Evaluative measures: The life satisfaction approach

In the first of these, people are asked how they feel about their life these days. The most common question is ‘Overall, how satisfied are you with your life nowadays?’Footnote ⁶ To answer it, individuals have to choose a score between 0 and 10, where 0 means ‘Not as all satisfied’ and 10 means ‘Very satisfied’. This is a question on life satisfaction. Alternatively (which gives very similar answers), people are offered a continuous line with the same answers at either end, and they are asked to mark where their answer lies in between. The results of using a ‘visual analogue scale’ like this are very similar to those from whole-number answers between 0 and 10 – both in terms of the score recorded and the factors that explain it.Footnote ⁷ The results are also similar when, the phrase ‘satisfied with your life’ is replaced by the phrase ‘happy with your life’.Footnote ⁸

In another variant, people are asked the following question, known as the Cantril ladder question:

The World Happiness Report uses the answers to the Cantril ladder question given when people reply in the Gallup World Poll.Footnote ⁹ This poll covers over a thousand people a year per country, in nearly every country. In surveys where people were asked both about life satisfaction and the Cantril ladder they were very highly correlated.Footnote ¹⁰ So for ease of language we shall describe the results of the Cantril ladder as ‘life satisfaction’.

The Gallup World Poll shows the huge spread of life satisfaction both within countries and between countries. For illustration, Figure 1.1 gives the spread in the United States and in India. Some key facts emerge:

• In each country there is a very wide spread of life satisfaction. This corresponds to our own experience, across the range of people we meet.

• The average of life satisfaction is much higher in the United States than in India.

There is indeed a huge variation of life satisfaction within any country, and this within-country variation actually accounts for 78% of the overall variance in life satisfaction across the world’s population.Footnote ¹¹ Exploring the within-country variation provides key information about the personal factors that matter most in human life (see Chapter 8). For example, as we shall see, income differences explain some 2% of the variance of life satisfaction within countries – much less than the 20% explained by mental health, physical health and human relationships. (Somewhat surprisingly, average life satisfaction in most countries is very similar for men and women.)

Figure 1.1 Percentage at each level of life satisfaction (0–10): United States and India

Source: Gallup World Poll 2016–2018, Cantril ladder

There are, however, also huge differences in average life satisfaction across countries. This is shown in Table 1.1,Footnote ¹² which reveals the extraordinary variation in the human condition around the world. Average wellbeing is evaluated at 7.5 or above in four Scandinavian countries and at 3.5 or below in war-torn countries like Syria, Yemen, Afghanistan, the Central African Republic and South Sudan. Explaining this variation provides many clues to the social factors that matter most in human life (see also Chapter 8).

One obvious question is, ‘Has wellbeing increased over time, as living standards have risen?’ The answer is in some cases yes and in others no. For the world as a whole, average wellbeing rose between the 1970s and 2007.Footnote ¹³ Since then, however, it has stagnated.Footnote ¹⁴ And in some countries wellbeing has not risen since research began. This is true of the United States (since the 1950s), West Germany (since the 1970s) and China (since 1990).Footnote ¹⁵ Figure 1.2 shows the figures for the United States.

If only we knew more about trends in wellbeing in earlier periods.Footnote ¹⁶ There is certainly strong evidence that most of the external conditions of life are now better than in most of human history. As Stephen Pinker has shown,Footnote ¹⁷ there is less violence, better human rights and so on. So in most parts of the world, life is probably as good now as it has ever been.

Figure 1.2 Average happiness in the United States (on a scale of 1–3)

Source: AIPO (1956–1970); NORC (1963–1976); GSS (1972–2018) grafted together onto the GSS scale based on overlapping years (see online Annex 1.1)

Hedonic measures based on ‘affect’

Our findings so far are based on people’s replies to a single question about how respondents evaluate their life ‘nowadays’.Footnote ¹⁸ But some people prefer a different approach to the study of human wellbeing – where we study more directly how people feel at the time. By comparison with this, the evaluative measures cover a longer unspecified period of time and involve an element of judgment (although they are generally completed within a few seconds).Footnote ¹⁹

