Interpreting pay gaps

The system showcased the Unadjusted pay gap for your dataset, the average salary difference among demographic groups, following the previous section where you created your compensation model and ran your pay equity analysis.

Once you’ve successfully conducted a pay equity analysis and developed a reliable compensation model, you can begin to explore your pay gap results in greater detail. In this section, we will take you through four key concepts and end with where to find the pay gap metrics in the platform:

• Comparing pay across demographic groups

• Measuring pay gaps: Unadjusted and Adjusted

• Demographics and reference groups

• Viewing your pay gaps in the platform

• Viewing your pay gaps by group

The Analysis overview result page is the first page you see after running a pay equity analysis. It is also the first place that displays your unadjusted and adjusted pay gaps in the insight tiles. For a deeper dive into your pay gaps, we recommend navigating to the Pay gaps tab within the same screen. This is the key location to view and interpret your pay gaps.

Pay gap insights

Comparing pay across demographic groups

When analyzing pay equity, it's important to recognize that not all salary comparisons are created equal. Our software allows you to look at compensation through three key lenses, where each tells a different story about equity in your organization.

Average comparison of salary

This is the simplest way to look at pay differences: it calculates the average salary for different demographic groups, for example, comparing the average salaries of men and women across the company, also known as the Unadjusted pay gap, or sometimes the gender pay gap.

What it highlights: Broad patterns and potential disparities at a high level.

Limitations: It doesn’t account for differences in job type, experience, or responsibilities, so gaps here may be due to factors other than pay inequity.

Comparison of salary while adjusting for objective factors, or comparing equal or similar work

This approach looks at people who are in the same or similar profiles, often using job families, grades, skills, locations, or classifications, and compares their pay and measures whether any demographic differences exist.

What it highlights: Whether people in the same profiles are being paid equally, regardless of demographic groups, also known as the adjusted pay gap, or sometimes the equal pay gap.

Limitations: It assumes job titles and descriptions are used consistently, which isn’t always the case.

Comparing work of equal value

This is a more advanced, and often legally significant, approach. It compares compensation for jobs that may not be identical, but are considered to require similar levels of skill, effort, responsibility, and working conditions. For more information, see Value-based comparison.

What it highlights: Whether job roles of equal value to the organization are being compensated fairly, even if the job functions are different.

Limitations: Determining “equal value” requires more analysis and judgment, but it helps surface deeper equity issues that other methods may miss.

Each method serves a purpose, and together, they help provide a more complete picture of pay equity in your organization. Next, we will cover how these terms relate to the measures of pay gaps.

Measuring pay gaps: Unadjusted and Adjusted

When analyzing pay equity, two key metrics come up again and again: the unadjusted pay gap and the adjusted pay gap. Each tells a different part of the story, and each is calculated in a different way.

Unadjusted pay gap

The Unadjusted pay gap, also known as the raw or gender pay gap, is the most basic way to measure pay differences. It's a simple comparison of the average salaries between two demographic groups—for example, men versus women across the entire company. This gap is calculated as the difference between these average salaries, expressed as a percentage of the reference group's average.

This metric highlights high-level disparities in how different groups are represented across an organization's roles, as it naturally includes the effects of varying job types, levels, and functions.

Example: Women earn 6% less than men on average in the company.

Adjusted pay gap

The adjusted pay gap , also called the equal pay gap, measures pay differences when controlling for factors that legitimately influence compensation, such as job roles, grades, experience, etc. The industry standard is to use a log-linear multivariate regression model, which estimates how various compensation drivers- such as job family, job grade, location, education, and tenure - contribute to pay. This is the method used within the PayAnalytics platform when running a Pay Equity analysis. When running such an analysis, you are building a compensation model that consists of all variables that contribute to pay. For more information, refer to Creating a compensation model.

Within PayAnalytics, we include the demographic variable (e.g., gender) in the regression model. The model's coefficient for the gender variable, after accounting for all the objective variables, is the adjusted pay gap. In a truly gender-neutral pay environment with no pay gaps, the coefficient for gender would be zero, as it does not contribute to or explain any differences in pay. The Adjusted pay gap relates to the comparison of demographics who perform equal or similar work.

