Content uploaded by Icek Ajzen
Author content
All content in this area was uploaded by Icek Ajzen on Jun 17, 2014
Content may be subject to copyright.
Psychology and Health
Vol. 26, No. 9, September 2011, 1113–1127
EDITORIAL
The theory of planned behaviour: Reactions and reflections
Icek Ajzen*
Department of Psychology, University of Massachusetts, Amherst, MA 01003-9271, USA
(Received 9 August 2011; final version received 9 August 2011)
The seven articles in this issue, and the accompanying meta-analysis in
Health Psychology Review [McEachan, R.R.C., Conner, M., Taylor, N., &
Lawton, R.J. (2011). Prospective prediction of health-related behaviors
with the theory of planned behavior: A meta-analysis. Health Psychology
Review,5, 97–144], illustrate the wide application of the theory of planned
behaviour [Ajzen, I. (1991). The theory of planned behavior. Organizational
Behavior and Human Decision Processes, 50, 179–211] in the health domain.
In this editorial, Ajzen reflects on some of the issues raised by the different
authors. Among the topics addressed are the nature of intentions and the
limits of predictive validity; rationality, affect and emotions; past behaviour
and habit; the prototype/willingness model; and the role of such
background factors as the big five personality traits and social comparison
tendency.
Keywords: theory of planned behaviour; review; future directions
Introduction
Since its introduction 26 years ago (Ajzen, 1985), the theory of planned behaviour
(TPB; Ajzen, 1991, in press) has, by any objective measure, become one of the most
frequently cited and influential models for the prediction of human social behaviour.
Its popularity is revealed by conducting a Google Scholar search for the keyword
‘theory of planned behavior OR theory of planned behaviour.’ From 22 citations in
1985, the number of citations per year has grown steadily to a total of 4550 in 2010
(Figure 1). Relying on a variety of indices, Nosek et al. (2010) found that my
programme of research ranks as having the highest scientific impact score among US
and Canadian social psychologists.
Yet, for all its popularity, or perhaps because of it, the TPB has also been the
target of much criticism and debate. Some researchers reject it outright as an
adequate explanation of human social behaviour. These investigators tend to deny
the importance of consciousness as a causal agent (Wegner, 2002; Wegner &
Wheatley, 1999) and view much human social behaviour as driven by implicit
attitudes (Greenwald & Banaji, 1995) and other unconscious mental processes (Aarts
& Dijksterhuis, 2000; Bargh, 1989; Bargh & Chartrand, 1999; Brandsta
¨tter,
Lengfelder, & Gollwitzer, 2001; Uhlmann & Swanson, 2004). Most critics, however,
accept the theory’s basic reasoned action assumptions but question its sufficiency or
*Email: aizen@psych.umass.edu
ISSN 0887–0446 print/ISSN 1476–8321 online
ß2011 Taylor & Francis
DOI: http://dx.doi.org/10.1080/08870446.2011.613995
http://www.tandfonline.com
Downloaded by [University of Massachusetts, Amherst], [Icek Aizen] at 09:03 20 September 2011
inquire into its limiting conditions (for a discussion, see Fishbein & Ajzen, 2010
Chapter 9). The articles in the present issue, as well as the meta-analysis of TPB
research published in Health Psychology Review (McEachan, Conner, Taylor, &
Lawton, 2011), are largely in the latter vein and I will take this opportunity to try to
clarify aspects of the theory and to engage with some of the critical issues raised
by the different authors.
Limits of predictive validity
Even when all TPB constructs are carefully assessed, they contain random
measurement error. Well-designed measures of attitude towards a behaviour of
interest, subjective norm, perceived behavioural control, intention and behaviour
rarely exhibit reliabilities in excess of 0.75 or 0.80. It follows that, even with good
measures, the most we can reasonably expect in terms of correlations among the
theory’s constructs are coefficients of about 0.60. Past syntheses of TPB research
have shown that, even when studies with questionable measures are included in the
meta-analysis, the observed mean correlations approach their theoretical limits. For
example, in a synthesis of the results of several previous meta-analyses, Sheeran
(2002) reported a mean overall correlation of 0.53 between intention and behaviour;
the mean correlation between perceived behavioural control and intention was found
to be 0.40 in a meta-analysis by Armitage and Conner (2001); and in meta-analytic
reviews covering a broad range of different behaviours (Armitage & Conner, 2001;
Cheung & Chan, 2000; Notani, 1998; Rivis & Sheeran, 2003; Schulze & Wittmann,
2003), attitudes, subjective norms and perceived behavioural control produced mean
multiple correlations with intentions that ranged from 0.59 to 0.66.
Meta-analysis made by McEachan et al. (2011) produced comparable results.
Correlations of attitudes, subjective norms and perceptions of control with intentions
ranged from 0.40 to 0.57, producing a multiple correlation of 0.67. The intention–
behaviour correlation of 0.43 and the perceived control–behaviour correlation of
0.31 were somewhat lower than in previous meta-analysis, most likely due to the fact
that the present synthesis was restricted to prospective studies that assessed
behaviour at some time after administering the TPB survey.
0
500
1000
1500
2000
2500
3000
3500
4000
4500
5000
1985 1990 1995 2000 2005 2010
Figure 1. Number of citations of the TPB in Google Scholar.
1114 I. Ajzen
Downloaded by [University of Massachusetts, Amherst], [Icek Aizen] at 09:03 20 September 2011
The intention–behaviour correlation, though usually quite substantial, can vary
considerably. The meta-analysis by McEachan et al. points to one moderator of this
relation, the temporal distance between measurement of intention and observation
of behaviour. It stands to reason that, as time passes, an increasing number of
intervening events can change people’s behavioural, normative or control beliefs,
modify attitudes, subjective norms or perceptions of control, thus generating revised
intentions. Changes of this kind will tend to reduce the predictive validity of intentions
that were assessed before the changes took place. Consistent with this argument,
shorter intervals between assessment of intentions and observation of behaviour (5
weeks or less) were associated with stronger correlations than longer time intervals.
