The social cognitive model of counselor training.

Although supervisors are ultimately interested in and responsible for positive client outcomes, the supervisor's sphere of influence is with the counselor, not the client. Likewise, psychologists are often responsible for a staff of counselors. Given that one important tool through which the client gets better is the counselor, supervisors need to pay close attention to models that describe how best to train counselors to be efficacious with clients.

At least two major lines of research have focused on the counselor rather than on the client. First, there is evidence that beginning-level counseling skills can be taught to counselors in a systematic way based on three well-researched training programs, namely, Human Relations Training (Truax & Carkhuff, 1967; Truax, Carkhuff, & Douds, 1964), Interpersonal Process Recall (Kagan, 1984), and Microcounselling (Forsyth & Ivey, 1980) (see Baker, Daniels, & Greeley, 1990, for a full review). Second, clinical and counseling psychologists continue to identify in a predominantly atheoretical way therapist variables that affect client outcomes (Beutler, Machado, & Nuefeldt, 1994).

Although the atheoretical research on therapist variables is essential, Beutler and colleagues (1994) expressed the need for more theory-driven research. A similar desire was expressed by Tracey (199 1), who cited Meehl's (1978) earlier critique that quantitative programmatic research in our field needed to be more theory driven with specific parameters specified in the theory. The basic notion is that systematic research will contribute to theory revision and expansion.

Beyond the research benefits of theory-driven research, counselor trainers and supervisors need direction as to which counselor variables to attend to given the limited time they have and the sheer number of therapist variables that are present. Moreover, supervisors recognize that the counselors' actions depend partly on their supervision and their clients. Counselors produce more effective actions with clients, in part, because of the quality of their supervision experience and the nature of their clients (e.g., presenting problems, readiness for therapy). Trainees receiving excellent supervision while seeing clients who are motivated to change and whose problems are manageable are in an ideal learning environment. Finally, supervisors are acutely aware of how supervisees perceive identical events quite differently. Astute clinicians know that supervisees filter their experiences differently. Most supervisors would say that attending exclusively to what the counselor is doing in the session is insufficient. The supervisee's motivation, thinking processes, confidence, and personality are relevant.

Yet, with the notable exception of the development models (e.g., Loganbill, Hardy, & Delworth, 1982; Stoltenberg, 1981; Stoltenberg & Delworth, 1987; Watkins, 1993), there is a scarcity of theories to conceptualize the complexity of the supervisor's jobs. Within the counselor training literature, previous writing has focused primarily on the trainee's skills and responses (see Baker et al., 1990, for a review) but has paid less attention to variables that may influence actions with clients such as the counselor's cognitive processes, goals, and counseling self-efficacy (CSE). Likewise, in supervision training, the focus is primarily on the development of the counselor over time and how the supervisor can facilitate her or his development. Less emphasis has been given to the dynamic processes of the counselor in session with the client (e.g., Mueller & Kell, 1972).

For example, as a beginning supervisor, I was looking for a conceptual map to frame my upcoming experience. On teaching the supervision didactic component over these past 12 years, I have found rich descriptions of the supervisory process (e.g., Mueller & Kell, 1972), the counselor's development (e.g., Loganbill et al., 1982; Stoltenberg & Delworth, 1987; Watkins, 1993), and relevant issues such as ethics (e.g., Bernard & Goodyear, 1992), multicultural issues (e.g., Martinez & Holloway, 1997; Ponterotto, Casas, Suzuki, & Alexander, 1995), and evaluation (e.g., Bernard & Goodyear, 1992). I was overwhelmed by the sheer volume of variables presented, each argued as important. I needed assistance in thinking about what the salient constructs were, how they were related, and when individual differences or the larger context became a barrier. For me, what was lacking was a heuristic model that would organize and synthesize disparate components.

Albert Bandura's (1977, 1982, 1986a, 1986b, 1988, 1989, 1990, 1991a, 1991b, 1993) social cognitive theory (SCT) seemed to provide a parsimonious framework. Bandura (1982, 1986a) posits that human beings exercise control over their thought processes, motivation, and actions. Because judgments and actions are partly self-determined, people can effect change in themselves and their situations through their own efforts. SCT concerns the self-determining mechanisms through which such changes are realized.

One of the most pervasive and central of these self-generated influences is self-efficacy estimation, defined as one's judgment about her or his capabilities to execute a desired course of action within a particular domain. Self-efficacy beliefs function as a major determinant of human action in a particular domain in that they affect choice of behavior, duration of behavior, effort expenditure, persistence, emotional reactions, and thought patterns (see Bandura, 1986a).

