1 Introduction

The International Society for Child Indicators is still a young scientific community. Comparative research using child indicators is also still in its infancy (Bradshaw 2013), and the study of children’s subjective well-being even more so. The UNICEF flagship The State of the World’s Children has been published only since 1980. The Health Behaviour of School Children survey began in 1983/84 (Currie et al. 2012). TIMMS began in 1995, PISA in 2000 and PIRLS in 2001. The Innocenti Report Card series on children in rich countries was first published in 2000. Although the OECD published some comparative data on children before, their Doing Better for Children only appeared in 2009 (OECD 2009) and our comparisons of child well-being in the EU, CEECIS and Pacific Rim regions first appeared in 2007, 2008 and 2010 respectively (Bradshaw et al. 2007; Richardson et al. 2008 and Lau and Bradshaw 2010). The European Commission has made much progress on indicators of child poverty and deprivation, but so far only very tentatively engaged with child well-being (but see TARKI 2010, European Commission Social Protection Committee 2008). The African Report on Child Well-being from the African Child Policy Forum (2013) is only in its second edition. UNDP publishes the World Development Index (http://hdr.undp.org/en) which has some child relevant indicators in it and this has been further developed by work in Bristol (Gordon et al. 2003) and Oxford (Alkire and Santos 2010). There are also exciting new developments such as UNICEF’s Moda project (www.unicef-irc.org/EU-MODA and De Neuborg et al. 2012) and of course the Children’s Worlds Survey (http://www.isciweb.org/Default.asp). The aim of this article is to outline the wider context within which the Children’s Worlds Survey is taking place, and to make the case for the importance of this and similar research. It is of course written in advance of the publication of the results from the Children’s Worlds Survey, including the papers in this collection. One of the outcomes of the Children’s Worlds Survey will be to test whether this kind of comparative investigation of child subjective well-being has resonance for social policy.

2 The Value of Comparative Studies

The studies outlined above, including the Children’s Worlds Survey, are primarily comparative in nature. Why do we bother with comparisons? After all, policy is made mainly at a national level. Comparison has something akin to experimental power—differences between countries can help to raise hypotheses about the explanations for these differences. Also, without comparison it is difficult to assess how well we are doing or how good we could be. To give an example, in the UK there have been ‘State of the Nation’ studies of child well-being since 1999, approximately every 2 years (the most recent Bradshaw 2011). Over the years children’s situations have sometimes been getting worse (under the Thatcher government in the 1980s) and more recently, until the financial crisis, they were getting better. This is a legitimate cause for optimism. But the comparative evidence tells us that we should not be too pleased. The latest UNICEF Report Card 11 (UNICEF 2013) still has UK children 16th in the OECD league table (or 14th if subjective well-being is included)—well below the commensurate position if child well-being was determined by GDP.

To make a case for the importance of studies such as the Children’s Worlds Survey, this paper will focus on the topic of child well-being and social policy. The question of whether we can engineer child well-being using social policies will be considered. In attempting to tackle this question the impact of social policy on objective child well-being will not be addressed; there is much more certainty that social policy can mitigate child poverty and deprivation, improve children’s health, increase educational participation and attainment, reduce pollution and improve housing and even reduce risks for children. Of course a major point of debate in each of these domains remains around which policies best address the issues. But a detailed discussion of that is beyond the scope of this paper.

Instead, the focus will be on the newer and perhaps more difficult issue of whether policy can make children happier? As we shall see objective well-being is related to subjective well-being and this gives us an indication that working on policies that influence objective well-being may well have impacts on subjective well-being.

3 Can Nations Make Their Children Happier?

Figure 1 presents the league table of subjective well-being derived from HBSC data (Klocke et al. 2014). It combines life satisfaction (measured using Cantril’s ladder, a single-item measure which asks children to rate their satisfaction with their life on an 11-point scale); an index of relationships with family and friends (the proportion of young people finding it easy to talk to father, mother and who found their friends kind and helpful); an index of subjective health (a combination of subjective health 1 = excellent to 4 = poor and the proportion of children in each country reporting two or more of eight psychosomatic health complaints); and an index of subjective education (liking school and feeling pressured by school work). It is far from a full representation of subjective well-being, but probably the best comparative data we have. It was used recently in Innocenti RC 11 (Bradshaw et al. 2013) at a macro level, and subsequently in a micro analysis of the HBSC (Klocke et al. 2014).

