Margherita Harris (LSE): “Model Robustness: Schupbach’s Explanatory Account of Robustness Analysis to the Rescue?”

In science, obtaining a “robust” result is often seen as providing further support for a hypothesis. The Bayesian should have something to say about the logic underpinning this method of confirmation. Schupbach’s recent explanatory account (2018) of robustness analysis (RA) is a welcome attempt to do so. Indeed, by having ‘as its central notions explanation and elimination’ (ibid., 286), this account seems to fit very nicely with many empirically driven cases of RA in science, thereby revealing why these cases are able to lend confirmation to a hypothesis. The subject of this talk, however, is Schupbach’s further claim that his account of RA ‘applies to model-based RAs just as well as it does to empirically driven RAs’ (ibid. 297), since when we arrive at this claim, he and I decisively part ways. I will argue that the application of Schupbach’s account to model-based RAs is considerably more complicated than he and others (such as Winsberg (2018)) suggest and relies on several non-trivial and often dubious assumptions. By making these assumptions explicit, I will show that Schupbach’s account of RA is inapplicable to many cases of model-based RA’s, contrary to what has been assumed in the literature.

 

Margherita Harris is a visiting fellow at LSE Philosophy, where she also completed her PhD. Margherita’s research interests lie in epistemology and the philosophy of science, with a special focus on modelling under uncertainty, robustness analysis and climate science.