Improve Your Diversity Measurement for Better Outcomes

Managers need better data collection practices if they are to gain a clearer picture of DEI and design more effective interventions.

Reading Time: 16 min 

Topics

Permissions and PDF

Michael Austin/theispot.com

If business leaders hope to move the needle on diversity, equity, and inclusion (DEI), it’s critical that they put measurements in place to track progress and hold managers accountable for results, something that few organizations are currently doing effectively.1

Indeed, the more than 2,200 executives who have signed the CEO Action for Diversity & Inclusion pledge have committed to driving “measurable action,” but recognizing that something should be measured is not the same as knowing how to measure it. And for many organizations, a lack of understanding of precisely what metrics to collect, how to do so, and how to interpret the data is a common stumbling block in their DEI efforts, contributing to underwhelming returns on their diversity investments.2

In this article, we draw upon more than three decades of collective research and practical experience to highlight six factors that should be considered when deciding how to measure and track DEI data.

Effective DEI Measurement Starts With Sound Goals

The first step to improving DEI measurement and the effectiveness of such efforts is to have a clear picture of the organization’s strategic goals and objectives for DEI. What is the point of collecting and tracking DEI-related metrics for the company? Does it want to improve representation for an underrepresented group, or ascertain whether there is pay inequity across employees and/or jobs? After determining broad goals, develop them into more specific ones that can provide an identifiable outcome and help frame plans and processes for measurement.3

One fundamental principle for such goals is that they should encourage steady improvement over time rather than simply target absolute numbers to be attained. Showing steady improvement compared with one’s past provides nuance that snapshot numbers alone fail to demonstrate, especially when making comparisons: At a certain point in time, organization A may have a higher percentage of employees from underrepresented groups than company B, but if A’s percentage has been steadily declining while B’s has been steadily growing, the seemingly less-diverse group is in fact making better progress. Tracking and comparing data over time can show that the organization is not only trying but succeeding at improving its diversity and/or inclusion.

After goals are in place, organizations can take a more strategic approach to collecting and analyzing DEI data. While there is no one-size-fits-all approach, we contend that any measurement of DEI efforts should encompass all levels of the organization, consider demographics of different geographic locations, account for multiple types of diversity, look at linkages across the data collected, and consider the impact of when data is gathered. Let’s look at each of these factors in detail.

Take a Multilevel View for a Clearer Picture

Numbers can be used strategically to mislead others, as Darrell Huff showed nearly 70 years ago in his classic book How to Lie With Statistics.4 In the context of DEI, summary statistics that obscure what’s happening at different levels of the organization can paint a rosy but inaccurate picture of progress. For example, we consulted with a university seeking to promote itself as highly diverse in terms of race and ethnicity. It supported this claim with several summary statistics about employee and student composition. Although the student body was indeed impressively heterogeneous, the university’s summarized employee data hid an unfortunately all-too-common reality: Faculty and senior administrators were overwhelmingly White, and their physical plant workers and administrative staff members were predominantly racioethnic minorities.

This particular pattern of stratification of diversity and inclusion is by no means unique to that university. A quick perusal of statistics from the U.S. Bureau of Labor Statistics shows that racial, ethnic, and gender diversity (the most easily trackable forms of visible difference) tend to be most pronounced at lower hierarchical levels and become progressively rarer as one moves up the corporate ladder in authority, responsibility, and rewards. Likewise, our research indicates that managers often perceive greater organizational support for diversity than do rank-and-file employees (a tendency that has negative implications for unit performance).5 Summary statistics aggregating across all hierarchical levels are likely to mask these over- and underrepresentations.

When unpacking summary statistics, companies need to pay particular attention to potential pockets of concentration. Nike, IBM, and Salesforce are among the companies that are doing this well. When reporting their DEI metrics, they indicate demographic representation separately for leadership roles, tech roles, and nontech roles. This type of examination of diversity helps to avoid overemphasizing diversity at lower levels as representative of the overall diversity within the company.

Organizational sociologists have long recognized the pitfalls associated with overrepresentation of racioethnic minority and female staff members in certain operating units within a company (as well as in some professions).6 Units and professions dominated by women and minorities tend to have fewer resources available and offer smaller financial rewards, less prestige, and fewer promotional opportunities. Though it is true that overconcentration alone does not necessarily indicate inequity, it certainly suggests a lack of equal opportunity for all. Our research shows that employees notice when organizational leader profiles diverge from those of the rank-and-file, which diminishes their trust in the organization and increases their perceptions that they have been mistreated.7 Accordingly, the identification of considerable underrepresentation at any organizational level should trigger a more focused investigation to determine its causes and consequences.

