The worst off in Europe - country differences and trends over time in (low) life satisfaction

Abstract

In recent years, policymakers and researchers have shown increased interest in subjective well-being across countries. While previous research primarily focused on country averages, measuring the distribution of subjective well-being through standard deviation has become more frequent. This article introduces a new approach to assess subjective well-being: focusing on the “worst off," or the group with the lowest levels of well-being. Based on several ethical and political theories, this measure is deemed the most relevant when assessing well-being levels in society. The study constructs new measures of low subjective well-being (the bottom 10%) to evaluate differences across countries, changes over time, and associations with economic growth, using data from 33 European countries from 2002 to 2018. The findings indicate significant variations in well-being for the worst off across countries, with improvements observed in almost all countries studied, particularly in Poland, Germany, and the Czech Republic. Improvements are generally larger for the worst off compared to the general population. Furthermore, both GDP per capita and financial satisfaction are positively associated with the subjective well-being of the worst off, both over time and when countries are compared cross-sectionally. The implications of these findings for future research and benchmarking quality of life are discussed.

Keywords:

• Subjective well-being
• worst off
• life satisfaction
• economic growth
• inequality

Introduction

Subjective well-being (SWB) concerns people’s life satisfaction (LS) and their experiences of positive and negative affect (Brülde 2007; Diener 1984). In this paper, we focus on LS as the measure of SWB. There are at least three viable approaches to benchmark quality of life in different countries (or other groups) based on SWB. The most common approach is to investigate average levels of SWB across countries. Studies based on this approach have found considerable variation in average levels of SWB between countries (Diener et al. 2009; Inglehart et al. 2008). According to the most recent World Happiness Report (Helliwell et al. 2023), people are most satisfied with their lives in Finland and least satisfied in Afghanistan. Another approach that has gained popularity recently is to assess inequality of SWB by comparing the standard deviation of SWB between countries (Veenhoven 2005). Several papers report that SWB-inequality is inversely correlated with average levels of SWB, but that this correlation is far from perfect (Ott 2005; Helliwell et al. 2016). For instance, the Netherlands have lower SWB-inequality than their average SWB-level would predict (Ott 2005; Helliwell et al. 2023).

In this paper we concentrate on a third approach that has largely been neglected in cross-country studies of SWB, namely, to focus on the worst off (in terms of SWB) in a population. A limited number of studies have examined SWB among the worst off using cross-country data. However, these studies typically emphasize differences between social groups within countries rather than between countries (Lelkes 2013; Clark et al. 2017; Flèche and Layard 2017). Alternatively, they analyze the situation of the worst off within the broader framework of happiness inequality (c.f. Ifcher and Zarghamee 2016; Clark et al. 2016). An exception is Krueger et al. (2009), who compared the amount of time individuals spend in a negative emotional state, as measured by the U-index, across France and the USA, and a recent study by Melios et al. (2023) who studied low SWB scores across the world by using the Gallup World Poll.

We argue that understanding SWB within this group is important from a normative point of view as seminal philosophical literature identifies this group as particularly deserving of concern. Yet, since contemporary cross-national research on SWB has not focused on this group we do not know if previous results are universally transferable across the SWB spectrum or not. Thus, our primary objective is twofold: to introduce theoretical arguments for focusing on the worst off individuals in society and to empirically analyze this group. The theoretical contribution primarily serves to justify why social scientists should study the worst off, outlining the normative relevance of this focus, rather than theorizing about the determinants of SWB for this group. Subsequently, our empirical investigation adopts a more explorative approach. We adapt the worst off perspective on three areas where previous SWB research has made strong empirical contributions in understanding cross country differences. First, we examine cross-country differences in the SWB of the worst off individuals. Second, we explore changes in SWB over time for this group. Finally, we analyze the relationship between economic growth and the SWB of the worst off.

Reasons for focusing on the worst off

We argue that most moral and political theories that concern subjective well-being support a focus on the worst off. These theories include different forms of utilitarianism, prioritarianism (Rawls theory of maximin being one example), and sufficientarianism.

Two caveats bear mentioning. First, not all theories above hold life satisfaction to be the standard by which to judge states of affairs, or the value to be maximized for the worst off. For instance, Rawls (1971) argues that we should maximize social primary goods for the worst off, whereas utilitarians typically value either happiness, or satisfaction of preferences (depending on whether one is a preference utilitarian or a hedonistic utilitarian). Second, utilitarianism doesn’t focus on the worst of qua worst off. Instead, utilitarians typically focus on the net maximization of either pleasure or preference satisfaction, depending on whether one is a hedonistic utilitarian or a preference utilitarian (Tännsjö 1996). Despite these caveats, we argue that the aforementioned theories support our position: from a moral standpoint, the most disadvantaged individuals constitute the most crucial group to study within the domain of SWB, particularly the life satisfaction dimension of SWB. From a theoretical standpoint, life satisfaction is conceptualized as an individual’s overall assessment of her life (Brülde 2007). To possess high subjective well-being in this sense means to be content with all significant facets of one’s life, with minimal or no inclination toward desiring change. Conversely, low life satisfaction is characterized by dissatisfaction with one or more important aspects of life, accompanied by a desire for significant changes or improvements.

Life satisfaction is important for two reasons. First, even though not necessarily the value identified as the most relevant by any of the aforementioned theories, life satisfaction can work as a proxy for other values, such as possession of primary goods, desire fulfillment, or hedonic well-being. Life satisfaction is likely the best existing empirical measure of desire fulfillment (or preference satisfaction¹), as it represents a subjective evaluation of how well one’s desires are fulfilled (Brülde 2007). Moreover, its strong correlation with affective well-being positions life satisfaction as a crucial indicator for hedonistic theories of well-being (Berlin and Fors Connolly 2019). In the context of Rawls’ theory, life satisfaction indirectly reflects the availability and utilization of primary goods—such as liberty, opportunity, income, and self-respect—since all these goods are correlated with high levels of life satisfaction (Steckermeier, 2021).

Second, the general nature of life satisfaction judgements is a pivotal advantage since it can function as a measure for unambiguous welfare comparisons across countries over time. This makes life satisfaction particularly useful for public policy purposes, where it can function as a criterion for cost-benefit analyses. Life satisfaction also resonates with ‘‘the Principle of Preference Autonomy’’ put forward by Harsanyi (1982), stating that when “…deciding what is good and bad for a given individual, the ultimate criterion can only be his own wants and his own preferences”.

Below, we describe each of the moral theories we believe support a focus on the life satisfaction of the worst off when assessing SWB.

Utilitarianism does not focus on the worst off explicitly, as the right action according to utilitarians is that which leads to more net happiness than any alternative action. It is irrelevant who experiences the happiness in question (Tännsjö 1996). However, the principle of diminishing marginal utility implies that increases in total utility is easiest to accomplish by improving the situation of those possessing the least of those things that generate utility (Diener et al. 2008).

Negative utilitarianism places a greater emphasis on the well-being of the worst off. The core idea of negative utilitarianism is that “an act is morally right if and only if it leads to less suffering than any available alternative” (Ord 2013). Even though not a (negative) utilitarian himself (Smart 1989:35), a quote from Karl Popper is often used to illustrate this position. Popper writes that “[H]uman suffering makes a direct moral appeal […] for help, while there is no similar call to increase the happiness of a man who is doing well anyway […] from the moral point of view, pain cannot be outweighed by pleasure, and especially not one man’s pain by another man’s pleasure. Instead of the greatest happiness for the greatest number, one should demand, more modestly, the least amount of avoidable suffering for all[…]” Popper (1945:235, Fn. 2).

