Dummy Variable Trap in Stata

Question

So basically I am using the wine.dta file and have a range of dummy variables for 6 regions. I have excluded 1 of the regions (regressing on 5 of the dummy variables) but for some reason Stata says there is still a collinearity problem and only gives 4 coefficients...

Why is there still a collinearity problem when I have already taken out one of the dummy variables to avoid the dummy variable trap.

You should paste in the commands you are using and resultant output. Particularly: tab region; xi. regress Y i.region — D L Dahly, Apr 01 '13 at 12:21
I'm voting to close this question as off-topic because the OP has not responded to comments and it has been years. — Peter Flom, Jul 17 '19 at 11:55

score 5 · Answer 1 · answered Apr 01 '13 at 14:28

There are many data sets that can be called "wine.dta." Just because you have used it does not mean we also know it. Please either provide more details on the data or a source where we can look at it if it's publicly available.

As for your problem, there could be many reasons. Your dummies might have been calculated incorrectly. Try add up your five chosen dummies and see if they do add up all to 1.

Another possibility is that you have a region that has no wine entry, causing it to contain only zero and subsequently thrown out from the regression model.

And lastly, in Stata, you do not need to compose your own dummies (unless you're told to do so by the assignment, etc.) You can use an "i." prefix to specify a variable being categorical:

sysuse census, clear
reg marriage i.region

Output:

. reg marriage i.region

      Source |       SS       df       MS              Number of obs =      50
-------------+------------------------------           F(  3,    46) =    0.21
       Model |  1.3626e+09     3   454209163           Prob > F      =  0.8874
    Residual |  9.8438e+10    46  2.1400e+09           R-squared     =  0.0137
-------------+------------------------------           Adj R-squared = -0.0507
       Total |  9.9801e+10    49  2.0368e+09           Root MSE      =   46260

------------------------------------------------------------------------------
    marriage |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
      region |
          2  |   2957.194   20398.63     0.14   0.885    -38103.13    44017.51
          3  |   10208.47    19274.9     0.53   0.599    -28589.89    49006.82
          4  |  -2900.991   20059.57    -0.14   0.886    -43278.82    37476.84
             |
       _cons |   44479.22   15419.92     2.88   0.006     13440.54    75517.91
------------------------------------------------------------------------------

Dummy Variable Trap in Stata

1 Answers1