Social Connections and Loneliness in OECD Countries

Report

16 October 2025

2. Social connections across OECD countries

Copy link to 2. Social connections across OECD countries

Abstract

Across OECD countries, most people report frequent social interactions and generally feel supported by their social network. However, a notable share of the population faces deprivations in social connectedness: on average, 10% feel unsupported, 8% report having no close friends and 6% experienced loneliness most or all of the time over the past four weeks. Different indicators provide different insights: strong outcomes in one area – such as frequent in-person interactions – do not necessarily translate into high-quality or supportive relationships, both across countries and for individuals.

Figure 2.1. Snapshot: Key social connections outcomes, 2022 or latest available year
Copy link to Figure 2.1. Snapshot: Key social connections outcomes, 2022 or latest available year

This chapter provides an overview of the structure, function and quality of social connections outcomes across OECD countries. It covers how people spend their time; who they spend it with; the composition of their social networks; experiences of loneliness; access to and receipt of different forms of social support; and various measures of relationship quality.

Figure 2.1 highlights results for a key set of such outcomes collected around 2022. People across all OECD countries reported frequent social interactions: on average, 67% say they interacted with friends or family at least daily over the past week. In 23 European OECD countries with available data, these interactions occurred much more often remotely – 29% of respondents contacted friends daily via telephone, text or social media – than in-person, with only 11% meeting up with friends at least daily over a given year.

On average across OECD countries, 90% of respondents have someone to rely on in times of need, and just 4% of people (across 25 OECD countries with available data) are dissatisfied with their personal relationships. Nevertheless, notable deprivations persist: on average 10% of people report having no one to support them should they need it, across 22 European OECD countries 8% of people have no close friends, and across 25 OECD countries with comparable data, between 3% and 11% of the population reported feeling lonely most or all of the time over the past four weeks.

These findings are based on international surveys fielded between 2022 and 2023, and therefore may still reflect the tail-end effects of the COVID-19 pandemic and its significant disruptions to (social) life. However, medium-term trend analysis (see Chapter 4) shows that the frequency of in-person social interactions have been declining since 2006, while remote social interactions have been rising, suggesting that the higher rates of remote interactions seen in this chapter are not pandemic-related anomalies but reflect longer-term shifts in social behaviour.1

A key insight is that strong performance in one area of social connection does not guarantee positive outcomes in others. For instance, there is no clear relationship between daily in-person socialising2 and dissatisfaction with relationship quality – indeed, some countries with low rates of daily socialising also have relatively low rates of dissatisfaction with relationship quality, while others have high rates of both (Figure 2.2).

This pattern also holds at the individual level (Table 2.1.).3 Frequency measures of social interactions (such as daily socialising or daily remote contact) are moderately correlated with one another, but show only weak correlations with perceptions of relationship quality (such as feeling lonely and dissatisfaction with relationships). This highlights the importance of measuring and analysing both quantitative and qualitative dimensions when assessing the state of social connections, as they provide distinct and complementary insights.

Figure 2.2. Strong performance in one area of social connections does not guarantee positive outcomes in others
Copy link to Figure 2.2. Strong performance in one area of social connections does not guarantee positive outcomes in others

Scatterplot of dissatisfaction with personal relationships and daily in-person interactions with friends, OECD EU-EFTA 23, 2022

	Get together with family on a daily basis	Get together with friends on a daily basis	Contact family (remotely) on a daily basis	Contact friends (remotely) on a daily basis	Felt lonely most or all of the time over the past four weeks
Get together with friends on a daily basis	0.23
Contact family (remotely) on a daily basis	0.27	0.13
Contact friends (remotely) on a daily basis	0.10	0.32	0.31
Felt lonely most or all of the time over the past four weeks	-0.01	-0.02	-0.02	-0.04
Dissatisfied with personal relationships	-0.02	-0.02	-0.05	-0.04	0.24

Note: Table displays weighted listwise Pearson correlation coefficients between quantitative and qualitative social connections outcomes from 23 European OECD countries in 2022. OECD EU-EFTA 23 refers to Austria, Belgium, Czechia, Denmark, Estonia, Finland, France, Greece, Hungary, Ireland, Italy, Latvia, Lithuania, Luxembourg, the Netherlands, Norway, Poland, Portugal, the Slovak Republic, Slovenia, Spain, Sweden and Switzerland.

Source: Eurostat (n.d.[2]), European Union Statistics on Income and Living Conditions (EU-SILC) – Scientific Use File (SUF) (database), 2022 six-yearly rolling module on “Quality of life”, https://ec.europa.eu/eurostat/web/microdata/european-union-statistics-on-income-and-living-conditions (accessed in October 2024).

How and with whom people spend their time

Copy link to How and with whom people spend their time

One way to understand social connections is to examine the basic patterns of people’s social interactions – such as how much time people spend with others versus alone, how often they interact in person compared to digitally, and the size of their social networks. While these indicators are based on self-reported data from household surveys,4 they are focused on the frequency and quantity of social interactions, making them quite different from indicators that address the perceived quality of people’s social lives – such as how supported people feel or how satisfied they are with their social relationships. These latter dimensions are explored later on in this chapter.

Frequency of social interactions

Frequent interactions with others provide important well-being benefits, including better physical and mental health (Holt-Lunstad and Smith, 2016[8]; Holt-Lunstad et al., 2015[9]; Hansen et al., 2017[10]; Amati et al., 2018[11]; van der Horst and Coffé, 2012[12]). They also help to strengthen social networks that offer access to labour market, educational and financial resources, in addition to emotional support (Lleras-Muney et al., 2020[13]; Chetty et al., 2022[14]). While social interactions take place in a variety of settings – including schools, the workplace and community spaces5 – this section focuses specifically on interactions with family and friends, given their relevance to the entire population (i.e., not just the school-aged, or those in the labour market).

Evidence from all OECD countries shows that a large majority of people frequently interact with friends and family (Figure 2.3). In 2022, on average 95% of respondents say they had at least one interaction – either in person or remotely – with friends or family they live with or who live nearby, over the past week. Two-thirds (67%) reported such interactions on a daily basis. Still, 5% of respondents reported no contact at all – neither in-person nor remote – with friends or family in the previous week, with this rate varying from as low as 1% to as high as 16% across countries. These findings highlight that although most people remain socially connected, a notable share of the population faces elevated risks of social isolation, and its associated negative impacts on well-being (see Chapter 3 for details on the groups most affected).

Figure 2.3. In all but three OECD countries, over half of the population reported some form of daily interaction with friends and family over the past week
Copy link to Figure 2.3. In all but three OECD countries, over half of the population reported some form of daily interaction with friends and family over the past week

Share of respondents who interacted (either in person or remotely) with friends or family who live nearby over the past 7 days, by frequency, OECD, 2022

Types of social interactions – in person vs. remote

In recent decades, a larger share of respondents have reported declining in-person social interactions with friends and family and increasingly frequent digital interactions – via phone and video calls, text messages, emails and social media platforms. The broader implications of this shift for the quality and nature of social relationships are explored in greater detail in Chapter 5. However, existing cross-country data outlined in this chapter already offer valuable insights into current patterns of in-person versus remote social engagement.

2022 data from 23 European OECD countries show that people are more likely to stay in touch with family and friends remotely than in person. On average, 72% of respondents report contacting family members (not living in the same household) at least weekly, and 68% say the same for friends. In contrast, only 45% of people report getting together weekly with family, and 47% with friends (Figure 2.4). National data from non-European OECD countries reveal similar patterns (Box 2.1). The higher frequency of remote interactions likely reflects, at least in part, the lasting effects of the COVID-19 pandemic, during which social distancing and confinement policies limited in-person contact and for some population groups, pushed many aspects of work, education and social life online. By 2022 – when these data were collected – many behaviours had returned to pre-pandemic norms, but some lingering effects may remain. Regardless, these patterns are consistent with long-term trends in the increasing use of digital communication, at a time when face-to-face interaction is falling (see Chapter 4 on trends).

Figure 2.4. In 23 European OECD countries, nearly half of people meet friends or family weekly, while over two-thirds stay in touch remotely
Copy link to Figure 2.4. In 23 European OECD countries, nearly half of people meet friends or family weekly, while over two-thirds stay in touch remotely

Share of respondents who get together with or contact family or friends, by frequency in a given year, OECD EU-EFTA 23, 2022

Box 2.1. National spotlight: Frequency of in-person vs. remote interactions in Japan and the United States

Copy link to Box 2.1. National spotlight: Frequency of in-person vs. remote interactions in Japan and the United States

Official data from the United States and Japan also confirm that remote interactions are more common than in-person contact. In the United States, as of 2024, nearly 80% of respondents reported getting together with friends or family fewer than three times per week – one threshold used to identify potential social isolation used by the United States Census Bureau in reporting data from its Household Pulse Survey (CDC, 2024[15]). In contrast, far fewer reported limited remote interactions: 47% talked on the phone and 21% texted others fewer than three times per week (Figure 2.5, Panel A).

