Enneagram accuracy: what research says and why your type changes

You were a Type 4 last time and a Type 5 today, every site gave you a different number, or your wing moved from 4w3 to 4w5. Stories like these are common, and they naturally raise two questions: “Is the Enneagram test even accurate?” and “What is my real type?”

The short answer is that a changing result is not unusual, and there are reasons for it. On the research side, a systematic review of 104 samples found mixed evidence for the Enneagram’s reliability and validity (Hook et al., 2021, from the abstract). There is also not one Enneagram test but many, each with its own items and scoring. And when your top two types score close together, a few different answers are enough to swap them.

This guide covers how researchers have judged the Enneagram’s accuracy, why different tests give different results, the research on whether people really fall into nine types, why close scores swap, and how to read a result that changed. It is less about handing you the right number than about reading your result with less confusion and more use. Why MBTI results change is covered in the guide to MBTI accuracy.

Accuracy in psychological testing is usually judged on two counts. Reliability asks whether the same person gets a similar result when tested again (test–retest reliability) and whether the items for one type move together (internal consistency). Validity asks whether the test measures what it sets out to measure, and whether its results connect meaningfully with other tests and with how people actually are.

The Enneagram did not start out as a measuring instrument. Gurdjieff (1872–1949) brought the nine-pointed figure to the West but did not describe nine personality types; Oscar Ichazo began attaching personality types to the figure, and the psychiatrist Claudio Naranjo developed this into a theory of personality. From the early 1980s, people who had learned from Naranjo began teaching the Enneagram to others (Sutton, 2007). Some accounts trace it to older traditions such as Sufism, but accounts of its origins differ, and Ichazo himself said his teaching did not come from the Sufis (Sutton, 2007). The nine types themselves are described in the 9 Enneagram types at a glance.

A systematic review that started from the observation that many clients use the Enneagram to understand their personality and relationship patterns examined 104 independent samples and found mixed evidence of reliability and validity. On the plus side, some factor-analytic work partly matched the theory, subscales related to other traits such as the Big Five in the ways the theory predicts, and several studies found the Enneagram helpful for personal or spiritual growth. On the minus side, factor analyses typically found fewer than nine factors, no study had yet used cluster analysis to look for the nine types, and there was little research supporting secondary ideas such as wings and movement between types (Hook et al., 2021, from the abstract).

That verdict means neither “there is no evidence at all” nor “science has settled it”. Results that fit the theory sit beside results that do not, and above all it has not yet been shown that people actually fall into nine types. The number that comes top on a test is therefore less a boundary confirmed by measurement than a convention: the highest-scoring type is called your type.

There is no official Enneagram test that everyone uses. Riso and Hudson, who created the RHETI, wrote that there is not, and never has been, a universally agreed-on interpretation of the Enneagram (Riso & Hudson, 2000, as cited in Scott, 2011). The RHETI asks you to choose the better-fitting statement in each of 144 pairs (Newgent, 2001, as cited in Scott, 2011); another test has you read nine short paragraphs and rank the ones that describe you best (Daniels & Price, 2009, as cited in Scott, 2011); and Korean research has used the 81-item, five-point Korean Enneagram Personality Type Indicator (Jin & Youn, 2009, from the abstract). With items and scoring this different, the same person can get different numbers on different tests.

The mind.io Enneagram test is none of these. It is not an adaptation of the RHETI or any commercial test but mind.io’s own test: 45 statements written by mind.io (five per type, two of them scored in reverse), each answered from 1 to 5. Each type is placed on a 0–100 scale between its lowest and highest possible totals, and the test’s retest reliability and validity have not been studied. The research below is therefore about Enneagram tests in general; none of its figures tells you how accurate a mind.io result is.

In a study that gave the RHETI to 287 people, internal consistency was broadly adequate, but support for construct validity was mixed, while support for its heuristic value was strong (Newgent et al., 2004, from the abstract). In Newgent’s doctoral dissertation, on which that article built, six of the nine type scales had internal consistency of 0.70 to 0.82, and the other three were lower, at 0.56 to 0.66 (Newgent, 2001, as cited in Scott, 2011).

