Bookmark and Share

Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at

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

Re: st: xttobit & stata 12: Initial values not feasible

From   Surender Komera <[email protected]>
To   [email protected]
Subject   Re: st: xttobit & stata 12: Initial values not feasible
Date   Wed, 4 Jul 2012 12:09:42 +0530

Hi Maarten,
Thanks for your patience and your time. I centered the explanatory
variables (egen new_var =  old_var - mean(old_var)) and run the
'xttobit'. The problem is same, it is still throwing out with the same
error message "Initial values not feasible".
 Is there any alternative way to handle this issue?  I don't have any
clue why this problem arose when i divided the data into subsamples, i
could run the model with my complete dataset. Kindly guide me to
handle this issue.

Thanks - Surender
On Tue, Jul 3, 2012 at 11:35 PM, Maarten Buis <[email protected]> wrote:
> On Tue, Jul 3, 2012 at 6:07 PM, Surender Komera wrote:
>> Ok. I'm not sure i got it completely what you are saying. As you see i
>> don't have any binary variable or any time invariant one as
>> explanatory variables. All my explanatory variables are ratios and
>> they are continuous. I didn't clearly get it how i can center them?
> I do not understand what you do not understand. You can create new
> variables with -generate-. These new variables will be equal to the
> old variables minus some number that makes sure that the value 0 is
> within the range of the data. After that you just add those new
> variables instead of the old variables. You should not do this with
> categorical variables, but since you claim not to have them, that is
> not a problem. The fact that your variables are not time invariant is
> irrelevant.
>> I appreciate if you can site some source for reading to understand it clearly.
> Any introductory text on multilevel modeling will deal with this.
>> However, do you think it can
>> potentially offer some solution to my problem of "Initial values not
>> feasible".
> As I explained before: yes.
> -- Maarten
> --------------------------
> Maarten L. Buis
> Institut fuer Soziologie
> Universitaet Tuebingen
> Wilhelmstrasse 36
> 72074 Tuebingen
> Germany
> --------------------------
> *
> *   For searches and help try:
> *
> *
> *
*   For searches and help try:

© Copyright 1996–2018 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   Site index