It has recently been reported that the people in the UK are getting happier. Based on data from its Personal Well-Being Survey, the ONS claims that the proportion of people rating their life satisfaction as 7 or more out of 10 rose from 75.9% to 77% between 2012 and 2013 (http://www.bbc.co.uk/news/uk-23501423). How meaningful is this data?
The data is collected by asking people to answer the question, on a scale of 0-10, 'Overall, how satisfied are you with your life nowadays?'. 0 indicates 'not at all satisfied' and 10 indicates 'completely satisfied' (http://www.ons.gov.uk/ons/rel/wellbeing/measuring-national-well-being/personal-well-being-in-the-uk--2012-13/sb---personal-well-being-in-the-uk--2012-13.html#tab-Methodology). Respondents are also asked, 'Overall, how happy did you feel yesterday?', again on a scale of 0-10. The data shows that in 2013 72% of people rate their happiness as 7 or above, compared with 71.5% in 2012 (http://www.ons.gov.uk/ons/rel/wellbeing/measuring-national-well-being/personal-well-being-in-the-uk--2012-13/sb---personal-well-being-in-the-uk--2012-13.html#tab-Overview).
First, people don't really know how happy/satisfied they are. It is cognitively very difficult to give an overall assessment of life satisfaction. Think about your own answer to the first question above. Most people will say 7 out of 10.
Second, people are liable to give normative responses to questions about satisfaction with their lives or with major aspects of their lives. Admission of dissatisfaction with one’s life can sound like admission of personal failure or the assignment of blame to others. Generally, features of the survey design, such as who asks the questions and how, or how the questions are worded, can systematically affect individuals responses. This is evidenced in the ONS's observation that 'Higher average ratings for the life satisfaction, worthwhile, happy yesterday questions and a slightly lower average for the anxious yesterday question were provided by respondents interviewed via the telephone compared with those who are asked personal well-being questions face-to-face.' (http://www.ons.gov.uk/ons/rel/wellbeing/measuring-national-well-being/personal-well-being-in-the-uk--2012-13/sb---personal-well-being-in-the-uk--2012-13.html#tab-Methodology). It seems that people may be reluctant to admit to being unhappy to a stranger over the telephone.
Third, and particularly relevant to policy, is the statistical difficulty of identifying the relationship between overall satisfaction and specific causal factors. Even if we accept that it is meaningful to conclude that overall happiness has increased, the causes of this increase may be almost impossible to identify using survey techniques.
Another problem is that the measurement scale not very sensitive, and differences between life satisfaction scores may not map directly onto real-life differences between individuals' levels of well-being. For example, there is a big difference between rating one's life satisfaction as 3 and rating it as 8, and someone whose life satisfaction drops from 8 to 3 obviously has some serious problems. But what can we conclude about a drop from 8 to 7? The problem is more acute in comparing between persons. Due to differences between individuals - in their expectations, their level of optimism, and even what they interpret 'satisfaction with life' to mean, we cannot meaningfully compare one person's happiness with another using life satisfaction data.
However, the purpose of a large-scale survey such as this is not to compare between individuals. Rather, it is to aggregate well-being scores and compare them across groups of people and within groups of people across time. But there is an important reason why it might be problematic to do this. This is that survey data does not directly tap into a commonly-held concept of 'life satisfaction' because of the differing 'norms' and 'expectations' of respondents. A good example of this is the case job satisfaction. Research has found that there is a u-shaped curve when average level of job satisfaction is plotted against income quintile. That is, on average those in the lowest income are the relatively satisfied with their jobs, with those in middle income groups are on average relatively less satisfied, and average satisfaction increases again amongst those in the highest income quintile. From this data we might conclude that on average those on low incomes have better, more satisfying jobs than those in middle income work. But that would be too quick. Brown et al. argue that many workers on low incomes are in fact in low quality jobs, and are only expressing satisfaction against a bench-mark on very low norms and expectations. They are, in effect, making the best of a bad situation. Brown et al. support their argument with data on New Labour's years in government, which show that job satisfaction increased while according to subjective measures of work effort and stress, there was little or no improvement in job quality; and with qualitative data which shows that low-paid workers have extremely low norms and expectations about work (see Brown et al., 'Job quality and the economics of New Labour: a critical appraisal using subjective well-being data', http://carecon.org.uk/QM/Conference%202008/Papers/Brown%20CJE.pdf). If this argument is sound, then, as Brown points out, 'it would be incorrect to interpret an aggregation across very different groups as tapping into the absolute value of true underlying job quality, given that it is implausible to argue that, for example, factory workers and merchant bankers have identical respective norms and expectations regarding work.'
A similar argument could be made with respect to the apparent increases in life satisfaction and happiness between 2012 and 2013. It may be that people's norms and expectations have changed, rather than that their objective life circumstances have improved. Perhaps people are adapting their expectations given the tougher economic climate. Perhaps they are getting used to the post-recession landscape, and have revised their expectations about their life prospects accordingly. Perhaps they are comparing themselves to others, in Britain or the wider world, who are worse off, and feel more satisfied as a result. If so, it may be that people's life circumstances are actually getting worse, while their life satisfaction is improving. This phenomenon, known in economic theory as 'adaptive preferences' is well documented. A good example is when people win the lottery. Many people think that winning the lottery will bring them great happiness. But due to adaptation – the fact that when people are in a state for a period of time the initial pleasure or pain of that state is attenuated – winning the lottery doesn’t bring people as much happiness as they think it will. People adapt their norms and expectations to their circumstances, so we cannot conclude from small improvements in happiness and life satisfaction data that people's circumstances have improved.
The ONS does not only make comparisons between the population across different times, it also makes comparisons between groups within the population. For example, it finds that on average women have higher life satisfaction than men. But we can't conclude that women lead better lives than men. It might be that their expectations are lower. Indeed, other evidence suggests this might be the case.
For these reasons, subjective well-being data such as that collected by the Personal Well-being Survey is not an appropriate basis to conclude that all is well. And it should not be used to supplant proper public debate, using a broad range of evidence, on well-being, living standards, and the impacts of austerity and the recession.