Therapist Effects: How Much Does the Person Delivering Treatment Matter?

When we test whether a treatment works, we typically focus on what the treatment is: CBT versus medication, exposure versus relaxation, twelve sessions versus six. But there is another variable that affects outcomes at least as much and receives far less attention: who delivers the treatment.

The size of the effect

A systematic review of therapist effects estimated that the therapist accounts for approximately 5% of the variance in patient outcomes [2]. Other estimates range from 3% to 8%, depending on the study design and setting. In routine practice settings, where therapists have more freedom in how they work, the estimates tend to be higher than in tightly controlled trials where therapists follow manuals.

Five percent of variance does not sound like much. But to put it in perspective: the difference between active treatments in psychotherapy comparison trials is often around 0–3% of outcome variance. The therapist effect is larger than many of the treatment differences researchers spend years and millions of dollars trying to detect.

In a large routine-practice dataset, Saxon and Barkham found that the most effective therapists achieved recovery rates roughly double those of the least effective therapists, even when case complexity was accounted for [6]. These are substantial practical differences masquerading as a small statistical percentage.

What makes some therapists better

If some therapists consistently outperform others, the obvious next question is why. The answer is disappointing: we don’t know very well.

The alliance. The therapeutic alliance — the quality of the working relationship, agreement on goals, sense of collaboration — is the most robustly identified predictor of outcome. A meta-analysis that accounted for therapist effects found that the alliance-outcome association, while significant, was modest [4]. Better therapists tend to form better alliances, but the alliance does not fully explain why they get better results.

Experience. Contrary to reasonable expectation, years of experience do not consistently predict better outcomes. A longitudinal study tracking therapist outcomes over careers found that, if anything, therapists showed slight deterioration over time on average [5]. This finding is counterintuitive and controversial, but it has been found in multiple datasets.

Training and orientation. Whether a therapist is trained in CBT, psychodynamic therapy or another approach has less predictive power than you might expect once you control for what they are treating. This fuels the “common factors” argument: that what matters is the relationship, warmth, expectation and structure rather than the specific techniques.

What is not well studied. Factors like therapist responsiveness (adjusting to the individual patient), moment-to-moment skill in delivery, and the ability to manage ruptures in the alliance have been proposed as important but are difficult to measure across large samples.

How therapist effects complicate trial evidence

Most randomized controlled trials of psychotherapy report outcomes as though patients within each therapist are independent. They are not. Patients treated by the same therapist share a common influence, and ignoring this statistical non-independence can lead to overly narrow confidence intervals and inflated significance levels [1].

In practical terms: if a trial compares CBT and relaxation therapy, and the CBT happens to use two particularly skilled therapists while the relaxation therapy uses two average ones, the apparent superiority of CBT may partly reflect who delivered it. Trials that use many therapists per condition reduce this risk, but many use only two to four per arm.

Multilevel modeling, which nests patients within therapists, is the appropriate analysis. Some journals now require it, but many influential older trials did not use it.

What this means for understanding anxiety treatment

The therapist effect has specific implications for how we evaluate anxiety treatments:

Exposure therapy. The effectiveness of exposure for anxiety depends heavily on implementation — pacing, timing, the therapist’s willingness to tolerate the patient’s distress, and how safety behaviors are handled. Therapist skill in managing these elements plausibly explains a large portion of outcome variance.

Treatment comparison trials. When two active treatments produce similar outcomes (as they often do in anxiety), the null difference may partly reflect therapist effects averaging out rather than genuine equivalence. If the best CBT therapist outperforms the best relaxation therapist but the worst CBT therapist underperforms the worst relaxation therapist, the aggregate comparison shows no difference.

Dissemination. Moving a treatment from a trial setting (where therapists are carefully selected, trained and supervised) to routine practice (where therapist skill varies widely) predictably reduces average effectiveness. This “voltage drop” is a therapist effect by another name.

How to read research with therapist effects in mind

1. How many therapists delivered each treatment? Fewer therapists mean more confounding.
2. Was a multilevel model used? If patients are not nested within therapists, confidence intervals may be too narrow.
3. Were therapists crossed across conditions? If each therapist delivers both treatments, between-therapist differences are controlled.
4. Was therapist competence or adherence measured? Some trials rate therapist fidelity to the manual. High fidelity reduces, but does not eliminate, therapist variability.

For a broader look at what makes a trial informative versus misleading, see what a Cochrane review tells you.

When to look for more support. If you have tried therapy for anxiety and it did not help, it may be worth trying again with a different therapist rather than concluding that therapy does not work for you. The fit between you and the therapist matters. Ask your GP for a referral or look for a therapist trained in evidence-based approaches for your specific condition. If you are in crisis, call or text 988 (US) or Samaritans on 116 123 (UK and Ireland).