The adoption of human resource practices to support employees affected by intimate partner violence: Women representation in leadership matters
Abstract
Intimate partner violence (IPV) is a global public health issue that negatively impacts organizations and their employees. Research suggests that organizations can play a supportive role to lessen this negative impact. However, it has been relatively silent on the conditions under which organizations choose to play such a role. Integrating social role and critical mass perspectives, we examine the extent to which organizations adopt human resource (HR) practices to support employees affected by IPV. Specifically, we argue that organizations are more likely to adopt IPV-related HR practices when they are led by female Chief Executive Officers (CEOs) and Top Management Teams (TMTs) with more female members. Furthermore, we argue that when women's representation reaches a critical mass plateau, appointing more women in TMTs has no incremental impact, and this non-linear relationship moderates the CEO gender effect. Overall, we found support for our hypotheses based on a survey study of HR professionals from 414 Australian organizations (Study 1) and an archival study using 2 years of the Workplace Gender Equality Agency data from 4186 Australian organizations (Study 2). Theoretical and practical implications on the influence of gender configurations in leadership positions on the adoption of diversity, equity, and inclusion-related HR practices are discussed.
Intimate partner violence (IPV), commonly referred to as domestic violence, is defined as behavior directed toward the goal of inflicting physical, psychological, emotional, and sexual harm to a former or current intimate partner (Krug et al., 2002). It is considered “a global public health problem of epidemic proportions” (World Health Organization [WHO], 2013, p. 3) and results in social problems affecting every life domain (Golding, 1999; Moe & Bell, 2004). Although IPV is commonly experienced by both men and women, more women suffer from it, and they incur more serious injuries that require medical attention (Rennison & Welchans, 2000). Similar to other countries, in Australia, the context of our research, one in six (one in four) women have been subjected to physical and/or sexual (emotional) abuse by a current or former partner since the age of 15 (Australian Bureau of Statistics, 2016; WHO, 2013). Regrettably, this problem has been exacerbated by COVID-19 lockdowns as partners spend more time together in confined spaces, as well as by the ensuing global economic recessions, the financial repercussions of which may make it even harder for IPV victims to leave abusive relationships (Boserup et al., 2020; Bradbury-Jones & Isham, 2020).
While IPV is an issue of societal relevance, it also has considerable relevance to organizations and the individuals who work in them. Consistent research evidence suggests that IPV negatively impacts organizations and their employees in a variety of ways (e.g., Deen et al., 2021; Duffy et al., 2004; LeBlanc et al., 2014; Tolentino et al., 2016; for a comprehensive review, see Deen et al., 2022). At the organizational level, IPV has been found to incur direct (e.g., illness, death, absenteeism, and turnover), indirect (e.g., reduced commitment, morale, and performance), and intangible costs (e.g., negative impact on company image and work climate; O'Leary-Kelly et al., 2008) for organizations. At the individual level, IPV has been found to negatively impact women's career advancement. Specifically, Tolentino et al. (2016) showed that IPV experienced at home significantly interfered with both employees’ in-role performance and citizenship behaviors at work. Importantly, however, this relationship was weaker for employees with high (as opposed to low) perceived organizational support, suggesting that organizations can play a crucial role in combating the negative crossover effects of IPV from home to work.
Despite the role organizations play in reducing the negative impact of IPV on employees in the workplace, it is unclear why organizations differ widely in their response to this pressing problem. Some organizations actively adopt human resource (HR) practices, such as offering paid leave to affected employees and training opportunities to help supervisors recognize IPV and support employees whose performance is impacted by IPV, while others remain completely unresponsive to this issue. For example, the Society for Human Resource Management (SHRM, 2013) reported that 65% of the companies surveyed did not have any formal workplace domestic violence prevention policy and 80% did not have formal training programs for domestic violence in place. Our research, therefore, focuses on the conditions under which organizations adopt IPV-related HR practices for two reasons. First, IPV is a public health problem that is potentially on the rise globally (WHO, 2021). Second, although research suggests that organizations, via their management of human resources, have a positive role to play in tackling this global challenge (e.g., Tolentino et al., 2016), organizations seem to be particularly divided in terms of seeing IPV as relevant to the workplace and vary in their levels of responsiveness in managing this issue.
We argue that a fruitful avenue to understanding the different organizational responses is through the lenses of leadership and gender. First, leaders have an important role to play in initiating and implementing practices to protect and support their employees affected by IPV (e.g., Liz Claiborne, Inc. as examined by O'Leary-Kelly et al., 2008). Second, female leaders, in particular, may be more likely to support and implement socially responsible and family friendly practices because their orientation towards leadership could partly be shaped by existing gender roles and expectations (e.g., Cook & Glass, 2016, 2017; Glass & Cook, 2018; see also de Jonge, 2018). This gender-based leadership perspective suggests that female leaders, who act in accordance with their gender role, may consider IPV as an issue that falls under their care and therefore choose to offer organizational support to employees who are victimized by it in their organizations (de Jonge, 2018; Wilcox et al., 2021). To this end, we examine how the gender of Chief Executive Officers (CEOs) and the gender representation of Top Management Teams (TMTs) separately and jointly shape the extent to which organizations adopt HR practices to support employees affected by IPV.
Our research contributes to the management and diversity literature in three important ways. First, despite a growing body of research on the negative impact of IPV on organizations and their employees (e.g., LeBlanc et al., 2014), our knowledge about how organizations respond to IPV is limited (Adhia et al., 2019). Given that employment can produce a critical path for IPV victims to exit abusive relationships (McFerran, 2011) and in light of the evidence that organizational actions can mitigate the detrimental impact of IPV on employees (Tolentino et al., 2016), we argue that understanding the conditions under which organizations are more or less likely to adopt HR practices to assist employees in these situations represents a moral imperative (cf. de Jonge, 2018; Wilcox et al., 2021). Our research also responds to the call for more empirical work that documents the effects of employer actions that support employees affected by IPV (Deen et al., 2022).
Second, we respond to Kulik and Metz's (2017) call for more research on how the gender diversity of a top management team (TMT) influences organizational outcomes, including the adoption of organizational practices that promote socially responsible behavior. To date, the few studies that have examined the impact of decision-makers’ gender on the adoption of equity practices to support women and other underrepresented groups have mainly focused on that of CEOs and/or board of directors (BoD; e.g., Cook & Glass, 2016; Glass & Cook, 2018; Srikant et al., 2020). While BoD and TMT are both responsible for the most important strategic decisions (e.g., Kulik & Metz, 2017), TMT is different in its reporting line to the CEO and its unique interface with middle management (Raes et al., 2011). The extent to which organizations adopt practices that support the well-being of their employees both in the workplace and at home may be influenced by TMTs’ understanding of employees’ needs via their unique interface with middle management. The support of such practices by both the CEO and TMT, as well as how they work together, as we will argue, shape the adoption of IPV-related HR practices. Consequently, we focus on two bodies of leadership, namely CEO and TMT to advance research on the gender composition of leadership and its impact on the adoption of diversity, equity, and inclusion (DEI)-related HR practices.
Third, we integrate social role theory (e.g., Eagly & Wood, 2012) and the critical mass perspective (Dahlerup, 2006; Kanter, 1977) to offer novel insights into the influence of CEO gender and gender representation in the TMTs on the adoption of DEI-related HR practices. Current research shows that women leaders tend to act in accordance with their social role expectations to make more other-oriented decisions, such as those related to corporate social responsibility (CSR; Cook & Glass, 2017). Research also shows that organizations with BoD with a critical mass of a minimum of three women have stronger CSR records than those with two or fewer women on their BoD (Cook & Glass, 2017). This line of research typically assumes that the positive impact of a critical mass of women on organizational outcomes to be linear (e.g., Abebe & Dadanlar, 2019; Cook & Glass, 2017; Konrad et al., 2008). However, we draw upon research on gender and group proportions and associated dynamics (e.g., Allmendinger & Hackman, 1995; Post et al., 2022) to theorize and empirically demonstrate that there is a plateau to this critical mass in our context. That is, the positive relationship between the percentage of women in TMTs and adoption of IPV-related HR practices levels off after this critical mass plateau is reached. Furthermore, we theorize and show how this non-linear effect moderates the relationship between CEO gender and the adoption of IPV-related HR practices. We distinguish the term “critical mass threshold” from “critical mass plateau” when describing women's numerical representation in the TMTs. We use the term “critical mass threshold” to describe the point at which women are able to exert influence (i.e., below this threshold, women are in a token position). We used the term “critical mass plateau” to describe the point at which adding another woman to the TMTs no longer increases the likelihood that IPV-related HR practices are adopted. While social role theory informs us of the likely nature of the decisions made by influential women versus men, we draw upon the critical mass perspective to extend the predictive reach of social role theory to reveal how the influence of CEO gender and gender representation in TMTs on DEI-related HR practices is more complex than originally theorized and demonstrated. In this way, our research meaningfully extends the literature on the largely linear influence of female representation in leadership positions on DEI-related practices and outcomes in organizations (e.g., Abebe & Dadanlar, 2019; Ali & Konrad, 2017; Cook & Glass, 2016; Dobbin et al., 2011; Glass et al., 2020; Glass & Cook, 2018).
