From Intentions To Actions
From Intentions To Actions
From Intentions To Actions
There appears to be general agreement among social psychologists that most hu-
man behavior is goal-directed (e.g., Heider, 1958; Lewin, 1951). Being neither ca-
pricious nor frivolous, human social behavior can best be described as following
along lines ofmore or less well-formulated plans. Before attending a concert, for ex-
ample, a person may extend an invitation to a date, purchase tickets, change into
proper attire, call a cab, collect the date, and proceed to the concert hall. Most, if not
all, ofthese activities will have been designed in advance ; their execution occurs as
the plan unfolds.To be sure, a certain sequence ofactions can become so habitual or
routine that it is performed almost automatically, as in the case of driving from
home to work or playing the piano. Highly developed skills ofthis kind typically no
longer require conscious formulation of a behavioral plan. Nevertheless, at least in
general outline, we are normally well aware of the actions required to attain a cer-
tain goal. Considersuch a relatively routine behavioras typing a letter. When setting
this activity as a goal, we anticipate the need to locate a typewriter, insert a sheet of
paper, adjust the margins, formulate words and sentences, strike the appropriate
keys, and so forth . Some parts of the plan are more routine, and require less con-
scious thought than others, but without an explicit or implicit plan to guide the re-
quired sequence of acts, no letter would get typed.
Actions, then, are controlled by intentions, but not all intentions are carried
out; some are abandoned altogether while others are revised to fit changing cir-
cumstances. The present chapter examines the relations between intentions and
actions: the ways in which goals and plans guide behavior, and the factors that
induce people to change their intentions, or prevent successful execution of the
behavior. The first part of the chapter deals with prediction and explanation of
behavior that is largely under a person's volitional control. A theory of reasoned
action is decribed which traces the causal links from beliefs, through attitudes
and intentions, to actual behavior. Relevant empirical research is reviewed, with
particular emphasis on the intention-behavior link and the factors that may pro-
duce changes in behavioral intentions. The chapter's second part deals with a be-
havioral domain about which much less is known. There, an attempt is made to
extend the theory ofreasoned action to goal-directed behaviors over which an indi-
vidual has only limited volitional control. First, internal and external factors that
may influence volitional control are identified. Next, a behavior-goal unit is de-
fined, and the theory of reasoned action is modified to enable it to predict and ex-
plain such goal-directed behavior. The modified theory, called "a theory ofplanned
behavior," differs from the theory of reasoned action, in that it takes into account
perceived as well as actual control over the behavior under consideration.
A great many behaviors of everyday life may be considered under volitional con-
trol in the sense that people can easily perform these behaviors if they are in-
clined to do so. To illustrate, under normal circumstances most people can, if
they so desire, watch the evening news on television, vote for the candidate of
their choice in an election, buy toothpaste at a drug store, pray before going to
bed, or donate blood to the Red Cross. The theory of reasoned action (Ajzen &
Fishbein, 1980; Fishbein & Ajzen, 1975) is designed to predict volitional behav-
iors of this kind and to help us understand their psychological determinants.
As its name implies, the theory of reasoned action is based on the assumption
that human beings usually behave in a sensible manner ; that they take account
of available information and implicitly or explicitly consider the implications of
their actions. Consistent with its focus on volitional behaviors, the theory postu-
lates that a person's intention to perform (or not to perform) a behavior is the im-
mediate determinant of that action. Barring unforeseen events, people are ex-
pected to act in accordance with their intentions. Clearly, however, intentions
can change over time ; the longer the time interval, the greater the likelihood that
unforeseen events will produce changes in intentions. It follows that accuracy of
prediction will usually be an inverse function of the time interval between mea-
surement of intention and observation of behavior.
Since we are interested in understanding human behavior, not merely in pre-
dicting it, we must next identify the determinants of intentions. According to the
theory of reasoned action, a person's intention is a function of two basic determi-
nants; one personal in nature and the other reflecting social influence. The per-
sonal factor is the individual's positive or negative evaluation of performing the
behavior; this factor is termed attitude toward the behavior. Note that the theory
of reasoned action is concerned with attitudes toward behaviors and not with the
more traditional attitudes toward objects, people, or institutions. The second de-
terminant of intention is the person's perception of the social pressures put on
him to perform or not perform the behavior in question. Since it deals with per-
ceived prescriptions, this factor is termed subjective norm . Generally speaking,
people intend to perform a behavior when they evaluate it positively and when
they believe that important others think they should perform it.
The theory assumes that the relative importance ofthese factors depends in part
on the intention under investigation. For some intentions, attitudinal considera-
From Intentions to Actions : A Theory of Planned Behavior 13
tions may be more important than normative considerations, while for other inten-
tions normative considerations may predominate. Frequently, both factors are im-
portant determinants of the intention. In addition, the relative weights of the atti-
tudinal and normative factors may vary from one person to another. The discussion
of the theory up to this point can be summarized symbolically as follows :
(1)
1 In addition, the theory of reasoned action requires that intention and behavior be operationally
defined so that they correspond in their target, action, context, and time elements (see Ajzen &
Fishbein, 1977).
2 Theories of a similar nature have been proposed by Edwards (1954), Rosenberg (1956), and ot-
hers (see Feather, 1959).
14 leek Ajzen
a given behavior will lead to mostly positive outcomes will hold a favorable atti-
tude toward performing the behavior while a person who believes that perform-
ing the behavior will lead to mostly negative outcomes will hold an unfavorable
attitude. The beliefs that underlie a person's attitude toward the behavior are
termed behavioral beliefs.
