Hypnotic Suggestibility Scales: What They Measure and What They Don't

The dinner-party sentence here is “some people are just hypnotizable and some aren’t,” usually said with a knowing look, as if there’s a switch in the brain and a test that reads it. I used to say the skeptic’s version: “it’s a gullibility test with a fancy name.” Both sentences assume the score measures something simpler than it does. Out of how many suggestions? On the classic scales, about a dozen, given once, on one afternoon.

I’ve already covered what hypnotizability is and what it predicts and the Stanford and Harvard scales themselves. This post is the follow-up I kept wanting to write: when you see “high hypnotizable” in a study, what exactly was counted, and what are people reading into that count that it can’t carry?

What a scale actually records

Most suggestibility scales follow the same recipe [1]:

1. A scripted induction, read identically to everyone, live or recorded.
2. A fixed list of test suggestions, running from easy (a hand feels heavy and drifts down) to hard (not hearing a sound, forgetting part of the session).
3. A pass or fail for each item, judged against a behavioral criterion, then summed.

That’s the core. Newer scales add a second layer: how real and how involuntary each response felt. The SWASH, a 10-item group scale tested in 418 people, scores both, and the experiential scoring turned up something I find quietly devastating for the old method: responses to the post-hypnotic suggestion item appear to be inflated when you score behavior alone [3]. Someone can do the thing without the experience the item is supposed to capture.

The scales are good at what they do. A 2016 study gave 230 people both the newer Elkins Hypnotizability Scale and the Stanford Form C. The Elkins scale had adequate internal consistency (α = .78) and the two scores correlated highly (ρ = .86) [2]. If you want a reliable count of how people respond to standardized suggestions, the field has several.

The critics’ point is about what the count throws away. A 2021 review argued that binary scoring, testing each suggestion only once, and some dated item content limit what the classic scales can tell us [1]. A person who almost passes an item and a person who doesn’t respond at all get the same zero.

Where “high” and “low” come from

The labels in study abstracts sound like categories. They’re cut points. By convention, researchers often call the top 10-15% of scorers in a sample “high” and the bottom 10-15% “low,” with everyone else in the middle [3]. Change the sample and the same raw score can land in a different band. Most people sit in the middle, responding to some suggestions and not others.

That convention has a practical consequence for reading trials. When an experiment recruits only the top scorers, it’s testing suggestion in the most responsive slice of people, and the effect it finds is likely to be bigger than you’d see in a waiting room. When a study compares highs with lows and leaves out the large middle, the contrast is sharpened on purpose. Neither design is wrong. Both are easy to over-read, and I over-read them for years in the other direction, dismissing every high-versus-low experiment as “testing the gullible.” The design isn’t testing gullibility. It is testing the extremes, and it says so in the methods section, which I wasn’t reading.

What it may be measuring instead of “hypnosis”

This is the part I didn’t know when I was making jokes at dinner, and it changed how I read every paper that sorts people into highs and lows.

By definition, hypnotizability is supposed to be the extra responsiveness that hypnosis produces. In practice, scales measure responsiveness to suggestion after an induction, which is not the same thing. Kirsch made the case in 1997: the correlation between suggestibility with and without hypnosis comes close to the scales’ own reliability, which means the scales mostly measure ordinary, waking suggestibility [4]. The induction adds something, but the score isn’t a clean measure of that something.

Two experiments backed this up. Suggestibility without hypnosis, motivation and expectancy together predicted 53% of the variance in behavioral response to hypnosis. The induction raised suggestibility a little on average, and for a substantial minority of participants it lowered it [5]. So when a study says “high hypnotizable,” a fair translation is often “high responder to imaginative suggestion, who expected to respond and was motivated to.”

That isn’t a debunking. Old me would have taken it as one. It’s a description of what the instrument does, and it matters for what you conclude from it.

Same score, different people

A total score also hides how someone got there. In one study, researchers split highly suggestible participants by their tendency to dissociate: 19 low-dissociative and 11 high-dissociative high scorers, compared with 21 low scorers. The high-dissociative group responded more to hallucination suggestions, reported more involuntariness and had weaker working memory capacity; the low-dissociative group showed better visual imagery and no observable deficits [6]. Two people with the same high score may be doing quite different things in their heads.

