conducting their research in a performance-oriented organizational culture (Ashikali & Groeneveld, 2015; Bass et al., 2003; Nederveen Pieterse et al., 2013). First, by inspirational motivation the transformational leader will create a collective sense of vision among research team members. The emphasis of a collective vision lies on elaboration of task-relevant information to produce synergetic and cooperative outcomes instead of outperforming each other. Second, by intellectual stimulating his or her PhD students, a transformational leader encourage these students to make full advantage of the broadened pool of resources in terms of skills and knowledge a diverse team has to offer. PhD students use the broadened resource pool to develop their own competences. Third, by showing individual consideration, PhD students feel more valued for their uniqueness and input. As a result, PhD students are more likely to
explore and elaborate different opinions and perspectives from dissimilar research team members to get a more thoroughly understanding of the task at hand (Ashikali &
Groeneveld, 2015; Bass et al., 2003; Kearney & Gebert, 2009; Nederveen Pieterse et al., 2013). In line with the CEM, a high learning goal orientation will exert positive effects on the relationship between diversity and performance (van Knippenberg et al., 2004). Therefore is expected:
H3: An increased learning goal orientation of a PhD student partially mediates the
moderating effect of transformational leadership on the positive relationship between diversity in research teams and the performance of PhD students.
Figure 1. Proposed conceptual model
Nationality and
gender diversity Performance of PhD students
Transformational leadership
3. Method
3.1 Sample and data collection
The data used in this research was obtained by an online survey. In 2017 there were 9319 PhD students affiliated with a Dutch university according to the ‘Vereniging van Universiteiten’ (VSNU, 2018). There were no restrictions to participate in this research, other than being enrolled in a PhD trajectory at a Dutch university. The PhD students were recruited with the help of ‘Promovendi Netwerk Nederland’. ‘Promovendi Netwerk Nederland’ is an interest organization for PhD students at a Dutch university.
‘Promovendi Netwerk Nederland’ helped recruiting participants in two ways. First, ‘Promovendi Netwerk Nederland’ offered the opportunity to present this research at their annual meeting. PhD students that made themselves available for participating were sent a link per email, which gave access to the survey. Second, ‘Promovendi
Netwerk Nederland’ helped distribute the survey among their members (approximately 1.000), also per email. A small number of participants were recruited through the
researcher’s own network. By recruiting participants through ‘Promovendi Netwerk Nederland’, a representative selection of PhD students at a Dutch university in the research sample was pursued. The final research sample consisted of 135 PhD students at Dutch universities.
Of these 135 PhD students, 65.9% were female and 34.1% were male. In 2017, 54.9% of all PhD students at a Dutch university were female (VSNU, 2018). Also, in this sample 26.7% of the participants had a non-Dutch nationality. In the academic world, 46.7% of all PhD students at a Dutch university had a Dutch nationality. Female and non-Dutch PhD students were underrepresented in this research sample. Therefore, gender and nationality of the PhD students was controlled for in the regression analyses. The Dutch Universities had to give permission to ‘Promovendi Netwerk Nederland’ to distribute the survey among their PhD students. The University of Maastricht, Tilburg University, University of Twente, Leiden University, Erasmus University Rotterdam and Utrecht University gave their permission to distribute the survey. As a result, 91.8% of the participants in the sample were from these six universities. Participants of these six
universities were overrepresented in the sample in comparison to the percentage of PhD students enrolled at these six universities in 2017 (40.8%)(VSNU, 2018). To
account for the differences in degree of the performance-oriented organizational culture in Dutch universities, organizational culture was added to the regression analyses as a control variable.
3.2 Measurements
Research team diversity. To measure research team diversity in terms of nationality and
gender, PhD students were asked to indicate the share of Non-Dutch and female colleagues in their research team (a research team consists of professors, associate professors, assistant professors and other PhD students). In large research teams, it is assumed that participants do not know the (dominant) nationality of each team
member. For this reason, a distinction between Dutch and non-Dutch team members is made, because most of the PhD students at a Dutch university are Dutch (56,9% in 2013)(Goede, Belder & Jonge, 2013). Nationality diversity is measured by the item “What is the share of non-Dutch professors and researchers (with or without PhD) in your research team?” using a seven-point scale ranging from (1) There are no non-Dutch to (7) There are only non-Dutch. The same reasoning is followed for gender diversity. Although it is easier to determine the gender of a research team member than his or her (dominant) nationality, accurate numbers regarding the distribution of men and women in large research teams will be hard to determine for participants. Therefore, the item “What is the share of female professors and researchers (with or without PhD) in your research team?” was used to measure gender diversity. Again a seven-point scale was used ranging from (1) There are no women to (7) There are only women.
Transformational leadership. Transformational leadership was measured by an adjusted
12-item survey questionnaire, derived from the research of Ashikali & Groeneveld (2015). Three items were used to measure each dimension of transformational leadership, as conceptualized by Bass et al. (2003). The items for idealized influence (TL1-TL3), inspirational motivation (TL4-TL6), intellectual stimulation (TL7-TL9) and individualized consideration (TL10-TL12) are presented in Table1. For each item a
5-point Likert scale was used: (1) strongly disagree, (2) disagree, (3) neutral, (4) agree and (5) strongly agree.
