How the engine works — and what we haven't proven yet.
Psychometrics is a field with a lot of pseudoscience. This page exists so buyers, researchers, and regulators can see exactly what we score, how we score it, what's licensed, and where our validity work stands — without marketing gloss.
Maintained by the EQIQs team. This page describes current practice honestly, including in-progress work. It is not a third-party certification, and no claim below should be read as peer-reviewed evidence unless it links to a published source.
Validity & Reliability — where we stand
Established psychometric publishers (Hogan, PI, Truity) publish test-retest reliability, predictive validity, and normative sample sizes. We are earlier in that journey. Here is what is shipping, what is in progress, and what is planned — with no fabricated numbers.
Publish per-scale internal-consistency evidence
Versioned per-scale reliability results, sample definitions and calculation methods are needed before making public reliability or population-norm claims. These results are not established by the score alone.
Coverage transparency, not false precision
Every read shows raw framework coverage (e.g. "14 of 21 frameworks") driven by item count, response consistency, and data-sufficiency thresholds (minimum 3 completed frameworks for compatibility scoring). Coverage stays visible everywhere the score does.
Auditable, reproducible scoring
Every score is deterministic given the input responses. The scoring engine is patented (U.S. Patent #9,002,961) and traceable back to the underlying instruments.
Test-retest reliability study
A longitudinal re-test cohort is being recruited across our Big Five, EI, and Attachment scales. Results will be published in the methodology whitepaper (target: within the next release cycle).
Convergent validity vs. established instruments
We are correlating EQIQs Big Five outputs against IPIP-NEO reference administrations and EQIQs EI outputs against published EI scales. Results will be published with sample sizes and correlation coefficients.
Outcome study (in progress — no predictive claims until published)
Working with pilot enterprise customers on team and retention outcomes linked back to compatibility scores. This is the study that separates a compatibility engine from a personality quiz — and we won't publish numbers until the sample size supports them.
Peer-reviewed methodology paper
Full methodology paper documenting item selection, scoring, reliability, and construct validity — targeting an appropriate peer-reviewed venue rather than a self-published PDF.
Honest limitation: until the test-retest and predictive-validity studies above are published, EQIQs scores should be treated as decision-support, not the sole basis for hiring, clinical, or high-stakes decisions. Every report ships with this disclaimer for that reason.
Reference psychometrics for the source instruments we adapt
Published reliability and validity numbers for the underlying, peer-reviewed instruments our scales are adapted from. These are the source-instrument benchmarks — not EQIQs's own numbers, which will be published alongside the methodology whitepaper. Cronbach's α of 0.70+ is generally considered acceptable for research; 0.80+ is strong.
| Source instrument | Reported α (range) | Test-retest | Reference |
|---|---|---|---|
| IPIP-NEO-120 (Big Five) | 0.77 – 0.86 per domain | ~0.83 (2–3 mo) | Johnson, J.A. (2014). J. Res. Personality, 51, 78–89. |
| ECR-R (Attachment) | 0.90 – 0.94 | ~0.90 (3 wk) | Fraley, Waller & Brennan (2000). J. Pers. Soc. Psychol., 78(2). |
| O*NET Interest Profiler (RIASEC) | 0.93 – 0.96 per scale | ~0.81 (1 mo) | Rounds et al. (2010). U.S. Dept. of Labor, O*NET Development. |
| Schwartz Value Survey | 0.61 – 0.75 per value | ~0.75 | Schwartz, S.H. (1992). Adv. Exp. Soc. Psychol., 25, 1–65. |
| MEQ (Chronotype) | 0.83 | ~0.83 | Horne & Östberg (1976). Int. J. Chronobiol., 4(2), 97–110. |
| Thomas-Kilmann conflict-mode literature | 0.43 – 0.71 per mode | ~0.61 (avg) | Thomas & Kilmann (1974); Womack (1988) meta-review. |
| Growth Mindset Scale (Dweck) | 0.82 – 0.94 | ~0.80 (2 wk) | Dweck (1999); De Castella & Byrne (2015). Eur. J. Psychol. Educ., 30. |
| WLEIS (Emotional Intelligence) | 0.83 – 0.90 per dim. | ~0.79 (4 wk) | Wong & Law (2002). Leadership Q., 13(3), 243–274. |
| Grit-S (Duckworth) | 0.73 – 0.83 | ~0.68 (1 yr) | Duckworth & Quinn (2009). J. Pers. Assess., 91(2), 166–174. |
| Self-Compassion Scale (SCS) | 0.92 (total) | ~0.93 (3 wk) | Neff (2003). Self & Identity, 2(3), 223–250. |
| Regulatory Focus Questionnaire | 0.73 – 0.80 | ~0.79 (2 mo) | Higgins et al. (2001). Eur. J. Soc. Psychol., 31(1), 3–23. |
| Rotter Locus of Control | 0.65 – 0.79 | ~0.72 (2 mo) | Rotter (1966). Psychol. Monographs, 80(1); Lange & Tiggemann (1981). |
| Rosenberg Self-Esteem Scale | 0.77 – 0.88 | ~0.85 (2 wk) | Rosenberg (1965); Robins et al. (2001). Pers. Soc. Psychol. Bull., 27. |
| Career Anchors Inventory (Schein) | 0.66 – 0.80 per anchor | ~0.70 (6 mo) | Schein (1990); Feldman & Bolino (1996). J. Occup. Psychol., 69. |
| Love Languages (5LLPS) | 0.68 – 0.87 per language | not established | Egbert & Polk (2006); Bunt & Hazelwood (2017). J. Fam. Comm., 17. |
| Enneagram (RHETI v2.5) | 0.56 – 0.82 per type | mixed evidence | Newgent et al. (2004). Meas. Eval. Couns. Dev., 36(4). Limited peer-reviewed base. |
| MBTI® (source instrument) | 0.75 – 0.90 per dichotomy | ~0.65 4-letter (5 wk) | Myers et al. (1998) MBTI Manual, 3rd ed. Note: type-stability limitations well documented. |
| DISC (self-report adaptations) | 0.71 – 0.85 per style | ~0.78 (3 wk, PPA) | Marston (1928); Thomas Int'l PPA tech. report (2009). Varies widely by vendor. |
EQIQs adapts item wording and scoring from these instruments; adapted-form α should be expected to sit at or slightly below the source benchmarks until our own confirmatory study publishes. We do not claim the source-instrument numbers as our own.
