[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]
st: RE: change values of e(b)
At 09:39 04/12/02 +0000, Nick Cox wrote:
The best place to look is probably the Stata Programming Manual, under -[P]
estimates-. I suspect -estimates repost- is part of the solution that
Hung-Jen Wang is looking for.
> > Revising history is best left
> > to science fiction writers and, indeed,
> > historians.
> Well put, Nick. But if I may, I want to emphasize that the reason I
> wish to change values of e(b) is only for the purpose of
> exploring the
> data and properties of an estimator which is a function of the
> predicted values. Please spare me with the details. (I know
> what I am
> So, from a pure technical point of view, is it possible to change
> values of e(b)?
I believe not. You can read from it but not
write to it directly.
I attempted to indicate from a Stata point of view why this is
the case. e(b) is a (transient) historical record: providing
a way to change it directly would be providing a tool with
which users could get confused, directly or indirectly.
There can be good reasons for wanting to change -e(b)-, although it is
usually also appropriate, in these circumstances, also to change -e(V)-. An
example from my experience is using -glm- to calculate confidence intervals
for algebraic means and their ratios. I outlined how to do this in my
review, in The Stata Journal, of Hardin and Hilbe's "Generalized Linear
Models and Extensions" (The Stata Journal 2001; 1(1): 98-100), and I still
have the do-file to demonstrate this, just in case anybody is interested.
I hope this helps.
Lecturer in Medical Statistics
Department of Public Health Sciences
King's College London
5th Floor, Capital House
42 Weston Street
London SE1 3QD
Tel: 020 7848 6648 International +44 20 7848 6648
Fax: 020 7848 6620 International +44 20 7848 6620
or 020 7848 6605 International +44 20 7848 6605
Opinions expressed are those of the author, not the institution.
* For searches and help try: