The Work to Make Government a DEI Model


The staff that drafted President Biden’s executive order on diversity, equity and inclusion should feel good about the coverage and understanding of what’s needed. At 12 pages it’s way too bureaucratic, but it addresses the important points. It’s a solid beginning.

The EO commits the administration to making federal employment fair and open to everyone. It recognizes that bias and discrimination can occur at each stage in a career, requiring agencies to develop strategies to eliminate barriers to equity during virtually all processes: “recruitment; hiring; background investigation; promotion; retention; performance evaluations and awards; professional development programs; mentoring programs or sponsorship initiatives; internship, fellowship, and apprenticeship programs; etc.”  

It’s not stated but implied that agencies should plan to assess the fairness of their practices at each stage. Toward that end, agencies are required to adopt an “evidence-based and data-driven approach” to employee management. This will be a challenge—government’s human capital practices have never been subject to rigorous evaluation in the past.

The EO directs agencies to develop standards based on leading policies and practices in the public and private sectors. Working with businesses to understand those practices and develop strategies to make them effective in government will take time. The EO mandate is expansive, including women, “communities of color,” employees with disabilities and LGBTQ+ employees. Assessing how human capital practices have affected each of the groups and then implementing corrective actions is a major undertaking. The impact of the changes, if successful, will change the work experience in every agency.

The Work Ahead

The EO recognizes that agencies may not have adequate HR resources available to support the goals. Resources here include technology, reliable employee data, and staff skills. Simply defining the necessary resources will take a month or two. Developing the necessary databases is likely to take longer than anticipated.

In a recent report on gender pay differences, the Government Accountability Office highlighted the problem of missing or incomplete employee data. The auditors noted, for example, that agencies do not retain complete information on individuals who apply for promotions. They also noted some information is voluntary and not verified. The work outlined in the EO cannot be undertaken until the needed data are available.

One of the first steps for agencies should be to create a position for a chief diversity officer, as recommended by the EO, although that title makes the position sound like an HR role. It should not be optional. The initial incumbents will need authority and broad support to succeed. Seniority (the word used in the EO) is not the issue. The incumbent needs the support to make change happen. Agency leaders will need to make it clear this position is a priority. Executives across government need to be accountable. 

Only two of the new requirements in the executive order are relatively straightforward (promote paid internships, and build and strengthen partnerships to facilitate recruiting).

The partnerships, both new and old, should help to open doors and improve outreach and recruitment of individuals from underserved communities. That suggests agencies will be expected to go far beyond posting vacancies on USAJobs. 

Lack of Data

The data problem is central to assessing recruiting experience. Unstated but relevant is the need to assess and improve government’s brand as an employer. That should be done agency by agency. It would be useful to survey college seniors, especially honor students, about their perceptions of federal agencies and federal employment.

The EO also requires agencies “to take steps to implement or increase the availability and use of diversity, equity, inclusion and accessibility training programs for employees, managers, and leadership.” That’s clearly important but agencies need to be aware that research shows that training on DEI issues can backfire. A reader comment following an earlier column on the EO illustrates the problem:

“This is nonsense and will not end well. It’s pure insanity that leadership is so ignorant. They will infect the federal workforce with the same mind virus that has destroyed liberal discourse in higher education.”

The point of course is that a segment of the workforce is likely to push back. The training must be done well or it could reinforce biases.

Human capital practices will need to be evaluated for bias and discrimination. For example, research shows that when employee performance is evaluated, raters’ comments differ significantly for men and women. Assessments will need to confirm the ratings are not discriminatory. The same is true for the handling of promotions. 

Agencies should consider creating teams of managers and/or employees, depending on the level of the decision making to guide the projects. That’s a widely used practice in higher education and healthcare. Employees need to live with the new practices and their involvement will help to build acceptance. That’s key to successful change.

Pay Equity is More Complex

Somehow the pay comparisons always get a lot of attention. It’s interesting that the EO focused narrowly on pay equity. It was silent on whether federal salaries are market competitive. The annual pay gap review does not disclose starting salaries for new grads.

There is an argument that the new graduates from underserved communities, who may have made sacrifices to get through college, should not be asked to accept a noncompetitive salary. Although peripheral, its suggested agencies and their employees should be aware of how starting salaries compare with market levels. The phrase “pay fairly” has also come into wide use.

The GAO study confirmed that men and women are paid differently—or more correctly stated, that their analyses were unable to fully explain the differences in pay. The statistical estimates should not be interpreted as evidence of discrimination and GAO was careful to state the gaps are unexplained.

Significantly, GAO used four somewhat different statistical models, each with different variables and each produced somewhat different answers. That’s important because it shows there is no “valid” model or set of variables. Pay equity is normally limited to jobs “substantially equal in skill, effort, responsibility and working conditions.” The GAO analysis was broader. Planning the analyses will be a core question.

Although it’s not always acknowledged, pay equity analyses narrowly focused on occupations would generate a new equation each time. That is to say, the statistical results based, for example, on accounting jobs would not be the same as one based on engineering jobs. The results would also be different if the jobs and incumbents are in the GS 12-15 range compared to employees in the GS 7-11 range. Regression analyses simply summarize the relationships between data points; they do not explain the statistics. 

GAO relied on a model and variables that was first introduced by comparable worth advocates in the 1970s. Two variables—education and age—are somewhat questionable. Age is essentially a proxy for years of experience. Yes, older workers generally have higher salaries. Education is similar. Employers make salary decisions based on college majors, the quality of the academic program and for recent grads, an applicant’s GPA. Not surprisingly higher levels of education are associated with higher salaries. The statistics will need to be carefully interpreted. Neither is a measure of employee value.

Agencies should plan similar analyses to understand the differences in cash awards, promotions, performance ratings, and the reclassification of jobs. It could, for example, be that male dominated jobs are reclassified higher more often than jobs held by women. It’s those…



Read More:The Work to Make Government a DEI Model