Dimensions of Wellbeing

Happy and flourishing? Hedonic and eudemonic wellbeing

Hedonism (happiness) and eudemonia (flourishing) are both important components of individual wellbeing. The European Social Survey (ESS) shows how these related but distinct dimensions of wellbeing vary across Europe.

Introduction

Questions regarding what exactly wellbeing consists of can be traced back to the philosophical debates in ancient Greece (Waterman, 1993). The hedonic school of thought, exemplified by Epicurus, believed a good life to be filled with happiness, and little pain. Aristotle dismissed this narrow conception, and instead proposed eudemonia, or flourishing, living in accordance with your true self, as a way to lead a good life (Keyes and Annas, 2009).

These distinct views on the nature of wellbeing have survived the ages, and still exist in contemporary research on wellbeing. Through economics, Bentham’s hedonic calculus, or the idea that the utility of everything can be calculated, has found its way into contemporary policy debates with subjective wellbeing as the ultimate outcome (Dolan and Peasgood, 2008). Hedonic conceptions of wellbeing, such as life satisfaction and happiness, are now well-known ways of investigating wellbeing.  Eudemonic wellbeing is closer to current psychological constructs, such as Maslow’s hierarchy of needs, which names self-development and autonomy as higher order needs, which emerge after more basic needs such as food, shelter and security are fulfilled (Ryff and Keyes, 1995). Another conceptualisation of what eudemonic wellbeing entails can be found in the capabilities approach (Nussbaum and Sen, 1993), which states that wellbeing is about the potential to choose as well as the actual opportunities taken. So while hedonic wellbeing is about feelings and the evaluation of one’s situation, eudemonic wellbeing is more about being in control and having a choice. The existence of different forms of wellbeing that are not reducible to one basic form has been recently discussed in academic literature on wellbeing (e.g. Clark and Senik, 2011; Kashdan, Biswas-DIener and King, 2008; Vanhoutte, 2014).

Although the philosophical conceptions of wellbeing may come from different angles, there is a lot to be gained from thinking about them alongside one another.  Large scale surveys such as the European Social Survey (ESS) which measure different aspects of wellbeing allow us to do this.  We can empirically evaluate the extent to which there is in fact a distinction between hedonic and eudemonic wellbeing as experienced by individuals in society today.  We can also look at whether and how levels of these two types of wellbeing vary across countries and between different sub-groups of the population within a country. Having reliable data that can distinguish between different conceptions of wellbeing in this way is beneficial to policymakers seeking to understand and address differences in wellbeing within society. Cross-national data such as the ESS are particularly useful in providing a comparative European perspective for policy.

Combining the theoretical insights discussed above with data from the rotating module on ‘Personal and Social Wellbeing’ fielded in Round 6 (2012/2013) of the ESS, this contribution aims to construct a multidimensional measure of wellbeing that reflects both eudemonic and hedonic aspects. It will go on to investigate differences in these two forms of wellbeing across 29 European countries and between men and women.

Using the ESS to investigate different forms of wellbeing

Round 3 and Round 6 of the ESS fielded a module of questions on ‘Personal and Social Wellbeing’ capturing both hedonic and eudemonic aspects of wellbeing using multiple items. This allows researchers, analysts and policymakers to distinguish between happiness and flourishing (Huppert et al., 2009; Huppert and So, 2013). Here, we use the latest data from Round 6 of the ESS that was fielded in 29 countries.¹ We focus on seven items which together capture aspects of both hedonic and eudemonic personal wellbeing. These are described in Figure 1. The module also contains a number of other measures of wellbeing including measures of affective wellbeing, psychological resources such as optimism and social wellbeing – some of which are discussed in other contributions on this site – but these are beyond the scope of this particular analysis.

Figure 1 : ESS Round 6 (2012/2013) questions included in the analysis

Hedonic Wellbeing

B20. All things considered, how satisfied are yo uwith your life as a whole nowadays? Please answer using this card, where ) means extremely dissatisfied and 10 means extremely satisfied.
C1. All things Taking all things together, how happy would you say you are?

Eudemonic Wellbeing

B16. I feel I am free to decide for myself how to live my life.
D17. In my daily life I get very little chance to show how capable I am.
D18. Mos days I feel a sense of accomplishment from what I do.
D23. I generally feel that what I do in my life is valuable and worthwile

D35. To what extent do yo ufeel that you have a sense of direction in your life?

Since wellbeing is a latent, unobserved characteristic, measured indirectly through multiple items, we use factor analysis to construct the measures of wellbeing to be used in analysis. As we explicitly want to test if we can meaningfully speak about hedonic and eudemonic wellbeing as different forms of wellbeing, we use a technique known as confirmatory factor analysis (see Brown, 2006) to test if our data fit better into our assumed two-dimensional model instead of a one-dimensional model.

