Theory of Planned Behavior

Abstract

The theory of planned behavior is a social-cognitive model, formulated by Icek Ajzen, that predicts deliberate behavior from a person's intention to perform it, where intention is the weighted sum of three determinants: attitude toward the behavior, subjective norm, and perceived behavioral control. It extends the theory of reasoned action, which Ajzen developed with Martin Fishbein, by adding perceived behavioral control, which captures the perceived ease or difficulty of acting and predicts behavior both indirectly, through intention, and directly. The model is one of the most cited frameworks in health and social psychology, accounting in meta-analyses for roughly a quarter to two-fifths of the variance in intentions and behavior. Its central limitation, the intention-behavior gap, is the reliable finding that intentions translate only partially into action.

Keywords: theory of planned behavior, behavioral intention, perceived behavioral control, subjective norm, intention-behavior gap

The theory of planned behavior (TPB) is a framework for predicting and explaining volitional human behavior from a small set of cognitive antecedents (Ajzen, 1991). It holds that the proximal cause of a behavior is the intention to perform it, and that intention is itself determined by three independent constructs: how favorably the person evaluates the behavior (attitude), how much social pressure they perceive to perform it (subjective norm), and how much control they believe they have over performing it (perceived behavioral control). The theory matters because it makes behavior tractable: rather than treating action as the product of stable personality traits, it locates the causes in modifiable beliefs, which is why it has anchored thousands of behavior-change interventions in health, environmental, and consumer domains (Godin & Kok, 1996). It is also one of the most heavily tested theories in the behavioral sciences, and that scrutiny has exposed both its predictive reach and its systematic shortfalls (Sheeran & Webb, 2016).

What the Theory of Planned Behavior Is

The theory of planned behavior grew directly out of the theory of reasoned action, which Martin Fishbein and Icek Ajzen developed in the 1970s to predict behavior from attitude and subjective norm alone. Reasoned action worked well for behaviors under complete volitional control but failed when performance depended on skills, resources, or opportunities the actor did not fully command. Ajzen closed that gap by adding a third determinant, perceived behavioral control, and renaming the expanded model the theory of planned behavior (Ajzen, 1985). The definitive statement appeared in 1991 and specified the causal structure that every later test has used: beliefs feed the three determinants, the determinants combine into intention, and intention — together with actual control — produces behavior (Ajzen, 1991).

The model is deliberately parsimonious. It does not claim that attitudes, norms, and control are the only influences on behavior; it claims that every other influence — personality, demographics, past experience, emotion — operates through these constructs and their underlying beliefs. This is the theory's sufficiency assumption, and it is both its greatest strength and the target of its sharpest critiques (Ajzen, 2020). A background variable such as socioeconomic status changes behavior only insofar as it changes what a person believes about the behavior's consequences, what they think others expect, or how capable they feel of acting.

Perceived behavioral control is the construct that distinguishes the theory from its predecessor, and it does two jobs at once. As a determinant of intention, it captures motivation: people rarely intend to attempt what they believe they cannot do. But it also has a direct path to behavior, because to the extent that perceived control accurately reflects actual control, it stands in as a proxy for the real resources and obstacles that determine whether an intention can be carried out (Ajzen, 1991). That dual role is why the theory can predict behavior even when intention is held constant.

The Three Determinants of Intention

Intention is the theory's central variable, and it is modeled as a weighted linear combination of the three determinants. The relative weights are not fixed by the theory; they are estimated empirically and vary by behavior and population. Across a broad meta-analysis, attitude and perceived behavioral control tend to carry larger weights than subjective norm, which is often the weakest of the three predictors (Armitage & Conner, 2001). Figure 1 shows the canonical path structure.

The demonstration below makes the weighted-sum rule concrete. Each determinant is rated on a seven-point scale, and the predicted intention is their weighted average using representative meta-analytic weights.

Intention is a weighted average, not a simple sum: each determinant enters scaled by an empirically estimated weight. With the representative weights shown, a favorable attitude (6), mild social pressure (3), and fair control (5) yield a predicted intention of 4.85, with attitude supplying about 48% of the total and subjective norm only about 17% — the weak-norm pattern typical of a private behavior. Because the weights are estimated per behavior, the same three ratings can produce very different intentions for a socially visible act (Ajzen, 1991; Armitage & Conner, 2001).

