FOR USE IN GPS 520

Leary, M. R., & Miller, R. S. (1986). Social psychology and dysfunctional behavior: Origins, diagnosis, and treatment, (pp. 147-163). New York: Springer-Verlag Inc.

Having examined the processes involved in clinical judgment and diagnosis, we are now ready to turn our attention to the process of therapy itself. During the 1960s, several writers began to examine the process of counseling and psychotherapy from social psychological perspectives. Although differing in emphasis, these analyses shared two central assumptions: that therapy and counseling involve interpersonal processes in which the therapist influences the thoughts, feelings, and behavior of the client, and that therapeutic change can be understood only within the context of the interpersonal dynamics of the counseling setting (Frank, 1961; Goldstein, 1966; Goldstein, Heller, & Sechrest, 1966; Strong, 1968). In fact, Strong (1968, p. 101) went so far as to suggest that "psychotherapy can be viewed as a branch of applied social psychology."

The notion that counseling and psychotherapy involve social influence processes is regarded as self-evident by most psychologists today, but Strong (1982) observed that this perspective was not widely accepted among most counselors and psychotherapists 20 years ago. Even today, many practitioners, particularly those who subscribe to humanistic or client-centered approaches, argue that psychologists should not directly influence their clients, but rather should provide a supportive environment in which the client can mobilize his or her own change processes.

From a social psychological perspective, such a view of psychotherapy and counseling is untenable. All human relationships involve mutual influence as the individuals act and react to one another, adjust their behavior on the basis of one another's actions, respond emotionally to each other, modify their beliefs and opinions as a result of what the other individual says and does, and so on. Given that counseling involves a social encounter between two or more persons, it is unavoidable that therapist and client influence one another during the course of therapy. Indeed, a therapeutic relationship in which influence does not occur is not a relationship at all, but merely an instance of two individuals coacting independently in the same place at the same time. Explicitly or implicitly, purposefully or accidentally, the effective counselor influences clients to adopt certain therapeutic goals, to think about themselves and others in more adaptive ways, to learn new social competencies, to apply cognitive and behavioral principles learned in therapy to their everyday lives, and so on.

In short, we accept as self-evident the premise that counseling and psychotherapy involve a process of mutual social influence in which a therapist and a client modify one another's beliefs, emotions, and behavior. Questions then arise regarding how this influence occurs and whether it contributes to desired changes in the client. These questions are the focus of this chapter.[1]

With the onset of World War 11, social psychologists became intensely interested in social influence processes within the context of such topics as attitude change, brainwashing, propaganda, and leadership. Since then, influence has come to be regarded as a component of all social interaction and has taken a central place in the study of interpersonal behavior.

Strong (1968) was one of the first to apply the vast social psychological literature on influence to an understanding of what happens when an individual seeks help from a psychologist or other mental health professional. In what has become a landmark paper, Strong noted close parallels between a speaker attempting to influence an audience and a counselor trying to help a client. In both cases, a communicator attempts to influence another's attitudes and behavior through verbal communication. Given the strong parallel, Strong argued that social psychology's understanding of attitude change was directly relevant to the counseling process. He demonstrated the viability of his claim by reviewing research dealing with the impact of communicator and audience characteristics on the influence process and by applying this research to an analysis of counseling (see also Goldstein et al., 1966).

Since that time, a tremendous body of research has examined the manner in which therapists influence their clients. This literature is quite extensive, so our coverage of this area will be necessarily selective. In determining which material to cover, we chose to deal chiefly with the research based most directly on social psychological perspectives and with findings in which we have the most confidence. Other reviews of this literature may be found in Strong (1968, 1982), Strong and Claiborn (1982), Corrigan, Dell, Lewis, and Schmidt (1980), and Maddux, Stoltenberg, and Rosenwein (in press).

Most research on influence in counseling and therapy may be traced directly or indirectly to Hovland, Janis, and Kelley's (1953) work on attitude change. In Communication and Persuasion, Hovland et al. (1953) provided an extensive review of the existing literature on variables that affect the influence process, classifying them into factors relevant to the source or communicator of a message, the target or audience of the communication, and characteristics of the message itself. This line of work has had a greater impact on research in counseling and psychotherapy than any other in social psychology. Building on the framework provided by Hovland et al., researchers in counseling and psychotherapy have studied characteristics of the counselor (source), client (target), and treatment (message) that affect therapeutic outcomes. It is to this literature that we now turn our attention.

