In the 8th century, in a small town just outside Baghdad, a seemingly lonely and perhaps socially isolated individual wrote the following melancholic message on a wall:

“May God water the days of togetherness with His rain and return every stranger to his home. There is no good in this world without togetherness and no joy in life without a loved one”.

Loneliness and social isolation know no borders; these are timeless human experiences. However, there is a perception that both loneliness, a subjective experience, and social isolation, an objectively quantifiable state, are on the rise. The public health implications of both these forms of social disconnection are well known. What is less well understood is how personal digital technologies (devices/services) might contribute to the onset, worsening, or alleviation of these states.

Differentiating loneliness and social isolation

Loneliness is an unpleasant feeling that accompanies the perception of unmet social needs (Hawkley & Cacioppo, 2010). It is not the same as “being alone”. Many individuals experience, and perhaps even enjoy, extended periods of solitude without feeling lonely (Hawkley & Cacioppo, 2010). Furthermore, a person may lead a seemingly active social life, yet still feel lonely. This reality is eloquently articulated in a quote attributed to the Swiss Psychologist, Carl Jung:

“Loneliness does not come from having no people about one, but from being unable to communicate the things that seem important to oneself”.

At its core, loneliness arises from a perceived disparity between desired and actual levels of social connection. Loneliness is a subjective reaction to perceived deficiencies in one's social world. Social isolation, on the other hand, is the objective reality of having relatively few social roles, relationships, and social interactions (World Health Organization, 2025)

Despite these differences, loneliness and social isolation correlate (Ge et al., 2017; Taylor et al., 2023). Individuals with fewer relationships, social roles and limited human connections or social support (the socially isolated) tend to experience loneliness. Likewise, those who feel lonely may be inclined to withdraw from others, leading to relative social isolation(National Center for Chronic Disease Prevention and Health Promotion (NCCDPHP), 2024). Loneliness and social isolation can also be enduring, persisting over long periods. In such cases, loneliness is viewed as a trait (van Roekel et al., 2018), and might be described as chronic or long-term loneliness. At precisely which time point does transient or short-term loneliness become chronic loneliness? This remains a matter of debate, with suggested durations spanning anywhere from 1 to 6 years (Wolska & Creaven, 2023). A similar distinction is made between transient (short-lived) and long-term social isolation.

Health implications of loneliness and social isolation

The WHO’s constitution proposes that health is “a state of complete physical, mental and social well-being and not merely the absence of disease or infirmity”. Social wellbeing is at the very heart of what it means to live a healthy life. Although social wellbeing has been relatively overshadowed, the health implications of poor social wellbeing (loneliness and social isolation) have become increasingly apparent in recent decades.

According to the US Centres for Disease Control and Prevention, both loneliness and social isolation (when chronic) are independently associated with an increased risk of experiencing mental and physical health problems, including heart disease, stroke, depression, and anxiety (National Center for Chronic Disease Prevention and Health Promotion (NCCDPHP), 2024). Research aimed at quantifying the mortality risk of chronic loneliness and social isolation suggests that they are comparable to obesity, physical Inactivity, alcohol consumption, and cigarette smoking (National Academies of Sciences et al., 2020). The World Health Organisation estimates that almost one in six people is affected by loneliness globally, and that loneliness and social isolation contribute to around 871,000 deaths annually (World Health Organization, 2025)

During the COVID-19 pandemic, levels of loneliness and social isolation rose significantly due to the social distancing measures (Allen et al., 2022). However, the growing public health concern around these issues certainly pre-dates the pandemic. In 2018, for instance, the UK government appointed Tracey Crouch as the minister for loneliness. This first-of-its-kind appointment followed an influential government report on loneliness and social isolation — the Jo Cox Commission on Loneliness. Among other statistics, the report suggests that around 200,000 older people in the UK hadn’t conversed with a friend or relative in more than a month. The report described the impact of loneliness and social isolation as being twice as harmful as obesity and comparable to smoking 15 cigarettes per day.(Jopling, 2017) The UK report also talked about elevated rates of loneliness among young people. University students who feel like they don’t fit in—spending days in relative isolation with nothing but college deadlines and digital devices for companionship and support(Jopling, 2017).