In the Larson and Daniels (1998 [this issue]) article, 32 studies were presented in which CSE was related to other important counselor training variables such as counselor performance, counselor anxiety, and the supervision environment (e.g., Johnson, Baker, Kopala, Kiselica, & Thompson, 1989; Ridgway & Sharpley, 1990; Sipps, Sugden, & Faiver, 1988). CSE is described as one's beliefs or judgments about one's capabilities to effectively counsel a client in the near future (Larson et al., 1992). Larson and Daniels synthesized the CSE literature by organizing current findings around SCT, even though the studies themselves focused exclusively on the self-efficacy construct apart from the larger theory.

SCT explains how learning complex actions like counseling occurs. Self-efficacy beliefs, along with the intervening cognitive, affective, and motivational processes, serve as the causal link between knowing what to do and executing the action. People with higher self-efficacy beliefs will tend to have more self-aiding thoughts; experience anxiety as challenging rather than debilitating; and set more realistic, moderately challenging goals. Bandura (1990) modified his earlier definition of self-efficacy beliefs to include these intervening processes: "Perceived self-efficacy is concerned with people's beliefs in their capabilities to mobilize the motivation, cognitive resources, and courses of action needed to exercise control over task demands" (p. 316). Self-efficacy beliefs, along with the intervening affective, cognitive, and motivational processes, are referred to by Bandura as personal agency (see Figure 1). They are the self-determining influences that make human beings able to adapt to ever-changing, complex environments. The way this process happens is called "emergent interactive agency" or what is more commonly referred to as triadic reciprocal causation. In triadic reciprocal causation, personal agency, action, and the environment operate as interacting determinants (see Figure 1). Any attempt to explain human actions must include self-generated influences as a contributing factor (Bandura, 1989). The two-way arrows in Figure 1 represent triadic reciprocal causation. This dynamic process does not occur in a vacuum. Rather, the counselor's stable characteristics (e.g., racial identity) provide the personal context, whereas the training environment and the larger social and/or cultural environment provide the larger context in which triadic reciprocal causation occurs (see Figure 1).

[Figure 1 ILLUSTRATION OMITTED]

For Bandura (1990), these domain-specific actions are complex and are not reducible to discrete microskills. Bandura (1986a) gives the example that driving a car is not an array of subskills (e.g., starting the ignition, checking rear view mirror every 10 seconds, parallel parking), but rather driving a car is about maneuvering busy arterial crossroads during rush hour (p. 397). To produce effective, complex actions requires "continuously improvising multiple subskills to manage ever changing circumstances most of which contain ambiguous, unpredictable, and often stressful elements" (p. 391).

This description also captures the essence of the complexity of effective counseling. The focus of this article is on applying SCT to counselor training. SCT is one model to explain how supervisors can train counselors to be efficacious with clients,

Consider the following example: For 23 weeks, beginning practicum trainees were followed beginning 3 weeks prior to practicum and continuing 3 weeks after practicum (Larson et al., 1993). All participants gave informed consent to being in a case study. I was in the unique position of serving as both the instructor to the counselors and serving as the supervisor to their respective supervisors. I observed both the counseling sessions and the supervisory sessions live throughout the semester. Their CSE beliefs, outcome expectancies, anxiety, self-evaluations of their counseling performances, and their supervisors' evaluations were assessed each week. Each week, supervisors, who were advanced doctoral students, observed their supervisees' counseling sessions as well as their colleagues' supervisees in session. At the end of the semester, trainees rated their skill level in comparison to peers. Their supervisors also rated the trainees' normative skill level.

Two trainees, fictitiously called Terry and Logan, were consistently viewed by their supervisors, the instructor, and other classmates, as being respectively-the stronger and weaker members of the practicum in terms of counseling skills. (To protect their anonymity, all identifying information has been removed from their descriptions; they and their supervisors are presented as gender neutral. All information was collected individually and kept confidential.) The two supervisors' mean evaluations of their performances over time resulted in Terry being rated I standard deviation higher than Logan. Also, the supervisors' mean evaluation of client outcomes showed Terry to be rated 6 points higher on a 12-point scale. Moreover, Terry's client(s) reported more successful experience(s), their goals were met, and they would recommend Terry to a friend. In contrast, Logan's client(s) prematurely terminated. As the instructor, I frequently observed their work with clients and noticed crucial differences. Terry tracked her or his clients well and the clients seemed to improve. By contrast, Logan struggled to track her or his clients and appeared judgmental and gave advice in sessions. In sum, based on the multiple sources, their counseling actions were more and less effective. Although Terry's performance and Logan's performance were clearly different, they both reported similar CSE estimations and self-evaluations overall. Moreover, their CSE beliefs both increased about I standard deviation. They saw about the same number of clients over time and were at the same level of training.