Fig. 1
figure 1

Overall subjective well-being. Source HBSC 2011/12 (Klocke et al. 2013)

Full size image

Consider the Netherlands. Children in this country have substantially better subjective well-being than their peers in other countries. The Netherlands comes top of the league table on three out of the four components. It is sixth from highest on subjective health. In successive HBSC surveys Dutch children regularly have the highest life satisfaction. Why? How do they achieve these results? The Netherlands is a rich country, but not the richest in this distribution. It is certainly not a big spender on social policy—in the OECD it comes 18th in spending on family benefits and services, below the average. On education spending it is 10th in the OECD, just above average. Thinking about conditions in the country, it is flat and watery, quite crowded, people live in small houses, it has liberal laws on drugs. It has had quite low labour participation rates of mothers (but there is no evidence that children whose mothers work have lower life satisfaction). It has perhaps achieved an unusually good balance in parental labour supply. Dutch children report liking their schools, they don’t have to wear uniforms (though uniforms may prevent bullying in very unequal countries), they have a lot of choice over what they study, they seem to have good relationships with their teachers and school mates, and yet they are successful in PISA terms. They report being, and seeing others as, kind in their relationships. But we remain unsure as to why.

Space precludes detailed consideration of each country, but some other aspects of the league table are worth noting. Most of the Nordics—Iceland, Sweden and Norway—are where they may reasonably be expected given their performance in other indicators—that is, towards the top of the league. But Finland is further down—amongst the causes of this are that Finnish children report lower enjoyment of school, although they perform best of all European countries on PISA attainment. Anglophone countries—the USA, Canada and the UK—are also where they might reasonably be expected to be—all below average. No clear pattern is evident in the positions of the southern EU countries—for example Spain is above average, whilst Italy is fourth from the bottom. The EU 10 countries are for the most part below average, though Slovenia comes 9th.

Moving on to an examination of well-being over time in the UK, Fig. 2 presents Bradshaw and Keung’s (2011, updated by authors) analysis of data from the British Household Panel Survey, a longitudinal survey concerned with an array of social issues in Britain. The Survey has asked a sample of young people aged 11–15 a set of questions about their subjective well-being in every year since 1994. The figure shows the mean composite score of children’s satisfaction with school work, appearance, family, friends and life as a whole—again, this is not a perfect measure of subjective well-being, but it does provide a rare opportunity for time series analysis. It shows evidence that subjective well-being is statistically higher at the end of the period than it was at the beginning. But as previously, pertinent questions remain around the causes of this improvement.

Fig. 2
figure 2

Subjective well-being 11–15 UK: BHPS

Full size image

Additional analysis by Bradshaw and Keung (2011) reveals interesting facts that arise from the data itself. At the beginning of the period girls were much less happy than boys (partly because they were less happy about their appearance). Both boys’ and girls’ subjective well-being improved significantly over the period, but girls improved faster and there was no difference in average well-being by gender at the end of the period. Again, questions are raised about the causes of these changing trends. Bradshaw and Keung looked at which components of subjective well-being had improved most, and found that they were views about school work and friendships. Some possible explanations for this include that the increased spending on schools by the Labour Government after 1999 led to improvements in satisfaction with school; or that spending on children’s benefits and services in general increased over this period—by 2009 the UK came second from top in the OECD on spending on family benefits and services. In regard to improvements to friendships especially for girls, suggestions include that this may be the result of social networking—for example texting, Twitter, email, Facebook. All of these make it much easier for girls to maintain friendships, whereas in the past they may have been restricted from meeting friends outside of school and other organised activities for fear of their safety. It may be significant that subjective well-being has now begun to fall as the austerity measures of the Coalition Government since 2008, which have hit families with children hardest (Cribb et al. 2013).

Thinking about these two observations together—that we have big international variations in child subjective well-being, and that we have clear evidence of improvement in well-being over time in the UK—the question will now be considered whether either can convincingly be attributed to policy.

