Behavior Change Frameworks Compared | The Behavioral Scientist

Behavior change frameworks compared: which model should you use?

Behavior change frameworks can help you examine a problem, design an intervention, or describe what you did. Those are different jobs. Before choosing a model, decide which job you need it to do.

The field has a framework problem: plenty of named approaches, with no common meaning of “works.” Behavior change theories, design frameworks and technique taxonomies answer different questions. A theory proposes relationships between psychological or environmental factors. A design framework organizes intervention development. A taxonomy supplies names for intervention content. Evidence that one tool classifies programs reliably cannot be compared directly with evidence that another predicts behavior or improves an outcome.

This comparison covers twelve widely discussed approaches, including COM-B, Fogg, EAST, and the Theory of Planned Behavior. It explains what each can contribute, where the evidence is weak, and how to choose among them.

Twelve behavior change frameworks and techniques at a glance

Approach	Main purpose	Key distinction
COM-B / Behaviour Change Wheel	Organize assessment and intervention design around capability, opportunity and motivation.	The full wheel adds nine intervention functions and seven policy categories. Classification reliability is not proof of superior outcomes.
BCT Taxonomy v1	Describe intervention content using 93 techniques in 16 groups.	A shared vocabulary does not identify which technique will work best.
MINDSPACE	Consider Messenger, Incentives, Norms, Defaults, Salience, Priming, Affect, Commitments and Ego.	Each influence needs its own evidence; a checklist is not a validated causal model of every behavior.
EAST	Generate options that make behavior Easy, Attractive, Social and Timely.	Trials of particular interventions do not establish the effectiveness of using the four-part checklist.
Nudge / choice architecture	Change how options are presented while preserving choice and avoiding substantial changes to economic incentives.	Defaults, reminders and other designs have different effects across settings; the broad label does not specify an intervention.
Fogg Behavior Model	Propose that motivation, ability and a prompt converge for an action to occur.	B=MAP is a design model. Successful applications alone do not establish universal validity.
Theory of Planned Behavior	Relate attitudes, subjective norms and perceived control to intention, and intention and control to behavior.	Predicting intention differs from demonstrating successful behavior change.
Social Cognitive Theory	Examine self-efficacy, goals, outcome expectations and environmental influences.	The theory extends beyond confidence and includes social and structural conditions.
Self-Determination Theory	Examine motivation quality and support for autonomy, competence and relatedness.	Extrinsic motivation can be autonomous when people endorse its purpose; the theory is not simply intrinsic versus extrinsic motivation.
Transtheoretical Model	Describe stages and processes of change, alongside self-efficacy and temptation.	The 1997 version includes termination as a sixth stage; many applications use five. Describing readiness and proving the benefit of stage matching are separate tasks.
Health Belief Model	Examine perceived susceptibility, severity, benefits and barriers, with cues to action and later self-efficacy extensions.	Threat is only part of the model. Relationships can vary across behaviors and measures.
Implementation intentions	Specify an if-then plan linking a situation to an intended response.	This is a planning technique. It does not by itself diagnose or remove all barriers.

A closer look at beliefs, confidence, motivation, and plans

Theory of Planned Behavior: which beliefs shape intention?

TPB separates three questions.

Attitude: does the person regard the behavior as worthwhile or unpleasant?
Subjective norm: do they think important people expect or approve of it?
Perceived behavioral control: do they believe they can perform it under the circumstances?

These factors help explain intention. Intention and perceived control then help explain behavior, although actual control can limit what someone does.

Armitage and Conner's 2001 review combined 185 independent studies and reported 39% of variance in intention and 27% in behavior accounted for by the model. These are summaries of prediction across the included studies. They do not mean that a TPB intervention changes 27% of someone's behavior or succeeds for 27% of participants.

Social Cognitive Theory: confidence has sources

Self-efficacy is confidence in being able to perform a particular action. Bandura's account identifies information from successful performance, watching others, encouragement, and physiological states as sources of that confidence. In a training program, guided practice can give someone an actual success to judge; a demonstration can show how another person performs the task. Neither is equivalent to telling an untrained person to believe in themselves. Check the skill, the person's judgment of it, and the environment in which they must use it.

