Belonging, Psychological Safety, and the Manager-Employee Relationship

Connection is the third of Frank Newman’s “5 Cs” of people management, and this page is the science and stakes vein — the evidence behind belonging, psychological safety, trust, and the manager’s outsized role in engagement, and the business case for getting it right. It reviews what research actually says, in a neutral, sourced register. For how a present manager builds connection day to day, see One-on-Ones, Check-ins, and Manager Presence; for the science of care behind it, see Compassion at Work; and for the wider cost of disconnection, see Workplace Mental Health and Burnout — Costs and Consequences.

One theme runs through the evidence below: the direction is far better established than the magnitude. The need to belong is verified science; psychological safety, trust, and engagement are reliably associated with better outcomes, but the relationships are largely correlational and the exact effect sizes vary by setting. The most-quoted numbers of all — “managers drive 70% of engagement,” “disengagement costs $8.8 trillion,” “loneliness costs $154 billion” — are proprietary vendor estimates, not verified constants. The confidence behind each claim therefore varies, so each section carries its own confidence label.

The human need to belong, and how it shows up at work

The need to belong — to form and keep “strong, stable interpersonal relationships” — is one of the best-supported ideas in social psychology, and a present, connected manager meets a real, evidenced human need. In a landmark 1995 review, Roy Baumeister and Mark Leary argued that humans have a fundamental need for frequent, non-aversive interactions inside an ongoing relationship they expect to continue, and that going without it is linked to measurable harm to health, adjustment, and well-being. That core thesis is verified: the paper is among the most-cited in the field, and belonging is treated as a basic motivation alongside needs like safety.

Where the evidence thins is the leap from “people need to belong” to “belonging at work produces these exact business results.” The widely circulated figures — for example a vendor claim that high belonging brings a 56% jump in performance, a 50% drop in turnover risk, and a 75% cut in sick days — come from a consultancy report (BetterUp, 2019) based on a US survey, not from peer-reviewed causal research. They should be read as vendor estimates: US-based, and directional at best.

Canadian data is preferable for a Canadian audience, and it exists — though it is descriptive, not proof of a belonging-to-profit chain:

• Statistics Canada (2023) reports average job-satisfaction scores by occupation ranging from about 6.9 to 8.9 out of 10; satisfaction fell to 6.9 for employees in households finding it very difficult to make ends meet, versus 8.1 for those finding it very easy.
• A 2021 Canadian employer survey (LifeWorks / Morneau Shepell), drawing on roughly 3,000 Canadian employees, found that 73% of respondents felt a sense of belonging at work before the pandemic, but only 65% did during the survey month — and on-site workers reported more belonging than remote or hybrid ones. (This is itself a vendor survey, not peer-reviewed.)

For a Kitchener-Waterloo small business, the takeaway is well grounded: the need that a present, connected manager meets is real and evidenced. The need is solid science; the dollar-figure “ROI of belonging” numbers are vendor-grade and should be hedged.

Source: Baumeister & Leary, The Need to Belong: Desire for Interpersonal Attachments as a Fundamental Human Motivation, Psychological Bulletin 117(3), pp. 497–529 (1995); BetterUp, The Value of Belonging at Work (report, 2019) (vendor); Statistics Canada, Are Canadians satisfied with their jobs? (Quality of Life Hub, 2023); LifeWorks (Morneau Shepell), Mental Health Index (2021) (vendor).

Confidence: verified for the core need to belong; the workplace “ROI of belonging” figures are directional vendor estimates, and the StatCan and LifeWorks data are descriptive.

What psychological safety is, exactly

Psychological safety is a shared belief, held across a team, that the group is safe for interpersonal risk-taking — that you can speak up, ask a question, admit a mistake, or challenge an idea without being humiliated or punished for it. It is a property of the group’s climate, not a personality trait, and it is distinct from trust and from simply being nice. Amy Edmondson introduced the term in a 1999 study of 51 work teams and defined it as “a shared belief held by members of a team that the team is safe for interpersonal risk taking.”

