Questions, scoring and translations for the Open Extended Jungian Type Scales (OEJTS), the open-source personality test behind openjung.org. Pure TypeScript, zero dependencies, ESM.
- 32-question OEJTS 1.2 with the 8-question quick mode and per-dimension mini tests
- Scoring that is deterministic and documented: raw scores, four-letter type, pole percentages
- Quality metrics: confidence per dimension, profile clarity, answer-consistency flags
- 44 locales on every question (5 original, 39 AI-assisted drafts; see TRANSLATIONS.md)
- PurrJung, a 16-question companion test for cats, sharing the same result shape
Based on the OEJTS 1.2 by Eric Jorgenson / Open Psychometrics.
npm install @openjung/coreRequires Node 20+ or any bundler that understands ESM. The package ships dist/ with declarations and source maps, plus src/ for readers.
import { sortedQuestions, generateResult, isTestComplete, type TestAnswers } from '@openjung/core';
// Present `sortedQuestions` (32 bipolar pairs, ordered by ID) and collect 1–5 answers:
// 1 = strongly the left trait, 3 = neutral, 5 = strongly the right trait
const answers: TestAnswers = { 1: 4, 2: 2 /* …all 32 question IDs */ };
if (isTestComplete(answers)) {
const result = generateResult(answers);
result.type; // "ENFP"
result.scores; // { EI: 18, SN: 30, TF: 20, JP: 32 } raw, 8–40 each
result.percentages; // { E: 69, I: 31, S: 31, N: 69, T: 38, F: 62, J: 25, P: 75 }
}Every question looks like this. Locale keys beyond en are present on all 44 supported locales:
{
id: 1,
dimension: 'JP',
title: { en: 'How do you track tasks?', zh: '你如何记录任务？', de: 'Wie behältst du deine Aufgaben im Blick?', /* … */ },
leftTrait: { en: 'Makes lists', zh: '制定清单', de: 'Ich mache Listen', /* … */ },
rightTrait: { en: 'Relies on memory', zh: '依靠记忆', de: 'Ich verlasse mich auf mein Gedächtnis', /* … */ },
}| Test | Questions | Per dimension | Score range | Midpoint | Low score → | High score → |
|---|---|---|---|---|---|---|
| Full | 32 | 8 | 8–40 | 24 | E, S, F, J | I, N, T, P |
| Quick | 8 | 2 | 2–10 | 6 | E, S, F, J | I, N, T, P |
| Single dimension | 8 | 8 | 8–40 | 24 | E / S / F / J | I / N / T / P |
| PurrJung (cats) | 16 | 4 | 4–20 | 12 | E, S, T, J | I, N, F, P |
- Scores are plain sums of the 1–5 answers. Unanswered questions count as neutral (3).
- A score above the midpoint resolves to the high-score pole; a score exactly on the midpoint resolves to the low-score pole.
- Percentages map the score range linearly onto 0–100 for the high-score pole; the opposite pole is the remainder. Each pair sums to 100.
- Note that TF runs F → T on the human tests (left traits are Feeling) and T → F on PurrJung.
Everything is exported from the package root. Types are listed at the end.
| Export | Description |
|---|---|
questions |
All 32 QuestionPairs, grouped by dimension |
sortedQuestions |
The same questions ordered by ID, for presentation |
dimensionQuestions |
{ EI, SN, TF, JP } → question IDs |
TOTAL_QUESTIONS |
32 |
QUESTIONS_PER_DIMENSION |
8 |
quickTestQuestionIds |
{ EI, SN, TF, JP } → the two most discriminating IDs per dimension |
quickTestQuestions |
Those 8 IDs, sorted |
QUICK_TEST_TOTAL |
8 |
QUICK_TEST_PER_DIMENSION |
2 |
| Function | Returns |
|---|---|
generateResult(answers) |
TestResult with type, scores, percentages |
calculateScores(answers) |
DimensionScores, 8–40 each |
determineType(scores) |
Four-letter type string |
calculatePercentages(scores) |
DimensionPercentages, one entry per pole |
isTestComplete(answers, total?) |
true when exactly total (default 32) answers exist. Count check only. |
generateQuickResult, calculateQuickScores, determineQuickType, calculateQuickPercentages and isQuickTestComplete mirror the full-test functions over the 8 quick-test questions.