When collecting and tracking demographic and other diversity-related data, it is important to consider factors such as job type, position, and hierarchy when analyzing and interpreting it. Such factors provide a clearer picture of how an organization is doing in terms of representation.

Account for Location

Organizations exist in very specific contexts: They are embedded in communities with living histories and characteristics that affect who lives there, what cultural norms prevail, and what it’s like to live there. Considering the surrounding social context and its influence on diversity dynamics within the organization is of paramount importance. Despite the growing number of employers allowing fully remote work, most people still live and work within the same general community. Some demographic characteristics, like gender, are relatively consistent from one community to another, but others, such as nationality, race, ethnicity, sexual orientation, and socioeconomic status, vary considerably more.

For instance, it is most likely easier to recruit Black talent in Memphis, Tennessee, where Black people make up almost 65% of the population, than it is in Salt Lake City, where Black people account for less than 3% of the population. Such racial differences in communities likely affect racial differences throughout the employment cycle, which may exacerbate inclusion issues if they are present. In addition to these compositional differences, collective attitudes related to differences such as race or sexual orientation vary across locales, resulting in some cultures and climates being more hospitable to members of underrepresented groups, whereas others are less tolerant.8

Not all organizations adequately recognize the implications that such differences between locations have for their DEI measurement efforts. For instance, some companies establish goals for the entire organization based on companywide benchmarks or national or regional market data. This approach sets expectations that an office in a relatively homogeneous location like Salt Lake City can achieve similar levels of racial diversity as an office in a more heterogeneous locale like New York City. Similarly, establishments in areas with histories of hostility toward particular groups are expected to be as successful in recruiting and retaining members of these groups as those based in welcoming communities. (Consider recruiting LGBTQ+ individuals in many Middle Eastern countries as opposed to San Francisco, for example.) DEI goals must be attainable if compensation is tied to reaching certain metrics. Holding managers accountable for unreasonable diversity goals is more likely to demotivate than inspire them.

Instead of ignoring local contexts, companies should consider community characteristics when setting targets and appraising progress toward DEI goals. Local labor markets provide valuable context about what is reasonable to expect regarding representation. Who lives in the surrounding community? If certain groups are underrepresented within that community, there may be reasons why (such as the cultural climate) that are difficult for an employer to influence. Are there colleges and universities that could serve as feeders, and, if so, what is the composition of their student bodies? Comparisons with competitors should also be local to ensure that targets are attainable. If a local rival is consistently more successful at attracting and retaining members of certain underrepresented groups, what makes it so much more enticing to job applicants?

Companies should consider community characteristics when setting targets and appraising progress toward DEI goals.

Locations also provide context about the meaning of a particular demographic characteristic, such as race or religion. It may be convenient for a multinational conglomerate to assume that a demographic characteristic that means one thing in a particular setting has a comparable meaning in another setting. Such an assumption, however, may be unrealistic for some characteristics, such as race, ethnicity, sexual orientation, or religion. For instance, Google and other companies have been criticized for employing U.S. Census Bureau racial and ethnic classifications in countries where these categories may not have the same meaning. Categories like Hispanic, Black, and African American are relatively unambiguous in the U.S., but there is less consensus regarding their meanings in other national contexts. It’s difficult to draw definitive conclusions from cross-contextual comparisons when there isn’t shared meaning underlying the metric.

That said, the absence of a universally agreed-upon categorization scheme should not be used as a justification to avoid measurement altogether. As the saying goes, don’t let the perfect be the enemy of good, and remember that DEI strategies and sustained progress require sound measurement and tracking. Companies should think about the DEI factors that are relevant and important for different locations and make adjustments that better capture data that is relevant to the location of interest.

Track Multiple Dimensions of Diversity

Diversity efforts should be comprehensive and consider groups that are underrepresented based on a variety of characteristics, including racioethnicity, gender identity, sexual orientation, and disabilities. It may be tempting for organizations to tout one form of diversity within their organization as evidence that the company is sufficiently diverse. For instance, a company may be doing well in terms of gender diversity but have very little racial diversity. Another company may be racially diverse overall but have little to no LGBTQ+ representation in its workforce (or assume as much because it isn’t actually measuring it). In any case, collecting and reporting metrics that pinpoint specific demographic groups as opposed to aggregating across diversity characteristics provides a more accurate picture of the state of diversity within the company, and it allows the organization to develop an action plan that is specific to its needs.