Sheeran and colleagues' 2016 review examined 204 experimental tests in which an intervention changed attitudes, norms, or self-efficacy relative to a control. Successful changes in these factors were accompanied by changes in health intentions and behavior. That provides intervention evidence about these particular targets. It is not a test of SCT as a complete system, and it does not mean every attempt to increase confidence will work.

Self-Determination Theory: why does the person want to act?

Two people can perform the same action for different reasons. One studies because the subject matters to their chosen career; another studies to avoid guilt or a threatened penalty. SDT distinguishes personally endorsed reasons from pressure, including pressure people impose on themselves. Enjoyment is one autonomous reason, but valuing an activity's purpose can also be autonomous even when the activity is not fun. The theory asks how autonomy, competence, and relatedness are supported.

A 2020 review by Sheeran and colleagues identified 65 independent tests of SDT interventions from 56 articles. The randomized health-behavior trials showed a small average benefit, with estimates reduced after adjustments for publication and small-sample bias. This supports testing concrete ways to support motivation, while setting modest expectations; it does not establish that SDT outperforms the other approaches in this table.

Implementation intentions: specify the moment and the action

“I intend to practice” leaves the occasion undecided. “If I finish lunch at home, then I will practice the opening passage for ten minutes” specifies a recognizable situation and a response. The useful design question is whether the cue will occur and the response will be feasible when it does.

Gollwitzer and Sheeran's 2006 meta-analysis covered 94 independent tests and found an average improvement in goal attainment. The review included initiation and other self-regulation tasks, such as protecting ongoing goal pursuit. Its average is evidence about those tests, not a forecast for a new program or proof that an if-then plan makes a demanding activity effortless.

What the Evidence Can Establish

Design and classification.

The original Behaviour Change Wheel study tested classification in tobacco control and obesity and called for research on whether using the wheel improves intervention design. BCTTv1 likewise provides evidence about identifying and reporting intervention content. Being good at describing an intervention does not tell you whether the intervention will help someone.

Theory use.

Prestwich and colleagues’ 2014 meta-analysis of 190 diet and physical activity interventions found generally weak and inconsistent relationships between theory use and intervention effectiveness. Programs reporting Social Cognitive Theory or the Transtheoretical Model were not more effective than programs with no reported theory basis. Putting a theory’s name on a program is no assurance of better results. These findings concern particular domains and imperfectly reported applications; they do not settle the value of every use of theory.

Stage matching.

A 2010 Cochrane smoking-cessation review found no clear advantage in four trials that directly compared stage-matched interventions with equivalent interventions without stage matching. That is narrower than saying all stage-based programs fail or that readiness never matters.

Health beliefs.

Carpenter’s 2010 review covered 18 studies. Perceived benefits and barriers were the strongest predictors among the variables examined, with results depending on outcome and timing. The result supports looking beyond threat alone. It provides no overall ranking against the other frameworks.

Fogg applications.

A 2025 scoping review included six intervention studies applying the Fogg Behavior Model. That is a small, varied literature. Useful applications are worth studying, but they do not yet tell us how much the model itself contributed or how broadly its predictions hold.

Effects at scale.

DellaVigna and Linos reported average effects of 8.7 percentage points in an academic trial sample and 1.4 percentage points across two nudge units. These were different collections of interventions. The comparison does not show that the same intervention loses five-sixths of its effect merely because it scales up.

How to choose a behavior change framework

Start before the framework: what outcome do you want, and which behavior could achieve it? Check whether that behavior fits the person’s abilities, interests, and circumstances. A beautifully organized intervention can still be aimed at an impractical action.

This site’s Behavior Market Fit argument puts behavior selection before the work of persuading people to perform it. If someone has no access to a swimming pool, another reminder to swim does not solve the problem. A different activity may be a better starting point.

Once you have a plausible behavior, identify the job: understanding barriers, predicting action, generating design options, or specifying intervention content. Then look for evidence relevant to your population and setting.

