This article is for use only of students enrolled in GPS 520: Advanced Social Psychology for purposes associated with this course and may not be retained or further disseminated.

Leary, M. R., & Miller, R. S. (1986).  Expectancies and behavior change.  In M. R. Leary & R. S. Miller, Social Psychology and Dysfunctional Behavior: Origins, Diagnosis, and Treatment (pp. 179-194).  New York: Springer-Verlag Inc.

Expectancies and Behavior Change

Envision an unattractive, wary, poorly educated client encountering a pessimistic, burned-out counselor who dreads coming to work each morning. Then contrast that scenario with that of an attractive, trusting, well-educated client meeting an enthusiastic, confident counselor who is glad to be of help. Both intuition and research results (Garfield, 1978) predict that therapeutic change is more likely in the latter case than in the former, but why exactly is that so? A number of explanations are possible, but this chapter focuses on the possibility that change is more likely in the second scenario because both counselor and client expect it to occur. Their expectations may influence the outcome of therapy by leading to behavior that produces desirable changes that, in the absence of those expectations, would not have occurred. In short, the belief that change will occur may be a powerful component in psychological therapy.

The expectations with which both client and counselor enter the therapeutic relationship have long been of interest to clinical researchers (e.g., Goldstein, 1962). There are many variables to forecast-whether the participants will like each other, how long therapy will last, what it will be like, whether it will be successful-and clients' global expectations determine whether they enter therapy to begin with (Tinsley, Brown, de St. Aubin, & Lucek, 1984). Furthermore, once therapy begins, clients' expectations of what it involves may determine whether they continue treatment or drop out (see Garfield, 1978, for a review). Clients who expect direct advice and rapid cures, for instance, are often disappointed, and it is valuable to prepare clients for therapy by showing them in advance exactly what it entails (e.g., Wilson, 1985). In addition, it is desirable that therapist and client share the same expectations regarding the nature of treatment (Garfield, 1978), and there is even evidence that certain therapies are better suited than others to clients with particular expectancies (Abramowitz, Abramowitz, Roback, & Jackson, 1974; Kilmann, Albert, & Sotile, 1975).

We focus in this chapter on one role of expectations in counseling and psychotherapy, drawing on social psychological studies that show how our beliefs can become behavioral realities. We examine the possibility that expectations of psychological change on the part of both therapists and clients can facilitate or even create such change. For their part, therapists may occasionally construct self-fulfilling prophecies, obtaining certain therapeutic outcomes largely because they expect to. Clients, too, make predictions regarding the outcome of treatment, and they may master their problems mainly because they believe they can. Indeed, we examine the argument that psychological therapies succeed by enhancing clients' estimates of their personal abilities, or judgments of self-efficacy. Finally, we address the possibility that the expectation of change underlies most successful therapies.  

Therapist Expectancies

For decades, clinical researchers have grappled with the methodological difficulties of demonstrating that psychological treatments have effects above and beyond the "nonspecific" or "placebo" effects of coming under a therapist's care (for thoughtful discussion of this issue see Critelli & Neumann, 1984; Frank, 1973; Lick & Bootzin, 1975; Rosenthal & Frank, 1956; and Shapiro & Morris, 1978). It is widely recognized that simply meeting an attentive, supportive therapist can have desirable effects regardless of the techniques the therapist employs. One influential component of these placebo effects is the therapist's expectation that he or she can help and that treatment will be successful. Even reviewers who prefer to make little of therapist expectancy effects grudgingly admit that therapy is dramatically affected by the therapist's beliefs about the likelihood of change (Wilkins, 1977).

Importantly, a therapist's expectation of improvement is not merely a predictor of beneficial change; such expectations can create change that would not otherwise occur. Berman and Wenzlaff (1983) randomly manipulated the expectations of 17 therapists at a university counseling center by informing them that pretests had indicated that one of their new clients would rapidly improve; for a second (control) client, the therapists were told that no firm prediction could be made. When each therapist and both of his or her clients were monitored through six counseling sessions, a clear effect of positive expectations was obtained. The therapists believed that clients whom they had expected to improve had improved more than the controls, and the clients shared those views. The positive-expectancy clients actually; reported substantially greater reductions in anxiety and depression (though the groups did not differ in interpersonal problems and somatic complaints). It seems undeniably true, as Goldstein (1962) suggested years ago, that all else being equal, prospective clients should be assigned to the therapists who hold the most favorable prognostic expectations for them.

