Introduction

Employers define human resources policies and practices and make employment decisions that, while based on legitimate business decisions, may have a disparate effect on different groups of employees. A hypothetical example of selections for a pro­posed reduction in force (RIF) illustrates how an evaluation can reveal a disparate impact of a protected class group. Unveiling potential disparities before the imple­mentation of the proposed RIF allows management to revise the goals, objectives, and the planning of the selections for termination, correcting a potential disparate effect of protected class groups, and minimizing the likelihood of legal disputes.

Employers, be it a company, a nonprofit organization, or a govern­ment agency, make employment decisions on a regular basis based on a broad variety of possible business reasons that may range from market changes to internal synergies and dynamics. Selection-based decisions, such as hiring, compensation changes, job assignment, pro­motions, disciplinary actions, and the like, based on sound and legiti­mate business reasons may lead to disparate treatment claims if they affect groups of employees differently. Evaluating whether employment decisions from the prospective of potential disparate treatment may seem like a straightforward topic. Reality, however, is quite complex, and the effects of employment decisions may be difficult to navigate. In many instances, sound and legitimate business decisions may result in potentially lengthy, complex, and expensive litigation because they may have adverse effects on different groups of employees.

One possible alternative is to evaluate the neutrality of an employ­ment practice with respect to protected class groups before the decision is implemented. This can allow the employer to evaluate the decision process, which in many instances can bring about the opportunity to introduce changes to ensure that the decision-making process is “neu­tral” with respect to groups of employees. Testing for potential adverse effects before certain employment decisions are made, allows the organization to achieve two goals: (1) reveal whether the employment decisions would have an adverse impact on protected or non-protected class groups; and (2) correct the potential adverse effect and potential disparities affecting class groups, if such disparities can be avoided.

What Defines a "Protected Class"

Title VII of the Civil Rights Act of 1964 is the most commonly applied federal anti-discrimination law, and it prohibits employers from discrimi­nating against employees or job applicants on the basis of their race, color, national origin, gender, and religion. An earlier legislation, the Equal Pay Act of 1963, protects men and women who perform simi­lar work for the same establishment from wage-based discrimination. Women and minorities are considered members of a “protected class” based on these federal laws. However, the definition of “protected class” has evolved over the years with the passing of further legislation, such as the Age Discrimination in Employment Act of 1967 and the Americans with Disabilities Act of 1990. Generally speaking, women, minorities, people over the age of 40, and people with disabilities fall into the category of “protected class” for employment discrimination purposes.

Federal anti-discrimination laws protect employees and job applicants from employment-based discrimination, and the Civil Rights Act of 1991, among other things, provides monetary damages for intentional employ­ment discrimination.1 Under the provision of federal legislation, employ­ment practices such as hiring and firing, compensation, job assignment, promotions, recruitment, training, eligibility for fringe benefits, and many other terms of employment, are to be “neutral” with respect to “protected class” status. Or in other words, employment decisions are not to be conducted on the basis of an employee’s “protected class” status.

It is important to note that, while federal anti-discrimination legisla­tion defines protected class groups, its intent is to protect all employees from discrimination. Reverse discrimination, i.e., employment actions that adversely affect employees and job applicants who are not in a protected class is also not allowable under federal anti-discrimination legislation.

Selection-Based Employment Decisions

Many of the employment decisions that are implemented by organiza­tions are selection-based. Hiring and termination practices are the most obvious selection processes that organizations initiate. Other selection-based decisions, however, are not as obvious. Employment practices that may involve a selection process could include job assignment, pro­motions, disciplinary actions, demotions, use of company resources and facilities, access to training, and the determination of merit increases or assignments to salary bands, among others.

Generally speaking, any employment practices that may affect an employee’s status and future opportunities within an organization can be evaluated from the prospective of a selection process.

