Expectation, Placebo and Hypnosis: Why Comparison Groups Matter

Suppose a group of people try hypnosis for a problem and most of them feel better afterwards. That observation alone says surprisingly little. Several explanations could produce the same improvement, and only some of them involve hypnosis. This explainer sets out those explanations, the comparison groups researchers use to rule them out, and what the evidence says about the question people most often ask: is hypnosis just placebo?

Why people improve without the specific treatment

Improvement after treatment is a sum of several parts, and most of them are not specific to the treatment.

• Spontaneous improvement. Many problems ease or fluctuate on their own. A meta-analysis of 37 three-armed trials, which randomized patients to no treatment, placebo or an active intervention, found that, on average, spontaneous improvement and placebo each accounted for a substantial share (24% and 20%) of the improvement seen in the actively treated groups [1].
• Regression to the mean. People tend to seek help, and enrol in studies, when symptoms are at their worst. Unusually extreme measurements tend to be followed by more typical ones, which can make natural variation look like real change [2].
• Placebo effects. A Cochrane review of 234 trials that randomized patients to placebo or no treatment found no important clinical effects of placebo in general, but effects on patient-reported outcomes, especially pain and nausea. The reviewers noted that it is hard to distinguish such effects from biased reporting [3].
• Attention and care. Time with an attentive practitioner, a calm setting and a structured routine can help in themselves.

None of this is a flaw in participants. These are ordinary features of human experience, and good study design aims to separate them from the effect of the intervention.

One result, two answers: the insomnia example

A 2015 meta-analysis of hypnotherapy for insomnia shows how much the comparison matters. Compared with a waitlist, hypnotherapy shortened the time it took to fall asleep (standardized mean difference −0.88). Compared with a sham intervention, there was no significant difference [4]. Eleven of the 13 studies had low methodological quality, with high risk of bias in blinding and in the design of control interventions [4].

Both statements are true. “Hypnotherapy beats doing nothing” and “hypnotherapy has not been shown to beat a credible sham” are different claims, and a headline that reports only the first is misleading.

The main types of comparison group

Comparison	What it rules out	What it leaves open
Wait-list or no treatment	Natural course, regression to the mean	Expectation, attention, reporting bias
Usual care	The same, in real-world conditions	Expectation, attention; usual care varies
Attention or education control	Contact time and attention	Credibility and expectation, if lower than hypnosis
Active treatment	Whether hypnosis beats a realistic alternative	Whether either has a specific effect
Sham hypnosis	Expectation and ritual, if the sham is credible	Whether the sham is truly inert
Same therapy with and without hypnosis	What the hypnotic element adds	Whether the label alone explains the gain

Wait-list and no treatment

These account for natural course and regression to the mean, but not expectation. They may also flatter the treatment. In a network meta-analysis of 49 depression trials of cognitive behavioural therapy, the odds of response with no treatment were 2.9 times higher than on a wait-list, suggesting that being told to wait may slow improvement [5]. The authors cautioned that the evidence quality was less than ideal and the finding did not survive adjustment for publication bias [5].

Attention controls and active comparisons

A three-arm trial in 328 veterans with chronic pain compared eight group sessions of hypnosis, mindfulness meditation or pain education [6]. All three groups improved by the end of treatment, with no overall difference in average pain intensity. At three and six months, the hypnosis group had greater reductions in pain intensity, pain interference and depressive symptoms than the education group, and no outcome differed between hypnosis and mindfulness at any point [6]. A trial with only a wait-list arm would have reported a much simpler story.

Sham hypnosis

A credible sham is the most direct way to address expectation. In a 2025 trial, 250 postmenopausal women used six weeks of either self-administered hypnosis audio or white noise presented as hypnosis, a control designed to match placebo and expectancy effects. Hot flash scores fell 53.4% with hypnosis and 40.9% with the sham, a significant difference [7]. The sham group’s large improvement is itself informative: it suggests that much of the improvement in the hypnosis arm, too, came from factors other than the hypnosis content.

Adding hypnosis to an existing therapy

Another design keeps everything constant except the hypnotic element. A meta-analysis of 18 such studies found that adding hypnosis to cognitive behavioural therapy improved outcomes enough that the average client receiving it did better than at least 70% of clients receiving the same therapy without hypnosis [8]. The authors noted how few procedural differences there were between the conditions [8]. That makes the result striking, but it also raises the question of what exactly the hypnotic framing contributed.

Same procedure, different label

Labels alone change responses. When the same induction was presented as “relaxation” it produced a modest rise in suggestibility; presented as “hypnosis” it produced a very significant one [9]. Any comparison in which one group is told it is receiving hypnosis and the other is not carries that difference in expectation with it.

Is hypnosis “just placebo”?

The honest answer is: partly expectation, not only expectation, and the boundary is hard to draw.

Evidence that expectation matters. In a laboratory pain study of 60 participants, imaginative suggestions given without hypnosis did not differ significantly from hypnotic suggestions, a placebo did not differ from either and beat no treatment only when given after them, and pain reduction was mediated by expectancy [10].

Evidence that hypnosis adds something. Sham-controlled trials such as the hot flash study still found a significant added effect [7].

Evidence that they are not the same trait. If hypnosis were simply placebo, the most hypnotizable people should be the strongest placebo responders. Two double-blind studies found this link inconsistent: hypnotic suggestibility did not significantly correlate with placebo pain relief induced through conditioning plus verbal suggestion (rs = 0.11), while non-hypnotic suggestibility showed a modest correlation (rs = 0.27) with placebo relief induced through verbal suggestion alone [11].

Several theories treat expectation and response to suggestion as central to how hypnosis works, not as side effects of it. On that view, “controlling for expectation” risks removing part of what is being tested. That does not make research impossible. It means each study needs to be clear which question it answers: does hypnosis beat doing nothing, beat a credible alternative, or do something beyond expectation? For how individual differences complicate this, see What Is Hypnotizability?.

Five questions to ask of any claim

1. Was there a comparison group, and was assignment to it random?
2. What did the comparison group receive, and did it seem as credible as hypnosis?
3. Did both groups get similar time and attention?
4. Were outcomes self-reported, objective or both?
5. Does the conclusion match what the comparison can actually show?

For the rest of the trial-reading checklist, see What Is a Randomized Controlled Trial? Reading Hypnosis Research.

What comparisons cannot settle

A well-chosen comparison answers a specific question for a specific population. Beating a wait-list does not show that hypnosis beats other active treatments. A result in one condition does not transfer to another. And no single study determines whether hypnosis is appropriate for an individual.

When to see a clinician first. If you are weighing hypnosis for a health condition, treat it as something to add to established care, not a replacement, and discuss it with the clinician treating you. See a doctor promptly about new, severe or worsening symptoms, pain without a diagnosis, or low mood with thoughts of self-harm, rather than waiting to see whether hypnosis helps.