This example illustrates the intent of this article. What may seem like a complex and confusing array of events in counselor training is clearly predicted and explained by applying SCT to counselor training. Specifically, the article is structured to meet seven goals. The first goal is to introduce SCT to the reader. Major assumptions of SCT will be identified and modifications in translating the model to counselor training will be noted. It is important to note that Bandura (1977, 1982, 1986a, 1986b, 1989, 1990, 1991a, 1991b, 1993), in his extensive presentation of SCT, did not address the domain of counseling or counselor training. Throughout the article, when the Social Cognitive Model of Counselor Training (SCMCT) is referred to, it is the translation of SCT to counselor training that is being presented. It is understood that the hypotheses were derived from SCT with some modifications.

The second goal is to describe the major components of the SCMCT as illustrated in Figure 1. The three primary components involve the counselor's personal agency, which are the self-determining influences, the counselor's effective actions in counseling and supervision, and the proximal environment: namely, the counseling session and the supervisory session, in which the learning of those counseling actions primarily occurs.

The third goal is to describe how the three major components presented in Figure 1 dynamically interact. Specifically, the goal is to describe how triadic reciprocal causation results in counselors learning to be efficacious in session with clients. SCT is complex, as is the SCMCT, and in this format, only key relationships will be highlighted. The process of how counselors regulate their actions and simultaneously generate new actions during counseling and supervision will be articulated. This process will be related to common language in the counselor training literature such as tracking the client and receiving feedback.

The fourth goal is to describe the context in which triadic reciprocal causation occurs. Relatively stable characteristics of the counselor (e.g., racial identity) will be discussed first because these attributes form the crucial internal context in which triadic reciprocal causation occurs (see Figure 1). Next, the relatively stable characteristics of the supervisor and client, the supervisory relationship, the client-counselor relationship, the training environment, and the larger context will be presented.

The fifth goal is to specify how these contextual variables may emerge as barriers to impede triadic reciprocal causation. Lent, Brown, and Hackett (1994) identified contextual variables as potential barriers (e.g., sexism, racism) that interfered with the development of interests. With respect to Figure 1, the barriers are stable characteristics of the counselor, client, and/or supervisor that may emerge to negatively affect the counselor learning to be efficacious with clients. Other barriers may be embedded in the larger context (e.g., the training environment). Likewise, barriers within the counselor (e.g., homophobia) or in the environment can emerge as obstacles to the counselor learning to be efficacious with clients.

The sixth goal is to identify a research agenda to guide future research. The intention is to demonstrate how fellow researchers could systematically and collectively examine the relationships among the major constructs in the model. The seventh goal is to begin the process of translating the SCMCT to practice, especially practicum training.

In meeting the seven goals, the metaphor of a drama will be used. The drama is about the counselor learning to be efficacious with clients. First, the uniqueness of the drama as well as its overlap with similar dramas in which the counselor is the central character will be described. Second, the cast will be introduced with the counselor in the leading role and the supervisor and client in supporting roles. The self-determining aspects of the counselor as a human agent will be described as well as or his efficacious actions in supervision and counseling. From the example above, two of those self-determining variables are CSE and self-evaluation of counselor performance. The supervisor and client are introduced in context of the two scenes in which the drama occurs, namely the supervision hour and the counseling hour.

Third, the plot is presented. The plot involves the counselor, as a self-determining agent, learning to be efficacious with clients through triadic reciprocal causation. The counselor primarily learns this complex action during supervision and while seeing clients. The plot unfolds in two interweaving scenes, the counseling hour and the supervision hour. The way the counselor learns to be efficacious with clients depends in large part on how the counselor cognitively mediates her or his interactions with the supervisor and the client(s).

Fourth, the internal context of the counselor (i.e., relatively stable characteristics) and the external context (e.g., the training environment) will be described to add depth and perspective to the plot. Fifth, the plot will be described when it goes awry and not go as planned. In this section, crucial stable counselor/supervisor/client characteristics are described as to how they may emerge as barriers to obstruct the plot. Sixth, seven research propositions are presented. Finally, applications of the SCMCT for practice will be presented.