In Japan, 2023 data show that 44% of respondents reported having in-person conversations with friends or family at least weekly over the course of a year. Slightly more – 47% – communicated this frequently via social media, and around one-third did so via phone calls (Figure 2.5, Panel B). However, nearly 10% of respondents say they never interact in person – a much higher share than in European OECD countries (Figure 2.4). Additionally, around 15% of Japanese respondents reported never calling or using social media to communicate with others (Figure 2.5, Panel B).

Figure 2.5. In the United States and Japan, remote interactions are more frequent than in-person contact – but sizable shares report little or no social connection
Copy link to Figure 2.5. In the United States and Japan, remote interactions are more frequent than in-person contact – but sizable shares report little or no social connection

The types of people we socialise with

Social circles extend beyond friends and family to include colleagues, classmates, neighbours, community members and even strangers – all of whom play important roles in building social capital and supporting social mobility (see next section). The Gallup World Poll’s Global State of Social Connections survey asked respondents whether they had interacted – either in person or remotely – with various types of people over the past week. Unsurprisingly, in nearly all OECD countries, the most common interactions were with nearby friends or family members, reported by 95% of respondents for the OECD average (Figure 2.6, Panel A; Figure 2.6, Panel B).

However, a majority of people across OECD countries also reported contact with a broader set of social ties: 85% interacted with neighbours, 83% with distant friends or family, 72% with members of shared-interest or belief groups, 71% with colleagues or classmates and 67% with strangers (Figure 2.6, Panel A). These patterns vary across countries: for example, the share of respondents interacting with strangers in the past week ranged from below 50% to over 80%; for colleagues and classmates, from under 65% to 80%; and for shared-interest groups, from 45% to 85% (Figure 2.6, Panel B).

Figure 2.6. Across OECD countries, people are most likely to interact with friends, family and neighbours either in person or remotely, and least likely to interact with strangers
Copy link to Figure 2.6. Across OECD countries, people are most likely to interact with friends, family and neighbours either in person or remotely, and least likely to interact with strangers

Size and composition of social networks

As discussed above, friends and family represent core components of most social networks, and positive relationships with both are key to realising the well-being benefits of social interaction.6 Data from 22 European OECD countries in 2022 show that people report, on average, having five close family members and four close friends (Figure 2.7, Panel A). However, a non-negligible share of the population reports social network deprivations. On average, 8% of people report having no close friends, ranging from 3% to as high as 13% across countries (Figure 2.7, Panel B). The share reporting to have no close family members is lower, at 3%, on average. Comparable data collected by the Pew Research Center in the United States show similar patterns: in 2023, 8% of adults reported having no close friends (Goodard, 2023[18]).

Network size and diversity – particularly beyond the strong ties with immediate family and close friends – play a crucial role in building social capital and supporting social mobility. For example, supportive relationships with colleagues are associated with higher productivity and greater job satisfaction (Patel and Plowman, 2022[19]), interacting with individuals from different socio-economic groups can encourage social mobility (Chetty et al., 2022[14]; Chetty et al., 2022[20]; Harris et al., 2025[21]), interactions with strangers or distant acquaintances (so-called “weak ties”) have been shown to improve societal inclusion (Granovetter, 1983[22]), and a diversity of social contacts – including with those with differing views – may foster greater tolerance and civic participation (Mutz, 2002[23]; Ikeda and Richey, 2009[24]; Quintelier, Stolle and Harell, 2012[25]).

Internationally comparable survey data on network diversity is scarce, but recent evidence from three OECD countries sheds light on selected dimensions (Figure 2.8). In 2022, 13% of respondents in the United States, 14% in France and 21% in Mexico reported having no friends with different political views. In both France and Mexico, 21% of respondents reported having no friends from a different religious background; in the United States the rate was only 13%. Official statistics from England further illustrate similar trends, showing that many people’s social networks are composed primarily of individuals who share similar characteristics (Box 2.2).

Figure 2.7. On average in 22 European OECD countries, people report having between four and five close friends or family members, while 8% report having no close friends at all
Copy link to Figure 2.7. On average in 22 European OECD countries, people report having between four and five close friends or family members, while 8% report having no close friends at all

Figure 2.8. Across three OECD countries the share of respondents who have no friends with different political or religious beliefs ranges from 13% to 21%
Copy link to Figure 2.8. Across three OECD countries the share of respondents who have no friends with different political or religious beliefs ranges from 13% to 21%

Share of respondents who report having no friends with political or religious beliefs that differ from their own, 2022

In addition to data collected on household or time use surveys, some researchers have used alternative approaches to measuring network composition using digital trace data from social networking platforms such as Facebook and LinkedIn to assess the size, structure and diversity of individuals’ social networks, as well as patterns of engagement with different network members (Bazzaz Abkenar et al., 2021[27]; Bailey et al., 2018[28]; Davis et al., 2020[29]). For example, research using Facebook data has shown that United States counties with more geographically dispersed social networks – measured by the share of friends living more than 100 miles away – tend to also have better socio-economic outcomes and higher social mobility (Bailey et al., 2018[28]). In the United Kingdom, Facebook friendships reveal high levels of homophily by social class (i.e., a tendency to associate with others of a similar class), but cross-class connections are linked to greater upward mobility (Harris et al., 2025[21]). These approaches provide new avenues for data sourcing, but come with important ethical considerations surrounding data use, privacy and consent – requiring careful consideration by analysts prior to data collection and analysis.

Direct mapping of social networks can be resource intensive, however understanding social networks can provide important information on social mixing and segregation. A different method of understanding network diversity, recently employed by Statistics Netherlands, leverages administrative data to estimate the size and diversity of social networks through indirect methods (Box 2.3).

Box 2.2. National spotlight: Social network composition in England

Copy link to Box 2.2. National spotlight: Social network composition in England

Data from the Community Life Survey in England provide insight into the composition of individuals’ social networks by asking how similar or dissimilar their friends are across a range of socio-demographic characteristics (Figure 2.9). In 2021-2022, over half of respondents reported that more than half of their friend group shared the same age (69%), education level (68%), religious background (56%) and ethnicity (80%). Additionally, more than 1 in 5 respondents said that all of their friends shared their own characteristics in terms of age (20%), education (22%), religion (22%), and ethnicity (37%). In contrast, only around one in ten respondents reported high diversity in their social networks – that is, having more than half of their friends differ from them across these dimensions.

Figure 2.9. In England, over half of respondents report their friends share similar socio-demographic characteristics
Copy link to Figure 2.9. In England, over half of respondents report their friends share similar socio-demographic characteristics

Share of respondents who report friendships with those who are similar/dissimilar to them, England, 2021-2022

Box 2.3. Methodological insight: Using administrative data to estimate the potential size and diversity of social networks in the Netherlands

Copy link to Box 2.3. Methodological insight: Using administrative data to estimate the potential size and diversity of social networks in the Netherlands

Information on social network composition – those with whom we live, study, work and socialise – provides detailed insight into societal segregation, including the socio-demographic profiles of who is most likely to (self-)segregate, and the types of places that may be more (or less) conducive to cross-group mixing. Getting detailed information on social network size and diversity can be difficult: either resource intensive, via official household surveys, or associated with complex data use considerations (and issues of non-representative samples), via social media data.

Statistics Netherlands (CBS) has recently employed a new method to indirectly measure social networks by using administrative data from tax, education and personal records of the entire registered population to identify potential neighbours, colleagues, family members, flatmates and classmates in an individual’s network (CBS, 2023[31]). Importantly, the network compositions in this dataset are based on administrative relationships, and therefore do not necessarily represent actual connections.¹ Therefore, CBS refers to the resulting information as a reservoir of potential contacts rather than the actual network of individuals. However, this approach illustrates how administrative data can illustrate macro-level patterns in social networks across Dutch society as a whole.

Research using these data has illustrated the socio-economic dynamics of social segregation (i.e., living closely to and having contact mainly with people similar to oneself). For example, one study finds that the dynamics of segregation vary for those born in the Netherlands compared to those born abroad (CBS, 2024[32]). For those born in the Netherlands, segregation levels are highest in rural areas and decrease with urbanisation; segregation levels also rise with income. The opposite patterns are true for those born outside of the Netherlands: individuals born abroad who live in urban areas are more segregated than those born abroad who now live in rural areas in the Netherlands. Similarly, individuals born abroad who have lower levels of income are more segregated than those born abroad who have high earnings (Figure 2.10).

Figure 2.10. Segregation rises with urban density and income for those of Dutch origin; the opposite is true for those born outside the Netherlands
Copy link to Figure 2.10. Segregation rises with urban density and income for those of Dutch origin; the opposite is true for those born outside the Netherlands

Linear regression coefficient, segregation score as dependent variable, by country of origin, 2009-2020

How people perceive their social relationships

Copy link to How people perceive their social relationships

Beyond the frequency and structure of social connections, the quality of interactions and the functional support they provide are essential elements to consider. This section captures these aspects through a range of indicators, including feelings of loneliness; the need for, receipt of, and provision of social support; satisfaction with personal relationships; and the perceived quality of relationships with friends and partners.