Subjective norms are also assumed to be a function of beliefs, but beliefs of a
different kind, namely the person's beliefs that specific individuals or groups
think he should or should not perform the behavior. These beliefs underlying the
subjective norm are termed normativebeliefs. Generally speaking, a person who
believes that most referents with whom he ist motivated to comply think he
should perform the behavior will perceive social pressure to do so. Conversely, a
person who believes that most referents with whom he is motivated to comply
think he should not perform the behavior will have a subjective norm that puts
pressure on him to avoid performing the behavior. The relation between norma-
tive beliefs and subjective norm is expressed symbolically in Equation 3. Here,
n
SNoc L: bjmj (3)
j=l
Empirical Support
The extent to which the theory of reasoned action predicts behavioral intentions
is usually evaluated by means of linear multiple regression analyses; the regres-
sion coefficients produced by these analyses serve as estimates of Wl and W2, the
weights of the attitudinal and normative predictors (see Equation 1). In addition,
whenever possible the intention-behavior correlation is reported, as are correla-
tions between direct (A B , SN) and indirect (Lb;e;, LAmj) measures of attitudes
and subjective norms. Finally, the sample of respondents can be divided into
those who did (or intented to) perform the behavior and those who did not. Dif-
ferences in behavior (or intentions) can then be explained by examining patterns
of differences in behavioral beliefs, outcome evaluations, normative beliefs , and
motivations to comply. (See Ajzen & Fishbein, 1980, for examples.)
Table 2.1 presents a sample of research findings on the major relationships
specified by the theory of reasoned action. It can be seen that the theory permits
highly accurate prediction in a wide variety of behavioral domains. Generally
16 leek Ajzen
speaking, people were found to act in accordance with their intentions. Strong
intention-behavior correspondence was expected, given the largely volitional na-
ture of the behaviors that served as criteria in these investigations. With one ex-
ception, all intention-behavior correlations exceeded 0.70. Interestingly, the ex-
ception ocurred in the prediction of having another child where the intention-be-
havior correlation, although significant, was only 0.55. Having another child is,
of course, only partially under volitional control since fecundity, miscarriage ,
and other factors also mediate attainment of the goal. We shall return to this is-
sue below.
The second column in Table 2.1 shows that, in each case, a linear combination
of attitudes and subjective norms permitted highly accurate prediction of inten-
tions. The relative importance of the two components is revealed by inspecting
the next four columns. Except for reenlisting in the military, where the regression
coefficient of subjective norm did not attain significance, both attitudes and sub-
jective norms made significant contributions to the prediction of intentions. In
seven of the nine studies listed in Table 2.1, the relative contribution of attitudes
exceeded that of subjective norms; but in two cases, the pattern was reversed.
Not surprisingly, women's decisions to have an abortion, and a couple's decision
to have another child, were strongly affected by perceived expectations of im-
portant others .
Finally, the last two columns of Table 2.1 report relations between direct and
belief-based measures of attitudes and subjective norms. It can be seen that the
results tend to support Equations 2 and 3: Behavioral beliefs and outcome evalu-
ations can be used to estimate attitudes toward a behavior (Column 7), while
normative beliefs and motivations to comply provide estimates of subjective
norms (Column 8).
Clearly then, the theory of reasoned action can afford highly accurate predic-
tion of intentions and behaviors that are under volitional control. By examining
closely the underlying belief structure, one can also gain a good understanding
of the factors that ultimately determine a person's decision to perform or not to
perform a given behavior. To illustrate, consider women 's use of birth control
pills. With respect to behavioral beliefs Ajzen and Fishbein (1980, pp.141-142)
summarized the important research findings as follows. "The major considera-
tions that entered into the women's decisions to use or to not use the pill re-
volved around questions of physiological side effects, morality, and effective-
ness. Although all women believed that using the pill leads to minor side effects
(such as weight gain), they differed in their beliefs about severe consequences.
The more certain a woman was that using the pill would not lead to such nega-
tive outcomes as blood clots and birth defects, the more likely she was to intend
using the pill. Also associated with intentions to use the pill were beliefs that this
was the best available method for preventing pregnancy." In addition, it was
found that women intended to use the pill only if they had no strong moral ob-
jections. "On the normative side, the women's major concerns centered on the
prescriptions of their husbands or boyfriends and doctors . They were highly
motivated to comply with these referents, and women who believed that their
husbands or boyfriends and doctors thought they should use the pill intended to
From Int ention s to Actions: A Theory of Planned Behavior 17
do so. By the same token, women who believed that these two referents opposed
their use of birth control pills formed intentions to not use them" (p.142).
To summarize briefly, many behaviors of interest to social psychologists ap-
pear to be under volitional control and can be predicted with a high degree of ac-
curacy from intentions to perform the behaviors in question. These intentions, in
tum, appear to be based on personal attitudes toward the behaviors and per-
ceived social norms. Attitudes are influenced by beliefs concerning a behavior's
likely outcomes and evaluations of those outcomes, while subjective norms de-
rive from normative beliefs regarding expectations of specific referent individu-
als or groups and motivations to comply with these referents. By examining dif-
ferences in behavioral and normative beliefs (and in associated outcome
evaluations and motivations to comply) we can go beyond prediction to provide
a detailed explanation of volitional behavior.
Changes in Intention
Many factors have been found to influence the stability of behavioral intentions.
Examination of these factors sheds light on the ways and means by which it may
be possible to prevent changes in intentions or modify predictions to take antic-
ipated readjustments into account.
From Intentions to Actions: A Theory of Planned Behavior 19
Effects in Time
Intentions change as time goes by. Some changes arguably occur simply as a
function of time while others depend on the emergence of new information.
New Information . Many changes in intentions, however, are the result of factors
other than the mere passage of time. In fact, changes that appear at first glance to
occur automatically may actually be mediated by internal processes (see Beck-
mann & Kuhl , 1984) or external factors . A multitude of unanticipated, and
sometimes unforeseeable, events can disrupt the intention-behavior relation. A
person's behavioral and normative beliefs are subject to change as events unfold
and new information becomes available . Such changes may influence the per-
son's attitude toward the behavior or his subjective norm and, as a result, pro-
duce a revised intention. To illustrate, consider a woman who intends to vote for
the Democratic candidate in the forthcoming senate race. After her intention is
assessed, she learns - by watching a television interview with the candidate a few
days before the election - that he opposes abortion and equal rights for women.
As a result, she "changes her mind " : she forms new beliefs concerning the conse-
quences of voting for the Democratic candidate, modifies her attitude toward
this behavior, decides to vote for the Republican candidate instead, and actually
does so in the election . Her voting choice corresponds to her most recent inten-
tion, but it could not have been predicted from the measure of intention obtained
at ail earlier point in time. .
