Improved generalized estimating equation analysis via xtqls for quasi–least squares in Stata
Justine Shults
Department of Biostatistics and Epidemiology
Center for Clinical Epidemiology and Biostatistics
University of Pennsylvania School of Medicine
Philadelphia, PA
[email protected]
|
Sarah J. Ratcliffe
Department of Biostatistics and Epidemiology
Center for Clinical Epidemiology and Biostatistics
University of Pennsylvania School of Medicine
Philadelphia, PA
|
Mary Leonard
Department of Biostatistics and Epidemiology
Center for Clinical Epidemiology and Biostatistics
University of Pennsylvania School of Medicine
Philadelphia, PA
|
Abstract. Quasi-least squares (QLS) is an alternative method for estimating the
correlation parameters within the framework of the generalized estimating
equation (GEE) approach for analyzing correlated cross-sectional and
longitudinal data. This article summarizes the development of QLS that
occurred in several reports and describes its use with the user-written
program xtqls in Stata. Also, it demonstrates the following
advantages of QLS: (1) QLS allows some correlation structures that have not
yet been implemented in the framework of GEE, (2) QLS can be applied as an
alternative to GEE if the GEE estimate is infeasible, and (3) QLS uses the
same estimating equation for estimation of Β as GEE; as a result, QLS
can involve programs already available for GEE. In particular, xtqls
calls the Stata program xtgee within an iterative approach that
alternates between updating estimates of the correlation parameter and then
using xtgee to solve the GEE for Β at the current estimate of
α. The benefit of this approach is that after xtqls, all the
usual postregression estimation commands are readily available to the user.
View all articles by these authors:
Justine Shults, Sarah J. Ratcliffe, Mary Leonard
View all articles with these keywords:
xtqls, correlated data, clustered data, longitudinal data, generalized estimating equations, quasi–least squares
Download citation: BibTeX RIS
Download citation and abstract: BibTeX RIS
|