Mixed -Logit Model Alternative-Specific Regressor Variation

[Juliana Lopex]

Hi all, I am trying to estimate Mixed Logit Model using fake insurance choice dataset "use https://www.stata-press.com/data/r18/inschoice" .
I tried to construct another alternative-specific variable that varies over alternatives but does not vary over cases.
Then, the estimation omits the new alternative-specific variable due to collinearity. I was wondering why collinearity occurs.

. tabulate insurance var1

           |                          var1
Insurances |      1000       2000       3000       4000       5000 |     Total
-----------+-------------------------------------------------------+----------
    Health |       250          0          0          0          0 |       250
     HCorp |         0        250          0          0          0 |       250
   SickInc |         0          0        250          0          0 |       250
    MGroup |         0          0          0        250          0 |       250
MoonHealth |         0          0          0          0        250 |       250
-----------+-------------------------------------------------------+----------
     Total |       250        250        250        250        250 |     1,250

[Hong Il Yoo]

Juliana Lopex: Does your model include alternative-specific constants? If yes, var1 is perfectly correlated with those constants: Note that your var1 does not vary within an alternative and only varies between alternatives.