Confirmation by Robustness Analysis: A Bayesian Account

Area of Science:

• Philosophy of Science
• Epistemology
• Scientific Modeling

Background:

• Minimal models are debated for their epistemic value.
• Robustness analysis is proposed for hypothesis confirmation but faces resistance.
• Bayesian frameworks offer a potential rationalization for robustness analysis.

Purpose of the Study:

• To provide a Bayesian rationalization for robustness analysis in confirming hypotheses.
• To explore the confirmatory potential of minimal models.
• To identify conditions under which robustness analysis is detrimental to confirmation.

Main Methods:

• Bayesian rationalization of robustness analysis.
• Case study from macroeconomics.
• Analysis of evidential variety.

Main Results:

• Robustness analysis over minimal models can indeed confirm hypotheses.
• The confirmatory value is context-dependent.
• Specific cases where robustness analysis hinders confirmation were identified and linked to evidential variety.

Conclusions:

• Robustness analysis, when applied to minimal models, can serve a confirmatory role.
• The epistemic benefits of robustness analysis are contingent on specific circumstances.
• Understanding evidential variety is crucial for assessing the impact of robustness analysis.