4 Why We Might Not Expect an Association Between Social Policy and Subjective Well-Being

There are many reasons why we might we not expect to find an association between policy and subjective well-being. These are now briefly detailed.

4.1 Difficult to Measure

The dependent variable, that is, the way we measure and represent subjective well-being, may not be good enough. Figure 3 shows a matrix of the elements of well-being in Rees et al.’s (2013) framework. International comparative data is really only available for the hedonic elements, and within that only the cognitive part. Even within that we only have data on a few of the domains of life satisfaction.

Fig. 3
figure 3

Components of self-reported well-being: Rees et al., Children’s Society 2013

Full size image

4.2 Lost in Translation

Our indices may not be reliable, at least for international comparison. This may be because of ‘lost in translation’ issues. Words like ‘life satisfaction’ or the ‘best possible life for you’, both commonly used in questions relating to subjective well-being, may have different meanings when translated into French or Korean. Many examples of this exist, of which the following two are illustrative:

  • The translation of the two positive answers “I like [school] a lot / I like it a bit” into Italian is “mi piace molto / mi piace abbastanza” in HBSC. “abbastanza” is an Italian quantifier with a really ‘woolly’ meaning) and very few Italian children check it—perhaps as a result Italian children appear to have very negative views of their schools.

  • In the PISA 2003 Student Questionnaire, Section D, Q27(f)

Where the original question in English is:

Q27 My school is a place where: (please tick only one box in each row)

(f) I feel lonely

In Japan, this is translated as:

My school is a place where

(f) it is boring all time

This has been cited as evidence that Japanese students feel lonely at school and have very low subjective well-being.

4.3 Cultural Response Bias

Then there is the argument that the measures may not be valid or reliable because of culturally determined unwillingness to express satisfaction or dissatisfaction with your lot. French teenagers come bottom of the league table on relationships with family or friends—by some margin—could this be attributed to cultural predispositions (stereotypically, Gallic anxt!), or does it reflect the reality of their lives? Similarly, it may be that Finnish teenagers score low on liking school a lot because they tend not to respond very enthusiastically (“a lot”) to anything; does this reflect a genuine lack of enthusiasm, or is it in their nature, or their cultural background, to be low key?

4.4 Adaptive Preferences

Associated with this, there is also some evidence that expressions of subjective well-being may be a function of false expectations or adaptive preferences (discussed by Hallerod 2006, in relation to poverty)—that people in what would commonly be considered deprived situations alter their preferences in lieu of being able to alter their circumstances, in order to avoid the pain associated with deprivation. Very deprived children may say that they are very satisfied with life because they know no better, or because they have become reconciled to their lot. There is certainly some evidence in poverty studies of poor children not complaining to their parents in order to protect them from guilt (Ridge 2002). However, in analysis of surveys conducted with children relating to material deprivation, Main (2013) found at best mixed evidence around whether children demonstrated adaptive preferences, suggesting that social factors as well as or instead of adapted preferences may determine children’s expressions of whether they want what they lack. An example of false expectations in the other direction would be a child being dissatisfied with his/her body or his/her clothing because he/she does not look like models he/she sees in the media.

4.5 Transience

Then there is the view (not upheld by British evidence) that subjective well-being is (merely) a volatile or a transient mood. Indeed a greater challenge comes from Cummins (2010) who has argued that happiness is largely the result of genetically determined homeostatic adaptation. Over the millennia the humans who have survived most successfully have been those who have had more capacity to adapt to their environment and the shocks of life. Humans, including children, have a natural resilience to bounce back to a predetermined happy state. However, Cummins and Cahill (2000) do note that whilst there may be a stable level of subjective well-being for most people, the experience of traumatic events can alter this level. Main (2014) argues that, particularly for children whose personality traits are less stable, the study of subjective well-being and how it can be influenced by policies is still valuable and pertinent, because it may have more profound impact on them.