Given the extensive public health implications of loneliness and social isolation, numerous national and international initiatives have emerged to address what is often referred to as "the loneliness epidemic." For instance, in the UK, there is the Campaign to End Loneliness; in Australia, Ending Loneliness Together; and in the United States, the Foundation for Social Connection, along with the World Health Organisation’s Social Isolation and Loneliness initiatives, and the Global Initiative on Loneliness and Connection (Taylor et al., 2023).

Measuring loneliness and social isolation

Most attempts to measure loneliness have focused on it as a persisting trait — an enduring pattern of experience — rather than a short-lived, transient state(Oughli & Lee, 2024). These measures of trait loneliness can be grouped into two categories: those that treat loneliness as unidimensional (a unitary, global experience) and those that treat it as multidimensional. Within the unidimensional view, loneliness varies in intensity or frequency from low to high. Multidimensional models, however, are more nuanced and, as a result, perhaps more contested. For instance, one might obtain a low score for “alienation”, as a proposed component of loneliness, while reporting relatively high levels of estrangement. However, a criticism of the multidimensional models is that the theoretical conceptualisation of the proposed components or types of loneliness lacks clarity and consensus.

Perhaps due to their brevity and ease of interpretation, unidimensional measures of trait loneliness have been the most frequently employed. The University of California, Los Angeles (UCLA) Loneliness scale is currently the most widely used among such metrics. Developed by researchers at UCLA, this scale was designed to be psychometrically adequate (valid and reliable) and easily administered. The scale has undergone several revisions, and there are now long (20-item) and short (3-item) forms. The UCLA-LS is a self-report measure where individuals are asked to indicate how frequently they felt, for example, “a lack of companionship” or “isolated from others”. Depending on the version of the scale - long or short form - items are scored from 1 to 4, where one equates to “never” having the experience and four reflects “often” feeling this way (Russell et al., 1978)

Standard methods of validation for both multi- and unidimensional measures of loneliness involve comparing populations known to experience higher levels of loneliness, so-called “at-risk” groups, with healthy controls. Such comparisons have been made between healthy college students and, for example, (a) patients experiencing depression, (b) people attending a remedial social skills workshop, and (c) divorcees. In each instance, the at-risk groups report significantly higher levels of loneliness. In addition to comparisons with known groups, measures of loneliness are also frequently validated against self-labelled and peer-reported loneliness.

In measuring social isolation, several potential objective indicators include marital or relationship status, living alone, and living with others. Additionally, self-report measures are used to quantify the size and closeness of a person’s social network by assessing the level of support they receive from family and friends (Veazie et al., 2019). Examples of such measures include the Social Network Scale and the Social Support Scale. These scales typically seek to quantify the levels of social support or isolation. The Social Network Scale, for example, has three items that ask respondents to quantify social interaction, such as, “How many relatives/friends do you see or hear from at least once a month?” Similarly, the six-item Social Support Scale asks respondents to report the number of people they can count on in response to questions such as: “Who can you count on when you need help?” and “Who can you count on to console you when you are very upset?”

Epidemiological studies of loneliness and social isolation

In 2024, the American Psychiatric Association suggested that 1 in 3 Americans are lonely; that is, 30% of US adults said they have experienced feelings of loneliness at least once a week over the past year(American Psychiatric Association, 2024). Similarly, in the same year, Gallup’s World Poll suggested that “Over 1 in 5 (23%) people worldwide feel lonely a lot”.(Dugan, 2024) Work by the WHO’s Commission on Social Connection puts the global rate at around 16%. Whatever way we slice it, there are a lot of lonely people out there.