INTRODUCING THE SCMCT

SCT is distinct from other psychological theories in that other theories have focused on the acquisition of knowledge or on the performance of the actions. Previously, counselor training has focused on how to best present the declarative knowledge via lecture classes (e.g., theories of counseling, ethical guidelines) and procedural knowledge (e.g., practica, devoted to skill development) to the trainees. Much of the research has focused on increasing effective counselor responses (Baker et al., 1990) or on counselor development over time (e.g., Stoltenberg & Delworth, 1987). SCT focuses on the self-referent thought that mediates the transformation between knowing what to do and executing the action. The intent of the SCMCT is to examine the interrelationship of the causal determinants of the transformation between the knowledge we impart to the trainee and the resultant actions by the counselors.

In the SCMCT, CSE beliefs would be viewed as the primary causal determinants of effective counseling action. Several studies have shown CSE to be a significant predictor of counselor action (Larson et al., 1992; White, 1996). Following from Bandura's work (Bandura, 1982, 1986a, 1989, 1990, 1993), in the SCMCT, CSE is hypothesized to affect choice of counselor responses, effort expenditure and persistence in the face of failures, and risk taking behavior. CSE beliefs, based on SCT, would be hypothesized to affect counseling actions through the mediating influences of other self-generated processes including affective processes, motivational processes, and other cognitive processes. Only one study has provided evidence that CSE, along with anxiety, significantly predicted counselor performance (Larson et al., 1992). Other studies have clearly shown an inverse relation between CSE and anxiety as predicted by SCT (Alvarez, 1995; Daniels, 1997; Friedlander, Keller, Peca-Baker, & Olk, 1986; Larson et al., 1993; Larson et al., 1992). Based on SCT (e.g., Bandura, 1991b), the SCMCT would assume that people with higher CSE would be more likely to view their anxiety as challenging, to set realistic, moderately challenging counseling and supervision goals, and to have thoughts in supervision and counseling that are self-aiding. These self-generated influences are the essence of what it means to be a human agent. In SCT, they are hypothesized to act collectively in what Bandura (1989) labels as personal agency (see Figure 1), a dynamic, interactive, complex system that allows humans to be both responsive to an ever-changing environment and proactive in determining that environment.

According to SCT, the way in which human beings are both reactive and proactive is through triadic reciprocal causation (see Figure 1). The triad of personal agency, the person's complex actions, and the proximal environment dynamically interact. This interactive emergent agency (Bandura, 1989) happens constantly as the person learns complex actions. The implication to the SCMCT is that this process is happening while therapists are learning the complex actions involved in counseling clients.

As human agents, translating from SCT, counselors (and supervisors) filter their perceptions of the counselor's actions, often arriving at different assessments as evident in the earlier example of Terry and Logan. SCT, with its emphasis on the cognitive mediation of events, differentiates between perceptions of actions and observable indicators of actions. As human beings, our actions are filtered through our own self-evaluation.

The SCMCT presents four sources thought to increase CSE in descending order of potency: (a) mastery, (b) modeling, (c) social persuasion, and (d) affective arousal (Bandura, 1989). CSE is hypothesized to be largely determined and modified through the counselor's cognitive appraisal of the four experiential sources. Mastery would include successfully seeing clients, and modeling would include participant modeling (viewing one's own successful counseling session on videotape). Social persuasion would include the supervisor supporting, encouraging, and structuring learning situations for the counselor to succeed with clients. Affective arousal would include the anxiety associated with seeing clients, especially for beginning practicum counselors. Logan, the less efficacious counselor, did not achieve mastery, although she or he may or may not have perceived that. Perhaps her or his supervisor needed to provide more modeling experiences and needed to be more socially persuasive. These sources will be presented in a somewhat different order. Mastery will be included under the counselor's effective counseling actions. Modeling and social persuasion will be described when the supervisor is presented. Affective arousal will be included as one of the dynamic processes within the counselor. Practicum includes all four sources.

MODIFICATIONS

Given the extensive knowledge base concerning the counseling endeavor, several modifications to SCT have been made in developing the SCMCT. First, Bandura, in his later writings (e.g., 1989, 1991b, 1993), conceived of action and motivation as including both a behavioral and cognitive component. His point was to …