Regression Omission


I have created a regression model in stata. It contains 5 explanatory variables a set of 5 dummy variables. How would I go about finding out which one of the set has the greatest and least affect on the dependent variable?
The problem I have is that I omitted the first dummy of the set for the base case, to avoid the dummy variable trap.

Last edited:
When you omit the 1st dummy, say, d1 then the coefficient on the constant in your model is the coefficient on the 1st dummy. It is thus straightforward to compare the coefficients.

Personnaly, I keep all the dummies and drop the constant term (reg varlist, nocons) because I don't like writing things like "test _cons = d2" instead of "test d1 = d2"

Hope this helps

Ok thanks.

Further to that if I have 5 different countries. Country 1 is the omitted dummy and countries 2-5 are the dummies 2-5. The topic is stadium prices in each the 5 countries. I'm trying to work out which country has the most expensive stadiums in terms of price. (Price is dependent variable)

For country 2 would I run a regression with just country 1 and country 2?

If you run the following regression:

reg price d1-d5, nocons

then the coefficient on dummy di (where i = 1 .. 5) is the mean of price in country i. Then you can test whether d1 is statistically different from d2.