HSP test accuracy: what research says and why your score changes

You scored 67, “Highly sensitive (HSP traits)”, last time and 63, “Sensitive in some situations”, today; your result after a busy month differed from the one after a holiday; or one site said “you are an HSP” and another said you are not. Stories like these are common, and they naturally raise two questions: “Can I trust a sensitivity test?” and “Am I an HSP or not?”

The short answer is that a changing result is not unusual, and there are reasons for it. The research scale for highly sensitive people has items that hang together well, and scores retaken a few months later come out fairly similar, yet the researchers who build these scales also say a self-test result should be read as a rough estimate. Sensitivity is not something you either have or lack: it runs from low to high, and “HSP” names one side of a line drawn across it. Near that line, a small difference changes the name. And because most items ask about feeling drained or overwhelmed, recent tiredness or stress can creep into the score.

This guide covers how researchers have checked the accuracy of the scale, why results differ from one scale to another, how to think about “HSP or not”, why your band can change on mind.io, and how to read a result that changed. What highly sensitive people are like is covered in the guide to the signs of a highly sensitive person, and how common it is in the guide to HSP percentages. Why MBTI, Enneagram and attachment results change is covered in the guide to MBTI accuracy, the guide to Enneagram accuracy and the guide to attachment test accuracy.

When people talk about how accurate a psychological test is, they usually look at two things. Reliability asks whether the same person gets a similar result when tested again (test-retest reliability) and whether the items move together consistently (internal consistency). Validity asks whether the test measures what it sets out to measure, and whether its results connect meaningfully with other tests or with real behaviour.

The Highly Sensitive Person Scale was built by the psychologists Elaine and Arthur Aron. They wrote 60 items from in-depth interviews with 39 adults who saw themselves as highly sensitive, introverted or easily overwhelmed by stimulation, then tested them on 604 psychology undergraduates and 301 people from the community and refined them into 27 items (Aron & Aron, 1997, as cited in Greven et al., 2019). Across seven studies, the scale was reported to show adequate reliability and validity (Aron & Aron, 1997, from the abstract).

Later samples also found the items hang together well. Internal consistency was .89 in 901 US university students (Lionetti et al., 2018), .904 in 1,773 Korean adults (Yang & Kwon, 2024) and above .84 in 750 South African university students (May et al., 2022). The French version reached .90 in 814 adults (Bordarie et al., 2022, from the abstract), and the Korean K-HSPS-18, which keeps 18 of the 27 items, .85 (Son & Kim, 2021). The Japanese HSPS-J19, cut to 19 items, reached .78 overall, with a low .57 for its aesthetic sensitivity group (Takahashi, 2016). High internal consistency means the items point the same way; it does not mean you will get the same score next time.

Few studies have measured the same people twice. In the K-HSPS-18 study of 925 Korean adults, just over half of them university students, 164 answered again about two months later, and the two scores correlated .78, with .66 to .76 for the separate groups of items (Son & Kim, 2021). A study that retested 173 of 1,222 Colombian adults, more than half of them aged 18 to 24, found .76 after six months (Rocha-Nieto et al., 2025). The French version was also reported to have high retest agreement, .889, but the abstract does not give the interval or how many people took part (Bordarie et al., 2022, from the abstract). We found no study that retested the same adults over several years. Researchers list how stable sensitivity is across development, and which experiences might change it, as questions still to be answered (Greven et al., 2019), and the authors of the newly revised scale write that its test-retest reliability has yet to be checked (Pluess et al., 2024, preprint).

The researchers who build these scales say both things. Their research site presents short tests that measure sensitivity easily and reliably, yet adds that the outcome of a self-report test is not very precise and should only be read as an estimate. It notes that other factors such as age, gender and nationality also shape the result, especially when a score is close to another sensitivity group (Sensitivity Research, n.d.). The HSP-R self-test sheet on Aron’s site likewise says the score is not a precise indication or a diagnosis but shows your likely position along a continuum (Aron, n.d.).

The mind.io sensitivity (HSP) test is none of these. It does not reuse Aron’s items; it is mind.io’s own check of 18 statements written by mind.io along the three groups often used in research (6 per group), answered from 1 to 5, with no reverse-scored items. The test’s retest reliability and validity have not been studied. So the figures above describe research scales; none of them tells you how accurate a mind.io result is.

