Abstract
Using data from a survey of large nonprofits across Canada, this study focuses on the determinants of the range of diversity (defined as the number of different ethnocultural and visible minority groups represented) on boards across the country. The determinants of diversity that the article examines include community, organizational, and general board characteristics as well as board diversity practices. We examine the extent to which these factors are related to an increased range of diversity on the boards. It appears that the diversity of the community that nonprofits operate in and efforts to institutionalize formal diversity-related policies are particularly significant determinants of diversity, although board size and reliance on interorganizational alliances in recruitment of board members also have a small relationship. The implications for theory and practice are examined.
Introduction
Research examining the determinants, impacts, and outcomes of demographic differences among members of groups and communities (or what we are calling diversity) has recently been increasing in a number of fields. In this article, we are focusing particularly on the determinants of diversity on nonprofit boards, although the impacts and outcomes associated with diversity also need further study. We believe the growing interest in demographic diversity reflects the changing North American population and workforce composition, as well as legislative and social changes. According to the 2001 Canadian census, for example, 28% of the total population was born outside Canada (Badets, 2003). The Conference Board of Canada predicts that by 2016, members of visible/racialized minorities will comprise approximately 20% of the population and approximately 18% of the workforce (Antunes, MacBride-King, & Swettenham, 2006). We would expect these demographic changes to be reflected in the makeup of boards of directors (Erhardt, Werbel, & Shrader, 2003; Shrader, Blackburn, & Iles, 1997), but there has been limited tracking of the composition of Canadian nonprofit boards. It is therefore difficult to determine the nature and speed of these shifts, or to learn what factors are associated with boards that are more representative of diverse communities. In particular, although there has been research on the gender of Canadian nonprofit board members (e.g., Bradshaw, Murray, & Wolpin, 1996), there has been less investigation of the extent to which the boards are reflecting the changing makeup of the population in terms of race, ethnicity, and color, or what Statistics Canada refers to as ethnic origin and visible minority status. 1
Researchers in the United States have found that nonprofit boards of directors are still overwhelmingly homogeneous, comprised primarily of white trustees (Abzug, 1996; Ostrower, 2002; Ostrower & Stone, 2006; Siciliano, 1996; Widmer, 1987). Ostrower (2007) recently did the first national survey of nonprofit governance in the United States, and she found that 86% of board members in the country are White, non-Hispanic; 7% are African American or Black, and 3.5% are Hispanic/Latino. In Canada, research tracking diversity in the voluntary sector within some of the largest charities and foundations based on revenues registered in the Greater Toronto Area shows that in 2011 visible minorities made up 9.9% of senior executives. This represents a decrease from the 2010 rate of 15.5% but a slight increase from the 2009 rate of 8.5%. At the board level, visible minorities made up 13.3% of members, a slight increase from 11.9% in 2010 but a decline from 13.7% in 2009 (Cukier, Yap, Aspevig, & Lejasisaks, 2011).
The advantages of diversity, within the context of boards of directors, are increasingly well articulated (Berthoud & Greene, 2001; Daley & Angulo, 1994; Daley & Marsiglia, 2001; Fink, Pastore, & Reimer, 2001; Fletcher, 1999; Fredette, Bradshaw, & Inglis, 2006; Miller, 1999). Some of the advantages that researchers have proposed include superior financial performance (Dobbin & Jung, 2011; Joy, Carter, Wagner, & Narayanan, 2011; Parris, Cowan, & Huggett, 2006), improved creativity and decision making at the group level (Ely & Thomas, 2001; Gonzalez, 2010; Yang & Konrad, 2011), increased responsiveness to community and clients (Siciliano, 1996), more effective executive leadership (Barta, Kleiner, & Neumann, 2012; Perkins & Fields, 2010), and increased cultural sensitivity driving more effective global relationships (Carter, Simkins, & Simpson, 2003; Cox & Blake, 1991; Robinson and Dechant, 1997). What has not been looked at as extensively are the determinants of diversity on boards.