Below, we explicate our research model, develop hypotheses, and report two field studies that test our hypotheses. In Study 1, we conducted a survey of HR professionals from 414 Australian organizations. Study 2 constructively replicates and extends Study 1 findings by providing evidence for stronger causal inferences via the analysis of two-waves of archival data (i.e., 2015–2016 and 2016–2017) from the Workplace Gender Equality Agency (WGEA).
1 THEORY AND HYPOTHESES
1.1 CEO gender
Social role theory (see Eagly & Wood, 2012 for a comprehensive review) provides an overarching theoretical framework that supports our argument that women in decision-making positions, such as CEOs, are more likely than their male counterparts to endorse and adopt HR practices to support employees affected by IPV. This theory focuses on gender differences and thus enables us to specify the different orientations (e.g., characteristics, values, priorities, and behaviors) women and men bring to an organization (see Eagly et al., 2000 for a review of meta-analytic studies that support this theorizing). While men are socialized by society for openly engaging in agentic behaviors designed to maximize self-interests, women are expected by society to take on a different role—a role that requires them to prioritize communal and relationship-building behaviors that often support others' interests (Eagly & Wood, 2012). Importantly, when women violate prescriptive and proscriptive gender roles in organizations, they often are penalized (e.g., Eagly et al., 1992; Rudman, 1998).
Irrespective of the origins of gender differences, social role theory, in the context of organizational decision-making, suggests that female and male leaders hold priorities and values that correspond to their gender norms (e.g., Eagly, Johannesen-Schmidt, & Van Engen, 2003; Mendelberg & Karpowitz, 2016). For instance, prior studies (e.g., Cook & Glass, 2016; Glass & Cook, 2018) drew on social role theory and demonstrated that female CEOs tend to prioritize relationship building, as well as equity and diversity (i.e., goals consistent with the female gender role) and thus were more likely to champion equity and inclusion policies. In a similar vein, conforming to the widely and firmly held expectation that women should be caring and nurturing, research has shown that women with decision-making powers exhibit greater care for their employees and undertake fewer reductions in workforce than their male counterparts (Matsa & Miller, 2013; Tate & Yang, 2015).
In terms of values, a study of over 500 board members in Sweden demonstrated that female directors scored higher on Schwartz's (1992) universal values of benevolence and universalism than their male counterparts (Adams & Funk, 2012). These values emphasize concern for the welfare of others and are consistent with the female gender role. Extending their study, Adams et al. (2011) found that these value differences predicted decisions and behaviors at work, with benevolence and universalism negatively related to support for shareholder wealth maximization. Together, the two studies suggest that female leaders are less likely to support a “share-holderist” strategy than male leaders via their female gender role consistent values. Indeed, Cook and Glass (2017) found that female directors made more stakeholder-oriented and less shareholder-oriented decisions than male directors. Hence, female CEOs not only are more likely to prioritize employee well-being than male CEOs, but they may also think about employees’ well-being in a more inclusive manner (i.e., stakeholder perspective). We argue that female leaders may believe that their leadership and duty of care for their employees transcend beyond the workplace to include providing support for employees who encounter hardships at home. In support of this argument, Crowder-Meyer (2021) found that women and men deviate on issues that pull gender roles apart. That is, women showed concern for vulnerable populations including children and the poor, while men were more concerned about finances and taxes.
Given that the decision to adopt IPV-related HR practices exemplifies a more caring and communal orientation that is more consistent with the female rather than the male gender role, we argue that organizations led by female CEOs are more likely to adopt such HR practices than those led by their male counterparts. As such, we hypothesize:
Hypothesis 1.Organizations led by female CEOs will be more likely to adopt IPV-related HR practices than organizations led by male CEOs.
1.2 Gender representation in the TMTs
CEOs do not make organizational decisions in a social vacuum. Many organizational decisions, including those related to the adoption of HR practices, are also shaped by the TMT (Arthur et al., 2014). TMT is defined as the “inner circle of executives who collectively formulate, articulate, and execute strategic and tactical moves of the organization” (Klenke, 2003, p. 1024; see also Hambrick & Mason, 1984). As with female CEOs, social role theory predicts that female TMT members are more likely to endorse HR practices that support employees affected by IPV than their male counterparts. However, social role theory does not inform us of how this gender-based influence manifests in mixed-gender groups where both women and men may compete to obtain their gender-infused preferences adopted by their groups. To extend the predictive reach of social role theory, we draw upon the critical mass perspective (Dahlerup, 2006) to provide a more nuanced understanding of the effect of the percentage of women in TMTs and its resulting team dynamics on the adoption of IPV-related HR practices.
Originating from nuclear physics, the term “critical mass” has been used to explain the quantity necessary to create an irreversible turning point into a new situation (Dahlerup, 2006). Drawing on Kanter's (1977) seminal work on how group proportions change group dynamics as a theoretical foundation, Dahlerup (1988) developed the critical mass perspective based on Nordic data in the political domain. Applied in our research, this perspective suggests that for women to exert influence in the TMT, they need to reach a critical mass threshold within the team. Furthermore, drawing on research on gender and group proportions and associated dynamics (e.g., Mendelberg & Karpowitz, 2016; Post et al., 2022), we argue that women's influence levels off after women reach a certain critical mass.
Women who have not reached the critical mass threshold or are in a token position within their group (typically 15% or less) generally exert little influence in their groups (Kanter, 1977; see also Turco, 2010). This is because they tend to self-censor, but even when they do express their unique perspectives, their contributions are more likely to be dismissed by men. Self-censoring and lacking influence can both be explained by the processes of visibility, polarization, and assimilation (Kanter, 1977). Token women and their differences are more visible in male-dominated groups, which leads them to be self-conscious of their token status and experience undue pressure to perform well. Token women's presence can also stimulate the dominant male group to view its within-group similarity and the differences between the two gender groups in an exaggerated manner, which produces polarization between the two groups. The dominant male group also attempts to fit token women's attributes into its pre-existing feminine stereotypes (i.e., assimilation) and evaluates token women's views with bias. To cope with these pressures, token women often stay as low profile as possible to appear non-intrusive (Allmendinger & Hackman, 1995). This means that token women are more likely to conform to the perspectives of their majority male colleagues. If they raise their unique views in team discussions, their views are likely to be ignored or met with resistance. Indeed, research shows that priorities placed on “care issues” (i.e., those related to family and the vulnerable) by token women in male-dominated groups may not receive a full hearing (Mendelberg & Karpowitz, 2016).
Team dynamics change, however, when token women are joined by others to reach the critical mass threshold (i.e., between nine to 35%).1 This subgroup of women is now more likely to share, test, and pursue their agenda to become a stronger political force in their TMTs (Zanna et al., 1987). There are several reasons for why the presence of other women enhances token women's ability to advance their agenda (e.g., Biswas et al., 2021; Cook & Glass, 2017). First, homophily theory suggests that those who are demographically similar tend to be attracted to each other (Byrne, 1971), have shared knowledge and values, and thus are more likely to support each other (McPherson et al., 2001). Indeed, women, rather than men, are more likely to support other women for championing socially progressive agenda, such as implementing DEI policies (Cook & Glass, 2016). Increased support would ensure women's ideas receive adequate attention in team discussions and deliberations, which enhances the influence of the idea and increases its adoption. Second, in addition to having greater social influence, having other women in a team plays a vital symbolic role. Decision-making research in political science shows that women's presence in leadership provides symbolic representation, which enhances the perception that women, as a group, are competent, worthy of respect and well-suited to exercise power (Mendelberg & Karpowitz, 2016). As more women join the TMT and reach the critical mass threshold, we expect women's legitimacy as decision-makers to increase. This in turn enhances each woman's effectiveness in advocating for their preferences (cf. Post et al., 2022), including the adoption of IPV-related HR practices. Indeed, Hewstone et al. (2006) empirically showed that polarization and assimilation, which cause token women to self-censor and their ideas to be dismissed, play less of a role in the functioning of a team with a minority of women.
