Hello,
I was reading about the command "teffects" which allows for estimating the treatment effects in a quasi-experimental setup. However, I could not find any command to estimate, say multinominal logit, for the outcome model. In other words, the command "teffects" is limited to logit, probit, flogit, linear, nonnegative and survival outcome model estimation.
The following is an example of my problem:
Suppose, I am estimating the treatment effects of a job training programme provided to students from a particular university. I have done a quick survey to get information on the outcome variable (a) whether they are employed; and (b) kind of job they are employed in. Now, there will be three types of categorical values (a) blue collar job; (b) white collar job and (c) not employed.
I would like to run an estimation where my "Y" is the variable which contains the three categorical values [blue collar, white collar and not employed]. Now the command "teffects" does not have option to run such as model. (Note: My question is specifically with respect to the outcome model, and not about the treatment assignment model where "mlogit" is possible.)
Can you please guide me on the following:
(a) whether the command "teffects" allows for such a model? If so, what would be the command be ?
(b) If not, is there a command which allows me to estimate a mlogit for the outcome model ?
Thanks
I was reading about the command "teffects" which allows for estimating the treatment effects in a quasi-experimental setup. However, I could not find any command to estimate, say multinominal logit, for the outcome model. In other words, the command "teffects" is limited to logit, probit, flogit, linear, nonnegative and survival outcome model estimation.
The following is an example of my problem:
Suppose, I am estimating the treatment effects of a job training programme provided to students from a particular university. I have done a quick survey to get information on the outcome variable (a) whether they are employed; and (b) kind of job they are employed in. Now, there will be three types of categorical values (a) blue collar job; (b) white collar job and (c) not employed.
I would like to run an estimation where my "Y" is the variable which contains the three categorical values [blue collar, white collar and not employed]. Now the command "teffects" does not have option to run such as model. (Note: My question is specifically with respect to the outcome model, and not about the treatment assignment model where "mlogit" is possible.)
Can you please guide me on the following:
(a) whether the command "teffects" allows for such a model? If so, what would be the command be ?
(b) If not, is there a command which allows me to estimate a mlogit for the outcome model ?
Thanks
