Method
The points behind every answer, the formula that turns 60 statements into four letters, and the reasons a letter can be wrong.
Last reviewed September 6, 2026 · Shailendra Zade, Founder & Developer
Scoring runs the same way for everyone, with no adjustment for age, gender, country, or how long you took. Each step below is one stage of that pipeline.
The five choices map to fixed numbers: −3, −2, 0, +2, and +3. Nothing else about your answer is recorded: not the order you answered in, not how long you hesitated.
Every statement belongs to one axis and points at one side of it. Agreeing with a forward-keyed statement adds points; agreeing with a reverse-keyed one subtracts them. That is the only reason a statement can pull against your agreement.
The four axes are scored separately and never influence each other. Your Social energy answers cannot change your Structure letter.
A positive total gives one letter, a negative total the other, and the size of the total becomes the percentage. That percentage is a strength of lean, not a ranking against other people.
Five choices, each worth a fixed number of points. The scale is symmetric around Neutral, so agreeing strongly is worth exactly as much as disagreeing strongly.
−3
Strongly disagree
−2
Disagree
0
Neutral
+2
Agree
+3
Strongly agree
The scale skips ±1: the mild options are worth 2 points rather than 1. That widens the gap between a genuine Neutral and a mild opinion, which we prefer, because a shrug and a mild preference are different answers. It also means the strongest answer is only 1.5× a mild one, so a single item can never dominate an axis.
A statement you never answer counts as Neutral: zero points, no pull in either direction.
Half the statements on each axis are written toward one pole and half toward the other. Without that, agreeing with everything would produce a confident result out of nothing but agreeableness.
The formula
contribution = answer points × keying direction
Every statement belongs to one axis and points at one side of it. Agreeing with a forward-keyed statement adds the answer points. Agreeing with a reverse-keyed one subtracts them. That product is the contribution, and it is the only reason a statement can pull against your agreement.
“I regularly start conversations with people I have just met.”
Forward, toward Extraverted
Agreeing moves you toward E.
Agree → E“I prefer quiet, reflective spaces over lively group collaboration.”
Reverse, toward Introverted
Agreeing moves you toward I.
Agree → IOnce every contribution on an axis is added together, two lines of arithmetic produce the letter and the percentage. This is the arithmetic the scoring code runs.
The letter
letter = total ≥ 0 ? positive pole : negative pole
Each axis has a positive and a negative pole. Positive is E, N, T, or J; negative is I, S, F, or P. A total of exactly zero resolves to the positive pole, an arbitrary tie-break, which is why we never report a zero as settled.
The percentage
percentage = min(100, max(50, ceil(|total| ÷ maxScore × 50) + 50))
maxScore is the largest total that axis can reach: 3 points times its item count. Dividing by it puts every axis on the same footing even though the axes hold different numbers of statements. The result is then expressed as a lean from 50% to 100%, so it always reads as a percentage toward the letter you were given.
| Axis | Letters | Statements | Max total |
|---|---|---|---|
| Social energy | E / I | 16 | ±48 |
| Information | N / S | 16 | ±48 |
| Decisions | T / F | 14 | ±42 |
| Structure | J / P | 14 | ±42 |
How maxScore is used
Two axes carry 16 statements and two carry 14, so their maximum totals differ. maxScore is 3 points times that count.
Scaling by it keeps a 70% on one axis comparable to a 70% on another. The result is then expressed as a lean from 50% to 100%, so it always reads as a percentage toward the letter you were given.
Take the Decisions axis, which holds 14 statements. Three of them, with the arithmetic spelled out:
Strongly agree
+3
Forward, toward Thinking
Contribution +3
Disagree
−2
Forward, toward Thinking
Contribution −2
Agree
+2
Reverse, toward Feeling
Contribution −2
| Your answer | Points | Keying | Contribution |
|---|---|---|---|
| Strongly agree | +3 | Forward, toward Thinking | +3 |
| Disagree | −2 | Forward, toward Thinking | −2 |
| Agree | +2 | Reverse, toward Feeling | −2 |
Notice the third example: agreeing produced a negative contribution, because the statement was written toward Feeling.
Do that for all 14 Decisions statements and suppose the contributions sum to −15.
The total is negative, so the letter is F.
The axis maximum is 3 × 14 = 42, so the ratio is 15 ÷ 42 ≈ 0.357.
15 ÷ 42 × 50 ≈ 17.86, rounded up to 18, plus 50, giving 68.
Your report reads 68% Feeling.
That 68% is above our close-call line, so the F is reported as settled. Had the total been −3 instead of −15, the same arithmetic would have produced 54%, and we would have flagged the letter and pointed you at the neighboring type instead.
A percentage near 50 means your answers landed close to the middle of that axis. The letter still has to be printed somewhere, but reporting it as settled would be a small lie.