Example: After adjusting for all objective factors in the model, for example, job level, tenure, and location, women earn 3% less than men.

Demographics and reference groups

Demographic pay gap

When analyzing pay equity, a key step is identifying which demographic groups you are comparing with. While gender is the most common demographic used in pay gap analysis, it's not the only option. The demographic variable you choose depends entirely on the purpose of your analysis, such as measuring the gender pay gap or the ethnicity pay gap, etc. The PayAnalytics system enables the inclusion of up to two demographic variables.

Some demographic variables include more than two groups. For example:

• Gender may include male, female, non-binary, and not-reporting.

• Race may include several categories, such as White, Black, Asian, Hispanic, and Indigenous.

When this occurs, the platform will compute separate pay gaps for each group compared to a reference group. This means that for each additional demographic group, there is an additional pay gap measure.

For example, if gender includes three categories, you would compute two pay gaps, one for each group compared to the reference (male): Female compared to Male pay gap and Non-binary compared to Male pay gap.

Reference groups

When running a gap analysis, you have the option to select a reference group. This is the demographic group that the other groups are compared against. The choice of reference group should be aligned with the goal of the analysis, legal requirements, or your organizational standards. It is important to document and communicate the rationale behind your choice to ensure accurate interpretation of the results.

The selection of a reference group is critical because it directly shapes how pay gaps are calculated. Without thoughtful choice and clear understanding of the rules, results can be misinterpreted."

Automatic selection

The PayAnalytics platform defaults to “Automatically select” for the reference group. This setting identifies eligible groups based on fixed criteria and ultimately selects the group with the highest adjusted pay. Following figure illustrates the selection logic used when automatically determining the demographic reference group:

Automatic selection logic

This process ensures that groups that are too small to serve as a benchmark are not eligible, and that the highest-paid eligible group is used as a fair baseline.

Manual selection

Alternatively, you can override the automatic reference group selection by manually choosing a group. This can be particularly useful in contexts such as regulatory reporting, where the reference group is predefined, e.g., using males as the reference even if females are the highest-paid group. Another scenario for manual selection is when Adjusted pay gaps fluctuate around zero, with different demographic groups alternately having the higher adjusted pay. In such cases, organizations may prefer to use a consistent reference group to maintain comparability over time.

To manually select a reference group, go to the Measure pay gap section and choose your preferred reference group from the Demographic variable drop-down list. When analyzing two demographic variables, you can specify a reference group for each variable individually, as presented in the following picture:

Manual selection of the demographic variable

Viewing pay gaps in the platform

Upon successful import and configuration of your data, you will be directed to the Employee Overview page. Here, you'll find the Data summary section (located either to the left or below the employee overview graph, depending on screen size). By default, the platform displays the average unadjusted pay gap (labeled as 'Gap'). You can customize this view by selecting alternative aggregates, such as the median Unadjusted pay gap, to suit your analytical needs, as presented in the following illustration:

Employee overview page

After running a pay equity analysis, you will be directed to the Analysis overview tab. When performing a single pay gap analysis (using one demographic variable), the insights tiles at the top of this section display both the Unadjusted and Adjusted pay gaps.

This is an attention message.

This option is not yest available for multiple demographic pay gaps, however, we will continuously be releasing updates to this section to improve pay gap clarity.

The Unadjusted and Adjusted pay gaps are presented in the key results tabs when you run your pay equity analysis without raise suggestions, as presented in the following illustration:

Analysis without raise suggestions

The Adjusted pay gaps, before and after raises, are presented in the key results tabs when you run your pay equity analysis, including raise suggestions, as illustrated in the following picture:

Analysis with raise suggestions

Within the Pay Equity Analysis page, under the dedicated Pay Gap tab, you will also find the Pay gaps by demographics section which includes the adjusted and unadjusted pay gaps metrics before raises and after raises. You can toggle between these two options. If you have only two demographic values (such as male and female) the graph will look relatively simple, as illustrated below. However for demographics with multiple values (for instance, ethnicity being White, Black, Asian, Hispanic) the graph will become more extensive.