There is more than one HSP scale. The original 27 items are answered from 1 to 7, and the research group reported that a 12-item short form for adults (HSP-12) measured sensitivity reliably and validly across three studies of 910 people (Pluess et al., 2023, from the abstract). In 2024 the group released an 18-item revised scale with six dimensions (HSP-R), based on 1,000 adults in the UK and the US; the version we read is a preprint, released before peer review (Pluess et al., 2024, preprint). The self-tests now on the researchers’ site and on Aron’s site use this revised scale (Sensitivity Research, n.d.; Aron, n.d.). Translations differ in length too: the Japanese HSPS-J19 keeps 19 of the 27 items (Takahashi, 2016) and the Korean K-HSPS-18 keeps 18 (Son & Kim, 2021).

How the items group together also differs from study to study. The Arons saw one core trait (Aron & Aron, 1997, from the abstract); three groups, ease of excitation, low sensory threshold and aesthetic sensitivity, later came into wide use (Smolewska et al., 2006, as cited in Greven et al., 2019), but the critical review points out that they were not designed in advance: they emerged from analysis (Greven et al., 2019). In US student data, a structure with one general sensitivity factor alongside the three groups fitted better than the others (Lionetti et al., 2018); the German version kept three groups after dropping 13 items (Konrad & Herzberg, 2019, from the abstract), the French version found four (Bordarie et al., 2022, from the abstract), the Korean version found four including one the authors called nunchi, a Korean word for reading the room (Son & Kim, 2021), and South African student data gave five (May et al., 2022). The researcher who built the Japanese version wrote that the debate over the number of groups has not been settled (Takahashi, 2016).

Wording can sway the score too. The team behind the revised scale wrote that the original ease-of-excitation and low-sensory-threshold items are almost all negatively worded, which may tilt the total toward the hard side of sensitivity and may partly explain its steady, moderate correlation with neuroticism (Pluess et al., 2024, preprint). The original also has no reverse-scored items, so it cannot filter out a habit of agreeing whatever the statement says (May et al., 2022). That does not mean the original measured the wrong thing: the same team found the new scale correlates highly with the original and took this to mean the original captured sensitivity despite some shortcomings (Pluess et al., 2024, preprint).

Answering habits differ between cultures as well. The authors of the study of 1,773 Korean adults suggested that East Asian respondents’ tendency to pick the middle of a 7-point scale rather than its ends may be why the medium group came out larger, while adding that there is not yet enough data to put the difference down to answering style alone (Yang & Kwon, 2024). The cut-offs used to call someone highly sensitive also vary: 3.71 and 4.66 in US students (Lionetti et al., 2018), 3.81 and 4.73 in Korean adults (Yang & Kwon, 2024) and 4.31 and 5.25 on the K-HSPS-18 (Son & Kim, 2021). The shares found in each study are set out in the guide to HSP percentages.

mind.io’s three areas, “Easily overwhelmed”, “Depth of processing” and “Sensory sensitivity”, are plain-language names for these three groups, but mind.io wrote its own items. So the research groupings and cut-offs above do not map directly onto mind.io scores. Why the “Depth of processing” score deserves a light reading is explained in the guide to the signs of a highly sensitive person.

The evidence increasingly treats sensitivity as a trait that, like height, runs from low to high. In US student data the scores came close to a normal distribution, and the three groups differed in how much of the same trait they had rather than in kind (Lionetti et al., 2018). Even the long-quoted figure of 20% was first proposed by analogy with research on infant temperament and reactivity, rather than as a line found by measurement (Greven et al., 2019).

So cut-offs should be read as approximations. The researchers who found the three groups in adults wrote that, because no self-report scale is perfectly reliable, their cut-offs are mere approximations and may well differ between populations (Lionetti et al., 2018). A recent review suggests that the dandelion, tulip and orchid metaphor risks oversimplifying a normally distributed trait and may unintentionally suggest better and worse, and that population norm scores showing a person’s relative position are more useful than cut-offs (Greven et al., 2026, authors’ manuscript). When the 2019 critical review was written, though, no population norms for sensitivity existed yet (Greven et al., 2019).

Sensitivity also overlaps with other traits. In a meta-analysis, adult sensitivity correlated r = .40 with neuroticism and r = .14 with openness, and was unrelated to extraversion (Lionetti et al., 2019, from the abstract). Even so, its links with existing temperament and personality traits stay small to moderate when they are considered together, and the critical review concluded there is reasonably good evidence that sensitivity is a distinct construct (Greven et al., 2019). How it differs from introversion is covered in the guide to HSP vs introvert.

