Making research inclusive doesn't end once the data are collected. The same structural barriers that lead to underrepresentation in healthcare and research also shape the data we end up with: who is missing, whose measurements are less reliable, and whose experiences are poorly captured by the variables available. Standard statistical assumptions often don't hold under these conditions, and analyses that treat a population as uniform can obscure real differences between groups. How should we account for uneven data quality? How can models be specified to answer the question we actually care about, for everyone it concerns? And how do we judge whether conclusions transfer fairly across groups? As with study design, these are statistical questions at heart, but each carries ethical, practical, and financial implications.

This is the second of two half-day online workshops run in partnership between Cancer Research UK and Data Science for Health Equity (DSxHE), bringing together methodologists, biomedical researchers, bioethicists, funders, and public contributors to work through these questions, and to keep building the practical, statistically grounded guidance that's currently missing.

Workshop 2 focuses on Equitable Statistical Data Analysis, covering:

• Data quality — how differences in missingness and measurement error across groups can be accounted for, and where methods such as imputation help, as well as the limitations that come with them

• Model design — how to specify a model that targets the question of interest, including careful handling of confounders and explicitly modelling heterogeneity, for example allowing a treatment effect to vary by demographic group rather than assuming one effect fits all

• Generalisability and fairness — how to evaluate differences in model performance across groups, using uncertainty quantification, validation, and group-level audits, and how to navigate the tensions this can surface between data constraints, cost-benefit trade-offs, and the transferability of findings

  
  

The workshop will include three expert panel discussions with live Q&A, and breakout discussions for attendees to dig into the challenges alongside panellists.

Workshop 2 builds on themes from Workshop 1: Inclusive Study Design, but attending the first session isn't required to take part

Schedule, panellists and chairs to be confirmed

Practical info

• This event is free and takes place entirely on Zoom

• Open to methodologists, biomedical researchers, bioethicists, funders, PPIE contributors, and anyone involved in writing or reviewing grant applications for inclusive research.

Any questions?

Please direct any queries to info@datascienceforhealthequity.com.