A decade of Understanding Society

What makes Understanding Society special?
Understanding Society gathers information about its participants every year, which gives an incredible richness to the data that is collected. It covers a broad range of topics – work, family, wellbeing, health and social activity – and that makes it interesting to people from a range of subject areas or policy perspectives. It has a large sample of 40,000 households, and the advantage of that is means you can dig into groups who are traditionally going to be too small for most standard surveys.

Finally, it is a household survey – it is not gathering details about one person, but a household. That opens up interesting questions about understanding the circumstances of one person by looking at the whole household they are living in.

What have been the most interesting findings from the study?
If you look back over the last 10 years, there has been a huge interest in understanding inequality and how it is changing, and the impact of inequality on people’s lives. There has been work into how children growing up in poorer households fare upon their entry into the labour market, and how different groups ‘sink or swim’ after periods of disadvantage in terms of their employment. Some people transition back into work quite quickly after losing their job, whereas others struggle. Understanding Society has been helpful in tracking that. The sample sizes have shone a light on different minority ethnic groups too.

What has been the study’s impact?
Understanding Society has a policy unit that is specifically tasked with outreach. That has underpinned some very interesting developments in how research can feed into policy. One of the challenges of this study is the data can be complex to use, and the policy unit has been successful in helping policymakers. Data from the study was important in helping to underpin some of the changes that were made to pension provision and auto-enrolment, for example. It was also used by the Treasury in assessing the impact of Covid-19 on working households’ incomes.

How has it adapted as a result of the pandemic?
Understanding Society has been trialling new ways of collecting data for a long time. It began as a paper-based interview study. It was already collecting data using computer-assisted personal interviewing, and when Covid-19 hit was using a mixture of online and face-to-face research. The team already had the technology set up, and it was simply a case of going out and getting the people who weren’t doing the survey online to start with. Before Covid-19, the data collection was annual, but for the Covid-19 study it has been done monthly, which has meant we can get data on the impact of the pandemic out to the research community very quickly

How do you see the survey developing in the years ahead?
There is an immediate question about transitioning back from Covid-19 and thinking about what has changed: is there anything we can learn from Covid-19 and the way we have done our monthly surveys that we might want to keep going with in the future? There is an ongoing issue of what questions we ask and what data we collect, and whether we do this in different ways, and there is always interest in how you can link survey data with routine administrative data, as you reduce the burden on participants and add to the richness of the data.

Why are longitudinal studies so important?
Longitudinal studies are well-suited to help us understand change. Unlike single research projects that tell you a lot about people’s lives at a single point in time, longitudinal studies are following people over time and collecting data. They are therefore building an incredibly rich picture of people’s lives. It is like the difference between having a single picture of a family and having a photo album of that family over the years showing how they are changing.

Is there a space for longitudinal research in a market research industry shifting towards quicker insights?
There is a need for both. The value of longitudinal data is it allows you to address questions you cannot get from cross-sectional data collection. For example, looking at how events in early childhood shape outcomes years later on, you can only do that robustly through a longitudinal study, as you are otherwise you are in the realm of asking people to recall events from long ago with all the biases that entails. In an intellectual sense, there is a clear need for longitudinal data.

Understanding Society in Covid-19 managed to collect high-quality, robust data, link it back to earlier data about the same people and get it out to the research community quickly. I don’t think it would have been possible to do that to the same quality threshold if you were starting from scratch.

What advice would you give researchers based on your experience of the study? 
Be really clear why you are doing it. What are the research questions you want to answer? That determines who you need to collect data from and what you need to collect. The challenge now is what is the most efficient way of collecting data. There is a push towards online, but we know that the response rate will be lower if you are entirely reliant on it.

You need to think carefully about the design of the first few rounds of data collection, and how you can maximise the feeling that people belong to the study. And do not let the desire to do it cheaply get in the way of the breadth and diversity of your sample.