HSP is not a diagnosis. Researchers treat sensory processing sensitivity as a temperament trait, not a disorder, and point out that it is too common to be one, with nearly a third of people in the high range (Greven et al., 2019). Small to moderate links with anxiety and depression are reported again and again, but many of these studies rely on self-reports at a single point in time (Greven et al., 2026, authors’ manuscript), so they do not show that sensitivity is the cause. In the Korean study too, the total score correlated .30 with anxiety and .21 with depression, while the aesthetic sensitivity group went the other way (Son & Kim, 2021). Research suggesting the same temperament raises the risk of stress-related problems in poor environments but gains more from supportive ones (Greven et al., 2019) is covered in the guide to the signs of a highly sensitive person.

Read mind.io’s three bands the same way. 0–39 “Takes things in calmly”, 40–64 “Sensitive in some situations” and 65–100 “Highly sensitive (HSP traits)” are reading aids, not cut-offs set in research or clinical practice. A score of 65 does not mean “confirmed HSP”, nor 64 “not an HSP”: each band is a short name for a position on a continuous score.

The mind.io sensitivity (HSP) test turns the average of its 18 answers into a score from 0 to 100. So, with all 18 answered, moving one answer by one step moves the total by about 1.4 points (1 or 2 after rounding), and the score of that area, which has only 6 items, by about 4 points. Someone at 64 who moves a single answer one step toward agreeing next time can reach 65 and become “Highly sensitive (HSP traits)”; the same goes for 39 and 40.

The result page shows your total, your band and the three area scores, but it shows no separate note when a score sits near a band’s edge. So rather than reading the band name alone, check for yourself how far your score is from 40 or 65. The “Most pronounced” label also appears only when one area scores higher than both others, so if two areas are one step (about 4 points) apart, their order may swap next time.

All mind.io statements are worded in the same direction, with no reverse-scored items. As the result page itself points out, leaning toward “agree” on most of them raises the total. Answering 3, the neutral point, to every statement gives a total of 50, “Sensitive in some situations”.

The same person’s state at the time of answering can also move the score. As the test’s own notes say, in a week of short sleep or heavy stress the “Easily overwhelmed” score can come out higher. For the research scale too, reviewers point out that, apart from the aesthetic items, most items are negatively worded and may not capture the experience of highly sensitive people who are getting on fine (Greven et al., 2019). Whether you pictured yourself at work or at home, as you usually are or as you have been lately, also changes the answers you pick.

One more thing: in October 2026, mind.io replaced nine of the sensitivity test’s statements. They asked about several situations at once, asked you to compare yourself with others, or were statements almost anyone would agree with, and five of them were “Sensory sensitivity” items. Scoring and bands stayed the same and saved results kept their scores, but if a result from before then differs from today’s, the new items may be part of the reason. When you compare two such results, the previous-result comparison on the result page notes that some items changed.

First, look at your score and its distance from the line, not the band name. If two results are 63 and 67, they sit close together even though the names differ. If they are far apart, say in the 30s and the 70s, first think back to how you were and what you pictured while answering.

Second, look at the order of the three areas. If “Sensory sensitivity” came out highest both times, that tells you more than a small rise or fall in the total. Where to start in each area is set out in the guide to coping as a highly sensitive person.

Third, retake it later under similar conditions. Retaking it several times straight away because you dislike the result makes it easy to remember the items and pick the answers you want. Answer as you usually are rather than as you feel today, jot down how you slept and how busy you were, and retake it a few weeks later at a similar time. If your earlier result is still in the same browser, the previous-result comparison on the result page shows how many points your total and each area changed, and adds that sensitivity is a fairly stable temperament trait, so small changes usually reflect your current state and surroundings.

Fourth, check from another angle. Setting your extraversion and emotional sensitivity scores from the Big Five test next to the three area scores of the sensitivity test shows more clearly where your sensitivity overlaps with other traits and where it does not. How sensitivity differs from introversion is covered in the guide to HSP vs introvert.

