Social Ecological Model: S-Tier Behavioral Designer’s Guide

Every public-health campaign that ever failed at scale failed in the same way: it targeted the level that was cheapest to reach instead of the level where the behavior actually lived. Print a poster. Run a 30-second ad. Send a text reminder. The intervention sits on the individual, which is convenient for the budget and useless for the outcome, because the behavior is being shaped three layers up by a workplace policy nobody on the campaign team has the authority to change.

The Social Ecological Model is the diagram everybody draws on the whiteboard right before they go pick the wrong level anyway. Five concentric rings: Individual, Interpersonal, Community, Organizational, Public Policy. The model has been the de facto map of public health since McLeroy, Bibeau, Steckler and Glanz formalized it in 1988, and it has been misused for almost the entire run.

This guide is the version I wish every behavioral designer, public-health officer, and product manager had read before their first multi-stakeholder intervention. We will not just describe the five levels. We will name the reason the model keeps producing campaigns that look correct on a slide deck and collapse on contact with the real world, and we will rebuild it as a design surface, a place where Octalysis Core Drives get assigned per level rather than spread evenly across all five.

Speed Run Notes

• The Social Ecological Model is a map, not a theory. It tells you where to look. It does not tell you what to do once you get there. Pair it with a theory of behavior, or it predicts nothing.
• Five levels, in order: Individual → Interpersonal → Community → Organizational → Policy. McLeroy 1988 is the public-health version. Bronfenbrenner 1979 is the developmental origin, and the two are not interchangeable.
• The textbook win is tobacco control: simultaneous moves at all five levels (individual cessation aids, peer norms, community smoke-free spaces, workplace bans, federal excise tax) cut US adult smoking from 42% in 1965 to 11.5% in 2021.
• The textbook failure is any single-level intervention. Schoolyard anti-drug talks. Lone “eat-five-a-day” posters. Personal-responsibility climate campaigns. Single-level moves underperform multi-level moves by roughly an order of magnitude in meta-analytic effect size.
• The map’s quiet defect is direction ambiguity: arrows flow both up (an individual choice aggregates into a norm) and down (a policy reshapes individual choices). Treating the rings as a fixed top-down hierarchy hides the most actionable leverage point, which is almost always one ring inward from where the campaign budget is being spent.
• The Octalysis pairing: each ring gets a different Core Drive lead. Individual = Development & Accomplishment plus Empowerment of Creativity. Interpersonal = Social Influence & Relatedness. Community = Epic Meaning & Calling. Organizational = removing anti-Core Drives (sludge, hidden costs). Policy = Loss & Avoidance defaults plus Scarcity regulatory levers.

About the Creator of the Octalysis Framework

Yu-kai Chou created the Octalysis Framework after studying gamification since 2003, years before the term entered mainstream vocabulary. As a Human-Systems Architect & Behavioral Designer, his framework has been applied by LEGO, Microsoft, Porsche, Coca-Cola, Salesforce, and MrBeast, impacting over 1.5 Billion Users.

Chou has taught the Octalysis methodology at Harvard, Stanford, Yale, Tesla, Google, BCG, and IDEO.

His work has been cited by Harvard, Stanford, MIT, Forbes, Wall Street Journal, Wired, US Department of Energy, NIST, NSF, NCBI, US Department of Education, ClinicalTrials.gov, and Google Scholar — with 3,700+ more academic publications. Explore his books here.

What Is the Social Ecological Model?

The Social Ecological Model (SEM) is a behavior-change framework that places any individual behavior inside nested rings of influence, from the person at the center outward to the public policy environment at the perimeter. The model claims one load-bearing thing: behavior is the product of multiple, interacting layers, and intervention that ignores any layer leaves predictable failure on the table.

The version most public-health programs use today comes from Kenneth McLeroy, Daniel Bibeau, Allan Steckler, and Karen Glanz, who published “An Ecological Perspective on Health Promotion Programs” in Health Education Quarterly in 1988. The McLeroy team adapted the model for health-behavior intervention by naming five practical levels: intrapersonal (the individual), interpersonal (family, friends, peers), institutional or organizational (schools, workplaces, healthcare systems), community (relationships among organizations, local cultural norms), and public policy (laws and regulations).

The deeper origin is older. Urie Bronfenbrenner published The Ecology of Human Development in 1979 with a different goal in mind. Bronfenbrenner was a developmental psychologist trying to explain how children become who they become; his rings were microsystem, mesosystem, exosystem, macrosystem, and (added later in his 1986 update) chronosystem, which threaded time across the other four. The Bronfenbrenner version is broader, more interactionist, and less actionable for public-health work. The McLeroy adaptation is narrower and explicitly designed for intervention planning. Confusing the two is the most common citation error in the entire SEM literature.

For the rest of this guide we will work with the McLeroy public-health version, because that is the version the World Health Organization, the Centers for Disease Control, the US Office of the Surgeon General, and most national behavioral-insights units actually use when they plan a program. We will revisit Bronfenbrenner explicitly when his frame answers a question McLeroy does not.

What separates the Social Ecological Model from earlier behavior-change frameworks is the structural commitment. The Health Belief Model lives inside the individual. The Theory of Planned Behavior lives inside the individual. Self-Determination Theory lives inside the individual. The Social Ecological Model is the first widely-adopted framework that says: even the best individual-level theory is incomplete if you stop there. The interesting questions are above the skull.

The Five Levels of Influence

The model’s contribution is the level taxonomy. Each level names a class of intervention targets that behave differently from the others, different timelines, different stakeholders, different evidence standards, different failure modes. Designing for one level using the playbook of another is one of the most common ways for an otherwise-thoughtful intervention to flatline.

Level 1: Individual (Intrapersonal)

The individual level covers everything inside one person’s skin: knowledge, attitudes, beliefs, self-efficacy, intentions, biology, history. This is where most behavioral interventions live because it is the cheapest to reach. Run an ad. Send a text. Push a notification. Add an in-app modal. Run an educational seminar. None of these require negotiating with anyone other than the person at the receiving end.

