This article is meant as a companion to help you better understand and apply Review your Base Analysis in PayGap.
WHEN TO USE THIS ARTICLE
Use this article when:
the mean and median figures on your Base Analysis dashboard show very different values and you want to understand why
you want to know what the quartile distribution view is telling you about your workforce
you are preparing to explain these numbers to leadership, HR business partners, or employees
you are dealing with a tricky pattern (for example, mean and median pointing in opposite directions) and want concrete examples
you want to decide which measure to highlight in internal communication, alongside what your jurisdiction requires for reporting
Important: Base Analysis is an unadjusted view. None of the measures described here control for role, seniority, tenure, or other legitimate pay factors. For that, see the Adjusted Pay Gap analysis.
KEY CONCEPTS
1. Mean unadjusted pay gap
The mean is the simple average: add all salaries together and divide by the number of employees. Calculated separately for men and women, the mean unadjusted pay gap is the percentage difference between the two averages:
Mean Gap % = ((Men's mean − Women's mean) / Men's mean) × 100
A positive % means men earn more than women. A negative % means women earn more than men. PayGap shows the calculation behind the figure when you hover over the ⓘ icon on the tile, with the actual values used for your data.
The mean reflects the total financial disparity between men's and women's pay. It is the measure required by most reporting frameworks, including the EU Pay Transparency Directive.
The mean has one important weakness: it is sensitive to outliers. A small number of very high salaries pulls the mean up; a small number of very low salaries pulls it down.
Mini-example. A team of 4 men earning €40k, €50k, €60k, €70k has a mean of €55k. Add one director at €200k and the mean jumps to €84k - even though four out of five people earn well below €84k. The same effect operates on the men's mean across an entire workforce.
2. Median unadjusted pay gap
The median is the middle salary when all salaries are sorted from lowest to highest. Calculated separately for men and women, the median unadjusted pay gap is the percentage difference between the two midpoints:
Median Gap % = ((Men's median − Women's median) / Men's median) × 100
Same sign convention as the mean: positive means men earn more, negative means women earn more.
Unlike the mean, the median is not influenced by extremes. Only the position in the sorted list matters, not the absolute value at the top or the bottom. This makes the median a more stable view of what a "typical" employee earns.
Mini-example. Using the same team as above (€40k, €50k, €60k, €70k, €200k), the median is €60k — the middle value. Remove the director and the median becomes €55k. The median moves a little, but nothing like the mean.
3. Why the mean and median usually differ
When outliers are concentrated on one side (typically men at the top of the salary distribution), the men's mean is pulled up but the men's median is not. The result: the mean pay gap is noticeably larger than the median pay gap.
The size of the difference between the two measures tells you something about who is driving the gap:
Pattern | What it usually means |
Mean noticeably higher than median | The gap is amplified by high earners - typically more men at the top of the salary scale. Focus on senior representation and pay at the top. |
Mean and median similar | The gap is reasonably consistent across the salary distribution, not driven by outliers. Look for structural drivers (role distribution, job levels). |
Median higher than mean | Less common. Usually means that a small number of very highly paid women (or very low-paid men) are offsetting the average. Worth a closer look. |
4. Quartile distribution
The quartile view splits your workforce into four groups of equal size, sorted by salary:
Upper Quartile - the 25% highest paid employees
Upper Middle Quartile - from the median up to the 25% highest paid
Lower Middle Quartile - from the median down to the 25% lowest paid
Lower Quartile - the 25% lowest paid employees
Each quartile contains roughly the same number of employees. What differs is the gender composition within each quartile.
Quartile distribution is a representation view, not a pay view. It answers the question: where in our salary distribution do men and women sit? When the gender split changes meaningfully across quartiles - typically fewer women as you move up - that imbalance is often a major driver of the unadjusted pay gap, even if pay is equal for equal work at every level.
Always read quartiles against your overall workforce ratio. If women make up only 30% of your headcount, you would not expect 50% women in all quartiles - that is mathematically impossible. The right benchmark is your overall ratio. The question is not "is each quartile 50/50?" but "does each quartile reflect the overall mix, or do certain quartiles deviate from it?"
All four quartiles around 30% women (matching the overall ratio) → balanced representation, no quartile-driven contribution to the unadjusted gap, even though women are a minority overall.
