Generalized Estimating Equations by Andreas Ziegler (auth.)

By Andreas Ziegler (auth.)

Generalized estimating equations became more and more renowned in biometrical, econometrical, and psychometrical functions simply because they triumph over the classical assumptions of facts, i.e. independence and normality, that are too restrictive for plenty of problems.

Therefore, the most aim of this e-book is to provide a scientific presentation of the unique generalized estimating equations (GEE) and a few of its additional advancements. therefore, the emphasis is wear the unification of varied GEE techniques. this is often performed by way of varied estimation options, the pseudo greatest chance (PML) process and the generalized approach to moments (GMM).

The writer info the statistical starting place of the GEE procedure utilizing extra common estimation thoughts. The e-book may well consequently be used as foundation for a path to graduate scholars in information, biostatistics, or econometrics, and should be precious to practitioners within the similar fields.

Show description

Read Online or Download Generalized Estimating Equations PDF

Best biostatistics books

Advances in Ranking and Selection, Multiple Comparisons, and Reliability

"S. Panchapakesan has made major contributions to score and choice and has released in lots of different components of records, together with order records, reliability concept, stochastic inequalities, and inference. Written in his honor, the twenty invited articles during this quantity mirror fresh advances in those fields and shape a tribute to Panchapakesan's impression and effect on those parts.

Variation in the Human Genome (Novartis Foundation Symposia)

The mapping of human genes is continuing swiftly. Genes linked to particular inherited illnesses are being pointed out, usually delivering perception into the molecular reason behind the disorder. in the interim, despite the fact that, little attention is being given to the difference found in various human populations.

Quasi-Least Squares Regression

Drawing at the authors' huge services in modeling longitudinal and clustered information, Quasi-Least Squares Regression offers a radical therapy of quasi-least squares (QLS) regression-a computational method for the estimation of correlation parameters in the framework of generalized estimating equations (GEEs).

Practical Guide to Clinical Computing Systems, Second Edition: Design, Operations, and Infrastructure

Even if informatics trainees and practitioners who imagine operational computing roles of their association can have kind of complicated knowing of theoretical informatics, many are unusual with the sensible subject matters - comparable to downtime approaches, interface engines, person help, JCAHO compliance, and budgets - for you to develop into the mainstay in their operating lives.

Extra info for Generalized Estimating Equations

Sample text

59), it can be shown that I(f, f ) ≥ 0 for any two densities f and f . The KLIC equals 0 if and only if f = f almost surely. The first term on the right side of the KLIC from Eq. 21 is constant so that the KLIC is minimized if the second term is maximized. , the global unique maximum of the ML approach based on the density β f , converges to β , the parameter value that minimizes the KLIC.

Thus, µi = F (β0 + xi β ) is the mean of yi given xi if the error of the linear model for the latent variable is distributed according to F (·), and the model on the observable level therefore is yi = µi + i . 5. Consider n stochastic vectors y 1 , . . , y n of length T × 1. X 1 , . . , X n are the corresponding T × p fixed or stochastic matrices of regressors. ), and IE(εi |X j ) = 0 for all i, j. Finally, assume that the matrix 1 n n i=1 X i X i con- verges to a non-stochastic regular matrix Q as n → ∞.

N let xi be a p × 1 vector of fixed and/or stochastic independent variables. Furthermore, let the T dimensional dependent variable y i given xi be T dimensionally normally distributed, specifically, y i |xi ∼ NT (µi , Σ). , g = id, is chosen as the link function and if B = (β 1 , . . , β T ) ∈ IRp×T is the matrix comprising the parameters of interest, one obtains the multivariate linear regression model by letting µi = η i = B xi . This model can also be formulated in standard notation by letting X i = xi ⊗ IT , which is of dimension T × T p, and the T p × 1 parameter vector β = (β 1 , .

Download PDF sample

Rated 4.07 of 5 – based on 45 votes