summary.lm with vcov. argument

summary.lm2.R

summary.lm <- function(object, correlation = FALSE, symbolic.cor = FALSE, ..., vcov. = NULL) {
  ss <- stats::summary.lm(object, correlation, symbolic.cor, ...)

  if (!is.null(vcov.)) {
    if (is.function(vcov.)) {
      vcov. <- vcov.(object, ...)
    }

    ss$cov.unscaled <- vcov./ss$sigma^2
    ss$coefficients[,2] <- sqrt(diag(vcov.))
    ss$coefficients[,3] <- ss$coefficients[,1]/ss$coefficients[,2]
    ss$coefficients[,4] <- 2 * pt(abs(ss$coefficients[,3]), ss$df[2], lower.tail = FALSE)
  }
  ss
}

This overloads stats::summary.lm to add the vcov. argument. It can be either a variance calculating function, or a variance/covariance matrix. E.g.

library(sandwich)
summary(mod, vcov. = vcovHC)
summary(mod, vcov. = vcovHC(mod, type = "HC1"))