Causal Diagnosis¶

← Back to Mechanisms by Solution Family

Solutions that distinguish symptoms from causes, compare explanations, localize a fault, or identify the intervention point responsible for an outcome.

17 mechanisms across 2 solution archetypes in this solution family. A mechanism inherits the primary family of the archetype it instantiates; family is about the move the solution makes, not the domain where it originated.

Multiple Causation and Explanatory Pluralism¶

Explain a complex outcome by coordinating multiple causal families and scales instead of reducing it to one master cause.

6 mechanisms · View full solution archetype

Causal Loop or Influence Diagram — Draws the causes as a network of signed arrows, gates, and feedback loops so interactions and cross-scale dependencies become visible instead of additive.
Cause-Role Worksheet — Forces each proposed factor to earn its place by pinning it to the bounded outcome and assigning it one defensible causal role — or striking it.
Counterfactual Sensitivity Probe — Removes, delays, or intensifies each factor in turn and asks whether the outcome would still hold, ranking causes by how much the result depends on them.
Cross-Disciplinary Causal Review — Convenes causal explanations from several disciplines or stakeholders on one outcome and fuses their partial accounts into a single weighted, uncertainty-marked synthesis.
Multicausal Factor Matrix — Lays every candidate cause into one grid — a row per factor, columns for family, scale, role, and weight — so the whole causal field can be compared at a glance.
Process-Tracing Evidence Table — Orders the case's evidence along its timeline and grades each piece by diagnostic strength, so a causal story must survive what actually happened, in sequence.

Residual-Driven Model Refinement¶

Subtract what the best current explanation predicts, then treat reproducible structure in the remainder as evidence about what the explanation still misses.

11 mechanisms · View full solution archetype

Autocorrelation and Whiteness Test — Checks whether residuals, read in order, are serially uncorrelated 'white noise'; leftover autocorrelation is evidence the model missed time- or sequence-dependent structure.
Control Chart on Residuals — Plots residuals over time against statistical control limits so a model that has drifted or broken shows up as an out-of-control signal, not a slow creep in average error.
Cross-Validated Error-Slice Report — Breaks out-of-sample error down by data slice and ranks it, so the segments where the model is quietly worst — invisible in the headline metric — become explicit targets.
Heteroscedasticity and Scale Test — Tests whether residual spread stays constant or grows with the fitted value or a predictor; scale-dependent variance means the model's error structure — not just its mean — is misspecified.
Influence and Leverage Diagnostic — Finds the individual observations whose presence most changes the fitted model — high-leverage, high-influence points — so a result resting on a handful of rows is exposed before it's trusted.
Model-Revision Experiment Log — A running record of every model revision — the residual pattern it targeted, the bounded change made, and whether held-out error actually improved — so refinement accumulates as evidence instead of drifting into overfitting.
Posterior-Predictive Residual Check — Simulates replicated datasets from the fitted model and asks whether the observed residuals look like data the model itself would produce.
Quantile-Quantile Residual Check — Plots ordered residuals against the quantiles of their assumed distribution, turning wrong tails and skew into a telltale bent line.
Residual Root-Cause Review — A structured review that works a flagged residual pattern through candidate causes with domain experts and commits to one bounded, testable model change.
Residual-versus-Fitted Plot — Plots each residual against the model's fitted value (or a predictor) so leftover curvature and changing spread show up as visible shape.
Subgroup Residual Heatmap — Tiles average residual across two crossed segmentations so a subgroup the overall fit hides lights up as a hot cell.