Sensitivity Analysis

Sensitivity analysis examines how model outputs or decisions change when inputs or assumptions vary. It may consider local changes, broad ranges, interactions, or alternative structures, depending on the question.

Sensitivity analysis asks how a model's output or recommended choice changes when an input or assumption changes. It can reveal a threshold where the preferred option reverses, or show that a conclusion survives a wide range. The important range is one that is plausible and relevant, not an arbitrary percentage around every input.

Changing one input at a time is easy to interpret but can miss interactions and correlated uncertainties. Some variables matter only together, and an influential input is not necessarily uncertain or controllable. Use the analysis to distinguish what drives the output, what remains unknown, and what additional information could change the actual decision.

When to use it

When a forecast or recommendation depends on uncertain assumptions and knowing its stability could improve the decision or direct further research.

How it can help

Identify decision-changing assumptions, choose plausible ranges, and distinguish sensitivity from uncertainty and controllability. Investigate combinations when inputs interact or move together.