Methodology
How the number gets made.
A psychometric score is only worth as much as the process behind it. This is that process for both instruments, including the parts that are still in progress.
- Instruments
- 2
- Questions
- 221
- Reverse-keyed
- 102
- Sub-constructs
- 67
- Population norms
- Not yet
- 01
Items, not questions
Each construct is measured by many short items rather than one long question. A single misread item cannot move a score, and the spread across items is itself informative. The entrepreneurial assessment uses a 5-point agreement scale across all 114 items; the yacht assessment mixes four formats.
- 02
Half of it written backwards
Reverse-keyed items state the opposite of the trait, so agreement counts against the score. 50 of the entrepreneurial items and 52 of the yacht items are keyed this way. A respondent agreeing with everything to look good produces a profile that contradicts itself.
- 03
Scored to sub-constructs
Headline construct scores are built from sub-constructs: 41 of them in the entrepreneurial assessment, 26 in the yacht assessment. This is why a report can say which part of resilience is thin rather than just that resilience is low.
- 04
Checked for whether they engaged
Alongside the trait scores, responses are checked for internal consistency. A respondent who clicked through without reading produces a pattern that does not hold, and the report says so rather than averaging it into a plausible-looking score. The mechanics of that check are deliberately not published. A respondent who knows exactly what is being compared can defeat it.
- 05
Read against thresholds, for now
Scores are currently interpreted against absolute thresholds set from the research the constructs are drawn from. Norm-referenced percentiles require a completed-assessment population we are still building. We would rather say that than publish percentiles derived from too few people.
Provenance
Where the constructs come from.
Both instruments draw their constructs and item styles from established psychometric research rather than being written from scratch. The yacht instrument was additionally cross-reviewed across multiple independent analyses before the item set was fixed, specifically to catch items that read differently to a second-language respondent.
Item wording follows a hard constraint on reading level: single-clause sentences, no phrasal verbs, no double negatives. An assessment that measures reading comprehension instead of the trait is worse than no assessment.
Limits
What we will not claim.
- No published external validation study. The constructs are grounded in prior research; the assembled instruments have not been independently validated.
- No predictive accuracy figure. Anyone quoting one on a population this size is extrapolating.
- No population norms yet. Thresholds are absolute until the completed-assessment sample supports percentiles.
These are the honest limits of a young instrument. They narrow as more assessments complete, and we will update this page rather than quietly changing the claim.