Introduction

Previous research has extensively examined whether residing in urban or rural areas results in higher subjective well-being, typically measured as life satisfaction. This inquiry can be reframed as whether life in rural or urban areas offers greater fulfilment, contingent upon the distinctive environmental and social conditions of each setting. It is also plausible that variability within urban and rural experiences is so substantial that no consistent pattern can be identified. It is also possible that the observed differences in SWB are not directly attributable to the urban or rural context itself, but rather to the unique amenities, services, and characteristics typical of each environment, which may drive these variations.

Differences in SWB between urban and rural areas also vary across countries. This may be partially explained by socioeconomic conditions, environmental factors, and cultural contexts. For example, findings from China (Ma & Chen, 2020; Su et al., 2021) report higher SWB levels in urban areas, which aligns with results identified by Counted et al. (2024) across 100 countries. However, in Europe and the Americas, research suggests that rural living is linked to higher SWB (Hoogerbrugge & Burger, 2021; Okulicz-Kozaryn, 2024; Sørensen, 2021). Similar, more recent findings were reported by Huang et al. (2023), who observed higher SWB in rural areas within China. However, some studies report mixed or statistically insignificant results, highlighting the need for additional research to better understand these variations (Chen et al., 2023; Navarro et al., 2020).

Research in this field spans various spatial scales, from specific municipalities and regions (Navarro et al., 2020), to national analyses (Counted et al., 2024; Hoogerbrugge & Burger, 2024; Rüger et al., 2023) and cross-country comparisons (Kim, 2018; Prati, 2024). These variations in research highlight the different spatial scales that can be applied. Studies relying on proprietary data tend to offer more detailed insights into local dynamics, whereas those leveraging extensive pre-existing datasets, such as the World Values Survey, the European Values Study, the European Social Survey, or EU-SILC enable broader comparative analyses across regions or countries.

The environment plays a significant role in shaping individuals’ well-being, particularly in light of the global trend of migration to urban areas (Chan, 2013; Ghio et al., 2023). This migration is largely driven by the pursuit of improved living standards, economic opportunities, and access to the various amenities that cities offer. A key factor contributing to this shift is the presence of agglomeration economies – benefits that arise when businesses and individuals concentrate in cities. These economies foster greater opportunities, more innovation, and an enhanced quality of life. They are characterized by economies of scale, knowledge spillovers, and labour market pooling, all of which increase business efficiency and stimulate local economic dynamism (Capello, 2009; Glaeser & Gottlieb, 2009). Consequently, cities experience enhanced profitability, faster knowledge exchange, and wider market access, all of which contribute to both economic and subjective well-being.

However, agglomeration economies also give rise to agglomeration diseconomies, including traffic congestion, pollution, and rising living costs. These negative effects can reduce overall quality of life as urban areas become overcrowded and strained by excessive demand (Meijers & Burger, 2010). Such conditions can lead to stress, declining physical health, and lower well-being in densely populated urban environments. This creates a paradox in which the benefits of urban living may be negated by its drawbacks, rendering SWB outcomes highly variable depending on how these costs and benefits are balanced. Rural areas offer alternative advantages, generally including more tranquillity, natural surroundings, and possibly stronger community ties, and SWB may be enhanced through a slower pace of life and lower cost of living. Overall, the urban–rural SWB dichotomy reflects trade-offs between economic opportunities and environmental satisfaction, social cohesion, and personal space.

The complex nature of the relationship between urban and rural life and subjective well-being contributes to mixed findings across different spatial scales and datasets. The traditional dichotomy of “urban versus rural” should not be overly simplified. This balance is shaped by crucial factors including individual attitudes, institutional settings, and the unique contextual dynamics in each region. As opposed to simple categorization into urban or rural, these key factors play a decisive role and can significantly influence the comparative results of SWB in different countries.

The key idea of our study is that the urban–rural divide varies significantly depending on whether economic benefits or environmental quality are prioritized. In less economically developed countries, cities may attract people primarily due to jobs, services, and income opportunities, who place less emphasis on environmental or privacy concerns. Conversely, in more developed countries, rural areas are often preferred for their privacy, tranquillity, and superior environmental conditions, as residents increasingly prioritize personal space and quality of life. This framework helps to contextualize the ‘urban paradox’ (Carlsen & Leknes, 2022; Morrison, 2021), which refers to the phenomenon in which cities, despite being centres of wealth and opportunity, often report lower subjective well-being than do rural areas. Our concept suggests that the urban paradox may manifest differently depending on local prioritization of cultural activities, amenities, services, and the environment benefits and the environment. The questions we consider include:

  • Does the urban paradox affect all European countries and lower income groups in particular?

  • Is the paradox more prominent in affluent countries where rural areas can offer more economic and environmental benefits?

  • What living environment factors (green spaces, noise, pollution, etc.) contribute to a positive subjective perception of a living area?

These questions motivate deeper exploration of the urban paradox and investigation of whether it is predominantly a phenomenon of more economically developed countries. This study uses a broad scale sample of European countries. However, a thorough understanding of results based on large representative datasets is often limited by the variables that are potentially important for explaining the variation in SWB. Typically, differences in SWB between rural and urban areas are reported as simple differences, controlled for a set of the socio-demographic characteristics of respondents (Prati, 2024). However, this view only reveals part of the complex relationship that is at play. The question remains whether urbanity/rurality per se explains the variation in SWB, or if this relationship is instead indirectly influenced by a set of the environmental characteristics in which individuals live. Essentially, is there a direct association between urbanity/rurality and SWB, or is there a set of characteristics closely related to urbanity/rurality that ultimately shapes SWB?

