BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Debiased regression adjustment in completely randomized experiment
 s with moderately high-dimensional covariates - Yuhao Wang (Tsinghua Unive
 rsity)
DTSTART:20250128T130000Z
DTEND:20250128T140000Z
UID:TALK227731@talks.cam.ac.uk
CONTACT:Qingyuan Zhao
DESCRIPTION:Completely randomized experiment is the gold standard for caus
 al inference. When the covariate information for each experimental candida
 te is available\, one typical way is to include them in covariate adjustme
 nts for more accurate treatment effect estimation. In this paper\, we inve
 stigate this problem under the randomization-based framework\, i.e.\, that
  the covariates and potential outcomes of all experimental candidates are 
 assumed as deterministic quantities and the randomness comes solely from t
 he treatment assignment mechanism. Under this framework\, to achieve asymp
 totically valid inference\, existing estimators usually require either (i)
  that the dimension of covariates p grows at a rate no faster than O(n3/4)
  as sample size n→∞\; or (ii) certain sparsity constraints on the line
 ar representations of potential outcomes constructed via possibly high-dim
 ensional covariates. In this paper\, we consider the moderately high-dimen
 sional regime where p is allowed to be in the same order of magnitude as n
 . We develop a novel debiased estimator with a corresponding inference pro
 cedure and establish its asymptotic normality under mild assumptions. Our 
 estimator is model-free and does not require any sparsity constraint on po
 tential outcome's linear representations. We also discuss its asymptotic e
 fficiency improvements over the unadjusted treatment effect estimator unde
 r different dimensionality constraints. Numerical analysis confirms that c
 ompared to other regression adjustment based treatment effect estimators\,
  our debiased estimator performs well in moderately high dimensions.
LOCATION:MR14\,  Centre for Mathematical Sciences\, Wilberforce Road\, Cam
 bridge
END:VEVENT
END:VCALENDAR