Excluding a category of a categorial variable for hausmann test (mlogit)

[Kim Schwartau]

Hi everyone,

I am conducting a Seemingly unrelated estimation for my mlogit model. I am doing it with a variable (Type_ENG_code) that has 4 categories (I coded them 1;2;3;4), in the first step I am trying to create 4 models, where one category is excluded for 4 models.

I am using this manual: https://www.stata.com/manuals13/rsuest.pdf

So in the first step I want to exclude category 1 and I am using this code:

mlogit Type_ENG_code NetZeroPledge ROA EMPL SlackR_ln firm_age i.Year i.industry if Type_ENG_code !="1":Type_ENG_code, base (2)
estimates store m1,title( Type_ENG_code!="1":Type_ENG_code)

the code works, but when I am looking at the table after running the logit, category one is still shown, in the table. I continued to do the same for the other categories except for the base outcome.

What can I do to actually exclude a category?
Thanks in andvance!

Kim

[Andrew Musau]

You just may be overcomplicating it.

if  Type_ENG_code!=1

[Clyde Schechter]

Amplifying Andrew Musau's excellent advice, -if Type_ENG_code != 1- will be the same as -if Type_ENG_code !="1":Type_ENG_code- if and only if the variable Type_ENG_code has a value label named Type_ENG_code and the value 1 is labeled with "1" therein.

Those conditions (suitably amended to the actual names in that example's data set) apply to the manual examples of -suest- that O.P. cites in #1, which may have prompted. But we don't know if they apply in O.P.'s data set. O.P. can find out by running -des Type_ENG_code- and -label list Type_ENG_code-.

[Kim Schwartau]

thanks for your help, I resolved the issue. I encountered another issue, following the aforementioned manual. The hausmann test I did turned out not well defined, so I tried to use suest, via suest m*,noomitted for my 4 models. Afterwards I used a series of the following command: test [m1_4=m2_4],cons with (varying models and categories of course) to compare all models. However, the chi2( 0) and Prob > chi2 values after each test remain empty. I cannot find my mistake.

Thanks again