Goal orientation. The goal orientation of the PhD students was measured with an
adjustment of the 12-items questionnaire developed by Elliot and Murayama
(2008)(Table 1). Six items were used to measure learning goal orientation (LGO1-LGO6) and six items were used to measure performance goal orientation (PGO1-PGO6). The items were again measured based on a 5-point Likert scale, ranging from (1) strongly
disagree to (5) strongly agree.
Performance. As stated earlier, as PhD trajectory is an ongoing process and the final
product is not graded, expected and experienced delay during the PhD trajectory and consideration of stopping with the PhD trajectory were chosen as performance
indicators of PhD students (Castelló et al., 2017; Gardner, 2009). The item “Did you ever consider stopping with your PhD trajectory” measured consideration of stopping with the PhD trajectory. The answer options were: (1) Never, (2) Sometimes and (3)
Regularly. Expected and experienced delay in the PhD trajectory was measured on a
2-item scale: “Are you on schedule with your PhD trajectory so far” and “are you expecting delay in your PhD trajectory. The answer options were yes or no. If answered no on the item “Are you on schedule with your PhD trajectory so far”, the participant was asked how many delay in months they experienced so far. If they answered yes on the item “are you expecting delay in your PhD trajectory”, the participants were asked how much delay in months they expect for completing their PhD trajectory. The number of months of expected and experienced delay was summed up for each participant to define the expected and experienced delay performance indicator.
Control variables. In this research gender, nationality, team size, task interdependence
and organizational culture were included as control variables. Female and non-Dutch PhD students were underrepresented in the sample, so were therefore included as control variables. As described in the theoretical section, team size, task
interdependence and organizational culture can influence the relationship between diversity and performance. Team size was measured by asking: “how many team
were categorized to a 6-point scale: only myself, 2 to 5 members, 6 to 10 members, 11 to 25 members, 26 to 100 members and ≥ 101 members. Task independence was
measured by an adapted 5-item questionnaire as used in earlier research (Kearney & Gebert, 2009; van der Vegt and Janssen, 2003)(Table 1: TI1-TI5). The participants were asked to indicate how strongly they agree or disagree on the same 5-point Likert scale as described above.
Organizational culture can be seen as something that evolves naturally and is not
changed easily (Santos, Gray, Dentsen & Cooper, 2005), as is the case for Dutch Universities. The organizational culture of Dutch universities has been performance-oriented for the last 25 years (de Boer et al., 2015; Waaijer et al., 2018). Therefore, performance-oriented organizational culture is seen as context and not as variable in this research. However, the degree to which the organizational culture of Dutch universities is performance-oriented might differ per university. To control for the differences in degree of performance-oriented culture, a 4-item scale of the
Organizational Culture Profile (OCP) was used to measure the performance-oriented culture (Hartnell, Kinicki, Lambert, Fugate & Doyle Corner, 2016; O’Reilly, Chatman & Caldwell, 1991; Sarros, Gray, Densten & Cooper, 2005). The four items are listed in Table 1 (OC1-OC4) and were rated on a 5-point Likert- type scale ranging from (1) strongly
Table 1. List of items
Transformational Leadership
Idealized influence
TL 1 “My promoter or supervisor considers my needs over his or her own needs”
TL 2 “I trust my promoter or supervisor”
TL 3 “My promoter is consistent in conduct with underlying ethics, principles and values”
Inspirational motivation
TL 4 “My promotor or supervisor emphasizes the importance of having a collective sense of
mission”
TL 5 “My promotor or supervisor behaves in ways that arouses individual and team spirit”
TL 6 “My promotor or supervisor behaves in ways that motivate, by providing meaning and
challenge to employees’ work”
Intellectual stimulation
TL 7 “My promotor or supervisor stimulates to be innovative and creative by questioning
assumptions, reframing problems and approaching old situations in new ways”
TL 8 “My promotor or supervisor seeks different points of view when solving problems”
TL 9 “My promotor or supervisor suggests new ways of working and different perspectives”
Individualized consideration
TL 10 “My promotor or supervisor recognizes individual differences in terms of needs and desires”
TL 11 “My promotor or supervisor helps employees to develop their strengths”
TL 12 “My promotor or supervisor pays attention to each individual’s need for achievement and growth by acting as a coach or mentor”
Goal orientation
Learning goal orientation
LGO1 “My aim is to completely master the academic skillset presented in the PhD trajectory” LGO2 “I am striving to understand the content of this PhD trajectory as thoroughly as possible” LGO3 “My goal is to learn as much as possible”
LGO4 “My aim is to avoid learning less than I possibly could”
LGO5 “I am striving to avoid an incomplete understanding of the course material in the PhD trajectory”
LGO6 “My goal is to avoid learning less than it is possible to learn”
Performance goal orientation
PGO1 “My aim is to perform well relative to other PhD students” PGO2 “I am striving to do well compared to other PhD students” PGO3 “My goal is to perform better than other PhD students” PGO4 “My aim is to avoid doing worse than other PhD students” PGO5 “I am striving to avoid performing worse than other students” PGO6 “My goal is to avoid performing poorly compared to others” Task independence
TI1 “I need information and advice from other members of my research team to perform my
job well, I have a one-person research”
TI2 “It is not necessary for me to coordinate or cooperate with others”
TI3 “I need to collaborate with other members of my research team to perform my job well”
TI4 “Other members of my research team need information and advice from me to perform
their jobs well”
TI5 “I regularly have to communicate with other members of my research team about
work-relate issues” Organizational culture
OC1 “The organizational culture of the university is achievement oriented”
OC2 “The organizational culture of the university is demanding”
OC3 “The organizational culture of the university set high expectations”
3.3 Analytical strategy
To analyze the acquired data and test the hypotheses multiple regression analyses were conducted. A principal component analysis (PCA) was performed prior to these
regression analyses to identify the dimensions in the acquired data. While the scales used to measure the variables in the conceptual model were derived from earlier research, to make the scales appropriate for this particular research, slight adjustments were made to the scales. Therefore, an exploratory factor analysis technique was chosen to analyze the data and to construct the measurements of the main concepts. To do so, the statistical software program SPSS was used.