A range of living environment attributes can act as full or partial mediators in the relationship between urbanity/rurality and SWB, and overlooking them can lead to inaccurate interpretations. In particular, when the relationship is not fully mediated and the indirect (mediated) effect and the direct effect exhibit opposing directions, they nullify each other, giving the impression of a seemingly negligible total effect. In such a case, simple differences in means would suggest no differences between urban and rural areas, while an alternative model would suggest either a positive or a negative relationship between urbanity/rurality and SWB.

This study applies mediation analysis to explain the relationships between various aspects of living environments and subjective well-being. We focus on two EU-SILC survey years: 2013 and 2018, covering 30 European countries. In both years, EU-SILC released a module focused on the well-being of the population, which paves the way for a more in-depth investigation of factors that influence well-being. The EU-SILC data is particularly valuable because it serves as the official source of information on living conditions across the European Union and allows for harmonized comparisons across member states. Further, the dataset is systematically broken down by degree of urbanization, which enables us to analyse the nuanced effects of urban versus rural living environments on subjective well-being. This comprehensive and reliable data enhances the validity of our findings and ensures that our research is aligned with current EU policies addressing living conditions and well-being.

In addition, the 2013 survey year included an extended set of questions useful for our study, including satisfaction with green spaces and access to amenities and services. These elements are crucial in shaping emotional and subjective well-being. Because urban and rural environments can influence well-being through variables including noise, pollution, crime, and the availability of green spaces and services, these environmental factors are included as potential mediators in our structural equation models. This approach sheds light on how these environmental factors mediate the relationships between urban/rural living and subjective well-being.

A key challenge in examining the urban–rural divide in subjective well-being lies in the absence of a universally accepted definition of urban areas; classifications vary significantly across national contexts and research objectives. This variability complicates cross-country comparisons and the interpretation of results. In our study, while we primarily employ a simple binary classification of urban and rural areas as officially adopted in EU statistics, we also conduct extensive robustness checks to address potential concerns related to this division. We utilize the NUTS2 regional variable from the EU-SILC dataset for a subset of countries, which offers a more refined categorization of urban and rural regions. We also incorporate more granular data from the Czech and Slovak subsamples, which include population size and density variables. Although these supplementary analyses are performed on a limited subsample of countries, they corroborate the key patterns observed in the full sample and ultimately reinforce the reliability and generalizability of our conclusions.

Conceptual Model

The living environment significantly shapes human values and behaviours, and impacts subjective well-being in both urban and rural settings (Mouratidis, 2017; Wong et al., 2021). This study builds on the framework proposed by Mouratidis (2017), which extends traditional categorizations by prioritizing economic benefits and environmental quality. As depicted in Fig. 1, the conceptual model posits that SWB outcomes are influenced not only by the physical characteristics of the environment, but also by how these characteristics align with the residents’ specific priorities.

Fig. 1
figure 1

Conceptual model

Full size image

Urban environments are associated with easy accessibility to services (Kyttä et al., 2015), public transport, and cultural facilities (Leyden et al., 2011). However, as urban populations increase, problems including environmental pollution (MacKerron & Mourato, 2009), crime (Ceccato & Brantingham, 2024), increased road traffic, and associated noise (Yuan et al., 2019) rise. Rural areas are assumed to offer more natural (green and blue) spaces, but generally poorer accessibility to public transport, health care, and services (Brereton et al., 2011; Smith et al., 2008).

In this study, we examine five environmental characteristics, categorized into two groups: (i) negative factors (noise, pollution, crime) and (ii) positive factors (green spaces, access to amenities/services). While negative factors could be recast as positive counterparts (e.g., tranquillity, cleanliness, safety) and vice versa, we chose this classification to facilitate clearer interpretation of the results.

Based on prior empirical research, we construct the following hypotheses, as illustrated in our conceptual model (Fig. 1):

  • H1 Urban residence relates to higher perceived noise, pollution, and crime, greater satisfaction with amenities, and lower satisfaction with green spaces. In contrast, we expect lower perceived noise, pollution, and crime, less satisfaction with amenities, and higher satisfaction with green spaces in rural areas.

  • H2.1 We posit a negative association between perceived noise, pollution, crime, and subjective well-being.

  • H2.2 We posit a positive association between satisfaction with access to amenities, services, and green spaces, and subjective well-being.

  • H3 Direct effects of residential type on subjective well-being vary across different national contexts.

  • H4 Urban-rural residence exerts an indirect effect on subjective well-being through its impact on the positive and/or negative attributes of the living environment.

Our primary methodological approach is structural equation modelling (SEM), which allows analysis of complex relationships between subjective well-being and urban/rural living. SEM has been widely applied in empirical contexts similar to this study (e.g., Chan & Wong, 2021; Gao et al., 2017; Mouratidis & Poortinga, 2020; van Campen & Iedema, 2007). While the original SEM framework proposed by Baron and Kenny (1986) has faced criticism from some researchers, it remains a prevalent tool in mediation analysis (Zhao et al., 2010). Critics argue that mediation analysis is unnecessary if there is no zero-order effect to mediate; however, Zhao et al. (2010) counter this perspective by illustrating that the zero-order effect is the sum of direct and indirect effects within the mediation framework. This is particularly pertinent to our study, as the direct and indirect effects can exhibit opposite signs, resulting in a null total effect – a phenomenon known as competitive mediation, as classified by Zhao et al. (2010).

