Spoke

performance-validity

Rating instrument diagnostics — Q1 convergence through Q6 change attribution plus MEI evidence bridges.

Character

You must prove whether appraisal forms behave like measurement instruments—not political theatre—with quantitative evidence ready for sceptical auditors.

Problem

External. Vendor “AI insights” trumpet predictive prowess without ΔR², reliability scaffolding, or change attribution artefacts.

Internal. Quarterly leadership reviews ambush you with survey complaints—yet spreadsheets cannot reproduce last cycle’s regressions reliably.

Philosophical. Ratings are measurements; diagnostics belong in substrate services alongside payroll APIs.

Guide

Stateless POST bundles assemble measurement-theory libraries: converge ratings vs milestones, multitrait-informed scorecards, stacked OLS incremental validity, trajectory slope diagnostics leveraging regression-to-mean awareness, PAT-159 calibration ROI overlays, paired change regressions plus structured UPSERTs funnelling explanatory mass into empirical MEI pillar recalibration. performance-validity honours PAT separation—call performance-calibration via HTTP summaries, never violating cross-spoke DB import bans while keeping MCP ergonomics symmetrical.

Abstract

Background. High-stakes merit and succession decisions presume instrument quality few organisations ever quantify reproducibly mid-flight.

Methodology. PAT-158 enumerates analytic question taxonomy from operational convergence diagnostics through explanatory decomposition, calibration ROI analytics requiring structured rollups, and change attribution regressions—all invoked through hardened Zod request contracts plus shared libs (measurement-theory, diagnostic-chain).

Scope. Computational service requiring caller-managed frames—not a survey platform capturing raw responses wholesale.

Contribution. MCP + HTTP parity exposes boutique consulting artefacts as typed automation—every call auditable similarly to MCP gate expectations.

Evidence / Provenance. PAT-158/159 README enumerates pairwise integration expectations with PAT-159 cycle summaries routed through HTTP—not SQL bridging.

Plan

  1. 01

    Pick analytic question cluster

    Dispatch the POST route aligning to Q1-Q6 hypotheses you need validated this sprint.

  2. 02

    Validate payloads

    Honour strict schemas so analytic columns decode predictably prior to regressions firing.

  3. 03

    Bridge empirical MEI pillars

    UPSERT pooled outcome summaries powering manager-effectiveness weight blends when empirical refresh approved.

  4. 04

    Narrate with calibration data

    Pair Q5 / Q6 responses with calibrated tuples from performance-calibration for holistic calibration ROI storytelling.

Call to Action

Direct. Inspect MCP descriptions for JSON exemplars gated by PAT-11 service keys ahead of scripted smoke runs.

Transitional. Digest PAT-158 memo enumerating methodological intent per analytic cluster.

Spoke I/O (visual language v1)

Every toolbox spoke shares the same abstract choreography: typed inputs on the left, distilled verbs in the center, typed outputs on the right, and (when relevant) cross-spoke HTTP composition along the bottom rail. Source package: @people-analytics-toolbox/spoke-illustrations.

Performance validityINPUTSMAIN ACTIONSOUTPUTSInstrument schemaRatingInstrumentRespondent ratingsRatingMatrixFit measurement modelsAttribute change + predictionsReliability + validityMeasurementReportPredictability surfacesOutcomeLinkageCardChange attributionDeltaDecompositionCOMPOSES WITHperformance-calibrationreincarnation

Try it now

Copy this curl. Paste in any terminal. Public read — no auth needed.

performance-validity.health

GET

POST diagnostics require payloads + PAT-11 key; heartbeat proves schema reachability for planning.

curl -sS "https://people-analytics-toolbox.vercel.app/api/spokes/performance-validity/health"

Vendor the contract

The Zod contract is the source of truth. Vendor a copy into your consumer app — you keep it; we don't break it underneath you. Re-vendor when the version bumps.

// Vendor canonical types:
// src/spokes/performance-validity/contracts/types.ts

Source path: src/spokes/performance-validity/contracts/types.ts · GitHub

Try in Dev ConsoleRead the explainerBack to all spokes

Failure

Legal or union scrutiny dismantles brittle rating narratives—you lack regression-ready diagnostics under subpoena timelines.

Success

Each performance cycle emits refreshable dossiers coupling psychometric vigilance with business outcome transparency.