The present study utilized an empirically derived coding system to identify content categories in the spontaneous verbalizations of social phobics in the context of a short-term cognitive-behavioral group therapy. The coding system was applied successfully to segments of transcriptions of eleven group sessions. Evaluation of changes in the content categories utilized by two subjects over the course of the life of the group indicates the usefulness of this coding system in identifying meaningful shifts in cognitive processes during treatment.
Social phobia is a significant problem among clinical and nonclinical populations (1), yet it has not received enough attention in the literature. Cognitive-behavioral group therapy is an effective treatment of social phobia for both adolescents and adults, showing positive effects on self-report measures of specific and generalized psychological distress as well as on behavioral measures of performance (2, 3). Cognitive-behavioral researchers believe that the key to alleviating social fears is a change in these patients' thought patterns, manifested in the content of their speech. However, they report only outcome data and do not apply process measures in their empirical research.
Monitoring the prevalence of content categories over the course of therapy could provide information about patients' maladaptive mental patterns, about the integrity of treatment (i.e., whether the discourse within sessions corresponds to the stated goals of the therapy), and about therapeutic progress. The specific content is likely to vary depending on the treatment approach as well as the patients' problems. Theoretically, in cognitive-behavioral treatments, the content of patients' discourse should reflect changes in their cognitive processing of their target problem; in successful therapy, a cognitive-behavioral therapist would expect a shift from preoccupations with the symptoms in early sessions to engaging in productive cognitive coping mechanisms in later sessions. In comparison, a psychodynamic therapist would expect a shift from discussions of symptoms to discussions of underlying conflicts that gave rise to the symptoms in the first place (4, 5).
There are several approaches to the use of content analysis of psychotherapy sessions. Most studies have used generic, predetermined categories in coding individual therapy sessions (6, 7). Similarly, most studies of discourse in group therapy have used theoretically based classification systems (8). Examples include: The Group Emotionality Rating System (GERS), based on Bion's writings; The Member-Leader Scoring System, originally developed by Mann; The System for Analyzing Verbal Interactions (SAVI), also based on Bion's theory of group dynamics; Structural Analysis of Social Behavior (SASB); and The Individual Group Member Interpersonal Process Scale, based on interpersonal circumflex to group processes. In contrast, our approach to analyzing content was empirically based. This approach is more in line with the linguistic approach labeled as decontextualizing. According to Tesch (9), this would involve first decontextualizing meaningful segments from the total body of text. Subsequent analysis and interpretation is then a process of recontexualizing, by putting the categories back together in new ways that are somewhat independent of the original context, based on their psychological meanings. Lofland and Lofland (10) describe tentative text-analytic categories as thinking units, emphasizing the contribution of the investigator's initial understanding even in empirically derived categories. Formalized as the constant comparative approach (11), these units are likely to be modified as the researcher is gaining more information by the close attention to the text. Thus, constructing a coding system empirically is a recursive process, in which observations are structured by preceding understandings and in turn also changing those understandings. …