Regardless of which side you represent in an employment discrimination case, getting the right personnel data can be difficult. Identifying and requesting personnel data that will help the attorney and their experts examine the allegations of employment discrimination can be an enormous task even if the defendant is your client.

In this article, we provide a primer on the types of personnel data that attorneys should collect in cases involving statistical analyses of employment discrimination allegations. The information should should be particularly helpful to employment attorneys faced with assembling a personnel database in an employment discrimination lawsuit, an EEOC administrative hearing, or a company initiated disparity ‘self-audit’.

Introduction

Regardless of which side you represent in an employment discrimination case, getting the right personnel data in these types of cases can be difficult. Identifying and requesting personnel data that will help the attorney and their experts examine the allegations of employment discrimination can be an enormous task, even if the defendant is your client.

One of the first things that many employment attorneys quickly find when attempting to collect personnel data in a discrimination case is that a great many employers do not have the personnel information that they need readily available. Instead, many find that to obtain the employee information that they need to effectively address the discrimination allegations in an administrative investigation or discrimination lawsuit, the relevant personnel data will have to be pieced together from multiple departments and human resource data systems. Depending on the employer, the complexity of the task of collecting the employment data from the employer can range from involved to colossus.

While it generally takes a significant amount of time and effort to correctly assemble the personnel data in a discrimination case, from a data analysis standpoint this is time that is well spent. Like the popular Michelin commercials that extol the importance of good tires by saying that ‘your life is riding on them’, the importance of good personnel data cannot be underestimated.

The wrong data in the best statistical model can easily and quite quickly sink a statistical discrimination analysis. In this article, we provide a primer on the data that is used to perform statistical analyses of employment discrimination. This article should be particularly helpful to employment attorneys who are faced with assembling a personnel database in an employment discrimination lawsuit, an EEOC administrative hearing or even a company initiated disparity ‘self audit’.The article begins with a description of a typical company’s personnel record keeping. In the second part of the article we provide a general discussion of the personnel records that are needed to analyze discrimination allegations concerning disparate treatment and disparate impact in a company’s application process, historical hiring record, company wide compensation and reduction in force policies.

Employer’s personnel records

For most mid-size to large employers, the responsibility of entering and updating employee records falls to the Human Resources (HR) department. The HR department may be in house, or as has become increasingly common, the functions of the HR department may be outsourced to a third party. The personnel data maintained by the Human Resources department will typically include records that show the employer’s hiring activity, employee compensation as well as employee level transactions such as job promotions.

In discrimination cases involving larger employers, the collection of personnel data will usually require the attorney to make discovery requests and/or to confer with multiple sections within the HR department and possibly multiple sections of the overall company. The process of collecting the personnel data may be more involved for employers that outsource all or part of the HR functions to third party agents. The personnel data the employment attorney will usually encounter will begenerated by sophisticated human resources database software often produced by companies such as PEOPLESOFT, ORACLE, and SAP. These HR software programs provide a menu driven user interface that allows the HR personnel to manage employee transactions such as job title, salary, and family status changes.

The employee transactions made by the HR personnel using these types of HR management software programs are typically recorded in an electronic ‘back-end database’. It is important to realize some human resource data, even in a relatively large company, may not be available in electronic form.

Using employer personnel data

Generally, transforming the personnel data into a format that will allow the attorney, or his or her experts, to examine and analyze the disparate impact or disparate treatment allegations in a case will require the attorney to do a number of related tasks. These tasks include the following:

A. Define the relevant employee universe and time period for the analysis.

From a statistical standpoint, in most situations you will typically need to consider the time period that sufficiently spans the employment discrimination allegations. In determining the sufficient time period, the analyst will need to consider both the facts surrounding the particular discrimination allegations and the actual business practices of the defendant-employer. The actual number of years used in the analysis may include only the year that the alleged discrimination occurred or it may span a number of years both before and after the alleged discrimination took place.

B. Request personnel data fields using data queries that provide select personnel information from the employer’s HR database software.