Loneliness

Loneliness is a subjective experience that arises when individuals feel undesirably alone or perceive that their interpersonal relationship needs are not being met. It is associated with a range of negative outcomes, including poorer physical and mental health (Holt-Lunstad and Smith, 2016[8]; Akhter-Khan et al., 2021[34]; Wang et al., 2018[35]), increased risk of premature mortality (Holt-Lunstad et al., 2015[9]), and reduced academic and workplace performance (Morrish, Mujica-Mota and Medina-Lara, 2022[36]; Matthews et al., 2019[37]; Maher et al., 2013[38]).

In 2022, the share of people reporting that they felt lonely “most of the time” or “all of the time” over the past four weeks ranged from 3% to 11% across 25 OECD countries,7 with an average of 6% (Figure 2.11).8 For a number of OECD countries using a slightly different measure – asking how lonely respondents feel in general rather than within a specific timeframe – results are broadly comparable, with the average share of people saying they feel lonely “always” or “often” ranging from 5% to 14% (Box 2.4).

Figure 2.11. Across OECD countries, the share of people who felt lonely most or all of the time over the past four weeks ranges from 3% to 13%
Copy link to Figure 2.11. Across OECD countries, the share of people who felt lonely most or all of the time over the past four weeks ranges from 3% to 13%

Share of respondents who felt lonely most or all of the time over the past 4 weeks, OECD 25, 2022-2023

Box 2.4. National spotlight: Loneliness in Canada, the United Kingdom, the United States and Japan

Copy link to Box 2.4. National spotlight: Loneliness in Canada, the United Kingdom, the United States and Japan

The most common approach to measuring loneliness in official data across OECD countries is a single-item question on whether individuals have felt lonely over the past 4 weeks (see Figure 2.11).

However, several countries use a slightly different measure, asking respondents how lonely they feel in general – a question included in nationally representative surveys in Canada, the United States, England and Japan. In 2023-2024, the share of people who reported feeling lonely “often” or “always” ranged from 5% in Japan to 7% in England, 12% in the United States and 14% in Canada (Figure 2.12).

Figure 2.12. The share of respondents in Canada, the United States, England and Japan who generally feel lonely ranges from 5% to 14%
Copy link to Figure 2.12. The share of respondents in Canada, the United States, England and Japan who generally feel lonely ranges from 5% to 14%

Share of respondents who feel lonely, Canada, the United States, England and Japan, 2023-2024

Loneliness is not a binary condition, and it can be further assessed by examining its intensity (Figure 2.13) and duration (Box 2.5). Data from 22 European OECD countries show that among those who reported experiencing some form of loneliness in the past four weeks, 43% described their loneliness as “very intense” (Figure 2.13).9 However, high prevalence of loneliness does not necessarily correlate with high intensity: some countries with elevated overall rates may have relatively lower shares of people experiencing intense loneliness (Schnepf, d’Hombres and Mauri, 2024[41]).

Figure 2.13. In 22 European OECD countries, of those who feel lonely, the share of *intense* loneliness exceeds 40%
Copy link to Figure 2.13. In 22 European OECD countries, of those who feel lonely, the share of <em>intense </em>loneliness exceeds 40%

Share of respondents who report their loneliness over the past 4 weeks was “very intense”, OECD EU 22, 2022

Box 2.5. National spotlight: Persistence of loneliness in Japan

Copy link to Box 2.5. National spotlight: Persistence of loneliness in Japan

Research shows that chronic loneliness – persistent feelings of loneliness over time – has particularly harmful impacts on long-term health and mortality (Shiovitz-Ezra and Ayalon, 2010[42]; Newall, Chipperfield and Bailis, 2014[43]; Sheftel, Margolis and Verdery, 2024[44]).

Data from the Government of Japan capture not only the overall prevalence of loneliness (Figure 2.14, Panel A), but also its duration (Figure 2.14, Panel B). Among the 81% of Japanese respondents who report experiencing some degree of loneliness (i.e., any response other than “never”), nearly half say these feelings have persisted for more than five years – by far the most common response.

This finding highlights that for many, loneliness is not a fleeting experience but a long-term condition with potentially serious consequences.

Figure 2.14. 5% of Japanese respondents feel lonely often or always, but of those who feel any form of loneliness, almost half say the feeling has lasted more than 5 years
Copy link to Figure 2.14. 5% of Japanese respondents feel lonely often or always, but of those who feel any form of loneliness, almost half say the feeling has lasted more than 5 years

As with all self-reported, perception-based indicators, cross-country differences in loneliness estimates may be influenced by various factors – including cultural response styles, social stigma surrounding loneliness, and differences in the socio-demographic composition of populations. Existing research shows how survey design, including the mode of data collection and potentially sampling strategy, can affect reported prevalence rates (Box 2.6).

Box 2.6. Methodological insight: Mode effects, sampling strategies and loneliness estimates

Copy link to Box 2.6. Methodological insight: Mode effects, sampling strategies and loneliness estimates

Survey mode can influence both how respondents interpret questions and how willing they are to disclose personal information. For example, interviewer-administered surveys (whether in person or by telephone) establish a different dynamic than self-administered formats; responses may also vary across in-person, telephone and web-based surveys. Research on mode effects in loneliness – the social connections outcome with the largest evidence base – can vary depending on the measurement approach used. For example, earlier research on the multi-item De Jong-Gierveld Loneliness Scale found no clear mode effect when comparing in-person vs. telephone, or interviewer-led vs. self-administered surveys (Tilburg and Leeuw, 1991[45]). However, a recent systematic review covering multiple measurement approaches finds that respondents report higher levels of loneliness in face-to-face surveys (40% for face-to-face, on average across studies) and lower levels in telephone-based surveys (15% for telephone) (Stegen et al., 2024[46]). On-going measurement efforts will provide data users with more detailed information on the psychometric properties of different loneliness measures, including mode effects (Paris et al., 2024[47]). Existing OECD measurement guidance for other thematic areas, such as subjective well-being, recommends that data producers always provide clear information on how data were collected, and to be consistent across survey rounds (OECD, 2025[48]).

Some recent reports suggest that sampling strategy may also affect loneliness estimates, in combination with – or perhaps in addition to – mode effects. The European Commission’s 2022 Loneliness in the EU project was composed of two surveys: while the main survey was administered via a consumer panel (a non-probability sampling method) across all 27 EU countries, a shorter version using probability sampling was conducted in four countries. The results showed lower loneliness estimates in the probability-based survey compared to the non-probability one (for example, 8% vs. 15% loneliness rates in France, within the EU4 vs. EU27 samples, respectively). Notably, survey mode also differed: the EU27 version was entirely web-based, while the EU4 version included telephone interviews for respondents unable to complete the survey online. Future research efforts by the team will seek to disentangle mode from sampling effects, to better understand how each may affect prevalence rates (Schnepf, d’Hombres and Mauri, 2024[41]).

The loneliness estimates shown in this report (Figure 2.11) and included in the OECD Well-being Database primarily draw from Eurostat’s European Union Statistics on Income and Living Conditions survey (EU-SILC), a household survey conducted using probability sampling by national statistical offices (European Commission, 2003[49]) and following the common framework for European statistics relating to persons and households (Regulation (EU) 2019/1700) (Eurostat, 2022[50]). In 2022 – the year in which these estimates were collected – 44% of participating countries used computer-assisted telephone interviews (CATI), 25% used computer-assisted personal interviews (CAPI), 21% used computer-assisted web interviews (CAWI), and 10% relied on paper-assisted personal interviews (PAPI). (Many countries using PAPI did so in combination with other collection methods; most all countries used mixed methods.) Full details on sampling and mode for all surveys referenced in this report, including national-level instruments, can be found in the Annex (Annex Table 2.A.1 and Annex Table 2.A.2).

Support received by and given to others

Social support can take many forms. The literature typically distinguishes between instrumental support (tangible aid, including financial), informational support (resources to deal with a problem or issue) and emotional support (warmth and nurture) (Tay et al., 2013[51]; Wills, 1991[52]). Social support can provide resilience against poor physical and mental health outcomes (Freak-Poli et al., 2021[53]; Wang et al., 2018[35]; OECD, 2023[54]), and has been associated with better performance in school and the workplace (Patel and Plowman, 2022[19]; Saeed et al., 2023[55]; Holahan, Valentiner and Moos, 1995[56]).

Data from 2022-2023 across all OECD countries show that, on average, 90% of people report having friends or family they can rely on in times of need. However, a notable minority – 1 in 10 – say they have no one to count on, with this figure reaching up to 20% in some countries (Figure 2.15).

Figure 2.15. Lack of social support affects between 2% and 20% of people across OECD countries
Copy link to Figure 2.15. Lack of social support affects between 2% and 20% of people across OECD countries

Share of respondents who report having no friends or family to count on in times of need, OECD, 2022-2023

Data from 2022 across 22 European OECD countries show that, on average, 62% of people report “always” having someone who loves them and makes them feel wanted, 60% have someone to help if they are sick and confined to bed, 58% have someone to do enjoyable activities with, and 57% have someone to confide in about private fears or worries (Figure 2.16, Panel A). In most countries, emotional support in the form of feeling loved is the most commonly reported, while emotional support involving personal disclosure is the least prevalent.

