Highly Sensitive Person Scale: 27 questions about how much you notice
Two people sit through the same meeting. One leaves with a headache and needs an hour alone. The other is fine. Aron and Aron built the Highly Sensitive Person Scale in 1997 to measure that difference, and they treated it as a temperament trait rather than a flaw or a disorder.
The scale came out of seven studies. Study 1 was a set of qualitative interviews with 39 people. Studies 2 through 4 tested larger samples, about 905 people in total. The final 27-item version was tested in Study 6 and cross-validated in Study 7.
One factor, three facets
Every study pointed to a single factor rather than separate traits. In Study 6 (n = 172) the first unrotated factor carried 54% of the common variance, and in Study 7 (n = 109) it carried 47%. Item loadings in Study 6 ran from .24 to .64. The authors tried two-factor rotations and could not give them a stable interpretation.
The items still group into three facets by content, and those are what a report shows:
| Facet | Items | What it covers |
|---|---|---|
| Ease of Excitation | 15 | Being overwhelmed by stimulation, time pressure, or too many things at once |
| Aesthetic Sensitivity | 6 | Noticing and being moved by art, music, and small changes in the environment |
| Low Sensory Threshold | 6 | Discomfort from loud noise, bright light, coarse fabric, strong smells |
Answers run 1 to 7, from Not at all to Extremely. Nothing is reverse-scored, so the total is a straight sum from 27 to 189, and a higher number means more sensitivity. The whole thing takes about 8 minutes.
Bands, and why to hold them loosely
The paper never set a cutoff. Aron and Aron described the top quartile of a sample as highly sensitive, which worked out to roughly 15% to 20% of the people they tested.
Later studies using latent class analysis produced item-mean cutoffs instead of total-score cutoffs, and they disagree with each other:
| Study | Sample | Low | Middle | High | Item-mean cutoffs |
|---|---|---|---|---|---|
| Lionetti et al. 2018 | UK university students | 30.5% | 40.3% | 29.2% | 3.14 / 4.14 / 5.02 |
| Yang & Kwon 2024 | Korean adults aged 20 to 80 | 22.0% | 45.3% | 32.7% | 3.81 / 4.73 |
| Bürger et al. 2024 | German adults (n = 1628) | 10% | 40% | 38% plus 12% | four groups fit better |
The Korean high-sensitivity group is larger than the Western estimate, which may say more about culture than about the trait. Bürger and colleagues found four groups fit better than three, though three remains the common approach.
Totals on this page are read against four descriptive ranges:
| Total | Band |
|---|---|
| 27-82 | Low |
| 83-110 | Moderate |
| 111-138 | High |
| 139-189 | Highly sensitive |
Average totals from the two validation samples land in the moderate to high range: 115.6 (SD 21.9) in Study 6 and 118.3 (SD 20.0) in Study 7. That is what an ordinary sample looks like, not a threshold.
Reliability
Internal consistency for the 27-item version was .87 in Study 6 and .85 in Study 7. Earlier versions ran lower: .64 to .68 across three classes in Study 2, .75 in Study 3, and .64 in a telephone survey in Study 4. The 19-item version that preceded the final one came in at .87, or .84 after removing three items that overlapped with another scale.
The original paper reported no test-retest data, so there is no stability estimate from the authors. Cross-cultural work has since reported alphas between 0.85 and 0.89, including .89 in a Mexican student sample of 1050 people.
What it relates to
Sensitivity overlaps with introversion and neuroticism without being either one. Correlations in the original studies were .64 with a low-stimulus-screening scale, .29 to .52 with an introversion measure, .41 with Big Five neuroticism and .65 with a self-report emotionality scale. Correlations with MBTI introversion (.14) and Big Five extraversion (.12) were not significant. Taken together, all five Big Five factors shared only .54 of the variance with the scale, leaving about 71% unaccounted for.
A 2024 German sample of 1628 adults reported neuroticism r = 0.41, extraversion r = -0.30, openness r = 0.26, agreeableness r = 0.19 and conscientiousness r = -0.01.
Sensitivity also tracks with distress. A 2024 review of six studies in university students found correlations with anxiety from 0.381 to 0.57 and with depression from 0.22 to 0.295. Those numbers cut both ways: the same trait that makes someone easier to overwhelm also means they register more of what is going on around them.
Two kinds of highly sensitive people
Cluster analysis in the original studies split the highly sensitive group in two. The smaller cluster, about a third of them, had unhappier childhoods and scored higher on social introversion and emotionality. The larger cluster, about two thirds, looked like everyone else apart from the sensitivity itself. Being sensitive is not the same as being fragile, and the data shows both patterns.
What this test is not
There is no clinical cutoff and the scale was not built to diagnose anything. No Chinese validation study exists yet, so the translation used here has not been checked against a Chinese norm sample. If a high score comes with anxiety, sleep problems, or trouble functioning at work, that is worth raising with a professional. This page is not a diagnosis and does not replace one.