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:Statistics
SUMMARY:Stable Weights that Balance Covariates for Causal
Inference and Estimation with Incomplete Outcome D
ata - JosÃ© Zubizarreta\, Columbia University
DTSTART;TZID=Europe/London:20150409T160000
DTEND;TZID=Europe/London:20150409T170000
UID:TALK58822AThttp://talks.cam.ac.uk
URL:http://talks.cam.ac.uk/talk/index/58822
DESCRIPTION:Weighting methods that adjust for observed covaria
tes\, such as inverse probability weighting\, are
widely used for causal inference and estimation wi
th incomplete outcome data. Part of the appeal of
such methods is that one set of weights can be use
d to estimate a range of treatment effects based o
n different outcomes\, or a variety of population
means for several variables. However\, this appeal
can be diminished in practice by the instability
of the estimated weights and by the difficulty of
adequately adjusting for observed covariates in so
me settings. To address these limitations\, this p
aper presents a new weighting method that finds th
e weights of minimum variance that adjust or balan
ce the empirical distribution of the observed cova
riates up to levels prespecified by the researcher
. This method allows the researcher to balance ver
y precisely the means of the observed covariates a
nd other features of their marginal and joint dist
ributions\, such as variances and correlations and
also\, for example\, the quantiles of interaction
s of pairs and triples of observed covariates\, th
us balancing entire two- and three-way marginals.
Since the weighting method is based on a well-defi
ned convex optimization problem\, duality theory p
rovides insight into the behavior of the variance
of the optimal weights in relation to the level of
covariate balance adjustment\, answering the ques
tion\, how much does tightening a balance constrai
nt increases the variance of the weights? Also\, t
he weighting method runs in polynomial time so rel
atively large data sets can be handled quickly. An
implementation of the method is provided in the n
ew package sbw for R. This paper shows some theore
tical properties of the resulting weights and illu
strates their use by analyzing both a real data se
t and a simulated example.
LOCATION:MR11\, Centre for Mathematical Sciences\, Wilberf
orce Road\, Cambridge
CONTACT:
END:VEVENT
END:VCALENDAR