Chapter 1 Introducing well-being data
Introduction to Understanding Well-being Data
This book seeks to advance understanding of the role of well-being in social and cultural policy, politics and research. It does this by focussing on the ideas, concepts and uses of well-being, as well as differences in types of well-being data. It was written primarily to offer practitioners a view ‘under-the-bonnet’ of data collection, analyses and uses to see how they actually operate, as well as what happens as a result of their very existence. Its accessible style aims to include students and a more general audience in discussions about data and those about well-being as two crucial issues of our time.
Understanding Well-being Data uses real-life examples, paying particular attention to the ways data are generated, analysed and used, to demonstrate how data practices respond to, and how they shape, society, culture, politics and policy. Its short and longer case studies make this an accessible learning curve, and one that is applicable to experts and novices of all sorts in all of our everyday lives. The book focuses on the uses of data in culture and society, and how they work as social policy, so that comparisons and contradictions are easy to see.
‘Following the data’ is a now-familiar phrase in the UK from its significant role in government communications about COVID-19. The phrase is important, because it demonstrates that the very idea of data is used to justify decisions and policies for the nation’s health and well-being. Many across the UK watched various press conferences in 2020 in which its prime minister and other advisors would refer to ‘the data’ as an objective thing that they were following, rather than various types of data and information that people learn how to use, deliberately collect and generate, and that they interpret and analyse.
The government broadcasts on managing the COVID-19 crisis also included graphs and other data visualisations. Some of these were designed to show a comparison across areas of the country to justify which were under restrictions and which were not. They were badly labelled, making them hard to interpret by those who are data literate, let alone ‘the public’ being broadcast to. Most people felt more alienated by these uses of data than comforted that they understood what the government was doing— and why. The last one of these press conferences that I personally saw, before finishing this book, was a few days before I was supposed to travel to spend Christmas with loved ones. The whole nation was told that this was no longer to be possible. We were told that the government had followed the data, but that the ‘science had changed’.
Of course, ‘the science’ had not changed at all. Instead, the decisions made, based on human interpretations of data about COVID-19, and other data about the economy and mental health, about schools and universities, about the inequalities of those who can work safely, and those who cannot, were all in a melting pot of pressures involved in decision-making at this level. It was the policy that had to change, not the science that had changed, and suddenly one set of data seemed more important than another to those in charge.
So, here we can clearly see that it is not that there is ‘the data’ as one indisputable thing, but these data are not neutral. By which we mean the data are not unbiased, nor impartial. They are collected, read, interpreted and presented and these processes involve many decisions. But, how can data themselves be biased? A good example of bias in data lies in the recent increase in algorithms that are trained using data to automate certain digital processes. Algorithms have actually been with us for centuries (an eighteenth-century happiness algorithm appears in Chap. 2). The word still refers to any form of automated instruction. The majority of algorithms are simpler than most people think and can be a single ‘if something is this, then do that’ statement that can then be actioned. Contemporary algorithms tend to be long sequences of these instructions. As you can imagine, with these many instructions and decisions, bias is likely to creep in.
One of the starkest instances of bias can be found in the search engine, which most of us now use all the time. It is a mundane part of our everyday lives that we don’t often think about. Search engines have been designed to learn to second guess what we are looking for, as they have a record of, or they ‘know’ all the searches we have made before this one, alongside all of everyone else’s searches. Safia Noble (2018) revealed how these guesses are biased in dangerous ways that are both racist and sexist. As recently as 2011, the first thing that would appear in searches with the term ‘black girls’ was a link to hardcore porn. You may try and explain this away as an algorithm prioritising some ads over others. Explaining these things away may be—in fact—a part of the problem, of course, when it comes to bias, sexism and racism. It therefore very much deserves attention.
Noble provides much more evidence than this example above, though. Noble shows a variety of ways that the search engine predicted the searcher was looking for derogatory images of black women, even apes, as well as pejorative character traits. Noble ‘followed the data’ to reveal how data practices are biased, but also revealed our own biases to us. People were shocked when Noble’s revelations were published. This shows us that not only are the search engines biased, but that we are. People are biased, in the way some want to believe that we live in a ‘post-racial’ society, and that we do not need to worry about racism any longer, when actually they are blinded to the fact they are consuming culture, through data, that are both biased and racist.