4.6 Difficult to Explain Variations

All of the above may explain why it is has been so hard to develop statistical models which explain variation in subjective well-being in terms of social structural characteristics or life events. In several surveys of child subjective well-being undertaken in the UK, we find that subjective well-being varies by age and gender, but few other characteristics have significant or sizable associations. In these surveys (see Rees et al. 2012 and Rees et al. 2013) 9 % of the variation in subjective well-being was explained using age, gender, ethnicity, number of siblings and disability (actually only age and gender were significant)—results are shown in Table 1. Using a child-derived index of material deprivation (see Main and Bradshaw 2012; Main 2014), the explanatory power of the model rose to 17 %; when family structure was controlled for this increased slightly to 19 %. Whilst it is important to acknowledge that the complexity of the social world limits the explanatory power likely to be achieved by statistical models, 19 % still leaves much to be explained. One possible explanation, noted above, is that children revert to their ‘normal’ level of subjective well-being relatively quickly after supposedly major life events.

Table 1 Multiple regression of subjective well-being: England (The Children’s Society 2012)
Full size table

4.7 Influence of Personality

In addition to the homeostatic explanation detailed above, there is the evidence that subjective well-being is not independent of personality. Goswami (2014) shows that adding a standard measure of personality (including extraversion, agreeableness, consistency, emotional stability and openness) to the regression above doubles the adjusted R-squared to 35 %. The trouble is that most of this comes from emotional stability, a trait which is arguably similar to, and may not be independent of, subjective well-being. If personality is indeed a major determinant of subjective well-being, the nature of the question changes to—can policy influence personality? This question is beyond the scope of this paper or indeed of most investigations into subjective well-being.

4.8 Elements of Subjective Well-Being may Not be Policy Amenable

A final point to note are the findings of work we have done with the Children’s Society (2012). We investigated which domains of the Good Childhood Index, a set of ten life domains,Footnote 1 (see Table 2) identified through qualitative and quantitative research with children as important to their subjective well-being. We assessed which contributed most to overall subjective well-being measured using a reduced version of Huebner’s Student’s Life Satisfaction Scale (Rees et al. 2010). It was found that the two most important domains are relationships within the family and the amount of choice that children have in their lives. The first is not surprising—children spend a great deal of time in the family context, and are dependent on families to meet many of their physical, psychological and emotional needs. The second is very interesting—research to date into children’s well-being has only minimally considered the issue of choice. But for the purpose of this paper they present the immediate question—what can policy do to influence family relationships or the amount of choice children have in their lives? With regards to the remaining eight domains, policy may be able to influence money and possessions, health, the future, even, school and home, but what can it do for time use or appearance?

Table 2 Associates of overall subjective well-being in rank order
Full size table

These arguments have convinced many people, perhaps even some of you readers, to either discount subjective well-being as not policy salient, or to treat is as a second order problem. Next, arguments which are more in favour of continued research effort in the measurement of children’s subjective well-being are detailed.

5 Why We Might Expect an Association Between Social Policy and Subjective Well-Being

This section puts together some evidence and arguments for there being an association between child subjective well-being and social policy.

5.1 Comparative Evidence

Perhaps the most powerful comparative evidence that policy can influence subjective well-being is that subjective well-being is associated with all the domains of objective well-being included in the UNICEF Report Card 11 on child well-being. Thus in Table 3 from analysis of the Health Behaviours of School-age Children survey (Bradshaw et al. 2013) overall subjective well-being is highly correlated at a macro level with all the other dimensions. This means that countries where well-being is better in the objective domains of material, health, education, behaviour, and housing tend to have happier children. Subjective well-being was most strongly associated with the material well-being and the housing and environment domains. Both of these can be and are a major focus of policy in rich countries.

Table 3 Correlation between overall subjective well-being and the objective domains of well-being OECD countries RC11
Full size table

The Fig. 4 shows the relationship between overall well-being, exclusive of subjective well-being, and subjective well-being. The association is strong and would be stronger without Romania.

Fig. 4
figure 4

Association between overall well-being excluding subjective well-being and overall subjective well-being (z scores on both axes)

Full size image

Although in UNICEF Report Card 11 (2013), we did not find an association between subjective well-being and spending on benefits and services for families with children as a proportion of GDP, if education spending is added there is an association, which would be stronger (see Fig. 5) if the Netherlands was not such an outlier: r = 0.48.