Direct evidence in support of the proposition that instability of intentions over time
can reduce their predictive validity was provided by Sheeran, Orbell, and Trafimow
(1999) and by Conner, Sheeran, Norman, and Armitage (2000).
However, intentions are sometimes found to be poor predictors of behaviour
even over relatively short time periods, as illustrated in the study by Kor and Mullan
(2011). Intentions were assessed with respect to three sleep-related behaviours in the
coming week: making bedroom/sleep environment restful, avoiding going to bed
feeling thirsty or hungry and avoiding anxiety and/or stress-provoking activities
before bedtime. One week later, the participants reported how often they had
performed each behaviour in the preceding week. Composite measures aggregated
across the three behaviours showed a correlation of only 0.17 between intention and
behaviour. (Perceived behavioural control predicted behaviour somewhat better,
with a correlation of 0.25.) A possible reason for the low intention–behaviour
correlation is revealed by the relatively strong effect of the participants’ general
capacity to override or inhibit impulses. Ability to inhibit responses, as assessed by
a visual Go/NoGo computer task, correlated 0.43 with behaviour. This finding
suggests that performance of the three sleep-related behaviours requires the ability to
self-regulate, an aspect of actual control over the behaviour. For example, many
people find it difficult to put distressing thoughts out of their minds and may
therefore be unable to avoid anxiety or stress-provoking activities before bedtime. In
the TPB, lack of actual control over a behaviour will tend to reduce the predictive
validity of intentions. The relatively low correlation between perceived behavioural
control and behaviour suggests that perceptions of control were not sufficiently
accurate to serve as a good proxy for actual control.
At its core, the TPB is concerned with the prediction of intentions. Behavioural,
normative and control beliefs as well as attitudes, subjective norms and perceptions
of behavioural control are assumed to feed into and explain behavioural intentions.
Whether intentions predict behaviour depends in part on factors beyond the
individual’s control, i.e. the strength of the intention–behaviour relation is
moderated by actual control over the behaviour. Barring methodological shortcom-
ings, a low intention–behaviour relation is a warning sign indicating that we may be
reaching the limits of reasoned action.
Affect, emotions and rationality in the TPB
Irrationality
A frequently voiced criticism of the TPB and other reasoned action models is that
they are too ‘rational,’ not taking sufficient account of cognitive and affective
Psychology and Health 1115
Downloaded by [University of Massachusetts, Amherst], [Icek Aizen] at 09:03 20 September 2011
processes that are known to bias human judgments and behaviour. It is true, of
course, that the TPB emphasises the controlled aspects of human information
processing and decision making. Its concern is primarily with behaviours that are
goal-directed and steered by conscious self-regulatory processes. This focus has often
been misinterpreted to mean that the theory posits an impassionate, rational actor
who reviews all available information in an unbiased fashion to arrive at a
behavioural decision. In reality, the theory draws a much more complex and nuanced
picture.
Importantly, there is no assumption in the TPB that behavioural, normative and
control beliefs are formed in a rational, unbiased fashion or that they accurately
represent reality. Beliefs reflect the information people have in relation to the
performance of a given behaviour, but this information is often inaccurate and
incomplete; it may rest on faulty or irrational premises, be biased by self-serving
motives, by fear, anger and other emotions, or otherwise fail to reflect reality.
Clearly, this is a far cry from a rational actor. However, no matter how people arrive
at their behavioural, normative and control beliefs, their attitudes towards the
behaviour, their subjective norms and their perceptions of behavioural control follow
automatically and consistently from their beliefs. It is only in this sense that
behaviour is said to be reasoned or planned. Even if inaccurate, biased or otherwise
irrational, our beliefs produce attitudes, intentions and behaviours consistent with
these beliefs (Geraerts et al., 2008).
Affect and emotions
Perhaps the most frequently mentioned biasing factors ostensibly neglected in the
TPB are affect and emotions (Conner & Armitage, 1998; Rapaport & Orbell, 2000;
Richard, de Vries, & van der Pligt, 1998; Wolff, Nordin, Brun, Berglund, & Kvale,
2011). This concern is based in part on the mistaken perception that the theory posits
a rational actor who is unaffected by emotions and in part on the standard
methodology that is typically used to operationalise the theory’s constructs. In the
TPB, affect and emotions enter in two ways. First, they can serve as background
factors that influence behavioural, normative and/or control beliefs. Thus, it is well
known that general moods can have systematic effects on belief strength and
evaluations. Compared to people in a negative mood state, people in a positive mood
tend to evaluate events (such as the consequences of a behaviour) more favourably
and to judge favourable events as more likely to occur (Forgas, Bower, & Krantz,
1984; Johnson & Tversky, 1983; Schaller & Cialdini, 1990). In addition, affective
states can also help to select the behavioural, normative and control beliefs that are
readily accessible in memory (Clark & Waddell, 1983; McKee, Wall, Hinson,
Goldstein, & Bissonnette, 2003). Thus, for example, McKee et al. (2003) reported
that, in a free-response elicitation session, participants in a negative mood state were
more likely to emit unfavourable beliefs about smoking compared to participants in
a positive mood state.
The research discussed above indicates that affect and emotions can have indirect
effects on intentions and behaviour by influencing the kinds of beliefs that are salient
in a given situation, as well as the strength and evaluative connotations of these
beliefs. However, it is often suggested that affect can influence behaviour in a more
direct fashion, and that this possibility is not sufficiently accounted for in the TPB.
1116 I. Ajzen
Downloaded by [University of Massachusetts, Amherst], [Icek Aizen] at 09:03 20 September 2011
Wolff et al. (2011) follow the lead of other investigators (Abraham & Sheeran, 2003;
Conner, Smith, & Mcmillan, 2003) who have argued that anticipated regret and,
more generally, anticipated affect can influence intentions and behaviour indepen-
dent of the other predictors in the TPB. Indeed, in a meta-analysis of 24 datasets,
Sandberg and Conner (2008) found that the inclusion of anticipated affect in the
prediction equation accounted for an additional 7% of the variance in intentions and
1% of the variance in behaviour.
