LGBTQ Leaders: The relationship a leader’s sexual orientation, perceived prototypicality, and effectiveness
Master Thesis
Student: Martin Noergaard-Solum/ Student no. 11375523
MSc. Business Administration, Leadership and Management Track University of Amsterdam, Faculty of Economics and Business
Supervisor: Dr. Claudia Buengeler
University of Amsterdam, Amsterdam Business School
Abstract
This paper investigates the relationship between a leader’s sexual orientation, perceived prototypicality, and perceived effectiveness. Building on the notion that LGBTQ leaders perceived as less prototypical and effective than heterosexual leaders, this paper investigates these two relationships, as well as whether perceived prototypicality mediates the relationship between leader’s sexual orientation and perceived effectiveness. Further, it investigates if certain leadership styles or organisational culture dimensions moderates the relationship between leader’s sexual orientation and perceived effectiveness. To test these relationships, data is used from 180 participants in 90 dyads, each dyad consisting of one leader and one subordinate. The results show that LGBTQ are perceived as both more prototypical and more effective than heterosexual leaders, and that the relationship between leader’s sexual
orientation and perceived effectiveness is positively moderated by transformational
leadership. These results stand in contrast to what extant literature suggests, and this paper adds to the almost non-existent body of literature on LGBTQ leadership.
Keywords: LGBTQ leadership, leader prototypicality, leader effectiveness,
transformational leadership, contingent reward leadership, participative leadership, parochial versus professional, open systems versus closed systems, employee-oriented versus work-oriented
Table of Contents
Statement of Originality 4
Introduction 5
Theoretical Background and Hypotheses 7
Leader Prototype 8
The Moderating Effect of Leader Behaviours 11
Transformational Leadership 12
Contingent Reward Leadership 15
Participative Leadership 16
The Moderating Effect of Organisational Culture 17
Methodology 22 Sampling Strategy 22 Independent Variable 25 Dependent Variable 26 Mediating Variable 26 Moderating Variables 27 Control Variables 28 Statistical Approach 29 Results 30 Descriptive Statistics 30
Regressions and Mediation Analyses 32
Regressions and Moderation Analyses 34
Additional Analyses 38
Discussion 40
Practical Implications 43
Limitations and Future Research 43
Conclusion 46
Statement of Originality
This document is written by student Martin Noergaard-Solum who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.
The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
Introduction
Following an outing scandal in British tabloids in the early months of 2007, former CEO of British Petroleum, John Browne, Baron Browne of Madingley, resigned from his post as chief executive of the oil giant. Browne, who had climbed the entire career ladder from bottom to top in the company over a 41-year period, and steered it into its “golden age” during his tenure as CEO (Browne, 2014), lost a legal battle against the Associated
Newspaper group regarding the publishing of an article detailing his personal life and former relationship with a male escort – and allegedly using corporate funds to pay for their
extravagant lifestyle (a statement that was later proven to be wrong). His reason for resigning from BP, he said, was “to avoid unnecessary embarrassment and distraction to the company” (Stewart, 2014). In his book, “The Glass Closet”, Browne later said that he did not initially want to disclose his sexuality, as he saw openness about his sexual orientation as detrimental to his career. According to him, in 2013 “there was not a single open gay chief executive in the Fortune 500” (Browne, 2014, p. 11) – a list of the 500 most valued companies in the world. With respect to the increasing focus on LGBTQ (Lesbian, Gay, Bi, Trans and Queer) rights such as marriage equality and anti-discrimination laws, why are there so few openly gay business leaders?
One possible explanation may be the way leaders are perceived. We all have a notion of what type of person ‘the perfect leader’ is, and leader prototypicality is the perception of what qualities and characteristics “typical” leaders hold, in the eyes of followers. One prototypical characteristic is age - higher age is associated with higher levels of leader
prototypicality (Buengeler, Homan, & Voelpel, 2016). Other broadly accepted characteristics are gender, i.e. the leader being male, as well as race, i.e. the leader being white. In John Browne’s case, he had all those traits. Yet, he chose to step down as his sexuality was painted all over the tabloids. Why did it matter? Does being openly gay deviate from the leader
prototype? Based on Browne’s actions, it may seem that he thought disclosure around non-heterosexuality did not go hand in hand with having a high position within a firm, certainly within the oil industry, which encompasses a very masculine culture (Mearns & Yule, 2009; Collins & Callahan, 2012) That raises a second question – are leaders who are open about their non-straight status at work perceived differently in different organisational cultures? These questions may lead to insights into what aspects of work life might marginalise openly LGBTQ leaders, and if so, what the underlying social and organisational norms that dictate this are?
Several studies have shown that many people choose to separate their sexuality from their working lives. Griffith and Hebl (2002) found that workers who are not open about their LGBTQ status at work are more likely to be dissatisfied in the workplace, yet that the more open the organisation is to diversity, the more likely gay and lesbian workers are to disclose their sexuality. They also found that Human Resource Development is important to counter workplace dissatisfaction in LGBTQ cases. Rumens and Kerfoot (2009) found that openly gay men struggle to be seen as equals to their heterosexual colleagues in the same professions. Embrick and Walther’s (2007) research focused on the question whether attitudes towards LGBTQ rights had progressed in recent years, yet they found that hidden animosities towards the LGBTQ community actually increased. These studies paint a dark picture for LGBTQ professionals who wish to be open about their sexuality.
The aim of this paper is to see how sexual orientation is related to the perception of leader prototypicality and effectiveness. Also, it will investigate whether perceived
effectiveness is moderated by factors such as leadership styles or organisational cultures. Consequently, this paper seeks to answer the following question:
How does sexual orientation affect leader prototypicality and effectiveness perceptions among followers, and is this relationship moderated by leadership styles or
organisational cultures?
After having established these relationships using regression analyses, this paper will discuss possible theoretical implications, contributing to the small body of research that exists on LGBTQ leadership. Additionally, it will look into possible practical contributions, i.e. what can LGBTQ people who find themselves in a leader position can do to counter the possibly negative bias they face being LGBTQ.
In the following section, I will take a look at what research has been conducted to date in order to establish the hypotheses this paper seeks to investigate, as well as attempting to fill the research gap in research. This gap is specifically the largely unexplored area of LGBTQ-specific research, both in terms of perceived effectiveness and prototypicality – whether LGBTQ leaders are perceived differently in these aspects than heterosexual leaders. Further, it will seek to fill the gap that is testing if certain leadership styles or organisational cultures have an effect on this relationship, as moderators.
Theoretical Background and Hypotheses
There has been very little research on LGBTQ leaders. In fact, less than a decade ago, it was stated that “there is almost nothing in the scholarly literature specifically regarding LGBT leadership issues” (Fassinger, Shullman, & Stevenson, 2010, p. 204). Also, in the research that has been conducted, the term LGBTQ comes in different variations. The abbreviation LGBTQ stands for Lesbian, Gay, Bi, Transgender and Queer/Questioning, and while various scholars use terms as LGB (Boxelaar, 2014) and LGBT (Fassinger et al., 2010) – which is also the most used term – this thesis will use the full abbreviation. While it is
important to note that Transgender mostly explains differences in gender identity while the others explain sexual identity, the purpose of this thesis is to explore the environment around leaders who see themselves as anything but heterosexual men or women, or “straight” people, emphasising the importance of both Transgender and Queer people in addition to the LGB ones.