Several studies have demonstrated the disruptive effects of unforeseen events.
For example, Songer-Nocks (1976a, 1976b) assessed intentions to choose the
noncompetitive alternative at the beginning of a 20-trial, two-person experimen-
20 leek Ajzen
tal game. Half of the pairs of players were given feedback after each trial which
informed them about the choices made by their partners and of the payoffs to
each player. The other pairs were given no such information. Feedback concern-
ing the partner's competitive or noncompetitive behavior may, of course, influ-
ence a player's own intentions regarding future moves. Consistent with this argu-
ment, Songer-Nocks reported that providing feedback significantly reduced the
accuracy with which initial intentions predicted actual game behavior.
Somewhat more indirect evidence regarding the disruptive potential of unan-
ticipated events is available from studies that have varied the amount of time be-
tween assessment of intention and observation of behavior. Since the probability
of unforeseen events will tend to increase as time passes, we would expect to find
stronger intention-behavior relations with short rather than long periods of de-
lay. Sejwacz, Ajzen, and Fishbein (1980) obtained support for this prediction in a
study of weight loss. A subsample of 24 college women indicated their intentions
to perform eight weight-reducing behaviors (avoid snacking between meals, par-
ticipate in sports on a regular basis, etc.) at the beginning of a two-month period,
and again one month later. Correlations were computed between initial inten-
tions and actual behavior over the two-month period, and between subsequent in-
tentions and actual behavior during the final month. As expected, the intention-
behavior correlations were stronger for the one-month period than for the two-
month period. For example, the correlation between intention to avoid long peri-
ods of inactivity and actual performance of this behavior (as recorded by the wom-
en) was higher when the time period was one month (r=O.72) than when it was
two months (r=0.47). Considering all eight behaviors, the average correlation
increased from 0.51 for the two-month period to 0.67 for the one-month period.
Temporal delay between assessment of intention and observation of behavior
thus tends to have a detrimental effect on behavioral prediction (see Hornik,
1970 and Fishbein & Coombs, 1974 for additional evidence in support of this
conclusion). As time passes, there is an increase in the likelihood ofunanticipat-
ed events and of concomitant changes in intentions. The result is a decline in the
correlation between observed behavior and intentions assessed before the
changes took place.
The discussion up to this point has dealt with rather drastic changes in intentions
that lead to a reversal of behavioral plans. Intentions, however, vary in strength
as well as direction, and changes can occur that would not be reflected in behav-
ior. Consider, for example, a voter who assigns a probability of 0.85 to his inten-
tion to vote for the Republican candidate, and a probability of 0.15 to his inten-
tion to cast his vote for the Democratic candidate. Exposure to new information
during the election compaign might reduce the perceived advantage of the Re-
publican candidate, changing the strength of these intentions to 0.65 and 0.35, re-
spectively ; even so, the person would still be expected to vote for the Republican
candidate.
From Intentions to Actions: A Theory of Planned Behavior 21
As a general rule, when an intention is held with great confidence (i.e., when
the intention is highly polarized), changes produced by new information will of-
ten be insufficient to reverse the planned course of action. In contrast, weak in-
tentions to perform (or not to perform) a behavior carry less of a commitment;
unanticipated events of relatively minor importance may influence such inten-
tions enough to bring about a change of mind. It follows that intention-behavior
correlations will usually be stronger when intentions are held with great, rather
than little, confidence.
Sample and Warland (1973, Warland & Sample, 1973) as well as Fazio and
Zanna (1978) have shown that attitudes held with high confidence are better
predictors of behavior than are attitudes held with low confidence. Ajzen, White,
and Timko (1982) examined more directly the effect of confidence on the inten-
tion-behavior relation. College students expressed their intentions to become
members of a psychology subject pool (on a 7-point "likely-unlikely" scale), and
indicated their confidence in their intentions on a 7-point scale that ranged from
"extremely certain" to "not at all certain. " At a later point in the experiment, they
were given an opportunity to actually sign up for the subject pool. Using the me-
dian score on the certainty scale as a dividing point, respondents were parti-
tioned into low and high confidence groups. As expected, the intention-behavior
correlation was significantly stronger among respondents who had high confi-
dence in their intentions (r= 0.73) than among respondents with low confidence
(r=0.47V
It is interesting to note that the very act of stating an intention may induce
hightened commitment to the behavior. In a series of experiments conducted by
Sherman (1980), respondents who, in response to a question, predicted that they
would act in a socially desirable manner were more likely to do so on a later oc-
casion than were respondents who were not asked to predict their own behavior.
Note, however, that the behaviors in question were of relatively little conse-
quence: writing a counter-attitudinal essay, singing the national anthem over the
telephone, and voluntering 3 hours to collect money for the American Cancer
Society. Merely stating an intention may have much less of an effect on such per-
sonally significant behaviors as having an abortion or reenlisting in the military
(see Table 2.1).
3 In addition to greater stability, confident intentions (being more extreme) will tend to exhibit
greater variance than intentions held with less confidence. Greater variability would also tend
to increase the predictive validity of confident intentions. The standard deviation of intentions
was, in fact, found to be significantly greater in the high (SD = 2.24) than in the low (SD = 1.36)
confidence group. It is thus not clear whether improved prediction in the high confidence group
was due to greater stabilit y of intentions or whether it was simply a statistical artifact produced
by a restriction of range in the low confidence condition .
22 Icek Ajzen
Individual Differences
Some people change their intentions more readily than others. According to
Snyder (1974, 1982), individuals differ in the extent to which their behavior is
susceptible to situational cues as opposed to inner states or dispositions. He de-
veloped the "self-monitoring scale" to assess the tendency for a person's behav-
ior to be guided by principle or inner disposition (low self-monitoring) on the
one extreme , and by situational contingencies or pragmatism (high self-monitor-
ing) on the other extreme. It stands to reason that the intentions of high self-
monitoring individuals, who are sensitive to external cues, will readily be influ-
enced by unanticipated events. The intentions of low self-monitors, however,
should be less affected by external events, and should thus be relatively stable,
since these individuals are sensitive primarily to internal states. We would thus
expect stronger intention-behavior correlations among low as compared to high
self-monitoring individuals."