The percentage of respondents reporting loneliness in the World Poll study varied significantly by territory, ranging from 45% in Comoros to 6% in Vietnam.(Dugan, 2024) The authors acknowledge that at least some of the international variation may stem from individuals in certain countries reporting that they spend parts of their day physically alone, rather than emotionally alone. Loneliness is arguably conceived and experienced differently across diverse cultural and linguistic groups, complicating the determination of global prevalence, especially when using single-item measures.

Establishing a global prevalence for problem loneliness is premature. This is primarily due to data scarcity, particularly in low-income countries. Additionally, across countries, Gallup found that reports of loneliness are consistently higher in web surveys than in traditional (in-person) modes of interviewing. (Dugan, 2024). Add to this the diverse ways of measuring and conceptualising problematic loneliness (methodological heterogeneity), and we begin to appreciate the challenge of establishing even a national, let alone International, prevalence trends for problematic loneliness.

Globally, adolescents (20.9%) and young adults (17.4%) appear to experience loneliness most, whereas social isolation is more common in older age groups, 25 to 34% (World Health Organization, 2025). For example, data from the Programme for International Student Assessment (PISA) suggests that across most of the 37 participating countries, there was an increase in the rates of loneliness at school (15 and 16-year-olds) between 2012 and 2018 (Twenge et al., 2021).

It is unclear whether societal levels of loneliness are increasing globally over time. Currently, there is insufficient evidence to confirm a rise in loneliness, although in some countries (e.g., the USA) and among certain age groups (18–29 years), this appears to be the case. Nevertheless, even if the rates remain relatively stable, loneliness continues to be a significant public health issue that has been underappreciated for too long.

There are several hard indicators that social isolation has risen: perhaps the most common is an increase in the proportion of the population living alone. In many nations, more people are living alone than at any point in recorded history (see Fig. 1).

Figure 1

Source (World Health Organization, 2025)

It is evident that to better understand the direction of travel, loneliness and social isolation should be incorporated into general health surveillance. Such initiatives need to have a broad geographical coverage, including low and middle-income countries and those without reliable access to the online world. This also calls for the consistent use of standardised and well-validated measurement tools (Dugan, 2024)

Digital Media Use and Loneliness

Although there is insufficient evidence to draw firm conclusions about a global rise in loneliness, there is a widespread perception that it is the case. Arguments about the causes of the posited increase in loneliness and social isolation tend to centre on the “lost community hypothesis”. Rooted in the work of the sociologist Ferdinand Tönnies, the lost community hypothesis proposes that the forces of modernity, such as urbanisation, industrialisation, and the rise of individualism, erode social connections. Digital technology is also viewed as one of the forces contributing to “community loss”. From the personal stereo and personal computer to the iPod, iPhone, and now personalised AI, tech typically nudges users towards hyper-individualism. Unsurprisingly, the relationship between digital media use and loneliness has received substantial research attention. For example, several studies have examined time series data, such as the Programme for International Student Assessment (PISA) and US national youth wellbeing surveys. These data frequently, but not invariably, report increases in loneliness and decreases in well-being that correspond with the popularisation of the smartphone (Heffer et al., 2019; Twenge, 2025; Twenge et al., 2021).

Much of this research depends on the rather crude metric known as screentime. Originating in early cinema, the term broadly refers to the amount of time spent viewing screen-based digital media. Current attempts to measure screentime may consider various types of screen use (television, social media, video games, general computer use) and may also try to differentiate dimensions such as active (chatting/commenting) versus passive (scrolling/viewing), and leisure versus occupational screentime. Studies examining the link between screentime and loneliness report mixed results (MacDonald et al., 2022; Tang et al., 2021). These inconclusive findings also extend more generally to the relationship between screentime and mental health (Santos et al., 2024; Tang et al., 2022). Much of the research in this field is correlational, so drawing causal conclusions remains premature. More detailed investigations of the motivations for screen use and the type of content consumed/produced might shed further light on the nature of the relationship between digital media use and loneliness. Ultimately, longitudinal and experimental studies are required to illuminate the possible mechanisms underlying any causal relationships.