Upper Quartile 10% women, Lower Quartile 50% women, overall 30% → meaningful deviation, women are concentrated at the bottom of the salary distribution and underrepresented at the top.
A workforce can have very few women and still have a balanced quartile distribution. It can also have a 50/50 overall split and still show heavy quartile imbalance. The two views are independent and need to be read together.
5. How quartile distribution connects to mean and median
Quartile distribution often explains the pattern you see in the overall gap tiles:
If the Upper Quartile is heavily male, your mean gap will be larger than your median gap - those top-paid men pull the men's average up.
If all four quartiles have similar splits, both your mean and median gaps are likely to be small (or have causes outside the salary distribution).
If imbalance is concentrated in a specific quartile (for example, only at the bottom), the driver is specific to that part of the distribution.
Used together, the three measures answer the question "where in the organization is the gap coming from?" before you start the deeper Drill-Down and Adjusted Pay Gap analysis.
WORKED EXAMPLES
These examples show how the same underlying numbers can produce very different headline figures, and how to read them.
Example 1: One high earner drives the entire mean gap
A 20-person department: 12 men, 8 women.
Men: 11 earn €50k, 1 earns €170k → mean €60k, median €50k
Women: 8 earn €49k → mean €49k, median €49k
Measure | Men | Women | Gap |
Mean | €60,000 | €49,000 | 18.3% |
Median | €50,000 | €49,000 | 2.0% |
Reading: The mean gap looks alarming, but the median tells the real story. The mean is pulled up by a single man earning €170k. If you remove that one person, the men's mean drops to €50k and the gap virtually disappears.
Action: Look at who that high earner is and why. If the role is legitimate and pay is justified, the gap is essentially a representation issue at the top - only senior representation will change it. Consider communicating both measures internally.
Example 2: Same overall numbers, different per-category story
Two companies have identical overall mean and median pay gaps - both around 7%. But once you split by category of worker, the picture is completely different.
Both have 70 men and 70 women, distributed equally across four levels.
Company A - gap concentrated in junior roles
Role | Men n | Women n | Men salary | Women salary | Mean gap | Median gap |
Junior | 30 | 30 | €40,000 | €32,800 | 18.0% | 18.0% |
Mid | 20 | 20 | €60,000 | €55,800 | 7.0% | 7.0% |
Senior | 15 | 15 | €85,000 | €85,000 | 0.0% | 0.0% |
Director | 5 | 5 | €130,000 | €130,000 | 0.0% | 0.0% |
Overall | 70 | 70 |
|
| 6.9% | 7.0% |
Company B - gap evenly spread across all roles
Role | Men n | Women n | Men salary | Women salary | Mean gap | Median gap |
Junior | 30 | 30 | €40,000 | €37,200 | 7.0% | 7.0% |
Mid | 20 | 20 | €60,000 | €55,800 | 7.0% | 7.0% |
Senior | 15 | 15 | €85,000 | €79,050 | 7.0% | 7.0% |
Director | 5 | 5 | €130,000 | €120,900 | 7.0% | 7.0% |
Overall | 70 | 70 |
|
| 7.0% | 7.0% |
Reading: From the overall gap tiles alone, these two companies look identical - same mean, same median. The comparable jobs view tells you they need completely different actions:
Company A has a localized problem. Women starting in junior roles are paid 18% less than their male peers - a clear systemic issue at entry level that compounds over time. Mid-level still has a small residual gap. Senior and director levels are clean.
Company B has a chronic, organization-wide problem. Every level shows the same 7% gap - small enough to look reasonable in isolation, but consistent enough that it indicates a structural issue (pay-setting practices, market-rate decisions, negotiation patterns).
Action: The remediation strategies are completely different. Company A should focus on junior pay-setting and entry salaries — fixing this stops the gap from compounding upward. Company B needs to review its pay structure organization-wide; a small uniform gap usually points to systemic factors like negotiation bias or market-rate adjustment policies.
Lesson: The overall mean and median tell you a gap exists. They cannot tell you where it is. Always cross-check with the comparable jobs / category-of-worker view before drawing conclusions or planning action.