In this article, we present the findings of a survey of nonprofit organizations from across Canada, which allows us to examine the current composition of boards and to explore factors that may be related to higher levels of diversity, or what we are calling determinants of diversity. We elected to measure the range of ethnocultural diversity on the boards in our sample, where range of diversity is defined as the number of different ethnocultural and visible minority groups represented on boards. There are different ways of looking at diversity, and often researchers examine the proportion of a board that is made up of individuals from nondominant and visibly different groups. As a consequence of this measure, however, a board that has half its membership composed of members from one ethnocultural group would be considered more diverse than a board in which a quarter of the membership came from three of four different groups. We look at range of diversity because it is both theoretically interesting and relevant to practice, and because such a measure represents a move toward what Harrison and Sin (2006) call an inclusive constitutive definition of diversity, or the collective amount of difference among individuals in a social unit. Although our research only looks at ethnocultural differences, we feel it takes an important step in capturing this aspect of diversity with the board as the unit of analysis. In our sample, analyses using other aspects of diversity (such as sexual orientation and ability) tended to yield less robust results (likely because such aspects of diversity are seldom found on these boards, and because they are often not visible, and are thus harder to track).
This article begins by exploring prior theorizing and empirical findings on determinants of diversity, to illuminate our choice of determinants to study. We then outline the research methods, present our findings, and conclude with a discussion of the implications of our research for both research and practice.
Literature Review
Many scholars argue that increasing the diversity on boards is important because it allows organizations to position themselves strategically, and hence better understand their increasingly diverse clients and members. Others point out that enhancing diversity is a response to imperatives created by stakeholders such as funders and legislators. Still others argue from a values stance that becoming more diverse is the correct or right thing to do. We are beginning to find some empirical support for the argument that greater diversity improves board performance. For example, Bradshaw, Fredette, and Sukornyk (2009a, 2009b) found a positive (although small) correlation between range of diversity and board effectiveness (using a scale composed of items relating to each board’s performance of its core functions—of fiduciary and financial oversight, safeguarding the mission, evaluation of the CEO, and ensuring that strategic planning is taking place). However, Gazley, Chang, and Bingham (2010), in a study of community mediation centers, found no association in any of their models between racial/ethnic or gender diversity and performance. In the corporate sector, Carter, D’Souza, Simkins, and Simpson (2010) similarly found no evidence of the business case for diversity, suggesting that the inclusion of women and ethnic minorities on boards had no negative or positive effects on financial performance. Van der Walt and Ingley (2003) concluded that although the literature on boards strongly supports diversity in the social and moral sense, the business case is not well supported by hard evidence. Jayne and Dipboye (2004), in summarizing the research on the impacts of diversity initiatives, pointed out that the results are often called a “double edged sword” with mixed results, but that if managed well, the benefits outweigh the negatives. In a similar vein, Richard, Kochan, and McMillan-Capehart (2002) said that the evidence of a positive relationship between diversity and performance will be determined by how proactively organizational leaders and participants respond to and manage diversity. In other words, a reinforcing circle of positive effects is possible, depending on the actions taken. Fredette and Bradshaw (working paper) have found that efforts to create more inclusive nonprofit boards, when the focus is on task or functional inclusion, result in a positive relationship between diversity and board performance. These studies suggest the need to do more research on boards and diversity, and that many questions remain about the drivers of diversity, its impacts, and the implications of how efforts to diversify boards are implemented. This article is an effort to contribute to this gap in our understandings of nonprofit boards and diversity by focusing on the determinants of the range of ethnocultural diversity.
Speculation about the determinants of diversity has ranged from macrostructural theories that are more sociological in nature (e.g., Sorensen, 2007) to mesolevel theories of discrimination, to micro- (individual) level (Smith, 2002). In early research on determinants of diversity, the focus was frequently on the impact of the external context on organizational composition. For example, Reskin, McBrier, and Kmec (1999) examined the composition of the qualified labor market, with the assumption that organizations that have race and gender neutral recruitment methods will have a composition of workers who reflect the labor pool in their communities. Tschirhart, Reed, Freeman, and Anker (2009) drew on this sociological tradition to explore the impact of labor markets of individuals with MBA and MPA degrees on the placement of graduates on various boards. Ostrower (2007) found that in the United States, boards that are outside metropolitan statistical areas (MSAs) tend to be predominantly White, potentially reflecting the impact of the labor market and the pool of people in those areas who boards can recruit to fill openings. Prior research has linked demographic changes in the community to the perception that greater diversity was a strategic priority in nonprofit organizations (Ostrower, 2002; Ostrower & Stone, 2006).