The effectiveness of attempts to change others' attitudes and behavior depends in large part on how the source of the persuasive communication is perceived by the target or recipient of the message. In discussing the effects of "source characteristics" on attitude change, Hovland et al. (1953) focused primarily on the impact of communicator "credibility." Two aspects of credibility were examined in their work: expertness (the degree to which a communicator is perceived as being a source of valid information) and trustworthiness (the degree to which a communicator is perceived as having honest intentions). Not surprisingly, laboratory research in social psychology has shown that, in a variety of domains, communicators who are viewed as expert and trustworthy are more influential than those who are not (e.g., Aronson, Turner, & Carlsmith, 1963; Eagly & Himmelfarb, 1978; Hovland et al., 1953; Mills & Harvey, 1972; see, however, Powell, 1965). Although this relationship between credibility and influence is often regarded as a given among social psychologists, research that has explored the effects of counselor credibility has yielded mixed results.

Studies of counselor expertness have usually introduced a therapist to a client in a manner that portrays the therapist as either highly trained and experienced or poorly trained and inexperienced. For example, counselors may be introduced to clients as Ph.D. psychologists with national reputations or as practicum students or undergraduates with little or no experience. Therapeutic outcomes are then assessed at the end of one or more counseling sessions.

At the risk of oversimplifying the results of this body of research, we can conclude that counselors can influence client change regardless of their presumed expertise, experience, or status. That is, some studies have found that even counselors described as inexperienced non-experts produce client change relative to non-treatment control groups (Greenberg, 1969; Sprafkin, 1970). Thus, perceived counselor expertness is not necessary for therapy to be effective.

This finding has been a source of consternation for many well-trained therapists, but four points can be made about it. First, it is possible that the counselor role per se may carry considerable influence regardless of the characteristics of the individual who occupies it (Corrigan et al., 1980; Strong, 1968). That is, merely adopting counselor and client roles may create a situation in which the person in the client role makes an effort to attend to and comply with the counselor's recommendations, regardless of the counselor's expertise or experience. Second, casual observation in everyday life shows that people are often influenced and helped by others who are not in any way viewed as expert helpers. Even untrained individuals may be quite helpful to distressed persons if their relationship is characterized by understanding, empathy, trust, warmth, and support (Emrick, Lassen & Edwards, 1977; Truax & Carkhauff, 1967). Third, the initial effects of a counselor's presumed expertness (or lack of it) are likely to weaken once the counselor and client begin to interact. Even untrained counselors may be perceived as competent helpers, whereas some supposedly expert counselors are not (see Heppner & Dixon, 1978; Strong & Schmidt, 1970a). Most research dealing with the effects of expertness on influence, within both social psychology and counseling, has been based on a single interaction between communicator and target (or counselor and client). Perceived expertise may become less important with extended contact.

Most importantly, the fact that inexpert helpers have been shown to be effective in analogue counseling settings does not imply that perceived expertise has no effect at all. On the contrary, although ostensibly inexperienced counselors are sometimes as effective as those who are supposedly well trained and experienced (Sprafkin, 1970), most studies show that counselors who are believed to be expert do have a greater effect on clients, at least under certain conditions (e.g., Beutler, Jobe, & Elkins, 1974; Binderman, Fretz, Scott, & Abrams, 1972; Greenberg, 1969; Heppner & Dixon, 1978; McKee & SmPuse, 1983; Strong & Schmidt, 1970a). Thus, although perceived expertise may not be absolutely necessary, a counselor's effectiveness is usually enhanced if the client perceives the counselor to be well trained and/or experienced.

But what of actual, rather than merely perceived, expertise? Are experienced therapists more effective than inexperienced ones? Exhaustive reviews of the literature by Auerbach and Johnson (1977) and Parloff, Waskow, and Wolfe (1978) concluded that the results relevant to this question are quite mixed. Although there is evidence that experienced therapists establish better relationships with their clients than less experienced ones, the relationship between therapist experience and therapeutic outcome is surprisingly weak (Auerbach & Johnson, 1977).

In any case, because perceived expertise often facilitates treatment effectiveness, several studies have explored factors that lead clients to infer that a therapist is expert. These studies show that perceived expertness is heightened by factors such as the possession of advanced degrees, the extent of the therapist's experience, the presence of diplomas in the therapist's office, the therapist's use of professional jargon, professional (as opposed to casual) clothing, and socially skilled nonverbal behavior (Atkinson & Carskaddon, 1975; Carkhuff 1969; Greenberg, 1969; Kerr & Dell, 1976; Scheid, 1976; Schmidt & Strong, 1970; Sprafkin, 1970; Strong & Schmidt, 1970a). Such findings raise the issue of whether counselors and psychotherapists should purposefully manage their impressions in ways that portray them as expert, intentionally using diplomas, clothing, nonverbal cues, and so on to suggest expertise (cf. Korda, 1976; Waxer, 1978). Many therapists find the thought repugnant, whereas others argue that if such simple strategies heighten their effectiveness with clients, they should be considered.

