Community Connection and Loneliness in Canada

ABSTRACT

The purpose of this study was to examine how loneliness relates to community size, participation and attitudes. We conducted two studies using three large‐scale Canadian datasets (total N = 20,071). Community size was determined by census postal code areas, and loneliness, community participation and attitudes were evaluated by self‐report ratings. In each cross‐sectional study, we use correlations, multiple regression and one‐way ANOVA analyses to evaluate the relationship of loneliness to urban–rural communities, group participation and ratings of connection and belongingness. In both studies, lower loneliness was predicted by higher feelings of connection in one's community. People who participated in groups were also less lonely, but the relationship was weak. Only Study 2 results showed a pattern of relationship between loneliness and urban/rural categories; participants living in urban communities identified higher loneliness. Attitudes about community connection are important predictors of loneliness where more physical variables of participation and size have a much smaller relationship. Measurement limitations and community characteristics are discussed.

1. Introduction

Our day‐to‐day experience is shaped by where we live and what we do, but the relationship of loneliness and these experiences is not fully clear. Canada has a wide variety of residential areas, from life in a bustling city, a quiet town, a sprawling farmland or a remote forest. Research on loneliness and community size has mostly focussed on older adults, with concerns about access to services in rural areas (Burholt and Scharf 2013; Havens et al. 2004). The activities of life in a city compared to smaller towns or remote locations are different for all ages, but few studies examine younger and middle‐aged adults. Scholars have proposed that as urban areas developed in North America, leisure activities and organizations and clubs became important in building community. Known as social capital, acts of community participation and feelings of belonging in one's neighbourhood have been shown to relate to loneliness (Coll‐Planas et al. 2017; Refaeli and Achdut 2022) and may be an important avenue for intervention. The present study focuses on how loneliness relates to Canadians' community size, participation and attitudes.

Most research on urbanization and loneliness has focussed on older adults. Findings largely show lower loneliness for those living in rural areas or areas with lower population density (Beere et al. 2019; Henning‐Smith et al. 2019; Pearlman‐Avnion et al. 2020; MacDonald et al. 2020; Scharf and de Jong Gierveld 2008). Havens et al. (2004) found that ratings of loneliness were similar between urban‐ and rural‐dwelling older adults, but that different rural and urban lifestyle variables changed the association with loneliness. For rural participants, loneliness was predicted by living alone, inadequate future income, feeling that seniors are not respected, low life satisfaction and the presence of multiple chronic illnesses. For urban participants, loneliness was predicted by being widowed and having multiple chronic illnesses. Burholt and Scharf (2013) reported that rural‐dwelling older adults participated in fewer social activities but had greater social resources; if access to family and friends was limited, loneliness increased. Few studies have looked at smaller differences in urbanization, finding similarly that those living in higher density locations were more likely to feel lonely (Finlay and Kobayashi 2018).

Research investigating the association of loneliness and urbanization in the wider age range of adulthood is scarce. Looking at urbanization and broader mental health, mood and anxiety disorders are more prevalent in urban centres and residents report fewer social supports, whereas rural inhabitants report a stronger sense of community and lower rates of depression (Karmakar and Raychaudhuri 2015; Peen et al. 2010; Romans et al. 2011). Interestingly, Bu et al. (2020) found that during COVID‐19 lockdowns, living in a rural area was a protective factor against loneliness. MacDonald et al. (2020) reported that in the Netherlands, loneliness was higher in very dense urban areas compared to rural areas, with incremental decreases in loneliness across five levels of urbanization. In contrast to a large country such as Canada, the Netherlands is a densely populated country such that even its most remote locations are < 200 km from an urban centre. Canada is geographically the second largest country in the world but has the 37th largest population. Overall population density for Canada is low; however, within major cities, density is comparable to cities such as Boston, Amsterdam and Chicago (Filipowicz 2018). When accounting for factors of sex, socioeconomic status and race, Canadians living in an urban environment were more likely to experience a major depressive episode (Wang 2004) and to report lower sense of belonging (Romans et al. 2011). To our knowledge, there has not been a published study on the association of loneliness with where Canadians live.

Where one lives plays a role in what one can do. Urban centres often provide many options for social activities, including large events, sports organizations or activity centres, but rural areas may offer communities with closer bonds. Glover (2018) argues that leisure is the starting point for social capital and that in North America, recreation activities were developed to enhance feelings of community in the early twentieth century as urbanization was growing and isolation increasing. Lochner et al. (1999) define social capital as the ‘ecological conditions’ that provide the framework for community development. Glover (2018) adds the idea that social capital is an investment in social relations that results in a benefit to the individual. While some researchers maintain that social capital is a characteristic of a social environment and not an individual characteristic, others have argued that the social capital of an individual can be measured (De Silva 2005). Participating in groups or organizations is known as ‘structural social capital′ or social participation. In a systematic review, De Silva (2005) reported that only 3 of 10 studies found significant relationships between structural social capital and mental illness. Loneliness is related to mental health and psychological conditions such as depression and anxiety; however, loneliness is distinct, and recent literature does suggest that people are less lonely when they report higher structural social capital (i.e., are more involved in organized groups; Baron‐Epel et al. 2022; Coll‐Planas et al. 2017; Refaeli and Achdut 2022).