External factors other than labor markets have also been examined in the context of boards, and the growth of a contingency perspective (e.g., Ostrower & Stone, 2010) reflects this focus. For example, Brown and Guo (2010) looked at environmental complexity and its impact on board roles. Another factor that has been examined is the impact of market-based incentives, for example the role of the clientele for an organization’s services and products, which may create a sex- or race-specific demand for certain workers to interact with customers (Reskin et al., 1999). Often called the “business case” for diversity, this approach might explain why Ostrower (2007) found that boards of organizations that serve a higher level of minorities are more diverse. She also found that funding source (specifically the percentage of funding from fees, from government, and from foundations) was positively associated with having minority group members on the board. Other researchers have discovered that greater proportions of government funding were linked to higher levels of gender diversity on nonprofit boards (Bradshaw et al., 1996). In a recent survey of Canadian public and private organizations, the top reasons for investing in diversity were ranked as follows: “To access a broader talent pool; to be an employer of choice; to show corporate social responsibility (do the right thing); to mirror the organization’s customer base; to improve or create customer/citizen relationships; to meet legal requirements; to mirror local demographics” (Parris et al., 2006, p. 11).
Researchers have also looked at various organizational and board characteristics such as complexity, size, and age in relation to board composition generally, and to diversity specifically (de Andrés-Alonso, Azofra-Palenzuela, & Romero-Merino, 2009). Theoretically, it has been speculated that larger-sized and older organizations will have more sophisticated personnel systems, and this will lead to more diversity. Ostrower (2007) found that boards that are predominately white are usually smaller in terms of budgets, tend to be smaller in size, and are less likely to represent organizations that are national in terms of geographic scope. Results in other studies have been mixed on some of these factors. For example, no significant relationship was found between the age of the nonprofit organization and gender diversity (Bradshaw et al., 1996; Burke, 2000; Zald, 1969) while in terms of overall diversity (including gender and ethnic differences), researchers have found that larger boards have higher levels of diversity (Carter et al., 2003) and tend to be more inclusive of stakeholders (Brown, 2002). Formalization, defined as the extent to which processes and practices are formally embedded in written policies or structures, has been found to be a determinant of diversity and it is speculated that it works to limit discretion, which can create bias in processes such as recruitment (Reskin et al., 1999). Some research has explored the impact of board formalization on nonprofit board performance (Bradshaw, Murray, and Wolpin, 1992), and their results suggest that formalization is positively correlated with board performance, and marginally related to the board’s ability to fulfill its mission. But overall, there has been limited investigation of formalization and its impact on diversity. Employment practices such as recruitment and hiring models have also been examined to determine their impacts on diversity. Many believe that the use of informal networks, for example, will reproduce an organization’s composition, because workers usually tell people who are similar to themselves about opportunities. Research in the nonprofit field has suggested that systematic recruitment practices may be positively associated with greater inclusivity at the board level (Brown, 2002). Ostrower (2007) concluded that the importance of racial and ethnic diversity as a formal board recruitment criterion is positively related to board diversity, while also emphasizing that friendship with existing board members is negatively associated with having board diversity. Research on the impact of nominations committees has begun, and has yielded mixed results in the context of diversity on corporate boards (Ruigrok, Peck, Tacheva, Greve, & Hu, 2006). The impact of top management team diversity has also been explored by Perkins and Fields (2010), and their findings reinforce those of Ruigrok et al., suggesting that there is a need to consider different dimensions of diversity simultaneously in future research. This is especially relevant when looking at recruitment and nominations of board members, given the need to balance diversity with different skills, perspectives, and backgrounds of board memberships.
This review highlights that there is not yet much systematic research examining the determinants of diversity on nonprofit boards. It also demonstrates that, although the current empirical and normative literatures are rich with suggestions, these ideas are still largely fragmented and untested. Given this review of the literature, we propose an integrated model including four determinants for range of diversity on boards. In particular, we examine community characteristics and external pressures, organizational characteristics, general board characteristics, and board diversity practices. The following section describes the sample and how we operationalized the variables in the model. A visualization of our model is provided in Figure 1. In essence, the model examines the effects on four categories of predictors on the empirical level of range of diversity in a board.
Determinants of the range of ethnocultural diversity on nonprofit boards.
