DISC test accuracy: what research says and why your style changes

Your DISC result at a company workshop said Steadiness and mind.io says Decision, or your top style changed when you took a test again a few weeks later. That can raise two questions: which one is really my style, and can a DISC test be trusted at all?

The short answer: “DISC” is not the name of one test but a framework that began with a theory of emotion published in 1928, and many different tests have been built on it (Marston, 1928; Wiley, 2026a, publisher’s own material). So accuracy has to be judged test by test. In material from Wiley, the publisher of DiSC, the scales of Everything DiSC correlated .85 to .88 when people took it twice two weeks apart (Wiley, 2026b, publisher’s own material). But independent studies outside the publisher are few and their results are mixed (Henkel & Wilmoth, 1992, from the abstract; Martinussen et al., 2001, from the abstract), and Wiley itself says DiSC results should not be used for decisions such as hiring or promotion (Wiley, 2026a, publisher’s own material). The mind.io DISC test is not the official DiSC but a 24-item test mind.io wrote itself, and its reliability and validity have not been studied.

This guide covers how DISC became a set of tests, the publisher’s reliability figures and the independent studies, the strengths and limits of picking one option out of four, how DISC scores overlap with the Big Five, why your mind.io result moves, and how to read it. How DISC differs from MBTI is covered in DISC versus MBTI, and how common each style is in DISC style shares.

In his 1928 book Emotions of Normal People, the psychologist William Moulton Marston placed the emotions on a circle and called the emotions at four points on it the “primary emotions”: Compliance, Dominance, Inducement and Submission (Marston, 1928). The four letters of DISC come from there, but even the words behind I, S and C were different from today’s. The book is a theory of emotion, and Wiley, which publishes DiSC, says Marston never created an instrument to measure his model (Wiley, n.d.-a, publisher’s own material).

According to Wiley’s account, the history of tests measuring DISC begins in the 1940s with the Activity Vector Analysis, which the industrial psychologist Walter Clarke built for personnel selection; it then passed through another self-description test made at Clarke’s firm, and in the 1970s Performax, the company John Geier founded, became the first publisher of a DiSC assessment. A version from an improvement effort launched in 1994 became today’s DiSC Classic, and in 2013 Everything DiSC moved to a format that asks more items depending on your answers. DiSC written with a lowercase i is a Wiley trademark (Wiley, n.d.-a, publisher’s own material). Wiley also calls the model itself DiSC, a behavioral model that many tests and workshops draw from, and sets Everything DiSC apart as its own product line built on it; it notes too that tools using the DiSC name are not all the same (Wiley, 2026a, publisher’s own material).

So the things called “DISC tests” differ in their items and their scoring. DiSC Classic presents 28 sets of four adjectives (in the US English version) and asks you to pick the one most like you and the one least like you in each (Inscape Publishing, 2005, publisher’s own material); Everything DiSC asks how much you agree with each statement on a five-point scale, adds items for a scale where your answers are inconsistent, and shows the result as a dot on a circle (Wiley, 2026b, publisher’s own material). The Korean short form from the Korea Educational Consulting Institute, used in a Korean study, asks you to pick, for each of 24 items, the one adjective of four that describes you best (Kim, 2018).

The mind.io DISC test is none of these. For each of 24 workplace situations you pick the one of four actions closest to yours; mind.io wrote the items itself on the DISC framework, and they are not Wiley’s items or a paid official assessment. So whether “the DISC test is accurate” depends on which test you mean, and the figures below describe those tests only: they do not show how accurate a mind.io result is.

A test’s accuracy is usually judged in two ways. Reliability asks whether the same person gets a similar result on a second try (retest reliability) and whether the items measuring one style move together (internal consistency); validity asks whether the test measures what it sets out to measure. Reliability is usually given as a number between 0 and 1, and the publisher’s report treats .70 or more as acceptable and .80 or more as good (Wiley, 2026b, publisher’s own material).

According to the publisher’s report on DiSC Classic, retest correlations were, one week apart (142 people, UK English version), .87 to .89; five to seven months apart (174 people, US English version), .73 to .84; and ten to fourteen months apart (138 people, US English version), .71 to .80. When it was revised in 1996, internal consistency in the US English version’s development sample of 812 people was .85 to .92, and in the development samples of the language versions it was, for the Korean version (1,009 people), .76 to .85 and, for the Japanese version (813 people), .84 to .90. The publisher itself notes that differences between results become more noticeable as the interval grows (Inscape Publishing, 2005, publisher’s own material).