From the perspective of the TPB, expectations that performing a behaviour will
lead to experiencing pain, pleasure, regret, fear, elation or other emotions are simply
behavioural beliefs, i.e. beliefs about the likely consequences of the behaviour, some
positive and others negative. It has been argued, however, that these kinds of
behavioural beliefs are not sufficiently represented in applications of the theory due
to the way in which salient beliefs are elicited (Conner & Armitage, 1998; Wolff et al.,
2011). In a typical elicitation session, participants are asked to list what they believe
to be the advantages and disadvantages of performing a behaviour under
investigation. This question tends to elicit instrumental rather than experiential or
affective consequences. However, there is nothing in the TPB that requires a focus
on instrumental outcomes. For example, in a study on five leisure activities,
Ajzen and Driver (1991) asked participants in formative research to list the benefits
and costs of each leisure activity as well as the things they liked and disliked about
each activity. The instrumental beliefs were found to predict an instrumental attitude
measure (e.g. useful–useless) better than an experiential measure (e.g. interesting–
boring), and the reverse was true for the affective beliefs. The two types of attitude
made independent contributions to the prediction of intentions. However,
equally strong predictions were obtained when the two attitudes were combined
into a single measure, suggesting that they can usefully be considered parts of the
same attitude.
In a recent study, Ajzen and Sheikh (in press) explored a possible reason for the
finding that anticipated affect adds strongly and independently to the prediction of
intentions. A review of the literature revealed a peculiar feature of most studies
dealing with the role of anticipated affect. Whereas the basic variables in the TPB are
assessed with respect to performing a behaviour of interest, anticipated affective
reactions are usually measured in relation to not performing the behaviour (Fishbein
& Ajzen, 2010, Chapter 9). In fact, some investigators have made this explicit in their
definition of anticipated affect. Thus, according to Abraham and Sheeran (2003),
‘Anticipated regret refers to beliefs about whether or not feelings of regret or upset
will follow from inaction’ (p. 496, emphasis added). In the study reported by
Abraham and Sheeran, participants expressed their attitudes, subjective norms,
perceived control and intentions with regard to exercising on a regular basis in the
next 2 weeks, but they were asked how much they would regret it and how upset they
would be if they did not exercise regularly in the next 2 weeks. If, as suggested by the
TPB, measures of anticipated affect are (partial) measures of attitude, then it can be
argued that studies on anticipated affect actually assess two kinds of attitude: a
general attitude towards enacting a given behaviour and an affective attitude
towards not performing the behaviour. Because not doing is not necessarily the
opposite of doing (Richetin, Conner, & Perugini, 2011), this focus on action as well
as inaction may be sufficient to account for the residual predictive validity of
anticipated emotions, irrespective of whether the alternative attitude is affective in
nature or not.
Psychology and Health 1117
Downloaded by [University of Massachusetts, Amherst], [Icek Aizen] at 09:03 20 September 2011
The study by Ajzen and Sheikh (in press) provided strong support for this
argument. It showed that when all variables in the TPB as well as anticipated affect
were assessed either with respect to performing a behaviour (drinking, eating fast
food) or with respect to avoiding the behaviour, anticipated affect made no
independent contribution to the prediction of intentions. Only when anticipated
affect was measured with respect to one alternative (action or inaction) and the TPB
variables were assessed with respect to the other alternative was there a significant
residual effect for anticipated affect. The study also demonstrated that this effect
does not require a measure of affect. When the criterion was intention to enact a
behaviour (drinking, eating fast food), adding measures of attitude, subjective norm,
and perceived behavioural control with respect to avoiding the behaviour signifi-
cantly increased explained variance just as did anticipated affect associated with
avoiding the behaviour. By the same token, when intention to avoid the behaviour
was the criterion, adding the TPB predictors in relation to enacting the behaviour
made a significant contribution comparable to anticipated affect with respect to
enacting the behaviour. In other words, it was not the addition of a measure of
anticipated affect that improved prediction of intentions but rather the addition of
measures addressing the alternative to the behaviour.
The results of the study by Wolff et al. (2011) are quite consistent with these
findings. Trying to predict intentions to take a genetic screening test for a
hypothetical disease, Wolff et al. assessed instrumental beliefs regarding this
behaviour as well as anticipated affect. Importantly, the latter measure had to do
with the affective consequences of taking the test, not with anticipated regret or other
emotions that might result from not taking the test. Consistent with my argument,
these affective beliefs did not emerge as a separate factor in a factor analysis; no
attempt was made to test whether they make a significant independent contribution
to the prediction of intentions.
Measurement context
The affective state of participants may be considered part of the measurement
context. The intention–behaviour relation can be disrupted if participants experience
one affective state when their intentions are being assessed and another when they
perform the behaviour. According to the TPB, readily accessible behavioural,
normative and control beliefs provide the cognitive foundation for attitudes,
subjective norms and perceived control, respectively. We saw earlier that affective
states can influence the behavioural, normative and control beliefs that are readily
accessible. When different beliefs are activated in the survey context and in the
behaviour contexts, they will produce different attitudes, subjective norms and/or
perceptions of control, resulting in different intentions. The intention assessed at the
survey stage will then be a relatively poor predictor of actual behaviour because a
different intention is active in the behavioural context (see Ajzen & Sexton, 1999 for
a discussion).
The study by Cooke and French (2011) nicely illustrates a related issue: the effect
of measurement context on the relative importance of attitudes, subjective norms
and perceptions of control. Again, as the context of measurement changes – from bar
to library in this study – the kinds of behavioural, normative and control beliefs that
are activated can also change. It stands to reason that with respect to binge drinking,
1118 I. Ajzen
Downloaded by [University of Massachusetts, Amherst], [Icek Aizen] at 09:03 20 September 2011
normative beliefs become especially accessible when drinking alcohol (e.g. in a bar),
compared to a neutral context (a library). Consistent with this argument, subjective
norms contributed more to the prediction of intentions to binge drink when
assessment occurred in a bar than when it was obtained in a library. In this study, the
investigators obtained direct measures of the TPB predictors. It would be interesting
to demonstrate more directly that context affects beliefs by eliciting behavioural,
normative and control beliefs in the two different contexts.