Method
Sample
This research is based on a 2008 survey of boards in the Canadian not-for-profit sector (Bradshaw et al., 2009a, 2009b). The 14-page survey was distributed in both French and English. We pretested the questionnaire in two focus groups of executive directors and revised it based on their feedback. Next, we sent the questionnaire to a total of 825 organizations on the membership mailing list of Imagine Canada (formerly the Canadian Centre for Philanthropy). Just under 30% of these organizations responded by having either their chief executive officer, executive director, or board chairperson complete the questionnaire. This yielded a final sample of 234 organizations. However, we limited our analyses to surveys completed by the executive directors or chief executive officers (N = 183) to gain a greater consistency in the organizational perspectives of our respondents. 2 We believe that this is a representative sample as the relative responses, when compared to all the organizations surveyed, were not obviously skewed toward a particular subsector, size, nor type of organization, although nonresponse bias always needs to be considered. All organizations received at least three mailings, except those in Quebec that only had two mailings because of the time it took to complete the French translation of the survey. This may have limited our response rate in Quebec but we believe the impact is minimal because the vast majority of responses were received based on the first two mailings. The first mailing was a complete package consisting of a cover letter, questionnaire, and postage-paid return envelope. We followed this with a thank you/reminder card about 1 week later. Organizations that did not return a completed questionnaire received a second questionnaire package, identical to the first but without the cover letter. Organizations in the sample were mainly large, with an average annual budget of $10,117,321 and a median number of 11 full-time staff. They were also mainly well established, with a mean age of 42.7 years. Almost 60% of the organizations were located in the province of Ontario, which means the sample is not geographically representative (see Table 1). However, a national survey by Statistics Canada found that Ontario commands a substantial share of total revenues (43%) and employees (47%) of Canada’s nonprofit organizations (Statistics Canada, 2003). Further, more than one half of responding organizations self-identified as primarily focused in either the health (25.1%) or social welfare (29.5%) sectors. For this reason, our findings speak particularly to larger and more established nonprofits, especially those with health or social welfare missions.
Geographic Distribution of Responding Organizations.
Note. χ2(4) = 194.7192.
p < .001.
Measures
Range of diversity in our study is a composite measure, drawing on responses to the question about the current composition of each organization’s board, and using Statistics Canada’s categories of ethnic origin and visible minority status. The term ethnic origin and visible minority is an approved Statistics Canada standard of measure, which refers to whether a person belongs to a group that is non-Caucasian in race or non-White in color. Visible minority status here reflects how many board members are identifiable as having origins from one of 13 broad-based ethnic communities including Aboriginal, Arab, Black, Chinese, Filipino, Japanese, Korean, Latin American, South Asian (e.g., East Indian, Pakistani, Sri Lankan, etc.), Southeast Asian (e.g., Vietnamese, Cambodian, Laotian, Malaysian, etc.), West Asian (e.g., Iranian, Afghan, etc.), White/Caucasian, or other.
Range of diversity was determined by counting the number of different ethnic groups represented on each board as reported by the respondents. In this measure, unfortunately, individuals who are bi- or multiracial are not separately identified and we suspect that respondents used the category of “other” to represent them. As a consequence, the range of diversity may be lower than it should be, and this is a weakness of this measure. Although our survey allowed us to measure diversity more broadly, we had concerns about the reported diversity in sexual orientation and physical ability, because these aspects of diversity are often invisible, and the reports could thus be unreliable.
Relative index of diversity was calculated by using the ratio of the ethnic diversity of the board to the ethnic diversity of the community. Based on the community data on ethnic diversity available to us through Statistics Canada, we made this calculation using the percentages of local population that each nonprofit in our sample is located in, for each ethnic minority group. In other words, a board that is composed of 50% ethnic or visible minority members operating in a community with a population that has 75% ethnic or visible minorities would have a relative diversity index of 0.66 (or 66%). This can be interpreted to mean that the board is 66% as diverse as the community in which it is located. Similarly, a board with 50% ethnic or visible minority members operating in a community that has a population comprised of 50% ethnic or visible minority people would have a relative diversity index of 1 (or 100%), which means that the board and the community have achieved balance (Haughton & Mukerjee, 1995). In our measure of relative diversity, as the ratio gets larger, the board becomes more diverse relative to the community. As the ratio gets smaller, the board is becoming less diverse relative to the community. This measure of relative index of diversity, although not included in the model we test later, provides a useful measure of the relationship between community demographics and board composition and could be used in future studies.