Fig. 5
figure 5

Spending on family benefits and services and education as a % GDP and subjective well-being (z scores on the y axis)

Full size image

5.2 National Evidence

There remains the evidence at a micro level in the UK (The Children’s Society 2012 and 2013) that external factors have an impact. For example

  • Children being looked after outside their families tend to have lower subjective well-being.

  • Life events such as a change in family structure or moving home or school are associated with lower subjective well-being.

  • The quality of relationships matters a lot, certainly family conflict matters more than family structure.

  • Relationships with others and involvement in decision making makes a difference.

  • As we have seen, whether children feel materially deprived matters.

  • We have found that being bullied has a big impact on life satisfaction—perhaps more than any other factor. In fact a child’s recent experiences of bullying explained roughly as much of the variation in overall well-being, as all the individual and family characteristics combined.

5.3 The Case of Bullying

There are big international variations in the prevalence of bullying (see Fig. 6) and considerable evidence that it can be influenced by policy (see for example Finland’s experimental strategy to tackle bullying in schools (http://www.kivaprogram.net/). Klocke et al. (2013) estimated that a number of countries could improve their subjective well-being if they reduced their bullying. Austria, rather than the Netherlands, would be top of the international league table on subjective well-being and Estonia would move from 22nd to 12th if they reduced their bullying rates to the average.

Fig. 6
figure 6

Bullied at least weekly (HBSC 2010)

Full size image

5.4 Normative and Legal Responsibilities

First there is the normative argument, put forward by Ben-Arieh (2005) that we have a moral obligation to listen to children and take seriously what they think and feel. The so-called objective domains in well-being research (such as educational attainment and participation) are often about well-becoming rather than well-being, and focus more on adult concerns about children’s development and future productive potential, than on children’s own concerns about their day-to-day lives. Surveys of subjective well-being are one way of giving children voice on their well-being in relation to their present lives, rather than in relation to future agendas. The United Nations Convention on the Rights of the Child provides a legal framework which obliges almost all countries in the world to listen to children and take their views into account.

5.5 Priorities for Child Well-Being

The Children’s Society (2013) have bravely come up with the matrix in Fig. 7 for the six priorities children’s well-being and one can think of policies that might help promote each of these. For example social security policy can influence Money; schools can influence Learning; housing policy can influence the local Environment and so on.

Fig. 7
figure 7

Six priorities for children’s well-being

Full size image

6 Discussion

Children would be happier if they live in decent houses, in safe neighbourhoods, are not bullied, enjoy and achieve in schools and are not materially deprived. This makes intuitive sense and is backed up by research to date. These aspects of children’s lives can all be influenced by policy. Family and other relationships may matter more than these things for subjective well-being and they may not be directly amenable to policy. But indirectly they could be—by for example

  • reducing the burdens of poverty and inequality on parents,

  • identifying and treating parental depression,

  • providing family friendly services.

However, research into children’s subjective well-being is in its infancy, and research on what social policies work in this field is even more so. Inevitably we need more research. Examples of the kinds of research which would be valuable include:

This last brings us back to the purpose of this article: that is, assessing the value of the Children’s Worlds Survey of Child Well-Being. What this article has highlighted is that in its current state, both national and international investigations into children’s well-being help us to answer some questions, but also pose many more questions around what to measure, how to measure it, and how to create measures which produce good data across national and international boundaries (Main 2014, notes the importance of continued research into children’s subjective well-being if we are to successfully address the issues raised by critics of the field). The pilot phase of the Children’s Worlds survey, which this article and Special Issue is concerned with, represents an attempt at beginning to answer these questions. Without such work, uncertainties will remain about the value and relevance of subjective well-being and its various domains to social policy. The survey is based on child-centred principles, qualitative and quantitative consultation with children, and researchers who have dedicated a large part of their work to understanding children’s subjective well-being. Whilst it will not present the final word in the relevance of subjective well-being in social policy considerations, it represents a positive development to what has gone before, and will no doubt provide valuable lessons both for national and international policies, and for future research.