According to Zhao et al. (2010), three distinct mediation scenarios are possible: complementary, competitive, and indirect-only mediation. The first two scenarios assume that both direct and indirect effects are significant, and that complementary mediation occurs when both effects have the same sign, indicating that they reinforce each other. In contrast, competitive mediation occurs when the direct and indirect effects have opposite signs, suggesting a more complex relationship. The third scenario, indirect-only mediation, posits that there is no significant direct effect of urban/rural living on SWB.

If complementary or competitive mediation is observed, this indicates that the impact of urban/rural living on SWB is only partially mediated by the factors examined in this study. This finding may suggest potential omission of other important mediators that could influence the relationship. Understanding the nuances of these mediation types is crucial, as they provide insights into how various environmental and satisfaction-related factors interact with urban and rural living conditions to affect SWB. Ultimately, clarifying these mediation dynamics enriches our analysis and offers a more comprehensive picture of the factors that influence subjective well-being in different living environments.

A key limitation of our study stems from the constraints of the dataset (EU-SILC). First, dichotomization of urban and rural areas is based on the DEGURBA classification, in which cities are defined as areas with more than 50,000 inhabitants, whereas similar studies often classify cities as those with populations exceeding 100,000. Additionally, the environmental variables included in the survey do not capture all potentially important factors that might explain differences in SWB between urban and rural areas.

To address these concerns, we perform extensive robustness checks using various approaches to operationalize urbanity and rurality. We employ the NUTS2 regional variable for a subset of countries to offer a more nuanced classification of urban and rural regions. We incorporate more granular data from the Czech and Slovak subsamples, which include population size and density, and allow us to evaluate the effects of these factors on SWB. These robustness checks ensure that our findings remain consistent and reliable across different definitions of urbanization, reinforce the validity of our conclusions, and mitigate concerns related to the limitations of the binary urban–rural classification used in our primary analysis.

Data and Methods

This section first outlines the data source and sample, then explains classification and variables, and presents our primary analytical methods. Control variables, PLS-SEM methodology, and indicator reliability/validity appear at the end of the section.

The analyses in this study are based on microdata from the 2013 and 2018 EU Statistics on Income and Living Conditions (EU-SILC, Cross UDB, 2020–09 version). The EU-SILC is a harmonized survey that collects data on income, living conditions, and social exclusion in the European Union and other European countries. The study utilizes a subsample of respondents who were interviewed, i.e., those aged 16 and older. The sample sizes of the subsets utilized in this study range from 2,860 observations in Iceland to 17,739 observations in Italy for 2013, and from 2,885 observations in Iceland to 24,040 observations in Greece for 2018.

The urban–rural classification used in our analyses follows the commonly accepted Eurostat approach, which categorizes spatial units based on the degree of urbanization (European Commission, 2021).Footnote 1 Because the degree of urbanization is one of the key variables in the current study, countries not reporting this variable (the Netherlands, Slovenia, and Germany in the 2018 dataset) were excluded.

We focus on the division of urban vs. rural residences (urban = 1, rural = 0) and exclude respondents living in towns and suburbs. To enhance the robustness of our analysis, however, we also conduct additional tests that include “towns and suburbs” (Subsection "Robustness Analysis"), which further reinforce the reliability of our conclusions.

First, we employ the NUTS2 regional variable in the EU-SILC survey to define metropolitan regions according to Eurostat’s classification (European Commission, 2021), selecting only countries with available NUTS2 data. Subsequently, we identify regions where all individuals in densely populated areas resided in cities with populations over 100,000, excluding those individuals living in towns and suburbs. This approach provides a more refined classification of respondents residing in urban areas. These robustness checks are applied to a subset of countries (indicated in Table 2), including Czechia, Finland, Slovakia, and Spain in 2013, and Czechia, Slovakia, and Spain in 2018.

Next, we take advantage of the more detailed structure of the Czech and Slovak subsamples of the dataset, which includes variables on the population size of respondents’ locations, rather than a simple binary classification of urban versus non-urban areas. Additionally, the Slovak subsample allows us to assign population density to each location, which we incorporate into further robustness analyses. These additional data enable a more granular analysis, and allow us to assess the effect of population size and density on subjective well-being, mediated by both negative and positive factors.

The outcome variable (SWB) is operationalized using a standard life satisfaction question: “Overall, how satisfied are you with your life these days?” (on a scale from 0: not at all satisfied to 10: completely satisfied). The mean values of SWB across urban, suburban, and rural residence are reported in Table 1.

Table 1 Mean values of SWB by urban/suburban/rural residence (2013 and 2018)
Full size table

The mediating variable reflecting negative characteristics is constructed from items representing potential problems at the respondent's location, as assessed by questions: “Do you have any of the following problems related to the place where you live: 1. too much noise in your dwelling from neighbours or from outside (traffic, business, factory, etc.); 2. pollution, grime or other environmental problems in the local area such as smoke, dust, unpleasant smells or polluted water; 3. crime, violence and vandalism in the local area? Yes/No.”

Both the 2013 and 2018 survey years include the variables above, however, the 2013 dataset uniquely provides two additional questions related to positive factors, representing further mediating variables used in our study: “Satisfaction with recreational or green areas” and “Satisfaction with living environment” referring to the respondent’s access to services (e.g., shops, public transport etc.) and the presence of cinemas, museums, theatres, etc. Both were measured on a scale from 0: not at all satisfied to 10: completely satisfied. These two additional variables allow us to explore further potential mediating effects in the relationships between urban vs. rural residence and subjective well-being, highlighting the possibility that positive environmental and amenity-based factors may impact this relationship differently across urban and rural contexts.

