[Keowmani Thamron]

Thank you both of you for the discussion above! So it seems that margins can't do it after all.

[Richard Williams]

There is a Wish List for Stata 15 thread. You might go there and post your wish for margins and link back to this thread. If there really is a way to already do what you want you may get a response back in that thread. This seems like a reasonable request to me and perhaps not that hard to do.

[Bruce Weaver]

If all that is wanted is a scatter-plot showing the 95% prediction (or forecast) interval bands, I've just learned from Dupont's (2009) book (Statistical Modeling for Biomedical Researchers) that this can be achieved by adding the stdf option to a twoway command. Here is Dupont's code:

*  Add 95% prediction interval bands to the preceding graph.
*
twoway lfitci glyco ph,  stdf ciplot(rline) color(gray)   ///
    || lfitci glyco ph, lpattern(solid)                 ///
    || scatter glyco ph                        ///
    , ylabel(0(50)300) xlabel(6.9 (.1) 7.5)             ///
      ytitle(Plasma Glycolate (mg/dl)) legend(off)

And here is the resulting graph:

The data file and Dupont's complete code can be found at the following links.

• http://biostat.mc.vanderbilt.edu/dup...2.12.Poison.do
• http://biostat.mc.vanderbilt.edu/dup....12.Poison.dta

Cheers,
Bruce

[Gregory Grant]

If all that is wanted is a scatter-plot showing the 95% prediction (or forecast) interval bands, I've just learned from Dupont's (2009) book (Statistical Modeling for Biomedical Researchers) that this can be achieved by adding the stdf option to a twoway command. Here is Dupont's code:

* Add 95% prediction interval bands to the preceding graph.
*
twoway lfitci glyco ph, stdf ciplot(rline) color(gray) ///
|| lfitci glyco ph, lpattern(solid) ///
|| scatter glyco ph ///
, ylabel(0(50)300) xlabel(6.9 (.1) 7.5) ///
ytitle(Plasma Glycolate (mg/dl)) legend(off)

And here is the resulting graph:

[ATTACH=CONFIG]n1375578[/ATTACH]

The data file and Dupont's complete code can be found at the following links.

• http://biostat.mc.vanderbilt.edu/dup...2.12.Poison.do
• http://biostat.mc.vanderbilt.edu/dup....12.Poison.dta

Cheers,
Bruce

Bruce thanks for this that's really nice!
But if it's so easy to draw the limits in a graph, it should be easy to extract the limites for a specified X value.
Is it?
Thanks again very helpful thread!