Dear Stata users,
I would like to estimate a model of sequential choices which corresponds exactly to the kind of models we can estimate with the seqlogit command, which means a sequence of binary choices.
The problem is that I cannot use the seqlogit command as I have time-varying variables for each step, which can be included in the seqlogit command but only if you observe all of them at each step. This is not my case, as I have some dropouts at each step (this is actually what I want to estimate).
I was then wondering if I could use the cmp command to estimate the same model "manually", as the cmp command seems more flexible in terms of missing values, with a sequence of probit models.
Just to give you a practical idea of the model, I want to estimate the probability to renew one's loan, considering the financial characteristics of the project, which chage over time and which are observed when the client takes a loan.
Thank you for your advice,
Mathilde Bauwin
I would like to estimate a model of sequential choices which corresponds exactly to the kind of models we can estimate with the seqlogit command, which means a sequence of binary choices.
The problem is that I cannot use the seqlogit command as I have time-varying variables for each step, which can be included in the seqlogit command but only if you observe all of them at each step. This is not my case, as I have some dropouts at each step (this is actually what I want to estimate).
I was then wondering if I could use the cmp command to estimate the same model "manually", as the cmp command seems more flexible in terms of missing values, with a sequence of probit models.
Just to give you a practical idea of the model, I want to estimate the probability to renew one's loan, considering the financial characteristics of the project, which chage over time and which are observed when the client takes a loan.
Thank you for your advice,
Mathilde Bauwin