The individual level is also where the deepest theoretical work has been done. Decades of psychology have given us granular accounts of what happens inside the person: belief evaluation (Health Belief Model), intention formation (Theory of Planned Behavior), self-efficacy mechanisms (Bandura 1977), habit formation (Wood & Neal 2007), implementation intentions (Gollwitzer 1999), need satisfaction (Self-Determination Theory).

The individual level’s signature failure is the assumption that good information plus good motivation reliably produces good behavior. Three decades of intention-behavior-gap research, including Sheeran’s 2002 meta-meta-analysis showing only 53% of intenders actually act, says it does not. People who know smoking causes cancer keep smoking. People who decide on Sunday to exercise four times this week routinely make it to one session. The individual is necessary but not sufficient, which is the entire reason the four levels above it exist.

The Octalysis-flavored design move at the individual level: build clear Core Drive 2 (Development & Accomplishment) progress representation, install Core Drive 3 (Empowerment of Creativity & Feedback) mastery loops, anchor Core Drive 4 (Ownership & Possession) over the user’s data and partial state. We will get to the per-level Octalysis mapping in detail below. The point here is that even the best individual-level design is a single ring of five.

Level 2: Interpersonal

The interpersonal level covers family, friends, peers, romantic partners, immediate coworkers, gym buddies, congregation members. This is the level at which the strongest social-influence effects in the entire behavior-change literature operate. When Christakis and Fowler tracked the Framingham Heart Study cohort across 32 years, obesity spread through friendship networks with measurable transmission probabilities, and so did smoking cessation, happiness, and loneliness. The closer the tie and the more reciprocal the friendship, the larger the effect.

Interpersonal effects come in two flavors that the public-health literature constantly confuses. The first is descriptive norms, what people like me actually do. “Seven of your ten closest friends got their flu shot this month” is a descriptive norm. The second is injunctive norms, what people like me approve of. “Your doctor strongly recommends you get vaccinated” is an injunctive norm. Reasoned Action Approach (Fishbein & Ajzen 2010) and decades of follow-up work consistently show that descriptive norms outperform injunctive norms when the referent group is genuinely “people like me,” and injunctive norms outperform descriptive norms when the referent is a high-status authority.

The interpersonal level is where Octalysis Core Drive 5 (Social Influence & Relatedness) does its heaviest lifting. Almost every product that has scaled a behavior across a social network, Strava for running, Duolingo for language learning, Headspace for meditation, has done so by deliberately routing the design through interpersonal cues rather than relying on individual willpower.

Level 3: Community

The community level is where the SEM diagram gets the fuzziest, and also where the McLeroy team made their sharpest contribution. “Community” here does not mean “the general public.” It means the relationships among formal and informal organizations within a defined locality, plus the collective identity and shared norms of that locality. The community level is the layer where local cultural meaning lives, the reason a smoking-cessation campaign works in Boston and dies in rural Mississippi even when the funding and the messaging are identical.

Community-level interventions are slow, expensive, and politically contested. They include coalition-building, neighborhood-level environmental modifications (sidewalks, parks, healthy-food retail), local media campaigns aligned with community leaders, and the rebuilding of community-level identity around a target behavior. The CDC’s REACH (Racial and Ethnic Approaches to Community Health) program is a textbook example: rather than running a one-size-fits-all health campaign, REACH funds local coalitions to identify community-specific levers and deploy them locally.

The Octalysis-relevant insight at the community level is that this is where Core Drive 1 (Epic Meaning & Calling) becomes accessible. Individual Core Drive 1 is fragile, most people do not wake up feeling like they are doing world-historical work. Community-level Core Drive 1 is stable, a person’s connection to their neighborhood, their congregation, their hometown identity is durable, and behavior framed as “what people in this community do” recruits a meaning structure that the individual could not have built alone.

Level 4: Organizational (Institutional)

The organizational level covers schools, workplaces, healthcare systems, religious institutions, military units, and any formal organization with policies that shape its members’ behavior. This level is the quiet powerhouse of the SEM diagram. Most adults spend most of their waking hours inside organizations, and the rules those organizations set, often without considering themselves behavior-change agents, do more to shape daily behavior than almost anything else on the map.

A workplace that bans smoking on the premises reduces employee smoking. A school district that switches its cafeteria default to water instead of soda reduces childhood sugar consumption. A healthcare system that builds vaccination reminders into the electronic-health-record workflow at every annual physical reaches a meaningful fraction of eligible patients. None of these moves require asking the individual to want anything different. They reshape the default architecture inside which the individual operates.

The organizational level is also where most public-health programs underinvest. Working with the individual is cheap. Working with the organization requires negotiating with leadership, often spending months building relationships before any intervention launches, and accepting that the organization has its own goals that may not align with the campaign’s. The leverage is huge. The activation energy is high. Many programs choose the cheaper option, build their intervention at the individual layer, and watch it sit at low single-digit effect sizes.

The Octalysis-relevant lens at the organizational level is mostly subtractive. This is where you remove the anti-Core Drives that the organization has accumulated by accident, the broken progress feedback in the benefits-enrollment flow, the multi-step exception-approval process that punishes employees for following the recommended behavior, the hidden costs of compliance that show up in psychological-cost-of-friction terms (the Sludge taxonomy formalized by Sunstein in 2022). The most measurable wins at the organizational level usually come from removing friction, not adding mechanics.

Level 5: Public Policy

The policy level covers laws and regulations at municipal, state, and federal scales. Smoking bans in public buildings. Excise taxes on tobacco. Trans-fat bans in restaurant food. Required calorie posting on chain-restaurant menus. Seatbelt laws. Helmet laws. Soda taxes. Drug-cessation insurance mandates. These are the interventions where the entire choice architecture for tens or hundreds of millions of people changes at once.

Policy moves have the largest possible per-dollar effect size in the entire SEM. The Centers for Disease Control estimates that comprehensive tobacco-control policies (federal excise tax increases, smoke-free air laws, mass-media counter-marketing) accounted for the largest single share of the US adult smoking decline since 1965, and the cost per quitter is a small fraction of the cost per quitter for individual-level cessation programs.