Findings

Firstly we test to see if the data confirm the dual nature of wellbeing implied by theory, i.e. whether we can meaningfully distinguish between two different dimensions of wellbeing. We do this by comparing the fit of a two-factor model (which assumes that the data are measuring two distinct concepts) vs. a one-factor model (which assumes that all our measures of wellbeing are capturing the same underlying theoretical concept). Comparing how well the two models fit the data suggests a very good fit between our theoretical model and the ESS Round 6 data.² In our two-factor model, the items referring to happiness and satisfaction load on (i.e. relate to) a hedonic dimension, while all other items load on a eudemonic dimension. The correlation between both factors is high (ρ=0.65) – meaning that people with a good score in terms of hedonic wellbeing also tend to score high on eudemonic wellbeing – but the two concepts are distinct from one another (see also e.g. Clark and Senik, 2011).

Having demonstrated that there is an important distinction between hedonic and eudemonic wellbeing we next investigate differences in individuals’ levels of both forms of wellbeing across countries. Figure 2 depicts the country mean for both types of wellbeing. The factor scores used to measure the two wellbeing dimensions are calculated over all respondents in ESS Round 6, with a mean of zero and a standard deviation of one; some countries have lower than average wellbeing compared to others (indicated by negative scores) whilst others have higher than average wellbeing (indicated by positive scores). It is clear from Figure 2 that wellbeing, especially hedonic wellbeing, does vary across countries. Hedonic wellbeing tends to be higher than average in Scandinavia and lower than average in parts of southern and eastern Europe. Second, although as shown above there is a distinction between the two types of wellbeing, in nearly all countries both forms of wellbeing seem to go in the same direction of the mean at the country level; countries that do well in terms of hedonic wellbeing also do well on eudemonic wellbeing.

It is important to keep in mind that these country level findings hide a significant amount of within-country variation. Analysis shows that that only 15% of the variation in hedonic and 7% of the variation in eudemonic wellbeing can be explained by the country with the remainder being explained at the individual level.³ Looking at variation in wellbeing by a range of socio-demographic characteristics (e.g. gender, age, education, employment, marital status etc.) both within and between countries can provide further insights into the distribution of wellbeing across populations. Here we examine if there is a gender gap in hedonic and eudemonic wellbeing and whether the size of this gap varies across countries.

Figure 3 shows the gender gap in hedonic and eudemonic wellbeing across countries.  A positive value indicates that men score higher on wellbeing, while a negative value indicates that on average women have a higher wellbeing score than men. There is a statistically significant gender gap in wellbeing in five countries (Czech Republic, Portugal, Estonia, Israel and Ukraine) for hedonic wellbeing and in two countries (Portugal and Iceland) for eudemonic wellbeing. The gender differences are, however, relatively small in most countries. Interestingly, the gender gap goes in different directions for different countries. While men in the Czech Republic and Portugal have significantly higher hedonic wellbeing than women, hedonic wellbeing is higher for women in Estonia, Israel and Ukraine. Eudemonic wellbeing is higher for men in Portugal, but higher for women in Iceland. Interpreting the gender gap in wellbeing at the country level is not straightforward: gendered expectations are culturally sensitive, and at the same time there are large differences across countries in the gender distribution of resources that produce wellbeing. For example, having a job might be a more important determinant of wellbeing for men than women in one country, while it might carry similar importance for both sexes in another.  Further analysis of the gender gap in different types of wellbeing can be found in the “Drivers of wellbeing” section of this site.

Conclusion

Analysis of data from Round 6 of the ESS reveals that hedonic and eudemonic wellbeing are two distinct concepts reflecting different theoretical aspects of subjective wellbeing.   These two distinct concepts are, however, highly correlated and cross-national comparison shows that most countries reporting high hedonic wellbeing also report high levels of eudemonic wellbeing. There is more variation at the country level in hedonic wellbeing than in eudemonic wellbeing, but the large majority of variation in both forms of wellbeing is at the individual level. Our analysis, for example, demonstrates that there are gender differences in both types of wellbeing in a number of countries. Analysing how individual socio-demographic characteristics are related to different levels of hedonic and eudemonic wellbeing can provide useful insights for policymakers seeking to understand and address differences in wellbeing within society.

Wellbeing: A Multidimensional Concept

The wide range of wellbeing measures included in Round 6 of the European Social Survey (ESS) provides a nuanced picture of the different dimensions of personal and social wellbeing. The data allow for a detailed consideration of how experiences of wellbeing vary both across countries and between different subgroups of the population within countries.