Additional 2022 data from Mexico, the United States and France highlight that instrumental financial support is typically less commonly available. Only 25% of respondents in France, 33% in Mexico and 39% in the United States reported “always” having someone to lend them money – compared to over 60% who had someone to help them if they were ill (Figure 2.16, Panel B).10

Figure 2.16. Prevalence of social support differs by type: Financial assistance is the least available form in all OECD countries with available data
Copy link to Figure 2.16. Prevalence of social support differs by type: Financial assistance is the least available form in all OECD countries with available data

The data highlighted in the preceding paragraphs are based on survey questions that ask about the theoretical availability of different types of support – whether respondents could receive help if needed. However, when asked about actual experiences – whether they received support in the past 30 days, or provided support to others – evidence from Mexico, France and the United States shows that people are far more likely to report giving support than receiving it (Figure 2.17). This could reflect a genuine mismatch between support given and support received. Alternatively, it may stem from response biases – such as a tendency to overreport prosocial behaviours (e.g., helping others), underreport personal need, a lack of awareness of when one is receiving support (i.e. taking support for granted, or a reflection of the fact that support received is ineffectual and therefore not noticed or counted) or discomfort in acknowledging vulnerability due to stigma.

Figure 2.17. More people report giving social support, than receiving it
Copy link to Figure 2.17. More people report giving social support, than receiving it

Share of respondents who “often” gave someone support or help, or received it, over the past 30 days, 2022

Relationship satisfaction

The quantity of social relationships – such as the size of one’s network or frequency of interactions – is only part of the equation: it is equally, if not more important, to understand the quality of these relationships. For example, while having a partner is one of the strongest predictors of reduced loneliness (Arpino et al., 2022[57]), the nature of that relationship matters greatly. Marital stress is associated with higher levels of loneliness (Hawkley, 2008[58]) and strained marital or family relationships – characterised by constant arguing, critical comments, or excessive demands – are associated with trauma and adverse health outcomes, especially harmful to children and young people (Thomas, Liu and Umberson, 2017[59]; Robles et al., 2014[60]; Alm, Låftman and Bohman, 2019[61]; Giletta et al., 2021[62]).

Satisfaction with relationships in general

In 2022, evidence from 27 OECD countries with comparable data show that respondents rate their satisfaction with personal relationships at an average of 8 on a scale from 0 (not at all satisfied) to 10 (completely satisfied) (Figure 2.18, Panel A, see also national data from Colombia and Mexico in Box 2.7). When looking at relationship dissatisfaction (defined as the share of respondents reporting below the scale midpoint of 5) – the data reveal that while countries with higher average satisfaction scores also tend to have lower shares of dissatisfied respondents, there is substantial cross-country variation (Figure 2.18, Panel B). Among 25 OECD countries with comparable data, on average 4% of respondents are dissatisfied with their personal relationships, with individual country scores ranging from 1% up to 9%.

Figure 2.18. Satisfaction with personal relationships in 27 countries ranges from 6.5 to 8.6 on a 0-10 scale; 4% of respondents across 25 OECD countries are dissatisfied with their relationships
Copy link to Figure 2.18. Satisfaction with personal relationships in 27 countries ranges from 6.5 to 8.6 on a 0-10 scale; 4% of respondents across 25 OECD countries are dissatisfied with their relationships

Box 2.7. National spotlight: Satisfaction with relationships in Mexico and Colombia

Copy link to Box 2.7. National spotlight: Satisfaction with relationships in Mexico and Colombia

Average satisfaction with personal relationships is generally high in Colombia and Mexico. In Colombia, 78% of respondents rate their relationship satisfaction a 4 or 5 on a 1-5 scale, and only 1% report feeling dissatisfied (Figure 2.19, Panel A). In Mexico, 66% of respondents rate their satisfaction with personal relationships a 9 or a 10, on a 0 to 10 scale, with just 1% reporting dissatisfaction (defined as a score below 5) (Figure 2.19, Panel B).

Figure 2.19. More than one-third of Colombians, and two-thirds of Mexicans, are satisfied with their personal relationships
Copy link to Figure 2.19. More than one-third of Colombians, and two-thirds of Mexicans, are satisfied with their personal relationships

Relationship satisfaction with partners and friends

The 2022 EU Loneliness Survey also explored satisfaction with romantic relationships, specifically. Across 22 European OECD countries, respondents rated their happiness with their partner at an average of 7.6 on a scale from 1 to 10 (Figure 2.20, Panel A).

The 2022 pilot survey of the Gallup World Poll included a set of questions assessing the quality of friendships, focusing on the frequency of meaningful interactions and interpersonal conflict. In terms of positive engagement, 37% of respondents in Mexico, 47% in the United States and 54% in France reported “often” having meaningful interactions with friends (Figure 2.20, Panel B). This implies that in both Mexico and the United States, more than half of respondents do not frequently have meaningful interactions with friends – and in France, the figure is close to half. On the other hand, reported explicit conflict within friendships is relatively rare: only 4% of respondents in Mexico, and 3% in both the United States and France, say they often experience conflict with friends (Figure 2.20, Panel C).

Figure 2.20. Evidence across OECD countries sheds light on satisfaction with relationships to romantic partners, and the dynamics of friendships
Copy link to Figure 2.20. Evidence across OECD countries sheds light on satisfaction with relationships to romantic partners, and the dynamics of friendships

References

[34] Akhter-Khan, S. et al. (2021), “Associations of loneliness with risk of Alzheimer’s disease dementia in the Framingham Heart Study”, Alzheimer’s & dementia: The journal of the Alzheimer’s Association, Vol. 17/10, pp. 1619-1627, https://doi.org/10.1002/ALZ.12327.

[61] Alm, S., S. Låftman and H. Bohman (2019), “Poor family relationships in adolescence and the risk of premature death: Findings from the Stockholm Birth Cohort Study”, International Journal of Environmental Research and Public Health 2019, Vol. 16, Page 1690, Vol. 16/10, p. 1690, https://doi.org/10.3390/IJERPH16101690.

[11] Amati, V. et al. (2018), “Social relations and life satisfaction: The role of friends”, Genus, Vol. 74/1, pp. 1-18, https://doi.org/10.1186/S41118-018-0032-Z.

[57] Arpino, B. et al. (2022), “Loneliness before and during the COVID-19 pandemic—are unpartnered and childless older adults at higher risk?”, European Journal of Ageing, Vol. 19/4, pp. 1327-1338, https://doi.org/10.1007/s10433-022-00718-x.

[65] AXA (2023), Mind Health Report 2023, AXA Global Healthcare, https://www.axaglobalhealthcare.com/en/wellbeing/emotional/mind-health-report/.

[28] Bailey, M. et al. (2018), “Social connectedness: Measurement, determinants, and effects”, Journal of Economic Perspectives, Vol. 32/3, pp. 259-80, https://doi.org/10.1257/JEP.32.3.259.

[27] Bazzaz Abkenar, S. et al. (2021), “Big data analytics meets social media: A systematic review of techniques, open issues, and future directions”, Telematics and Informatics, Vol. 57, p. 101517, https://doi.org/10.1016/J.TELE.2020.101517.

[32] CBS (2024), Herkomstsegregatie in Nederland: een netwerkanalyse, Centraal Bureau voor de Statistiek, https://www.cbs.nl/nl-nl/longread/statistische-trends/2024/herkomstsegregatie-in-nederland-een-netwerkanalyse?onepage=true.

[68] CBS (2023), Sociale samenhang en welzijn, Centraal Bureau voor de Statistiek, https://www.cbs.nl/nl-nl/onze-diensten/methoden/onderzoeksomschrijvingen/korte-onderzoeksomschrijvingen/sociale-samenhang-en-welzijn (accessed on 12 March 2025).

[31] CBS (2023), Tijdreeks persoonsnetwerkbestanden: overzicht van de verschillen met de eerste versie van het persoonsnetwerk, Centraal Bureau voor de Statistiek, https://www.google.com/search?q=Tijdreeks+persoonsnetwerkbestanden%3A+overzicht+van+de+verschillen+met+de+eerste+versie+van+het+persoonsnetwerk&rlz=1C1GCEA_enFR1125FR1125&oq=Tijdreeks+persoonsnetwerkbestanden%3A++overzicht+van+de+verschillen+met+de++eerste+versie+van+het+persoonsnetwerk&gs_lcrp=EgZjaHJvbWUyBggAEEUYOdIBBzIxOGowajSoAgCwAgE&sourceid=chrome&ie=UTF-8 (accessed on 22 March 2025).

[15] CDC (2024), Lack of Social Connection, National Center for Health Statistics, Centers for Disease Control (CDC), https://www.cdc.gov/nchs/covid19/pulse/lack-socialconnection.htm.

[14] Chetty, R. et al. (2022), “Social capital I: Measurement and associations with economic mobility”, Nature, Vol. 608/7921, pp. 108-121, https://doi.org/10.1038/s41586-022-04996-4.

