Reason of error occurrence : omitted because of collinearity?

[Park Hwanggyu]

Hello, I am modelling ordered logit or ordered probit.
Result is like as following. y is the dependent variable, and x1~x5 are independent variables which are all categorial.

I don't have any idea why this 'omitted because of collinearity' comes out...
For now, I just think that my code is wrong, rather than the data because it is so small that I can check if there is any doubtful value.
Actually I've searched about this but even still don't know so with which that variable(x5) has collinearity.
I attached the data I used to calibrate the model. I would be very appreciate if you give me a hand...

[Carlo Lazzaro]

Park:
is this issue related to the so called "dummy trap" (https://en.wikipedia.org/wiki/Dummy_...(statistics))?

[Park Hwanggyu]

Carlo Lazzaro Hmm...I don't know. Definately some materials explaning 'dummy trap' are pointing out why I must not encode categorical variable into 0,1,2, but I am not sure if this problem is because of that.
I am about to do one-hot encode, but currently I don't know so how to do that practically.

[Carlo Lazzaro]

Park:
1) dummy trap is something related to regression machinery. Therefore, we cannot avoid to take the reference category (or whatever other level of our categorical variable) out;
2) -fvvarlist- notation does it for you automatically;
3) if you have a => three-level categotical variable and you do not use the -fvvarlist- -i.- prefix, Stata considers it as a continuos predictor (this is the possible reason why Stata does not return any message about collinearity if you do not ho -fvvarlist' notation.

[Park Hwanggyu]

Carlo Lazzaro Well, I got what you're saying. So I am trying to search about usage of 'fvvarlist', but no material tells about it so far...Isn't it a command?
I'm sorry but If my original code(which outputs above message) is

oprobit y i.(x1 x2 x3 x4 x5)

could you tell me how I can add it in here?

[Carlo Lazzaro]

Park:
1) see -help fvvarlist-;
2) your code adopts a possible way of including -fvvarlist- notation, that is equivalent to -i.x1- and so on, as you can see from the following toy-example (please also note that Stata automatically omits one level of the predictor to protect the regression from the so-called dummy trap):

. use https://www.stata-press.com/data/r17/fullauto.dta
(Automobile models)

. oprobit rep78 i.(foreign), allbase

Iteration 0:   log likelihood = -93.692061  
Iteration 1:   log likelihood = -79.126404  
Iteration 2:   log likelihood =  -78.99278  
Iteration 3:   log likelihood = -78.992642  
Iteration 4:   log likelihood = -78.992642  

Ordered probit regression                               Number of obs =     69
                                                        LR chi2(1)    =  29.40
                                                        Prob > chi2   = 0.0000
Log likelihood = -78.992642                             Pseudo R2     = 0.1569

------------------------------------------------------------------------------
       rep78 | Coefficient  Std. err.      z    P>|z|     [95% conf. interval]
-------------+----------------------------------------------------------------
     foreign |
   Domestic  |          0  (base)
    Foreign  |   1.721234   .3311954     5.20   0.000     1.072103    2.370365
-------------+----------------------------------------------------------------
       /cut1 |  -1.741284   .3225549                      -2.37348   -1.109088
       /cut2 |  -.8271202   .2027111                     -1.224427   -.4298138
       /cut3 |   .7298698    .196289                      .3451504    1.114589
       /cut4 |   1.847476   .2781064                      1.302397    2.392554
------------------------------------------------------------------------------

. oprobit rep78 i.foreign, allbase

Iteration 0:   log likelihood = -93.692061  
Iteration 1:   log likelihood = -79.126404  
Iteration 2:   log likelihood =  -78.99278  
Iteration 3:   log likelihood = -78.992642  
Iteration 4:   log likelihood = -78.992642  

Ordered probit regression                               Number of obs =     69
                                                        LR chi2(1)    =  29.40
                                                        Prob > chi2   = 0.0000
Log likelihood = -78.992642                             Pseudo R2     = 0.1569

------------------------------------------------------------------------------
       rep78 | Coefficient  Std. err.      z    P>|z|     [95% conf. interval]
-------------+----------------------------------------------------------------
     foreign |
   Domestic  |          0  (base)
    Foreign  |   1.721234   .3311954     5.20   0.000     1.072103    2.370365
-------------+----------------------------------------------------------------
       /cut1 |  -1.741284   .3225549                      -2.37348   -1.109088
       /cut2 |  -.8271202   .2027111                     -1.224427   -.4298138
       /cut3 |   .7298698    .196289                      .3451504    1.114589
       /cut4 |   1.847476   .2781064                      1.302397    2.392554
------------------------------------------------------------------------------

.

[Park Hwanggyu]

Carlo Lazzaro Thank you! I'll check it