Trust in a communicator is a primary determinant of attitude change in any encounter (e.g., McGuire, 1969), and it is particularly important in counseling settings. As Tyler (1965, p. 16) put it, "the client's confidence in the counselor, the assumption that he can believe what this person tells him, is the essential foundation for the whole counseling process." Not only are clients unlikely to heed the advice of counselors they distrust, but they are less likely to risk openly disclosing and confronting their difficulties when they do not trust the therapist. Unfortunately, despite the central role of trust in therapeutic relationships, little research has been conducted either on the factors that lead clients to trust their therapists or on the effects of trust on psychological change within therapy.

Strong (1968) suggested that whether client perceives a counselor as trustworthy depends upon four factors: the counselor's reputation for honesty trustworthiness implied by the counselor's social role (being a minister or a physician, for example), the sincerity and openness of the counselor's interactions with the client, the perceived lack of motivation for the counselor's personal gain. The few studies in the area have not specifically examined these determinants of trustworthiness, but have combined several within a single study. Studies show that interviewers who are inconsistent, inaccurate, unethical (violating confidentiality), or even overweight appear less trustworthy to clients (e.g., McGee & Smouse, 1983; Rothmeier & Dixon, 1980). In general, perceptions of trustworthiness appear to be affected more by the therapist's general manner than what he or she says (Kaul & Schmidt, 1971; Roll, Schmidt, & Kaul, 1972).

If social psychological research can be generalized to the counseling set ting, we would predict that trustworthiness is a necessary (though not sufficient) precondition for successful counseling, perhaps more important than expertness (Kelman & Hovland, 1953; Walster, Aronson, & Abrahams, 1966; see also Corrigan et al., 1978). Although a few counseling outcome studies have obtained effects of trustworthiness (e.g., Rothmeier & Dixon, 1980), others have not (Strong & Schmidt, 1970b), and much more work on the topic is needed.

Attractiveness refers to the degree to which a client likes the therapist. Research both in social psychology and in analogue therapy settings reveals that interpersonal attraction mediates social influence under some circumstances. However, considerably more research has been conducted on the factors that lead clients to like their therapists than on the effects of liking on therapy outcome.

As social psychological studies would predict (e.g., Berscheid & Walster, 1974), physically attractive counselors are generally rated more positively than unattractive ones. For example, Lewis and Walsh (1978) found that attractive counselors were judged to be more professional, interesting, and competent than unattractive counselors (see also Cash, Begley, McGown, & Weise, 1975; Cash & Kehr, 1978). The nature of the relationship between physical attractiveness and liking for a counselor is interesting; there is evidence that being unattractive reduces a client's liking for the counselor, but that being attractive does not necessarily increase liking (Corrigan et al., 1980).

Not surprisingly, warm and attentive counselors are liked better than cold and aloof ones (Patton, 1969; Schmidt & Strong, 1971). Similarly, nonverbal behaviors indicating attentiveness and interest (such as eye contact, nodding, smiling, forward lean) create greater liking for counselors (Claibom, '1979; Kleinke, Staneski, & Berger, 1975; LaCrosse, 1975; Strong, Taylor, Bratten, & Loper, 1971).

Social psychological research also demonstrates that similarity enhances attraction (Byrne, 1971), and, again, counseling research supports this pattern. When clients are told that they and their counselor are similar or that they share certain attitudes, their liking for the counselor increases (Goldstein, 1971; Hogan, Hall, & Blank, 1972). However, initial information about similarity may be offset by new information gleaned during future therapy sessions. The effects of attitudinal similarity on attraction weaken beyond the initial counseling session (Goldstein, 1971).

Thus, counselor attractiveness is affected by various cues regarding appearance, nonverbal behavior, warmth, and similarity, but the question arises whether attraction toward a therapist enhances the outcome of therapy. Although research in experimental social psychology suggests that better liked communicators are more influential (Berscheid, 1966; Byrne, 1971; McGuire, 1969), the findings in counseling studies are mixed. Some studies show positive effects of counselor attractiveness on treatment effectiveness (Patton, 1969; Strong & Dixon, 1971), whereas others do not (Dell, 1973; Schmidt & Strong, 1971; Sell, 1974; see Abramowitz, Berger, and Weary, 1982, Corrigan et al., 1980, and Dom, 1984, for reviews).