The PCA on transformational leadership, organizational culture, task interdependence, learning goal orientation and performance orientation items was conducted with an oblique rotation and resulted in five components with an eigenvalue greater than 1, which explained 58.0% of the variance. The Kaiser-Meyer-Olkin measure of sampling adequacy was .80 and Bartlett’s test of sphericity was significant (χ2 (528) = 2518.68, p < .05). There were no cross-loadings between the items, so variance in the data is
attributable to the components of interest. This implies that common method variance is non-significant. By common method bias is meant that the method or instrument to measure concepts causes variation in the data (Podsakoff, MacKenzie, Lee & Podsakoff, 2003). This confirms the validity of the measurement scales. Therefore, the components deriving from the PCA can be interpreted as the intended measurement scales of the variables.
The item TL9 (intellectual stimulation) was removed from the transformational
leadership scale, because this item did not meet the criteria of having a primary factor loading of .6 or above. As a result, the transformational leadership scale was constructed out of the items TL1-TL8 and TL10-TL12. From the learning goal orientation scale, the following three items were removed as a result of a factor loading below .6: LGO1, LGO2 and LGO3. So the scale of learning goal orientation was made up of the items LGO4-LGO6. The items PGO4-PGO6 were removed from the performance goal orientation scale, also due to factor loadings below the criteria of .6. As a result, the items PGO1-PGO3 were used to construct the scale of performance goal orientation.
A principal component analysis with the remaining items of transformational leadership organizational culture, task interdependence, learning goal orientation and performance orientation was conducted with an oblique rotation. The principal component analysis resulted in five components with an eigenvalue greater than 1 and explained 62.2% of the variance. There were no cross-loadings between the items and all items met the primary factor loading criteria of .6 and above.
The assumptions for performing multiple regression analyses were not violated in the regression model: the residuals were normally distributed, the model was
homoscedastic and the regression model was linear. On this basis was concluded that it was acceptable to further perform the regression analysis.
The variables team size, gender diversity and nationality diversity and consideration of stopping with the PhD trajectory were measured on a categorical scale. To conduct regression analyses with these variables, dummy variables for each category of the variables were computed. The dummy variables for team size are: team size of 1 (the students works solitarily), 2-5 team members, 6-10 team members or 11 or more team members. The gender diversity dummy variables are as follow: no gender diversity (only men or only women in the research team), full gender diversity (there are about as many men as women in the research team), more women than men and more man than women. For the variable nationality diversity, the following dummy variables were computed: no non-Dutch in the research team (least diverse), a small minority of the research team was non-Dutch, about half are non-Dutch and a large majority or only non-Dutch in the research team (most diverse).
To conduct logistic regression analyses with the categorical dependent variable consideration of stopping with the PhD trajectory, two dummy variables were computed. The dummy variables are: no consideration of stopping with the PhD trajectory and some or regular consideration of stopping with the PhD trajectory. Furthermore, the assumptions to conduct a logistic regression were also not violated. The performance variable is binary after computing dummy variables, the model fits
correctly, error terms are independent and the independent variables are linear. So, it was concluded that is was acceptable to further perform the analyses.
The scores for both performance variables, expected and experienced delay and consideration of stopping with the PhD trajectory, are recoded. So, a higher score on both variables indicates a higher performance of the PhD students. Descriptive statistics of all variables are included in table 1.
All of the central concepts of the theoretical model are measured on the individual level. The degree of diversity in research team, transformational leadership and goal
orientation are based on the perception of the participating PhD students, but refer to the group level. By testing theories regarding diversity and leadership on the individual level, insights in meeting the PhD students’ goals and interests as a result of the
organizational culture in relation to enhancing a synergetic group outcome can be obtained.