Negative utilitarians view the importance of negative well-being overriding the value of positive well-being, regardless of quantities (Knutsson 2021). Thus, they view the well-being of the worst off as deserving the most attention when evaluating wellbeing in a population. Negative utilitarianism also implies that the absolute quantity negative well-being is of greatest moral importance. For example, in a country where the average life satisfaction has risen, and the relative number of people with negative life satisfaction has decreased, the situation is viewed as an improvement only if the absolute quantity of negative life satisfaction has decreased.

The basic idea of prioritarianism is that “[b]enefiting people matters more the worse of these people are” (Parfit 1995:19). The relative importance ascribed to changes in different groups or individuals is not specified by the theory, but the basic principle is clearly supporting a focus on the worst off. John Rawls famously proposed an extreme priority view where benefitting the worst off in society is lexically prior to benefitting any other group in society. This implies that even very small benefits to the worst off are to be prioritized over very large improvements to any other group in society (Rawls 1971).

Sufficientarianism is a moral theory that in its original form states that “what is important from the point of view of morality is not that everyone should have the same, but that each should have enough. If everyone had enough, it would be of no moral consequence whether one had more than others” (Frankfurt 1987:134). There are many versions of sufficientarianism, for example, Nussbaum’s (2006) capabilities approach and Elizabeth Anderson’s (1999, 2010) theory of democratic equality. There are even libertarian versions of sufficientarianism (e.g. Wendt 2018). Although different, the common denominator within sufficientarianism is that of a threshold or ‘par’. When people are below par, it is important to help them reach it, and if people reach par is reached, it is important that they don’t fall below it. When par is reached, however, it is either of no, or of lesser importance to benefit people further (Shields 2012). The theories above have different approaches and starting points but have in common that a focus on the worst off is important.

We have witnessed a surge in analyzes of trends in, and determinants of, SWB-inequality at the macro level through cross-country analyses (e.g., Ott 2005; Ovaska and Takashima 2010; Veenhoven 2005). Veenhoven (2005) motivates this shift by claiming that most people dislike inequality and for this reason inequality of SWB should complement average levels of SWB when assessing well-being in society. We agree that the inequality approach is an improvement to only assessing averages. However, as we argue below, studying the worst off either captures the underlying intuitions regarding the badness of inequality or provides a complementary perspective.

In what follows we present an argument presented by prominent philosophers such as Parfit (1995). Even if one does not agree with the arguments presented below, the worst off approach can still be motivated as an important complement to existing measures of SWB (i.e., inequality, average levels). According to Parfit (1995) the problem with inequality is not the fact that some people are better off than others, but rather that some people are worse off than they could be. At the face of it, it might seem that the two measures (inequality vs. the worst off) are describing the same phenomenon in different wordings. This, however, is not the case. We will illustrate why with the example of SWB (on a scale from 0 to 10) in two populations (A and B), consisting of five individuals each. In population A the SWB scores are 4, 8, 8, 9, 9. In population B, the scores are 6, 6, 6, 10, 10. The mean SWB is the same in both populations, but the inequality as measured by the standard deviation, is larger in population B. However, we believe that distribution B is better than distribution A, since the worst off are better off in that distribution. This example also illustrates a point made in the theory section. When viewing these two distributions, what is morally undesirable is not the fact that some people are very well off, thus making some people bad off relative to them (but not in absolute terms). Rather, what is morally undesirable is the fact that someone is bad off in absolute terms. Further, if inequality was intrinsically bad, distribution A would in some way be improved if the SWB of every individual scoring 9 and 8 would decrease to 4. The same goes for distribution B, if every individual scored 6 instead of some individuals scoring 10. This seems intuitively implausible and is known as the leveling down objection (Parfit 1995). Since the well-being of the worst off is one part of the equation that determines the inequality of SWB, inequality could serve as a proxy for the value we are interested in, but when the direct measure is available, that measure should be used instead of the proxy for it.

Second, it is possible that intuitions regarding the undesirability of inequality stem from the realm of material distributions, since the distribution of material goods at a given point in time is a zero-sum game; most material goods are finite, meaning that the consumption of a good means that others are excluded from consuming that same good. The case of life satisfaction is somewhat different since only some determinants of LS may have a zero-sum character (eg. material goods, having a partner) while others do not (eg. having meaningful relationships, a sense of purpose in life). This makes inequality a less relevant measure for life satisfaction.

Our contribution

We aim to contribute to the literature on subjective well-being across countries by examining (1) cross country variation in SWB for the worst off and (2) changes over time for this group and (3) the association of socio-economic conditions and the SWB of the worst off over time. Most studies of life satisfaction, whether assessing changes within countries over time or cross-national differences, focus on average levels or inequality of life satisfaction, while all effectively ignore directly studying the worst off in terms of life satisfaction. Therefore, we examine the worst off’s life satisfaction, and country differences as well as changes within countries. The reasons for focusing on both current country differences and changes is that both perspectives are important in order to attain a basic understanding of the worst off in terms of subjective well-being. Examining changes over time could reveal time variant societal factors affecting the worst off in society. Focusing on country differences at a given point in time, on the other hand, could reveal relatively time invariant societal factors affecting the worst off in society, which is the reason for examining current country differences. Focusing on the well-being of the worst off is essential from a policy perspective, as it allows for a better understanding of the factors affecting their life satisfaction, enabling the creation of policies designed to improve the lives of the least advantaged members of society.

As possible determinants of differences in LS for the worst off within and between countries, we find economic factors to be the most obvious to examine. From a macro-perspective, the economy is the most volatile factor known to substantially predict LS (Sacks et al. 2012). Economic growth can have varying effects on the worst off in society. It could be claimed that economic growth does not always benefit the worst off, as the distribution of wealth may primarily favor middle and upper-class individuals. This can lead to increased economic inequality and a decline in the relative standing of the worst off. Conversely, the principle of marginal utility suggests that the worst off stand to gain the most from economic growth, even if their absolute gains are smaller than those of wealthier citizens. These two perspectives are not necessarily contradictory, but it remains an open question as to which of these effects plays a more significant role in influencing the life satisfaction of a country’s worst off.

Among possible variables to examine, we used GDP per capita and self-reported financial satisfaction of the worst off population. Regarding the former measure, we focus on GDP per capita since previous studies have shown that it is a robust predictor of LS both within and between countries (Sacks et al. 2012). Also, GDP per capita is one of the most frequently used standards to measure prosperity and has as such been highly influential in policy making across most, if not all, countries in the world. As for self-reported financial satisfaction, it is a more direct measure which, as opposed to GDP, specifically captures the economic situation of the population on which we focus in this study.

Method and analytical strategy

Data comes from the European Social Survey (ESS). ESS is a cross-sectional survey covering a representative sample of persons aged 15 and older in Europe. In each wave and country, face-to-face interviews with a random sample of approximately 1500-2000 respondents were conducted. The data cover the first nine waves of ESS (2002 to 2018) and include 33 countries. In the descriptive analysis we only use the 17 countries that participated in the first two waves as well as the last two waves. We apply this criterion to ensure a large enough number of respondents for each measuring point, as well as to avoid wave-specific events that could disproportionately impact the data. In order to maximize the statistical power in the explanatory longitudinal analysis, all 33 countries that have participated in more than one ESS wave were included. Tables A1–A4 present the means, standard deviations, and number of observations for all survey measures, disaggregated by country and survey wave. Table A5 provides the corresponding GDP per capita data, also organized by country and wave, with GDP per capita expressed as its natural logarithm.