In part, this complex pattern of results may reflect the fact that attractiveness may heighten counselors' influence only under certain conditions. For example, if a therapist is already perceived as competent and trustworthy, whether the client likes the therapist may be irrelevant because the client may participate fully in therapy anyway. However, if the therapist is viewed as less competent, clients may be less committed to the therapeutic relationship unless they are attracted to the therapist as an individual. As Goldstein (1971) pointed out, attraction may affect therapeutic outcomes by increasing the likelihood that the client will return for future sessions and participate actively and willingly in them. The authors know a woman who remains in therapy despite the fact that she doesn't view her therapist as particularly effective simply because she likes her therapist. Indeed, Strong and Dixon (1971) found that attractiveness increased counselors' influence only when they were viewed as inexpert rather than expert.

It seems, then, that counselors who are perceived as experts are usually more effective than non-experts even though non-experts are often effective; that trustworthiness may contribute to therapeutic effects, though more work is needed; and that attractiveness may help inexpert counselors but is of limited value to expert counselors. Thus, these interpersonal variables affect counseling outcomes, but the results are not as straightforward as either social psychological research or clinical wisdom would lead one to expect. There are several possible reasons for this.

First, the experiments designed to test the impact of counselor characteristics upon the counselor's effectiveness usually employ analogue counseling situations involving a single contact between the counselor and client. As such, they tell us only about the impact of counselor characteristics during the first 60 minutes (or less) of an initial counseling session. The effects of some counselor variables may strengthen over time, whereas others may attenuate. For example, liking may become a more important determinant of therapeutic outcome as therapy progresses over time and the client persists in therapy, despite urges to quit, because he or she likes the counselor. On the other hand, reputed experience may be important only initially, until the client forms his or her own impressions of the counselor. Although existing research highlights potential contributors to the outcome of treatment, more work is needed in ongoing, long-term relationships between counselors and clients.

Second, many source characteristics may have their effects by increasing the receiver's attention to the message (Hovland et al., 1953). For example, people may pay more attention to credible and attractive sources than to less credible communicators. Although this attentional difference may have a strong effect on counselor effectiveness in real counseling encounters, it may be less important in research settings where "clients" usually know they are subjects in an experiment. Being subjects, they may attend closely to all the details of the session, regardless of the characteristics of the counselor, thereby making inexpert counselors appear more effective than they would normally be.

Third, the kinds of changes targeted in counseling are usually more involving and complex than those of interest in basic social psychological research. It may be that the psychological changes of interest in counseling and psychotherapy typically require stronger manipulations of credibility and attractiveness over a longer period of time.

In sum, although expertness, trustworthiness, and attractiveness do affect treatment outcomes, the complexity of the results should prompt additional research, preferably in clinical field studies, with increased attention to improved theory and methodology.

In comparison to the extensive research on counselor characteristics, message and client characteristics have received short shrift in counseling research. Message characteristics that social psychologists have studied include one-sided versus two-sided messages, messages that elicit fear, the degree of discrepancy between the message and the target's initial opinion, whether the communicator draws explicit conclusions, repetition of the message, and message style (the degree to which the presentation of the message is dynamic versus subdued, for example). Within counseling and psychotherapy, most attention has been devoted to message discrepancy, fear appeals, and stylistic factors.

For every person who seeks counseling or psychotherapy, there is a gap between how the individual perceives his or her circumstances and how the individual would like things to be. Part of the role of the counselor is to help the client minimize this discrepancy by various cognitive and behavioral means (Strong & Claiborn, 1982), including direct suggestions regarding how the client might develop more functional ways of thinking and acting. Given that a therapist wishes to induce a cognitive or behavioral change in a client, how discrepant from the client's initial position should the therapist's recommendation be?

In general, social psychological research has demonstrated a positive, linear relationship between the magnitude of the discrepancy and the amount of resultant change (see McGuire, 1969). That is, the greater the discrepancy between the position advocated by a communicator and that initially held by a target, the greater the eventual change in the target's position. There is an upper limit to this effect, however, a point beyond which increasingly discrepant messages are dismissed outright and are less influential.

