Pareto Plots

[Michel Friesenhahn <michel.friesenhahn.b@bayer.com>]

I'm using the pareto() function on Splus 6.2 for windows as follows:

pareto(tmplm,type="tvalues")

where tmplm is an lm object.  It appears to draw the vertical significance line
at the incorrect level in that it doesn't seem to identify the same terms as
significant as indicated by the p-values in summary(tmplm).

Looking at the code for pareto.lm() I noticed the following near the end of the
function:

      if(type == "tvalues") {
            sig.level <- summ.lm$sigma * qt(1 - sig/2, fit$df.residual)
            attr(sig.level, "sig") <- sig
            names(sig.level) <- "mse"
            xmax <- max(xmax, sig.level)
      }

Shouldn't this be:

      if(type == "tvalues") {
            sig.level <- qt(1 - sig/2, fit$df.residual)
            attr(sig.level, "sig") <- sig
            names(sig.level) <- "mse"
            xmax <- max(xmax, sig.level)
      }

Please let me know if I've overlooked something.

Thanks!

Mike