To evaluate life satisfaction (LS), participants were asked to rate their level of satisfaction with their life as a whole on a scale from 0 to 10, with 0 being ‘Extremely dissatisfied’ and 10 being ‘Extremely satisfied.'

There is no obvious criterion for categorizing the “worst off”. For this reason, we chose two different criteria to capture a range of possible interpretations of the term. When creating these categories, we considered the issue of sample size. For instance, using the first percentile as a criterion would require a larger sample than that available in ESS and most other surveys. The “worst off” in this paper are defined as either the 10th percentile and below or the 50th percentile and below. We chose two different relative measures, the first measuring the 10% worst off in the population which is clearly at the bottom of the distribution while still being a large enough group not to be overly sensitive to survey bias/attrition. Our second measure is broader and assesses the 10-50% of the population who are worst off. The reason for including this measure is that it captures general tendencies in society which may affect societal phenomena such as social cohesion, trust, and the stability of the social structure.

GDP per capita was assessed by using data taken from the World Bank with estimates adjusted for purchasing power in each country. We use the natural logarithm of GDP per capita to account for substantial differences between countries.

To measure average financial satisfaction, we use the following question from the ESS: Which of the descriptions on this card comes closest to how you feel about your household’s income nowadays?' The possible responses are ranging from ‘Living comfortably on present income’ to ‘Finding it very difficult on present income’ on a four-grade scale. The variable was averaged per country to capture perceived financial satisfaction (Min = 1.25; Max = 3.25).

As an additional test we ran a series of two-level mixed models where we nested observations in countries. As mixed models are not restricted by missing case at certain waves (e.g. Gabrio et al. 2022) we incorporated all countries that had data for at least two waves (N = 33), spanning across all nine ESS rounds conducted between 2002 and 2018 (max = 9). The choice to exclude countries with only one observation was made because such observations contribute only to cross-country differences and not to within-country changes, which are the main focus of this analysis.

To be able to parse out differences between effects of changes in, e.g., GDP per capita from country-level cross-country differences, we used two measures of the independent variable. The first one is the country grand mean (between-country effects) over all waves. The second one is the deviation from the country grand mean at each wave (within-country effects). These variables are orthogonal in relation to each other and capture between-country and within-country differences, i.e., average differences between countries versus change within countries. All models were run as fixed effects models with a random slope for intercept and wave. The covariance structure was chosen based on model fit and parsimony (compound symmetry). Put in formula, the analyses can be described as follows:

1. Y_it = b₀ + b₁Wave_it + b₂$\overline{G}$_i + b₃(G_it − $\overline{G}$)+ ε_it

2. b₀​ = γ₀₀​ + u₀

3. b₁ = γ₁₀ + u₁

Life satisfaction in country i at time t is a function of wave and the independent variable(s) G (GDP per capita and/or financial satisfaction). The between part (b2G̅i) is the grand-country mean, and the within part (b3(Git − G̅)) is the deviation from the country mean at time t. Two denotes the fixed intercept (γ00) plus the random intercept (u0). Three denotes the fixed and random slopes for time.

Results

We begin by investigating country differences for the worst off by analyzing mean levels of life satisfaction for the bottom 10%, the group between 10% and 50%, and mean values for the full samples (whole populations) in ESS wave 8 and 9 (pooled). We examine the full samples since most previous studies on subjective well-being do this, which makes it a natural reference point. As shown in Figure 1, we observe large differences between countries for the bottom 10%, average life satisfaction ranging between 4.68 in the Netherlands, and 1.53 in Portugal. As a reference, the corresponding range between the extreme points in the full sample is 8.15 (Switzerland) and 6.24 (Hungary). Based on the information shown in the figure, the Netherlands stands out with 0.29 units higher life satisfaction for the bottom 10% compared to the country scoring second highest, Finland with 4.39. The average difference between the six countries following the Netherlands - Finland, Switzerland, Norway, Sweden, Belgium and Austria (in that order) being 0.11. Below Austria by 0.37 units, we find Spain (3.37) followed by Ireland (3.31), Germany (3.21), Czechia (3.19), Slovenia (2.94), Poland (2.92) and the United Kingdom (2.74). We then observe a third gap down to France (1.90) and Hungary (1.81). The country with the worst average life satisfaction for the bottom 10% is, as already stated, Portugal (1.53).

Figure 1. Country differences for the worst off by mean levels of life satisfaction for the bottom 10%, the group between 10% and 50%, and the full sample.

Regarding the next measure of the worst off, i.e. the group between the 10th and 50th percentile, the difference between the highest and lowest scoring countries is smaller than for the lowest 10%, being 2.47 units. Further, the exact rank order between countries differs somewhat compared to the bottom 10% measure. However, the correlation between the two measures is very strong (r = .96); hence, both measures overlap to a large extent and depict largely a similar picture regarding the situation of the worst off across both measures in the countries examined.

In the second step of our analysis, we examine the change in life satisfaction of the worst-off individuals across countries. Figure 2 presents the change in mean values of life satisfaction for the bottom 10%, the 10-50% of the sample, as well as the mean value for the full sample. In almost all countries, we observe a rise across all three measures in life satisfaction. Among the three measures we observe the largest improvement for the bottom 10%, followed by the group 10-50% and (hence) the smallest rise for the full sample. At the most general level, we note a large improvement in some countries (Poland, Czechia, and Germany) while others saw only small, or even no changes (Sweden, Finland and the UK). Based on the data, we divide countries into four groups according to the rise in life satisfaction for the 10% worst off (first, second, third, and fourth). The criterion for inclusion in the first group is an improvement of the 10% worst off above 1 point in life satisfaction, for the second group improvement is larger than 0.5 but smaller than 1, for the third improvements that are statistically significant, but smaller than 0.5, and finally for the fourth group, everything is virtually identical between the two time points.

Figure 2. Changes in life satisfaction from first waves (2002-2004) to the last waves (2016-2018) for the for the bottom 10%, the group between 10% and 50%, and the full sample.

At least two things are worth noting regarding the first group. The improvement in Poland for the lowest 10% stands out since this group’s life satisfaction improved by almost two points compared to a one-unit improvement for the whole population. Improvements among the bottom 10% in Czechia and Germany are not quite as large as in Poland but still noteworthy. Regarding the second group, we observe improvements of almost one unit in the Netherlands, Slovenia and Hungary, while life satisfaction increased in Austria and France by just over 0.5 points among the lowest 10%. Improvements for the lowest 10% in Belgium, Spain, Switzerland, and Norway are relatively small but still statistically significant. Finally, we observe almost no improvements in life satisfaction of the 10% worst off in Finland, Sweden, Portugal, and the United Kingdom. Ireland even has a slight decrease for this group.

A general pattern in the data is that we also observe improvements for the group 10-50% in almost all countries, and improvements for the group 10-50% are larger in countries where the improvements of bottom 10% were also large (r = 0.84). Regarding within-country changes in life satisfaction for the full samples we observe improvements in almost all countries, but these improvements were smaller than the improvements for the bottom 10% and 50%.

