The association between alliance (at a given point in time or aggregated across several sessions) and outcome is one of the most consistent findings in psychotherapy research1, 2. However, the mechanism underlying this association is one of the most controversial. Some theorists and researchers believe that alliance is therapeutic in itself; others argue that it is a by‐product of effective treatment or of a trait‐like patient ability to benefit from treatment3, 4. For many years, the debate has been confined mainly to the domain of theory. Recently, several studies have applied advanced analytic strategies to explore the mechanism behind the alliance‐outcome association.
The argument that alliance is simply a by‐product of successful treatment has been previously addressed by studies controlling for early symptomatic change when examining the ability of alliance to predict outcome. Some of these studies suggest that alliance is indeed a by‐product of early symptomatic change, while others indicate that it can predict outcome even after controlling for that change1. However, previous studies treated alliance as a static variable, and ignored the fact it can change across treatment, which may have contributed to the mixed results. Recent studies used statistical methods such as autoregressive cross‐lagged modeling to examine whether alliance levels precede symptomatic levels, session by session over the entire course of treatment. The findings show that alliance indeed precedes symptom reduction over the course of treatment in both psychotherapy5, 6, 7 and psychopharmacotherapy8, suggesting that it is a true predictor of outcome.
The other challenge to the argument that alliance is therapeutic is the proposition that alliance is a by‐product of a patient's general trait‐like ability to benefit from treatment. Individuals who are more capable of forming strong and satisfying relationships with others may also have a better chance of forming a strong and satisfying alliance with their therapist. Alliance cannot be said to be therapeutic in itself if it is a trait‐like characteristic of the patient. Recently developed detrending and centering methods9 have made it possible to explore empirically the theoretical distinction between the state‐like and trait‐like components of alliance and determine which of the two predicts outcome. Studies show that patients’ pre‐treatment interpersonal characteristics can predict alliance as it develops across treatment10 and that the alliance trait‐like component can significantly predict outcome7, 11. However, studies also suggest that state‐like changes in alliance over treatment can have a significant effect on outcome5, 7, 11.
If state‐like changes in alliance can bring about therapeutic change, manipulating these characteristics is expected to influence outcome. One recent study has examined this question empirically, randomizing patients to either a feedback condition, in which therapists received feedback on the alliance to assist them in strengthening its state‐like component, or to a control condition in which no feedback was provided. The study found a greater effect of the state‐like component of alliance on outcome in the feedback condition7, suggesting that the effect of this component of alliance on outcome can indeed be manipulated. Furthermore, another recent study suggests that when therapists detect poor alliance with their patients, and have sufficient time to work on strengthening the state‐like component of alliance, this component is associated with a better outcome12.
The groundbreaking methodologies recently applied in psychotherapy research bring new insights to our understanding of the question of whether alliance is therapeutic. These methodologies are poised to play a critical role in future research, focusing on diverse populations and therapeutic orientations, and may lead to the development of even more advanced models of moderation‐mediation analyses.
Sigal Zilcha‐Mano Department of Psychology, University of Haifa, Haifa, Israel
References
- 1. Crits‐Christoph P, Gibbons MBC, Hamilton J et al. J Consult Clin Psychol 2011;79:267‐78. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. Wampold BE. World Psychiatry 2015;14:270‐7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. DeRubeis RJ, Brotman MA, Gibbons CJ. Clin Psychol Sci Pract 2005;12:174‐83. [Google Scholar]
- 4. Zilcha‐Mano S, Barber J. J Contemp Psychoanal 2014;50:58‐88. [Google Scholar]
- 5. Falkenström F, Granström F, Holmqvist R. J Couns Psychol 2013;60:317‐28. [DOI] [PubMed] [Google Scholar]
- 6. Zilcha‐Mano S, Dinger U, McCarthy KS et al. J Consult Clin Psychol 2014;82:931‐5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Zilcha‐Mano S, Errázuriz P. J Couns Psychol (in press). [DOI] [PubMed] [Google Scholar]
- 8. Zilcha‐Mano S, Roose SP, Barber JP et al. Psychother Psychosom 2015;84:177‐82. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Wang LP, Maxwell SE. Psychol Methods 2015;20:63‐83. [DOI] [PubMed] [Google Scholar]
- 10. Zilcha‐Mano S, McCarthy KS, Dinger U et al. Psychotherapy 2014;51:424‐33. [DOI] [PubMed] [Google Scholar]
- 11. Hoffart A, Øktedalen T, Langkaas TF et al. J Couns Psychol 2013;60:471‐82. [DOI] [PubMed] [Google Scholar]
- 12. Zilcha‐Mano S, Solomonov N, Chui H et al. J Couns Psychol (in press). [DOI] [PMC free article] [PubMed] [Google Scholar]