The sufficiency assumption
Several of the studies in this issue address, at least in part, an issue known as the
TPB’s sufficiency assumption. According to the theory, we should be able to predict
performance of a behaviour from intentions to perform the behaviour and from
perceived behavioural control. Intentions, in turn, should be predictable from
attitude towards the behaviour, subjective norm and perceived behavioural control.
Addition of other variables should not improve prediction of either intention or
behaviour. Generally speaking, the TPB does indeed permit quite accurate prediction
of intentions and behaviour, often – as I argued earlier – coming close to the
theoretical limit. Nevertheless, it has been proposed that the constructs contained in
the theory may not be sufficient to fully explain people’s intentions and actions
(Conner & Armitage, 1998). Indeed, one of the most frequently addressed questions
in research with the theory has to do with the prospect of increasing the amount of
explained variance in intentions or behaviour by adding one or more predictors.
In earlier treatments of the theories of reasoned action and planned behaviour
(Ajzen, 1991; Ajzen & Fishbein, 1980), the possibility of adding more predictors was
explicitly left open. In fact, the TPB was developed in this fashion by adding
perceived behavioural control to the original theory of reasoned action and, more
recently, by adding descriptive norms to the normative component (Fishbein &
Ajzen, 2010). Proceeding in this vein, Kor and Mullan (2011) as well as Norman and
Cooper (2011) investigated the role of past behaviour, and the latter also examined
the related habit construct; Rivis, Sheeran, and Armitage (2011) asked whether
prototype similarity affords predictive validity over and above intentions; Wolff
et al. (2011) included uncertainty avoidance motive in the prediction equation;
Hassandra et al. (2011) considered the role of self-concept; and Kor and Mullan
added perceived autonomy support.
For the sake of parsimony, additional predictors should be proposed and added
with caution, and only after careful deliberation and empirical exploration. Fishbein
and Ajzen (2010, Chapter 9) described some of the criteria that should be met by any
proposed addition to the theory. First, like the theory’s existing predictors, the
proposed variable should be behaviour-specific, conforming to the principle of
compatibility. That is, it should be possible to define and measure the proposed
factor in terms of the target, action, context and time elements that describe the
behavioural criterion. Second, it should be possible to conceive of the proposed
variable as a causal factor determining intention and action. Third, the proposed
addition should be conceptually independent of the theory’s existing predictors.
Fourth, the factor considered should potentially be applicable to a wide range of
behaviours studied by social scientists. Finally, the proposed variable should
Psychology and Health 1119
Downloaded by [University of Massachusetts, Amherst], [Icek Aizen] at 09:03 20 September 2011
consistently improve prediction of intentions or behaviour if it is to be made part of
the theory.
Past behaviour and habit
The dictum that ‘past behavior is the best predictor of future behavior’ is supported
by much empirical evidence. The finding of a strong correlation between past and
later behaviour attests to the temporal stability of the particular behaviour and its
antecedents. Moreover, it is often found that a measure of past behaviour
contributes to the prediction of future behaviour even after the predictors in the
TPB have been accounted for (Ajzen, 1991; Conner & Armitage, 1998; Ouellette &
Wood, 1998). For example, in a secondary analysis of data reported by Abraham
and Sheeran (2003), the amount of variance in physical activity explained by the TPB
increased from 36% to 53% with the addition of past physical activity. As mentioned
earlier, Kor and Mullan (2011) predicted a composite of three sleep-supportive
behaviours. They reported a correlation of 0.32 between past behaviour and
behaviour measured 1 week later. Although not as strong as in other studies, it
exceeded the intention–behaviour correlation (r¼0.17) as well as the correlation
between perceived control and behaviour (r¼0.25). In fact, when past behaviour
was added to the regression equation, the regression coefficients for intention and
perceived control were no longer significant.
One possible reason for findings of this kind is methodological, having to do with
the nature of the intention and behaviour measures. Whereas behaviour at the two
time points was assessed in terms of frequency of performance (e.g. ‘Over the past
week, how many days did you make your bedroom/sleep environment restful?’), the
measures of intention and perceived control relied on assessments of the likelihood
or subjective probability of performing the three behaviours (e.g. ‘I intend to make
my bedroom/sleep environment restful over the next week’, strongly disagree–
strongly agree; ‘I am confident that I can avoid having anxiety and stress provoking
activity before bedtime everyday’, strongly disagree–strongly agree). There was thus
greater scale compatibility (Courneya & McAuley, 1993) between the measures of
past and future behaviour than between the measures of intention and perceived
control on one hand and either past or future behaviour on the other. The greater
shared method variance between measures of past and later behaviour may have
been at least in part responsible for the relatively strong correlation between them.
Some support for this argument was also reported by Conner, Warren, Close, and
Sparks (1999).
It should be noted that past behaviour fails to meet one of the criteria for
inclusion in the TPB, namely the requirement that it constitute a causal antecedent of
intention. It is difficult to argue that the performance of a behaviour in the past
directly causes a person’s current intention. Instead, past behaviour is usually
considered a proxy for habit strength: The more frequently a behaviour has been
performed in a stable context, the more it is said to habituate and come under the
direct control of external stimulus cues at the expense of intentions (Ajzen, 2002).
Consistent with this argument, Norman and Cooper (2011), studying breast self-
examination, obtained an independent measure of habit strength in addition to
assessing past behaviour as well as the extent to which the behaviour is being
performed in a stable context, a condition required for habituation. This approach is
1120 I. Ajzen
Downloaded by [University of Massachusetts, Amherst], [Icek Aizen] at 09:03 20 September 2011
greatly superior to much previous research in which it has simply been assumed that
frequency of past behaviour can be equated with habit strength. Unfortunately,
contrary to previous findings, in this study frequency of past behaviour did not
contribute independently to the prediction of later behaviour and it was thus
impossible to test the hypothesis that its effect is mediated by habit strength. In fact,
the results of this study were quite weak. Even after adding past behaviour, context
stability and the interaction between these two variables, the TPB accounted for only
15% of the variance in future breast self-examination.