Data collected by Ajzen, Timko, and White (1982) support this prediction.
The study was conducted at the time of the 1980 presidential election in the
United States. Shortly before the election , college students completed question-
naires that, among other questions, included Snyder's self-monitoring scale and
measure s of intention to vote in the election and intention to smoke marijuana in
the next 3 or 4 weeks. About 2 weeks after the election, participants were contact-
ed by telephone and asked to report their behavior. As expected, individuals who
scored below the median on the self-monitoring scale exhibited significantly
stronger intention-behavior correlations than did individuals who scored above
the median on the scale. The intention to vote predicted actual voting with a
correlation of 0.59 for high self-monitors, and with a correlation of 0.82 for low
self-monitors. The corresponding correlations with respect to the number of oc-
casion s on which the respondents reported to have smoked marijuana were 0.42
and 0.70.
Long-Range Prediction
It should be clear by now that many factors can influence the stability of inten-
tions and hence the strength of the observed intention-behavior relation. One so-
lution to this problem is to assess intentions immediately prior to observation of
the behavior. Clearly, the shorter the delay, the less time and the fewer the oppor-
tunities for change. In practice , however, it may be neither feasible nor of much
practical value to measure the intention in close temporal proximity of the be-
havior. Imagine, for example, that we are trying to predict behavior during such
natural disa sters as floods , tornados, earthquakes, or fires burning out of control.
4 Snyder and Swann (1976) and Zann a, Olson, and Fazio (1980) have reported stronger att itude-
beh avior correlations for low self-monitors as compared to high self-monitors; but Zuckerm an
and Reis (1978) failed to obt ain similar results.
From Intentions to Actions : A Theory of Planned Behavior 23
5 The reason that Bumpass and Westoff(1969) found such close intention-behavior correspon-
dence at the aggregate level is that the proportion of women who overproduced was almost ex-
actly the same as the proportion of women who underproduced. Had there been a bias toward
having more or fewer children than intended, the aggregate-level prediction would have been
less accurate.
24 leek Ajzen
The fact that intentions can change over time forces us to recognize their provi-
sional nature. Strictly speaking, all an individual can say is that, as ofnow, he in-
tends to perform a given behavior, and can assign a certain degree of confidence
(subjective probability) to that intention. Assuming the behavior is under voli-
tional control (and the person is prepared to exert maximum effort), failure to
act in accordance with the intention would indicate that the person had a change
of mind.
Further complications enter the picture as we turn to behaviors that are not fully
under volitional control. Failure to enact a behavior of this kind may occur either
because of a change in intention or because performance of the behavior failed.
A clear example is provided by the many smokers who intend to quit but either
change their intentions or, when they do try, fail to achieve their goal. A number
of investigators have in recent years turned their attention to the question of voli-
tional control (e.g., Bandura, 1977, 1982; Kuhl, 1981, 1982; Semmer & Frese,
1979), and at least one attempt has been made to extend the theory of reasoned
action to the prediction of partly nonvolitional behavior (Warshaw, Sheppard, &
Hartwick, in press). The present discussion will draw on some of these analyses.
At first glance, the problem of volitional control may appear rather limited in
scope. Its relevance is readily apparent whenever people try to overcome such
powerful habits as smoking or drinking, or when they set their sights on such dif-
ficult to attain goals as marrying a millionaire or reducing weight. Closer scru-
tiny reveals, however, that even very mundane activities, which can usually be
performed (or not performed) at will, are sometimes subject to the influence of
factors beyond one's control. Consider, for example, a person who intends to
spend the evening at home watching a movie on television. As she turns on the
set, a puff of smoke indicates that its useful life is over, putting an end to her
plans for the evening. Some behaviors are more likely to present problems of
control than others, but we can never be absolutely certain that we will be in a
position to carry our intentions. Viewed in this light it becomes clear that, strictly
speaking, every intended behavior is a goal whose attainment is subject to some
degree of uncertainty. We can thus speak of a behavior-goal unit; and the inten-
tion constitutes a plan of action in pursuit of the behavioral goal. We shall return
to these issues. First, however, we must briefly consider some of the factors that
influence volitional control over a behavioral goal.
From Intentions to Actions: A Theory of Planned Behavior 25
Internal Factors
our abilities; yet new situations arise frequently, and failure to achieve our goals
due to lack of requisite skills is the order of the day.
The idea that behavioral achievement is a function not only of intention (or
motivation) but also of ability is of course not very original. Heider (1958) made
it a cornerstone of his "naive analysis of action," and Jones and Davis (1965) in-
corporated it into their theory of correspondent inferences. According to Heider,
a person is viewed as responsible for his action (i. e., the action is attributed to the
person) if he is believed to have tried and to have had the ability to produce the
observed effects. Evidence for this analysis in the context of achievement-related
attributions has been provided by Weiner (1974) and his associates (e.g ., Weiner,
Frieze, Kukla, Reed, Rest, & Rosenbaum, 1971), as well as by other investigators
(e.g., Anderson, 1974).
It seems self-evident that successful performance of an intended behavior is
contingent on the presence of required information, skills, and abilities; and per-
haps it is for this reason that few investigators have bothered to demonstrate em-
pirically the interactive effect of intention and ability on actual behavioral per-
formance.
what related concept of state versus action orientation. A person's state or action
orientation is viewed both as a relatively stable predisposition and as dependent
on a variety of situational factors. Action-oriented individuals are assumed to fo-
cus their attention on action alternatives and to make use of their knowledge and
abilities to control their performance. In contrast, state-oriented persons are like-
ly to focus their attention on their thoughts and feelings (their present, past, or
future state) rather than taking action consistent with their intentions. Kuhl
(1982) has developed a scale to assess action orientation and, using this scale, has
found higher intention-behavior correlations among action-oriented as opposed
to state-oriented individuals. He has also investigated a variety of situational fac-
tors (e.g., a failure experience) that may result in state or action orientations, and
thus influence execution of intended behavior (see also Mischel's, 1974, work on
situational determinants of delay of gratification).
