Dear users,
I am trying to estimate a multiple hurdle model using the cmp command by David Roodmand (Estimating fully observed recursive mixed-process models with cmp Stata J., 11 (2) (2011), pp. 159-206). This is my first time working on this command, and I am still learning how to implement it.
I have a three stage decision process: first stage is a decision to contribute (0/1), second stage is a decision to add to an initial contribution (intensity of contribution, could be 0, 1, 2 or 3) and last stage is the final contribution ($ amount).
I fit the following model:
Is this the right way of fitting the model?
I am interested in finding out the marginal effects of the last stage, say the impact of x3 on final contribution.
1. My question is how do I find the difference in average contribution of someone who responded and decided to add to his donation once versus someone who responded, and decided to add to his contribution twice.
2. If I used the following command:
, does it take in to account the decisions of the first two stages, or does it produce results as if the stages are independent of each other?
Thanks,
Anwesha
I am trying to estimate a multiple hurdle model using the cmp command by David Roodmand (Estimating fully observed recursive mixed-process models with cmp Stata J., 11 (2) (2011), pp. 159-206). This is my first time working on this command, and I am still learning how to implement it.
I have a three stage decision process: first stage is a decision to contribute (0/1), second stage is a decision to add to an initial contribution (intensity of contribution, could be 0, 1, 2 or 3) and last stage is the final contribution ($ amount).
I fit the following model:
Code:
cmp(respond = x1 x2 x3)(multiple_add = x2 x3 x4)(ln_contribution= x3 x4 x5), ind($cmp_probit $cmp_oprobit $cmp_cont) nonrtolerance vce(robust)
I am interested in finding out the marginal effects of the last stage, say the impact of x3 on final contribution.
1. My question is how do I find the difference in average contribution of someone who responded and decided to add to his donation once versus someone who responded, and decided to add to his contribution twice.
2. If I used the following command:
Code:
margins, dyex(*) predict(equation(ln_contribution) pr)
Thanks,
Anwesha
Comment