Table 2 presents the descriptive statistics of the central concepts of the conceptual model and the Cronbach’s alphas of the measurement scales are showed. The nationality diversity variable “about half non-Dutch” was positively significant related to expected and experienced delay in months in the PhD trajectory. None of the other diversity variables were significantly correlated to either one of the performance variables; expected and experienced delay or consideration of stopping with the PhD trajectory. Transformational leadership, contrary to the expectations, was negatively related to the performance variable consideration of stopping with the PhD trajectory. Learning goal orientation was not correlated to one of the performance variables. Performance goal orientation was marginally negatively significant related to consideration of stopping. The control variable organizational culture was positively significant related to
consideration of stopping with the PhD trajectory. Task interdependence was not significantly related to both performance variables. Transformational leadership was not related to both learning goal orientation and performance goal orientation.
Note. * p < .05, ** p < .01.
Measurement reliability: Cronbach’s alphas are shown on the diagonal
a 0 = Male, 1 = Female. b 0 = Dutch, 1 = Non-Dutch c 0 = Sometimes or Regularly, 1 = Never
Variable M SD 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 1. Gendera 0.66 0.48 – 2. Nationalityb 0.27 0.44 .02 – 3. Task interdependence 2.47 0.87 -.02 -.05 (α = .84) 4. Organizational culture 2.04 0.60 -.04 .02 .23 ** (α = .76) 5. Team size 1 0.06 0.24 .11 -.01 .08 .05 – 6. Team size 2-5 0.36 0.48 -.01 -.11 .16 .09 -.19* – 7. Team size 6-11 0.27 0.45 .20* .16 -.09 -.07 -.15 -.46** – 8. Team size ≥ 11 0.30 0.46 -.24** -.03 -.12 -..04 -.17 -.50** -.41** – 9. No gender diversity 0.16 0.37 .11 -.04 .07 -.07 .23 .08 -.05 -.16 –
10. Less than half
women 0.22 0.42 -.26
** .16 .12 .12 .02 .00 -.05 .03 -.24** –
11. More than half
women 0.38 0.49 .08 -.09 -.09 -.03 -.07 .02 -.03 .05 -.34 ** -.42** – 12. Full gender diversity 0.24 0.43 .07 -.02 -.07 -.03 -.14 -.10 .13 .05 -.25 ** -.30** -.43** – 13. No non Dutch 0.26 0.44 .18* -.24** .02 .03 .14 .22** -.02 -.28** .43** -.32** -.01 -.05 – 14. Small minority non Dutch 0.29 0.45 .04 -.16 -.17 * .07 -.02 -.07 -.03 .11 -.10 -.18 .25** -.01 -.38** –
15. About half non
Dutch 0.24 0.43 -.19
* .29** .03 -.05 .01 -.17 .13 .05 -.15 .29** -.08 -.07 -.33** -.36** –
16. Large majority or
only non Dutch 0.21 0.41 -.04 .13 .14 -.06 -.13 .02 -.08 .04 -.18
* .24** -.18* .13 -.31** -.33** -.29** – 17. Transformational leadership 2.33 0.76 -.02 -.07 .20 ** -.21* .05 -.01 -.02 -.00 .27** -.04 -.10 -.08 -.07 .04 .02 .01 (α = .92) 18. Learning goal orientation 2.51 0.76 -.12 .00 .00 .06 -.05 -.04 -.03 .09 -.02 -.03 .05 -.02 -.01 .01 .08 -.08 .08 (α = .73) 19. Performance goal orientation 2.49 0.87 .10 .08 -.01 .04 -.02 -.11 .18 * -.05 -.08 .04 .07 -.05 -.11 .12 .05 -.06 .01 .25** (α = .87) 20. Expected and experienced delay 3.70 5.70 -.02 .13 -08 .04 -.12 .06 .09 -.09 .00 .16 -.14 .00 -.10 -.07 .27 ** -.09 -.12 .00 .00 – 21. Consideration of stoppingc 0.46 0.50 -.04 -.02 -.06 .25** -.08 -.02 .03 .03 -.08 -.04 .08 .02 .04 -.07 .06 -.03 -.28** -.01 -17* -17* –
4. Results
To test hypothesis 1 and the direct effects of diversity in research teams on the
performance of PhD students, a hierarchical regression analysis was conducted for the performance variable expected and experienced delay in the PhD trajectory. A logistic regression was conducted for the performance variable consideration of stopping with the PhD trajectory. Separate analyses were performed for both gender diversity and nationality diversity. The control variables gender, nationality, organizational culture, task interdependence and team size were entered in the first step. The direct effects of diversity and transformational leadership were added in the second step of the
hierarchical regression analysis. In Table 3, the results of the regression analysis on the performance variable expected and experienced delay are displayed for both gender diversity (Model 1) and nationality diversity (Model 2). The results for the logistic regression analyses on the performance variable consideration of stopping with the PhD trajectory are shown in Table 4 (gender diversity) and Table 5 (nationality diversity). The reference category of the dummy variables for the control variable team size in the analyses is a team size of 1 (so the PhD students works solitary). The dummy variables regarding gender diversity in the analyses are referring to the category of no gender diversity (only men or only women in the research team). The dummy variables used in the analyses concerning nationality diversity are referring to the category of no non-Dutch in the research team.