We further propose that the positive impact of greater women representation in TMT on IPV-related HR practices levels off after they reach a critical mass plateau. By the time they have reached the critical mass plateau, women as a group would have gained symbolic representation (Mendelberg & Karpowitz, 2016). This means that the TMT already values women's contributions, is open to gender diversity, and has established its support of the IPV agenda. Beyond the critical mass plateau, her role as a “woman” becomes less important than her role as an “individual” to influence the TMT (cf. Biddle's, 2013 role theory). As such, appointing another woman will not shift the TMT's decisions on IPV-related HR practices significantly. Consequently, there is a diminishing return in appointing another woman to the TMT on the adoption of additional IPV-related HR practices after the critical mass plateau is reached. In support of this argument, Post and her colleagues (2022) found that appointing a female member to the TMTs shifted the TMTs’ overall cognition to become more change-oriented, which in turn increased firms’ investments in R&D. However, when there were four or more women in the TMTs (i.e., when women's TMT representation was beyond approximately 35%), the influence of appointing another woman on the TMTs’ change-oriented cognitions waned. This is because the change-oriented cognitions of the TMT have already been significantly altered by the former three female incumbents when they were first appointed. Firms’ investments in R&D, in turn, did not increase when women reached this critical mass plateau in their TMTs.
In the context of our theorizing, when women representation goes beyond the critical mass plateau in the TMTs, we expect the diminishing return in appointing another woman to the TMT to cause the positive relationship between women in the TMTs and the adoption of IPV-related HR practices to level off. Following this line of reasoning, we predict a non-linear relationship between women's representation in the TMTs and IPV-related HR practices adoption. We propose:
Hypothesis 2.There will be a non-linear relationship between the percentage of women in the TMTs and the adoption of IPV-related HR practices, such that there will be no effect when the percentage of women in the TMTs is below the critical mass threshold and above the critical mass plateau, and a positive relationship when the percentage of women in the TMTs is between the critical mass threshold and critical mass plateau.
1.3 The moderating role of gender representation in TMTs on the CEO gender effect
Simsek et al.'s (2017) review of CEO-TMT interface research reported that TMT links and interacts with CEO in two ways; namely, in a sequential and/or reciprocal interdependent manner. The former is when the CEO is dependent on the TMT members, or vice-versa. The latter involves mutual interdependencies between the CEO and TMT, where there are back-and-forth communication and coordination between the two. For both interfaces, the resources that are exchanged between the parties include advice, information, psychological counsel, and task and social support (e.g., Arendt et al., 2005; Cao et al., 2009; Lin & Rababah, 2014). Although the type of interface between the TMT and the CEO may vary across organizations, both types of interdependencies enable TMT to exert influence on CEO's decisions (e.g., Arendt et al., 2005). Consequently, we will argue below that the percentage of women in the TMT would also shape the effect of having a female CEO on an organization's adoption of IPV-related HR practices.
As argued earlier, women representation in TMT influences the extent to which organizations adopt IPV-related HR practices in a non-linear fashion. Extending this argument, we propose that this influence can also extend to the CEO's effect on the adoption of IPV-related HR practices because of the interdependence between the CEO and the TMT. First, when the CEO is sequentially dependent on the TMT, the TMT provides advice to the CEO regarding which organizational issues need to be prioritized and adopted by the CEO (Arendt et al., 2005). The likelihood that IPV issues are considered in the CEO's agenda may depend largely on whether TMT believes these issues are important. When IPV issues do not get on the CEO's agenda, HR practices designed to support employees affected by IPV are less likely to be adopted by the organization regardless of the CEO's gender. Second, when the CEO is reciprocally interdependent with the TMT, the TMT influences the CEO primarily through advocacy, feedback, mutual influence, and team discussion (Lin & Rababah, 2014). In this case, the degree to which TMT supports IPV-related HR practices can shape the extent to which a CEO is able to implement their own preference on the issue. A TMT which does not support IPV-related HR practices may nudge the CEO to focus attention away from IPV issues, such as by questioning the legitimacy of the organization's involvement in employees’ family life and/or by raising other pressing issues and urging the CEO to deploy limited organizational resources into those issues. As such, regardless of a CEO's gender, the extent to which TMT supports IPV-related HR practices impacts the likelihood that the CEO would adopt IPV-related HR practices.
The above discussion suggests that the CEO gender effect on the adoption of IPV-related HR practices hypothesized earlier is likely to be weaker when women are in a token position. In contrast, the presence of a critical mass of women in the TMTs will likely provide female CEOs with important resources and support to overcome constraints stemming from their token status (Kanter, 1977) and to successfully advance their goal to support employees affected by IPV. In this way, the positive relationship between female CEOs and organizations' adoption of IPV-related HR practices will become stronger as more women are represented in the TMTs. As theorized in Hypothesis 2, there is a diminishing return in appointing another woman in the TMTs on the adoption of IPV-related HR practices when women representation passes the critical mass plateau. We, therefore, expect the positive moderating role of percentage of women in the TMTs to level off when women representation reaches the critical mass plateau. Based on these theoretical considerations, we propose:
Hypothesis 3.The percentage of women in the TMTs will moderate the relationship between CEO gender and the adoption of IPV-related HR practices, such that the effect of CEO gender on IPV-related HR practices (or the difference between the female CEOs-IPV-related HR practices and male CEOs-IPV-related HR practices relationships) will be stronger between the critical mass threshold and critical mass plateau and there will be no effect of CEO gender when the percentage of women in the TMTs is below the critical mass threshold and above the critical mass plateau.
2 STUDY 1
2.1 Method
2.1.1 Sample and procedure
A survey was distributed in 2017 to all members of the Australian Human Resource Institute (AHRI). AHRI members consist of mainly HR professionals and some line managers engaged in HR-related responsibilities. Members also consist of some student and academic members, who are not eligible to participate in the study. Three reminder emails were sent after the initial email invitation. Respondents completed the survey regarding their organization's adoption of HR practices to support employees affected by IPV and characteristics of the organization (i.e., CEO gender and percentage of women in the TMT). Data collected from 414 HR professionals from different Australian organizations were used to test H1-H3. We assert that these HR professionals are in the best position to complete the survey because they were responsible for and/or involved in the implementation of HR practices in their organizations. The average age of the respondents was 44.36 years (SD = 10.55) and most respondents were women (79%). In terms of academic qualifications, 21% of the respondents have a graduate certificate, 32% have a bachelor's degree, and 24% have a master's degree or a PhD (1%).
2.2 Measures
2.2.1 CEO gender and percentage of women in TMT
Respondents were asked to indicate the gender of their CEO (1 = female, 0 = male) and the percentage of women in the TMT in their organization (i.e., [number of female TMT members/total number of TMT members] x 100). Of the 414 organizations, 8.7% have no women in the TMT, and 13% of these organizations have more than 50% women in the TMT. The percentage of women in the TMT is not normally distributed (Kolmogorov–Smirnov [414] = 0.46, p < 0.01). The distribution is skewed to the left (i.e., there are more organizations with a percentage of women in the TMT between zero and 50, and very few organizations with more than 50% women in the TMT). Given the non-normal distribution, we centered this variable to enhance robustness in further analyses.2
2.2.2 IPV-related HR practices
After reviewing articles on IPV-related HR practices in academic and professional journals and newspapers and engaging in meetings with the AHRI Board, we selected the following five HR practices: (1) IPV leave, (2) training for supervisors to recognize IPV victims, (3) training for supervisors to help IPV victims to disclose IPV, (4) counseling for IPV victims, and (5) flexible work arrangements for IPV victims. Respondents were asked which of these practices currently exist in their organizations (response scale: “yes” or “no”). Following previous research (Boon et al., 2019), we calculated the total number of HR practices adopted in an organization (total number of ‘yes’) and treated this measure as a quasi-interval scale.
2.2.3 Control variables
To rule out alternative explanations in our findings, we controlled for (the log of) size and type of the organization (public/not-for-profit vs. profit) because female CEOs are more often appointed in smaller, public, and not-for-profit organizations (Gondhalekar & Dalmia, 2007; Khan & Vieito, 2013). Type and size of the organization are also related to the number of HR practices within an organization (Delaney & Huselid, 1996). We also controlled for the size of TMT because TMT size affects the degree of influence CEOs and individual TMT members have and is typically accounted for in research involving TMT demographic composition (Post et al., 2022).3 Respondents were asked to indicate the size of their organization (1 = non-employing/sole proprietor; 2 = small [2–49 employees]; 3 = medium [50–249 employees]; 4 = large [250 and more employees], type of the organization (0 = public/not-for-profit; 1 = for profit), and of TMT in absolute numbers).
