I am testing the following probit model:
violence by state actor in a crisis (1,0) = duration of the actor's involvement in the crisis + democracy score + contiguity, + etc....
the observations are state actors in crises from 1918 to 1994.
the probit results indicate that duration has a strong positive impact on violence by the actor; i.e. the longer an actor is in a crisis, the greater the chances it will use violence.
the problem is that I suspect duration is endogenous. when I do a hazard model of duration, violence has a strong positive impact on duration.
so long crises lead to violence and violence leads to long crises.
how can I sort this out and make sense of these relationships? I want to do tests like Hausman and Smith-Blundell but will they work with duration data?
thanks in advance
[email protected]
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
