Circling Science

Among the therapies

Lineage · Anatomy · Among the therapies · Side by side

Several of the earlier practices on the Origins page come from therapy or sit close to it: Gestalt therapy; Focusing, which grew out of Carl Rogers’s client-centred therapy; and the encounter groups, linked above all with Rogers. Gestalt, Focusing-oriented and client-centred therapy belong to what research calls the humanistic-experiential family. This page places that family among the other kinds of therapy on two questions: how they compare on the evidence, and how they differ in what the therapist does.

Everything here concerns the treatment of clinical problems, given by therapists to clients. None of it is evidence about Circling, which isn’t a therapy, and none of the sources we read mentions it. The last section says what carries over and what doesn’t.

Four families

The grouping is ours; the reviews cited below each draw their own lines.

Cognitive-behavioural therapies. The therapist directs the session, sets homework, teaches skills for coping with symptoms, explains the disorder and its treatment, and keeps the focus on the client’s thinking and on what comes next. In comparative studies of what happens in sessions, these six activities set this family apart from psychodynamic therapy (Blagys & Hilsenroth, 2002).

Psychodynamic and interpersonal therapies. The therapist attends to feelings and how they are expressed, to what the client avoids, to patterns that recur, to the past, to relationships, to the therapy relationship itself, and to wishes, dreams and fantasies (Blagys & Hilsenroth, 2000; Shedler, 2010). Both lists come from comparing these two families with each other, so neither sets a family apart from every other approach.

Humanistic-experiential therapies. Client-centred (or person-centred), emotion-focused, Gestalt, Focusing-oriented and related therapies. Rogers described the therapist’s skill as going into a climate in which clients can do their own work, built by showing them that their feelings are understood and accepted. On the therapist’s side his conditions are genuineness, empathic understanding and unconditional positive regard (Rogers, 1946, 1957). A 2021 review of the family notes that all of its therapists guide the client’s process to some degree, and the emotion-focused, Gestalt, psychodrama and motivational-interviewing ones more than the rest (Elliott et al., 2021).

Two that cut across them. Mindfulness-based programmes teach formal meditation, with practice at home; that is how one large review defines them (Goldberg et al., 2018). Group therapy gives many of these treatments to several clients at once (Burlingame et al., 2016).

On our reading, Circling’s moves sit closest to the humanistic-experiential family, which also works with present experience and with the relationship. Of the psychodynamic list they share two items: attention to feeling, and to the relationship in the room. Of the cognitive-behavioural list they share almost nothing. A circle sets no homework, teaches no coping skills, explains no disorder, follows no plan made in advance and stays in the present.

How they compare on outcomes

In depression, a network meta-analysis of 331 trials with 34,285 adult patients found no significant differences between cognitive-behavioural therapy, behavioural activation, problem-solving therapy, third-wave, interpersonal and psychodynamic therapies (Cuijpers et al., 2021). Each did better than usual care, by half to three-quarters of a standard deviation (standardised mean differences of 0.50 to 0.73). Non-directive supportive counselling did less well, at 0.32, and helped a smaller share of patients than any other therapy. In the trials at low risk of bias that gap was no longer significant, and the authors point out that counselling was often the control condition, so it may not have been given at its best. Their category isn’t the same as client-centred therapy, which the humanistic-experiential review counts separately.

The limits are large. Of the 331 trials, 211 tested cognitive-behavioural therapy. The authors write that almost every trial comparing two therapies is far too small to settle the question, and they rate nearly all their estimates as low or very low in certainty.

Other reviews point the same way. In 23 trials, manual-guided psychodynamic therapy came out equivalent to treatments of established efficacy, within a margin fixed in advance and with researchers from both camps taking part (Steinert et al., 2017). Mindfulness-based programmes did as well as evidence-based treatments, overall and for anxiety and depression (28 comparisons, d = −0.004; Goldberg et al., 2018). The same treatment worked as well in a group as one to one (46 studies, g = −0.01; Burlingame et al., 2016). And the humanistic-experiential therapies came out level with other therapies in head-to-head studies (d = −.08), with large change in clients from the start of therapy to its end (d = .86; Elliott et al., 2021).