Carlson and Wu (2012), and more recently Sturman, Sturman, and Sturman (2022), warned about the improper use of control variables. Carlson and Wu argued that controls are in general weakly related to the focal variables and rarely influence the interpretation of study results. Sturman et al. (2022) further found that the use of control variables can increase effect sizes and the probability of inappropriately detecting statistical significance. Thus, we follow their advice to conduct and report the following analyses with and without control variables.
2.3 Data analytic approach
Prior studies that have tested a moderating effect on a non-linear relationship include Baer and Oldham (2006), Kluger et al. (1994), Janssen (2001) and more recently, Yang et al. (2022). For instance, Janssen examined the role of fairness perceptions on the inverted U relationship between job demands and job performance. They found the effect of fairness perceptions on performance (or the difference between high fairness perceptions-performance and low fairness perceptions-performance relationships) to be strongest at intermediate levels of job demands. Similarly, Yang et al. studied the cross-level moderating effect of team task support on the nonlinear relationship between employee proactive personality and reflective learning. Following these studies' procedure of hypotheses testing, we mean centered the dependent and independent variables (Aiken & West, 1991), we added the controls in model 1, and added the gender of the CEO to test H1 in model 2. Next, we added the percentage of women in TMT to control for linear trend and the quadratic effect of the percentage of women in TMT in model 3 to test H2. In model 4, we added the two and three-way interaction terms (i.e., Gender CEO X percentage of women in the TMT and Gender CEO X percentage of women in TMT2) to test H3.
Although the organizations are nested within industries and differences in HR practices adoption between industries can be expected (see Study 2), we calculated the intraclass correlation (ICC1; Bliese, 2000) to understand this influence. The ICC1 did not show that the IPV-related HR practices differ across industries (ICC1 = 0.02). This finding may have been influenced by the high percentage of respondents who selected “other industry” (40%).4 Therefore, the data were analyzed using ordinary linear regression (OLR).
2.4 Results
2.4.1 Descriptive statistics and zero-order correlations
Table 1 shows that organizations have, on average, 2.29 IPV-related HR practices (SD = 1.58) in place out of the five practices we provided. The most common IPV-related HR practices include counseling (70%), flexible work arrangements (65%), and leave for IPV victims (53%). Few organizations have in place practices to train their supervisors on how to recognize IPV victims (18%) and how to help victims disclose IPV (15%). The correlation table shows a positive relationship between the percentage of women in TMT and IPV-related HR practices (r = 0.12, p < 0.01). For organizations led by female CEOs, there are more IPV-related HR practices (r = 0.14, p < 0.01). IPV-related HR practices are also more common in larger organizations (r = 0.13, p < 0.01), organizations with more TMT members (r = 0.12, p < 0.01) and not-for-profit organizations (r = 0.19, p < 0.01). As expected, organization size is moderately correlated with size of the TMT (r = 0.59, p < 0.01).
| Variables | Mean | SD | 1. | 2. | 3. | 4. | 5. |
|---|---|---|---|---|---|---|---|
| 1. IPV-related HR practices | 2.29 | 1.58 | |||||
| 2. CEO gender (1 = female; 0 = male) | 0.23 | 0.42 | 0.14** | ||||
| 3. % women in TMT | 37 | 22 | 0.12** | 0.47** | |||
| 4. Organization size | 3.52 | 0.69 | 0.13** | −0.24** | −0.15** | ||
| 5. Size TMT | 7.07 | 2.63 | 0.12** | 0.14** | −0.17** | 0.59** | |
| 6. Type of organization (1 = for profit) | 0.37 | 0.48 | −0.19** | −0.30** | −0.25** | −0.18** | −0.09 |
- ** p < 0.01;
2.4.2 Hypotheses testing
Table 2 shows the standardized coefficients of the regression analyses. To test H1, organizations led by a female CEO tend to adopt more IPV-related HR practices—the effect of CEO gender was added in step 2. In support of H1, this coefficient was significant (β = 0.17, p < 0.01).
| Variables | Step 1 | Step 2 | Step 3 | Step 4 | Step 5 (without controls) |
|---|---|---|---|---|---|
| Organization size | 0.04** | 0.04** | 0.04** | 0.04** | |
| Size of TMT | 0.04* | 0.04* | 0.04* | 0.04* | |
| Type of organization | −0.03 | −0.03 | −0.03 | −0.03 | |
| Gender CEO (1 = female; 0 = male; H1) | 0.17** | 0.17** | 0.17** | 0.21** | |
| % women TMT | 0.08** | 0.08** | 0.09** | ||
| % women TMT2 (H2) | −0.11** | −0.11** | −0.10* | ||
| Gender CEO x % women TMT | −0.02 | −0.01 | |||
| Gender CEO x % women in TMT2 (H3) | −0.05* | −0.04* | |||
| Explained variance | 0.07 | 0.11 | 0.18 | 0.21 | 0.12 |
- ** p < 0.01;
- * p < 0.05.
For H2, the non-linear effect of the percentage of women in TMT—both main effect (β = 0.08, p < 0.01) and the quadratic effect (β = −0.11, p < 0.01) were significant. We calculated the critical mass plateau by computing the first derivative of the main and quadratic effect and found the critical mass plateau is 34%. Before this critical mass plateau, the relationship between the percentage of women in the TMTs is 0.21 (p < 0.01), after the plateau, it is 0.04 (n.s.) supporting the hypothesized plateau effect. There is no significant difference in the steepness between the organizations with a percentage of women in the TMTs below the critical mass threshold/token position (β = 0.22, p < 0.01) and the ones with a percentage of women in the TMTs between the critical mass threshold and the critical mass plateau (β = 0.19, p < 0.01; t = 0.5, n.s.). Thus, H2 is partially supported in that there is a significant non-linear effect but rather than having no effect, organizations with percentages of women below the critical mass threshold are positively associated with IPV-related HR practices adoption.
To test H3, which proposes a moderating influence of the non-linear effect of the percentage of women in TMT on the relationship between CEO gender and IPV-related HR practices, we entered the two-way (CEO gender x percentage of women in TMT) and three way-interaction (CEO gender x quadratic effect of percentage of women in TMT) terms in step 4. In support of H3, this interaction is significant (β = −0.05, p < 0.01). Figure 1 displays the nature of the interactive effect in the first part of the non-linear relationship up until when there are 50% women and men in TMT. The figure shows that the CEO gender effect (or the differential impact of CEO gender) on IPV-related HR practices becomes stronger before and plateaus after the critical mass plateau. Indeed, the data show that the positive relationship between female CEO and IPV-related HR practices before the critical mass threshold is significantly stronger (simple slope = 0.29, p < 0.01; t = 2.01, p < 0.05) compared to the relationship between male CEO and IPV-related HR practices (simple slope = 0.10, p < 0.05). We also analyzed the steepness between the organizations with the percentage of women in the TMTs below the critical mass threshold, for female and male CEOs separately (β = 0.20, p < 0.01 for females and β = 0.12, p < 0.05 for males) and the ones with the percentage of women in their TMTs between the critical mass threshold and the critical mass plateau (β = 0.18, p < 0.01 for females and β = 0.11, p < 0.05 for males). These differences are not significant for females (t = 0.4, n.s.) and males (t = 0.2, n.s.).
We calculate the critical mass plateau for organizations led by female versus male CEOs. For female CEOs who are already more likely to support equity-related practices according to social role theory (e.g., Cook & Glass, 2016), fewer women in TMTs will be needed to form a critical mass to shift the dial in terms of IPV-related HR practices. For organizations led by male CEOs, in contrast, the adoption of IPV-related HR practices will be less prevalent according to social role theory and more women in TMTs will be required to create a critical mass to change the male CEOs' minds to move the needle. Thus, we expect that the former plateau to be lower than that of the latter. As expected, this analysis shows that organizations led by a female CEO have a lower critical mass plateau (i.e., 0.32 or 32% women in the TMT) than that of organizations led by a male CEO (i.e., 0.38 or 37.5% of women in the TMT).5
In step 5, we present the results without control variables. The results are largely consistent with those with controls. They also support our hypotheses in the same way.