Score one dimension on its own, for a focused mini test:
import {
generateDimensionResult,
isDimensionTestComplete,
getDimensionQuestionIds,
} from '@openjung/core';
getDimensionQuestionIds('EI'); // [3, 7, 11, 15, 19, 23, 27, 31]
if (isDimensionTestComplete(answers, 'EI')) {
generateDimensionResult(answers, 'EI');
// { dimension: 'EI', score: 18, preference: 'E', leftPercent: 69, rightPercent: 31 }
}Also exported: calculateDimensionScore, determineDimensionPreference, calculateDimensionPercentages, and the constants DIMENSION_QUESTIONS_COUNT (8), DIMENSION_SCORE_MIN (8), DIMENSION_SCORE_MAX (40), DIMENSION_THRESHOLD (24).
How clear and how internally consistent a result is:
import { calculateTestConfidence, checkTestConsistency, getConfidenceLabel } from '@openjung/core';
const confidence = calculateTestConfidence(result.scores);
confidence.EI; // { dimension: 'EI', level: 'moderate', distance: 6, percentage: 38 }
confidence.clarityIndex; // 0–100, how far the whole profile sits from the midpoints
getConfidenceLabel(confidence.EI.level, 'E'); // "Moderate E preference"
const consistency = checkTestConsistency(answers);
consistency.overallConsistent; // false when adjacent answers in a dimension diverge by 3+
consistency.warnings; // ["Inconsistent answers in SN dimension"]Levels: strong (distance ≥ 12 from the midpoint), moderate (≥ 6), slight (≥ 2), balanced. Per-dimension variants: calculateDimensionConfidence, checkDimensionConsistency, getConfidenceLevel.
import { SUPPORTED_LOCALES, getLocalizedText, type Locale } from '@openjung/core';
SUPPORTED_LOCALES; // ['en', 'zh', 'ja', 'ko', 'zh-tw', 'ms', 'de', 'fr', …] (44)
getLocalizedText(question.leftTrait, 'de'); // falls back to `en` for unknown localesPURRJUNG_LOCALES lists the five locales available on the cat test. DEFAULT_LOCALE is 'en'.
The 39 locales added in September 2026 are AI-assisted drafts without native-speaker review. TRANSLATIONS.md records their provenance and the phrases most in need of review; corrections are welcome through the Translation correction issue template.
A 16-question cat temperament test on the same four axes (Social/Solitary, Routine/Novelty, Independent/Bonded, Structured/Spontaneous):
import { sortedPurrjungQuestions, generatePurrjungResult } from '@openjung/core';
generatePurrjungResult(catAnswers); // { type: 'ISFP', scores: { EI: 16, … }, percentages: { … } }Also exported: purrjungQuestions, purrjungDimensionQuestions, PURRJUNG_TOTAL_QUESTIONS (16), PURRJUNG_QUESTIONS_PER_DIMENSION (4), PURRJUNG_SCORE_MIN (4), PURRJUNG_SCORE_MAX (20), PURRJUNG_THRESHOLD (12), calculatePurrjungScores, determinePurrjungType, calculatePurrjungPercentages, isPurrjungTestComplete, getPurrjungDimensionQuestionIds.
TestAnswers, Dimension, DimensionScores, DimensionPercentages, TestResult, SingleDimensionResult, QuestionPair, MultilingualText (alias BilingualText, deprecated), DimensionQuestions, Locale, ConfidenceLevel, DimensionConfidence, TestConfidence, ConsistencyResult, TestConsistency.
The repository ships a skill that teaches Claude Code the openjung.org HTTP API and this package's scoring model:
npx skills add https://github.com/openjung/core --skill openjung-apinpm ci
npm run check # typecheck, format check, tests, build, package lintSee CONTRIBUTING.md for the contracts that must not change silently, and for the release process.
Code is MIT. The OEJTS questionnaire items and all adaptations of them, including the 44 locales and the quick subset, are CC BY-NC-SA 4.0 by Eric Jorgenson / Open Psychometrics; see LICENSE-DATA and the original questionnaire.