Emotions and Compulsions. Skills, abilities, and will power may present prob-
lems of control, but it is usually assumed that , at least in principle, these prob-
lems can be overcome. In contrast, some types of behavior are often viewed as
controlled by forces that are largely beyond our control. People sometimes ap-
pear unable to cease thinking or dreaming about certain events, to stop stutter-
ing, or to hold a tick in check. These compulsive behaviors are performed despite
intentions and concerted efforts to the contrary.
Emotional behaviors seem to share some of the same characteristics. Individu-
als are often not held responsible for behaviors performed under stress or in the
presence of strong emotions. We usually attribute little control to a person who is
"overcome by emotion". Violent acts and poor performance are expected under
such conditions, and there seems to be little a person can do about it. Some, how-
ever, have argued that emotional behaviors are not all that different from other
types of behavior, and that their antecedents are very similar to those of nonemo-
tional acts (e.g ., Averill, 1980; Solomon, 1976). If this view is correct, then we
should find relatively strong relations between intentions and emotional behav-
ior, although problems of control may well prevail , especially in the case of in-
tense emotional experiences.
In conclusion, as we move beyond intentions, various internal factors influ-
ence successful performance of an intended behavior. It may be fairly easy to
gain control over some of these factors, as when we acquire the information or
skills needed to perform a behavior. Other factors, such as intense emotions,
stress, or compulsions, are more difficult to neutralize. Whatever the nature of
the internal factor, however, it will tend to influence our control over the behav-
ioral goal.
External Factors
Also impinging on a person's control over behavioral goals are external or situa-
tionalfactors. These factors will be discussed under two headings: (1) time and
opportunity and (2) dependence on other people.
28 leek Ajzen
tentions were significantly better predictors of behavior for females (r= 0.68)
than for males (r= 0.39). Clearly, females in our society find it relatively easy to
obtain the cooperation of males when they attempt to execute their intentions to
engage in premarital sexual intercourse. By comparison, males often have great-
er difficulties in finding willing partners.
As in the case of time and opportunity, inability to carry out an intention be-
cause of dependence on others may have little effect on the underlying motiva-
tion. Often a person who encounters difficulties related to interpersonal depen-
dence may be able to perform the desired behavior in cooperation with a
different individual. Sometimes, however, this may not be possible as in the case
of dependence on one's spouse. A wife's adamant refusal to have more children,
for example, will usually cause the husband eventually to abandon his plan to
enlarge the family, rather than shift his efforts to a different partner.
I have tried to show that time, opportunity, and dependence on others often
lead only to temporary changes in intentions. When time is the constraining fac-
tor, the behavior may simply be delayed ; when circumstances prevent perfor-
mance of a behavior, the person may wait for a better opportunity; and when an-
other person fails to cooperate, a more compliant partner may be sought.
However, when repeated efforts to perform the behavior fail, more fundamental
changes in intentions can be expected.
The above discussion makes clear that many factors can obstruct the intention-
behavior relation. Although volitional control is more likely to present a problem
for some behaviors than for others, personal deficiencies and external obstacles
can interfere with the performance of any behavior. Given the problem's ubiqui-
ty, a behavioral intention can best be interpreted as an intention to try perform-
ing a certain behavior. A father 's plan to take his children fishing on the forth -
coming weekend, for example, is best viewed as an intention to try to make time
for this activity, to prepare the required equipment, secure a fishing license, and
so forth. Successful performance of the intended behavior is contingent on the
person's control over the various factors that may prevent it. Of course, the con-
scious realization that we can only try to perform a given behavior will arise pri-
marily when questions of control over the behavior are salient. Thus , people say
that they will try to quit smoking or lose weight, but that they intend to go to
church on Sunday. Nevertheless, even the intention to attend Sunday worship
services must be viewed as an intention to try since factors beyond the individu-
al's control can prevent its successful execution.
These observations have important implications for the prediction of behavior
from intentions. Strictly speaking, intentions can only be expected to predict a
person's attempt to perform a behavior, not necessarily its actual performance. If
our measure of intention fails to predict attempted behavior, it is possible that
the intention changed after it was assessed (see the earlier discussion of the pre-
diction of volitional behavior). However, if the intention does predict whether or
30 leek Ajzen
not a person attempts to perform the behavior, but fails to predict attainment of
the behavioral goal, it is likely that factors beyond the person's control prevented
the person from carrying out his intention."
Consider, for example, the study of weight reduction mentioned earlier (Sej-
wacz, Ajzen, & Fishbein, 1980). The college women in this study expressed their
intentions to lose weight over a two-month period, and reported on their perfor-
mance of various dietary behaviors and physical activities during that same peri-
od. The intention to lose weight was found to have a nonsignificant correlation
of 0.16 with actual weight lost, but a significant correlation of 0.49 with an aggre-
gate measure of behavior. It seems reasonable to argue that engaging in dietary
behaviors and physical activities constitute attempts to reduce weight. Intentions
were thus better predictors of attempts to reduce weight, than of actual changes
in body weight. Clearly, losing weight depends not only on one's intention to do
so, but also on other factors, such as will power and physiological variables that
are only partly under volitional control. That the correlation between intentions
and attempted behavior was no higher than 0.49 can perhaps be attributed to
changes in intentions that may have occurred over the two-month period. Alter-
natively, it may have been due to a lack of will power, or to a state (as opposed to
action) orientation (Kuhl, 1982).
Additional support for the present argument can be found in a study by Poma-
zal and Jaccard (1976). The intentions of college students to "donate blood at the
upcoming blood drive" had a correlation of 0.46 with actual blood donations.
However, it was noted that several participants in the study who came to donate
blood were rejected for medical reasons or because of overcrowding. As expect-
ed, when showing up to donate blood (i. e., attempted behavior) was used as the
behavioral criterion, the correlation with intentions increased to 0.59.
In short, behavioral intentions will often be better predictors of attempted
than actual behavior. To insure accurate prediction in such instances, we would
not only have to assess intentions but also obtain some estimate of the extent to
which individuals are apt to exercise control over the behavior in question.
