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

st: -ml display- behaves differently in Stata 8 & 9

From   Richard Williams <[email protected]>
To   [email protected]
Subject   st: -ml display- behaves differently in Stata 8 & 9
Date   Thu, 09 Jun 2005 19:29:01 -0500

Here is an esoteric problem & solution that may only affect me - but just in case others might be interested, I'll pass this along.

Every time I think my -gologit2- program is ready to be officially released, I find some weird problem. (The output from -gologit2- looks similar to the output from -mlogit-) Most recently, I noticed that the -or- (Odds Ratio) option was not working right in Stata 9, although it works fine in Stata 8. Specifically, the coefficients for the first equation would get exponentiated, but the coefficients for later equations did not. I thought this was a "bug" in Stata 9, but it turns out it is an undocumented "feature" (and unfortunately version control does not avoid the problem). Here is what tech support said:

Stata 8, -ml display- would just exponentiate all of the equations, causing
the user-programmer to use the -first- and/or -neq()- options.

In Stata 9, -ml- now looks for -e(k_eform)- to determine how many equations to
to exponentiate when one of the -eform()- type options are specified.  This
feature was added to facilitate the bootstrap or jackknife variance estimation

If `neq' contains the number of equations in your model, then just add the
following line in your -gologit2- command:

        ereturn scalar k_eform = `neq'
        ereturn local cmd gologit2

You have pointed out that we didn't really document this.
Thank you, we will try to fix this omission soon.
There have been a couple of instances now where user-written programs have been "broken" by Stata 9. Most of these seem to be post-estimation commands, and hence you have to use version control on the original estimation command to avoid the problem. In my case, you would think that version control would have avoided the problem, but it didn't; but luckily the fix was very simple.

I do think it would be nice if Stata had a FAQ for "un-breaking" programs that have been zapped by the changes in Stata 9. Version control alone isn't always doing the job.

Richard Williams, Notre Dame Dept of Sociology
OFFICE: (574)631-6668, (574)631-6463
FAX: (574)288-4373
HOME: (574)289-5227
EMAIL: [email protected]
WWW (personal):
WWW (department):

* For searches and help try:

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