The variables included in step 1 and 2 did not significantly explain the variance in results for the expected and experienced delay variable (Table 3) in model 1, F(11, 123) = 1.14, p > .05, R2 = .09, and model 2, F(11, 123) = 1.61, p > .05, R2 = .13. The predictors in
step 1 and 2 did marginally significant explain the variance in scores for consideration of stopping with the PhD trajectory for gender diversity (Table 4), χ2(11) = 19.59, p < .10,
R2 = .18. The variables in logistic regression analyses with direct effects of nationality
diversity did significant predict the results for consideration of stopping (Table 5), χ2(11) = 20.83, p < .05, R2 = .19. Hypothesis 1 states that diversity in research teams will
have a positive effect on the performance of PhD students. However, fully or almost fully diverse research teams in terms of gender (b = -0.99, p > .05) and nationality (b = 0.23, p > .05) do not have an effect on the expected and experienced delay in months for PhD
students compared to teams that are homogeneous. Research teams with more women than men (b = -1.85, p > .05) or more men than women (b = 1.05, p > .05) also did not significantly affect the expected and experienced delay of PhD students in their PhD trajectory. Diverse research teams in terms of nationality, whereby the share of non-Dutch in the research team was a small minority (b = 0.79, p > .05) also did not affect the expected and experienced delay significantly. Research teams with about half non-Dutch team members significantly predict the expected and experienced delay in the PhD trajectory (b = 4.11, p < .01). Teams that were completely or almost completely diverse in terms of gender (b = -0.18, p > .05) and nationality (b = -0.05, p > .05) had no
significant effect on the consideration of stopping with the PhD trajectory. The share of more (b = 0.00, p > .05) or less (b = -0.43, p > .05) women than men in the research team did not significantly predict the consideration of stopping. A small minority non-Dutch (b = -0.45, p > .05) in research teams did not significantly influence the consideration of stopping. If a research team consists of half non-Dutch members, there is also no
significant relationship between nationality diversity and performance (b = 0.34, p < .05). Although there is one significant relationship, it turns out in general that both nationality and gender diversity are not associated with the performance of PhD students. Therefore, hypothesis 1 is rejected.
Despite finding no significant relationship between diversity in research teams and performance of PhD students, further analyses were performed to examine the possible relationship between this null finding and the moderating effect of transformational leadership. To test hypothesis 2, which states that transformational leadership
positively moderates the positive relationship between diversity in research teams and the performance of PhD students, the interaction terms of transformational leadership and gender and nationality diversity were added in step 3 of the hierarchical regression analysis and the logistic regression analysis. These interaction terms are added to explain the variance in both analyses as a result of the moderating influence of transformational leadership on the relationship between diversity and both
performance variables. The variables and interaction terms added in step 1,2 and 3 of the hierarchical regression analysis (Table 3) did not significantly explained the variance in Model 1 (gender diversity), F(14, 120) = 1.06, p > .05, R2 = .11, but did
p <.10, R2 = .17. For the logistic regression analyses, the variables and interaction terms
added in step 1, 2 and 3 did marginally explain the variance in the model concerning gender diversity (Table 4), χ2(14) = 21.39, p < .10, R2 = .20, and nationality diversity,
χ2(14) = 21.28, p < .10, R2 = .20. The regression coefficients of transformational
leadership were both non-significant regarding gender diversity (b = -0.95, p > .05) and nationality diversity (b = -0.68, p > .05) for expected and experienced delay during the PhD trajectory. Transformational leadership had no significant effect on consideration of stopping, in the logistic regression model of gender diversity (b = -0.60, p > .05) (Table 4). Transformational leadership in regard to nationality diversity was significantly associated to consideration of stopping (b = -0.64, p < .05)(Table 5). Contrary to the described theoretical framework, higher degrees of transformational leadership will lead to more often consideration of stopping with the PhD trajectory. The interaction between transformational leadership and both forms of diversity did not yield
significant regression coefficients concerning expected and experienced delay for fully or almost fully diverse teams in terms of gender (b = 1.72, p > .05) or nationality (b = -2.03, p > .05), or with teams with less women than men (b = 0.89, p > .05) or vice versa (b = 1.16, p > .05), and teams with a small minority non-Dutch team members (b = 1.41,
p > .05) or about half non-Dutch (b = 2.45, p > .05). Transformational leadership did also
not moderate the relationship for teams with less women than men (b = 1.01, p > .05), less men than women (b = 0.81, p > .05), and fully diverse teams in terms of gender (b = 0.92, p > .05), teams with a small minority non-Dutch (b = 0.57, p > .05), teams with about half non-Dutch (b = 0.26, p > .05), or almost fully diverse teams in terms of nationality (b = 0.32, p > .05) regarding the consideration of stopping with the PhD trajectory. Although a higher degree of transformational leadership regarding
nationality diversity will decrease the performance in terms of consideration of stopping with the PhD trajectory, it will not moderate the relationship between diversity in
research teams and performance. Hypothesis 2 is therefore not accepted.