Overall, experimental and correlational investigations in counseling settings support this pattern-the greater the initial difference between the client's position and that advocated by the therapist, the greater the change that is likely to ultimately occur (Beutler, 1971; Beutler, Jobe, & Elkins, 1974; Beutler, Johnson, Neville, & Workman, 1972; Binderman et al., 1972; Patton, 1969). However, consistent with basic research (e.g., Aronson, Turner, & Carlsmith, 1963; Bergin, 1962), there is an upper limit to how discrepant a counselor's communication may be and still produce change (Strong and Dixon, 1971).

Throughout history, a common means of changing others' attitudes and behavior has been to instill in them the fear of dire consequences that will result if one's advice or commands are not followed. For example, physicians describe the adverse effects of failure to comply with medical regimens, evangelists detail the horrors of hell to potential converts, and parents stress to their children the painful consequences of their misdeeds.

Several studies have examined the impact of fear-laden messages during attempts to persuade subjects to change dysfunctional behaviors such as smoking, improper dental care, and failing to use seat belts (see Higbee, 1969; Leventhal, 1970; McGuire, 1969; for reviews). This research shows that messages that induce fear can facilitate attitude and behavior change, as long as the fear produced by the message is relatively strong, the target believes that the fearful event has a high probability of occurring, and the target thinks that the fearful events will be avoided by changing his or her attitudes or behavior (Mewborn & Rogers, 1979; Rogers, 1975). Although therapists may be reluctant to use fear therapeutically, there are instances (such as those in which a client's behavior is dangerous to himself or herself or to others) in which other modes of persuasion may be buttressed by the use of fear to the client's benefit.

A few studies have examined the effects of communicator style on social influence, seeking to determine whether active, dynamic communicators are more effective than less active, subdued ones. Their findings are mixed, the results seeming to depend on the attributions the target makes about the communicator on the basis of message delivery.

For example, an early study found that "dynamic" presentations were less effective in producing attitude change and were more likely to be viewed by subjects as propaganda, suggesting that subjects inferred that dynamic communicators were trying to manipulate them (Dietrich, 1946). On the other hand, communicators who talk quickly are sometimes more effective than slow talkers, possibly because fast speech connotes knowledgeability and honesty, whereas slower paced talking seems overly cautious and suspicious (Apple, Streeter, & Krauss, 1979; Maclachlan, 1979; Miller, Maruyama, Beaber, & Valone, 1976).

Within the context of therapy, such stylistic factors have been found to be related to therapist-client interactions, but rarely to the outcome of therapy (see Pope, 1977, for a review). For example, Pope concluded that, although a high activity level eases the strain of the interaction between therapists and clients, there is no evidence that a high activity level is more therapeutic.

However, the degree to which a therapist is directive (offering explicit interpretations and advice, structuring and guiding the client's train of thought) versus nondirective does appear to have an impact on therapeutic outcomes. Specifically, clients who have an internal locus of control, who are defensive at the outset of therapy, or who are more highly educated prefer and benefit most from a nondirective therapist, whereas externally oriented, non-defensive, less educated clients do better with directive approaches (see Berzins, 1977; Goldstein, 1971; Pope, 1977). Thus, contrary to the schools of therapy that recommend either a directive or a nondirective therapeutic style (such as rational-emotive versus client-centered approaches), the scanty research available suggests that each approach may be optimal for certain client groups.

A therapist's effectiveness is also affected by his or her nonverbal behaviors (Waxer, 1978). Researchers have identified a set of nonverbal behaviors that enhance clients' ratings of the counselor and therapeutic effectiveness: a high degree of eye contact, a forward body lean, an attentive and concerned facial expression, maintaining a close interpersonal distance to the client (within three feet or so), smiling at the appropriate times, spontaneous (rather than "mindless") head nodding, and a relaxed, open body position (Lee, Hallberg, Kocsis, & Haase, 1980; Lee, Zingle, Patterson, Ivey, & Haase, 1976; Smith-Hanen, 1977).

The impact of these sorts of behaviors upon therapeutic outcome appears to be mediated by the inferences clients draw from them. Counselors who use attentive, socially skilled nonverbal behaviors are perceived as more expert, trustworthy, attractive, and persuasive and, thus, stimulate clients to be more open and to adopt the counselor's suggestions (Waxer, 1978). Indeed, in one study counselors who used "inattentive" nonverbal behaviors were rated only half as positively as "attentive" counselors (Krumboltz, Varenhorst, & Thoresen, 1967).

