Dear friends,
My stata code is at the following:
---xtset Stkcd Year
---xtlogit Firm_Envir Post_Treat i.Year [iweight=Weight], fe
and I got an error message as follows:
---weights must be the same for all observations in a group
In fact, my "Weight" equals to "weight in the Stkcd level” times "weight in the Year level”.
I have tried command ---svy, but it cannot solve the problem.
And I have also searched for some other commands about logit, such like clogit, mlogit, cmclogit and so on. It seems that conditional logit and individual weight can not exist at the same time. Can someone kindly resolve the above issue? In fact, there may be some feasible directions.
First, Whether there is a command that can include two-way fixed effect and complex weight in logit regression?Command ---logit Firm_Envir Post_Treat i.Year i Stkcd [iweight=Weight] is not desirable because of bias.
Second, Whether there is a command that can include multidimensional fixed effect in logit regression? As I think, ---reghdfe and ---ppmlhdfe are multidimensional fixed effect in linear regression and possion regression respectively and they can also be used with complex weight. So, maybe a command like that is feasible.
Third, someone says ---reghdfe can be used in logit regression if it is multidimensional. Is that true? Is command ---reghdfe feasible in my regression?
This issue has borthered me for a long time.
Thank you very much.
My stata code is at the following:
---xtset Stkcd Year
---xtlogit Firm_Envir Post_Treat i.Year [iweight=Weight], fe
and I got an error message as follows:
---weights must be the same for all observations in a group
In fact, my "Weight" equals to "weight in the Stkcd level” times "weight in the Year level”.
I have tried command ---svy, but it cannot solve the problem.
And I have also searched for some other commands about logit, such like clogit, mlogit, cmclogit and so on. It seems that conditional logit and individual weight can not exist at the same time. Can someone kindly resolve the above issue? In fact, there may be some feasible directions.
First, Whether there is a command that can include two-way fixed effect and complex weight in logit regression?Command ---logit Firm_Envir Post_Treat i.Year i Stkcd [iweight=Weight] is not desirable because of bias.
Second, Whether there is a command that can include multidimensional fixed effect in logit regression? As I think, ---reghdfe and ---ppmlhdfe are multidimensional fixed effect in linear regression and possion regression respectively and they can also be used with complex weight. So, maybe a command like that is feasible.
Third, someone says ---reghdfe can be used in logit regression if it is multidimensional. Is that true? Is command ---reghdfe feasible in my regression?
This issue has borthered me for a long time.
Thank you very much.