The mind.io sensitivity test is a check for self-understanding, not a validated clinical tool. It cannot pick out conditions such as anxiety disorders, ADHD or sensory processing difficulties. If sound, light or touch make daily life hard, or anxiety or a low mood has lasted more than two weeks, don’t put it all down to sensitivity: look at how you are with the anxiety check-in (GAD-7) or the depression check-in (PHQ-9), then talk to a professional. These check-ins are not diagnoses either. Where to turn when you feel worn out and overwhelmed is set out at the end of the guide to coping as a highly sensitive person.

Don’t use it to label people. Asking a child, partner or colleague to take the test so you can call them an HSP, or using it in hiring or appraisals, is not what it is for. Explaining every difficulty with “I’m just sensitive” also makes it easy to miss what you can change, such as sleep, stress or a relationship.

Keep the limits of the research in mind too. A recent review notes that sensitivity research has grown fast but still often relies on single-time-point studies and self-report, on unrepresentative or convenience samples, and on an over-representation of educated women (Greven et al., 2026, authors’ manuscript). Gaps in the evidence do not mean there is no evidence, but they do not mean the question is settled either. Read your result only as far as the research can take it.

Q. Is an HSP test scientifically reliable? A. The 27-item scale (including translations) shows internal consistency of about .89 to .904 (Lionetti et al., 2018; Yang & Kwon, 2024), and the shortened versions .78 for the Japanese HSPS-J19 and .85 for the Korean K-HSPS-18 (Takahashi, 2016; Son & Kim, 2021). The Korean version correlated .78 when retaken about two months later (Son & Kim, 2021). But the researchers who build these scales also say a self-test result should be read as a rough estimate. The mind.io test uses its own items and has no reliability data yet, so read your area scores rather than the band name.

Q. Every site tells me something different about being an HSP. Which one is right? A. Scales differ in their number of items, their answer options and the cut-off they use for “HSP”, so the same person can get different results. Even in the US student study, a separate sample that answered on a 5-point scale showed the three groups less clearly (Lionetti et al., 2018). Rather than any one site’s verdict, look at where your score keeps landing and which areas keep standing out.

Q. Does a score of 65 or more mean I am an HSP? A. mind.io’s 65 is simply the edge of a band drawn to make results easier to read; it is neither a research cut-off nor a top percentage. 64 and 65 can be one answer apart. Shares and cut-offs from the studies are in the guide to HSP percentages.

Q. Is being an HSP an illness? Can it be diagnosed? A. No. Researchers treat sensory processing sensitivity as a temperament trait, not a disorder (Greven et al., 2019). But if feeling drained is wearing down your daily life, look at that difficulty itself with a professional, whatever your sensitivity.

Q. Does sensitivity stay the same for life? A. Scores retaken a few months apart were fairly similar (Son & Kim, 2021; Rocha-Nieto et al., 2025). But we found no study that retested the same adults over several years, and researchers still see this as an open question (Greven et al., 2019). The twin study on how much is inherited is covered in the guide to the signs of a highly sensitive person.

Q. Is sensitivity the same as neuroticism (emotional sensitivity)? A. They overlap but are not the same. In a meta-analysis of adults the correlation was r = .40 (Lionetti et al., 2019, from the abstract), and the team behind the revised scale suggests the original scale’s negative wording may partly explain that correlation (Pluess et al., 2024, preprint). Looking at it next to the Big Five trait of emotional sensitivity makes the difference clearer.

Q. Do MBTI, Enneagram and attachment results change for the same reason? A. In one respect, yes: near a dividing line, a small difference changes the name. More is in the guide to MBTI accuracy, the guide to Enneagram accuracy and the guide to attachment test accuracy.

