SAIL Databank news. Published 01704/2025
A research methods paper, which was started as part of the Population Data Science Summer Internship Programme in 2023, has recently been published in BMC Public Health. This collaborative work by SAIL Databank researchers at Swansea University highlights the valuable opportunities and contributions of the summer internship programme and its role in producing impactful, high-quality research and development of the next generation of Researchers & Data Scientists.
Why this study is important
This paper serves as an important resource for researchers using the Welsh Index of Multiple Deprivation (WIMD) in health data studies and is relevant to most research projects using SAIL. The team created a single citable resource for researchers using the WIMD in research, along with highlighting the effect of ignoring a commonly overlooked source of endogeneity bias, and offers a methodology to mitigate this bias.
By addressing these challenges, the paper offers valuable insights for researchers working with the WIMD to study various domains, including:
- Health.
- Education.
- Housing.
- Physical environment.
- Income & employment.
- Community safety.
- Access to services.
This work reinforces the importance of rigorous research methods in data-driven studies and strengthens the foundation for evidence-based policy and practice.
How the study utilised SAIL data
The study linked WIMD 2019 scores provided by the Welsh Government to 2.76 million individuals in the SAIL Databank using their Lower-layer Super Output Area (LSOA) of residence. Researchers explored how different methods of calculating WIMD scores affected estimates for a health study using routinely-collected data. One method removed the health domain, while another used only the income domain. Statistical techniques, including Spearman’s correlation and Cohen’s kappa, measured how WIMD rankings changed across these methods. Additionally, binary logistic regression was used to assess how these adjustments impacted the association between WIMD quintiles and diabetes mellitus prevalence, providing age-adjusted odds ratios and confidence intervals. The study found a statistically significant difference between the prevalence estimates, depending on the choice of WIMD method, highlighting the effect of ignoring this potential source of bias.
Researcher insights
Dr Hywel Turner Evans, Senior Data Scientist & User Support Lead at SAIL Databank, shared his thoughts on the impact of this research:
“If you use the Welsh Index of Multiple Deprivation (WIMD) in research studies, then this paper is for you. We created a one-stop shop for researchers to summarise information on the domains and indicators which comprise the WIMD 2019, along with evidencing that studies related to WIMD domains should consider recalculating the original score.”
We also spoke with 2023 summer intern Shamsudeen Mohammed, who reflected on the significance of this research methods paper:
“We wrote this for anyone interested in using the Welsh Index of Multiple Deprivation (WIMD) in their research. It provides practical guidance to assess whether including LSOA ranks based on all the domains of WIMD in your analysis may introduce bias, depending on your outcome measure, and suggests methods to identify and address any potential biases.”
Following the summer internship, Shamsudeen now works as a Research Fellow at the London School of Hygiene and Tropical Medicine.
Read the full paper here.
Interested in developing your skills and experiences and making an impact like Shamsudeen? Applications for the 2025 Population Data Science Summer Internship Programme are now open. Find out more and apply here.