Nyqvist et al. (2016) found that the association between loneliness and structural social capital was only significant for middle adulthood (ages 30–49 years) when accounting for covariates such as friendships, neighbours and another form of social capital known as ‘cognitive social capital′. Cognitive social capital is the attitudes about one's community with respect to trust, reciprocity and cooperation (Bhandari and Yasunobu 2009). Cognitive social capital is also negatively related with loneliness (Achdut and Refaeli 2021; Domènech‐Abella et al. 2017; Nyqvist et al. 2016). Hill and MacGillivray (2024) found that loneliness and sense of community have many common predictors that have similar magnitudes of importance. Further, sense of belonging mediates the relationship of loneliness to well‐being (Samuels and Jeong 2024). Cognitive social capital has been found to mediate the relationship between structural social capital and loneliness (Sun and Lu 2020). Increasing structural social capital has been proposed as a mental health treatment (Williams et al. 2020) and may be particularly effective for loneliness. Coll‐Planas et al. (2017) reported upon an intervention where groups of elderly people were assigned to visit local group events at libraries, museums and community centres with specific activities (such as crafts, storytelling, etc.) and found that after 2 years, almost half of the participants maintained contact with at least one person and that the intervention had positive effects on loneliness.

Social capital, both structural and cognitive, is emerging as a significant factor in understanding loneliness from what people choose to do. The theory of social capital fits well with models of loneliness such as the filtration model (Hawkley et al. 2008) where structural factors in society impact loneliness through the cognitive perceptions of relationships. Structural factors such as education, income and health are considered ‘distal factors’ influence loneliness by providing social opportunity, which results in social relationships, thereby determining loneliness (Hawkley et al. 2008). While some research has been done about neighbourhood characteristics related to social capital (e.g. Refaeli and Achdut 2022), the relationship of social capital variables and loneliness in Canada has not yet been studied to our knowledge with respect to where citizens live. Buck‐McFadyen et al. (2019) reported that while civic engagement and sense of belonging were higher in rural Canadians, there was no difference between rural and urban participants on their self‐rated mental health. Findings about social capital and mental health have varied, but studies about social capital and loneliness have consistently shown that activity in organized groups and feelings of trust and reciprocity do relate to loneliness.

The present study addresses gaps in the community psychology literature by examining loneliness and urbanization in a large spectrum of adults and using multiple categories of urban and rural community sizes. The present study also examines participation in community activities (structural social capital) and as well as attitudes about community (cognitive social capital) including within the context of urban categories, which has not been presented in published studies to the authors' knowledge.

2. Study 1

Given the unique geographic and demographic composition of Canada, how do structural and cognitive social capital components relate to loneliness in Canadians? We expected that:

Hypothesis 1

Loneliness will be related to both participating in their community (structural), as well as attitude towards their community (cognitive) with stronger associations with cognitive social capital compared to structural, consistent with previous studies and with the filtration model of loneliness (Hawkley et al. 2008; Nyqvist et al. 2016; Sun and Lu 2020).

We also examined the relationships between social capital and self‐rated mental health, anxiety and depression and how these correlations compared to the association between social capital and loneliness. We expected that:

Hypothesis 2

Social capital components would be more closely related to loneliness than broader mental health concerns.

Next, we asked how living in urban and rural contexts (i.e., community size) related to loneliness. Given the previous research that suggests individuals in rural areas are less lonely, we expected similar results to MacDonald et al. (2020) that:

Hypothesis 3

As urbanization increased, so would loneliness.

Looking at community size and social capital, we expected that results from our study would be similar to that of Buck‐McFadyen et al. (2019):

Hypothesis 4a

Participants living in more rural areas had more involvement in organized groups.

and

Hypothesis 4b

Participants living in more rural areas felt a greater sense belonging in community. Finally, we explored whether social participation impacts loneliness in urban populations differently than rural.

3. Method

3.1. Participants

The data for this study came from the 2022 Network of Economic and Social Trends (NEST) Omnibus Survey. The survey was a probability sample of adult residents of Canada, selected based on age, gender and region. The survey included items contributed by researchers at the University of Western Ontario with interest in attitudes and opinions on public issues in Canada. The data were collected online through the Qualtrics platform where participants provided informed consent. Participants were 2303 Canadians (50.32% women, 48.85% men, and 0.65% nonbinary or other gender, 0.22% did not specify; age range 18–100 years old). Reported ethnicity was 76.25% White, 1.50% Black, 5.8% Chinese, 1.1% Filipino, 2.95% South Asian, 1.26% Southeast Asian, 0.78% Indigenous, 2.51% “other visible minority”.

4. Measures

4.1. Loneliness

The survey included the Three‐Item Loneliness Scale (TILS; Hughes et al. 2004). The items for the TILS were taken from the Revised University of California Los Angeles Loneliness Scale (R‐UCLA; Russell et al. 1980) to develop a shorter scale (the full version has 20 items) that remained a reliable and valid measure of loneliness. In the present study, participants were given instructions to indicate how often each of the statements are descriptive of them: ‘I lack companionship’, ‘I feel left out’ and ‘I feel isolated from others’. Each item was rated on a 3‐point scale of 1–3 (Hardly ever, Sometimes, Often). The internal consistency was good (Ω = 0.85).