We are now ready to consider a possible expansion of the theory of reasoned
action that will allow us to include consideration of nonvolitional factors as de-
terminants of behavior. Equation 4 shows how the strength of a person's attempt
to perform a behavior (BI ) interacts with the degree of his control (C) to deter-
(4)
mine the likelihood of actual performance of the behavior (B). The harder the
person tries, and the greater his control over personal and external factors that
may interfere, the greater the likelihood that he will attain his behavioral goal,"
For some behaviors, a low level of effort (BI) is sufficient, and successful perfor-
8 Another possibility which cannot be discounted is that the person changed his intention while
trying to perform the behavior.
9 The notion ofcontrol is used here quite similar to Triandis's (1977) concept of "facilitative con-
ditions" (F) in his model of interpersonal behavior. However, in Triandis's model, Finteraets
with intentions and habits to determine the likelihood of a behavior.
From Intentions to Actions: A Theory of Planned Behavior 31
mance of the behavior depends largely on the level of control. Good examples
are highly skilled activities such as typing or driving a race car. Increased effort
on such tasks is less important than a high level of skill. For other behaviors, a
minimal level of control is enough, and successful performance of the behavior
varies with degree of effort. For instance, jogging 10 or 15 minutes every day re-
quires only low levels of control; for most people, achievement of this behavioral
goal depends largely on their willingness to try.
The question of control is often tied up with development of an adequate plan
that will enable performance of the behavior. A plan usually consists of a set of
intentions Which, if carried out, are expected to result in the desired behavioral
goal. It may also contain contingency plans; that is, alternative plans of action in
case the intended sequence of behaviors is blocked. Often, plans are developed
only in general outline: The initial behaviors may be clearly specified and later
parts of the plan are to be developed, depending in part on the success of earlier
actions (see Miller, Galanter, & Pribram, 1960). The behavioral attempt (Bd con-
stitutes the initiation of the plan; it is designed to overcome a perceived discrep-
ancy between the present state and the desired goal. Successful execution de-
pends on the adequacy of the plan itself, and on the various personal and
external factors discussed earlier that may influence control over the behavior.
Consistent with the theory of reasoned action, the immediate determinant of a
person's attempt to perform a behavior is his intention to try doing so (I/); and
this intention is in turn a function of attitude toward trying (A/) and subjective
norm with regard to trying (S N/). These relations are expressed symbolically in
Equation 5.
(5)
Here, as in Equation 1, the wavy line between B/ and 1/ indicates that expressed
jntentions to try performing a behavior can change before the behavioral attempt
is observed, and Wl and W2 are empirically determined weights for the two pre-
dictors of 1/. Thus, the more favorable a person's attitude toward trying to per-
form a behavior, and the more he believes that important others think he should
try, the stronger his intention to try.
We must go beyond the theory of reasoned action, however, when we consider
the determinants of A/, the attitude toward trying to perform a behavior. Clearly,
the attitude toward trying and succeeding (i. e., the attitude toward the behavior)
will usually differ from the attitude toward trying and failing. Whenever the pos-
sibility of failure is contemplated, therefore, the attitude toward trying will be de-
termined not only by the attitude toward (successful) performance of the behav-
ior (As) but also by the attitude toward a failed attempt (AtJ. This idea is
expressed in Equation 6, where the attitudes toward behavioral success and
A/O( [PsAs+pfAf);Ps+PF 1.0 (6)
failure are weighted by the respective subjective probabilities of these events (Ps
and Pf)' The subjective probabilities of success and failure should be related to
beliefs about the presence or absence of personal and external factors discussed
earlier that may facilitate or inhibit behavioral performance. This formulation is
32 leek Ajzen
(7)
10 Warshaw, Sheppard, and Hartwick (in press) have also proposed a similar formulation. In
their analysis, attitude toward pursuing a goal is determined by attitude toward pursuit with
success and attitude toward pursuit with failure , each weighted by the expectancy (probabili-
ty) of success or failure, respectively. In addition, they proposed that attitudes toward action s
involved in pursuit of the goal also be considered. The present analysis assumes that these lat-
ter attitudes are reflected in attitudes toward successful and unsuccessful behavioral attempts.
From Intentions to Actions: A Theory of Planned Behavior 33
p I b e, _ _ p A
5 S. I 5 5
I
are influenced by some of the same factors. The correlation between behavioral
expectation and actual behavior (B), however, will depend on correspondence
between the person's belief in his control over the behavior (be)and the degree of
his actual control (C). To the extent that the person's assessment of his skills, will-
power, and other requisite personal factors, and of the presence of favorable or
unfavorable external factors is realistic, his behavioral expectation will predict
his actual behavior. However, when his assessment fails to accurately reflect real-
ity, prediction of behavior from behavioral expectation will suffer. In any case, it
must be kept in mind that unlike behavioral intentions, behavioral expectations
may have no causal effect on actual behavior.
From a practical point of view, it will be very important to identify factors that
correlate with realistic perception of behavioral control. In addition to past expe-
rience, these factors may include confidence in one's subjective judgment of
control, availability of a detailed plan of action, and general self-knowledge. To
the extent that perceived control is likely to be realistic, it can serve as an esti-
mate of actual control and, together with intention to try, it can be used to predict
the probability of a successful behavioral attempt.
Subjective perceptions of control may, of course, influence attempts to per-
form behavior regardless of their accuracy. A person who has a pessimistic view
of his control over the behavior may never try and may thus fail to find out that
he was wrong. As a result, perceived control will usually correlate with behavior-
al performance. Again, however, this correlation will tend to be strong only when
perceived control corresponds reasonably well to actual control.
Bandura's (1977, 1982) work on "self-efficacy" provides support for the rela-
tion between perceived control and behavioral performance, usually in the form
of overcoming certain phobias or strong habits . For example, in one study (Ban-
dura, Adams, & Beyer, 1977), adult snake phobics following a period of treat-
ment rated the strength of their expectations that they could perform various
behaviors in relation to snakes . Correlations between these estimates of self-effi-
cacy (behavioral control) and subsequent behavior toward a snake were 0.83 and
0.84 for two groups of participants who had received performance-based as op-
posed to vicarious treatment, respectively. Experiences during treatment (and
perhaps prior to treatment) seemed to have created quite accurate perceptions of
self-efficacy. Since all participants were there to try to overcome their phobias,
attainment of the behavioral goal was dependent primarily on their (perceived)
control over the behavior involved.
