Explanatory factors

Under System parameters ▸ Explanatory factors an administrator can define factors that are based on employee data and may explain part of the pay gap in Group insights. The list starts empty. While it is empty the whole gap is unexplained, and nothing changes in the analysis.

Sources and effects should be reviewed by a lawyer. The page shows this as information. It does not stop you from saving.

Allowed sources

Tenure, counted in years from the hire date.
Performance (PERFORMANCE, a number).
Level (LEVEL). The number in the level is used, or your own value per level.
A custom field of the number, choice or yes/no type. Free-text, file and date fields can never be chosen.
- A number field is used as a number.
- A choice field is used either as an ordered scale, with a number for at least two of its options, or as a category: one of the options.
- A yes/no field is used as a category: Yes or No.

Blocked sources. They can never be chosen:

gender and anything derived from gender;
birth date and age;
parental leave and absence, sick leave;
part-time and employment rate;
salary fields;
contact, address and bank details;
nationality, ethnicity, religion and disability;
union membership.

A custom field is blocked if its name or sensitivity tag shows such a ground. A custom field that is not blocked gives a warning.

Fields with restricted visibility. A field that is hidden from some users, by the field’s own visibility or by a role, can be used. Group insights is then calculated on the server, also when you filter, so everyone who sees the analysis gets the same result. The page shows a line about this. Such a field is only used for groups, never per person in Individual analysis.

If a custom field is deleted, every factor based on it is deactivated. This is saved as a new version and logged. The list shows a warning on the factor. A factor whose field has changed so that it is no longer allowed is not applied, and Group insights says so.

Every factor needs a described objective reason and a confirmation that the source is objective and gender-neutral.

How it is calculated

Group insights compares two roles: a female-dominated role and a higher-paid comparator role, in the same group or between groups.
For each factor, the mean is calculated in both roles. A mean is only calculated when at least as many people as the minimum group size have a value.
Explained = effect × (the comparator role’s mean − the female-dominated role’s mean). With the direction “a lower value explains more”, the difference counts the other way.
For a category, the mean is the share of people who have the category. The effect is what is explained at a full difference in share: everyone in the comparator role has the category and nobody in the female-dominated role. A difference of half the share explains half.
Explained is never below 0 and never above the factor’s cap. The cap is at most 5 percentage points per factor.
The sum over all factors is never more than 10 percentage points and never more than the gap itself.
Unexplained = the gap − the sum explained, but never below 0. A factor can therefore never make a gap larger.
The caps of active factors may add up to at most 10 percentage points. This is checked both when you save and in the calculation.

Group insights shows one label per factor that explained something, both within and between groups. It also shows suggested supporting text for a justification. The suggestion is never saved automatically, and no gap counts as justified on its own (art. 10). The pay gaps in the EU report (art. 9) are never affected by the factors.

Individual analysis. The factors are also used per person. The personal target becomes the target × (1 − explained / 100). Explained is counted against the mean of the group whose median is the target, and the same caps apply. That group must have at least as many values as the minimum group size. The row shows the amount labelled as a factor-adjusted estimate. Who is flagged and the actual gap do not change. The action plan costs the factor-adjusted estimate.

Justification with a factor. In Individual analysis you can choose the justification Explanatory factor and the factor that justifies the gap. The justification stores the factor’s id and version. The factors only suggest. A person always chooses the justification.

Every change and every deactivation is saved as a new version and recorded in the audit log. A deactivated factor can be reactivated. The annual snapshot stores a full copy of every active factor with its version, and of every factor version a justification refers to. A later change therefore never affects an archived year.

The feature is on by default. The Explanatory factors permission in the roles controls who may see and change the factors. Employees do not have it. Factors affect the whole organisation’s results, so changing them requires Admin or a custom role with write access and the Access Area All. A role with a restricted Access Area can see the list but not change it.