PhD Student Session: Charles Beasley and Fabian Beigang

This event will take place online via Zoom. 

Everyone is welcome to join using a computer with access to the internet and Zoom. To take part just follow these instructions:

• Download Zoom
• Join the event using this link: https://lse.zoom.us/j/88937835129

Please note that these events are routinely recorded, with the edited footage being made publicly available on our website and YouTube channel. We will only record the audio, the slides and the speaker and will not include the Q&A section. However, any question asked during the talk itself will feature in the final edit.

Charles Beasley: “Replication Pluralism”

Over the past decade, systematic replication failures have placed the status of large bodies of scientific knowledge into doubt. Understandably, this has given rise to a number of hard questions. For example, is there widespread fraud, bias, and/or misaligned incentive structures in the sciences? Do p-values need to be reformed? Is there a genuine replication crisis that spans across the sciences? Is there a crisis at all? These pressing questions, however, are all downstream from a more fundamental one; what is a replication?

On the account that I will put forward, X is a replication of Y, iff X is relevantly similar to Y, where relevant similarity is determined across multiple dimensions, within the context Z of holistically evaluating X alone.

The majority of my talk will be spent clarifying, defending, and expanding upon this definition in light of mainstream accounts, such as direct and conceptual replication (e.g. Crandall and Sherman 2019; Pashler and Harris 2012), as well as recent singular and general accounts, such as the resampling (Machery 2020) and the diagnostic accounts (Nozek and Errington 2020). In doing so, I will further specify both the scope of replication conceived broadly and the particular claim of replication pluralism, before moving on to a discussion of why replications must be evaluated holistically and not in terms of their constituent parts. I will then address the specific role that replication plays in science and why it should be treated as a unique trust-inducing practice.

Fabian Beigang: “Yet another impossibility theorem in algorithmic fairness”

In recent years, there has been a surge in research addressing the question which properties predictive algorithms ought to satisfy in order to be considered fair. Three of the most widely discussed criteria of fairness  are the criteria called equalized odds, predictive parity, and counterfactual fairness. In this talk, I will present a new impossibility result involving these three criteria of algorithmic fairness.  In particular, I will argue that there are realistic circumstances under which any predictive algorithm that satisfies counterfactual fairness will violate both, equalized odds and predictive parity. This impossibility result forces us to give up one of four widely held assumptions about algorithmic fairness. I will explain the four assumptions and discuss which of them can plausibly be given up in order to circumvent the impossibility.”