You are on page 1of 9

Pers Soc Psychol Rev OnlineFirst, published on May 26, 2010 as doi:10.

1177/1088868310368802

Personality and Social Psychology Review

Outcome Expectancy and Self-Efficacy: XX(X) 1­–9


© 2010 by the Society for Personality
and Social Psychology, Inc.
Theoretical Implications of an Reprints and permission:
sagepub.com/journalsPermissions.nav

Unresolved Contradiction DOI: 10.1177/1088868310368802


http://pspr.sagepub.com

David M. Williams1

Abstract
According to self-efficacy theory, self-efficacy—defined as perceived capability to perform a behavior—causally influences
expected outcomes of behavior, but not vice versa. However, research has shown that expected outcomes causally influence
self-efficacy judgments, and some authors have argued that this relationship invalidates self-efficacy theory. Bandura has
rebutted those arguments saying that self-efficacy judgments are not invalidated when influenced by expected outcomes.
This article focuses on a contradiction in Bandura’s rebuttal. Specifically, Bandura has argued (a) expected outcomes cannot
causally influence self-efficacy, but (b) self-efficacy judgments remain valid when causally influenced by expected outcomes.
While the debate regarding outcome expectancies and self-efficacy has subsided in recent years, the inattention to this
contradiction has led to a disproportionate focus on self-efficacy as a causal determinant of behavior at the expense of
expected outcomes.

Keywords
expectancies, motivation, goals, attitudes

When deciding on a course of action, people consider what self-efficacy judgments and argued that this relationship called
they stand to gain or lose from performing the behavior. This into question the validity of self-efficacy theory (Borkovec,
weighing of expected consequences is the central tenet of 1978; Corcoran, 1991, 1995; Eastman & Marzillier, 1984;
expectancy-value theories (for a review, see Feather, 1982), Kazdin, 1978; Kirsch, 1982, 1985, 1986, 1995; Teasdale,
including social learning theory (Rotter, 1954). Bandura’s 1978, Wolpe, 1978). A central point in Bandura’s (1982,
(1977) insight—that people’s behavior is also influenced by 1986a, 1986b, 1995, 1997, 2007) rebuttals to these critiques
their perceived capability to perform the behavior—was a was the concession that expected outcomes of behavior can,
seminal addition to social learning theory and has had tre- indeed, influence self-efficacy ratings but that this relation-
mendous influence on psychological theory and research. ship does not invalidate such judgments of self-efficacy.
Indeed, self-efficacy is a robust predictor of behavior (for Bandura’s rebuttal, however, creates a contradiction in
meta-analyses, see Holden, 1991; Holden, Moncher, Schinke, his writings on self-efficacy theory. Specifically, Bandura
& Barker, 1990; Moritz, Feltz, Fahrbach, & Mack, 2000; has argued (a) expected outcomes cannot causally influence
Multon, Brown, & Lent, 1991; Sadri & Robertson, 1993; self-efficacy (Bandura, 1978, 1984, 1986c, 1995, 1997,
Stajkovic & Luthans, 1998), has been incorporated in social 1998, 2004, 2006b), but (b) self-efficacy judgments remain
learning models of behavior that previously focused mainly valid when causally influenced by expected outcomes (e.g.,
on expected outcomes (Ajzen, 1985; Maddux & Rogers, Bandura, 1982, 1986a, 1986b, 1995, 1997, 2007). This con-
1983; Rosenstock, Strecher, & Becker, 1988), and has been tradiction has not received attention among self-efficacy
the target of numerous behavior change interventions (for researchers. Indeed, in the past decade the debate regarding
reviews, see Bandura, 1984; Hyde, Hankins, Deale, & Marteau, the causal influence of outcome expectancies on self-efficacy
2008; Marks, Allegrante, & Lorig, 2005). has mostly subsided (for exceptions see Bandura, 2007;
Bandura introduced self-efficacy in the context of an
explanatory model of human behavior, in which self-efficacy
causally influences expected outcomes of behavior, but 1
Brown University, Providence, RI
not vice versa (Bandura, 1978, 1984, 1986c, 1995, 1997,
Corresponding Author:
1998, 2004, 2006b). However, in the two decades follow- David M. Williams, Brown University, Program in Public Health, Institute
ing the introduction of self-efficacy theory, several research- for Community Health Promotion, Box G-S121-8, Providence, RI 02912
ers showed that expected outcomes causally influenced Email: david_m_williams@brown.edu

