PA Instruments

The measurement & decision primitives

A is one composable measurement or decision primitive, exposed as a single standalone tool. They are the ingredients: each does one thing well, statelessly, and can be called on its own or wired together. The toolbox's products — the meals a buyer meets — are built by composing them.

The ingredients

Thirteen instruments across five spokes today. Each links to its spoke's explainer.

Preference Modeler

explainer →
  • MaxDiff — generate

    Build a deterministic set of best/worst choice tasks from a list of items.

    preference-modeler.maxdiff.generate

  • MaxDiff — score

    Turn best/worst responses into per-item and per-variable preference weights.

    preference-modeler.maxdiff.score

  • Present↔future gap

    Score the gap between where something is today and where you want it — the change intent.

    preference-modeler.present-future

Performance Validity

explainer →
  • Rater alignment

    Measure how closely raters converge, item by item, on the same target.

    performance-validity.alignment

  • Directional alignment

    Score a leader's alignment with those above, below, and beside them.

    performance-validity.directional-alignment

  • Distribution fit

    Fit a statistical distribution to a sample and report how well it actually matches.

    calculus.distribution-fit

  • Importance reconcile

    Reconcile what people say matters against what the data implies matters.

    calculus.importance-reconcile

Forecasting

explainer →
  • Interval scoring

    Grade past interval estimates against what actually happened — coverage and sharpness.

    forecasting.interval-scoring

  • Bayesian combine

    Merge several independent estimates of one quantity into a single calibrated posterior.

    forecasting.bayesian-combine

  • Measurement recommend

    Given a thing to measure and your constraints, recommend how to measure it.

    forecasting.measurement-recommend

  • Measurement catalog

    The menu of available measurement methods and the ranges they produce.

    forecasting.measurement-catalog

Factor Models

explainer →
  • Analytics plan

    Rank models by priority relevance, then order measurement by value of information.

    factor-models.analytics-plan

  • Empirical importance

    Derive each item's importance from the variance of its observed outcomes.

    factor-models.empirical-importance

Two instruments worth naming for what they unlock: (only gather data that would change the decision) and (merge many estimates into one calibrated answer). The preference instruments use and capture; the fit instrument reports .

Composed into products

The same instruments, wired together over typed contracts, become the products a buyer actually meets.