Introduction

Improving our overall experience of life, i.e. improving our wellbeing, is an aim that will appeal to most people – especially when wellbeing is understood to refer to something more than happiness alone. Many would argue that government has a duty to prioritise creating the societal conditions which enable citizens to achieve high wellbeing (Diener and Seligman, 2004). If individuals and policymakers are interested in understanding what those conditions should be, it is useful first to understand peoples’ current experiences of their lives, and to measure these alongside the range of factors which might have some influence on those experiences. Only with such measurement can we begin to understand which features are most likely to contribute to higher wellbeing, and to identify groups within society which might benefit most from interventions designed to increase wellbeing. But how can such a complex concept as wellbeing be measured systematically?

A common approach is to use surveys to ask people to evaluate their overall experience of life (Diener et al., 1985). This often takes the form of asking questions such as: “taking all things together, how happy would you say you are?” or “all things considered, how satisfied are you with your life as a whole?” However, several studies now suggest that, whilst offering a fairly good overview of wellbeing, using a single, catch-all measure might also hide some of the interesting details that a more multidimensional approach can reveal (Michaelson et al., 2009; OECD, 2013; Seligman, 2011).

The ‘Personal and Social Wellbeing’ module included in Round 6 (2012/2013) of the European Social Survey (ESS) includes 39 questions specifically designed to measure the different possible dimensions of wellbeing, enabling a much more nuanced understanding of the concept both within and between countries.

Exploring the different dimensions of wellbeing across countries

Using a combination of theoretical models and statistical analysis, we grouped 35 of the items included in the Personal and Social Wellbeing ESS module into six distinct wellbeing dimensions.

The six wellbeing dimensions are:

• Evaluative wellbeing, which covers individuals’ overall estimations of how well their life is going, including feeling satisfied with life and feeling happy overall.
• Emotional wellbeing, which includes positive day-to-day feelings such as happiness and enjoyment of life, and lack of negative feelings such as anxiety and depression.
• Functioning, which includes feelings of autonomy, competence, engagement, meaning and purpose, self-esteem, optimism and resilience.
• Vitality, which includes sleeping well, feeling energised and feeling able to face the challenges that life presents.
• Community wellbeing, which covers an individual’s feelings about the community in which they live, including trust in other people, feeling supported by members of the community, and experiencing a sense of neighbourliness.
• Supportive relationships, which relates to individuals feeling that there are people in their lives who offer support, companionship, appreciation, and with whom they can discuss intimate matters.

For each wellbeing dimension, we then created an index which combines the scores for all the survey items included within that wellbeing dimension into a single score to represent the dimension as a whole.

The wellbeing dimension scores for each of the 29 countries that took part in Round 6 of the ESS are displayed in Figure 1.¹ Results for different countries can be displayed by selecting or deselecting from the list of countries shown next to the chart. We see that where respondents within a country report high scores on one wellbeing dimension, they tend to also report high scores on the other wellbeing dimensions (for example, in Switzerland); whilst countries where respondents report lower scores on one wellbeing dimension, also tend to report lower scores on the other wellbeing dimensions (for example, in Ukraine). However, this pattern is not always found. For example, in Hungary, whilst people report quite low wellbeing across most of the dimensions, they report a much higher score for the community wellbeing dimension. Similarly, in Russia, the average score for the functioning dimension is much lower than the scores for each of the other five dimensions.

We can see from this analysis that where citizens report experiencing high or low wellbeing according to one dimension, this usually, but crucially not always, predicts similar scores according to the other wellbeing dimensions.

Where respondents within a country report high scores on one wellbeing dimension, they tend to also report high scores on the other wellbeing dimensions

Exploring the dimensions of wellbeing across age and income groups

As well as comparing the overall level of wellbeing on different dimensions within a country, it is also important to consider how that wellbeing is distributed across the population and whether there are notable inequalities in wellbeing between different groups. The ESS contains a wide range of demographic background variables allowing for a detailed analysis of how different population sub-groups score on the different wellbeing dimensions.

Differences by age group

Figure 2 shows the average wellbeing dimension scores for ESS respondents broken down by age group. This analysis includes respondents from all 29 of the countries included in the survey except Russia.²

It demonstrates that the relationship between age and wellbeing varies by dimension. For example, whilst vitality and emotional wellbeing decline steadily with age, functioning scores remain fairly consistent across age groups, before declining in the oldest age group. Supportive relationships and evaluative wellbeing scores are both at their lowest amongst the middle age group. Interestingly, the community wellbeing scores show a very different trend, increasing fairly consistently and steeply with age.

