Dear statalist,
I would like to estimate an endogenous switching regression model in stata where the "y" or outcome variable is binary. The "movestay" ado seems perfect for what I would like to do, but it appears to require a continuous outcome variable. Is there a way in stata to estimate an endogenous switching regression model when the outcome takes a standard 0/1 binary structure?
If useful, here's more detailed background on my study:
I am testing the effects of founders' prior experience on new venture survival when founders pursue either of two business models. I wish to account for the selection effects associated with business model choice, and then test for whether the prior experience of founders has effects on venture survival that vary between the two business model regimes. I have tried using a continuous outcome variable (days to survival), but the regression never converged due to non-concavity. Also, I tried to use the "difficult" command with "movestay" to facilitate convergence, but it was said to not be allowed. Finally, I ran "movestay" using two binary outcomes (for both the selection and survival regressions). This converged and produced results with face validity, however I presume they are biased and should not be reported if I am correct in believing that "movestay" requires a continuous outcome variable. Could you confirm this presumption, as well?
I found a similar question posted a few years ago, but I was unable to find a reply to it (Thu, 16 Sep 2004 14:12:53 -0700: Maclean, Aldair).
Sincerest thanks in advance for any help you might be able to offer!
Best,
- Chad
_________________________________________________________________
Get more out of the Web. Learn 10 hidden secrets of Windows Live.
http://windowslive.com/connect/post/jamiethomson.spaces.live.com-Blog-cns!550F681DAD532637!5295.entry?ocid=TXT_TAGLM_WL_domore_092008
*
* 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/
