Dummy Variable Regression
What are Dummy Variables
• Dummy variables are also known as binary variables that takes the
value of 0 and 1.
• These variables are important because they are used to incorporate
qualitative information or variables in a regression.
• The example of a qualitative variable is gender. It’s either you are a
male or female.
• The group that is assign a value of 0 is a benchmark group and the
benchmark group is always omitted in the regression
Faculty name - Economics and finance
Department name – Economics
Dummy Variable Regression
• A regression model that consist • The Example of an ANOVA
of dummy variables only as model
independent variables is called
ANOVA and ANOVA stands for
the Analysis of variance
• Where and are dummy variables
that takes the value of 0 and 1
• represents wage rate
Faculty name - Change on Master Slide (View>Slide Master)
Department name – Change on Master Slide (View>Slide Master)
Dummy Variable Regression cont’
Where and are dummy variables that takes the value of 0 and 1
Assuming that is a gender dummy and is a race dummy.
• We assign a value of 1 if a person is female and 0 if a person is male.
• We also assign a value of 1 if a person is African and 0 otherwise
Faculty name - Economics and finance
Department name – Economics
Dummy Variable Regression cont’
• The benchmark group on the above model is male that is non African.
which is usually known as an intercept parameter is called the mean or the
average for the benchmark group in the context of the dummy variables regression
represent mean differentials between gender
represent mean differentials between
race
Faculty name - Economics and finance
Department name – Economics
Dummy Variable Regression cont’
• Determining the overall mean value in the model
)
)
It turns out that the expectation of a constant is equal to the constant
and the expectation of the error term is zero therefore the mean value
will be:
Faculty name - Economics and finance
Department name – Economics
Dummy Variable Regression cont’
• The mean value for the benchmark group is:
By substituting the value of 0 that is assigned for the benchmark group
in the model we have:
The mean value of the benchmark group is equal to
Faculty name - Economics and finance
Department name – Economics
Dummy Variable Regression cont’
• The mean value for the female holding other factors fixed is:
By substituting the value of 1 that is assigned if a person is female in
the model we have:
The mean value for female is equal to
Faculty name - Economics and finance
Department name – Economics
Dummy Variable Regression cont’
• The mean value for the race dummy holding other factors fixed is:
By substituting the value of 1 that is assigned if a person is African in
the model we have:
The mean value for African is equal to
Faculty name - Economics and finance
Department name – Economics
Interpretation of the model
• Interpreting dummy variables differs from the normal interpretation
of the model with quantitative variables.
• Suppose we are given the following model:
Where: 1 if a person is female and 0 otherwise. 1 if a person is
African and 0 otherwise.
Faculty name - Economics and finance
Department name – Economics
Interpretation of the model cont’
• Ceteris paribus, females earn less than males by 0,20 or 20%
The overall wage rate for female is
• Ceteris paribus, Africans earn less than non African by 0,10 or (10%)
The overall wage rate for African is
Faculty name - Economics and finance
Department name – Economics
ANCOVA regression or model
• ANCOVA stands for the Analysis of the Covariance
• ANCOVA model is a model that consist of both qualitative and
quantitative variables and it is an extension of an ANOVA model
For example:
are dummy variables (qualitative variables)
INFL which stands for inflation is a quantitative variables with a normal
interpretation
Faculty name - Economics and finance
Department name – Economics
Interaction Effects Using Dummy
Variables
• Now we are going to make our analysis even more interesting by
including multiplicative effect of our dummy variables.
For example:
The description remain the same where: is a gender dummy 1 if a
person is female and 0 if male. is a race dummy 1 if a person is African
and 0 otherwise.
Faculty name - Economics and finance
Department name – Economics
Interaction Effects Using Dummy
Variables cont’
• Given the above model the benchmark group is male non African
• The mean value for female African is:
Faculty name - Economics and finance
Department name – Economics
Interpretation of the model
(0,36) (0,15)
Where F stands for female which is a gender dummy.
• Interpretation
Holding inflation constant, females that are African earn less by 0,8
compared to the benchmark group which is male non African
Faculty name - Economics and finance
Department name – Economics
Exercises
• =
Se for female = 4,29
Se for black = 12,71
Se for female.black= 18,15
What is the benchmark group on the above model08:35
(i) What is the estimated difference in sat between black male and non black male
(ii) What is the estimated difference in sat between black female and non black female
(iii) Is there a strong evidence that female.black should be included in this model?
Faculty name - Economics and finance
Department name – Economics
End of the Topic
Thank You!!!!
Faculty name - Economics and finance
Department name – Economics
