BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Memento EPFL//
BEGIN:VEVENT
SUMMARY:Assumption-lean inference for generalised linear model parameters
DTSTART:20201106T151500
DTEND:20201106T163000
DTSTAMP:20260502T091155Z
UID:dc575a889564e3e69f8f0a756185edac72e10c5828f70260e8d5e5d0
CATEGORIES:Conferences - Seminars
DESCRIPTION:Oliver Dukes\, Ghent University\nInference for the parameters 
 indexing generalised linear models is routinely based on the assumption th
 at the model is correct and a priori specified. This is unsatisfactory bec
 ause the chosen model is usually the result of a data-adaptive model selec
 tion process\, which may induce excess uncertainty that is not usually ack
 nowledged. Moreover\, the assumptions encoded in the chosen model rarely r
 epresent some a priori known\, ground truth\, making standard inferences p
 rone to bias\, but also failing to give a pure reflection of the informati
 on that is contained in the data.\n\nInspired by developments on assumptio
 n-free inference for so-called projection parameters\, we here propose nov
 el nonparametric definitions of main effect estimands and effect modificat
 ion estimands. These reduce to standard main effect and effect modificatio
 n parameters in generalised linear models when these models are correctly 
 specified\, but have the advantage that they continue to capture respectiv
 ely the primary (conditional) association between two variables\, or the d
 egree to which two variables interact (in a statistical sense) in their ef
 fect on outcome\, even when these models are misspecified.\n\nWe achieve a
 n assumption-lean inference for these estimands (and thus for the underlyi
 ng regression parameters) by deriving their influence curve under the nonp
 arametric model and invoking flexible data-adaptive (e.g.\, machine learni
 ng) procedures.\n\nThis is joint work with Stijn Vansteelandt\n 
LOCATION:zoom https://epfl.zoom.us/j/81403546538
STATUS:CONFIRMED
END:VEVENT
END:VCALENDAR