BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Nonlinear Shrinkage Estimation in Quadratic Inference Function Ana
 lysis for Correlated Data - Clifford Lam (London School of Economics)
DTSTART:20180215T110000Z
DTEND:20180215T120000Z
UID:TALK101098@talks.cam.ac.uk
CONTACT:INI IT
DESCRIPTION:Quadratic inference function (QIF) analysis is more efficient 
 than the generalized estimating equations (GEE) approach when the working 
 covariance matrices for the data are misspecified. Since QIF naturally req
 uires a weighting matrix which is the inverse of a sample covariance matri
 x of non-identically distributed data\, finite sample performance can be g
 reatly affected when the number of independent data points is not large en
 ough\, which is usually the case in cluster randomized trials or many long
 itudinal studies. While nonlinear shrinkage is very successful in regulari
 zing the extreme eigenvalues of a sample covariance matrix\, the method is
  only restricted to independent and identically distributed data. We propo
 se a novel nonlinear shrinkage approach for a sample covariance matrix of 
 non-identically distributed data\, which improves finite sample performanc
 e of QIF\, and gives room for increasing the potential number of working c
 orrelation structures for even better performance. Together with a nonline
 arly shrunk weighting matrix\, we derive the asymptotic normality of the p
 arameter estimators\, and give an estimator for the asymptotic covariance 
 matrix. We demonstrate the performance of the proposed method through simu
 lation experiments and a real data analysis.<br><br><br><br>
LOCATION:Seminar Room 2\, Newton Institute
END:VEVENT
END:VCALENDAR