Community characteristics include indicators of the demographic diversity in communities where our survey respondents are located, as well as their perception of the pressures for increased diversity on their boards from stakeholders. We measured demographic characteristics by looking at the percentage of the local population that identified as a visible minority in the Canadian census. We operationalized community as the census tract in which each organization resides, using the postal codes provided by the respondents. Canadian census tracts are small, relatively stable geographic areas that usually have populations between 2 and 8000. Again we used the Statistics Canada definition of visible minority—a person who is non-Caucasian in race or non-White in color.
Second, using responses from the questionnaires, we examined the extent to which the respondents perceive various external community-based factors exerting pressure on their board to embrace diversity. Such factors included competitive pressures, expectations of funders, and donors, changing demographics of the community, clients, or members, and pressure from government priorities or regulations. We conducted principal axis-factor analysis with varimax rotation on the items concerning external pressures to embrace more diversity. In doing so, we found a one-factor solution accounting for 44.3% of the variance. All four items had high factor loadings (see Table 2).
Factor Loadings Based on a Principal Axis Analysis of External Pressures for Diversity.
Organizational characteristics included data on each organization’s size, age, and staffing. Organizational age and size were assessed by asking respondents to identify their organization’s age in years (M = 43.0), and the number of full-time equivalent employees (i.e., paid staff) in the 2008 fiscal year (M = 66.5).
General board characteristics involved an assessment of each board’s size and operations. Respondents gave information on the extent to which their board had various policies and practices in place. Factor analysis revealed two factors relating to what we are calling formalization (see Table 3). The first is a general board formalization scale, which addresses the extent to which the board has detailed printed materials on how it operates, on committee responsibilities, key role descriptions, conflicts of interest, and discrimination policies. This factor accounts for 25.3% of the common variance. Board size was a straightforward measure of how many current members each board has (M = 12.7).
Factor Loadings Based on a Principal Axis Analysis of Formalization Measures.
Board diversity practices indicate the boards’ practices and policies relating to diversity and inclusiveness. One indicator of a board of directors’ commitment to diversity is the existence of a formal definition of diversity. We asked whether the boards had a formal definition of what diversity means for them, as well as the range of categories included in the definition (e.g., aboriginal status, age, country of origin, ethnicity, gender, immigrant or refugee status, language, religion, sexual orientation). Some 41.9% of respondents indicated that their board had its own working definition of diversity. We created a composite measure to account for the scope of each board’s diversity definition, based on counting the number of categories included in their definition.
Another dimension of board diversity practices is the degree of formalization in diversity-focused processes or policies at the board level. The factor analysis revealed a second factor relating to formalization of board level diversity practices that is also included in Table 3. This factor accounts for 25.3% of the common variance and includes dimensions such as whether the board has specific policies regarding racial, ethnic, and gender representation on the board, board self-assessments that include diversity considerations, and whether diversity is explicitly incorporated in the board’s work plans.
Recruitment practices are the final dimension of board diversity practices we examined. We drew these dimensions from the practice literature, in which various strategies for recruiting more diverse board members are frequently outlined. From the reported lists of recruitment strategies, our factor analysis revealed three key types of approaches (see Table 4). The first approach we labeled “print and publication,” which is related to use of media to find board members (e.g., publishing board opportunities on a web site, in e-communications, or through advertising in a newspaper or ethno-specific publication). The second approach related to use of interorganizational alliances (e.g., partnering with ethnocultural organizations to help identify potential board members) and the third approach concerned activation of interpersonal networks (mobilizing board member contacts or targeting individuals active in the community). Cumulatively, these three factors explain 42.6% of the common variance.
Factor Loadings Based on a Principal Axis Analysis of Board Recruitment Practices.