Problematic Digital Media Use and Loneliness

We shape our tools, and thereafter, our tools shape us.

[Frequently attributed to Marshall McLuhan]

There is no doubt that individuals can develop problematic relationships with digital media. However, the debate revolves around the best way to conceptualise these issues: behavioural addiction, compulsion, masked depression or maladaptive coping strategy? Clinical concern and published research interest in problematic digital media use and technology-related disorders emerged alongside the popularisation of the World Wide Web in the mid-1990s. In the earliest scholarly work on this topic, Griffiths draws insights from existing research on pathological gambling and describes technology-related disorders as non-chemical (behavioural) addictions involving human-machine interaction. Echoing Griffiths, Young proposed the inclusion of Internet Addiction Disorder [IAD] within revisions to the 4th edition of the American Psychiatric Association’s Diagnostic and Statistical Manual [DSM-IV-TR] (American Psychiatric Association, 2022). Young’s conceptualisation of IAD was broad, encompassing numerous subtypes that reflected different facets of problematic internet use, such as cyberrelationship [social media] addiction, cybersexual [pornography] addiction, gaming addiction, and more. Ultimately, IAD was not included in the DSM-IV-TR. However, gaming addiction, renamed internet gaming disorder [IGD], was incorporated into the revised manual. IGD was recognised as a condition warranting further research, a status it still holds in DSM-5-TR. The proposed DSM-5 criteria for IGD include preoccupation/obsession, withdrawal, tolerance, loss of control, anhedonia, continued overuse, deception, mood repair, and social/occupational impairment. It is stipulated that at least 5 of the 9 symptoms must be present for at least 12 months to meet the criteria for the proposed diagnosis. Other research teams have adapted the same criteria for “social media disorder.”(van den Eijnden et al., 2016)

Going beyond the American Psychiatric Association (APA), the World Health Organization [WHO] has officially recognised gaming disorder as a diagnostic entity within its classification system 25. In 2018, the WHO included gaming disorder in the behavioural addictions section of the International Classification of Diseases, 11th Revision [ICD-11]. Other internet-related issues, such as problematic social media use, can also be diagnosed under the section’s residual categories: “disorders due to addictive behaviours, unspecified,” and “other specified disorders due to addictive behaviours”(Brand et al., 2020; Lindenberg et al., 2022). Extending beyond Europe and North America, many Chinese clinicians utilise the Chinese Classification of Mental Disorders, the CCMD-3, which is the most widely used psychiatric diagnostic system in China (Tejeiro et al., 2016; Zou et al., 2008). The CCMD-3 permits the diagnosis of gaming disorder within the section on habit and impulse disorders (code 61), alongside pathological gambling (Tejeiro et al., 2016).

Despite these relatively recent developments, attempting to integrate problematic technology use within medical/psychiatric frameworks, the diagnostic utility of concepts such as gaming disorder and internet addiction disorder remains widely contested (Musetti et al., 2016). Similarly, as new digital technologies emerge and social norms evolve, what constitutes problematic will require conceptual re-evaluation (Ellis, 2019).

Measuring Problematic Digital Media Use

Drawing on models of behavioural addiction, rooted in earlier work on pathological gambling, most measures of problematic digital media use attempt to assess “addiction” symptoms such as preoccupation, tolerance, withdrawal, persistence, mood modification, deception, displacement, and conflict. Typically, such symptoms (five or more under DSM criteria) have persisted for at least 12 months.

Table adapted from A critical review of “Internet addiction” criteria with suggestions for the future (Van Rooij & Prause, 2014)

Measuring these problematic behaviours typically involves self-report inventories that assess the presence, severity or frequency of such symptoms. The inventories most often used to screen individuals for problematic social media use typically factor in most, if not all, of the symptoms mentioned above. One of the most widely used screening instruments for PSMU is the 9-item Social Media Disorder Scale (SMD9). The SMD9 short form uses a (Yes/No) response scale. Drawing on the APA’s proposed criteria for internet gaming disorder, the presence of 5 out of 9 symptoms is taken as the screening cut-off. Respondents are asked about their behaviour over the past 12 months, for example, “I often used social media to escape from negative feelings” (mood regulation) and “I tried to spend less time on social media but failed” (relapse/control).