As a third and final step, we examine the relationship between economic conditions and the worst off. We model the effects of different economic conditions on life satisfaction and changes in those conditions on changes in life satisfaction for both the 10% worst off and the full samples. Given that the descriptive analyses showed the greatest variability among the 10% worst off, we focus on this group and compare the estimates with the ones obtained for the full sample. While GDP per capita and financial satisfaction measure different aspects of economic well-being, they are strongly correlated. Therefore, to avoid issues with collinearity, we run separate models.

Both models including only time (Model 0a and 0b) show, in accordance with previous analyses, that there is a slight increase in life satisfaction over time for both the worst off and the full samples. Focusing on GDP per capita and the average perceived financial satisfaction, we see that both groups are more satisfied with their lives in more prosperous countries, regardless of whether the prosperity is measured objectively (GDP per capita) or subjectively (through aggregated individual perception of their financial satisfaction). Table 1 shows that when GDP per capita and the average perceived financial satisfaction increase life satisfaction increases. More importantly, the effect sizes for the worst off (1a and 1b) are more than twice as large compared to the average level of LS (1b and 2b) for both GDP per capita and the perceived financial satisfaction, despite the fact that the average level of life satisfaction does include the worst off. In sum, the results from Table 1 show that improving economic conditions goes together with increased life satisfaction, but that such economic changes benefit the worst off the most. This pattern is consistent across countries as there are no significant random slopes of GDP per capita or financial satisfaction (not displayed in the table to keep the more parsimonious analyses).

Table 1. Mixed model.

	Model 0a Time	Model 1a GDP per capita	Model 2a Financial satisfaction	Model 0b Time	Model 1b GDP per capita	Model 2b Financial satisfaction
	10% worst off			Full sample
Intercept	2.17**	−23.24**	−4.66**	6.56**	−14.18**	1.34**
Wave	0.06**	−0.06	0.05**	0.05**	−0.01	0.04**
GDP per capita between		2.54**			2.04**
GDP per capita within		1.58**			0.67**
Financial satisfaction between			2.96**			2.26**
Financial satisfaction within			1.42**			0.74**
Repeated CS diagonal	0.13**	0.12**	0.10**	0.04**	0.04**	0.03**
Repeated CS covariance	2.14**	0.65**	0.34**	1.12**	0.25**	0.14**
Random: Intercept + wave	0.007*	.004*	0.003	0.003*	0.002*	0.001

** p < .001.

* p < .05.

Life satisfaction for the 10% worst off and life satisfaction for the full sample.

Discussion

Whereas most cross-national analyses of subjective well-being focus on average levels of life satisfaction, or alternatively, inequality of life satisfaction, our study focuses on the worst off. To our knowledge this is the first study that systematically investigates the bottom distribution of life satisfaction across countries as well as within countries over time. Given the normative importance of improving the situation for the worst off in society, we believe that our study provides an important addition to previous macro-level studies on subjective well-being and the broader literature on quality of life and well-being. Our study also provides a philosophical account for the importance of focusing on the worst off which can be taken as a point of departure for future research on subjective well-being. In addition, our study shows that economic growth is an important determinant of the life satisfaction of the worst off.

In our descriptive analysis covering 17 European countries we found large differences between countries in average life satisfaction of the worst off, especially when examining the 10% worst off. For instance, the 10% worst off in Portugal, Hungary and France display levels of life satisfaction close to the negative end of the scale (below 2) whereas the worst off in the Netherlands display life satisfaction scores close to the middle point of the scale (just under 5). A similar pattern across countries was observed for the 10-50% worst off but country differences were generally smaller for this measure compared to the narrower 10% measure. However, an even more striking result was observed for the full samples where country differences were even smaller than for both measures of the worst off.

We observed that the worst off have seen an improvement in life satisfaction in most of the 17 countries analyzed and the increase in life satisfaction was most pronounced for the 10% worst off. In some countries the improvement for this group was particularly strong. For instance, in Poland the 10% worst-off improved their life satisfaction by more than 1.5 units during the 15-year span analyzed, a truly remarkable outcome. To put this into perspective, this difference over time for the worst off in Poland is roughly equal to the difference in life satisfaction between the employed versus the unemployed reported in previous research using ESS-data (Ervasti and Venetoklis 2010). Compared to the 10% worst off, improvements were also substantial for the 10-50%, but much weaker for the full samples of each country.

The improvement in life satisfaction for the worst off, in most countries, aligns with previous research that has found an overall increase in life satisfaction across most European countries since the 1980s (c.f. Inglehart et al. 2008; Helliwell et al. 2016). However, our study extends these findings by suggesting that the observed improvements may be primarily driven by a disproportionate increase in life satisfaction among the worst off, as we found substantially larger gains in this group compared to the full samples.

Our analysis of 33 European countries reveals that GDP per capita and financial satisfaction are significantly correlated with higher life satisfaction among the 10% worst off. This relationship holds true both across countries and within countries over time. Additionally, the link between economic factors and life satisfaction is markedly stronger for the 10% worst off compared to the overall population (full samples).

Our focus on the worst off revealed several interesting empirical insights that contribute to the broader literature on subjective well-being and quality of life. First, the pronounced differences between countries regarding the well-being of the worst off suggest that societal factors may play a particularly significant role in the life satisfaction of the worst off. Second, our study highlights the large potential for improvement in life satisfaction among the worst off over time. The observed improvements in life satisfaction, particularly among the 10% worst off, suggest that life satisfaction can change quite dramatically for this group. This finding is encouraging, as it indicates that policies can make a meaningful difference in the lives of those who are struggling the most. Third and relatedly, our analysis underscores the strong influence of economic factors on the life satisfaction of the worst off. The strong correlation between GDP per capita, financial satisfaction, and higher life satisfaction among the 10% worst off suggests that economic policy and growth have the potential to significantly impact the well-being of the worst off. This finding aligns with the principle of diminishing marginal utility, which posits that the benefits of increased income and consumption are greater for those with less to begin with. As such, policies aimed at stimulating economic growth or improving financial satisfaction may be particularly beneficial for the worst off, over and above improvements for the population in general.

Limitations

Our study has several limitations. First, our descriptive analysis was based on data from 17 European nations which is not fully representative for Europe as a whole. However, in order to obtain a reasonably large sample size of the 10% worst off in each country as well as study trends over time, an exclusion of some countries was necessary. Further, other international datasets containing a larger selection of countries, like the World Values Survey, were not suitable given the importance of a large enough sample size.

Second, our cross-national comparisons may suffer from various forms of measurement issues. Even though single item indicators of life satisfaction seem to work reasonably well across countries according to some studies (c.f. Fors and Kulin 2016) various forms of methodological issues cannot be ruled out. For instance, the comparison between the worst off and the full samples may be problematic due to ceiling effects on the life satisfaction scale. Not least since the mean value of the full samples was close to 8 in many countries (10 being the maximum value on the scale). Further, we also note vast differences in life satisfaction scores among the worst off when comparing countries that are rather similar in terms of poverty rates (Nolan and Whelan 2011) and other objective indicators of misery (i.e., the Netherlands vs. France). These findings may seem counterintuitive and warrant further investigation. However, even if our cross-cross country comparisons may suffer from measurement issues, our time series analysis should be less affected by such issues since it is unlikely that reporting biases change significantly within countries over the fairly short time period studied. However, our times series analysis may be affected by other shortcomings. For instance, we acknowledge that our samples may not be fully representative of the population of the worst-off due to self-selection into participation. Since response rates vary over time and decline in most countries it is unclear to what extent the trends over time in life satisfaction may be affected by response bias.