Downloaded from psr.sagepub.com at PENNSYLVANIA STATE UNIV on April 8, 2016


2 Personality and Social Psychology Review XX(X)

Cahill, Gallo, Lisman, & Weinstein, 2006). Meanwhile, on. For behaviors that require self-regulatory skills rather
self-efficacy has gained prominence in psychological theory than specialized physical or mental skills, self-efficacy is not
and research. merely perceived capability to carry out the necessary motor
The purpose of the present article is to rekindle the debate response but involves perceived capability to perform the
regarding the causal influence of expected outcomes on self- behavior in the context of competing demands and impedi-
efficacy, with particular focus on the conceptual contradic- ments (Bandura, 1997). For example, exercise self-efficacy
tion outlined above. It is argued herein that the inattention to may be assessed by asking respondents to rate their capabil-
this conceptual contradiction has led to a disproportionate ity to exercise when it is raining, when they feel they have
focus on self-efficacy as a causal determinant of behavior at little time to spare, and so on (Bandura, 2006a). Self-efficacy
the expense of expected outcomes. It is further argued that for discrete motor acts (e.g., self-efficacy for basketball
the vast literature supporting the predictive power of self- shooting) is often referred to as task self-efficacy (e.g.,
efficacy does not immunize the construct from this con- Woodgate, Brawley, & Weston, 2005), whereas self-efficacy
ceptual critique. Finally, it is argued that to resolve the for overcoming obstacles or impediments to regulate behav-
contradiction, either (a) the operationalization of self-efficacy ior over time (e.g., self-efficacy for exercise) is referred to as
must be revised such that self-efficacy judgments are free self-regulatory efficacy (Bandura, 1982, 1997) or barrier
from the influence of expected outcomes or (b) self-efficacy self-efficacy (e.g., Blanchard et al., 2007). Importantly,
theory must be revised such that expected outcomes are regardless of the self-efficacy subtype, self-efficacy generally
acknowledged as an important causal determinant of self- refers to perceived capability to perform a behavior (Bandura,
efficacy. In making these arguments, I (a) summarize the 1977, 1986c, 1997).
central tenets of self-efficacy theory, (b) review previous The other central construct in self-efficacy theory is out-
conceptual critiques and associated empirical findings come expectancies. “An outcome expectancy is defined as a
regarding the causal influence of expected outcomes on self- person’s estimate that a given behavior will lead to certain
efficacy, (c) summarize Bandura’s rebuttal to those critiques, outcomes” (Bandura, 1977, p. 193). According to self-efficacy
(d) discuss the conceptual contradiction inherent in Bandu- theory, outcome expectancies are clearly distinguished from
ra’s rebuttal, (e) discuss the theoretical and practical implica- self-efficacy because self-efficacy is perceived ability to do
tions of this contradiction, and (f) provide directions for a behavior, whereas outcome expectancies are judgments
future research. about the likelihood of outcomes that flow from behavior
(Bandura, 1977, 1986c, 1997).1
For a theory to be useful in explaining behavior, it must
Self-Efficacy Theory specify the causal pathways among the posited determinants
Self-efficacy theory—a component of social cognitive of behavior (Weinstein & Rothman, 2005). In self-efficacy
theory—focuses on the interrelationships among self-efficacy, theory, self-efficacy causally influences outcome expectan-
outcome expectancies, and behavior (Bandura, 1997). Self- cies, but not vice versa (Bandura, 1978, 1984, 1986c, 1995,
efficacy involves people’s perceptions that they are capable 1997, 1998, 2004, 2006b). This theorized relationship
of successfully performing a behavior (Bandura, 1977, 1986c, between the two constructs has been depicted in graphi-
1997). Thus, self-efficacy can be distinguished from inten- cal representations (e.g., Bandura, 1998, p. 629; Bandura,
tions because intentions involve willingness to do a behav- 2004, p. 146) and is a widely accepted tenet of self-efficacy
ior, whereas self-efficacy involves belief that one can do a theory (Maddux, 1991, 1993; Maddux & Stanley, 1986;
behavior (Bandura, 1997, p. 43). Self-efficacy is situation S. L. Williams, 1992). Indeed, in response to the proposal
specific; for example, people’s self-efficacy to resist cake is that outcome expectancies may causally influence self-
independent of their self-efficacy to solve math problems. efficacy judgments (Eastman & Marzillier, 1984), Bandura
Self-efficacy is involved with perceived capability rather (1997) stated:
than actual capability (Bandura, 1997). Finally, according to
self-efficacy theory, self-efficacy judgments are derived The outcomes people anticipate depend largely on
from four sources: mastery experiences (i.e., previous suc- their judgments of how well they will be able to per-
cessful behavioral performance), modeling (i.e., observation form in given situations. To claim . . . that people visu-
of successful behavioral performance), verbal persuasion, alize outcomes and then infer their own capabilities
and physiological and affective states at the time of the self- from the imagined outcomes is to invoke a peculiar
efficacy rating (Bandura, 1997). system of backward causation in which the outcomes
According to self-efficacy theory, self-efficacy should be that flow from actions are made to precede the actions.
assessed with respect to various “gradations of challenge” (p. 21)
(Bandura, 1997, 2006a). For example, basketball shooting
self-efficacy might be assessed by asking respondents to rate Thus, according to self-efficacy theory, outcome expectancies
their capability to make 5 of 10 shots, 6 of 10 shots, and so are not a causal determinant of self-efficacy.