The weights matter as much as the ratings. A behavior for which subjective norm carries little weight — private exercise, say — will barely move as social pressure changes, whereas a socially visible behavior may be dominated by it. Because the theory estimates rather than assumes these weights, it accommodates the fact that the same three constructs are assembled differently for different behaviors.

From Beliefs to Attitudes: The Expectancy-Value Foundation

Each of the three determinants is not a primitive; it rests on an expectancy-value substructure inherited from the theory of reasoned action (Ajzen, 1991). Attitude toward a behavior is the sum, across a person's salient behavioral beliefs, of the subjective probability that the behavior produces a given outcome multiplied by the evaluation of that outcome. Formally, attitude is Σ bᵢ eᵢ, where bᵢ is belief strength and eᵢ is outcome evaluation. Subjective norm has the parallel structure over normative beliefs (what specific referents think) weighted by motivation to comply, and perceived behavioral control over control beliefs (the presence of facilitating or impeding factors) weighted by their perceived power.

This substructure is what makes the theory useful for intervention. A campaign cannot change an attitude directly, but it can target the beliefs beneath it — either the perceived likelihood that a behavior leads to an outcome or the value attached to that outcome (Conner & Armitage, 1998). The demonstration below builds an attitude from three salient beliefs, showing how a single strongly negative belief can outweigh several mild positives.

attitude = Σ b×e = 16 (favorable)

Attitude is the sum of belief strengths weighted by how good or bad each outcome is. Because evaluations can be negative, a single strongly disliked outcome that the person thinks is likely can pull an otherwise positive attitude down or even negative — which is why a campaign changes attitude most efficiently by targeting the one belief carrying the largest adverse product, not by piling on new positives (Ajzen, 1991).

The expectancy-value model also explains why measured attitudes sometimes diverge from a person's global feeling about a behavior: the sum runs only over salient beliefs — those accessible in memory at the moment of judgment — so a belief that is true but not currently salient contributes nothing (Ajzen, 2020).

Intention and the Intention-Behavior Gap

Intention is the theory's proximal predictor of behavior, but the link is far from perfect. Meta-analyses of prospective studies place the intention-behavior correlation around 0.5, meaning intention explains roughly a quarter of the variance in later behavior — strong for a single predictor, but leaving most of the variance unexplained (Sheeran, 2002; McEachan et al., 2011). This shortfall is the intention-behavior gap, and its structure is asymmetric: almost everyone who does not intend to act does not act, but a large share of those who do intend nonetheless fail to follow through — the inclined abstainers (Sheeran & Webb, 2016).

The gap is not a flaw in measurement; it is where the interesting behavioral action lies. Perceived behavioral control narrows it, because an intention backed by genuine control is more likely to be enacted, which is exactly why the theory routes control directly to behavior (Ajzen, 1991). More recent work locates additional moderators in the properties of the intention itself — its stability, certainty, and accessibility — so that two people reporting the same intention strength can differ sharply in whether they act (Conner & Norman, 2022). The demonstration below models how realized behavior falls short of intention as actual control declines.

realized behavior B = I × (0.3 + 0.7 × C/100) = 40.8 · gap = 39.2 (large)

Intention sets a ceiling; actual control decides how much of it is reached. Holding intention at 80, low control (30) lets only 40.8 of it become behavior — a gap of roughly 39 — while raising control to 90 closes the gap to about 6, all without strengthening the intention itself. That is the model’s central lesson and the rationale for its direct control-to-behavior path: an intention backed by genuine control is the one that gets enacted (Sheeran, 2002; Sheeran & Webb, 2016).

Interventions that target the gap directly — implementation intentions, the if-then plans specifying when, where, and how to act that Gollwitzer introduced as a distinct self-regulatory mechanism (Gollwitzer, 1999) — can boost enactment without changing intention strength at all, which is strong evidence that the gap is a distinct causal stage rather than merely weak intention (Sheeran & Webb, 2016).

Predictive Validity Across Domains

The theory's empirical record is unusually large, and it is uneven across behaviors. Table 1 summarizes representative meta-analytic findings. The pattern is consistent: the three determinants predict intention well, intention and perceived behavioral control predict behavior moderately, and subjective norm is typically the weakest determinant (Armitage & Conner, 2001; McEachan et al., 2011).

Note. Representative values pooled across health-behavior meta-analyses (Armitage & Conner, 2001; McEachan et al., 2011); exact figures vary by behavior and study.