Third, our conceptual decision to focus on life satisfaction. Subjective well-being is commonly conceptualized as having two major components: life satisfaction and affective well-being (affect). We restricted our analysis to life satisfaction since this component of subjective well-being is closer to values central to political-philosophical theories where the worst off play a crucial role, such as the one proposed by John Rawls, and because cross-national time series data on SWB is mostly restricted to measures of life satisfaction (rather than affect). However even if life satisfaction and affective well-being are strongly correlated across countries (Fors and Kulin 2016), cross-national comparisons of each measure are not interchangeable since some countries score higher on life satisfaction than affect (and vice versa).

Future studies

Based on the results from our study, we suggest several avenues for future research. First, factors driving the large variation in life satisfaction scores for the worst off across European countries over time should be studied further. Researchers may ask questions such as: What accounts for the remarkable improvement in life satisfaction for the worst-off in Poland? Why does the life satisfaction of the worst-off improve in most countries but not in others, such as Portugal? Although our analysis points to economic factors as one important explanation for improvements of the worst off, possible confounders and mediators in this relationship should be investigated. Second, our approach of investigating the worst-off in terms of life satisfaction could be extended to a regional level time series analysis. Analysis on the regional level should be especially suitable for assessing possible macro level determinants of the worst-off since the number of cases on the regional level are much larger than at the country level. Future research should also investigate determinants of life satisfaction of the worst off at the individual level and investigate if these determinants differ between different cultural and institutional contexts (such as countries). Another suggestion for future studies is to assess the SWB of the worst off across countries using other measures of SWB. For instance, to what extent do measures of affect mirror measures of life satisfaction across countries and over time? A focus on affect is warranted since it clearly is of normative importance and since measures of affect (especially momentary affect) may be less susceptible to cross-cultural reporting bias (Oishi 2010).

Conclusion

One aim of this study was to provide valuable insights and information that policy makers can utilize in their efforts to improve the lives and well-being of the most disadvantaged members of society. Moreover, we hope that our findings will contribute to ongoing discussions on how to effectively measure and assess the quality of life and welfare within a society from a policy standpoint. It is clear that most policy applications of SWB-measures (c.f. OECD 2011; Helliwell et al. 2016) have focused on average levels of whole populations rather than the worst-off. We do not suggest that a focus on the well-being of whole populations should be abandoned, but that it at least be complemented by an assessment of the worst-off. Further, focusing on the worst off will decrease SWB-inequality but in a way that avoids the leveling down objection.

Compliance of ethical standard statement

Data used in our article involved human subjects who consented to participate in the European Social Survey.

Disclosure statement

The authors declare that there is no conflict of interest.

Data availability statement

The data utilized in this study can be accessed publicly via the European Social Survey’s official website at https://www.europeansocialsurvey.org/.

Additional information

Funding

No funding was received for this article.

Notes

1 In this context we view desire fulfillment and preference satisfaction as interchangeable.

Appendix

Table A1. Means, standard deviations and number of observations for life satisfaction across countries and groups in pooled samples of ESS wave 1-2 and 8-9.

	Wave 1 and 2 (pooled)									Wave 8 and 9 (pooled)
	Full sample			10% worst off			10-50% worst off			Full sample			10% worst off			10-50% worst off
Country	m	sd	n	m	sd	n	m	sd	n	m	sd	n	m	sd	n	m	sd	n
AT	7.48	2.12	4474	3.13	1.38	448	6.57	1.08	1789	7.77	1.84	4503	3.73	1.38	451	7.14	0.80	1801
BE	7.43	1.87	3659	3.43	1.56	366	6.81	0.90	1464	7.50	1.65	3527	3.90	1.39	353	7.01	0.76	1411
CH	7.98	1.72	4176	4.09	1.38	418	7.47	0.68	1670	8.15	1.67	3063	4.37	1.38	307	7.59	0.60	1225
CZ	6.37	2.26	4325	1.88	1.09	433	5.24	1.00	1730	6.87	1.86	4651	3.19	1.01	466	6.02	0.82	1860
DE	6.76	2.30	5774	1.94	1.15	578	5.75	1.13	2309	7.59	1.99	5198	3.21	1.44	520	6.94	1.00	2079
ES	7.03	1.96	3356	3.00	1.33	336	6.20	0.81	1342	7.36	1.93	3613	3.36	1.56	362	6.59	0.88	1445
FI	7.95	1.60	4018	4.29	1.48	402	7.54	0.61	1607	8.01	1.57	3676	4.39	1.48	368	7.60	0.58	1470
FR	6.37	2.47	3306	1.33	1.12	331	5.20	1.14	1322	6.51	2.27	4066	1.90	1.13	407	5.46	1.09	1626
GB	7.03	2.07	3933	2.76	1.23	394	6.10	0.95	1573	7.29	2.07	4158	2.73	1.31	416	6.55	0.97	1663
HU	5.62	2.48	3153	0.97	0.86	316	4.29	0.90	1261	6.24	2.26	3249	1.80	1.02	325	5.11	1.00	1300
IE	7.58	1.92	4304	3.53	1.53	431	6.88	0.96	1721	7.25	1.92	4947	3.30	1.39	495	6.40	0.81	1979
NL	7.56	1.67	4237	3.80	1.58	424	7.13	0.66	1695	7.84	1.43	3352	4.68	1.41	336	7.48	0.56	1340
NO	7.71	1.74	3793	3.96	1.27	380	7.13	0.86	1517	7.90	1.67	2947	4.22	1.24	295	7.35	0.68	1179
PL	6.01	2.59	3803	1.03	0.88	381	4.62	1.06	1521	7.11	2.09	3160	2.90	1.28	316	6.10	0.86	1264
PT	5.67	2.13	3537	1.49	0.89	354	4.65	0.93	1415	6.39	2.36	2318	1.53	1.25	232	5.30	0.96	927
SE	7.82	1.72	3939	4.03	1.26	394	7.27	0.78	1576	7.91	1.71	3078	4.09	1.34	308	7.39	0.69	1231
SI	6.73	2.26	2936	2.08	1.38	294	5.68	0.90	1174	7.20	2.08	2620	2.94	1.44	262	6.29	1.01	1048

Table A2a. Means, standard deviations and number of observations for financial satisfaction of the 10% worst off in life satisfaction across countries (wave 1 – 5).

