Seminar: Bias, random error, and the variety of evidence thesis

THE UNIT FOR HISTORY AND PHILOSOPHY OF SCIENCE
Held in conjunction with the Sydney Centre for the Foundations of Science
SEMESTER one
RESEARCH SEMINAR SERIES
MONDAY 13th MARCH 2017

Barbara Osimani, PhD
Assistant Professor
Munich Center for Mathematical Philosophy
Ludwig-Maximilians-Universität München
Bias, random error, and the variety of evidence thesis.

In this talk I analyse the two main research strategies advocated by opposing schools in medical methodology (“evidence elitism” and “methodological pluralism”) and underscore their epistemological underpinnings, with a particular focus on the role of reliability and varied evidence in the two camps.

Since the latter strategy can be made more general by appealing to the Variety of Evidence Thesis, I analyse this thesis and its diverse versions, by delving in particular on the one presented by Bovens and Hartmann (2003), where the interaction of reliability and replication in hypothesis confirmation has an essential role in defining the epistemic value of varied evidence vs. replication.

I then present Claveau’s variation of this model (2013), which models unreliability as systematic error (bias), and go on to propose a model (Osimani, Landes forthcoming), where a distinction is made between random and systematic error. This delivers results that contrast with both Bovens and Hartmann (2003), and Claveau (2013): when evidence is highly biased relatively speaking (bias much larger than random error), then confirmation is greater for varied evidence. This is in conflict with Bovens and Hartmann results where the VET failed for unreliable evidence (in their sense of unreliability). Furthermore, when evidence is only weakly biased, then the model favors replication; and for low values of both kind of errors, the area where VET fails become negligible.

Although the VET fails in all models, it does so under different conditions in each of them, which are especially linked to how reliability, dependence of observations, and consistency are modeled. This demands for a further clarification of these notions both in scientific practice and in formal epistemology.
DATE: Monday 13th March 2017
TALK TIME: 5:30 PM
LOCATION- CCANESA MEETING ROOM, MADSEN BUILDING
CAMPERDOWN CAMPUS
Best access to CCANESA is from the Eastern Avenue entrance of the Madsen Building. When you enter you will be on the 3rd floor. Please proceed across the foyer and take the stairs on the right up one floor. The door to CCANESA will be straight ahead on this landing

All Welcome | No Booking Required | Free
PLEASE CHECK OUR WEBSITE FOR ANY CHANGES TO VENUE OR TIME
sydney.edu.au/science/hps/