Home / Methodology
How Haliscope measures 11 traits
Where the eleven traits come from, how responses are scored, how occupational matching is done, and what the method cannot do.
1. Foundations
Haliscope sits on two research traditions that developed separately for most of the twentieth century.
HEXACO (Ashton & Lee, 2004) is a six-factor model of personality drawn from lexical studies across several languages. It revises the Big Five and adds honesty-humility as a factor of its own. RIASEC (Holland, 1959) is a six-dimension model of vocational interest — Realistic, Investigative, Artistic, Social, Enterprising, Conventional — and remains the organising framework for most career guidance. Both are explained in full here.
O*NET is the occupational database published by the U.S. Department of Labor’s Employment and Training Administration. It carries structured profiles for over 900 occupations — tasks, work activities, work styles, work contexts and RIASEC interest codes. Because it is coded in RIASEC terms, matching becomes a scoring problem rather than an editorial one. Haliscope matches against the roughly three hundred occupations for which O*NET publishes every input the scoring needs.
The eleven Haliscope traits are our own, developed from both models rather than taken from either. The history of both threads is here.
2. The eleven traits
The eleven scales span both models in a single vocabulary. Each is reported as a position on a continuum, never as a category, and the order they are drawn in on your radar chart spells the product’s own name.
| Trait | What it measures |
|---|---|
| Operating | Measures your preference for hands-on, practical work. High scorers thrive in action-oriented roles with tangible outputs and prefer doing over theorizing. |
| Analyzing | Measures your drive to understand systems and solve problems. High scorers are knowledge-seeking, strategic thinkers who enjoy investigating and finding solutions. |
| Creating | Measures your inclination toward original thinking and artistic expression. High scorers generate innovative ideas and find rigid structure limiting. |
| Helping | Measures your orientation toward supporting and serving others. High scorers find fulfillment in teaching, counseling, and contributing to others' success. |
| Influencing | Measures your drive to lead, persuade, and shape outcomes. High scorers are confident communicators who naturally take charge and direct others. |
| Organizing | Measures your preference for structure, detail, and systematic processes. High scorers excel at creating order, optimizing workflows, and maintaining precision. |
| Exploring | Measures your openness to new experiences and curiosity about the unknown. High scorers seek novelty, enjoy experimentation, and value the journey itself. |
| Harmonizing | Measures your focus on group cohesion and interpersonal peace. High scorers build consensus, facilitate cooperation, and prioritize team harmony. |
| Persevering | Measures your persistence and resilience in the face of challenges. High scorers push through obstacles, maintain long-term focus, and value grit as both means and end. |
| Law-Abiding | Measures your commitment to ethics, rules, and fairness. High scorers value honesty, follow procedures, and prefer environments with clear moral guidelines. |
| Safety-Seeking | Measures your preference for stability and predictability. High scorers value security, prefer familiar environments, and carefully weigh risks before acting. |
Several traits draw on more than one source dimension. That is deliberate: the two models were never built to be orthogonal, and forcing them apart would give us cleaner scales that describe people worse.
3. What the assessment captures
Forty-five questions in four phases, about twenty minutes, in several formats. More than one format exists because each has its own failure mode, and using several reduces the chance that all of them fail in the same direction.
Written responses
Twenty-four of the forty-five, and the ones that carry the analysis. What you write and how you write it are both read.
Two timed exercises
A sixty-second and a twenty-second task. These measure what you do under a clock rather than how you describe yourself.
Rankings and rated choices
Ordering options against each other rather than agreeing with each in isolation, which blunts the pull towards whatever sounds best.
Résumé and background
Your actual history, used to ground the pathways in what you have already done. Describe it in your own words if you would rather not upload anything.
How long you spend on each answer and how often you revise it are recorded too. They do not move your trait scores — they feed the confidence attached to the result, so a rushed assessment is reported as a less certain one rather than a different one.
4. How scoring works
Each of the eleven traits carries a written definition, fixed in advance, versioned, and identical for everyone. Those definitions are the instrument. They say what counts as evidence for a trait and what does not — the definition of law-abiding, for instance, notes that nearly everybody describes themselves as honest, so saying so is not evidence.
Your completed transcript is then read against all eleven definitions in one pass, and a position from 0 to 1 comes back for each. There is no item-by-item scoring key underneath that, and no weighted sum of individual answers. We could describe one, and several products do; it would not be what this one does.