Result section

The unadjusted and adjusted pay gaps mentioned above are available for any given pay equity analysis, including all employees in the analysis.

Pay gaps in subgroup analysis

When you conduct a subgroup analysis, you’ll be directed to an overview page listing all analyses and their key results. For each subgroup, the system displays:

• The Adjusted pay gap

• The Unadjusted pay gap

• The p-value, which indicates the statistical significance of the adjusted pay gap.

This is an note/information message.

A smaller p-value signifies higher statistical significance.

In addition, the system calculates an overall adjusted pay gap for all employees. This is based on the weighted average of the Adjusted pay gaps across the subgroups. However, the unadjusted pay gap for all employees is calculated using the full employee dataset, not a weighted average of the subgroups. An example of such calculation is presented in the following illustration:

Example calculation of Adjusted and Unadjusted pay gap for all employees in the organization.

If the system's automatic reference group selection is enabled, it will choose a reference group independently for each subgroup analysis. For instance, 'Female' may be selected as the reference group in one subgroup, while 'Male' could be chosen in another, depending on the criteria for that specific subgroup.

Unadjusted to adjusted pay gap

To conduct a more in-depth analysis of pay gaps within your organization, particularly understanding how different variables influence the adjusted pay gap, navigate to the Pay gap tab after completing a pay equity analysis.

The initial section, From Unadjusted to Adjusted Pay Gap, presents the variables incorporated into your model and details their individual impact on the pay gap. The accompanying graph visually illustrates whether the inclusion of a specific variable leads to an increase or decrease in the gap. This provides crucial insight into potential inequities present within your pay structure.

In the following illustration, the adjusted pay gap is in favor of male employees:

'From Unadjusted to Adjusted Pay Gap' section

In the example provided, the adjusted pay gap favors male employees, leading to the following interpretations:

• When a variable reduces the unadjusted pay gap (e.g., grade group): This indicates that, on average, there is a higher concentration of male employees in more highly paid grade groups. Similarly, for management responsibility, male employees, on average, hold more management responsibilities. When you adjust for these factors (e.g., grades or management responsibility), the pay gap decreases. This suggests that a portion of the unadjusted pay gap can be explained by differences in the distribution of employees across these variables.

• When a variable increases the pay gap (e.g., education), in cases where the gap favors male employees: This implies that, on average, male employees receive higher compensation for their level of education compared to female employees.

• When the adjusted pay gap is higher than the unadjusted pay gap: In such scenarios, the overall distribution of genders across the pay structure is relatively equal, meaning men and women hold similar pay levels. However, when comparing salaries within comparable roles or with similar qualifications, a pay gap emerges, indicating underlying discrepancies even with seemingly balanced distributions.

Pay gaps by groups

This is a tip message.

To discover more about pay gaps by groups within the Pay gaps tab, an important requirement under the EU Pay Transparency Directive, we recommend that you move to the following article Measuring & closing pay gaps by group.

To further analyze the pay gaps in groups within the variables in the competition model, go to the Categorical variables section of the Compensation model tab. For each categorical group, the system is now displaying the unadjusted pay gaps and the estimated adjusted pay gaps.

These pay gaps by group are calculated using only the employees within each respective group. The adjusted pay gap is estimated without using regression coefficients by computing the average outlier percentage for each gender and comparing the difference. This method provides an approximation of the adjusted pay gap.

Consider the example below that shows the estimated adjusted pay gap in India. In this particular case, female employees earn, on average, 3.3% less than male peers, all else being equal.

Example of estimated pay gap

In this section (i.e. Categorical variables) of the compensation model results , you can explore and analyze where the most significant pay equity issues exist within the groups included in your compensation model.

To examine pay gaps across other groups, referring to categories not included in the model, go to Pay gap breakdown by other groups section. Upon selecting a column of interest, you can view both the unadjusted and estimated adjusted pay gaps for these groups. This provides deeper insights into specific teams, business units, or any other defined groups, helping you identify whether and where any pay disparities exist.

Pay gap breakdown by other groups