A general definition of organisational leadership is said to be “the ability of an
individual to influence, motivate, and enable others to contribute toward the effectiveness and success of the organizations of which they are members” (Simonton, 1994, p. 411). Three key words can be drawn from this definition, namely: influence, motivate, and enable. The ways in which these concepts are put into use are numerous, and their effectiveness depend on the leaders – and followers.
Leader Prototype
According to Rosch (1978), humans create categories to organise and process information efficiently, which is known as traditional categorisation theory. These categories have very distinct features and properties, and the people who fit into each category are similar, yet non-identical. From these categories, prototypes evolve, as standards or averages of the categories they stem from. Based on the concept of basic categorisation theory, leader categorisation theory was developed. In their research, Lord, Foti, and De Vader (1984) found that there are several traits that are attributed to characterise a leader in various situations, and that the way one judges these traits to be leader-typical or not in that situation may affect one’s perceptions of the qualities of a given leader. Hence, one can argue that in the process of categorisation, people will evaluate the traits a leader holds against a pre-set notion of qualities a typical leader should hold, and determine to which extent they see this person as a leader or not. Lord et al (1984, p. 374) identified the important implications of being perceived as a leader in their
research: “For example, frequently individuals with effective ideas may not succeed in getting them implemented because they are not perceived as leaders. Being perceived as a leader may have many other important symbolic functions that produce acceptance of organizational goals and compliance, commitment, and positive affect for subordinates”.
Being perceived as a leader means that a leader needs to have certain qualities or
characteristics that are said to be prototypical characteristics, i.e. characteristics that form the image of the prototypical leader (Rosette et al., 2008, p. 759). In their research, Rosette et al (2008) found that race is one such characteristic, specifically being white enhances followers’ perception of leader prototypicality. Buengeler, Homan, and Voelpel (2016) found that age is another characteristic, and that young managers are seen as both less effective and less prototypical by their followers, as opposed to their senior counterparts. Another study, conducted by Ryan, Haslam, Hersby, and Bongiorno (2011), found that female leaders consistently experience lower performance evaluations than their male colleagues for identical work. These studies paint a picture that the general perception of a leader, and in particular an effective leader, is middle-aged, white, and male.
Research has shown that in the workplace, openly LGBTQ people are seen as inferior to their heterosexual peers. In particular, LGB people experience significantly higher career stagnation and distrust in their employers (Collins & Callahan, 2012), gays and lesbians are victims of so-called White male solidarity in highly masculine industries; where they
experience gay bashing and banter, and will not be considered to fill vacant positions if they are suspected of being LGBTQ (Embrick et al., 2007). They are also more likely to report discrimination when they work in mainly heterosexual groups (Ragins & Cornwell, 2001). In his article, Herek (2007) talks about the sexual stigma that non-heterosexuals face in form of inferior status and relative powerlessness that society exerts on them due to their
legitimised the inferior status of sexual minorities, and that it continues to do so today. Firstly, the assumption that everyone is “straight”, so-called heterosexual assumption, and secondly that when non-heterosexuality is disclosed, it is problematized as being unnatural, requiring explanation, and ultimately, gives rise to differential treatment and discrimination. Herek (2007) also stated that “heterosexuals are regarded as prototypical members of the
category people, and heterosexual behaviour and different-sex relationships are presumed to be normal and natural” (Herek, 2007, p. 908) Drawing lines from research on LGBTQ people to the already established notion of leaders deviating from the leader prototype, one may argue that a prototypical leader is also straight. This is an aspect that this thesis seeks to examine. Based on the inferior status LGBTQ people have, and the preference for
heterosexuality in managerial positions as evidenced by the research presented above, the following hypothesis is presented:
Hypothesis 1a: LGBTQ leaders are perceived to be less prototypical than
heterosexual leaders.
This thesis will also examine whether or not non-heterosexuality is related to perceived leader effectiveness. Prior research has shown that prototypical leaders are perceived as more effective (Hogg, 2001; Van Knippenberg & Hogg, 2003; Pierro, Bonaiuto, Van Knippenberg, & Kruglanski, 2005). At the base of this notion is the social identity theory, which states that members of a group treat the group, and by extension, the group prototype, as a source of social reality, and that they are more open to influence from the prototypical leader (De Cremer, Van Dijke, & Mayer, 2010). Because of the increased openness to influence, they are more likely to endorse the prototypical leader, and the prototypical leader is perceived as more effective than non-prototypical leaders (Pierro et al, 2005). These effects of prototypical
leadership are direct results of the social identity theory, as has been shown by laboratory and field studies (Giessner & Van Knippenberg, 2008; Platow & Van Knippenberg, 2001; Ulrich, Christ, & Van Dick, 2009). This close relationship between prototypicality and effectiveness gives rise to the second and third part of hypothesis 1:
Hypothesis 1b: LGBTQ leaders are perceived to be less effective than heterosexual
leaders.
Hypothesis 1c: The relationship between a leader’s sexual orientation and perceived
effectiveness is mediated by perceived prototypicality.
The Moderating Effect of Leader Behaviours
If LGBTQ leaders are at a disadvantage in the way they are perceived as effective or not due to their sexual orientation, as hypothesised, it would be interesting to investigate whether they would be able to counter this disadvantage by adopting certain leadership styles.
Similarly, knowing under which leadership styles they could possibly be perceived as even
less effective could be useful to LGBTQ leaders as they would know which styles of
leadership to adopt or avoid. In other words, certain leadership styles might moderate the relationship between sexual orientation and perceived effectiveness. In the course if its history, leadership research has tried to establish which forms of leadership have the greatest impact on various organisational and leadership outcomes – and what constitutes a good leader. In the earliest year of the field, research focused on different traits leaders could have that would differentiate the good ones from the bad ones. While some physical appearance traits such as height (Judge & Cable, 2004), and personality traits such as the Big Five (DeRue, Nahrgang, Wellman, & Humphrey, 2011) are shown to have positive correlations with leadership emergence and effectiveness, the field of research went on to look at
leadership styles, such as Initiating Structure & Consideration (Judge, Piccolo, & Ilies, 2004), and Autocratic/Democratic leadership (Lewin & Lippitt, 1938). From there, the research started focusing on the contingency approach, before moving on to charismatic,
transformational and transactional leadership, which will be of focus in the following section.
Transformational Leadership
Charismatic leadership is a term coined by sociologist Max Weber, as a source of legitimacy on the basis that a charismatic leader would be perceived by followers to have exceptional qualities (Conger & Kanungo, 1987), and therefore be able to exert influence. Weber`s theory was later expanded by several scholars. Wiliner (1984) stated that "It is not what the leader is but what people see the leader as that counts in generating the charismatic relationship" (Wiliner, 1984, p. 14), suggesting that charisma is a notion of perception. This notion was supported by Meindl, Ehrlich, and Dukerich (1985), who introduced the
perspective of romance of leadership. This perspective states that people are inclined to glorify leadership as a category, as a response to the need to make sense of complex organisational phenomena. In doing so, they over-attribute organisational outcomes, both good and bad, to the leader, proposing that the leader has more power over the organisation that what might be the case. Conger and Kanungo (1987) concluded that charismatic
leadership is a two-way interaction between the leader and the follower, as “charisma is both a set of dispositional attributions by followers and as a set of leaders' manifest behaviours" (Conger & Kanungo, 1987, p. 645). Despite the disagreement around the theoretical concept on which way Charismatic leadership works, there is consensus that charisma works via two mechanisms, namely through idealised influence (charisma) and inspirational motivation (Bass, 1990).