Test–retest figures, which show how similar results are on a second sitting, also vary widely by test and sample. In a Turkish study that developed a new inventory with an online sample of 3,531 people, 91.3% of them women, 200 answered again within four weeks; the type scales’ retest correlations were 0.289 to 0.512, which the authors considered satisfactory. In the same sample, though, a short Big Five measure had similarly low retest correlations of 0.292 to 0.501, so these figures cannot show that the Enneagram is less stable than the Big Five (Yanartaş et al., 2022). The Korean Enneagram Personality Type Indicator used in Korean research is described as having had a retest reliability of 0.89 when it was developed, but that is the developer’s own figure, cited in a journal article the developer co-wrote (Jin & Youn, 2009, from the abstract), and we found no retest data from an independent study.

Is a type set by an expert more accurate? Sutton, who researched the Enneagram, summed up earlier work as showing that types set by questionnaires are generally less stable than types set in workshops (Sutton, 2007). Yet in a study where experts trained by Naranjo or his students typed people from interview videos, the experts agreed with one another on only about half the interviews, and when they watched the same videos again 2.5 to 3 years later, the agreement of 7 experts with their own first judgements was 0.598 for the more experienced and 0.434 for the less experienced (Gamard, 1986, as cited in Scott, 2011). Workshops and experts are not a guaranteed answer key either.

If the theory is right, people’s answers should split into nine groups. Factor analysis is the statistical method that shows how many strands item answers fall into; according to the systematic review, factor analyses typically found fewer than nine factors, and no study had yet used cluster analysis to look for the nine types (Hook et al., 2021, from the abstract). In Newgent’s dissertation on RHETI data, too, each type’s answers split into two strands, which was read as failing to support the nine-type structure (Newgent, 2001, as cited in Scott, 2011).

There are results on the other side. In her doctoral dissertation, Scott rewrote the RHETI as six-point items, collected 6,401 people’s answers online and obtained nine factors, one for each type, which she called the “first and only empirical evidence” of the nine types. Scott reported that the nine factors and their internal consistency held up whether or not participants said they already knew their type, but her participants chose to take part through The Enneagram Institute’s website, and 41.9% said they already knew their type. Scott herself listed the self-selected sample and the lack of a retest among her study’s limitations, and qualified her conclusion with “at least among the population of people interested in the Enneagram” (Scott, 2011).

Studies comparing Enneagram types with trait scores such as the Big Five show a fair number of links in the directions the theory predicts. In Sutton’s study of 416 people who had settled on their type through a week-long intensive course or at least three weekend courses (75% women, mostly from Britain and the United States), 21 of the 28 hypotheses about the Big Five were supported. Type 1s, for example, scored higher than the rest of the sample on conscientiousness, Type 4s lower on emotional stability and higher on openness, Type 5s lower on extraversion, and Type 7s higher on extraversion (Sutton, 2007). In the Turkish study, conscientiousness was most strongly linked with the Type 1 scale (r = 0.622; Yanartaş et al., 2022). What these overlaps look like in MBTI letters is covered in our guide to the Enneagram and the MBTI.

But overlap like this cannot show whether the nine types are clearly distinct from one another or cover every kind of personality. Scott made this point in reviewing earlier studies, and also noted as a weakness that participants who already know their type are familiar with its description and may answer “that’s me” to match it (Scott, 2011). In Sutton’s study, too, Enneagram type predicted job attitudes about as well as measures of values and motives, but not as well as the Big Five (Sutton, 2007).

The mind.io Enneagram test has five statements per type, so if you answered all five, moving one answer by one step moves that type’s score by 5 points. When your top two types are within 5 points of each other, then, changing a single answer next time can swap them. The wider the gap, the harder it is for a few answers to reverse the order.

That is why the result page shows the scores for all nine types, not just the top number. When the top two are tied or within 5 points, it names both as candidates and asks three questions that compare them, such as what you deeply want and what you deeply want to avoid. The questions are only there to help you think; they do not re-score the result, and the saved result stays the same. A gap of 6 to 10 points gets a one-line note about the close second. None of this means the test got it wrong: it tells you that the two types’ motivations came through about equally in these answers.