[20] Chetty, R. et al. (2022), “Social capital II: Determinants of economic connectedness”, Nature, Vol. 608/7921, p. 122, https://doi.org/10.1038/S41586-022-04997-3.

[78] Ciciolla, L. et al. (2017), “When mothers and fathers are seen as disproportionately valuing achievements: Implications for adjustment among upper middle class youth”, Journal of Youth and Adolescence, Vol. 46/5, pp. 1057-1075, https://doi.org/10.1007/S10964-016-0596-X.

[63] DANE (n.d.), Encuesta Pulso Social (database), Departamento Administrativo Nacional de Estadística, https://www.dane.gov.co/index.php/estadisticas-por-tema/encuesta-pulso-social.

[26] Data for Good at Meta (2022), Social Connections Survey, https://data.humdata.org/dataset/social-connections-survey.

[29] Davis, J. et al. (2020), “Networking via LinkedIn: An examination of usage and career benefits”, Journal of Vocational Behavior, Vol. 118, p. 103396, https://doi.org/10.1016/J.JVB.2020.103396.

[39] DCMS (2024), Community Life Survey 2023/24: Loneliness and support networks, Department for Culture, Media & Sport, https://www.gov.uk/government/statistics/community-life-survey-202324-annual-publication/community-life-survey-202324-loneliness-and-support-networks--2.

[30] DCMS (2023), Community Life Survey 2021/22: Identity and social networks, Department for Culture, Media & Sport (DCMS), https://www.gov.uk/government/statistics/community-life-survey-202122/community-life-survey-202122-identity-and-social-networks.

[17] e-Stat (2023), 人々のつながりに関する基礎調査, Government of Japan, https://www.e-stat.go.jp/stat-search/files?page=1&toukei=00000004&metadata=1&data=1.

[49] European Commission (2003), Commission Regulation (EC) No 1982/2003 of 21 October 2003 implementing Regulation (EC) No 1177/2003 of the European Parliament and of the Council concerning Community statistics on income and living conditions (EU-SILC) as regards the sampling and tracing rules, https://eur-lex.europa.eu/eli/reg/2003/1982/oj/eng.

[3] European Commission and Joint Research Centre (JRC) (2024), EU Loneliness Survey, Joint Research Centre (JRC) [Dataset] PID:, https://doi.org/10.2760/2083329.

[50] Eurostat (2022), EU-SILC EU Comparative Quality Report 2022, https://ec.europa.eu/eurostat/web/income-and-living-conditions/quality.

[71] Eurostat (2020), Ageing Europe - Statistics on social life and opinions, Statistics Explained, https://ec.europa.eu/eurostat/statistics-explained/index.php?title=Glossary:Community_survey_on_ICT_usage_in_households_and_by_individuals;%20https://ec.europa.eu/eurostat/statistics-explained/index.php?title=Ageing_Europe_-_statistics_on_social_life_and_opinions&oldid=581876#Contacts_between_older_people.2C_family_and_friends.

[70] Eurostat (2017), Meeting up with friends and relatives, https://ec.europa.eu/eurostat/web/products-eurostat-news/-/ddn-20170725-1.

[69] Eurostat (2010), Social participation and social isolation, European Commission, https://ec.europa.eu/eurostat/web/products-statistical-working-papers/-/ks-ra-10-014.

[2] Eurostat (n.d.), European Union Statistics on Income and Living Conditions (EU-SILC) – Scientific Use File (SUF) (database), https://ec.europa.eu/eurostat/web/microdata/european-union-statistics-on-income-and-living-conditions (accessed on October 2024).

[53] Freak-Poli, R. et al. (2021), “Social isolation, social support and loneliness as predictors of cardiovascular disease incidence and mortality”, BMC Geriatrics, Vol. 21/1, pp. 1-14, https://doi.org/10.1186/S12877-021-02602-2/.

[1] Gallup (2023), Global State of Social Connections, https://www.gallup.com/analytics/509675/state-of-social-connections.aspx.

[4] Gallup (n.d.), Gallup World Poll, https://www.gallup.com/analytics/318875/global-research.aspx.

[62] Giletta, M. et al. (2021), “A Meta-Analysis of Longitudinal Peer Influence Effects in Childhood and Adolescence”, Psychological Bulletin, Vol. 147/7, pp. 719-747, https://doi.org/10.1037/BUL0000329.

[67] Goebel, J. et al. (2019), “The German socio-economic panel (SOEP)”, Jahrbücher für Nationalökonomie und Statistik, Vol. 239/2, pp. 345-360, https://doi.org/10.1515/JBNST-2018-0022.

[18] Goodard, I. (2023), How many close friends do Americans have? , Pew Research Center, https://www.pewresearch.org/short-reads/2023/10/12/what-does-friendship-look-like-in-america/.

[22] Granovetter, M. (1983), “The strength of weak ties: A network theory revisited”, Sociological Theory, Vol. 1, pp. 201-233, https://www.jstor.org/stable/202051.

[10] Hansen, L. et al. (2017), “Associations between the structural and functional aspects of social relations and poor mental health: A cross-sectional register study”, BMC Public Health, Vol. 17/1, pp. 1-10, https://doi.org/10.1186/S12889-017-4871-X/FIGURES/4.

[21] Harris, T. et al. (2025), “Social capital in the United Kingdom: Evidence from six billion friendships”, https://doi.org/10.31235/OSF.IO/KB7DY_V1.

[58] Hawkley, L. (2008), “From social structural factors to perceptions of relationship quality and loneliness: The Chicago health, aging, and social relations study.”, The Journals of Gerontology Series B: Psychological Sciences and Social Sciences, Vol. 63/6, pp. 375-S384, https://doi.org/10.1093/geronb/63.6.s375.

[56] Holahan, C., D. Valentiner and R. Moos (1995), “Parental support, coping strategies, and psychological adjustment: An integrative model with late adolescents”, Journal of Youth and Adolescence, Vol. 24/6, pp. 633-648, https://doi.org/10.1007/BF01536948/.

[8] Holt-Lunstad, J. and T. Smith (2016), “Loneliness and social isolation as risk factors for CVD: Implications for evidence-based patient care and scientific inquiry”, Heart, Vol. 102/13, p. 987, https://doi.org/10.1136/HEARTJNL-2015-309242.

[9] Holt-Lunstad, J. et al. (2015), “Loneliness and social isolation as risk factors for mortality: A meta-analytic review”, Perspectives on Psychological Science, Vol. 10/2, pp. 227-237, https://doi.org/10.1177/1745691614568352.

[24] Ikeda, K. and S. Richey (2009), “The impact of diversity in informal social networks on tolerance in Japan”, British Journal of Political Science, Vol. 39/3, pp. 655-668, https://doi.org/10.1017/S0007123408000574.

[77] INEGI (2025), Módulo de Bienestar Autorreportado (BIARE) 2025, Instituto Nacional de Estadística y Geografía (INEGI).

[64] INEGI (n.d.), Bienestar subjetivo - BIARE Básico, Instituto Nacional de Estadística y Geografía, https://www.inegi.org.mx/investigacion/bienestar/basico/#documentation.

[66] Insee (2022), L’enquête Statistiques sur les ressources et conditions de vie, Institut national de la statistique et des études économiques, https://www.insee.fr/fr/information/4230346.

[13] Lleras-Muney, A. et al. (2020), “Party on: The labor market returns to social networks and socializing”, NBER Working Paper Series, Vol. 27337, https://doi.org/10.3386/W27337.

[74] Luthar, S., S. Barkin and E. Crossman (2013), ““I can, therefore I must”: Fragility in the upper-middle classes”, Development and Psychopathology, Vol. 25/4pt2, pp. 1529-1549, https://doi.org/10.1017/S0954579413000758.

[38] Maher, B. et al. (2013), “Medical school attrition-beyond the statistics: A ten year retrospective study”, BMC Medical Education, Vol. 13/1, p. 13, https://doi.org/10.1186/1472-6920-13-13/.

[76] Mahoney, J. et al. (2024), “Measuring social connectedness in OECD countries: A scoping review”, OECD Papers on Well-being and Inequalities, No. 28, OECD Publishing, Paris, https://doi.org/10.1787/f758bd20-en.

[37] Matthews, T. et al. (2019), “Lonely young adults in modern Britain: Findings from an epidemiological cohort study”, Psychological Medicine, Vol. 49/2, p. 268, https://doi.org/10.1017/S0033291718000788.

[80] Ma, Y., A. Siu and W. Tse (2018), “The role of high parental expectations in adolescents’ academic performance and depression in Hong Kong”, Journal of Family Issues, Vol. 39/9, pp. 2505-2522, https://doi.org/10.1177/0192513X18755194.

[5] Ministerio de Desarrollo Social y Familia (2021), Encuesta de Bienestar Social, http://observatorio.ministeriodesarrollosocial.gob.cl/encuesta-bienestar-social.

[36] Morrish, N., R. Mujica-Mota and A. Medina-Lara (2022), “Understanding the effect of loneliness on unemployment: Propensity score matching”, BMC Public Health, Vol. 22/1, https://doi.org/10.1186/S12889-022-13107-X.