Hypothesis 3 states that an increased learning goal orientation of a PhD student partially mediates the moderating effect of transformational leadership on the positive
relationship between diversity in research teams and the performance of PhD students. The learning goal orientation variable and interaction terms between all diversity variables and the learning goal orientation are added to step 4 of the hierarchical
regression analysis for both Model 1 and Model 2 (Table 3). These variables are also added to step 4 for the logistic regression analyses for gender diversity (Table 4) and nationality diversity (Table 5). To test the mediated moderation of learning goal orientation, the first step is to test the direct effect of learning goal orientation on the performance variables. If there is a significant effect between the learning goal
orientation and the performance variables, the second step is to determine if there are significant interaction effects between the diversity variables and learning goal
orientation concerning performance. If the significant interaction effects between learning goal orientation and diversity diminish the interaction effects of diversity and transformational leadership, the moderation is mediated by learning goal orientation. The variables in the regression analysis did not significantly explain the variance for expected and experienced delay in Model 1 (gender diversity, Table 3), F(18, 116) = 0.93, p > .05, R2 = .13. The variance for expected and experienced delay in Model 2
(nationality diversity, Table 3) was marginally significant explained by the variables in the regression analysis, F(27, 150) = 1.53, p < .10, R2 = .19. As the logistic regression
analyses regards, the variance in consideration of stopping was not significantly
explained by the models for both gender diversity (Table 4), χ2(18) = 24.29, p > .10, R2 =
.22, and nationality diversity (Table 5), χ2(18) = 21.78, p > .10, R2 = .20. The regression
coefficients of the direct effect of learning goal orientation in Model 1 (gender diversity, Table 1)(b = -0.09, p > .05) and Model 2 (nationality diversity, Table 1)(b = -0.20, p < .05) were not significant for expected and experienced delay. Learning goal orientation did also not predict significantly consideration of stopping in both logistic regression
analyses regarding gender diversity (b = 0.06, p > .05)(Table 4) and nationality diversity, (b = -0.05, p > .05)(Table 5). The regression coefficients of the interaction terms between learning goal orientation and diversity in relationship to expected and experienced delay (Model 1, Table 3) in teams with less women than men (b = 1.06, p > .05), teams with less men than women (b = 2.72, p > .05) and teams with equally distributed men and women (b = 1.20, p > .05) were all non-significant. This is also the case for
interaction variables between learning goal orientation and teams with about half non-Dutch team members (b = 1.89, p > .05), or almost fully diverse teams in terms of nationality (b = 1.81, p > .05). The regression coefficient of the interaction between learning goal orientation and teams consisting of a small minority non-Dutch was
marginally significant (b = 3.47, p < .10). There were also no significant relationships found between the interaction variables and consideration of stopping with the PhD trajectory for both logistic regressions regarding gender (Table 4) and nationality diversity (Table 5). The most fully diverse teams in terms of nationality (b = -0.22, p > .05) and gender (b = 0.57, p > .05) showed no significant regression coefficients in
interaction with learning goal orientation. Teams with more women then men (b = -0.70,
p > .05) and vice versa (b = -0.48, p > .05) also reported the same non-significant
statistics. Research teams with a minority share of non-Dutch members (b = -0.41, p > .05) and about half Dutch team members (b = 0.02, p > .05) also yield
non-significant results in relationship to consideration of stopping, interacted by learning goal orientation. An increased learning goal orientation does not enhance the
performance directly nor partially mediating the moderating influence of
transformational leadership on the relationship between diversity and performance in research teams. As a result, also hypothesis 3 is rejected.
Note. N = 135. Unstandarized regression coefficients are reported (with standard errors in parentheses).