We realize that it can be difficult to collect demographic data on concealable identities that have historically received little legal or organizational protection, such as sexual orientation and gender identity. Organizations may leverage employee resource groups to help determine the needs and interests of specific populations. In addition, organizations that have the resources could contract with outside diversity partners to collect employee data that can be connected to organizational data in order to provide information on how well the organization is managing the concerns of particular underrepresented groups. Partnering with outside consultants may help alleviate employees’ potential fears of being outed or of experiencing backlash if concealable stigmatized identities can be linked to specific individuals by others within their organization.

In addition to not focusing on only one or two types of diversity, organizations should more carefully attend to the role of intersecting identities in understanding true compositional makeup and potential bias within the company. Research has demonstrated that the workplace impact of a demographic characteristic like race is often influenced by other characteristics, such as gender, and vice versa.9 For instance, the experiences of Black women differ from those sharing only their race (like Black men) or gender (like White women).

What does it look like when we dive deeper into the data on the percentage of women hired and investigate on the basis of their race as well? Take Google, for instance: Its 2022 annual diversity report showed an increase in the number of women it hired in 2021, with women making up 37.6% of new employees in its U.S. workforce, compared with 2020, when women made up 33.1% of new hires.10 Though Google saw an increase in hiring women of all races, Black, Latinx, and Native American women accounted for a noticeably lower percentage of 2022 new hires (4.3%, 3.2%, and 0.3%, respectively) compared with White women (14.8%). If we consider individual demographics only in isolation, we gain at best an incomplete picture — and, more likely, an inaccurate one. Taking a more fine-grained, intersectional approach to data tracking allows organizations to gain a better understanding of where there may be gaps that indicate bias or other issues for certain groups that might otherwise be overlooked.

Integrate Data Systems to Better Track Outcomes

Too often, organizational plans for DEI measurement begin and end with the choice of which demographics should be measured. But a myopic focus only on how many people from underrepresented groups are present in the workforce will leave managers blind to how those demographic characteristics could lead to inequities in opportunities and outcomes.

More progressive organizations are presently tackling this issue by both identifying and beginning to address simple demographic disparities. What we mean by “simple” is that there is a difference in the outcomes (such as compensation and promotions) for one group, like gay employees, and for that of another along the same dimension, like straight employees. Companies can link available demographic information on employees to their human resource records and look for correlations between identity markers and outcomes like selection rates, compensation, performance appraisal ratings, job assignments, and promotions.

Those activities are a good start, but it is imperative that organizations go further. Metrics can also capture demographic differences in rates of association that amount to differential returns or penalization. Say, for instance, that autistic employees received less than half the return in financial compensation (raises and bonuses) for accumulating five years of managerial experience compared with their neurotypical counterparts. Solely considering the simple effect of disability on compensation could miss this important source of inequity that requires the organization to actively look for and consistently track differential returns.

Our own work illustrates the potential consequences of not effectively integrating data systems and interpreting HR data across the employment cycle. For instance, we were initially contracted by a chemical company to help update its employee selection system to improve the diversity of new hires only to find that it did not engage in any forms of targeted recruitment. Redesigning its selection system would have proved fruitless (or underdelivered, at best) if we hadn’t first helped the company recognize the importance of investing in targeted recruitment of applicants from underrepresented groups. Integrating its data on who is applying and who is chosen would help the company recognize how shortfalls in recruitment often influence selection outcomes. In short, you can’t select people who will help increase diversity if they don’t apply.

In addition, even companies at the vanguard of human resource analytics and data mining often fail to sufficiently integrate their various data management information systems. For example, in working with a large retailer that had been commended for its diversity management efforts, we discovered that personnel records on lateness and absenteeism were located in one system, whereas performance appraisal data was in another, and the company had no means of integrating the two. This was troublesome because our own research has shown that Black employees tend to be punished more harshly for tardiness than White employees, but the company’s siloed HR information systems made it impossible to discover whether this was happening.11 Similarly, the failure to examine these kinds of connections means small issues can snowball into large ones before they are identified and addressed.

One linkage that is ignored all too often is the connection between recruitment and retention. When examining the full spectrum of the employee pipeline, we’ve found that organizations doing well on sourcing and hiring rates for some underrepresented groups often have high attrition rates for these same groups. This could be indicative of inclusion issues, such as demographic differences in people’s perceived sense of belonging, that should be addressed.

Factor In Time of Year When Collecting Data

In some sectors, there may be seasonal impacts on hiring that shape an organization’s demographic profile, which can thus vary based on time of year. Unfortunately, many organizations merely take a snapshot of demographics at the same time each year. That data can be used to disingenuously suggest a commitment to DEI that, upon closer scrutiny, is revealed to be misleading.

For example, some organizations strategically take this snapshot during the December holiday shopping season, when retailers employ seasonal workers. In many cases, these are low-wage workers — a group in which women and people of color are disproportionately represented.12 Capturing an annual DEI profile at this time might intentionally, or unintentionally, make the company’s progress on diversity seem greater than it truly is.