[23] Mutz, D. (2002), “Cross-cutting social networks: Testing democratic theory in practice”, American Political Science Review, Vol. 96/1, pp. 111-126, https://doi.org/10.1017/S0003055402004264.

[43] Newall, N., J. Chipperfield and D. Bailis (2014), “Predicting stability and change in loneliness in later life”, Journal of Social and Personal Relationships, Vol. 31/3, pp. 335-351, https://doi.org/10.1177/0265407513494951.

[48] OECD (2025), OECD Guidelines on Measuring Subjective Well-being (2025 Update), OECD Publishing, Paris, https://doi.org/10.1787/9203632a-en.

[73] OECD (2024), How’s Life? 2024: Well-being and Resilience in Times of Crisis, OECD Publishing, Paris, https://doi.org/10.1787/90ba854a-en.

[54] OECD (2023), How to Make Societies Thrive? Coordinating Approaches to Promote Well-being and Mental Health, OECD Publishing, Paris, https://doi.org/10.1787/fc6b9844-en.

[75] OECD (2023), PISA 2022 Results (Volume II): Learning During – and From – Disruption, PISA, OECD Publishing, Paris, https://doi.org/10.1787/a97db61c-en.

[72] OECD (2021), COVID-19 and Well-being: Life in the Pandemic, OECD Publishing, Paris, https://doi.org/10.1787/1e1ecb53-en.

[7] OECD (n.d.), How’s Life? Well-being Database, http://data-explorer.oecd.org/s/fu.

[47] Paris, B. et al. (2024), “Evaluating Loneliness Measurements across the European Union”, PsyArXiv Preprints, https://doi.org/10.31234/OSF.IO/TK68S.

[19] Patel, A. and S. Plowman (2022), “The increasing importance of a best friend at work”, Gallup Workplace, https://www.gallup.com/workplace/397058/increasing-importance-best-friend-work.aspx.

[25] Quintelier, E., D. Stolle and A. Harell (2012), “Politics in peer groups: Exploring the causal relationship between network diversity and political participation”, Political Research Quarterly, Vol. 65/4, pp. 868-881, https://www.jstor.org/stable/41759320.

[79] Rizwan, M., M. Talha and X. Qi (2020), “Cultural impact of perceived parental expectations on students’ academic stress”, Annals of Social Science and Perspective, Vol. 1/2, pp. 53-65, https://doi.org/10.52700/ASSAP.V1I2.25.

[60] Robles, T. et al. (2014), “Marital quality and health: A meta-analytic review”, Psychological Bulletin, Vol. 140/1, pp. 140-187, https://doi.org/10.1037/A0031859.

[55] Saeed, K. et al. (2023), “How social support predicts academic achievement among secondary students with special needs: The mediating role of self-esteem”, Middle East Current Psychiatry, Vol. 30/1, pp. 1-9, https://doi.org/10.1186/S43045-023-00316-2/.

[41] Schnepf, S., B. d’Hombres and C. Mauri (2024), Lonliness in Europe: Determinants, risks and interventions, Springer Cham, https://link.springer.com/book/10.1007/978-3-031-66582-0.

[44] Sheftel, M., R. Margolis and A. Verdery (2024), “Loneliness Trajectories and Chronic Loneliness Around the World”, The Journals of Gerontology: Series B, Vol. 79/8, https://doi.org/10.1093/GERONB/GBAE098.

[42] Shiovitz-Ezra, S. and L. Ayalon (2010), “Situational versus chronic loneliness as risk factors for all-cause mortality”, International Psychogeriatrics, Vol. 22/3, pp. 455-462, https://doi.org/10.1017/S1041610209991426.

[40] Statistics Canada (2025), Loneliness by gender and other selected sociodemographic characteristics, https://open.canada.ca/data/dataset/277e3275-5b97-4b2b-bf59-59af72541bd7.

[6] Stats NZ (2024), Wellbeing statistics: 2023, https://www.stats.govt.nz/information-releases/wellbeing-statistics-2023/.

[46] Stegen, H. et al. (2024), “Loneliness prevalence of community-dwelling older adults and the impact of the mode of measurement, data collection, and country: A systematic review and meta-analysis”, International Psychogeriatrics, Vol. 36/9, pp. 747-761, https://doi.org/10.1017/S1041610224000425.

[51] Tay, L. et al. (2013), “Social relations, health behaviors, and health outcomes: A survey and synthesis”, Applied Psychology: Health and Well-being, Vol. 5/1, pp. 28-78, https://doi.org/10.1111/APHW.12000.

[59] Thomas, P., H. Liu and D. Umberson (2017), “Family Relationships and Well-Being”, Innovation in Aging, Vol. 1/3, https://doi.org/10.1093/GERONI/IGX025.

[45] Tilburg, T. and E. Leeuw (1991), “Stability of scale quality under various data collection procedures: Mode comparison on the ’De Jong-Gierveld Loneliness Scale’”, International Journal of Public Opinion Research, Vol. 3/1, pp. 69-85, https://doi.org/10.1093/IJPOR/3.1.69.

[16] US Census Bureau (2024), Household Pulse Survey, https://www.census.gov/programs-surveys/household-pulse-survey.html.

[12] van der Horst, M. and H. Coffé (2012), “How friendship network characteristics influence subjective well-being”, Social Indicators Research, Vol. 107/3, pp. 509-529, https://doi.org/10.1007/S11205-011-9861-2/.

[33] Van Der Laan, J. et al. (2023), “A whole population network and its application for the social sciences”, European Sociological Review, Vol. 39/1, pp. 145-160, https://doi.org/10.1093/ESR/JCAC026.

[35] Wang, J. et al. (2018), “Associations between loneliness and perceived social support and outcomes of mental health problems: A systematic review”, BMC Psychiatry, Vol. 18/1, pp. 1-16, https://doi.org/10.1186/S12888-018-1736-5/.

[52] Wills, T. (1991), “Social support and interpersonal relationships”, in Clark, M. (ed.), Prosocial Behavior, Sage Publications, Inc., https://psycnet.apa.org/record/1991-97117-010.

Annex 2.A. Technical annex

Copy link to Annex 2.A. Technical annex

Data sources

Copy link to Data sources

Data in this report are pulled from various high-quality surveys. Annex Table 2.A.1 provides detailed information about the multi-country, internationally comparable data sources – including official statistics such as the European Union Statistics on Income and Living Conditions survey (EU-SILC) – while Annex Table 2.A.2 provides relevant information for official surveys fielded by OECD Member states.

Survey	Years	Country coverage	Sample size	Sampling methodology	Survey mode
European Union Statistics on Income and Living Conditions Survey (EU-SILC) Eurostat (n.d.[2])	Meet with friends or family at least monthly for a drink or a meal: 2009, 2013-2022 Get together with friends / family; Contact friends / family: 2006, 2015, 2022 Satisfaction with personal relationships: 2013, 2018, 2022 Feeling lonely†: 2018, 2022	Member states: AUT, BEL, CZE, DNK, EST, FIN, FRA, DEU, GRC, HUN, ISL, IRL, ITA, LVA, LTU, LUX, NLD, NOR, POL, PRT, SVK, SVN, ESP, SWE, CHE, TUR, GBR Accession candidate countries:* BGR, HRV, ROU Data for DEU in 2022 are not included due to high non-response rates (DEU is therefore also not included in any time series analysis); data for SWE is not included in time series analysis due to a break in methodology between 2018 and 2022 making outcomes not comparable between the years. Additional information on data quality can be found on Eurostat’s website.	Exact sample sizes vary, but range from around 2 900 to 50 000 respondents per country, per year	In line with EU regulation 2019/1700, the EU-SILC data collected are based on representative samples. These are drawn from sampling frames set up at the national level which allow persons or households to be selected at random, with a known probability of selection. The sampling frames aim to identify and cover the target population exhaustively, with the usual accepted coverage error. They are regularly updated. The sampling frames contain all the necessary information for the sample design, such as information needed for stratification purposes and for contacting the persons or households.	A combination of computer-assisted telephone interviews (CATI), computer-assisted personal interviews (CAPI), computer-assisted web interviews (CAWI) and paper-assisted personal interviews (PAPI); rates of each vary by country and year
Gallup World Poll Gallup (2023[1])	Annual since 2005	Member states: AUS, AUT, BEL, CAN, CHL, COL, CRI, CZE, DNK, EST, FIN, FRA, DEU, GRC, HUN, ISL, IRL, ISR, ITA, JPN, KOR, LVA, LTU, LUX, MEX, NLD, NZL, NOR, POL, PRT, SVK, SVN, ESP SWE, CHE, TUR, GBR, USA Accession candidate countries: ARG, BRA, BGR, HRV, IDN, PER, ROU, THA	Around 1 000 respondents per country, per year Gallup World Poll data are pooled to 2- or 3-year averages in OECD analyses, to ensure sufficiently large sample sizes	Nationally representative probability sampling of the non-institutionalised, adult population Sampling design varies by geographic region (depending on the availability and granularity of existing population information – for example from pre-existing census data), and survey mode (in-person household sampling vs. telephone survey sampling)	A combination of telephone and face-to-face interviews. In general telephone surveys are used in North America, Western Europe and East Asia; face-to-face surveys are used in Central and Eastern Europe and much of Latin America and the former Soviet states.
Global State of Social Connections Gallup (2023[1])	2022	Member states: AUS, AUT, BEL, CAN, CHL, COL, CRI, CZE, DNK, EST, FIN, FRA, DEU, GRC, HUN, ISL, IRL, ISR, ITA, JPN, KOR, LVA, LTU, LUX, MEX, NLD, NZL, NOR, POL, PRT, SVK, SVN, ESP SWE, CHE, TUR, GBR, USA Accession candidate countries: ARG, BRA, BGR, HRV, IDN, PER, ROU, THA	Around 1 000 respondents per country	Same as annual Gallup World Poll (see above)	Same as annual Gallup World Poll (see above)
Social Connections Survey Data for Good at Meta (2022[26])**	2022	Member states: FRA, MEX, USA	2 000 respondents per country	Probability-based, nationally representative of the 15+ population living in a household	Telephone (France, United States) and face-to-face (Mexico)
EU Loneliness Survey European Commission, Joint Research Centre (JRC) (2024[3])	2022	Member states: AUT, BEL, CZE, DNK, EST, FIN, FRA, DEU, GRC, HUN, IRL, ITA, LVA, LTU, LUX, NLD, POL, PRT, SVK, SVN, ESP, SWE Accession candidate countries: BGR, HRV, ROU	Around 1 000 respondents per country (aside from Luxembourg, which has 500 respondents)	Online survey of respondents recruited from established consumer panels; quotas were used to ensure representativeness by age, gender, education level, geographic distribution and economic characteristics	Online self-administered survey
Mind Health Survey AXA (2023[65])	Annual since 2022	Member states: BEL, FRA, DEU, IRL, ITA, JPN, MEX, ESP, CHE, TUR, GBR, USA Accession candidate countries:* THA	Around 2 000 respondents per country	Quota method sampling conducted via Ipsos. Quotas are used to ensure representativeness by gender, age, occupation and region	Online self-administered survey

