The Empirical Evidence of Team Diversity on High Performance Outcomes

Abstract

Research into diversity and inclusion in the workplace is not a new topic. However, the focus has largely been on theoretical or sociological impacts of diversity and has tended to be anecdotal. Typically, diversity and inclusion at organizations is viewed through the lens of it being the “right” thing to do. This paper builds on that premise while also leaning into some of the quantitative impacts that diverse teams have on organizations. While there have been other articles that begin to explore quantitative nuances to diversity in a corporate environment, this analysis uses the Shannon Index methodology from biological sciences to offer a novel framework for evaluating the correlation of diversity on positive business outcomes such as retention and team engagement.

While diversity can often — and correctly — be viewed through smaller lenses such as gender, ethnicity, and age, this analysis attempts to view diversity through a holistic and global lens. It is also an analysis using a singular organization’s data which necessitates that the Shannon Index framework is used by others to evaluate the same proposed impacts at their organizations, larger organizations and potentially in a cross-organization or industry study. This Shannon Index model uses gender, age, tenure — as a proxy for diversity of thought — as well as regional representation to distill individual team diversity into a singular, composite score. That score is then used in panel regression analysis to determine the impact diversity has on voluntary attrition within smaller teams as well as self-reported senses of belonging and engagement.

Ultimately, it is found that the top 25th percentile of diverse teams demonstrate statistically significant advantages in retention (certainly over the bottom 25th percentile). Those same groups also have demonstrably higher results in team engagement (which can be a more stable driver of retention likelihood).

Prior Research into the Empirical Impact of Team Diversity

Analysis into the benefits of a diverse workforce has laid the foundation for this study. Prior research from Elizabeth Foma, in Impact of Workforce Diversity written in 2014 explains that “the fundamental missing word to social justice within business is to make certain that the workplace remains a place of diversity. Not only does this ensure recognition of basic rights and liberties, but it also creates an atmosphere of trust that is often lacking between corporation and consumer.” (Foma, 2014) Additionally, Foma touches on how beneficial diversity in the workplace is from an economic standpoint. “Some economic benefits include increase in the pool of qualified personnel from different backgrounds, and it widens the scope of eligible candidates for hire. Another benefit of a diverse workplace is it improves communication with an organization’s clients. Additionally, a sense of harmony is recreated when a company recognizes and accommodates the differences within its diverse workforce” (Foma, 2014). This was derived from data analysis from PepsiCo’s article from their online website. Diversity is a main contributing factor for prospective Generation Y and Generation Z candidates in determining employment decisions. This research demonstrates that there is more engagement in more diverse companies and teams.

In a qualitative study from 2013, Priscilla Dike found that the size of a company, as well as prioritization and financial backing, play a massive part in differentiating whether or not a company implements practices to strengthen the diversity in their workforce. “Workplace diversity has contributed to more productivity but some factors such as differential treatment could hinder its successful implementation and hence company success. Big companies are more passionate about diversifying their workforce and see its implementation as a norm and continuously strive to improve diversity management, whereas small companies see it as a choice and evitable when they feel it a burden or cannot effectively manage it.” (Dike, 2013). However, these specific focus groups did not have empirical studies to validate the authors qualitative findings.

Some additional insights found in research of Harold Patrick and Vincent Kumar, in Managing Workplace Diversity: Issues and Challenges (2012) present the quantitative connection between team diversity and team performance focused research on these several components:

• Strategies to enhance workplace diversity
• Frequently Encountered barriers
• Mean and Standard Deviation for Strategies to Increase Inclusiveness
• Mean and Standard Deviation for Strategies to Increase Awareness About Workplace Diversity
• Indicating Mann–Whitney U and t-Test for Strategies to Increase Inclusiveness Between Gender
• Indicating Kruskal–Wallis Test and ANOVA for Strategies to Increase Inclusiveness Between Age Groups

The most prevalent outcomes of higher diversity prove to be unleashing creativity and increasing employee morale. Having the freedom to be innovative, combined with improving the experience of team members proves to be successful. On the opposite end, attempting to decrease employee complaints and litigation proves to be a rather restrictive approach to handling workplace diversity. (Patrick and Kumar, 2012).

While they have identified the most and least effective strategies into boosting workplace diversity, Patrick and Kumar made sure to identify the frequently encountered barriers in their study. Among the top of the list was found to be discrimination and prejudice. These concepts stem from lack of education and an unwillingness to move away from stereotypes and ideology that are learned through exclusion. Surprisingly, the least prevalent among the barriers was backlash. This demonstrates that the least common barrier when it comes to their study, has very little to do with the subjects’ reaction from improving diversity (Patrick and Kumar, 2012).

