如何在R中制作回归估计器线性组合的置信区间？

Question

In this regression:

I know confit() from the package glht can do confidence interval of every estimator.

But how to make confidence interval of a linear combination of coefficients, such as confidence interval of β3+2*β5 in R?

Added this

Answer 1

You can do that with linearHypothesis in the car package:

library(car)

dat <- data.frame(
  y = rnorm(100),
  x1 = rnorm(100),
  x2 = rnorm(100)
)

fit <- lm(y ~ x1 + x2, data = dat)

# enter linear hypothesis as a matrix
linearHypothesis(fit, t(c(0,2,2)), 0)
# enter linear hypothesis as a string
linearHypothesis(fit, "2*x1 + 2*x2 = 0")

Or with glht in the multcomp package, which also provides a confidence interval for the linear combination:

library(multcomp)

lh <- glht(fit, linfct = t(c(0,2,2)))
confint(lh)
# Linear Hypotheses:
#        Estimate lwr     upr    
# 1 == 0  0.1258  -0.4398  0.6914