For Everything DiSC, in 599 people who took it twice two weeks apart, the scales correlated .85 to .88, and in data from 50,646 people internal consistency was .79 to .89, with a median of .83. Among the people with the most inconsistent answers, the median was .62 with the base items alone and rose to .77 once the extra items were added; on average 24% of respondents received such extra items on a scale (Wiley, 2026b, publisher’s own material). The French version, in 384 working adults on training programs, had a median of .83 (Wiley, 2014, publisher’s own material), and the Japanese version, in 124,469 working adults, a median of .81 (Wiley, n.d.-d, publisher’s own material). We did not find figures for a Korean version of Everything DiSC.

All of these figures were produced by the publisher that sells the tests, not by independent, peer-reviewed studies. Studies from outside the publisher are few. Research on the Personal Profile System (PPS) as it was before 1994 was reviewed in a 1988 report, which concluded that several studies supported its criterion-related validity, but that there were few reliability data and that construct-validity studies gave inconsistent results (Henkel & Wilmoth, 1988, from the abstract). When the same researchers factor-analysed the answers of 1,045 senior non-commissioned Air Force officers, four factors emerged, but they concluded that the results did not completely justify the publisher’s claims (Henkel & Wilmoth, 1992, from the abstract).

Korean research is in a similar position. A study of nursing students reported the reliability of the Korean short form as a single figure, .84 (Kim, 2018), and a review of 91 Korean DISC papers concluded that, although the research had matured in quantity and quality, the accurate use of measurement tools was a task still to be done (Lee, 2019, from the abstract). Above all, every one of these figures describes its own test, so none of them tells you how accurate the mind.io test is.

Listing only desirable words and asking you to choose among them, as DiSC Classic does, is called forced choice. The publisher’s case for it is that, since every option is a positive word, it reduces answers given to look good (Inscape Publishing, 2005, publisher’s own material). Researchers also see such comparative answers as a way to reduce several response biases common in rating scales, where you rate each statement on its own (Brown & Maydeu-Olivares, 2013, from the abstract).

There is a cost, though. Once you pick one option you cannot pick the others, so when the picks are simply counted, everyone’s four scores add up to the same total. That makes it impossible for every style to come out high, or every style low, and it can distort the relations between the scales and the reliability estimates (Brown & Maydeu-Olivares, 2013, from the abstract). A 1988 paper warned about the misuse of personality tests of this kind in recruitment and selection in the United Kingdom: not that they have no use at all, but that the claims made for their reliability, their validity and their use for comparing people are misleading (Johnson, Wood & Blinkhorn, 1988, from the abstract). Another analysis saw substantial advantages in forced choice, which can control some of the response biases of rating scales, but found that its reliability results hold up only with many scales, around thirty, with low correlations between them, and that standard statistical analyses are inappropriate when there are few scales (Baron, 1996, from the abstract). A forced-choice DISC test has four. When the same job applicants answered in both formats, the relations between the scales and the choice of whom to select differed sharply (Meade, 2004, from the abstract).

One study looked at a DISC test directly. When 103 university students took both a forced-choice and a rating version of a DISC test (DISCUS), the two versions did not give the same results, and the researchers advised caution when judging validity with the forced-choice version (Martinussen et al., 2001, from the abstract). In the DiSC Classic data, Decision (D) and Steadiness (S) correlated −.82 and Influence (I) and Conscientious (C) −.71, moving strongly in opposite directions, which the publisher reads as support for its model (Inscape Publishing, 2005, publisher’s own material). But forced-choice scoring makes one scale fall when another rises, which can distort the relations between scales (Brown & Maydeu-Olivares, 2013, from the abstract), so these figures alone are hard to judge. On the other hand, in Everything DiSC, which is answered on a five-point scale, Influence and Conscientious also moved in opposite directions at −.70, and Decision and Steadiness at −.67, which the publisher takes as evidence for its circle structure (Wiley, 2026b, publisher’s own material).

Ways around the problem have been studied too. Scoring forced-choice answers with a statistical model that fits the way people choose (Thurstonian item response theory) can give scores free of these distortions (Brown & Maydeu-Olivares, 2013, from the abstract). The mind.io DISC test does not use such a model; it counts your picks. So mind.io’s four shares show the order of the styles you picked most often within yourself; they do not measure how much more of a Decision style you have than someone else.