The alternative measures are known as hedonic measures (from the Greek word hedone meaning pleasure). Psychologists use the word ‘affect’ to describe feelings of different sorts. Some types of affect are positive (happiness, enjoyment, laughter) and some are negative (worry, sadness, anger and stress). There are three main ways to capture these feelings. One is in real time by beeping people and asking how they are feeling just now (Ecological Momentary Assessment). The second is to sit people down a day later and have them record how they felt on the previous day hour by hour – and what they were doing and who they were with (the Day Reconstruction Method). The third, also used by the Gallup World Poll, is to ask them to summarise their different feelings over the previous day by asking Yes/No questions like ‘Yesterday did you experience a lot of happiness?’

How people feel when they are doing different things is hugely interesting. A team led by Daniel Kahneman used the Day Reconstruction Method to investigate the feelings of around 900 Texan women during the course of the previous day.Footnote ²⁰ Table 1.2 shows what they enjoyed most and what they enjoyed least.

Such information is highly relevant to how each of us can improve our daily life.Footnote ²¹ And, in principle, hedonic measures could be used to produce a single measure of wellbeing over a period of time. You could either pick a single question (such as How happy are you right now, or yesterday) or you can calculate an ‘affect balance’ by adding up all the positive scores and subtracting all the negative ones.Footnote ²² Or, altogether novel, you can analyse people’s social media activity (especially Twitter) and calculate the balance between positive and negative words or remarks.Footnote ²³ Or you can analyse Google searches.

Comparison of evaluative and hedonic measures

So which of the two preceding criteria is best for use as a guide to the choice of policies? As we argued at the beginning of the chapter, it self-evidently matters how people feel. This is extremely obvious when it comes to pain versus contentment/enjoyment (the ‘hedonic’ dimension of experience). But it also applies to how people ‘feel about their life’ (the ‘evaluative’ dimension).

As a measure of wellbeing, life satisfaction has many good features. First, it covers more than a day or two.Footnote ²⁴ Second, researchers and policymakers know what it means – it is based on a single question that they themselves could answer. Third, it is democratic – it does not require analysts to compile a list of indicators and then use their own arbitrary weights to produce an overall index.Footnote ²⁵ Instead, each citizen applies her own weights and tells us the result – how she feels about her life. And fourth, policymakers are used to asking people how satisfied they are with their services, so why not ask people how satisfied they are with their lives?

Hedonic measures are an even more direct measure of experience, but they are much more difficult to collect over a long enough period to be relevant to many basic social issues. They are, however, essential in analysing brief experiences – like a country walk or a game of tennis or a gig. And, as science develops, they will become easier and easier to measure (including via the use of biomarkers – see Chapter 5). So, in this book, we use both life satisfaction and hedonic measures but mainly life satisfaction. When possible, we report where the two approaches yield different results.Footnote ²⁶

Eudaimonic measures

The third wellbeing measure in common use is usually called ‘eudaimonic’ after the word ‘eudaimonia’ used by the Greek philosopher Aristotle. Eudaimonia is difficult to translate but it means literally having a ‘good demon’, which Aristotle understood, roughly, as having a rounded and virtuous character.Footnote ²⁷ So it includes the idea that virtue ought to be included in any measure of wellbeing. A typical question is ‘Do you feel that the things you do in your life are worthwhile?’

There are, however, two objections to this approach. First, virtue is difficult to measure. If you want to know whether someone is virtuous, you cannot find out by asking that person – the Nazis felt that what they did was positive and worthwhile. Second, in the wellbeing approach, virtue is a means to an end.

We want people to be virtuous for two reasons: it will raise the wellbeing of others, and it will often (though not always) raise the wellbeing of the virtuous person. So, for example, in Figure I.1, my wellbeing is higher if

1. (i) the people I meet – my ‘social connections’ – are virtuous, and

2. (ii) I myself have good altruistic ‘values’.

But these are just some of the many things that determine wellbeing – they are not part of the wellbeing outcome itself. To include virtue in the outcome is to confuse means with ends. We should study virtue deeply, but we should not include it in the measure of wellbeing.Footnote ²⁸ So, in this book, we do not use the eudaimonic concept of wellbeing.Footnote ²⁹ We rely mainly on evaluative and hedonic measures of wellbeing, which we also sometimes call happiness. As an aide memoir, Box 1.1 summarises the preceding discussion.

