Loading…
The Science
Built on measured ground.
TIGER stands on two research traditions, and is honest about the difference between them. This page states what we can show, what has not been proven yet, and how a reading should be used — in the same type size as everything we promise.
- Not yet
- Independently validated. We say so before anyone has to ask.
- None
- Norms, percentiles or predictive claims published — we have not earned them
- Every sitting
- Keeps its item-level answers, so validation work can run on real data
- Same in, same out
- Scoring is fixed and versioned. Nothing here learns from who got hired
Foundations
Two traditions, weighted honestly
Neither tradition is ours, and they do not carry equal evidential weight. Saying which is which is the least a science page can do.
Personality, in the six-factor tradition
The HEXACO model describes personality as six broad factors, each with facets underneath it, and it comes out of decades of peer-reviewed lexical research across languages. Of the two traditions here it is the academically stronger, and it is the one our own gaps are measured against. Kirnova also offers it as an assessment in its own right.
Behavioural style, in the DISC tradition
A practitioner tradition nearly a century old that describes observable behavioural style — Dominance, Influence, Steadiness, Conscientiousness. It has strong workplace legibility and a thinner peer-reviewed base than the six-factor work. We use it for what it is good at: a vocabulary people recognise, captured in both a natural framing (who you are when nothing is at stake) and an adaptive one (who you become at work).
What Kirnova adds is not a new theory of personality
Our items are original. No licensed instrument is copied and no equivalence to any branded assessment is claimed. TIGER itself is the layer on top: a fixed, versioned method for reading those two signals as five workplace dimensions — Trust, Influence, Grit, Execution and Resilience.
What is not on this page, and why
The weights, the cutoffs and the rules that turn answers into a reading are not published here. That is not because they are shaky; it is because they are the product. What is here is everything a reader needs to judge whether to trust a result: what it stands on, what it has not proven, and how it must be used. If any of that is missing, tell us and we will add it.
Our approach
Four properties a reading always has
These hold for every Kirnova assessment, every sitting, every time. They are the part of the method that a buyer, a candidate or a regulator can hold us to.
Deterministic, and it does not learn
The same answers always produce the same reading. There are no machine-learned weights and no model that drifts between one person and the next. Nothing in the scoring path is fed by what happened after an assessment — whether somebody was hired, or how they did — because a model that learns from its own past decisions produces numbers that look credible and are not.
Absolute, not relative
Band labels are fixed. There is no norm sample and no percentile, so a reading is never a comparison against other people, and no result carries a position in a queue.
Suppressed when it cannot be trusted
If an answer pattern shows too little variation to mean anything, the result is withheld rather than published with a caveat attached. Unusually fast completions are flagged. We would rather say a sitting cannot be trusted than hand somebody a confident-looking number built on it.
Versioned, and never quietly rewritten
Every result carries the version of the method that produced it. When the method changes, past results are superseded and kept, not overwritten — so nobody's result changes overnight without a record of what it used to say.
Validation
What is proven, what is not, and what we are doing about it
Most vendors write this page in the passive voice and hope nobody reads it. We would rather you read ours twice.
The honest position
TIGER’s ingredients rest on published research we did not do: the six-factor personality tradition and the behavioural-style tradition. TIGER’s recipe — that these five dimensions, combined our way, describe workplace behaviour — has not been tested by anyone. Not by us, and not by anyone selling a comparable promise.
What that means concretely
We publish no reliability coefficients, no factor analysis, no population norms, no fairness statistics and no predictive-validity claims — because we have not earned them. Publishing unearned statistics is how this industry lost its credibility, and there is no version of this page where we borrow it back.
Raw responses are kept
Every completion stores its item-level answers and the record that lets the sitting be reconstructed — not just the derived scores. Reliability, factor and fairness analysis can therefore run on real data, retroactively, the day the sample is large enough.
Assembly is reproducible
A stored sitting regenerates the exact form a person sat. That is what makes proper test-retest and parallel-form designs possible later.
Scoring is versioned
Every historical result stays tagged to the method that produced it. When the method changes, old results are superseded and kept, never silently rewritten.
The programme, in order
Each stage is gated on an adequate sample, and the order is not negotiable — you cannot claim prediction before you can show the thing measures anything consistently. Results will be published here as they land, including the ones that force us to change the model.
- 01
Internal consistency
Reliability per dimension and item-total analysis, on real responses rather than simulated ones.
- 02
Structure
Exploratory, then confirmatory factor analysis — at the sample sizes those methods actually require, not before.
- 03
Convergent and discriminant validity
Measured against established open instruments, so our numbers can be checked against numbers that already have standing.
- 04
Fairness
Differential item functioning across demographic groups. Until this is complete we tell customers directly: TIGER is not appropriate as a selection gate, anywhere.