It is crucial to distinguish between frequent use and problematic use; simply spending a lot of time on social media (screen time) does not necessarily indicate problematic usage. Many people use social media very often, for long periods, without any negative effects. Such individuals do not show any signs of behavioural addiction. They are not regularly using social media to escape from an unpleasant mood. They do not become unduly angry or distressed when they cannot access their preferred platform. Most importantly, their use does not negatively affect their social (relationships) or occupational responsibilities. These individuals might be described as intense or heavy users, but their usage is not inherently problematic.

Have you jeopardized or lost an important relationship, job or an educational or career opportunity because of your gaming activity?

An Item from the Internet Gaming Disorder Scale 34

A nearly identical method is used to assess problematic gaming or gaming disorder. For example, the internet gaming disorder scale includes nine items that reflect symptoms of behavioural addiction related to gaming. It also asks about gaming activities over the past 12 months and is based on the APA’s proposed criteria for internet gaming disorder. The IGDS includes questions such as “Have you continued your gaming activity despite knowing it was causing problems between you and other people?” and “Have you jeopardised or lost an important relationship, job, or educational or career opportunity because of your gaming activity?” The nine proposed criteria for internet gaming disorder are listed below.

Proposed criteria for Internet Gaming Disorder as listed in the American Psychiatric Association’s diagnostic manual, DSM-5-TR

Beyond the presence of at least 5 of the nine symptoms, the APA also propose that these symptoms must cause "significant impairment or distress" in several aspects of a person's life.

Epidemiology: Prevalence of Problematic Media Use

Like research on loneliness, studies of problematic media use face similar issues of conceptual and measurement heterogeneity. Even the names proposed for problem use vary widely, from smartphone addiction to social media disorder. Despite conceptual diversity, attempts have been made to quantify problem prevalence (variously conceived) and to explore sociodemographic risk factors. One meta-analysis combining 62 studies, including 34,798 respondents, reported a prevalence of 5% for problematic social media use (PSMU)(Cheng et al., 2021). This figure was based on the most stringent screening criteria, counting only those classified as experiencing “very severe symptoms”. Using more relaxed criteria, the prevalence rose to 13%. Notably, in this analysis, the highest rates of PSMU were recorded among the youngest age group (adolescents).

This meta-analysis also spanned 32 countries. Nations were grouped based on the degree to which each society emphasised individualistic versus collectivist values (Hofstede, 2001). Participants in collectivist nations (e.g., Japan, KSA, Taiwan) reported significantly higher rates of PSMU than their relatively individualistic counterparts (e.g., USA, UK, Australia). This is an area for future research. However, it might be that the obligations associated with interdependence, such as compliance with social norms (fitting in) and maintaining close kinship connections, drive greater social media use in collectivist societies, perhaps leading to higher rates of PSMU. With gender in focus, another meta-analysis explored 51 independent studies including both problematic social media use and gaming disorder. The authors suggest that females are more likely to report PSMU, whereas males are more likely to meet the proposed screening criteria for gaming disorder (Su et al., 2020).

Meta-analytic studies specifically examining gaming disorder typically suggest a rate of around 3% (Chiang et al., 2022; Kim et al., 2022). One study spanning 17 countries, including close to a quarter of a million participants, reported a prevalence of 3.05%, with the issue more common in males at a ratio of about 3:1 (Stevens et al., 2021). However, the gaming industry has been actively working to encourage more females into gaming. Market research suggests that female gamers are increasing, particularly mobile phone gamers (Newzoo, 2020); the male-female ratio may shift in the coming years. Similarly, gaming disorder is currently associated with younger age groups (children and adolescents). However, this may also be an aspect of the phenomenon that evolves with sociocultural changes.

