Overview and Development of the Instrument
The Family Therapy Alliance Rating Scale (FTARS) was developed to fill an important gap, to estimate the strength of the working alliance in conjoint family treatment. To date, no single indicator of the therapeutic process has been shown to be more powerful in predicting client outcome than the therapeutic alliance (Horvath & Symonds, 1991). Not only is the alliance a meaningful predictor of therapeutic success across theoretical orientations (Bachelor, 1991), but also clients’ perspectives on the alliance early in treatment consistently predict outcome weeks or months later (Horvath & Symonds, 1991).
For individual psychotherapy, the alliance can be assessed using a variety of measures -- client- or therapist-reported as well as observer-reported (Tichenor & Hill, 1989). In the couples and family literature, however, there is only one published instrument, Pinsof and Catherall’s (1986) integrative alliance scales. Research on this measure, which is based on Bordin’s (1979) tri-partite conceptualization of the alliance, has been shown to be reliable and predictive of session impact (Heatherington & Friedlander, 1990) and treatment outcome for both couples and families (Bourgeois, Sabourin, & Wright, 1990; Johnson & Talitman, 1997; Pinsof & Catherall, 1986; Quinn, Dotson, & Jordan, 1997).
Like the individual alliance measures, Pinsof and Catherall’s (1986) instrument focuses far more on therapist behavior than on family members’ behaviors. Indeed, little is known about what observable behaviors contribute to a strong alliance in couples and family therapy (Friedlander & Tuason, 2000). In the absence of this knowledge, therapists (and supervisors) can only rely on clinical judgment in assessing the strength of the alliance. To fill this gap, we developed an observer rating scale of client behaviors for this treatment modality.
Observer scales of the individual therapy alliance could not be adapted to family therapy because the presence of multiple family members, children as well as adults, makes this format unique in several respects. Two dimensions of our instrument reflect this uniqueness. These dimensions are operationally defined as the Shared Sense of Purpose Within the Family scale and the Safety scale. The former was included because family members do not always value the therapy equally. It has been demonstrated that a “split” alliance (i.e., some family members are more involved and more strongly aligned with the therapist than others) occurs frequently (Heatherington & Friedlander, 1990). Our instrument reflects Pinsof’s (1994) recommendation that alliance measures include the within-system alliance -- family members’ collaboration with one another in relation to the therapy. The latter dimension, safety, reflects another unique aspect of family treatment (Friedlander & Tuason, 2000). In several respects, family therapy involves more and different risks for clients than does individual therapy. When people come for help with their families, they risk having secrets or shameful events revealed against their will. Some may find out that their spouse or partner is planning to leave them. Others may be punished or physically hurt by family members who are angry about what took place in the session. Indeed, clients’ most important relationships can be put at risk when they seek professional help in a family context.
To construct the instrument, we began by reviewing the theoretical and empirical literature on the therapeutic relationship in family therapy. This process resulted in the identification of a set of descriptors illustrating both a positive and a negative alliance. These descriptors were then used to create an initial pool of items illustrating different aspects of client involvement and collaboration. Guiding the item construction was the requirement that each item be an observable behavior. Thus, for example, rather than, “Family members are interested in each other’s perspective,” the item reads, “Family members ask each other for their perspective.”
To refine the initial item pool, our research team (N = 4) observed 12 videotaped family sessions for which we had the clients’ self-reported perspectives on the alliance from an earlier study (Heatherington & Friedlander, 1990), i.e., data from Pinsof and Catherall’s (1986) integrative alliance scales. With the knowledge of each family member’s perspective on the alliance, we searched for individual and family behaviors that might be indicative of their feelings and cognitions. Item editing occurred through repeated comparisons and negotiation of our ratings. Discrepancies prompted us to clarify items and discard those that were difficult to rate.
Originally, we attempted to rate each behavioral indicator on a 7-point Likert scale for every family member. Because this process was difficult and laborious, we clustered similar items and inductively identified 4 underlying dimensions. Then, the first 4 authors independently wrote definitions of these constructs, compared and integrated them, and decided which items logically related to each dimension. Testing this new procedure by rating 6 different videotapes indicated that the instrument’s interjudge agreement could be improved considerably.
Next, we used a sorting task to assess the content validity of the instrument. To do so, we randomly ordered the 44 items and asked a sample of family therapy process researchers in the U.S. and Canada to indicate which items reflect each of the 4 underlying constructs. The item, “Family members ask each other for their perspective,” for example, is located in the cluster reflecting the Shared Sense of Purpose Within the Family dimension. If at least 3/4 (75%) of our respondents selected the same dimension for a given item, it was retained; if not, it was eliminated or moved to a different cluster. (Participants were invited to select more than one dimension but to circle the most important one.) Participants were also asked to comment on the items or suggest additional ones.
The sorting task was also completed by eight Spanish family researchers. The FTOARS was translated into Spanish, and the participants performed the same task and commented on the cultural appropriateness of the items and the four underlying dimensions.
Results of both this sorting task by participants in North America and in Spain indicated a high degree of consistency, and most of the items were viewed as behavioral exemplars of the dimensions we had originally selected. Indeed, the majority of items were viewed similarly by well over 75% of the judges in both samples. Based on the criteria described above, several items were moved from one dimension to another or modified in wording. (A couple of items were removed that did reached the 75% criterion in only one sample; we decided that it was important to make the English and Spanish instruments identical.) None of the items needed to be eliminated because of a lack of cultural appropriateness for Spanish clients.
The FTOARS is designed to be completed by trained judges while observing a videotaped family therapy session.. At a minimum, the judges should be graduate students in a mental health specialty, but a great deal of clinical experience is not necessary. It is recommended that at least three judges observe and rate the same family session for purposes of assessing interjudge reliability.