2.1 Core measures

Loneliness was assessed using the DeJong Gierveld Emotional and Social Loneliness scales [47], comprising two distinct subscales: emotional loneliness (reflecting feelings of emptiness or emotional isolation) and social loneliness (indicating a perceived absence of a supportive social network). This scale has been thoroughly validated in various countries and settings [47]. Participants rated their agreement with six statements reflecting their current feelings, using the response options “Yes”, “More or less”, or “No”. The three emotional loneliness items were: (1) “I miss having people around”, (2) “I often feel rejected”, and (3) “I experience a general sense of emptiness”. Responses indicating emotional loneliness (“Yes” or “More or less”) were scored as 1, while “No” responses were scored as 0. The emotional loneliness subscale score was calculated as the sum of these three items, resulting in a total ranging from 0 to 3, with higher scores reflecting greater emotional loneliness. The three social loneliness items were reverse-scored and included: (1) “There are plenty of people I can rely on when I have problems”, (2) “There are many people I can trust completely”, and (3) “There are enough people I feel close to”. Responses of “No” or “More or less” indicated social loneliness and were scored as 1; responses of “Yes” were scored as 0. Summing these three items provided the social loneliness subscale score (range: 0–3), with higher scores indicating greater social loneliness.

Social engagement was assessed by asking participants to report the frequency with which they had engaged in 20 distinct social activities during the past three months. Activities included greeting a neighbor or stranger, having an in-person conversation, checking in via text or private message, phone conversations with friends or family, group video chats, walking with someone, meeting for meals or drinks, playing video or board games, visiting friends, visiting family, volunteering in the community, helping neighbors or friends with chores, attending community meetings, participating in online discussion groups, engaging in group exercise, attending religious services, making new friends, hugging someone, kissing someone, and having sex. In Study 1, participants rated each activity using an 8-point numerical scale coded as 0 (Never), 1 (More than three months ago), 2 (Within the past three months), 3 (Within the past month), 4 (Within the past two weeks), 5 (Within the past week), 6 (Within the past few days), and 7 (Today). Studies 2 and 3 added two extra activities: to extend an invitation and been invited. Frequency of engagement, rather than recency, was used in these two studies, ranging from 0 (“Not in the past three months”) to 6 (“Daily or almost daily”).

Social network size was measured by asking participants to estimate the number of individuals with whom they had spent time socializing in the previous three months, categorized into seven relationship types: (1) close friends, (2) casual friends, (3) classmates, (4) coworkers, (5) friends of friends, (6) acquaintances, and (7) other social connections. Participants were instructed to assign each social contact to only one category, selecting the one that best described their primary relationship to that individual. The number of individuals in each category was recorded, and summed to compute the total social network size, representing the breadth of social connections.

Time spent socializing was measured by asking participants to report the total number of hours they spent interacting socially with different groups during the past week. Participants provided numeric estimates of their hours spent with (1) family members, (2) friends, (3) coworkers, (4) classmates, (5) neighbors, (6) acquaintances, and (7) strangers. These data were used to evaluate participants’ weekly investment in social activities across diverse social contexts.

Sociodemographic variables included self-reported gender (restricted to men in this analysis), age (later split into decades), ethnicity (categorized as White, East Asian, South Asian, Indigenous, and Other racialized groups), and household income in Canadian dollars (CAD) recoded into numeric values reflecting midpoints of income categories. Apart from those mentioned above, each wave contained unique variables that will be further described under “Measures” of each corresponding study.

2.2 Sample description

The demographic characteristics and loneliness levels for the samples are described in Table 1 and briefly summarized within each corresponding section.

