Social Connection and Behavior Change

---

1. Social Connection as a Health Variable

1.1 The Mortality Evidence

The case for treating social connection as a major health variable begins with mortality. Holt-Lunstad, Smith & Layton (2010) conducted a meta-analysis of 148 prospective studies (N = 308,849, average follow-up 7.5 years) and found that adequate social relationships were associated with a 50% increased likelihood of survival (OR = 1.5) compared to poor or insufficient social relationships.

The magnitude of this effect is frequently cited for context: it exceeds the mortality risk associated with obesity (OR ≈ 1.2), physical inactivity (OR ≈ 1.2), and excessive alcohol consumption, and is comparable to smoking 15 cigarettes per day. This is not a marginal effect. Social isolation is a major risk factor for premature mortality.

A subsequent meta-analysis by Holt-Lunstad et al. (2015) focused specifically on loneliness and social isolation: both predicted mortality independently (loneliness HR = 1.26; social isolation HR = 1.29; living alone HR = 1.32), and their effects were not explained by preexisting health conditions. The effect held across demographic groups, countries, and follow-up periods.

1.2 Mechanisms: Why Social Connection Affects Health

The mortality association reflects multiple mechanistic pathways operating in parallel:

Physiological stress buffering: social support reduces HPA axis reactivity to stressors. Cohen et al. (1997) demonstrated that people with more diverse social networks had significantly lower rates of clinical illness after experimental rhinovirus exposure and lower cortisol responses to acute stress. Perceived social support predicts more rapid HRV recovery after stressors (Oveis et al., 2009).

Behavioral pathways: social relationships influence health behaviors directly — through modeling, norm transmission, accountability, and direct encouragement or discouragement of specific behaviors (Umberson & Montez, 2010). This is the primary mechanism relevant to behavior change platforms.

Psychological pathways: loneliness increases threat vigilance and sympathetic nervous system activation (Cacioppo & Hawkley, 2009), producing chronic low-grade stress responses that accumulate over years. Social connection supports self-regulation capacity — a sense of belonging reduces ego threat and supports more flexible, long-term-oriented decision-making.

Neuroendocrine pathways: oxytocin, released in social bonding contexts, reduces cortisol reactivity, promotes trust, and modulates reward circuitry in ways that support prosocial behavior and reduce anxiety (Heinrichs et al., 2003). Chronic loneliness is associated with altered dopaminergic and serotonergic signaling.

1.3 Loneliness vs. Social Isolation

These are distinct constructs with distinct health implications:

• Social isolation: objective lack of social contacts and relationships
• Loneliness: subjective perception that social connection is inadequate relative to desired level

Both predict mortality independently (Holt-Lunstad et al., 2015). A person can be socially isolated but not lonely (hermits, monastics); conversely, highly socially active people can be profoundly lonely. For behavior change, both dimensions matter: objective social support resources (who can help you stick to this?) and subjective sense of belonging (do you feel part of a community practicing these behaviors?).

---

2. Social Influence Mechanisms in Behavior Change

2.1 Social Modeling: We Do What We See

Bandura's (1986) social cognitive theory established that much human learning and behavior regulation is observational — we learn by watching others, adjust our behavior to match in-group norms, and regulate effort based on social comparisons. In health behavior, social modeling operates at multiple levels:

Network contagion: Christakis & Fowler (2007, NEJM, N=12,067, follow-up 32 years) demonstrated that obesity spreads through social networks: if a close friend becomes obese, an individual's own risk increases by 57%; a sibling, by 40%; a spouse, by 37%. The effect extended to three degrees of separation (friend of a friend of a friend). The mechanism is norm transmission and modeling, not biological contagion.

The same group found analogous effects for smoking cessation (Christakis & Fowler, 2008, NEJM): smoking cessation spread in clusters, with close contacts of quitters having substantially higher cessation rates than matched controls. Geographic clusters of cessation emerged, with friends, siblings, spouses, and coworkers all showing linked cessation patterns. The network structure of quitting mattered as much as individual motivation.

