Skip to contentScholarGate
LibraryBookshelfDeskReview StudioAssistant
Sign in
On this page
IntuitionHow it worksWhen to use itStrengths & limitationsCommon pitfallsApplicationsFrequently asked🔒 Read the full methodSourcesRelated methods
Cite this pageSpotted an issue on this page? Report or suggest a fix →
Home›Simulation›Bayesian Sensitivity Analysis — Prior-informed uncertainty propagation and output sensitivity assessment
Process / pipelineSimulation / optimization

Bayesian Sensitivity Analysis — Prior-informed uncertainty propagation and output sensitivity assessment

Also known as: BSA, Bayesian SA, Bayesian robustness analysis, prior sensitivity analysis

Bayesian Sensitivity Analysis (BSA) combines Bayesian inference with sensitivity analysis to systematically quantify how uncertain model inputs — expressed as prior probability distributions — propagate through a model and influence outputs. It identifies which parameters most drive output variability, supporting robust conclusions under genuine uncertainty.

ScholarGate
  1. Process / pipeline
  2. v1
  3. 2 Sources
  4. PUBLISHED
Cite this page →
Tools & resources
Download slides
Learn & explore

Read the full method

Members only

Sign in with a free account to read this section.

Sign in

Method map

The neighbourhood of related methods — select a node to explore.

Bayesian Sensitivity Analysis
Bayesian Dynamic Program…Bayesian Markov ModelMarkov ModelMONTE-CARLO-SIMULATIONStochastic Sensitivity A…Bayesian Monte Carlo Sim…Bayesian Scenario Analys…

When to use it

Use Bayesian Sensitivity Analysis when uncertainty in model parameters cannot be ignored and probabilistic quantification is needed — especially in health economic models, climate projections, reliability engineering, and policy analysis with sparse data. It is the right choice when expert knowledge must be formally incorporated as priors and when the robustness of conclusions to those assumptions needs verification. Do NOT use it as a drop-in replacement for classical one-at-a-time sensitivity analysis when a full probabilistic model is unavailable or computationally infeasible. Avoid it when no reasonable prior can be specified and when stakeholders require fully frequentist inference, as Bayesian outputs (credible intervals) carry different interpretive commitments.

Strengths & limitations

Strengths
  • Formally incorporates prior knowledge and expert beliefs, making uncertainty assumptions explicit and auditable.
  • Provides full posterior distributions over outputs rather than point estimates, enabling probabilistic decision-making.
  • Identifies which parameters most reduce output uncertainty, guiding efficient data collection and research investment.
  • Naturally handles small samples and sparse data where frequentist methods lack power.
  • Supports robustness analysis across a class of priors, revealing which conclusions are fragile and which are stable.
  • Coherently propagates uncertainty through complex, multi-step models including simulations and decision trees.
Limitations
  • Requires specification of prior distributions, which can be controversial and time-consuming when expert elicitation is needed.
  • Posterior computation for high-dimensional or non-conjugate models demands MCMC or variational methods, which are computationally expensive.
  • Results and their interpretation depend on the chosen prior class; a poorly chosen prior class may miss important sensitivity.
  • Communicating Bayesian outputs (posteriors, credible intervals) to non-statistical audiences is more challenging than classical p-values or confidence intervals.
  • Validating that the MCMC sampler has converged requires additional diagnostics, adding methodological overhead.

Frequently asked

How is Bayesian Sensitivity Analysis different from standard (frequentist) sensitivity analysis?

Standard sensitivity analysis typically varies inputs deterministically or samples them without a probability model, reporting how outputs change. Bayesian Sensitivity Analysis instead encodes input uncertainty as probability distributions updated by data, propagates the full posterior through the model, and explicitly tests robustness to the choice of prior — producing probabilistic output distributions rather than scalar change metrics.

Do I need a large dataset to apply Bayesian Sensitivity Analysis?

No — in fact, BSA is especially valuable with small datasets where priors carry meaningful weight. However, with very sparse data the choice of prior dominates the posterior, making prior sensitivity analysis even more critical and its conclusions more contingent on expert assumptions.

What software can I use for Bayesian Sensitivity Analysis?

Common tools include Stan (via RStan or PyStan) for MCMC posterior computation, the R package 'sensitivity' for Sobol indices on posterior samples, BUGS/JAGS for hierarchical models, and Python libraries such as PyMC and ArviZ for posterior diagnostics. SALib in Python handles global sensitivity indices.