The key word is shared. Psychological safety is a property of the group’s climate — the way things are done here — not a personality trait and not someone’s mood on a given day. In Edmondson’s original study, the higher-performing teams reported more errors, not because they made more, but because they felt safe enough to surface and discuss mistakes instead of hiding them.

Two distinctions matter, because both are easy to get wrong:

• It is not the same as trust. Trust is about whether you can rely on a specific other person. Psychological safety is about the climate of the team as a whole. You can distrust one colleague and still work on a psychologically safe team. (How trust itself is built is a separate question — see the trust section below.)
• It is not “being nice,” lowered standards, or the absence of accountability. In The Fearless Organization (2018), Edmondson pairs safety with accountability: high accountability without safety produces anxiety; high safety without accountability produces a comfort zone; only high safety and high standards produce a learning team. She names the common failure mode as being polite in the meeting and candid only afterward in the hallway — politeness that buries the very signals safety is meant to surface.

The construct itself is well established: it has a validated survey measure, decades of replication, and broad agreement on what it means. Whether, and how much, it improves performance is a separate and more cautious question — see the next section.

In a small Kitchener-Waterloo business, the “team” is often the whole company, or the handful of people reporting to one owner-manager. That makes the climate almost entirely a product of how that one person reacts to bad news. The concept transfers cleanly to a small workplace; only the size of the performance payoff needs care.

Source: Edmondson, Psychological Safety and Learning Behavior in Work Teams, Administrative Science Quarterly 44(2), pp. 350–383 (1999); Edmondson, The Fearless Organization (Wiley, 2018).

Confidence: verified.

Does psychological safety improve team performance, and by how much?

Yes, with care about how much, and why. Psychological safety is reliably and positively associated with the behaviours that drive performance; it stands on weaker ground when the claim becomes that it causes a specific performance lift.

The strongest evidence is a 2017 meta-analysis by Frazier and colleagues, pooling 136 independent samples (roughly 22,000 individuals and 5,000 groups). Corrected correlations were moderate to strong: about .43 for task performance, .45 for work engagement, .52 for information sharing, .62 for learning behaviour, .53 for job satisfaction, and .48 for organizational commitment — and weaker, .13, for creativity. Psychological safety also predicted these outcomes over and above related factors like good leader relationships and engagement.

Two honest caveats keep this from being oversold:

• Direction of cause is not established. These are almost entirely cross-sectional correlations. “Associated with” is solid; “causes, by X%” is not. High-performing teams may generate safety as much as safety generates performance.
• More is not always better. A 2023 study by Eldor, Hodor and Cappelli found that routine task performance rises with psychological safety only up to a point, then declines when there is no matching accountability — empirically echoing Edmondson’s point that safety has to be paired with high standards.

The most-quoted claim of all — Google’s Project Aristotle naming psychological safety the number-one driver of team effectiveness — should be treated as corroborating colour, not proof. It is a single company’s internal study of 180+ teams, not peer-reviewed, with proprietary methods.

For an Ontario employer, one more wrinkle: the effect sizes vary by national culture (Frazier and colleagues found the task-performance link substantially stronger in some cultures than others), so US and global averages will not transfer to a Waterloo Region workforce with precision. Treat the direction as robust and the exact number as indicative. The accountability half of the picture — useful, honest feedback — is covered in the feedback-and-performance research.

Source: Frazier, Fainshmidt, Klinger, Pezeshkan & Vracheva, Psychological Safety: A Meta-Analytic Review and Extension, Personnel Psychology 70(1), pp. 113–165 (2017); Eldor, Hodor & Cappelli, The limits of psychological safety: Nonlinear relationships with performance, Organizational Behavior and Human Decision Processes 177:104255 (2023); Google re:Work, Understand team effectiveness (Project Aristotle) (2015) (internal, non-peer-reviewed).

Confidence: industry-consensus for the association; the evidence is largely correlational, the relationship can reverse without accountability, and the Google result is a single non-peer-reviewed internal study.

How is manager-employee trust built?