Why do therapists' expectations have these positive effects? As we saw in chapter 8, one possibility is that the improvement is more apparent than real; an expectation of change may lead both counselor and client to overestimate how much change has actually occurred (Conway & Ross, 1984). The benefits of therapy are often far from illusory, however. How might positive expectations become behavioral realities?

Self-Fulfilling Prophecies

By acting in accord with their favorable expectations, therapists may set into motion self-fulfilling prophecies. They may treat clients more warmly, encourage them more, praise small successes more highly, and in general behave very differently than they would have had their expectations been less positive. Such generous, attentive treatment may then elicit from clients more desirable change than would have been obtained had the therapists not been so supportive. In short, the therapists' (potentially false) expectations of success may become true largely because the therapists' beliefs lead them to behave in ways that make them come true.

A wide variety of social psychological studies demonstrate that such processes occur. Perhaps the best known example is Rosenthal and Jacobson's (1968) study of teacher expectations and student achievement. These researchers led elementary school teachers to believe that randomly selected students, whatever their prior achievements, would improve substantially during the school year. Indeed, at the end of the year, those students had improved more than their classmates. Further study showed that such remarkable effects do not always occur (Rosenthal & Rubin, 1978), but when they do they are often sizeable. In turn, students' expectations regarding the quality of the teaching they will receive can affect their teachers' behavior as well (Feldman & Prohaska, 1979; Feldman & Theiss, 1982).

The teacher expectation studies demonstrate that self-fulfilling prophecies can have happy endings, but, unfortunately, negative expectations can also come true. We saw in chapter 8, for example, that men who believed their female partners were unattractive elicited drab behavior from them (Snyder et al., 1977). Similarly, when men know they are interacting with women they often elicit more feminine, sex-typed behavior from them than they do when they falsely believe their female partners are other men (Skrypnek & Snyder, 1982).

In fact, it is possible for negative expectations to become regrettably persistent realities. Snyder and Swann (1978b) showed that behavior elicited from a person in response to a self-fulfilling prophecy can mold that person's responses in later situations as well. In their study, pairs of male subjects competed on a reaction-time task, and on alternating trials each man had an aversive "noise weapon" at his disposal. One of the subjects, the "perceiver," was told that his opponent, the "target," was either sensitive, kind, and cooperative (in short, nonhostile) or aggressive, competitive, and cruel.

Both subjects were also informed that extensive use of the noise weapon was usually either a dispositional characteristic, depending on the type of person one was, or a situational strategy, depending on the play of one's opponent. In any case, the "perceivers" who believed they were playing a hostile, cruel opponent began the game by delivering higher levels of noise than did those who thought their partners were nonhostile, and in line with a selffulfilling prophecy, the "hostile" targets responded with more noise of their own. In brief, perceivers who expected hostility got it.

The most provocative results of this study, however, came when the perceivers were replaced with new naive subjects and the "hostile" targets again played the game. If they had been led to believe that their frequent use of the noise weapon was a dispositional characteristic, they continued to act in a hostile manner, delivering more noise than did those who believed their earlier aggression had been largely situational. Under some circumstances, then, the targets of prejudicial personal evaluations not only come to act in a manner that confirms those judgments, but continue to act that way even when no one expects them to (cf. Fazio et al., 198 1). One can imagine how it could be hard for a "mental patient" who is widely feared or rejected to act in a manner that is neither threatening nor withdrawn. People often elicit from others the behavior they expect, whether or not it is desirable.

Darley and Fazio (1980) have delineated the specific steps by which self-fulfilling prophecies may occur (Figure I I - 1). Their analysis focuses on the preconceptions and expectations of one member of a dyad, examining how that person's judgments can ultimately affect the other's behavior. (They do recognize, however, that both members of a dyad may have preconceptions that come to influence the behavior of their partners.)

First, the person, or perceiver, forms an expectancy about the other person, or target, that predicts how the target will behave. Various information about the target, such as age, race, sex, physical attractiveness, and social class, may affect the perceiver's judgments in ways of which the perceiver is unaware, and if the perceiver is able to observe the target's behavior, the personalistic bias makes dispositional attributions likely (see chapter 8).

