Assessment scoring methodology

This page answers the question we get most often about the Yareta Assessment: how do individual question answers become a trait score? It covers the response scale, which items feed which trait, and — because it is the most common misconception — why a trait score is not a simple average of its questions.

The building blocks: items

Every question in the Yareta Assessment is called an item. Each item belongs to exactly one of 26 scales:

25 personality traits — the sub-traits shown on a profile (Obsession, Integrity, Risk Tolerance, and so on)
1 validity scale (“Sensitivity”) — a social-desirability check described below, which is not one of the 25 reported traits

There are 127 items in total across those 26 scales, ranging from 4 to 6 items per trait:

Trait	Items	Trait	Items
Obsession	6	Positivity	5
Constant Desire to Succeed	4	Magnetism	5
Continuous Learning	6	Breaking Boundaries	6
Growing Through Adversity	6	Resourcefulness	5
Desire to never give up	4	Power	6
Integrity	5	Self-Confidence	4
Ownership	4	Collaboration	5
Attention to Detail	5	Embracing Differences of Opinion	4
Comfortable being Uncomfortable	5	Patience with Others	5
Drive to Win	6	Empathy	5
Decisiveness	4	Self-Awareness	4
Intuition	4	Sensitivity (validity scale, not a trait)	4
Risk Tolerance	5

How an answer is recorded

Every item is answered on the same 1–6 scale (a 6-point Likert scale — there is no neutral midpoint, which forces a lean toward agree or disagree). Two adjustments happen to a small number of items before anything is scored:

Reverse-scored items — a handful of items are phrased so that agreeing indicates a lower level of the trait. Those raw answers are flipped (7 − answer) before scoring, so higher always means “more of the trait” across the whole instrument.
Response-collapsed items — a couple of items collapse the bottom two response options into one category, because respondents rarely used the full 1–6 range on those specific items in calibration data.

Both adjustments are applied automatically; a respondent never sees them — they simply answer 1–6 on every item.

How a trait score is calculated (it is not an average)

This is the part people most often assume works like trait score = average of its N items. It does not — for a good reason: a plain average treats every item as equally informative and every respondent’s answer pattern as equally reliable, and it can’t be compared meaningfully across people who saw items of different difficulty.

Instead, each trait is scored with a psychometric model called a Graded Response Model (GRM) — a standard, published technique for instruments with ordered response options (used widely in validated personality and aptitude testing, not something Yareta invented from scratch):

Each item was calibrated in advance against a reference sample, giving it two properties: how strongly it discriminates people at different levels of the trait, and where along the trait scale each response threshold (1→2, 2→3, and so on) sits. Some items are simply more informative than others, and the model already knows this.
A respondent’s whole pattern of answers on that trait’s items is combined into a single estimate of their underlying trait level, using those item properties — an approach called Expected A Posteriori (EAP) estimation. This is why two people who both averaged, say, 4.3 out of 6 on the same items can still end up with different trait scores: the pattern of which items they agreed with most strongly matters, not just the arithmetic mean.
That estimate is standardized against a reference sample (converted to a percentile via the normal distribution) and reported on a 0–100 scale, so trait scores are comparable to each other and to industry/unicorn-founder benchmarks.

In short: conceptually similar to “the questions for this trait roll up into one number,” but the roll-up is a calibrated statistical model, not a simple average — that is what lets scores stay comparable across respondents and across the two instrument variants Yareta uses (see Yareta Assessment for the 127-item developer API version).

From 25 traits to 6 domains

The 25 trait scores are then combined — again through a weighted model, not a simple average — into the 6 higher-order domains shown on a Yareta DNA profile: Discipline, Execution, Grit, Self Belief, Leadership, and Vision & Purpose. Several traits contribute to more than one domain, reflecting that traits like Resourcefulness or Risk Tolerance are relevant to more than one higher-order construct.

The validity scale (“Sensitivity”)

Four items belong to a Social Desirability scale, shown as “Sensitivity.” It is not one of the 25 reported personality traits — it exists to flag response patterns that look like impression management (answering to look good rather than answer honestly), the same way most validated personality instruments include a lie/consistency check. It is excluded from the 25 trait scores and from the 6 domain scores.

What this page does not cover

This page describes how an answer becomes a score. It does not publish the exact item wording, the calibrated item parameters, or the domain weighting coefficients — those are the proprietary calibration behind the instrument (the same way a published test’s exact item bank and scoring key aren’t usually public alongside a description of the method). If your use case needs that level of detail (for example, mapping Yareta trait scores against an external dataset), talk to your Yareta contact about what can be shared under your agreement.

Yareta Assessment — what the instrument is, how it’s delivered, and the API for embedding it
Success Frameworks — how the 25 traits and 6 domains feed into the six components of a framework
How scoring works — the end-to-end scoring pipeline this instrument feeds into
Glossary — quick definitions of terms used here