In the publisher’s material, Everything DiSC scores were related to the NEO PI-R, a test of the five-factor (Big Five) model, in the directions the theory expects. In data from 694 people, Influence (i) correlated .69 with extraversion, Conscientious (C) −.63 with extraversion, Steadiness (S) .67 with agreeableness and Decision (D) −.58 with agreeableness (Wiley, 2026b; Inscape Publishing, 2012, publisher’s own material). These are correlations, so neither side causes the other, but in the publisher’s data they mean Everything DiSC scores line up with the traits used in personality research in the expected directions.

One point deserves attention. Conscientious (C) and Big Five conscientiousness, despite the similar names, correlated only .11, and the publisher explains that its C scale was built to measure a reserved, methodical, analytical disposition rather than a preference for order directly (Wiley, 2026b, publisher’s own material). So, on these data at least, a high Conscientious share should not be read as high Big Five conscientiousness. In an independent study too, the four dimensions of one DISC test (DISCUS) appeared as combinations of Big Five traits rather than as traits in their own right (Martinussen et al., 2001, from the abstract).

Can it be used for hiring? Wiley states plainly that DiSC results should not be used for hiring, screening, promotion, compensation or similar decisions, but to support development, communication and team agreements (Wiley, 2026a, publisher’s own material). It also explains that DiSC is not recommended for pre-employment screening because it does not measure the specific skills or aptitudes a position needs (Wiley, n.d.-b, publisher’s own material). The mind.io DISC test is not validated as a basis for hiring, selection or evaluation either.

That does not mean DISC results are unrelated to anything. In an observational study of 132 Australian physiotherapy students whose styles were measured with Everything DiSC, students with an Influence (i) style and with a Conscientious/Steadiness (CS) style had 3.96 and 4.34 times higher odds of failing a clinical placement than the whole sample, but among Bachelor’s students on their own, only students with Steadiness (S) and Conscientious (C) styles failed, and that association was not statistically significant either. An observational study cannot show causes, and the researchers proposed using the result to spot early the students who may need support, so that students and clinical educators can prepare together (Milne et al., 2019).

On the mind.io DISC test, each share is the share of your 24 answers that picked the style, so one answer is a twenty-fourth, about 4.2 points. Switching one answer to another style lowers one style by about 4.2 points and raises another by about 4.2, so the gap between those two styles changes by about 8.3 points at once (calculated from mind.io’s scoring). If your top two styles are one answer apart, say 9 answers (38%) and 8 (33%), a single different answer next time can swap them.

The lines where the reading changes are also one answer apart. The result page reads 40% or more as a clear tendency and under 20% as a style that rarely shows up; with 24 questions, 40% or more means 10 answers or more (42%), while 9 answers is 38%. Under 20% means 4 answers or fewer (17%), while 5 answers is 21%. So the same style can be read as a clear tendency one time and as a moderate one the next.

When your first and second styles were picked the same number of times, the result page notes that they got the same number of answers. Styles picked equally often are shown in the order Decision, Influence, Steadiness, Conscientious (D, I, S, C): pick each of the four styles 6 times and Decision appears as your primary style, with a “D/I” note beside it. When only your second and third styles are equal, no such note appears. And because the shares are rounded, the four can add up to anywhere from 99% to 102% (calculated from mind.io’s scoring).

The scene you picture changes the result too. The same person acts differently in different scenes, so thinking of yourself at work and thinking of yourself with friends can bring out different styles. Wiley, too, warns against reducing people to one style when they behave differently across contexts (Wiley, 2026a, publisher’s own material). You may also choose differently on a busy, tired day than on a relaxed one; the publisher’s report also notes that someone tired at one testing and well rested at the next, or in a different mood, may answer differently (Wiley, 2026b, publisher’s own material). The publisher’s tests are no different: DiSC Classic’s retest correlations were lower ten to fourteen months apart than one week apart, and the publisher explains that when results change, a profile’s level usually changes rather than its basic shape (Inscape Publishing, 2005, publisher’s own material).

On the mind.io test the questions always come in the same order, and although each question’s options are shuffled, they are shuffled the same way every time. If you retake it right after seeing your result, it is easy to pick from memory. One more thing: in October 2026 the options were shuffled and the label showing which style each option belonged to was hidden, the score changed to a share of your answers, and four options that read as another style were rewritten. If a result from before then differs from today’s, that can be part of the reason.