Results
For our boards, the relative index of diversity ranged from 0 (in cases in which the board had no ethnic or visible minority members) up to 16.72—which is a very diverse board (60% of that board is composed of ethnic minorities in a very homogeneous community, which resulted in it being much more diverse than the community in which it works in, hence the relative index of diversity was well over 1). It should also be noted that this exceptionally diverse board is one of the smaller boards. Our sample had a relative index of diversity with a mean of 1.2, and a SD of 7.9. This means that taken as an average, the board compositions of our sample are relatively close to the ethnic composition of the communities in which they are located. However, the standard deviation of 7.9, which is much larger than the mean itself, indicates that there is a great deal of variation in the relative index scores. Some boards were much less diverse than their communities, and some boards were much more diverse than their communities, which averaged out across our sample. We believe this measure has potential value in future research, but it is important to note that the following analysis relies on the range of diversity measure to examine the factors that are related to diversity on boards.
In terms of the range of ethnocultural diversity (the number of different groups represented on the board), the median number was two. Looking more closely at the range of diversity among board members, 42.6% of the organizations selected had only one group present, 25.0% had two groups, 17.6% had three groups, 9.1% had four groups, and 5.7% had five or more groups represented among their board members. Although there is some progress in attracting board members from different ethnic groups, there is more progress in terms of gender diversity. Women in this sample held almost 44.2% of seats on boards (no statistically significant difference from the results of the survey of Canadian nonprofit boards conducted by Bradshaw et al., 1996). In addition, 61.5% of those who completed the survey (Executive Directors or CEOs) were women. Of the boards in the sample, nine were all female boards, and four were all male. Almost three-quarters of the organizations (70.6%) indicated that there was no person with a disability on their board, and 74.7% of those surveyed said they did not have a board member who was openly lesbian, gay, or bisexual.
Our measure of range of diversity is related to several other dimensions of diversity within the boards and we used a chi-square measure, instead of a correlation, as it is better suited to categorical variables such as the one used to assess range of diversity. Range of diversity is significantly related to the percent of reported disabled members on the board (χ2 = 12.41, p = .03) and to the percent of reported lesbian or gay members of the board (χ2 = 14.14, p = .028). The percent of females on the board is also significantly related to the range of diversity (χ2 = 244.8, p = .004).
Correlation and Regression Analysis
The means, standard deviations, and zero-order correlations among the study variables are presented in Table 5. The range of ethnocultural diversity among board members correlated significantly with community diversity (i.e., boards located in more diverse communities tended to be more diverse which is consistent with the findings using the relative index of diversity), and also with board size, the two formalization scales (general board formalization and formal diversity-related processes), and scope of the definition of diversity. One of the recruitment measures relating to interorganizational alliances was significantly correlated, and a moderate positive correlation was found between range of diversity and the external pressures for diversity scale.
Zero-Order Correlation Coefficients, Means, and Standard Deviations.
p < .05, **p < .01.
We performed a series of standard multiple regressions with range of ethnocultural diversity on boards as the criterion (findings of all regressions are reported in Table 6, see Models 1-7). The first regression included community characteristics and the external pressures for diversity scale as predictors of diversity. Results (see Model 1) revealed a significant model, F(2, 138) = 3.773, p = .02, accounting for 4.7% of the variability in range of board diversity. Community diversity was a significant predictor, that is, more ethnically diverse communities tended to have boards that reflect a larger range of diversity although external pressures were also significant. The second regression (see Model 2) examined the influence of organization size and age in predicting the range of diversity. These two predictors accounted for 3.8% of the variance in range of diversity, F(2, 171) = 3.377, p = .036, although organizational size was the only variable that was significant. In the third regression (see Model 3), we included two predictors: number of active board members, and the general board formalization scale. These predictors accounted for 9.8% of the variance in range of diversity, F(2, 160) = 8.689, p < .001. Both general board formalization and the number of board members were significant predictors. The fourth regression (see Model 4) examined board practices with respect to diversity as predictors of range in diversity. Specifically, we included broad-based definition of diversity, formal board diversity practices scale, and the three recruitment practices measures with range of diversity used as the criterion. These predictors accounted for 26.7% of the variance in range of board diversity, F(5, 115) = 8.395, p < .001. However, the presence of formal diversity practices was the only significant predictor.
Results of Multiple Regression and Hierarchical Analysis on Range of Diversity.
p < .10, **p < .05, ***p < .01, ****p < .001.