Policy is also the slowest, most politically contested, and most opaque level of the SEM. It requires legislative or regulatory action, which means coalitions, lobbying, public testimony, electoral pressure, and patience measured in years or decades rather than quarters. The pandemic-response literature from 2020 to 2024 contains many bitter case studies of policy moves that were behaviorally correct and politically infeasible, or politically feasible and behaviorally incoherent.

The Octalysis-relevant pairings at the policy level are Core Drive 6 (CD6): Scarcity & Impatience, regulatory levers that create binding constraints (caps, quotas, deadlines), and Core Drive 8 (CD8): Loss & Avoidance, the default-architecture mechanism that powers most behavioral-economics policy. Opt-in versus opt-out for organ donation. Auto-enroll versus opt-in for retirement savings. The mandatory-disclosure regime in nutrition labeling. These are all CD8 design moves at the policy ring, and they collectively produce the largest behavioral effects per implementation dollar that have ever been measured in any field.

A Word on the Sixth Level: Society and Culture

Several SEM variants add a sixth ring above policy: the broader societal or cultural environment. Some versions call it “macrosystem” (the Bronfenbrenner echo), others “structural determinants,” others “cultural norms.” This sixth ring covers the deep cultural assumptions of an entire society about gender, race, age, work, health, family, time. The reason it does not always appear in the public-health canon is that it is genuinely unactionable on intervention timescales. You cannot run a campaign that changes the broad cultural assumptions of a society in five years. You can run a campaign that, accreting with thousands of other campaigns, contributes to a generational shift.

Where the sixth ring earns its place is in honest acknowledgment of the floor every other level rests on. Some behaviors are downstream of structural inequality that no organizational policy and no community coalition can fully fix on the timescale of any single program. The wise SEM practitioner names this honestly rather than pretending an individual-level intervention can substitute for a structural one.

What McLeroy and Bronfenbrenner Got Right

It is fashionable in recent behavior-change writing to treat the SEM as an obvious idea that did not deserve the citation count it has. The fashion is wrong. Three things the original authors got right are worth naming explicitly because they keep being rediscovered, badly, by people who did not read the source papers.

First, they got the nesting structure right. The decision to draw the levels as concentric rings rather than as a flat list was a real intellectual move. A flat list of “things that affect behavior” is a checklist, and a checklist invites the user to address each item once and call it done. Concentric rings communicate that the inner rings sit inside the outer rings, that an individual is shaped by their interpersonal context, which is shaped by their community, which is shaped by organizations, which are shaped by policy. Once you accept that visual, you cannot keep designing single-level interventions with a clean conscience.

Second, they got the bidirectional causality right, even if many later users of the model do not. Both McLeroy and Bronfenbrenner explicitly stated that influence flows both inward (policy shapes the individual) and outward (individuals aggregate into communities that lobby for new policy). Most public-health pedagogy collapses this into a one-way top-down picture because it is easier to teach, but the bidirectionality is exactly what makes the model useful for organizing intervention strategy. Some leverage points are inward-flowing; others are outward-flowing; the audit asks where the highest-leverage flow direction is.

Third, they got the institutional level on the map at all. Before McLeroy’s 1988 paper, most behavior-change theory either lived entirely inside the individual or jumped from individual to society without naming organizations as their own intermediate level. Naming organizations as a discrete level surfaces an enormous category of intervention design that earlier frameworks left implicit. Most of the durable, scalable wins in modern behavioral design, from workplace wellness to default-enrollment retirement savings to electronic-health-record nudges, live at exactly this institutional level. McLeroy made them visible.

Where the Social Ecological Model Falls Apart

The Social Ecological Model has three structural problems that explain most of the campaigns that fail despite invoking it correctly. Naming them is more useful than apologizing for them.

The Map-Without-Theory Problem

The SEM tells you where to look. It does not tell you what to do once you get there. A campaign that says “we are using the Social Ecological Model” has not yet committed to any specific account of how behavior changes. The model is compatible with operant conditioning at the individual level, with Cialdini-style social proof at the interpersonal level, with diffusion-of-innovations dynamics at the community level, with default architecture at the organizational level, and with regulatory economics at the policy level. Because it is compatible with everything, it predicts nothing on its own.

This is not a knockdown criticism. It is a usage warning. The SEM is a strategy framework, not a tactical framework. A program design that names the five levels and stops there has done the easy part. The harder part is choosing, for each level, the underlying behavior-change theory that will actually drive the intervention. Without that pairing, “we used the SEM” is a slide-deck decoration. With that pairing, the SEM becomes a real planning tool.

The Single-Level Bias

The model’s own evidence base shows multi-level interventions outperform single-level interventions by large margins, and yet single-level interventions remain the modal design choice in public-health programming. The reason is structural: budget cycles, grant timelines, agency silos, and political ownership all push toward single-level interventions because single-level interventions can be funded by a single funder, executed by a single agency, evaluated on a single metric, and credited to a single principal investigator.

Multi-level interventions require cross-agency coordination, multi-year timelines, complex evaluation designs, and shared credit, which is to say they require the entire program-management infrastructure that most funders are constitutionally unable to provide. The SEM names the right answer and the funding architecture funds the wrong one. This is the single largest reason public-health programs systematically underperform their theoretical ceilings, and the SEM diagram itself cannot fix it.

The Causal Direction Ambiguity

The bidirectionality McLeroy and Bronfenbrenner got right is also the model’s quiet defect. Without a clear theory of which direction dominates in a specific case, “everything affects everything” becomes the practitioner’s working assumption, and that assumption is paralyzing. Should we invest in the individual ring, where the dollar cost per touch is lowest? Or the policy ring, where the leverage is highest but the political timeline is longest? The SEM does not answer this. It cannot. The answer depends on the behavior, the population, the budget, the timeline, and the political environment, none of which the diagram itself encodes.

In practice, most teams resolve this ambiguity by reaching for the level they were already going to use. The advertising agency lobbies for an individual-level campaign. The legislative staffer pushes for a policy bill. The community-coalition consultant proposes a coalition. The SEM diagram on the wall ratifies whichever choice the team would have made anyway. The model deserves to be more than aesthetic cover for pre-existing preference, and the way to make it more is to add an explicit leverage-assessment step at the front of every audit, which we will sketch below in the Practical Steps section.

