Lorem ipsum dolor sit amet, consectetur adipiscing elit. Vivamus convallis sem tellus, vitae egestas felis vestibule ut.

Error message details.

Reuse Permissions

Request permission to republish or redistribute SHRM content and materials.

Common Reasons for Flagged Pay Differences

Statistical analysis can reveal pay equity risks

A woman in glasses is looking at a piece of paper at her desk.

Some employers have a practice of periodically conducting statistical analyses of employee compensation, under attorney-client privilege, to identify potential areas of risk related to pay equity concerns. These analyses are usually focused on gender and race or national origin.

Through these statistical analyses, employees are placed into comparator groups and the compensation of the employees in those groups is analyzed for differences that remain after controlling for relevant factors available in the data set. Such factors may include data points such as job title, tenure with the company, time in position, location/market, and performance ratings.

Once the analysis has controlled for the available factors, the statistical model will flag those comparator groups where the pay differences are statistically significant and adverse to a particular demographic.

The analysis work does not end there. Further work may be needed to determine whether pay differences of those within flagged groups can be explained by legitimate, nondiscriminatory factors—either within the data set or outside the data set. This is critical because the statistical models can only control for data that is available within the data set.

There are many factors that may legitimately explain pay differences but that cannot be used in the statistical analysis. This requires a "deeper dive" analysis of those comparator groups that have been flagged. Below are some of the common factors that may explain differences that are often encountered during this deeper-dive work.

Data Errors and Misalignments

Any analysis that relies on data is susceptible to issues caused by inaccuracies or errors in the data. When conducting deeper-dive analyses, one may find data inaccuracies, such as incorrect job titles, misaligned pay grades, or missing information. Thus, employers may want to take steps to ensure and confirm the accuracy of their data, ideally before the next pay equity analysis. This will improve the value of the statistical analysis.

Related to this issue, there may be situations where the data is correct but the employee is not properly aligned with the apparent comparators. For example, the data may report the correct job title that the employee holds, but that job title does not accurately reflect the employee's job responsibilities—which may be at a higher level or lower level than others holding the same job title. This may result in one or more employees being inaccurately flagged as underpaid or overpaid.

Prior Work Experience and Education

Comparators may have pay differentials that are legitimately based on prior, relevant work experience or education. These are often two key factors that determine an employee's starting pay with an employer and they may legitimately explain pay differences as an employee progresses with the company. Yet, employers may not have this information available in their human resource information systems to use for statistical analysis of compensation.

Some employers may have this information available but may not consider it to be reliable because of how it is entered into the system. For example, some companies rely on employees to input and update their education information. Since this information is not entered with a consistent and reliable process, its accuracy cannot be assured and therefore cannot be used.

Mergers and Acquisitions

Employers that have been involved in recent or frequent mergers with and acquisitions of other employers often find that the merger and/or acquisition alone may result in otherwise unexplained pay differences. This is due to the many challenges of fully integrating a new group of employees into an existing workforce—considering the many potential differences in organizational structure, job titles, job duties and, of course, compensation.

Even if an employer is focused on trying to reconcile these many differences as quickly as possible, there will be some delay between the time of the transaction and the time when all employees are fully integrated.

Of course, a company that has just merged with or acquired another company has many issues to address in closing on the transaction and otherwise integrating the businesses, and integrating and aligning employee compensation is just one such factor.


Statistical analysis alone may not account for all legitimate factors that impact pay. So, the deeper dive is an important element of the analysis that helps to distinguish those situations where pay differences may be justified by legitimate factors and those where they are not justified—which enables employers to focus remediation efforts on making adjustments to the pay of those employees whose underpayment cannot be explained.

In addition, the deeper dive is often the phase of a proactive pay equity analysis that can help employers uncover the root causes behind adverse statistical findings and behind those pay differences that cannot be explained. This information is critical to helping employers identify and implement changes or practices that will make lasting improvement in their pay equity efforts.

Sarah J. Platt is a shareholder at law firm Ogletree Deakins in Milwaukee and co-chair of the firm's Pay Equity Practice Group. Liz S. Washko is a s shareholder in the Nashville office and a managing director for the firm, and a co-chair of Pay Equity Practice Group. © 2022, Ogletree, Deakins, Nash, Smoak & Stewart, P.C. All rights reserved. Republished with permission


​An organization run by AI is not a futuristic concept. Such technology is already a part of many workplaces and will continue to shape the labor market and HR. Here's how employers and employees can successfully manage generative AI and other AI-powered systems.