Not every review agrees. Counting only comparisons between bona fide treatments, one found cognitive-behavioural therapy ahead of psychodynamic therapy, though not of interpersonal or supportive therapies, and only for anxiety and depression (26 studies; Tolin, 2010). The humanistic-experiential review found an uncertain edge of .26 in cognitive-behavioural therapy’s favour, mostly in trials run by cognitive-behavioural researchers against versions of the humanistic therapies that weren’t bona fide (Elliott et al., 2021). “All therapies work equally well” is one reading of this evidence, not something it has shown.

The relationship, across every kind of therapy

The third task force on the therapy relationship, drawn from divisions of the American Psychological Association, commissioned a meta-analysis of each element it studied, published in 2018, and rated the evidence (Norcross & Lambert, 2018, 2019). On our reading, these are the elements closest to Circling:

Element Studies Pooled effect The task force’s rating
The working alliance, in adult therapy 295 r = .28 Demonstrably effective
Empathy 82 samples r = .28 Demonstrably effective
Positive regard 64 g = .28 Demonstrably effective
Cohesion in group therapy 55 r = .26 Demonstrably effective
The real relationship, person to person 16 r = .38 Probably effective
Repairing ruptures in the alliance 11 r = .29 Probably effective
Genuineness 21 r = .23 Probably effective
Self-disclosure and immediacy 21 none: a qualitative synthesis Promising, too little research

An r is a correlation and a g a standardised difference, so positive regard’s g of .28 can’t be read against the two r’s of .28 above it.

What these show: in treatment, better relationships go with better outcomes, across kinds of therapy. What they don’t show: that the relationship causes the improvement. All but two of the reviews found no randomised trial able to test that.

The ratings are the task force’s own judgement, and they don’t follow the size of the effect: the real relationship, with the largest correlation here, is rated only probably effective. The alliance also fills three of the nine top ratings: in individual therapy, with children and adolescents, and with couples and families. That the relationship matters as much as the method, or more, is the task force’s position, and Wampold (2015) argues much the same for the factors all therapies share. A later review concludes that no factor, shared or specific, has yet been shown to be how therapy works (Cuijpers, Reijnders & Huibers, 2019).

Therapeutic presence, which researchers from the humanistic-experiential family have described and measured, isn’t one of the task force’s elements. The evidence we found is a questionnaire (Geller, Greenberg & Watson, 2010); a randomised study in which a five-minute centring exercise before sessions raised therapists’ sense of presence, and clients rated those sessions as more effective though not their therapists as more present (Dunn et al., 2013); and recent studies of trainee counsellors in China, in which the presence trainees reported in one session predicted a stronger alliance in the next (Wu et al., 2026).

What carries over to Circling, and what doesn’t

None of the effect sizes carry over. They measure the treatment of clinical problems. The relationship findings are about a therapist and a client, or a group led by a therapist; the outcome comparisons set recognised treatments against one another; presence is a therapist’s, rated by the therapist or the client. Circling is none of these.

What does carry over is narrower, and still useful:

  • Words researchers already use, with measures. Alliance, empathy, genuineness, cohesion and presence are all measured with questionnaires, such as the Working Alliance Inventory and the Therapeutic Presence Inventory. Versions for peers in a circle could be tried (Active ingredients).
  • The size of therapist effects. Differences between therapies are small next to differences between therapists: about d = .20 at most between treatments, against therapist effects of d = .35 in trials and .55 in routine practice (Wampold, 2015). Who leads a circle may matter more than the school they trained in. In the Stanford study of encounter groups, harm clustered under one style of leading, and a leader’s school said little about how they led (Yalom & Lieberman (1971); Safety, facilitation and harm).
  • A control for future studies. If a well-led circle is good for people, the first explanation to rule out is what many approaches share: attentive, accepting company. A study that wants to credit Circling itself would compare it with something that offers that company too (A controlled trial of Circling).

Sources

We read Cuijpers et al. (2021), Elliott et al. (2021), Goldberg et al. (2018), Shedler (2010), Wampold (2015) and Rogers (1946) in full, and the rest from their abstracts. The task force’s ratings come from its summary as reproduced by Findings, a research digest.