From the perspective of the TPB, more interesting than the effect of past on
future behaviour is the frequently reported finding that a measure of past behaviour
contributes independently to the prediction of intentions, over and above attitudes,
subjective norms and perceived behavioural control. In three meta-analytic syntheses
(Albarracı
´n, Johnson, Fishbein, & Muellerleile, 2001; Rise, Sheeran, & Skalle, 2006;
Sandberg & Conner, 2008), the addition of past behaviour to the prediction equation
raised the proportion of explained variance in intentions by between 9.65% and
13%. One possible interpretation of such findings is that the TPB’s sufficiency
assumption is invalid. In other words, intentions may be determined not only by
attitudes, norms and perceived control but also by one or more additional variables,
and these additional variables are captured, at least in part, by measures of past
behaviour. This explanation implies that if we could identify and assess these
additional variables, then the direct residual effect of past behaviour on intentions
would disappear. Fishbein and Ajzen (2010, Chapter 9) examined the possibility that
the two most frequently proposed additions to the theory – self-identity and
anticipated affect – constitute the missing components. After reviewing the available
research, they concluded that although each of these variables has been found to add
predictive validity (see also Hassandra et al., 2011), neither could account for the
residual effect of past behaviour on intentions. This issue is still unresolved, begging
for additional research.
Prototype similarity vs. intention
According to the prototype/willingness model (Gibbons, Gerrard, Blanton, &
Russell, 1998), the reasoned action processes described in the TPB are only one
possible path to arrive at a behaviour. In the second path, behaviour is more
spontaneous, reactive on the immediate situation and heavily influenced by perceived
similarity to a behavioural prototype. When people find themselves in situations that
encourage certain behaviours, especially risk-taking behaviours such as smoking, it is
not their preconceived intentions that determine their actions but rather their
willingness to engage in the behaviours, i.e. their openness to the opportunity. Their
willingness, in turn, is determined by the extent to which they see themselves as
similar to the prototypical person who performs the behaviours in question.
The idea that a relatively spontaneous mode of operation stands in contrast to the
TPB rests on a misunderstanding of reasoned action. There is no assumption in the
TPB that people carefully and systematically review all available information before
they form an intention to engage in a behaviour. On the contrary, the theory
recognises that most behaviours in everyday life are performed without much
cognitive effort. Consistent with contemporary theorising in social psychology
(Carver & Scheier, 1998; Chaiken & Trope, 1999; Petty & Cacioppo, 1986), it is
Psychology and Health 1121
Downloaded by [University of Massachusetts, Amherst], [Icek Aizen] at 09:03 20 September 2011
assumed that the amount of information processing that people engage in prior to
performing a behaviour varies along a continuum, from shallow to deep (Ajzen &
Sexton, 1999). In-depth processing is reserved for important decisions and behaviours
in novel situations that demand careful consideration of the behaviour’s likely conse-
quences, the normative expectations of significant others, and the obstacles that may
be encountered. When it comes to routine, everyday behaviours like eating breakfast,
taking one’s vitamin supplements, going to work, watching the news on TV and so
forth, no careful deliberation is required or postulated. Attitudes, subjective norms
and perceptions of control as well as intentions in relation to these kinds of behaviours
are assumed to guide behaviour implicitly without cognitive effort and often below
conscious awareness (see Ajzen & Fishbein, 2000, for a discussion of these issues).
The major remaining questions in relation to the prototype/willingness model,
then, are whether a measure of willingness predicts spontaneous kinds of behaviour
better than a measure of intention, and whether including perceived similarity to a
behavioural prototype in the TPB improves prediction. In their discussion of the first
question, Fishbein and Ajzen (2010, Chapter 2) concluded that the empirical
evidence neither gives an advantage to willingness over intention, nor supports the
idea that adding a measure of willingness improves prediction of behaviour.
Behavioural intentions are indications of a person’s readiness to perform a
behaviour. This readiness to act can be operationalised by asking whether people
intend to engage in the behaviour, expect to engage in the behaviour, are planning to
engage in the behaviour, will try to engage in the behaviour, and indeed, whether
they are willing to engage in the behaviour. These various expressions of behavioural
readiness are best considered manifest indicators reflective of the same latent
underlying construct, i.e. intention. The results of the study reported by Matterne,
Diepgen, and Weisshaar (2011) are consistent with this line of reasoning. Comparing
the TPB and the prototype/willingness model, the investigators found a strong
correlation between their measures of willingness and intention to adopt skin
protection measures, but intention was a better predictor of behaviour (r¼0.49) than
was willingness (r¼0.36).
Rivis et al. (2011) addressed the second question. Using a within-subjects design,
they examined the predictive validity of prototype similarity for a set of 14 health-
related behaviours. According to the prototype/willingness model, perceived
prototype similarity should have a direct impact on behaviour, unmediated by
intention. In fact, the data showed that perceived prototype similarity significantly
improved prediction of behaviour over and above the predictive validity of
intentions, accounting for an additional 6% of the variance. However, the measure
of prototype similarity or identification in this study was virtually identical to a
measure of self-reported behaviour. The question employed, ‘In general, how similar
are you to the type of person your age who (performs behaviour x)?’ is likely to
produce much the same information as asking, ‘Are you the kind of person who
performs behaviour x?’; i.e. ‘do you perform behaviour x?’ Clearly, this could
account for the observed relation between perceived prototype similarity and
behaviour. Unfortunately, it is not clear how one can obtain a measure of prototype
similarity that is clearly different from self-reported behaviour. Recognising this
difficulty, Rivis et al. (2011) refer to prior research in which prototype similarity had
an effect on behaviour even after controlling for a measure of past behaviour. Their
argument would be more convincing if the investigators had controlled for past
behaviour in their own study.
1122 I. Ajzen
Downloaded by [University of Massachusetts, Amherst], [Icek Aizen] at 09:03 20 September 2011
Wolff et al. (2011) adapted a general attitude towards uncertainty scale to their
specific context, i.e. uncertainty connected to taking a genetic screening test. As such,
this scale constituted an alternative measure of attitude towards taking such a test
and it thus fails the criterion that factors added to the TPB should be conceptually
independent of the existing constructs. As it turned out, this attitude measure was
superior to one based on the summed products of behavioural beliefs and outcome
evaluations in that it was the best predictor of intentions. However, it should be
noted that this study raises concerns regarding compatibility among the measures
employed. Whereas intentions to take a genetic screening test specified two context
elements (taking the test on one’s own initiative and on the advice of a physician),
measures of the other TPB constructs did not mention these elements.