Note: * Country participation varies across years; each figure depicting an OECD average using these data clarifies which countries are included in the average. ** This pilot survey, a collaboration between Meta and Gallup, informed the subsequent Global State of Social Connections survey by Meta and the Gallup World Poll in 2022. † Eurostat elected to not publish data on feeling lonely in 2022 due to non-response rates in nine countries (including non-OECD countries) exceeding 40%. In this report, whenever figures show country-level Eurostat data, a figure note indicates if non-response rates exceed 40% to clearly note this to the reader. Broader analysis on trends and sub-group analysis are conducted on OECD average outcomes (i.e., not at the country level) to ensure sample sizes are as robust as possible.

Survey	Years	Country coverage	Sample size	Sampling methodology	Survey mode
Canadian Social Survey Statistics Canada (2025[40])	Annual, quarterly since 2021	Canada (non-institutionalised respondents aged 15+ living off-reserve)	20 000 respondents (one randomly selected per household)	Probabilistic two-stage stratified sampling	Self-response via electronic questionnaire with computer-assisted telephone interview follow-up in case of non-response
Social Pulse Survey Encuesta Pulso Social DANE (n.d.[63])	Monthly, between July 2020 and March 2023	Colombia (23 departmental capital cities and their metropolitan areas)	Varies across rounds, for an average of ~10-12 000 completed surveys per round	Sub-sample of the labour market household survey, Gran Encuesta Integrada de Hogares (GEIH), which uses a probabilistic multi-stage stratified sampling design	Telephone interviews
Social Welfare Survey Encuesta de Bienestar Social Ministerio de Desarrollo Social y Familia (2021[5])	2021, 2023	Chile	20 800 respondents (2021); 12 400 respondents (2023)	Probabilistic two-stage sampling design, based on a sub-sample of La Encuesta de Caracterización Socioeconómica Nacional (Casen), a bi- or triennial household survey	Telephone interviews
Community Life Survey Department for Culture, Media & Sport (2023[30])	Annual since 2012/13, quarterly estimates available in 2023/24	England	Around 180 000 respondents in most recent round (to enable local area estimates); around 10 000 respondents in previous years	Stratified random sampling	Web time series from 2013-2014
The Statistics on Resources and Living Conditions Survey L'enquête Statistiques sur les ressources et conditions de vie Insee (2022[66])	Annual	France (beginning in 2022, extended to French overseas departments excluding Mayotte)	Around 25 000 to 38 000 respondents per year (higher sample sizes in recent years)	Nationally representative probability sampling of individuals living in private households	Face-to-face (aside from 2021, when all surveys were done via telephone)
The German socio-economic panel (SOEP) Sozio-oekonomische Panel (SOEP) Goebel et al. (2019[67])	Annual	Germany	Around 30 000 respondents in 22 000 households per year	Multistage stratified sampling	Mixed-mode approach: computer-assisted telephone interviews (CATI), computer-assisted personal interviews (CAPI), computer-assisted web interviews (CAWI) and paper-assisted personal interviews (PAPI) and self-administered questionnaire (SAQ). In recent years CATI and CAPI are most common.
Basic Survey on People's Connections 人々のつながりに関する基礎調査 e-Stat, Government of Japan (2023[17])	Annual since 2021	Japan	Around 12 000 respondents aged 16+, randomly selected from the population-wide Basic Resident Ledger	Random sampling	Mixed method: respondents can choose to respond to the survey online, or fill out the survey and submit it by mail
Subjective Well-being – BIARE Basic survey Bienestar subjetivo - BIARE Básico INEGI (n.d.[64])	Quarterly since 2013; Annually beginning 2025	Mexico	Around 2 400 households per round; Beginning in 2025, 4 818 households per round	Probabilistic multi-stage stratified sampling. Prior to 2025, BIARE allowed for estimates of urban aggregates for the 32 largest cities. Beginning in 2025, the survey samples the 12+ population; sample size provides national coverage for both urban and rural areas	Personal interview on mobile device
Social cohesion and well-being Sociale samenhang en welzijn CBS (2023[68])	Annual	the Netherlands	Around 7 500 respondents per year	Weights are applied to ensure the sample is representative based on gender, age, origin, marital status, urbanisation, province, region, household size, income and survey month	Mixed-mode approach: respondents are asked via letter to respond online, computer-assisted web interviewing (CAWI); non-responses are re-contacted by telephone (computer-assisted telephone interviewing (CATI)), and if no telephone number is available, the respondent is contacted in person by an enumerator (computer-assisted personal interviewing (CAPI))
General Social Survey Statistics New Zealand (2024[6])	Biennially (most recently in 2022/2023 – the GSS will not be fielded in 2025/26 but is planned for 2026/2027)	New Zealand	Around 12 000 respondents per round, achieved sample size of 7 800 respondents in 2023	Probabilistic three-stage sampling of 15+ usually resident population living in private dwellings	Face-to-face, computer-assisted personal interviews (CAPI); in 2023 computer-assisted video interviews (CAVI) were introduced as an option
Household Pulse Survey US Census Bureau (2024[16])	High-frequency data collection (beginning every two weeks, moving to monthly) between April 2020 and September 2024	United States	Between 50 000 and 60 000 respondents per round	Multi-stage probabilistic sampling design, based on the Census Bureau’s Master Address File of sampled housing units	Internet questionnaire; respondents are invited to complete the survey online via email or text message

Note: The specific modules and variables included in each survey may vary across years. “Sample size” typically refers to the total number of completed surveys, however reporting on sample sizes varies across national statistical offices and so in some cases may refer to the total number of respondents selected for inclusion in the survey (rather than completed surveys).

Additional statistical analysis

Copy link to Additional statistical analysis

Coding answer options for frequency of social interactions (in-person vs. remote)

Throughout this report, data on in-person vs. remote social interactions come from the European Union Statistics on Income and Living Conditions survey (EU-SILC). Respondents are asked how frequently they (1) get together in person or (2) contact (via remote forms of interaction, such as messaging, email, video chat, etc.) both family members and friends in an average year. The answer options are as follows: daily, every week (not every day), several times a month (not every week), once a month, at least once a year (less than once a month) and never. The full range of answer options is shown in Figure 2.4, however when conducting correlation analysis (Figure 2.2 and Table 2.1), looking at inequalities by population group (Chapter 3) and trends over time, including trends for population groups (Chapter 4) outcomes are shown for respondents who select “daily”. (In Chapters 3 and 4, outcomes for “daily” are contrasted with “never”.)

Reporting outcomes in this way – showing the full set of answer options, but comparing “daily” and “never” when discussing outcomes for specific population groups – mirrors the approach Eurostat has taken in its statistical publications describing the same set of indicators (Eurostat, 2010[69]; 2017[70]; 2020[71]).

For readers who are concerned that “daily” is too narrow a frequency grouping, in this Annex we present results for “at least weekly” – combining answer options for “daily” and “every week (not every day)”. As is shown in Annex Figure 2.A.1 and Annex Table 2.A.3 below, results are consistent regardless of the frequency grouping shown (compare to Figure 2.2 and Table 2.1, respectively).