Exercise contagion: Aral & Nicolaev (2017, Nature Communications, N=1.1 million) analyzed a running platform dataset and found significant social contagion in running behavior: users ran more on days their contacts ran more; the effect was asymmetric and stronger for same-gender pairs; less active individuals were influenced more by active contacts than vice versa.

Modeling mechanism: seeing others perform a behavior reduces the perceived difficulty and social cost of the behavior, provides information about how to perform it, and normalizes it as "something people like me do."

2.2 Social Norms

Social norms operate through two pathways (Cialdini et al., 1990):

Descriptive norms: what most people actually do ("Most people in your neighborhood recycle"). These provide information about what is normal and cue conformity.

Injunctive norms: what most people approve of ("Most people think recycling is important"). These provide social approval/disapproval pressure.

Both predict behavior, but their interaction matters. A descriptive norm pointing in the wrong direction (most people in your gym skip stretching) can undermine behavior. Effective norm messaging communicates both that a healthy behavior is common and that it is approved of.

The Opower effect: Schultz et al. (2007) randomized energy-use feedback to include or exclude a social comparison component. Households receiving information about their energy use relative to efficient neighbors reduced consumption. The effect size was modest (d ≈ 0.2) but highly replicable — Opower deployed it at scale with consistent results across 60+ utilities. Importantly, households already below the norm increased consumption (a boomerang effect) — mitigated by adding an injunctive component (a smiley face for efficient households). The intervention is now the largest behavioral energy conservation program in history.

Health behavior norms: perceived norms about exercise, diet, and sleep among one's social reference group predict individual behavior. College students overestimate peer drinking frequency and underestimate peer sleep duration — correcting these misperceptions via social norm feedback reduces risky behavior (Perkins et al., 2005 on drinking; Murnane et al., 2015 on sleep).

2.3 Social Support: Types and Evidence

Cohen & Wills (1985) distinguished support types:

• Emotional support: expressions of care, empathy, understanding
• Informational support: advice, guidance, feedback
• Instrumental support: practical help (driving to the gym, cooking healthy meals)
• Appraisal support: feedback useful for self-evaluation (telling you your form is improving)

The stress-buffering hypothesis (Cohen & Wills, 1985) proposes that social support reduces the pathogenic effects of stress on health, rather than having a direct main effect. Meta-analyses support the buffering model: support effects on health outcomes are strongest under high stress conditions (Uchino et al., 1996).

For behavior change specifically, instrumental support (practical help with logistics) and informational support (knowing how to do the behavior correctly) are more predictive of behavior initiation; emotional support is more predictive of maintenance (Thoits, 2011).

2.4 Accountability and Commitment Devices

Social accountability — the public commitment to a behavioral goal, combined with a socially aware tracking mechanism — is a specific application of social influence with its own evidence base.

Public commitment: Cialdini's (2007) consistency principle: public commitment increases follow-through because people are motivated to appear consistent with their stated identities and commitments. Deutsch & Gerard (1955) demonstrated that public, written commitments produce more durable behavior than private or unwritten ones.

Commitment contracts: Volpp et al. (2009, NEJM) randomized 878 employees to financial incentives (up to $750 for 6-month cessation) vs. information-only. The incentive group had substantially higher 9-month cessation rates (14.7% vs. 5.0%). Social accountability mechanisms that include commitment contracts or wagers amplify effects further. stickK.com and Beeminder operationalize commitment contracts at scale.

Accountability partnerships: Wing & Jeffery (1999, International Journal of Obesity): weight loss participants recruited with friends lost significantly more weight than those recruited alone (5.7 vs. 4.4 kg at 16 weeks) and showed substantially higher 16-week maintenance (66% vs. 24% of initial loss maintained). The accountability structure of sharing the goal with a known friend was the operative mechanism.

