Dr Raheela Awais, Senior Lecturer at the University of Liverpool, explains why an intersectional lens in data analysis is essential.

Data tells stories but only if we know how to ask the right questions. 

At Advance HE’s 2025 Equality, Diversity and Inclusion Conference, my colleagues Dr Maria Limniou, Professor Gita Sedghi and I had the pleasure of sharing our findings on “Amplifying Diverse Voices: An Intersectional Approach to Curriculum Inclusivity in Higher Education.” 

It was a meaningful moment, not just for us personally, but as part of a wider conversation on what intersectionality is and why the intersectionality lens in data analysis really matters. One key takeaway from both our research outcomes and the conference discussions is that applying an intersectional lens isn’t only a nice-to-have, it’s essential if we truly want to understand who’s being left behind and why. We have to look beyond averages and see deep under the layers to unravel the realities that people face. 

After all, leaving no one behind isn’t just a phrase; it’s a principle rooted in the Sustainable Development Goals (SDGs). And that starts with how we look at data. 

So, what is intersectionality really? 

Coined by legal scholar Kimberlé Crenshaw (1997), Intersectionality describes how different aspects of a person’s identity, such as race, gender, socioeconomic status, disability, and other individual characteristics, interact to create unique experiences of advantage or disadvantage. As Anne Sisson Runyan (2018) points out, this is not a theoretical luxury; it’s a necessary framework for recognising that people don’t live single-issue lives. Intersectionality helps us track how systems (like education, health, or law) can reproduce inequality when they fail to see beyond a single demographic variable. 

In practice, it’s the difference between thinking “this programme isn’t working for women” and asking, “why is this programme failing Black disabled women in particular?”. 

Why an intersectional lens in data analysis matters 

Let’s say student attainment data shows that disabled students are underperforming compared to their peers. That’s a starting point, but it’s not the full story. Which disabled students? Are they neurodivergent? From racially minoritised (RM) backgrounds? Studying part-time? Experiencing financial hardship? These intersecting factors significantly shape their educational journey. Without this deeper analysis, we risk masking real barriers and overlooking the most effective solutions. 

Kapilashrami and Hankivsky (2018), writing on global health, argue that without disaggregating data along intersecting identities like race, class and gender, we end up with one-size-fits-all policies that disadvantage the most vulnerable. 

Their study gives the example of cardiovascular disease to illustrate this point, which is the leading cause of death worldwide. While disparities by geography, socioeconomic status, race and sex are well documented, less attention is paid to how causes, progression and outcomes vary among different subgroups. Although global death rates have declined over the past decade, age-specific reductions have been greater for men than for women, reinforcing the need for intersectional data analysis. This insight became especially evident during the Covid-19 pandemic. Lara Maestripieri’s work (2021) showed how immigrant women of colour, often in frontline jobs, bore the brunt of both virus exposure and economic fallout, facts hidden in average national statistics. 

In short, intersectionality in data analysis isn’t about complicating things. It’s about telling the truth. 

What we discovered through our disaggregated data 

While analysing data from an online survey exploring student perspectives on curriculum inclusivity, our initial focus was on statistically significant patterns. One unexpected finding stood out: RM students reported a stronger sense of belonging and closer staff connections than white students – a trend that contrasts with much of the existing research. 

Curious to dig deeper into this anomaly, we began to disaggregate the data by intersecting identities. That’s when the disparities emerged. 

Among white students, those who were disabled, or disabled and working, or disabled and mature, reported markedly lower levels of connection to staff. The aggregated data had masked these nuanced experiences. 

As we examined RM students more closely, the sample size naturally reduced from 64 students overall, to 18 who identified as both RM and disabled, and ultimately to just four once working status was also factored in. Although these small numbers limited statistical generalisation, the open comments revealed powerful insights: stories of exclusion, resilience, and unmet needs that would have remained invisible in the aggregate. 

These insights were only possible because we chose to go beyond the surface of aggregated data. Disaggregation allowed us to see and hear those students often rendered invisible and to shape more precise, meaningful interventions in response. 

The challenge of small sample sizes 

As the earlier examples show, applying an intersectional lens by disaggregating data has its limits. The more variables we add, the smaller the subgroups become, often to the point where the data is statistically unreliable or risks breaching confidentiality. 

Many institutions follow a standard rule: if fewer than 10 individuals fall into a category, the data can’t be reported. This protects privacy, especially in small or identifiable populations. But while ethically important, it comes with trade-offs: 

• some groups remain invisible in public data, simply because their numbers are small
• policy-makers may overlook these groups, not due to lack of need, but lack of visibility
• silence in the data can be misread as a sign that there’s no issue to address. 

Let’s not allow small numbers to justify inaction 

Here are a few recommendations to apply an intersectional lens in the educational sector:  

1. Use pooled data over time: where sample sizes are too small for annual reporting, consider aggregating data across multiple years to identify trends without compromising anonymity.
2. Complement quantitative data with qualitative insights: using mixed methods, such as interviews, focus groups and case studies, can enrich understanding and validate patterns observed in small datasets.
3. Adopt flexible reporting practices: share data even when subgroup sizes are small, but include caveats.  Emphasise patterns and directionality rather than relying solely on statistical significance.
4. Practice transparent reporting: when data cannot be disaggregated, explicitly acknowledge these gaps and their implications. Transparency builds trust and encourages systemic improvements. 

Seeing fully 

To build systems that truly serve people, we must start by seeing them fully in all their complexity. That means embracing diverse identities, even when it makes our data less tidy. One-size-fits-all solutions fall short. Real change demands approaches that reflect the lived realities of those we aim to support. Tailored, inclusive strategies aren’t just preferable, they are essential for lasting impact. 

How are you making sure intersectionality stays part of the picture, even when the data gets small? 

Raheela Awais is a Senior Lecturer and Director of the Integrated Master of Biological Sciences in the School of Biosciences at the University of Liverpool.  You can find out more about her work on inclusive and personalised teaching in lab-based complex environment and intersectionality here.