Letters and percentages: how to read MBTI results without guessing

How are the four letters and percentages on an MBTI result calculated? Work through a 52% by hand, learn how to read axes near 50%, why ties get no letter, and find your neighboring type.

SCHROE Editors

Saturday morning, the first meeting of a neighborhood book club. The host hands out name stickers and asks everyone to add their MBTI type. Maya and Jordan both write ENFP without a second thought. Delighted, they pull out their phones to compare results, and the bars look nothing alike. Jordan’s first line reads “Extraversion (E, 88%).” Maya’s reads “Extraversion (E, 52%).” Jordan laughs: “So you’re basically half an I.” Maya goes quiet. Same four letters. Is her E even a real E?

Maya and Jordan are made up for this article, and Maya’s 52% is a worked example we’ll calculate below. The math, though, is exactly what the schroe.kr MBTI-style test uses to score answers. Here is where the four letters and the percentages on your result screen come from, why a number like 52% is not a verdict that you’re “on the fence,” why a letter can be left blank, and how to find the type that sits one letter away from yours.

Each line on the result screen packs in three things

Every axis on the result screen has a bar with a letter at each end (I and E, for example), a fill that grows from the center line toward one side, and a caption such as “Extraversion (E, 52%)” underneath. That caption holds three pieces of information:

The name (Extraversion): the side your answers leaned toward. The screen uses Extraversion/Introversion, Sensing/Intuition, Thinking/Feeling and Judging/Perceiving.
The letter (E): the letter that goes into your four-letter type.
The percentage (52%): how far that side leaned, shown on a scale from 50% to 100%. That is why the result screen never shows a number below 50%.

If both sides score exactly the same, the caption changes to “Evenly matched (50% / 50%)”. If there are no answers to score or the saved values don’t add up, you’ll see “Score unavailable”. All three are the exact wording on the site.

One thing matters more than anything else here: the percentage is not a ranking against other test takers. The note at the bottom of the result screen says so directly: “Scores summarize responses; they are not measures of ability, probabilities, or population percentiles.” Maya’s 52% doesn’t compare her with anyone. It describes where her own answers landed in this one sitting.

How one answer turns into a score

The test asks 10 questions per axis, 40 in total. Each question has a seven-step slider: 0 in the middle, −3 and +3 at the ends. The scoring code turns each answer into points like this:

Points for one question = answer (−3 to +3) × direction (1 or −1) × weight

Direction handles questions that are worded the other way around. On the same E/I axis, the right end of one question may point toward E while another points toward I; the direction value lines up the signs. After that, points are always positive for E, S, T, J and negative for I, N, F, P.

Weight is how much each question counts. The question bank holds 25 questions per axis: 10 with weight 1, 6 with 0.8, 5 with 1.2, and 4 with 1.5 (out of 25). When you start, 10 of the 25 are drawn at random.

Adding up the points from those 10 questions gives the axis score. At the same time, the code tallies a maximum score: the sum of the drawn questions’ weights times 3, which is what you would get by pushing every slider all the way to one end.

Maya’s 52%, worked out by hand

Here are Maya’s ten E/I answers. Each bar in the chart is one question’s share.

Step by step:

Add the points: +3 −2 +1 −1 +2 −2.4 +2.4 −1.2 +1.2 −1.5 = +1.5. Positive, so it leans E.
Maximum score: the weights add up to 1+1+1+1+1+0.8+0.8+1.2+1.2+1.5 = 10.5. Times 3, that’s 31.5.
The formula: the code uses ((score + max) ÷ (max × 2)) × 100. So (1.5 + 31.5) ÷ 63 × 100 = 52.38%, measured toward E.
What the screen shows: the leaning side, rounded: “Extraversion (E, 52%).”

All the formula does is stretch the score range, −31.5 to +31.5, onto 0% to 100%, then mark where +1.5 falls. A score of 0 sits at exactly 50%. Had Maya scored −1.5 instead, the same formula would give 47.62% toward E, and the screen would show the other side: “Introversion (I, 52%).”

What 52% says, and what it doesn’t

Back to Jordan’s “half an I.” Once you follow the math, 52% says one thing only: in these ten questions, Maya’s E answers and I answers carried almost equal weight, and E came out 1.5 points ahead.

Look at the chart again and her answers aren’t vague at all. She went all the way to the E end (+3) on questions 1 and 7, and all the way to the I end (−3) on question 6. Her answers didn’t pile up in the middle; they split clearly in both directions and cancelled out to near zero. She may be someone who seeks people out in some situations and firmly wants to be alone in others. A 52% can’t tell “someone who stands in the middle” apart from “someone who moves between both sides,” so there’s no reason to read it as “I’m wishy-washy.”

Likewise, Jordan’s 88% doesn’t mean Jordan is more outgoing than Maya as a person. It means Jordan’s answers that day lined up on one side. As the site says, this is a reflection tool built on SCHROE’s own questions, not the official MBTI assessment.

Same score, different percentage: it depends on which questions you got

The formula doesn’t only use your score. It also uses the maximum score, and that changes from person to person and from one attempt to the next, because 10 questions are drawn at random from 25 and their weights differ.

Draw the ten lightest questions (six at 0.8, four at 1) and the maximum is 26.4. Draw the ten heaviest (four at 1.5, five at 1.2, one at 1) and it’s 39. The same +3 becomes 55.68% on the first set and 53.85% on the second. With heavier questions, the same gap sits on a wider scale, so the percentage moves less.

That’s why comparing percentages with a friend point by point doesn’t tell you much: you may not have answered the same questions. Percentages are most useful within a single result, for spotting which of your four axes sits closest to the middle.