3.1 Measures

Social wellbeing knowledge was assessed using eight binary items designed to evaluate participants’ awareness of the health impacts associated with social connection, loneliness, and social isolation. Participants responded “Yes, I already knew this” or “No, I did not know this” to each of the following statements: (1) “social connection was important for your physical health”, (2) “social connection was important for your mental health”, (3) “loneliness and social isolation are associated with higher risk for disease, such as heart disease and cancer”, (4) “loneliness and social isolation are associated with a higher risk for premature death”, (5) “loneliness and social isolation are as bad for your health as air pollution”, (6) “loneliness and social isolation are as bad for your health as being obese”, (7) “loneliness and social isolation are as bad for your health as living a sedentary lifestyle”, and (8) “loneliness and social isolation are as bad for your health as consuming alcohol or tobacco”. Each response was coded as a binary variable (1 = Yes, 0 = No), and responses were summed to form a composite knowledge score ranging from 0 to 8, with higher scores indicating greater overall knowledge regarding the health implications of social well-being.

3.2 Data analysis

Descriptive statistics, including means (M) and standard deviations (SD), were computed for all study variables to characterize the sample and provide context for regression analyses. Statistical analyses involved two regression models with 5000 bootstrap samples to examine associations between social wellbeing knowledge scores (independent variable) and loneliness, controlling for age, ethnicity, and household income. Percentile-based bootstrap confidence interval (CI) were also calculated. Model assumptions, including linearity, normality, homoscedasticity, and multicollinearity, were examined using diagnostic plots and variance inflation factors.

3.3 Results

Participants (n = 77 men) had a median age of 37 years (range: 18 to 79). Most identified as White (71.4%) and had an approximate median annual household income of CAD 55,000. Other demographic characteristics can be found in Table 1. Table 2 shows the proportion of participants who “knew” about each specific statement. Overall, men demonstrated moderate social wellbeing knowledge (M = 5.47, SD = 2.30). Most participants knew social connection was important for physical (87%) and mental health (86%), whereas fewer knew about associations with obesity (53%) or substance use (57%).

Social wellbeing knowledge was negatively associated with social loneliness (β = −0.18, 95% CI = −0.31 to −0.04) but not emotional loneliness (β = −0.03, 95% CI = −0.15 to 0.08).

4.1 Measures

Participants ranked 17 factors according to how impactful they believed each factor was for physical health, daily wellbeing, disease risk, and longevity. Rankings ranged from 1 (“most impactful”) to 17 (“least impactful”). The specific factors participants ranked were: (1) diet (e.g., food intake, fruit and vegetable consumption), (2) supplement usage (e.g., vitamins, minerals), (3) chemical exposures (e.g., microplastics, pesticides), (4) air quality (e.g., particulate matter, pollution), (5) physical activity (e.g., exercise), (6) time spent sitting, (7) sleep, (8) work-related stress, (9) alcohol use, (10) tobacco use, (11) cannabis use, (12) use of other drugs, (13) financial situation (e.g., income, cost of living), (14) quality of romantic and family relationships, (15) amount of social interactions, (16) stress from relationships, and (17) genetic factors. To quantify Perceived Importance of Social Connection, we combined rankings for items (14) “quality of romantic and family relationships”, and (15) “amount of social interactions”. Specifically, each item’s rank was reverse-coded by subtracting the original rank from the highest possible score so that higher numbers indicate greater perceived importance. These two reverse-coded values were then summed to produce a composite score with possible values ranging from 2 (lowest perceived importance) to 34 (highest perceived importance), with higher scores indicating that participants viewed social connections as more impactful relative to other health factors.

4.2 Data analysis

Like in Study 1, we employed bootstrap regression modelling with 5000 sample repetitions to investigate the associations between the rated importance of the two social health factors and loneliness scores while adjusting for age, ethnicity, and household income.

A Principal Component Analysis (PCA) with varimax rotation was conducted on the social activity items to identify underlying patterns of social engagement. Participants received standardized PCA factor scores for each component, which were subsequently used as dependent variables in regression analyses to explore associations with the perceived importance of social health factors.

4.3 Results

The study sample consisted of 37 men with a median age of 53 years (range: 20 to 73). The majority identified as White (64.9%) and the median household income was CAD 70,000 (see Table 1 for other sociodemographic characteristics). Supplementary Table 1 provides other sample characteristics regarding network size and shared time with said network. The mean score for the perceived importance of relationship quality and social interactions to physical health was 21.27 (SD = 7.37, range = 6–33). The distribution was slightly left-skewed (skewness = −0.35), indicating a modest tendency toward higher scores. The 25th and 75th percentiles were 17 and 27, respectively.

