I work on how science ought to be conducted to produce evidence that best facilitates decision-making under uncertainty in policy or private contexts. To date, most of my research discusses how scientists could use effect sizes of tested interventions to inform rational decision-making. I also have ongoing interests in how scientists can produce evidence for long-run decision-making, and on which causal properties scientists should study to facilitate rational decision-making. More recently, I have started to work on how uncertainty attitudes (such as risk aversion) influence scientific knowledge.
Australasian Journal of Philosophy (2026)
We often want to choose the intervention that most effectively achieves our goals. But what are the most effective interventions? Here, we often turn to advice from scientists who report the effectiveness of tested interventions, using the mean difference: the difference in expected outcomes given a tested intervention and given a control intervention. I identify a problem with this practice: even perfectly accurate mean differences omit information about how interventions change the probability of an outcome of interest, and this omitted information can be vital for decision-makers to choose the best intervention—the one they would rationally choose if they were to learn all the information available in studies about how the interventions change the probability distribution over an outcome of interest. These interventions’ mean differences omit some of this information and, in doing so, may fail to inform the agent sufficiently to identify the best intervention. That is the bad news. The good news is: I identify sufficient conditions for mean differences to inform rational agents sufficiently (specifically, those agents abiding by expected utility theory or risk-weighted expected utility theory). Based on my results, I advise how researchers and decision-makers can use mean differences to facilitate rational decision-making.
Philosophy of Science (2023)
Absolute and relative outcome measures measure a treatment’s effect size, purporting to inform treatment choices. I argue that absolute measures are at least as good as, if not better than, relative ones for informing rational decisions across choice scenarios. Specifically, this dominance of absolute measures holds for choices between a treatment and a control group treatment from a trial and for ones between treatments tested in different trials. This distinction has hitherto been neglected, just like the role of absolute and baseline risks in decision-making that my analysis reveals. Recognizing both aspects advances the discussion on reporting outcome measures.
Metascience (2026)
Please feel free to e-mail me for a draft or slides for these.
co-authored with Nicholas Makins
Synopsis: We prove that both absolute and relative effect sizes can fail to sufficiently inform almost any expected utility maximizer (i.e. we establish necessary and sufficient conditions for such effect sizes to fail to sufficiently inform expected utility maximizers). We argue that this result motivates a context-sensitive approach to using absolute and/or relative effect sizes in evidence-based decision-making.
Synopsis: I prove that standardising mean differences risks stripping away information that rational decision-makers care about, even when facing measurement uncertainty. I also prove that this risk can be avoided in two kinds of scenarios. To establish these results, I develop a model for rational updating under measurement uncertainty. These findings advance ongoing debates in statistics on when researchers should standardise mean differences.
Synopsis: I argue that using effect sizes to inform decision-makers often risks providing fewer opportunities to gain decision-relevant knowledge for more risk averse people, and thereby for those who tend to be already disadvantaged. In this way, using effect sizes to communicate effectiveness gives rise to distributive epistemic injustice. More generally, the paper introduces differences in risk attitudes to recent debates on distributive epistemic justice in science.
Synopsis: I show that correcting observed effect sizes for research biases (such as publication bias) can harm rational decision-makers if the corrected effect sizes would fail to sufficiently inform these agents. I argue that such harms constitute a pro tanto moral reason against correcting observed effect sizes, and prove results that allow us to rely on generic information about research biases (such as the direction of bias) to identify who would be exposed to harm from correcting an effect size.
Synopsis: I argue that research on the long-run effects of interventions in the biomedical and social sciences often leaves evidential gaps on how causal effects persist. Yet, understanding how causal effects persist is vital for good long-run decision-making. To argue for this view, I develop an account of which properties of causes facilitate persistent effects.
This project is funded by the German Academice Exchange Service (DAAD), via the PRIME postdoctoral fellowship programme. I'm happy to hear from people who work on related questions!
Scientific knowledge is shaped by scientists making decisions, and these decisions typically have uncertain consequences. When facing such uncertainty, scientists may exhibit varying inclinations to take risks, and such attitudes may influence their decisions. For instance, two scientists may not differ in their judgements about how desirable the consequences of accepting or rejecting a hypothesis are. And yet, one might accept the hypothesis while the other rejects it, merely because they have different attitudes towards the uncertainty involved in accepting or rejecting the hypothesis. Science may be laden not just with judgements about how desirable consequences are, but also with ones arising from uncertainty attitudes, such as risk aversion, ambiguity aversion and unawareness aversion.
Philosophical discussions on values in science, however, have overlooked the distinct role attitudes towards uncertainty may play in science. According to the orthodox view in philosophy, scientists’ moral views, political opinions and personal preferences—their non-epistemic value judgements—influence many, if not all, choices scientists make when establishing research findings. Much of this literature discusses such non-epistemic value judgements as judgements about the desirability of consequences, without distinguishing these judgements from uncertainty attitudes. Yet, in decisions where the consequences are uncertain, decision-makers’ attitudes towards uncertainty play a distinct, though similarly important, role as their evaluations of consequences.
This influence of uncertainty attitudes on science raises a vital challenge: science ought to discover knowledge, not personal opinion, and produce results trustworthy for everyone, not just those sharing the scientists’ inclinations to take risks. But how can science live up to these ambitions if scientists rely on their attitudes towards uncertainty when establishing their findings? Philosophers have overlooked this challenge. Instead, they focus only on how to manage scientists’ non-epistemic value judgements about consequences so as to uphold these ambitions of science. But managing uncertainty attitudes may interact non-trivially with managing such evaluations of consequences. Certain uncertainty attitudes can conflict with certain moral judgements of consequences, entail such judgements, or be entailed by them. As a result, to properly manage uncertainty attitudes, we need to pay attention to their distinct influence on science.
This project develops the first account of the proper role of uncertainty attitudes in science: arguing that uncertainty attitudes constitute a distinct influence shaping science, and proposing ways to manage this influence.