The two mechanisms were adopted as part of the expanded Transformational leadership framework by Bass, which also consisted of two other dimensions, namely intellectual
stimulation and individualised consideration (Hoch, Bommer, Dulebohn, & Wu, 2016). Through idealised influence, which is the extent to which a leader behaves in manners that will make followers identify with the leader, a charismatic leader will appeal to his or her followers on an emotional level, showing confidence and taking a stand. Inspirational motivation entails a leader projecting an attractive vision that appeals to and inspires followers. Leaders who possess this talent project optimistically talk about how to reach future goals and provide meaning for the tasks to come. Intellectual stimulation is the extent to which leaders challenge assumptions, take risk and seek followers’ ideas. Stimulation and encouragement of creativity are talents displayed by leaders who possess this trait. Finally, individualised consideration entails the leader taking a coaching role to help the followers reach their individualised goals, attending to their needs and paying attention to their concerns (Judge & Piccolo, 2004). In summary, transformational leadership is about forgoing the short-term orientation and immediate self-interests by transforming the follower and the
organisation, creating identification with the leader among the followers, elevating followers’ interests and making them go the extra mile for the good of the group (Den Hartog, Muijen, & Koopman, 1997).
Transformational leadership has been shown to have a positive relationship with leader effectiveness, as stated by Hoch, Bommer, and Dulebohn (2016). It was also stated by Hater and Bass that “the dynamics of transformational leadership involve strong personal
identification with the leader, joining in a shared vision of the future, or going beyond the self-interest exchange of rewards for compliance” (Den Hartog & Koopman, 200, p. 176). Judge and Piccolo (2004) found that transformational leadership has a strong correlation with
different leadership criteria, some of which are follower job satisfaction, follower leader satisfaction, and rated leader effectiveness.
Bass and Avolio (1989) found in their study that leaders that are highly transformational are more likely to be seen as prototypical than transactional leaders. Research has also found that women leaders are in general more transformational than their male counterparts (Bass, Avolio, & Atwater, 1996). For example, Eagly and Crowley (1986) found that women are more nurturing than men, while Eagly and Johnson (1990) described female leaders as more interested in others and socially sensitive. Both aspects are part of being a transformational leader. The reason for why women leaders are more transformational than men, as argued by Bass (1990), may be “anti-feminine bias and disadvantage as a plausible explanation for finding that women are somewhat more transformational and therefore more likely to make effective leaders” (Bass, 1990, p. 17). These findings from women leaders are transferable to LGBTQ leaders (specifically gay men), as findings from Salter and Liberman (2016) showed that gay men are seen as more similar to heterosexual women than heterosexual men.
Further, if this paradoxical explanation for why women are more transformational and effective is due to their prototypical nature, one can argue that the same goes for non-heterosexual leaders, and their source of transformationalism being due to “anti-gay bias” and similar non-prototypical disadvantage. Research showing that transformational leadership has a positive relationship with effectiveness (Judge & Piccolo, 2004; Hoch et al., 2016) suggests that leaders who are highly transformational are perceived as more effective than those who are not. Based on the presented literature on transformational leadership, the following hypothesis is proposed:
Hypothesis 2: Transformational leadership will positively moderate the relationship
who score high on transformational leadership are perceived to be more effective than those who score low.
Contingent Reward Leadership
In contrast to transformational leadership, transactional leadership relies on a series of bargains and exchanges between the leader and the follower (Howell & Avolio, 1993).
Contingent reward is one such form of transactional leadership, where a leader and a follower agree upon an expected level of performance through negotiation. In this negotiation process, they will also agree what level of reward the follower can expect once this performance has been achieved. When achieved, the leader will reward the follower in a manner on which they agreed, be it monetary rewards or other rewards that the leader can give. And as long as they both find this exchange to be mutually rewarding, they will continue this transactional relationship (Howell & Avolio, 1993).
In their study, Buengeler et al. (2016) found that for non-prototypical leaders (non-prototypical based on their young age), contingent reward was a leadership style that was effective, as it reduced voluntary turn-over among followers of younger leaders (i.e. followers leaving due to perception of poor leadership). Contingent reward increased involuntary turnover, i.e. the leader dismissing followers who did not perform as expected. These effects were experienced by young leaders, non-prototypical in power of their age. They argued that contingent reward would be an effective leadership style due to its anchoring in the leader’s
position, and not status-related characteristics of the leader, such as age, race or gender.
In line with the findings of Buengeler et al. (2016) that contingent reward leadership works well for non-prototypical leaders, one might argue that the same results can be seen when tested in a setting where non-heterosexuality is the non-prototypical characteristic. Thus, the following hypothesis is presented:
Hypothesis 3: Contingent reward leadership will positively moderate the relationship
between leader’s sexual orientation and effectiveness, such that LGBTQ leaders who score high on contingent reward leadership are perceived to be more effective than those who score low.
3.2.3 Participative Leadership
Participative leadership is defined as “joint decision making or at least shared influence in decision making by a superior and his or her employees” (Koopman & Wierdsa, 1998, p. 297). The leader seeks input from subordinates in important decisions and in the
implementation of these decisions, and values his or her employees’ points of view. It has been shown that a leader who solicits new ideas on existing problems from subordinates has been linked to high performance (Somech & Wenderow, 2006). The decisions that the leader makes in partnership with his/her subordinates is found to be of increased quality (Scully, Kirkpatric, & Locke, 1995).
In his study, Sauer (2011) investigated how the perceived status of a new leader (in terms of team tenure) influenced subordinates’ reactions to different leadership styles. The two styles in question were participative and directional leadership – a style in which the leader seeks team members’ compliance with directions about how to solve problems. A leader’s personal power, which stems from the perceived status the leader holds among the members of the team, is more important in determining managerial effectiveness than position power, which comes from the authority the leader role demands. Participative leadership is a leadership style that relies on personal power, while directive leadership is a style that relies on position power. The study found that low-status leaders were perceived as more effective when they used a directive leadership style than when they used a participative leadership
style, and that high-status leaders were more effective when they used a participative leadership style than a directive leadership style.
This evidence is supported by Buengeler et al. (2016), who, in addition to their findings on contingent reward leadership, also found that participative leadership was not effective for young leaders, as it is a leadership style that requires recognition and even identification with the leader on a personal level, and that the experience, status and competence of the leader are not ‘accepted’ by the followers in young leaders as compared to their older counterparts. Younger leaders lacked the natural status cue required to be effective as participative leaders, a status cue that older leader held in power of their age. In accordance with the theoretical framework by Herek (2007), which states that, at a societal level, non-heterosexual
individuals suffer from perceptions of inferior status and power compared to heterosexuals, and the findings of Sauer (2011) and Buengeler et al. (2016), one may argue that being LGBTQ is a cue for lower status perception within a team. Hence, for a LGBTQ leader to employ a participative leadership style will be detrimental for his or her perceived
effectiveness, because of his low-power status. Based on this, the following hypothesis is presented:
Hypothesis 4: Participative leadership negatively moderate the relationship between
leader’s sexual orientation and effectiveness, such that LGBTQ leaders who score high on participative leadership are perceived to be less effective than those who score low.
