Therapeutic Alliance

Definition

The therapeutic alliance, or working alliance, is the quality of collaboration between therapist and client. Bordin's formulation, which remains standard, has three components:

• Goals: agreement on what the work is for.
• Tasks: agreement on what will be done.
• Bond: the affective relationship.

It is a property of the pair, not of the therapist or the client alone.

History

The concept has psychoanalytic roots but was made pantheoretical by Bordin in 1979, which is what allowed it to become the common ground for comparing treatments that agree on almost nothing else.

Theory

The alliance sits at the centre of the common-factors argument: that what different psychotherapies share may matter more than what distinguishes them.

The central unresolved question is causal direction. Alliance is measured during treatment, and clients who are already improving may report a better alliance as a result of improvement rather than as a cause of it.

Disentangling this requires measuring alliance and symptoms repeatedly and separating within-client from between-client variance. Studies that do this find a smaller but still present effect.

Research

The alliance-outcome correlation is around r = .28 across hundreds of studies, remarkably stable across modalities, populations and measurement approaches.

Better designed studies find that the therapist's contribution to alliance predicts outcome while the client's contribution predicts it less, which is evidence against the pure reverse-causation account.

Alliance ruptures and their repair have their own literature, and repaired ruptures are associated with better outcomes than sessions with no rupture at all.

Applications

• Routine outcome monitoring with alliance feedback improves outcomes for clients who are not progressing.
• Rupture repair training is now a component of several supervision models.
• Alliance is a reasonable early indicator: a poor alliance at session three is worth addressing directly rather than waiting.

Criticism

• Causal direction remains contested, and the headline correlation is often reported as though it were settled.
• Common method variance. The same client often rates both alliance and outcome, and shared method inflates the correlation.
• The effect size is modest. r = .28 is around 8 percent of outcome variance, which is meaningful but is not the whole of therapy.
• It can become an excuse. Attributing all change to relationship can be used to avoid the question of whether a technique works.

References

• Bordin, E. S. (1979). The generalizability of the psychoanalytic concept of the working alliance. Psychotherapy, 16(3), 252-260.
• Flückiger, C., et al. (2018). The alliance in adult psychotherapy: A meta-analytic synthesis. Psychotherapy, 55(4), 316-340.
• Zilcha-Mano, S. (2017). Is the alliance really therapeutic? American Psychologist, 72(4), 311-325.
• Eubanks, C. F., et al. (2018). Alliance rupture repair: A meta-analysis. Psychotherapy, 55(4), 508-519.