Downloaded from psr.sagepub.com at PENNSYLVANIA STATE UNIV on April 8, 2016


Williams 3

The Critique: Expected Outcomes capability, respondents automatically perceive a context for
Causally Influence Self-Efficacy Ratings the behavior, which elicits expected outcomes of performing
the behavior in that context (Eastman & Marzillier, 1984;
Since Bandura’s (1977) introduction of self-efficacy, some Wolpe, 1978). The influence of expected outcomes on self-
authors have challenged the notion that self-efficacy is free efficacy judgments is made explicit in measures typically
from the causal influence of expected outcomes (Borkovec, used to assess self-regulatory efficacy for repetitive behav-
1978; Corcoran, 1991, 1995; Eastman & Marzillier, 1984; iors (Bandura, 2006a). As noted above, when formulating
Kazdin, 1978; Kirsch, 1982, 1985, 1986, 1995; Teasdale, their exercise self-efficacy
1978, Wolpe, 1978; Maddux, 1999). Indeed, a number of
experimental studies have demonstrated that outcome expec- individuals judge how well they can get themselves to
tancies do influence self-efficacy. In Kirsch’s (1982) semi- exercise regularly under various impediments, such as
nal study, snake-fearful college students were asked to rate when they are under pressure from work, are tired or
their perceived capability (i.e., self-efficacy) to approach and depressed; in foul weather; or when they have other
interact with a snake and to throw wadded-up paper into a commitments or more interesting things to do. (Bandura,
basket from increasing distances. Participants were then 1997, p. 43)
offered hypothetical incentives (i.e., expected outcomes) for
performing each behavior. Results showed that self-efficacy Responses to these self-efficacy items are a function of
for both behaviors increased as a result of increasing mon- expected outcomes of exercising when one expects that
etary incentives, with self-efficacy to approach the snake exercise will result in falling further behind at work, becoming
increasing significantly more than for the paper-throwing task more exhausted, getting cold and/or wet from exercising in
as a result of the increasing incentives. Kirsch’s (1982) find- bad weather, or missing an opportunity to do something
ings were replicated among light to moderate smokers. Spe- more interesting (see Figure 1). Interitem variability on
cifically, hypothetical incentives led to greater increases in these self-efficacy scales is an indication that perceived
self-efficacy for quitting smoking over escalating time periods capability to perform the behavior changes as a function of
than for making basketball shots over increasing distances expected outcomes of performing the behavior in the various
(Corcoran & Rutledge, 1989), thus indicating that expected contexts delineated in each item (Corcoran, 1995; Eastman
outcomes of behavior can influence self-efficacy and that this & Marzillier, 1984; Wolpe, 1978).
effect is stronger for behaviors that involve regulation of Expected outcomes influence self-efficacy ratings even
behavior (i.e., approaching a snake, quitting smoking) rather when respondents are not explicitly encouraged to consider
than specialized physical skills (i.e., basketball shooting). specific contexts for the behavior. As noted by Rhodes
The effects of expected outcomes on self-efficacy ratings and Courneya (2003), “Unmotivated individuals towards
have also been shown using actual incentives. In one study, a behavior will likely respond that a behavior is difficult, and
college students were randomly assigned either to receive lack confidence/capability to perform it, regardless of whether
(actual) escalating monetary incentives for longer exposure they perceive they have the skills, opportunities, or resources
to pain via the cold-pressor task or to not receive incentives to enact the behavior” (p. 81). For example, in a study con-
(Baker & Kirsch, 1991). Participants who were offered the ducted among chronic pain patients, participants were asked
incentives had higher ratings of self-efficacy and tolerated to rate their self-efficacy to perform a series of functional
the pain for longer periods. Importantly, both the presence of behaviors (Council, Ahern, Follick, & Kline, 1988). Self-
incentives and expected pain intensity were predictive of efficacy was a strong predictor of functional behavior. How-
self-efficacy for longer pain tolerance, which in turn ever, when questioned about the reason for their submaximal
influenced actual pain tolerance. Although the relationship ratings of self-efficacy (83% of responses reflected less than
between expected pain and self-efficacy to tolerate pain was 100% self-efficacy), participants cited expected pain (83%),
merely associative, the impact of expected monetary rewards fear of injury (12%), and lack of ability (5%), suggesting that
on self-efficacy to tolerate pain was causal because the pres- “although self-efficacy expectancies ultimately determined
ence of monetary incentives was manipulated based on performance, pain response expectancies may have exerted
random assignment. The same findings were obtained in their influence at an earlier point in a causal sequence”
a similar study in which snake approach behavior was tar- (Council et al., 1988, p. 328). Similarly, when college stu-
geted (Schoenberger, Kirsch, & Rosengard, 1991): Manipu- dents were asked to give the reason or reasons for their self-
lation of expected monetary rewards affected self-efficacy to efficacy for exercise, they indicated that expectations of
approach the snake, which in turn mediated the effects of improved health (30%), enjoyment (19%), and motivation
outcome expectancies on snake approach behavior. (37%) influenced their self-efficacy judgment (Rhodes &
In addition to experimental data, some researchers have Blanchard, 2007).
argued that outcome expectancies consistently influence Taken together, these studies show that outcome expec-
ratings on self-efficacy scales because, when rating their tancies causally influence self-efficacy ratings. Moreover,

Downloaded from psr.sagepub.com at PENNSYLVANIA STATE UNIV on April 8, 2016


4 Personality and Social Psychology Review XX(X)

types of evidence showing that self-efficacy is a more impor-


Outcome tant predictor of behavior than outcome expectancies. The
Self-Efficacy Behavior
Expectancies first type of evidence involves studies that have assessed
both self-efficacy and outcome expectancies and have shown
self-efficacy to be a vastly superior predictor of behavior (for
I expect that exercise I am capable of a review, see Bandura, 1995). The second type of evidence
will result in: exercising when:
involves correlational studies showing that self-efficacy is
• Falling further • I am under predictive of behavior even when the effects of outcome
behind at work pressure from work
• Becoming more • I am tired or
expectancy are statistically controlled (for reviews, see
exhausted depressed Bandura, 1984, 1991; S. L. Williams, 1992, 1995). These lines
• Getting cold and/or • There is foul
wet from weather
of evidence, however, do nothing to support or refute the
exercising in bad experimental evidence that self-efficacy judgments are caus-
weather
• Failing to follow • I have other
ally influenced by expected outcomes.
up on my other commitments The second theme of Bandura’s rebuttal is directed at
commitments
• Missing an • I have more
Kirsch’s interpretation of the findings that outcome expec-
opportunity to do interesting things tancies causally influence self-efficacy. Specifically, Bandura
something more to do
interesting
concedes that self-efficacy judgments can be causally influ-
(Adapted from enced by outcome expectancies but argues that, contrary to
Bandura, 1997, p.43) Kirsch and others (Cahill et al., 2006; Corcoran, 1995;
Kirsch, 1982, 1985, 1986, 1995), such a relationship does
Figure 1. Example of the causal influence of outcome not invalidate those self-efficacy judgments (Bandura, 1982,
expectancies on self-efficacy judgments 1986a, 1986b, 1995, 1997, 2007; also see Cervone &
Scott, 1995, p. 363). Indeed, in response to Kirsch’s (1982,
1985, 1986, 1995) critiques, Bandura (1986b) states,
assessments of self-efficacy in the studies reviewed
above are consistent with Bandura’s (1997, 2006a) guide- He [Kirsch] seems to assume that if one can boost
lines for constructing self-efficacy scales and with the people’s beliefs in their efficacy by the prospect of
vast body of self-efficacy research in which Bandura’s escalated benefits, this somehow challenges the status
guidelines have been followed. Thus, the findings imply of self-efficacy measures in phobic domains of func-
that judgments of self-efficacy in previous research in tioning, as though self-efficacy judgments are authen-
similar domains of functioning (e.g., anxiety-related avoid- tic only if unalterable by social influences. (p. 390;
ance behavior; pain-related behavior; health behaviors, also see Bandura, 1986a, p. 367)
including smoking and exercise) have likely been influ-
enced by outcome expectancies. It is reasonable to expect Thus, according to Bandura (e.g., 2007), self-efficacy ratings
that the same relationship occurs in additional behavioral are valid indicators of self-efficacy even when influenced by
domains. outcome expectancies.
Kirsch (1982, 1985, 1986, 1995) and others (Cahill et al.,
2006; Corcoran, 1995) have concluded from the data reviewed
above that self-efficacy ratings are invalid (i.e., do not actu- The Contradiction
ally reflect respondents’ perceived capability to perform the Although Bandura’s rebuttal to Kirsch and others preserves
behavior) when influenced by outcome expectancies because, the validity of existing operationalizations of self-efficacy
according to Bandura’s self-efficacy theory, outcome expec- (see Bandura, 1997, 2006a), it represents a conceptual con-
tancies cannot causally influence self-efficacy judgments tradiction in his writings on self-efficacy theory. That is, if
(Bandura, 1978, 1997, 2006b). Kirsch’s interpretation implies self-efficacy ratings remain valid when influenced by out-
that assessments of self-efficacy in the voluminous literature come expectancies, then empirical findings showing such a
examining self-efficacy in similar behavioral domains (e.g., relationship should also be considered valid. Such a relation-
anxiety-related avoidance behavior, pain-related behavior, ship, however, is contrary to Bandura’s repeated statements
health behavior), as well as the conclusions drawn from such that outcome expectancies cannot causally influence self-
literature, have also been invalid.2 efficacy (e.g., Bandura, 1978, 1997, 2006b; see the above
section on self-efficacy theory).
The status quo in self-efficacy research is consistent
The Rebuttal with the conceptual contradiction described above. That is,
Bandura’s rebuttal to the critique discussed above has had researchers use existing operationalizations of self-efficacy
two themes. First, Bandura (1984, 1991, 1995) cites two (which are causally influenced by expected outcomes) but do