Case Study: Potential Disparate Impact in a Hypothetical Case of RIF

Consider the following hypothetical scenario:

Under severe economic pressure, XYZ Corporation plans a reduction in force (RIF) to enhance productivity and eliminate duplicative positions. After an extensive review of employee performance, roles and responsibilities, and compensation, managers and human resource professionals determine that certain projects will be cancelled. As a result, two divisions will be targeted for layoffs and the staff assigned to the cancelled project will be the ones terminated. From the reorganization prospective, this initiative makes good business sense.

XYZ Corporation employs 550 people, 200 women and 350 men. The two divisions that would be targeted for layoffs employs 350 people, 100 women and 250 men. Women represent roughly 36 percent of the company’s workforce, and roughly 29 percent of the employees in the targeted divisions combined. The decision makers of XYZ Corporation determine that in order to meet the financial and reorganization goals, 85 employees will be laid off from Division A and Division B, 35 women and 50 men.

While the decision to initiate a RIF may be based on legitimate business reasons, one of the questions to be answered is whether or not the selection process may impact men and women in the targeted divisions proportionately to their representation in the workforce. If the selection process results in a disproportionate number of women or men being terminated, XYZ Corporation may be made susceptible to a disparate treatment lawsuit by either group. A RIF that was initiated by financial and market pressures may result in legal expenses that could completely offset the cost savings originally intended by the reorganization.

Claims of employment discrimination can be brought about through allegations of either “disparate treatment,” when an employee or group of employees may allege that they were treated differently from the rest due to their protected class status, or “disparate impact,” where an employee or group of employees may allege that a policy or practice impacted them differently as member(s) of a protected class. However, with just a small amount of insight and some additional planning, the likelihood of a lawsuit could be prevented, minimized, or, at least, foreseen.

The first question to be answered in this example is whether the RIF selections would actually adversely impact women or men, and whether or not the selection process can be considered neutral with respect to gender. If it is determined that the selection process would adversely impact women or men as members of a protected class or a non-protected class, the second question to be answered is whether XYZ Corporation could introduce any changes to the reorganization plans before the lay-offs are announced to ensure that the outcome is neutral with respect to gender.

What a Statistical Test Revealed

Several statistical tests and econometrics techniques are available and are generally accepted in conducting analysis of selection-based employment decisions.2 The selection of a particular statistical test (such as chi-squared, which is applied in the hypothetical example discussed), or a more sophisticated econometrics model involving regression analy­sis, depends both on the complexity of the data available, as well as on the complexity of the employment decisions being analyzed, among other factors.

For purposes of simplification, the illustrative exercise presented through the above hypothetical scenario utilizes the principles of the chi-squared test. Once a chi-squared statistic is computed, a probability table is consulted to determine whether or not any disparities may be associated with chance occurrence. Generally, a chi-squared statistic associated with a probability value of 5 percent or less is considered to be statistically significant, and potentially indicative of a disparity that did not occur by chance, all else being equal.

Table 1 reports simple counts on the workforce information for our hypothetical XYZ Corporation. As indicated earlier, of the 550 staff members of XYZ Corporation, 350 are employed in Divisions A and B. The 200 employed by XYZ Corporation represent roughly 36 percent of the company’s workforce, while the 350 men represent 64 percent of the company’s workforce. Division A employs 120 men and 80 women. The gender representation in Division A is 60 percent and 40 percent, respectively. Division B employs 150 staff, 130 men and 20 women. The gender representation in Division B is 87 percent and 13 percent, respectively.

Table 2 presents the results of the proposed RIF affecting Division A and Division B combined. According to the information in Table 2, 11 more women would be terminated than expected based on their repre­sentation in the combined workforce of the two divisions analyzed. This results in a chi-squared test statistic of 8.74, which is associated with a probability of less than one percent of occurring by chance, and consid­ered statistically significant. This implies that the total number of women selected for lay-off from Division A and Division B combined is signifi­cantly higher than what would be expected in a selection process that is neutral with respect to gender. It could be argued that the selection process for purposes of this RIF has a disparate impact on women as a protected class, and could lead to claims of disparate treatment.