---

3. Group-Based Behavior Change

3.1 Why Groups Work

Group-based behavior change programs consistently outperform individual programs across domains including weight loss (Paul-Ebhohimhen & Avenell, 2009), smoking cessation (Stead & Lancaster, 2017), exercise adherence (Burke et al., 2006), and alcohol reduction. Effect advantages of groups over individual approaches are typically d = 0.2–0.4.

Mechanisms:

• Shared identity: membership in a group creates in-group identification that supports norm adoption
• Social facilitation: presence of others during exercise increases effort (Triplett, 1898 — the oldest finding in social psychology)
• Accountability: others know your goals and whether you achieved them
• Social comparison: upward comparison ("others are doing better, I can too") is motivating when perceived as achievable; downward comparison ("I'm doing better than others") reduces motivation
• Information sharing: peers who have succeeded provide practical knowledge not available in expert sources

3.2 Group Size and Composition

Group effectiveness depends on composition:

• Homophily: groups of people with similar starting points (fitness level, body weight, lifestyle) show better outcomes than heterogeneous groups — perceived similarity increases modeling credibility
• Optimal size: 5–12 members balances intimacy (individual accountability, close relationships) with diversity (varied experience, reduced dependence on any one member). Very large groups (>20) reduce individual accountability and increase social loafing
• Group climate: cohesion — the degree to which members feel bonded and mutually committed — predicts adherence better than group size or meeting frequency (Carron et al., 1996)

3.3 Competition vs. Cooperation

Head-to-head competition in health behavior apps (step competitions, weight loss contests) has mixed evidence. Patel et al. (2018, JAMA Internal Medicine, N=304): four conditions for step increase — individual monitoring, collaborative team, competitive team, combination. Competitive conditions outperformed collaborative and individual conditions at 26 weeks (step increase 2x vs. collaborative). However, competition effects are highly individual — people who lose early in a competition may disengage, producing worse outcomes for the bottom half.

Cooperative framing (shared team goal rather than individual ranking) generally shows more equitable outcomes across participants and lower dropout among lower-performing members.

---

4. Online Social Support: Evidence and Limits

4.1 What Online Communities Can Do

Online health communities have grown substantially as a behavior change medium. Evidence across multiple domains:

Exercise: Cavallo et al. (2012) RCT: Facebook group support for physical activity produced significant increases in self-reported activity vs. control at 12 weeks. Aral & Nicolaev (2017) running data: social influence effects on exercise were large (up to 0.3 additional miles/run per additional mile by contacts).

Weight loss: Pinto et al. (2013) found online peer support groups produced equivalent weight loss to in-person support groups at 12 months — an important finding that legitimizes digital community as a delivery mechanism, not just a lesser substitute.

Smoking cessation: Cobb et al. (2011): participants in a smoking cessation Facebook group had significantly higher 7-day point-prevalence abstinence at 60 days vs. a website-only control (14.4% vs. 5.0%).

Mental health: Firth et al. (2017) meta-analysis of online peer support for depression: significant but small effects (d = 0.22), with higher-quality studies showing smaller effects — consistent with a pattern of modest but real benefit.

4.2 Limits and Risks

Attenuation vs. in-person: online support consistently shows smaller effect sizes than in-person equivalents across domains. The dose matters — asynchronous text posts are less potent than synchronous video or face-to-face interaction.

Social comparison anxiety: fitness apps with visible leaderboards or social feeds can increase anxiety and reduce motivation for users with lower performance. This is especially documented in diet and weight-loss apps, where social comparison to idealized bodies produces negative body image (Fardouly et al., 2015). The same features that motivate competitive users demotivate lower-performing or self-conscious users.

Performative health: public tracking can shift motivation from intrinsic (doing this for myself) to extrinsic (doing this because others are watching). Per self-determination theory (SP-11), this shift predicts reduced long-term adherence. Platforms that create pressure to log and share can undermine the autonomous motivation that sustains behavior change.

Echo chamber norms: online health communities can reinforce dysfunctional norms — extreme dietary restriction, overtraining, supplement overuse — when communities self-select on extreme behaviors.

