Dear All,
I'm trying to run an ivprobit using ml. The binary dependent variable is success_dummy, I have some exogenous variable exo and endogenous variables event_dummy and time_since_event, which depend on random assignment to group.
The idea is that event_dummy taking the value of 1 might change the probability of success_dummy being 1 and i want to estimate by how much. But I want to allow for a possibility that this impact depends on time_since_event: the impact of event_dummy on success_dummy (or the underlying latent variable which determines if success_dummy is 1 or 0) is expected to decay exponentially with the passage of time_since_event.
So, i suppose I could assign some arbitrary value to event_time (e.g. 0) when event_dummy is 0 and use sth like
xi: ivprobit success_dummy exo (event_dummy event_time = i.group)
but then the effect of time_since_event on the latent variable would be linear. So, i'm trying to manipulate the ml code but i got stuck. Any advice? thanks in advance!
I'm trying to run an ivprobit using ml. The binary dependent variable is success_dummy, I have some exogenous variable exo and endogenous variables event_dummy and time_since_event, which depend on random assignment to group.
The idea is that event_dummy taking the value of 1 might change the probability of success_dummy being 1 and i want to estimate by how much. But I want to allow for a possibility that this impact depends on time_since_event: the impact of event_dummy on success_dummy (or the underlying latent variable which determines if success_dummy is 1 or 0) is expected to decay exponentially with the passage of time_since_event.
So, i suppose I could assign some arbitrary value to event_time (e.g. 0) when event_dummy is 0 and use sth like
xi: ivprobit success_dummy exo (event_dummy event_time = i.group)
but then the effect of time_since_event on the latent variable would be linear. So, i'm trying to manipulate the ml code but i got stuck. Any advice? thanks in advance!