Downloaded from psr.sagepub.com at PENNSYLVANIA STATE UNIV on April 8, 2016


Williams 5

not acknowledge that outcome expectancies influence self- causal independence of self-efficacy from outcome expec-
efficacy. The result is that (a) self-efficacy ratings are highly tancy (Rhodes & Blanchard, 2007; Rhodes & Courneya,
predictive of behavior because they encompass the causal 2003, 2004).
influence of expected outcomes (see Figure 1) and (b) out- The findings reviewed above show some promise for
come expectancies are not acknowledged as a source of operationalizing self-efficacy in a way that is independent of
self-efficacy judgments because this is inconsistent with outcome expectancies and thus consistent with self-efficacy
self-efficacy theory. It is this approach to self-efficacy theory. However, in using this assessment approach, all out-
research that has led to the disproportionate attention to self- come expectancies may not be adequately held constant
efficacy at the expense of attention to outcome expectancies with the phrase “if I wanted to” (Rhodes & Blanchard,
in the context of theoretical models and as targets of behav- 2007). Although Rhodes and Blanchard (2007) see this
ior change interventions. issue as a potential barrier to be overcome by further refine-
ment of measurement methodology, Marzillier and Eastman
(1984) see it as a more serious obstacle:
Reconciling Theoretical and
Operational Definitions of Self-Efficacy Since human activity can be seen as a continuous inter-
The conceptual contradiction discussed above can be change between behavior and environmental response
addressed in two ways. Either the operational definition of (outcome), any assessment about how well one is
self-efficacy must be modified such that expected outcomes going to perform a particular activity must take into
cannot influence self-efficacy (consistent with current con- account not only one’s own estimation of competence
ceptualizations of self-efficacy theory) or self-efficacy the- but also an assessment of what likely outcomes may
ory must be modified such that outcome expectancies can occur. (p. 259; also see Feather, 1992, p. 114; Teasdale,
influence self-efficacy (consistent with empirical findings 1978, p. 214)
using current operationalizations of self-efficacy). These
approaches are discussed below along with potential prob- Altering Self-Efficacy Theory to Be Consistent With Empirical
lems with each approach. Findings. If the existing operationalization of self-efficacy is
to be considered valid despite the fact that such ratings are
Altering Operational Definitions of Self-Efficacy to Be Consistent influenced by expected outcomes (e.g., Bandura, 2007),
With Self-Efficacy Theory. If the existing operational definition then self-efficacy theory must be modified to allow for the
of self-efficacy is considered invalid because it is influenced causal influence of outcome expectancy on self-efficacy.
by outcome expectancies (Cahill et al., 2006; Corcoran, 1995; However, parallel to the difficulty in operationalizing
Kirsch, 1982, 1985, 1986, 1995), then the operational defini- self-efficacy as independent from outcome expectancies (see
tion of self-efficacy must be modified to be consistent with above), self-efficacy may be difficult to conceptually and
self-efficacy theory. In such an operational definition, self- empirically disentangle from the influence of expected out-
efficacy ratings would not be causally influenced by expected comes if the current operational definition of self-efficacy is
outcomes. There is some evidence to suggest that such a maintained. For example, in interpreting someone’s self-
measurement approach is possible. For example, in a study efficacy to perform a breast self-examination, it would be
on use of the “broken-record technique” in assertiveness critical to determine whether any perceived incapability is a
training, participants were asked to rate their self-efficacy function of expected negative outcomes of such an exam or
by indicating agreement with the statement “I could use the perceived lack of skills to perform the exam. However, if there
broken-record technique if I wanted to [emphasis added]” are no attempts made to hold expected outcomes constant in the
(Maddux, Norton, & Stoltenberg, 1986, p. 785), thus assessment of self-efficacy, then assessments of self-efficacy
attempting to hold outcome expectancies constant. As a will encompass expected outcomes of the exam, making it dif-
result of this methodology, self-efficacy and outcome expec- ficult to conceptualize, label, and extract the variance in self-
tancies were uncorrelated but independently predictive of efficacy that is independent of outcome expectancies.
behavioral intention. Rhodes and colleagues (Rhodes & An additional problem with preserving the current
Blanchard, 2007; Rhodes & Courneya, 2003, 2004) also operationalization of self-efficacy is that, when using this
assessed self-efficacy with the qualifier “if I wanted to” and approach, respondents are able to impose their own defini-
found that relationships between self-efficacy and outcome tion of capability vis-à-vis expected outcomes. For example,
expectancies decreased relative to a traditional, unqualified imagine that two dieters with equal expected outcomes for
assessment of self-efficacy. In addition, participants had cake eating (e.g., expected deliciousness of the cake) are
higher mean ratings on self-efficacy items with the “if asked to rate their capability to resist eating a piece of cake
I wanted to” qualifier than self-efficacy items without the (i.e., self-efficacy for resisting cake). Dieter 1 personally
qualifier, and the qualified self-efficacy ratings had sig- defines capability in such a way that expected outcomes of
nificantly weaker relationships with intentions, suggesting cake eating cannot influence his capability to resist the cake