4.2. Community Participation

Participants were asked if they were a member or participant in any of the following: a political party or group; a sports or recreational organization; a union or professional association; cultural, educational or hobby organization; a religious‐affiliated group; a school group, neighbourhood, civic or community association; a service club; a senior's group; a youth organization; ethnic or immigrant association or club; a co‐operative organization; other type of organization. Participants were asked to select all that applied to them. Most participants (56%) did not select any organizations, and sample sizes in each category were small. Responses were therefore used to create a binary variable that indicated whether the participant was involved in any kind of organization.

4.3. Attitude Towards Community

The survey included two questions to measure participants' attitude to their community: ‘I feel a sense of being connected with other members of my community’ and ‘I feel good about being a member of my community’. These items were rated on a scale of 1 = Strongly disagree, 2 = Somewhat disagree, 3 = Somewhat agree and 4 = Strongly agree. These two items were combined to create a variable that indicated participants' subjective ratings of their cognitive social capital.

4.4. Rurality

The study also collected postal code data, which was used to determine neighbourhood classification for how rural or urban the participants' area of residence is. We used the Postal Code Conversion File Plus Version 8 A from December 2022 and the variable ‘Community Size’, which uses 2021 census population data in each census metropolitan area or census agglomeration to categorize locations into five community size classifications. The classifications are found in Table 1, along with the number of participants in each category.

Table 1.

Community size classification and cities.

Community size	N	Population	Cities
1	819	1,500,000 +	Toronto, Montreal, Vancouver
2	546	500,000–1,499,999	Ottawa‐Gatineau, Edmonton, Calgary, Québec, Winnipeg, Hamilton
3	418	100,000–499,999	18 census metropolitan areas + 7 large census agglomerations
4	227	10,000–99,999 (any CMA/CA < 100,000)	106 census agglomerations
5	268	< 10,000 (any non‐CMA/CA)	Rural and small town
9	25	Missing

Abbreviation: CMA/CA = Census Metropolitan Area or Census Agglomeration.

4.5. Mental Health

The survey included one question on general mental health: ‘In general, would you say your mental health is…?’ to which participants chose from response options of ‘poor’, ‘fair’, ‘good’, ‘very good’ and ‘excellent’. The other item for comparing loneliness to mental health was: ‘During the past week, about how often did you feel depressed or anxious?’ to which participants responded on a scale of 1–5 (None of the time to All of the time).

4.6. Covariates

We controlled for covariates of gender and marital status. Participants identified whether they were a woman, a man, nonbinary or another gender (and asked to specify). Participants specified their current marital status as: ‘married’, ‘living with a partner’, ‘divorced’, ‘separated’, ‘widowed’ and ‘never married’. We coded these as 1 = ‘married’ and ‘living with a partner’, and 0 = ‘all other responses’.

5. Analytic Procedure

We conducted the analyses using R version 4.2.2 (R Core Team 2022). Missing data were addressed with listwise deletion. To address H1, we conducted separate regression analysis with loneliness as the dependent variable, and community participation and attitudes as independent variables, while controlling for age, gender and marital status. For H2, we compared correlations of loneliness with both community variables to those of the self‐rated mental health with community variables by using confidence intervals. We addressed H3 using one‐way ANOVA analyses with categories of community size and loneliness and addressed 4a, 4b with one‐way ANOVA analyses with categories of community size and community participation, as well as community attitude. To explore whether community participation impacts loneliness in urban populations differently than rural, we examined the intraclass correlation coefficient (ICC). The ICC is the ratio of the between cluster variance to the total variance, which reveals the proportion of the total variance in loneliness that is accounted for by the clustering of urbanization categories. We calculated the ICC by estimating a linear mixed‐effects model (using lme4 package; Bates et al. 2015). If the ICC was a small correlation or greater (r > 0.10), we would use multilevel modelling to examine the intercept and slope differences in each group.

Given the large sample size in this study, we report effect sizes and confidence intervals along with p values. Specifically, we give emphasis to correlations of r = 0.10 and greater, as, according to a meta‐analysis by Gignac and Szodorai (2016), r = 0.10 corresponds to the 25th percentile in individual differences research and is considered a small effect size. Medium and large effect sizes are considered at r = 0.20 and r = 0.30, respectively. We use the same values for partial correlations. For t‐tests, we also estimate Cohen's d, using guidelines of 0.20 = small, 0.50 = moderate and 0.80 = large. Data are available at: 2022 NEST Omnibus Survey—Network for Economic and Social Trends (NEST) Dataverse. This study was preregistered on the Open Science Framework: https://osf.io/j39x5.

6. Results

Table 2 contains the descriptive statistics and correlations between the study variables. General mental health had a small relationship with community participation and this relationship was stronger than between loneliness and community participation (z = 5.45, p < 0.001). The relationship of frequency of depressed or anxious feelings and community participation was negligible, and the magnitude was not significantly different than between loneliness and community participation (z = 0.25). Community attitude had a moderate positive correlation with general mental health, and this correlation was not significantly different from loneliness in magnitude (loneliness had a negative correlation with mental health; z = 0.71). The correlation of community attitude with general frequency of depressed and anxious feelings was significantly smaller than the correlation of community attitude with loneliness (z = 2.10, p < 0.05).

Table 2.

Correlations between study variables.

