A check of whether a study's conclusions hold up under different analytic assumptions.
Sensitivity analysis re-runs a study's statistical analysis using alternative assumptions, methods, or data subsets (e.g. excluding outliers, changing missing-data handling, varying a cut-off) to see whether the main findings remain stable. If results change substantially under reasonable alternatives, the original conclusions are considered less robust. It's a marker of methodological rigor but does not by itself prove a finding is correct.
This guide was auto-drafted and is pending editorial review.