Downloaded from psr.sagepub.com at PENNSYLVANIA STATE UNIV on April 8, 2016


6 Personality and Social Psychology Review XX(X)

and thus rates his perceived capability to resist the cake (i.e., to mean that self-efficacy—independent of expected
self-efficacy) as 100%, despite his high expectations that the outcomes—is an important predictor of behavior. Likewise,
cake will taste delicious. Dieter 2 has a more liberal personal previous self-efficacy research must be interpreted with
definition of capability, in which expected outcomes of cake the knowledge that participants have been free to refer to
eating can influence his capability, and thus rates his capa- their own definitions of capability when responding to
bility to resist the cake (i.e., self-efficacy) as 10%, based on self-efficacy items.
his equally high expectation that the cake will taste deli- In conducting future research on self-efficacy, research-
cious. An observer concludes that Dieter 1 has stronger self- ers must be clear about their theoretical position regarding
efficacy for resisting cake than Dieter 2. However, the self-efficacy and outcome expectancies (i.e., self-efficacy
apparent variability in cake-resisting self-efficacy is an arti- either is or is not causally independent from outcome expec-
fact of the two dieters’ differing definitions of capability. tancies). Likewise, researchers should operationally define
It can be seen from this example that ratings of self- self-efficacy in a way that is consistent with their theoretical
efficacy are highly influenced by the way in which capabil- perspective. If self-efficacy is theorized to be independent of
ity is defined by respondents.3 Thus, self-efficacy assessments outcome expectancies, this may involve qualifying perceived
that do not specify the definition of capability vis-à-vis capability items with the phrase “if I wanted to” (Maddux
expected outcomes provide a test of how people use lan- et al., 1986; Rhodes & Blanchard, 2007; Rhodes & Courneya,
guage. The outcomes of that language test, for which there 2003, 2004) as well as explicitly informing respondents that
are no predefined right or wrong answers, are implicitly perceived capability is independent of expected outcomes
incorporated into self-efficacy ratings. In short, some partici- (Cahill et al., 2006). Alternatively, if ratings of perceived
pants may incorporate expected outcomes into their self- capability (i.e., self-efficacy) are theorized to incorporate
efficacy ratings, whereas others may believe that it is against the causal influence of outcome expectancies, then research-
natural language conventions to do so. ers should explicitly inform respondents that they should
Indeed, Cahill and colleagues (2006) showed that partici- consider expected outcomes when rating their self-efficacy.
pants who were given vignettes illustrating the independence The above recommendations are relevant for all self-efficacy
of capability from expected outcomes had greater increases subtypes and regardless of whether or not outcome expectan-
in subsequent ratings of self-efficacy than participants not cies are also assessed.
given the vignettes. The findings show that respondents’ Finally, both proposed alternatives to reconciling self-
self-efficacy ratings are dependent on the meaning of the word efficacy conceptualization and operationalization have poten-
capability vis-à-vis expected outcomes. Moreover, the find- tial pitfalls. Thus, depending on one’s perspective, research
ings suggest that at least some respondents will consider the is needed to (a) improve methodology for assessing self-
influence of expected outcomes in rating their self-efficacy efficacy as independent from outcome expectancies or
unless explicitly instructed not to do so. If meanings of the (b) conceptually and empirically disentangle self-efficacy from
words (i.e., capability) that define a psychological construct the influence of expected outcomes.
(i.e., self-efficacy) are moving targets, then how can one test
hypotheses based on that construct?
Practical Implications
Regardless of whether one believes that ratings of per-
Summary and Recommendations ceived capability are valid indicators of self-efficacy when
for Future Research influenced by outcome expectancies, such ratings of per-
The findings reviewed herein indicate that currently used ceived capability are indeed influenced by outcome expec-
operational definitions of self-efficacy are likely to be influ- tancies (Baker & Kirsch, 1991; Corcoran & Rutledge, 1989;
enced by expected outcomes. Thus, it is critical that research- Council et al., 1988; Kirsch, 1982; Rhodes & Blanchard,
ers either (a) modify operationalizations of self-efficacy to 2007; Shoenberger et al., 1991) and highly predictive of
be independent of expected outcomes or (b) acknowledge behavior (Holden, 1991; Holden et al., 1990; Moritz et al.,
the causal influence of outcome expectancies on self-efficacy. 2000; Multon et al., 1991; Sadri & Robertson, 1993; Stajkovic
In addition, it is recommended that researchers discontinue & Luthans, 1998). Thus, understanding people’s expected
current practices in which respondents are allowed to refer outcomes of behavior is critical to understanding what they
to their personal definitions of capability when responding say they are capable of (regardless of whether this is a valid
to self-efficacy items—definitions that may or may not be indicator of their self-efficacy), which in turn has been shown
consistent with those of other respondents and researchers. to be highly predictive of behavior. To return to the example
Moreover, these issues must be considered when inter- of cake eating, consider a dieter who reports low self-
preting prior research on self-efficacy. Specifically, prior efficacy to resist cake based on his expectation that the cake
research on self-efficacy, which shows self-efficacy to will taste delicious and thus eats the cake. Depending on
be highly predictive of behavior, must not be interpret­ed one’s perspective, the dieter’s self-efficacy rating may or