Trust between a manager and an employee is built less through grand gestures than through three steadily-read signals: ability, benevolence, and integrity, filtered through the truster’s own propensity to trust. The dominant research model — Mayer, Davis and Schoorman (1995) — defines trust as a willingness to be vulnerable to someone else’s actions, and says we judge a person’s trustworthiness on three things:

• Ability — do they have the competence and skill relevant to the situation?
• Benevolence — do they genuinely care about my interests, not just their own?
• Integrity — do they hold to principles I find acceptable, consistently?

These judgments are filtered through the truster’s own propensity to trust — a stable disposition that matters most when information about the other person is still ambiguous, and matters less once their trustworthiness is clear. A useful practical wrinkle from the original work: integrity tends to dominate early in a relationship, while benevolence builds over time. This ABI model is verified — it is the standard framework in organizational-trust research, widely measured and replicated.

One popular idea deserves a warning label. The claim that “oxytocin is the trust molecule” traces to a 2005 Nature experiment (Kosfeld and colleagues) and was carried to business audiences as brain-chemistry proof of trust. It has not held up. A 2015 review concluded the finding “has not replicated well,” and a high-powered 2020 registered replication (about 677 participants, co-authored by one of the original authors) found no effect of oxytocin on trusting behaviour in the closest-matched condition. Treat the “trust molecule” story as contested, not established.

For a Waterloo Region owner-manager, the defensible model is also the practical one: show competence, show genuine care for the employee’s interests, and be consistent and principled. It needs no neuroscience, and it travels across cultures. Do not cite the oxytocin claim as fact.

Source: Mayer, Davis & Schoorman, An Integrative Model of Organizational Trust, Academy of Management Review 20(3), pp. 709–734 (1995); Kosfeld, Heinrichs, Zak, Fischbacher & Fehr, Oxytocin increases trust in humans, Nature 435, pp. 673–676 (2005); Nave, Camerer & McCullough, Does Oxytocin Increase Trust in Humans? A Critical Review, Perspectives on Psychological Science 10(6) (2015); Declerck, Boone, Pauwels, Vogt & Fehr, A registered replication study on oxytocin and trust, Nature Human Behaviour 4(6), pp. 646–655 (2020).

Confidence: verified for the ABI model; the “oxytocin is the trust molecule” claim is contested and failed a high-powered 2020 replication.

Why do people say the manager drives roughly 70% of engagement?

The widely-quoted “70%” is Gallup’s estimate that managers account for at least 70% of the variance between teams in employee-engagement scores — a proprietary, single-source vendor figure, not an independently audited constant. The figure comes from Gallup’s State of the American Manager (2015), where Randall Beck and Jim Harter wrote that “managers account for at least 70% of the variance in employee engagement scores across business units.” Read it carefully: it is a statement about why some teams inside the same company are more engaged than others — it is not a claim that a manager controls 70% of any one person’s engagement. Gallup presents the figure as stable (one vendor source puts re-analyses between 67% and 72%, though that stability claim is itself uncorroborated), and its broader Q12 engagement program (a 2020 meta-analysis spanning 112,312 work units) links engagement to outcomes like productivity, turnover, and safety.

The caveat is about sourcing. This is Gallup-proprietary, vendor-produced research. The Q12 instrument and the underlying client data are not openly available; the “70%” claim has not been independently replicated on Gallup’s data by outside researchers; and the method is not fully auditable from outside the firm. The work is methodologically serious — very large samples, corrections for measurement error — but it is in-house. So “70%” is best treated as a single-source vendor estimate that is plausible, and points the same direction as the independent academic literature on leadership and team climate, rather than as a verified scientific constant.

For a Kitchener-Waterloo small business, the spirit of the finding is well supported even where the precise number is not independently confirmed: the direct manager is the highest-leverage point for a team’s engagement. When you quote “70%,” attribute it to Gallup and flag that it rests on a single source.

Source: Gallup (Beck & Harter), State of the American Manager / Managers Account for 70% of Variance in Employee Engagement (2015); Gallup, The Relationship Between Engagement at Work and Organizational Outcomes: 2020 Q12 Meta-Analysis (10th Edition) (2020).