What’s Really Happening Inside the Brain

The Social Ecological Model is not a neural model, but the cognitive science of the last twenty years has clarified why multi-level interventions outperform single-level ones in ways the original authors could not have anticipated.

The individual ring corresponds, roughly, to the conscious deliberation system that Daniel Kahneman labeled System 2 in Thinking, Fast and Slow. Individual-level interventions try to recruit deliberative cognition: read this, weigh this, decide this, intend this. The system is expensive, slow, and easily depleted. Most of the time, most behavior is not under System 2 control, which is why individual-level interventions that rely on conscious choice consistently underperform their theoretical ceilings.

The interpersonal ring corresponds to social cognition, especially the descriptive- and injunctive-norm machinery that runs largely below conscious awareness. The fusiform face area and the medial prefrontal cortex do most of the work here, integrating cues from familiar faces about what is normal and approved. Interventions at the interpersonal level recruit a different and cheaper substrate than individual interventions because they ride on the social-monitoring system that most people leave running 24/7 anyway.

The community and organizational rings correspond to environmental cuing and what behavioral economists call choice architecture. The basal ganglia handle the procedural-habit substrate; the prefrontal cortex handles goal maintenance; the environment supplies the cues that determine which procedural routine fires. Interventions at these levels work by changing the cues the procedural system receives, not by asking the deliberative system to override the procedural system, which is what individual-level interventions implicitly request.

The policy ring corresponds to the choice-architecture environment at the largest scale: the defaults, the prices, the friction, the salience of options at the moment of decision. Behavioral economics has documented over and over that the default option captures a disproportionate share of decisions even when the cost of switching is trivial, because default-following is the brain’s energy-conservation strategy in any non-critical decision. Policy moves work, neurally, by changing the default that the system encounters.

This neural mapping reinforces the SEM’s central design insight: as you move outward through the rings, the cognitive cost to the individual of the intervention decreases. Individual-level interventions cost the user mental effort. Policy-level interventions cost the user nothing, they encounter a different environment, and the new environment makes the desired behavior the path of least resistance. This is also why outer-ring interventions tend to be more equitable than inner-ring interventions: inner-ring interventions advantage people with more cognitive bandwidth, more education, more time, and more individual agency, while outer-ring interventions distribute the behavior change across populations without requiring those resources.

SEM vs Other Behavior-Change Theories

The Social Ecological Model is not a substitute for the individual-level theories of behavior. It is a planning frame that those theories sit inside. Knowing where each theory lives is most of what separates a thoughtful application from a confused one.

SEM vs the Health Belief Model

The Health Belief Model (Rosenstock 1974) is an individual-level theory. It decomposes behavior into Perceived Susceptibility, Perceived Severity, Perceived Benefits, Perceived Barriers, Self-Efficacy, and Cues to Action. Every one of those constructs lives inside one person’s head. The Health Belief Model is excellent for understanding why a specific individual does or does not adopt a health behavior. It is silent on how to design the workplace, community, or policy environment around that individual. The SEM is the larger map; the Health Belief Model is what you use at the individual ring of that map. Programs that use the Health Belief Model alone often underperform because they have implicitly assumed the outer four rings of the SEM do not need attention.

SEM vs the Theory of Planned Behavior

The Theory of Planned Behavior (Ajzen 1991) is also an individual-level theory. It predicts behavior via Attitude, Subjective Norm, and Perceived Behavioral Control feeding into Intention, which then maps imperfectly onto Behavior. The Theory of Planned Behavior is the most-cited behavior-change model ever written, and it explains roughly 39% of variance in intention and 27% of variance in behavior in Armitage and Conner’s 2001 meta-analysis. Like the Health Belief Model, the Theory of Planned Behavior lives in the individual ring of the SEM. The Subjective Norm construct gestures at the interpersonal ring but stops there. The Theory of Planned Behavior and the SEM are complements, not competitors: the SEM tells you which ring to design, and the Theory of Planned Behavior tells you what to do once you commit to designing the individual ring.

SEM vs the COM-B Behavior Change Wheel

The COM-B model (Michie, van Stralen & West 2011) and its surrounding Behaviour Change Wheel are the closest theoretical cousins to the SEM in modern practice. COM-B decomposes behavior into Capability, Opportunity, and Motivation, and the surrounding Wheel adds intervention functions and policy categories. Where the SEM gives you a structural map (where to look), COM-B gives you a functional map (what to look for). Many UK behavioural-insights teams now use both: SEM for stakeholder mapping and intervention-target selection at the strategy stage, COM-B for diagnostic depth on a chosen target. They are layered, not competing.

SEM vs Diffusion of Innovations

Diffusion of Innovations (Rogers 1962) is a population-level model of how new behaviors spread through a social system over time. It introduces the S-curve adoption diagram with Innovators, Early Adopters, Early Majority, Late Majority, and Laggards. Diffusion lives mostly at the community and interpersonal rings of the SEM but operates over time in a way the SEM diagram is silent on. Most well-run SEM-guided campaigns layer in the Diffusion model at the community ring to choose which sub-populations to seed first and how to use early-adopter visibility to recruit the early majority.

The Social Ecological Model in the Real World

The model has been deployed at scale across enough domains that the pattern of where it works and where it fails is now visible. Four cases worth examining closely: tobacco control, workplace wellness, climate-behavior interventions, and the pandemic response.

Tobacco Control: The Textbook Win

US adult smoking prevalence fell from 42.4% in 1965 to 11.5% in 2021, the largest sustained behavior change in modern public-health history. No single intervention produced that decline. A simultaneous multi-decade push at every level of the Social Ecological Model produced it. At the individual level: nicotine replacement therapy, prescription cessation aids, and quitlines. At the interpersonal level: anti-smoking messaging that reframed smoking as socially undesirable rather than glamorous. At the community level: smoke-free restaurants, smoke-free bars, smoke-free parks, and visible enforcement that signaled the new local norm. At the organizational level: workplace smoking bans that removed the convenient smoking break from the structure of the workday. At the policy level: federal excise tax increases that raised the per-pack price to the point where price-sensitive smokers quit, the 1998 Master Settlement Agreement that defunded most tobacco-industry marketing, and the FDA Family Smoking Prevention and Tobacco Control Act of 2009 that gave the FDA direct regulatory authority.

