New to the game: statistics, methods, research and me
Six months ago I joined the Methodological Research Hub in the Office for National Statistics’ Methodology Division, with a focus on developing methods for producing official statistics using administrative data.
Coming from a psychology background, my first few weeks were quite difficult. I felt way behind in my statistical knowledge, and I couldn’t imagine ever getting to a point where I understood anything. However, it soon became clear that a lot of the skills and knowledge I had developed during my undergraduate and master’s degrees were transferable.
One of my main projects involves developing Government Statistical Service-wide quality guidance to help those using administrative data for statistical purposes. The guidance aims to help the user think about and address the extensive range of quality issues associated with this kind of work. The content of this was all new to me – I’ve had to learn a lot about quality, administrative data, and statistical processing. However, skills I already possessed such as effective searching strategies, gap analysis, and critical analysis have all been a big help in terms of my approach.
It has also been really interesting reaching out across the Government Statistical Service (GSS) with a collaborative approach, trying to understand the challenges that other statistical professionals encounter. Collaboration is central to all the work we do in our hub, where we engage with academics, other National Statistical Institutes, and other government and non-government organisations. This feature of my role is something I enjoy, as it means I get to meet and work with a wide range of interesting people with different perspectives – I’m always learning something new.
I’m also working on researching and testing a statistical method that combines latent class modelling and multiple imputation. The method is used to measure and estimate classification uncertainty across several administrative datasets, and to yield imputed classification where there is person-level uncertainty.
This is quite a novel approach and draws heavily on my research skills, methodological knowledge, and collaboration and communication skills. Whilst my existing statistical knowledge has provided a good basis for this kind of research, I have had to develop this beyond what I learned at university, and the biggest challenge has been learning to code in R!
As well as these projects, I have been helping to organise the this year’s Government Statistical Service Methodology Symposium (taking place on the 2 and 3 November 2021) and horizon scanning for new qualitative methods. The latter is really exciting to me, as it draws upon approaches and research I am familiar with from my psychology background!
Working at Office for National Statistics (ONS) has helped me to see statistics in a different light. At university, they were something we did a bit begrudgingly as a small part of larger assignments. At ONS, however, they’re the heart of what we do. They aren’t just passive numbers, but important outputs with real utility and impact on policy and people’s lives.
This aspect of the work is something I really enjoy. It means I must think critically about statistical methodology to identify gaps and opportunities to address them, with the ultimate aim of keeping up to date with new methods, theory, and data sources to ensure decisions made using such sources are based on sound data and methods. So far, this role has definitely been a challenge, but a positive one that I feel builds upon my previous experience and has ultimately pushed me to develop, learn, and grow.
If you would like to know more about the research methods we have mentioned we are exploring in our article, or that is ongoing in our Methodological Research Hub or across the Methodology division, or if you are interested in working or collaborating with us, please contact: Methods.Research@ons.gov.uk.
We are especially interested in research relating to administrative data, alternative data sources and official statistics.