2.5 Study 1 discussion
In Study 1, we found general support for the three hypotheses. These results suggest that organizations led by a female CEO are more likely to have IPV-related HR practices in place than those led by a male CEO, supporting social role theory. There is also partial support for the critical mass perspective in that we found a nonlinear relationship between women's representation in the TMTs and organizations’ adoption of IPV-related HR practices. That is, the more female members a TMT has, the more IPV-related HR practices are adopted. This positive relationship plateaus when female members reach the critical mass plateau of 34% in TMT. Furthermore, the CEO gender effect strengthens below and levels off beyond the critical mass plateau. We found that organizations led by a female CEO have a lower plateau (32%) than those led by a male CEO (38%).
Although Study 1 provided empirical support for our hypotheses, the major limitations of this study are that we utilized a cross-sectional research design and self-report data. Wright et al. (2005) argued that although numerous HR studies have shown a positive effect of HR practices on firm performance (e.g., Guthrie, 2001), this research often lacks sufficient methodological rigor to claim causality for this relationship. Drawing on Cook and Campbell (1976), Wright and his colleagues argued that causality can only be inferred when (a) the cause and effect are correlated, (b) the cause occurs before the effect, and (c) the relationship between the cause and effect is not spurious, in that there are no other explanations for the existence of the cause-and-effect association. Cross-sectional survey research, therefore, can never claim that HR practices causally influenced performance, or in our case that CEO gender and the percentage of women in TMT cause the adoption of IPV-related HR practices. Moreover, we cannot claim that the Study 1 results showed the adoption of IPV-related HR practices caused CEO gender or the percentage of women in the TMT. We, therefore, conducted Study 2 to address these limitations.
Study 2 aims to replicate Study 1 results and provide evidence for stronger causal inferences by analyzing two waves of data (2015–2016 and 2016–2017) from the Workplace Gender Equality Agency (WGEA). We also aim to constructively replicate the Study 1 findings by using a different operationalization of IPV-related HR practices. Lindsay and Ehrenberg (1993) highlighted that the cumulative effect of replication is vital to generalization such that findings not subjected to replication are “virtually meaningless and useless” (p. 219). Along similar lines, the converging results between two operationalizations can further bolster our claim that our study findings are robust and not due to methodological artifacts (Bouchard, 1976).
3 STUDY 2
3.1 Method
3.1.1 Data
Under the Workplace Gender Equality Act 2012, employers are required to complete and submit an annual report covering standardized reporting matters to the WGEA. The WGEA dataset contains information related to gender equality issues from non-public sector organizations with 100 or more employees in Australia. The 2015–2016 WGEA dataset contains 4697 organizations, while the 2016–2017 dataset contains 4621 organizations.6 The combined dataset contains 4189 organizations, with an average of 876 employees (SD = 3579). Thirteen percent of the organizations are in healthcare and social assistance, 12% in manufacturing, 10% in education and training, and 10% in professional, scientific, and technical services industries. The remaining are in accommodation and food services (9%), financial and insurance services (9%), retail (8%), electricity, gas, water, and waste services (8%), administrative and support services (5%), mining (5%), agriculture, forestry, and fishing (4%), arts and recreation services (4%), and transport, postal, and warehousing (4%) industries.
3.2 Measures
3.2.1 CEO gender and percentage of women in the TMT
This information was collected from the 2015–2016 WGEA dataset. Organizations were asked to indicate the gender of their CEO (1 = female; 0 = male) and the percentage of women in the TMT within their organization. In this dataset, while 18% of the organizations have no women in the TMT, eight percent of the organizations have more than 50% women in the TMT. Similar to Study 1, the percentage of women in TMT is not normally distributed (Kolmogorov–Smirnov [4148] = 0.36, p < 0.01) and is skewed to the left (i.e., there are more organizations with a percentage of women in the TMT between zero and 50, and very few organizations with more than 50% women in the TMT). Consistent with Study 1, we centered this variable so that it is more robust for further analyses.7
3.2.2 IPV-related HR practices
The dataset for the two waves contains 14 “support mechanisms that organizations have in place to support employees who are experiencing family or domestic violence” (see Appendix A). Respondents were asked to indicate whether they have each of the 14 HR practices in place to support employees who are experiencing family or domestic violence (“yes” or “no”). The items were targeted to support employees affected by IPV. As with Study 1, we counted the number of practices an organization has in place (Boon et al., 2019).
3.2.3 Control variables
As with Study 1, we controlled for the size of the organization (log) and the size of TMT in Study 2. However, we did not control for the type of organization because the WGEA data consist only of non-public sector organizations.
To strengthen our causal claims, one of the methods Wright et al. (2005) suggested is to collect data for the dependent variable over time. In this way, one can “explore the relative relationships between our focal variables [i.e., CEO gender and percentage of women in the TMT] and the performance variables [i.e., IPV-related HR practices] assessed prior to, concurrent with, and following the assessment of these focal variables” (p. 421). Our dataset allowed us to control for IPV-related HR practices (2015–2016), which is prior to the period of our study. In using this method, our confidence in the hypothesized effect of CEO gender and percentage of women in TMT (assessed in 2015–2016) on the adoption of IPV-related HR practices (2016–2017) increases (Wright et al., 2005). This evidence, however, does not necessarily prove that there is no reverse causality.
As with Study 1, we report the results of our analyses with and without controls (Carlson & Wu, 2012; Sturman et al., 2022).
3.3 Data analytic approach
We followed the same approach as Study 1 to test H1–H3. Since organizations are nested in industries (ICC1 = 0.09; Bliese, 2000) and the percentage of women in the industry is a higher order construct (Snijders & Bosker, 1994), we analyzed the data using hierarchical linear modeling (HLM) with industry at the higher level and organizations at the lower level.
The model fit for multi-level models is calculated in SPSS, mixed models using the −2 log likelihood, and four information criteria, including the Akaike's Information Criteria. These statistics follow the “small-is-better” interpretation, meaning that the model with the smaller value is the one with the better overall fit (see Bickel, 2007, p. 93). This means that by adding more variables (and interaction terms), the model fit becomes better in every model (see also Cavanaugh, 2005). To determine whether the model fit is significantly better than the previous model, the deviance difference between the two models can be calculated with the degrees of freedom as the additional variables in the model. The deviance difference is distributed (approximately) in a similar manner as a Chi-Square distribution (Bickel, p. 94). The deviance in the model fit for model 1 is χ2(2) = 934 and is calculated by subtracting the model fit for model 1 (i.e., 11,941) from that of the empty model (i.e., model without any variable, 12,874). The deviance in the model fit of model 5 (i.e., the model without the controls) is calculated in relation to the empty model (i.e., χ2[5] = 6372; 12,874–6502).
3.4 Results
3.4.1 Descriptive statistics and zero-order correlations
In Table 3, we reported means, standard deviations, and zero-order correlations among the study variables. There is a positive relationship between HR practices in wave 1 (2015–2016) and in wave 2 (2016–2017): r = 0.58, p < 0.01. Mirroring the Study 1 findings, organizational size and size of the TMT were positively related to the adoption of IPV-related HR practices (r = 0.14, p < 0.01; r = 0.13, p < 0.01). Female CEOs (2015–2016) are more likely to adopt IPV-related HR practices (2016–2017; r = 0.14, p < 0.01). Percentage of women in TMTs (r = 0.06, p < 0.05) in 2015–2016 were related to the adoption of IPV-related HR practices in 2016–2017.
| Variables | Mean | SD | 1. | 2. | 3. | 4. | 5. | 6. | 7. |
|---|---|---|---|---|---|---|---|---|---|
| 1. IPV-related HR practices | 0.27 | 0.22 | |||||||
| 2. CEO gender (1 = female; 0 = male) | 0.14 | 0.20 | 0.14** | ||||||
| 3. % women in TMT | 31 | 47 | 0.06* | 0.19** | |||||
| 4. Organization size | 876 | 3579 | 0.14** | 0.03 | −0.02 | ||||
| 5. Size TMT | 8.44 | 2.65 | 0.13** | 0.10* | 0.09* | 0.55** | |||
| 6. IPV related HR practices 2015–2016 | 0.22 | 0.20 | 0.58** | 0.13** | 0.05** | 0.14** | 0.14** | ||
| 7. % women in the organization | 47 | 24 | 0.05* | 0.51** | 0.18* | −0.02 | 0.04 | 0.04* | |
| 8. % women in the industry | 30 | 30 | 0.10** | 0.29** | 0.23** | −0.01 | 0.08 | 0.09* | 0.25** |
- ** p < 0.01;
- * p < 0.05.