Five distinct social engagement dimensions were identified, explaining 67% of the total variance. These were, community and Organized Group Activities (e.g., volunteering, community meetings, discussion groups, religious services, group exercise); Informal and Socializing Activities (e.g., meeting for coffee, extending or receiving social invitations, going for walks, texting, phone calls); Casual Daily Interactions (e.g., greeting neighbours, casual in-person conversations, chatting informally); Intimate Physical Contact (e.g., sexual activities, kissing, intimate interactions); Family Interactions (e.g., visiting and spending time specifically with family members). All loadings are available in Supplementary Table 2.

Across all models (Table 3), the associations between perceived importance of social connection and indicators of social engagement, network size, and time spent socializing were small, with most 95% bootstrap confidence intervals including zero, suggesting considerable uncertainty and no statistically robust evidence of association. A higher perceived importance of social connection was associated with slightly higher emotional loneliness (β = 0.05), but the 95% CI (−0.02 to 0.13) includes zero. The association with social loneliness was weaker (β = 0.02; 95% CI: −0.05 to 0.11).

Regarding social activities, the largest estimate was for structured ones (β = 0.05; 95% CI: −0.01 to 0.09), while the opposite pattern emerged for casual interactions (β = −0.06; 95% CI: −0.14 to 0.01), suggesting that valuing social connection may be more closely aligned with organized or purposeful social contexts than with spontaneous or informal encounters. Associations with time spent socializing or network size were generally weak and nonsignificant, except for classmates, indicating a very small but possibly real effect in that specific category.

Taken together, the findings suggest a paradoxical role of valuing social connection. While the perceived importance of connection was weakly and inconsistently related to actual social behavior, network size, or time spent socializing, it was not associated to decreases in loneliness levels. If anything, the direction of the effect showed a modest positive association with emotional loneliness. This pattern may reflect a form of motivational tension: men who place a higher value on connection may be more attuned to unmet relational needs, which translates into a search for companion. However, this search does not necessarily translate into meaningful connection. These interpretations remain tentative given the limited strength of the observed effects.

5.1 Measures

Participants evaluated their perceptions of twelve preliminary Canadian Social Connection Guidelines (See Supplementary Table 3 for text of guidelines and questionnaire items), each designed to foster social wellbeing and connection. Guidelines 1–6 focused on individual behaviors, whereas guidelines 7–12 focused on community-level actions. Each guideline was followed by six items assessing participant agreement regarding its importance, perceived benefit, achievability for a typical person, personal achievability, common knowledge, and personal adherence to the guideline. Responses were captured on a 7-point Likert scale ranging from 1 (“Strongly disagree”) to 7 (“Strongly agree”).

5.2 Data analysis

We computed descriptive statistics and Pearson correlations to explore associations between composite variables. Structural equation modeling (SEM) was employed using lavaan [48] to test relationships among perceptions of guideline importance, benefit, common knowledge, achievability, and reported adherence. Multivariable linear regression analyses further explored whether living in accordance with the guidelines was associated with reductions in emotional and social loneliness.

5.3 Results

Among 122 participants, the median age was 45 years (range: 19 to 85), with a median income of CAD 65,000 and most (61.3%) identified as White (See Table 1 for other sociodemographic characteristics). In general, participants reported high agreement with statements that portrayed the guidelines as beneficial (M = 5.66, SD = 0.83) and achievable for typical individuals (M = 5.24, SD = 0.89). They moderately agreed that the guidelines were common knowledge (M = 4.57, SD = 1.01). Self-reported adherence was also moderate (M = 4.28, SD = 1.08). Correlational analyses indicated strong associations between perceived importance and benefits (r = 0.93, p < 0.001). As shown in Fig. 1, moderate to strong correlations were found between perceptions of achievable typicality and adherence (r = 0.44, p < 0.001), as well as between common knowledge and adherence (r = 0.42, p < 0.001).