In an important second step, the perceiver acts, usually in accord with his or her expectancy. The perceiver's judgments are not always reflected in his or her behavior; if subjects expect hostility from another person, for instance, they may begin interaction by being especially pleasant, hoping to divert the other's wrath (Ickes, Patterson, Rajecki, & Tanford, 1982). However, it is usually hard for people to keep from subtly communicating what they really think about others (DePaulo et al., 1985). For example, when Word, Zanna, and Cooper (1974) asked white subjects to interview both white and black job applicants (all of whom were highly trained confederates), they found that the black applicants received shorter interviews and heard more speech errors from the interviewers, who sat further away. The whites' relative uneasiness during the interviews with the black applicants was manifest, whether they knew it or not. Indeed, comprehensive meta-analyses of 136 expectancy studies by Harris and Rosenthal (1985) have shown that perceivers' expectancies influence a wide range of their nonverbal behavior toward their targets. Perceivers with favorable expectations, for instance, interact longer and more often with their targets, sharing more eye contact, sitting closer, smiling more, asking more questions,- and encouraging more responses than do perceivers with less positive expectancies. Our earlier supposition that therapists with favorable expectations are likely to be more encouraging, enthusiastic, and supportive may be close to the mark; perceivers with positive expectations create altogether warmer interpersonal climates for their targets.

The third step of Darley and Fazio's (1980) model involves the target's interpretation of the perceiver's behavior. If the perceiver's behavior is seen as being caused by the perceiver's personality, the self-fulfilling prophecy created in this interaction can generalize to entirely new interactions with the perceiver later on. Alternatively, the target may believe that he or she is the cause of the perceiver's action; in this instance, if the target feels the perceiver has misread him or her, the target may try to change the perceiver's judgments, blocking the self-fulfilling prophecy (as we will shortly see). In any case, the target draws some conclusion from the perceiver's behavior. In a second interview study, Word et al. (1974) trained two confederates to portray the same relaxed or uneasy interviewing styles that white interviewers had displayed with white and black applicants, respectively. Real subjects were then interviewed by the confederates, and whether they were black or white, the applicants judged the "uneasy" interviewer to be unfriendly. The subjects were clearly aware of how the interviewer was acting, and, like most observers, they attributed his behavior to his dispositions.

In the fourth step, the target responds to the perceiver. Unless the target is actively trying to dispel the perceiver's expectancy, this response is usually similar to the perceiver's action. Enthusiasm is usually met with interest (Snyder et al., 1977), hostility with counteraggression (Snyder & Swann, 1978b). In the study by Word et al. (1974), subjects who interacted with "uneasy" interviewers sat further away, spoke more haltingly, appeared more nervous, and came off less positively than those speaking with interviewers who were relaxed (cf. Harris & Rosenthal, 1985).

Such is the nature of a self-fulfilling prophecy that, as the perceiver interprets the target's response in the fifth step of the model, the perceiver is unlikely to recognize the role he or she played in producing it. Perhaps without a word being said, the perceiver may find in the target the behavior he or she expected; what better evidence is there that the expectancy was accurate?

Consider a real-life analogue of the Word et al. (1974) study. A white vocational counselor with a stereotypical conception of blacks is uneasy at the thought of interviewing them, and, without meaning to, acts that way. Black job-seekers faced with a distant, apparently unfriendly interviewer become uneasy themselves and appear inept. The interviewer finds, once again, that his or her expectations have been confirmed and becomes even more certain that, "It's too bad, but they just can't cut it." What is more, it is possible, in a final step of Darley and Fazio's (1980) model, for a target to interpret his or her actions in the same way as the perceiver does, internalizing them and continuing to behave that way on other occasions (Fazio et al., 1981; Snyder & Swann, 1978b).

Thus, a perceiver's expectations can substantially influence a target's behavior as both parties base their actions on their judgments of the other and on their prior interaction. As well as describing the perpetuation of social stereotypes, self-fulfilling prophecies help explain the effects of therapists' expectations. It is likely that the positive expectations of the therapists in Berman and Wenzlafrs (1983) study created positive outcomes for their clients by directly influencing the "nonspecific" therapy the clients received. When counselors held favorable expectations, clients probably received more positive regard, were encouraged to attempt more by a more enthusiastic and attentive counselor, and were urged to internalize their successes to a greater degree. They might even have spent more time with their counselors overall (Fehrenbach & O'Leary, 1982).

Given the potential benefits of positive expectations and the corresponding harmful potential of negative expectations, a therapist's optimism is a desirable component of any strategy of behavior change. Of course, positive expectations do not always produce results, and interpersonal prophecies do not always fulfill themselves, as any clinician knows.  

When Prophecies Fail

Although others' expectations for us have powerful effects, it may be quite difficult to convince us that we are something we firmly feel we are not. While our self-concepts are open to change, they are not so malleable that any stranger with a new opinion of us can substantially alter our self-views. Indeed, we tend to refute and reject information about ourselves that is inconsistent with our self-concepts, preferring instead to hear things that verify and support what we already think about ourselves (Swann, 1985). For instance, when Swann and Hill (1982) provided social feedback that was discrepant-with subjects' self-evaluations of dominance or submissivenesstelling a submissive person, for instance, that she really appeared to be forceful and dominant-the subjects went out of their way to be even more dominant or submissive than usual, proving they were not what the evaluator believed.

