Re: st: joint estimation of two equations using GLM

[Shehzad Ali <drshehzad_ali@yahoo.com>]

Thanks, Nick. Yes, my previous post was probably less specific in terms of the models; hence, I couldn't work out how best to implement the suggestions. I thought if I specified the actual equations I may get more specific response (I should have may be responded to the same thread). Apologies for double posting. However, I would still appreciate listers to consider this question. 

Many thanks
Shehzad



----- Original Message -----
> From: Nick Cox <njcoxstata@gmail.com>
> To: statalist@hsphsun2.harvard.edu
> Cc: 
> Sent: Friday, 4 November 2011, 9:55
> Subject: Re: st: joint estimation of two equations using GLM
> 
> You asked essentially the same question on 14 October and got various
> replies. I imagine that people following this matter would have the
> same suggestions and/or would expect that you would be reporting why
> those suggestions are not appropriate. It is difficult to see the
> answers would be different three weeks later.
> 
> Nick
> 
> On Fri, Nov 4, 2011 at 9:31 AM, Shehzad Ali <drshehzad_ali@yahoo.com> 
> wrote:
> 
>>  I am working with glm models for two dependent variables that may be 
> correlated. My equations are:
>> 
>>  glm y1 $xvars1, family(binomial) link(logit)
>> 
>> 
>>  glm y2 $xvars1, family(gamma) link(log)
>> 
>> 
>>  Is there a way that I could jointly estimate the two equations to allow for 
> correlation? I believe -sureg- estimates equations jointly but assumes normality 
> which is not tenable in this case. Hence, I would prefer to use glm but would 
> like to allow for correlation. Any help would be appreciated.
> 
> *
> *   For searches and help try:
> *  http://www.stata.com/help.cgi?search
> *  http://www.stata.com/support/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/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/