First, look at the four shares and how many answers separate your first and second styles, rather than at the name of the top style. The result page shows how many answers went to each style. If they are one or two answers apart, keep both styles as candidates and check which style stays near the top across your two results.

Second, answer with one scene in mind. Choose a scene you want to understand, such as this quarter’s meetings, and pick what you actually did there; the result is more useful that way. Rather than retaking the test straight away because you dislike the result, take it again a few weeks later, on an ordinary day, with the same scene in mind.

Third, use the result as a hypothesis to check rather than an answer. Wiley also advises treating results as hypotheses to test against real behavior and context, and talking about “what helps me work well with you” rather than “what you are” (Wiley, 2026a, publisher’s own material). It lists pigeonholing colleagues by style as something to avoid (Wiley, n.d.-c, publisher’s own material).

Fourth, do not compare share numbers with a colleague. mind.io’s shares show the order of the four styles within you, so a higher Decision share than a colleague’s does not mean you are the more decisive person. It is better to use your results to talk about how each of you acts in which scenes.

Fifth, cross-check with a test that works differently. The Big Five test asks how true each statement is of you instead of making you pick one, and shows your traits as scores. How to read DISC alongside an MBTI result is covered in DISC versus MBTI, and why a description that would fit anyone can feel like it is about you is explained in the Barnum effect guide.

Q. Is DISC a scientifically validated test? A. DISC is not one test but a framework many tests share, so it depends on the test. In the publisher’s material, Everything DiSC’s two-week retest correlations were high, at .85 to .88 (Wiley, 2026b, publisher’s own material), but independent studies outside the publisher are few, and one showed the limits of the forced-choice version (Martinussen et al., 2001, from the abstract). It is fairer to place it somewhere between “fully validated” and “no evidence at all”, and the mind.io test has no studies of its own.

Q. My company wants me to take a DISC test. Is it used for hiring or evaluation? A. Wiley, which publishes the test, says DiSC results should not be used for hiring, screening, promotion or compensation decisions (Wiley, 2026a, publisher’s own material). The use the publisher recommends is talking, in training or team workshops, about the ways each of you prefers to communicate. If you are the one hiring, keep in mind the publisher’s own point that DiSC does not measure the specific skills or aptitudes a position needs (Wiley, n.d.-b, publisher’s own material).

Q. My result at work differs from my mind.io result. Which one is right? A. They are different tests, with different items and scoring. DiSC Classic asks for the most and least like you in sets of adjectives, Everything DiSC asks for a five-point answer to each statement, and mind.io asks you to pick one option in each of 24 workplace situations. The scene you pictured while answering may also have differed. Rather than calling one of them right, look at the styles that come near the top in both.

Q. Are DiSC and DISC different things? A. DiSC with a lowercase i is a Wiley trademark and refers to the tests Wiley publishes, such as DiSC Classic and Everything DiSC (Wiley, n.d.-a, publisher’s own material). DISC is closer to the name of the model these tests, and other publishers’ tests, are built on, though Wiley itself also writes DiSC for the general model (Wiley, 2026a, publisher’s own material). The mind.io DISC test is not a Wiley DiSC assessment.

Q. My four shares do not add up to 100%, and my first and second styles have the same share. A. The shares are rounded, so they can add up to anywhere from 99% to 102%. When your first and second styles were picked the same number of times, the result page notes that they got the same number of answers, and styles picked equally often are shown in the D, I, S, C order. If your first two shares are equal, treat both styles as your main styles.

Q. If my result changes, has my personality changed? A. Not necessarily. On mind.io one answer is about 4.2 points, so a few different answers can change the order, and the scene you pictured and how you felt that day matter too. The comparison with your previous result on the result page also notes that a difference does not mean your personality changed; it can reflect your recent state, the context or how you framed your answers.

Q. Which is more accurate, DISC or MBTI? A. Both group results into a few styles or types, so in both a result near a boundary can change. Why MBTI results change is covered in the MBTI test accuracy guide. If you want to see traits as scores rather than types, the Big Five test is the better fit.