The Moderating Effect of Organisational Culture
John Browne of BP chose to resign from his position as chief executive in 2007 as his sexual orientation became known (Browne, 2014). In the same industry only four years later, Colby Melvin (now an LGBTQ activist) was fired from his position as vice president of
Mobile, allegedly for the same reason according to Melvin himself (the exact reason for he was fired is not known by the author) (Queerty). Research has concluded that the oil and gas industry is indeed a masculine industry (Mearns & Yule, 2009; Collins & Callahan, 2012). In fact, according to Collins and Callahan, “it is common for higher-ranking oil and gas
executives to own ranches or farms where they can have private getaways”, and that “narrative accounts from people within the industry seem to indicate playing out masculine ‘cowboy’ behaviours is indicative of executives, supervisors and employees throughout oil and gas companies” (2012, p. 461). They further argue that this masculine behaviour is transferred from the private sphere into the office, where it works against openly gay men – who are seen as less masculine because of their sexual orientation. Based on the arguments by Herek (2007) that all non-heterosexual minorities suffer from differential treatment and lower status, one may assume that Collins and Callahan’s (2012) findings can be transferred to all sexual minorities, beyond the scope of just gay men.
Even in more female-dominated industries, such as the Health Care industry, which is seen as more “gay-friendly”, homosexuality is not a non-issue. While it is certainly a more forgiving climate, even here gay men take measures to counter what they see as disadvantages of being not only gay, but perceived as gay – such as carefully choosing how to dress and shaping their bodies to give the impression of enhanced masculinity (Rumens & Kerfoot, 2009).
What do industries that are more likely to marginalise LGBTQ minorities have in common? One possible explanation may be that they have similar cultural values. Geert Hofstede, a pioneer in the research on national and organisational cultures, stated that “culture is the collective programming of the mind which distinguishes the members of one group or society from those of another” (Hofstede, 1984, p. 82). In his early work, he identified several dimensions of national culture. His research was ground-breaking, resulting in thousands of
empirical studies that sought to test the theories he had presented (Taras, Kirkman, & Steele, 2010). However, the dimensions of national cultures are not suitable to measure the culture in an organisation, due to the fact that national cultures are embedded in values, while
organisational cultures are embedded in practices (Boella & Goss-Turner, 2013). In his later work, Hofstede identified dimensions that were suitable to measure culture at the
organisational level, each with a set of expected and desirable behaviours (Hofstede, 1998). Of the dimensions he identified, three have been selected in this thesis to test whether they moderate the relationship between a leader’s sexual orientation and perceived effectiveness, based on identification with the leader, fit theory, and people-oriented leadership.
The first dimension is the parochial versus professional dimension. This dimension seeks to establish where people draw their identities from, whether it is from the organisation they work for or the job they do. In a parochial culture, people identify with the organisation’s norms and behaviours, and they feel that it matches those they are used to from home. When hiring in a parochial culture, social and family background is taken into consideration in addition to competencies that the job requires. Parochial cultures are short term-oriented, and identification with the leader is also central to this type of culture (Hofstede, 1998). This leader identification aspect is similar to that of transformational leadership, where it is one of the four mechanisms through which transformational leadership works. It can be argued that the leader identification works the same way in parochial cultures as it does in
transformational leadership. Opposite of the parochial dimension is the professional one. Here, people identify with the type of job they do, and they believe that people’s private lives are their own business, in which the latter can be seen in accordance with the findings of Griffith and Hebl (2002) showing that LGBTQ professionals who are not open at work show higher levels of dissatisfaction than those who disclose their sexuality at work. Thus, the following hypothesis is proposed:
Hypothesis 5: The parochial versus professional dimension of organisational culture
moderates the relationship between leader’s sexual orientation and perceived effectiveness such that in organisations with a parochial culture, LGBTQ leaders are perceived to be more effective than in organisations with a professional culture.
The second cultural dimension is the open systems versus closed systems. This dimension captures the communication climate in the organisation, and the difficulties newcomers face when integrating in the organisation (Hofstede, 1994). In an open culture, most people would fit into the organisation, and newcomers quickly feel at home. Communication is open and informal in this culture, both internal and external. Here, one might draw upon
person-organisation fit theory. This is the concept that an person-organisation attracts candidates with similar values as the organisation and selects the ones who seem most fit. However, and this is
perhaps most important in the case of open versus closed systems, through attrition candidates will see first-hand how compatible they are with the organisation and choose to stay or leave. P-E fit has shown to have several positive organisational outcomes, perhaps most importantly, it has shown to increase job performance (Caldwell & O’Reilly, 1990). In contrast, a closed systems culture holds the belief that only very special people fit into the organisation. Newcomers that do come in take at least a year to feel at home, and people in such a culture are believed to be closed and secretive. Considering the potential for fit and the time it takes before newcomers to feel at home, open systems cultures can be argued to have a positive effect on effectiveness ratings. As with the professional culture dimension, this closed and secretive culture can be seen in accordance with the findings of Griffith and Hebl (2002), who stated that non-disclosure around LGBTQ sexuality at work is related to dissatisfaction. Thus, the following hypothesis is proposed:
Hypothesis 6: The open systems versus closed systems dimension of organisational
culture moderates the relationship between leader’s sexual orientation and perceived effectiveness, such that in organisations with an open systems culture, LGBTQ leaders are perceived to be more effective than in organisations with a closed systems culture.
The final cultural dimension that is suggested to matter in the relationship between leader’s sexual orientation and perceived effectiveness is the employee versus work
dimension, which captures a concern for people as opposed to a concern for getting the job done. This dimension is related to Blake and Mouton’s (1964) managerial grid. In an employee-focused culture, people’s personal problems are taken into consideration, and the organisation takes responsibility to maintain the group’s welfare. This can be translated to a typical people-oriented environment. On the other side of the spectrum, a work oriented culture exerts a pressure to get the job done, which can be translated to a task-oriented environment (Hofstede, 1998). When people arrive at work, they ought to leave whatever domestic problems they might have at the door in order to be effective at work. Klenke (2004) found that people-oriented leadership style is considered a typical feminine leadership style. In line with the finding of Salter and Liberman (2016) in which gay men are seen as more equal to heterosexual women than men, one may argue that gay men have a higher cue
validity for being people oriented than task-oriented. For women leaders, Klenke (2004) made no distinction between being heterosexual and lesbian. Thus, the following hypothesis is proposed:
Hypothesis 7: The employee oriented versus job oriented dimension of organisational
effectiveness, such that in organisations with an employee oriented culture, LGBTQ leaders are perceived to be more effective than in organisations with a job oriented culture.