Table 3 reports the results of the analysis when Division A and Division B are analyzed separately. The purpose of an evaluation of the proposed layoffs in each division separately is two-fold: (1) to isolate the potential problem areas in the selection process; and (2) to evaluate sim­ilarly situated employees. The results presented on Table 3 indicate that while the proposed layoff selection affecting Division A does not result is gender-based disparities that are statistically significant, the story for Division B is slightly different. It appears that the proposed layoffs for Division B would result in a larger than expected number of women being laid off from this division, and that such disparity is statistically significant. If the proposed layoff is implemented, XYZ Corporation may be faced with a gender-based dispute on claims of discrimination.

An Alternative Proposal for XYZ Corporation

Now that YXZ Corporation has conducted an analysis of the pro­posed layoffs before they happen, it can evaluate its policy and the potential exposure to an employment discrimination dispute. Under these circumstances, the organization can evaluate the business reasons for the selections that are proposed, or can attempt to introduce changes to the selection process in order to ensure that the selection process is neutral with respect to gender.

In this hypothetical example, the management of XYZ Corporation concludes that the qualifications, performance, and experience of employees in Division A and Division B are relatively similar, and there­fore the financial and reorganizational goals of the RIF can be achieved through the reassignment of staff to the different projects in addition to the RIF. Although relatively time consuming, management evaluates the possibility of reassigning staff to the different projects conducted by Division A and Division B. Once this evaluation is completed and before it is implemented, management proceeds to re-analyze the RIF selections. Again, the goal of the analysis is to determine whether the business reasons leading to the cancellation of certain projects would have an adverse effect on women as a protected class group.

The results of such an analysis are reported on Table 4. The chi-squared statistics computed for Division A and Division B presented on Table 4 are associated with probability values that are higher than 30 percent. These results are interpreted to mean that the difference in the number of women or men who would be terminated through this proposed RIF is not statistically different from what would be expected in a process that is neutral with respect to gender.

While not a simple process, and one that required time and dedica­tion and additional up-front costs, the result is slight modifications to the originally proposed RIF selections that allow the organization to pro­ceed with its consolidation and its financial and reorganizational goals through a RIF process that is neutral with respect to gender.

Understanding Protected Class Groups

The hypothetical example discussed above considers only one of the several dimensions of protected class classification: gender. In reality, evaluating employment decisions from the prospective of potential dis­parate treatment claims typically involves giving consideration not only to the gender dimension of the workforce composition, but also to the race, national origin, age, and possible disability status representation in the workforce.

Taking into consideration all the possible dimensions of protected class groups may require considerable effort—as in many instances, protected class groups overlap. For example, a significant percentage of the men in Division A are over the age of 40, and a smaller per­centage of them were not born in the United States. The overlap of protected class groups may be a more common occurrence, one that renders the analysis of disparate impact significantly more challenging. The more complex the workforce composition is, the more important it becomes to evaluate potential disparate impact on protected class groups.

Understanding the composition of the workforce of the entire organi­zation or establishment, as well as of the portion of the organization that may be targeted by a particular set of employment decisions is not only the first step in the process, but a key one. Without an understanding of the workforce composition, an analysis of potential disparate impact would be faulty at best.

Charges of Disparate Treatment May Be Preventable

Situations that may potentially lead to lengthy and costly employ­ment disputes can often be preventable, or, at a minimum, could be anticipated. The hypothetical scenario just described is a much simpli­fied version of what reality can be like for many organizations engag­ing in similar processes, but it illustrates the point of how a review of the employment decision from the prospective of potential exposure to discrimination claims ahead of its implementation may alter the likeli­hood of a legal dispute. Just like our fictitious XYZ Corporation, real companies can engage in evaluations when policies and practices are being designed and before they are implemented.

