Supporting the Office for National Statistics (ONS) with new statistical methods

Alexander Amaral-Rogers

The Methodological Research Hub is responsible for researching new statistical methods to see whether they can be used in the production of official statistics. The Hub is part of the Methods and Quality Directorate at the Office for National Statistics (ONS).

We have recently been looking into the use of a technique called Structural Equation Modelling (SEM). This is a flexible type of statistical model that combines regression analysis and factor analysis. It includes at least one latent, or “hidden”, variable based on several observed and directly measurable indicators.

The work we have done is a continuation of research that the University of Southampton had previously done on our behalf. We’re taking it a step further by applying the theory to real-life case studies, while quality assuring the work with the help of our academic partners.

So why were we doing this? Well, part of the reason is to look at how well the method serves our needs in the ONS generally. But we’re also using it to support the Social Statistics Admin First (SSAF) Housing team with their project around developing property floor area statistics based on administrative data.

This is an important project as having good measures of floor area can help us to estimate how overcrowded buildings and properties are. But the team needed more information on the quality of the data. By supporting the project we have been able to provide the insight needed as well as giving us the opportunity to develop our research for the benefit of the wider ONS.

I had no previous knowledge of housing statistics or the data sources the SSAF Housing team were using. As such, it was important for me to understand more about this to be able to build the structural equation model.

I have a background in working with SEM in academia and sociology, but using the method for estimating measurement error in different administrative data sources was new to me. So this piece of work involved using it for a completely new purpose and I worked closely with the team to agree the model that suited the data best.

During the process we quality assured each other’s work through regular check-ins. This helped both sides increase the quality of our respective methods. We also agreed to publish our papers on the same day, so the final work was fully joined up for readers. We provided links between each paper for users to follow. We published both the SEM paper and admin-based floor area statistics research on 26 October 2022. We are currently talking to other teams about interesting ways to continue to develop this method.

If you’re interested in knowing more about the research we’re doing or would like to work with us in the Methodological research hub, please contact us at

Nina Sommerland
Alexander Amaral-Rogers
Nina Sommerland is a Methodologist in the Methodological Research Hub in the ONS. She holds a PhD in sociology and before joining the ONS in 2021, she worked as a researcher in Sweden, Belgium and South Africa.