The tobacco-control case is the strongest empirical evidence the Social Ecological Model has. No single level would have produced the decline. Probably no pair of levels would have. The combination, sustained for decades, is what worked. Every public-health practitioner trained in the last twenty years has been told this story, and yet most public-health programs continue to fund single-level interventions because the cross-level coordination required is beyond what most agency structures can sustain.

Workplace Wellness: The Cautionary Case

Workplace wellness programs are a useful counterexample. Most large US employers now run some form of wellness program, biometric screenings, health-risk assessments, gym subsidies, smoking-cessation benefits, sometimes premium discounts tied to participation. The first large randomized evaluation, the BJ's Wholesale Club study published in JAMA in 2019 by Song and Baicker, found no significant effect on clinical health measures or healthcare spending after 18 months.

The diagnosis from a Social Ecological Model perspective is straightforward: workplace wellness programs typically operate at the organizational and individual levels and ignore the interpersonal, community, and policy levels that would make them stick. Offering an employee a discount on a gym membership without changing the workplace culture around long hours, the surrounding community’s food environment, or the broader policy context around health-insurance design produces small, transient changes that wash out at follow-up. The same dollars invested across multiple SEM levels, including the organizational redesign of work itself, would likely produce different results, but the budget structure of corporate wellness funds single-level investments because that is how procurement is built.

Climate Behavior: The Mismatched Level

The climate-behavior literature shows the single-level bias at its most visible. Most consumer-facing climate campaigns target the individual level: turn off lights, take shorter showers, drive less, fly less, eat less meat. The aggregate effect of these individual moves, while not zero, is dwarfed by the policy-level moves available at the same time (carbon pricing, building codes, vehicle-emission standards, grid-decarbonization regulation). Decades of meta-analytic work consistently find that individual-level climate behavior change accounts for a small fraction of the total behavioral leverage and that the framing of climate as a personal-responsibility problem actively obstructs the policy-level moves that would matter more.

From a Social Ecological Model standpoint, this is a classic case of intervening at the wrong ring. The cheapest ring to reach gets the campaign budget. The ring where the actual leverage sits gets a fraction of the attention. The model itself does not prescribe the answer, but the framework’s central commitment, multi-level interventions outperform single-level ones, is a constant reminder that the individual ring is rarely where the biggest behavior change lives.

Pandemic Response: The Real-Time Stress Test

The COVID-19 pandemic was an inadvertent natural experiment in Social Ecological Model design. Some jurisdictions intervened across all five rings simultaneously: individual-level public-health messaging, interpersonal-level recommendations about household contacts, community-level local restrictions, organizational-level workplace and school protocols, and policy-level laws on masking, distancing, and vaccination. Other jurisdictions intervened at only one or two rings. The differential outcomes across jurisdictions, by every published behavioral and epidemiological measure, tracked the multi-level versus single-level distinction.

The pandemic also exposed the model’s limits. Sustained behavior change at all five rings requires sustained political coherence at all five rings, and that coherence eroded almost everywhere within twelve to eighteen months. The Social Ecological Model is a planning frame for intervention design; it does not, and cannot, guarantee the political conditions required to keep the design coherent over time. That is a different kind of problem, and one the model honestly does not solve.

The Elephant in the Room

The elephant is this: most people who invoke the Social Ecological Model in 2026 use it the way mid-century academics used the word “interdisciplinary.” It signals seriousness without committing to anything. A program proposal that names all five levels has performed the ritual that funders expect, and then quietly returns to designing whatever single-level intervention the team was going to design anyway.

This is the failure mode the framework cannot fix from inside itself. The Social Ecological Model can name where to look; it cannot make practitioners look. The clearest sign of a healthy SEM-guided program is not the presence of the five-ring diagram in the proposal but the presence of cross-level operational decisions in the work plan. Does the program have a community-level coalition stood up? Does it have organizational MOUs with the schools, workplaces, or healthcare systems in scope? Does it have a policy advocacy track running in parallel? If those operational artifacts are missing, the SEM diagram is decoration.

A second elephant: the model is, for many practitioners, a way to feel like they are taking structural causes seriously without doing the structural work. Naming “policy” as the fifth ring on a slide is not policy advocacy. Naming “community” as the third ring is not coalition-building. The model invites cheap performance of seriousness, and the most rigorous SEM practitioners treat the diagram as a checklist of operational commitments, not as a vocabulary list.

A third elephant: the bidirectional causality McLeroy and Bronfenbrenner got right is also where Black-Hat design lives. The same model that explains how prosocial behaviors spread upward and outward also explains how harmful behaviors do. Tobacco companies in the 1950s through 1990s ran an inadvertent SEM-style campaign in reverse: individual-level marketing to recruit new smokers, interpersonal-level cultural production to make smoking look attractive, community-level placement of cigarette ads at sports events and bars, organizational-level workplace smoking cultures, and policy-level lobbying to prevent regulation. Every public-health win since 1965 has been a counter-campaign against an industry-funded SEM-style intervention. Naming this out loud is part of taking the model seriously rather than romantically.

How to Apply the Social Ecological Model with the Octalysis Framework

The Social Ecological Model tells you where to intervene. The Octalysis Framework tells you which psychological lever to pull at that location. The two frameworks are complements: SEM is the structural map, Octalysis is the per-ring design surface. Used together, they convert “we should run a multi-level intervention” from a slogan into a build-able specification.

Below is the per-ring Octalysis assignment I run when I am asked to audit an SEM-guided program. The point of the assignment is not that every Core Drive belongs at every ring, it is that each ring has a small set of Core Drives that do most of the work, and trying to use the wrong drive at the wrong ring is one of the most reliable ways for a campaign to produce small, transient, or counter-productive results.

