The approach, and the papers behind it.

BFI-2 at facet resolution through Rasch analysis (BFI-2-R)

Rasch validation of the Big Five Inventory-2’s 15 facets in 5,362 Italian adults — facet scores can satisfy fundamental measurement properties.

HEXACO honesty-humility is separable from Big Five agreeableness

Four German samples (N=3,976) find two-factor models fit better; latent ρ=.33-.51 and incremental validity favor a distinct honesty-humility dimension.

Detecting construct-irrelevant variance with network psychometrics (HEXACO-PI)

Exploratory Graph Analysis + Unique Variable Analysis flagged 110 CIV items on the 240-item HEXACO-PI; 89% overlapped with traditional methods.

AAS-9: a two-factor attachment model for emerging adults (anxiety / avoidance)

In 307 US emerging adults 18-29, CFA selected a two-factor model (anxiety α=.90, avoidance α=.74) over three- and four-factor alternatives.

Attachment styles out-predict Big Five for personal network structure

Secure attachment → dense, supportive, noncentralized networks; avoidant → fragmented, less supportive. Attachment explained more variance than the Big Five.

Big Five facet-level items inspired by public-domain IPIP patterns

How we author 1-5 Likert items per facet, route by id prefix to trait aggregates, and what the BFI-2-R Rasch validation says about facet-level scoring.

Schwartz refined values: 19-value structure

From coarse 6-bucket to the contemporary 19-value model. Why finer distinctions matter for dating compatibility.

HEXACO honesty-humility: not just agreeableness

Across four German samples (N=3976), honesty-humility and agreeableness are correlated but distinguishable — and they increment each other in predicting outcomes.

Attachment — AAS-9 two-factor model (anxiety / avoidance)

Two-factor CFA wins for emerging adults (N=307, TLI .985, CFI .993, RMSEA .044). We mirror this structure in our relationship context items.

ECR-R conceptual basis for relationship-context items

Experiences in Close Relationships — Revised provides the conceptual scaffolding for our relationship-context items (not a verbatim validated scale).

Gottman — bids, repair, and the four horsemen

Bids → responses and repair attempts predict relationship outcomes better than trait similarity alone. Our relationship items probe biddability and repair.

EQ — ability vs. trait distinction

Mayer-Salovey-Caruso ability EQ and trait EQ (Bar-On, Petrides) measure different things. Our items lean toward self-reported trait-EQ with explicit provenance.

Safety, consent, and red-flag recognition — community framework

Our safety items draw from community-standard consent frameworks (enthusiastic consent, revocability, coercion-spectrum). Not a single validated scale; aggregated from healthy-relationship curricula.

Product-authored items — our methodology for novel constructs

Some items probe constructs that no public-domain validated scale covers well (lifestyle compatibility, daily-rhythm fit, shared-goal timing). We author them explicitly and label the provenance.

Psychometric Research Deep Dive: Facets, Routing, and Uncertainty

A public explanation of the evidence, routing choices, and limitations behind PsiMatch compatibility signals.

What Agent-Driven MindSync Completion Looks Like

A consent-first walkthrough of how an MCP agent can identify unanswered MindSync areas, ask the next questions, and submit only the user’s answers.