Moreover, the Slovak dataset (both 2013 and 2018) provides an exceptional level of detail, including data on local services and amenities – covering pharmacies, a variety of healthcare providers (physicians, paediatricians, dentists, gynaecologists), educational institutions, and cultural resources like cinemas, theatres, libraries, museums, and galleries. This rich dataset uniquely positions us to explore the mediating impact of amenities and service accessibility in the relationship between urban versus rural residence and subjective well-being (SEM 6-SK).

We present results from two types of analyses. The first analysis examines the relationship between urban vs. rural residences and SWB using OLS. However, because OLS assumes equal numerical distance between categories, it may be unsuitable for ordinal data. To address this issue, we conduct two robustness checks.

  1. 1.

    We apply a transformation method proposed by Van Praag (2007), which adjusts discrete choice variables to vary continuously over the real axis. This transformation enables the use of OLS as an alternative to ordered probit, often referred to as probit-adapted OLS.

  2. 2.

    We estimate ordered probit models, treating the dependent variable as ordered, to confirm the robustness of our results.

In the second analysis, we employ structural equation modelling for empirical analysis of the conceptual framework illustrated in Fig. 1. This analysis includes a series of models designed to examine the mediating effects of environmental and service-related factors on the outcomes of interest. We estimate five models, along with an additional model utilizing an extended dataset from the Slovak subsample:

  • SEM 1 investigates the mediating effects of pollution, crime, and noise.

  • SEM 2 focuses on the mediating role of satisfaction with access to amenities and services.

  • SEM 3 combines the mediating effects of pollution, crime, noise, and satisfaction with access to amenities/services.

  • SEM 4 assesses the mediating role of satisfaction with green spaces.

  • SEM 5 incorporates all mediators, and examines pollution, crime, noise, access to amenities/services, and satisfaction with green spaces.

  • SEM 6-SK examines the mediating effects of local-level availability of amenities.Footnote 2

The regression models are controlled for a standard set of household- and individual- level characteristics: age, gender, income, education, health (suffer from any chronic illnesses), tenure status, economic activity, marital status, and deprivation. Please refer to Table S1 in the Supplementary Information for details regarding the respective questions, categories, and recoding of the categories. The mean values of the key variables used in this study are presented in Tables S2 and S3 of the Supplementary Information.

To capture the complex interrelationships between environmental factors, services, and subjective well-being, we adopt a partial least squares structural equation modelling (PLS-SEM). We use partial least squares (Hair, 2017) with SmartPLS software (Ringle et al., 2022). Widely used in exploratory research, PLS-SEM is a second-generation approach known for its ability to handle measurement errors, to model complex structures, and to accommodate non-normal data distributions (Hair et al., 2022).

It is possible that the indicators may not fully capture underlying environmental characteristics, so we assess the reliability and validity of indicators using reflective measures. As shown in Table S4 in the Supplementary Information, most indicators exhibit outer loadings above 0.6, while the threshold of 0.708 indicates that the construct explains more than 50% of the variance in these indicators (Hair et al., 2019). However, according to Hair et al. (2022), indicator loadings between 0.40 and 0.70 can be retained if they hold conceptual importance for the construct. In evaluating scale reliability, we note that Cronbach’s alpha and Rho A measures fall below 0.6 in some instances, likely due to the small number of binary indicators, which can result in lower coefficients, as these scales may not fully capture the construct’s variability.

Results

Geography of Differences in SWB Between Urban and Rural Areas

Our first question asks whether individuals living in urban areas generally experience greater life satisfaction than those in rural areas. To analyse this, we apply ordinary least squares (OLS) regression to assess differences in subjective well-being across urban and rural areas, controlling for socio-demographic and economic characteristics. However, the pattern of subjective well-being across rural and urban areas cannot be uniformly applied across Europe, because there are variations in economic histories, cultural practices, degrees of urbanization, and social policies. To account for these differences, the findings are categorized into Western, Northern, Southern, and Eastern (post-communist) European countries, in accordance with the geographic regions outlined by the United Nations Statistics Division. Countries with a post-communist legacy are often a part of Eastern Europe regardless of their geographical location. We have adopted this categorization to highlight the distinct characteristics of their socio-political transformation.

Figures 2 and 3 illustrate differences in SWB between urban and rural areas, with Fig. 2 focusing on mean SWB levels and their variation across two time periods (2013 and 2018), and Fig. 3 breaking down these differences by income group and educational attainment in 2018, both before and after adjusting for socio-demographic and economic variables.

Fig. 2
figure 2

Differences in SWB between Urban and Rural Areas (2013 and 2018). Panels A and B display the mean levels of SWB in response to the question: “Overall, how satisfied are you with your life these days?” (0: not at all satisfied; 10: completely satisfied). Panels C and D present the differences in SWB between urban and rural residents, where grey [black] dots correspondent to differences based on models without [with] control variables (age, deprivation status, marital status, education, housing tenure, sex, economic activity, self-reported health, and income). Error bars indicate 95% confidence intervals. See Table 3 for list of country abbreviations. Source: Authors’ calculations based on EU-SILC Cross 2013 and 2018, 2022–09 version

