Re: st: From: Anna Zakharova <annazakharova87@googlemail.com>

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

Hi Anna,

Several thoughts here. If you can assume that being loyal is exogenous to the decision of upgrading but that the decision of upgrading depends n whether a customer is loyal or not, you can estimate a simple -probit- including loyal as an explanatory variable. You can also add randomness to the effects of loyalty in upgrading by running mixed probit estimation with loyal both in the equation and as the random cluster.

You can also create a new independent variable with four options: not loyal not upgrade, not loyal upgrade, loyal not upgrade, loyal upgrade. With these four options you can run a multinomial logit or probit to see how the different explanatory variables explain the different options. But even better for your purposes, I think, is that with the options that way, you can nest the options so you can run a nested logit model where the first level is whether the customer is loyal or not, and from each node of that decision you spread two more branches to reflect whether the client upgraded or not.

I hope this helps.

Alfonso Sanchez-Penalver

> On Mar 13, 2014, at 8:38 AM, owner-statalist@hsphsun2.harvard.edu wrote:
> 
> dear statalist
> 
> I have two equations with DVs customer's loyalty (1,0) and customer's
> upgrade (1,0), both equations having the same IVs, and I tried estimating
> the two equations simultaneously with -biprobit-. Now, I would like to
> know when do customers upgrade while being loyal. Which command do I
> need? is there any possibility to estmate both equations
> simultaneously, so that one is conditional on the other?
> Thanks in advance
> *
> *   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/