- 05
Criterion studies
The long game — partnered, consented, time-lagged studies linking profiles to real outcomes. Last, because everything above has to hold first.
Evidence can change the implementation. It cannot change the honesty policy.
Known limits
What this assessment cannot do for you
These are not hypothetical risks. Each one is true of the assessment people sit today, and we publish them because a limitation you find yourself costs us more than one we told you about.
It is self-report, and self-report has a ceiling
Every reading comes from what a person said about themselves in one sitting. We do not observe anyone at work, and no questionnaire can. That is a limit of the method, not a fault we intend to engineer away.
We publish no reliability, structure or fairness statistics
Not because they would be unflattering — because we have not run them at the sample sizes those methods need. The programme above is the order we intend to earn them in, and results will be published here as they land, including the ones that force us to change the model.
Role resemblance is judgment, not evidence
Role suggestions are authored mappings between a reading and the demands of a role. We can point to published support for only one of those links. Treat them as a place to start a conversation, never as a match score.
A reading describes now, not always
People move and so does their context, which is why a TIGER result carries a validity window rather than standing forever. A reading taken during one job, at one pressure, is evidence about that period and should be re-taken rather than assumed.
Using a result
Do's and don'ts
The product is advisory by construction: there is no pass mark, no ranking and no automated gate anywhere in it. These rules are how a human keeps the rest of the decision honest.
Do
Use the assessment to create better conversations and better role context.
- Use a reading as one structured input alongside interviews, work samples, references, and context.
- Compare a reading against role demands instead of treating every high reading as universally better.
- Discuss it with the person, so interpretation includes lived experience and current constraints.
- Revisit it over time when role pressure, team design, or development focus changes.
Don't
Avoid turning a bounded, partly inferred signal into a simple gate.
- Do not use a reading as an automatic hiring, promotion, or rejection rule.
- Do not treat behavioural tendencies as fixed identity labels or clinical diagnoses.
- Do not compare people without considering the role, manager, culture, and team mix.
- Do not present a derived or blended signal as if it were a standalone validated instrument.
Research
Research posture: separate from marketing, by rule
A finding and a sales claim are different objects and must never be presented as one. This section holds only the first kind.
Foundational literature
The item bank and the interpretation layer draw on public personality-psychology work: the six-factor lexical tradition and its research on personality and work outcomes, the open tradition of publicly available personality items, and the behavioural-styles literature descending from Marston. A reading list with full citations is available on request.
Kirnova studies
None published yet. Our first will be the internal-consistency and structure analyses from the programme above, released with their limitations attached. Until then this shelf stays almost empty — an empty shelf is a promise, a faked full one is a fraud.
Research partnerships
If you run assessments at scale, we partner with employers and institutions on properly designed, consented studies — the time-lagged criterion research this industry talks about and rarely runs. Talk to us about research.
FAQs
Straight answers
Is TIGER scientifically validated?
Not yet, and we would rather be the ones to say it. The research traditions underneath it are extensively studied; our interpretation layer is in a staged programme that has not started producing results. What we claim today is deterministic, reproducible and transparent scoring. What we do not claim is reliability coefficients, norms, or predictive power.
Which frameworks does TIGER read?
Two. Behavioural style, in the DISC tradition, and personality, in the six-factor HEXACO tradition. Both are asked inside the same sitting. What TIGER adds is the reading on top of them — and that reading is ours, which is why we hold it to a standard we did not have to meet for the traditions we borrowed.
Can TIGER predict who will succeed in a job?
No. Prediction claims need time-lagged criterion studies, and we have not run them. They sit at the end of our programme, behind reliability, structure and fairness, which is where they belong.
Can I use TIGER to screen out candidates?
No, and there is nothing to configure that would let you. There is no pass mark, no ranking and no automated gate anywhere in the product. Until fairness analysis is done we advise against using any assessment, ours included, as a selection gate.
Is this like Myers-Briggs?
No. There are no fixed types — every reading carries a range rather than a binary — the method is stated rather than implied, and the validation roadmap is published instead of hinted at. Nothing here sorts anybody into a box with a name on it.
Will my score change if I retake it?
It can. You are not a constant and neither is your context, which is why results carry a validity window. What never changes quietly is the scoring: every result stays tagged to the exact version that produced it, and old versions are kept rather than rewritten.
Why should I trust a test that admits all this?
Because the alternative is trusting one that does not. These gaps exist in every assessment on the market; the difference is whose website you can read them on. A number is only worth having if it deserves to be believed.
Before you use a result
Decision support, not decisions.
TIGER results are decision support, not decisions. They estimate workplace tendencies from self-reported responses; they are not a clinical, medical, diagnostic, or legal instrument, and they are never a substitute for interviews, references, work samples, or human judgment.