	W1			W2			W3			W4			W5
Country	m	sd	n	m	sd	n	m	sd	n	m	sd	n	m	sd	n
AT	2.45	0.94	220	2.79	0.99	219	2.81	0.93	233
BE	2.51	0.86	184	2.34	0.91	176	2.48	0.97	179	2.46	0.88	175	2.34	0.88	170
BG							1.42	0.61	138	1.36	0.60	222	1.33	0.62	242
CH	2.84	0.88	202	2.67	0.91	212	2.77	0.93	180	2.71	0.91	180	2.71	0.92	149
CY							2.36	0.79	99	2.48	0.95	120	2.23	0.94	105
CZ	2.05	0.79	133	1.93	0.81	268				2.18	0.83	199	1.94	0.93	235
DE	2.38	0.87	290	2.40	0.90	283	2.27	0.89	288	2.42	0.90	273	2.47	0.88	301
DK	3.09	0.88	148	3.24	0.86	147	3.19	0.90	146	3.35	0.73	160	3.13	0.86	156
EE				1.75	0.80	198	2.15	0.75	147	2.01	0.72	164	2.05	0.82	179
ES	2.35	0.90	167	2.55	0.95	164	2.65	0.83	185	2.50	0.88	253	2.50	0.97	188
FI	2.59	0.83	198	2.52	0.84	201	2.61	0.77	189	2.54	0.83	219	2.47	0.88	187
FR							2.40	0.80	199	2.48	0.80	207	2.36	0.83	173
GB	2.71	0.92	204	2.57	0.98	188	2.63	0.92	238	2.55	0.93	233	2.38	0.90	239
GR	1.87	0.81	254	1.90	0.86	240				2.07	0.92	206	1.67	0.78	270
HR										2.12	1.01	146	2.17	0.91	162
HU	1.84	0.73	167	1.93	0.81	148	1.67	0.73	149	1.78	0.82	152	1.75	0.78	155
IE	2.61	0.92	199	2.87	0.95	226	2.75	0.91	175	2.45	0.92	176	2.35	0.97	254
IL	2.17	0.94	242							2.18	1.01	238	2.23	0.97	206
IS				2.88	0.94	56
IT	2.78	0.90	118
LT													2.18	0.84	164
LU	2.88	0.96	152	2.80	0.99	162
LV										1.69	0.79	192
NL	2.86	0.90	234	2.69	0.92	186	2.59	0.94	188	2.77	0.91	176	2.54	1.01	181
NO	2.90	0.92	204	2.85	0.94	176	3.03	0.93	175	3.17	0.89	155	3.17	0.81	155
PL	1.89	0.63	209	2.02	0.66	171	2.20	0.69	171	2.26	0.72	160	2.18	0.74	172
PT	2.16	0.86	147	2.13	0.87	202	2.27	0.92	215	1.97	0.90	231	2.22	0.93	211
RU							1.59	0.73	236	1.82	0.79	244	1.89	0.77	254
SE	3.06	0.82	199	2.85	0.94	193	2.91	0.90	191	3.07	0.89	183	3.01	0.89	150
SI	2.47	0.90	150	2.64	0.91	143	2.53	0.94	145	2.68	0.92	127	2.57	0.98	138
SK				1.71	0.81	147	1.79	0.79	172	2.11	0.85	178	2.01	0.81	183
TR				2.05	0.90	185				1.82	0.81	232
UA				1.44	0.64	197	1.51	0.68	194	1.57	0.68	175	1.42	0.67	186

Table A2b. Means, standard deviations and number of observations for financial satisfaction of the 10% worst off in life satisfaction across countries (wave 6 – 9).

	W6			W7			W8			W9
Country	m	sd	n	m	sd	n	m	sd	n	m	sd	n
AT				2.78	0.91	179	2.62	0.84	200	2.58	0.89	247
BE	2.24	0.98	187	2.36	0.92	177	2.55	0.83	176	2.49	0.89	176
BG	1.25	0.53	225							1.40	0.69	215
CH	2.77	0.96	149	2.70	0.93	152	2.77	0.86	152	2.78	0.95	151
CY	1.81	0.79	111							2.13	0.92	77
CZ	1.92	0.84	194	2.35	0.88	209	2.20	0.90	223	2.37	0.91	231
DE	2.43	0.89	295	2.59	0.85	302	2.62	0.91	283	2.70	0.88	234
DK	3.14	0.85	163	3.13	0.84	150				3.30	0.83	155
EE	1.87	0.80	237	2.03	0.76	205	2.08	0.80	202	2.30	0.77	191
ES	2.11	0.91	187	2.3	0.94	192	2.30	0.89	194	2.54	0.95	164
FI	2.57	0.85	218	2.61	0.76	208	2.60	0.87	192	2.61	0.82	175
FR	2.40	0.77	197	2.41	0.78	191	2.28	0.83	206	2.34	0.74	199
GB	2.42	1.00	226	2.43	0.92	225	2.71	0.93	195	2.62	0.88	219
GR
HR										2.40	1.02	179
HU	1.62	0.75	198	1.97	0.78	166	2.24	0.85	156	2.41	0.90	158
IE	2.18	0.89	259	2.20	0.98	234	2.48	0.88	274	2.61	0.95	216
IL	2.32	0.98	238	2.16	1.01	251	2.31	0.99	251
IS	2.66	1.02	74				2.99	0.96	87	3.22	0.82	86
IT	2.40	0.88	92				2.09	0.83	253	2.22	0.93	265
LT	2.07	0.86	206	1.95	0.85	220	2.16	0.82	206	2.03	0.83	180
LU
LV										1.97	0.82	91
NL	2.57	0.92	183	2.48	0.95	190	2.75	1.04	167	2.87	0.99	165
NO	3.11	0.88	162	3.08	0.97	143	3.10	0.83	154	3.14	0.82	140
PL	2.19	0.74	187	2.39	0.64	160	2.53	0.73	167	2.48	0.68	146
PT	1.93	0.81	211	2.05	0.83	126	2.11	0.94	127	2.13	0.84	104
RU	1.96	0.83	240				1.84	0.87	232
SE	2.76	0.99	182	2.97	0.95	177	3.07	0.96	154	3.10	0.80	153
SI	2.56	0.90	124	2.39	0.86	122	2.56	0.96	130	2.62	0.98	131
SK	1.85	0.87	183							2.19	0.75	105
TR
UA	1.56	0.69	209

Table A3a. Means, standard deviations and number of observations for the 10% worst off in life satisfaction across countries (wave 1 – 5).

	W1			W2			W3			W4			W5
Country	m	sd	n	m	sd	n	m	sd	n	m	sd	n	m	sd	n
AT	3.14	1.48	224	3.11	1.28	224	3.17	1.42	239
BE	3.27	1.66	189	3.61	1.43	178	3.27	1.62	180	3.12	1.46	176	3.90	1.34	171
BG							0.14	0.34	139	0.09	0.29	222	0.21	0.41	243
CH	4.05	1.44	204	4.13	1.34	214	4.22	1.31	181	4.05	1.41	182	4.50	1.46	151
CY							3.82	1.63	100	3.26	1.30	121	2.80	1.33	107
CZ	1.87	0.99	135	1.89	1.13	298				2.3	1.02	200	2.00	0.97	238
DE	2.00	1.14	292	1.89	1.15	286	2.03	1.15	292	2.12	1.12	275	2.57	1.21	303
DK	4.87	1.64	150	5.23	1.44	149	5.17	1.65	150	5.51	1.49	161	4.90	1.58	158
EE				1.65	1.04	199	2.13	1.14	150	1.86	1.11	166	1.87	1.19	180
ES	2.85	1.19	171	3.17	1.45	166	3.75	1.50	187	3.54	1.26	256	3.76	1.39	189
FI	4.10	1.39	200	4.48	1.54	202	4.53	1.41	190	4.41	1.30	220	4.48	1.28	188
FR	1.09	1.08	150	1.52	1.12	181	1.20	1.03	199	1.36	0.99	207	1.29	1.04	173
GB	2.61	1.25	205	2.92	1.18	189	2.79	1.32	239	2.57	1.29	235	2.67	1.21	242
GR	1.38	1.11	256	1.73	1.09	241				1.14	0.9	207	0.95	0.82	271
HR										1.66	1.10	148	1.37	1.03	164
HU	0.90	0.89	167	1.06	0.82	149	0.47	0.59	151	0.60	0.72	154	1.16	0.84	156
IE	3.15	1.55	203	3.86	1.44	228	3.34	1.46	179	2.69	1.28	177	2.11	0.99	258
IL	0.82	0.87	248							2.69	1.53	245	2.79	1.22	225
IS				5.55	1.77	58
IT	2.44	1.31	119
LT													0.59	0.58	167
LU	3.46	1.78	155	3.38	1.74	164
LV										1.21	0.83	196
NL	3.91	1.55	236	3.68	1.62	188	3.81	1.34	189	4.20	1.56	178	4.39	1.55	183
NO	4.11	1.19	204	3.78	1.34	176	4.01	1.25	175	4.15	1.19	155	4.34	1.29	155
PL	0.89	0.86	210	1.29	1.00	171	1.76	1.15	172	2.05	1.26	161	2.32	1.21	174
PT	1.41	0.94	150	1.54	0.86	204	1.33	0.77	219	1.34	0.82	233	1.43	0.95	214
RU							0.47	0.58	242	0.83	0.85	250	1.08	0.89	259
SE	3.96	1.28	200	4.11	1.23	195	3.92	1.46	193	4.05	1.43	183	4.17	1.18	150
SI	1.62	1.25	150	2.70	1.50	144	2.50	1.42	147	2.66	1.31	128	2.59	1.34	140
SK				0.63	0.75	150	1.61	1.01	175	1.93	1.11	179	1.85	1.14	184
TR				0.17	0.38	185				0.01	0.11	234
UA				0.11	0.31	201	0.00	0.00	197	0.00	0.00	178	0.27	0.45	190