If people reject straightforward feedback about themselves with which they do not agree, what happens when a perceiver's more subtle expectancy about a target contradicts that target's self-concept? Can the perceiver elicit from the target behavior that just does not fit the target's self-view? Swann and Ely (1984) examined this issue, asking female subjects to evaluate the introversion or extraversion of people who viewed themselves as either extraverted or introverted. Each subject was equipped with either a strong or a weak expectancy that contradicted the self-description of the target she would meet, and was asked to test that expectancy using the hypothesis testing procedure of Snyder and Swann (1978a; see chapter 8). Judges' ratings of three separate trials revealed that the perceivers' expectancies became self-fulfilling prophecies only when strong expectations in the perceivers were pitted against rather uncertain self-conceptions in the targets. If both perceivers and targets held uncertain judgments, or if the targets were sure of themselves, the perceivers failed to elicit the behavior they expected; the targets continued to describe themselves in ways that were consistent with their self-concepts, despite the implicit demands from their partners to describe themselves as something else.

Swann and Ely's (1984) study was not a strong test of self-fulfilling prophecies because it only involved verbal communication between perceivers and targets who could not see one another. Nevertheless, it indicates that even strong expectations may fail to influence behavior that is central to a person's self-concept. Whether or not it is dysfunctional, behavior that helps affirm a person's self-image may be unlikely to change in response to others' expectations, leaving the opposing prophecies of others uniformly unfulfilled.

In many therapeutic situations, of course, the therapist has no wish to contradict the client's self-conceptions completely. Therapy often involves an effort to provide willing clients with new interpretations of their feelings and abilities, not wholesale revisions of their self-views. Even if a therapist has such revisionist aspirations, his or her credibility, prestige, and attractiveness might make therapeutic expectancies much more influential than Swann and Ely's (1984) study would suggest (see Aronson et al., 1963, and chapter 9). Still, it is likely that therapists' expectations have the greatest impact on relatively interested, motivated clients who care what their therapists think and who are receptive to their advice. In the absence of those facilitating conditions, even the most favorable expectations may be unlikely to influence resistant clients. And even if desirable change is achieved in the controlled therapeutic environment, it may attenuate when clients have the chance to reaffirm their self-images on their own (cf. Swann and Hill, 1982). For all the value and potential benefit of therapists' positive expectations, clients' expectations of success are probably more vital still. If clients do not believe in themselves, lasting change may be difficult to achieve.

Backfiring Expectations

High expectations can have hidden costs. People sometimes "choke" under pressure, performing poorly when the importance of their actions is magnified. In an intriguing series of studies, Baumeister (1984) and his colleagues showed that high expectations from an audience can backfire, causing people to do less well than they otherwise would. For example, an archival analysis of championship games in major league baseball and basketball demonstrated that it may actually be disadvantageous to play the final game of a championship series before an adoring home crowd (Baumeister & Steinhilber, 1984). Home teams win over 60% of the first two games of the World Series, but they lose over 60% of the decisive seventh games. Similar results occur in basketball, and analysis of team performances reveals that the home teams "choke" -committing errors, missing free throws-to a greater extent than the visiting teams excel. Curiously, the performance decrement is associated not with the fear of failure, but with the chance to achieve success; when facing elimination in a sixth game of a World Series, home teams win 73% of the time, but when they have the chance to win the championship instead, they do so only 38% of the time.

The key to these remarkable results may be the extent to which the performers share the audience's lofty expectations. Baumeister, Hamilton, and Tice (1985) found that an audience's expectations of success increase the pressure on a performer and are likely to harm performance unless they are so credible, persuasive, and compelling that the performer privately expects success as well. In the Baumeister et al. studies, a person's private expectation of success invariably facilitated that person's skillful performance. By contrast, if a person was privately sure he or she would fail, high expectations from an audience made matters worse, debilitating the person further. If the person came to share a credible audience's positive expectations, however, performance improved. Thus, high expectations from others (e.g., therapists) can have either a beneficial or a detrimental impact, depending upon the individual's own beliefs. It is very likely that most therapists are credible enough that their confident projections of success have a desirable influence on their uncertain clients. However, if the clients are unwilling participants in therapy or are convinced of their own incompetence, high therapist expectations may backfire, impeding progress. Once again, clients' expectations are of fundamental importance to the therapeutic enterprise.

