Empirical Methods II (API202A)
Kennedy School of Government
Harvard University
1
Lecture Notes 6
Nonlinear Relationships – Interactive Terms
Nonlinear regression can take on more than one form:
1.
Logs and Quadratics (last class)
−
The predicted change in Y associated with a change in X
1
depends on the
value of X
1
2.
Interaction between two variables – i.e. X1*X2 (dummies and/or nondummies)
−
The predicted change in Y associated with a change in X
1
depends on the
value of another variable
X
2
Example of dependence on another variable
: Does the data suggest that the relationship
between wages and years of experience is the same for men and women?
Wages
=
0
ˆ
1
ˆ
Female
+
2
ˆ
Education
+
ˆ
. reg wage female educ, robust
Regression with robust standard errors
Number of obs =
526
Rsquared
=
0.2588


Robust
wage 
Coef.
Std. Err.
t
P>t
[95% Conf. Interval]
+
female 
2.273362
.2702033
8.41
0.000
2.804179
1.742545
educ 
.5064521
.0598956
8.46
0.000
.3887867
.6241176
_cons 
.6228168
.7286843
0.85
0.393
.8086909
2.054324

Note:
To simplify the presentation we will use wages and not Log (wages) as LHS variable.
QUESTION #1
: Can we answer our question with this regression?
This preview has intentionally blurred sections. Sign up to view the full version.
View Full Document