The concept of ‘perceived autonomy support’ proposed by Kor and Mullan
(2011) as a possible addition to the TPB in the context of sleep-related behaviours
also fails to meet the criterion that it be conceptually independent of the other
predictors in the model. The sample scale item provided (‘People who are important
to me provide me with choice and options with regard to my sleep habits’) suggests
that this construct is part of the theory’s normative component. When included in
the prediction equation, its regression weight was not significant and it failed to
substantially improve prediction of intentions.
Background factors
In the TPB, the most detailed substantive information about the determinants of a
behaviour is contained in a person’s behavioural, normative and control beliefs. The
theory does not specify where these beliefs originated; it merely points to a host of
possible background factors that may influence the beliefs people hold – factors of
a personal nature such as personality and broad life values; demographic variables
such as education, age, gender and income; and exposure to media and other sources
of information. Factors of this kind are expected to influence intentions and
behaviour indirectly by their effects on the theory’s more proximal determinants.
Most empirical studies assess a few demographic characteristics if only considered as
control variables. Some studies, however, focus on one or more background factors
that, for intuitive or theoretical reasons, are considered to be relevant to the
behaviour under investigation.
A good case in point is reported by the study of Manning and Bettencourt (2011).
The investigators used the TPB as their conceptual framework to examine adherence
to a medical regimen. Unlike Kor and Mullan (2011) who dealt with their behavioural
category of sleep-related activities by assessing the TPB constructs in relation to each
behaviour, Manning and Bettencourt aggregated several regimen adherence behav-
iours and then assessed the TPB constructs with reference to the category as a whole.
Intentions to adhere were predicted very well, but the theory accounted for only a
small proportion of variance in behaviour, perhaps due to the long time lag between
the TPB survey and the behaviour. In addition to measuring the TPB constructs, the
investigators also assessed depressive symptoms as a possibly relevant background
factor. Degree of depression correlated negatively with intentions and reported
adherence to the medical regimen. However, consistent with the TPB, these effects of
depressive symptoms were found to be mediated by the theory’s predictors.
As have other investigators in the past (Courneya, Bobick, & Schinke, 1999;
Rhodes & Courneya, 2003), Rivis et al. (2011) examined the role of the ‘big five’
Psychology and Health 1123
Downloaded by [University of Massachusetts, Amherst], [Icek Aizen] at 09:03 20 September 2011
personality traits – openness, conscientiousness, extroversion, agreeableness and
neuroticism (Costa & McCrae, 1985) in the context of the TPB. In addition, they also
assessed the general tendency to compare oneself to important others. Interestingly,
however, rather than postulating a simple effect of these kinds of background factors
on intentions and behaviour, their within-subjects methodology allowed them to
examine the possibility that these variables influence the predictive validity of
intentions relative to perceived prototype similarity (see the earlier discussion of this
construct). Although the effects of the big five personality traits and social
comparison tendency were quite small, this investigation – like earlier studies by
Trafimow and Finlay (1996), Sheeran, Norman, and Orbell (1999) and others –
shows that there may be stable individual differences that influence the relative
weights of the different predictors in the TPB.
Discussion
Research on the TPB has made considerable progress since the theory was introduced
some two dozen years ago. Initial studies were mostly attempts to test the theory’s
predictive validity in various behavioural domains. The combined weight of much
empirical evidence, perhaps best captured in such meta-analytic syntheses as the one
included in the current set of articles, lends clear support to the theory. Satisfied that
the TPB does in fact predict intentions and behaviour quite well, investigators turned
their attention to more sophisticated questions, although straight-forward applica-
tions to new behaviours or behaviours in novel settings continue to appear in print.
The questions raised in the current issue are representative of some of the questions
that occupy contemporary investigators. Among other things, the present studies
were designed to gain a better understanding of the role of automatic or spontaneous
processes involved in habitual behaviour, processes that may be in play side-by-side
with more reasoned modes of operation; to explore impulsivity and the ability to
inhibit it when required for self-regulation; to examine the utility of making detailed
plans as a way to improve ability to act on intentions; to test the ideas that adding
anticipated affect or the motive to avoid uncertainty may improve prediction of
intentions; to demonstrate individual differences in the relative weights assigned to
the predictors in the TPB; and to study the role of such background factors as
personality traits and depression. I have tried to show that some of these variables and
processes, such as willingness to perform a behaviour or social support that appear to
go beyond the TPB can actually be accommodated within it, whereas others, such as
habit formation and various background factors, can expand and enrich our
understanding of human social behaviour.
References
Aarts, H., & Dijksterhuis, A. (2000). Habits as knowledge structures: Automaticity in goal-
directed behavior. Journal of Personality and Social Psychology, 78, 53–63.
Abraham, C., & Sheeran, P. (2003). Acting on intentions: The role of anticipated regret.
British Journal of Social Psychology, 42, 495–511.
Ajzen, I. (1985). From intentions to actions: A theory of planned behavior. In J. Kuhl &
J. Beckman (Eds.), Action-control: From cognition to behavior (pp. 11–39). Heidelberg:
Springer.
1124 I. Ajzen
Downloaded by [University of Massachusetts, Amherst], [Icek Aizen] at 09:03 20 September 2011
Ajzen, I. (1991). The theory of planned behavior. Organizational Behavior and Human Decision
Processes, 50, 179–211.
Ajzen, I. (2002). Residual effects of past on later behavior: Habituation and reasoned action
perspectives. Personality and Social Psychology Review, 6, 107–122.
Ajzen, I. (in press). The theory of planned behavior. In P.A.M. van Lange, A.W. Kruglanski,
& E.T. Higgins (Eds.), Handbook of theories of social psychology. London: Sage.