Confidence: single-source — a proprietary Gallup vendor estimate, plausible and directionally consistent with the academic literature but not independently replicated.

Is it true that people leave managers, not companies?

The slogan oversimplifies. Turnover meta-analyses show the manager relationship genuinely predicts quitting — but it is one driver among several, and not the strongest.

Rubenstein and colleagues (2018, Personnel Psychology) — the most comprehensive turnover meta-analysis to date (316 studies, ~1,800 effect sizes) — aggregated leader–member exchange, consideration and transformational style into a single “Leadership” construct and found a corrected correlation of ρ = −.24 with voluntary turnover: better leadership, less quitting, but not the top factor. The strongest predictors are proximal withdrawal states — quit intentions (ρ ≈ +.56) and job search (ρ ≈ +.40) — followed by organizational commitment (−.29), job satisfaction (−.28), and job embeddedness (−.26). Notably, pay level is a weak direct predictor (ρ ≈ −.17) — weaker than leadership, satisfaction, commitment and embeddedness. Griffeth, Hom and Gaertner (2000) found the same pattern decades earlier.

So the cliché has a true core — relationships matter — but the exclusive “people leave managers, not companies” version is not supported. A 2025 Journal of Vocational Behavior study that directly tested it concluded the boss adage “has a true core,” but turnover is multi-determined, and boss issues actually surface less often in anonymous surveys than the slogan implies.

This also explains why “pay” dominates exit interviews: departing employees tend to give safe, socially acceptable reasons, and exit data under-report relational causes — the stated reason (pay) and the actual antecedents (satisfaction, commitment, embeddedness, leadership) diverge.

For a small Ontario employer, the practical implication holds even though the cliché is overstated: a frontline manager who erodes satisfaction, commitment and embeddedness raises quit risk, and the leaver will probably cite pay. But it is not “all the manager” — a hot local labour market (perceived alternatives, ρ ≈ +.23) and weak embeddedness are independent, comparably-sized drivers. The dollar cost of replacing a leaver is handled separately (Frank’s “Rule of Three”), as is the cost of disengagement below.

Source: Rubenstein, Eberly, Lee & Mitchell, Surveying the forest: a meta-analysis of the antecedents of voluntary employee turnover, Personnel Psychology 71(1), pp. 23–65 (2018); Griffeth, Hom & Gaertner, A meta-analysis of antecedents and correlates of employee turnover, Journal of Management 26(3) (2000); Hom, Lee, Shaw & Hausknecht, One hundred years of employee turnover theory and research, Journal of Applied Psychology 102(3) (2017); Turnover reasons are more complex than people quit bosses: an approach-avoidance perspective, Journal of Vocational Behavior (2025).

Confidence: industry-consensus — the manager/leadership relationship is a genuine but moderate predictor of turnover, and the exclusive “leave managers, not companies” version is not supported.

What does low engagement or disengagement actually cost a business?

Higher engagement at the work-unit level is reliably associated with better business results — but the link is correlational and modest, and the headline dollar figures are vendor estimates.

The peer-reviewed anchor is Harter, Schmidt and Hayes (2002, Journal of Applied Psychology), a meta-analysis of 7,939 business units across 36 companies that found generalizable unit-level relationships between employee engagement and customer satisfaction, productivity, profit, turnover, and safety. Gallup has re-run this analysis repeatedly; by its later editions the reported true-score correlation between engagement and a composite performance measure was about .43, with top-quartile units differing from bottom-quartile units by roughly 20% in productivity and profitability and substantially in turnover. Two cautions: the 2002 paper is peer-reviewed, but the later “editions” are Gallup-published and not independently peer-reviewed; and this is a correlation at the unit level — engagement may partly drive performance, but precise causal claims are not licensed.

The much-quoted cost numbers — that disengagement costs the world “$8.8–8.9 trillion, equal to 9% of global GDP” — come from Gallup’s State of the Global Workplace and are proprietary vendor estimates. The method for monetizing lost productivity from survey scores is not transparently published or peer-reviewed; these are modeled global aggregates, not a measured loss for any one firm.

