It’s a great time to be involved with health studies at a time of significant public health challenges: an ageing population, health inequalities and multiple disadvantage, the challenges of obesity, of poor mental health and so on. And all against the backdrop of the magnifying impact on those issues caused by the COVID-19 pandemic.
Data collection, and in particular social surveys, are crucial to measuring and monitoring the impact of health issues and behaviours.
I sometimes rather unfairly split those involved with data collection into “traditionalists” and “modernists”, when in reality we are on a continuum. There are those who lean more towards favouring traditional data collection approaches, that advocate for the traditional approach to the census and typically embrace the social survey including conducting face-to-face interviews. They believe there has been and remains good logic to why we use our existing data collection methods.
And then there are those that believe we need to change and modernise our approaches as people live their lives differently, for example with an increased online existence. They lean towards making more use of administrative sources and that there may be great opportunities ahead in harvesting less structured or unstructured data using the power of technology to inform people of what is happening right now, not in two years’ time.
When face-to-face interviewing was paused in March 2020, many of us inevitably asked: is this the turning point, a moment of significant change, for social surveys? Is this when administrative and big data sources really come to the fore? Is this when multi-mode survey approaches come good and face-to-face interviewing is set aside as no longer necessary or practical?
It appears that the answer to that question is to be a resounding “no”. It’s probable that face-to-face interviewing will return for most of the complex national surveys, including the health ones. There is even advocacy for a traditional census in five years’ time. The need for longitudinal data has become more important, both during and no doubt after the pandemic, not less.
Did you see that wonderful “word cloud” of social media data telling us how many COVID-19 cases there are in real time? No, nor did I. With so many asymptomatic cases, we know that we can’t get accurate data on prevalence from administrative data, anymore than we currently can for obesity or mental health issues, because not all cases are diagnosed.
Whatever our individual preferences and beliefs might be around data collection in the future, we should be prepared to turn to the evidence and what our pandemic experience has told us about our data collection methods at this point in time.
When the numbers and trends in infection cases became so important, where did we turn? The ONS COVID-19 Infection Study (CIS), a whopping great national survey of the population, conducted face-to-face. Why? Because we needed to be sure or as sure as we can be of the data. We have supplemented it using other sources, of course, but there is good reason why the National Statistician turned to the CIS as the primary underpinning source.
Health-focused social surveys have other advantages too - you can ask the exact questions you want to ask of the respondent, you can tell broadly how accurate the data is and adjust for biases. And more to the point, the turnaround in the data collected throughout the pandemic has been at times fast, countering one of the perceived big weaknesses of the social survey - slow turnaround. The Opinions and Lifestyles Survey being a very good example of this.
The amount of innovation on national surveys in adapting to the pandemic has been great and I would be keen to see this continue post-pandemic. A lot of it has happened in close collaboration with customers and users. In many ways, there’s so much we can be proud of, from our commissioners to academics, in rising to the challenge of the pandemic from data collection to analysis and dissemination.
It’s interesting to see that the so-called “knock to nudge” approach is producing quite solid response rates in most cases across surveys and that it is (only slightly) cheaper - is this an alternative we will continue to use post-pandemic? My expectation is that social survey practitioners need to be able to offer commissioners a suite of products going forward depending on the study, the uses to be made of the data and available budgets.
We’ve also seen that some technological innovations are not needed in some contexts - e.g. we have tested video interviewing and found that there is limited interest among participants for it if telephone interviewing is available on some surveys. It does seem to have a place in longitudinal studies and we are using the European Social Survey to see what happens if there are no alternatives in a Postcode Address File (PAF) based survey.
In many ways though, the pandemic itself has shown the real importance of surveys - such as the CIS, but also the Mental Health of Children and Young People survey, our longitudinal studies (like Understanding Society) and so on - in understanding the pandemic and responding to it at pace. Relevance is important in surveys and, while social data is not stock market data and we may only need annual data most of the time, we still need to get quicker at delivery of results, however complex and inter-linked our data becomes.
Most of the methodological weaknesses in terms of the different survey modes have been identified and proven throughout this pandemic, such as: the lack of good quality sampling frames (telephone and web), lower response rates, mode effects, alternative modes that are not quite as cheap yet as we had hoped as a community. We are still looking at the data that has been produced by alternative modes but first impressions are that some groups are less likely to respond (say renters, compared with home owners). This can be corrected for, but can make samples less efficient, and the issues of getting the harder to reach groups to participate are still paramount irrespective of mode.
Good policy-making is essential in a democracy and trusted, relevant data is crucial to help inform decisions. It’s a time for us to be bold about the importance of our research studies in business cases and cost-benefit analyses and seek new and additional funding streams. Health is the greatest example of this at this very moment with evidence-based policy-making essential. If policymakers are going to succeed in helping Government level up the UK, then data and research is vitally important and we must be bold in conveying that to those in power.
This blog was taken from the keynote speech at the Health Studies User Conference 2021, hosted by the UK Data Service, in partnership with UCL and NatCen.