Methodology
The primary goal of this paper is to test the relationship between a leader’s sexual
orientation and perceived effectiveness, mediated by perceived prototypicality. Furthermore, it seeks to test the moderating effect of different leadership behaviours and organisational culture variables might have on the perceived prototypicality and effect. The following section will in detail provide an overview of the research design and how the research was conducted.
Sampling Strategy
Data for the study was collected together with the thesis supervisor and one other master’s student who was also writing a thesis about leader’s sexual orientation. Because the study measured perception of different aspects of leadership, it was necessary for the data to be collected in dyads. It was agreed that a minimum of 80 dyads would be collected, whereof 40 of the leaders would be heterosexual and 40 of the leaders would be LGBTQ. For this thesis, the number of dyads achieved was N=90. To qualify as a leader in the study, one would have to be organisationally, or hierarchically responsible for at least one subordinate.
Consequently, any person with leadership responsibilities would qualify, with the level of leadership ranging from line-manager to CEO.
Participants were recruited mainly through the authors’ personal networks, where they were asked if they wanted to participate in a master thesis study. Participants were also recruited outside the authors’ personal networks, for example through LGBTQ-specific Facebook pages, or through LGBTQ organisations, such as the Amsterdam Gay Pride
Foundation and the COC. Snowballing was also used, asking participants to state the name and email address of people they thought would want to participate.
If they agreed to participate, participants received an email (see Appendix) with information about the study, a link to a survey that was created for the study, as well as a personalised code they would fill in before commencing the survey. The code was necessary to match the members of a dyad. Each author was free to distribute codes of his or her own choosing, as long as participants were identifiable. I used a combination of my own initials, the initials of the dyadic member I knew personally or had contacted, as well as a number to indicate whether the participant was a leader or a subordinate. For example, the leader of one dyad was given the code MNSNAW01, and the subordinate member of the dyad was given the code MNSNAW02. Coding was also used to track the progress of any given dyad, as only complete dyadic responses could be used in the study. If one person had completed the survey and the other had not, the author responsible for recruiting that dyad could send a reminder.
Specifically, the data was collected in an online survey on the platform Qualtrics. The survey was constructed by the authors and overseen by the supervisor. Scales to measure different variables were found and if necessary, translated, in academic journals, such as the Journal of Applied Psychology. Depending on the role of the participant, either leader or subordinate, the respondent selected the appropriate role and was directed to the appropriate part of the survey. The survey included a number of questions relating to the workplace, the participant’s relation to and perception of the leader, or self-perception if the respondent was a leader, as well as the culture in the workplace.
Because of the sensitivity of the LGBTQ topic, the purpose of the study was only revealed to the leader participants, either when they were contacted about participation, or additionally through the email in which they received the link and the personalised code. The reason for this was that we did not want to “out” anyone at their workplace as people might be open
about their sexuality in their private lives, but not at work. For example, a leader who is openly gay in private but not open at work would not even remotely have suspicions raised about his or her sexuality among his or her subordinates - by them receiving an invitation to participate in a study about LGBTQ leaders. For this reason, the purpose of the study was not revealed to subordinate participants. This was reflected by the way the survey was designed so that subordinates would not be able to draw any conclusions about their leader’s sexuality from the questions. Consequently, any question regarding the aspect of sexuality was
duplicated with a questions about gender, age, and culture (see Appendix). Anonymity in participation was guaranteed.
The survey was created in English. When it was complete, it was translated by the two master’s students and the supervisor to Dutch, Norwegian, and German respectively. Then, it was back-translated to English by persons will good knowledge of the English academic language. These persons were selected by the students and the supervisor, each for their respective language. The reason for translating the survey was because it was thought it might attract more respondents if it was offered more languages. The reason for back-translating it to English was to make sure that the no aspects of any question were lost in the first
translation, so that the questionnaire was equally representative in Dutch, Norwegian and German as it was in English. As a final step, each translation was overseen by an academic native to each language, and if necessary, changes were made to the translation.
Of the 236 total responses, 90 dyads were complete, resulting in a final sample size of 180. The respondents were Dutch (57.2%), Norwegian (27.8%), German (5%), American (2.2%), British (2.2%), while the remaining were Swedish, Thai, Irish, Canadian, Polish, Danish, Persian, Moroccan and Chinese (5% altogether). Participants were between 18 and 65 years old, with leaders having a mean age of 42.89 (SD – 10.49), and subordinates having a mean age of 37.93 (SD – 12.19). The majority of leaders were men (71.1%), a minority were
women (25.6%), while some were part-time men (1.1%), part-time women (1.1%) and what was self-reported as “gender fluid/demiguy” (1.1%). Among the subordinates, the majority were women (63.3%), and the rest were men (36.7%). Among the leaders, 20% had
completed high school, 28.9% had completed a bachelor level education, 43.3% had
completed a master level education, and 7.8% had completed a PhD level education. For the subordinates, 20% had completed high school, 2.2% had completed vocational training. 45.6% had completed a bachelor level education, 30% had completed a master level education, while 2.2% had completed a PhD level education. Industry wise, respondents typically worked in the hospitality industry (18.7%), the educational industry (12.1%), and for NGOs (8.8%). Among the leaders, 47.8% viewed themselves as exclusively heterosexual, 2.2% as mostly heterosexual, 1.1% as bisexual, 6.7% viewed themselves as mostly lesbian or gay, while 42.2% saw themselves as exclusively lesbian or gay. Among the subordinates, 70% viewed themselves as exclusively heterosexual, 14.4% as mostly heterosexual, 2.2% as bisexual, 2.2% as mostly lesbian or gay, while 11.1% viewed themselves as exclusively lesbian or gay. Of the non-heterosexual leaders, 91.5% were men.
Independent Variable
The independent variable in the study is a leader’s sexual orientation. To measure it, the items from Brewster, Velez, Mennicke, and Tebbe’s (2014) study on transgender
employees were used. Both leaders and subordinates were asked to indicate their sexual orientation, ranging from exclusively heterosexual, mostly heterosexual, bisexual, mostly
lesbian or gay, exclusively lesbian or gay, and other self-specified sexual orientation, the
latter in which respondents wrote a sexual orientation that was not measured by the former alternatives, to capture the full range of the LGBTQ statuses. This was later recoded to a dichotomous variable where 0 = heterosexual orientation (exclusively heterosexual) and 1 =
non-heterosexual orientation (the rest of the alternatives). The recoding into a dichotomous variable was necessary in order to analyse the results, as this study compared heterosexual leaders on one hand to LGBTQ leaders on the other. Thus, leaders who were not exclusive heterosexual were in this study considered LGBTQ.
Dependent Variable
The dependent variable in this study is a leader’s perceived effectiveness. To measure it, the seven items used by Fielding and Hogg (1997) were used. The subordinate participants were asked to rate their supervisor on a 7-point Likert, from 1= “strongly disagree” to 7= “strongly agree”, based on the following items; “my supervisor…” a) helps the team achieve its goals; b) helps the team make decisions; c) encourages and supports team members; d) is aware of others’ needs; e) helps to motivate the team; f) communicates well with team members; and g) is effective in influencing the team.