The next few analyses demonstrate quantitative testing of strategies to increase inclusiveness and awareness of workplace diversity. Patrick and Kumar use Mann–Whitney U and t-Tests as well as Kruskal–Wallis tests and ANOVA tests. These tests were implemented to measure the most effective strategies for increasing inclusiveness among genders and age groups. For context, all tests were used to determine if any intervention had a statistically significant impact versus the null hypothesis. When looking at mean and standard deviation for strategies to increase inclusiveness and awareness about workplace diversity, it was found that the most preferred strategies are by learning about cultural difference. Understanding the importance of self-awareness and the ability to delve into the differences in culture is pivotal. (Patrick and Kumar, 2012). They also found that dismissing myths about diversity when in a group of individuals is the least preferred strategy. Being vulnerable in admitting that not everyone is the same and peer-to-peer education is critical to positive growth. (Patrick and Kumar, 2012).

The final test for analysis on increasing inclusivity between groups, shows how tenure — as a proxy for diversity of thought — plays an integral role in how team members perceive inclusivity; teams often think of diversity of thought differently than factors such as age and gender.

Results show high variability in preferences and expected output between tenure, gender, and age. The biggest takeaway from the study is that preferences in working model and expected output differs based on the demographic group that an individual is in. An optimal team borrows from all types of backgrounds to maximize the distribution of preferences and increase the likelihood of high performance (Patrick and Kumar, 2012).

Previous data analyses allow a view into the wide range of components that factor into the correlation of team diversity and strategies to increase team diversity and inclusivity. The data illustrates quantitative strategies to improve diversity, as well as barriers that can possibly hinder the success in improving diversity. These analyses did not explicitly measure any correlation with engagement, team satisfaction, self-reported innovation or team retention. These are critical next-steps to understanding the business value of inclusive and representative teams.

The Shannon Index Methodology to Evaluate Workplace Diversity

The Shannon Index is a framework used in biodiversity to measure the relative variety of species and subspecies in an ecosystem. In biodiversity, there is a need to compare two separate environments and see what is most densely populated with different flora or fauna. In doing so, researchers are able to showcase how assortments differ between those two (or more) environments and where you are more likely to see a greater representation of subjects. Explaining the Shannon Index, Radek Dušek and Renata Popelková from the University of Ostrava, Faculty of Science, Department of Physical Geography and Geoecology state, “Shannon’s diversity index is frequently used in the determination of landscape diversity. Its indisputable advantage is a possibility to obtain numeric values that can subsequently be easily compared.” (Dušek & Popelková, 2012).

This is compelling because in corporate environments, diversity is usually viewed on a “binary scale” or one that is difficult to compare across a range of demographics. In the graphic below (Figure 1), we see how the Shannon Index score varies in an environment with three different subspecies of trees. The more equal the representation across those three subspecies, the higher the Shannon Index score. Whereas those environments with little representation of one or more of those species tends to have lower scores:

The first and third sets of trees differ in their breakdown of tree type. Because they both are populated by 50% of one tree and 25% for each other tree, they hold the same Shannon Index score. The value of the Shannon Index score is even greater when measuring diversity in environments with more than one species. This provides a great taxonomy for viewing environments with more than two binary demographics.

The figure below (Figure 2) demonstrates how this methodology can be applied across two frames: trees and animal types. Each grouping gets a distinct Shannon score which can be aggregated (assuming equal weights as is the case in this example):

This graphic clearly depict that the more diversity across the board in multiple binary demographics, the higher index score for each group. Consider the first example in Figure 2, which shows the combined Shannon Index score is highest when equal distribution of the species is present. However, in the subsequent example, even though the plant demographic displays equal distribution, the animal demographic distribution brings the combined Shannon index score down a substantial amount.

Using this method in a corporate diversity framework gives the ability to analyze and compare the relative diversity of teams; however, instead of analyzing tree or animal demographics, the scales are gender, ethnicity, age and region (to account for locational representation in a global environment). In doing so, the analysis can give a singular quotient, which can be used for regression analysis against positive outcome metrics such as team retention, team performance, and innovation.

As a final example, using the corporate taxonomy outlined above, consider three teams as follows:

It can be difficult to determine which of these teams is the “most diverse” across the entire spectrum of considered variables; however, the list above shows the Shannon Index score for each component of those three teams alongside the aggregation. This aggregated score makes it clear which of the teams is most and least diverse; the singular metric gained from the aggregation lends itself to greater analysis.

In this study, the focus was creating a Shannon Index score for each component variable in the corporate diversity model. Each component score was appropriately weighted, to create an indexed aggregate score. For example, gender — in this analysis — is weighted at 70% whereas age, tenure, and region are each only weighted at 10%. The table below shows the relative weightings of each component variable in the aggregate Shannon Score presented in this particular model. To determine those optimal weights, a regression analysis was done between the component Shannon Index score and positive outcomes in team retention. After evaluating each distinctly and alongside corporate goals in diversity, equity, and inclusion, the weighting for this model was determined.

The final analysis took teams aggregate Shannon scores and correlated them with positive outcomes using OLS (Ordinary Least Squares) regression modeling. The dependent variables used in this analysis are:

• Retention — based on 18 months of turnover data
• Sense of belonging, engagement, and innovation — self reported by teams assessed through quarterly pulse surveys

The results from that analysis are shared in the subsequent section.