Full size image
Fig. 3
figure 3figure 3

Differences in SWB between urban and rural areas broken down by income group and highest education attained (2018). Panels A through C present the differences in SWB between urban and rural residents, disaggregated by income group (based on equivalized disposable household income, with thresholds estimated at the country level). Panels D through F show the differences in SWB, broken down by the highest level of education attained. Grey [black] dots correspondent to differences based on models without [with] control variables (age, deprivation status, marital status, education, housing tenure status, sex, economic activity, self-reported health, and income). Error bars indicate 95% confidence intervals. See Table 3 for list of country abbreviations. Source: Authors’ calculations based on EU-SILC Cross 2013 and 2018, 2022–09 version

Full size image

Panels A and B of Fig. 2 demonstrate that life satisfaction is typically higher in Northern and Western European countries than in Southern and Post-communist countries. A comparison between Panels A and B shows that the patterns of variation between countries are consistent across the two time periods (2013 and 2018). Panels C and D of Fig. 2 further illustrate the differences in SWB between urban and rural areas, employing two distinct methodological approaches. The grey dots represent unadjusted differences, while the black dots depict differences after controlling for a set of standard socio-demographic and economic characteristics. Detailed outputs of the regression analyses, including control variables, are available in Tables S5 and S6 of the Supplementary Information.

Panels C and D show that, in Northern and Western Europe, rural living is generally associated with higher life satisfaction, while in post-communist countries, urban residents report greater life satisfaction. However, after adjusting for socio-demographic and economic factors (income, age, education, health, etc.), these differences reduce and even reverse in some cases. The findings from Southern European countries align more closely with those of Western and Northern Europe, where, after adjusting for control variables, urban living is typically associated with lower SWB than is rural living. While urban areas offer higher incomes and better educational opportunities, the “urban paradox” persists, as urban residents often report lower SWB. The OLS regression results show no significant effect of education on SWB differences between urban and rural contexts, indicating that factors beyond education may play a more prominent role here. As shown in Panels A, B, and C of Fig. 3, educational attainment does not significantly explain these differences, nor does income.

To gain deeper understanding of the mechanisms underlying the differences described in this subsection, we next turn to the role of living environment characteristics. In the following section, we employ a series of structural equation models to explore how various environmental factors mediate the relationships between urban and rural residence and subjective well-being across diverse European contexts, as outlined in the conceptual framework.

Structural Equation Model Analysis

The previous section shows that SWB differences between urban and rural areas vary across countries and also show spatial patterns. Adjusting for socio-demographic variables results in minor changes, but the core patterns remain. Income significantly predicts SWB, but the main trends persist. Some country groups show higher rural SWB, while others show the opposite, reflecting local conditions and access to services. However, the insights gained from the OLS regression models may be influenced by factors and relationships not directly observable in the model results.

This section further investigates the mediating role of living environment characteristics in the relationships between urban/rural residence and subjective well-being. To explore the mechanisms that shape the relationships between urban living and SWB, we develop a series of structural equation models based on the conceptual framework outlined in Fig. 1. Each SEM offers a unique perspective on how various environmental factors shape the relationships between urban/rural residence and subjective well-being across different European contexts.

SEM 1: Mediating Effects of Pollution, Crime, and Noise

The results presented in Panels A and B of Fig. 4 suggest that urban residence is generally associated with higher degree of perceived crime, pollution, and noise than rural residence. These adverse environmental factors are negatively associated with SWB. The total indirect effect of urban versus rural residence on SWB, mediated by these negative characteristics, exhibits a consistent pattern across countries and survey years, though the effect is marginal in Slovakia in 2018 (p = 0.082).

Fig. 4
figure 4

The direct and indirect effects of the structural model. Panels A and B present the total indirect effects of urban residence (compared to rural) on SWB mediated by negative factors (noise, pollution, crime). Panels C and D present the direct effects of urban residence (compared to rural) on SWB. The models are controlled for age, deprivation status, marital status, education, housing tenure status, sex, economic activity, self-reported health, and income. The error bars represent 95% confidence intervals. See Table 3 for list of country abbreviations. Source: Authors’ calculations based on EU-SILC Cross 2013 and 2018, 2022–09 version

Full size image

In contrast, the direct effect of urban versus rural residence on SWB does not exhibit a clear pattern. The results presented in Panels C and D of Fig. 4 indicate that, while the direct effect is positive in a few countries, it is negative in others, and in some countries, no statistically significant effect is observed. Changes in the direction of this effect are noted in a few countries between the two survey periods. Overall, the findings related to the direct effect of urban versus rural residence on SWB do not provide sufficient evidence to support any generalizable conclusions. However, the patterns observed in this set of results align with those from the OLS models (Panels C and D of Fig. 2).

Detailed results from the SEM models, including control variables, are presented in Tables S7 and S8 in the Supplementary Information.

SEM 2: Mediating Effects of Satisfaction with Access to Amenities/Services

We next explore the mediating effects of positive characteristics of living environments, specifically respondents’ satisfaction with access to amenities and services and green areas in their residential surroundings.

Respondents’ satisfaction with access to amenities and services was approximated by a question regarding their satisfaction with the overall quality of the environment. The objective of this question was to capture satisfaction with services and the availability of amenities such as cinemas, museums, and other cultural and public services in their area. However, it is important to note that the phrasing of this question varied across countries (see Table S9 in the Supplementary Information). In some countries, the question was posed in general terms, asking respondents about their overall satisfaction with the quality of the environment, while in others, the question explicitly addressed satisfaction with access to amenities, cultural events, and public services. As demonstrated in Fig. 5 (see Table S10 in the Supplementary Information for details), the mediating effect of satisfaction with access to amenities and services exhibits substantial variation between the two distinct forms of question phrasing. As a result, subsequent analyses in this paper focus exclusively on countries where the question explicitly addressed satisfaction with access to amenities and services.