Downloaded from psr.sagepub.com at PENNSYLVANIA STATE UNIV on April 8, 2016


Williams 7

may not be considered valid because it is dependent on For example, a single mother may say that she “cannot”
outcome expectancies. Either way, the dieter’s expected exercise 5 days per week because she must care for her chil-
outcomes of eating or resisting the cake will influence dren, complete household chores, and work long hours to
his self-efficacy rating (valid or not) and ultimately his have enough money for food and housing. Although most
behavior. If such outcome expectancies are not thoroughly would agree that these are good reasons for not exercising,
assessed, it will appear as though the dieter’s low per- they, nonetheless, reflect a lack of motivation to exercise
ceived capability—independent of expected outcomes—is based on expected consequences, including expectations that
the critical determinant of his behavior. Thus, reports of her children would be neglected, her home would be in disar-
low self-efficacy should prompt clinicians to consider ray, and she would ultimately lose her job. The extent to
assessment of and intervention on expected outcomes of which these expected outcomes of exercise can validly influ-
the target behavior. ence self-efficacy is up for debate. Either way, such expected
In doing so, outcome expectancies should be conceptu- outcomes must be recognized as the crucial mechanism of
alized and operationalized broadly. Truncated conceptual- change so that theoretical frameworks can be appropriately
izations of outcome expectancies have led to a focus on revised, research findings can be consistently interpreted,
(a) temporally distal, but not temporally proximal outcomes; and novel interventions can be designed.
(b) instrumental, but not affective outcomes; (c) positive, but
not negative outcomes; and (d) outcomes of the target Acknowledgments
behavior, but not outcomes of salient alternative behaviors I would like to thank Richard Winett and Beth Lewis as well as
(D. M. Williams, Anderson, & Winett, 2005; for some the anonymous reviewers for their comments on previous versions
exceptions, see Gilovich, Griffin, & Kahneman, 2002; Hsee, of this article. Additional thanks to Eileen Anderson and Robert
Hastie, & Chen, 2008; Lawton, Conner, & McEachan, 2009; Stephens for stimulating my interest in this topic and tolerating
Mellers, 2000; van der Pligt & de Vries, 1998; D. M. Williams, lengthy debates on the causal precedence of outcome expectancy
2008). Indeed, in changing most behaviors that are the target and self-efficacy.
of self-regulatory interventions, people must overcome expec-
tations of immediate, negative affective outcomes associated Declaration of Conflicting Interests
with the adaptive behavior (e.g., nicotine withdrawal symp- The author(s) declared no potential conflicts of interests with
toms, displeasure of resisting high-calorie foods, aversive respect to the authorship and/or publication of this article.
phobic stimuli) and resist expectations of immediate positive
affective outcomes of maladaptive behaviors (e.g., the plea- Financial Disclosure/Funding
sure of smoking and indulging in high-calorie foods and the The author(s) received no financial support for the research and/or
relief of avoiding phobic stimuli). Alternatively, expected authorship of this article.
positive outcomes of adaptive behaviors are often distal and
uncertain (e.g., reduced risk of disease and early death). Notes
Such lopsided expectations favoring maladaptive behavior 1. The preceding definitions of self-efficacy and outcome expec-
are what lead people to not want to perform adaptive behav- tancy are consistent with self-efficacy theory as articulated in
iors, thus making behavioral change notoriously difficult to the most recent publications by Bandura (e.g., 1997, 2006a).
adopt and maintain. Some researchers have been critical of alleged changes over
time in Bandura’s definitions of self-efficacy (e.g., Corcoran,
1995; Eastman & Marzillier, 1984) and narrow definition of
Concluding Remarks outcome expectancy (e.g., Kirsch, 1985, 1995). Although these
The notion that self-efficacy as currently operationalized issues are relevant to this article, they are not necessary for the
(valid or not) is a function of outcome expectancies is poten- conceptual contradiction that is the focus of this article. That
tially controversial because it implies that failure to engage in is, the purpose of this article is to highlight a previously un-
adaptive behavior is often a result of lack of motivation based explored conceptual problem in Bandura’s writings that exists
on perceived consequences rather than incapability indepen- when defining self-efficacy and outcome expectancy in a way
dent of perceived consequences. This may be especially dis- that is consistent with his most recent writings on self-efficacy
concerting for clients with addictions or phobias (Bandura, theory.
1986a, 2007; S. L. Williams, 1995). However, expected out- 2. Kirsch (1982, 1985, 1986, 1995) also argues that the influence
comes of behavior need not be viewed as trivial reasons for of outcome expectancies on self-efficacy ratings is an indica-
performing (or not performing) a behavior. Indeed, “in real tion that such self-efficacy ratings actually reflect behavioral
life, alternatives may come with a heavy price in emotional, intentions and that this phenomenon occurs when the target
financial, or physical consequences, which may result in other behavior is a “choice” behavior rather than a “skill” behavior.
avenues being pursued; nevertheless, those alternatives are These additional arguments are often a target of Bandura’s
available” (Corcoran, 1995, p. 204). (e.g., 1995) rebuttals; however, they are unnecessary to the

