Analyzing Compensation Data A Guide To Three Approaches

Why Analyze Compensation Data?

• Awareness, to identify the elements that actually influence salary, wages and other types of compensation
• Fairness, to determine if the elements are uniformly applied; and, if so what the impact is on minorities and women
• "An ounce of prevention is worth a pound of cure"

This guide describes three approaches that Federal contractors may use to analyze their compensation systems. Such analyses may be useful in determining if there are patterns of discrimination in the workforce. The focus of this Guide is on analyses of salaries or wages. However, the procedures described here can be used to analyze other forms of compensation as well. The analytical approaches described in the Guide are only three of many ways that salary or wage data may be analyzed. Depending on the factors used by your company to establish compensation, other approaches not included here may be more appropriate to use.

The approaches described in the Guide provide indicators that can be used by a contractor to assess the impact of compensation decisions on minorities and women. As you undertake a compensation analysis, it is important to remember that completeness and accuracy of data will affect results. Finally, if as a result of your assessment, you find problem areas, it is important that they be corrected.

The procedures in the Guide have been developed to assist Federal contractors that wish to analyze their compensation programs to ensure that all employees are treated fairly. The Office of Federal Contract Compliance Programs (OFCCP) may do additional or different analyses of compensation data during the conduct of a compliance evaluation. During a compliance evaluation, OFCCP also may examine a range of other employment practices such as hiring and promotions to determine compliance with the regulations administered by OFCCP. In addition to the regulations, OFCCP's Interactive Compliance Assistance Advisor and the Federal Contract Compliance Manual provide detail information about OFCCP requirements and procedures.

BASIC DATA REQUIREMENTS

• Data elements that are identified as affecting compensation
• Current data for each of the identified elements

The first step in conducting a compensation self-assessment is to determine what factors are used to establish compensation levels for each job within the company. These factors vary from company to company and from industry to industry. Typically. many companies include such factors as

• length of time at the company or in the position or both
• grade or pay level
• research on the market value of the position
• performance evaluation ratings
• educational degrees or certifications required
• related experience in previous employment

Many times factors to be identified at this step of the process are included in various procedural manuals that also include criteria for entry in various levels, definitions of terms used, and ranges for bonus payments or salary increases at given grade levels.

THREE TOOLS FOR COMPENSATION ANALYSIS

• Median Approach
• Average Approach
• Table and Sort Features on the PC

The three approaches that are described are ones that are currently used by OFCCP compliance officers as they conduct compliance evaluations of Federal contractor establishments. The median approach is one that is easy to use and provides much useful information; however, more sophisticated statistical tests do not usually use the median for comparisons. The average approach permits the comparison of averages and also extension to statistical tests. The discussion of tables and sorts using a personal computer describes the easiest and simplest set of analyses. However, even using this approach can indicate areas where there may be discrimination. Although each approach is described separately, they can all be done as part of a compensation analysis program.

The examples that follow all concentrate on salary analysis. A person's salary may represent only one part of his or her total compensation and it may be necessary to analyze other elements such as commissions or bonuses. In addition to analyzing salary, it may be necessary to analyze other factors, such as experience or time on the job, in order to explain differences in the salary between groups of similarly situated employees.

MEDIAN COMPENSATION ANALYSIS

The median is one way to describe the mid point of the group of salaries being analyzed. It is the salary figure that is larger than or equal to half of the other salaries and equal to or smaller than half of the other salaries. In identifying the median, it is easiest if the items are arranged in ascending or descending order. For example, in the following group of salaries

the median is $19,334, since $19,334 is the number at the middle of the distribution. If there were an even number of salaries in the distribution, the median is computed by computing the average of the middle two salaries. For example, if the group of salaries was

the median would be $32,465, which is the average of $33,022 and $31,900, which are the middle two salaries./P>

The median is useful for analysis because it is not affected by very high or very low salaries in the distribution. In many cases, it is a more revealing measure than the average, which is described next. It is usually more revealing when the data being analyzed contains a relatively small number of salaries with values at one extreme, either very high or very low.

In conducting a compensation analysis, the median value for example, for men and for women is calculated and compared for each element affecting compensation. The medians are then compared, both individually by element and for combinations of elements, such as length of service and pay, or performance ratings and pay. In making these comparisons between elements, one might ask:

• Does the group with higher median pay also have longer median periods of service?
• Does the group with higher median pay also have higher median performance evaluation ratings?
• Are there other business factors that explain the higher median pay of a group?

The size of the difference should also be taken into consideration. For example, does only a few months difference in seniority warrant several thousand dollars difference in compensation? Or, are the differences in compensation slight, less than $100 a year so that they might easily be explained by other factors not included in the analysis.

The following table provides an example of how a median analysis might be presented for those salary grades where important differences were found. The columns show the salary grade, the median salary for the grade, the median years in the grade and a comparison of the male to female differences.