3.4.2 Hypotheses testing
Table 4 reports the HLM results. To test H1, the effect of CEO gender was added in step 3. In support of H1, this coefficient was significant (β = 0.14, p < 0.01). To test H2, the non-linear effect of the percentage of women in TMT, we added the main effect and quadratic effect of the percentage of women in TMT in step 3. Both the linear (β = 0.07, p < 0.01) and the non-linear effect (quadratic: β = −0.09, p < 0.01) were significant thereby supporting H2. We calculated the critical mass plateau by computing the first derivative of the main and quadratic effect and the plateau is 33%. The simple slopes analyses show that before this critical mass plateau, the relationship between the percentage of women in the TMT is 0.24 (p < 0.01), and after the plateau, it is 0.02 (n.s.) supporting the hypothesized plateau effect. There is no significant difference in the steepness between the organizations with percentages of women below the critical mass threshold (token position: 0.24, p < 0.01; t = 0.2, n.s.) and the ones with percentages of women between the critical mass threshold and plateau (0.25, p < 0.01). Thus, H2 is partially supported in that the non-linear effect is significant, but rather than having no effect, organizations with percentages of women below the critical mass threshold are positively associated with IPV-related HR practices adoption.
| Variables | Step 1 | Step 2 | Step 3 | Step 4 | Step 5 (without controls) |
|---|---|---|---|---|---|
| Organization size | 0.04** | 0.04** | 0.04** | 0.04** | |
| Size of TMT | 0.03* | 0.03* | 0.03* | 0.03* | |
| IPV-related HR practices (2015–2016) | 0.58** | 0.58** | 0.58** | 0.58** | |
| Gender CEO (1 = female; 0 = male; H1) | 0.14** | 0.14** | 0.14** | 0.25** | |
| % women TMT | 0.07** | 0.07** | 0.15** | ||
| % women TMT2 (H2) | −0.09** | −0.09** | −0.11* | ||
| Gender CEO x % women TMT | 0.01 | 0.04 | |||
| Gender CEO x % women in TMT2 (H3) | −0.05* | −0.07* | |||
| Model fit | 11,941 | 9942 | 7524 | 5917 | 6502 |
| Deviance in model fit | χ2(2) = 934** | χ2(1) = 1999** | χ2(2) = 2418** | χ2(2) = 1607** | χ2(5) = 6372** |
| Variance | |||||
| Variance at the organizational level | 0.05* | 0.01* | 0.01* | 0.01* | 0.02* |
| Variance at the industry level | 0.95** | 0.60** | 0.59** | 0.57** | 0.61** |
| Explained variance | 0.28 | 0.39 | 0.40 | 0.42 | 0.37 |
- ** p < 0.01;
- * p < 0.05.
To test H3 (i.e., the moderation of the non-linear effect of percentage of women in TMT on the CEO gender effect), we entered the relevant two way- and three way-interaction (CEO gender x quadratic effect of percentage of women in TMT) terms in step 4 (see Study 1). In support of H3, this interaction is significant (β = −0.05, p < 0.01). Figure 2 displays the nature of the interactive effect before there are 50% women and men in the TMT. The CEO gender effect (or the differential impact of CEO gender) on IPV-related HR practices becomes stronger before the critical mass plateau and levels off after the plateau. Indeed, the relationship between female CEO and IPV-related HR practices before the critical mass plateau is stronger (simple slope = 0.33, p < 0.01; t = 3.47, p < 0.01) in comparison to the relationship between male CEO and IPV-related HR practices (simple slope = 0.11, p < 0.05). We also analyzed the steepness between the organizations with a percentage of women in the TMTs below the critical mass threshold, for female and male CEOs separately (β = 0.32, p < 0.01 for females and β = 0.12, p < 0.05 for males) and the ones with a percentage of women in their TMTs between the critical mass threshold and the critical mass plateau (β = 0.33, p < 0.01 for females and β = 0.10, p < 0.05 for males). These differences are not significant for both females (t = 0.07, n.s.) and males (t = 0.02, n.s.).
As expected and consistent with Study 1, organizations led by a female CEO have a lower critical mass plateau of 0.32 (i.e., 32% women in the TMT) than that of organizations led by a male CEO (i.e., 0.35 or 35% of women in the TMT).
In step 5, we conducted the same analysis without controls. The results are largely consistent with the results with controls and support our hypotheses in the same way.8
3.4.3 Robustness checks
Percentage of women in the organization and percentage of women in the industry
In this robustness check, we controlled for the percentage of women in the organization and the percentage of women in the industry because these percentages can explain the gender of the CEO, the percentage of women in the TMTs, and the adoption of IPV-related HR practices. The presence of larger percentages of women in both the organization and the industry will likely be related to more female CEOs and a higher percentage of women in the TMTs (e.g., Cook & Glass, 2014). The presence of more women in the organization and the industry will also likely translate into a stronger need for IPV-related HR practices (cf. Shi et al., 2017).
The percentage of women in the organization was already available in the WGEA data set. The percentage of women in the industry was collected from the Australian Bureau for Statistics (ABS, 2015; 2016) and matched with WGEA industry data for each organization, and subsequently added to the Study 2 WGEA dataset. The ABS provides the percentage of workers of 19 industries by gender. Women made up 45.6% of the total workforce and men comprised 54.4%. The percentages of female workers varied across the industries. For instance, healthcare and social assistance have 79% female employees, while financial and insurance services and construction have 48.1% and 12% female employees, respectively.
The descriptive statistics and correlations for the percentage of women in the organization and the percentage of women in the industry are reported in Table 3. While both variables are significantly related to the gender of CEO (r = 0.51, p < 0.01; r = 0.29, p < 0.01, respectively), the percentage of women in the TMT (r = 0.18, p < 0.05; r = 0.23, p < 0.01, respectively), and the adoption of IPV-related HR practices (r = 0.05, p < 0.05; r = 0.10, p < 0.01, respectively), controlling for the percentages of women in the organization and in industry had no effect on the effect sizes of the focal variables. As such, this did not change the support of our hypotheses.
Comparison of IPV-related HR practices between study 1 and study 2
The content (and the number) of IPV-related HR practices items in the WGEA dataset are somewhat different from the content (and the number) of IPV-related HR practices items in Study 1. A direct and stringent comparison of the results from the two studies is thus difficult to make. As another robustness check, we, therefore, compared the two lists of items and reanalyzed the Study 2 data with only the items that overlap with the HR function themes that were covered by Study 1 items: (1) paid and unpaid domestic violence leave (equivalent to IPV leave in Study 1), (2) training for key personnel (i.e., training for supervisors to recognize IPV victims and to help IPV victims to disclose IPV in Study 1), (3) Employee Assistance Program (i.e., counseling for IPV victims in Study 1), and (4) flexible working arrangements. While the effect sizes differ between the main and this robustness check analyses, the three hypotheses were supported similarly.9
4 GENERAL DISCUSSION
Intimate partner violence is not only a societal problem, it also negatively impacts organizations and their employees (e.g., Duffy et al., 2004). Organizations can and should ameliorate the situation by adopting, for instance, HR practices to support employees affected by IPV (Tolentino et al., 2016; cf. de Jonge, 2018; Wilcox et al., 2021). Indeed, a recent meta-analysis showed that there are benefits to such workplace interventions, including supervisors' and allies' increased knowledge in recognizing IPV, willingness to intervene, and likelihood to provide resources to those affected by IPV and to take a leadership role in stopping IPV (Adhia et al., 2019). Despite the potency of HR practices in combating the negative effects of IPV, there is limited research on why organizations vary in the extent to which they adopt IPV-related HR practices (e.g., SHRM, 2013). To better address this question, we conducted two studies using a survey and two waves of archival data from Australian organizations. Drawing on research showing the important role women leaders play in supporting socially responsible practices (e.g., Glass & Cook, 2018), we examined the influence of women's representation in leadership positions on the likelihood of organizations adopting IPV-related HR practices.
Three main findings were consistent across the two studies. First, organizations led by female CEOs are more likely to adopt IPV-related HR practices than organizations led by male CEOs. Second, the percentage of women in the TMTs is positively associated with the adoption of IPV-related HR practices. However, this positive relationship levels off when women's representation in the TMT reaches a critical mass plateau (i.e., the average of Studies 1 and 2: 33.5%), beyond which more women in TMTs have no additional statistical impact on our dependent measures. Third, this non-linear effect moderates the CEO gender effect such that the latter strengthens below and levels off after the critical mass plateau has been reached. Organizations led by female CEOs have a lower plateau (average of Studies 1 and 2: 32%) than those led by male CEOs (36.3%), which suggests that fewer women in the TMTs are needed to shift the dial when an organization is led by a female CEO.