A practical decision guide

Your question	Approaches to consider	What to check
Why is this action difficult?	COM-B; Fogg for a focused look at motivation, ability, and prompts	Actual skills, resources, time, access, and desire. Do not assume the issue is low motivation.
What beliefs might explain intention?	Theory of Planned Behavior; Health Belief Model for relevant health beliefs	Whether the beliefs predict the specific behavior, and whether changing them changes action.
What role do confidence and motivation play?	Social Cognitive Theory; Self-Determination Theory	Confidence versus skill; personally endorsed reasons versus pressure.
How could this interface or decision be easier?	EAST; Fogg; choice architecture	The actual task and its friction. Enrollment and sustained participation are different outcomes.
How can someone remember to act at a particular moment?	Implementation intentions	A recognizable situation, a feasible response, and an intention to act.
How do we describe a multicomponent program?	BCT Taxonomy v1	What was actually delivered, by whom, and how often.

These are starting points for judgment. They have not been tested as rules for choosing the best framework for every problem.

Match the model to the kind of behavior

A one-time enrollment decision differs from taking an action every morning. Both differ from learning a difficult skill. An if-then plan may help someone begin a practice session; it does not supply the skill or effort needed during the session. For a sustained activity, examine whether the person can and wants to keep doing it, as well as what starts it.

For product designers, this often means studying the actual flow and making the task easier. For health or organizational teams, it may mean examining several barriers and specifying a multicomponent program. A job title alone does not determine which model to use.

One behavior, three different investigations

Suppose an adult wants to practice a musical instrument for ten minutes after lunch. This is a hypothetical comparison, not a tested combined intervention.

TPB: ask whether they expect practice to be useful, what social expectations matter, and whether they believe the schedule and task are manageable. Someone who sees no value in the exercise presents a different problem from someone who values it but expects interruptions.
SCT: watch them attempt the passage. Do they lack the technique, underestimate a skill they already have, or struggle because the task is too advanced? A demonstration and a simpler practice passage address different needs from reassurance alone.
SDT: ask why this instrument and this practice goal matter to them. A learner who chose the goal for personal enjoyment may need different support from one practicing only under family pressure. Offer meaningful choices within the lesson and explain the purpose of difficult exercises.

If lunch is unpredictable or the instrument is unavailable, investigate that practical problem too. None of these questions substitutes for checking whether the selected behavior fits the person's life.

How the frameworks can work together

COM-B may help structure an assessment. TPB, SCT or SDT may help formulate more specific hypotheses. EAST or MINDSPACE may support idea generation. BCTTv1 can describe the selected content, while an if-then plan may be one component. These are possible combinations, not a tested universal recipe.

For each choice, specify the proposed mechanism, delivery, comparison and outcome. Assess whether the intended population can carry out the action and whether the setting permits it. Measure implementation and unintended effects as well as the target behavior.

A correlation, proportion of explained variance, standardized mean difference and percentage-point change answer different questions. Do not put them into a single effectiveness ranking. Prefer evidence with comparable behaviors, populations, controls, outcomes and follow-up periods.

Common questions about behavior change frameworks

What is the best behavior change framework?

There is no supported overall winner in this comparison. COM-B can organize an assessment; Fogg can focus attention on an action’s immediate conditions; EAST offers design prompts. Choose for the problem and test the resulting intervention. The number of categories or citations is not an effectiveness score.

What is the difference between COM-B and the Fogg Behavior Model?

COM-B separates capability, opportunity, and motivation, and distinguishes reflective from automatic motivation. The Behaviour Change Wheel connects that model to intervention functions and policy categories. Fogg’s B=MAP brings motivation, ability, and a prompt together at the moment of action. Its emphasis on making an action easier can be useful in product design. The models organize some overlapping concerns differently; their labels are not interchangeable.

How many behavior change frameworks exist?

Davis and colleagues’ scoping review identified 82 theories of behavior and behavior change. The original Behaviour Change Wheel paper examined 19 intervention frameworks. Those are counts from particular reviews, not a current total of every framework in use.

Do behavior change frameworks actually work?

Ask what outcome was tested. Reliable classification, prediction of intention, and an improvement in behavior are different achievements. The theory-use and stage-matching reviews above give reasons to be skeptical of broad promises. A framework can still help you formulate a question or organize work; you need outcome evidence for the intervention you choose.

Sources and evidence

Richard H. Thaler and Cass R. Sunstein (2008): Nudge: Improving Decisions about Health, Wealth, and Happiness