Mediating Variable
The mediating variable in this study is a leader’s perceived prototypicality. To measure this, subordinate respondents were presented a list of 28 attributes and asked how well each of the attributes matched their image of their supervisor on a 7-point Likert Scale from 1 = “strongly disagree” to 7 = “strongly agree”. 21 of the items were taken from the revised questionnaire on implicit leader theory by Epitropaki and Martin (2004), and were for example clever, knowledgeable, dynamic, understanding, and intelligent. Since all the items measured leader prototypicality, seven filler items were added in order to avoid social
desirability bias – so respondents wouldn’t suspect what was measured and answer according to what would look good. These seven items included attributes such as urbane, providential, and outdoorsy.
Moderating Variables
The first of the moderating leadership variables in this study is transformational
leadership. To measure it, subordinate respondents were asked to indicate to which extent
they agreed with 11 statements about their leader, on a 7-point Likert scale from 1 = “strongly disagree” to 7 = “strongly agree”. The 11 items were from the CLIO (Charismatic Leadership in Organisations), by De Hoogh, Den Hartog, and Koopman (2004), and included statements such as “My supervisor…” ”has a vision and imagination of the future”; “displays conviction in his/her ideals, beliefs, and values”; “encourages subordinates to be independent thinkers”; and ”talks with subordinates about their important values and beliefs”.
The second moderating leadership variable in this study is contingent reward leadership. To measure this, the subordinate respondents were asked to indicate to which extent they agreed with five statements about their leader, on a 7-point Likert scale from 1 = “strongly disagree” to 7 = “strongly agree”. The items were taken from Podsakoff, MacKenzie, Moorman, and Fetter (1990), and included statements such as “My supervisor…”: “always gives me positive feedback when I perform well”; “commends me when I do a better than average job”; and “personally compliments me when I do outstanding work”.
The third and last moderating leadership variable in this study is participative leadership. This was measured using five statements in which respondents were asked to indicate to which extent they agreed with the statements, on a 7-point Likert scale from 1 = “strongly disagree” to 7 = “strongly agree”. The items were taken from Indvik (1986, 1988), which were based on House and Dressler (1974), as published in Buengeler et al (2016). The items included statements such as “My supervisor…”: “consults with subordinates when facing a problem;” ”asks for suggestions from subordinates concerning how to carry out assignments”; and ”asks for suggestions on what assignments should be made”.
The first moderating organisational culture variable is the parochial versus professional dimension. Respondents were asked to which extent they agreed with 3 statements, on a 7-point Likert scale from 1 = “strongly disagree” to 7 = “strongly agree”. The items were taken from Hofstede (1998), and were as follows: “Where I work…”: “people’s private life is their own business”; “job competence is the only criterion in hiring people”; and “people tend to think three years ahead or more”.
The second moderating organisational culture variable is the open systems versus closed
systems dimension. To measure this, respondents were asked to which extent they agreed with
3 statements, on a 7-point Likert scale from 1 = “strongly disagree” to 7 = “strongly agree”. Taken from Hofstede (1998), the statements were as follows: “Where I work…”: “only very special people fit in the organisation”; “new employees need more than a year to feel at home”; and “the organisation and people are closed and secretive”.
The third and final moderating organisational culture variable is the employee versus work dimension. This was measured with 3 statements to which respondents were supposed to answer the extent they agreed with the statements. Taken from Hofstede (1998), the items were as follows: “Where I work…”: “important decisions are made by individuals”; “the organisation is only interested in the work people do”; and “there is little concern for personal problems of employees”.
Control Variables
Three control variables were used in this research, namely leader’s age, gender, and subordinate’s level of education. Leader’s age has been shown to be an important attribute of leader prototypicality, as shown by Buengeler et al (2016), where older leaders are perceived as more prototypical than younger leaders. And due to the close link between prototypicality and perceived effectiveness (Hogg, 2001; van Knippenberg & Hogg, 2003; Pierro, Bonaiuto,
van Knippenberg, & Kruglanski, 2005), leader’s age might influence effectiveness ratings. Thus, a control was added for age.
The gender of the leader has been shown to influence effectiveness ratings. Van Fleet and Saurage (1984) stated that the performance of female leaders is frequently subjectively evaluated less favourably than identical performance by males. Eagly, Makhijani, and
Klonsky (1992) also found in their meta-analysis that there is an overall tendency for subjects to evaluate female leaders less favourably than male leaders. Thus, a control was added for gender.
The third control variable was subordinates’ level of education, as research has shown that people’s level of education influences their attitudes towards LGBTQ people, where people with low levels education have more negative attitudes and people with high levels of
education have more positive attitudes (Herek, 1996; Smith, 1998). Thus, a control was added for subordinate’s level of education.
Additionally, since this study research was a multi-lingual and multi-country one, it was important to make sure that the translations of the survey correct – that content equivalence was guaranteed in the translated versions. To ensure this, back-translation was used, a
commonly used method of translation introduced by Brislin (1970). Bilingual speakers would first translate the survey from English to Dutch, Norwegian and German, and then other bilingual speakers would translate it back to English. The accuracy of the back-translated version was considered an indicator of the accuracy of the target translation.
Statistical Approach
In this research, the statistical analysis was executed using the statistics tool SPSS Statistics. First, a frequency check was run to make sure there were no missing values. Then the data was checked for normality. Further, descriptive statistics were provided to describe
the data sample, after which hierarchical regressions were run to test the relationship between leader’s sexual orientation, and leader’s perceived prototypicality and leader’s perceived effectiveness. Then, moderation effects were tested using hierarchical regression, and finally mediation, moderation, and moderated mediation tests were run using the SPSS add-on PROCESS.
Results
This section contains the findings of this research, and the structure is as follows: first, descriptive statistics will be provided, followed by hierarchical regressions to test Hypotheses 1 a through c to see if leader’s sexual orientation is able to predict leader’s perceived
prototypicality and perceived effectiveness, as well as the mediating effect of perceived prototypicality, while controlling for age, gender, and subordinates’ level of education Then, moderated regressions will be run to test moderation effects, as proposed in Hypotheses 2 through 7. As a final step, mediation, moderation and mediated moderation tests will be run in PROCESS.
Descriptive Statistics
The data was checked for missing cases, in which only the variables transformational
leadership and perceived leader effectiveness were missing responses. In both cases, it was
one item measuring the variables that were missing. This was dealt with by excluding cases list-wise, meaning that only cases without missing data were analysed for measuring
reliability, where the two aforementioned variables were measured with N=89, whereas the other variables were measured with N=90 – which was the total number of dyads. However, the scale means and other descriptive statistics for all variables were computed using all dyads (N=90). Counter-indicative items were recoded.
The skewness and kurtosis of the variables were tested to test for the normality of the distribution of the data. As a rule of thumb, if the absolute value of the skewness is between -1 and +-1, the data is within reasonable limits of normal distribution. The skewness for
perceived leader effectiveness was -2.107, perceived leader prototypicality was -.096, transformational leadership was -1.364, participative leadership was -.623, contingent reward leadership was -1.072. , parochial versus professional was .337, open systems versus closed systems was .318, and employee versus work was .294. By the rule of +/-1 for absolute
value, perceived leader effectiveness, transformational leadership, and contingent reward
leadership needed adjusting in order to meet the criteria of normal distribution. This was done
using negative (reflective) log10, in which the new skewness of each variable was .309, -.005, and -.409 respectively after the correction, falling into the +/-1 rule of thumb.
