Validation philosophy
We prioritize construct clarity, reliability, criterion validity, and longitudinal sensitivity. Methods include classical test theory and IRT as appropriate, with transparent reporting.
Science
Fairness & bias monitoring
We evaluate differential item functioning and adverse impact, and monitor model‑mediated dialogue for biased suggestions.
Privacy & data minimization
Only essential data are collected for scoring and feedback. De‑identification and aggregation are standard for research metrics.
Transparent roadmap
Planned disclosures include measurement invariance studies, technical manuals, and release notes for any model changes affecting scores.