Can we believe self-reports?

As time passes, we shall undoubtedly get better at measuring wellbeing. Neuroscience will improve, and we shall also use more and more big data from sources like Twitter and Google. But do the questionnaires we currently use really supply any useful information? When we ask people these questions, are their answers accurate? Do they really mean anything, or do different people interpret the question and the scales so differently that their answers cannot really be compared? In other words, are the measures ‘reliable’ and ‘valid’ evidence on the thing they purport to measure.

On reliability, the question is ‘Do people give consistent answers when retested?’ In one study, the correlation between the two sets of answers two weeks apart was 0.55 for life satisfaction and 0.64 for net affect.Footnote ³⁰ This is not bad, but it shows that there is some noise in the data.

A quite different issue is whether the scale is a ‘valid’ representation of what we want it to represent, including whether different people use the scale in the same wayFootnote ³¹ There are at least four reasons to believe that people’s answers provide significant objective information about how their subjective wellbeing compared with that of other people in the same country.

The reporting scale

A different issue is exactly how people use the reporting scale they are offered (0,1, … ,9,10). Do they treat it like a metre rule, where the difference between 3 and 4 centimetres is the same size as the difference between 8 and 9 centimetres? This practice would make the scale an ‘interval scale’ (or what economists call a ‘cardinal scale’). Or do the points on the scale simply reflect a ranking of different mental states (making it an ‘ordinal scale’)?

So what would it mean to say the scale is cardinal? If we want to measure differences in the level of a sensation, the standard basic unit of difference is one that is just noticeable by the person experiencing it. In other words, the natural unit for sensations is the just-noticeable difference (JND).Footnote ³⁷ So we would call a scale an interval scale (or cardinal) if the number of JNDs between the answer 3 and the answer 4 were the same as the number of JNDs between 8 and 9.

Are they? There is a simple empirical test of whether the scale is cardinal: when people are asked and then re-asked to record their life satisfaction (0–10), the average absolute difference in replies should be similar at all points on the scale. It is.Footnote ³⁸ So there is good reason to suppose that people use the scale in a cardinal way, as we have defined it – apparently, respondents naturally reply as though the difference between 3 and 4 is as noticeable as the difference between 8 and 9.

We have already mentioned another relevant fact. Sometimes people are asked to score a variable not by selecting an integer but by selecting a point on a continuous scale (the ‘visual analogue’ method). Studies of the kind give very similar regression results to those using integer scales.Footnote ³⁹

So, when psychologists study sensations, they generally assume people treat the scale as cardinal (see Box 1.2 for further details). By contrast, it is difficult to see how respondents could use the scale (0–10) in an ordinal fashion. There would have to be a ranking of all their possible states of wellbeing, and they would then somehow divide this up into eleven ranges (0–10). However, since there is no concept of distance between one state and another, it is extremely difficult to see how they would undertake this process in any way that is consistent over time or across people.Footnote ⁴¹

Box 1.2 ‘Reporting functions’ in psychology

The study of sensations assumes that humans report their sensations in a cardinal fashion. It then addresses the question: How does their sensation relate to the intensity of external phenomena – like brightness? In other words, what is the ‘reporting function’ that relates the reported sensation to the external stimulus (measured in its own units)? For example, how does the reported brightness of light relate to its actual brightness measured in lumens per square metre of receiving surface?

In this case, the just-noticeable difference (JND) in brightness corresponds to a given proportional change in actual brightness (measured in lumens per square metre). Thus the ‘reporting function’ for brightness is

$\text{Sensation}=alog\text{Brightness}$

This is known as the Weber–Fechner effect. Similar ‘reporting functions’ have been found for the intensity of sound and indeed for many other things that can vary over a huge range from 0 upwards. But when the range of variation is narrow relative to the average (as with human height), the reporting function becomes roughly linear.Footnote ⁴⁰

These reporting functions are telling us how people report on an external phenomenon. When it comes to wellbeing, we are talking about something quite different – how people record an inner state. There is no obvious reason why they would not do this in the cardinal way we have described, and the science of psycho-physics assumes they do so.