Insights from the Shannon Methodology on Individual Team Retention, Engagement & Belonging

Using the methodology discussed in the prior section, the Shannon scores for disparate teams at a midsized organization will be correlated with those same team’s retention over an 18-month period as well as the team members’ reported sense of belonging and engagement.

To get a better understanding of outcome differences in teams and how that effects the areas mentioned above, the analysis includes results among the top 25th, middle 50th, and bottom 25th percentile of teams based on their Shannon Index scores.

The results demonstrate a statistically significant relationship between more diverse teams and a higher reported sense of belonging. More diverse teams also have less attrition and have a demonstrable relationship with retention.

The chart below (Figure 3) shows the composite Shannon Index scores (using the methodology stated in the previous section) for teams. Of course, higher scores for higher performing teams is expected given this methodology, but it is insightful to see the marginal differences in aggregate score between teams in the bottom 75th percentile in terms of diversity compared to those in the top 25th. As mentioned, these same groupings were then correlated with team attrition (over an 18-month window to account for volatility in the labor market) as well as a self-reported sense of belonging assessed via survey.

Avg. Shannon Index Score — (25–50–25 Percentile group)

Figure 3

Avg. Annual Vol. Attrition — (25–50–25 Percentile group)

Figure 4

In regards to attrition, the bottom 25th percentile teams have experienced higher volatility in comparison to those in the top 75th percentiles. That said, there is no statistically significant difference between those in the top 25th and middle 50th percentiles when analyzing team retention.

Sense of belonging score (pulse survey)

Figure 5

Every quarter, a sense of belonging is assessed via pulse survey and these results are trended over time. This sense of belonging is critical to teams because is it a key driver in retention (assessed via survey correlation coefficients). This is why a positive sense of belonging is so critical to teams. The bottom 25th percentile of teams in this analysis has 76% belonging. This is a 5% drop between the least diverse teams within this company versus the top 25th percentile. While this may seem marginal, this difference is statistically significant and equates to a considerable difference in cultural experience within the organization and the impact on overall team engagement.

It is reasonable to assume that these metrics do differ within more local functional areas within the organization. Expanding the above analysis beyond the overall company requires looking at the same set of data but through functional area. The functions include:

• G&A (Finance, HR, IT, Legal)
• Engineering (Products and Technology)
• GTM (Customer Excellence, Marketing, Sales and Services)

Avg. Annual Vol. Attrition — by functional area (25–50–25 Percentile group)

Figure 6

Sense of Belonging (Pulse Survey) — by functional area (25–50–25 Percentile group)

Figure 7

By functional area (Figure 6), it is evident that retention likelihood is most prevalent in the most diverse G&A teams and is not as significant for Engineering or Go-to-Market teams. Unsurprisingly, G&A teams are also the groups in software organizations more likely to have greater diversity (from a gender and immediate candidate availability perspective). Over the last 18 months, some attrition values can be skewed do the competitiveness of labor markets, particularly in India and APJ, this also negatively impacts the results for teams that have higher Shannon indexed scores because they are heavily leveraged across the globe.

That said, the sense of belonging results (Figure 7) shows a positive correlation across all functions except for G&A, where sense of belonging is fairly equivalent regardless of the diversity score of each team and there is no statistically significant difference. However, the Engineering and Go-to-Market teams most diverse teams have a greater sense of belonging than any teams in G&A. As previously stated, a positive sense of belonging does have a positive correlation with engagement and retention likelihood. Actual retention outcomes did not see a discernible difference for the Engineering or Go-to-Market teams; however, these are the groups that have experienced some of the greater volatility during the 18-months prior to this study which may have confounded the intent to stay from the actuality.

Concluding thoughts

Diversity in the workplace is a necessity for every company at each step of their employees’ journey. From recruiting pipelines and initiatives, to ERGs (employee resource groups), to employee retainment, to open conversations about career development and opportunities to be innovative in their work environment. This analysis provides an additional framework to analyze the quantitative, or perceived, impact of diversity on team and organization performance metrics. The confined analysis in a singular organization indicates the most diverse teams experience greater retention for most groups and express a higher sense of team belonging (which can drive engagement). The diversity model in this analysis includes several demographics that take a more extensive look at who individuals are, other than just race and gender. These include age, tenure, and regional breakdown in addition to the standard ways’ companies look at diversity.

The data has imperfections and is limited by the depth in which the study could go since it’s at the aggregate level. Given the volatility in labor market conditions over the 18 months prior to this analysis, some of the results are likely confounded. Regardless, this is valuable when looking at assembling high performing and highly retained teams. While this data is not causal, it is relevant and shows positive relationships between aggregate, holistic diversity and positive business outcomes. It can also spark interest into additional studies and analyses using a similar methodology or with a larger, cross-industry dataset. A good follow up analysis would be to look at the impact of these same diversity scores when it comes to quota-attainment for quota carrying teams in sales functions.

Workplace diversity is at the forefront of many discussions in the corporate world and with good reason. The story the data tells through this analysis only adds to the argument that more diverse teams and companies may have a competitive advantage for retention and engagement and will certainly be better poised to address the increasingly diversity-focused corporate landscape.