4.1 Theoretical and practical implications
Our findings have implications for theory and practice. First, by focusing on organizational responses to IPV, our research reveals that the influence of CEO gender and gender representation in TMTs on DEI practices is more complex than originally theorized and demonstrated in previous research. Prior research generally shows that female CEOs and the percentage of women on boards are positively related to the adoption of DEI policies and practices, such as LGBT-friendly policies and employee benefits addressing work-family concerns (e.g., Cook & Glass, 2016; Glass & Cook, 2018). Integrating social role theory and the critical mass perspective, our theorizing extends this research by empirically demonstrating that having both female CEOs and women on the TMTs matter until the critical mass plateau of 33.5% (average from Studies 1 and 2) is reached, beyond that, a greater percentage of women in TMTs does not add statistically significant influence on the adoption of IPV-related HR practices. Our research provides novel insights to the literature on the effect of gender diversity on organizations' strategy in that the benefits of having more women in leadership positions may not be perfectly linear as previous research implies (e.g., Abebe & Dadanlar, 2019; Cook & Glass, 2017; Glass & Cook, 2018; Konrad et al., 2008).
Of note, our research findings do not in any way imply that organizations should refrain from recruiting more than 33.5% women on their TMTs. Rather, our findings suggest that strategic decisions may no longer be driven as much by homophily processes (within social demographic groups in which leaders belong) after women have reached the critical mass plateau. After the plateau, strategic decisions may be based more on other factors and processes, such as leaders' individual characteristics and vision for the organization, as well as the stronger interdependence and mutual support between women and men as their relationship stabilizes (cf. Allmendinger & Hackman, 1995). Striving to have an equal number of women and men (and other social groups) in the TMTs to bring in diverse perspectives to enhance organizational outcomes is still of significance (e.g., van Knippenberg et al., 2004). The two studies presented build on and complement each other. They allow us not only to establish the validity of our theorizing but also to illustrate the criticality of having women in positions of decision-making power.
Second, our findings contribute to research that reveals the boundary conditions of token theory (e.g., Yoder, 1991). The results of our two studies consistently showed that women are still able to push forward their agenda and influence their organizations to adopt IPV-related HR practices even when their numerical representation is below the critical mass threshold. This suggests the negative dynamics usually experienced by token women may not be experienced by our groups of women. The arguments forwarded in token theory focus on group proportions (e.g., dominant vs. token; minority vs. majority; Kanter, 1977), without consideration of contextual and social factors, such as the relative status of the social groups in which tokens and dominants belong (e.g., male vs. female) and the extent to which organizations value egalitarianism (e.g., Yoder, 1991; Zimmer, 1988). Our samples of high-status women are influential individuals (without evoking gender identities). Recent empirical studies have shown that token women who are of high status are still able to exert their influence in both work and political domains (e.g., Bratton, 2005; Cook & Glass, 2017; Elstad & Ladegard, 2012). This may be so because high-status female board directors have been found to be less conforming and traditional and more risk-loving than both high-status male board directors and the general population of women (Adams & Funk, 2012). Moreover, McDonald et al. (2004) found that when increasing the status (via education) of female tokens, the negative expectations associated with tokenism decreased. Our research suggests that when considering token dynamics, scholars should consider the extent to which such dynamics apply to and how they may manifest in their research contexts.
Third, there is some evidence in our data that suggest that the findings for IPV-related HR practices may generalize to the adoption of other DEI practices that are more consistent with the female, rather than male, gender role (e.g., care for all employees and stakeholders versus maximizing self- and shareholders' interests). For example, Study 1 findings show that organizations' adoption of IPV-related HR practices is associated with that of anti-bullying policy, inclusion and diversity practices, mentoring programs, and mental health issue practices. However, in considering how our research expands existing research on female leaders and adoption of DEI practices, it is useful to recognize there are two distinct types of DEI programs in organizations. The identity-conscious programs target specific identity groups (e.g., preferential hiring or “special measures” in the Australian context—positions identified for women or for Aboriginal and Torres Strait Islander people), while the identity-blind type does not target any specific groups (e.g., Equal Employment Opportunity practices, flexible work practices available to all; Konrad & Linnehan, 1995). We view IPV-related HR practices as situated very closely to identity-blind DEI programs because employees belonging to any identity groups can be victimized by IPV (e.g., Deen et al., 2022). Given that past research on the role of female leaders on DEI-related organizational outcomes has mainly focused on identity-conscious issues and practices, such as those supporting LGBTQI, women, and ethnic minorities (e.g., Abebe & Dadanlar, 2019; Ali & Konrad, 2017; Cook & Glass, 2016), our research adds nuance to this literature by demonstrating female leaders also promote identity-blind practices that go beyond benefiting traditionally marginalized identity groups. Furthermore, given the well-documented general lack of support for identity-conscious DEI practices by both beneficiaries and non-beneficiaries (e.g., Leslie, 2019), we suspect that organizations may need an extra push by requiring a higher female representation in their TMTs when it comes to the adoption of such practices. Thus, future research can compare the influence of female representation in leadership on the adoption of different types of DEI programs.
Finally, our research has important implications for upper echelons theory. This theory posits that the decisions made, and practices adopted by organizations are driven by top executives' personalized interpretation of situations, which is in turn is a function of their characteristics (e.g., tenure, age, sex), values, and cognitive frames (Hambrick & Mason, 1984). We contribute to this line of research by introducing a novel dependent variable, the adoption of IPV-related HR practices, as well as by demonstrating that this important HR decision can be shaped by gender differences in the social role expectations of top executives and gender composition of decision-making bodies in organizations. We highlight the importance of the TMT-CEO interface in upper echelons theory (Simsek et al., 2017). The interaction between the two leadership bodies suggests that the benefits of appointing female CEOs on the adoption of IPV-related HR practices are enhanced by having more women in the TMTs.
4.2 Limitations and future research
While the current research advances the literature on gender diversity in leadership and equity and inclusion HR practices, our studies have some limitations that can be addressed in future research. First, the data for the current studies come from public and for-profit firms of various sizes in Australia. Given that macro-level factors, such as federal and state laws, public sentiments and political attitudes, media coverage, and gender representation in the industry (Chuang et al., 2011; Dobbin et al., 2011; Wald et al., 1996) influence the adoption of DEI policies and practices, the focus on Australian data may limit our ability to fully generalize our findings to organizations in other countries.
Consider, for instance, the White Ribbon organization in Australia (https://www.whiteribbon.org.au/), which is a part of a global social movement working to stop gendered violence, has a Workplace Accreditation Program that supports and recognizes workplaces that take steps to respond to and prevent IPV that occurs inside and outside of work. Furthermore, data for both studies were collected when two legislative reforms providing stronger employment protections for IPV victims were under national consideration. The two reforms include the Australian Capital Territory becoming the first state to consider those subjected to domestic violence a protected class in April 2017 and IPV victims' entitlement to five days of unpaid domestic violence leave per year for all employees was passed nationally in August 2018 (see Roff, 2020 for details of these legislations). Moreover, in 2018 Australian courts seemed to begin moving closer to holding employers liable for acts of domestic violence at work (Guthrie & Babic, 2021). These legislative forces hold organizations more accountable and therefore can have significant implications on how IPV victims are treated and supported in the workplace. Indeed, Johns (2017) highlighted that variables and relationships vary in their contextual sensitivity and that this issue should be considered when theorizing and designing empirical research. Therefore, it is possible that the adoption of IPV-related HR practices is more sensitive to these legislative forces than women's representation in leadership, such that the former effects take over or serve as alternatives to the latter effects (cf. Dobbin et al., 2011). It is also possible that countries with a higher level of economic, social, and legislative support for IPV issues may encourage organizations to adopt more innovative and effective types of IPV practices, such as advance pay or loan options to IPV victims and a focus on preventative measures by contributing to educational projects on “Respectful Relationships”. Future research can therefore attempt to tease apart the relative influence of various high level contextual forces and women representation in leadership on the adoption of IPV-related HR practices.10 Future research can also investigate whether and how gender representation in leadership positions, including CEOs and TMTs in organizations outside of Australia influence the adoption of quantity and “quality” of IPV-related HR practices.