Social loafing: in collaborative group challenges, individual effort decreases as group size grows — members reduce effort when individual contributions are not identifiable (Latané et al., 1979). Design must maintain individual identifiability within groups.

---

5. Individual Variation in Social Responsiveness

5.1 Who Benefits Most from Social Behavior Change Interventions

Social responsiveness to behavior change interventions varies substantially across individuals. Predictors of high social responsiveness:

Attachment style: securely attached individuals use social support more effectively and show larger behavior change responses to social accountability mechanisms than insecurely attached individuals (Pietromonaco & Collins, 2017).

Existing network quality: individuals with strong existing social networks gain less incremental benefit from platform-delivered social support; socially isolated individuals show the largest gains (Thoits, 2011).

Competitiveness: individual differences in competitiveness predict whether competition-based features increase or decrease motivation. Competitive individuals are more responsive to leaderboards and challenges; non-competitive individuals may disengage.

Social anxiety: individuals high in social anxiety often experience social comparison features as threatening rather than motivating; private tracking produces equivalent or better outcomes for this group.

5.2 The Opt-In Principle

Given large individual variation, the evidence supports designing social features as opt-in rather than opt-out or mandatory. Users who self-select into social accountability mechanisms may show larger effects than those passively enrolled, and those who opt out avoid the harms (comparison anxiety, performative motivation, shame from lapses).

A/B testing of social vs. private conditions in real health apps (Lim et al., 2020; Franklin et al., 2019) consistently shows that social features improve outcomes for users who choose them and are neutral-to-harmful for users who do not.

---

6. Social Connection as a Behavior Change Domain

6.1 Social Connection as a Target, Not Just a Tool

Most behavior change platforms treat social features as mechanisms for improving adherence to other behaviors (exercise, diet, sleep). But social connection is itself a health behavior — its cultivation and maintenance is something a personal science platform can directly support.

Evidence-based targets for improving social connection:

• Frequency of meaningful interaction: not total contact hours, but quality exchanges (Sandstrom & Dunn, 2014: even brief conversations with strangers improve well-being)
• Social diversity: maintaining a diverse social network (colleagues, friends, family, community) predicts better health outcomes than relying on a single relationship (Cohen et al., 1997)
• Acts of giving: prosocial behavior (helping others, volunteering) produces larger well-being effects than receiving help — and is more controllable (Layous et al., 2012)
• Quality over quantity: the subjective quality of relationships matters more for health outcomes than relationship count (Hawkley & Cacioppo, 2010)

6.2 N=1 Experiments in Social Connection

Social connection is underused as a personal science domain because it is harder to measure objectively than sleep or HRV. But tractable experiments exist:

• Conversation depth experiment: compare daily well-being ratings after days with primarily surface-level conversations vs. days with at least one substantive personal conversation; Mehl et al. (2010, Psychological Science): people who reported more substantive conversations had higher well-being independent of extraversion
• Gratitude expression: send one specific, personal thank-you message per day for 2 weeks; measure subjective well-being and connection ratings (Algoe et al., 2013)
• Alone vs. with others for exercise: compare performance, enjoyment, and next-day motivation when exercising alone vs. with a partner for 4-week blocks
• Digital social media reduction: reduce passive social media consumption (scrolling feeds) while maintaining active social contact; measure loneliness and subjective well-being
• Community participation: join a group activity (class, club, team) for 8 weeks; measure social belonging and general well-being

---

7. Platform Design Principles

Make social features opt-in by default. The evidence on individual variation and social comparison harms is clear: forced social features benefit competitive, non-anxious users and harm socially anxious or less-performant users. Opt-in preserves benefits while limiting harms.

Distinguish accountability from comparison. Accountability (did you do what you said you'd do?) is reliably motivating across individuals. Social comparison (how do you rank against others?) is motivating for some and demotivating for others. Design these as separate features with separate opt-in paths.