We also used regression based on hierarchical analysis to demonstrate the partial effects of independent variables on our dependent variable (Cohen, Cohen, West, & Aiken, 2003). This type of sequential or step wise addition of the variables as described in our model (see Figure 1) allows us to see the change in the R2 with the addition of each block of variables as well as the overall R2 for each model. Thus in the first step (see Model 5), we entered control variables (i.e., community diversity and external pressures scale), and then subsequently added each set of independent variables in blocks (see Models 6 and 7). A change in the explained variance was used to determine levels of significance. Results (see Model 5) revealed that community characteristics and external pressures accounted for 7.3% of the variance in range of diversity (F(4, 135) = 2.954, p < .01) in the presence of organization size and age. Although organizational characteristics and general board formalization accounted for an additional 6% variance in the range of diversity (Model 6). Looking at the change in R2, we see that the addition of formal board diversity practices (see Model 7) accounted for a further 23.5% of the variance in range of diversity, F(11, 94) = 4.557, p < .001. Specifically, the formal diversity practices scale, board size, community diversity, and the extent to which they used interorganizational alliances (e.g., partnering with ethnocultural organizations to help identify potential board members) to increase diversity on the board, were significantly associated with the range of ethnocultural diversity.
Discussion
Our results show that although women are strongly present on the larger and older Canadian nonprofit boards that compose our sample, the representation of visible minorities and different ethnic groups remains low, although reflective of the communities in which they are located. Although this type of research cannot determine the direction of causality, our results do identify factors that are significantly related to a greater range of ethnocultural diversity on boards. For example, community diversity, as measured by Statistics Canada, is associated with the range of ethnocultural diversity on the boards in our sample. This is not really surprising from an intuitive perspective, but from a theoretical perspective it is still interesting. Is this a manifestation of Reskin et al.’s (1999) argument that organizations having race- and gender-neutral recruitment methods will have a composition of workers which reflects the labor pool in their communities? It may suggest that time and the increasing community diversity will gradually result in more representative boards.
Although organizational size in terms of the number of paid staff was a significant determinant of diversity, the age of the organization was not significant in the regression analysis. This is consistent with earlier research on gender of board members where Bradshaw et al. (1996) found no relation between organizational age and the number of women on the boards. General board formalization was significantly related with range of diversity in some of the regressions; however, the formalization of board diversity processes was the most consistently significant scale. This suggests that boards that are more institutionalized, with better-developed policies and practices, and in particular with policies relating to diversity (such as including diversity goals in the planning processes and board self-assessments) have more success in attracting a range of ethnoculturally diverse groups to the boards. The results also suggest that larger boards are more diverse. There have been different findings about the impact of the size of boards on diversity (Bradshaw et al., 1996; Burke, 2000). In our study, we found a statistically significant relationship between board size and board diversity, which echoes the findings of Carter et al. (2003) that larger boards tend to be more diverse. Perhaps larger boards are able to be both more inclusive of different skills sets while also embracing people from more diverse communities. This may help boards who see a tradeoff between getting the skills they need (e.g., legal and human resources) as well as representation from diverse communities.
We looked at the use of different recruitment practices and found little evidence to support the notion that relying on print or other publications (e.g., posting board openings on a web site, e-communicating with potential candidates, or advertising in newspapers about openings), or on activating personal networks had any relationship with actual range of ethnocultural diversity on the boards in our sample. However, developing interorganizational alliances (e.g., partnering with ethnocultural organizations to identify qualified board candidates) was positively associated with range of ethnocultural diversity.
Conclusions
Our study suggests that macrocontextual/environmental factors may have impacts on board diversity. Particularly, it appears that the diversity of the community had a significant impact, but we did not find a robust relationship between external pressures from funders, competitors, or government priorities. These interactions deserve more investigation. Ostrower (2007), for example, did find positive associations between minority group membership on the boards in her sample and funding sources. Resources dependence theory (Pfeffer and Salancik, 1978), for example, suggests that funders will have an impact on decision making because organizational survival hinges on the ability of the organization to obtain critical resources from its external environment. If funders make their funding allocations to dependent organizations contingent on changes in the composition of the board (to reflect the diversity of the community), then we would expect such requirements to impact board membership as the dependent organizations respond in ways that are designed to reduce uncertainty and increase access to scarce resources (Casciaro & Piskorski, 2005; Mizruchi & Yoo, 2002; Sherer & Lee, 2002).
