The panel opens through the Analysis parameters button in the Input data section. Its most important function is selecting the regression-adjustment variant, which determines which control variables enter the calculation of the adjusted gender pay gap.
Two tabs
Tab | Content |
Basic parameters | Time, unit and regression settings. |
Code lists | Custom education, seniority and grading types. |
Code lists tab
This tab allows the client to define its own levels for three key attributes used in employee data. The application does not impose fixed categories; the client can define what fits its internal structure.
Education types
For each education level, the client can enter an ID or name and the number of years in that education cycle.
Example ID / name | Number of years |
Primary school | 9 |
Secondary school | 13 |
University | 17 |
Doctorate | 20 |
The value from the Education and training column in employee data is mapped to these definitions. For regression, the number of years is the important value because it is used as a continuous education variable.
Professionalism / seniority
The client can define seniority levels within positions, for example Junior, Medior and Senior. The scale may also be Trainee / Specialist / Lead / Principal or any other internal structure.
Custom grading
If the client has its own pay grade system and wants to use it directly without the Work group estimator, grades are defined as ID + name pairs. These definitions are then used in Level 1 / Level 2 / Level 3 in employee data.
Basic parameters tab
Field | Type | Default / example | Description |
Reference year | Integer | 2025 | Year for which the pay gap is calculated. |
Standard working week | Integer | 40 | Standard weekly working time. |
Normative annual hours | Integer | 2,088 | Annual working-hour fund, for example 40 × 52.2. |
Age format | Drop-down | Age in years | Unit for the age field: age in years, date of birth or start date. |
Service format | Drop-down | Years of service | Unit for the Years of service column. |
Currency | Drop-down | CZK | Currency of all remuneration fields. |
Regression parameters — six predefined variants
A separate drop-down field above Save values contains six predefined variants of control variables that enter the linear regression.
What is a variant
A variant is the list of regressors used to adjust pay for structural factors. The coefficient of the Gender variable after this regression represents the unexplained pay difference, i.e. the adjusted gender pay gap. The selected variant can materially change the result. More regressors mean a higher chance that part of the difference is explained by structure, reducing the adjusted gap.
Example of the broadest variant
One broad scenario includes:
Length of employment.
Years of education.
Years of experience.
Seniority.
Employee scarcity.
Employee evaluation.
Employee contribution.
This is one of the broadest scenarios and uses all attributes entered in Manage data.
Criteria for selecting a variant
The selected variant must meet two conditions:
Relevance — attributes must be meaningful for the company and must not double-count the same logic, for example if years of experience and length of employment overlap.
Completeness — all attributes included in the variant must be filled in the data. If a selected column is empty, the model may work with zeros instead of real values and produce an incorrect regression result.
Reproducibility
The Simple overview report repeats the list of regression parameters used in the run. This makes the result transparent and auditable even without access to the settings panel.
Saving
The Save values button writes the parameters. Changing parameters does not recalculate the analysis automatically; the user must run the analysis again.
Link to the analysis
Analysis parameters belong to the configuration of the calculation run, not to employee data. The same dataset can be analysed using different regression variants, allowing sensitivity of the result to be assessed. Together with three grading systems, the client can obtain up to 18 views of the same dataset: six regression variants times three grading systems.