If we group countries into broad geographical regions we see that the scores on the various wellbeing dimensions follow broadly similar trends by age across the different regions. A notable difference between regions is that in south-eastern, central and eastern Europe, scores on the functioning dimension peak earlier than in the other regions, and then decline more steeply with age.

Differences by income

Figure 3 shows the strength of the correlation between the different wellbeing dimensions and total net household income. We can see that all of the wellbeing dimensions correlate positively with total net household income, except community wellbeing, where the correlation is not significant. This suggests that people with higher total net household incomes tend to report higher scores across the wellbeing dimensions. It is interesting, however, that the extent to which this relationship holds varies across the dimensions, with evaluative wellbeing scores increasing the most as total net household income increases, and vitality scores being less strongly associated with income. In this analysis, no significant relationship (at the 0.05 level) was found between total net household income and the scores that people report in terms of community wellbeing.   The lack of an overall correlation between community wellbeing and income across countries may perhaps partly be explained by variation in the relationship between community wellbeing and income across different regions of Europe.

At the regional level, in three of the six regions – Scandinavia, southern and western Europe – we find a weak but nevertheless statistically significant correlation between scores on the community wellbeing dimension and household income.  However, whereas this correlation is positive in Scandinavia and western Europe it is negative in southern Europe. This suggests that in Scandinavia (or western Europe), we would expect people with higher household income to report higher community wellbeing scores than the regional average, whilst in southern Europe, we would expect those with higher household incomes to report lower community wellbeing scores than the regional average.

Conclusion

The analysis described here demonstrates the importance of taking a nuanced approach to wellbeing that allows us to differentiate between related but distinct dimensions of wellbeing.  We see that scores on the different wellbeing dimensions vary both between countries and between different population groups within countries. Such detailed analysis – made  possible by the ESS rotating module on ‘Personal and Social Wellbeing’ – can offer policymakers insight into how best to design effective policy by showing how different determinants influence different wellbeing dimensions. For example, the analysis presented here suggests that, whilst increasing income might help to increase many aspects of wellbeing, in some regions of Europe increasing income seems unlikely to contribute to higher community wellbeing. By shining a light on the experiences of different segments of the population, such analysis can also help policymakers to identify those sections of the population which might benefit most from policies designed to improve wellbeing.  For example, the findings presented here suggest there may be value in developing policies to address the tendency towards declining emotional wellbeing and vitality among older age groups.

What is a decent society? Europeans’ subjective evaluations

A ‘decent society’ – i.e. one which promotes the wellbeing of its citizens – has a number of different dimensions. The European Social Survey (ESS) is a rich source of data with which to understand the extent to which Europeans think they live in a ‘decent society’.

Introduction

As well as being interested in individual wellbeing, researchers and policymakers are concerned with wider societal wellbeing and how this can be influenced by contextual economic, social, political and environmental factors. The Decent Society Model uses sociological theory to specify what it is that promotes the wellbeing of residents and provides the conditions for all to thrive (Abbot et al, forthcoming). Whether a country can be considered a ‘decent society’ will depend of course on the objective conditions in that country. However, it is also relevant to ask whether countries are perceived by their residents as fit for purpose. The European Social Survey (ESS) is a rich source of data on people’s experiences and beliefs, helping us understand the extent to which Europeans think they live in a ‘decent society’. Data from Round 6 (2012/2013) of the ESS are used here to show how European countries compare with each other in terms of their subjective evaluations of society.

Constructing a Subjective Index of decent society

We have followed the Social Quality approach (Beck et al., 2001; van der Maesen and Walker, 2012) to define what constitutes a ‘decent society’. The Social Quality Model identifies four broad requirements for a ‘decent society’ as shown in Figure 1: economic security, social cohesion, social inclusion and empowerment.  The model does not assign priority to any single one of these ‘quadrants’; instead it implies that each of them affects all the others. The indicators used to measure Social Quality and how they are grouped into domains within the quadrants depend on the availability of suitable data across the desired range of countries.

Figure 1: The Social Quality Model

To construct a Subjective Index of how people perceive their society we selected those questions from ESS Round 6 (2012/2013) which provide indicators of people’s attitudes and evaluations towards the processes or requirements identified for each of the quadrants in Figure 1. We make use of items from the core questionnaire (repeated every round) as well as items from the Round 6 rotating modules on ‘Understandings and Evaluations of Democracy’ and ‘Personal and Social Wellbeing’. Indicators were grouped into ‘domains’ within quadrants on theoretical grounds and using principal component analysis as a guide to clustering.

