How it works
How Relatrix works and where its claims stop
Relatrix is a stateless calculation core. Consumers run assessments and keep their data; Relatrix supplies versioned content and calculates self-describing results from what it is given.
From responses to results
Versioned content
A consumer resolves an assessment from the model catalog: Item Tasks, response formats and localized presentations, each pinned to an exact version and digest. The consumer administers it and keeps the responses.
Profile
Responses are scored with versioned classical weighted scoring and normalized to a scale from 0 to 1. Each dimension estimate carries its measurement uncertainty, evidence coverage, evidence quality and origin.
Exact Fit
Two profiles are evaluated in one of seven Fit Contexts: marriage, dating, long-term relationship, cohabitation, co-parenting, friendship and colleagueship. The Fit Model defines which dimensions matter, how they interact and in which direction.
Retrieval projection
The same context yields query and candidate vectors. A consumer can index them to shortlist profiles by distance and then evaluate each pair with Exact Fit. How closely vector distance follows the Exact Fit order is measured for every context.
Self-describing results
Every result names its exact model set, lifecycle, validation applicability, evidence status, claim level and warnings. The same input and model version always produce the same output.
Scientific claim boundaries
These limits are part of every result. The interface and the API keep them visible; they are not fine print.
Estimates, not verdicts
A profile is an estimate with stated measurement uncertainty. When evidence is too sparse, a dimension is reported as unavailable instead of being guessed.
Experimental models
All current models are experimental, so their validation applicability is not evaluated. A model becomes active only after source review, reliability, validity, fairness and locale adaptation evidence.
Comparisons across groups
People who answered in different languages or belong to different populations are compared with the same mechanism. Until measurement invariance has been shown, such results carry a cross-group comparison warning.
Declared intended use
Results are produced for a declared intended use: self-insight, dyadic insight, support for a consumer’s matching suggestions, or research. High-stakes uses such as employment selection are excluded, and the final decision always stays with people.
Retrieval is a shortlist
Vector distance helps find profiles worth evaluating. The fit of a pair is defined by Exact Fit, not by vector similarity.
Relatrix-native derived views
Derived views are Relatrix’s own artifact-defined interpretations. They claim no equivalence with external typologies or instruments and never replace the canonical profile.
A fit index is not a probability
A compatibility value such as 0.78 is a computed fit index on a stated scale. It does not mean a 78 percent probability or chance of a successful relationship, and Relatrix makes no calibrated prediction.
Sources and licenses
Every model version records its sources with exact editions, citations and usage rights. The main external foundations are:
IPIP–HEXACO scales
Ashton, M. C., Lee, K., & Goldberg, L. R. (2007). The IPIP–HEXACO scales: An alternative, public-domain measure of the personality constructs in the HEXACO model. Personality and Individual Differences, 42(8), 1515–1526. https://doi.org/10.1016/j.paid.2006.10.027
Public-domain items of the International Personality Item Pool. The proprietary HEXACO-PI-R inventory is not used.
IPIP Interpersonal Circumplex (IPIP-IPC)
Markey, P. M., & Markey, C. N. (2009). A brief assessment of the interpersonal circumplex: The IPIP-IPC. Assessment, 16(4), 352–361. https://doi.org/10.1177/1073191109340382
Public-domain items of the International Personality Item Pool.
Refined theory of basic individual values
Schwartz, S. H., Cieciuch, J., Vecchione, M., Davidov, E., Fischer, R., Beierlein, C., Ramos, A., Verkasalo, M., Lönnqvist, J.-E., Demirutku, K., Dirilen-Gumus, O., & Konty, M. (2012). Refining the theory of basic individual values. Journal of Personality and Social Psychology, 103(4), 663–688. https://doi.org/10.1037/a0029393
Relatrix-native Item Tasks operationalize the theory; the items of its published questionnaires are neither copied nor paraphrased.
O*NET® data
This page includes information from the O*NET 31.0 Database by the U.S. Department of Labor, Employment and Training Administration (USDOL/ETA). Used under the CC BY 4.0 license. O*NET® is a trademark of USDOL/ETA. Relatrix has modified all or some of this information. USDOL/ETA has not approved, endorsed, or tested these modifications.
This page includes information from the O*NET 30.2 Database by the U.S. Department of Labor, Employment and Training Administration (USDOL/ETA). Used under the CC BY 4.0 license. O*NET® is a trademark of USDOL/ETA. Relatrix has modified all or some of this information. USDOL/ETA has not approved, endorsed, or tested these modifications.
The Work Styles taxonomy of the O*NET 31.0 Database and the Work Values taxonomy of the O*NET 30.2 Database inform the colleagueship Item Tasks as construct theory; occupation ratings are never used as scores or norms for people.
The public-domain status of an item pool does not validate Relatrix scoring, fit or derived interpretations; each of them needs its own evidence.