Clearly, more research is required before solid conclusions regarding the impact of message characteristics in counseling can be drawn and applied with confidence to therapeutic settings. Future research dealing with therapeutic style should pay close attention to the inferences clients draw from therapist behaviors, because the impact of therapist style is probably mediated by such inferences. Further, because there are great cross-cultural differences in how various stylistic and nonverbal behaviors are interpreted, researchers and practitioners should be cognizant of the limitations of this line of research (see Yuen & Tinsley, 1981).

A good deal of social psychological research has examined how characteristics of the recipients of a persuasive communication mediate the degree to which they are influenced by it, but this topic has not been explored seriously in counseling and psychotherapy research. Instead, most research has studied how similarities between clients and therapists affect the counseling session.

The strongest data relating client characteristics to the therapeutic process are provided by Goldstein's (1971) study of "YAVIS" clients (young, attractive, verbal, intelligent, and successful), who are generally more similar to their therapists in background, values, and life-style than are non-YAVIS clients. This similarity may affect not only the client's perceptions of the therapist but also the therapist's reactions to the client. Studies have shown, for example, that lower-class individuals are less likely to be accepted for counseling and psychotherapy, less likely to receive intensive therapy, and more likely to terminate treatment prematurely (Lorion, 1974; Parloff, Waskow, & Wolfe, 1978). Further, the socioeconomic status of a therapist's family of origin is related to his or her willingness to see lower-class patients (Kandel, 1966).

When therapists see non-YAVIS clients, their attitudes may be conveyed to the client in subtle ways that have an indirect effect on the outcome of therapy. For example, counselors tend to rate more positively clients who are physically attractive, are verbally fluent, and participate more fully in their therapy (Lewis, Davis, Walker, & Jennings, 1981); this attitude thus favors YAVIS clients. Not surprisingly, YAVIS and non-YAVIS clients react differently to counselors with varying characteristics (Goldstein, 1971).

The effects of clients' personalities and, specifically, the degree of similarity between clients' and therapists' personality profiles have received a modicum of attention. A variety of personality measures have been used, including the MMPI, the Myers-Briggs Type Indicator, the Fundamental Interpersonal Relations Orientation Scale (FIRO-B), the Personality Research Form, and a number of specific trait scales (see Abramowitz et al., 1982; Berzins, 1977; Parloff et al., 1978, for reviews).

The most consistent results have been obtained using the FIRO-B. The FIRO-B assesses the compatibility of two people in terms of their needs to originate and receive behaviors along three dimensions: inclusion, control, and affection. Without going into specific details, studies using the FIRO-B show that increased compatibility on some dimensions is related to treatment outcome. In the Goldstein (197 1) study mentioned above, for example, non-YAVIS clients were more attracted to their therapists over a 12-week period when they were highly compatible on the FIRO-B. Apparently, compatible therapist-client dyads find it easier to communicate than do incompatible dyads, thereby facilitating the therapeutic relationship and the outcome of therapy.

Finally, a few studies have examined the relationship between client personality and responsiveness to particular kinds of treatment under the assumption that people may be predisposed to prefer certain therapeutic approaches over others. We have already noted that internally controlled and well-educated clients prefer nondirective therapies, whereas externally controlled and less well educated persons prefer therapists who provide more structure and direction (Berzins, 1977; Goldstein, 1971). Further, subjects classified as "thinking" types on the Myers-Briggs Type Indicator rate rational-emotive therapy (which is highly cognitive) more highly than do "feeling" types, and indicate that they would be more likely to choose a rational-emotive therapist in the future (Stoltenberg, Maddux, & Pace, 1986).

The sparse research on client characteristics suggests that they interact with counselor characteristics and types of therapy to affect the therapeutic relationship and, in some instances, the outcome of therapy. The available data do not identify which client variables are most important, but the existing studies should sensitize researchers and practitioners to the fact that not all therapeutic approaches are equally effective on all types of clients (Leary, in press-b). This point was discussed by Kiesler (1966), who argued that researchers and therapists implicitly operate under a "myth of uniformity" that assumes that clients constitute a relatively homogeneous group. Future research is needed to identify important client characteristics that are related to therapist influence and client change as a function of specific difficulties and treatments.

As we reviewed the research based upon the social influence model of counseling and psychotherapy, we purposefully refrained from critiquing specific studies or the model upon which they are based. Our review completed, it is now important to discuss three general issues regarding the social influence model of counseling and psychotherapy.

Although many of the findings of counseling research are consistent with those in experimental social psychology, some inconsistencies have emerged both between the counseling and the social psychological studies and among several of the counseling studies themselves. For example, the effects of counselor attractiveness on clients' acceptance of therapeutic " messages" is not as strong or as straightforward as experimental social psychological research would suggest. Similarly, within counseling research, some studies have found effects of counselor trustworthiness, but many others have not.