	M (SD)	1	2	3	4	5	6	7
1. Sex	0.49 (0.50)	1.00
2. Age	49.29 (17.91)	0.15 [0.10, 0.19]	1.00
3. Marital status	0.60 (0.49)	0.12 [0.08, 0.15]	0.22 [0.18, 0.27]	1.00
4. Loneliness	1.89 (1.83)	−0.11 [−0.15, −0.08]	−0.29 [−0.33, −0.25]	−0.28 [−0.32, −0.24]	1.00
5. General mental health	2.25 (1.10)	0.13 [0.08, 0.18]	0.27 [0.23, 0.30]	0.17 [0.13, 0.21]	−0.52 [−0.55, −0.48]	1.00
6. Depressed/anxious	1.15 (1.10)	−0.16 [−0.21, −0.12]	−0.30 [−0.34, −0.26]	−0.18 [−0.23, −0.14]	0.58 [0.55, 0.61]	−0.60 [−0.63, −0.57]	1.00
7. Community participation	0.44 (0.50)	0.04 [0.00, 0.08]	0.10 [0.06, 0.14]	0.04 [0.00, 0.08]	−0.04 [−0.08, 0.00]	0.12 [0.08, 0.17]	−0.06 [−0.10, −0.01]	1.00
8. Community attitude	3.86 (1.35)	−0.05 [−0.09, −0.01]	0.15 [0.11, 0.18]	0.09 [0.05, 0.13	−0.21 [−0.25, −0.17]	0.23 [0.19, 0.27]	−0.15 [−0.19, −0.10]	0.19 [0.15, 0.23]

Note: Sex is coded 1 = woman, 2 = man; estimation of 95% confidence intervals is based on 1000 bootstrap samples.

The hierarchical regression analysis of community participation and attitudes is found in Table 3. Age, gender and marital status were all statistically significant predictors of loneliness (ps < 0.01); partial correlations indicated a moderate relationship with age and marital status where people who were younger and living without a partner were more likely to be lonely. Women were also more likely to be lonely, but the effect size was negligible. The amount of variance in loneliness explained by the Model 1 variables was 14%. In Model 2, adding in the community participation variable did not significantly increase the amount of variance explained by the model (ΔR ² = 0.00); participation in a group did not predict loneliness scores compared to those who did not participate in any groups (p = 0.298, semi‐partial r = 0.02). Community attitude was a significant predictor of loneliness over and above demographic covariates and structural social capital with a small‐to‐moderate effect size (p < 0.001, semi‐partial r = −0.17) and added 3% of the total variance explained. People who rated feeling good about and feeling more connected to their community had lower loneliness scores.

Table 3.

Hierarchical regression evaluating structural and cognitive social capital as predictors of loneliness.

Predictor variables	Model 1	Model 2	Model 3	Semi‐partial r
Intercept	3.65	3.66	4.39
Age	−0.02*	−0.02*	−0.02*	−0.21*
Gender	−0.20*	−0.20*	−0.25*	−0.07*
Marital status	−0.82*	−0.82*	−0.78*	−0.21*
Structural social capital		−0.04	0.08	0.02
Cognitive social capital			−0.23*	−0.17*
R 2	0.14*	0.14*	0.16*
ΔR 2		0.00	0.03*

Note: Regression coefficients are unstandardized. Gender, 0 = woman, 1 = man; marital status: 1 = married or living with partner, 0 = all other responses; structural social capital, 1 = at least one group, 0 = no groups.

^* p < 0.01.

The one‐way ANOVA to examine whether loneliness ratings differ between levels of urbanization was significant, F(4, 2273) = 3.31, p = 0.010. Post hoc pairwise t‐tests (using pooled standard deviations with Holm–Bonferroni correction) revealed one significant relationship; community size 2 (CS2) and CS3 were significantly different, with a small effect size, indicating that CS2 had higher ratings of loneliness (d = 0.22, p = 0.006).

Comparing levels of urbanization to community participation and attitude with separate one‐way ANOVAs identified no significant differences across levels of urbanization on participation [F(4, 2273) = 1.43, p = 0.219] or attitude [F(4, 2273) = 0.65, p = 0.628].

The ICC = 0.005 for community participation suggesting that 0.50% of the variance in loneliness ratings is accounted for by community size. This value is a negligible effect size and indicates that the relationship between community participation and loneliness is not significantly different across different community sizes. Similarly, for community attitude and loneliness by community size, the ICC was 0.006.

7. Discussion

Overall, the results point to a moderate connection between loneliness and community attitudes, and a weak relationship between loneliness and community participation. There is little evidence for a relationship of loneliness to community size.

Evaluating our first hypothesis, loneliness was predicted by community attitude with a moderate effect size, over and above demographic variables of age, marital status and gender; lower loneliness was predicted by higher ratings of feeling connected with and good about their community. Participating in community organizations was not a significant predictor of loneliness and its bivariate correlation was negligible.

Positive attitude towards one's community connection was more closely related to low loneliness than to reports of low depression and anxiety, but its relationship with general mental health was of a similar magnitude to that of its relationship with loneliness. H2 was not supported as we expected loneliness to be more closely related to community attitude than broader mental health; however, there was a distinction that showed loneliness with a stronger relationship to feelings of belonging in a community than depression and anxiety.

