confint_robust: The confint function adapted for vcovHC

Description

The confint.lm uses the t-distribution as the default confidence interval estimator. When there is reason to believe that the normal distribution is violated an alternative approach using the vcovHC() may be more suitable.

Usage

confint_robust(
  object,
  parm,
  level = 0.95,
  HC_type = "HC3",
  t_distribution = FALSE,
  ...
)

Value

matrix A matrix (or vector) with columns giving lower and upper confidence limits for each parameter. These will be labelled as (1-level)/2 and 1 - (1-level)/2 in

Arguments

object: The regression model object, either an ols or lm object
parm: a specification of which parameters are to be given confidence intervals, either a vector of numbers or a vector of names. If missing, all parameters are considered.
level: the confidence level required.
HC_type: See options for vcovHC()
t_distribution: A boolean for if the t-distribution should be used or not. Defaults to FALSE. According to Cribari-Nieto and Lima's study from 2009 this should not be the case.
...: Additional parameters that are passed on to vcovHC()

References

F. Cribari-Neto and M. da G. A. Lima, "Heteroskedasticity-consistent interval estimators", Journal of Statistical Computation and Simulation, vol. 79, no. 6, pp. 787-803, 2009 (tools:::Rd_expr_doi("10.1080/00949650801935327"))

Examples

Run this code

n <- 50
x <- runif(n)
y <- x + rnorm(n)

fit <- lm(y~x)
library("sandwich")
confint_robust(fit, HC_type = "HC4m")