The resulting Subjective Index should be taken as indicative of people’s views only. Combining data from such a large number of indicators means that, cumulatively, there is a considerable number of cases for which there is some missing data, which may distort representativeness.¹ Despite those limitations, however, the Subjective Index constructed provides an innovative approach to the detailed measurement of different dimensions of societal wellbeing and an exploration of how complex subjective perceptions of wellbeing vary across Europe.

Findings

Some of the key patterns in ESS respondents’ perceptions of whether their country constitutes a ‘decent society’ according to the Social Quality Model are discussed below.

Perceptions of a ‘decent society’ across Europe

Figure 2 shows the overall scores on the Subjective Index for 28 of the countries that participated in ESS Round 6.² The country with the highest Subjective Index score – i.e. where people have the most positive views of their society – is Norway, closely followed by the other Scandinavian countries together with Switzerland. At the bottom lies Ukraine, together with most of the other former socialist states of central and eastern Europe. Portugal and Cyprus are also part of this lowest scoring group. In the middle, at or just above the mean, falls the rest of western Europe plus Israel; Germany and the Netherlands score highest within this group, and Italy, Poland, Estonia and Slovenia are at the lower end of the middle grouping.

There is a significant correlation between the Index score and headline indicators of wellbeing such as subjective life satisfaction (ρ=0.58) and happiness (ρ=0.54).  However, these headline measures share only a quarter of their variance with the Index. So more is at stake than just whether the country offers a satisfactory life in general; substantial variance remains to be explained even after controlling for happiness/general satisfaction.

As well as variation across countries, there is also variation in people’s perceptions across different sub-groups within countries. Men are significantly more likely than women to experience their society as decent, although the difference between the sexes is small. There are significant and steady increases in positive perceptions as income rises and as education level increases (see Abbott et al. (forthcoming) for more details).

Comparing different quadrants of the Social Quality Model

What is of interest for policy and practice is not only the overall score or ranking, but also how it is made up – which of the different conditions for wellbeing the country is seen to meet by its residents (see Table 2 for scores on each of the four quadrants across countries).

Figure 3 demonstrates that countries vary in terms of their relative scores on the four quadrants of the Social Quality Model. Results for different quadrants can be displayed by selecting or deselecting them in the legend.

Norway, Denmark and Sweden rank high across all four quadrants. Switzerland and Finland perform well in terms of Economic Security and Social Cohesion in comparison to most other countries but score relatively low in terms of Social Inclusion. Iceland, on the other hand, scores relatively low on Economic Security and Empowerment but high on Social Inclusion. Residents in Germany, Israel, Italy, the Czech Republic, Albania and Bulgaria rate Social Inclusion in their countries higher than any of the other dimensions.

The correlations between a country’s score on the overall Subjective Index and individual quadrant scores are high and significant (Table 3). It is notable however that the correlation with Social Inclusion is noticeably lower than for the other three quadrants. There are also significant and reasonably strong correlations between each of the four individual quadrants, although again the correlation between Social Inclusion and the other three quadrants is noticeably lower. Whilst there is a strong element of ‘general approval’ across different aspects of society – scores on one quadrant are fairly predictive of scores on the others at the country level – Social Inclusion appears to be evaluated differently (sometimes better, sometimes worse) to other aspects of a ‘decent society’ in many countries. Further work is needed to understand why this might be the case.

‘Drilling down’ further beyond the four quadrants, it is possible to analyse which of the constituent domain scores making up each quadrant are having most effect on overall perceptions of a ‘decent society’ and which individual indicators are important. For example, many of those countries which rank lower on Social Inclusion than we might expect given their overall ranking (Norway, Denmark, Sweden, France, UK) have relatively low rates of self-reported church attendance whilst other countries which perform relatively well in terms of Social Inclusion compared with their overall ranking (Ukraine, Bulgaria) are distinguished by relatively high rates of church attendance. This suggests an interesting area for further research – the church may be promoting social inclusion and welfare where this function is not taken on by governments. Our data suggest, however, that a low rate of church attendance may be partly compensated by high social activity/involvement in France and by high level of tolerance of minority groups in the UK. Low levels of political inclusion, on the other hand, appear to decrease the overall national Social Inclusion score (e.g. in Portugal and Cyprus) despite the relatively high rate of church attendance.

Conclusion

The ESS provides valuable insights into what constitutes a ‘decent society’ by allowing us to construct detailed measures of the extent to which people perceive their society to be a decent one  and to explore how these perceptions vary across countries.  In combination with more objective indicators on how far the conditions in a country meet the requirements for a ‘decent society’, such insights can offer potentially useful guidance to policymakers seeking to identify what aspects of their political and social regime may be in need of reform in order to promote citizens’ wellbeing.