Re: st: specifying the nested design for xtmelogit

[Alfonso Sanchez-Penalver <alfonso.statalist@gmail.com>]

Hi Dana,

My first impression is that your specification would lead to biased estimates because there are bound to be crossed effects between the firms' effects and the industry effects. I wanted to check my Rabe-Hesketh, Skondral book of Multilevel and Longitudinal Modeling Using Stata (Stata Press) to suggest an appropriate specification but it is at home and I won't get out of the university until late at night.

I'll check on this later and get back to you if no-one else has done so before.

Best

Alfonso Sanchez-Penalver

> On Nov 6, 2013, at 12:38 PM, Dana Shills <shills52@hotmail.com> wrote:
> 
> My data is of the following format - I have multiple years of observations on firms in different industries. The industries are defined at a very dis-aggregated level so each firm is not associated with just one industry but with multiple industries and this classification doesn't change over time. For instance: Firm A is associated with 4 different industries and we have three years of data, so there are 12 observations on firm A. My dependent variable, Y, is  0/1 (Did the firm make an announcement in that industry in that year or not). X is a vector of covariates.
> 
> I am using the following nested random effects model accounting for the multiple observations per firm and the multiple firms in each industry.
> 
> xtmelogit Y X ||industry: ||firm: 
> 
> Is this correct given that each firm is not just associated with just one industry?
> 
> Thanks
> 
> Dana                         
> *
> *   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/