+ = p < .10, * p < .05, ** p < .01, *** p < .001
Model 1: Gender diversity Model 2: Nationality diversity
Variable Step 1 Step 2 Step 3 Step 4 Step 1 Step 2 Step 3 Step 4
Step 1: Control variables
Male 0.42 (1.09) -0.03 (1.11) -0.02 (1.12) -0.19 (1.16) 0.42 (1.09) -0.27 (1.07) 0.24 (1.10) 0.14 (1.11) Dutch nationality -1.51 (1.12) -1.04 (1.14) -0.92 (1.15) -1.04 (1.17) -1.51 (1.12) -0.45 (1.20) -0.47 (1.20) -0.69 (1.22) Organizational culture -0.21 (0.85) -0.58 (0.87) -0.48 (0.89) -0.29 (0.92) -0.21 (0.85) -0.37 (0.87) -0.42 (0.86) -0.25 (0.87) Teamsize 2-5 3.05 (2.30) 3.43 (2.22) 3.67 (2.26) 4.68+ (2.43) 3.05 (2.30) 3.41 (2.16) 3.35 (2.14) 4.65* (2.29) Teamsize 6-10 2.99 (2.25) 3.44 (2.31) 3.69 (2.37) 4.64 + (2.51) 2.99 (2.25) 2.68 (2.22) 2.66 (2.19) 3.79 (2.31) Teamsize ≥ 11 1.53 (2.27) 2.10 (2.33) 2.40 (2.42) 3.45 (2.58) 1.53 (2.27) 1.45 (2.28) 1.29 (2.25) 2.52 (2.39) Task interdependence -0.49 (0.59) -0.44 (0.61) -0.53 (0.62) -0.55 (0.63) -0.49 (0.59) -0.38 (0.62) -0.063 (0.62) -0.47 (0.63) Step 2: Main effects
Less women than men (GD1) 1.05
(1.71) 0.92 (1.82) 0.78 (1.84)
More women than men (GD2) -1.85
(1.54) -1.96 (1.66) -2.17 (1.68)
Full gender diversity (GD3) -0.99
(1.69) -1.41 (1.82) -1.62 (1.84) Small minority non-Dutch
(ND1)
0.79
(1.36) 0.68 (1.35) 0.63 (1.36)
About half non-Dutch (ND2) 4.11**
(1.55) 3.95
**
(1.53) 3.87
**
(1.54) Large majority or only
non-Dutch (ND3) 0.23 (1.54) 0.30 (1.51) 0.05 (1.54) Transformational leadership (TL) -0.93 (0.71) -0.95 (0.73) -0.91 (0.75) -0.86 (0.68) -0.68 (0.68) -0.61 ** (0.70) Step 3: Interactions GD1 X TL 0.89 (1.95) 1.24 (2.02) GD2 X TL 1.16 (1.85) 1.27 (1.87) GD3 X TL -1.72 (2.12) -1.57 (2.14) ND1 X TL 1.41 (1.79) 0.60 (1.86) ND2 X TL 2.45 (2.06) 1.75 (2.14) ND3 X TL -2.03 (1.91) -2.52 (1.97) Step 4: Mediators
Learning goal orientation (LGO) -0.09 (0.69) -0.20 (0.66) GD1 X LGO 1.06 (2.38) GD2 X LGO 2.72 (2.14) GD3 X LGO 1.20 (2.47) ND1 X LGO 3.47+ (2.06) ND2 X LGO 1.89 (2.02) ND3 X LGO 1.81 (2.01) R2 .05 .09 .11 .13 .05 .13 .17 .19 ΔR2 .05 .05 .02 .02 .05 .08* .05+ .02 F 0.90 1.14 1.06 0.93 0.90 1.61 1.77+ 1.53+
Note. N = 135. Unstandarized regression coefficients are reported (with standard errors in parentheses).
+ = p < .10, * p < .05, ** p < .01, *** p < .001
Gender diversity
Variable Step 1 Odds ratio Step 2 Odds ratio Step 3 Odds ratio Step 4 Odds ratio
Step 1: Control variables Male 0.10 (0.40) 0.91 0.19 (0.43) 0.83 0.13 (0.43) 0.88 0.14 (0.44) 0.87 Dutch nationality 0.18 (0.42) 0.83 0.18 (0.44) 0.84 0.14 (0.45) 0.87 0.15 (0.45) 0.86 Organizational culture 1.08** (0.34) 0.34 0.92 ** (0.36) 0.40 0.98 (0.36) 0.38 0.89 (0.37) 0.41 Teamsize 2-5 0.68 (0.82) 0.51 0.56 (0.84) 0.57 0.72 (0.87) 0.49 0.51 (0.95) 0.60 Teamsize 6-10 0.96 (0.85) 0.38 0.90 (0.88) 0.41 1.13 (0.93) 0.32 0.95 (1.00) 0.39 Teamsize ≥ 11 0.83 (0.85) 0.44 0.77 (0.89) 0.47 0.99 (0.94) 0.37 0.78 (1.01) 0.46 Task interdependence -0.28 (0.22) 1.33 -0.12 (0.24) 1.13 -0.13 (0.24) 1.14 -0.14 (0.24) 1.15
Step 2: Main effects Less women than men (GD1)
-0.43
(0.65) 1.54 -0.68 (0.70) 1.97 -0.65 (0.70) 1.91
More women than men (GD2)
0.00
(0.58) 1.00 -0.26 (0.65) 1.29 -0.20 (0.65) 1.22
Full gender diversity (GD3) -0.18 (0.64) 1.20 -0.44 (0.70) 1.55 -0.37 (0.72) 1.45 Transformational leadership (TL) -0.66** (0.28) 1.93 -0.60 (0.29) 1.82 -0.62 (0.30) 1.86 Step 3: Interactions GD1 X TL 1.01 (0.86) 0.36 1.09 (0.90) 0.34 GD2 X TL 0.81 (0.82) 0.45 0.84 (0.86) 0.43 GD3 X TL 0.92 (0.91) 0.40 0.89 (0.98) 0.41 Step 4: Mediators Learning goal orientation (LGO) 0.06 (0.30) 0.94 GD1 X LGO -0.48 (1.03) 1.61 GD2 X LGO -0.70 (0.97) 2.01 GD3 X LGO 0.57 (1.16) 0.57 χ2 12.75+, df = 7 19.59 +, df = 11 21.39 +, df = 14 24.29, df = 18 Nagelkerke R2 12% 18% 20% 22%
Note. N = 135. Unstandarized regression coefficients are reported (with standard errors in parentheses).