Individual Ring: Core Drives 2, 3, and 4

At the individual ring, three Core Drives carry most of the load. Core Drive 2 (CD2): Development & Accomplishment handles the visible progress representation that converts abstract behavior change into felt forward motion. Every individual-level intervention needs a CD2 surface, a progress bar, a streak, a level, a milestone unlock, that makes the user’s behavior legible to themselves. Without CD2, even a motivated user loses the felt sense of progress within days. Core Drive 3 (CD3): Empowerment of Creativity & Feedback is where Self-Efficacy is built, through Bandura’s mastery-experience mechanism: short loops, low stakes, rapid feedback. CD3 is the substrate under any “practice mode” or low-stakes onboarding loop. Core Drive 4 (CD4): Ownership & Possession anchors the user’s data and partial state, making the intervention feel like an asset the user has built rather than a service they are consuming.

What the individual ring should not lean on, at least not as the primary drive: Core Drive 8 (Loss & Avoidance) deployed atmospherically (constant fear messaging fatigues), or Core Drive 6 (Scarcity & Impatience) deployed as a default rhythm (urgency framing corrodes trust in a sustained behavior-change context). Save CD6 and CD8 for the policy ring, where the structural defaults can carry them without exhausting the user.

Interpersonal Ring: Core Drive 5, Split Two Ways

The interpersonal ring is almost entirely Core Drive 5 (CD5): Social Influence & Relatedness, but with a split that most public-health design ignores. CD5 has two operational modes that the Reasoned Action Approach work formalized in 2010: descriptive norms (“seven of your friends got their flu shot this month”) and injunctive norms (“your doctor strongly recommends the flu shot”). Descriptive norms outperform injunctive norms when the referent group is genuinely “people like me.” Injunctive norms outperform descriptive norms when the referent is a high-status authority figure.

The design move at the interpersonal ring is to choose the right CD5 mode for the specific behavior and referent. A teenager being asked to quit vaping responds more to descriptive norms from peers than to injunctive norms from parents. An older adult being asked to get a vaccine responds more to injunctive norms from their primary-care physician than to descriptive norms from acquaintances. Picking the wrong CD5 mode is the most common design error at the interpersonal ring; the fix is a referent-group analysis at the planning stage, not a post-hoc data-cut after the campaign launches.

Community Ring: Core Drives 1 and 5 Together

The community ring is where Core Drive 1 (CD1): Epic Meaning & Calling becomes accessible in a way it is not at the individual ring. Most individuals do not wake up feeling like their daily behavior matters at world-historical scale; that is too heavy a frame to install in one person’s head. But a person’s connection to their neighborhood, their congregation, their hometown, their team is durable and CD1-saturated almost by default. Designing the community-ring intervention to recruit existing CD1 structures, “people in this neighborhood take care of their elders” rather than “you, as an individual, should care for your elder”, converts a fragile individual drive into a stable community-level one.

The community ring also leans heavily on Core Drive 5 (CD5), especially the descriptive-norm side. Community-level interventions work when the visible behavior of the community itself signals the new norm. A neighborhood where the visible signs are sidewalk-friendly running shoes, bikes locked outside coffee shops, and farmers’-market produce communicates a different set of community defaults than a neighborhood where the visible signs are drive-through traffic and convenience-store storefronts. The design move is to make the desired behavior visible at the community level so that the descriptive-norm machinery in the brain has accurate data to work with.

Organizational Ring: Subtract Anti-Core-Drives Before Adding Anything

The organizational ring is the ring where I almost always recommend subtractive design over additive design. Most organizations have accumulated, by accident over years, a long list of anti-Core-Drives, friction in workflows, hidden costs of compliance, broken progress feedback, ownership-erosion from lost partial state, that obstruct the desired behavior more than any new mechanic would promote it. The Sludge audit, formalized by Cass Sunstein in 2022, is the most rigorous version of this subtractive lens.

What a subtractive audit looks like at the organizational ring: walk the desired behavior end to end as if you were a new employee, a new patient, a new customer. Count the steps that exist for institutional convenience and not user benefit. Count the steps that require the user to re-enter data the institution already has. Count the steps where partial progress is lost. Count the steps where the cost of compliance is borne entirely by the user with no compensating help from the institution. Each one is a candidate for removal. The behavior change available from removing these obstructions is almost always larger than the behavior change available from adding a new gamification mechanic on top.

If additive design is the right call at the organizational ring, the Core Drive that usually wins is Core Drive 2 (CD2) via institution-supplied progress visibility, the electronic-health-record “your care team” view, the employee “your benefits this year” dashboard, the school “your child’s portfolio” report. These are CD2 surfaces that the institution can build once and serve to every member of the population, which is the kind of leverage the organizational ring is supposed to produce.

Policy Ring: Core Drives 6 and 8 as Structural Levers

The policy ring is the ring where Core Drive 6 (CD6): Scarcity & Impatience and Core Drive 8 (CD8): Loss & Avoidance reach their highest leverage, because policy is the layer at which the choice architecture for an entire population can be set. CD6 at the policy level is regulatory scarcity: caps on harmful products, quotas on emissions, deadlines on compliance. CD8 at the policy level is default architecture: the opt-out default that gives organ-donation rates of 80% or above in countries that use it versus 15% or below in countries that use opt-in defaults; the auto-enrollment default in retirement savings that has lifted participation rates from 40% to over 90% in employer-sponsored plans; the disclosure mandate that raises the salience of harmful options at the moment of choice.

Used at the policy ring, CD6 and CD8 produce population-level behavior changes that no individual-ring or interpersonal-ring intervention can match. Used at the wrong ring, CD8 deployed atmospherically across an individual-level campaign, CD6 deployed as artificial urgency in a community-level coalition, they corrode trust and burn out the audience. The discipline of the Octalysis-plus-SEM audit is matching each Core Drive to the ring where it does its best work.

The SEM × Octalysis Audit (Six Steps)

The audit turns the model from a vocabulary list into a planning instrument. Run it at the start of any new intervention.

First, name the target behavior in operational terms: who, doing what, how often, under what conditions, observed how. “Increased physical activity” is not a target behavior. “Adults aged 45-65 in this zip code walking at least 20 minutes per day, measured by a phone-based pedometer, sustained for 90+ days” is a target behavior.