Table A3b. Means, standard deviations and number of observations for the 10% worst off in life satisfaction across countries (wave 6 – 9).

	W6			W7			W8			W9
Country	m	sd	n	m	sd	n	m	sd	n	m	sd	n
AT				2.93	1.51	180	3.76	1.30	201	3.72	1.46	250
BE	3.53	1.46	187	3.71	1.38	177	3.96	1.33	177	3.85	1.46	177
BG	0.04	0.19	225							0.47	0.50	218
CH	4.42	1.59	150	4.33	1.41	154	4.31	1.26	153	4.44	1.48	154
CY	1.86	1.26	111							2.62	1.49	78
CZ	2.22	1.14	198	2.60	1.22	215	2.99	1.11	227	3.46	0.96	239
DE	3.06	1.50	296	3.00	1.34	305	3.14	1.41	285	3.30	1.46	236
DK	5.28	1.90	165	4.62	1.61	150				5.19	1.60	157
EE	1.55	1.11	238	1.99	1.10	205	2.76	1.19	202	3.06	1.46	191
ES	1.93	1.30	189	2.46	1.33	193	3.30	1.57	196	3.45	1.56	167
FI	5.03	1.45	220	4.33	1.46	209	4.27	1.49	193	4.52	1.45	176
FR	1.34	1.12	197	1.39	1.04	192	1.93	1.11	207	1.87	1.14	200
GB	2.82	1.24	227	2.61	1.33	227	2.94	1.31	196	2.59	1.34	221
GR
HR										1.62	1.25	181
HU	1.04	0.86	200	1.65	1.02	169	2.02	1.04	161	1.59	0.96	165
IE	2.15	1.27	262	2.55	1.34	239	3.36	1.46	276	3.24	1.30	220
IL	3.05	1.64	249	2.68	1.33	254	3.69	1.47	255
IS	4.25	1.37	75				4.20	1.67	88	4.95	1.56	86
IT	1.59	1.35	95				2.18	1.43	259	2.44	1.30	272
LT	1.76	0.89	209	1.48	1.02	224	1.94	0.99	209	1.69	1.05	182
LU
LV										1.62	1.30	92
NL	4.33	1.58	185	4.36	1.33	192	4.41	1.47	168	4.95	1.30	168
NO	4.63	1.37	163	4.17	1.40	144	4.19	1.26	155	4.26	1.23	141
PL	2.34	1.35	189	2.24	1.23	161	2.93	1.32	168	2.87	1.25	148
PT	1.87	0.98	214	0.70	0.85	126	1.33	1.21	127	1.76	1.26	105
RU	1.44	1.04	246				1.31	1.01	242
SE	4.15	1.31	185	4.2	1.33	179	3.86	1.45	155	4.34	1.20	154
SI	2.59	1.17	126	1.70	1.20	122	2.50	1.30	131	3.42	1.49	132
SK	1.96	1.08	184							2.00	0.89	108
TR
UA	0.49	0.50	213

Table A4a. Means, standard deviations and number of observations for full samples of life satisfaction across countries (wave 1 – 5).

	W1			W2			W3			W4			W5
Country	m	sd	n	m	sd	n	m	sd	n	m	sd	n	m	sd	n
AT	7.55	2.13	2236	7.41	2.10	2238	7.53	2.07	2383
BE	7.44	1.94	1882	7.43	1.80	1777	7.41	1.87	1795	7.27	1.90	1755	7.51	1.65	1702
BG							4.66	2.67	1389	4.31	2.59	2220	4.74	2.63	2421
CH	7.96	1.72	2038	8.01	1.72	2138	8.03	1.69	1803	7.91	1.75	1818	8.14	1.65	1505
CY							7.41	1.75	995	7.00	1.87	1208	7.16	2.00	1069
CZ	6.27	2.25	1348	6.41	2.27	2977				6.57	2.08	1995	6.30	2.18	2378
DE	6.81	2.29	2916	6.70	2.30	2858	6.71	2.24	2912	6.84	2.25	2746	7.11	2.15	3027
DK	8.44	1.58	1494	8.46	1.48	1484	8.48	1.50	1497	8.52	1.42	1609	8.35	1.55	1572
EE				5.89	2.24	1985	6.38	2.21	1496	6.2	2.23	1653	6.52	2.29	1791
ES	6.93	1.98	1703	7.12	1.94	1653	7.44	1.78	1864	7.26	1.82	2552	7.32	1.75	1885
FI	7.91	1.65	1999	8.00	1.54	2019	7.99	1.51	1893	7.94	1.54	2195	7.94	1.52	1876
FR	6.36	2.54	1500	6.37	2.41	1806	6.32	2.48	1986	6.25	2.45	2070	6.21	2.45	1728
GB	7.02	2.12	2045	7.03	2.02	1888	7.13	2.03	2389	7.02	2.13	2345	7.10	2.09	2411
GR	6.26	2.38	2554	6.39	2.19	2404				5.98	2.31	2063	5.65	2.33	2707
HR										6.43	2.34	1477	6.15	2.37	1637
HU	5.61	2.48	1668	5.63	2.49	1485	5.31	2.58	1504	5.31	2.56	1537	5.84	2.48	1555
IE	7.46	2.05	2026	7.69	1.80	2278	7.48	2.02	1790	7.14	2.12	1762	6.46	2.28	2572
IL	6.42	2.68	2476							7.40	2.21	2441	7.41	2.19	2249
IS				8.46	1.45	571
IT	6.86	2.13	1190
LT													5.19	2.56	1670
LU	7.80	2.07	1542	7.82	2.11	1633
LV										5.88	2.41	1954
NL	7.62	1.65	2358	7.48	1.70	1879	7.48	1.61	1887	7.62	1.55	1772	7.69	1.50	1828
NO	7.76	1.69	2036	7.66	1.80	1757	7.76	1.67	1746	7.89	1.66	1548	7.93	1.64	1548
PL	5.84	2.63	2093	6.23	2.51	1710	6.67	2.41	1713	6.87	2.3	1607	7.01	2.22	1740
PT	5.76	2.25	1498	5.61	2.04	2039	5.46	2.14	2181	5.60	2.28	2329	5.87	2.26	2133
RU							5.19	2.57	2419	5.34	2.48	2496	5.60	2.44	2583
SE	7.80	1.73	1994	7.84	1.70	1945	7.82	1.76	1924	7.86	1.72	1830	7.91	1.68	1496
SI	6.57	2.38	1497	6.90	2.12	1439	6.97	2.18	1470	6.93	2.14	1274	6.97	2.15	1394
SK				5.58	2.57	1496	6.08	2.29	1748	6.37	2.24	1786	6.41	2.24	1839
TR				6.35	3.06	1850				5.52	2.88	2331
UA				4.34	2.44	2002	4.28	2.60	1964	4.03	2.63	1778	4.66	2.53	1894

Table A4b. Means, standard deviations and number of observations for full samples in life satisfaction across countries (wave 6 – 9).