Framework licensing — what's what
EQIQs synthesizes 21 frameworks. Some sit on public-domain research (Big Five/IPIP, RIASEC, Attachment). Two commonly recognized ones — MBTI® and DISC® — are trademarks held by external publishers, and we are transparent that we produce our own approximations, not licensed administrations of those instruments. The rest are either public models with our own item banks, or EQIQs-developed composites.
Scored using IPIP (International Personality Item Pool) items — public domain and free for commercial use.
Public-domain item bank. EQIQs response scores are not population percentiles; published normative calibration is not established.
Interest inventory items adapted from public-domain O*NET Interest Profiler.
Adapted from Schwartz Value Survey structure with plain-language items.
Item structure informed by ECR-R public research literature.
Self-report scales in the public research tradition; not the Mayer-Salovey MSCEIT (which is licensed).
Structure informed by the public Thomas-Kilmann conflict-mode literature; items are our own.
EQIQs-developed rubric mapping self-report language patterns to four behavioral clusters.
EQIQs-developed items for remote/async/structure preferences.
We are not licensed by The Myers-Briggs Company. We produce a four-letter type using our own item bank; results are not interchangeable with an official MBTI® assessment.
DISC® is a family of trademarks held by several publishers. We produce a DISC-style four-quadrant read using our own items; results are not interchangeable with a licensed DISC assessment.
The nine-type model is public. Our item bank and scoring are our own.
The five-category model is public. Items and scoring are our own.
EQIQs-developed composite that reads across cognitive, emotional, and behavioral signals.
Adapted from the public MEQ (Horne-Östberg) literature; items are our own.
EQIQs-developed rubric for speed vs. deliberation and data vs. intuition preferences.
Growth vs. fixed mindset structure from Dweck's published work; items are our own.
EQIQs-developed pattern read across coping, recovery, and reactivity signals.
Cultural framework used strictly as a low-weight, dampened proxy. Never used in isolation.
Cultural framework used strictly as a low-weight, dampened proxy. Never used in isolation.
Cultural framework used strictly as a low-weight, dampened proxy. Never used in isolation.
Trademark note. MBTI® is a registered trademark of The Myers & Briggs Foundation. DISC® is a family of trademarks held by various publishers. EQIQs is not affiliated with, endorsed by, or licensed by these organizations. Where we generate a four-letter type or a four-quadrant behavioral read, we describe it as a "type" or "style" — never as a licensed MBTI or DISC result.
How compatibility is scored
Composite, not any single framework
Compatibility is a weighted synthesis across up to 21 frameworks. No single instrument — including MBTI-style type or zodiac — can move a score on its own. Cultural/proxy frameworks (numerology, zodiac) are heavily dampened by design.
Context-weighted by relationship type
Weights shift by context — Business, Romantic, Co-parenting, Friendship. Conscientiousness and communication weigh more for Business; attachment and values weigh more for Romantic. The re-weighting is deterministic and inspectable.
Data-sufficiency floor
A compatibility score requires at least 3 completed frameworks per person. Below that, the UI shows raw coverage ("N of 21 frameworks"), flags the read as directional, and blocks high-stakes report downloads.
Reproducible and explainable
The same inputs always produce the same score. Every report can be traced back to the underlying framework outputs and item responses.
Patent citation
A method of opening communication channels between two people using match suggestions derived from a weighted composite across psychometric type, Chinese zodiac, and Western astrology — with the ability to weight two of the three categories higher than the third. Today's EQIQs engine generalizes that weighted-composite-plus-dampened-proxy pattern across 21 frameworks; the patent covers the underlying scoring construction, not the wider framework list.
We cite the patent because it is real, granted, and enforceable — not as a substitute for validity evidence. Patent grant reflects novelty, not psychometric accuracy. See the Validity & Reliability section above for how we treat the two separately.
Production proof
You do not have to take our word for any of this. The pages below are live and updated as we ship — not marketing copy.
Live system status
Real probes against the web app, auth, API, OpenAPI spec, shared reports, and extension update feed — run in your browser, not a static badge.
Open /statusPublic product changelog
Dated entries for every shipped feature, improvement, and fix — with an RSS feed. What we ship is on the record.
Open /changelogTrust Center & security
Data handling, subprocessors, retention, incident contact, and current security controls in one place.
Open /trustDoing enterprise due diligence?
Request the current methodology packet — internal-consistency values per scale, item-bank sourcing, and the status of in-flight validity studies. We would rather answer hard questions than lose to them.