The relationship of loneliness and levels of urbanization was not a linear pattern as has been seen in previous research (e.g., MacDonald et al. 2020). Loneliness was rated as highest in the second most populated category (population 500,000–1,499,999) and was only significantly higher than the middle community size (population 100,000–499,999; H3). There were no significant differences in group involvement or sense of belonging in the community across community size groupings, failing to support H4a and H4b.

Finally, comparing the relationship of community participation and loneliness in different community sizes revealed no significant differences in the impact of participation on loneliness when clustering communities by population size.

The results of Study 1 indicate that loneliness is linked with positive attitudes about community membership and general mental health. More objective lifestyle data such as belonging to groups or population density of one's community does not emerge as significant factors in comparing loneliness among Canadians. The difference in community size was small and only emerged in one comparison, so replication of that result would be beneficial before we make definitive conclusions about the relationship of loneliness and community size.

8. Study 2

The second study extends and replicates the investigation of loneliness, community attitude and community size using data from Statistics Canada. The measures used are different from Study 1, so results that replicate will add to the strength of the conceptual understanding of loneliness and community characteristics. We expect, consistent with Study 1 and previous research (Nyqvist et al. 2016; Sun and Lu 2020), that:

Hypothesis 1

Loneliness will be negatively related to variables related to cognitive social capital. These variables in particular look at trust in others and community belonging.

Also replicating Study 1, we predict that:

Hypothesis 2

Both loneliness and overall mental health will be related to cognitive social capital to a similar magnitude.

Finally, we will examine differences among levels of community size for loneliness and cognitive social capital.

9. Method

9.1. Participants

The data for this study came from the Canadian Social Survey Wave 7, 2022 and Wave 8, 2023. The Canadian Social Survey is conducted by Statistics Canada every 3 months. Data are collected by electronic questionnaire, targeted at any noninstitutionalized person ages 15 years and older. Participants are recruited via mail invitation, provide informed consent and may complete the survey electronically or by mail. The survey may be completed in either English or French. Wave 7 was collected from October 21 to December 4, 2022, and is from 9684 participants (4390 men, 5294 women) with an average age of 52.82 years (SD = 17.65). The Wave 8 data were collected from April 21 to June 4, 2023, and are from 8084 participants (3775 men, 4309 women) with an average age of 54.64 years (SD = 17.36).

9.2. Measures

9.2.1. Demographics

Participants entered their age in years and identified their sex at birth as ‘male’ or ‘female’. The survey asked participants' marital status with response options of ‘Married’, ‘Living common law’, ‘Never married and not living common law’, ‘Separated and not living common law’, ‘Divorced and not living common law’, ‘Widowed and not living common law’. We coded these as 1 = ‘Married’ and ‘Living common law’, and 0 = ‘all other responses.

9.2.2. Loneliness

The survey used a single item of frequency of feeling lonely: ‘How often do you feel lonely?’ to which participants responded on a 5‐point scale of 1 = Never, 2 = Rarely, 3 = Sometimes, 4 = Sometimes, 5 = Often.

9.2.3. Attitude Towards Community

The survey included two items that related to community attitude (cognitive social capital). Sense of belonging was measured by the item, ‘How would you describe your sense of belonging to your local community?’ with response options of 1 = Very weak, 2 = Somewhat weak, 3 = No opinion, 4 = Somewhat strong, 5 = Very strong. Trust in others was measured by a binary choice item for participants to choose whether 1 = ‘Most people can be trusted’ and 2 = ‘You cannot be too careful in dealing with people’.

9.2.4. Rurality

The study included a variable of ‘Community Size and Metropolitan Influence Zones’ that was derived from participants' geographic location. The classifications are found in Table 4 along with the number of participants in each category. The first four community size classifications are the same as they were in Study 1, and 5–7 are an expanded category of Non‐CMACA areas, depending on their proximity and influence from a metropolitan area, that is, Non‐CMACA; Strong Metropolitan Influence Zone (MIZ) is an area that is a rural area or small town but is near a larger metropolitan area.

Table 4.

Community Size and Metropolitan Influence Zone classification and cities.

Community size	N 2022	N 2023	Population	Cities
1	2706	2153	1,500,000+	Toronto, Montreal, Vancouver
2	1760	1606	500,000–1,499,999	Ottawa‐Gatineau, Edmonton, Calgary, Québec, Winnipeg, Hamilton
3	2152	1843	100,000–499,999	18 census metropolitan areas + 7 large census agglomerations
4	1304	1114	10,000–99,999 (any CMACA < 100,000)	106 census agglomerations
5	654	426	Non‐CMACA; Strong MIZ	Rural and small town
6	728	569	Non‐CMACA; Moderate MIZ	Rural and small town
7	380	373	Non‐CMACA; Weak/No MIZ	Rural and small town

Abbreviations: MIZ = Metropolitan Influence Zone; CMA/CA = Census Metropolitan Area or Census Agglomeration.

9.2.5. General Mental Health

The survey asked participants about their mental health overall with a single item: ‘In general, how is your mental health?’ Response options were 1 = Poor, 2 = Fair, 3 = Good, 4 = Very good, 5 = Excellent.