Downloaded from psr.sagepub.com at PENNSYLVANIA STATE UNIV on April 8, 2016


8 Personality and Social Psychology Review XX(X)

main thesis of the present article—that outcome expectancies Borkovec, T. D. (1978). Self-efficacy: Cause or reflection of behav-
can causally influence self-efficacy ratings. ioral change? Advances in Behaviour Research and Therapy,
3. When assessing perceived capability (i.e., self-efficacy), per- 1, 163-170.
ceptions of capability are at the discretion of the perceiver (e.g., Cahill, S. P., Gallo, L. A., Lisman, S. A., & Weinstein, A. (2006).
Bandura, 1997), but it is necessary that researchers and respon- Willing or able? The meanings of self-efficacy. Journal of
dents are all using the same definition of capability. Social and Clinical Psychology, 25, 196-209.
Cervone, D., & Scott, W. D. (1995). Self-efficacy theory of behav-
References ioral change: Foundations, conceptual issues, and therapeutic
Ajzen, I. (1985). From intentions to actions: A theory of planed implications. In W. T. O’Donohue & L. Krasner (Eds.), Theories
behavior. In J. Kuhl & J. Beckmann (Eds.), Action control: From of behavior therapy: Exploring behavior change (pp. 349-383).
cognition to behavior (pp. 11-39). New York, NY: Springer. Washington, DC: American Psychological Association.
Baker, S. L., & Kirsch, I. (1991). Cognitive mediators of pain per- Corcoran, K. J. (1991). Efficacy, “skills,” reinforcement, and choice
ception and tolerance. Journal of Personality and Social Psy- behavior. American Psychologist, 46, 155-157.
chology, 61, 504-510. Corcoran, K. J. (1995). Understanding cognition, choice, and
Bandura, A. (1977). Self-efficacy: Toward a unifying theory of behavior. Journal of Behavior Therapy and Experimental Psy-
behavioral change. Psychological Review, 84, 191-215. chiatry, 26, 201-207.
Bandura, A. (1978). Reflections on self-efficacy. Advances in Corcoran, K. J., & Rutledge, M. W. (1989). Efficacy expectation
Behaviour Research and Therapy, 1, 237-269. changes as a function of hypothetical incentives in smokers.
Bandura, A. (1982). Self-efficacy mechanism in human agency. Psychology of Addictive Behaviors, 3, 22-29.
American Psychologist, 37, 122-147. Council, J. R., Ahern, D. K., Follick, M. J., & Kline, C. L. (1988).
Bandura, A. (1984). Recycling misconceptions on perceived self- Expectancies and functional impairment in chronic low back
efficacy. Cognitive Therapy and Research, 8, 231-255. pain. Pain, 33, 323-331.
Bandura, A. (1986a). The explanatory and predictive scope of self- Eastman, C., & Marzillier, J. S. (1984). Theoretical and method-
efficacy theory. Journal of Social and Clinical Psychology, 4, ological difficulties in Bandura’s self-efficacy theory. Cognitive
359-373. Therapy and Research, 8, 213-229.
Bandura, A. (1986b). Fearful expectations and avoidant actions as Feather, N. T. (1982). Expectations and actions: Expectancy-value
coeffects of perceived self-inefficacy. American Psychologist, models in psychology. Hillsdale, NJ: Erlbaum.
41, 389-391. Feather, N. T. (1992). Values, valences, expectations, and actions.
Bandura, A. (1986c). Social foundations of thought and action. Journal of Social Issues, 48, 109-124.
Upper Saddle River, NJ: Prentice Hall. Gilovich, T., Griffin, D., & Kahneman, D. (2002). Heuristics and
Bandura, A. (1991). Human agency: The rhetoric and the reality. biases: The psychology of intuitive judgment. New York, NY:
American Psychologist, 46, 157-162. Cambridge University Press.
Bandura, A. (1995). On rectifying conceptual ecumenism. In Holden, G. (1991). The relationship of self-efficacy appraisals to
J. E. Maddux (Ed.), Self-efficacy, adaptation, and adjustment: subsequent health related outcomes: A meta-analysis. Social
Theory, research, and application (pp. 347-375). New York, Work in Health Care, 16, 53-93.
NY: Plenum. Holden, G., Moncher, M. S., Schinke, S. P., & Barker, K. M. (1990).
Bandura, A. (1997). Self-efficacy: The exercise of control. New York, Self-efficacy of children and adolescents: A meta-analysis. Psy-
NY: Freeman. chological Reports, 66, 1044-1046.
Bandura, A. (1998). Health promotion from the perspective of Hsee, C. K., Hastie, R., & Chen, J. (2008). Hedonomics: Bridging
social cognitive theory. Psychology and Health, 13, 623-649. decision research with happiness research. Perspectives on Psy-
Bandura, A. (2004). Health promotion by social cognitive means. chological Science, 3, 224-243.
Health Education and Behavior, 31, 143-164. Hyde, J., Hankins, M., Deale, A., & Marteau, T. M. (2008). Inter-
Bandura, A. (2006a). Guide for constructing self-efficacy scales. In ventions to increase self-efficacy in the context of addiction
F. Pajares & T. C. Urdan (Eds.), Self-efficacy beliefs of adoles- behaviours: A systematic literature review. Journal of Health
cents (pp. 307-337). Greenwich, CT: Information Age. Psychology, 13, 607-623.
Bandura, A. (2006b). Toward a psychology of human agency. Per- Kazdin, A. E. (1978). Conceptual and assessment issues raised
spectives on Psychological Science, 1, 164-180. by self-efficacy theory. Advances in Behaviour Research and
Bandura, A. (2007). Much ado over a faulty conception of per- Therapy, 1, 177-185.
ceived self-efficacy grounded in faulty experimentation. Kirsch, I. (1982). Efficacy expectations or response predictions:
Journal of Social and Clinical Psychology, 26, 641-658. The meaning of efficacy ratings as a function of task characteris-
Blanchard, C. M., Fortier, M., Sweet, S., O’Sullivan, T., Hogg, W., tics. Journal of Personality and Social Psychology, 42, 132-136.
Reid, R. D., et al. (2007). Explaining physical activity levels Kirsch, I. (1985). Self-efficacy and expectancy: Old wine with
from a self-efficacy perspective: the physical activity counsel- new labels. Journal of Personality and Social Psychology, 49,
ing trial. Annals of Behavioral Medicine, 34, 323-328. 824-830.