In most cases, you will not be asking the HR department to provide you with the employer’s entire HR database. Instead, most employment cases will involve asking for certain data fields from the employer’s personnel information. For instance, a typical data query will ask for information such as the employee id, the year end salary, and job location for every employee over a given time period. In order to determine what data fields are needed, the attorney should ask for a listing of available data fields. This listing is commonly referred to as a data dictionary.

C. Collect and data-enter relevant hardcopy data.

Not all personnel data will be available in electronic format. Any data entry that must be performed should be done carefully and incorporate multiple data quality checks.

Data issues in employment discrimination cases

In most employment cases involving either disparate impact or disparate treatment, the statistical analysis will ultimately involve analyzing claims of discrimination concerning the employer’s hiring, compensation, promotion, or termination record. The specific data issues that need to be addressed for each type of case are discussed in more detail in the following sections.

Applicant flow data

Discrimination cases that allege that a company discriminates against a protected employee group in a hiring case typically begins with an analysis of the company’s applicant flow data. The application flow data for a company contains the files and applications of individuals that responded to an employment opening at the company. Depending on the manner in which the employer receives its employment applications, the application data may contain individual hardcopy resumes and/or standardized electronic application data. For instance, many larger companies use a number of different recruiting sources such as on-campus recruiting, newspaper ads and trade industry magazine want-ads, computer job search databases, media ads, and word of mouth to fill job openings. Before starting the applicant flow data collection process, it is crucial to fully understand the employer’s hiring process.

As a result, the application contained in the data from one source may be somewhat standardized and include the same set of information for all applicants. Other application sources may only result in an individually prepared (and nonstandard) resume being sent in by the applicant. Since a common application form is usually required for government and public employers, the application data for these types of organizations tends to be more uniform than that of private employers.

Overall, regardless of the employer, the applicant data information will most likely be the most incomplete and varied of the employment data that the typical employment attorney will encounter in a discrimination case. It is common for companies to maintain only minimal documentation on the individuals that applied for a given job position. This practice is common partially because the large majority of the applications received by firms are submitted by individuals who are not qualified for the position or who showed only a passing interest in the post job position.

A. Applicant flow data will typically consist of a number of hardcopy documents.

It is fairly safe to assume that the creation of an applicant flow database will involve manually (or at a minimum scanning) converting a number of hardcopy paper documents to electronic format.

B. The set of work related information concerning the employer’s job applicants is typically significantly less inclusive for applicants who were not actually hired by the employer-defendant.

For instance, at the time of application, many individuals do not provide a full, complete, and verified work history that contains all the relevant dates and a description of the experience obtained at each job. Many times this type of information is completed during later phases of the hiring process. Some information is only collected if an individual is hired by the employer. These data limitations should be kept in mind when considering the questions to be addressed in the analysis of the employment discrimination allegations.

C. Each application flow source presents a unique set of issues for the attorney or expert who is examining the discrimination allegations.

For instance, the applicant pool for a job position that is primarily advertised through online job databases will look substantially different from a job position that generates the majority of its job applications from on campus recruiting efforts. In some circumstances, the applicant flow information that is collected from the employer may have to be augmented with government labor force data to provide a complete analysis of the company’s applicant flow.

Hiring data

The actual hiring outcomes of the employer are often used in conjunction with the applicant flow data to analyze allegations of hiring discrimination. In contrast to applicant flow data which contains information on the individual job applicants, an employer’s hiring data will generally contain data on the individuals whom the employer considered for a given job opening which in many circumstances does not include all the individuals that applied for a particular job. The hiring data contains information on the outcomes of the employer’s different phases of the hiring process.

These data are distinctly different from the application data and are often maintained separately from the applicant flow database. The hiring data will typically provide the analyst with detailed information on relevant hiring related issues such as the job candidate’s scores on job related pre-employment tests and standardized interview tests. The hiring data may also show which individuals voluntarily or involuntarily opted out at some point during the hiring process.