References: Baron, H. (1996), Strengths and limitations of ipsative measurement, Journal of Occupational and Organizational Psychology 69(1), 49-56, https://doi.org/10.1111/j.2044-8325.1996.tb00599.x; Brown, A., & Maydeu-Olivares, A. (2013), How IRT can solve problems of ipsative data in forced-choice questionnaires, Psychological Methods 18(1), 36-52, https://doi.org/10.1037/a0030641; Henkel, T. G., & Wilmoth, J. N. (1988), Theoretical considerations for extracting meaning from Personal Profile System data: the need for independent construct validity studies, ERIC ED324344, https://eric.ed.gov/?id=ED324344; Henkel, T. G., & Wilmoth, J. N. (1992), Factor analysis of the Personal Profile System, The Journal of Experimental Education 60(3), 271-280, https://doi.org/10.1080/00220973.1992.9943880; Inscape Publishing (2005), DiSC validation research report (DiSC Classic), https://www.everythingdisc.co.uk/EverythingDiSC/media/SiteFiles/Help-Classic/MS2010/DiSCClassicValidationResearchReport.pdf; Inscape Publishing (2012), Everything DiSC research report for adaptive testing assessment (2007-2012 edition), https://www.itda.com/Wiley/Everything_DiSC/Everything_DiSC_Research_Report.pdf; Johnson, C. E., Wood, R., & Blinkhorn, S. F. (1988), Spuriouser and spuriouser: the use of ipsative personality tests, Journal of Occupational Psychology 61(2), 153-162, https://doi.org/10.1111/j.2044-8325.1988.tb00279.x; Kim, H. R. (2018), The effect of DISC behavioral style on nursing students’ knowledge and clinical performance [in Korean], Journal of the Korea Academia-Industrial cooperation Society 19(11), 60-67, https://doi.org/10.5762/KAIS.2018.19.11.60; Lee, Y. J. (2019), DISC behavior style research in Korea: trends and challenges [in Korean], Journal of Learner-Centered Curriculum and Instruction 19(12), 277-294, https://doi.org/10.22251/jlcci.2019.19.12.277; Marston, W. M. (1928), Emotions of normal people, London: Kegan Paul, Trench, Trubner, https://archive.org/details/emotionsofnormal032195mbp; Martinussen, M., Richardsen, A. M., & Vårum, H. W. (2001), Validation of an ipsative personality measure (DISCUS), Scandinavian Journal of Psychology 42(5), 411-416, https://doi.org/10.1111/1467-9450.00253; Meade, A. W. (2004), Psychometric problems and issues involved with creating and using ipsative measures for selection, Journal of Occupational and Organizational Psychology 77(4), 531-551, https://doi.org/10.1348/0963179042596504; Milne, N., Louwen, C., Reidlinger, D., Bishop, J., Dalton, M., & Crane, L. (2019), Physiotherapy students’ DiSC behaviour styles can be used to predict the likelihood of success in clinical placements, BMC Medical Education 19, 379, https://doi.org/10.1186/s12909-019-1825-2; Wiley (n.d.-a), History of DiSC, DiSC Profile (Personality Profile Solutions, a Wiley company), https://www.discprofile.com/what-is-disc/history-of-disc; Wiley (n.d.-b), Is DiSC a good tool for hiring?, DiSC Profile (Personality Profile Solutions, a Wiley company), https://www.discprofile.com/everything-disc/hiring; Wiley (n.d.-c), The do’s and don’ts of DiSC, https://www.everythingdisc.com/dos-and-donts/; Wiley (n.d.-d), Japanese validation addendum to the Everything DiSC research report for the Everything DiSC Workplace assessment, https://www.everythingdisc.com/EverythingDiSC/media/SiteFiles/Facilitator-Resources/Workplace/Japanese-Validation-Addendum.pdf; Wiley (2014), French validation addendum to the Everything DiSC research report for the Everything DiSC Workplace assessment, https://www.everythingdisc.com/EverythingDiSC/media/SiteFiles/Facilitator-Resources/Workplace/French_07-Workplace-Validation-Addendum.pdf; Wiley (2026a), Everything DiSC vs DiSC: what’s the difference? (updated 4 August 2026), https://llms.everythingdisc.com/everything-disc-vs-disc; Wiley (2026b), Everything DiSC research report for adaptive testing assessment, https://www.everythingdisc.co.uk/EverythingDisc/media/SiteFiles/Assets/LeadGenAssets/Everything-DiSC-Research-Report.pdf