Fig. 5
figure 5

Mediated effect of satisfaction with services on SWB (2013). The figure reports the total indirect effects of urban residence (compared to rural) on SWB mediated by satisfaction with services in the environment. The ‘extended question’ specifically inquired about opportunities related to the living environment (e.g.: How satisfied are you in general with the shopping opportunities, access to public transport and cultural offerings in the area where you live?). The ‘general question’ asked about the respondents’ living environment in general (e.g.: How satisfied are you overall with the quality of the environment in which you live?) The models are controlled for age, deprivation status, marital status, education, housing tenure status, sex, economic activity, self-reported health, and income. Error bars represent 95% confidence intervals. See Table 3 for list of country abbreviations. Source: Authors’ calculations based on EU-SILC Cross 2013 and 2018, 2022–09 version

Full size image

The mediating effect of satisfaction with access to amenities and services is positive, and suggests that residing in urban areas is associated with higher levels of satisfaction compared to rural areas (see Panel A, Fig. 6). This increased satisfaction with access to amenities and services is positively correlated with greater subjective well-being. In contrast, the direct effect of urban residence on SWB is negative. These findings indicate that, after isolating the indirect effect of urban living on SWB, mediated by satisfaction with amenities and services, the remaining direct effect of urban residence becomes negative. This result is statistically significant across all European countries where the extended question was posed.

Fig. 6
figure 6

Mediating effect of negative factors, satisfaction with access to amenities/services, and green areas on SWB. The figure reports the indirect and direct effects of urban residence (compared to rural) on SWB mediated by satisfaction with access to amenities/services (Panel A); negative factors and satisfaction with services (Panel B); satisfaction with green spaces (Panel C); and the combination of all three (Panel D). The models are controlled for age, deprivation status, marital status, education, housing tenure status, sex, economic activity, self-reported health, and income. Error bars represent 95% confidence intervals. See Table 3 for list of country abbreviations. Source: Authors’ calculations based on EU-SILC Cross 2013 and 2018, 2022–09 version

Full size image

SEM 3: Mediating Effects of Pollution, Crime, and Noise, and Satisfaction with Access to Amenities/Services

When we incorporate negative characteristics of the living environment, including pollution, noise, and crime, alongside satisfaction with access to amenities and services (Panel B, Fig. 6 and Table S11 in SI), the qualitative pattern of the results remains consistent with those observed in Panel A. These findings suggest that the positive influence of satisfaction with access to amenities and services outweighs the negative impact of environmental stressors. When both negative environmental factors and satisfaction with services are considered as mediators in the urban/rural residence–SWB relationship, the total indirect effect becomes statistically insignificant in Luxembourg and Iceland. Nonetheless, the overall pattern of results demonstrates a consistent trend across most countries.

SEM 4: Mediating Effects of Satisfaction with Green Spaces

We also examine satisfaction with green spaces as a potential mediator in the relationship between urban–rural residence and SWB (Panel C, Fig. 6). Unlike the consistent indirect effects observed for negative environmental factors and access to services, the mediating effects of satisfaction with green areas vary across countries. The effect is statistically insignificant in two countries, negative in eight countries, and positive in three countries. As detailed in Table S12 of the Supplementary Information, the relationship between satisfaction with green areas and SWB is generally positive and statistically significant across all EU countries. However, the relationship between urban residence and satisfaction with green areas is negative in all EU countries except Finland, Croatia, and Serbia.

SEM 5: Mediating Effects of Pollution, Crime, and Noise and Satisfaction with Access to Amenities/Services and Green Spaces

Finally, when we consider all three sets of characteristics – negative environmental factors, satisfaction with services, and satisfaction with green areas – as mediators in the urban/rural residence–SWB relationship (Panel D, Fig. 6 and Table S13 in SI), the total indirect effect becomes statistically insignificant in six countries, positive in five countries, and negative in two countries. Nevertheless, in alignment with the partial models depicted in Panels A-C, the direct effect of urban residence on SWB remains negative in all countries, although it is statistically insignificant in Denmark and Iceland.

These findings provide robust evidence of the complex and context-dependent nature of the urban–rural residence–SWB relationship, and highlight the importance of considering multiple mediators to fully understand these dynamics.

Robustness Analysis

In our main analyses, we use a binary urban–rural classification (urban = 1, rural = 0), excluding respondents from towns and suburbs. To address potential concerns about this exclusion, we conduct robustness checks including a “towns and suburbs” category. The results, detailed in Supplementary Table S14 and Panel A of Table S15, for subsections "Geography of Differences in SWB Between Urban and Rural Areas" and "Structural Equation Model Analysis", remain consistent with the main findings, confirming the robustness of our classification approach even when intermediate categories are included.

We conduct further robustness checks using different urban–rural classifications to assess the reliability of our key findings, especially those derived from structural equation modelling (Table S16). We employ the NUTS2 regional variable for a subset of countries, which provides a more refined classification of urban areas. These checks, applied to Czechia, Spain, and Finland (2013), and Czechia and Spain (2018), show that the results are consistent with the main analysis.

The results remain consistent across partial models (SEM1–SEM4), and detailed regression results for robustness analyses are available from the authors upon request.

Additionally, we leverage the more detailed data from the Czech and Slovak datasets, which includes population size and density information. This allows for a more granular examination of the effects of population size and density on subjective well-being, with both positive and negative mediators. The results, shown in Tables S17 – S19, align with the initial binary classification results, and demonstrate the robustness of our conclusions across different definitions of urbanization.