When a score is exactly zero, the letter stays blank

Rarely, an axis adds up to exactly 0. When that happens, the scoring code doesn’t assign a letter at all. A 50% doesn’t get an E or an I stuck on it by default.

In the code, a zero axis goes onto a list of tied axes (tiedDimensions), its letter is set to null, and if any axis is tied, the four-letter type itself is null. Because some weights are decimals like 0.8 and 1.2, a computer can produce tiny rounding errors; the code treats those as 0 too, so a rounding error can’t invent a preference.

Instead of a type, the result screen shows:

> Some preferences are evenly matched > We have not assigned a preference where the two sides have equal scores. Instead of forcing a four-letter type, consider your usual behavior and review your responses in a new assessment.

and the tied axis is labeled “Evenly matched.” The reasoning is simple: picking a letter from a score of zero would be a coin toss.

How the “neighboring type” is chosen

The site’s code also includes a rule for picking your closest neighboring type (getClosestNeighborType). It comes down to three steps:

For each axis, measure how far the E/S/T/J-side percentage is from 50 (|percentage − 50|).
Pick the axis with the smallest distance and flip that one letter. That’s the neighboring type.
If two or more axes are equally close, the one that comes first in the order E/I → S/N → T/F → J/P wins.

Say Maya’s other axes were Intuition 70%, Feeling 64% and Perceiving 75%. Measured toward E, S, T and J, that’s 52.38, 30, 36 and 25, so the distances from 50 are 2.38, 20, 14 and 25. E/I is closest, so Maya’s neighbor is INFP, with 50 + 2.38 rounded to 52 recorded alongside it.

And Jordan? With Extraversion 88%, Intuition 80%, Feeling 61% and Perceiving 77%, the closest call is T/F, so Jordan’s neighbor is ENTP. Two ENFPs, two different neighbors. A neighboring type isn’t “the right answer if your result is wrong.” It’s your type with its least settled letter flipped. Reading both descriptions side by side helps you notice the situations where you drift that way.

Find your own neighbor

Open your result and find the line whose percentage is closest to 50%. The screen rounds, so if two lines show the same number, take the one that comes first in the order above (E/I → S/N → T/F → J/P); in rare cases the unrounded values decide differently. Then read across from your type to that axis.

| Your type | Closest call on E/I | on S/N | on T/F | on J/P | |---|---|---|---|---| | ISTJ | ESTJ | INTJ | ISFJ | ISTP | | ISFJ | ESFJ | INFJ | ISTJ | ISFP | | INFJ | ENFJ | ISFJ | INTJ | INFP | | INTJ | ENTJ | ISTJ | INFJ | INTP | | ISTP | ESTP | INTP | ISFP | ISTJ | | ISFP | ESFP | INFP | ISTP | ISFJ | | INFP | ENFP | ISFP | INTP | INFJ | | INTP | ENTP | ISTP | INFP | INTJ | | ESTP | ISTP | ENTP | ESFP | ESTJ | | ESFP | ISFP | ENFP | ESTP | ESFJ | | ENFP | INFP | ESFP | ENTP | ENFJ | | ENTP | INTP | ESTP | ENFP | ENTJ | | ESTJ | ISTJ | ENTJ | ESFJ | ESTP | | ESFJ | ISFJ | ENFJ | ESTJ | ESFP | | ENFJ | INFJ | ESFJ | ENTJ | ENFP | | ENTJ | INTJ | ESTJ | ENFJ | ENTP |

To get a feel for the percentage itself, subtract 50 and double it. That tells you what share of the maximum score the axis leaned by.

| Result screen % | Lean, as a share of the maximum | In points, if the max is 31.5 | |---|---|---| | Evenly matched (50%) | 0% | 0 (no letter assigned) | | 52% | about 4% | about 0.9–1.6 | | 55% | about 10% | about 2.8–3.5 | | 60% | about 20% | about 6.0–6.6 | | 70% | about 40% | about 12.3–12.9 | | 85% | about 70% | about 21.7–22.4 | | 100% | 100% | 31.5 (all ten sliders at one end) |

Points are given as ranges because the screen rounds. Maya’s 52%, 1.5 out of 31.5, fits right in.

Back at the book club

On the way home, Maya opens the INFP description. She nods at the part about recharging by sorting out her thoughts alone, then opens ENFP and nods at the part about ideas growing in conversation. Both were her. Her 52% was a record that those two sides showed up with similar weight in this test, not a ruling that she’s a blur.

Jordan, meanwhile, gets more curious about the 61% T/F axis than the 88% E/I one, reads the ENTP description, and jots down: “This is more like me at work.”

In short:

The four letters are the side each axis leaned toward. If any axis scores exactly 0, its letter stays blank and no four-letter type is assigned.
The percentage comes from ((score + max) ÷ (max × 2)) × 100, and the screen shows the leaning side, rounded. It is not a percentile or a skill rating.
An axis near 50% means your answers on that axis carried similar weight on both sides. It doesn’t label you as indecisive.
Your neighboring type flips the letter on the axis closest to 50%. The table above lets you find it yourself.

Curious what your four lines look like? Try the MBTI-style personality test.

MBTI results are for fun and self-reflection, not diagnosis or evaluation. The calculations follow the schroe.kr MBTI-style test scoring code as of September 30, 2026. Maya, Jordan, their answers and percentages are illustrative examples. Question counts (10 drawn from a bank of 25 per axis) and weights come from the site’s question files.

Try it yourself

Take the MBTI-style test