Worked Example

Consider a person deciding whether to start a regular exercise routine, with each determinant measured on a seven-point scale. Suppose attitude is 6 (they view exercising very favorably), subjective norm is 3 (they perceive only mild social pressure), and perceived behavioral control is 5 (they feel fairly capable of fitting it in). Using representative meta-analytic weights of 0.49 for attitude, 0.34 for subjective norm, and 0.43 for perceived behavioral control, the predicted intention is the weighted average:

intention = (0.49 × 6 + 0.34 × 3 + 0.43 × 5) / (0.49 + 0.34 + 0.43) = 6.11 / 1.26 ≈ 4.85.

The predicted intention of 4.85 on the seven-point scale is moderately strong. Decomposing the numerator shows attitude contributing about 48% of the total, perceived behavioral control about 35%, and subjective norm only about 17% — the weak-norm pattern typical of a private health behavior. Now suppose this person's actual control is low: an illustrative translation in which realized behavior is intention scaled by control gives, for an intention expressed as 80 on a 0–100 scale and actual control of 30, a realized behavior of 80 × (0.30 + 0.70 × 0.30) ≈ 40.8 — a gap of roughly 39 points between what they intend and what they do. Raising control to 90 shrinks the gap to about 6 points. The arithmetic makes the theory's central lesson concrete: strengthening intention helps, but without control the intention leaks away before it becomes action.

Discussion

The theory of planned behavior earned its dominance by being simple, measurable, and falsifiable in a field short on all three. Its constructs are operationalized by standardized elicitation and questionnaire procedures, its predictions are quantitative, and its sufficiency claim is testable — one can always ask whether an added variable improves prediction over the core three (Ajzen, 2020). That testability has produced a large and largely supportive literature, alongside a catalog of well-characterized limits.

Three critiques recur. First, the theory is a model of deliberation and says little about habitual or automatic behavior, where past behavior often out-predicts intention. Second, subjective norm is persistently the weakest determinant, which has prompted proposals to split it into injunctive and descriptive components or to add moral and self-identity constructs (Conner & Armitage, 1998). Third, and most fundamentally, the intention-behavior gap means that even a fully specified model predicts action only moderately, because intention is a necessary but insufficient cause (Sheeran, 2002). These limits led some critics to argue that the theory had reached the end of its usefulness and should be retired in favor of more dynamic models that incorporate automatic processes and change over time (Sniehotta et al., 2014). Defenders respond that the theory never claimed to predict behavior beyond the reach of intention, and that the gap is a substantive discovery the theory helped to locate rather than a refutation of it (Ajzen, 2020).

Current Directions

Contemporary work has shifted from asking whether the theory predicts behavior to asking when and how much, using methods that were not available when the model was first tested. Meta-analytic path analysis now pools correlation matrices across hundreds of studies to test the full model simultaneously rather than one path at a time, confirming the core structure while quantifying how much each path varies across behavioral domains (Hagger et al., 2016). A second line of work dissects intention itself: rather than treating a single self-report as the predictor, researchers measure intention stability, certainty, and strength, and find that unstable or weakly held intentions account for much of the gap — two respondents with identical intention scores can diverge in action because one intention is firm and the other provisional (Conner & Norman, 2022). A third strand targets the gap with self-regulatory tools such as implementation intentions, which convert a goal intention into an if-then plan and reliably raise enactment without altering the underlying motivation (Sheeran & Webb, 2016). Together these developments treat the theory less as a finished predictor and more as a scaffold onto which moderators of translation are being added.

Common Misconceptions

The theory predicts what people actually do.

Its proximal prediction is intention, not behavior. Intention predicts behavior only moderately (r ≈ 0.5), and the residual is the intention-behavior gap (Sheeran, 2002). Treating a strong intention as a guarantee of action is the single most common misreading of the model.

Perceived behavioral control is the same as actual control.

The construct is a perception. It substitutes for actual control only to the extent that the perception is accurate; when people misjudge the obstacles they face, perceived control predicts intention but not behavior (Ajzen, 1991).

Subjective norm just means what most people do.

In the original model, subjective norm is injunctive — what important referents think one ought to do, weighted by motivation to comply — not the descriptive norm of what others actually do. Conflating the two is one reason the norm construct has been proposed for expansion (Conner & Armitage, 1998).

Glossary

Attitude toward the behavior.

A person's overall favorable or unfavorable evaluation of performing the behavior, formed from salient behavioral beliefs and their outcome evaluations.

Behavioral belief.

A subjective probability that performing the behavior will lead to a given outcome; the building block of attitude.

Behavioral intention.