Downloaded from psr.sagepub.com at PENNSYLVANIA STATE UNIV on April 8, 2016


Williams 9

Kirsch, I. (1986). Response expectancy and phobic anxiety: Rhodes, R. E., & Courneya, K. S. (2003). Self-efficacy, control-
A reply to Wilkins and Bandura. American Psychologist, 41, lability and intention in the theory of planned behavior: Mea-
1391-1393. surement redundancy or causal independence? Psychology and
Kirsch, I. (1995). Self-efficacy and outcome expectancy: A conclud- Health, 18, 79-91.
ing commentary. In J. E. Maddux (Ed.), Self-efficacy, adaptation, Rhodes, R. E., & Courneya, K. S. (2004). Differentiating motiva-
and adjustment: Theory, research, and application (pp. 341-345). tion and control in the theory of planned behavior. Psychology,
New York, NY: Plenum. Health, and Medicine, 9, 205-215.
Lawton, R., Conner, M., & McEachan, R. (2009). Desire or reason: Rosenstock, I. M., Strecher, V. J., & Becker, M. H. (1988). Social
Predicting health behaviors from affective and cognitive atti- learning theory and the health belief model. Health Education
tudes. Health Psychology, 28, 56-65. Quarterly, 15, 175-183.
Maddux, J. E. (1991). Self-efficacy. In C. R. Snyder & D. R. Forsyth Rotter, J. B. (1954). Social learning and clinical psychology.
(Eds.), Handbook of social and clinical psychology: The health Englewood Cliffs, NJ: Prentice Hall.
perspective (pp. 57-78). New York, NY: Pergamon. Sadri, G., & Robertson, I. T. (1993). Self-efficacy and work related
Maddux, J. E. (1993). Social cognitive models of health and exer- behavior: A review and meta-analysis. Applied Psychology: An
cise behavior: An introduction and review of conceptual issues. International Review, 42, 139-152.
Journal of Applied Sport Psychology, 5, 116-140. Schoenberger, N. E., Kirsch, I., & Rosengard, C. (1991). Cogni-
Maddux, J. E. (1999). Expectancies and the social-cognitive per- tive theories of human fear: An empirically derived integration.
spective: Basic principles, processes, and variables. In I. Kirsch Anxiety Research, 4, 1-13.
(Ed.), How expectancies shape experience (pp. 17-39). Stajkovic, A. D., & Luthans, F. (1998). Self-efficacy and work-
Washington, DC: American Psychological Association. related performance: A meta-analysis. Psychological Bulletin,
Maddux, J. E., Norton, L. W., & Stoltenberg, C. D. (1986). Self- 124, 240-261.
efficacy expectancy, outcome expectancy, and outcome value: Teasdale, J. D. (1978). Self-efficacy: Toward a unifying theory
Relative effects on behavioral intentions. Journal of Personality of behavioural change? Advances in Behaviour Research and
and Social Psychology, 51, 783-789. Therapy, 1, 211-215.
Maddux, J. E., & Rogers, R. W. (1983). Protection motivation van der Pligt, J., & de Vries, N. K. (1998). Expectancy-value mod-
and self-efficacy: A revised theory of fear appeals and atti- els of health behavior: The role of salience and anticipated
tude change. Journal of Experimental Social Psychology, 19, affect. Psychology and Health, 13, 289-305.
469-479. Weinstein, N. D., & Rothman, A. J. (2005). Commentary: Revi-
Maddux, J. E., & Stanley, M. A. (1986). Self-efficacy theory in talizing research on health behavior theories. Health Education
contemporary psychology: An overview. Journal of Social and Research, 20, 294-297.
Clinical Psychology, 4, 249-255. Williams, D. M. (2008). Exercise, affect, and adherence: An inte-
Marks, R., Allegrante, J. P., & Lorig, K. (2005). A review and synthe- grated model and a case for self-paced exercise. Journal of
sis of research evidence for self-efficacy-enhancing interventions Sport and Exercise Psychology, 30, 471-496.
for reducing chronic disability: Implications for health education Williams, D. M., Anderson, E. S., & Winett, R. A. (2005). A review
practice (part II). Health Promotion Practice, 6, 148-156. of the outcome expectancy construct in physical activity research.
Marzillier, J. S., & Eastman, C. (1984). Continuing problems with Annals of Behavioral Medicine, 29, 70-79.
self-efficacy theory: A reply to Bandura. Cognitive Therapy and Williams, S. L. (1992). Perceived self-efficacy and phobic dis-
Research, 8, 257-262. ability. In R. Schwarzer (Ed.), Self-efficacy: Thought control of
Mellers, B. A. (2000). Choice and the relative pleasure of conse- action (pp. 149-176). Washington, DC: Taylor and Francis.
quences. Psychological Bulletin, 126, 910-924. Williams, S. L. (1995). Self-efficacy and anxiety and phobic disor-
Moritz, S. E., Feltz, D. L., Fahrbach, K. R., & Mack, D. E. (2000). ders. In J. E. Maddux (Ed.), Self-efficacy, adaptation, and adjust-
The relation of self-efficacy measures to sport performance: ment: Theory, research, and application (pp. 69-107). New York,
A meta-analytic review. Research Quarterly for Exercise and NY: Plenum.
Sport, 71, 280-294. Wolpe, J. (1978). Self-efficacy theory and psychotherapeutic change:
Multon, K. D., Brown, S. D., & Lent, R. W. (1991). Relation of A square peg for a round hole. Advances in Behaviour Research
self-efficacy beliefs to academic outcomes: A meta-analytic and Therapy, 1, 231-236.
investigation. Journal of Counseling Psychology, 38, 30-38. Woodgate, J., Brawley, L. R., & Weston, Z. J. (2005). Maintenance
Rhodes, R. E., & Blanchard, C. M. (2007). What do confidence cardiac rehabilitation exercise adherence: Effects of task and
items measure in the physical activity domain? Journal of self-regulatory self-efficacy. Journal of Applied Social Psychol-
Applied Social Psychology, 37, 759-774. ogy, 35, 183-197.

Downloaded from psr.sagepub.com at PENNSYLVANIA STATE UNIV on April 8, 2016

You might also like