varest: Estimates of the covariance.
In kloke/jrfit: Joint ranking estimates

Description Usage Arguments Details Value Author(s) References Examples

Estimates of the variance covariance of the scores. Used in estimating variance covariance of the linear model estimates and also in the score test.

1	tcs(ehat, X, block, ahat = NULL, scores = wscores, eps = 1e-04)

`ehat`	N x 1 vector of residuals
`X`	N x p design matrix
`block`	N x 1 vector denoting block membership
`ahat`	optional vector of scores evaluated at the residual (if null then is computed using ehat and scores)
`scores`	score function used in the fit of ehat
`eps`	used in tcs to ensure positive definite

Estimates vhat = sum_k^m X_k^T varphi_k X_k. See Kloke, McKean, Rashid (2009) for details.

p x p matrix

John Kloke <kloke@biostat.wisc.edu>

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

##---- Should be DIRECTLY executable !! ----
##-- ==>  Define data, use random,
##--	or do  help(data=index)  for the standard data sets.

## The function is currently defined as
function (ehat, X, block, ahat = NULL, scores = wscores, eps = 1e-04) 
{
    if (is.null(ahat)) 
        ahat <- getScores(scores, ehat)
    nvec <- tapply(ehat, block, length)
    M <- sum(choose(nvec, 2)) - ncol(X)
    myfunc <- function(ak) {
        Ak <- tcrossprod(ak)
        Ak[upper.tri(Ak)]
    }
    rhohat <- sum(unlist(tapply(ahat, block, myfunc)))/M
    K0 <- -1/(max(nvec) - 1)
    if (rhohat < K0) 
        rhohat <- K0 + eps
    if (rhohat > 1) 
        rhohat <- 1 - eps
    myfunc <- function(i, x, rhohat) {
        xi <- as.matrix(x[i, ])
        ni <- nrow(xi)
        sk <- matrix(rhohat, nrow = ni, ncol = ni)
        diag(sk) <- 1
        t(xi) %*% sk %*% xi
    }
    Reduce("+", tapply(1:nrow(X), block, myfunc, X, rhohat))
  }