Causal Map¶
Artifact — instantiates Relation Mapping
Diagrams hypothesized or validated cause-and-effect edges among factors, each carrying its evidence basis, a confidence label, and the conditions under which it holds — so plausible-looking arrows cannot pass as proven ones.
A Causal Map is a diagram whose edges are claims that one factor produces a change in another — X drives Y — and whose defining discipline is that each such claim must carry the evidence it rests on and a mark of how much confidence that evidence warrants. Any relation map can draw an arrow; a causal map refuses to let an arrow mean "causes" unless it can say why it believes so. Correlation, temporal sequence, and a compelling story are explicitly not enough. It is the one relation-mapping mechanism whose central anxiety is being wrong about direction and mechanism, so alongside each edge it labels confidence and records the regime or window within which the causal link even operates.
Example¶
A bottling plant sees its bottle-seal defect rate jump. A quality engineer builds a causal map of candidate drivers: a new resin supplier, ambient humidity, a recently rotated operator, and machine-head wear, each drawn as an arrow into "seal defect rate." Rather than trust the tidiest story, she tags every edge with its basis. The humidity → defect edge earns high confidence — a bench test reproduces it — and it is marked to hold only during the summer heat wave (a temporal window: the edge is live June–August, dormant otherwise). The new-supplier → defect edge is downgraded to "suspected," because although defects rose the week the supplier changed, that is sequence, not evidence, and the hold-batch test has not been run. The operator edge is struck out once the data show the defect spans all shifts. The map's payoff is that it stops the plant from tearing out the new supplier — the plausible culprit — and points instead at a dehumidifier fix, because the evidence, not the narrative, carried the weight.
How it works¶
- Draw candidate cause → effect edges among the factors, each initially a hypothesis.
- Attach a basis to every edge — controlled experiment, natural comparison, known mechanism, expert judgment, or bare correlation. The source is mandatory, not a footnote.
- Label confidence and sign per edge, distinguishing validated from merely hypothesized, and marking whether the effect is positive or negative.
- Record the validity window — the regime, season, or dose range in which the cause actually operates — so a conditional cause is not read as a permanent law.
- Prune ruthlessly. An edge resting only on correlation or sequence stays dashed, or comes off the map.
Its distinguishing move against a plain graph: an edge is provisional until evidenced, and it can be reopened when the regime changes.[n1]
Tuning parameters¶
- Evidence bar — how strong the support must be before an edge is drawn solid rather than dashed. A high bar yields a sparse, trustworthy map; a low bar, a rich but speculative one.
- Confidence scale — binary (hypothesized / validated), ordinal, or probabilistic. Finer scales carry more but tempt false precision.
- Effect detail — sign only (+/−) versus magnitude or elasticity on each edge.
- Validity-window granularity — always-on edges versus conditional, seasonal, or dose-dependent ones.
- Feedback allowance — permit loops (reinforcing and balancing) or restrict the map to acyclic causal chains.
When it helps, and when it misleads¶
Its strength is that it forces the difference between "we see these move together" and "we have reason to believe one moves the other," and it keeps the conditions of the claim attached rather than lost.
Its failure mode is that causal edges are the easiest to assert and the hardest to prove: a confident-looking arrow can encode a bias, a reversed direction, or a lurking common cause, and the diagram makes any of them look established. The classic misuse is building the map to justify an intervention already chosen — motivated causal reasoning dressed as analysis — or reading a reinforcing loop as destiny. The guarding discipline is to keep hypothesized edges visibly dashed until they are tested, label confidence honestly rather than optimistically, and reopen an edge when the regime its validity window names has changed.
How it implements the components¶
evidence_and_source_basis— every edge names how the causal claim is known; this evidentiary demand is the map's whole reason for being.relation_strength_indicator— confidence and uncertainty labels, plus edge sign and magnitude, separate a strong validated link from a weak suspected one.temporal_validity_window— each edge records the regime, season, or dose range within which the cause operates, so a conditional cause is not mistaken for a standing one.
A causal map asserts and evidences cause → effect; it does not record atomic reliance links or trace their blast radius — the relation_instance_link and consequence_annotation machinery of downstream impact belongs to Dependency Map, its nearest twin. One line apart: a dependency map says "A relies on B" and asks what breaks if B fails; a causal map says "A changes B" and must defend that it truly does.
Related¶
- Instantiates: Relation Mapping — the causal variant, which adds evidence discipline to the drawn edge.
- Sibling mechanisms: Dependency Map · Relationship Graph · Ownership Map · Stakeholder Map · Adjacency Matrix · Service Dependency Inventory · Knowledge Graph · Data Lineage Map · RACI Matrix
Editorial Notes¶
Form Classification¶
Form family: Representation, Specification & Plan
Rationale: The mechanism records candidate cause-effect edges with evidence basis, sign, confidence, and validity window in a diagram, so its operative form is a causal evidence map.
Nearest alternative: Analysis, Modeling & Optimization — Analytic judgment grades edges, but the persistent graph externalizing those claims is the mechanism.
Review outcome: Adjudicated after independent review; high confidence.
Origin Attribution¶
Primary origin: Systems Thinking & Cybernetics
Origin pattern: Cross-disciplinary synthesis
Present-day reach: Multi-domain
Rationale: Systems thinking and cybernetics established maps of directed influences among interacting factors, including signed effects, feedback, boundaries, and regime-dependent behavior.
Related originating lineages:
- Philosophy — Philosophy of causation contributes the distinction between plausible, assumed, and warranted causal claims and their validity conditions.
- Statistics & Experimental Design — Causal graphical modeling contributes formal directed-edge semantics, identification assumptions, and evidence-based causal inference.
Review resolution: The reviewers split between causal statistics and systems cybernetics. Systems sources establish causal maps and loops as system representations, while Pearl's work gives causal diagrams formal statistical semantics. Because the mechanism includes feedback and signed, regime-bound edges, systems practice governs; evidence and confidence per edge make the result a synthesis with statistics and philosophy.
Attribution caveat: A causal DAG is a strong alternate ancestor, but this mechanism permits feedback, sign, and operating regimes rather than requiring acyclicity; its mandatory evidence and confidence labels are a synthesis layered onto the systems map.
Encyclopedia synthesis: The exact catalogued form synthesizes established practice rather than reproducing a single standard historical label.
Review outcome: Researched adjudication after independent review; high confidence.
Sources consulted:
- UK Government: Systems Thinking Toolkit
- WHO Alliance: Course Package on Causal Loop Diagrams
- Judea Pearl: Causal Diagrams for Empirical Research
Notes¶
[n1] Modern causal inference — notably Judea Pearl's causal graphical models — makes precise the old warning that correlation does not imply causation: a diagram licenses a causal reading only under stated assumptions about what else was held fixed. A causal map that omits its evidence basis is exactly the diagram that framework warns against. ↩