Support dyadic accountability partnerships. The Wing & Jeffery evidence on friend-recruited pairs, Volpp commitment contracts, and the simplicity of one committed partner suggest that 1:1 accountability relationships are the highest-leverage social feature. They scale better than groups, avoid social loafing, and avoid comparison harms.

Correct misperceived norms. Users systematically misestimate what their peers are doing. Surface accurate social norms: "Among users tracking sleep in your age range, the median sleep duration is 7h 12m" is more useful than a leaderboard. Add injunctive reinforcement for users who are already at or above norms (a simple positive acknowledgment, not a smiley face).

Track social connection as a health metric. Build daily or weekly prompts for quality of social interaction alongside sleep, exercise, and mood. Make the relationship between social connection ratings and other health outcomes visible in the user's own data. This positions social connection as a behavioral target, not just a platform feature.

Design for recovery after social lapse. Relationship ruptures and periods of isolation are common and predictable. Platforms should prompt reconnection without shame — "It's been a while since you checked in with your accountability partner" is better than silence or a streak reset.

Protect against performative tracking. Allow users to track privately by default. Make sharing an explicit, friction-added choice rather than the default. Users who share should understand they are choosing to do so and why — ideally tying the share to a specific accountability function (weekly check-in with partner) rather than ambient audience broadcasting.

---

8. Individual Variation in Social Connection Effects

The wellbeing effects of social connection are not uniform across individuals. Several well-documented individual differences determine the form, dose, and type of social interaction that is beneficial versus costly for a given person.

Introversion and extraversion constitute the clearest moderator in the literature. Extraverts derive energy from social interaction and show larger wellbeing gains from increased social contact, as well as larger wellbeing costs from isolation. Introverts show non-linear dose-response effects: moderate levels of social connection are beneficial and necessary for wellbeing, but high-frequency or high-intensity social interaction produces fatigue and wellbeing decrements (Helgoe, 2013). The optimal social dose for introverts is meaningfully lower than for extraverts — attempting to match an extravert's social frequency is not a path to higher wellbeing for introverts and may produce the opposite. This means "more social connection" is not a universally valid prescription; the right dose depends on the individual's trait profile.

Attachment security predicts both the average level and the variability of wellbeing effects from social interaction. Securely attached individuals show consistent wellbeing benefits from close social contact across a range of interaction quality. Anxiously attached individuals show high variability: positive social interactions boost their wellbeing more than average, but negative interactions, perceived rejection, or social uncertainty hurt more. Avoidant individuals show blunted social effects in both directions — less wellbeing gain from close contact and less loss from conflict — because their regulatory strategy involves down-regulating social information. For anxiously attached individuals in particular, the quality of individual interactions matters more than total social frequency, and high-risk social environments (competitive apps, comparison-heavy communities) can be actively harmful.

Social comparison orientation is a stable trait that moderates the wellbeing effects of social features in apps and communities. Gibbons and Buunk (1999) identified social comparison orientation (SCO) as an individual difference with approximately 30% of adults scoring high. For high-SCO individuals, any social connection feature that includes performance information — leaderboards, public streak counts, visible progress comparisons — reliably reduces wellbeing through upward comparison processes. Pure social affiliation without status or performance information is beneficial for all groups; comparison-embedded social features are beneficial only for low-SCO individuals and harmful for high-SCO individuals.

Network size versus quality tradeoff differs systematically by extraversion level. For high extraversion scorers, network breadth (number of social contacts, diversity of social contexts) predicts wellbeing; they benefit from a wide social network even if individual relationships are less deep. For low extraversion scorers, relationship depth and closeness quality predict wellbeing more strongly than network breadth. The commonly used metric "number of close friends" is a poor universal predictor of wellbeing because it conflates these two dimensions and systematically misidentifies the optimum for both profiles.

