Defining Your Diversity and Inclusion Goals - and Achieving Them Through Analytics

Print this Article | Send to Colleague

Sarah Babineau, MHR, PHR, SHRM-CP is the Managing Partner of Compass Metrics, LLC, a certified woman-owned, disability-owned business enterprise. Compass Metrics is the only consultancy in the U.S. fluent in both Diversity Analytics and Affirmative Action.

Diversity Analytics

The first step in solving a problem is to ask ourselves: What, exactly, is the problem we are trying to solve? Many diversity initiatives fall short because this question is not explored in sufficient depth prior to implementing a solution. Having an accurate problem statement ensures that interventions we design have a better chance of working the way we intend them to. According to Iris Bohnet in What Works, "I cannot overstate the importance of testing and measuring."[i]

Without some evidence that we are working from the root cause of the problem, any intervention, no matter how well-intentioned, stands a fair chance of failing. Not because the intervention is not valuable, but because it wasn’t designed to address the true nature of the problem. When we implement interventions that don’t get the results we wanted, we lose credibility as problem solvers. And there is only so much time, money and energy stakeholders will invest before they expect a positive and dramatic return. In effect, getting the problem right is critical to designing the right solution. To do that, we have two types of analytics we can use before recommending changes: lag analytics and lead analytics.

The difference in lag and lead analytics can be compared to the difference in the roles of doctor and coroner. Both have similar training and goals but different expectation of outcomes. The coroner, lag analytics, performs an examination to discover what went wrong in the past. No amount of good work from the coroner will change the outcome for the patient.

The doctor, lead analytics, uses indicators to predict what problems might occur and what might be done to avoid them. There is no absolute guarantee that the problem will occur without the intervention, but data from both the coroner and the doctor can be reviewed together to assess the likelihood of a negative outcome without intervention compared to the likelihood with interventions. Using multiple data sets, multiple analyses and multiple perspectives allows medical professionals to design protocols that have a greater chance of being effective than if they only used lag analytics.

Lead analytics are also likely to show signs of change faster than lag analytics. And, as it is with human patients, demonstrating measurable change is critical to gaining support for making sure the changes are implemented. Making progress towards goals motivates employees to become and stay engaged in the process of change.

Lag Analytics

Many HR departments already conduct lag analytics workforce trends like representation, hiring, turnover, attrition, promotions and transfers. Some also do comparative analyses of these trends on the basis of immutable characteristics such as gender, race/ethnicity, and status as a person with a disability or protected veteran. Most large U.S. law firms have also incorporated measures of how lesbian, gay, bisexual and transgender (LGBT) attorneys fair in these activities compared to their colleagues. Lag analytics are precise, historical and based in fact. They are also slow to change and do not afford the opportunity to circumvent a problem, since in order to be captured, the problem has already transpired.

Lag analytics are useful in narrowing down the areas in which one could search for the root cause of the problem, but they don’t provide enough information to guide us to solutions that are definitively related to the problem. For example, the People Operations (Human Resources) team at Google discovered that women had a higher turnover rate than men. They could have stopped the analysis there and instituted workshops for women, training for managers, or launched a new employee resource group. Instead they analyzed the data more rigorously and found that turnover of talented women was more closely correlated to employees with young children than it was to gender. What appeared to be a gender problem, was actually a problem for parents. Google used those findings to redesign procedures that had the immediate effect of reducing turnover of parents of young children to the same levels as their peers.[ii]

Extrapolating from that example, an organization could follow up with analyses of caregiver status to determine if retention is an issue for all employees with caregiver responsibilities, including those who care for older children or adults. Had the data showed a positive correlation, new procedures and policies could be developed to benefit all caregivers in the workforce. The deeper data analysis would then result in inclusive interventions with a greater return on investment.

Lead Analytics

Lead analytics are predictive, like weather forecasts. By nature, they are more subjective than lag analytics because the outcome that supports the hypothesis is the one we’re trying to change. Lead analytics offer us the opportunity to change potentially undesirable outcomes before they become facts.

In another example from Google, an analysis of key performance indicators (KPIs) that lead to career success found that an employee’s technical prowess was not the best predictor of future success. Instead, high proficiency in soft skills – communication, empathy, ability to collaborate and judgement, for example – was a much stronger predictor of career success. The organizational development and training teams were then able to direct resources to programs focused on increasing proficiency in these KPIs rather than increasing focus on technical skills.[iii]

An "opportunity analysis" is another example. Qualitative and quantitative data are reviewed for KPIs, patterns or trends that historically result in success. Access to these KPIs is then measured in employees from historically disenfranchised groups and compared to their "non-diverse" peers (creating a control group). Should the analysis determine that the KPIs are not what were previously thought, or that employees do not have equal access to them, interventions can be designed that have a high probability of distributing opportunities more evenly.

For example, in law firms with lock-step, up-or-out promotion processes, one of the strongest KPIs is billable hours. Lawyers who do not meet their target hours for more than a year or two are more likely to depart the firm. If lawyers from historically disenfranchised groups are not getting as many opportunities to bill hours, they will not be likely to stay with their firms. Over time, firms that do not take steps to ensure equal access to opportunities to develop these KPIs will become less diverse, particularly at the senior associate and partnership levels. This could be part of the reason that law firms are lately being publically called out for having shown little progress in advancing attorneys from historically disenfranchised backgrounds.[iv]

This trend is starting to result in firms inadvertently alienating corporate clients that insist on a diverse slate of experienced attorneys staffed on their matters.[v]While progress has been made in advancing women and attorneys of color in recent years, we have seen that, in U.S. law firms, diverse attorneys are more likely to drop out of law firms around the mid-career level than are their similarly situated counterparts.[vi]^,[vii]As only one example, the legal industry demonstrates the importance of being able to forecast outcomes for diverse talent. KPIs will vary by industry, company and even position requiring a dedicated effort to collect the right data to ensure analytics provide useful insights.

Back to NEHRA News