jrfit: Joint ranking estimates for cluster correlated data.

In kloke/jrfit: Joint ranking estimates

Description

Returns R estimates for the linear models with cluster correlated errors. Also returns objects useful for inference and diagnostics.

Usage

 jrfit(x, y, block, yhat0 = NULL, scores = wscores, fitint = NULL, 
var.type = 'sandwich',fitblock=FALSE,tuser=NULL,...) 

Arguments

x	N by p design matrix
y	N by 1 response vector
block	N by 1 vector denoting cluster (block) membership
yhat0	initial fit (defaults to computing l1 fit)
scores	which score function to use (defaults to wscore (Wilcoxon scores)
fitint	should an intercept be fit in addtion to the regression parameters (adds a col of 1s to x)
var.type	one of sandwich (default), cs for compound symmetry, or user for user defined
fitblock	should blocks be fit as (nuisance) fixed effects
tuser	optional function to compute V_varphi. used when var.type='user'
...	additional arguments. currently unused.

Details

Solves the rank based minimization problem using Jaeckel's (1972) dispersion function. That is the ranks are taken over the entire dataset. Results are presented in Kloke, et. al. (2009).

scores are available in coderfit.

If fitint is set to TRUE, a column of ones is added to the design matrix. If it is set to NULL then, if 1 is not in the column space of the x then a column of ones is added.

The default behavior for fitblock is to set to TRUE when var.type is set to 'cs' and FALSE otherwise. It is not recommended to set fitblock to TRUE and use var.type as 'sandwich'. Setting fitblock to FALSE when using var.type as 'cs' may be useful at times.

var.type specifies how the variance covariance matrix of the parameter estimates should be estimated. The default is to use a sandwich estimate. Another option we have developed is the compound symmetry estimate (see Kloke, et. al. 2009). The user is welcome to supply his or her own variance covariance function using the option 'user'. This requires the user define tuser.

Author(s)

John Kloke [e-mail]

References

Hettmansperger, T.P. and McKean J.W. (2011), Robust Nonparametric Statistical Methods, 2nd ed., New York: Chapman-Hall.

Jaeckel, L. A. (1972). Estimating regression coefficients by minimizing the dispersion of residuals. Annals of Mathematical Statistics, 43, 1449 - 1458.

Jureckova, J. (1971). Nonparametric estimate of regression coefficients. Annals of Mathematical Statistics, 42, 1328 - 1338.

Kloke, J.D., McKean, J.W., Rashid, M.M. (2009). Rank-based estimation and associated inferences for linear models with cluster correlated errors. Journal of the American Statistical Association, 104, 384-390.

See Also

rfit

Examples

# a simple simulated example w/ 4 equal sized blocks #
x<-matrix(rnorm(40),ncol=2); y<-rnorm(20) ; g<-rep(1:4,each=5)
jrfit(x,y,g)

kloke/jrfit documentation built on May 20, 2019, 12:34 p.m.