Practical implications for self-experimentation: Identify your introversion/extraversion baseline using any validated measure (Big Five extraversion facet, or simply tracking energy level after social interactions over two weeks — energy gain vs. depletion is the empirical signal). Track energy level after social interactions systematically to find your personal dose-response curve before attempting to increase social frequency. For individuals with high social comparison orientation, disable competitive features in apps, avoid passive social media feed consumption during low-wellbeing periods, and design social accountability around direct dyadic relationships rather than group comparison formats. The social connection target is the quantity and type that is optimal for your profile, not the quantity that is optimal for the median extravert.

The introvert/extravert taxonomy, like the glycemic index, is a population-derived description being applied as an individual prescription. Trait extraversion is real, heritable, and predictive at the group level — but the optimal social configuration for any specific individual cannot be read off a trait score. It must be measured. Someone who scores low on extraversion may find that small intimate dinners are consistently energizing while large social events are depleting — yet may have been avoiding both under an introvert label that lacked the resolution to distinguish between them. The label is doing work it was never designed to do. The self-report "I am an introvert" is a prior, not a finding. It reflects accumulated experience and received identity, both of which can be wrong in systematic ways. The energy-tracking data from the experiment below is the finding. If the data confirms the label, the label was a useful prior. If the data reveals a more granular pattern — energized by 1:1 contact, drained by groups, irrespective of the labeled identity — update the identity to match the data.

---

N=1 Experiment Protocols

These protocols are designed for individual self-experimentation. Each uses a within-person design to generate personalized evidence that population averages cannot provide.

Social dose-response experiment (3 weeks). Week 1: ≤2 social interactions per day; Week 2: 3–5 social interactions; Week 3: >5 social interactions. Log energy level after each interaction (1–10) and end-of-day wellbeing (1–10). Decision: identify the weekly pattern that maximizes end-of-day wellbeing without producing interaction fatigue — this is your personal social dose.

Interaction quality vs. quantity crossover (4 weeks). Weeks 1–2: prioritize volume (more frequent but shorter interactions); Weeks 3–4: prioritize depth (fewer but longer, more substantive interactions). Same total social time budget. Measure: weekly loneliness rating (UCLA Loneliness Scale, 3 items) and connection satisfaction (1–10). Decision: whichever condition produces lower loneliness scores = your optimal social structure.

Digital vs. in-person comparison (2 weeks). Log all social interactions as digital (text, call) or in-person. Rate each for "felt connected" (1–5). Compute average by type after 14 days. Use the difference to calibrate your digital/in-person ratio.

Social label validation experiment (6 weeks). Before starting: write down your current social identity label — introvert, extrovert, social butterfly, homebody, or any descriptor you use to guide your social choices. This is your prior. For 6 weeks, rate your energy level (1–10) within 30 minutes after every distinct social interaction, logged by interaction type: large group (5+ people), small group (2–4 people), 1:1, digital-only (text/call), stranger, close friend, acquaintance. At the end of week 6, compute your average post-interaction energy by type. Decision criterion: if the pattern that emerges matches your label — introverts drained across the board, extraverts energized — the label was a useful prior and can be retained. If the pattern is more granular than the label predicted (energized by 1:1s and small groups, drained by large events, regardless of stated introversion), update the label to match the data. The point is not to discard categories but to replace a population-derived identity with a personally measured one. The data determines the prescription; the label does not.

---

9. Conclusion

Social connection is health infrastructure, not merely a feel-good auxiliary to the more measurable behavioral variables. The mortality evidence (OR ≈ 1.5 for adequate vs. inadequate social relationships) is among the largest effect sizes in population health research, and the mechanisms — stress buffering, behavioral modeling, norm transmission, accountability — are directly actionable through platform design.

The central tension in building social features for a health platform is between social influence's power and its double-edged quality. The same features that motivate some users — social comparison, visibility, competitive challenges — demotivate, shame, or harm others. The solution is not to avoid social features but to build them as opt-in, purpose-specific, and individually calibrated. A platform that offers 1:1 accountability partnerships as its primary social structure, with comparison and community features as deliberate opt-in additions, captures most of the evidence-based benefit while minimizing the documented harms.