Fig. 1.Correlations between guideline beliefs and adherence.

Fig. 2 and Table 4 show our SEM results. The model demonstrated an excellent fit to the data: χ² (2) = 3.10, p = 0.21; comparative fit index = 0.996; Tucker-Lewis index = 0.986; Root Mean Square Error of Approximation = 0.067; and Standardized Root Mean Square Residual = 0.009. The perceived importance of the guidelines strongly predicted perceived benefit (Standardized β = 0.93, p < 0.001), whereas perceptions that the guidelines were common knowledge was not a significant predictor of benefit (β = −0.01, p = 0.82). However, both perceived achievability for typical individuals/communities (β = 0.26, p = 0.018) and perceptions that the guidelines were common knowledge (β = 0.31, p < 0.001) significantly predicted self-reported guideline adherence. The perceived benefit was not associated with adherence (β = 0.10, p = 0.35).

Fig. 2.SEM path diagram. n.s.: non-significant.

Greater adherence to the guidelines was significantly associated with lower emotional loneliness (β = −0.21, p = 0.002, R² = 0.34, Adjusted R² = 0.27), after adjusting for age, ethnicity, and household income. Living in accordance with the guidelines was also significantly associated with lower social loneliness (β = −0.29, p < 0.001, R² = 0.26, Adjusted R² = 0.19) in the fully adjusted model.

Study 3 reveals that men’s adherence to social connection guidelines is driven less by abstract beliefs and internal motivation and more by perceptions of social norms (“common knowledge”) and practical achievability. Encouragingly, those who adhered to the guidelines experienced lower emotional and social loneliness, reinforcing the utility of such behavioral frameworks.

6.1 Measures

Measures for this study were the same as used in Studies 1 and 2. The primary variable of interest for this study was one’s self-assessed effort to connect with others, which was measured by asking: “How much effort do you put into intentionally connecting with others?” Responses were ordinally recoded from 1 (“No effort at all”) to 5 (“A great deal of effort”).

6.2 Data analysis

We conducted multivariable linear regressions to examine whether men’s effort to connect was associated with each outcome. All models adjusted for age, ethnicity, and household income. The ggstatsplot package was used to display and describe differences between groups.

6.3 Results

A total of 1681 took part in this study during the third wave of the COVID-19 pandemic. The median age was 32 years (range: 16 to 89), with a median income of CAD 47,500. Most participants identified as White (63.9%). This wave was still marked by lockdowns and an overall imbalance in the dimensions of loneliness compared to the other studies. The sample was also considerably younger (See Table 1 for sociodemographic characteristics).

Fig. 3 shows the number of participants according to their effort (x axis) as well as group differences in social and emotional loneliness (y axis). Significant dose-dependent differences can be observed. These unadjusted results align with the adjusted estimates: greater effort to connect was significantly associated with higher emotional loneliness (B = 0.053, p = 0.024) and lower social loneliness (B = −0.171, p < 0.001). These results suggest a dual reality. On one hand, behavioral activation may help men expand their networks and reduce the sense of social scarcity. On the other hand, that same effort may signal emotional longing and unfulfilled relational depth. Men who strive to connect may do so because they feel emotionally disconnected, not necessarily because they are successful in forming meaningful bonds. This finding mirrors the emotional dissonance observed in Study 2, where valuing connection did not guarantee fulfillment.

Fig. 3.Box/Violin plots comparing loneliness levels by connecting efforts. HDI: Highest Density Interval; JZS: Jeffreys-Zellner-Siow; BF: Bayes Factor; CI: Confidence Interval; obs: observations.

7.1 Recommendations for public health practitioners and policy makers

Public health campaigns should go beyond transmitting knowledge and engaging men in social activities to incorporate strategies to foster deeper emotional connections, particularly men-to-men, as this narrative is contested by unhealthy masculinity models [4]. Messages that emphasize that healthy social behaviors are achievable and align with common sense and existing social norms are more likely to be effective at improving both social and emotional loneliness. Previous research has also found that messages that directly frame men as lonely may trigger avoidance, particularly among older men [6], therefore an adequate balance is required. Additionally, campaigns encouraging men to increase efforts to connect must be aware of the potential emotional costs of unmet social needs, coupling their campaigns with appropriate social training or strategies to help men manage the emotional strain associated with striving to connect [41, 57]. Framing socialization as an activity to help others may give men a sense of responsibility and pride that is compatible with their views of masculinity [6].