References: Aron, E. N., & Aron, A. (1997), Sensory-processing sensitivity and its relation to introversion and emotionality, Journal of Personality and Social Psychology 73(2), 345-368, https://doi.org/10.1037/0022-3514.73.2.345; Greven, C. U., Lionetti, F., Booth, C., Aron, E. N., Fox, E., Schendan, H. E., Pluess, M., Bruining, H., Acevedo, B., Bijttebier, P., & Homberg, J. (2019), Sensory Processing Sensitivity in the context of Environmental Sensitivity: A critical review and development of research agenda, Neuroscience & Biobehavioral Reviews 98, 287-305, https://doi.org/10.1016/j.neubiorev.2019.01.009; Lionetti, F., Aron, A., Aron, E. N., Burns, G. L., Jagiellowicz, J., & Pluess, M. (2018), Dandelions, tulips and orchids: Evidence for the existence of low-sensitive, medium-sensitive and high-sensitive individuals, Translational Psychiatry 8, 24, https://doi.org/10.1038/s41398-017-0090-6; Yang, W., & Kwon, M. (2024), Cut-off scores for sensitivity interpretation of the Korean Highly Sensitive Person Scale, PLoS ONE 19(9), e0309904, https://doi.org/10.1371/journal.pone.0309904; May, A. K., Norris, S. A., Richter, L. M., & Pitman, M. M. (2022), A psychometric evaluation of the Highly Sensitive Person Scale in ethnically and culturally heterogeneous South African samples, Current Psychology 41(7), 4760-4774 (published online 2020), https://doi.org/10.1007/s12144-020-00988-7; Bordarie, J., Aguerre, C., & Bolteau, L. (2022), Validation and study of psychometric properties of a French version of the Highly Sensitive Person Scale (HSPS-FR), European Review of Applied Psychology 72(4), 100781, https://doi.org/10.1016/j.erap.2022.100781; Son, O.-S., & Kim, J. (2021), Revalidation of the Korean version of the Highly Sensitive Person Scale (K-HSPS-18) [in Korean], The Korean Journal of Counseling and Psychotherapy 33(3), 1049-1075, https://doi.org/10.23844/kjcp.2021.08.33.3.1049; Takahashi, A. (2016), Development of Japanese version of the 19-item Highly Sensitive Person Scale (HSPS-J19) [in Japanese], Japanese Journal of Research on Emotions 23(2), 68-77, https://doi.org/10.4092/jsre.23.2_68; Rocha-Nieto, L. M., González Gutiérrez, J. L., & López-López, A. (2025), Highly Sensitive Person Scale: Adaptation and validation in the Colombian population, Behavioral Psychology/Psicología Conductual 33(2), https://doi.org/10.31083/bp45946; Pluess, M., Aron, E. N., Kähkönen, J. E., Lionetti, F., Huang, Y., Tillmann, T., Greven, C. U., & Aron, A. (2024), Evolution of the concept of sensitivity and its measurement: The Highly Sensitive Person Scale-Revised, PsyArXiv preprint (not peer reviewed), https://doi.org/10.31234/osf.io/w7bqu; Sensitivity Research (n.d.), Sensitivity tests, https://sensitivityresearch.com/self-tests/ (accessed October 2026); Aron, E. N. (n.d.), Are you highly sensitive? Self-test, https://hsperson.com/test/highly-sensitive-test/, with the printable HSP-R self-test (© 2024 Pluess, Aron, Kähkönen, Lionetti, Huang, Tillmann, Greven & Aron), https://hsperson.com/wp-content/uploads/2025/02/New-Self-Test-PDF.pdf (accessed October 2026); Pluess, M., Lionetti, F., Aron, E. N., & Aron, A. (2023), People differ in their sensitivity to the environment: An integrated theory, measurement and empirical evidence, Journal of Research in Personality 104, 104377, https://doi.org/10.1016/j.jrp.2023.104377; Smolewska, K. A., McCabe, S. B., & Woody, E. Z. (2006), A psychometric evaluation of the Highly Sensitive Person Scale: The components of sensory-processing sensitivity and their relation to the BIS/BAS and “Big Five”, Personality and Individual Differences 40(6), 1269-1279, https://doi.org/10.1016/j.paid.2005.09.022 (cited in Greven et al., 2019); Konrad, S., & Herzberg, P. Y. (2019), Psychometric properties and validation of a German High Sensitive Person Scale (HSPS-G), European Journal of Psychological Assessment 35(3), 364-378, https://doi.org/10.1027/1015-5759/a000411; Greven, C. U., Trupp, M. D., Homberg, J. R., & Slagter, H. A. (2026), Sensory processing sensitivity: Theory, evidence, and directions, Trends in Cognitive Sciences 30(6), 530-545 (published online November 2025; authors’ manuscript https://doi.org/10.31234/osf.io/wkdhg_v3), https://doi.org/10.1016/j.tics.2025.10.007; Lionetti, F., Pastore, M., Moscardino, U., Nocentini, A., Pluess, K., & Pluess, M. (2019), Sensory Processing Sensitivity and its association with personality traits and affect: A meta-analysis, Journal of Research in Personality 81, 138-152, https://doi.org/10.1016/j.jrp.2019.05.013