Problematic Digital Media Use & Relationship with Loneliness

As mentioned earlier, the link between digital media use (screen time) and loneliness remains unclear (mixed findings). However, there is much greater clarity about the connection between problematic digital media use and loneliness. Furthermore, longitudinal studies indicate a two-way relationship, where problematic technology use can precede loneliness, and loneliness can also precede problematic use.

One meta-analytic study examining longitudinal investigations of problematic digital media use and loneliness reviewed 26 studies —19 of which used the UCLA-LS (Zhang et al., 2023). The authors concluded that adults and adolescents experiencing loneliness are at a greater risk of developing problematic internet use, and similarly, that those with problematic internet use are at a higher risk of later experiencing loneliness. These findings are explained through Reinforcement theory: tech use alleviates loneliness (negative reinforcement), causing lonely individuals to increase their internet use to preserve social benefits (entertainment, online connections), eventually leading to problematic internet use. An alternative, and complementary, explanation is the Internet displacement hypothesis: prolonged PIU results in degraded in-person social connections, leading to loneliness. It is plausible that these proposed initial mechanisms converge, creating a vicious cycle (mutual exacerbation), in which PIU displaces social connections, causing loneliness, which in turn fuels further PIU as an attempt to escape or avoid this unpleasant state.

Another systematic review focused on loneliness, social anxiety and social media included 52 previously published studies (O’Day & Heimberg, 2021). The review concluded that Individuals with high levels of social anxiety (excessive/problematic shyness) and loneliness are more prone to PSMU. The review's authors suggest that this link is due to the socially anxious seeking social support on social media, perhaps to make up for the lack of in-person support. Furthermore, loneliness appears to be a risk factor for PSMU. Several studies have examined loneliness and PSMU over time, finding that loneliness at time one predicts increases in social media use at a later time point. Research in this area has also highlighted differences between active (interacting, commenting, posting) and passive social media use, such as aimless scrolling through the timeline. Unsurprisingly, passive use is most reliably preceded by feelings of loneliness (Verduyn et al., 2015)

One might expect different findings in collectivist societies where families are larger and more closely connected. However, a study exploring this issue among adults from Saudi Arabia and Kuwait reported that problematic internet and social media use was significantly linked to increased feelings of loneliness in these societies (Alheneidi et al., 2021). There was also a dose–response relationship; that is, greater loneliness predicted higher problematic internet use. A study conducted in Lebanon reported a similar link between PSMU and loneliness (Youssef et al., 2020).

Correlation, of course, is not causation. However, given the impact of loneliness on health and well-being, such correlational evidence signals concern and the need for further research. This study utilises data from Sync’s global digital well-being survey to investigate the relationship between loneliness and problematic technology use, specifically problematic social media use and internet gaming disorder, as outlined in DSM-5. This study explores these relationships across 35 countries spanning seven world regions aiming to further examine these posited links across diverse cultures and populations.

Sample description across 35 countries and seven world regions

The data reported here are from Sync’s global digital wellbeing survey. These data comprise 35,000 adult respondents, with 1,000 respondents per territory. Based on pre-existing panels, the sample broadly represents the Internet-using adults in each participating nation. Table 1 presents the raw count and percentage of participants identified as lonely based on the UCLA three-item loneliness scale. The study also explored gaming disorder symptoms and problematic social media using reliable, widely used and well validated scales. All these measures are detailed below.

The UCLA Loneliness Scale (UCLA-LS-3)

The University of California, Los Angeles (UCLA) Loneliness scale is currently the most widely used measure of Loneliness. This scale was designed to be psychometrically adequate (valid and reliable) and easily administered (Russell et al., 1978). In the present study we used the short (3-item) form of the scale. The UCLA-LS-3 is a self-report measure which includes three items: (1) I lack companionship, (2) I feel left out, and (3) I feel isolated from others. Respondents answered these items in terms of frequency: hardly ever, some of the time, and often, scored 1, 2, and 3, respectively. In the present study, we use the recommended cut-off, that is, loneliness as scores of 6 or higher. The scale demonstrated good internal reliability (α = 0.79).