In analyzing this chart:

• It is assumed that time-in-grade is more significant than company seniority in determining current salary.
• In grade 14, further investigation seems indicated since the median annual salary for females is several thousand dollars less than the median annual salary of males and females have less than a one year difference in median time-in grade.
• In grade 10, further investigation does not seem indicated since there is only a small difference in the median annual salary of females compared to that of males and the median years in grade for females is less than the median years in grade for males.
• In grade 6, further investigation seems indicated since the median annual salary of males is less than the median annual salary of females, but there is less than one year difference in median time in grade between the two groups.

Thus, further investigation of the salaries in Grades 14 and 6 seems indicated and, if all other factors are equal and the differences in salary still exist, it is possible that upward adjustment to the compensation of certain individuals is needed.

AVERAGE COMPENSATION ANALYSIS

The average or arithmetic mean is another way of describing the mid point of a group of salaries being analyzed. The average is computed by adding the salaries in a distribution of salaries being analyzed, and then dividing their sum by the number of salaries. The average is useful for comparing the sizes of the terms in two groups, for example comparing the average salary of men to the average salary of women. One can then ask questions such as the ones suggested below:

• Is the average salary roughly the same for each group or is one much larger than the other?
• Does the group with the higher average salary also have longer average periods of service?
• Does the group with the higher average salary also have higher average performance ratings?
• Are there other business factors that explain the higher average pay of the group?

For each element affecting compensation, the average for men and for women is calculated and compared. These should be done for each salary level, grade level, time in position, etc. For example, in the table below, the male average salary is shown in the second column, the third column shows the difference between the male average salarv for each grade level and the female average salary. The fourth column shows the difference in the average seniority of females compared to malcs and the fifth column shows the average difference in performance evaluation scores.

In analyzing the above table, one might make the following observations:

• Although there is a large difference in the average female salary in grade 9, females also have fewer years of seniority and lower performance evaluations, which may explain the differences in salary.
• The average difference between male and female salaries in grade 8 is very small and there are no differences in the average seniority or performance evaluations. This difference seems small enough so that no further investigation is warranted.
• Although there is a relatively large difference in the average female salary in grade 6, females have many more years of seniority and higher performance evaluations, which may explain the differences in salary.

"TABLE" AND "SORT" FEATURES ON THE PC

Spreadsheet software includes a variety of procedures that can be used to analyze and present data. The sorting features found in a spreadsheet can be used to give a quick overview of the workforce arrayed by various dimensions. For example, just a few of the ways that the sort feature may be used, include:

• sorting by salary identifies the highest paid minority group member or woman and the level at which there is no further representation of either group
• sorting by hire date gives an indication of whether particular group members are new hires or have long tenure
• sorting by current job date can be used in an initial assessment of length of time since last promotion
• sorting by previous job title can identify feeder groups
• sorting by department or other organizational unit can indicate concentrations or absences of members of a particular demographic group

Software that includes a subtotal function can be used to give counts, for example to count the number of individuals with an identical job title. It may also provide the maximum and minimum values of a field. The pivot table and filter functions are also useful adjuncts for these types of analyses, since they isolate fields of interest, which can then be examined apart from the entire data spreadsheet. The sorting features are particularly useful in giving an overall picture of the composition of the workforce as arrayed using different criteria.

Charts and graphs are another useful way to present data. The chart below shows the percentage distribution of minorities and women in various sales positions arrayed from the lowest paying job title, Accountant 1, to the highest, Financial Manager.

In looking at this chart, one can make the following observations:

• Women and minorities are 2.5 times as likely to be in the entry Accountant I job classification
• 30 percent of the total Accounting Department is in the lowest two classifications
• 60 percent of the women and 45 percent of the minorities are in the lowest two classifications
• Almost 25 percent of the total Accounting Department is in the top two classifications
• Less than 6 percent of the women and less that 8 percent of the minorities are in those two top classifications

While this chart does not provide a complete picture of the Accounting Department, and there is no consideration of length of employment, for example, the chart does raise questions about the compensation levels of minorities and women that warrant further investigation. Translating tables into graphs is a relatively easy operation, but one that can provide a vivid picture of a potential problem area.

• Wage differentials are not discriminatory or unlawful when based on neutral job related factors.
• Analysis of compensation data is a useful tool to assure fairness in compensation practices.
• The analysis of compensation data may identify problems that can be corrected without being the subject of a complaint, compliance review finding or litigation.
• In its initial phases, the analysis of a compensation system may be done simply without "expert" or outside assistance.
• The OFCCP strongly encourages Federal contractors to conduct analyses of their compensation systems in accordance with the self-audit responsibilities under Executive Order 11246, in order to eliminate or prevent discriminatory policies and practices in this very important aspect of employment.

• Regional Activity Calendars
• Construction Contractor TA Guide
• Census 2000 Special EEO File
• Sample AAP for Small Employers
• Frequently Asked Questions
• EEO Poster
• EEO-1 Report
• Disability Resources
• Veterans Resources
• VETS 100

• ADA
• E. O. 11246
• Section 503
• 38 USC 4212
• 41 CFR Chapter 60
• Final Rules
• Notices
• Proposed Rules & Comments Received

• National & Regional
• District & Area