Perhaps our most notable finding is that nonprofits who want to increase the range of ethnocultural and visible minority representation on their boards, regardless of organizational age or size, benefit from embedding and institutionalizing formal diversity related practices (e.g., policies regarding representation of diverse groups on the board such as including diversity considerations when doing board self-assessments, and incorporating diversity goals into work plans of the board). Rather than suggesting a risk of bureaucratic rigidity from formalization, our findings suggest that policies and practices that are formalized actually coincide with improvements in the range of boardroom diversity. Why would this be? One reason may be that with institutionalization, diversity comes to be seen as good board practice. It becomes “just the way we do things around here,” as the policy is routinized. Embedding the policies and practices ensures that they continue and perhaps reduces the tendency to introduce personal bias and idiosyncratic approaches to boardroom processes.
Our findings relating to the three facets dealing with recruitment practices indicate that only the interorganizational alliances construct had a significant relationship with board-level range of diversity. We failed to support a common normative prescription often heard in Canada, namely that many proactive recruitment strategies are key to improving diversity, and this finding gave us reason to pause. Ostrower (2007) also found that emphasizing friendship with existing board members as a recruitment criterion was negatively associated with having any minority members. This is an area warranting future study, as it may be that board members’ efforts to recruit and attract new members from diverse communities is constrained by the dynamics of exclusion and homosocial reproduction (Elliott & Smith, 2001), and negatively impacted, by board members’ own limited networks and connections. This may suggest the opportunity to develop a social capital view of diversity (Fredette & Bradshaw, 2012; Van der Walt & Ingley, 2003).
Despite the contributions of our research, several important limitations should be noted. First, our sample is predominantly composed of larger, older organizations based in the health and social welfare domains of the Canadian nonprofit sector. Second, although we surveyed more than 800 organizations and had a 30% response rate, it may be possible that responding organizations were those most familiar and comfortable with issues of diversity and inclusion. One result of this might be to give the appearance of greater diversity than is actually present in the sector. Also, we relied on a single informant (all the informants being senior executive leaders) at one point in time, which limits the generalizability of our findings and restricts our ability to determine the direction of causality. Finally, despite investing in professional translation services, the potential exists for deviations in understanding resulting from the translation of the survey from English to French.
From a practical perspective, our study suggests some possible levers that those interested in enhancing the range of diversity on their boards might consider utilizing, and also suggests which levers may not be effective. For example, having a larger board may allow more diversity, perhaps as it creates opportunities to recruit both members with the necessary skills and members who come from diverse communities. On smaller boards, these different criteria for selecting members may be seen as competing priorities. Embedding and formalizing diversity oriented policies also appears related to diversity outcomes, and although we do not know whether these practices result in more diversity or if it is diversity that leads to formalization of practice, the connection is strongly and statistically significant. Although external pressures did not appear to have a strong impact within our sample, being located in a diverse community was clearly related to board diversity.
Our model, we believe, can be further tested in other contexts to begin to more systematically assess what is driving board diversity (with a focus beyond ethnocultural diversity). A growing base of studies across countries (e.g., Ostrower, 2007) will be helpful to assess the generalizability of various factors. More research is needed, but this study makes a contribution from a Canadian sample, and will help us track changes over time. Future research is also needed that is longitudinal in nature so that the directions of causality can be determined. Given the impact of diversity in the community where the nonprofit is located, future research that uses a measure of the relative index of diversity (e.g., the ratio of diversity in the nonprofit board divided by the diversity of the community) would be helpful in exploring this dynamic more fully (Haughton & Mukerjee, 1995).
From a theoretical perspective, our study reinforces the possibility to examine the determinants of diversity from multiple theoretical perspectives and from various levels of analysis. For example, institutional theory and resources dependence theory at a macrolevel may help explain the impacts of the community diversity and formalization. Similarly, our findings seem to support the work of Reskin et al. (1999), who suggest that the composition of the external labor pool, if policies and practices are race and gender neutral, will lead to more diversity. Examining these issues from the group or mesolevel dynamics may be enhanced by using a social capital perspective. Although our model attempts to bring multiple factors together, more research is required.
Footnotes
Acknowledgements
The authors thank the support of the Institute for Governance of Private and Public Organizations, Imagine Canada, and SSHRC. The authors also thank Heather Krause, Lisa Fiksenbaum, Yan Yan Wu, Julie Ann Israel, The Institute for Social Research at York University, and particularly Liza Mercier.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.