+ = p < .10, * p < .05, ** p < .01, *** p < .001
Nationality diversity
Variable Step 1 Odds ratio Step 2 Odds ratio Step 3 Odds ratio Step 4 Odds ratio
Step 1: Control variables Male 0.10 (0.40) 0.91 0.01 (0.42) 0.99 -0.01 (0.43) 1.01 0.03 (0.43) 0.97 Dutch nationality 0.18 (0.42) 0.83 0.44 (0.48) 0.65 0.39 (0.49) 0.68 0.41 (0.49) 0.67 Organizational culture 1.08** (0.34) 0.34 1.00 ** (0.37) 0.37 1.01 ** (0.37) 0.37 1.02 ** (0.38) 0.36 Teamsize 2-5 0.68 (0.82) 0.51 0.62 (0.83) 0.54 0.65 (0.83) 0.52 0.53 (0.89) 0.59 Teamsize 6-10 0.96 (0.85) 0.38 0.91 (0.85) 0.40 0.91 (0.85) 0.40 0.82 (0.90) 0.44 Teamsize ≥ 11 0.83 (0.85) 0.44 0.85 (0.88) 0.43 0.88 (0.88) 0.42 0.75 (0.94) 0.47 Task interdependence -0.28 (0.22) 1.33 -0.20 (0.24) 1.22 -0.20 (0.25) 1.22 -0.22 (0.25) 1.25
Step 2: Main effects Small minority non-Dutch (ND1)
-0.45
(0.53) 1.56 -0.43 (0.54) 1.54 -0.43 (0.54) 1.53
About half non-Dutch (ND2) 0.34 (0.61) 0.71 0.35 (0.62) 0.70 0.34 (0.62) 0.71 Large majority or only non-Dutch (ND3) -0.05 (0.60) 1.05 -0.04 (0.61) 1.04 -0.03 (0.61) 1.03 Transformational leadership (TL) -0.62* (0.27) 1.86 -0.64 * (0.28) 1.90 -0.63 * (0.28) 1.88 Step 3: Interactions ND1 X TL 0.48 (0.74) 0.62 0.57 (0.77) 0.57 ND2 X TL 0.20 (0.85) 0.82 0.26 (0.89) 0.77 ND3 X TL 0.30 (0.80) 0.74 0.32 (0.83) 0.73 Step 4: Mediators Learning goal orientation (LGO) -0.05 (0.28) 1.05 ND1 X LGO -0.41 (0.86) 1.51 ND2 X LGO 0.02 (0.86) 0.99 ND3 X LGO -0.22 (0.85) 1.24 χ2 12.75+, df = 7 20.83 *, df = 11 21.28 +, df = 14 21.78, df = 18 Nagelkerke R2 12% 19% 20% 20%
5. Discussion and conclusion
This research has addressed the link between diversity in research teams and the performance of PhD students by examining the moderating role of transformational leadership. To do so, the following question was answered: does transformational
leadership moderate the relationship between diversity in research teams and performance of PhD students in a performance-oriented organizational culture?
Contrary to the expectations, the results showed no significant relationship between both gender and nationality diversity and the performance of PhD students, except for the relationship between teams consisting of about half non-Dutch and experienced and expected delay in months compared to teams with no non-Dutch members. Further analyses were performed to examine the possible relationship between this null finding and the moderating effect of transformational leadership. It was hypothesized that transformational leadership would reinforce the relationship between diversity in research teams and the performance of PhD students. However, the results showed no significant moderating effect of transformational leadership on the relationship between diversity and performance. It is notable that transformational leadership even yielded negative direct effects on the performance of PhD students in terms of consideration of stopping with the PhD trajectory and expected and experienced delay in the PhD
trajectory. Despite the absence of a moderating effect of transformational leadership on the relationship between diversity in research teams and performance, learning goal orientation did also not partially mediate the role of transformational leadership or directly affect the relationship between diversity in research teams and the performance of PhD students.
5.1 Theoretical implications
Earlier research showed inconsistent findings in the direct effects of diversity on performance (Bell, Villado, Lukasik, Belau & Briggs, 2011; Horwitz & Horwitz, 2007; Guillaume, Dawson, Otaye-Ebede, Woods & West, 2017). In this particular research, only when research teams consisted of about half non-Dutch members compared to teams with no non-Dutch team members, PhD students expected and experienced less delay in months during their PhD trajectory. Besides of this result, no significant relationship