Second, list every existing intervention or environmental factor that currently shapes this behavior, sorted into the five rings. Most teams discover that they already have four or five interventions piled up at the individual ring and zero or one at the policy ring. Surfacing this imbalance is half the value of the audit.

Third, identify the highest-leverage ring for new investment given the team’s budget, timeline, and political access. The highest-leverage ring is rarely the cheapest ring. Choosing it explicitly is the audit’s central decision.

Fourth, for the chosen ring, name the underlying behavior-change theory. Without this, the ring designation is empty. Use the Health Belief Model or the Theory of Planned Behavior at the individual ring; use the Reasoned Action Approach social-norms machinery at the interpersonal ring; use diffusion dynamics at the community ring; use choice architecture at the organizational and policy rings.

Fifth, for the chosen ring and theory, map the Octalysis Core Drives that do the heaviest work. Use CD2-CD3-CD4 at the individual ring, CD5 split injunctive/descriptive at the interpersonal ring, CD1-CD5 at the community ring, anti-Core-Drive subtraction plus CD2 at the organizational ring, CD6-CD8 at the policy ring.

Sixth, build the smallest viable intervention that exercises the chosen ring, theory, and Core Drives. Measure. Adjust. Then, and only then, add a second ring to the program. Do not try to launch multi-ring interventions on day one; the operational overhead will collapse the program before any single ring’s design has matured.

Practical Steps to Run a Social Ecological Audit

1. Write the target behavior in operational form. Name the population, the action, the frequency, the observation method, and the success threshold. If you cannot write this in a single paragraph, the rest of the audit will be vague in proportion.
2. Inventory existing interventions and environmental factors across all five rings. Include interventions that other agencies, departments, or organizations are running, not just your own. The map is the territory you are entering, not the territory you wish you were entering.
3. Identify the dominant constraint. Ask: which ring, if it changed, would unlock the largest behavior change? This is rarely the ring with the most existing investment. It is the ring with the largest leverage relative to the resources required to move it.
4. Choose your behavior-change theory for the chosen ring. Do not skip this step. “Using the SEM” without a paired behavior-change theory is the most common reason multi-level interventions fail in practice.
5. Run the Octalysis × SEM Core Drive assignment for the chosen ring. Pick the two or three Core Drives that match the ring’s leverage. Resist the temptation to spread all eight across all five rings; that is how programs end up with no concentration anywhere.
6. Pilot at the smallest scale that lets you observe the target behavior. Most SEM-guided programs go from PowerPoint to statewide roll-out without a real pilot, which is why so many of them collapse on first contact with the actual population.
7. Layer in a second ring only after the first ring’s design has stabilized. Multi-ring interventions are the goal, but launching them all simultaneously is one of the most reliable ways to dilute the design at every ring. Sequence the build.

The Social Ecological Model Was the Beginning, Not the End

McLeroy, Bibeau, Steckler, and Glanz did the field a real service in 1988 by formalizing a vocabulary the field had been grasping for. Forty years later, the model is everywhere, taught in every public-health curriculum, cited in every program proposal. The risk now is that ubiquity has eroded discipline. The SEM has become a slide that people put on slides, not a planning instrument that people use to make decisions.

The recovery move is to treat the model as the beginning of an audit, not the end of one. Name the five rings. Then name the behavior-change theory inside each ring. Then name the Core Drives that fit the theory. Then name the operational commitments, coalitions, MOUs, advocacy tracks, evaluation designs, that turn the ring designations into work plans. The model rewards practitioners who use it as the start of the planning process; it punishes practitioners who use it as a vocabulary for sounding rigorous after the planning has already been done.

The strongest version of the framework, the version that survives serious use over time, is the version paired with an underlying theory at each ring and a design surface (Octalysis or otherwise) for translating that theory into specific behavioral interventions. That is the version that produced the tobacco-control win. It is the version that, if applied with discipline, could produce the next one.

Frequently Asked Questions

What is the Social Ecological Model in simple terms?

The Social Ecological Model is a behavior-change framework that places any individual behavior inside five nested rings of influence: the individual, their interpersonal relationships, their community, the organizations they belong to, and the public policy environment around them. It argues that behavior is shaped by all five rings simultaneously, and that the most effective behavior-change interventions operate at multiple rings at once rather than at any single ring in isolation.

Who developed the Social Ecological Model?

The public-health version most widely used today was published in 1988 by Kenneth McLeroy, Daniel Bibeau, Allan Steckler, and Karen Glanz in Health Education Quarterly. The deeper conceptual origin is Urie Bronfenbrenner’s 1979 book The Ecology of Human Development, which used a similar nested-rings structure to describe developmental contexts in children. The McLeroy adaptation is more action-oriented and more directly applicable to behavior-change programming.

What are the five levels of the Social Ecological Model?

The five levels, from innermost to outermost, are: (1) Individual or intrapersonal, knowledge, attitudes, beliefs, biology, self-efficacy; (2) Interpersonal, family, friends, peers, immediate social network; (3) Community, relationships among local organizations, neighborhood-level norms and cultural identity; (4) Organizational or institutional, schools, workplaces, healthcare systems, religious institutions; (5) Public policy, laws and regulations at municipal, state, and federal scales. Some variants add a sixth ring for broader societal or cultural context.

How is the Social Ecological Model different from the Health Belief Model or the Theory of Planned Behavior?

The Health Belief Model and the Theory of Planned Behavior are individual-level theories. They describe what happens inside one person’s head when they decide whether to perform a health behavior. The Social Ecological Model is a structural map that sits above those individual-level theories. The relationship is layered, not competitive: the Social Ecological Model tells you which ring to design an intervention at, and the individual-level theories tell you what to do once you commit to designing the individual ring.

Why do single-level interventions usually underperform?

Because behavior is being shaped by multiple rings simultaneously, an intervention at any single ring is fighting against the inertia of the four rings it ignored. The classic example is an individual-level smoking-cessation campaign run in a workplace that still allows smoking breaks, surrounded by friends who smoke, in a community where smoke-free public spaces are scarce, in a state without tobacco taxes. Even motivated quitters relapse under that aggregate environmental pressure. Multi-level interventions succeed by reducing the aggregate pressure across rings simultaneously.