9.3. Analytic Procedure

Missing data were addressed with listwise deletion. We use hierarchical regression, comparison of correlations using confidence intervals, and one‐way ANOVAs to address the hypotheses as described in Study 1. When we use p values for significance for t‐tests, we set α at 0.002 given a Bonferroni correction for 21 comparisons. We otherwise focus on effect size to determine significant results. This study was preregistered on the Open Science Framework: https://osf.io/vhmjf.

10. Results

Table 5 contains the means, standard deviations and intercorrelations for all study variables. Loneliness and mental health had large, negative correlations in both the 2022 and 2023 samples. Trust in people was more highly correlated with mental health than with loneliness in both years, though magnitudes for both relationships were small (z = 2.13, p < 0.05 and z = 2.59, p < 0.01, respectively). Sense of belonging in one's community was correlated highly with loneliness and mental health; in 2022, the relationship with loneliness was a lower magnitude compared to mental health, but in 2023, the correlations were not significantly different (z = 0.82).

Table 5.

Descriptive statistics and intercorrelations of study variables.

	2022	2023
	M (SD)	M (SD)	1	2	3	4	5	6	7
1. Sex	1.55 (0.50)	1.53 (0.50)	1.00	0.01 [−0.01, 0.03]	−0.09 [−0.11, 0.07]	0.07 [0.05, 0.09]	−0.08 [−0.10, 0.05]	0.00 [−0.02, 0.03]	0.01 [−0.01, 0.03]
2. Age	52.82 (17.65)	54.64 (17.36)	0.00 [−0.02, 0.02]	1.00	0.07 [0.05, 0.09]	−0.13 [−0.15, −0.10]	0.18 [0.16, 0.20]	0.04 [0.02, 0.06]	0.16 [0.14, 0.18]
3. Marital status	0.61 (0.49)	0.60 (0.50)	−0.06 [−0.08, −0.04]	0.08 [0.06, 0.10]	1.00	−0.29 [−0.31, −0.27]	0.12 [0.10, 0.15]	0.07 [0.05, 0.10]	−0.03 [−0.05, 0.01]
4. Loneliness	2.45 (0.99)	2.36 (1.03)	0.10 [0.08, 0.12]	−0.10 [−0.12, −0.08]	−0.30 [−0.32, −0.28]	1.00	−0.44 [−0.46, −0.43]	−0.11 [−0.13, −0.08]	−0.32 [−0.35, −0.31]
5. Mental health	3.47 (1.06)	3.49 (1.05)	−0.08 [−0.10, −0.06]	0.20 [0.18, 0.22]	0.12 [0.10, 0.14]	−0.43 [−0.44, −0.41]	1.00	0.15 [0.12, 0.17]	0.33 [0.31, 0.35]
6. Trust	1.49 (0.50)	1.46 (0.50)	−0.02 [−0.04, 0.00]	0.04 [0.02, 0.06]	0.09 [0.07, 0.11]	−0.12 [−0.14, −0.10]	0.15 [−0.14, −0.17]	1.00	0.17 [0.15, 0.19]
7. Belonging	3.10 (1.25)	3.15 (1.24)	0.00 [−0.02, 0.02]	0.14 [0.12, 0.16]	0.07 [0.05, 0.09]	−0.30 [−0.32, −0.29]	0.34 [0.32, 0.35]	0.17 [0.15, 0.19]	1.00

Note: Sex was coded 1 = men, 2 = women. Estimation of 95% confidence intervals is based on Fisher's r–z transformation with bias adjustment. 2022 is below the diagonal and 2023 is above.

The hierarchical regression analysis is found in Table 6 for 2022 and Table 7 for 2023. In both 2022 and 2023, age and sex explain 2% of the variance of loneliness in Model 1, and adding marital status increased the amount of variance explained by 8%. Semi‐partial correlations for Model 3a and 3b in both years had negligible values for age and sex (with a small negative semi‐partial correlation for age in 2023 Model 3b). Controlling for age, sex and either sense of belonging or trust in others, marital status consistently had a moderate relationship with loneliness.

Table 6.

Hierarchical regression models predicting loneliness from 2022 sample.

Predictor variables	Model 1	Model 2	Model 3a	3a semi‐partial r	Model 3b	3b semi‐partial r
Intercept	2.46	2.81	3.36		3.06
Age	−0.01	0.00	0.00	−0.04	0.00	−0.08
Sex	0.19	0.15	0.15	0.08	0.15	0.08
Marital status		−0.59	−0.55	−0.27	−0.57	−0.28
Sense of belonging			−0.22	−0.27
Trust in people					−0.18	−0.09
R 2	0.02	0.10	0.18		0.11
ΔR 2		0.08	0.08		0.01

Note: All coefficients, R ² and ΔR ² are significant at p < 0.001. Regression coefficients are unstandardized. Sex, 1 = male, 2 = female; Marital status, 1 = married or living with partner, 0 = other responses.

Table 7.

Hierarchical regression models predicting loneliness from 2023 sample.

Predictor variables	Model 1	Model 2	Model 3a	3a semi‐partial r	Model 3b	3b semi‐partial r
Intercept	2.55	2.92	3.50		3.15
Age	−0.01	−0.01	0.00	−0.06	−0.01	−0.10
Sex	0.15	0.10	0.10	0.05	0.10	0.05
Marital status		−0.59	−0.55	−0.26	−0.58	−0.27
Sense of belonging			−0.25	−0.29
Trust in people					−0.17	−0.08
R 2	0.02	0.10	0.18		0.11
ΔR 2		0.08	0.08		0.01

Note: All coefficients, R ² and ΔR ² are significant at p < 0.001. Regression coefficients are unstandardized. Sex, 1 = male, 2 = female; Marital status, 1 = married or living with partner, 0 = other responses.

