Diagnostic scoring formulas, examples, and limits

The instrument is intended for executives and senior leaders describing current organizational conditions from their own perspective. It does not independently observe behavior or verify evidence. Role, visibility, recall, response style, recent events, and social desirability may affect results.

The bank contains 114 questions across eight domains. A session selects 35–50 questions; the default is 40. After sector and organization-size eligibility rules, the selector prioritizes weighted construct coverage, adds value for at least one full-weight primary item per construct, and then balances domains. It interleaves selected questions across domains.

Different respondents may receive different item sets. Scores can be compared descriptively, but no equivalence study has established that every generated form has the same measurement properties.

A metric takes the equal mean of its observed input construct scores. Unobserved constructs are excluded. Item weights affect the construct scores beneath a metric, but input constructs are equally weighted in the metric mean.

A metric unlocks only when its question-number threshold and minimum observation count are met and every observed input construct has at least 1.0 weighted evidence. An unlocked metric is marked provisional when any observed input has less than 3.0 weighted evidence.

Bands and unlock thresholds are design rules. They were not derived from outcome data, normative samples, sensitivity or specificity analysis, or externally observed breakpoints.

An unanswered item contributes nothing. An entirely unobserved construct is excluded from a metric rather than assigned zero. Because inputs enter as they are observed, a partial metric can change composition during the assessment.

Evidence sufficiency is answer coverage, not statistical confidence. The session labels are Early for 0–9 answered questions, Developing for 10–20, Established for 21–34, and Complete for 35 or more. “Complete” does not mean definitive, accurate or validated.

Evidence fill uses weighted input evidence, not raw observations: min(weighted input evidence ÷ (minimum observations × 2), 1). A metric can display a high overall fill and still be provisional if one observed input construct is thin.

The separate self-directed 24-item Organizational Readiness Index offers “Insufficient evidence.” It excludes that response from condition means and requires at least three substantiated items in a condition before displaying its score. Read the ORI scoring note.

The self-diagnostic does not estimate a causal constraint. It ranks unlocked, recommendation-eligible metrics other than Builder Index. Healthy-oriented metrics use (maximum − score) ÷ range; inverted metrics use (score − minimum) ÷ range. The greatest normalized severity supplies the “Address First” recommendation.

Unlock order can influence a partial result because a metric cannot be ranked before it unlocks. The normalization also assumes positions on different scales are comparable. That is a design assumption, not an empirically established property.

“Primary constraint” therefore means indicated area to investigate first, not proof that the metric caused an outcome or that improving it will produce the greatest effect.

Team reports require at least three respondents. For each unlocked metric, the engine reports the arithmetic mean, minimum, maximum, range, and population standard deviation across available respondent scores. It also counts provisional contributions.

A divergence zone is flagged when the observed range exceeds 1.5 points on a 1–5 metric, or the equivalent 37.5 points on a 0–100 index. This is a design rule, not an empirically validated threshold. Divergence describes disagreement; it does not establish its cause.

A team-diagnostic participant may choose, after submission, to contribute answers to instrument research. Refusal has no cost. Browser-only instruments do not make this offer because their answers are not sent.

A contributed record excludes direct identifiers such as name, email address, personal title, organization name, team identifier, and response identifier. It retains item responses, instrument id and fingerprint, broad organization and role categories, elapsed time, contribution month, and an opaque organization key. This design reduces linkability; it does not claim that re-identification risk is zero in every context. Retention is five years, and the one-time withdrawal code deletes the record when presented.

Collection makes future analysis possible; it does not itself validate the instrument. No claim on this page changes until an analysis is completed and reported with its sample, methods, date, and instrument fingerprint.

Automated tests establish that the software follows its specified scoring, selection, and aggregation rules. They do not establish measurement validity.

No published study has yet established internal consistency, test–retest reliability, inter-form equivalence, factor structure, convergent or discriminant validity, criterion validity, predictive calibration, measurement invariance, normative bands, or meaningful-change thresholds for this instrument. Until that work is completed, every output should be treated as developmental and directional.

The frozen fingerprint is a hash of every item's stem, construct tags, and five behavioral anchors. The public validation status names each planned analysis, minimum sample, and result that would count against the instrument.

This documentation follows the principle that interpretation must be tied to purpose and evidence. See the APA Guidelines for Psychological Assessment and Evaluation.