Regression Analysis

The Regression analysis helps you systematically examine pay differences between men and women. In contrast to descriptive analysis, it takes into account relevant influencing factors that can explain pay differences.

What does regression analysis do?

Regression analysis shows how strongly various factors are related to pay – for example, location, position level, or performance. The model compares all employees at the same time and calculates which pay differences can be explained by these factors.
The remaining difference between men and women that cannot be explained by objective characteristics is referred to as the adjusted pay gap.

How do you determine which factors to include?

Use your internal pay structure as a guide:

  • How is pay currently determined in your organization?
  • Which factors significantly influence the level of pay? Typically, these include:
    • hierarchical level
    • professional field,
    • professional experience,
    • or individual performance.

How do you determine the quality of the model?

A key measure of quality of analysis is the so-called Adjusted R² value. It indicates how well the model can explain pay differences.
Interpretation of Adjusted R²:

  • The closer the value is to 1, the better the model explains pay differences.
  • If the value increases significantly, the added factor contributes substantially to explaining pay differences.
  • If the value remains unchanged, the factor provides little explanatory power.

How do you interpret the results, and what should you pay attention to?

A negative sign of the value shown in the table means that women earn less on average than men.

If the calculated pay gap changes after including additional factors, it is because these factors differ between women and men and help explain compensation differences.

If the pay gap does not change after including another factor, there may be several reasons:

  • The factor has no influence on pay.
  • The factor is distributed similarly between men and women.
  • The factor is strongly correlated to another factor already included in the model and offers no additional explanatory power.

How do you determine which pay components to analyze using regression?

We generally recommend conducting this type of analysis for all pay components to gain the best possible understanding of your potential pay gaps.
An analysis is especially useful for pay components that:

  • are granted to many employees,
  • are awarded regularly and transparently,
  • vary sufficiently variation to make differences visible,
  • Examples: base salary, target and actual variable compensation

However, regression analysis is less suitable for other variables. This is the case, for example, if only very few employees receive these pay components or if the variation in these pay components is little (e.g., with a defined spot bonus). In these cases, we recommend alternative analysis methods. Please feel free to contact us.