How does the Social Ecological Model relate to the Octalysis Framework?

The Social Ecological Model tells you where to intervene; the Octalysis Framework tells you which psychological lever to use at each location. The standard pairing assigns Core Drive 2 (Development & Accomplishment), Core Drive 3 (Empowerment of Creativity & Feedback), and Core Drive 4 (Ownership & Possession) to the individual ring; Core Drive 5 (Social Influence & Relatedness) split into descriptive and injunctive norms at the interpersonal ring; Core Drives 1 (Epic Meaning & Calling) and 5 at the community ring; subtractive anti-Core-Drive work plus Core Drive 2 at the organizational ring; and Core Drives 6 (Scarcity & Impatience) and 8 (Loss & Avoidance) at the policy ring.

What is the most common misuse of the Social Ecological Model?

Citing it as a vocabulary in a program proposal and then proceeding to design a single-level intervention anyway. The model is supposed to drive operational decisions about which rings the program will actually touch, with budget, timeline, and stakeholder commitments at each ring. When the model appears only as a diagram and not as a set of operational commitments, coalitions stood up, organizational MOUs signed, policy advocacy tracks funded, it is decoration.

Where is the highest behavioral leverage typically found?

It depends on the behavior, the population, and the political environment, but as a working heuristic, the highest per-dollar leverage usually sits at the organizational or policy ring rather than at the individual ring. This is because outer-ring interventions reshape the choice architecture for entire populations at once, while inner-ring interventions ask each individual to expend cognitive effort to override their environment. The tobacco-control case is the canonical example: federal excise taxes and smoke-free-air laws produced larger per-dollar effects than individual cessation programs.

Can the Social Ecological Model be used for non-health behaviors?

Yes. While the model’s public-health framing is the most documented, the same nested-rings logic applies to consumer behavior, financial behavior, civic behavior, environmental behavior, and any other domain where individuals act inside contexts shaped by larger structures. Climate-behavior researchers, financial-inclusion programs, civic-engagement campaigns, and product-design teams have all adopted the model. The Octalysis-style per-ring Core Drive assignment translates across domains with minor adjustments to the underlying behavior-change theory paired at each ring.

What is the single biggest improvement a team can make to its Social Ecological Model practice?

Stop listing all five rings as a generic checklist. Pick the highest-leverage ring for the specific behavior, population, and budget at hand, and concentrate the design there with a paired behavior-change theory and a specific Core Drive lineup. After that ring’s design has stabilized and is showing measurable behavior change, layer in a second ring. Most programs fail by trying to be at all five rings at once and ending up with a watered-down presence at every ring; the discipline is sequencing the build, not spreading thin.

References

• McLeroy, K. R., Bibeau, D., Steckler, A., & Glanz, K. (1988). An ecological perspective on health promotion programs. Health Education Quarterly, 15(4), 351-377.
• Bronfenbrenner, U. (1979). The Ecology of Human Development: Experiments by Nature and Design. Harvard University Press.
• Bronfenbrenner, U. (1986). Ecology of the family as a context for human development: Research perspectives. Developmental Psychology, 22(6), 723-742.
• Stokols, D. (1996). Translating social ecological theory into guidelines for community health promotion. American Journal of Health Promotion, 10(4), 282-298.
• Sallis, J. F., Owen, N., & Fisher, E. B. (2008). Ecological models of health behavior. In K. Glanz, B. K. Rimer, & K. Viswanath (Eds.), Health Behavior and Health Education: Theory, Research, and Practice (4th ed., pp. 465-485). Jossey-Bass.
• Glanz, K., Rimer, B. K., & Viswanath, K. (Eds.). (2015). Health Behavior: Theory, Research, and Practice (5th ed.). Jossey-Bass.
• 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.
• Fishbein, M., & Ajzen, I. (2010). Predicting and Changing Behavior: The Reasoned Action Approach. Psychology Press.
• Ajzen, I. (1991). The theory of planned behavior. Organizational Behavior and Human Decision Processes, 50(2), 179-211.
• Rosenstock, I. M. (1974). Historical origins of the Health Belief Model. Health Education Monographs, 2(4), 328-335.
• Rogers, E. M. (2003). Diffusion of Innovations (5th ed.). Free Press.
• Michie, S., van Stralen, M. M., & West, R. (2011). The behaviour change wheel: A new method for characterising and designing behaviour change interventions. Implementation Science, 6, 42.
• Thaler, R. H., & Sunstein, C. R. (2008). Nudge: Improving Decisions About Health, Wealth, and Happiness. Yale University Press.
• Sunstein, C. R. (2022). Sludge: What Stops Us from Getting What We Want and What to Do about It. MIT Press.
• Bandura, A. (1977). Self-efficacy: Toward a unifying theory of behavioral change. Psychological Review, 84(2), 191-215.
• Gollwitzer, P. M. (1999). Implementation intentions: Strong effects of simple plans. American Psychologist, 54(7), 493-503.
• Sheeran, P. (2002). Intention-behavior relations: A conceptual and empirical review. European Review of Social Psychology, 12(1), 1-36.
• Song, Z., & Baicker, K. (2019). Effect of a workplace wellness program on employee health and economic outcomes: A randomized clinical trial. JAMA, 321(15), 1491-1501.
• Chou, Y.-K. (2015). Actionable Gamification: Beyond Points, Badges, and Leaderboards. Octalysis Media.

Related Reading

• The Octalysis Framework: The Complete Guide to the 8 Core Drives of Human Motivation, the eight-Core-Drive design surface this guide pairs with at every ring.
• Health Belief Model: S-Tier Behavioral Designer’s Guide, the individual-ring health-behavior theory that pairs cleanly with the Social Ecological Model.
• Theory of Planned Behavior: S-Tier Behavioral Designer’s Guide, the most-cited individual-ring behavior-change model.
• Sludge (Sunstein): S-Tier Behavioral Designer’s Guide, the subtractive audit lens for the organizational ring.
• The Behavioral Framework Library, the full library of behavioral and psychological frameworks every designer should know.

WOULD YOU LIKE YU-KAI CHOU TO WORK WITH YOUR ORGANIZATION?