R help - Jul 2001 - hypothesis testing in models

Dear all,

Being an economics student, I am trying to put together a little
tutorial/FAQ/... for those who have (more) background in econometrics rather
than in statistics -- just as I have. I'v been looking for a hypothesis
testing tool in R (just linear or also nonlinear) for model parameters, but
could not find anything so far. I did find a similar but unanswered question
in the archives, but nothing apart from that. Is it there or am I missing
something? If it is not there, is there a specific reason for that?
I am aware that the required code is not special or long, but the one I could
come up with is rather ugly and a pain to use:

htest  <- function(object, resmat, target = 0, level = 0.95, n = 0)
UseMethod("htest")
  # Testing liner restrictions of the form: resmat %*% cfs = target
  # resmat: matrix/vector of restriction multipliers (i x k) ARRAY!!
  # target: restriction value vector/scalar (k)

htest.lm <- function(object, resmat, target = 0, level = 0.95)
{
  tsu <- summary(object, corr=FALSE)
  cfs <- as.array(object$coefficients)
  dfr1 <- nrow(resmat)
  dfr2 <- object$df.residual
  S <- aperm(resmat %*% aperm(cfs) - target) %*% solve(resmat %*%
tsu$cov.unscaled %*% aperm(resmat)) %*% (resmat %*% aperm(cfs) - target) /
(nrow(resmat) * tsu$sigma^2)
  cat("  F(",dfr1,",",dfr2,")
=",qf(level,dfr1,dfr2), "  vs.  S =",S,"  [ p =
",df(S,dfr1,dfr2),"]\n")
}

Thanks for any comments,
Zsombor
-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: r-help-request at
stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._

Dear all,

A couple of days ago I had a question,

On Wed, Jul 18, 2001 at 07:24:02PM +0100, Zsombor Cseres-Gergely wrote:
> than in statistics -- just as I have. I'v been looking for a hypothesis
> testing tool in R (just linear or also nonlinear) for model parameters, but
but no answer arrived. The answer is (just for the sake of hystorical record):
yes, there is such a procedure in John Fox's "car" package at
http://socserv.socsci.mcmaster.ca/jfox/Misc/car/index.htm that implements the
F test in a similar fashion that I proposed (but in professional packaging).
Unfortunately it deals only with linear restrictions. It is a tool along with
some others in that package that could well belong to "base" IMHO.

Best regards,
Zsombor
-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: r-help-request at
stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._