How many prior variants should I test to claim robustness?

At minimum, compare a weakly informative, a moderately informative, and a strongly informative prior. In formal robustness analysis, a geometric class or epsilon-contamination class spanning the plausible prior space is swept. Conclusions described as robust should hold across all plausible variants, not just two or three hand-picked cases.

Can Bayesian Sensitivity Analysis be combined with Monte Carlo Simulation?

Yes — Monte Carlo samples drawn from the posterior distribution are the standard mechanism for propagating uncertainty in BSA. The distinction is that in BSA the sampling distribution is the posterior (data-updated beliefs) rather than an assumed fixed distribution, making the uncertainty estimates coherent with observed evidence.

Sources

  1. Berger, J. O. (1994). An overview of robust Bayesian analysis. Test, 3(1), 5–124. DOI: 10.1007/BF02562676 ↗
  2. Saltelli, A., Ratto, M., Andres, T., Campolongo, F., Cariboni, J., Gatelli, D., Saisana, M., & Tarantola, S. (2008). Global Sensitivity Analysis: The Primer. Wiley. ISBN: 9780470059975

How to cite this page

ScholarGate. (2026, June 3). Bayesian Sensitivity Analysis — Prior-informed uncertainty propagation and output sensitivity assessment. ScholarGate. https://scholargate.app/en/simulation/bayesian-sensitivity-analysis

Related methods

Bayesian Dynamic ProgrammingBayesian Markov ModelMarkov ModelMONTE-CARLO-SIMULATIONStochastic Sensitivity Analysis

Which method?

Set this method beside its closest kin and read them side by side — the library lays the books on the table; the choice is yours.

  • Bayesian Dynamic ProgrammingSimulation↔ compare
  • Bayesian Markov ModelSimulation↔ compare
  • Markov ModelSimulation↔ compare
  • MONTE-CARLO-SIMULATIONDecision-making↔ compare
  • Stochastic Sensitivity AnalysisSimulation↔ compare
Compare side by side →

Referenced by

Bayesian Markov ModelBayesian Monte Carlo SimulationBayesian Scenario Analysis

Similar methods

Bayesian Monte Carlo SimulationBayesian Scenario AnalysisRobust Sensitivity AnalysisBayesian Sensitivity Analysis for CausalityRobust Bayesian InferenceRobust Monte Carlo SimulationPolicy Scenario Sensitivity AnalysisBayesian Statistical Inference

Related reference concepts

Prior Elicitation and Sensitivity AnalysisBayesian Inference FoundationsPrior DistributionsSensitivity Analysis in Economic EvaluationBayesian Model Comparison and SelectionEmpirical Bayes Methods

Spotted an issue on this page? Report or suggest a fix →

ScholarGate — Bayesian Sensitivity Analysis (Bayesian Sensitivity Analysis — Prior-informed uncertainty propagation and output sensitivity assessment). Retrieved 2026-07-20 from https://scholargate.app/en/simulation/bayesian-sensitivity-analysis · Dataset: https://doi.org/10.5281/zenodo.20539026
Quick facts
Originator
Berger, J. O. (Bayesian robustness); Saltelli et al. (global SA integration)
Year
1984–1994
Type
Uncertainty propagation and sensitivity quantification
DataType
Probabilistic inputs, prior distributions, model parameters
Subfamily
Simulation / optimization
Related methods
Bayesian Dynamic ProgrammingBayesian Markov ModelMarkov ModelMONTE-CARLO-SIMULATIONStochastic Sensitivity Analysis
ScholarGate

A content-first reference library for research methods — what each one is, how it works, and where it comes from.

Open data (CC-BY)

Explore

  • Library
  • Search the library…
  • Browse by field
  • Fields
  • Journey
  • Compare
  • Which method?

Reference

  • Subjects
  • Atlas
  • Glossary
  • Methodology
  • Philosophy

Your tools

  • Bookshelf
  • Desk
  • Chat

Company

  • About
  • Pricing
  • Contact
  • Suggest a method

Entries are compiled from published sources for reference. Verifying the accuracy and suitability of any information for your own use remains your responsibility.

© 2026 ScholarGate · A research-method reference library
  • Privacy
  • Cookies
  • Terms
  • Delete account