[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: st: Re: using coefficients from model

From   Chris Witte <>
Subject   Re: st: Re: using coefficients from model
Date   Tue, 31 Mar 2009 09:44:46 -0700 (PDT)


Hmm, I think I understand where you are coming from, but by adjusting the constant in the equation, I consistently get the same values as _predict_... put another way, if I use the coefficients and constant given in the _fracpoly_ output, I get values that are always off of the _predict_ values by the same amount - and this can be fixed by adjusting the constant by the amount that I am off.  This seems like a half-arsed way of doing things, I know, but it does seem to work.

--- Chris Witte wrote:
> I want to report the regression equation in my manuscript 
> so the reader can plug in their independent variable values
> and get the correct estimate.  I thought that calculating
> the difference between the _fracpred_ estimated values and
> the values estimated using the coefficients from the 
> _fracpoly_ output would be the fastest way to find the 
> "real" constant for reporting purposes.  Is there a better
> (i.e. easier) way?  Am I misunderstanding the process?

Point is that -fracpoly- does not adjust the constant but
the covariates. I would actually report the formulas in 
a footnote or an appendix (together with the necessary
describtive statistics of the covariates, that are used 
in the formulas). This way your result will become less
of a "black box".

-- Maarten

Maarten L. Buis
Institut fuer Soziologie
Universitaet Tuebingen
Wilhelmstrasse 36
72074 Tuebingen


*   For searches and help try:

© Copyright 1996–2015 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index