Beyond vertical levels - horizontal segregation: The same logic applies horizontally. Men and women may sit at the same seniority level but in different job families - for example, women concentrated in HR or marketing roles, men in engineering or sales. If the families dominated by one gender are systematically better paid, that produces a gap even when within-role pay is equal. The overall tiles will not show this; Drill-Down Analysis is designed exactly to surface these horizontal patterns alongside the vertical ones.
Example 3: Median higher than mean (the rare case)
A boutique professional-services firm: 30 men, 20 women.
Men's salary range: €40k to €120k → mean €68k, median €70k
Women's salary range: €45k to €95k, with 2 partners earning €180k each → mean €72k, median €60k
Measure | Men | Women | Gap |
Mean | €68,000 | €72,000 | −5.9% |
Median | €70,000 | €60,000 | 14.3% |
Reading: The mean shows a "negative" gap (women earning more). The median shows a positive 14% gap (women earning less). Both are mathematically correct - but they describe different parts of the population.
Two highly paid female partners pull the women's mean above the men's mean, while the typical (median) woman earns less than the typical man.
Action: Do not use the mean alone in this case. The median is more representative. Investigate whether the gap at the median level is driven by role distribution or by within-role pay differences.
Example 4: Mean and median point in opposite directions
A 40-person team where:
The 5 highest-paid people are all men (so the men's mean is pulled up)
Men cluster in the middle of the distribution, while a few women are at the very bottom (so the men's median is below the women's median)
Result: mean gap +8% (favors men), median gap −4% (favors women).
Reading: This almost always signals concentration effects - outliers at the top on one side, outliers at the bottom on the other. Neither figure on its own tells you what is going on.
Action: Look at quartile distribution and the comparable jobs table together. Treat both figures with caution and explain the pattern when you communicate it. Drill-Down Analysis will clarify the underlying drivers.
Example 5: Small female population - read with care
A 200-employee tech firm with only 12 women across the workforce.
Quartile distribution: Upper Quartile 95% men / 5% women, Lower Quartile 88% men / 12% women.
Mean gap: 22%. Median gap: 18%.
Reading: The numbers are real, but with only 12 women in total the per-quartile percentages move dramatically when one woman changes role or leaves. A single hire or departure can shift the Upper Quartile from 5% to 10% women - a near-doubling, without anything structural changing.
Action: Treat these figures as directional, not precise. Track them over time rather than reading too much into a single snapshot. In Drill-Down Analysis, focus on the characteristics of individual women's roles rather than aggregate percentages.
Example 6: Three quartile patterns side by side
Three companies with similar mean gaps (around 22%) and the same overall workforce mix (40% women / 60% men), but very different quartile patterns.
Reading (benchmark: 40% women overall):
Quartile | Company A: top-heavy male | Company B: bottom-heavy female | Company C: stable share |
Upper | 85% M / 15% F | 70% M / 30% F | 63% M / 37% F |
Upper Mid | 70% M / 30% F | 70% M / 30% F | 61% M / 39% F |
Lower Mid | 50% M / 50% F | 70% M / 30% F | 59% M / 41% F |
Lower | 35% M / 65% F | 30% M / 70% F | 57% M / 43% F |
Company A - classic vertical segregation. Women's share drops well below the 40% benchmark in the Upper Quartile and sits above it at the bottom. The 22% mean gap is driven mostly by male concentration at the top.
Company B - the issue is at the bottom. Women are heavily overrepresented in the lowest-paid 25% (70%, far above their 40% overall share), often operational or part-time roles. The gap reflects who is doing low-paid work, not who sits at the top.
Company C - stable share across quartiles, all close to the 40% benchmark. Quartile distribution is not a major driver, even though women are a minority. The 22% gap must come from within-role pay differences or specific high-impact roles. Look at the comparable jobs table.
Action: Each pattern leads to a different remediation strategy. Same mean, three completely different stories. Note Company C: women are a minority overall (40%), but their representation is balanced across the salary distribution - so headcount imbalance alone is not what drives the gap.
KEY INFORMATION
Most pay gap reporting regimes require both the mean and the median. PayGap shows both by default.
The mean is the measure most commonly cited in public reporting - including under the EU Pay Transparency Directive - but is the most easily distorted by a handful of high earners.
The median is often considered a better measure of the "typical" employee experience and a better anchor for internal communication.