Client Expectations

Clients' expectations that their therapy will be successful probably do much more than just make them willing, compliant participants in their treatments. Reviewers of "nonspecific" therapies and placebo effects have noted several ways that clients' positive expectations facilitate change. First, hopeful expectations contribute to positive moods and higher motivation, which may help clients learn new skills or understand new frames of reference (Bootzin, 1985; Lick & Bootzin, 1975). In a similar fashion, expectations of improvement may reduce clients' anxiety about their problems, calming the worrying that may exacerbate them (Ross & Olson, 1982; see chapter 2).

Second, a hopeful outlook may lead clients to "accentuate the positive," noticing and attaching importance to small successes they would otherwise overlook (Ross & Olson, 1982). They may even test their expectations of improvement in a biased, confirmatory manner so that their hopes are supported, encouraging them to continue treatment (Lick & Bootzin, 1975). If they fail to find any factual evidence of success, they may resolve the resulting cognitive dissonance by convincing themselves they are feeling better anyway (Shapiro & Morris, 1978).

Third, to the extent a client has enjoyed prior success with therapy, the therapeutic situation itself may be a conditioned stimulus that promotes improvement, alleviating anxiety and decreasing depression (Ross & Olson 1982; Shapiro & Morris, 1978). Finally, Ross and Olson speculated that positive expectations might produce physiological changes that have palliative effects (stimulating the production of morphine-like endorphins in the brain, for instance).

In addition to these potential benefits, the expectation that one's behavior will change may actually make one's behavior more likely to change. Merely predicting how we will act increases the likelihood that we will do what we predicted (a personal self-fulfilling prophecy). Sherman (1980) found that when people were asked to volunteer to collect money for the American Cancer Society, only 4% complied. However, if people instead were asked to predict whether or not they would volunteer, 48% believed they would, and when that entire prediction group was later asked to volunteer, 31% complied. The different rates of compliance reflect the channeling effects of behavioral predictions, and expecting that one will stop smoking, begin dating, be more assertive, or control one's temper may make those behaviors more likely to occur. Thus, having clients make public predictions about the desirable changes they hope to achieve during therapy may be beneficial.

 Of course, one's beliefs about what one can do will substantially influence one's beliefs about what one will do, and that brings us to consideration of the most important client expectancies of all.

 Self-Efficacy

 In a provocative and heuristic analysis of behavior change, Bandura (1977, 1978, 1982, 1984) has suggested that clients' estimates of their specific competencies largely determine how they cope with their problems. For Bandura, a person's "efficacy expectancies" describe that person's beliefs about his or her particular skills and capabilities, and they determine how the person reacts behaviorally, cognitively, and emotionally to problematic events. Perceptions of self-efficacy determine what activities people attempt, how hard they try, and how long they persist in the face of failure. Burdened with low efficacy expectations, clients are likely to avoid their problems, expend little effort, give up prematurely, dwell on their inadequacies, and suffer considerable anxiety and stress. By contrast, those with high efficacy expectations are likely to set challenging goals, persevere in the face of frustration, approach therapy tasks without anxiety, and, in short, attain more (Bandura, 1984).

 Bandura (e.g., 1982, 1984) makes an important distinction between an efficacy expectancy, which is the subjective probability that one can successfully execute a certain behavior, and an outcome expectancy, which is the subjective probability that a particular behavior will lead to particular outcomes. Your conviction that you can run an entire marathon would be an efficacy expectancy, for example, whereas your anticipation of the accompanying praise and respect from your spouse and friends would be an outcome expectancy. Although they are not always easy to discriminate, the distinction between the two types of expectancy is important because either may be a source of dysfunctional behavior (see chapter 6). A shy person may consciously shy away from rewarding interaction either because she believes she lacks the skills to engage in small talk or because she believes that her partner will reject her no matter how well she does (Leary & Atherton, 1986). Thus we may fear failure either because of perceived personal inadequacy or because of an unresponsive, punitive environment that we believe will leave our best efforts unrewarded.

 In fact, Bandura (1982) posited four sets of interactive effects of the combination of independent efficacy and outcome expectancies. If both types of expectancy are high, people are sure of themselves and believe the environment is facilitative, and assured, confident behavior should result. If both expectancies are low, dispirited apathy and helplessness result (Devins et al., 1982). When self-efficacy is high but the environment is thought to be unresponsive, frustration, protest, and active efforts to change the environment are likely. Perhaps the worst case of all, however, is the combination of personal inefficacy and positive outcome expectancies; success seems available if only one were competent enough to attain it, in which case depression, despondency, and diminished self-esteem are all likely.