Wings can swap the same way. The wing on the result page is whichever of your type’s two neighbours scored higher, so when the neighbours are close it can switch between 4w3 and 4w5, and when they tie both are shown as candidates. Research supporting the idea of wings is still thin (Hook et al., 2021, from the abstract). In an older study, people who had settled on their type in workshops were asked again up to a year later: on average 85% chose the same type, and more than half of those who changed moved to a neighbouring number (Wagner, 1981, as cited in Sutton, 2007). Sutton saw this as early support for wings, but it is a different kind of evidence from retaking a scored questionnaire.

Your state at the time also sways your answers. Tired or rested, picturing yourself at work or at home, answering as the person you would like to be: each can change which statement you pick. Sutton argued that questionnaires can only tap the self-image you are aware of, so questionnaires aimed at the unconscious motives the Enneagram describes may show lower reliability (Sutton, 2007). Scott, listing the lack of a retest as a limitation, also wrote that it matters whether a typing instrument measures lasting traits rather than passing states (Scott, 2011).

One more thing: in October 2026, mind.io replaced seven of the Enneagram test’s statements with new ones. Saved results are unchanged, but if a result from before then differs from one you get now, the changed statements may be part of the reason.

First, look at the spread of scores rather than the number. Check how far your top type is from the second and third, and which types come near the top in both results. If the same two or three types keep coming near the top, the question “Am I a 4 or a 5?” turns into a more accurate description: “I use the motivations of both 4 and 5 a lot.”

Second, read the candidate types side by side and tell them apart by motivation rather than behaviour. The Enneagram treats the same behaviour as different types when the motivation differs, so asking “What do I most want to avoid?” helps. The three questions on the result page work the same way. When you compare the two descriptions, set aside sentences that would fit anyone and see which of the sentences that separate the two types sounds more like you. Why vague sentences can still feel personal is explained in the guide to the Barnum effect. Asking someone who knows you well which one sounds more like you also helps.

Third, give it time before you retake it. Retaking it straight away because you dislike the result makes it easy to remember the items and pick the answers you want. A few weeks later, on an ordinary day when you are neither worn out nor elated, answer with your usual self in mind rather than one particular situation. In the meantime, notice which type’s description fits when you actually go through a conflict or recover from one.

Fourth, cross-check with scores. The Big Five test does not sort people into types; it shows five traits as scores. Big Five scores will not tell you your Enneagram number, but they can help someone on the boundary see themselves from another angle.

The Enneagram can be a useful language for reflecting on your own motives, but it is not suited to selecting people. Sutton, who researched the Enneagram and rated its usefulness highly, wrote in her 2007 doctoral thesis that a truly reliable and valid questionnaire to identify Enneagram type had yet to be developed, and that the Enneagram is clearly unsuitable for selection, because finding one’s type can be slow and the typology in no way assesses the ability to do particular jobs (Sutton, 2007). An interview that asks for your Enneagram number or favours certain types is far from this researcher’s conclusion.

Nor is the Enneagram a diagnostic tool. It cannot assess or explain difficulties such as depression or anxiety. If low mood or worry is making daily life hard, look at that difficulty itself before your type. The depression check-in (PHQ-9) and the anxiety check-in (GAD-7) can help you put your current state into words before talking to a professional. These check-ins are not diagnoses either.

What about the idea that it helps people grow? Several studies have reported benefits for personal or spiritual growth (Hook et al., 2021, from the abstract), but studies check such effects in different ways. In a study whose authors included researchers from an Enneagram training school, ego development was measured in 122 course participants before training and 18 months later: the overall average change was not statistically significant, and only one of eight training groups (23 people) rose significantly. The authors read this as showing growth in some participants, but there was no comparison group that received no training; they themselves listed the quasi-experimental design and the self-selected participants as limitations and called for a more controlled study (Daniels et al., 2018). People’s sense that it helped may be real, but whether the Enneagram is the reason needs stricter research.

Finally, don’t use numbers as labels that settle who someone is. People with the same number still differ, and a result is most useful as a starting point for understanding each other’s motivations. Mixed evidence is not the same as no evidence, and it is not the same as settled evidence either.

Q. Is the Enneagram a scientifically validated test? A. A systematic review of 104 samples found mixed evidence for its reliability and validity (Hook et al., 2021, from the abstract). Some studies show links with the Big Five that fit the theory, but it has not yet been shown that people actually fall into nine types. So it is a different kind of tool from trait measures such as the Big Five, which were refined through large-scale research.