The Slovak dataset also provides valuable insights into the availability of local services and amenities, allowing us to explore their mediating role in the relationship between urban–rural residence and SWB (Tables S18 and S19). The analysis from model SEM 6-SK indicates that better access to services positively mediates SWB, reinforcing the importance of local amenities as a factor in urban–rural disparities in well-being, as reflected in Panel A of Fig. 6.

Another concern arises from the fact that, in the SEM 2 through SEM 5 models, we regress subjective well-being on satisfaction with access to amenities, services, and green areas, effectively regressing one subjective phenomenon on another. The issue with this approach is the potential influence of a third factor that may affect both the dependent variable (SWB) and the independent variable (satisfaction) (Bartram, 2021). To address this concern, we mitigate the potential confounding effect by controlling for a variable that captures respondents’ emotional states, specifically their reported feelings of happiness over the prior four weeks. This question reflects a general disposition towards positive emotions, which may influence both an individual’s level of SWB and their perceived satisfaction with their environment, including access to amenities and services.

By including this control variable, we aim to isolate the specific relationship between the subjective variables under investigation, accounting for the influence of this potential underlying common factor. Results presented in Panel B of Table S15 in the Supplementary Information indicate that controlling for this third factor reduces the magnitude of the total indirect effect in the majority of countries; however, the direction of the effect remains unchanged, and it remains statistically significant. A similar concern applies to the regression of SWB on negative characteristics of the environment (pollution, noise, and crime), which are also self-reported and thus potentially influenced by the same third factor. As demonstrated in the tables, the total indirect effect remains robust even when we account for this potential bias in the case of negative environmental characteristics. This approach helps to ensure the validity of our findings by addressing the potential confounding influence of dispositional factors, thereby strengthening the robustness of the results.

The primary findings in subsection "Geography of Differences in SWB Between Urban and Rural Areas" are based on OLS estimates, with subjective well-being measured on an ordinal scale from 0 to 10. Because OLS assumes equal numerical distances between categories, which may not be appropriate for ordinal data, we conduct two robustness checks to validate the results. First, we use Van Praag (2007) transformation method, which adjusts discrete choice variables for continuous analysis. Second, we estimate ordered probit models to account for the ordinal nature of the dependent variable. Both methods, as detailed in Table S14, confirm the robustness of our results.

Discussion

The results of this study align with prior research on the relationships between urban and rural environments and subjective well-being, and offers new insights into the complex nature of the relationship between living environments and SWB. Previous studies have presented mixed evidence regarding whether SWB tends to be higher in urban or rural areas, emphasizing the role of environmental and social conditions in shaping these outcomes. By testing three central hypotheses, our study advances the debate using a rigorous methodological approach: leveraging mediation analysis through structural equation modelling to isolate the mediating effects of environmental factors including noise, pollution, crime, service availability, and access to green spaces on SWB.

One central question is whether income and education explain the differences in SWB between urban and rural areas. Recent studies by Lenzi and Perucca (2023) and Morrison (2021) suggest that higher education can help mitigate urban challenges. These studies propose a mechanism by which well-educated individuals benefit more from city life, via access to better employment opportunities and cultural amenities, which weakens the “urban paradox”. In contrast, less educated individuals may experience fewer of these benefits and may also face greater competition for resources along with negative externalities including higher living costs, congestion, and pollution. While income and education can indeed be important factors, our findings from this study do not demonstrate a significant impact on the urban–rural SWB gap.

Our hypothesis H1 proposes that urban residents experience greater exposure to the negative environmental factors of noise, pollution, and crime, and are generally more satisfied with amenities but less satisfied with green spaces than are rural residents. The findings of our structural equation models are in strong agreement with H1 in different European contexts. The results support earlier findings that urban living often correlates with increased exposure to noise, pollution, and crime, all of which are associated with lower levels of SWB, as documented by MacKerron and Mourato (2009) and Fisher et al. (2022). On the other hand, rural environments, characterized by greater tranquillity and access to green spaces, support higher SWB levels through the provision of natural amenities, which is also consistent with previous research (Brereton et al., 2011). These findings highlight the duality of urban and rural living environments and the distinct attributes that each setting offers. Our contribution here lies in the structural modelling approach, which allows a detailed breakdown of these environmental stressors as individual mediators, something prior research has not explicitly tested. The mediation models offer a refined causal pathway, and provide new clarity into how each factor independently and collectively influences well-being in urban settings.

Our study demonstrates that the relationship between urbanity/rurality and SWB is not straightforward and can be better understood through the lens of mediation. Hypothesis H2.1 highlights the negative environmental factors, especially in urban settings, that act as mediators to reduce SWB. We test this hypothesis using SEM 1 (and SEM 3), which include noise, pollution, and crime as mediators. The findings from these models support the hypothesis that urban living, characterized by these adverse factors, directly diminishes SWB despite potential economic or social advantages. This approach aligns with the broader literature on agglomeration economies, in which the benefits of economic and social concentrations in cities are often accompanied by negative externalities that undermine well-being, as proposed by Glaeser and Gottlieb (2009) and Meijers and Burger (2010).

