This function computes the gap statistic which is used to test for an outlier using Daniels method
This function computes the gap statistic which is used to test for an outlier using Daniels method
gapstat(beta, pse)
beta |
input - vector of coefficients from saturated model fit to the data |
pse |
input - Lenth's PSE statistic calculated from the elements of beta |
returned gap statistic
John Lawson
## The function is currently defined as
function (beta, pse)
{
p <- length(beta)
psehe <- pse
sel <- beta >= 0
betap <- beta[sel]
betap <- sort(betap)
betas <- betap[1]
sel <- beta < 0
betan <- beta[sel]
nn <- length(betan)
betan <- sort(betan)
betal <- betan[nn]
zl <- qnorm((nn - 0.375)/(p + 0.25))
zs <- qnorm((nn + 1 - 0.375)/(p + 0.25))
gap <- ((betas - betal)/psehe)/(zs - zl)
return(gap)
}Please choose more modern alternatives, such as Google Chrome or Mozilla Firefox.