TSGMM_Poi: Two-Step Generalized Method of Moments, Longitudinal Count...

In lalondetl/GMM:

Description

This function calculates the Generalized Method of Moments (GMM) parameter estimates and standard errors for longitudinal count (Poisson) responses. This is modeled similarly to a Poisson Regression using a log link. The function allows for unbalanced data, meaning subjects can have different numbers of times of observation. Both time-independent covariates and time-dependent covariates can be accommodated. Time-dependent covariates can be handled either by specifying the type of each time-dependent covariate, or by allowing the data to determine appropriate moment conditions through the extended classification method. Data must be organized by subject, and an intercept term is assumed. The function outputs a list with parameter estimates betaHat along with parameter covariance estimates covEst.

Usage

TSGMM_Poi(yvec, subjectID, Zmat, Xmat, Tvec, N, mc = "EC",
  covTypeVec = c(-1))

Arguments

subjectID	The vector of subject ID values for each response ((sum(Tvec)) x 1).
Zmat	The design matrix for time-independent covariates ((sum(Tvec)) x K0).
Xmat	The design matrix for time-dependent covariates ((sum(Tvec)) x Ktv).
Tvec	The vector of times for each subject.
N	The number of subjects.
mc	The method of identifying appropriate moment conditions, either 'EC' for extended classification (default) or 'Types' for user-identified types.
covTypeVec	The vector indicating the type of each time-dependent covariate, according to the order of the columns of Xmat.
y	The vector of count responses. This vector must be organized by subject, and by time within subject ((sum(Tvec)) x 1).

Examples

TSGMM_Poi()

lalondetl/GMM documentation built on May 30, 2019, 11:40 p.m.