Data play a large role in society. Critical data studies, like Noble’s and throughout this book, where we ‘follow the data’ to see how it works in context, reveal truths about both data and society. We need to learn from these revelations about data to improve well-being and society.
Subjective and objective data
But what if we return to data used by politicians, surely this does not contain evidence of the same biases? A good example is ‘the poverty line’. When a politician talks about ‘the poverty line’, we think that this is an absolute thing. Not necessarily a real thing, like picturing people living under a power line, but that the line represents a measure from data which are objective.
Objective measures of poverty are objective by name, but they are not entirely neutral. So, does that mean they are actually objective? There is no measure of poverty that is conclusive: while it means not having enough resources to cover essential needs, this is a subjective valuation of the words ‘essential’ and ‘enough’. The subjective nature of the word essential has also gained prominence in the UK, as politicians have used it to avoid making clear decisions on what COVID-19 restrictions should entail— despite their data expertise. Instead, people are forced into making their own evaluations on what counts as ‘essential’ travel, work or food, and therefore what is lawful behaviour under parallel lockdown restrictions in different areas of the UK at different points in time.
Returning to the issue of poverty, in the UK and in most countries, ‘enough for essential’ tends to mean around 60% of the nation’s median income . This is classed as ‘relative poverty’, and it fluctuates. Absolute poverty is adjusted in line with inflation, rather than average living standards. These two different metrics can be used to paint two pictures of the same story, as a topical case demonstrates in Prime Minister’s Questions in UK Parliament.
The UK government refused to commit continued support of free school meals in the 2020 summer holidays. This policy decision about children’s well-being led to a high-profile campaign and a U-turn (that was repeated again in the Autumn). This controversy and debate included a wider discussion of the current government’s impact on child poverty. The leader of the opposition cited that 600,000 more children were living in relative poverty than in 2012 . Given that the Conservative Coalition took office in 2010, the implication here was that the Conservative governments of the last decade are responsible, and with serious negative effects. The prime minister retorted, ‘There are 400,000 fewer families living in poverty now than there were in 2010’ . How can one politician use poverty data to make a claim and the other use poverty data to claim the opposite?
How can data on poverty from the same time period, and cited in such an important setting as parliament, paint such contrary pictures? Each party leader chose slightly different timeframes within this ten-year period and they chose different poverty data. The leader of the opposition chose the poverty data and timeframe that told a story of the greatest negative impact, while the prime minister is thought to have possibly chosen a different timeframe and the other index to argue the exact opposite . These different indices aren’t intended to be fiddled with by politicians, but, actually, some measures will subjectively suit some arguments more than others. This does not mean that they cannot offer a more objective appraisal in other contexts, but as you can see, expert judgements can be subjective when deciding which objective data to use about people’s well-being, and in which context.
This use of poverty data is a good example of how well-being data have been used for centuries. Their collection and analysis are motivated by the need to track the health and wealth of society and evaluate the success and progress of social projects and policies. Indeed, these underlying assumptions have been the backbone of social science, statistical and policy work for the last 200 years. Yet, these data are not neutral or entirely objective. They can be used and misused as evidence in forums in which important decisions are made, and yet, we do not often ‘follow the data’ to appreciate these inconsistencies ourselves.
Understanding well-being data means looking at instances and inconsistencies of their use. It is generated to inform decision-making, which also means it can be used to hold others—particularly those in power—to account. It is also gathered on far smaller scales to appreciate the impact of aspects of society on us: our weight, our work, our children and their schooling. Major events, such as COVID-19, enable the power of well-being data to come to the fore. But these are data about us and are used to evaluate what to do next in a crisis. That is why everyone should feel able to access tools to help them better understand how this all works in society, should they want to; that is why this book tries to offer something for everyone.