The Gaming Disorder Scale (Short Form)

The short form of the Gaming disorder scale (IGDS9-SF) is a nine-item measure where each statement reflects one of the DSM-5’s proposed criteria for Internet Gaming Disorder (IGD). Respondents are asked to consider their gaming experience over the past twelve months and then respond to items (symptom descriptions) such as "played in order to temporarily escape or relieve a negative mood” and “continued your gaming activity despite knowing it was causing problems between you and other people?”. The scale invites Yes/No responses, and endorsing 5 of 9 symptoms takes the respondent above the proposed screening cut-off, representing possible gaming disorder. The IGDS9-SF has been widely used and well validated(Feng et al., 2017; Pontes & Griffiths, 2015). Its reliability in the current study was good (α = 0.803).

The Social Media Disorder Scale (Short Form)

The short form (nine items) of the Social Media Disorder Scale (SMD9-SF) is derived from the original 27-item version (van den Eijnden et al., 2016). It is extrapolated from the DSM-5’s proposed criteria for Internet Gaming Disorder(American Psychiatric Association, 2013). Respondents are asked about their social media use over the past 12 months, example items include “…often used social media to escape from negative feelings?” and “tried to spend less time on social media but failed”. The SMD9-SF uses a Yes/No response scale. Endorsing 5 out of 9 symptoms takes the respondent above the proposed screening cut-off and is deemed to represent problematic social media use. The SMD9-SF has good convergent and criterion validity along with sufficient sensitivity, specificity, and test-retest reliability(van den Eijnden et al., 2016). In the current study, internal reliability was also good, α = 0.895

Data collection procedure

PSB Insights, a global analytics consultancy with extensive experience in multinational polling services, managed the data collection for the 30-nation digital wellbeing survey (DWS). Materials were translated and back-translated from English into the majority language of each participating nation. The survey was undertaken online. Based on existing participant banks (panels), the survey obtained nationally representative samples of the adult internet-using population in each participating territory. Participants were pre-registered survey panelists in their respective countries. Potential participants received invitations via email. The survey response rate was 19.35%. Automated data quality checks ensured that those who failed to complete the survey were excluded from the analysis, as were those who completed the materials with an overly stereotyped response pattern (e.g., answering yes to everything). Similarly, automated data quality checks removed those who completed the survey too quickly (speeding). The mean exclusion rate was 17%; however, oversampling ensured that each nation had 1000 valid participants. The final sample (N = 35,000) comprised a thousand respondents from each country. All data were collected between July 12th and July 26th, 2023. The study was reviewed and approved by the research ethics committee of King Abdulaziz Centre for World Culture (IRS 202371).

Overall rates of loneliness

Across all participating territories, there were individuals for whom loneliness was a significant issue. The highest rates for individuals scoring above the UCLA loneliness scale cut-off (scores of six or more). were observed in Pakistan and Bangladesh, where more than half of respondents reported loneliness (58%). The lowest rates were reported for China, however, even here around 23% of respondents scored above the cut-off. Across the entire sample, 39.45% were classified as lonely according to the recommended cut-off. Even when the cut-off was raised, and set to the maximum score of nine, 5.03% of respondents were identified as lonely using this stringent threshold.

Figure 2

Item-level analysis

The loneliness scale includes three items: (1) I lack companionship, (2) I feel left out, and (3) I feel isolated from others. Respondents reported the frequency of experiencing such feelings, hardly ever, some of the time, and often, scored 1, 2, and 3, respectively. Lacking companionship (Item 1) was the most frequently endorsed (see Table 3)

The heatmap (Fig. 3), "I lack companionship," is the most strongly endorsed in 31 of the 35 countries.

Figure 3

For additional analyses of loneliness, please see Appendix 1

Indicators of social isolation

Although social Isolation was not assessed explicitly, we did explore known risk factors. Based on previous research, we combined four demographic variables to arrive at a high-risk profile for socially Isolated Individuals. The indicators included not having children, not having a strong connection to a religion, not being employed or in education/training, and not having completed college. Data such as marital status and household occupancy were not available.

