I am exercising the examples of the user packages betafit, zoib, dirifit and fmlogit, of Maarten Buis (more here), which are discussed in this presentation.
My interest is in fitting a fractional multinomial logit model but I fail to run the post-estimation command dfmlogit.
If you can repeat this code (I run Stata 15.1, ver. 10-15-2018):
It might make sense that the post-estimation command is incompatible with the latest version of Stata.
However, the ado-file dates: *! version 1.3.0 MLB 14Feb2017, which suggests to me that it should run in some version mode.
This I tried to do, like with:
but it will not work (on my system).
So, three questions:
My interest is in fitting a fractional multinomial logit model but I fail to run the post-estimation command dfmlogit.
If you can repeat this code (I run Stata 15.1, ver. 10-15-2018):
Code:
* Get the packages: ssc install dirifit , replace ssc install fmlogit , replace * Get the data file: use http://fmwww.bc.edu/repec/bocode/c/citybudget.dta, clear * Run the model fitting a Dirichlet distribution by maximum likelihood as well as the post-estimation command: dirifit governing safety education recreation social urbanplanning, /// mu(minorityleft noleft houseval popdens) ddirifit, at(minorityleft 0 noleft 0 ) // Results are omitted here, but all should run fine. * Run the model fitting a Dirichlet distribution by maximum likelihood as well as the post-estimation command: fmlogit governing safety education recreation social urbanplanning, /// eta(minorityleft noleft houseval popdens) dfmlogit, at(minorityleft 0 noleft 0 ) // Results are omitted here, but only the model runs on my system but not the post-estimation command, which prompts the user with an error statement (text is from the ado file): // dfmlogit only exists for older versions of Stata // use margins instead
However, the ado-file dates: *! version 1.3.0 MLB 14Feb2017, which suggests to me that it should run in some version mode.
This I tried to do, like with:
Code:
version 8: dfmlogit, at(minorityleft 0 noleft 0 )
So, three questions:
- Can anyone reproduce the same error prompt?
- Would anyone know if and how it is possible to run dfmlogit?
- And, if that is not possible, can someone show me how margins can give the same post-estimation results as dfmlogit would provide, as explained on pages 32 and 33 of the presentation.