Another issue worth considering at this point is the influence of past behavior
on present performance. Some investigators have suggested that past perfor-
mance of a behavior exerts an influence on present behavior that is independent
of behavioral intentions, attitudes, or subjective norms. For example, Bentler
and Speckart (1979) reported a study in which college students' use of alcohol,
marijuana, and hard drugs at one point in time made a significant contribution
independent of intentions to the prediction of the performance of these behav-
iors two weeks later. According to the present analysis, past performance of a be-
havior may be correlated with control over the behavior (C). In the case of addic-
tive behaviors, such as drinking alcohol or using drugs, frequent past behavior is
From Intentions to Actions : A Theory of Planned Behavior 35
likely to be associated with lack of control. A person who is addicted may intend
not to take hard drugs or drink excessively, but lacks the control to achieve his
behavioral goal. As a result of its correlation with control, past behavior would in
fact be expected to influence present behavior over and above the effect of inten-
tion (see Equation 4). According to the theory of planned behavior, past perfor-
mance should have no independent effect on present behavior only when a per-
son has complete control, that is, when dealing with a volitional behavior.
Drinking and use of drugs clearly do not qualify.
Examination of Equations 4 through 8 and Fig.2.1 shows that the theory of
reasoned action discussed at the beginning of this chapter is a special case of the
theory of planned behavior. The special case occurs when the subjective proba-
bility of success or perceived control (Ps or be) and actual degree of control over
the behavioral goal (C) are at their maxima . This can be seen clearly if both are
scaled from 0 to 1. When Ps and C equal 1, the likelihood of a behavioral attempt
is equivalent to the likelihood of performing the behavior (see Equation 4), atti-
tude toward trying reduces to attitude toward the behavior (Equation 6), and
subjective norm with respect to trying is the same as subjective norm regarding
performance of the behavior (Equation 7). In addition, when be equals 1, behav-
ioral expectation becomes equivalent to behavioral intention (Equation 8). Of
course, when Ps or be are equal to 1 (failure is not even considered as a possibili-
ty) and when actual control (C) is perfect , the behavior is under complete voli-
tional control and the theory of reasoned action applies. However, when the pos-
sibility of failure is salient and actual control is limited, then it becomes
necessary to go beyond the theory of reasoned action. It is here that the theory of
planned behavior will prove most useful.
References
Ajzen, I. (1971). Attitudinal vs. normative messages : An investigation of the differential effects
of persuasive communication on behavior. Sociometry. 34, 263-280.
Ajzen, I. & Fishbein, M. (1970). The prediction of behavior from attudinal and normative vari-
ables . Journal ofExperimentalSocial Psychology, G,466-487.
Ajzen, I., & Fishbein, M. (1977). Attitude-behavior relations : A theoretical analysis and review
of empirical research. Psychological Bulletin, 84, 888-918.
Ajzen, I., & Fishbein, M. (1980). Understanding attitudes and predicting social behavior. Engle-
wood-Cliffs, N .J .: Prentice-Hall.
Ajzen, I., Timko, C; & White , J. B. (1982) Self-monitoring and the attitude-behavior relation.
Journal of Personality and Social Psychology, 42, 426-435.
Ajzen, I., White, J. B., & Timko, C. (1982). Direct experience, confidence, and the attitude-be-
havior relation. Unpublished manuscript, University of Massachusetts.
Anderson, N. H. (1974). Cognitive algebra : Integration theory applied to social attribution. In
L. Berkowitz (Ed .), Advances in experimentalsocial psychology, Vol. 7. New York : Academic
Press.
Averill, J.R. (1980). A constructivist view of emotion. In R.Plutchik & H.Kellerman (Eds.) ,
Emotion: Theory, research, and experience. Volume 1: Theories of emotion. New York : Aca-
demic Press.
Bandura, A. (1977). Self-efficacy : Toward a unifying theory of behavioral change. Psychological
Review. 84, 191-215.
Bandura, A. (1982). Self-efficacy mechanism in human agency. American Psychologist, 37,
122-147.
Bandura, A., Adams, N. E., & Beyer, J. (1977). Cognitive processes mediating behavioral
change. Journal ofPersonality and Social Psychology, 35, 125-239.
Beckmann, J., & Kuhl, J. (1984). Deforming information to gain action-control : Functional
aspects of human information processing in decision making. Journal of Researchin Person-
ality, 18,224-237.
Bentler, P. M ., & Speckart, G. (1979). Models of attitude-behavior relations. Psychological Re-
view, 86, 452-464.
Brown, J. S. (1948). Gradients of approach and avoidance responses and their relation to moti-
vation. Journal of Comparative and Physiological Psychology, 41,450-465.
Bumpass, L., & Westoff, C. (1969). The later years of child bearing. Princeton, NJ : Princeton
University Press.
Edw ards, W. (1954) . The theory of decision making. Psychological Bulletin, 51, 380 -41 7.
Fazio, R.J ., & Zanna, M. (1978). Attitudinal qualities relating to the strength of the attitude-be-
havior relationship. Journal of ExperimentalSocial Psychology, 14, 398-408.
Feather, N .T. (1959). Subjective probability and decision under uncertainty. Psychological Re-
view, 66, 150-164.
Fishbein, M. (1966). Sexual behavior and propositional control. Paper read at the Psycho nomic
Society meetings.
Fishbein, M., & Ajzen, I. (1975). Belief, attitude, intention, and behavior: An introduction to the-
ory and research. Reading, Ma ss. : Addison-Wesley.
Fishbein, M., & Coombs, F. S. (1974). Basis for decision : An attitudinal analysis of voting be-
hav ior. Journal ofAppliedSocial Psychology, 4,95-124.
Heider, F. (1958). Thepsychologyofinterpersonal relations. New York : Wiley.
Hornik, J.A. (1970). Two approaches to individual differences in an expanded Prisoner's Di-
lemma game. Unpublished master's thesis, University of Illinois.
Jones, E. E., & Davi s, K. E. (1965). From acts to dispositions : The attribution process in person
perception. In L. Berkowitz (Ed .), Advances in experimental social psychology, Vol.2 . New
York: Academic Press.
