Stata The Stata listserver
[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

st: Re: graph prediction of transformed variable from regression

From   Steve <[email protected]>
To   [email protected]
Subject   st: Re: graph prediction of transformed variable from regression
Date   Thu, 19 Aug 2004 06:36:08 -0500


Stata 8.2, exec: 23 Jul 2004; ado: 27 Jul 2004

I've run a multiple regression with -reg-.  One of the independent variables
is natural log transformed.  I want to plot the prediction of that variable
from the regression with 95% CIs along with the scatterplot of adjusted
residuals with the x-axis in the original untransformed units.  This should
plot as a trio of logarithmic curves.

I've no problem getting the adjusted residual scatterplot in the
untransformed units.  My problem is graphing the predicted curve and its 95%
CIs in the original units.

I've tried various combinations of code from the "books" at the UCLA site
without success.  -findit graph nonlinear- also located Jeroen
Weesie's -graphf- from Vol 10 of STB, which might do the trick except that
it always returns the error: "unrecognized command:  enumopt".  I take that
to be a command used internally in the ado file which cannot be run or
found.  The same error is returned when I prefix the command for Stata
7: -version 7: graphf exp exp-.

In lieu of a patch that will permit this command to work in current Stata,
can someone please suggest the steps necessary to manually make the partial
regression plot using the untransformed units of an initially transformed
independent variable in a multiple regression?

Thanks for your help.
Best regards,
Steve Rothenberg
Instituto Nacional de Salud P�blica
Cuernavaca, M�xico

*   For searches and help try:

© Copyright 1996–2024 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index