 The obvious scope of Bandura's analysis and his argument that perceptions of self-efficacy play a central, causal role in mediating behavior change have drawn critics from far and wide. We will examine the most common criticisms of self-efficacy theory as a means of introducing the volumes of studies that bear on the theory.

 Is self-efficacy a redundant construct? Bandura's early studies of self-efficacy (e.g., Bandura, Adams, & Beyer, 1977) showed that as subjects with snake phobias progressed through therapy, the changes in their self-efficacy expectancies closely tracked their actual behavioral accomplishments. Selfefficacy was a very good predictor of performance, but Borkovec (1978), Eysenck (1978), and Wolpe (1978) argued that one need not resort to a construct of self-efficacy to explain such results. Changes in conditioned anxiety and arousal could also be responsible, they reasoned, and would provide a more parsimonious explanation. Bandura and Adams (1977) were able-to show, however, that once phobic subjects were completely desensitized, entirely eliminating their anxiety, their efficacy expectancies still closely predicted the (widely variable) behaviors they were able to perform. (This ability to predict the differential response of individuals receiving the same type of treatment, and the levels of change attained by different treatments [Bandura et al., 1977; Bandura, Adams, Hardy, & Howells, 19801, is a press; Stanley & Maddux, 1985, 1986). For that matter, it is not always easy to differentiate efficacy expectations from outcome expectations (Maddux et al., 1982); for example, a potential marathoner with low efficacy expectations may expect very different outcomes (e.g., ripped muscles, lasting metabolic disorders) than another runner with higher self-efficacy (e.g., sore muscles, temporary fatigue) as a result of their different efficacies, not their perceptions of the rewards available.

 What should we make, then, of the concept of self-efficacy? Some of its conceptual links to other determinants of behavior need clarification, and it needs to be considered as only one of several causes of a person's actions, as Bandura (e.g., 1984) has repeatedly~ emphasized. Still, efficacy expectancies are probably vitally important to the amelioration of dysfunctional behavior, since many troubled clients already possess the behavioral competencies needed to help their problems, but do not implement them because they do not believe they can. Arnkoff and Glass (1982) recalled the story of Dumbo, the flying elephant who could remain aloft as long as he was holding what he thought was a magic feather. This was a workable arrangement until Dumbo dropped the feather; he then started to plummet to the ground, and not until he was (hurriedly) convinced that he could fly did he again start using the skills he had possessed all along. Human clients may face similar dilemmas, and the successes of insight, client-centered, and rational-emotive therapies, for example, may depend largely on the extent to which they are able to influence clients' conceptions of the life skills they already have. When specific new skills are taught to clients in participant modeling or relaxation training, for instance, clients must still develop the confidence and certainty in their newfound talents that enable them to employ those skills when needed. Clients' expectations for their own successes appear to be of fundamental importance, and influencing clients' efficacy expectancies may be an indispensable part of any therapeutic endeavor.

Enhancing self-efficacy expectations. Indeed, Bandura (1977, 1982) suggested that, whatever their specific procedures, all effective strategies of behavior change have beneficial effects in part by influencing clients' perceptions of self-efficacy. He posed four broad ways a person's self-efficacy can be changed.

 The most influential source of efficacy expectations is a client's own behavioral accomplishments. By actually demonstrating their mastery of a new skill or an old fear, clients gain straightforward evidence that they are capable, at least under some circumstances, of successfully enacting important behaviors. For example, Williams et al. (1985) showed that guided mastery of specific tasks in an eight-story parking garage was of greater and more lasting benefit to subjects with acrophobia than was a desensitization treatment in which the subjects merely faced the feared eight-story drop. The extent to which efficacy expectations are affected by performance accomplishments, of course, depends on the person's attributions for his or her success; attributions to internal, stable, global, and, especially, controllable factors are likely to lead to the greatest change in self-efficacy (see chapter 2).

 Vicarious experience through the observation of others is a second source of efficacy information. Although not as influential as enactive mastery (Bandura et al., 1977), modeling treatments can have reliable effects on self-efficacy depending on the similarity of the models to the client, the number of models employed, and other situational variables (Bandura, 1977). Witnessing the successes of others can influence one's expectancies of performing similar actions.

