"myags"<-
function(formula, id, weights = NULL, cor.met = NULL, family, alpfun = NULL, 
	scalefun = LZ.scalefun, wcorigen = identni, tol = 0.0001, contrasts = 
	NULL, corstr = c("independence", "exchangeable", "ar1", "unstructured"),
	maxiter = 25)
{
	## myags same as yags, except 
	##         replace    yags.pmat.fit with
	##                    myags.pmat.fit

	prog.error <- NULL
	call <- match.call()
	m <- match.call(expand = F)
	m$contrasts <- m$tol <- m$wcorigen <- m$scalefun <- m$alpfun <- m$
		family <- m$cor.met <- m$id <- m$corstr <- m$maxiter <- NULL
	m[[1]] <- as.name("model.frame")
	m <- eval(m, sys.parent())
	Terms <- attr(m, "terms")
	y <- as.matrix(model.extract(m, response))	
	# 1.5.1.1 B : add weight-processing specifically for binomial matrices
# as in glm()
	if(is.null(weights))
		weights <- rep(1, nrow(y))
	else if(length(weights) != nrow(y))
		stop("lengths of y and weights incompatible")
	if(ncol(y) == 2) {
		n <- apply(y, 1, sum)
		y <- y[, 1, drop = F]/n
		weights <- n/weights
	}
	else weights <- 1/weights
	x <- model.matrix(Terms, m, contrasts)
	Y <- load.clustered.outcome(y, id)
	X <- load.clustered.design(x, id)
	if(is.null(alpfun) & length(corstr) > 1) {
		warning("no alpfun provided and corstr has several elements; corstr[1] used"
			)
		corstr <- corstr[1]
	}
	if(!is.null(alpfun) & length(corstr) == 1)
		warning("both alpfun and corstr provided; corstr will be ignored"
			)
	else if(corstr[1] == "exchangeable" & is.null(alpfun)) {
		alpfun <- LZ.exchalp
		wcorigen <- excoriput
	}
	else if(corstr[1] == "ar1" & is.null(alpfun)) {
		if(is.null(cor.met))
			stop("must supply times (cor.met) with ar1")
		alpfun <- prop.ar1alp
		wcorigen <- ar1.coriput
	}
	else if(corstr[1] == "unstructured" & is.null(alpfun)) {
		if(is.null(cor.met))
			stop("must supply times (cor.met) with unstructured")
		alpfun <- prop.unstruc.alp
		wcorigen <- unstr.coriput
	}
	else if(corstr[1] == "independence" & is.null(alpfun)) {
		alpfun <- function(a, b, c, d, p)
		NULL
		wcorigen <- identni
	}
	ans <- myags.pmat.fit(X, Y, id, weights, cor.met, family, alpfun, 
		scalefun, wcorigen, tol, maxiter, prog.error)
	ans$call <- call
	class(ans) <- c("yags", "glm", "lm")
	ans
}