Q. Every site gives me a different number. Which one is right? A. There is no official test that everyone uses, and tests differ in their items and scoring, so it is hard to treat any one site’s number as the right answer. Treat the two or three types that keep coming near the top as candidates, and sort them out yourself by reading their core motivations and fears.

Q. Is a type set by an expert or in a workshop more accurate? A. Types set in workshops have been summed up as generally more stable than types set by questionnaires (Sutton, 2007), but in a study where experts typed people from interview videos, the experts agreed with one another on only about half the interviews (Gamard, 1986, as cited in Scott, 2011). Whoever sets your type, it takes time checking it against your own experience.

Q. Is there evidence for wings and arrows (directions of stress and growth)? A. Research supporting wings and movement between types is thin (Hook et al., 2021, from the abstract). The mind.io result page also labels them as interpretations from Enneagram theory. Use them only as a map for noticing how you differ when you are worn out and when you are at ease.

Q. Can a company or school sort people by Enneagram type? A. It is not suited to selecting or evaluating people. Sutton, who has researched the Enneagram, said it is unsuitable for selection because finding one’s type takes time and it does not assess the ability to do a job (Sutton, 2007). Using it as a conversation tool to understand each other’s motivations is a different thing from evaluating people.

Q. Which is more accurate, the MBTI or the Enneagram? A. The two tests look at different layers. The MBTI looks at preferences such as where you get your energy and how you make decisions; the Enneagram looks at the motivations and fears beneath them. Both sort people into types, though, so in both a result near a boundary can change. Why MBTI results change is covered in the guide to MBTI accuracy, and if you would rather see scores than types, the Big Five test is the one to take.

References: Hook, J. N., Hall, T. W., Davis, D. E., Van Tongeren, D. R., & Conner, M. (2021), The Enneagram: a systematic review of the literature and directions for future research, Journal of Clinical Psychology 77(4), 865-883, https://doi.org/10.1002/jclp.23097; Scott, S. A. (2011), An analysis of the validity of the Enneagram, doctoral dissertation, College of William and Mary, https://doi.org/10.25774/w4-b0e9-4630; Sutton, A. (2007), Implicit and explicit personality in work settings: an application of Enneagram theory, doctoral thesis, University of Leeds, https://etheses.whiterose.ac.uk/id/eprint/2869/; Newgent, R. A., Parr, P. H., Newman, I., & Wiggins, K. K. (2004), The Riso-Hudson Enneagram Type Indicator: estimates of reliability and validity, Measurement and Evaluation in Counseling and Development 36(4), 226-237, https://doi.org/10.1080/07481756.2004.11909744; Yanartaş, Ö., Malakcıoğlu, C., Acarkan, İ., & Akça, E. (2022), Validity and reliability of Enneagram personality types and subtypes inventory in a Turkish sample, Anadolu Kliniği Tıp Bilimleri Dergisi 27(3), 248-259, https://doi.org/10.21673/anadoluklin.1104082; Jin, J.-Y., & Youn, Y.-S. (2009), A study of the Korean Enneagram Personality Type Indicator based on item response theory [in Korean], Enneagram Research 6(1), 9-30, https://www.earticle.net/Article/A115589; Daniels, D., Saracino, T., Fraley, M., Christian, J., & Pardo, S. (2018), Advancing ego development in adulthood through study of the Enneagram system of personality, Journal of Adult Development 25, 229-241, https://doi.org/10.1007/s10804-018-9289-x; Riso, D. R., & Hudson, R. (2000), Understanding the Enneagram: the practical guide to personality types, New York: Houghton Mifflin (cited in Scott, 2011); Newgent, R. A. (2001), An investigation of the reliability and validity of the Riso-Hudson Enneagram Type Indicator, doctoral dissertation, University of Akron (cited in Scott, 2011); Daniels, D., & Price, V. (2009), The essential Enneagram, New York: HarperCollins (cited in Scott, 2011); Gamard, W. S. (1986), Interrater reliability and validity of judgments of Enneagram personality types, doctoral dissertation, California Institute of Integral Studies (cited in Scott, 2011); Wagner, J. (1981), A descriptive, reliability and validity study of the Enneagram personality typology, doctoral dissertation, Loyola University of Chicago (cited in Sutton, 2007)