A bigger bank is not automatically a better instrument
PsiMatch has an active MindSync core and a larger inactive extended bank. The extended bank exists to improve coverage and support future experiments; its size alone is not evidence that it predicts relationship outcomes.
We keep that distinction visible because a long questionnaire can create false confidence just as easily as a short one.
Start with a construct and a boundary
Each item begins with a defined compatibility concept, such as a Big Five facet, relationship context, emotional-intelligence tendency, values pattern, or everyday lifestyle rhythm.
The authoring record also states what the item cannot establish. A question about conflict repair can describe a self-reported tendency; it cannot diagnose attachment or guarantee how someone behaves under pressure.
Route every answer to a named namespace
Question identifiers carry their destination namespace. Prefixes distinguish Big Five facets, relationship context, emotional intelligence, values, safety, and other product-authored domains before the server aggregates responses.
This prevents a new item from silently disappearing into a legacy catch-all category and lets the product show where confidence is strong, partial, or still developing.
- Forward-scored Likert items use a consistent 1–5 direction.
- Facet and domain names remain attached through aggregation.
- Raw answers stay private and are not exposed as profile content.
Review for ambiguity, redundancy, and harm
Before a batch is merged, items are reviewed for unclear wording, double-barrelled claims, unnecessary reverse framing, construct-irrelevant detail, and semantic redundancy.
Safety and relationship-context items receive additional framing so the result is not presented as a clinical diagnosis, a moral judgment, or a declaration that someone is safe.
Keep inactive items inactive until the product is ready
The extended bank is currently inactive for the core MindSync instrument. It can support targeted micro-questions and agent-guided completion without silently changing the scoring contract for everyone.
That separation gives us room to inspect coverage and authoring quality before changing the primary assessment experience.
What this work does and does not prove
The current bank demonstrates a disciplined product-authoring and routing process. It does not constitute a validated clinical instrument, establish measurement invariance across cultures, or prove that a particular score predicts a lasting relationship.
The useful output is a clearer starting hypothesis: what may be shared, where differences could matter, and what two people should ask each other next.