 Third, verbal persuasion affects self-efficacy. Although persuasion is not presumed to be as great an influence as performance-based treatments (since a single behavioral failure may undo all that is accomplished through persuasion), it can have desirable impact, particularly when it leads a client to attempt or persist at activities in which success is later obtained. Persuaders may also be able to remind clients of past successes and to point out internal causes for success that clients have overlooked (Goldfried & Robins, 1982, 1983). The credibility and attractiveness of the persuaders will likely determine how much persuasive influence they have (see chapter 9).

 Finally, physiological states of arousal may affect perceptions of self-efficacy because people generally read signs of arousal in problematic situations as evidence of their vulnerability (cf. Leary & Atherton, 1986; Leary, Atherton, Hill, & Hur, in press). When they are cool and calm in the face of calamity, clients may be more confident of their abilities to cope with their predicaments. Biofeedback and symbolic desensitization may thus be useful treatments, removing arousal as a cue of low self-efficacy.

 As Goldfried and Robins (1983) suggested, many therapies probably provide more than one source of efficacy information. A relaxation procedure not only reduces arousal, but also teaches a new behavioral skill that clients gradually master. Cognitive restructuring persuades a client of a new point of view but also reduces worried arousal. With these multiple effects in mind, we prefer, instead of contrasting therapies, to note their many commonalities as a means of summarizing this chapter. Bandura's analysis suggests that diverse treatments all share a common means of facilitating behavior change by influencing perceptions of self-efficacy.

 Common Features of Strategies of Behavior Change

 Observers of psychotherapies generally agree that there are several features shared by even disparate treatments (Brady et al., 1980; Critelli & Neumann, 1984; Frank, 1973; Goldfried, 1980; Strong & Claiborn, 1982). Examination of that list underscores the reliance of therapies upon the expectations of both therapist and client in influencing behavior.

 First, psychotherapies involve a therapist who usually seems competent, key advantage of a self-efficacy perspective.) Williams and his colleagues (Williams, Turner, & Peer, 1985; Williams & Watson, 1985) have since shown that anxiety and perceived danger are much poorer predictors of phobics' behavior than is self-efficacy.

 Does self-efficacy have causal impact? The alert reader may have already anticipated a criticism like that raised by Eysenck (1978), Kazdin (1978), and Wolpe (1978): Given the close correlation between self-efficacy and behavior, do perceptions of self-efficacy causally affect behavior, or do they simply follow from witnessing one's actions? Bandura (1982) admitted that, as an estimation of one's competencies, self-efficacy is influenced by one's performance. Efficacy expectations are based on one's interpretation of one's performances, however, and because of the inferential ambiguities involved (with external aid, task difficulty, one's effort, etc., all affecting the conclusions one draws) judgments of self-efficacy are not simple reflections of behavioral accomplishments (Bandura, 1984). More importantly, efficacy expectancies do, in turn, direct behavior. Self-efficacy is a better predictor of future performance than is past performance, accurately predicting performance on tasks people have never attempted before (Bandura, Reese, & Adams, 1982). Manipulations of self-efficacy not only determine subjects' coping behavior but affect their heart rates, blood pressures, and levels of plasma epinephrine and norepinephrine as well (Bandura, et al., 1982; Bandura, Taylor, Williams, Mefford, & Barchas, 1985). Introducing low efficacy expectancies creates a depressive mood, but introducing a depressive mood does not create low efficacy expectancies (Stanley & Maddux, 1985). It seems, overall, that the causal link between self-efficacy and behavior is reciprocal, with each partially determining the other.

 Are outcome or efficacy expectations more important? In two different ways self-efficacy theory has been criticized for neglecting the importance of outcome expectancies in directing behavior. The stronger form of this argument is again that self-efficacy is an unnecessary construct; Borkovec (1978) and Teasdale (1978) suggested that quantification of the outcomes and incentives that follow operant behavior is sufficient to explain behavior change, and that we need not propose cognitive mediators such as self-efficacy. However, a wide range of studies have demonstrated that efficacy expectancies are much more closely related to one's actual behavior than are outcome expectancies. For instance, Barling and Abel (1983) examined 12 components of subjects' tennis performance (such as their anticipation, concentration, footwork, competitiveness, and power and spin) and found that the subjects' perceptions of self-efficacy were closely linked to their performances whereas their outcome expectancies were not. Similar findings have been obtained in studies of subjects' assertiveness (Lee, 1984a), ability to handle snakes (Lee, 1984b), and persistence after failure (Jacobs, Prentice-Dunn, & Rogers, 1984).