Generally, the employment attorney should attempt to obtain data that describes each phase of the employer’s typical hiring process. The hiring phases include basic job candidate qualification checking, job candidate phone interviews, job testing, and offers.

Issues to consider

Overall, the employment attorney should find that the hiring data is relatively well documented and substantially more complete than the application data. This is partially because some employers use pre-hire data to validate the efficacy of their employee selection process and to benchmark employee performance later in an employee’s tenure with the company. It is not uncommon for the hiring data to be a mix of electronic data including spreadsheets and hardcopy documents such as hand graded written tests. Applicant flow data, each set of hiring data presents issues that should be well conceptualized before beginning the data collection process. Perhaps the largest concern is ensuring that all the data concerning the job applicant’s employment relevant qualifications are collected and included in the hiring data. For instance, if there is a basic education requirement for the job or jobs in question then the data needs to incorporate and include educational information.

Salary Data

In order to analyze either disparate impact or disparate treatment in a case involving salary discrimination, the most obvious first step is to gather the relevant personnel data on employee compensation for the relevant group of employees. As with hiring and applicant flow issues, the appropriate universe and time period for the analysis will depend on the facts surrounding the particular discrimination allegations and the actual business practices of the defendant-employer.

The actual number of years used in the analysis may include only the year that the alleged discrimination occurred or it may span a number of years both before and after the alleged discrimination took place. Similarly, the analysis may or may not include the salary data for employees that worked outside of the issues where the alleged discrimination occurred.

Generally, an employer-defendant’s compensation job data includes the information about how the company’s employees are compensated and descriptions of the jobs that they perform. The information contained in these databases includes the employee’s base salary, commission salary (if applicable), salary grade and performance bonuses received. The salary data will also typically contain job codes that describe the employee’s job at the company. The basic salary information for this type of data will almost always be available in some type of electronic format. Compensation data, which is one of the more commonly outsourced human resource functions, is typically kept in a separate database and updated when changes are made to the employee’s compensation.

Compensation databases are commonly referred to as transactional databases because they are updated when a new transaction, such as a promotion or other compensation related event occurs. In practice, a firm’s compensation database may include several spreadsheet-like tables all of which are related by a common identifier such as an employer generated identification number or the employee social security number. Generally, databases that contain data tables that are related by a unique identifier of some sort are referred to as relational databases. (For more discussion on the different types of employee databases see the addendum on the last page of this article.)

It is important to note that a given employer may ‘lease’ some or all of their employees from companies as well as hire a significant number of independent contractors. While in day-to-day terms a leased employee or independent contractor may perform many of the same jobs of other regular employees at the company, the future job opportunities as well as other relevant job related factors may differ substantially from a regular employee. In any event, it is important to keep the specific employment arrangement of the company’s workforce in mind as the data is being collected for the statistical analysis.

Supplemental salary data

In addition to the basic compensation records, there are a number of other salary related data records that could be useful in a statistical analysis of employment discrimination allegations. These records include other employ non-compensation information such as the employee’s marital status, dependent information, and time away from the job taken for sick leave and vacation. For example, in some instances the employee’s time spent away from the company can provide useful insights about the employee’s job related soft skills such as the employee’s dedication to the particular job, which is often important but usually difficult to measure.

Conclusion

Regardless of which side you represent in an employment discrimination case, getting the right personnel data in these types of cases can be difficult. Depending on the employer, the complexity of the task of collecting the employment data from the employer can range from involved to colossus.

While it generally takes a significant amount of time and effort to correctly assemble the personnel data in a discrimination case, from a data analysis standpoint this is time that is well spent. The wrong data in the best statistical model can easily and quite quickly sink a statistical discrimination analysis. In this article, we have provided a primer on the data that is used to perform statistical analyses of employment discrimination allegations concerning hiring, salary, and employee terminations.

This article should be particularly helpful to employment attorneys who are faced with assembling a personnel database in an employment discrimination lawsuit, an EEOC administrative hearing or even a company initiated disparity ‘self audit’.