The alliance construct in psychotherapies: from evolution to revolution in theory and research

The construct of alliance (alternatively addressed as therapeutic, working or helping) was first formulated within psychoanalytic circles, before it was reconsidered in trans-theoretical terms and became recognized as an integrative variable, common factor, and generalizable change process or “principle of” change^{1, 2}.

Much has been written over the years about the role of alliance in the adherence to various specific treatment tasks defined as critical to change (e.g., emotional insight and skill development), but also about alliance development as effecting change or “curative” in and of itself³.

In the analytic literature, the evolution of the construct can be traced from Freud to Greenson, with a number of notable contributions in between. The construct was developed to highlight the importance of collaboration and the real and human aspects of the patient-therapist interaction. Interestingly, it did not receive much attention in the interpersonal and humanistic literatures, where these aspects were always central.

The construct complemented transferential considerations of patient-therapist relationship and provided a ground for technical flexibility, i.e., for departing from the idealized stance of therapist abstinence and neutrality. It did not come, however, without criticism and concern regarding its orientation towards patient identification or compliance with the analyst’s agenda³.

Bordin⁴ broke boundaries with his seminal reformulation of the alliance as comprised of “purposeful collaboration” (patient-therapist agreement on the tasks and goals of treatment) and their “affective bond” (that is, mutual respect and trust, as well as emotional attunement), thus introducing the application of the construct to other orientations.

This coincided with or contributed to the psychotherapy integration movement that attempted to identify common change processes and that in turn adopted the alliance construct as its poster variable. With its emphasis on mutuality and orientation towards negotiation, Bordin’s reformulation permitted greater attention to therapist participation and subjectivity.

Part of the post-modern turn or relational revolution that chal­lenged the rigid demarcation between subjectivity and ob­jectivity, and recognized the inextricable relationship between the observer and the observed, Safran and Muran⁵ provided an intersubjective elaboration that concentrated on the person of the therapist and the negotiation of existential dialectics around agency/communion and subject/object in the alliance. According to this elaboration, the resolution of these dialectics in the context of the alliance represents an opportunity for change – that is, a new relational or corrective experience.

Beginning in the 1970s, the alliance construct became the focus of the psychotherapy research community, in large part due to Bordin’s reformulation, which led to the development of many measures and a proliferation of research demonstrating the predictive relationship of alliance to outcome (see Norcross and Lambert⁶ for a meta-analysis of 306 alliance studies, N=30,000).

This generation of research did not come without some controversy: multiple measures not surprisingly resulted in some definitional imprecision or confusion, and much of the research was observational and correlational, failing to address the question of causality. However, there have been more recent mediational analyses to establish the causal relationship of the alliance as a “change mechanism”⁷. There has also been some (though limited) research on patient and therapist factors or characteristics that moderate the quality of the alliance^{1, 6}.

An extension of the research on the alliance-outcome relation included analyses of alliance patterns based on repeated post-session ratings to identify “v-episodes” (precipitous drops and then returns to recovery), and pre- to post-session ratings to identify “sudden gains” (significant increases) as proxies for alliance rupture repair. A meta-analysis of eleven such studies (N=180) has demonstrated that precipitous “drops” or ruptures are quite prevalent (15-80% of sessions) and subsequent “gains” or repairs predict outcome⁶.

There is also research that directly assessed the presence of rupture, and found that patients report rupture in 20-40% of sessions, therapists in 40-60% of sessions, and third-party observers in 40-100% of sessions⁶. These direct assessments included self-reports (by patients or therapists) of any “tension or problem, misunderstanding, conflict or disagreement”, and observations (by third parties) of “confrontation” (movements against other) and “withdrawal” (movements away from self or other) behaviors that mark ruptures. The prevalence of rupture that these studies demonstrate highlight the inherent messiness and conflict in human relations, including patient-therapist interactions^{5, 8}.

These efforts (despite limitations in number and other methodological concerns) have been integral to a “second generation” of alliance research, particularly aimed at the construct of rupture (generally defined as deteriorations or breaches in relatedness) and the clarification of repair processes^{3, 6, 9}. This second generation has included mixed method (quantitative and qualitative) efforts or task analyses (six small-scale studies) that have yielded clinically useful “when/then” data and defined stage-process models of rupture repair as a “change event”.

More specifically, these efforts defined specific tasks to carry out in the face of rupture, beginning with an acknowledgement of the rupture and including an exploration of rupture experience and sometimes some renegotiation of the work of therapy and/or a formulation of the patient’s presentation (all of which can be construed as resulting in a new or corrective experience). These efforts also led to experimentally designed studies that evaluated the effect of alliance-focused trainings aimed to advance therapist abilities to address ruptures (six studies, N=276), which provided limited but promising support^{6, 9}.

Future directions for consideration regarding the alliance construct include the need for: a) more definitional clarification and consensus on alliance and rupture (both suffer from too many definitions and methodological translations that seem too removed from the original conceptualization); b) more research on the causal relation of alliance development and rupture repair (more study of how each of these effect overall change); c) more research on patient (personal characteristics, intervention responsiveness) and therapist (personal characteristics, technical interventions) factors (specifically how these variables moderate alliance development and rupture repair).

In addition, there is a need for: d) more research on rupture repair processes, and more efforts to develop observer-based measures and to apply mixed method studies to explore what processes (i.e., specific patient and therapist behaviors and interactions) are essential to repair, and e) more experimental research on alliance-focused trainings (protocols designed to develop therapist abilities to negotiate alliance) and their potential effect on psychotherapy process and outcome.

These second-generation efforts could significantly address the risk of failure posed by alliance rupture and consequently redress the rates of failure in psychotherapy, including premature termination and poor adherence to treatment protocol.