BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//talks.cam.ac.uk//v3//EN
BEGIN:VTIMEZONE
TZID:Europe/London
BEGIN:DAYLIGHT
TZOFFSETFROM:+0000
TZOFFSETTO:+0100
TZNAME:BST
DTSTART:19700329T010000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=-1SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:+0100
TZOFFSETTO:+0000
TZNAME:GMT
DTSTART:19701025T020000
RRULE:FREQ=YEARLY;BYMONTH=10;BYDAY=-1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
CATEGORIES:Causal Inference Reading Group
SUMMARY:Confounder selection via iterative graph expansion
  - Richard Guo (University of Cambridge)
DTSTART;TZID=Europe/London:20231027T153000
DTEND;TZID=Europe/London:20231027T170000
UID:TALK207514AThttp://talks.cam.ac.uk
URL:http://talks.cam.ac.uk/talk/index/207514
DESCRIPTION:Confounder selection\, namely choosing a set of co
 variates to control for confounding between a trea
 tment and an outcome\, is arguably the most import
 ant step in the design of observational studies. P
 revious methods\, such as Pearl's celebrated back-
 door criterion\, typically require pre-specifying 
 a causal graph\, which can often be difficult in p
 ractice. We propose an interactive procedure for c
 onfounder selection that does not require pre-spec
 ifying the graph or the set of observed variables.
  This procedure iteratively expands the causal gra
 ph by finding what we call "primary adjustment set
 s" for a pair of possibly confounded variables. Th
 is can be viewed as inverting a sequence of latent
  projections of the underlying causal graph. Struc
 tural information in the form of primary adjustmen
 t sets is elicited from the user\, bit by bit\, un
 til either a set of covariates are found to contro
 l for confounding or it can be determined that no 
 such set exists. We show that if the user correctl
 y specifies the primary adjustment sets in every s
 tep\, our procedure is both sound and complete.("p
 aper":https://arxiv.org/abs/2309.06053)
LOCATION:Centre for Mathematical Sciences\, MR12
CONTACT:
END:VEVENT
END:VCALENDAR