Such tactics can backfire if employees believe that the organization is engaging in DEI efforts for exploitative purposes, such as to reach quotas, rather than to promote equality.13 For example, one organization we worked with had received criticism from its Black and Latinx members about underrepresentation. The organization countered by showing that it compared reasonably well (albeit at slightly lower levels) against others in its peer and aspirant group. What management failed to consider was that the representation of both groups was on a downward trajectory and their populations were at 20-year lows for the organization.


We’re encouraged that executives, human resource managers, and data analysts increasingly recognize the importance of good measurement for effective diversity management. But the devil is in the details of organizational DEI metrics. While the considerations we’ve reviewed in this article apply universally, each organization must tailor its own approach — in some cases, multiple approaches — on the basis of its unique characteristics. We have identified five considerations (hierarchical levels, location, types of diversity, linkages across data, and time of year) that we hope will help guide conversations among the growing number of executives pledging their commitment to DEI and the strategists charged with actualizing this commitment.

Taking the time to strategically plan and set DEI goals will help improve measurement because it provides greater focus on what metrics need to be collected and how they should be analyzed. Metrics can help leaders understand both their current position and where they are going on the path to helping their organization become more diverse, equitable, and inclusive. To maximize the utility of DEI metrics, executives and HR specialists should consider the tips provided here that will help create a more accurate picture of the overall DEI assessment.

Senior-level commitment is an important step toward making an organization more inclusive. As executives begin to prioritize DEI in ways their companies have not in the past, they will seek to increase managerial accountability and likely use metrics to do so. By taking these considerations into account, they can make these metrics more meaningful and increase the returns on their DEI investments.

Topics

References

1. E.J. Kennedy, “Can Data Drive Racial Equity?” MIT Sloan Management Review 62, no. 2 (winter 2021): 9-11.

2. P. Newkirk, “Diversity, Inc.: The Failed Promise of a Billion-Dollar Business” (New York: Bold Type Books, 2019).

3. E.N. Ruggs and D.R. Avery, “Linking Good Intentions to Intentional Action,” MIT Sloan Management Review 62, no. 4 (summer 2021): 95-96.

4. D. Huff, “How to Lie With Statistics” (New York: W.W. Norton, 1954).

5. P.F. McKay, D.R. Avery, and M.A. Morris, “A Tale of Two Climates: Diversity Climate From Subordinates’ and Managers’ Perspectives and Their Role in Store Unit Sales Performance,” Personnel Psychology 62, no. 4 (winter 2009): 767-791.

6. B.F. Reskin, D.B. McBrier, and J.A. Kmec, “The Determinants and Consequences of Workplace Sex and Race Composition,” Annual Review of Sociology 25 (1999): 335-361.

7. A.P. Lindsey, D.R. Avery, J.F. Dawson, et al., “Investigating Why and for Whom Management Ethnic Representativeness Influences Interpersonal Mistreatment in the Workplace,” Journal of Applied Psychology 102, no. 11 (April 2017): 1545-1563.

8. M.J. Gelfand, L.H. Nishii, and J.L. Raver, “On the Nature and Importance of Cultural Tightness-Looseness,” Journal of Applied Psychology 91, no. 6 (December 2006): 1225-1244.

9. A.N. Smith, M.B. Watkins, J.J. Ladge, et al., “Making the Invisible Visible: Paradoxical Effects of Intersectional Invisibility on the Career Experiences of Executive Black Women,” Academy of Management Journal 62, no. 6 (June 2019): 1705-1734.

10.Google Diversity Annual Report 2022,” PDF file (Mountain View, California: Google, 2022), https://static.googleusercontent.com.

11. A. Luksyte, E. Waite, D.R. Avery, et al., “Held to a Different Standard: Racial Differences in the Impact of Lateness on Advancement Opportunity,” Journal of Occupational and Organizational Psychology 86, no. 2 (April 2013): 142-165.

12. F. Hanleybrown, L. Iyer, J. Kirschenbaum, et al., “Advancing Frontline Employees of Color: Innovating Competitive Advantage in America’s Frontline Workforce,” PDF file (Boston and Oakland, California: FSG and PolicyLink, January 2020), www.fsg.org.

13. P.F. McKay and D.R. Avery, “Warning! Diversity Recruitment Could Backfire,” Journal of Management Inquiry 14, no. 4 (December 2005): 330-336.

Reprint #:

64212

More Like This

Add a comment

You must to post a comment.

First time here? Sign up for a free account: Comment on articles and get access to many more articles.

No comments