Annex Figure 2.A.1. Strong performance in one area of social connections does not guarantee positive outcomes in others
Copy link to Annex Figure 2.A.1. Strong performance in one area of social connections does not guarantee positive outcomes in others

Scatterplot of dissatisfaction with personal relationships and at least weekly in-person interactions with friends, OECD EU-EFTA 23, 2022

	Get together with family at least weekly	Get together with friends at least weekly	Contact family (remotely) at least weekly	Contact friends (remotely) at least weekly	Felt lonely most or all of the time over the past four weeks
Get together with friends at least weekly	0.26
Contact family (remotely) at least weekly	0.35	0.18
Contact friends (remotely) at least weekly	0.15	0.45	0.33
Felt lonely most or all of the time over the past four weeks	-0.05	-0.07	-0.07	-0.08
Dissatisfied with personal relationships	-0.07	-0.07	-0.10	-0.09	0.24

Additional correlational analysis

Annex Table 2.A.4 and Annex Table 2.A.5 serve as a robustness check to the correlation analysis presented in Figure 2.2 and Table 2.1, using data from the Global State of Social Connections survey and EU Loneliness Survey, respectively: in all three surveys considered, including the EU-SILC survey referenced in the main text, quantitative (structural) measures of the frequency of social interactions are only weakly correlated with self-assessments of social connections (loneliness, social support, relationship satisfaction).

	Interact with friends / family who live close by daily over the past week	Interact with friends / family who live far away daily over the past week	Feel lonely
Interact with friends / family who live far away daily over the past week	0.17
Feel lonely	-0.10	-0.02
Feel supported	0.13	0.05	-0.16

Note: Further definitional details for “interaction” are not specified (i.e., whether in person or remote). Table displays weighted listwise Pearson correlation coefficients between quantitative and qualitative social connections outcomes from all 38 OECD countries in 2022.

Source: Gallup (2023[1]), Global State of Social Connections, https://www.gallup.com/analytics/509675/state-of-social-connections.aspx.

	See family at least daily	See friends at least daily	Contact family at least daily	Contact friends at least daily	Felt lonely most or all of the time over the past four weeks
See friends at least daily	0.38
Contact family at least daily	0.32	0.21
Contact friends at least daily	0.21	0.38	0.41
Felt lonely most or all of the time over the past four weeks	-0.01	0.02	-0.03	0.01
Dissatisfied with relationship	-0.03	-0.01	-0.04	-0.04	0.17

Note: “See” refers to meeting up face-to-face, either pre-arranged meetings or by chance encounters. “Contact” refers to remote interactions via phone, Internet or social media. Table displays weighted listwise Pearson correlation coefficients between quantitative and qualitative social connections outcomes from 22 European OECD countries in 2022.

Source: European Commission, Joint Research Centre (JRC) (2024[3]), EU Loneliness Survey. European Commission, Joint Research Centre (JRC) [Dataset] PID: http://data.europa.eu/89h/82e60986-9987-4610-ab4a-84f0f5a9193b.

Notes

Copy link to Notes

← 1. Shorter time series are available for qualitative outcomes in social connections – such as feeling lonely and relationship dissatisfaction – making it more difficult to establish whether recent declines are pandemic-related, or began beforehand. However, supplemental evidence from national surveys shows that, for countries with more frequently collected data that extend closer to present day, loneliness and other qualitative outcomes declined over the course of the COVID-19 pandemic and have not yet rebounded to pre-pandemic levels (see Chapter 4 for an extended discussion).

← 2. Refer to the Annex for a discussion of how answer options are grouped for frequency of interacting in person vs. remotely, throughout this report. Analysis using an alternate grouping (“at least weekly” instead of “daily”) is presented, to show that results are consistent regardless of grouping.

← 3. Figure 2.2 and Table 2.1 use data from the European Union Statistics on Income and Living Conditions survey (EU-SILC). Additional analysis in the Annex shows that this pattern – weak correlation between quantitative outcomes in social connections (e.g. frequency of socialising) and functional or qualitative outcomes (e.g. feeling lonely, lack of perceived support or relationship dissatisfaction) holds across multiple data sources. See Annex Table 2.A.4 and Annex Table 2.A.5.

← 4. Time use surveys can also provide useful data on structural aspects of social connections, such as the amount of time spent interacting with others. All data used in this chapter, however, come from household surveys.

← 5. While school and the workplace are not the primary focal settings for outcomes presented in this chapter, recent surveys have collected data on the quantity and quality of social connections in these environments. For example, evidence from the 2022 EU Loneliness Survey found that in 22 European OECD countries, 3% of people reported feeling isolated at work vs. 42% who never feel isolated at work, whereas 8% of students reported feeling isolated at school and 20% reported never feeling so. In general, respondents are more likely to feel supported by their peers (21% of workers always feel supported by colleagues, and 17% of students always feel supported by classmates) than their superiors (19% of workers are always supported by managers, and 15% of students report always having support from teachers) (European Commission and Joint Research Centre (JRC), 2024[3]). Additionally, data from the 2022 OECD Programme for International Student Assessment (PISA) reveal that 16% of 15-year-olds reported feeling lonely at school, 17% felt like an outsider and 21% felt awkward or out of place. On average across OECD countries, sense of school belonging deteriorated slightly between 2018 and 2022, however individual country trajectories varied (OECD, 2023[75]).

← 6. Close relationships with friends and family serve an important protective role, and provide resilience against poor health and socio-economic outcomes. However, poor quality relationships can actively cause harm. For example, evidence from a long-term study in Sweden tracking individuals from birth to age 65 shows that poor quality family relationships in childhood are associated with premature mortality, even when other common risk factors (including childhood poverty, parental mental illness, etc.) are controlled for (Alm, Låftman and Bohman, 2019[61]). Other evidence shows how parental expectations and pressure imposed on children to succeed academically can induce anxiety, depression, stress and burnout in young people (Ciciolla et al., 2017[78]; Rizwan, Talha and Qi, 2020[79]; Ma, Siu and Tse, 2018[80]; Luthar, Barkin and Crossman, 2013[74]).

← 7. There are different approaches to measuring loneliness, including a single-item question asking respondents how lonely they feel (either in general, or in reference to a specific time period), as well as multi-item scales: the two most common are the University of California Los Angeles (UCLA) 3-item scale and the De Jong-Gierveld (DJG) 6-item scale. The single-item question referenced in the main text of this chapter is the most commonly-used loneliness question on official surveys across OECD countries (Mahoney et al., 2024[76]), and asks respondents how lonely they felt over the past four weeks, with responses ranging from: all of the time, most of the time, some of the time, a little of the time and none of the time. The most common approach when scoring these answers is to define “lonely” respondents as those who answered either “all of the time” or “most of the time” (Schnepf, d’Hombres and Mauri, 2024[41]; Mahoney et al., 2024[76]).

← 8. In 2025, the Mexican statistical office began collecting national data on feelings of loneliness over the past four weeks. In 2025, 8.3% of Mexicans reported feeling lonely most or all of the time (INEGI, 2025[77]). This data point is not included in Figure 2.11 because the year of data collection (2025) is too far removed from the other data included (2022-23).

← 9. Intense loneliness is defined as a score exceeding 6 on a scale from 1 (not very intense) to 10 (very intense).

← 10. These data were collected in 2022, when the COVID-19 pandemic was still greatly impacting many aspects of society in OECD countries, and the cost-of-living crisis led to financial insecurity (OECD, 2021[72]; 2024[73]). This may have influenced responses regarding the availability of certain types of social support: for example, the ability to obtain financial support from friends or family (pandemic job disruptions, inflation and cost-of-living all affecting household finances), or the availability of aid should one be sick and confined to bed (a likely recent actual occurrence for many respondents, given the high rates of COVID throughout 2020 and 2021).

Featured topics

Agriculture and fisheries

Climate change

Development

Digital

Economy

Education and skills

Employment

Environment

Finance and investment

Governance

Health

Industry, business and entrepreneurship

Regional, rural and urban development

Science, technology and innovation

Society

Taxation

Trade

Energy

Nuclear energy

Transport

Featured topics

Agriculture and fisheries

Climate change

Development

Digital

Economy

Education and skills

Employment

Environment

Finance and investment

Governance

Health

Industry, business and entrepreneurship

Regional, rural and urban development

Science, technology and innovation

Society

Taxation

Trade

Energy

Nuclear energy

Transport

Countries A - C

Countries D - I

Countries J - M

Countries N - R

Countries S - T

Countries U - Z

Regional and global engagement

Countries

Countries A - C

Countries D - I

Countries J - M

Countries N - R

Countries S - T

Countries U - Z

Regional and global engagement

Publications

Publications

Featured publications

Data

Data

Featured data

News & events

News & events

Featured events

About OECD

About

Engage with us

Work with us

Featured topics

Agriculture and fisheries

Climate change

Development

Digital

Economy

Education and skills

Employment

Environment

Finance and investment