Hypothesis H2.2 addresses the positive mediating effects of environmental characteristics, including access to services and green spaces, which have been shown to enhance SWB. We test this hypothesis through SEM 2, SEM 4 (and SEM 3). SEM 4 specifically focuses on green spaces as a mediator and confirms that access to green spaces significantly enhances SWB, particularly in rural settings where natural amenities are more abundant. This finding supports the appeal of rural environments for their environmental quality. SEM 5, which integrates both positive factors (green spaces, amenities) and negative factors (pollution, noise, crime), offers a comprehensive view of how these elements interact. This supports the hypothesis (H4) that urban–rural residence exerts an indirect effect on subjective well-being through its influence on the positive and/or negative characteristics of the living environment. Findings from SEM 5 indicate that, while services and amenities contribute to well-being in urban areas, the positive impact of green spaces on SWB is especially pronounced in rural settings. This aligns with prior observations by Wang and Wang (2016), and emphasizes that environmental quality is a crucial factor in enhancing SWB in rural settings.

The evidence of complementary and competitive mediation in our findings reflects the complexities of the relationship identified by Zhao et al. (2010), where the direct and indirect effects of urbanity/rurality on SWB can oppose each other. Hypothesis H3 proposes that the impact of urban versus rural living on SWB varies by national and regional contexts, with more affluent countries showing a stronger preference for rural living. We test this using SEM 6-SK, which incorporates detailed Slovak local data on amenities and population density. Findings confirm that, in more developed countries, rural areas tend to offer greater privacy, tranquillity, and environmental benefits, and contribute to higher SWB. This more detailed modelling approach demonstrates that the urban paradox – where cities, despite their economic advantages, often report lower SWB than rural areas (Carlsen & Leknes, 2022; Morrison, 2021) is driven by a trade-off between the economic opportunities and the environmental downsides of urban living. Our results also support the hypothesis that, in more developed countries, rural areas tend to offer greater privacy, tranquillity, and environmental benefits, leading to higher SWB.

The divergence in subjective well-being patterns between urban and rural areas across Europe, spanning all income levels, points to a more profound narrative. It suggests that the urban–rural divide is less about economic capability and more about the pursuit of quality of life. In Western and Northern Europe, rural areas are valued for their tranquillity and connection to nature, reflecting broader societal preferences that extend beyond material wealth. In contrast, in post-communist countries, urban areas are more closely associated with aspirations for economic progress and improved access to services, reflecting a stage of development in which urbanization is equated with advancement.

While the findings of this study provide valuable insights into the factors influencing SWB in urban and rural environments, it is important to acknowledge several limitations that may affect interpretation of the results. This study relies on the EU-SILC dataset, which offers both advantages and limitations. One of the dataset’s primary strengths is its large sample size and comprehensive representation of national populations, which enable robust comparisons between urban and rural areas. However, the restricted range of available variables is a key limitation. Another limitation of our study is the simplified classification of rural and urban residences based on Eurostat’s approach, which may not fully capture the complexities of different living environments.

A significant limitation lies in the simplified classification of urban and rural residences based on Eurostat’s approach. While a threshold of 50,000 inhabitants was used to distinguish urban and rural areas, and additional robustness analyses were conducted at 100,000 inhabitants, this classification does not fully capture the diversity of urban environments. Urban areas vary considerably in terms of infrastructure, economic activity, population density, and social dynamics, all of which can influence SWB. Even among cities exceeding 100,000 inhabitants, substantial differences persist. For instance, megacities like London and Paris, along with large metropolitan areas such as Berlin, Barcelona, Madrid, and Rome, each with populations over one million, differ fundamentally from smaller urban settlements. These large cities often exhibit distinct socioeconomic patterns, levels of inequality, and access to services that can shape well-being in ways that smaller cities cannot replicate.

Finally, due to potential endogeneity issues such as self-selection, the data cannot support explicit causal claims. While this limits causal inference, our study effectively describes associations in disparities between urban and rural well-being. Future research should address causal identification to enhance understanding of the urban–rural divide in subjective well-being.

Concluding Remarks

Consistent with the urban paradox hypothesis, we find that subjective well-being is generally higher in rural areas. In Western, Northern, and Southern Europe, SWB in rural areas consistently exceeds that in urban areas across both survey years (2013 and 2018). However, in Post-communist European countries, no consistent pattern emerges over the same period. These findings challenge the simplistic assumption that urbanization inherently results in greater life satisfaction. The persistence of this pattern across all income groups suggests that the urban paradox is not merely a reflection of economic disparities but may instead be driven by cultural and contextual factors.

Importantly, structural equation modelling results that examine the mediating roles of both negative and positive factors in the relationships between urban–rural residence and SWB reveal a significant negative direct effect of urban residence on SWB. This further supports the urban paradox, indicating that, after accounting for potential mediators, urban living is negatively associated with SWB. These results underscore the importance of both sensory and cognitive dimensions of living environments. Place-related characteristics, such as noise and pollution, are perceived as critical detractors from well-being. However, our findings also reveal that positive factors – such as the availability of services and amenities – can enhance the quality of life, though they are often underappreciated by respondents in SWB surveys.

Looking ahead, future research should explore more granular classifications of urban areas, distinguishing between small towns, mid-sized cities, and large metropolitan regions, to better understand how urban diversity influences well-being. Longitudinal studies and causal identification techniques would further deepen understanding of the dynamic relationships between living environments and SWB. Additionally, cross-continental comparisons could test the generalizability of these findings in non-European contexts, where urbanization patterns and cultural factors differ.

In conclusion, this study underscores that the urban–rural divide in well-being cannot be fully explained by economic opportunities alone. Instead, it reflects a complex concurrency of environmental quality, social dynamics, and cultural preferences. Recognizing this complexity is essential for designing living environments that promote sustainable well-being across diverse urban and rural contexts.