Scores are reported in five bands — Low, Below Average, Average, Above Average, High — at fixed cut-points on that scale. They are not percentiles against other people who have taken the assessment, and we don’t present them as such; section 7 says why.
5. What the AI does and doesn’t do
A model reads you. Arithmetic ranks the jobs.
What the model does: reads your transcript and produces the eleven scores; writes the prose of your report from them; chooses your three pathways out of the top thirty ranked careers; and evaluates any careers you name for yourself. It is the scorer, not a layer of decoration over one.
What it does not do: decide what each occupation requires, set the weights or tolerances used to match you against it, or rank the occupations. Every occupation’s eleven-trait profile is derived arithmetically from O*NET’s published interest and work-style data, with no model in the loop, and the ranking that follows is the same formula with the same numbers for everyone.
So the honest division is this: judgement about you is made once, against definitions you can read, and everything downstream of it is arithmetic you could reproduce.
6. Occupational matching
Six of the eleven traits map onto RIASEC’s interest areas and five onto named O*NET work styles, which is what makes an occupation profile and a person’s profile comparable at all. Every matchable occupation is then scored on four weighted components.
| Component | Weight | What it is |
|---|---|---|
| Personality | 0.50 | Your eleven scores against the occupation’s own eleven, as both the shape of the profile and its distance from it. |
| Skills | 0.25 | The occupation’s five most important O*NET skills, each stood in for by the traits that tend to go with it. Nothing here measures a skill directly. |
| Preferences | 0.15 | Physical demands against what you told us you want. The rest of this component is a constant today. |
| Market viability | 0.10 | The occupation’s outlook, its resilience to automation, and its median wage. A property of the job, not of you. |
Near-ties are common enough that an exact ordering would imply a precision the arithmetic doesn’t have, so matches are grouped into bands rather than presented as a leaderboard.
Three pathways are then chosen from the top of that ranking — a natural next step, the highest earning potential, and a bold move — because a single “best match” would imply the same false precision. You can also name careers of your own, and each is evaluated against your profile with the traits that support or work against it set out explicitly.
7. Known limitations
Seven, in descending order of how much they should affect the way you read your report.
Personality explains only part of career satisfaction
Pay, management, colleagues, commute, health and luck account for a great deal of whether work goes well, and none of them are measured here. Trait–outcome relationships in the literature are real but modest. Treat the report as one input, not a verdict.
A language model does the scoring
Your eleven scores come from a model reading your transcript against published definitions — held to them, but not replaced by arithmetic. That is a real dependency and we would rather state it than bury it. What we can say for it is that it is stable: re-running the same transcript moves a score by 0.02 to 0.05, against a spread between different people of 0.15 to 0.33.
There are no population norms yet
“High” and “Low” are fixed cut-points on the 0–1 scale, not percentiles against other people who have taken the assessment. Calibration so far rests on seven written personas rather than a user sample, which is enough to catch a systematic offset and nowhere near enough to describe a distribution.
Some of the matching weight barely varies
The preferences component moves within a few hundredths across the whole occupation pool, so most of its fifteen per cent cannot separate one job from another. The skills component infers aptitude from personality rather than testing anything. Both are known and tracked.
The weights are judgement calls
The split across the four components, and the split inside market viability, were chosen by us and have not been validated against outcomes. Because near-ties are common, matches are reported in bands rather than as an exact ordering.
O*NET is American and lags the market
Occupation profiles describe U.S. work and are updated on a rolling schedule rather than continuously. Emerging roles may be missing or described as they were several years ago, and salary figures are national.
It is not a clinical instrument
Haliscope measures normal-range personality and vocational interest. It diagnoses nothing, it is not validated for hiring or screening, and it should not be used to select or reject candidates.
If something in your report seems wrong, we want to hear about it — get in touch. Disconfirming cases are the only thing that improves an instrument.
8. Data handling
Your answers, anything you upload, the analysis drawn from them and your finished report are stored so that the report stays available to you in your account and so that the assessment can be improved. Your results are yours; they are not shown to an employer and Haliscope is not sold for hiring or screening.
You can have all of it deleted whenever you like — write to us and we act within thirty days. Some records, payments chiefly, we are required to keep.
We do not sell your personal data, and we do not sell the inferences we draw from it. Running the product does mean a small number of processors handle parts of it — payments, analytics, and the model that reads your answers — and the Privacy Policy names them, along with what is kept and for how long.
Now check it against yourself
Twenty minutes, and a free preview of your results before you decide anything. The FAQ answers the rest.
Start Assessment