In small populations (very few men or very few women), all three measures can move dramatically with single hires or departures. Treat them as directional.
Quartile boundaries are recalculated with every new upload. Quartiles are calculated on the salary measure currently selected in the dashboard - if you change the pay component filter, the quartile view recalculates accordingly.
COMMON QUESTIONS / TROUBLESHOOTING
Which measure should I focus on internally?
Under the EU Pay Transparency Directive, the gaps that trigger formal action are adjusted gaps of 5% or more in a category of workers that cannot be justified by objective, gender-neutral factors. That is the regulatory threshold to focus on.
But the other unadjusted figures - overall mean, overall median, and per-category gaps below the 5% threshold - will also be disclosed and your employees will see them. They will ask questions: "Why is the company-wide mean gap 12%? Am I personally underpaid?" You need to be ready to explain where each number comes from and what it does (and does not) mean.
In practice this means you should:
Treat the 5%-in-category adjusted gaps as your action list under the directive
Treat the overall and per-category unadjusted gaps as your communication list - figures you must understand well enough to explain to employees, managers and works councils. Some drafts (and the directive itself) indicate that, on an employee's request, the employer will be required to explain or remediate any unexplained pay gap - not only the ones that cross the 5% threshold.
Use the mean and median together: the difference between them is often the easiest way to explain whether a gap is driven by outliers, representation or structural factors
For everyday internal messaging, the median often resonates better because it reflects the typical employee. But you must understand both - because both will be public.
A female employee saw the mean gap and asked for a raise to match. How do I respond?
This is a common situation after public reporting, an employee earning the median salary may see the mean figures and interpret them as evidence that she is personally underpaid. Example 1 above is a good illustration for the conversation: the mean gap can be driven by a single high earner and does not mean every woman is underpaid relative to every man. Show the median alongside the mean, and if needed, walk through the comparable job view so the employee sees where her specific role sits.
Can the mean and median point in different directions?
Yes - see Example 4 above. It happens when outliers are concentrated on different ends of the distribution for men and women. When you see this pattern, do not rely on either figure alone. Use the quartile view and the comparable jobs table to understand the underlying structure.
My adjusted gap is much smaller than my unadjusted mean gap. What does that tell me?
It tells you that most of the unadjusted gap is explained by pay factors - typically role, seniority, location, or experience. The adjusted gap is the part that the model could not explain with those factors. A small adjusted gap with a large unadjusted gap usually points to representation as the main issue (men and women are in different roles) rather than pay-for-equal-work.
My mean gap is huge - over 40%. Should I trust the figure?
Possibly, but verify the data first. Very large mean gaps can result from:
One or two outlier salaries (an executive, a founder) skewing the mean - check Drill-Down to see if a few records dominate.
Data quality issues - incorrect FTE values, missing pay components, or salaries entered in the wrong currency.
Genuine extreme imbalance - for example, a fully male executive layer in an otherwise mixed organization.
Compare with the median: if the median gap is significantly lower and only the mean is extreme, the issue is almost certainly outliers or data quality.
Does PayGap calculate these measures the same way every country does?
The formulas themselves are standard, but some countries require specific standardizations (for example, full-time equivalent, or specific pay components). Your official annual report will follow the local rule. The dashboard uses PayGap's standard definition.
Why aren't my quartiles split 50/50 by gender?
Quartiles group employees by salary, not by gender. Each quartile contains the same number of employees, not the same share of men and women. The split in each quartile also depends on the overall gender ratio in your organization - if women make up only 30% of the workforce, you would not expect 50% women in all quartiles. What matters is whether their share is stable across the four quartiles, or rises or falls as you move from the Lower to the Upper Quartile.
Why can a company with no pay discrimination still show an unadjusted gap?
Because unadjusted gaps reflect both pay differences within roles and who sits in which roles. A company that pays fairly within every role can still show a large unadjusted gap if men are overrepresented in higher-paid roles.
What does a negative pay gap mean?
A negative pay gap means women earn more than men on that measure. The sign follows the standard formula (men's salary minus women's salary, divided by men's salary).
How exactly are quartile boundaries calculated?
See the methodology documentation for the full calculation approach. The specific method can differ slightly from country-level official methods - your official annual report will follow the local rule, while the dashboard uses PayGap's standard definition.