As a next step, the variables were checked for their reliability indicating how consistent the measures are. The Cronbach’s alpha of each variable has been added to a correlation matrix, showing the variables’ means, standard deviations and correlations, as can be seen in Table 1 below.
Table 1
Means, Standard Deviations, Correlations, and Reliabilities
Variable M SD 1 2 3 4 5 6 7 8 9 10 11 12
1. Age 25.89 10.49
-2. Gender (1 = female, 2 = male) 1.83 .753 -.72 -3. Education 3.21 .868 .187 .054 -4. Sexual orientation (0 = exclusively
heterosexual, 1 = non-heterosexual) .522 .502 -.206 .262* -.194 -5. Perceived prototypicality 4.969 .499 .074 .111 -.067 .405** .707 6. Transformational leadership .368 .154 .222* -.013 .0183 -.356** -.039 .914 7. Contingent reward leadership .459 .181 -.067 .102 .147 .103 .055 -.103 .900 8. Participative leadership 5.327 .784 -.015 -.120 .118 -.004 -.216* -.580** .172 .670 9. Parochial vs professional 3.693 1.168 .056 .014 .112 .209* .062 -.284** -.203 .217* .508 10. Open systems vs closed systems 2.411 .871 -.191 .168 -.025 .189 .165 .036 .183 .111 -.077 .612 11. Employee vs work 2.841 1.012 -.001 -.016 .006 -.048 .190 .325** -.187 -.326** -.041 .028 .631 12. Perceived leader effectiveness .323 .160 .222* -.052 .099 -.275** .035 .724** -.083 -.556** .062 -.045 .221* .915
Note. Coefficient alpha reliability estimates are on the diagonal. N = 90.
*. Correlation is significant at the .05 level (2-tailed) **. Correlation is significant at the .001 level (2-tailed)
According to DeVellis (2003), the coefficient should ideally be above 0.7.
Participative leadership, parochial versus professional, open systems versus closed systems,
and employee versus work have a reliability of .670, .508, .612 and .631 respectively, which might suggest that they are inconsistent. However, for measures with less than 10 items, it is not uncommon to see low Cronbach’s alpha (Pallant, 2011). This is the case for all of the variables, which have 3, 3, 3, and 5 items respectively.
Looking at the correlations, we see that there is a significant positive correlation between sexual orientation and perceived leader prototypicality. Further, there is a significant negative correlation between participative leadership and perceived leader prototypicality, and between participative leadership and perceived leader effectiveness. There is also a significant positive correlation between transformational leadership and perceived leader
effectiveness. Lastly, we see that there is a significant positive correlation between the
employee versus work dimension of organisational culture and perceived leader effectiveness.
Regressions and Mediation Analyses
Hypothesis 1a stated that LGBTQ leaders are perceived as less prototypical than heterosexual leaders. A simple linear regression was calculated to predict perceived
prototypicality based on leader’s sexual orientation. The result can be seen in Table 2 (Model 2), and reveals that, after controlling for leader’s age, gender, and subordinate’s level of education, leader’s sexual orientation positively predicted perceived prototypicality. This effect is statistically significant from 0, t = 3.7597, p < 0.001, 95% CI [0.192, 0.623]. Thus, Hypothesis 1a is rejected – LGBTQ are leaders are not perceived to be less prototypical that heterosexual leaders.
Hypothesis 1b stated that LGBTQ leaders are perceived as less effective than heterosexual leaders. To test this, a simple linear regression was performed to predict perceived
effectiveness based on leader’s sexual orientation. The result of this regression can be seen in Table 2 (Model 4), and reveals that, after controlling for leader’s age, gender, and
subordinate’s level of education, leader’s sexual orientation positively predicted perceived effectiveness. This effect is statistically significant from 0, t = 3.353, p < 0.001, 95% CI [0.286, 1.118]. Thus, Hypothesis 1b is rejected – LGBTQ leaders are not perceived as less effective than heterosexual leaders.
Hypothesis 1c predicted that the relationship between leader’s sexual orientation and perceived effectiveness would be partly mediated by perceived prototypicality. The four-step procedure for mediation described by Kenny, Kashy, and Bolger (1998) was followed, and as in the previous analyses, leader’s age and gender, and subordinate’s level of education was controlled for. As a first step, leader’s sexual orientation should be related to perceived effectiveness. This requirement was supported by the regression run testing Hypothesis 1 (Table 2, Model 2). The second step requires leader’s sexual orientation to be related to perceived prototypicality. This requirement was supported by the regression run testing Hypothesis 2 (Table 2, Model 4). The third step requires perceived prototypicality to be related to perceived effectiveness. To test this, a regression was testing the relationship between the two variables. The test revealed that, after controlling for leader’s age and gender, and subordinate’s level of education, perceived prototypicality did not predict perceived effectiveness ([beta]= .05, p = .65, 95% CI [-0.315, 0.504]).
Because the third requirement was not met, this would suggest that no mediation effect was present. However, the procedure for testing mediation as described by Baron and Kenny (1986), Judd and Kenny (1981), and James and Brett (1984) is slightly different is step 3 than that of Kenny et al. (1988), where both the independent variable (sexual orientation) and mediating variable (perceived prototypicality) should be included in the third step, because the mediator and dependent variables may be correlated because they are both caused by the
independent variable. Hence, the third step was re-run including both the mediating and the independent variables to see if the result would differ. As can be seen in Table 2 (Model 6), there was no difference in significance, and thus, Hypothesis 1c is rejected – the effect of leader’s sexual orientation on perceived effectiveness is not mediated by perceived prototypicality.
Regressions and Moderation Analyses
To test Hypotheses 2 through 7, moderated hierarchical regressions were performed to explore the relationship between the independent variable, sexual orientation, and the
dependent variable, perceived leader effectiveness, after controlling for leader’s age, gender, and subordinate’s level of education. Moreover, in these regressions, the main relationship was tested for a moderation effect by the different leadership styles and organisational culture dimensions. Two regressions were performed, one with the leadership style as moderators, the other with the organisational culture dimensions as moderators.
Before the regressions were performed, the numeric variables (that were not binary) were standardised. These included the moderator variables (transformational leadership, contingent reward leadership, participative leadership, parochial versus professional, open systems versus closed systems, and employee versus work), as well as the control variables.