A person's readiness to perform a behavior, understood as the weighted sum of attitude, subjective norm, and perceived behavioral control; the theory's proximal predictor of action.

Control belief.

A belief about the presence of a factor that facilitates or impedes the behavior, weighted by its perceived power; the basis of perceived behavioral control.

Descriptive norm.

The perception of what most other people actually do, distinguished from the injunctive norm of what they approve of; a proposed extension to the norm construct.

Expectancy-value model.

The principle that an evaluation is the sum of beliefs each weighted by the value of what is believed; the substructure beneath all three determinants.

Implementation intention.

An if-then plan specifying when, where, and how a goal will be enacted; a self-regulatory tool that narrows the intention-behavior gap without changing intention strength.

Inclined abstainer.

A person who forms an intention to act but fails to carry it out; the asymmetric majority contribution to the intention-behavior gap.

Injunctive norm.

The perception of what important referents think one ought to do, weighted by motivation to comply; the original content of subjective norm.

Intention-behavior gap.

The reliable shortfall between what people intend and what they do; the theory's most-studied limitation.

Perceived behavioral control.

The perceived ease or difficulty of performing the behavior; the construct that distinguishes the theory from the theory of reasoned action, predicting behavior both through intention and directly.

Salient beliefs.

The limited set of beliefs accessible in memory at the moment of judgment, which alone contribute to a determinant's expectancy-value sum.

Subjective norm.

Perceived social pressure to perform or not perform the behavior, formed from normative beliefs weighted by motivation to comply.

Sufficiency assumption.

The claim that all external influences on behavior operate through the theory's core constructs, so adding other variables should not improve prediction.

Theory of reasoned action.

The predecessor model that predicted intention from attitude and subjective norm alone, for behaviors under full volitional control.

Key Researchers

Icek Ajzen (b. 1942). Professor Emeritus of Psychology at the University of Massachusetts Amherst; originated the theory of planned behavior by adding perceived behavioral control to the theory of reasoned action. Faculty Page - ORCID

Christopher J. Armitage. Professor of Health Psychology at the University of Manchester; authored the influential 2001 meta-analytic review establishing the theory's predictive validity across domains. Faculty Page - ORCID

Mark Conner. Professor of Applied Social Psychology at the University of Leeds; extended and meta-analytically tested the theory in health contexts and on the role of intention stability. Faculty Page - ORCID

Martin Fishbein (1936-2009). Professor at the University of Pennsylvania Annenberg School for Communication; co-developed the theory of reasoned action, the expectancy-value framework from which the theory of planned behavior grew. Wikipedia - Wikidata

Paschal Sheeran. Stephen J. Walsh Distinguished Professor of Psychology and Neuroscience at the University of North Carolina at Chapel Hill; quantified the intention-behavior gap and developed interventions to close it. Faculty Page - ORCID

Frequently Asked Questions

What is the theory of planned behavior in simple terms?
It is a model holding that deliberate behavior follows from the intention to perform it, and that intention is driven by three things: how favorable the behavior seems, how much social pressure surrounds it, and how much control the person feels over doing it (Ajzen, 1991).

How does the theory of planned behavior differ from the theory of reasoned action?
It adds a third determinant, perceived behavioral control, so that it can predict behaviors that are not under complete volitional control, whereas reasoned action used only attitude and subjective norm (Ajzen, 1985).

What are the three determinants of intention?
Attitude toward the behavior, subjective norm, and perceived behavioral control; intention is modeled as their weighted sum, with weights estimated empirically for each behavior (Ajzen, 1991).

How well does the theory predict behavior?
The three determinants explain about 40% of the variance in intention, and intention correlates with later behavior at roughly 0.5, explaining about a quarter of behavior variance (Armitage & Conner, 2001).

What is the intention-behavior gap?
It is the reliable finding that intentions translate only partly into action; most non-intenders do not act, but many intenders also fail to follow through (Sheeran, 2002).

Why is perceived behavioral control important?
It captures whether a person feels able to act and, to the extent it reflects real control, predicts behavior directly rather than only through intention (Ajzen, 1991).

What are the main criticisms of the theory?
That it explains deliberation but not habit, that subjective norm is a weak predictor, and that intention only moderately predicts behavior (Conner & Armitage, 1998).

How is the theory used to change behavior?
Interventions target the beliefs beneath the determinants, and self-regulatory tools such as implementation intentions target the translation of intention into action (Sheeran & Webb, 2016).

References