Although social psychological and counseling researchers purport to study many of the same variables (credibility, fear-inducing communications, and so on), the studies themselves differ so markedly that it is nearly impossible to compare them directly. As a result, it is difficult to know whether failures to conceptually replicate basic social psychological findings using counseling paradigms are due to the validity of the hypotheses being tested, the generalizability of the effects from one domain to another, or differences in methods and subjects used. Although many of the findings of basic attitude change research do generalize to real and analogue counseling settings, the inconsistencies are troublesome and suggest that tighter lines should be drawn between the conceptualizations and research strategies used in basic and applied research.

Researchers and practitioners in counseling and psychotherapy are understandably interested in whether the findings of research studies generalize to real therapy settings. Three considerations relevant to this issue emerge from the social influence literature.

First, although most counseling studies contain essential ingredients of real therapy settings (at minimum, a "counselor" and a "client"), they differ from real-world therapy in important ways that affect the generalizability of their results. As noted earlier, most counseling studies involve a single contact between a counselor and client, and client change is often assessed as early as 20 minutes after the start of the session. Thus, though acceptably designed and well controlled, these studies are most relevant to the initial stages of therapy. (This is not to say that they are not relevant to later stages of therapy, only that their relevance is unclear.) Future research should investigate factors that affect counseling processes later in the therapeutic relationship, possibly drawing on models of relationship development in social psychology (e.g., Levinger, 1980).

Second, most social influence research, whether in social or counseling psychology, deals with psychological changes that are less complex than most of the issues for which people seek professional help. Although attempts have been made to address behaviors relevant to real counseling settings, it is difficult to study the remediation of problems as complex and intense as those seen regularly by practicing psychologists. This is not to berate the work that has been done, but to highlight its limitations and the need for studies of more serious client difficulties.

Third, most social influence research assumes, as a convenient fiction, that influence in therapy is unidirectional and static. In part, this is so because the social influence model was developed in the context of studies of mass communication in which the assumption of unidirectional influence from a speaker to an audience is tenable. However, when dealing with dyadic interactions, such as those found in therapy, this approach portrays only one aspect of the interpersonal process. Although dynamic, "systems" models of behavior that incorporate interdependence and mutual influence between the individuals are complex and typically difficult to test, such approaches may be needed to capture more faithfully the interpersonal dynamics of therapy (see Strong & Claiborn, 1982). We will return to this point momentarily.

Having discussed issues dealing with the generalizability of the social influence studies, we should make two other points. First, contrary to what is commonly assumed, the external validity of a study does not depend upon the degree to which the experimental setting mirrors the situation to which one would like the results to generalize (i.e., mundane realism). Regardless of the experimental setting or the subjects used, the external validity of a study is always an empirical question that requires conceptual replications in other settings using other subjects. Thus, a clinical field study using real therapists and a clinical sample is not inherently more generalizable than a laboratory experiment on social influence using artificial tasks and college sophomores (see Mook, 1983).

To take the point a step further, a strong case can be made that the results of specific studies should never be generalized to the real world, no matter where the study was conducted, what methods were used, or what kinds of subjects were employed. Generally speaking, research studies such as those reviewed in this chapter are conducted to evaluate hypotheses derived from theory. Only after repeated tests of theoretical propositions confirm a theory's merit should the theory be applied toward understanding, predicting, and controlling events in the real world. Viewed in this manner, any empirical finding is too suspect to be generalized outside of the immediate study in which it was obtained. The problem, however, is that little of the research discussed in this chapter was designed to test theoretical notions about social influence and psychological change.

The greatest weakness of the social influence literature, both within social psychology and counseling, is its generally atheoretical nature. By and large, research findings in the area represent a loosely organized collection of empirical

generalizations, unconnected by an overriding theoretical framework. To some degree, this problem can be traced to the original work by Hovland et al. (1953), which was only loosely based on a formal theoretical model. Studies of social influence have been revealing and useful, but as a whole they have done little to contribute to a deeper understanding of the processes involved in psychological change. Thus, we devote the last section of this chapter to a brief examination of three integrative theoretical models of social influence.

A complete examination of theories of attitude change and social influence relevant to counseling and psychotherapy would go far beyond the objectives of this book, but a brief overview of three important models may help explicate the interpersonal processes discussed in this chapter.