Ajzen, I., & Driver, B.L. (1991). Prediction of leisure participation from behavioral,
normative, and control beliefs: An application of the theory of planned behavior. Leisure
Sciences, 13, 185–204.
Ajzen, I., & Fishbein, M. (1980). Understanding attitudes and predicting social behavior.
Englewood-Cliffs, NJ: Prentice-Hall.
Ajzen, I., & Fishbein, M. (2000). Attitudes and the attitude-behavior relation: Reasoned and
automatic processes. In W. Stroebe & M. Hewstone (Eds.), European review of social
psychology (Vol. 11, pp. 1–33). Chichester: Wiley.
Ajzen, I., & Sexton, J. (1999). Depth of processing, belief congruence, and attitude-behavior
correspondence. In S. Chaiken & Y. Trope (Eds.), Dual-process theories in social
psychology (pp. 117–138). New York, NY: Guilford.
Ajzen, I., & Sheikh, S. (in press). Action versus inaction: Anticipated affect in the theory of
planned behavior. Journal of Applied Social Psychology.
Albarracı
´n, D., Johnson, B.T., Fishbein, M., & Muellerleile, P.A. (2001). Theories of reasoned
action and planned behavior as models of condom use: A meta-analysis. Psychological
Bulletin, 127, 142–161.
Armitage, C.J., & Conner, M. (2001). Efficacy of the theory of planned behavior: A meta-
analytic review. British Journal of Social Psychology, 40, 471–499.
Bargh, J.A. (1989). Conditional automaticity: Varieties of automatic influence in social
perception and cognition. In J.S. Uleman & J.A. Bargh (Eds.), Unintended thought
(pp. 3–51). New York, NY: Guilford.
Bargh, J.A., & Chartrand, T.L. (1999). The unbearable automaticity of being. American
Psychologist, 54, 462–479.
Brandsta
¨tter, V., Lengfelder, A., & Gollwitzer, P.M. (2001). Implementation intentions and
efficient action initiation. Journal of Personality and Social Psychology, 81, 946–960.
Carver, C.S., & Scheier, M.F. (1998). On the self-regulation of behavior. Cambridge:
Cambridge University Press.
Chaiken, S., & Trope, Y. (Eds.). (1999). Dual-process theories in social psychology. New York,
NY: Guilford Press.
Cheung, S.-F., & Chan, D.K.-S. (2000). The role of perceived behavioral control in predicting
human behavior: A meta-analytic review of studies on the theory of planned behavior.
Unpublished manuscript, Chinese University of Hong Kong.
Clark, M.S., & Waddell, B.A. (1983). Effects of moods on thoughts about helping, attraction
and information acquisition. Social Psychology Quarterly, 46, 31–35.
Conner, M., & Armitage, C.J. (1998). Extending the theory of planned behavior: A review and
avenues for further research. Journal of Applied Social Psychology, 28, 1429–1464.
Conner, M., Sheeran, P., Norman, P., & Armitage, C.J. (2000). Temporal stability as a
moderator of relationships in the theory of planned behaviour. British Journal of Social
Psychology, 39, 469–493.
Conner, M., Smith, N., & Mcmillan, B. (2003). Examining normative pressure in the theory of
planned behaviour: Impact of gender and passengers on intentions to break the speed
limit. Current Psychology: Developmental, Learning, Personality, Social, 22, 252–263.
Conner, M., Warren, R., Close, S., & Sparks, P. (1999). Alcohol consumption and the theory
of planned behaviour: An examination of the cognitive mediation of past behavior.
Journal of Applied Social Psychology, 29, 1675–1703.
Cooke, R., & French, D. P. (2011). The role of context and timeframe in moderating relation-
ships within the theory of planned behaviour. Psycholoogy and Health, 26, 1225–1240.
Psychology and Health 1125
Downloaded by [University of Massachusetts, Amherst], [Icek Aizen] at 09:03 20 September 2011
Costa Jr, P.T., & McCrae, R.R. (1985). The NEO Personality Inventory manual. Odessa,
FL: Psychological Assessment Resources.
Courneya, K.S., Bobick, T.M., & Schinke, R.J. (1999). Does the theory of planned behavior
mediate the relation between personality and exercise behavior? Basic and Applied Social
Psychology, 21, 317–324.
Courneya, K.S., & McAuley, E. (1993). Predicting physical activity from intention: Conceptual
and methodological issues. Journal of Sport and Exercise Psychology, 15, 50–62.
Fishbein, M., & Ajzen, I. (2010). Predicting and changing behavior: The reasoned action
approach. New York, NY: Psychology Press.
Forgas, J.P., Bower, G.H., & Krantz, S.E. (1984). The influence of mood on perceptions of
social interactions. Journal of Experimental Social Psychology, 20, 497–513.
Geraerts, E., Bernstein, D.M., Merckelbach, H., Linders, C., Raymaeckers, L., & Loftus, E.F.
(2008). Lasting false beliefs and their behavioral consequences. Psychological Science, 19,
749–753.
Gibbons, F.X., Gerrard, M., Blanton, H., & Russell, D.W. (1998). Reasoned action and social
reaction: Willingness and intention as independent predictors of health risk. Journal of
Personality and Social Psychology, 74, 1164–1180.
Greenwald, A.G., & Banaji, M.R. (1995). Implicit social cognition: Attitudes, self-esteem, and
stereotypes. Psychological Review, 102, 4–27.
Hassandra, M., Vlachopoulos, S.P., Kosmidou, E., Hatzigeorgiadis, A., Goudas, M., &
Theodorakis, Y. (2011). Predicting students intention to smoke by theory of planned
behaviour variables and parental influences across school grade levels. Psychology and
Health, 26, 1241–1258.
Johnson, E.J., & Tversky, A. (1983). Affect, generalization, and the perception of risk. Journal
of Personality and Social Psychology, 45, 20–31.
Kor, K., & Mullan, B.A. (2011). Sleep hygiene behaviours: An application of the theory of
planned behaviour and the investigatior of perceived autonomy support, past behavior
and response inhibition. Psychology and Health, 26, 1208–1224.
Manning, M., & Bettencourt, B.A. (2011). Depression and medication adherence among
breast cancer survivors: Bridging the gap with the theory of planned behaviour.