In Model 3a, sense of belonging added an additional 8% for a total amount of explained variance of 18% in both years. Controlling for the demographic and marital status variables, sense of belonging had a moderate‐to‐large negative partial correlation with loneliness.

For both 2022 and 2023, trust in people added 1% for a total variance explained of 11% in Model 3b. The semi‐partial correlations identified a negative, but negligible effect size relationship with loneliness.

The one‐way ANOVA to examine whether loneliness ratings differ between levels of urbanization in 2022 was significant, F(6, 9678) = 5.23, p < 0.001). Loneliness was highest in large cities (Community Size 1; CS1) and generally had a pattern of decreasing loneliness as population decreased. Welch's t‐test results revealed that CS1 had significantly higher ratings of loneliness than CS5 and CS6 (p < 0.001, d = 0.20 and 0.14, respectively). CS2 and CS3 each also had higher ratings of loneliness than CS5 (p < 0.001, d = 0.18 and 0.15, respectively).

In the 2023 data, the one‐way ANOVA of loneliness by community size was also significant, F(6, 8078) = 3.81 (p < 0.001), and the pattern was similar with highest loneliness in large cities, and lower loneliness in non‐CMACA areas. Post hoc Welch's t‐tests identified that CS1 and CS2 each had significantly higher loneliness than CS6 (p = 0.001 and 0.002; d = 0.15 and 0.16), using our corrected cutoff of p = 0.002.

Comparing levels of urbanization to cognitive social capital with separate one‐way ANOVAs revealed significant differences with a sense of belonging in the 2022 data (F(6, 9678) = 22.35, p < 0.001) and 2023 = (F(6, 8078) = 18.68, p < 0.001). In 2022, the largest three communities had significantly lower ratings of belonging compared to CS4, CS6 and CS7 (p < 0.001, ds = 0.18–0.42). CS4 rated feelings of belonging as significantly lower than CS7 (p < 0.001, d = 0.22). CS5 rated feelings of belonging significantly lower than CS6 and CS7 (p < 0.001, d = 0.020 and 0.030). In 2023, CS1 had significantly lower belonging ratings than all other community sizes (p = 0.001 to p < 0.001), with small effect sizes compared to CS4–CS7 (d = 0.20–0.39). CS2 had significantly lower ratings of belonging compared to CS4, CS6 and CS7 (p < 0.001) with small effect sizes compared to CS6 (d = 0.30) and CS7 (d = 0.24). For trust in people, the ANOVA was significant for the 2023 data (F (6, 8078) = 4.00, p < 0.001); however, the pattern of mean differences was not consistent from high to low community sizes as it was with loneliness and belonging. CS3 had significantly lower trust than CS1 (p = 0.002, d = 0.10), CS4 (p = 0.002, d = 0.12) and CS7 (p = 0.001, d = 0.18). Effect sizes for these three were all considered to be small. The 2022 ANOVA for trust in people was not statistically significant (F (6, 9678) = 1.87, p = 0.081).

11. Discussion

Overall, the results showed a consistent moderate to strong relationship between loneliness and community attitude, particularly with feelings of belonging. Some of the comparisons showed that general mental health had a slightly stronger relationship with cognitive social capital variables than loneliness, and others showed that the magnitudes were similar. Sense of belonging in one's community was a significant predictor of loneliness controlling for demographic variables and had a similar magnitude to marital status. Trust in people was much lower of a factor in predicting loneliness.

When evaluating community size, there was an overall pattern in both years of data where larger communities reported significantly higher rates of loneliness. Sense of belonging in one's community had a similar pattern where larger communities reported significantly lower sense of belonging than smaller communities, but the effect sizes showed small to moderate differences. When it comes to trust in people, there were almost no differences between the community sizes in both years of data. It is important to note that the effect sizes for these differences are quite small, which suggests that the differences that citizens experience or observe in real life may not be very noticeable.

12. General Discussion

Loneliness and cognitive social capital are consistently related over and above other demographic variables, and both demonstrated patterns in relationship to community size. Individuals who report higher levels of connectedness and feelings of belonging are more likely to also report lower loneliness. This relationship is highly similar to the relationship between cognitive social capital and overall mental health. While we initially hypothesized that loneliness would have a stronger relationship with cognitive social capital than overall mental health, results from both studies indicated that the magnitudes were comparable or slightly stronger with overall mental health. These results suggest that not only does a lack of sense of belonging go hand‐in‐hand with loneliness, but it permeates one's overall mental health; that is, the loneliness that accompanies feeling disconnected is inextricably linked to general psychological wellbeing. The inverse must also be considered such that poor mental health may also be related to a negative perception of belonging in the community. The link between sense of community belonging and mental health is consistent with previous research in Canada by Kitchen et al. (2012) who found that sense of belonging to the local community was predicted by mental health when controlling for physical health, life stress and demographic variables such as age, sex, education and income. Outside of Canada, Nyqvist et al. (2016) also found a strong relationship between loneliness and community belonging and trust. In the present study, trust in others had a small relationship with loneliness, suggesting that among social capital concepts, feelings of belonging in a community have a larger connection with overall loneliness.