For a 20–200-person Kitchener-Waterloo employer, the takeaway is the direction, not the dollar amount: more engaged teams tend to perform better and quit less. A global-GDP figure means nothing for a 60-person firm, and no honest source can tell an owner “disengagement is costing you $X.” The per-departure cost of turnover is handled separately (Frank’s “Rule of Three”), not restated here.

Source: Harter, Schmidt & Hayes, Business-unit-level relationship between employee satisfaction, engagement, and business outcomes: A meta-analysis, Journal of Applied Psychology 87(2), pp. 268–279 (2002); Gallup, The Relationship Between Engagement at Work and Organizational Outcomes: Q12 Meta-Analysis report (2020); Gallup, State of the Global Workplace (source of the $8.8T / 9%-of-GDP figures) (2023) (vendor).

Confidence: industry-consensus for the engagement–outcomes association (correlational and modest); the $8.8T / 9%-of-GDP cost figures are proprietary Gallup vendor estimates and must never be asserted as fact.

What do workplace loneliness and isolation cost, especially on remote teams?

Lonelier employees measurably underperform, and remote work raises the risk — but the economy-wide dollar figures are vendor estimates.

The peer-reviewed anchor is Ozcelik and Barsade (2018, Academy of Management Journal), a time-lagged, multi-rater field study of 672 employees and 114 supervisors. Greater workplace loneliness was significantly related to lower supervisor-rated job performance (zero-order r = −.28), partly because lonely employees were rated less approachable by coworkers and reported lower affective commitment. The authors are explicit they cannot confirm causality (the loop may run both ways) — so this is one strong study, not a settled body of work.

The broader public-health harms of loneliness are well established but are not business-cost numbers: the U.S. Surgeon General’s 2023 advisory, drawing on Holt-Lunstad’s mortality meta-analyses (3.4 million people; ~26–32% higher mortality odds), famously likens social disconnection to “smoking up to 15 cigarettes a day.” That concerns mortality, not output — do not conflate the two.

The figures with dollar signs are vendor-generated: Cigna’s index (an insurer, with Ipsos) put workplace loneliness at “more than $154 billion a year” in stress-related absenteeism and “61% of US adults lonely” — the latter resting on a scale cutoff other researchers would not necessarily call “lonely.” Label these directional. Peer-reviewed work does link more remote work to greater loneliness and professional isolation, but a clean remote-specific loneliness→performance effect size is not yet established.

For a Kitchener-Waterloo employer — where hybrid and remote work are common in the tech sector — the defensible claim is that lonely, isolated employees are measurably less approachable, less committed, and rated lower on performance. The US $154B and global figures do not scale to a 60-person firm; the appropriate local anchor is Statistics Canada’s finding that about 13% of Canadians reported always or often feeling lonely in 2021 (general population, not workplace-specific). The practice side of keeping a dispersed team connected is covered in the hybrid and remote connection guide.

Source: Ozcelik & Barsade, No Employee an Island: Workplace Loneliness and Job Performance, Academy of Management Journal 61(6), pp. 2343–2366 (2018); Office of the U.S. Surgeon General, Our Epidemic of Loneliness and Isolation (advisory, 2023); Holt-Lunstad et al., Loneliness and Social Isolation as Risk Factors for Mortality: A Meta-Analytic Review, Perspectives on Psychological Science 10(2) (2015); Statistics Canada, Canadian Social Survey: Loneliness in Canada (2021); Cigna, Loneliness and the Workplace: 2020 U.S. Report (2020) (vendor).

Confidence: industry-consensus for the loneliness–performance link (one strong, multi-rater study, correlational); the mortality evidence is well established but separate from output; the Cigna $154B / 61% figures are directional vendor estimates.

---

This page is general information drawn from published research, not professional or legal advice. Effect sizes and study findings come with the caveats noted in each section — much of the foundational evidence is US-based and cross-sectional, and the most-quoted dollar figures are proprietary vendor estimates — so treat the figures as directional and confirm any specific claim against the cited source before relying on it.