.png)
When it comes to compensation, most organisations think they know what they’re rewarding. For example, it’s common to take criteria like performance, experience, education level and tenure into account.
But often, there are also other factors at play — things like unconscious bias, structural inequities or even outright discriminations. These operate under the surface and can influence compensation in ways that aren’t always obvious.
Here’s the problem: with so many legitimate variables to account for, how can you tell whether a pay gap is justified or discriminatory?
That’s where multiple variable regression analysis comes in. This might sound like a highly technical concept — and it is. But at its core, it’s a straightforward tool that can help you cut through the noise and understand what’s really driving pay at your organisation. Read on to find out what you need to know.
What is multiple variable regression analysis?
Multiple variable regression is a statistical technique used to understand how different factors influence a single outcome. In the context of pay equity, it’s used to help employers uncover which variables are driving pay differences across the organisation.
By running a multiple variable regression analysis, you can control for legitimate business factors like tenure, job grade, job family, experience and performance when assessing whether gender — or any other potentially discriminatory factor — is influencing pay.
This allows you to answer questions like:
- Is there a statistically significant difference in men’s and women’s pay when other factors are accounted for?
- What is the impact of each year of tenure on pay when you control for age?
- Is education valued the same across different departments?
- Is there a gender pay gap among new hires, even when education and prior experience are accounted for?
- Do employees from different ethnic backgrounds receive equal pay when controlling for qualifications and role level?
- Does parental leave history impact pay progression differently for men and women?
The benefits of multiple variable regression analysis
There’s no sugar-coating it: multiple variable regression is a complex statistical process. Running this kind of analysis takes time, effort and investment. You may even need to invest in tools or bring in external support to help you properly interpret the results.
But the payoff is worth it — here’s why:
- Goes beyond averages: Simple averages can hide a lot. Multiple variable regression gives you a more detailed view of your pay gap, helping you understand how different factors interact. That means you can see the full picture and target your interventions where they’ll make the biggest difference.
- Enables fair comparison: This method ensures you’re comparing like for like, enabling more accurate comparisons. For example, you might discover that a gender pay gap disappears once you control for variables like experience or job role. Alternatively, you may uncover disparities that averages alone wouldn’t reveal.
- Provides actionable insights: Rather than just identifying the existence of a pay gap, multiple variable regression shows you what’s driving it. That gives you the insight you need to design meaningful, data-backed interventions.
- Builds trust and transparency: A robust analysis of your pay systems — and clear communication about the results — helps build trust with your team. It shows you’re serious about fairness, not just ticking a box. (We’ll share more on how to communicate the results of your analysis with employees later on.)
How to conduct a multiple variable regression analysis
Running a multiple variable regression analysis can be complex — especially if you’re working with large datasets or unfamiliar tools. It may involve partnering with external experts or investing in specialist software.
But even if you’re not doing the analysis yourself, it’s important to understand the core steps involved. To help you, here’s an overview of what the process typically looks like.
1. Choose the variables you want to analyse
Multiple variable regression lets you assess the impact of different factors on pay at the same time. Start by listing the legitimate business factors you expect to influence compensation — such as:
- Experience
- Performance ratings
- Education level
- Tenure (years of service)
- Job level
- Job family
- Location
You should also include variables that shouldn’t be driving pay differences — but might be. Under the EU Pay Transparency Directive, gender is key here. You may also want to include other protected characteristics such as age, race, disability or sexual orientation, where data is available and legally permissible to collect.
2. Gather your data
Much of the data you’ll use for your analysis is likely already captured in your HRIS. For example, this is probably where you’ll find each employee’s job level, salary, location, tenure and more. Depending on the variables you’ve selected, you may also need to pull in data from other systems like your performance management platform, compensation tool or ATS.
3. Choose a tool and run your analysis
Next, decide how you’ll run the analysis. Data experts may use tools like Excel, R or Python, but these are complex and require specific expertise. These days, tools like Figures simplify the process for HR and compensation teams, allowing you to input data and interpret results without writing code or managing formulas manually.
4. Interpret your results
Once the analysis is complete, you’ll see which factors are statistically significant drivers of pay. Look closely at any unexplained differences linked to gender, ethnicity or other protected characteristics. But also check whether the variables driving pay align with your compensation philosophy. For example, is performance having the expected impact? What about tenure or education level?
5. Take action on your findings
If pay disparities remain after controlling for legitimate factors like job level, you’ll need to take action. That might mean reviewing how pay decisions are made, revising policies, or making pay adjustments to close unjustified gaps. The goal is to translate your insights into meaningful, measurable change.
Comprehensive or granular? Two approaches to regression analysis
Multiple variable regression is a widely accepted method for analysing pay equity — but there’s still some debate about how best to apply it. Broadly, there are two main approaches:
- The comprehensive approach: This involves analysing large groups of employees — for example, all sales professionals in a region — while controlling for relevant factors that influence pay, such as tenure or education.
- The granular approach: This breaks the workforce down into smaller clusters, often called Similarly Situated Employee Groups (SSEGs). In large organisations, this can mean running dozens or even hundreds of separate analyses across job types or departments.
There are arguments for both sides. For example, a comprehensive approach may be more effective in testing how documented pay philosophies and policies are applied across an organisation. On the other hand, a more granular approach may be more useful in fighting pay equity lawsuits, which are usually argued at a granular level.
Remember: it’s a spectrum, not a binary choice. Most organisations will need to strike a balance based on their structure, workforce size and overall goals.
When is multiple variable regression analysis not a good idea?
While multiple variable regression analysis is a powerful tool, it’s not the right approach for every organisation. For example, it’s generally not suitable for companies with fewer than 25–30 employees, as the sample size is too small to draw meaningful conclusions. Even in larger organisations, the analysis may not be reliable if there aren’t enough employees in each comparison group.
You may also want to reconsider this approach if:
- Your organisation lacks sufficient data on pay factors
- Roles are highly specialised and not comparable
- Your organisation lacks the expertise to interpret results
- Your organisation is going through major change
- You have no clear framework or philosophy guiding pay
If one of the above describes your situation, you might want to explore other tools or start with simpler forms of pay analysis such as average comparisons or cohort analysis.
Multiple variable regression analysis in the pay transparency era
With the EU Pay Transparency Directive just around the corner, it’s more important than ever for employers to understand the pay dynamics within their organisations. Multiple variable regression analysis can play a key role in meeting the directive’s requirements — and in ensuring equal pay for work of equal value.
But beyond legal compliance, the directive is ultimately about giving employees clearer access to information about how pay decisions are made. That means it’s worth thinking not just about running the analysis, but also about how to share the results.
Our advice? Keep it clear and focused. While transparency is essential, overwhelming employees with data can backfire. Instead of publishing every detail, focus on:
- Explaining the factors you included in your analysis (and why)
- Sharing aggregate results, rather than individual-level comparisons
- Using visuals to make your findings easier to understand
- Creating opportunities for discussion and questions
Most importantly, be open about any disparities you’ve uncovered — and what you plan to do about them. This kind of accountability goes a long way in building trust, boosting engagement, and showing that you’re serious about fairness and retention.
Where to learn more about pay transparency
In the interest of full transparency, this article was inspired by a great post by HR and payroll expert Anita Lettink over on LinkedIn. For more insights on pay transparency, pay equity and more, check out Anita’s Equal Pay newsletter — or stick around on this blog!
.png)
.avif)