So we shall assume throughout this book that different individuals use the scale in the same way, that the scale used is stable from period to period and that the scale is a normal one (like a metre rule).

What causes wellbeing?

So what determines their wellbeing? If we want to improve wellbeing (for ourselves or others), we have to know what affects it and by how much. To find this out we have two main sources of evidence. The first is surveys in which people are asked about their wellbeing – but also about many other aspects of their life. The relationship between wellbeing and these other things tell us how much different things matter to people. But precise causality is always difficult to establish from such surveys, nor do they tell us in any detail what we can do to make things better. For this, we need evidence from experiments where some people have received a particular treatment and their change in wellbeing can then be compared with that of a control group. In the last 40 years, evidence of both types has developed at such a rate that we now have the new science of wellbeing, whose findings are the subject of this book.

Wellbeing helps us achieve other valuable objectives

There are, however, many people who question the idea of wellbeing as the objective. We shall discuss this issue in Chapter 2. But even people who do not value wellbeing for its own sake should value wellbeing because it is an important means to many other objectives that they do value. Here are some striking examples.

• Education. Making children happier makes them learn better.Footnote ⁴⁴

• Health. Your wellbeing has a powerful effect on your longevity.Footnote ⁴⁵

• Productivity. Greater wellbeing increases productivity and helps with problem-solving.Footnote ⁴⁶

• Family/Social cohesion. Happy people create more stable families, and happy people are more pro-social.Footnote ⁴⁷

• Political stability. Wellbeing affects election outcomes more than the economy does. And unhappy people tend to support populist parties.Footnote ⁴⁸

• Charitable giving. Happy people give more to others.Footnote ⁴⁹

These consequences of wellbeing are of major importance. So throughout this book we shall discuss the two-way relation between wellbeing and the different aspects of human life.

To re-cap, the two relations are these:

• One relation is the effect of each aspect of life on overall wellbeing. This was displayed in Figure I.1, with the arrows pointing towards wellbeing.

• The other relation is the effect of wellbeing on each aspect of life. This involves a diagram just like Figure I.1 but with the arrows pointing in the opposite direction.

In this book, we look at both sets of effects, but our strongest interest is in the first relation: how wellbeing itself is determined.

Human needs

So let’s examine some basics about how wellbeing is determined and use this to explain the structure of the book. There is evidence that all human beings have similar fundamental needsFootnote ⁵⁰ and that their wellbeing can be (at least partly) explained by how well these needs are met.

A typical list of these needs includes:

• Food and shelter,

• Safety from attack,

• Love and support,

• Respect and pride,

• Mastery of what you do,

• Autonomy in what you choose to do.

Since 2005, the Gallup World Poll has surveyed representative samples of people in every country in each year. It asks about their life satisfaction and also whether their different needs are met.Footnote ⁵¹ The analysis confirms that satisfaction with life increases when more of these needs are met – and people are more satisfied both when their own needs are met and those of their fellow citizens. This confirms a common finding that wellbeing is contagious.Footnote ⁵²

But are some of these needs prior in importance to others? For example, if your life is physically hard, how important is it to experience respect and pride? According to Maslow,Footnote ⁵³ there is a hierarchy of needs (proceeding upwards from Physiological to Safety, Love and Belonging, Esteem and ultimately Self-actualisation): and the best strategy is to satisfy the needs by progressing up the sequence. Thus ‘respect is a disposable luxury when compared with food or safety’. Though hugely famous, there is little supporting evidence for Maslow’s hierarchy. The evidence from the Gallup World Poll is that when each need is satisfied, wellbeing improves. But having one need satisfied has little effect on the value of satisfying another need.Footnote ⁵⁴

How people interact with their environment

This book is about what determines how far our needs are satisfied. A first answer to that question is that it depends on the interaction between our genes (with which we enter the world) and our external environment – meaning by external environment the whole social and physical world outside ourselves. This is illustrated in Figure 1.3.

