Re: st: comparing regression coefficients across two models with the same dependent variables

[James Bernard <jamesstatalist@gmail.com>]

Thanks Nick,

The command you suggested solved the issue. But, may I check if the
-glm with the proper link function etc is a legitimate substitute for
-xtpoisson?

Thanks,
James

On Fri, May 3, 2013 at 7:04 PM, Nick Cox <njcoxstata@gmail.com> wrote:
> Try -glm, link(log) f(poisson)- instead.
>
> Many earlier comments (e.g. from David Hoaglin) still apply, it seems to me.
>
> Nick
> njcoxstata@gmail.com
>
> On 3 May 2013 11:18, James Bernard <jamesstatalist@gmail.com> wrote:
>
>> Thanks for your helpful suggestion. It seems -suest- is the
>> solution.... but I am getting errors:
>>
>> But before I elaborate: I got some info form this link and tried the
>> steps:     http://www.ats.ucla.edu/stat/stata/code/suest.htm
>>
>> I am using -suest- on a panel Poisson model
>>
>> I run model 1 and store its estimates using est store [name 1]
>>
>> then model 2 and store its estimates as [name2]
>>
>> and then using -suest  [name1] [name2]
>>
>> this is the error I get:
>>
>> unable to generate scores for model low
>> suest requires that predict allow the score option
>> r(322);
>>
>> Please note that low is the name of the stored estimates
>
>  On Thu, May 2, 2013 at 10:34 PM, John Antonakis <John.Antonakis@unil.ch> wrote:
>
>>>
>>> This is easily and correctly done with Stata's -suest- command. You can test
>>> for differences in coefficients across equations as well as sets of
>>> coefficients, including intercepts.
>
> On 02.05.2013 16:12, James Bernard wrote:
>
>>>> This may seem simple, but the seems to be no consensus on it. I have
>>>> two regressions with the same dependent variables
>>>>
>>>> I want to compare the same independent variable across these two models:
>>>>
>>>> Y=a+b0X+b1Z
>>>> Y=a+b0X+b1W
>>>>
>>>> Now, I want to compare b0 across these two models. Any idea if this
>>>> can be done in Stata?
>>>>
>>>> Comparing two coefficients in the same model is rather discussed and
>>>> clear, but how about the case I mentioned? (the reason I am doing this
>>>> is that I am splitting my sample instead of using interaction effects)
>>>>
> *
> *   For searches and help try:
> *   http://www.stata.com/help.cgi?search
> *   http://www.stata.com/support/faqs/resources/statalist-faq/
> *   http://www.ats.ucla.edu/stat/stata/
*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/