Second, Study 1 is limited by the self-report and cross-sectional nature of the data. Because our survey is not a formal reporting requirement, the HR professionals may have reported incomplete or out-of-date data about the availability of IPV-related HR practices, as well as the gender composition of the TMTs in their respective organizations. However, these limitations are mitigated by Study 2. Data from Study 2 originate from formal organizational reporting that must pass data quality standards to comply with the Workplace Gender Equality Act 2012. We also drew on two years of WGEA data to more confidently infer that the gender composition of the two bodies of leadership influenced the adoption of IPV-related HR practices (Wright et al., 2005). Future research can utilize a longitudinal design with at least three time points to infer causality more confidently, as well as to address complex research questions related to possible change dynamics between gender representation in leadership and the adoption of IPV-related practices (cf. Ployhart & Vandenberg, 2010). For instance, one could test a potential “Matthew effect” in which organizations starting with a high representation of women in the TMTs will adopt more DEI HR practices, including IPV-related ones, which in turn will increase women representation in the TMTs (cf. Gallardo-Gallardo et al., 2013). The same dynamics may apply to organizations that begin with low women representation in the TMTs, but in a downward spiral.
Third, we did not collect or have data for general HR practices (e.g., general training, performance management) or bundles of HR practices, such as high-performance work systems (HPWS; Appelbaum et al., 2000) in the two studies. Therefore, we cannot completely rule out the possibility that our findings could be confounded by the extent to which organizations invest in general HR practices. However, we controlled for organization size and type in Study 1 and organization size in Study 2. These two variables may influence how much organizations invest in their general HR practices. The fact that controlling for these variables did not change the pattern of our results gives us additional confidence that the adoption of IPV-related HR practices is not simply a function of how much resources organizations devote to HR practices but reflects gender dynamics at the decision-making levels of organizations.
Fourth, although we drew on social role theory to understand gender differences and the adoption of IPV-related HR practices, we acknowledge that gender differences can also be explained through other theoretical lenses, including gender belief system model (Deaux & Kite, 1987), gender schema theory (Bem, 1981), cognitive social learning theory (Bussey & Bandura, 1999), evolutionary psychology (Buss, 1995), and the interactionist perspective (i.e., an interplay between biological mechanisms and social-cultural influences; e.g., Eagly & Wood, 2013). Future research may draw upon these theoretical lenses to further explain how gender differences shape leaders' strategic decisions to adopt certain HR practices.
Fifth, the current research did not measure the processes through which gender representation in leadership positions is associated with the adoption of IPV-related HR practices. That is, we were unable to measure the proposed team dynamics, including homophily that shapes the non-linear relationship between the percentage of women in the TMTs and IPV-related HR practices. As recommended by Kulik and Metz (2017), scholars can contribute to the gender diversity literature by disentangling the mediating mechanisms that explain the relationship between gender diversity in leadership positions and organizational outcomes. Future research can draw on qualitative methodology, including interviews with CEOs and TMT members and participant observation of TMT meetings, to further provide in-depth insights into the proposed processes.
Finally, our research focuses on the effect of gender representation in leadership on IPV-related HR practices and social roles. Yet, extant research suggests that other social attributes, including ethnicity, religiosity, and social economic status may also impact individuals' attitudes, decisions, and behaviors within organizations (e.g., Chan-Serafin et al., 2013). In this respect, intersectionality may also matter as the experiences of white heterosexual women and men are likely to be very different from those of women and men of color from the LGBTQI community (e.g., García Johnson & Otto, 2019). Moreover, gender is socially constructed and can be experienced and expressed in nuanced and complex manners (West & Zimmerman, 1987). Yet, our measure of gender, which comes from third-party reports (i.e., survey and WGEA data), may overlook these complex issues, such as an individual's psychological androgyny (Bem, 1974). Future researchers can theorize and analyze the intersections of gender and other social categories, such as ethnicity and sexual orientation, as well as consider the fluidity of leaders’ gender on the adoption of HR practices to support employees affected by IPV.
5 CONCLUSION
Some practitioners and academics argue that IPV should be treated exclusively as a domestic concern because they see it as a private family matter, outside the purview of organizations (O'Leary-Kelly et al., 2008). However, IPV can occur within the confines of organizations in the form of perpetrators physically entering the workplace or using communication channels to engage in acts of aggression while their partners are at work. Moreover, most IPV victims and perpetrators are in paid employment (McFerran, 2011). We thus contend that IPV is a critical issue that organizational scholars must attend to. In light of the increased reports of IPV associated with COVID-19 lockdowns and in response to the #MeToo movement, whereby bystanders and hopefully IPV victims are more likely to speak up, our research calls for management scholars and HR practitioners to seize this moment and investigate how organizations can better support their employees impacted by IPV.
ACKNOWLEDGMENTS
We acknowledge Anne O'Leary-Kelly for her helpful comments in an earlier version of the manuscript. Open access publishing facilitated by University of New South Wales, as part of the Wiley - University of New South Wales agreement via the Council of Australian University Librarians.
Endnotes
APPENDIX A: MEAN AND SD OF IPV-RELATED HR PRACTICES (STUDY 2)
| 2015–2016 (n = 4621; α = 0.85) | 2016–2017 (n = 4697; α = 0.84) | ||||
|---|---|---|---|---|---|
| HR practices to support employees affected by IPVa | Mean | SD | Mean | SD | |
| 1. | Access to paid domestic violence leave | 0.07 | 0.19 | 0.10 | 0.23 |
| 2. | Access to unpaid leave | 0.26 | 0.28 | 0.34 | 0.29 |
| 3. | Employee Assistance Program (EAP)/access to psychologist, chaplain, or counselor | 0.66 | 0.47 | 0.73 | 0.45 |
| 4. | Training of key personnel | 0.11 | 0.31 | 0.14 | 0.35 |
| 5. | Domestic violence clause in enterprise/workplace agreement | 0.07 | 0.25 | 0.10 | 0.31 |
| 6. | Workplace safety planninga | 0.08 | 0.28 | 0.14 | 0.34 |
| 7. | Confidentiality of matters disclosed | 0.41 | 0.49 | 0.54 | 0.50 |
| 8. | Flexible working arrangements | 0.49 | 0.50 | 0.61 | 0.49 |
| 9. | Provision of financial support (e.g., advanced pay/bonus payment) | 0.11 | 0.32 | 0.16 | 0.40 |
| 10. | Emergency accommodation assistance | 0.04 | 0.20 | 0.06 | 0.24 |
| 11. | Access to medical services (e.g., doctor/nurse) | 0.10 | 0.30 | 0.15 | 0.35 |
| 12. | Protection from any adverse action or discrimination based on the disclosure of domestic violence | 0.21 | 0.41 | 0.29 | 0.45 |
| 13. | Offer change of office location | 0.14 | 0.34 | 0.20 | 0.40 |
| 14. | Referral of employee to appropriate domestic violence support services for expert advice | 0.27 | 0.44 | 0.35 | 0.48 |
- a The HR practices items were all targeted to support employees affected by IPV. The question stem for each item is: “Do you have any support mechanisms in place to support employees who are experiencing family or domestic violence?”
- b See example here: https://humanrights.gov.au/sites/default/files/Annex%20B%20safety_planning.pdf.
Biographies
Suzanne Chan-Serafin is an Associate Professor of Organizational Behavior at the UNSW Business School. She uses experimental and survey methodologies to study social justice issues in organizations. By examining these issues from a people and systems perspective, she aims to create more moral, diverse, and inclusive workplaces where people, especially those who belong to disadvantaged groups, can thrive.
Karin Sanders is the Professor of Human Resource Management and Organizational Behavior at the UNSW Business School. Her research focuses on the Human Resource (HR) process approach, in particular the impact of employees' perceptions, understanding and attributions of HR on their attitudes and behaviors, such as their informal learning activities.
Lu (Nick) Wang is an Associate Professor of Management. His primary research focuses on emotions in an organizational context, where he seeks to understand how emotions enhance individuals' ability to work together and, relatedly, how organizational context shapes the type of emotions individuals experience and express at work. Lu's research uses both experimental and survey methodology to examine these questions.
Simon Lloyd D. Restubog is a Full Professor in the School of Labor and Employment Relations and Department of Psychology (courtesy appointment) at the University of Illinois, Urbana-Champaign, USA. His research interests include employment relationships, the dark side of human behavior in organizations, and career development.
Open Research
DATA AVAILABILITY STATEMENT
The data that support the findings of this study are available from the corresponding author upon reasonable request.