Figure 1.3 How genes and external environment determine wellbeing

But this gives the impression that the individual herself is a passive player in the drama, with the environment simply interacting with the person’s genes to produce the outcome. Yet in fact every human being has some degree of agency. We influence our own wellbeing in two key ways:

• By our behaviour, we significantly influence the situations we experience and the behaviour of others towards us.

• By our thoughts, we influence how these experiences affect us – both through our attitudes and how we think about our life.

Figure 1.4 tries to capture some of this. Outside us there is a given social environment, but how we experience it is hugely affected by what we bring to it. So the individual’s behaviour interacts with the environment to produce the individual’s experience. This experience then affects the individual’s wellbeing – but how much depends also on how she thinks. This process is repeated over and over again. At each phase, the person’s wellbeing feeds back into her character and behaviour in the next phase of her life (as shown in the dotted lines).

Figure 1.4 How the individual and the external environment interact to produce wellbeing

From this analysis it is clear that our wellbeing does not only depend on our social environment. It also depends on what we bring to the table, in particular

• our behaviour,

• our thinking and

• our genes.

In Chapters 3–5, we discuss each of these processes, before we discuss the details of the environment in Part III of the book.

What kind of a subject is wellbeing science?

Finally, what kind of a subject is wellbeing science? It is a new, emerging field, which is totally interdisciplinary. Many of the most important subjects that have emerged in recent years are interdisciplinary – subjects like molecular biology or human geography. But this one is rather special, because it provides a rationale for each of the separate social sciences: they are only important because they help to explain wellbeing. If you look at Figure I.1, you will see what we mean. For example, international relations are important because of how they contribute to human wellbeing, and so on. In this sense, wellbeing science could be the queen of the social sciences: it gives a role to each science by showing quantitively how its own outcomes contribute to the overall good.

But, as we shall see in the chapters that follow, there are four key disciplines that are most central to the study of wellbeing: psychology, sociology, economics and statistics.

• Psychology. The study of wellbeing started in psychology.Footnote ⁵⁵ Psychologists study how human nature works at the individual level, using both surveys and experiments. They study how individual personality, family experience and education affect wellbeing. They also study how to make things better for individuals (clinical psychology) and at work (occupational psychology). However, except for social psychology, psychologists tend to ignore the role of social norms and social structures upon human wellbeing.

• Sociology. That is where sociologists come in. They study, above all, the way people interact in groups and how that affects their wellbeing.

• Economics. Economics brings four main contributions. First, economists have always focused on wellbeing as the overarching outcome (calling it ‘utility’), even though they often have a rather narrow view of it and what causes it. Second, they discuss policy in terms of maximising that single outcome. Third, they distinguish clearly between the things people choose and the things that just happen to them (which they call ‘externalities’). And fourth, they bring understanding of markets, income and unemployment.

• Statistics. This is central to the preceding disciplines. When it is used to analyse the distribution of characteristics in a population, it is sometimes called epidemiology.

There are three other fields of study, which, though relevant, are not the same as the science of wellbeing.

• Behavioural science. In recent years, economists have been developing, jointly with psychologists, a deeper understanding of how people actually behave – a subject sometimes known as behavioural economics. But the study of wellbeing is a different endeavour. It does not study how to get people to behave in a certain way. It studies the results of behaviour and thus which behaviours would maximise social wellbeing.

• Health and wellbeing. This phrase, increasingly used, signifies a welcome extension of health science to include mental states as well as physical conditions. But it does not typically view wellbeing as the overarching good, with health as just one of many influences upon it. Wellbeing science, in contrast, studies the effects of everything in life upon the wellbeing of the population.

• Finally, philosophy. ‘What is wellbeing?’, ‘Is it all that matters?’, ‘How should it be distributed?’, and so on, are, of course, philosophical issues. We turn to these issues in Chapter 2. After that, the rest of the book is about the science of what causes wellbeing and how it can be improved.

Questions for discussion

1. (1) Is there anything more important than how we feel? Is wellbeing the overarching goal? If not, what is, and how should we weight multiple goals?

2. (2) What is the best way to measure wellbeing? Are these measurements ‘cardinal’?

3. (3) How does wellbeing science relate to other subjects you have studied?