The samples are small, so I wouldn’t build a theory on them. But they’re a good reminder that “the highs” in a trial are not a single type.

What the score does not tell you about treatment

Here’s the gap I care most about, because it’s where the score gets oversold.

Test day versus treatmentWhat a suggestibility scale measures, and what it can't promise

On the scaleIn treatment

A count of standardized suggestions passed, given once

→

Whether hypnosis helps a specific condition

Responsiveness to imaginative suggestion, with or without an induction

→

How much the induction itself adds for this person

Top and bottom 10-15% of a sample labeled high and low

→

Most patients sit in the middle band

Strong predictor of lab pain relief

→

Weak, imprecise predictor of clinical outcomes, r = .24

In the laboratory, suggestibility predicts a lot. A meta-analysis of 85 experimental pain studies in 3,632 healthy volunteers found pain relief of 42% in high and 29% in medium suggestible participants, and minimal benefit in low scorers [7].

In clinics, much less. A meta-analysis of 10 clinical studies with 283 participants found a correlation of r = .24 between suggestibility and outcome, which works out to about 6% of the variance [8]. The 95% confidence interval ran from -0.28 to 0.75, so the true relationship could be anything from slightly negative to strong. Smaller studies reported larger effects, and the authors questioned whether assessing suggestibility in clinical settings is worth it. I read that confidence interval twice, hoping I’d misread it. I hadn’t.

And often, the clinical trials don’t measure suggestibility at all. A post hoc analysis of 448 people in a gut-directed hypnotherapy trial for IBS, one of the best evidence bases this field has, looked for predictors of who responded. The characteristics it examined were symptoms, demographics, quality of life and anxiety and depression scores [9]. Suggestibility wasn’t among them. The treatments with the strongest evidence mostly weren’t selected or tailored by a scale score.

A suggestibility score tells you	It does not tell you
How many standardized suggestions someone passed on test day	Whether hypnosis will help their condition
How responsive they are to imaginative suggestion, with or without an induction	How much the induction itself added
With newer scales, how real and involuntary responses felt	Which kind of high scorer they are
A stable, reliable individual difference	Anything about gullibility, willpower or character

Limitations of the evidence

Limitation	Consequence
Binary, single-trial scoring	Near misses and full misses look the same
Behavioral criteria only on older scales	Compliance can pass as experience
Scales track waking suggestibility closely	“Hypnotizability” may be a misleading label
Small subtype studies	Heterogeneity among high scorers is real but poorly mapped
Few clinical trials measure suggestibility	We don’t know who benefits from what
Wide clinical confidence interval	The clinical prediction could be large or near zero

What remains uncertain

What I’d like to see is boring and specific: large clinical trials that measure suggestibility with an experiential scale before treatment and report outcomes by score, with a credible control group. Until then, the lab result (strong prediction) and the clinic result (weak and imprecise) will keep sitting side by side, and people will keep quoting whichever one suits them.

What this means in practice

A suggestibility scale is a reliable instrument with a narrow job: counting responses to a standard set of suggestions, and on newer scales, rating how those responses felt. It is a good research tool for studying who responds to suggestion. It is a weak tool for deciding who should get hypnosis for a health problem, and it says nothing about anyone’s character. If a clinic offers to test you, it’s fine to be curious about the result. It’s not fine for anyone to tell you a low score means you can’t benefit, or that a high score guarantees you will.

For the stability and training evidence, see What Is Hypnotizability?. For why a control group matters when highs and lows are compared, see Expectation, Placebo and Hypnosis: Why Comparison Groups Matter.

Using research for a health decision. A suggestibility score is an assessment tool, not a diagnosis or a treatment recommendation. Whether hypnosis suits a health problem depends on the condition and the evidence for it. Discuss it with a qualified clinician, and see a doctor first about new, severe or worsening symptoms.