The range of employment policies, practices, and decisions that may potentially lead to legal disputes is broad, and sometimes difficult to identify and anticipate. A RIF such as the one discussed in the hypo­thetical example above can be more easily identified and isolated, as it is not a regular occurrence in most organizations. However, employ­ment decisions that are more frequent and that may result in adverse impact may not be evaluated, as their potential effect on different groups of employees may not be as evident. An example of a more dif­ficult employment set of decisions that may lead to claims of disparate treatment and that may not be so evident is the availability of training opportunities. Organizations may sometimes provide certain training opportunities because they see them as beneficial to their ultimate business goals. However, if the representation of a particular group of employees in the segment of the workforce that may benefit from such training is significantly different from their representation in the overall organization’s workforce, claims of discrimination may arise, particularly if additional training ultimately affects employees’ future compensation and future prospects within the organization.

Employment practices ranging from hiring to layoffs, to the deter­mination of merit increases, to the assignments to salary bands, sales territories, and the like, can all be evaluated from the prospective of the disparate impact of protected classes of employees. Such evaluations can be conducted by organizations as self-imposed employment prac­tices audits, which can assist the organization in identifying potential problem areas, and allow for corrections if needed, or to be prepared for potential legal disputes.

The Role of Counsel

The advisory role of counsel begins with raising the awareness of the organization towards the potential presence of difficult to detect disparities, and plays a central role in the process of navigating employment decisions such as the one in the hypothetical illustrative example discussed above. Legal counsel can equip management with recommendations in the decision-making process, and can be instrumental in initiating the involvement of a qualified professional team to evaluate the potential impact of employment decisions. Such a team of professionals, made up of legal counsel, management, and employment consultants, should work together to evaluate the business reasons leading to employment changes, and together can evaluate the presence of potential disparities and the likelihood of legal exposure for the organization.

The involvement of legal counsel could ensure that all the goals of the process are achieved. Legal counsel would ensure that federal legislation is considered in all steps of the process, that the necessary analytical consulting team is selected, that communications between the organization and the consulting team stay on target, and that the necessary steps of the process are completed in a timely fashion.

Conclusions

Organizations make employment decisions that sometimes may adversely affect different groups of employees. General awareness of potential discrimination disputes is not sufficient to avoid the presence of disparities in employment decisions. However, many employment policies, practices, and decisions could be evaluated from the prospec­tive of their potential disparate impact on groups of employees before they are implemented. Conducting such evaluations, which typically requires the involvement of counsel and the retention of an economic expert, may allow the organization to navigate the employment deci­sions, policies, or practices and assist in the determination of possible disparities before they occur.

One example of employment decisions that may have a disparate adverse effect on different groups of employees is a layoff process, also refereed to as reduction in force, or RIF. Through a hypothetical example, the above discussion illustrated how potential disparities can be identified through the use of statistical tools, and illustrated the point that the goals of a reorganization can sometimes be achieved while ensuring that the protected class composition of the targeted segments of the workforce is consistent with a process that is neutral with respect to protected class groups.

Which statistical tools are employed for the evaluation of employment decisions varies depending on several factors, ranging from budgetary constraints, to the complexity and reliability of available data, and to the complexity of the employment decision or set of decisions analyzed. While the statistical tools and techniques available vary in methodologi­cal aspects, they all accomplish a common goal: they reveal the potential effect of the particular decision or set of decisions that are implemented, and identify adverse disparities.

Notes

1US Equal Employment Opportunity Commission (EEOC).

2The scope of the hypothetical example is to illustrate what statistical analyses may reveal and what possible alternatives may be available to the organization when such analyses are conducted prior to the implementation of a set of employment decisions. While there are several different kinds of economic and statistical analyses that can be conducted to evaluate employment decisions, a discussion of such analyses is beyond the scope of the illustrative example presented and discussed in this article. What type of economic and statistical analysis is conducted depends on many factors, including, but not limited to, the size of the organization, the data collected and available to the consultant, and the financial resources available to be dedicated to the analysis.

Posted from the Employee Relations Law Journal, Vol. 36, No. 3 Winter 2010, with permission from Aspen Publishers, a Wolters Kluwer Company, New York, NY 1-800-638-8437, www.aspenpublishers.com. For more information on the use of this content, contact Wright’s Media at 1-877-652-5295.


SUCCESS IS PERSONAL Click here to learn more about our brand