	W6			W7			W8			W9
Country	m	sd	n	m	sd	n	m	sd	n	m	sd	n
AT				7.38	2.07	1792	7.66	1.85	2008	7.85	1.83	2495
BE	7.44	1.76	1868	7.45	1.71	1769	7.48	1.62	1763	7.53	1.67	1764
BG	4.34	2.68	2248							4.92	2.52	2173
CH	8.19	1.67	1491	8.08	1.70	1532	8.15	1.69	1524	8.15	1.65	1539
CY	6.90	2.39	1110							7.02	2.14	777
CZ	6.54	2.20	1979	6.64	2.09	2141	6.71	1.89	2265	7.02	1.81	2386
DE	7.48	2.05	2956	7.42	2.02	3041	7.52	2.01	2846	7.66	1.96	2352
DK	8.57	1.50	1648	8.36	1.62	1500				8.50	1.48	1565
EE	6.18	2.38	2373	6.40	2.22	2045	6.78	2.05	2019	7.02	1.97	1903
ES	6.90	2.34	1884	6.96	2.15	1922	7.30	1.94	1951	7.42	1.92	1662
FI	8.11	1.39	2194	7.92	1.58	2085	7.98	1.60	1923	8.04	1.52	1753
FR	6.40	2.46	1967	6.39	2.44	1911	6.54	2.27	2068	6.47	2.28	1998
GB	7.28	2.07	2268	7.16	2.12	2261	7.34	2.02	1957	7.24	2.12	2201
GR
HR										6.96	2.51	1803
HU	5.59	2.43	1996	5.83	2.26	1681	6.38	2.19	1603	6.10	2.31	1646
IE	6.72	2.23	2618	6.94	2.10	2381	7.26	1.89	2752	7.24	1.96	2195
IL	7.53	2.11	2487	7.38	2.18	2540	7.79	1.91	2544
IS	7.96	1.65	748				8.03	1.67	878	8.15	1.48	859
IT	6.68	2.35	948				6.82	2.10	2586	6.89	2.03	2712
LT	5.82	2.22	2085	5.81	2.30	2231	5.98	2.14	2081	6.23	2.34	1819
LU
LV										6.54	2.37	913
NL	7.77	1.56	1845	7.60	1.50	1918	7.77	1.51	1679	7.90	1.35	1673
NO	8.14	1.57	1622	7.94	1.70	1433	7.94	1.68	1543	7.86	1.66	1404
PL	7.10	2.27	1887	6.93	2.28	1607	7.16	2.11	1680	7.04	2.06	1480
PT	5.96	2.14	2136	5.80	2.55	1260	6.26	2.40	1268	6.55	2.31	1050
RU	5.79	2.33	2458				5.67	2.34	2413
SE	7.87	1.70	1844	7.90	1.68	1788	7.89	1.77	1543	7.94	1.64	1535
SI	6.98	2.20	1253	6.57	2.37	1219	7.06	2.18	1306	7.33	1.98	1314
SK	6.55	2.24	1834							6.14	2.23	1073
TR
UA	5.04	2.51	2128

Table A5. Log-Transformed GDP per Capita by country and wave.

	W1	W2	W3	W4	W5	W6	W7	W8	W9
AT	10.30	10.38	10.46	10.58	10.62	10.70	10.78	10.82	10.90
BE	10.27	10.34	10.41	10.51	10.54	10.62	10.68	10.74	10.83
BG	8.85	9.04	9.24	9.46	9.56	9.66	9.72	9.82	9.97
CH	10.51	10.54	10.61	10.81	10.86	10.94	11.00	11.07	11.11
CY	10.04	10.10	10.23	10.40	10.43	10.41	10.32	10.36	10.54
CZ	9.78	9.88	10.00	10.18	10.23	10.28	10.34	10.43	10.57
DE	10.25	10.30	10.37	10.50	10.51	10.66	10.71	10.77	10.88
DK	10.29	10.33	10.44	10.57	10.61	10.70	10.75	10.80	10.92
EE	9.24	9.48	9.72	10.01	9.93	10.12	10.23	10.29	10.43
ES	10.04	10.13	10.23	10.39	10.38	10.37	10.39	10.46	10.58
FI	10.23	10.28	10.38	10.54	10.55	10.62	10.63	10.66	10.77
FR	10.22	10.25	10.33	10.44	10.45	10.53	10.58	10.62	10.71
GB	10.23	10.32	10.39	10.48	10.46	10.52	10.60	10.66	10.73
GR	9.95	10.08	10.15	10.28	10.32	10.17	10.17	10.20	10.28
HR	9.35	9.51	9.64	9.87	9.90	9.94	9.99	10.04	10.20
HU	9.49	9.65	9.75	9.85	9.94	10.04	10.11	10.20	10.29
IE	10.39	10.50	10.61	10.75	10.64	10.71	10.78	11.14	11.26
IL	10.12	10.08	10.12	10.22	10.22	10.33	10.44	10.48	10.57
IS	10.37	10.39	10.52	10.62	10.64	10.62	10.70	10.80	10.92
IT	10.24	10.28	10.31	10.44	10.45	10.51	10.50	10.52	10.64
LT	9.15	9.40	9.58	9.86	9.81	10.04	10.19	10.27	10.43
LU	10.93	11.00	11.13	11.34	11.32	11.43	11.47	11.55	11.63
LV	9.11	9.31	9.54	9.81	9.74	9.90	10.04	10.13	10.26
NL	10.41	10.44	10.54	10.69	10.71	10.75	10.80	10.83	10.92
NO	10.54	10.56	10.77	10.93	10.92	11.04	11.11	11.01	11.07
PL	9.32	9.42	9.54	9.73	9.87	10.04	10.11	10.20	10.31
PT	9.88	9.94	10.03	10.15	10.18	10.20	10.24	10.30	10.41
RU	8.90	9.13	9.38	9.72	9.87	10.03	10.17	10.09	10.17
SE	10.31	10.37	10.44	10.62	10.60	10.71	10.74	10.80	10.86
SI	9.85	9.96	10.08	10.22	10.22	10.27	10.31	10.36	10.51
SK	9.42	9.56	9.72	9.96	10.05	10.17	10.24	10.31	10.31
TR	9.14	9.18	9.39	9.61	9.66	9.89	10.02	10.16	10.24
UA	8.44	8.63	8.85	9.06	8.96	9.09	9.28	9.23	9.38