Hello,
from Verzani, simpleR (pdf), p. 80, I created the following function to
test the coefficient of lm() against an arbitrary value.
coeff.test <- function(lm.result, var, coeffname, value) {
# null hypothesis: coeff = value
# alternative hypothesis: coeff != value
es <- resid(lm.result)
coeff <- (coefficients(lm.result))[[coeffname]]
# degrees of freedom = length(var) - number of coefficients?
n <- df.residual(lm.result)
s <- sqrt( sum( es^2 ) / n )
SE <- s/sqrt(sum((var - mean(var))^2))
t <- (coeff - value )/SE
2 * pt(t,n,lower.tail=FALSE) # times two because problem is two-sided
}
E.g. it needs to be called coeff.test(lm(N ~ D), D, "D", 70) if I want
to test the probability slope of the linear model being 70.
Is there a more elegant way to avoid passing D and "D" both as
parameters?
Also, as a non-professional, I would like to know whether the function
is valid for all coefficients of lm(), e.g. coeff.test(lm(N ~ D + H), H,
"H", 70). I am aware that Verzani gives a different formula for testing
the intercept.
Thanks!
Jan Rheinländer
______________________________________________
R-help@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.