Kuhl, J . (1981). Motivational and functional helplessness : The moderating effect of state versus
action orientation. Journal of Personality and Social Psychology, 40, 155-170.
Kuhl, J. (1982). Action- vs. state-orientation as a mediator between motivation and action. In
W. Hacker, W. Volpert, & M. von Cranach (Eds .), Cognitive and motivationalaspectsofaction.
Amsterdam , NL : North-Holland Publishing Co .
38 leek Ajzen
Lefcourt, H. M. (1982). Locus ofcontrol: Current trends in theory and research (2nd Ed.). Hills-
dale, NJ: Erlbaum.
Lewin , K. (1946). Action research and minority problems. Journal ofSocial Issues, 2,34-46.
Lewin, K. (1951). Fieldtheory in socialscience. New York : Harper. '
Lewin, K., Dernbo, T., Festinger, L., & Sears, P.S. (1944). Level of aspiration. In J . McV. Hunt
(Ed.), Personality and the behaviordisorders. New York : Ronald Press .
Manstead, A. S. R., Proffitt, C., & Smart, J . L. (1983). Predicting and understanding mothers' in-
fant feeding intentions and behavior: Testing the theory of reasoned action. Journal of Per-
sonality and Social Psychology, 44,657-671 .
Miller, G . A., Galanter, E., & Pribram, K. H . (1960). Plans and the structure of behavior. New
York : Holt.
Miller, N. E. (1944). Experimental studies of conflict. In J. McV. Hunt (Ed.), Personality and the
behaviordisorders. New York : Ronald Press .
Mischel, W. (1974. Processes in delay of gratification. In L. Berkowitz (Ed.), Advances in experi-
mental socialpsychology (Vol. 7). New York : Academic Press.
Pomazal, R.J ., & Jaccard, J.J. (1976). An informational approach to altruistic behavior. Journal
ofPersonality and Social Psychology, 33, 317-326.
Rapoport, A., & Chammah, A. M. (1965). Prisoner's dilemma: A study on conflict and coopera-
tion. Ann Arbor: University of Michigan Press .
Rosenberg, M.J. (1956). Cognitive structure and attitudinal affect. Journal ofAbnormal and So-
cial Psychology, 53, 367-372.
Rotter J. B. (1966). Generalized expectancies for internal versus external control of reinforce-
ment. Psychological Monographs. 80(1, Whole No. 609).
Sample, J ., & Warland, R. (1973). Attitude and prediction of behavior. (Social Forces, 51,
292-303.
Semmer, N., & Frese, M. (1979). Handlungstheoretische Implikationen fur kognitive Therapie.
In N. Hoffman (Ed .), Grundlagen kognitiver Therapie: Theoretische Modelle und praktische
Anwendungen. Bern : Huber.
Sejwacz, D., Ajzen, I., & Fishbein, M. (1980). Predicting and understanding weight loss : Inten-
tions, behaviors, and outcomes, In I. Ajzen & M. Fishbein, Understanding attitudes and pre-
dictingsocial behavior. Englewood Cliffs , N.J. : Prentice-Hall.
Sherman, S.J . (1980). On the self-erasing nature of errors of prediction. Journal of Personality
and Social Psychology, 39, 211-221.
Shtilerman, M. (1982). The influence of decision making process on the relationship between at-
titudes, intentions and behavior. Unpublished master's thesis. Tel-Aviv University.
Smetana, J . G., & Adler, N . E. (1980). Fishbein's value x expectancy model : An examination of
some assumptions. Personality and Social Psychology Bulletin. 6, 89-96.
Snyder, M. (1974). Self-monitoring of expressive behavior. JournalofPersonality and Social Psy-
chology, 30, 526-537 .
Snyder, M. (1982). When believing means doing : Cre ating links between attitudes and behavior.
In M. P. Zanna, C. P. Herman, & E.T. Higgins (Eds.), Consistencyin social behavior: The On-
tario Symposium, (Vol.2). Hillsdale, N.J. : Erlbaum.
Snyder, M ., & Swann, W. B., Jr. (1976). When actions reflect attitudes : The politics of impression
management. Journal of Personality and Social Psychology, 34, 1034-1042.
Solomon, R.C. (1976). The passions. New York : Anchor Press.
Songer-Nocks, E. (1976). Situational factors affecting the weighting of predictor components in
the Fishbein model. Journal ofExperimentalSocial Psychology, 12. 59-69. (a)
Songer-Nocks, E. (1976). Reply to Fishbein and Ajzen. Journal ofExperimentalSocial Psycholo-
gy, 12,585-590. (b)
Strickland, B. R. (1978). Internal-external expectancies and health-related behavior. Journal of
Consulting and Clinical Psychology, 46, 1192-1211.
Triandis, H.C. (1977). Interpersonal behavior. Monterey, CA: Brooks/Cole.
Vinokur-Kaplan, D . (1978). To have - or not to have - another child : Family planning attitudes,
intentions, and behavior. Journal ofApplied Social Psychology, 8, 29-46.
Warland, R. H., & Sample, J . (1973). Response certainty as a moderator variable in attitude mea-
surement. Rural Sociology, 38, 174-186.
From Intentions to Actions : A Theory of Planned Behavior 39
Warshaw, P. R., Sheppard, B. H., & Hartwick, J. (in press) . A general theory of intention and be-
havioral self-prediction. In Bagozzi (Ed.), Advances in Marketing Communication. Green-
wich, CT : JAI Press
Weiner, B. (1974). Archievement motivation and attribution theory. Morristown, N.J .: General
Learning Press.
Weiner, B., Frieze, I., Kukla, A., Reed, L., Rest, S., & Rosenbaum, R. M. (1971). Perceiving the
causesofsuccess andfailure. New York: General Learning Press.
Zanna, M. P., Olson, J. M., & Fazio, R. H. (1980). Attitude-behavior consistency : An individual
difference perspective. Journalof Personality and Social Psychology, 38, 432-440.
Zuckerman, M., & Reis, H.T. (1978). A comparison of three models for predicting altruistic be-
havior. Journalof Personality and Social Psychology, 36,498-510.