There is also a general pattern of decreasing loneliness with decreasing community size; however, the differences were small and observable only in a very large sample. The results differed between the two studies, where in Study 1, there was a slight difference showing lower loneliness for the second‐largest community sizes of cities such as Winnipeg, Hamilton, or Calgary compared to the third‐largest community size. With Study 2, there was a consistent pattern for both years of data with the largest three community sizes having highest loneliness, with a decrease towards the smallest rural communities. The inverse pattern was also seen in ratings of sense of belonging in community such that smaller communities had a higher sense of belonging. The differences between community sizes were more pronounced for feelings of belonging than for feelings of loneliness. This pattern has been found with other Canadian research finding that citizens in urban environments reported lower sense of belonging, social support and trust in others in their community (Buck‐McFadyen et al. 2019; Romans et al. 2011). Research from other Statistics Canada analyses using the same community size variable as Study 2 has found higher ratings of life satisfaction in rural areas with low urban influences (Thomson et al. 2025). The explanations for this trend include family networks, rootedness and disamenities. Turcotte (2005) examined Canadian survey data comparing different community sizes and identified that people in rural areas have more family in their regular social contacts. Having a close relationship with relatives can provide stability and a sort of ‘automatic′ belonging rather than having to work to develop friendships and acquaintances. People in rural areas are more likely to know and trust their neighbours than urban residents, which is a core part of cognitive social capital.

Turcotte (2005) also identified that individuals in rural and small towns are likely to have lived in the same place for 5 years or more compared to urban citizens. Simply living in one place for a long period of time is related to feelings of ‘rootedness′, where people report strong community ties, and a high rate of belongingness (Schellenberg et al. 2018). Finally, in addition to features of rural living fostering feelings of community, there are features of urban residency that impede community cohesion. Disamenities in urban settings include higher crime rates, increased pollution and lack of access to safe public places (Weiss et al. 2011). Any of these features may decrease feelings of safety and trust in the community and may discourage regular social interaction with fellow citizens.

Study 1 showed that the relationship to loneliness and participating in an organization was minimal and better accounted for by demographic variables and feelings of connection to one's community. These findings suggest that participation alone is not an indicator of loneliness, but rather a sense of community and belonging needs to accompany involvement to demonstrate any difference in reports of loneliness. While other research results have varied, most have found that the cognitive component of social capital is the important factor in changes in loneliness and that the overall relationship between community participation and loneliness is weak (De Silva 2005; Sun and Lu 2020). This view of social capital and loneliness is consistent with the filtration model of loneliness whereby the structures that provide social opportunity impact loneliness through the mechanism of the cognitive perception of relationships and belonging (Hawkley et al. 2008).

13. Limitations and Future Directions

The measurement of social capital is one that is difficult due to differences in concept, context and perspective (Claridge 2021). The measures in the present study did not encompass broad perspectives on the definitions of social capital but were focussed on community participation (structural) and feelings of trust and belonging (cognitive). These measures are similar to other studies of social capital (e.g., Baron‐Epel et al. 2022; Domènech‐Abella et al. 2017; Nyqvist et al. 2016); however, the use of wider ranging social indicators may provide a more nuanced picture of social capital with how it relates to loneliness (e.g., Buck‐McFadyen et al. 2019). Further, using more robust measures of sense of belonging and trust may provide additional evidence and insight in addition to the results of the present study.

By using five (Study 1) and seven (Study 2) categories for urbanization, the present study extends previous research which largely looks at a dichotomous urban/rural distinction. Future studies could examine urbanization further with more specific descriptions such as whether participants reside in a highly dense urban core, commuter neighbourhoods, or agricultural community.

The present study uses a cross‐sectional design, which cannot make conclusions about causation. Longitudinal designs to evaluate changes in loneliness or feelings of belonging because of community participation will provide a more conclusive answer about the temporal relationship between loneliness and community variables. Future studies may include an experimental design for loneliness based on community participation, examining effects over several months and whether those effects are maintained in longer term such as 1–2 years.

14. Conclusion

In summary, the present study has extended prior research by examining loneliness in multiple community sizes and across adulthood using large representative datasets. We have also shown how the concept of social capital may be understood in association with the filtration model of loneliness. Loneliness in Canadians shows little direct relationship to structural community factors of size and participation in the present study, but sense of belonging and community connection are key factors regardless of where one resides.

Ethics Statement

Study 1 Network for Economic and Social Trends: Ethical approval provided by the Office of Human Research Ethics at the University of Western Ontario. Study 2 Statistics Canada: Data are collected under the authority of the Statistics Act, Revised Statutes of Canada, 1985, Chapter S‐19.

Peer Review

The peer review history for this article is available at https://www.webofscience.com/api/gateway/wos/peer-review/10.1002/jcop.70042.

Baerg MacDonald, K. , and Schermer J. A.. 2025. “Community Connection and Loneliness in Canada.” Journal of Community Psychology 53: 1–12. 10.1002/jcop.70042.

Data Availability Statement

Data for Study 1 are available here: https://borealisdata.ca/dataverse/NEST. Data for Study 2 are available from Statistics Canada. Restrictions apply to the availability of these data.