Journal of Consulting and Clinical Psychology 1986, Vol. 54, No, 1,27-31 Copyright 1986 by the American Psychological Association, Inc. 0022-006X/86/S00.75 Structural Analysis of Social Behavior: Application to a Study of Interpersonal Process in Differential Psychotherapeutic Outcome William P. Henry, Thomas E. Schacht, and Hans H. Strupp Vanderbilt University Research strategies employing Structural Analysis of Social Behavior (SASB) were demonstrated in a study comparing 4 therapists, each of whom saw a good and a poor outcome case (N = 8), on interpersonal process variables in the third session. SASB represents complex interactive processes in a manner that is both theoretically cogent and empirically sound. Good versus poor therapeutic outcome was differentiated on the basis of the following interpersonal process variables: greater levels of "helping and protecting" and "affirming and understanding*' and significantly lower levels of "blaming and belittling" were associated with high-change cases. Patient behaviors of "disclosing and expressing" were significantly more frequent in high-change cases, whereas "walling off and avoiding" and "trusting and relying" were significantly more frequent in low-change cases. Additionally, negative comple- mentarity was greater in poor outcome cases. Implications for research methodology and interpersonal theory are discussed. Psychotherapy process and outcome factors are often split into three categories: (a) patient antecedents; (b) therapist techniques; and (c) relationship (so-called "nonspecific") variables. Research to date has emphasized antecedent patient and relationship vari- ables, although generally failing to demonstrate unique effects of therapist technique, leading to unsettling speculations that perhaps therapist techniques are unimportant (e.g.^ Parloff, Waskow, & Wolfe, 1978). Traditional research methods foster this view of therapy as composed of discrete patient, therapist, and relationship vari- ables. However, alternate approaches are possible. For example, Kiesler (1982) argued against considering patient and therapist variables isolated from their reciprocal dyadic context. Kiesler elaborated how problems in living appear as the sequelae of rigid, self-defeating communication patterns that collectively compose an interpersonal evoking style. The central antecedent patient variable is the client's interpersonal evoking style, and the main therapist technique variable is the therapist's manner of re- sponding to this style. Thus conceived, patient and therapist variables join in a final irreducible pathway: the therapeutic re- lationship. In this view, interpersonal transactions in the therapy dyad should become the fundamental unit of psychotherapy process analysis. Accordingly, technique is not distinct from patient and relationship variables; instead, maintaining and exploring the relationship is therapy's central technical task (not simply an enabling background condition for application of techniques). Therapists do not simply supply a good or bad relationship; rather, they use it technically as both the context for and the substance of psychological change. This work supported in part by National Institute of Mental Health Grants MH-16247 and MH-20369, Hans H. Strupp, principal investi- gator. Correspondence concerning this article should be addressed to William P. Henry, 134 Wesley Hall, Department of Psychology, Vanderbilt Uni- versity, Nashville, Tennessee 37240. Unfortunately, most studies of patient-therapist interaction rely on crude measures of unilateral behaviors that neglect the exact nature of the dyadic transactions while also failing to ar- ticulate with any particular interpersonal theory. The present study reports a methodological demonstration designed to in- crease precision and/or ties to interpersonal theory in the study of psychotherapy transactions. Circumplex models, from a psychometric standpoint, are the most sophisticated and theoretically coherent models of inter- personal behavior (cf. Wiggins, 1982, for review). Structural Analysis of Social Behavior (SASB; Benjamin, 1974, 1982) is the most detailed, conceptually rigorous, and empirically vali- dated of current models. SASB was selected for the present study in accordance with Schaffer's (1982) guidelines for therapy pro- cess research. Specific advantages of SASB along these lines in- clude the following: (a) It provides a research method congruent with theoretical premises about interpersonal process in psycho- therapy; (b) it permits extremely fine-grained analysis of virtually any interpersonal event, and (c) it uses small rating units judged by methods requiring relatively low inference and permitting high specificity. In the present demonstration, the SASB model was applied to 15-min segments of early therapy sessions drawn from four pairs of cases, each pair containing a high-change case and a low-change (i.e., good and poor outcome) case, treated by the same therapist under controlled conditions (described in Strupp, 1980a, 1980b, 1980c, 1980d). The following four hypotheses were designed to illustrate the kinds of clinically and theoretically meaningful questions addressable via the SASB model. 1. Communications should fall into different categories of in- terpersonal action (represented by SASB clusters) in high-change as compared to low-change cases. 2. Hostile and controlling therapist behavior (SASB Cluster 6, Surface 1) should predict poor outcome, consistent with Truax's (1970) observation that therapist's criticism was related to poor outcome and with Strupp's (1980a) finding that negative 27