What this competency actually requires

Sensitivity Analysis is the professional capacity to measure how model outputs and decisions respond to uncertain inputs, structural choices, and scenarios. Competence is visible when a practitioner can explain where the evidence came from, why the method fits the decision, which assumptions carry the conclusion, and how uncertainty should affect action.

The intellectual foundation includes Saltelli’s global sensitivity framework and Morgan and Henrion’s treatment of uncertainty in policy models (Saltelli et al., 2008) (Saltelli et al., 2020).

Research foundations

Good analysis begins before a model is selected. A reproducible workflow therefore preserves source data, documents transformations, versions code or formulas, and separates exploratory work from confirmatory claims (Morgan & Henrion, 1990).

Uncertainty has several layers. Sampling error is only one. They report practical magnitude and plausible ranges, examine exceptions, and identify conditions under which the recommendation would change (Sobol, 2001).

Ethical reasoning is part of analytical quality.

A defensible working method

Identify the decision output, uncertain inputs, plausible ranges, dependencies, and model alternatives. Begin with screening, then use local or global methods appropriate to nonlinearity and interactions. Present thresholds at which the preferred choice changes, not merely a tornado chart of parameter movement.

Before delivery, another qualified person should be able to reproduce the main result and challenge the strongest assumption. It should state limitations in language proportionate to the decision rather than burying them in technical notes.

1. Sensitivity Analysis framing: define the decision, outcome, unit of analysis, time horizon, and cost of error.
2. Audit: verify provenance, definitions, permissions, missingness, and measurement quality.
3. Analyze: establish a baseline, choose methods by assumptions, and retain a transparent trail.
4. Challenge: test alternatives, edge cases, sensitivity, and subgroup performance.
5. Act and learn: assign ownership, monitor consequences, and revisit the model when conditions change.

Failure modes and professional judgment

Varying one input at a time can miss interactions and impossible combinations, while arbitrary plus-or-minus percentages create a comforting picture unrelated to real uncertainty. In Sensitivity Analysis, these defects can redirect resources, obscure risk, and distribute harm.

A mature practitioner actively searches for disconfirming evidence. When evidence is weak, the correct output may be a smaller decision, a reversible pilot, better measurement, or no recommendation yet (Pannell, 1997).

Credibility grows from traceability rather than certainty theater.

Evidence of mastery

Evidence of Sensitivity Analysis mastery includes a well-scoped question, data dictionary, assumption register, reproducible analysis, appropriate validation, plain-language decision memo, and post-decision monitoring plan. It also shows restraint: the practitioner knows which conclusions are supported, which are provisional, and which remain outside the evidence (Briggs & Sculpher, 2006).

For leaders, the practical test is whether the competency improves decisions over repeated cycles.

Questions for review

Before approving work in Sensitivity Analysis, reviewers should ask: Is the decision question explicit? Are the data and definitions fit for that question? Could another analyst reproduce the result? Were reasonable alternatives tested? Are uncertainty and limitations visible where the recommendation appears? Is the action proportionate to the strength of evidence? For Sensitivity Analysis, these questions turn quality assurance into a practical discipline rather than a ceremonial sign-off.

The review record should preserve the original question, source dates, exclusions, transformations, analytical choices, contrary findings, reviewer comments, and final disposition.

Evidence and operating controls

Sensitivity analysis investigates how uncertainty in inputs, assumptions, and model structure changes outputs or decisions. It is more than increasing every forecast by ten percent. Local one-at-a-time changes describe behavior near a base case but can miss interactions and nonlinear thresholds. Global methods explore inputs jointly across defensible ranges and can apportion output variance when assumptions permit (Saltelli, 2008). Define uncertainty distributions or bounds from evidence, distinguish uncertainty from deliberate scenarios, and include structural alternatives where the model form is disputed. The central question is often whether the recommended action changes, not merely how many decimals the output moves.

Applied review protocol and limitations

Build an assumption register with source, range, dependence, owner, and update trigger. Screen inputs, then examine the influential variables and interactions with appropriate sampling. Plot response shapes and decision-switching boundaries; verify numerical convergence and rerun with alternative plausible model structures. Communicate which uncertainties are reducible through research and which reflect irreducible variation or values. Results are only as credible as the ranges and dependence assumptions. Treating correlated inputs as independent can create impossible cases, while narrow expert ranges can manufacture robustness. Do not vary convenient numbers while holding contested policy assumptions fixed.

Worked decision test

Decision test: rank assumptions by their effect on the actual choice, not only on the model output. For the most influential assumption, identify whether additional measurement could narrow uncertainty and calculate whether that information is worth obtaining before acting. If the decision changes repeatedly across plausible combinations, recommend a reversible option, staged commitment, or explicit risk acceptance instead of reporting the base case as robust.