The underutilized opportunity is treating social connection itself as a behavioral target, not just a platform architecture. A personal science approach to social connection — tracking quality of social interactions, running controlled experiments on conversation depth or community participation, building evidence-based practices around relationship quality — extends the platform's scope from individual biology to the relational context in which behavior change actually happens.

---

References

Algoe, S. B., Fredrickson, B. L., & Gable, S. L. (2013). The social functions of the emotion of gratitude via expression. Emotion, 13(4), 605–609.

Aral, S., & Nicolaev, C. (2017). Exercise contagion in a global social network. Nature Communications, 8, 14753.

Bandura, A. (1986). Social Foundations of Thought and Action: A Social Cognitive Theory. Prentice-Hall.

Burke, S. M., Carron, A. V., Eys, M. A., Ntoumanis, N., & Estabrooks, P. A. (2006). Group versus individual approach? A meta-analysis of the effectiveness of interventions to promote physical activity. Sport & Exercise Psychology Review, 2(1), 19–35.

Cacioppo, J. T., & Hawkley, L. C. (2009). Perceived social isolation and cognition. Trends in Cognitive Sciences, 13(10), 447–454.

Carron, A. V., Brawley, L. R., & Widmeyer, W. N. (1996). The measurement of cohesiveness in sport groups. In J. L. Duda (Ed.), Advances in Sport and Exercise Psychology Measurement (pp. 213–226). Fitness Information Technology.

Cavallo, D. N., Tate, D. F., Ries, A. V., Brown, J. D., DeVellis, R. F., & Ammerman, A. S. (2012). A social media–based physical activity intervention. American Journal of Preventive Medicine, 43(5), 527–532.

Christakis, N. A., & Fowler, J. H. (2007). The spread of obesity in a large social network over 32 years. New England Journal of Medicine, 357(4), 370–379.

Christakis, N. A., & Fowler, J. H. (2008). The collective dynamics of smoking in a large social network. New England Journal of Medicine, 358(21), 2249–2258.

Cialdini, R. B. (2007). Influence: The Psychology of Persuasion (rev. ed.). Harper Business.

Cialdini, R. B., Reno, R. R., & Kallgren, C. A. (1990). A focus theory of normative conduct: Recycling the concept of norms to reduce littering in public places. Journal of Personality and Social Psychology, 58(6), 1015–1026.

Cobb, N. K., Graham, A. L., & Abrams, D. B. (2011). Social network structure of a large online community for smoking cessation. American Journal of Public Health, 100(7), 1282–1289.

Cohen, S., Doyle, W. J., Skoner, D. P., Rabin, B. S., & Gwaltney, J. M. (1997). Social ties and susceptibility to the common cold. JAMA, 277(24), 1940–1944.

Cohen, S., & Wills, T. A. (1985). Stress, social support, and the buffering hypothesis. Psychological Bulletin, 98(2), 310–357.

Deutsch, M., & Gerard, H. B. (1955). A study of normative and informational social influences upon individual judgment. Journal of Abnormal and Social Psychology, 51(3), 629–636.

Fardouly, J., Diedrichs, P. C., Vartanian, L. R., & Halliwell, E. (2015). Social comparisons on social media: The impact of Facebook on young women's body image concerns and mood. Body Image, 13, 38–45.

Firth, J., Torous, J., Nicholas, J., Carney, R., Pratap, A., Rosenbaum, S., & Sarris, J. (2017). The efficacy of smartphone-based mental health interventions for depressive symptoms. World Psychiatry, 16(3), 287–298.

Hawkley, L. C., & Cacioppo, J. T. (2010). Loneliness matters: A theoretical and empirical review of consequences and mechanisms. Annals of Behavioral Medicine, 40(2), 218–227.

Heinrichs, M., Baumgartner, T., Kirschbaum, C., & Ehlert, U. (2003). Social support and oxytocin interact to suppress cortisol and subjective responses to psychosocial stress. Biological Psychiatry, 54(12), 1389–1398.