 Thus, behavior can be predicted from people's perceptions of what they can do as well as from their perceptions of the rewards and punishments available to them. If people feel, rightly or wrongly, that they cannot perform a certain action well, they are unlikely to do well whatever the rewards available to them. Self-efficacy provides predictive power that outcome expectancies do not provide, a fact that seems particularly important when clinically relevant behaviors are considered. Efficacy expectancies significantly predict clients' ability to stop smoking, identifying those who are likely to remain nonsmokers and accurately predicting situations in which relapses are likely to occur (Coelho, 1984; Condiotte & Lichtenstein, 1981; DiClemente, Prochaska, & Gibertini, 1985). Similarly, they predict who will drop out of weight-loss programs (Mitchell & Stuart, 1984) and how much weight the remaining clients will have lost after 6 weeks and 6 months (Bernier & Poser, 1984). Efficacy expectancies even predict university faculty research productivity (Taylor, Locke, Lee, & Gist, 1984). Thus, self-efficacy is far from redundant with outcome expectancies; efficacy expectancies can tell us much about people's behaviors that rewards and costs do not.

 A weaker form of the efficacy-versus-outcome criticism suggests that self-efficacy does have effects, but that Bandura's theory overlooks just how influential outcome expectancies really are (Eastman & Marzillier, 1984; Kirsch, 1985; Marzillier & Eastman, 1984). Even though, as we have just seen, efficacy expectations are much more closely linked to the parameters of actual performance than are outcome expectations (e.g., Barling & Abel, 1983; Jacobs et al., 1984), this criticism has some merit. Maddux and Barnes (1985) suggested that because self-efficacy reflects perceived skillfulness, clients' intentions to engage in certain behaviors, rather than their performance of those behaviors, are a better test of the relative contributions of efficacy and outcome expectations. Research shows that both types of expectations independently predict behavioral intentions, but when both are experimentally manipulated, only changes in outcome expectancies cause significant changes in subjects' intentions (Maddux, Norton, & Stoltenberg, in press; Maddux, Sherer, & Rogers, 1982). Efficacy expectations still add to our ability to predict intentions even when outcome expectancies are accounted for, however, and both appear to provide important, nonredundant information about people's plans of action (Maddux et al., in press). We may conclude, then, that both the competencies people perceive and the outcomes they expect influence what they believe they will do, whereas efficacy judgments alone are the primary determinants of how much they actually accomplish (Eccles, Adler, & Meece, 1984).

 Maddux and his colleagues suggested the positive or negative value of potential outcomes as* a third independent predictor of behavior, apart from the outcome expectancy that certain behaviors will obtain the outcomes (Maddux & Bames, 1985; cf. Kirsch, 1985), but so far it has proven difficult to disentangle outcome value from outcome expectancy (Maddux et al., in caring, and confident. Second, there exists a relationship between therapist and client and a therapeutic setting that is a crucible for change. Third, the therapy provides a credible rationale for change that allows an expectancy of improvement. Finally, the therapy involves new experiences in the form of effortful procedures that both enhance the client's self-confidence and implicitly demand improvement. Together, these characteristics provide social support, successful experiences, and hope for improvement that, rather than mere nuisance variables to be controlled in effectiveness studies, may be a fundamental basis for the effectiveness of psychological treatment (Critelli & Neumann, 1984; Frank, 1973).

 Along with whatever specific procedures are used to provide new experiences to clients, these common features are all likely to influence clients' beliefs that change is possible and that they will get better. Nearly all therapies involve a persuasive therapist who believes that change will occur, an interactive relationship that allows the therapist to communicate those expectancies, and an acceptable explanation for why change is attainable. Given the fact that many therapies have yet to demonstrate that they are any more effective than these positive expectancy-inducing features alone (Critelli & Neumann, 1984; Prioleau, Murdock, & Brody, 1983), the vital contribution of clients' positive expectations of change and the facilitating role of therapists' favorable expectations cannot be taken lightly. Whether or not it is explained in terms of self-efficacy, intentional or unintentional impact on clients' beliefs that favorable change is possible seems to be a universal factor in strategies of behavior change.

Conclusions

 A therapist's confident expectations of success are a boon to any strategy of behavior change, frequently setting in motion self-fulfilling prophecies that -increase the likelihood of psychotherapeutic change. The therapist's expectations probably have these beneficial effects in part by influencing the client's private expectations of what he or she can achieve. The client's expectations are of foremost importance, as they have direct influence on the client's behavioral capacities. Thus, enhancing clients' expectations of mastery, or self-efficacy, may be a vital part of any therapeutic enterprise. Happily, whatever their specific procedures, most psychological therapies share common elements that are likely to enhance a client's self-efficacy expectations.