Psychotherapy could not be conceived without the existence of a client-therapist relationship. The therapeutic meeting between patient and practitioner plays an essential role in the psychotherapeutic process. Psychotherapy researchers rate the relative contribution of relationship factors as accounting for 30% of the total change that patients make (Lambert, 1992). Therapeutic alliance has been identified as the main factor contributing to such psychotherapeutic relationship. Horvath and Greenberg (1994) explain that in the current notion of working alliance, «collaboration» between therapist and client is the key element. At its best, the working alliance provides a safe environment for clients to explore themselves and a relationship in which clients’ «key relational issues» are defined. The role of therapeutic alliance in promoting and facilitating therapeutic change was initially highlighted by psychoanalytically oriented psychotherapists (for a review, see Bordin, 1994), but is nowadays acknowledged by most theoretical approaches. Bordin (1979) defined the working alliance as consisting of three components: agreement on overall goals, agreement on tasks that lead towards achieving these goals, and emotional bond between the therapist and client.
Thus, the construction and validation of scales and inventories designed to assess key dimensions of effective therapeutic relationships has been one of the goals of psychotherapy researchers in recent decades. One of these questionnaires is the Working Alliance Theory of Change Inventory (WATOCI; Duncan and Miller, 1999).
The Working Alliance Inventory was developed by Horvath and Greenberg (1989). It is a self-report measure consisting of 36 items and three subscales of 12 items each representing three dimensions of working alliance (Bond, Goals, Tasks). A counselor and a client form are used to collect information regarding the reported strength of the working alliance. Horvath and Greenberg (1986) reported internal consistency estimates with alphas of .93 for the client total score and .87 for the therapist total score. The internal consistencies for the client sub-scales were reported as alphas of of .90, .88, and .91 for task, bond, and goal respectively. It was also found to have good convergent and divergent validity using a multitrait-multimethod analysis. The original 36-item WAI is also available in a short form version (WAI-S), comprised of 12 items (Tracey and Kokotovic, 1989).
The WATOCI is a 17-item pencil and paper version of the well-established Working Alliance Inventory (WAI). The WAI (short version, Tracey and Kokotovic, 1989) was developed and validated in relation to Bordin’s transtheoretical model of alliance (Bordin, 1994)—see Horvath and Greenberg (1986) for further information regarding the validation of the WAI. The shortened version of the WAI is a 12-item self-report measure that uses a 7-point Likert rating scale (1= never, 7= always) to yield both an overall score for alliance quality and three summed sub-scale scores (4 items per each sub-scale in the WAI short version):
- Bond: the emotional bond of trust and attachment between patient
and therapist. Some of the facilitative conditions that help to create such
a bond are: mutual understanding between patient and therapist, a caring attitude
on the therapist’s side, and the patient’s perception that the therapist likes
him or her.
- Goals: the degree of agreement concerning the overall goals
of treatment. Thus, the client is aware that such goals are relevant and he
or she identifies him or herlself with the themes made explicit and implicit
during the therapeutic process. The therapist has some direct or indirect
evidence that the goals established in the therapeutic relationship are shared
with and accepted by the client.
- Tasks: the degree of agreement concerning the tasks relevant
for achieving these goals. Both therapist and client feel that the tasks agreed
upon during the therapeutic process are rational, reachable, and closely related
to the therapeutic goals (Horvath and Greenberg, 1986).
Tracey and Kokotovic’s (1989) factor analysis of the WAI yielded an alliance overall main factor and three other factors that accounted for the three alliance sub-scales. The WAI-S (short version) derived from this study and it yielded a Chronbach’s alpha of .94. In the same study, the internal consistency reliabilities were .90, .84 and .88 for task, bond and goal scores, respectively (Tracey and Kokotovic, 1989).
Duncan and Miller (1999) added 5 items dealing with the agreement between patient’s and therapist’s theory of change to the reduced version of the WAI. This brief report presents the psychometric properties of the Spanish version of the WATOCI (see Table 1).
Research participants were 102 adults (79 women and 23 men) receiving outpatient psychotherapy. The mean age for the participants was 30.0 years (SD= 8.74). Their presenting complaints were: anxiety (41.7% of the sample); mood disorders (39.9%); and interpersonal/relational difficulties (17.5%). All of them completed the Spanish version of the WATOCI at the end of the third psychotherapy session.
Instrument’s coefficient of reliability based on the internal consistency assessed by Cronbach’s Alpha is .93. The internal consistencies for the sub-scales were reported as alphas of .91, .85, .86,and .0,82 for task, bond, goal, and theory of change respectively.
Kaiser-Meyer-Olkin Measure of Sampling Adequacy and Bartlett’s Test of Sphericity indicated that the factor model was appropiate. We also performed an exploratory Principal Component Analysis (with Varimax rotation), and it extracted three factors (eigenvalues over 1) with eigenvalues of 5.66 (first factor); 2.81 (second factor); and 2.59 (third factor). The three factors account for 65% of the total variance.
As can be seen in Table 2, the first factor is made up by a combination of (a) all items belonging to the tasks sub-scale; (b) two items belonging to the goals sub-scale, and (c) four items belonging to the theory of change sub-scale. The second factor is made up by a combination of (a) two items belonging to the goals sub-scale, and (b) one item belonging to the theory of change sub-scale. The third factor is made up by three of the four items belonging to the bond sub-scale. Thus, the first factor is made up of items belonging to all subscales, the second one is made up of some items belonging to the goals subscale and one belonging to the theory of change subscale, and the third one is made up of items belonging to the bond subscale.
Thus, it seems that the tasks, goals, theory of change subscales, and one item belonging to the bond subscale account for the first two factors, whereas three of the four items belonging to the bond subscale account for the third factor.
The internal consistency reliabilities for the total score, and subscales were good and agree with Tracey and Kokotovic’s results (Tracey and Kokotovic, 1989).
A closer look at the Principal Component Analysis results reveals that the Spanish version of the WATOCI does not fit well with the structure that could be expected from Bordin’s (1994) theory of therapeutic alliance. Factors obtained in the analysis do not discriminate between items belonging to different subscales; items from all subscales are attributed to the first factor, items from two different subscales are attributed to the second one, whereas some items in the bond subscale are attributed to the third factor.
These results seem to suggest (a) that the la theoretical structure of the instrument should be further refined, and (b) that therapeutic alliance may be a more unified construct than is usually thought of. While it is theoretically possible to divide it into different components, clinical research with psychotherapy patients seems to indicate that goals, tasks, bond and theory of change are closely related and are not orthogonal factors.
Thus, the results of this study are partially in agreement with those of the one by Tracey and Kokotovic (1989) mentioned before n which they reported finding a first factor made up of items belonging to all subscales. Our own results validate the notion that therapeutic alliance as assessed by the WATOCI is a coherent construct, but they also cast some doubts on the supposedly threefold structure of such a construct. Also, Duncan and Miller’s (1999) addition of the «theory of change» sub-scale does not add any meaningful clarification to the structure of the previous WAI short version.
This work was carried out while Sergi Corbella was holder of a Grant awarded by the Spanish Ministry of Education and Culture (AP 98).