As a people or Diversity and Inclusion (D&I) Leader, you want to promote an inclusive culture in the workplace by fostering an environment of professionalism, respect for personal differences, open communication, and neutral and objective criteria to avoid subjective employment decisions based on personal stereotypes or hidden biases. Not only is it an urgent and necessary social issue, but data also shows that diverse businesses continuously outperform their less diverse competitors.
Almost all organizations want to determine if they're doing a good job in terms of diversifying their workforce. Measuring inclusion can be more challenging. How do you measure how supportive your culture is, and whether your environment is one where everyone feels valued, represented, and equal?
As an HR professional, creating reports and audits that measure diversity or surface bias can be very challenging, because disparate data is often hard to source, and at the same time, analyzing data from multiple sources is a complex problem to solve. In most cases, data must be shared with an external vendor and this comes with high risks and costs.
Introducing Diversity Analytics in Oracle Fusion HCM Analytics: a solution for diversity based on machine learning, driven by a combination of parametric and non-parametric hypothesis testing for classification and scoring that continuously monitors your current employment practices such as hires, terminations, promotions, and salary by gender, ethnicity, and age to ensure that you can proactively take preventive action as required at the right granularity. Diversity Analytics feeds on data provided from HCM (Human Capital Management) Fusion and Recruitment, the data remains within your ADW (Oracle Autonomous Data Warehouse), and you can obtain statistical evidence that indicates bias or not via data visualizations in OAC (Oracle Analytics Cloud).
At 9% selection rate, only 20 Black or African American persons out of 200 are selected whereas for Caucasians, the selection rate is 20%. Statistical tests such as z test, Chi-square test, and Fisher's probability test can help uncover the reason and provide statistical evidence.
Our null hypothesis is “Employment decisions are non-discriminatory, or Selection Rates are Equal” and when we test our hypothesis with data, we ideally want to see a low standard deviation to validate our hypothesis.
Referring to the image above, the Chi-square test and z test clearly indicate that there's a major deviation in the selection of not only Black or African Americans but also Asians as compared to Caucasians. Look at how dispersed the data is in relation to the mean.
Because our hypothesis has been rejected, let's drill down further and understand how deep the bias flow within the organization. The detailed report provides sufficient evidence that the bias has existed over years.
We can extend our analysis with payroll data to find statistical evidence that salaries of Asians and Black or African Americans are also much below the median. This further weakens our initial null hypothesis. The detailed report provides us with enough evidence that the bias has existed over years. We can investigate further if needed, and understand the depth and breadth of this bias flow through the organization.
Using salary for comparing the compensation of employees within the same reporting establishment, we can further analyze the data to understand the selection rates and standard deviations for various ethnicities across multiple years. In this example, it's clearly established using various statistical attributes (such as z-test and Fischer’s exact probability) that the standard deviation for ethnic groups such as African American and Asians are high in comparison with the other ethnicities during a specific year (for example, 2015).
On further analysis, it's observed that for a specific job category of engineers, the standard deviations are high for the same ethnic groups. This substantiates the argument that there are potential indicators of biases against specific ethnic groups in the organization. These analysis act as the guiding principles for the D&I leaders, people leaders, and organizations in developing fair practices internally and monitoring them at regular intervals.
Similar examples of comparing the selection rates and standard deviations can be applied across various ethnicities, age groups, and gender for HR events such as hiring, promotions, and terminations.
Whether you're a senior executive, people or D&I Leader, or an HR consultant, all employment practices appear neutral. However, considering the far-reaching impact of bias in an organization and understanding its origins and how to address those enables organizations to not just better serve customers but also employees.
Tune into this podcast on how to productize D&I as a framework.