The percentage of people in each of the four risk groupings was as follows.

The percentage of people with all four risk Indicators (high risk for social Isolation) was 1.84.

Figure 5

Social Isolation risk scores were correlated with UCLA Loneliness. Scores. They were also associated with age, with older Individuals tending towards higher social Isolation risk scores.

Overall rates of Problematic Social Media Use

Of those who used social media over the past 12 months, 9.54% reported five or more symptoms, scoring above the symptom cut-off on the social media disorder scale (SMD).

Figure 6

Loneliness was a concern across all nations in the study. Even in the nations reporting the lowest levels of loneliness, around 1 in 5 people were lonely based on the UCLA-LS scores. In some nations (Japan, Pakistan, Bangladesh, Malaysia, Ghana, and India), rates were as high as 1 in 2. The well-established and extensively documented links between loneliness and an increased risk of physical and mental health problems underscore the public health implications of social disconnection (WHO, 2025).

One idea proposed to explain the perceived rise in loneliness is the “lost community” hypothesis. Within this formulation, increasing urbanisation and individualism are implicated in the erosion of social connections and the rise of loneliness. However, in the present study, nations reporting the highest levels of loneliness would traditionally be considered to have relatively collectivist national cultural values (Hofstede, 2001). It might be that urbanisation, industrialisation, and the creeping influence of individualistic values are most strongly associated with the erosion of social connections during the transitional phase, as people migrate from rural lifestyles to urban living and from traditional values to mindsets shaped by globalisation and information technology, ushering in a period of cultural dissonance.

Problematic social media use and gaming disorder symptoms were also observed across all nations. Even if the rate of 10% is a significant overestimate, the popularity of social media and gaming renders problem use worthy of further research attention and preventative intervention. Both problematic social media use and gaming disorder symptoms were correlated (medium effect size) with loneliness. This was the case across all 35 nations, although the strength of the relationship varied widely, from large effects (strong positive correlations) for Egypt and Kuwait to small effects (weak positive correlations for Japan. Even after controlling for all other demographic correlates such as age, gender and education level, gaming disorder symptoms, and problematic social media use were the strongest predictors of loneliness.

Numerous previous studies report similar associations between problematic technology use and loneliness, with several longitudinal studies reporting bidirectional relationships, that is, loneliness at time one predicts problematic technology use at time two, and vice versa (Zhang et al., 2023). This suggests that initiatives targeting loneliness may also reduce the risk of problematic technology use, and initiatives targeting problematic technology use may attenuate the risk of loneliness. At least one intervention study (controlled trial) reports such an effect (Thomas et al. In Prep).

Although the present study did not directly assess social isolation, known risk factors associated with social isolation were quantified (e.g., not in employment, training or education; not a parent). These individuals, those with fewer social roles and opportunities for connection, were present across all nations, however to a far lesser degree than loneliness. Unsurprisingly social isolation risk was correlated with loneliness as has previously been documented (Ge et al., 2017; Taylor et al., 2023). Similarly, social isolation risk was also associated with problematic technology use, but to a far lesser extent than loneliness showing small effect sizes (weak positive correlations).

The present study has the usual limitations associated with cross-sectional and correlational survey research. The correlational nature of the study means we cannot ascribe problematic technology use a causal role in the onset, maintenance or worsening of loneliness. However, a clear strength of the current study was the ability to identify the existence of the problematic technology use – loneliness relationship across each of the 35 participating countries using large representative samples within each territory.

While correlation is not causation, it is a cause for concern especially when the WHO estimate that loneliness and social isolation contribute to close to a million deaths annually (World Health Organization, 2025). Successful attempts to reduce loneliness and social isolation will contribute greatly to the overall mental, physical, and social health of society. This study represents a modest contribution towards a better understanding of loneliness and social isolation and the possible role that problematic technology use might play.