While there are positive effects of activity-based interventions, designing interventions that are compatible with the values of hegemonic masculinity risks attending to social, but not emotional loneliness. Educational efforts should be complemented by programs that foster the interpersonal skills, enhance emotional literacy, and prompt the psychological readiness needed to meaningfully engage with others. Separated men in particular may feel the need of gender balance in group activities [6], meaning men would benefit from learning how to approach and sustain friendship with all genders. Finally, policy makers should recognize emotional loneliness in men as a distinct target and offer tailored support that recognizes the unique ways in which men connect [1] through mental health services, relationship counseling, and peer support initiatives that address their deeper emotional needs as an integral part of the overarching strategy against loneliness.

7.2 Recommendations for clinicians, counselors, and social workers

For practitioners, it is essential to assess both social and emotional loneliness in their clients, recognizing that quantity of social interactions alone does not capture the full picture of loneliness. The interventions should focus not only on imparting knowledge but also on enhancing emotional skills, cultural sensitivity, and relational literacy, thus addressing the emotional and subjective aspects of loneliness more effectively. In line with previous research, practitioners should also adopt strategies that actively build and sustain trust, respect, and understanding from the outset, using clear, goal-focused structures, validating language while being sensitivity to each client’s unique gender socialization, and construction of masculinity [58]. Previous research has shown that appealing to specific archetypes like a “sage” or helping others may help men engaging in tasks related to socialization [6].

7.3 Strengths, limitations, and future research

Several overarching limitations span across the four studies. First, all data were cross-sectional, limiting our ability to infer causal relationships or directional pathways among knowledge, beliefs, behaviors, and loneliness. Longitudinal studies are essential to determine how these domains influence each other over time, especially in response to public health interventions. Second, reliance on self-reported measures—particularly for knowledge, values, perceived norms, and behavioral effort—raises concerns about social desirability and recall bias. While measures were pragmatic and face-valid, some constructs (e.g., effort, knowledge, adherence) were assessed using single-item or composite indices that may oversimplify complex psychological and behavioral processes. Sample sizes varied considerably, with Studies 1 and 2 drawing from small, demographically limited cohorts. We addressed this by using bootstrap methods that provide more robust estimates of standard errors and confidence intervals, allowing for more reliable inference despite the limited sample size. Although Study 4 used a much larger sample, the generalizability of all findings may still be constrained by non-probabilistic sampling and overrepresentation of White, Canadian men, and by the particular normativity around the COVID-19 pandemic. One important limitation related to sample size is that our analysis collapsed distinct life stages (ranging from 16 to 89 years) into a single analytical frame. Loneliness in later life may be shaped by bereavement, retirement, or health decline, whereas for younger men, it may intersect more with identity development, independence, and career transitions [59]. In addition, our analyses did not include explicit measures that would allow us to explicitly evaluate masculine norms neither did we examine women’s experiences as a comparator group, limiting the extent to which gendered patterns of loneliness could be assessed. Future research would benefit from age-targeted sampling or stratified analyses of bigger samples, explicitly attending to these developmental and social distinctions. Across all studies, we were limited in our capacity to assess the emotional texture, context, and quality of social connections—highlighting the need for future mixed-methods or qualitative inquiry. Despite these limitations, each study advances understanding in a meaningful way and builds sequentially on the others, presenting several key findings that contribute to a cohesive theoretical message about the role of social instead of individual and the need of working knowledge to improve social wellbeing. Experimental evidence will be particularly useful in helping us understand how individuals with different baseline levels of social and emotional loneliness respond to information, social training, therapy, and community-based interventions in the broader context of social connection as an important predictor of wellbeing [31, 32].