![]() |
COOKIES: By using this website you agree that we can place Google Analytics Cookies on your device for performance monitoring. | ![]() |
University of Cambridge > Talks.cam > Cambridge Statistics Discussion Group (CSDG) > Connecting the False Discovery Rate to shrunk estimates
Connecting the False Discovery Rate to shrunk estimatesAdd to your list(s) Download to your calendar using vCal
If you have a question about this talk, please contact Peter Watson. Science is currently facing a ‘replication crisis’ – a concern that many scientific findings reported are difficult or impossible to reproduce. A major cause of this is the availability of technology that permits the exploration and testing of very large numbers of hypotheses, some of which will almost certainly show large or significant effects by chance, even when no real effects are present: this is the ‘multiplicity’ or ‘multiple testing’ problem. The tools available to address this problem include: shrunk estimates, which reduce the estimated effect in relation to each hypothesis from the observed value towards the null value, and the False Discovery Rate (FDR), which relates to the subset of the hypotheses tested for which the discovery of an effect is announced, and states the proportion of these ‘discoveries’ that is expected to be false. This talk will first examine the conceptual basis for each of these tools, then consider how they are connected. Though shrunk estimates and the FDR are both conventionally presented in the frequentist statistical framework, they can both also be presented in empirical-Bayesian terms, with the prior distribution being provided by: the distribution of effect sizes over the full set of hypotheses (in the case of shrunk estimates), and the distribution of significance-test p-values over the subset of hypotheses giving ‘discoveries’ (in the case of the FDR ). Based on this connection, a formal relationship between shrunk estimates and FDR values, for a normally-distributed response variable, will be illustrated. The talk will conclude by considering which of the two tools is the more appropriate in different practical circumstances. This talk is part of the Cambridge Statistics Discussion Group (CSDG) series. This talk is included in these lists:
Note that ex-directory lists are not shown. |
Other listsAPDE Magdalene Society of Medievalists Biodiversity and genomicsOther talksCésar Milstein Lecture: Mechanisms driving the rapid evolution of genomes AI Education in Finland: technology for the 'AI generation' Using Maths to Decode the Universe CNHS Field Studies 2024 A Slippery Signifier: Sea Ice in the Norwegian National Imaginary |