Does Changing Behavioral Intentions Engender Behavior Change? A Meta-Analysis of the Experimental Evidence

Abstract

Numerous theories in social and health psychology assume that intentions cause behaviors. However, most tests of the intention- behavior relation involve correlational studies that preclude causal inferences. In order to determine whether changes in behavioral intention engender behavior change, participants should be assigned randomly to a treatment that significantly increases the strength of respective intentions relative to a control condition, and differences in subsequent behavior should be compared. The present research obtained 47 experimental tests of intention-behavior relations that satisfied these criteria. Meta-analysis showed that a medium-to-large change in intention (d = 0.66) leads to a small-to-medium change in behavior (d = 0.36). The review also identified several conceptual factors, methodological features, and intervention characteristics that moderate intention-behavior consistency.

Full text

Does Changing Behavioral Intentions Engender Behavior Change?
A Meta-Analysis of the Experimental Evidence
Thomas L. Webb
The University of Manchester Paschal Sheeran
The University of Sheffield
Numerous theories in social and health psychology assume that intentions cause behaviors. However,
most tests of the intention–behavior relation involve correlational studies that preclude causal inferences.
In order to determine whether changes in behavioral intention engender behavior change, participants
should be assigned randomly to a treatment that significantly increases the strength of respective
intentions relative to a control condition, and differences in subsequent behavior should be compared.
The present research obtained 47 experimental tests of intention–behavior relations that satisfied these
criteria. Meta-analysis showed that a medium-to-large change in intention (d⫽0.66) leads to a
small-to-medium change in behavior (d⫽0.36). The review also identified several conceptual factors,
methodological features, and intervention characteristics that moderate intention–behavior consistency.
Keywords: intention, behavior change, intervention, meta-analysis
Intentions are self-instructions to perform particular behaviors or to
obtain certain outcomes (Triandis, 1980) and are usually measured by
endorsement of items such as “I intend to do X!” Forming a behav-
ioral or goal intention signals the end of the deliberation about what
one will do and indicates how hard one is prepared to try, or how
much effort one will exert, in order to achieve desired outcomes
(Ajzen, 1991; Gollwitzer, 1990; Webb & Sheeran, 2005). Intentions
thus are assumed to capture the motivational factors that influence a
behavior (Ajzen, 1991). Theories of attitude–behavior relations, mod-
els of health behavior, and goal theories all converge on the idea that
intention is the key determinant of behavior (summaries by Abraham,
Sheeran, & Johnston, 1998; Austin & Vancouver, 1996; Conner &
Norman, 1996; Eagly & Chaiken, 1993; Gollwitzer & Moskowitz,
1996; Maddux, 1999). However, reviews of intention–behavior rela-
tions to date have relied on correlational evidence and do not afford
clear conclusions about whether intentions have a causal impact on
behavior. The present review integrates for the first time experimental
studies that manipulate intention and subsequently followup behavior.
In so doing, the review quantifies the extent to which changes in
intention lead to changes in behavior across studies.
The Role of Intention in Theories of Social
and Health Behaviors
Models of attitude–behavior relations such as the theory of
reasoned action (Fishbein, 1980; Fishbein & Ajzen, 1975), the
theory of planned behavior (Ajzen, 1985, 1991; Ajzen & Madden,
1986), and the model of interpersonal behavior (Triandis, 1977,
1980) each accord intentions a key role in the prediction of
behavior. An important impetus for the development of these
models was a review by Wicker (1969) that showed that general
attitudes (e.g., X is good/bad) only weakly predicted specific
behaviors. Models of attitude–behavior relations attempted to ex-
plain this attitude–behavior discrepancy by pointing to the impor-
tance of measuring attitudes and behavior at the same level of
specificity and by elucidating how attitudes combine with other
factors to influence behavior. For instance, the theory of reasoned
action (TRA; Fishbein & Ajzen, 1975) proposes that two addi-
tional constructs are needed to explain the relationship between
attitude and behavior. First, a favorable attitude toward a behavior
might not be translated into action because of social pressure from
significant others not to perform the behavior. The theory therefore
suggests that measures of subjective norm (e.g., Most people who
are important to me think that I should/should not do X) should be
taken alongside attitude measures in order to capture both social
and personal influences on behavior. Second, Fishbein and Ajzen
suggested that attitudes and subjective norms affect behavior by
promoting the formation of a decision or intention to act. That is,
the TRA proposes that behavioral intention is the proximal deter-
minant of behavior and mediates the influence of both the theory’s
predictors (attitude and subjective norm) and external variables
(e.g., personality and demographic characteristics). Thus, accord-
ing to the TRA, intention is the most immediate and important
predictor of behavior.
The TRA was designed to predict volitional behaviors, or be-
haviors over which the individual has a good deal of control.
However, many behaviors require resources, skills, opportunities,
or cooperation to be performed successfully (Liska, 1984). Con-
sequently, a person may not (a) intend to perform a behavior unless
the behavioral performance is perceived as under personal control
or (b) enact their behavioral intention successfully unless they
possess actual control over the behavior. To take account of these
Thomas L. Webb, School of Psychological Sciences, The University of
Manchester; Paschal Sheeran, Department of Psychology, The University
of Sheffield.
We thank Amanda Rivis and Vikkie Buxton for coding the study
characteristics, Paul Norman and Richard Cooke for coding assessed
control, and Gaston Godin and Ian Kellar for providing considerable
additional information about their research.
Correspondence concerning this article should be addressed to
Thomas L. Webb, School of Psychological Sciences, The University of
Manchester, Oxford Road, Manchester M13 9PL, United Kingdom.
E-mail: thomas.webb@manchester.ac.uk
Psychological Bulletin Copyright 2006 by the American Psychological Association
2006, Vol. 132, No. 2, 249–268 0033-2909/06/$12.00 DOI: 10.1037/0033-2909.132.2.249
249
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.