Although the original studies reviewed by Hovland et al. (1953) were not designed to construct or test a systematic theory of persuasion, they were based loosely on learning theory. In this framework, attitude change is regarded as a learning process, and change itself depends upon factors that facilitate the learning of the persuasive message. According to this view, whether an attitude is learned depends upon the degree to which the message is attended to, comprehended, and accepted. Thus, the various source, message, and recipient factors that faciliate attitude change may lead people to attend to, understand, and/or accept persuasive messages.

For example, people may pay greater attention to credible than noncredible sources and evaluate their arguments less critically, increasing the probability of acceptance. Similarly, fear-inducing messages heighten both attention and acceptance. The usefulness of this framework within counseling settings has not been fully evaluated, but it is at least plausible and simplistic enough to serve as a rough rule of thumb. When in doubt, counselors and psychotherapists may wish to proceed in a manner that promotes attention, comprehension, and acceptance of their communications to the client.

As noted earlier, traditional approaches to psychological change view social influence as unidirectional. Arguing that linear models of influence do not capture the richness of actual therapy relationships, Johnson and Matross (1977) have presented a "dynamic" approach to psychotherapy. In this model, counselor and client are said to have an interdependent relationship in which they negotiate the definition and achievement of mutual goals. The relationship between counselor and client involves mutual influence, with each affecting the other's beliefs, emotions, and behavior. Further, Johnson and Matross apply research on mixed-motive situations to analyze the process through which influence occurs.

According to this model, influence occurs because both the counselor and the client have resources upon which the other is dependent. Consistent with social psychological models of influence, Johnson and Matross maintain that influence requires that one party be dependent upon the other for valued resources (Thibaut & Kelley, 1959). To state their thesis in a single proposition:

A therapist influences a client to the extent that the therapist furnishes resources needed by the client for the accomplishment of highly valued goals and to the extent that the client cannot obtain these resources at a lower cost from other relationships. (Johnson & Matross, 1977, p. 405).

From this perspective, the factors that mediate social influence are those that cause the client to believe that the therapist possesses the resources needed for accomplishment of the client's goals.

An important feature of this model is that it acknowledges the dynamic nature of therapeutic relationships, focusing on the relationship over time and taking into account the many variables involved, the bidirectional nature of influence, and the possibility of feedback loops. Although the dynamic approach has not been widely studied or used, it highlights factors not included in other models of influence.

The newest theory to be applied to psychotherapeutic change is the elaboration likelihood model (ELM; Petty & Cacioppo, 1981). According to this approach, there are two distinct routes by which attitudes change as a result of persuasive communication. The central route involves attitude change that occurs as the result of conscious consideration of the merits of a communication. For central-route processing to occur, however, the individual must be both motivated and able to assess the merits of the received message. If these two conditions are met, whether the message will be influential depends primarily upon the individual's analysis of its merit.

However, if the individual lacks either the motivation or the ability to think about the message, attitudes may be affected by extraneous factors, such as characteristics of the source or message, or by pre-existing biases to accept or reject the message. Such factors may affect attitude change via the peripheral route without the individual having to do any extensive cognitive manipulation of the incoming message. According to Petty and Cacioppo (1981), attitude change that occurs via the peripheral route is less enduring than change that occurs centrally.

Although the ELM has been tested in only a handful of studies, it has the potential for making sense out of the complex patterns of results described in this chapter. Effects of counselor, message, and client factors may be inconsistent because other relevant factors (such as the client's motivation and ability to think about the message) were not taken into account. For example, according to the ELM, an important determinant of whether a communication is processed in a central or peripheral manner is the involvement of the target in the issue under consideration. Highly involved targets are more likely to process a message centrally, whereas less involved targets should use peripheral cues. Thus, the theory identifies an important factor not included in other models.

Both laboratory experiments (e.g., Petty, Cacioppo, & Goldman, 1981) and analogue counseling studies (Stoltenberg, Cacioppo, Petty, & Davis, 1985; Stoltenberg, & McNeill, 1984) provide empirical support for ELM, although the full extent of its usefulness has not yet been assessed.

That counseling and psychotherapy involve interpersonal influence processes is beyond question. What is not clear, however, is how this process should best be conceptualized. Although a huge body of research has identified factors that contribute to counselor influence, there is a pressing need for a theoretical model that can account for the available results. In our view, this model must deal both with the intricacies of the relationship between counselor and client and with the processes within the client that mediate cognitive, emotional, and behavioral change.

[1] Although we recognize that the client and the therapist influence one another, the emphasis in this chapter is only on the processes through which the therapist influences the client.

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.