BLOG - Getting to grips with administrative data analysis: My experience of the IADRA course
Having recently joined SCADR, I have been both excited and overwhelmed by the prospect of working with administrative data. Administrative data differs from other forms of data collected for research, such as surveys in the social sciences, and comes with its own benefits and challenges. To learn more about administrative data, its use in research and the practical approaches to analysing large quantities of this particular data type, I recently took part in the Introduction to Administrative Data Research & Analysis course (IADRA).
The course, delivered by SCADR, covers a range of theoretical, practical and ethical aspects of working with administrative data. It focuses on data in a UK context, across the four nations, so it’s particularly useful to researchers working in and with those contexts. It is structured into learning sessions throughout four weeks and contains a mix of self-paced learning using videos and completing a practical R coding script. Each week, participants can meet online for a catch-up, attend a Q&A session with administrative data researchers, and follow a live tutorial of the weekly practical coding exercise. For the practical sessions, participants work with synthetic datasets which imitate real-world data but do not contain actual information about individuals.
Week 1: An Introduction to Administrative Data
Week 1 started with a welcome session. For course participants who were able to come to Edinburgh, SCADR hosted an in-person welcome event followed by a networking lunch. The first set of video lectures covered a comprehensive introduction to administrative data including its diverse functions and its benefits for use in research. The live sessions included getting-to-know participants and a Q&A session with SCADR researchers on their wide-ranging areas of administrative data research in a Scottish context. The practical session of week 1 focused on data manipulation using the tidyverse package in R.
Week 2: Data Provenance and Record Linkage
The second week focused on the complexities of data provenance and record linkage. The video lectures highlighted the importance of understanding how data errors and irregularities can occur and introduced several strategies for addressing these challenges, such as consulting with researchers who have previously used the dataset, engaging with data custodians, and understanding how policy influences data collection and coding. The process of linking records was a key focus of another lecture, which differentiated between deterministic and probabilistic approaches to record linkage.
The lectures also covered the Researcher Access Service, a digital service provided by Research Data Scotland (RDS), which helps researchers secure access to linked administrative data, as well as an introduction to ADR Northern Ireland. For the weekly practical session, participants learned how to handle and join data frames from multiple sources.
Week 3: Ethical Considerations and Data Visualisations
In week 3, the course discussed the ethical considerations within the legal foundations of administrative data research and the importance of engaging with the public and communities to build trust and transparency in data research.
The weekly practical coding introduced participants to data visualisation in R, which can help make large and complex datasets easier to interpret and outputs more impactful.
Week 4: Analytical Approaches and Challenges
In the final week, the lectures showcased methods for analysing administrative data, using real-world case studies to illustrate methods including panel data analysis, cross-over designs and latent class analysis. A short series of videos provided a comprehensive overview of the analytical techniques commonly employed in administrative data research and their potential and limitations.
In the final Q&A session, SCADR researchers reflected on their experiences with accessing and analysing administrative data. They emphasised that successful administrative data research requires patience, persistence and collaboration. In the practical session, participants worked on conducting inferential data analysis with a focus on modelling under consideration of statistical disclosure control. This final hands-on exercise tied together many of the skills and concepts covered throughout the course.
Concluding thoughts
The IADRA course provided me with a comprehensive overview of the use, benefits and challenges of administrative data and increased my confidence in handling and analysing this particular type of data. I especially enjoyed the mix of general lectures and specific research showcases which covered foundations and illustrated key concepts using actual research projects. The self-paced setting of the lectures enabled me to go back and forth and allowed me flexibility when engaging with the course contents. Personally, I tried to find time to complete the practical coding sessions prior to the live tutorial because I could figure out the coding on my own and have questions prepared for the tutorial.
Despite some technical glitches on this occasion, I strongly recommend the IADRA course to anyone working with or intending to work with administrative data for research. Some of the video lectures were longer than anticipated, so I would advise planning plenty of time to engage in and complete these materials, especially since it may be challenging to watch pre-recorded video lectures for long periods of time. Thankfully, the lecture slides are provided, which helped when revising the covered topics.
My personal highlights of the course were the weekly practical coding sessions and the networking event in Edinburgh. At the end of the four weeks, I have gained a deepened understanding of administrative data research and thank the many people who contributed to this helpful course.
To register for the course or find out more, visit IADRA webpage.
This article was published on 11 Dec 2024