Psychology and Health, 26, 1173–1187.
Matterne, U., Diepgen, T.L., & Weisshaar, E. (2011). A longitudinal application of three
behaviour models in the context of skin protection behaviour in individuals with
occupational skin disease. Psychology and Health, 26, 1188–1207.
McEachan, R.R.C., Conner, M., Taylor, N., & Lawton, R.J. (2011). Prospective prediction of
health-related behaviors with the Theory of Planned Behavior: A meta-analysis. Health
Psychology Review, 5, 97–144.
McKee, S.A., Wall, A.-M., Hinson, R.E., Goldstein, A., & Bissonnette, M. (2003). Effects of
an implicit mood prime on the accessibility of smoking expectancies in college women.
Psychology of Addictive Behaviors, 17, 219–225.
Norman, P., & Cooper, Y. (2011). The theory of planned behaviour and breast self-
examination: Assessing the impact of past behaviour, context stability and habit strength.
Psychology and Health, 26, 1156–1172.
Nosek, B.A., Graham, J., Lindner, N.M., Kesebir, S., Hawkins, C.B., Hahn, C., ..., Tenney,
E.R. (2010). Cumulative and career-stage citation impact of social-personality psychology
programs and their members. Personality and Social Psychology Bulletin, 36, 1283–1300,
doi: 10.1177/0146167210378111.
Notani, A.S. (1998). Moderators of perceived behavioral control’s predictiveness in the
theory of planned behavior: A meta-analysis. Journal of Consumer Psychology, 7,
247–271.
Ouellette, J.A., & Wood, W. (1998). Habit and intention in everyday life: The multiple
processes by which past behavior predicts future behavior. Psychological Bulletin, 124,
54–74.
1126 I. Ajzen
Downloaded by [University of Massachusetts, Amherst], [Icek Aizen] at 09:03 20 September 2011
Petty, R.E., & Cacioppo, J.T. (1986). The elaboration likelihood model of persuasion.
In L. Berkowitz (Ed.), Advances in experimental social psychology (Vol. 19, pp. 123–205).
New York, NY: Academic Press.
Rapaport, P., & Orbell, S. (2000). Augmenting the theory of planned behaviour: Motivation
to provide practical assistance and emotional support to parents. Psychology and Health,
15, 309–324.
Rhodes, R.E., & Courneya, K.S. (2003). Relationships between personality, an extended
theory of planned behaviour model and exercise behaviour. British Journal of Health
Psychology, 8, 19–36.
Richard, R., de Vries, N.K., & van der Pligt, J. (1998). Anticipated regret and precautionary
sexual behavior. Journal of Applied Social Psychology, 28, 1411–1428.
Richetin, J., Conner, M., & Perugini, M. (2011). Not doing is not the opposite of doing:
Implications for attitudinal models of behavioral prediction. Personality and Social
Psychology Bulletin, 37, 40–54.
Rise, J., Sheeran, P., & Skalle, S. (2006). The role of self-identity in the theory of planned
behavior: A meta-analysis. Unpublished manuscript. Norwegian Institute for Alcohol
and Drug Abuse, Oslo, Norway.
Rivis, A., & Sheeran, P. (2003). Descriptive norms as an additional predictor in the theory of
planned behaviour: A meta-analysis. Current Psychology, 22, 218–233.
Rivis, A., Sheeran, P., & Armitage, C.J. (2011). Intention versus identification as determinants
of adolescents’ health behaviours: evidence and correlates. Psychology and Health, 26,
1128–1142.
Sandberg, T., & Conner, M. (2008). Anticipated regret as an additional predictor in the theory
of planned behavior: A meta-analysis. British Journal of Social Psychology, 47, 589–606.
Schaller, M., & Cialdini, R.B. (1990). Happiness, sadness, and helping: A motivational
integration. In E.T. Higgins & R.M. Sorrentino (Eds.), Handbook of motivation and
cognition: Foundations of social behavior (Vol. 2, pp. 265–296). New York, NY:
Guilford Press.
Schulze, R., & Wittmann, W.W. (2003). A meta-analysis of the theory of reasoned action and
the theory of planned behavior: The principle of compatibility and multidimensionality
of beliefs as moderators. In R. Schulze, H. Holling, & D. Bo
¨hning (Eds.), Meta-analysis:
New developments and applications in medical and social sciences (pp. 219–250). Seattle,
WA: Hogrefe & Huber.
Sheeran, P. (2002). Intention-behavior relations: A conceptual and empirical review.
In W. Stroebe & M. Hewstone (Eds.), European review of social psychology (Vol. 12,
pp. 1–36). Chichester: Wiley.
Sheeran, P., Norman, P., & Orbell, S. (1999). Evidence that intentions based on attitudes
better predict behaviour than intentions based on subjective norms. European Journal of
Social Psychology, 29, 403–406.
Sheeran, P., Orbell, S., & Trafimow, D. (1999). Does the temporal stability of behavioral
intentions moderate intention-behavior and past behavior-future behavior relations?
Personality and Social Psychology Bulletin, 25, 721–730.
Trafimow, D., & Finlay, K.A. (1996). The importance of subjective norms for a minority of
people: Between-subjects and within-subjects analyses. Personality and Social Psychology
Bulletin, 22, 820–828.
Uhlmann, E., & Swanson, J. (2004). Exposure to violent video games increases automatic
aggressiveness. Journal of Adolescence, 27, 41–52.
Wegner, D.M. (2002). The illusion of conscious will. Cambridge, MA: MIT Press.
Wegner, D.M., & Wheatley, T. (1999). Apparent mental causation: Sources of the experience
of will. American Psychologist, 54, 480–492.
Wolff, K., Nordin, K., Brun, W., Berglund, G., & Kvale, G. (2011). Affective and cognitive
attitudes, uncertainty avoidance and intention to obtain genetic testing: An extension of
the theory of planned behaviour. Psychology and Health, 26, 1143–1155.
Psychology and Health 1127
Downloaded by [University of Massachusetts, Amherst], [Icek Aizen] at 09:03 20 September 2011