Holt-Lunstad, J., Smith, T. B., & Layton, J. B. (2010). Social relationships and mortality risk: A meta-analytic review. PLOS Medicine, 7(7), e1000316.

Holt-Lunstad, J., Smith, T. B., Baker, M., Harris, T., & Stephenson, D. (2015). Loneliness and social isolation as risk factors for mortality: A meta-analytic review. Perspectives on Psychological Science, 10(2), 227–237.

Latané, B., Williams, K., & Harkins, S. (1979). Many hands make light the work: The causes and consequences of social loafing. Journal of Personality and Social Psychology, 37(6), 822–832.

Layous, K., Chancellor, J., Lyubomirsky, S., Wang, L., & Doraiswamy, P. M. (2012). Delivering happiness: Translating positive psychology intervention research for treating major and minor depressive disorders. Journal of Alternative and Complementary Medicine, 17(8), 675–683.

Mehl, M. R., Vazire, S., Holleran, S. E., & Clark, C. S. (2010). Eavesdropping on happiness: Well-being is related to having less small talk and more substantive conversations. Psychological Science, 21(4), 539–541.

Oveis, C., Cohen, A. B., Gruber, J., Shiota, M. N., Haidt, J., & Keltner, D. (2009). Resting respiratory sinus arrhythmia is associated with tonic positive emotionality. Emotion, 9(2), 265–270.

Patel, M. S., Small, D. S., Harrison, J. D., Fortunato, M. P., Oon, A. L., Rareshide, C. A. L., ... & Volpp, K. G. (2018). Gamification of health behaviors using social incentives and loss aversion. JAMA Internal Medicine, 178(12), 1663–1670.

Paul-Ebhohimhen, V., & Avenell, A. (2009). A systematic review of the effectiveness of group versus individual treatments for adult obesity. Obesity Facts, 2(1), 17–24.

Perkins, H. W., Haines, M. P., & Rice, R. (2005). Misperceiving the college drinking norm and related problems. Journal of Studies on Alcohol, 66(4), 470–478.

Pietromonaco, P. R., & Collins, N. L. (2017). Interpersonal mechanisms linking close relationships to health. American Psychologist, 72(6), 531–542.

Pinto, A. M., Fava, J. L., Raynor, H. A., & Wing, R. R. (2013). Development and preliminary evaluation of an internet-based healthy living programme. Journal of Obesity, 2013, 931569.

Sandstrom, G. M., & Dunn, E. W. (2014). Social interactions and well-being: The surprising power of weak ties. Personality and Social Psychology Bulletin, 40(7), 910–922.

Schultz, P. W., Nolan, J. M., Cialdini, R. B., Goldstein, N. J., & Griskevicius, V. (2007). The constructive, destructive, and reconstructive power of social norms. Psychological Science, 18(5), 429–434.

Stead, L. F., & Lancaster, T. (2017). Group behaviour therapy programmes for smoking cessation. Cochrane Database of Systematic Reviews, CD001007.

Thoits, P. A. (2011). Mechanisms linking social ties and support to physical and mental health. Journal of Health and Social Behavior, 52(2), 145–161.

Uchino, B. N., Cacioppo, J. T., & Kiecolt-Glaser, J. K. (1996). The relationship between social support and physiological processes: A review with emphasis on underlying mechanisms and implications for health. Psychological Bulletin, 119(3), 488–531.

Umberson, D., & Montez, J. K. (2010). Social relationships and health: A flashpoint for health policy. Journal of Health and Social Behavior, 51(S), S54–S66.

Volpp, K. G., Troxel, A. B., Pauly, M. V., Glick, H. A., Puig, A., Asch, D. A., ... & Audrain-McGovern, J. (2009). A randomized, controlled trial of financial incentives for smoking cessation. New England Journal of Medicine, 360(7), 699–709.

Wing, R. R., & Jeffery, R. W. (1999). Benefits of recruiting participants with friends and increasing social support for weight loss and maintenance. Journal of Consulting and Clinical Psychology, 67(1), 132–138.