Table 2
Regressions of Leader's Sexual Orientation on Perceived Prototypicality and Effectiveness, and the Mediating Effect of Perceived Prototypicality
Variables β SE β SE β SE β SE β SE β SE
Age .076 .053 .147 .050 -.182 .101 -.116 .097 -.182 .101 -.103 .099 Gender (1 = female, 2 = male) .243* .119 .076 .122 -.117 .227 -.270* .236 -.117 .227 -.263* .237 Education .045 .053 .076 .050 -.034 .101 -.006 .096 -.034 .101 .001 -.096 Sexual Orientation (0 = exclusively heterosexual, 1 = non-heterosexual) .410*** .108 .375** .209 .413** .226 Perceived Prototypicality -.093 .210 Model F 2.108 5.356** 1.492 4.063** 1.492 3.379** R2 0.608 .201 .049 .160 .049 .167 N 90 90 90 90 90 90 Statistical significance: *p<.05; **p<.01; ***p<.001
Perceived Prototypicality Perceived Effectiveness Perceived Effectiveness Model 5 Model 6 Model 1 Model 2 Model 3 Model 4
subordinate’s level of education). Second, the independent variable and the moderator variables were entered (transformational, contingent reward, and participative leadership in the first regression, and parochial versus professional, open systems versus closed systems, and employee versus work in the second regression). In the third step of the model, the interaction terms between the independent variable leader’s sexual orientation and the different moderators were added in the respective regressions. Tables 3 and 4 display the results of these regressions.
Hypothesis 2 predicted that transformational leadership will positively moderate the relationship between leader’s sexual orientation and perceived effectiveness such that LGBTQ leaders who score high on transformational leadership are perceived to be more effective than those who score low. From the interaction effects in Table 3 (Model 3), we see that the transformational leadership style has a positive, significant moderating effect on the relationship between leader’s sexual orientation and perceived leader effectiveness. Thus, for LGBTQ leaders as compared with heterosexual leaders, the positive relationship with
perceived effectiveness is stronger when transformational leadership is high rather than low. In order to look at the transformational leadership moderator further, the unstandardised regression coefficients were plotted to look at the slopes. The result is presented in Figure 1.
The regression model shows a positive beta for sexual orientation. This can lead one to think that LGBTQ leaders have higher perceived effectiveness. However, since
transformational leadership is a moderator, one has to analyse the difference on perceived effectives for LGBTQ vs heterosexual leaders across levels of transformational leadership. The positive effect of LGBTQ leaders is only observed in high transformational leadership, while under low transformational leadership heterosexual leaders are perceived as more effective which can be seen in Figure 1. The moderator effect of transformational leadership implies that the perceived effectiveness of heterosexual and LGBTQ leaders is different
across the levels of transformational leadership. When transformational leadership is low, heterosexual leaders have higher perceived effectiveness, whereas under high levels of transformational leadership LGBTQ leaders have higher perceived effectiveness compared to heterosexuals. Moreover, the difference between heterosexual and LGBTQ leaders on
perceived effectiveness is reversed across the levels of transformational leadership. As a last step to analyse transformational leadership as a moderator, bootstrapping was run using the SPSS add-on PROCESS, in order to see at which levels of the moderator the effect was significant. The results indicate that that only high levels of transformational leadership were significant, p = .075, 95% CI [.170, 1.069]. From the analyses, we can say that Hypothesis 2 is supported.
There was no support for any of the other moderator hypotheses, as can be seen in Tables 3 and 4, as none of the interaction terms were significant (all had significance p>.05). Hence, Hypotheses 3 through 7 are rejected.
Table 3
Regressions and Moderation Analysis of Leader's Sexual Orientation on Perceived Effectiveness
Variables β SE β SE β SE
Age -.174 .102 -.029 .073 -.037 .072
Gender (1 = female, 2 = male) .019 .133 .036 .097 .050 .093
Education -.047 .102 -.068 .071 -.059 .068
Sexual Orientation (0 = exclusively
heterosexual, 1 = non-heterosexual) .069 .160 .078 .153 Transformational Leadership -.557*** .097 -.771*** .130
Contingent Reward Leadership -.027 .071 -.108 .095
Participative Leadership .232* .091 .225 .135
Transformational x Sexual Orientation .323* .187
Contingent Reward x Sexual Orientation .102 .140
Participative x Sexual Orientation .021 .176
Model F 1.076 14.829*** 12.619***
R2 .036 .559 .615
N 90 90 90
Statistical significance: *p<.05; **p<.01; ***p<.001
Perceived Effectiveness
Model 1 Model 2 Model 3
Table 4
Regressions and Moderation Analysis of Leader's Sexual Orientation on Perceived Effectiveness
Variables β SE β SE β SE
Age -.174 .102 -.147 .097 -.148 .103
Gender (1 = female, 2 = male) .019 .133 -.037 .130 -.031 .134
Education -.047 .102 -.068 .097 -.067 .098
Sexual Orientation (0 = exclusively
heterosexual, 1 = non-heterosexual) .203 .203 .191 .207
Parochial vs Professional .245* .097 .387* .156
Open systems vs Closed Systems -.004 .098 .024 .162
Employee vs Work -.202* .093 -.249* .138
Parochial vs Professional
x Sexual Orientation -.166 .201
Open systems vs Closed Systems
x Sexual Orientation -.035 .211
Employee vs Work x Sexual Orientation .113 .192
Model F 1.076 2.978** 2.221*
R2 .036 .203 .219
N 90 90 90
Perceived Effectiveness
Additional Analyses
In addition to the analyses pertinent to the hypothesis tests, some additional analyses were run. These include a moderated mediation of the effect transformational leadership has on the relationship between a leader’s sexual orientation and perceived effectiveness,
mediated by perceived prototypicality, and regression analyses of each of the leadership style moderators with the other leadership styles as additional control variables.
Starting with the former, a moderated mediation analysis was run using bootstrapping in PROCESS to test the moderated mediation for transformational leadership and its effect on the relationship between leader’s sexual orientation and perceived effectiveness, mediated by perceived prototypicality. The reason to test for this is the nature of the transformational leadership style, with its emphasis on identification with the leader. There is precedence to believe that having a transformational leadership style will make a person seem more
prototypical as detailed in the section 2.2.1. The results of the bootstrapping showed that there was no significant moderated mediation present for any level of the moderator. Low level
Figure 1
Transformational Leadership as Moderator on the
Relationship Between Sexual Orientation and Perceived Effectiveness
1 2 3 4 5 6 7 Low Transformational
Leadership High Tranformational Leadership
P er ce ive d E ff ec ti ve n es s Heteresexual LGBTQ
showed a 95% CI [-0.1954, 0.151], medium level showed a 95% CI [-.1541, .1366], and high level of the moderator showed a 95% CI [-.1474, .1203].
In the additional regression analyses, each leadership style was tested as a moderator on the main effect between a leader’s sexual orientation and perceived effectiveness, although at this stage, the other leadership variables were added as control variables. The reasons for why this was done at this stage, and not earlier, was to see if any of the leadership variables might have any intervening effect on the particular moderator – to see if there would be a difference in significance. The second reason this was not done in main analyses is that too many control variables might mask the effect of the main relationship. For transformational leadership the result was p = .003, 95% CI [.156, .719], for contingent reward leadership the result was p = .441, 95% CI [-.172, .392], and for participative leadership the result was p = .073, 95% CI [-.518, .024]. To summarise, none of the leadership styles had resulted in any change in significance of the respective other leadership styles when included as control variables.
The same was done for the organisational culture dimensions. For parochial versus professional the result was p = .256, 95% CI [-.598, .161]. For open versus closed the result was p = .722, 95% CI [-.328, 472.]. Lastly, for employee versus work the result was p = .352, 95% CI [-.194, .540]. To summarise, none of the organisation culture dimensions had resulted in any change in significance of the respective other organisational culture dimensions when included as control variables.