The rule
55%
at or below
At 55% or below, we mark the letter as too close to call, say so on your results, and name the type you would have been given if that one letter had flipped. You can read that neighboring type beside yours and decide which description fits.
The part most tests leave out
52%·99%
print the same letter
This is the part most tests leave out: a 52% and a 99% print the same letter, and only one of them is telling you something.
Because reliability moves from sample to sample, a letter decided by a thin margin can flip on a retake.
When every axis is too close
If all four axes land at or near the midpoint, for example when someone answers Neutral throughout or agrees with everything, we say the printed code is a tie-break rather than a reading of you. The report is still built on it, so treat it lightly, and we suggest retaking it with more decisive answers.
Acquiescence bias, the tendency to agree with whatever a survey puts in front of you, is a documented way for a self-report test to produce a confident result out of noise. The item bank is built so that it cannot.
Because each axis holds an equal number of forward- and reverse-keyed statements, and every statement carries the same weight, identical answers all the way down cancel out exactly. Answer Strongly agree 60 times and all four totals come out at zero. The same is true of Strongly disagree, and of Neutral.
Social energy
8 forward · 8 reverse
Information
8 forward · 8 reverse
Decisions
7 forward · 7 reverse
Structure
7 forward · 7 reverse
| Axis | Forward | Reverse | Balanced |
|---|---|---|---|
| Social energy | 8 | 8 | Yes |
| Information | 8 | 8 | Yes |
| Decisions | 7 | 7 | Yes |
| Structure | 7 | 7 | Yes |
Verified in the build
This is not a claim we ask you to take on faith. Those counts are read from the live item bank rather than typed by hand, and a script in the repository replays every uniform answer pattern on each build. If an axis ever stopped being balanced, the site would fail to build.
This is the single most misread number on any personality report, so it is worth being blunt about.
68% Feeling does not mean you are more feeling-oriented than 68% of people. It means your answers leaned toward Feeling on our own scale, at 68% on a range that runs from 50% (a dead tie) to 100%.
We have not collected a representative sample to rank you against, and we would rather say so than imply a precision we do not have. We score you against the maximum the scale allows, not against other people.
The axes are preferences between two things, not quantities. A high Thinking score does not mean you feel less; it means that when the two pull against each other, you reported reaching for logic first.
The result is then expressed as a lean from 50% to 100%, so it always reads as a percentage toward the letter you were given. That percentage is a strength of lean, not a ranking against other people.
Retake the test in a different mood and the percentages will move. Letters near the middle can flip outright. Because reliability moves from sample to sample, a letter decided by a thin margin can flip on a retake.
That is the honest behavior of a self-report instrument, not a bug in ours. At 55% or below we say so and name the neighboring type rather than presenting a coin flip as a finding.
A personality result can change how someone sees a job, a partner, or themselves, which is exactly why the limits belong on the same page as the method. None of the following is fine print.
Nothing here identifies a disorder, a condition, or a mental health need, and no result should be read as one. If you are struggling, a licensed professional can offer something this page cannot. We are publishers, not clinicians.
The four-letter format comes from a long tradition of Jungian type description, but this is our own instrument with our own statements and our own scoring. It has not been through commercial certification or psychometric validation, and we do not claim otherwise.
Every answer is you describing yourself, on one day, in your own words about yourself. Mood, recent events, and the version of yourself you would like to be all leak into that. Nobody has observed how you actually behave.
Traits are continuous; letters are not. Two people with the same four letters can differ enormously, and two people who differ by one borderline letter can be nearly identical. Publishing percentages and flagging close calls is our attempt to soften that, but the format has a hard limit.
We score you against the maximum the scale allows, not against a representative sample of people, because we have not gathered one. Any site that hands you a percentile owes you an explanation of the sample behind it.
We have not run studies linking these results to job performance, relationship success, or any other outcome, so we will not imply that they do. The career and relationship chapters describe patterns people recognize; they are not forecasts.
Do not use this to screen candidates, assign work, or settle an argument about a partner. Type descriptions are a language for talking about yourself, and they are far too blunt to justify a decision about another person's life.
The statements assume contexts (offices, parties, plans) that carry different weight across cultures, and translation would shift their meaning. We have not tested how the instrument behaves outside English.
Sources: Pittenger (2005) (opens in a new tab); The Myers-Briggs Company (opens in a new tab)
If any of this reads as us undercutting our own test, that is the point. A result you can trust is one that told you where it stops.
The formula on this page is only half the record. The 60 statements and a live run of the same arithmetic are one click away.
If you came here while reading how the test works, these pages are the next useful stop.
Every statement in the bank, grouped by axis, each labeled with which letter agreeing with it moves you toward.
See the bankAbout 10 minutes. No account needed. Results save in this browser.
Take the testYour type report is saved in this browser. Open the full profile anytime on this device.
See your resultsWhat the research supports, what it does not, and the limits we keep on every result.
Read the standards