Title: “A flexible sensitivity analysis for sample selection bias”
Abstract: Selection bias can occur when a sample differs systematically from the population from which it was drawn. This can distort statistical quantities and lead to erroneous inferences. Selection bias can be difficult to address when there is limited information available on the population. In this talk, I will describe a sensitivity analysis for selection bias which is able to flexibly incorporate a wide variety of population-level information (e.g. summary statistics, negative controls, shape constraints), while providing valid statistical inference. Applications include 1) estimating the effect of education on income in UK Biobank and 2) estimating risk factors for Covid-19 infection within a sample of individuals who have received a Covid-19 test. Please see the pre-print for more details: https://arxiv.org/abs/1906.10159
This will be a free virtual seminar. If you would like to join, please register here: BSU Seminar – Matt Tudball Tickets, Thu 10 Feb 2022 at 14:00 | Eventbrite
All registered attendees will receive the joining information prior to the seminar.